Kenneth C.

Mills
Carl-Åke Däcker

The Casting
Powders Book
The Casting Powders Book
Kenneth C. Mills Carl-Åke Däcker
•

The Casting Powders Book

123
Kenneth C. Mills Carl-Åke Däcker
Royal School of Mines Swerea KIMAB AB
Imperial College London Kista
London Sweden
UK

ISBN 978-3-319-53614-9 ISBN 978-3-319-53616-3 (eBook)

DOI 10.1007/978-3-319-53616-3
Library of Congress Control Number: 2017933543

© Springer International Publishing AG 2017

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein or
for any errors or omissions that may have been made. The publisher remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.

Printed on acid-free paper

This Springer imprint is published by Springer Nature

The registered company is Springer International Publishing AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Foreword

Swedish steelmakers have a long-standing commitment to product development.

This demands that improved skills must be developed in the steel shop, in order to
manufacture steel products free from defects and process problems. This is par-
ticularly important in the casting operation where it is necessary to optimise the
process parameters especially, for the new products under development. The mould
powder is the key factor affecting casting process control and product quality and
has attracted considerable research effort in Sweden and other parts of the world
over the last 15–20 years. However, many of the parameters affecting continuous
casting are interactive; thus, changing one casting parameter can have a knock-on
effect on other parameters. Furthermore, there is an absence of collated and
structured data for casting powders; this has proved a major obstacle to researchers
in steel plants, research institutes and universities. This is particularly apparent
when new researchers enter the complex world of mould powders and continuous
casting.
The Swedish Steel Producers identified the need for a comprehensive book on
continuous casting powders. This was discussed in 2012 at a meeting of the
Swedish Steel Producers’ Association by its CEO, Bo-Erik Pers, the Chairman
of the Technical Area on Casting and Solidification, Bo Rogberg, the research
manager, Lars-Henrik Österholm and Carl-Åke Däcker, Manager of Process &
Material Department, Swerea KIMAB. Following this meeting, Carl-Åke Däcker
was tasked with contacting Prof. Ken Mills (Imperial College) with a view to
writing a book on casting powders.
The project of writing a book on mould powders and slags was discussed at a
meeting in London in August 2012 between Ken Mills and Carl-Åke Däcker, who
have both participated in a significant number of ESCS and RFCS projects
involving mould powders. The idea of a combined effort was persuasive since it
would bring together Ken Mills’ knowledge of academic research on mould
powders with Carl-Åke Däcker’s more practical experience in Swedish steel plants.
The Swedish Steel Producers’ Association (Hugo Carlsson Foundation) provided
the necessary funding to take on the project and to provide supervision of the
progress by a reference group.

v
vi Foreword

The book is not meant to be read from binder to binder and each chapter is stand-
alone with its own abstract, introduction and reference list. For that reason, the
authors have tried to minimise the amount of searching the reader must do, through
the various chapters, for figures, tables, etc. However, providing all the relevant
information comes at the expense of some repetition.
Acknowledgements

We wish to thank the Swedish steelmakers and Swerea KIMAB for their continual
support. We also acknowledge the additional funding and enthusiastic support
of the Technical Area on Casting and Solidification (TO 24).
We also thank Swerea KIMAB, and its CEO, Staffan Söderberg, for funding and
support and the Materials Department, Imperial College, London for their support.
We would like to give special thanks to the reference group that was formed at the
beginning of the book project. They attended a number of meetings and provided,
support, critical discussion and important technical input throughout the project:

– Tomas Sohlgren and Anders Lagerstedt, (SSAB Special Steels in Oxelösund).

– Arashk Memarpour, (Sandvik Materials Technology).
– Fatemeh Shahbazian, (Swerea KIMAB).

We also wish to thank the following for their help:

• Klaus Schultz and Dirk Eckhard (Imerys) and Claudio Valadares and Omar
Afrange (Carbox, Brazil) for their inputs on Chap. 8 “Manufacture of mould
fluxes”.
• KCM would like to thank the late, Dr. Adrian Normanton, Vince Ludlow,
Dr. Shahid Riaz and Dr. Bridget Steward ( Tata Steel/Corus) for many infor-
mative discussions.
• Pavel Ramirez-Lopez (Swerea MEFOS), Peter Andersson (Swerea KIMAB),
Brian Thomas (University of Illinois), Masahiro Susa and Miyuki Hayashi
(Tokyo Institute of Technology), Qian Wang and Bing Xie (Chonquing
University), Masayuki Kawamoto and M Hanao (NSSMC), Koichi Tsutsumi
(JFE Steel Corporation), JW Cho (Gift, POSTECH) and Dr. Begona Santillana
(Tata Steel, IJmuiden) for providing important information and valuable dis-
cussions on various points arised during the project.
• Mariana Ursu Däcker for valuable discussions, information and literature
regarding silicate chemistry.

vii
viii Acknowledgements

Finally, Ken Mills would like to thank his wife Margaret for her constant love,
support and understanding throughout the project.
Kenneth C. Mills
Carl-Åke Däcker
Contents

1 Introduction and Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 The Continuous Casting Process for Steel . . . . . . . . . . . . . . . . 2
1.3 The Introduction of Casting Powders . . . . . . . . . . . . . . . . . . . . 4
1.4 Mould Powder Behaviour in the Mould . . . . . . . . . . . . . . . . . . 6
1.5 Slag Film and Slag Rim Characteristics . . . . . . . . . . . . . . . . . . 7
1.5.1 Slag Film . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.5.2 Slag Rim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.6 Casting Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.6.1 Casting Speed (Vc) . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.6.2 Metal Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.6.3 Mould Dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.6.4 Oscillation Characteristics . . . . . . . . . . . . . . . . . . . . . 12
1.6.5 Steel Grade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.6.6 Ar Flow Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.7 Physical Properties of Mould Slags . . . . . . . . . . . . . . . . . . . . . 13
1.8 Fluctuations in the Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.9 Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.9.1 Powders, Slags, Fluxes . . . . . . . . . . . . . . . . . . . . . . . 14
1.9.2 Powder Consumption Terms . . . . . . . . . . . . . . . . . . . 14
1.9.3 Temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.9.4 Viscosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2 Slag Infiltration, Lubrication and Frictional Forces . . . . . . . . . . . . . 19
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.2 Powder Consumption (Q) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.2.1 Various Powder Consumption Terms . . . . . . . . . . . . 22
2.2.2 Measurement of Powder Consumption . . . . . . . . . . . 23
2.2.3 Methods Used to Understand Slag Infiltration
Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 23

ix
x Contents

2.2.4 Problems Arising from Poor Powder

Consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.2.5 Optimum Casting Conditions . . . . . . . . . . . . . . . . . . 30
2.2.6 Factors Affecting Powder Consumption . . . . . . . . . . 32
2.3 Slag Infiltration During the Oscillation Cycle . . . . . . . . . . . . . . 41
2.4 Empirical Equations for Calculating Powder Consumption . . . 44
2.4.1 Frictional Forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.4.2 Factors Affecting Frictional Forces in the Mould . . . 48
2.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3 Heat Transfer in the Mould and Shell Solidification . . . . . . . . . . . . 59
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.1.1 Heat Flux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.2 Horizontal Heat Flux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.2.1 Heat Transfer Mechanisms Involved
in Horizontal Heat Transfer. . . . . . . . . . . . . . . . . . . . 63
3.2.2 Interfacial Thermal Resistance (RCu/Sl) . . . . . . . . . . . 65
3.2.3 Factors Affecting the Horizontal Heat Flux . . . . . . . . 71
3.2.4 Measurement and Calculation of Heat Fluxes . . . . . . 80
3.3 Shell Solidification and Growth . . . . . . . . . . . . . . . . . . . . . . . . 85
3.4 Variability in Heat Flux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.4.1 Variations in Heat Flux (qHor) During the
Oscillation Cycle . . . . . . . . . . . . . . . . . . . . . . . . . . .. 88
3.4.2 Thermal Gradient Variations Arising from Metal
Flow and Other Causes . . . . . . . . . . . . . . . . . . . . . . . 88
3.4.3 Mould Level Variations . . . . . . . . . . . . . . . . . . . . . . 91
3.4.4 Carbon Content of Steel . . . . . . . . . . . . . . . . . . . . . . 92
3.4.5 Thermal Gradients in the Mould . . . . . . . . . . . . . . . . 94
3.4.6 Fracture of Slag Films. . . . . . . . . . . . . . . . . . . . . . . . 95
3.5 Vertical Heat Flux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
3.5.1 Heat Transfer Mechanisms Involved in Vertical
Heat Transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
3.5.2 Factors Affecting Vertical Heat Transfer . . . . . . . . . . 97
3.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
4 How to Manipulate Slag Behaviour in the Mould . . . . . . . . . . . . . . 109
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
4.2 Vertical Heat Flux and Thermal Insulation of Bed . . . . . . . . . . 111
4.2.1 Vertical Heat Flux . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
4.2.2 Thermal Insulation of the Bed . . . . . . . . . . . . . . . . . 114
4.2.3 Measurements of Thermal Insulation of Powders . . . 116
4.2.4 Ways of Improving the Thermal Insulation
of the Bed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 116
Contents xi

4.3 Melting Rate of the Powder (QMR) . . . . . . . . . . . . . . . . . . . .. 116

4.3.1 The Effect of Mould Powder Properties
on Melting Rate . . . . . . . . . . . . . . . . . . . . . . . . . . .. 118
4.3.2 The Effect of Casting Conditions on Melting
Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
4.3.3 Ways of Increasing Melting Rate . . . . . . . . . . . . . . . 119
4.4 Depth of Molten Slag Pool . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
4.4.1 Molten Slag Pool . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
4.4.2 Importance of Depth of Molten Slag Pool . . . . . . . . 122
4.4.3 Factors Affecting Slag Pool Depth . . . . . . . . . . . . . . 122
4.4.4 The Effect of Casting Speed and Oscillation
Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 124
4.4.5 The Effect of Thermal Insulation of Bed
on Pool Depth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
4.4.6 Ways of Increasing the Melting Rate . . . . . . . . . . . . 125
4.5 Powder Consumption (Q) and Liquid Film Thickness (dl) . . . . 125
4.5.1 Reasons for Controlling Powder Consumption . . . . . 126
4.5.2 Factors Affecting Powder Consumption . . . . . . . . . . 127
4.5.3 Ways of Controlling the Powder Consumption . . . . . 131
4.6 Solid Slag Film and Horizontal Heat Flux . . . . . . . . . . . . . . . . 131
4.6.1 Reasons for Control of Slag Film Thickness
and Horizontal Heat Flux . . . . . . . . . . . . . . . . . . . .. 132
4.6.2 Factors Affecting of Slag Film Thickness
and Horizontal Heat Flux . . . . . . . . . . . . . . . . . . . .. 132
4.6.3 Measurement of Horizontal Heat Flux . . . . . . . . . .. 135
4.7 Crystallinity in Slag Film . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 135
4.7.1 Importance of Crystallinity to the Casting
Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
4.7.2 Factors Affecting fcrys . . . . . . . . . . . . . . . . . . . . . . . . 137
4.7.3 Ways of Increasing Crystallinity in Slag Film . . . . . . 139
4.8 Delaying Solidification and Shortening the Length of Shell . . . 139
4.8.1 Factors Affecting Shell Length . . . . . . . . . . . . . . . . . 140
4.8.2 Ways of Controlling the Length
of Meniscus/ Shell . . . . . . . . . . . . . . . . . . . . . . . . .. 140
4.9 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 142
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 142
5 Effect of Casting Variables on Mould Flux Performance . . . . . . . . 147
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
5.2 Mould Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
5.2.1 Mould Dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . 148
5.2.2 Mould Length (Lmould) . . . . . . . . . . . . . . . . . . . . . . . 150
5.2.3 Mould Taper (Lmould) . . . . . . . . . . . . . . . . . . . . . . . . 150
5.2.4 Mould Coatings . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
xii Contents

5.3 CastingSpeed (Vc) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 150

5.3.1 Effect of Casting Speed on Powder Consumption . .. 151
5.3.2 Effect of Casting Speed on Heat Transfer . . . . . . . .. 151
5.3.3 Effect of Casting Speed on Metal Flow
Turbulence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 152
5.3.4 Effect of Casting Speed on Negative Strip Time . . .. 152
5.4 Oscillation Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . .. 153
5.4.1 Effect of Oscillation Characteristics on Powder
Consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 154
5.4.2 Effect of Oscillation Characteristics on Heat Flux . .. 155
5.4.3 Effect of Oscillation Characteristics
on Oscillation Mark Depth (DOM) . . . . . . . . . . . . . . . 156
5.5 Mould-Level Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
5.6 Metal Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
5.7 Fluctuations in Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
5.8 Application of Electromagnetic Devices . . . . . . . . . . . . . . . . . . 162
5.8.1 Electromagnetic Stirring (EMS) . . . . . . . . . . . . . . . . 162
5.8.2 Level Magnetic Field (LMF) . . . . . . . . . . . . . . . . . . . 163
5.8.3 Electromagnetic Casting (EMC) . . . . . . . . . . . . . . . . 164
5.8.4 Electromagnetic Braking (EMBr) . . . . . . . . . . . . . . . 165
5.9 Steel Grade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
5.9.1 Peritectic Steels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
5.9.2 High-Al Steels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
5.10 Water Flow Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
5.11 Argon Flow Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
6 Different Types of Mould Powders . . . . . . . . . . . . . . . . . . . . . . . . . . 177
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
6.1.1 Functions Carried Out by Mould Powder . . . . . . . . . 179
6.1.2 Criteria Affecting Selection of Mould Powders . . . . . 180
6.2 Selection of Mould Fluxes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
6.2.1 Conventional Mould Powders . . . . . . . . . . . . . . . . . . 187
6.2.2 Pre-melted Fluxes . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
6.2.3 Starter Powders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
6.2.4 Exothermic Fluxes . . . . . . . . . . . . . . . . . . . . . . . . . . 196
6.2.5 Fluoride-Free Powders . . . . . . . . . . . . . . . . . . . . . . . 198
6.2.6 Reduced F-Powders . . . . . . . . . . . . . . . . . . . . . . . . . 205
6.2.7 C-Free Powders. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
6.2.8 Powders for High-Speed Casting and Thin
Slab Casting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
6.2.9 Powders for Casting Round Billets . . . . . . . . . . . . . . 208
6.2.10 Powders for Casting Beam Blanks . . . . . . . . . . . . . . 208
6.2.11 Non-Newtonian Powders. . . . . . . . . . . . . . . . . . . . . . 209
Contents xiii

6.2.12 Powders for Casting TRIP and TWIP Steels . . . . . . . 210

6.2.13 Powders for Casting Stainless Steels . . . . . . . . . . . . . 217
6.2.14 Powders for Casting Steels with Rare Earths . . . . . . 218
6.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
7 Fluxes for Ingot Casting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
7.1 The Ingot Casting Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
7.1.1 Classification of Ingot Cast Steels . . . . . . . . . . . . . . . 225
7.1.2 Ingot Casting of Killed Steels . . . . . . . . . . . . . . . . . . 228
7.2 Aspects of Importance for Ingot Casting Quality . . . . . . . . . . . 231
7.2.1 Surface Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
7.2.2 Inner Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
7.2.3 Macro Segregation (Hot Top Insulation) . . . . . . . . . . 242
7.3 History of the Development of Mould Powders
for Ingot Casting (and CC) . . . . . . . . . . . . . . . . . . . . . . . . . .. 245
7.3.1 Development of Mould Powders for Continuous
Casting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 250
7.3.2 Development of Synthetic Mould Powders . . . . . . .. 251
7.3.3 Development of Granulated Powders . . . . . . . . . . .. 252
7.3.4 Today’s Situation Regarding Mould Powders
for Ingot Casting . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
7.4 Selection of Mould Powders for Ingot Casting. . . . . . . . . . . . . 253
7.4.1 Important Properties of the Mould Powder . . . . . . . . 253
7.4.2 Important Properties of the Mould Powder Slag . . . . 255
7.4.3 Selection of Mould Powders in Regard to Steel
Grade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 258
7.5 Application Techniques for Mould Powders. . . . . . . . . . . . . .. 262
7.6 Use of Mould Powders to Minimise Defects and Process
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
7.6.1 Laps and Ripple Marks . . . . . . . . . . . . . . . . . . . . . . . 266
7.6.2 Entrapped Oxides . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
7.6.3 Slag Patches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
7.6.4 Porosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
7.6.5 Cracks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
7.6.6 Bottom-End Defects . . . . . . . . . . . . . . . . . . . . . . . . . 268
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
8 Manufacture of Mould Fluxes . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 271
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 271
8.2 Raw Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 272
8.2.1 Selection of Carbon Additions to Mould
Powders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 274
8.2.2 Reactions During Melting and Cooling of Mould
Powders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 275
xiv Contents

8.3 Manufacturing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276

8.4 Quality Control at the Manufacturer . . . . . . . . . . . . . . . . . . . . . 280
8.5 Information Provided by the Manufacturer . . . . . . . . . . . . . . . . 280
8.6 Delivery Control by the Steel Makers . . . . . . . . . . . . . . . . . . . 282
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
9 Properties of Mould Fluxes and Slag Films . . . . . . . . . . . . . . . . . .. 285
9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 287
9.2 Structure of Slags . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 287
9.2.1 Effect of Individual Slag Components
on Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
9.2.2 Parameters to Represent the Structure of Slags . . . . . 292
9.2.3 Effect of Cations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
9.2.4 Effect of Temperature on Properties . . . . . . . . . . . . . 297
9.3 Crystallisation in Mould Fluxes . . . . . . . . . . . . . . . . . . . . . . . . 299
9.3.1 Importance of Crystallisation to the Process . . . . . . . 299
9.3.2 Crystalline Phases Formed in Slag Films . . . . . . . . . 300
9.3.3 Crystallisation Process. . . . . . . . . . . . . . . . . . . . . . . . 303
9.3.4 Crystallisation Kinetics . . . . . . . . . . . . . . . . . . . . . . . 305
9.3.5 Effects of Crystallisation . . . . . . . . . . . . . . . . . . . . . . 308
9.3.6 Methods of Determining Fraction of Crystalline
Phase in Slag Films . . . . . . . . . . . . . . . . . . . . . . . .. 310
9.3.7 Tests to Simulate fcrys Formed in Slag Film . . . . . .. 313
9.3.8 Empirical Rules to Calculate the Crystal Fraction
in Slag Films. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 313
9.3.9 Data for fcrys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 314
9.4 Physical Properties of Mould Slags . . . . . . . . . . . . . . . . . . . .. 315
9.4.1 Thermodynamic Properties and Liquidus
Temperatures (Tliq) . . . . . . . . . . . . . . . . . . . . . . . . . . 315
9.4.2 Break Temperature (Tbr) . . . . . . . . . . . . . . . . . . . . . . 318
9.4.3 Glass Transition Temperatures (Tg) . . . . . . . . . . . . . . 320
9.4.4 Viscosities (η) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
9.4.5 Thermal Conductivities . . . . . . . . . . . . . . . . . . . . . . . 327
9.4.6 Interfacial Tension (cmsl) and Surface
Tension (cs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 341
9.4.7 Density (q) and Thermal Expansion
Coefficient (a) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 350
9.4.8 Heat Capacity (Cp) and Enthalpy (HT–H298) . . . . . . . 354
9.5 Optical Properties of Mould Slags . . . . . . . . . . . . . . . . . . . . . . 360
9.5.1 Refractive Indices (n) [53, 55, 206, 278, 279] . . . . . 361
9.5.2 Absorption Coefficients (a*)
[53, 55, 56, 59, 110, 206, 211, 212, 280, 281] . . . .. 361
9.5.3 Reflectivity, Transmissivity and Emissivity . . . . . . .. 362
Contents xv

9.6 Thermomechanical Properties of Mould Slags . . . . . . . . . . . .. 363

9.6.1 Thermomechanical Tests . . . . . . . . . . . . . . . . . . . . .. 363
9.6.2 Stress Relaxation . . . . . . . . . . . . . . . . . . . . . . . . . . .. 364
9.7 Dissolution of Oxides, Nitrides and Carbides in Mould
Slags . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 364
9.7.1 Origin of Inclusions . . . . . . . . . . . . . . . . . . . . . . . .. 365
9.7.2 Mechanism of Inclusion Removal . . . . . . . . . . . . . .. 366
9.7.3 Transport of Inclusions to the Slag/Metal
Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 366
9.7.4 Transport Through Slag/Metal Interface . . . . . . . . . . 369
9.7.5 Dissolution of Inclusions . . . . . . . . . . . . . . . . . . . . . . 370
9.8 Other Tests Used on Mould Powders . . . . . . . . . . . . . . . . . . . . 372
9.8.1 Bulk Density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 372
9.8.2 Flowability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373
9.8.3 Permeability Index . . . . . . . . . . . . . . . . . . . . . . . . . . 374
9.8.4 Thermal Insulation . . . . . . . . . . . . . . . . . . . . . . . . . . 374
9.8.5 Measurement of Moisture and Hydrogen . . . . . . . . . 376
9.9 Comparison of Properties of Powders Used in Ingot- (IC)
and Continuous Casting (CC). . . . . . . . . . . . . . . . . . . . . . . . .. 376
9.9.1 Differences in Properties of Mould Powders
Used in CC and IC . . . . . . . . . . . . . . . . . . . . . . . . .. 377
9.9.2 Tasks Carried Out by Powders Used in
Continuous- and Ingot Casting . . . . . . . . . . . . . . . .. 379
9.9.3 Properties and Characteristics of Powders
Used in Continuous and Ingot Casting . . . . . . . . . .. 379
9.9.4 Conclusions from Comparison of CC and IC
Mould Powders . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 379
9.10 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 382
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 383
10 Selection of Mould Fluxes and Special Mould Fluxes
for Continuous Casting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 393
10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 394
10.2 Selection of Mineral Compositions of Mould Powder
for Given Casting Conditions . . . . . . . . . . . . . . . . . . . . . . . . .. 395
10.2.1 Effect of Mould Geometry on Mould Powder
Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 396
10.2.2 Effect of Casting Conditions on Mould Powder
Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 398
10.2.3 Effect of Steel Grade on Mould Powder
Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 399
10.2.4 Routines to Differentiate Between Steel Grades . . .. 400
10.2.5 Plots of Tbr as a Function of Slag Viscosity . . . . . .. 403
10.2.6 Other Casting Conditions Affecting Powder
Consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 404
xvi Contents

10.3 Selection of Carbon Components of Mould Powders . . . . . . . . 405

10.4 Mould Powder Selection for Special Conditions . . . . . . . . . . . 406
10.4.1 Thin-Slab Casting . . . . . . . . . . . . . . . . . . . . . . . . . . . 408
10.4.2 Round Billets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409
10.4.3 Mould Powder Selection for Moulds
with Large “Corner” Regions . . . . . . . . . . . . . . . . . . 409
10.4.4 Casting High-Al (Trip, Twip) Steel Grades . . . . . . . . 410
10.4.5 Fluoride-Free Powders . . . . . . . . . . . . . . . . . . . . . . . 411
10.4.6 Reducing SEN Erosion Rates . . . . . . . . . . . . . . . . . . 412
10.4.7 Minimising Carbon Pick-up . . . . . . . . . . . . . . . . . . . 412
10.4.8 Minimising Scale Formation . . . . . . . . . . . . . . . . . . . 413
10.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
11 Using Mould Fluxes to Minimise Defects and Process
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 417
11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 418
11.2 Longitudinal Cracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 418
11.2.1 Type of Steel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 419
11.2.2 Heat Flux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421
11.2.3 Lubrication and Powder Consumption . . . . . . . . . . . 425
11.2.4 Metal Flow, Use of EMBr, EMC and EMS . . . . . . . 425
11.2.5 Causes and Mechanisms . . . . . . . . . . . . . . . . . . . . . . 426
11.2.6 Ways of Dealing with Longitudinal Cracking . . . . . . 428
11.3 Longitudinal Corner Cracking . . . . . . . . . . . . . . . . . . . . . . . . . 430
11.3.1 Published Information . . . . . . . . . . . . . . . . . . . . . . . . 430
11.3.2 Causes, Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . 430
11.3.3 Ways of Dealing with Longitudinal Corner
Cracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 433
11.4 Sticker Breakouts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434
11.4.1 Factors Affecting Sticker Breakouts . . . . . . . . . . . . . 435
11.4.2 Causes, Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . 442
11.4.3 Ways of Dealing with Sticker Breakout . . . . . . . . . . 443
11.5 Oscillation Marks (OM’s) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 446
11.5.1 Characteristics of Oscillation Marks . . . . . . . . . . . . . 446
11.5.2 Mould Oscillation . . . . . . . . . . . . . . . . . . . . . . . . . . . 448
11.5.3 Factors Affecting Depth of OM’s (DOM) . . . . . . . . . . 449
11.5.4 Causes, Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . 456
11.5.5 Ways of Dealing with Deep OMs . . . . . . . . . . . . . . . 459
11.6 Transverse and Corner Cracking . . . . . . . . . . . . . . . . . . . . . . . 461
11.6.1 Factors Affecting Transverse Cracking . . . . . . . . . . . 463
11.6.2 Ways of Dealing with Transverse and Corner
Cracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 467
Contents xvii

11.7 Star Cracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469

11.7.1 Factors Affecting Star Cracking . . . . . . . . . . . . . . . . 469
11.7.2 Ways of Dealing with Star Cracking. . . . . . . . . . . . . 471
11.8 Depressions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471
11.8.1 Longitudinal Depressions . . . . . . . . . . . . . . . . . . . . . 471
11.8.2 Transverse Depressions . . . . . . . . . . . . . . . . . . . . . . . 476
11.8.3 Off-Corner Depressions . . . . . . . . . . . . . . . . . . . . . . . 477
11.9 Overflows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 479
11.9.1 Factors Affecting Overflows . . . . . . . . . . . . . . . . . . . 479
11.9.2 Causes, Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . 480
11.9.3 Ways of Dealing with C-Type Effects. . . . . . . . . . . . 480
11.10 Slag, Gas Entrapment and Sliver Formation . . . . . . . . . . . . . . . 480
11.10.1 Metal Flow Conditions Leading to Entrapment . . . . . 481
11.10.2 Slag Entrapment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
11.10.3 Gas Entrapment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 500
11.10.4 Inclusion Capture, Sliver Formation . . . . . . . . . . . . . 506
11.11 Formation of Scales . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 511
11.11.1 Factors Affecting Scale Formation . . . . . . . . . . . . . . 512
11.11.2 Causes, Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . 513
11.11.3 Ways of Dealing with Scaling . . . . . . . . . . . . . . . . . 515
11.12 Carbon Pick-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515
11.12.1 Factors Affecting Carbon Pick-up . . . . . . . . . . . . . . . 515
11.12.2 Causes, Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . 516
11.12.3 Ways of Dealing with Carbon Pick-up . . . . . . . . . . . 517
11.13 SEN Erosion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 519
11.13.1 Factors Affecting SEN Erosion Rates . . . . . . . . . . . . 520
11.13.2 Causes, Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . 522
11.13.3 Ways of Dealing with SEN Erosion . . . . . . . . . . . . . 524
11.14 Fluorine Emissions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 525
11.14.1 Factors Affecting Fluoride Emissions . . . . . . . . . . . . 525
11.14.2 Ways of Dealing with Fluoride Emissions . . . . . . . . 526
11.15 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 526
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 528
Chapter 1
Introduction and Overview

Abstract This chapter provides an introduction into the use of casting powders in
the continuous casting of steel. It includes brief descriptions of the following:
(i) The continuous casting process for steel;
(ii) The various changes that the mould powder undergoes as it transforms into a
liquid slag and then into a slag film;
(iii) The various tasks carried out by the mould powder and the slag formed from
it;
(iv) The factors affecting the formation and performance of slag films and slag
rims;
(v) The effect of changes in casting conditions (a) casting speed, (b) metal flow,
(c) mould dimensions, (d) oscillation characteristics, (e) steel grades, (f) Ar
flow rate on powder consumption, heat flux, etc. and explains how mould
powder can be modified to deal with these changes;
(vi) The important physical properties of mould slags and
(vii) It points out how the continuous casting mould is in a continual state of
fluctuation and how this affects casting performance.
Finally, definitions are given of various terms used in this book.

Symbols, Abbreviations and Units

A Area (m2)
dl Thickness of liquid slag film (m)
ds Thickness of solid slag film (m)
F Friction force (N)
f Frequency (Hz or cycles min−1)
f* Fraction casting powder forming slag
qhor Horizontal heat flux (Wm−2)
QMR Melting rate (kg min−1 or kgs−1)
Qs Powder consumption (kgm−2)
R* (surface area/volume) of mould
s Stroke length (m)

© Springer International Publishing AG 2017 1

K.C. Mills and C.-Å. Däcker, The Casting Powders Book,
DOI 10.1007/978-3-319-53616-3_1
2 1 Introduction and Overview

T Temperature (oC)
Tbr Break (or solidification) temperature
t Thickness (mould, m) or time (sec, min)
tcycle Time for one cycle (s)
tn Negative strip time (s)
tp Positive strip time (s)
Vc Casting speed (m min−1)
Vm Velocity of mould (m min−1 or ms−1)
w Width of mould (m)
η Slag viscosity (dPas)
q Density (kg m−3)
C/S Basicity = %CaO/%SiO2
SEN Submerged entry nozzle

Superscripts
powd
Refers to powder
slag
Slag formed from powder

1.1 Introduction

The continuous casting process for steel is a highly successful process; more than
90% of the world steel production (>1 billion tonne p.a.) is cast using this process.
It will be seen below that the performance of the casting powder has been one of the
key factors in the success of the continuous casting process.

1.2 The Continuous Casting Process for Steel

The continuous casting process is shown in Fig. 1.1. Molten steel from the ladle
pours through a submerged nozzle into the tundish; the tundish provides a constant
head for molten steel to flow into the mould. The steel passes into the mould via a
submerged entry nozzle (SEN). The copper mould is water-cooled and the bottom
of the mould is initially sealed by a steel dummy bar. Thus, steel freezes against the
cold, mould wall and forms a shell of solidified steel, the name, shell, comes from
its relation to an egg shell since it consists of a thin, solid containing a liquid inside.
The dummy bar is then removed and the shell is pulled gradually through the mould
(where it continually thickens as it moves gradually down the mould) and then
passes into a spray chamber (where it thickens even more). When the shell is thick
1.2 The Continuous Casting Process for Steel 3

Fig. 1.1 Schematic diagram

showing the continuous
casting process (courtesy of
AB Fox)

enough, the slab is bent. Later, when the steel is completely solidified, the slab is
cut off by an oxy-acetylene torch. The casting speed (Vc) (i.e. the rate of withdrawal
of the steel strand) is a key factor affecting the performance of casting powders.
However, the shell would stick to the mould unless action is taken to prevent it
sticking. Two measures are taken to prevent the shell from sticking to the mould.
The first measure involves oscillating the mould vertically, and typical values for
the stroke length (s) are 5–8 mm and frequencies (f) are ca. 2 Hz (or 120 cpm) [1,
2]. However, in practice, values of Vc, f and s are selected using Eq. 1.1 [3] so as to
maintain a specific negative strip time (tn) (i.e. the time when the mould is
descending faster than the shell (Vc)). The positive strip time (tp) constitutes the
remaining portion of the cycle time (tcycle) as shown in Eq. 1.2. This action is due to
the fact that oscillation marks are formed on the steel surface and the depths of these
marks (dOM) are linearly related to tn and transverse cracking tends to increase with
increasing dOM:
   
60 Vc
tn ¼  arc cos ð1:1Þ
pf psf

tp þ tn ¼ tcycle : ð1:2Þ

The second measure taken to prevent sticking of the shell to the mould was to
feed oil, or casting powders, to the top of the mould. Liquid (oil or mould slag)
infiltrates between the shell and the mould and provides some lubrication to the
moving shell.
The oil and mould powder are expected to perform the following tasks:
1. To prevent the steel surface from oxidising;
2. To provide thermal insulation to the top of the steel meniscus and thereby
prevent it from freezing;
3. To infiltrate between the shell and the mould and provide lubrication to the shell
and
4 1 Introduction and Overview

4. To control the heat extraction from the steel shell.

Initially, rape seed oil was fed continuously onto the top of the mould; the oil
helps to create a reducing atmosphere and prevents the oxidation of the molten
steel. The heat given out during the combustion of the oil provides some thermal
insulation of the steel surface. Furthermore, a thin oil layer infiltrates between the
steel shell and the mould; this helps to prevent the shell from sticking to the mould.
This oil film evaporates and forms a thin gaseous layer between shell and mould
which provides a thermal barrier to heat extraction from the shell to the mould.

1.3 The Introduction of Casting Powders

Mould powders were first used in bottom-pouring, ingot casting in Belgium in 1958
and were, subsequently, applied to continuous casting in 1963 (see Sect. 7.4).
These mould powders were based on fly ash, a waste product from power stations,
which contained high contents of SiO2, Al2O3 and unburnt carbon to which CaO
and various fluxes (e.g. Na2O and CaF2) were added. The powders heat up as they
descend the mould and eventually melt to form a slag pool (Fig. 1.2). This slag pool
provides a reservoir of liquid slag which infiltrates between the shell and the mould
during the period when the mould is descending (which occurs throughout tn and
into early tp). This liquid slag film partially solidifies to form a slag film, consisting
of a solid layer (1–2 mm thick) and a liquid layer (ca. 0.1 mm thick) [4]. The initial
slag film is predominantly glassy (because of the high cooling rate) but over time,

Fig. 1.2 Schematic diagram

showing various phases and
their location in the mould;
note the solid layer of the slag
film is shown as grey and the
liquid layer (and the slag
pool) as white (courtesy of
Ramirez-Lopez [5])
1.3 The Introduction of Casting Powders 5

the fraction of crystalline phases (fcrys) gradually increases until it reaches a

steady-state value. The thickness of the liquid slag film (dl) determines the lubri-
cation supplied to the shell and the thickness of the solid layer (ds) and fcrys control
the amount of heat extracted from the shell [4].
Casting powders were preferred over oil because they provided better thermal
insulation of the steel surface and allowed steelmakers to operate with significantly
lower levels of superheat (DT) for the steel. The slag films formed between shell
and mould also provided better control of the horizontal heat extracted from the
shell (cf. the gaseous layer formed in oil casting). Furthermore, the liquid slag pool
sealed off the steel from the atmosphere and prevented oxidation. The slag pool also
provided an extra benefit (cf. oil casting) by absorbing some of the non-metallic
inclusions (e.g. Al2O3) in the molten steel.
The level of lubrication provided by the liquid slag layer is usually described in
terms of the powder consumption, Qs in kg (slag) m−2 (of mould), which is equal to
q.dl where q is the density of liquid slag. The required powder consumption also
increases with increasing surface area of the shell (or mould). Thus, Qs must be
adjusted for the mould surface area; this adjustment is achieved by control of the
melting rate (QMR, which is related to Qs). In practice, the melting rate is adjusted
by the content (%) and size of the free carbon in the powder; thus, typical Cfree
values for slabs and billets (with large and small surface areas) are 4% and 25%,
respectively.
There is considerable serendipity associated with the use of fly ash, since not
only were the powder manufacturers paid to remove the fly ash but the unburnt
carbon in the fly ash aided the control of the melting rate and the formation of a
reducing atmosphere.
The constituents of mould powders can be divided into four classes: (i) network
formers (SiO2, Al2O3), (ii) network breakers (CaO, MgO), (iii) Fluxes (Na2O, K2O,
CaF2, B2O3) and (iv) Carbon particles with different sources and size (e.g. lamp-
black, coke breeze, etc.). In addition, mould powders contain (i) some minerals
which are present as carbonates and these decompose on heating giving off CO2(g)
and (ii) some impurities present in the minerals (e.g. TiO2, FeO). The minerals and
carbon particles are mixed into a slurry and are then spray-dried to form granules.
More details on powder manufacture are given in Chap. 8.
Some physical properties (such as viscosity and thermal conductivity) are very
dependent upon the degree of polymerisation of the slag which increases with
increasing SiO2 and Al2O3 contents and decreasing amounts of CaO, MgO, CaF2
and Na2O (see Chap. 9). The basicity (%CaO/%SiO2) is a useful measure of the
de-polymerisation. Glassy slags are highly polymerised but increasing basicity
results in both decreasing polymerisation and increasing crystallisation of the slag
(i.e. fcrys increases). In recent years, there has been a movement to significantly
reduce fluoride emissions and this has led to the development of F-free fluxes.
Furthermore, novel mould slags have been developed to cast certain steel grades
6 1 Introduction and Overview

(such as the calcium aluminate slags developed to cast high-Al (TRIP) steels) [6–9].
Further details of different types of mould powders are given in Chap. 6.
The performance of the mould fluxes has improved gradually over the last
50 years. Fly ash powders were gradually replaced by synthetic mould powders to
ensure better quality control. However, fly ash powders are still used in ingot
casting (described in Chap. 7). These improvements were brought about mostly
through ad hoc research. Scientific understanding has tended to trail behind the
developments resulting from this empirical research. However, some recent inno-
vations have been based on our scientific knowledge, although some empirical (ad
hoc) research was needed in their development (e.g. the development of
(i) non-Newtonian slags [10, 11] and (ii) the calcium aluminate slags for casting
high-Al steels mentioned above.

1.4 Mould Powder Behaviour in the Mould

Mould powders are dispensed onto the top of the steel meniscus in the mould and
form a powder bed. As the mould powders work their way down the powder bed,
they heat up and undergo various reactions. The bed contains three layers, namely,
powder, sintered and liquid layers (Fig. 1.2). The various reactions and events
occur in the following sequence (i.e. as the mould powder heats up):
(i) Any moisture in the flux evaporates at the top of the bed.
(ii) Carbonates (such as CaCO3 and Na2CO3) decompose to form oxide and
CO2(g) at a temperature of around 500 °C; these are endothermic reactions.
(iii) Carbon in the powder combusts with oxygen and CO2 (g) to form CO (g) in
those regions of the bed lying between 500 and 900 °C; these reactions are
exothermic.
(iv) In this temperature range the mineral particles also start to sinter.
(v) Finally, the solid slag begins to melt at temperatures above 900 °C and forms
a liquid slag pool; any remaining carbon particles float in this pool (see
Fig. 1.2). This slag pool serves as a reservoir of molten slag for subsequent
infiltration into the shell/mould channel; the liquid pool should have a depth
of at least 10 mm to ensure satisfactory infiltration of liquid slag.
The slag pool is very important to the process since it carries out the following
tasks:
(a) It seals off the steel meniscus from the atmosphere and prevents oxidation of the
steel.
(b) It provides a constant stream of liquid slag to lubricate the shell throughout the
length of the mould.
(c) It keeps the floating carbon particles away from the shell and thereby minimises
C-pick-up by the steel.
1.5 Slag Film and Slag Rim Characteristics 7

1.5 Slag Film and Slag Rim Characteristics

The slag film and slag rim are shown in Fig. 1.2.

1.5.1 Slag Film

The shell and its characteristics are central to the success of the continuous casting
process. The slag film is important because it is our main tool to control the shell.
The slag film controls both the lubrication and the heat transfer from the shell, and
so it determines the characteristics of the shell (e.g. shell length and thickness).
Hence, the characteristics of the slag film are the key to the success of the casting
process.
The lubricating properties of the slag increase with increasing liquid slag
thickness (dl) (this term is related to the powder consumption, Qs = q dl).
Lubrication is inversely related to the liquid friction force (Fl) and the parameters
affecting liquid friction are given in Eq. 1.3, where A is the surface area of the shell
(or mould), η is the slag viscosity and Vm is the velocity of the mould:

Fl ¼ AgðVm  Vc Þ=dl : ð1:3Þ

Thus, liquid friction decreases with decreasing values for mould surface area,
viscosity and (Vm − Vc) and increasing slag film thickness. The plot of Fl versus Vc
exhibits a minimum with (Vm − Vc) dominating at low casting speeds and Vc at
higher speeds. In practice, the slag liquid thickness (dl) can be increased by
reducing the solidification temperature (Tsol or Tbr) of the mould slag.
It was mentioned above that the heat extraction from the shell was controlled by
both the solid layer thickness (ds) and the fraction crystalline phase (fcrys) in the slag
film. The crystalline phases reflect and scatter radiant energy leaving the shell; thus
qhor decreases with increasing fcrys and with increasing crystal size [12, 13]. The
initial slag film formed is probably glassy because of the high cooling rate involved,
but it crystallises over time. Crystals have higher densities than glasses and thus
crystallisation is accompanied by shrinkage. This shrinkage, in turn, results in the
creation of (i) porosity in the slag film and (ii) an interfacial thermal resistance
(associated with an “air or gas gap”) at the Cu/slag interface, RCu/sl, which is
sometimes denoted Rint (see Chap. 3). Increasing porosity and RCu/sl both lead to
decreases in qhor [14–19]. The heat flux also decreases with increasing solid layer
thickness (ds) which, in turn, increases with increasing solidification temperature
(Tsol or Tbr) [19, 20]. In practice, ds, fcrys and RCu/sl all increase with increasing
basicity (= %CaO/%SiO2, denoted (C/S)) so these parameters can be increased by
increasing the basicity of the mould powder.
8 1 Introduction and Overview

Fig. 1.3 Photographs of slag films taken from the mould, showing a pores due to crystallisation at
one-third position from the mould (left) side and large pores in glassy phase due to CO
(g) formation on the shell (right) side (courtesy of Carboox [22]) and b large dendritic crystals
formed near the shell side and fine crystals formed on the mould side (permission granted, ISIJ
[24])

In conventional, mould powders (i.e. containing fluorides) cuspidine

(3CaO2SiO2CaF2 denoted as, C3S2Fl) crystals are usually the first to precipitate
(i.e. at high temperatures). Thus crystals are precipitated at the mould (cold) side of
the slag film (Fig. 1.3a) due to the undercooling of the liquid [21–23]. However, it
can be seen from Fig. 1.3b) that the large dendritic crystals can form on the shell
side of the slag film; this occurs because crystal growth is promoted by high
temperatures. Thus, small, cubic crystals form on the mould (cold) side [24] but
crystals grow in high-temperature region grow to form dendritic crystals.
The crystals in the slag film scatter IR radiation and thereby reduce the rate of
heat extraction from the shell. Thus the degree of crystallisation in the slag film [12,
13] is a key factor in the control of the horizontal heat flux. The slag film does
change over time by either slow downward movement of the slag film [25, 26] or
by fracture of the slag film and repair with liquid slag [27]. However, any such
changes are slow and the original slag film formed may persist throughout the cast
for those regions of the film [27] adjacent to the copper mould [28].
The movement of both the shell and the mould impose stresses on the slag film
and crystalline slag films have a tendency to fracture [25, 27]. Glassy slag films
form a super-cooled liquid (scl) above 600 °C and this scl will move when stress is
applied. Thus glassy slags, despite high viscosities, do provide some measure of
lubrication. Crystalline slags tend to fracture in the upper mould and glassy slags
tend not to fracture, but if they do, it is usually near the bottom of the mould [25].
The fractured slag film is repaired by liquid slag filling the gap and solidifying but
this may take time in the lower mould where the liquid slag flow is low [28–30].
1.5 Slag Film and Slag Rim Characteristics 9

1.5.2 Slag Rim

Slag films are formed by the infiltration of molten slag into the shell/mould channel.
Slag rims are their equivalent but are formed higher up in the mould (Fig. 1.2) and
are formed by the “painting” mechanism (see Fig. 11.92a, b) [31] as the mould
travels through the powder, sinter and liquid slag layers. The contents of the slag
film reflect this and contain unmelted mould powder, sinter, glassy carbon and
molten slag. An examination of a rim revealed it contained a series of layers
bounded by Na2O-rich frozen liquid which provided evidence for the painting
mechanism [27, 32]. The low melting, Na2O-rich liquid was considered to act as a
glue [27, 32]. The carbon comes from both the powder bed [33] and from the C-rich
layer formed by carbon particles floating at the top of the slag pool (see Fig. 1.2).
A slag rim is shown in Fig. 1.4a; typically the bulge of the slag rim is 3–5 mm
wide. The slag rim is thought to act as a piston and pushes liquid slag into the
mould/strand channel. No study relating the size of the slag rim to powder con-
sumption has been identified by the authors.
However, excessively large slag films, like those shown in Fig. 1.4b, can be
formed; these are frequently referred to as “ropes” or “bears”. The ropes can have
girths up to 60 mm [27]. Large rims are promoted by

Fig. 1.4 Photographs of slag rims formed with a a glassy rim formed by a low-basicity mould
slag and b a crystalline rim formed by high-basicity mould slag (courtesy of C-Å Däcker, Swerea
Kimab)
10 1 Introduction and Overview

• Large mould level variations;

• Using a mould slag with a high solidification (or break) temperature (e.g.
high-basicity slags);
• Mould powders with high, free carbon contents (for the given casting condi-
tions) which lead to low melting rates [27];
• Low meniscus temperatures, aided by low casting speeds, low superheat and
low melting steels (e.g. austenitic stainless steels or high-Al steels);
• Large Al2O3 pick-up by the slag which increases the melting temperature, the
viscosity and the fraction of crystalline phase (fcrys) (see Fig. 6.22 [34]); this is a
particular problem when casting high-Al steels where Al2O3 pick-up can reach
35%;
• Low metal flow velocities, which decrease the horizontal heat flux; frequently,
slag rims are not formed on the narrow wall because of the high metal flow
velocities [35]; single-roll flow systems reduce the meniscus temperature and
thus will favour slag rim growth. (cf. double-roll flow);
• Low-argon flow rates, since increased Ar flow increases both the vertical heat
flux and the meniscus temperatures; and
• Poor thermal insulation in the powder bed; thermal insulation can be improved
using (i) a thicker bed, (ii) smaller powder granules or (iii) exothermic agents in
the powder.
Highly crystalline slag rims tend to fracture easily [25]. Fractured slag rims can
(i) cause depressions in the surface of the steel [36–38], (ii) get trapped in the shell/
mould channel and block off slag infiltration which results in sticker breakouts and
(iii) create false alarms in the sticker detection system. Thus the formation of ropes
can be harmful to continuous casting.

1.6 Casting Conditions

The mould powder composition is selected for the given casting conditions; these
include (i) the casting speed, (ii) the mould dimensions, (iii) the oscillation char-
acteristics, (iv) the nature of the steel being cast and (v) the Ar flow rate. Empirical
rules have been developed to calculate the required values of powder consumption
(Qs), viscosity (η) and break (or solidification) temperature (Tbr) [39]. For further
detail, see Chap. 10.

1.6.1 Casting Speed (Vc)

The demands for ever-increasing productivity have resulted in the use of higher
casting speeds. This has imposed increasing demands on the mould powders which
1.6 Casting Conditions 11

have always been expected to be “forgiving” or “flexible”, i.e. that they will readily
accommodate changes in casting conditions.
Increases in casting speed result in the following, simultaneous, responses:
(i) Decreased powder consumption (i.e. less lubrication of the shell);
(ii) Increased heat flux density (qhor) from the shell;
(iii) Shorter residence time in the mould leading to both a decrease in the total
heat (qtot) leaving the shell and a thinner shell;
(iv) Increased metal flow turbulence, including increased metal velocities and the
formation of standing waves and vortices, leading to increased slag and gas
entrapment; and
(v) Shorter negative strip time (tn) leading to shallower oscillation marks.
Mould powders are usually selected to minimise the incidences of both “lon-
gitudinal cracking” and “sticker breakout” and involves careful selection of vis-
cosity, Tbr and fcrys for the slag [39]. Slag entrapment can be reduced by increasing
slag viscosity but this adversely affects the powder consumption (see Chap. 11).

1.6.2 Metal Flow

The metal flow pattern (Fig. 1.5) is an important factor in continuous casting. The
double-roll pattern is the preferred pattern and provides satisfactory casting con-
ditions. The single-roll flow pattern tends to cause both a decrease in meniscus
temperature (which can lead to the freezing of the steel meniscus) and high levels of
inclusions in the steel. Asymmetric flows pose many problems and can lead to high
levels of slag entrapment.
At high-casting speeds, turbulent metal flow occurs. This involves high metal
velocities and the formation of both standing waves on the metal/slag interface and

Fig. 1.5 Various metal flow patterns established in the mould a Double roll b single roll and
c Meniscus or Asymmetric roll; (permission granted, UNESID [47])
12 1 Introduction and Overview

vortices [40, 41]. All of these lead to slag entrapment and the incidence of longi-
tudinal cracking has been linked with the height of the standing waves [42].
Electromagnetic braking (EMBr, see Sect. 5.8.4 for more details) has proved useful
in dealing with turbulent flow. The upper coils (see Fig. 5.14) retard the velocity of
the metal leaving the SEN and the lower coils reduce the penetration of the metal
into the mould [43–45].

1.6.3 Mould Dimensions

It was mentioned in Sect. 1.3 that the lubrication supplied by the liquid slag
increases with increasing surface area of the mould and the required powder
consumption is in the hierarchy, Qslabs
s > Qblooms
s > Qbillets
s . The demand for slag is
regulated through the melting rate which is mainly controlled through the free
carbon content (typically, 4% C for slabs and 25% C for billets). For more details,
see Chap. 2. Mould dimensions also affect the flow patterns and slag entrapment
tends to be worse with very wide slabs [48] (see Sect. 11.9.2.8).

1.6.4 Oscillation Characteristics

Most steel works tend to operate with a fixed negative strip time. Thus when the
casting speed is increased, the frequency and stroke are adjusted (Eq. 1.1) to
maintain the same value for negative strip. Although there is some dispute about the
effects of oscillation parameters on powder consumption, most plant trial data
indicate that Qs increases with decreasing frequency and stroke. The depths of
oscillation marks increase with increasing values of stroke and negative strip times.

1.6.5 Steel Grade

Longitudinal cracks occur when the horizontal heat flux exceeds a certain critical
value (qhor crit) resulting in an overly thick shell. The shells of steels undergoing the
peritectic reaction (Eq. 1.4) are especially vulnerable to cracking because of the 4%
differences in thermal shrinkage coefficients for austenite and ferrite. Consequently,
to avoid longitudinal cracking, it is necessary to form a thin, uniform shell; this is
achieved by ensuring qhor is low by creating a thick, crystalline, slag film (i.e. with
high values of Tbr and fcrys):
1.6 Casting Conditions 13

FeðliqÞ þ Fed ðferriteÞ ! Fec ðausteniteÞ: ð1:4Þ

Sticker breakouts are associated with the formation of a thin, weak shell. Thus,
these breakouts can be minimised by creating a thick, strong shell and this is
achieved with a high qhor by creating a glassy, thin, slag film (i.e. with low values of
Tbr and fcrys). Thus, the nature of the steel affects the required properties of the
mould slag.
In recent years, the mould slags have had to respond to the challenges of casting
new steel grades, such as new high (>1% Al) steels where the slag has to absorb
large amounts of Al2O3 formed by metal/slag reactions (Eq. 1.5) where the
underline denotes in the metal. New families of mould slags have been developed to
meet this challenge but further development is still required:

2Al þ 1:5SiO2 sl ¼ Si þ Al2 O3 sl : ð1:5Þ

1.6.6 Ar Flow Rate

Argon is flowed through the SEN to reduce SEN clogging (i.e. the accumulation of
Al2O3 in the ports and the interior of the SEN). However, the argon is also the
source of “pinholes” in the steel. Furthermore, Ar bubbling is helpful in removing
inclusions from the steel to the slag pool. Argon flows also tend to oppose the
effects of casting speed and to cushion the metal flow. However, excessive Ar flow
rates (>5 l min−1) can transform a double-roll flow pattern in the steel to single-roll
flow pattern [49, 50], which provides less satisfactory casting conditions.

1.7 Physical Properties of Mould Slags

The mould slag must provide good lubrication (i.e. powder consumption) and the
key properties here are the fluidity (or reciprocal viscosity) and the break temper-
ature (Qs" as η# and Tbr#) [4]. The key properties for the control of the heat flux
(qhor) are Tbr (which determines the thickness of the solid layer of the slag film) and
fcrys, the fraction crystalline phase in the slag film [4]. However, the thermal con-
ductivities of the glass and crystalline phases and the optical properties of the slags
are also important but not as important as Tbr and fcrys. In slag and gas entrapment,
the key properties are the metal/slag interfacial tension (cmsl) and the viscosity (η).
The principal factor affecting cmsl is the S content of the steel but reducing the
B2O3, Na2O and CaF2 contents of the slag would help to increase cmsl. Further
details of the physical properties of mould slags are given in Chap. 9.
14 1 Introduction and Overview

1.8 Fluctuations in the Process

When you look at the top of the mould it looks very peaceful. However, the powder
bed frequently hides a mass of turbulent flow in the mould; surface waves and
vortices are formed on the steel meniscus and these are in a state of permanent
movement. The turbulence increases with increasing casting speed. The conditions
are continually fluctuating and are continually reacting to the current conditions.
Under such conditions the mould slag needs help. This can be achieved by (i) op-
timising parameters, like the immersion depth of the SEN and SEN port design and
(ii) using electromagnetic devices to reduce metal flow turbulence and to homo-
genise temperatures and steel composition. Nevertheless, the variation in the depths
of oscillation marks is an indication of the fluctuations. If all the bad casting
conditions were to coincide, there is the possibility of a disaster occurring (the
“butterfly effect”) [51].

1.9 Definitions

1.9.1 Powders, Slags, Fluxes

A number of terms are used to describe mould powders (e.g. casting powders,
mould powders, mould fluxes, etc.) which can lead to some confusion. In this work
we use the following terms to define the casting powder:
Powder implies that the flux is in its “as-received” state from the manufacturers,
i.e. it contains carbon and carbonates in pulverised or granule form; for instance, it
could be referred to as either casting powder or mould powder.
Slag (e.g. mould slag, casting slag or slag film) denotes the composition of the
slag pool (i.e. where carbon and carbonates have been removed during heating but
Al2O3, ZrO2, TiO2, TiN etc.) have been picked up by the slag.
The term, flux, is not extensively used here, where it is used, its meaning is
synonymous with “powder”.

1.9.2 Powder Consumption Terms

The lubrication supplied by the slag to the shell is usually cited in terms of powder
consumption. Not all of the powder forms (lubricating) slag since carbon particles
are oxidised and carbonates undergo gaseous decomposition. Consequently, when
defining specific powder consumption terms, it is necessary to specify whether this
1.9 Definitions 15

refers to powder or slag (e.g. Qslag

s (kg.m−2) or Qpowd
t (kg tonne−1)). The key factor
here is the fraction (f*) of the powder forming slag (Eq. 1.6); typical values of f* are
>0.9 and ca. 0.75 for casting powders used to cast slabs and billets, respectively:

Qslag
s ¼ f  Qpowd
s : ð1:6Þ

1.9.3 Temperature

In this work, temperature (T) is normally expressed in °C since this is the scale
normally used in the steel industry and is denoted by the symbol, T. However, it is
necessary to use the thermodynamic temperature scale (K) when describing ther-
modynamic functions and using Arrhenius relationships for viscosity and diffusion
coefficient. In these cases, the thermodynamic temperature will be used and is
denoted as TK.
Various temperatures associated with the slag film and the slag pool are used
here:
Glass transition temperature (Tg) is the temperature where a frozen glass
transforms into a super-cooled liquid (scl); the viscosity is considered to be 1013.4
(dPas) at Tg.
Deformation temperature (Td) is the temperature where the sample collapses
when measuring, for instance, the thermal expansion.
Critical temperature (Tcrit) is the temperature where the thermal conductivity
(k) of partially glassy samples decreases dramatically with increasing temperature
which occurs at ca. 770 °C for most mould slags; Tcrit is probably identical or
linked to the deformation temperature, Td.
Liquidus temperature (Tliq) is a thermodynamic entity and is the equilibrium
temperature where the mould slag becomes completely liquid.
Solidification or Break temperature (Tsol or Tbr) is the temperature where
crystals are first precipitated during a cooling cycle; the break temperature refers
specifically to the temperature where there is sudden increase in viscosity during
cooling; Tsol and Tbr are not equilibrium values and tend to decrease with increasing
cooling rate, and thus it is customary to associate them with the cooling rate (e.g.
10 °C min−1).
Crystallisation temperature (Tcrys) is not used in this text because it has been
used in the literature to describe different entities, e.g. for temperatures where
(i) crystallisation occurs on heating cycles [52] and (ii) where solidification occurs
in viscosity measurements during cooling (i.e. Tbr or Tsol) [53].
16 1 Introduction and Overview

1.9.4 Viscosity

In the continuous casting fraternity, the dynamic viscosity is usually cited in dPas
(which is the equivalent of the old unit, poise) but some workers cite values in Pas;
1 dPas (= 0.1 Pas) is used here.

References

1. P. Andrzejewski, A. Drastik, K. U. Köhler, W. Pluschkell, Proc. 9th Process Technol. Conf.,

1990, p. 173.
2. M. Wolf, Discussion Group on Continuous casting of mould fluxes, London, 1984, (Inst. Of
Metals, London, 1984).
3. ES Szerkeres, Iron and Steel Engineer, 73 (7), 29, (1996).
4. KC Mills, AB Fox, ISIJ Intl., 43, 1479, (2003).
5. P. Ramirez- Lopez, “Modelling shell and oscillation mark formation during continuous
casting via Explicit incorporation of slag infiltration” PhD Thesis, Imperial College, London,
2010.
6. JW Cho, KE Blazek, MJ Frazee, HB Yin, JE Park, SW Moon, ISIJ Intl., 53, 62, (2013).
7. K Blazek, HB Yin, G Skoczylas, M McClymonds, M Frazee, Iron and Steel Technology, 8
(3), 232, (2011).
8. TS Kim, JH Park, ISIJ Intl., 54, 2031, (2014).
9. Q Liu, GH Wen, JZ Li, XJ Fu, P Tang, W Li, Ironmaking and Steelmaking, 41, 292, (2014).
10. K Tsutsumi, K Watanabe, J Kubota, S Hatori, Y Miki, T Suzuki, T Omoto, Proc. 7th
Europ. Cont. Casting Conf., Dusseldorf, 2011, (VDEh, Dusseldorf, 2011) Session p. 1.
11. K Watanabe, K Tsutsumi, M Suzuki, H Fujita, S Hatori, T Omoto, ISIJ Intl., 54, 865, (2014).
12. M Susa, A Kushimoto, H Toyota, M Hayashi, R Endo, Y Kobayashi, ISIJ Intl., 49, 1722,
(2009).
13. M Susa, A Kushimoto, R Endo, Y Kobayashi, ISIJ Intl., 51, 1587, (2011).
14. J. W. Cho, H. Shibata, T. Emi,M. Suzuki, ISIJ Intl., 38 (5), 440 (1998).
15. K Watanabe, H. Okamoto, M. Suzuki, H. Kondo, T. Shiomi,. Proc. 79th Steelmaking Conf.,
1996, (ISS, Warrendale, PA, 1996) p. 265.
16. H Shibata, JW Cho, T Emi, M Suzuki, Proc. 5th Intl. Conf. Molten slags, fluxes and salts,
Sydney, 1997, (ISI–AIME,Warrendale, PA, 1997) p. 771.
17. H Nakada, K Nagata, ISIJ Intl., 46, 441, (2006).
18. M Hanao, M Kawamoto, ISIJ Intl., 48, 180, (2008).
19. M Hanao, M Kawamoto, A Yamanaka, ISIJ Intl., 52, 1310, (2012).
20. A Yamauchi, K Sorimachi, T Yamauchi, Ironmaking and Steelmaking, 29, 203, (2002).
21. M Hanao M Kawamoto, M Hara, T Murakami, H Kikuchi, K Hanazaki, Tetsu-to-Hagane, 88,
23, (2002).
22. MV Fonseca, ODC Afrange, A Lavinas, AA Ramos, CA Valadares, Proc. 5th Intl. Conf.
Molten slags, fluxes and salts, Sydney, 1997 (ISS, Warrendale, PA, 1997) p 851.
23. Z Li, KC Mills, MC Bezerra, Proc. XXXV Semin. De Fusao Refino e Solidifacao Metals,
Salvador, Brazil (2004) p. 281.
24. H Nakada, K Nagata, ISIJ Intl., 46, 441, (2006).
25. YA Meng, BG Thomas,. Met. Mater. Trans. B, 34B, 707, (2003).
26. C-Å Däcker, A Salwén, P Andersson, C Eggertsson, Proc. 7th Europ. Conf. Continuous
Casting, Düsseldorf, 2011, (VDEh, Dusseldorf, 2011) Session 12.
References 17

27. JA Kromhout, PhD Thesis, “Mould powders for the high speed continuous casting of steel”,
Univ of Delft, (2011).
28. E Lainez, Proc. 3rd Europ. Conf. Continuous Casting, Madrid, 1998, (UNESID, Madrid,
1998) p. 155.
29. RJ O’Malley, J Neal, Proc. Intl. Conf. New Developments in Metallurgical Process Technol.,
Dusseldorf, 1999, METEC Congress (VDEh, Dusseldorf, 1999) p. 73.
30. TJ Billany AS Normanton, KC Mills, P Grieveson, Ironmaking and Steelmaking, 18, (1991)
403.
31. C Perrot, JN Pontoire, C Marchionni, MR Ridolfi, LF Sancho, Proc. 5th Europ. Conf.
Continuous Casting, Nice, 2005, (La Rev. Metall., Paris, 2005,) p. 36.
32. J Kromhout, RS Schimmel, Proc 8th Europ. Conf. Continuous casting, Graz, 2014, (Austrian
Metals Soc., Vienna, 2014).
33. JJ Macho, G Hecko, B Golinmowski, M Frazee, Preprints 33rd McMaster Symp. Iron and
Steelmaking, Hamilton, Ont, 2005, (McMaster Univ. Hamilton, 2003), p. 131.
34. M Hanao, Y Tsukaguchi, M Kawamoto, Proc. 4th Intl. Cong. Sci. Technol. Steelmaking,
Gifu, Japan, 2008 (ISIJ, Tokyo, 2008), p. 94.
35. CA Dacker, M Glaes, SP Andersson, A Salwen, C Eggertsson, Proc. 6th Europ. Conf.
Continuous Casting, Riccone, Italy, 2008, (AIM, Milan, 2008), 8 pdf.
36. JW Kim, S. K. Kim, D. S. Kim, Y. D. Lee, J. I. Eum, E. S. Lee; Proc. 78th Steelmaking
Conf.1995 (ISS, Warrendale, PA,1995) p. 333
37. M Jenkins, BG Thomas, WC Chen, RB Mahapatra, Proc. 77th Steelmaking Conf. 1994, (ISS,
Warrendale, PA., 1994) p. 337.
38. M S Jenkins, BG Thomas, Proc. 80th Steelmaking Conf., Chicago, IL, 1997, (ISS,
Warrendale, PA, 1997) p. 285.
39. KC Mills, AB Fox, PD Lee, S Sridhar, Proc. Sci. Technol. Steelmaking, Swansea, 2001,
(IOM, London, 2001) p. 445.
40. LC Hibbeler, BG Thomas, Proc. AISTech Conf 2010, Pittsburgh (ISS, Warrendale. PA, 2010)
p. 1215.
41. LC Hibbeler, BG Thomas, Iron and Steel Technology, 2013 (Oct), 121, (2013).
42. M Hanao, M Kawamoto, M Hara, T Murakami, H Kikuchi, A Yamanaka, Proc. 5th
Europ. Conf. Continuous Casting, Nice, 2005 (La Revue Metall., Paris, 2005), see also
Tetsu-to Hagane, 88, (1), 23, (2002).
43. E Takeuchi, JOM, 1995 (May), 42, (1995).
44. E Takeuchi, H Harada, H Tanaka,T Ishii, T Toh, M Zeze, M Hojo, K Shigematsu, Nippon
Steel Technical Report, 61, 29, (1994).
45. M Washio, M Sugizawa, S Moriwaki, K Kariya, S Idogawa, S Takeuchi, Revue de
Metallurgie, CIT, 90 (April), 507, (1993).
46. SG Kollberg, HR Hackl, PJ Hanley, Iron and Steel Engineer, 73 (7), 24, (1996).
47. D Gotthelf, P Andrezjewski, E Julius, H Haubrich, Proc. 3rd Europ.. Conf. Continuous
casting, Madrid, 1998, (UNESID, Madrid, 1998), vol. 2, p. 825.
48. W Emling, T A Waugaman, SL Feldbauer, AW Cramb, Proc. 77th Steelmaking Conf. (ISS,
Warrendale, PA., 1994) p. 371.
49. PE Ramirez- Lopez, PN Jalali, J Bjorkvall, U Sjostrom, C Nilsson, ISIJ Intl. 54, 342.
(2014) and Proc. 8th Europ. Conf Continuous Casting, Graz, Austria, 2014 (ASMET, Vienna,
2014).
50. E van Vliet, DW van der Plas, SP Carless, A A Kamperman, AE Westendorp, Proc.. 7th
Europ. Conf. Cont. Casting, Dusseldorf, 2011, (VDEh, Dusseldorf, 2011) Session 4.
51. PD Lee, PE Ramirez-Lopez, KC Mills, B Santillana, Ironmaking and Steelmaking 39(4), 244,
(2012).
52. JV Dubrawski, JM Camplin, J Thermal Analysis, 40, 329, (1993).
53. JW Kim, J Choi, OD Kwon, IR Lee YK Shin and JS Park, Proc. 4th Intl. Conf. Molten slags
and fluxes, Sendai,1992,. (ISIJ, Tokyo, 1992) p. 468.
Chapter 2
Slag Infiltration, Lubrication
and Frictional Forces

Abstract It is essential to lubricate the shell since inadequate lubrication leads to

defects in the steel product (e.g. longitudinal cracks, sticker breakouts and star
cracks). The liquid layer of the slag film, formed between the shell and the mould,
lubricates the newly formed shell; the lubrication increases with increasing liquid
slag thickness (dl). Lubrication is usually represented by the powder consumption
(Qs in units of kg slag (or powder) m−2) which is related to liquid film thickness
(dl). However, there are several terms used for powder consumption and these terms
are interrelated (e.g. Qs, Qt and QMR). The frictional forces acting on the shell are
highest in the centre of slabs and thus slabs need more lubrication. The required
powder consumption, Qs increases with increasing distance from the corner and
thus Qslab
s > Qbloom
s > Qbillet
s . The required powder consumption can be calculated
from the relation, Qreq s = 2/(R* – 5) where R* = {2(w + t)/w
t} = (surface
area/volume) of the mould. However, the powder consumption, Qs, is also affected
by other parameters, namely, the casting speed (Vc), slag viscosity (η), the break
temperature of the slag and the oscillation frequency (f) and stroke (s). There is
general agreement that Qs decreases with increasing casting speed and viscosity
sreq = 0.55/ η
(e.g. empirical rules, Qslag
Vc). There is some dispute with regard to
0.5

the effect of f, s and Tbr but most plant studies indicate that Qslag
sreq decreases as f, s
and Tbr increase. The required values of powder consumption and viscosity can be
calculated for the given casting conditions using empirical rules. The predictions of
a mathematical model indicate that slag infiltration into the model/ strand channel
occurs when the mould and slag rim are descending but little powder consumption
occurs when the mould is ascending. The changes in mould direction are accom-
panied by periods of confused flow in the mouth of the channel and little slag
infiltration occurs in these periods. Frictional forces and the factors affecting them
are also discussed; it was found that liquid friction increased with increasing mould
dimensions, slag viscosity, casting speed and (Vm − Vc). Plots of liquid friction (Fl)
versus casting speed exhibit a minimum since Fl increases with increasing Vc but
decreases with decreasing (Vm − Vc).

© Springer International Publishing AG 2017 19

K.C. Mills and C. Däcker, The Casting Powders Book,
DOI 10.1007/978-3-319-53616-3_2
20 2 Slag Infiltration, Lubrication and Frictional Forces

Symbols, Abbreviations and Units

A Area (m2)
%Cfree Percentage of free carbon
%Ctotal Percentage of total carbon
%LOI Percentage of loss on ignition
DC Mean particle size of the carbon
Dcorn Distance mould corner to centre (m)
Dl Thickness of liquid slag film (m)
Fl Frictional force (N)
f Frequency (Hz or cycles min−1)
f* Fraction of powder forming slag
Qcycle Powder consumption (kg m−1 cycle−1)
QMR Melting rate (kg/min or kg/s)
Qs Powder consumption (kg/m2)
Qt Powder consumption (kg/tonne−1)
R* Mould (surface area/volume) (m−1)
s Stroke length (m)
T Temperature (oC)
Tbr Break (or solidification) temperature
t Time (s) or thickness of mould (m)
tcycle Time for one cycle (s or min)
tn Negative strip time (s)
tp Positive strip time (s)
Vc Casting speed (m/min)
Vm Velocity of mould (m/min or m/s)
w Width of mould
η Slag viscosity (dPas)
q Density (kg/m3)

Superscripts
powd
Refers to powder
slag
Slag formed from powder

2.1 Introduction

The newly formed steel shell is lubricated as it progresses down the mould by a
flow of liquid slag from the slag pool (Fig. 2.1). This flow of slag helps to reduce
the frictional forces acting on the shell. The liquid frictional force (Fl) operating on
the shell can be calculated using Eq. 2.1 where Vm and Vc are the velocities of the
mould and the casting speed, respectively, A is the active surface area of the mould,
η = slag viscosity and dl is the liquid slag film thickness.
2.1 Introduction 21

Fig. 2.1 Schematic drawing showing the lubrication of the sausage-shaped shell by liquid slag,
which is shown in light blue (permission granted, ISS/AIST, [1])

Fl ¼ AgðVm  Vc Þ=dl ð2:1Þ

Thus the liquid frictional force decreases as the:

• Thickness of the liquid film increases.
• Surface area of the mould decreases.
• Viscosity decreases.
• With a decrease in the difference between the velocities of mould and steel
strand.
It will be seen below that the thickness of the liquid slag film (dl) is related to the
powder consumption Qs (with units of kg/m2). In addition to the liquid friction
forces there are also solid friction forces which tend to occur in the lower mould.
The powder consumption, Qs (in kg/m2 of mould) is often used as a measure of the
lubrication supplied by the liquid slag since it is linearly related to dl (see Eq. 2.4).

2.2 Powder Consumption (Q)

The powder consumption (Q) can be expressed in several ways, each term having
different units. The most common form for powder consumption is Qt (in kg/t steel)
which is the mass of powder consumed per tonne of steel cast. This provides a
measure of the cost of the casting powders in continuous casting.
22 2 Slag Infiltration, Lubrication and Frictional Forces

It should be pointed out that the mould powder contains carbon, carbonate and
other volatiles which burn off and, hence, do not contribute to lubrication by the slag.
Some billet fluxes contain up to 25% free carbon and consequently, it is necessary to
differentiate between mould powder and slag. This is denoted here by attaching the
superscripts, powd and slag, respectively, to the various consumption terms. It is
possible to calculate one term from the other by calculating the fraction of powder
forming slag (f*) by Eq. 2.2 and using either the free carbon (%Cfree) and
total carbon contents of the powder (%Ctotal) or alternatively, the loss on ignition
(%LOI) which are usually supplied by the powder manufacturer.

f  ¼ ½100  ð%Cfree Þ  fð44=12Þð%Ctotal  %Cfree Þg=100 ¼ ð100  %LOIÞ=100

ð2:2Þ

The powder consumption of slag can be calculated from the relation:

Qslag
t ¼ f  Qpowd
t : ð2:3Þ

2.2.1 Various Powder Consumption Terms

As mentioned above, the powder consumption can be expressed in various

ways, each with different units; the various terms can be calculated from one
another, e.g. Qs and Qt by Eq. 2.4.

  
Qslag
s kg/m 2
¼ f 7:6 Q powd
t =R ¼ q dl  2550 dl : ð2:4Þ

where R* = (surface area/volume) of mould = {2(w + t)/ w
t (and has units = of
m−1) w = width of the mould (m), t = thickness of the mould (m) and q = the
density (kg/m3) of the molten slag, 7.6 is the density of steel in t/m3. Equation 2.4
also shows the link between Qs and the thickness of the liquid slag film (dl).
Furthermore, the melting rate of the mould powder, Qpowd
MR (in units of kg/min or
kg/s) can be calculated from Qslags , by Eq. 2.5 where Vc is the casting speed
(m/min).

 powd
Qslag
MR ðkg/sÞ ¼ f QMR ¼ 2ðw þ tÞQs Vc =60 ð2:5Þ

The melting rate should match the powder consumption (Qs) needed to provide
good lubrication of the shell. There are a number of variables which affect the
melting rate but it is primarily controlled through the free carbon content (%Cfree)
and the mean particle size of the carbon ðDCfree Þ of the powder (i.e. Qslag
MR " as %Cfree#
and DCfree " ).
The powder consumption per oscillation cycle, Qslag cycle (kg/cycle) can be calcu-
lated from Qs via Eq. 2.6 where f = oscillation frequency (in Hz)
2.2 Powder Consumption (Q) 23

1


Qslag
cycle kg cycle ¼ 2ðw þ tÞQs Vc =60 f ð2:6Þ

All of the above parameters can be derived by considering a liquid slag film of
uniform thickness (dl) distributed around the mould and by assuming
Ashell = Amould.

2.2.2 Measurement of Powder Consumption

The powder consumption, Qpowd t , is frequently determined as the number of bags (N) of
known mass of casting powder (m) used in the entire cast, for which the total mass of
steel cast (msteel) is known. Thus Qpowd
t can be calculated (Qpowd
t = Nm/ msteel).
Powder consumption rates for different periods during a cast can be determined
by measuring the mass of powder dispensed from the hoppers in a known time
period. However, these measurements are affected by the height of the powder in
the hopper and thus measured rates are affected by recharging the hopper; for true
significance, the measurements should refer to the same height of powder in the
hopper. There is some variability in the consumption values. Some of this vari-
ability probably arises from small variations in the casting conditions (e.g. changes
in casting speed) through the cast.
The most precise method for measurement of mould powder consumption,
when, for example, performing plant trials with new mould powders, is to use a
bucket with known amount of mould powder and count the number of buckets
needed for the casting.
Powder consumption data for 32 casts of the same steel under the same con-
ditions (where any casting speed variations were <±10%) is shown in Fig. 2.2.
The mean Qt value was 0.63 kg/t, the maximum range in values was 0.24 kg/t and
the uncertainty was calculated as ±12% [1, 2].
Studies where the powder supply was monitored at nine positions between the
centre and the edge of the mould showed that consumption (i) was highest at the
edge (presumably because it has to feed the narrow face too) and (ii) went through a
minimum located midway between the edge and the centre [3].
The powder consumption at different periods of the cast is shown in Fig. 2.3, it
can be seen that the powder consumption appears to be higher at the beginning of
the cast and gradually decreases to a steady value.

2.2.3 Methods Used to Understand Slag Infiltration

Mechanisms

Several techniques have been used to gain an understanding of the mechanisms

responsible for slag infiltration, they are:
24 2 Slag Infiltration, Lubrication and Frictional Forces

Fig. 2.2 Histogram of powder consumption (kg/t) in similar trials where a variation of <±10% in
casting speed was allowed (courtesy of Fox [2])

Fig. 2.3 Powder consumption at different periods of the cast (courtesy Fox [2])

2.2.3.1 Analysis of Plant Data

Plant trials are carried out where powder consumption is measured as one variable
(for instance, as a function of casting speed or slag viscosity) is changed while all the
other variables are held constant. Most of the empirical relations were derived in this
manner (see Table 2.3). One major problem with this approach lies in determining
the effects of oscillation characteristics where it is common practice to operate with a
fixed negative strip ratio. Thus if the frequency is increased, it is common practice to
alter the stroke (s) or casting speed to ensure that negative strip time remains the
same. Thus there are few plant trial data carried out where only one oscillation
parameter (say frequency, f) is varied systematically.
2.2 Powder Consumption (Q) 25

An alternative approach was taken by Saraswat et al. [4]; they carried out a
statistical analysis for a large database of powder consumption data from various
steel plants in order to identify those variables (Vc; η, f, s, Tbr) which had a sta-
tistically significant effect on Qs.

2.2.3.2 Physical Modelling Studies

In physical models, the effect of changing one variable is studied by observation of

the slag flow. The physical models can be divided into three types, given below,
and details of the various studies are given in Table 2.1:
(a) Cold (or water) models which are usually carried out at near-ambient tem-
peratures. In some cases the slag rim is represented by a solid ledge [5] and
other cases by freezing of the water representing the slag [6]. The main problem
with these tests is that they tend to be isothermal whereas the conditions in the
mould are not isothermal.

Table 2.1 Details of some physical modelling studies of slag infiltration and factors affecting
powder consumption
Reference Type Metal Slag Comments, findings
Anzai [12] CM Rubber Paraffin Pressure measured in mould/shell gap;
belt mould = oscillating plate,
shell = moving belt
Jenkins [6] CM Ice Gycerol/water Rim = solid ledge; Measures pressure
(p) in gap. Predicts (a) Qs" as f# as s"
and η#. (b) P versus η plot goes
through maximum at 2 dPas
Tsutsumi [7] HM Sn/Pb Stearic acid Slag infiltration occurs in both tn and tp
but more in tp—slag inflow occurs at
end of tn. Qs" as Vc# η# f# s f" Tsol#
Friction measured
Itoh [7], HM Cu Mould flux Qs dependent upon (tn + 0.5tp). No η
Nebeshima [7] dependence for Qs.
Kajitani [5] CM Moving Silicone oil Mould = oscillating acrylic plate;
polyester Mould/ shell gap monitored. Gap
belt profile is important. If gap narrows (as
go downward) Qs" as Vc", η ";
if gap widens Qs# as Vc", η" suggests
gap widens downward
Badri [9, 10] M Sim Steel Mould slag Oscillating mould. Principally focused
on shell profile, heat flux and OM
formation
Ko [11] M Sim Steel Mould slag Oscillating mould. focused on shell
profile, heat flux and OM formation
CM cold model; HM hot model; M Sim mould simulator
26 2 Slag Infiltration, Lubrication and Frictional Forces

(b) Hot models use low-melting metals, like copper or tin, to represent the steel
shell and the effects of changing variables are observed visually [7, 8].
(c) Miniature continuous casters [9–11] are, bench-top versions of a continuous
casting machine; they are fully instrumented to provide information about
powder consumption, heat flux, friction, etc. The apparatus (see Sect. 3.2.4.2)
consists of (i) a water-cooled copper mould which is inserted in a bath of steel
covered with mould powder (the mould is in inverse mode, i.e. where the mould
is surrounded by a slag film and a steel shell) (ii) an extraction system to
provide continuous casting (iii) mould level control (iv) sinusoidal oscillation
of the mould and (v) sensors to provide mould and steel temperatures, shell and
mould displacements [9–11].

2.2.3.3 Mathematical Modelling Studies

The importance of the slag infiltration process is reflected in the number of

mathematical models developed to predict the powder consumption; these models
mostly make use of the Navier–Stokes equation [1, 13–25]. Billany et al. [26] and
Gray [27] compared the predictions of the earlier models with plant data and
observations. They found that:
• The models predicted the correct trends with casting speed, viscosity, etc.
• The models tended to slightly underestimate powder consumption.
However, one common problem with the model predictions is that they predict
that Qs increases with increasing frequency whereas most experimental data indi-
cate the reverse relation, i.e. Qs and Qt decrease with increasing frequency (see
Sect. 2.2.6.4) (Fig. 2.4).

Fig. 2.4 Powder 0.7

consumption as a function of
frequency [28] (permission
granted ISIJ [28], re-drawn)
Q t , kg tonne -1

0.6

0.5

0.4
60 80 100 120 140 160 180
Frequency, f, cpm
2.2 Powder Consumption (Q) 27

There are two approaches to modelling powder consumption. The first considers
the powder consumption in its entirety. In the second approach, the various con-
tributions to powder consumption are calculated individually [7, 19, 20] and col-
lated. For instance, Meng and Thomas [19] consider that the powder consumption
(Qs) is made up of the following contributions:
• Liquid slag consumption (Ql).
• Solid slag consumption (Qsol) caused by any downward movement of the solid
slag film.
• Powder consumption (QOM) caused by liquid slag trapped in the oscillation
mark.
In some cases, Qsol + Ql = Qlub is assumed [7]. The powder consumption which
is trapped in the oscillation mark (QOM) has been derived from calculations of the
mean depth and the pitch of the oscillation mark [19] or from the relation,
QOM = 0.001f tnql (tn + 0.5tp) [20].
Itoyama et al. [20] divided the powder consumption (Qs) into the following
contributions:

Qs ¼ Qm þ Qg þ Qf þ QOM : ð2:7Þ

where Qm (=qdl/2) and Qg (=gq2d3l /12ηVc) are associated with the pressure drop for
parallel plates and the effect of gravity on the flow, respectively, Qf is the flow asso-
ciated with mould oscillation (Qf = (Lu/Le)qsf Le sin (b/Vc) and QOM = AOM q f/ Vc
where b is the mould taper, Le = effective length of mould) and Lu = relative,
decreasing, distance when mould is ascending and AOM = area of oscillation mark.
Saraswat et al. [4] calculated the values of QOM (denoted Qcalc OM) and compared
them with a database of powder consumption values recorded in different steel-
works (Qtotal
s ). It was found that all of these models tended to seriously
over-calculate QOM, i.e. (Qcalc
OM/Qs
total
) > 1 and frequently with values of 2 or more
for this ratio.
Details of the various mathematical models are summarised briefly in Table 2.2.

2.2.3.4 Empirical Rules

A number of empirical equations have been deduced form analysis of plant data or
from observations made in simulation tests. These are discussed in detail in
Sect. 2.2.6. The equations and the effects of changes in the various factors are
summarised in Table 2.3.
Table 2.2 Details of some published mathematical models covering slag infiltration
28

References Model details Comments, findings

req
Niggel [13] Initially oscillation excluded. Assumes solid slag film adheres to Two regimes (i) QMR  Qreq MR; (ii) QMR  QMR
mould; liquid moves down. Thermo-mechanical approach Model predicts Qs versus Vc exhibits a maximum
Schwerdtfeger Based on N–S eqn. for slag flowing between parallel plates. Needs Shear stress and Friction calculated from velocity profile. Predicts
[18] data for ds+l at meniscus and parallel plate regions shell bent away from mould during descent of mould and towards
Calculates velocity profile and P in gap mould during ascent of the mould
Kor [14] Based on N–S eqn. for slag flowing between two parallel plates— Must provide value of ds+l and no account of thermal effects in the
(1) moving at constant velocity and (2) oscillating slag film
Hill [29] s
Model to calculate slag flow, d-OM. Max.vol flow rate, Qmax = (2/3) Predictions for dOM: dOM# as Vc" as s# as η#
{η/ Dqg}0.5 V1.5–N–S Eqn.
Nippon Steel Values of ds and dl not needed. Assumes mould oscillation affects Predicts (i) dl" as η" and as f# (ii) Qs" as Vc# as η# [26]. In
[15, 16, 30] viscosity. Rheological approach to calculate slag film thickness and agreement with plant data
velocity profile
Meng [19] Couples transient slag flow model with solid slag stress and FD Models used to check effects of Vc and η. Fl gives negligible
model of heat transfer. Uses Slag fracture and TTT data. Validated stresses. Slag film Fracture occurs near meniscus for crystalline and
via Friction measurements on plant near mould exit for glass phase
Okazawa [17] Uses Reynolds eqn, not N–S eqn. Steady state Mould/shell gap widen as mould ascends; gap narrows during the
Unsteady state ascent
Gap opened during ascent and closed during descent of mould
Ojeda [22] Couples, FF and HT. Values assumed for ds, dl; men Steel overflows at end of tp. Rim exerts positive pressure. Positive
shape = Bikerman Qs during (tn to early tp) and negative Qs during tp. when Vm is high
Ramirez– Couples FF, HT and solidification. No assumptions re: ds, dl or Follows Qs, q, dshell, ds, dl through osc.cycle, Qs increases as
Lopez [21] men. shape mould/rim descends with max. in early tp. Related to direction of
flow in slag pool- downward-descending; outward-ascending
Jonayat [25] Couples transient flow of metal and slag phases with heat transfer Parametric data: Qs" as Vc#, as s" as a′" relation with f complex;-
and solidification models Q (kg m−1cycle−1)# as f"
Validated against plant data for Qs
MR is required Qs value; a′
FF fluid flow in slag and metal; HT heat transfer; men meniscus; osc oscillation cycle; N–S Navier–Stokes; P pressure; Qreq
modification ratio
2 Slag Infiltration, Lubrication and Frictional Forces
Table 2.3 The predicted effect (increase " or decrease, #) on Qs with an increase in parameter (e.g. f)
References Equation, Qs Vc η f Tsol Tliq s tp tn tcycle qslag dl dOM
Wolf [62] P Qs = 0.1+ 0.55 (60/fη0.5 Vc) # #
Ogibayashi [42, 43] P Qpowd
s ¼ 0:60=gVc # #
Wolf [41, 42], Fox [1] P Qpowd
s ¼ 0:70=g0:5 Vc # #
modified Qpslag
s = 0.55/η0.5 Vc
Kwon [51] Ps Qs = 0.22 + 0.4{60/(0.5 s)0.3fη0.5Vc} # # # #
Maeda [49] P Qs = 0.15 f
tp/ (η
Vc) # # " # "
0.116
2.2 Powder Consumption (Q)

Nakajima [52] P Qs = 205400V−0.628

c T−0.866
liq
s0.341 t−0.076
cycle tp # # " " #
Sridhar [45] P Qs = 0.3/ (η0.045Vc) # #
Wolf 5 [63] Qs = 0.5 (s
fη Vc/1000)−0.5 # # # #
Jenkins [6] Sim Qs = (0.433/V0.5 c ) {1 + (0.0283/ηVc)} # #
2
Fox-modified [2] M Qs = (0.369/v0.5
c )(1 + 0.1564/ηVc ) – 0.123
Nebeshima [7] Sim Qs = (ql
dOMdxf/2Vc) + (qldl) # " " " "
Tsutsumi [53, 64] Sim Qs = (kbs0.4/Tsolη0.5Vc) cos−1 [1000Vc/2pfs] # # # # "
Emi [65] Qt = 0.6η−0.15 #
Nakato [66, 67] Qs = qslag(0.143–0.003η) # # "
Kitagawa [57] Qs = 0.0085 tp
f/ Vc # " "
Noguchi [68] Qs = 10−3(1952 − 246Vc − 44η − 1.07Tliq.) # # #
Kobayashi [69] Qs = 0.003(tcycle + tp)1.5
f/Vc # " " "
Shimizu [70] Qs = f tp{0.0184(Ts–Tliq)–2.58}/10Vc # " "
Itoyama [20] Qs = Qm + Qg + Qf + QOM
Shin [23] P Qs = [0.025 ql
k1.43{(2Dc/ Dqg)0.5}0.556
t0.389n

v−1.49
s + 0.507e3.59tp] (f/vs) where vs = Vc103/60
Saraswat [4] StA Qs = e28.81/Vcη0.46
f0.49
s1.37
T3.48 br # # # # #
when casting a LC steel where f frequency; s stroke; tn negative strip time; tp positive strip time; Qm, Qf, Qg, Qs defined in Sect. 2.2.3.1 (iii); P analysis of plant
trial data; Sim simulation expts; StA statistical analysis of plant data
29
30 2 Slag Infiltration, Lubrication and Frictional Forces

2.2.4 Problems Arising from Poor Powder Consumption

Inadequate powder consumption has been reported to cause:

• Longitudinal cracking [31].
• Sticker breakouts (which are associated with a thin shell and poor lubrication).
• Deep oscillation marks [32, 33].
• Transverse cracking [34].
• Off-corner cracking [35, 36].
• The formation of depressions (when poor mould level control leads to the
capture of the slag rim and the simultaneous cutting off of slag infiltration)
[37–39].
Furthermore, heat flux (or mould temperature) variations were found to increase
with decreasing powder consumption (Fig. 2.5) [40]. One of the principal causes of
heat flux variations is the creation of a corrugated shell which is prevalent when
casting peritectic, MC steels. The powders used to cast these steels are designed to
give a thick, crystalline slag film to reduce the horizontal heat transfer. This is
achieved by increasing Tbr for the slag. However, a high Tbr slag tends to create in a
thin liquid slag film (dl) which results in a low Qs value. Corrugated shells are less
common in other steel grades and so slags tend to have lower Tbr values and hence
higher Qs values. Thus mould temperature fluctuations tend to exhibit an inverse
relation with powder consumption (Fig. 2.5).

2.2.5 Optimum Casting Conditions

Wolf [41, 42] proposed that there was an optimum range for casting (i.e. the
minimum friction and the optimum horizontal heat flux) which was defined in terms
of the parameter (ηV2c ). This is shown in Fig. 2.6. The optimum conditions were
found to occur when (ηV2c ) = 5 ± 3 dPas (m min−1)2.

Fig. 2.5 Fluctuations in 160

mould temperature as a
Mould Temp fluctua on Index

function of powder
consumption when using 120
different casting powders
(permission granted,
ISS/AIST, [40]) 80

40

0
0 0.1 0.2 0.3
Qs , kg.m-2
2.2 Powder Consumption (Q) 31

Fig. 2.6 Schematic drawing showing mould friction and horizontal heat flux as function of the
parameter (ηV2c ) (permission granted, VDEh/Stahl Eisen, [41, 42])

Ogibayashi et al. [43] reported that fluctuations in molten slag infiltration and
mould temperature (heat flux) were at a minimum when the parameter, ηVc, had a
value in the range 1–3 dPas m min−1 (Fig. 2.7a, b). A similar figure, involving
fluctuations in the frictional force, also exhibits a minimum in this region
(Fig. 2.7c) [44]. Thus the minimum fluctuations (i.e. optimum casting) occurs when
ηVc = 2 ± 1 dPas m min−1.
It can be seen from Figs. 2.5 and 2.6 that the equations, ηV2c = 5 ± 3 dPas
(m min−1)2 and ηVc = 2 ± 1 dPas (m min−1) that powder consumption increases
with decreasing viscosity. A plot of the viscosity of the casting slag as a function of
the casting speed for mould fluxes used in a large number of steel plants revealed
that the bulk of the data fell between the curves representing the Wolf [42] and
Ogibayashi [43, 44] relations (Fig. 2.8 [45]). The outliers in the figure all relate to
high-viscosity powders used in high-speed billet casting and this behaviour will be
discussed in Sect. 2.2.6.10.

(a) (b) (c)

Mould heat flux varia on,

0.4 5
15
Film varia on, mm

Fric on force, ton

0.3 4
Mcalm-2 hr -1

10 3
0.2
2
5
0.1
1

0 0 0
0 2 4 6 0 2 4 6 0 1 2 3
ηVc, dPas. m min -1 ηVc, dPas. m min -1 ηVc, dPas. m min -1

Fig. 2.7 Fluctuations in a Liquid slag film thickness on broad face (dl) [43] b mould heat transfer
(i.e., horizontal heat flux) [43] and c frictional force [44] as functions of the parameter, ηVc
(permission granted a, b Nippon Steel Sumitomo Metal Corp. [43] and c ISIJ [44])
32 2 Slag Infiltration, Lubrication and Frictional Forces

Fig. 2.8 Viscosity of mould slag at 1300 °C as a function of casting speed used in casting at a
number of steel plants [45]; dashed line Wolf relation (ηV2c = 5 ± 3 dPas (m min−1)2); solid line
Ogibayashi relation, (ηVc = 2 ± 1 dPas (m min−1)); dotted lines bounds of above relations;
D = slabs; = thin slabs; o = blooms; ◊ = billets; + = rounds; X, ◊ = high-speed billets
(permission granted, UNESID, [45])

2.2.6 Factors Affecting Powder Consumption

There are a number of factors affecting powder consumption, these are outlined
below. These factors can be classified as following: (i) casting conditions,
(ii) characteristics of the mould powders and slags and (iii) conditions arising from
process problems.

2.2.6.1 Mould Dimensions

It was noted above in Eq. 2.1 that the frictional force increased as the surface area
of the mould increased. Neumann et al. [46] proposed the powder consumption (Qs)
could be represented as function of the parameter, R* (Fig. 2.9a, Eq. 2.8). This term
represents the ratio of the (surface area/volume) for the mould. Alternatively, it can
be regarded as the ratio of (the surface area of shell/volume of steel in the mould). It
can be calculated from Eq. 2.9 where w and t are the width and thickness of the
mould. The parameter, R*, was used in Eq. 2.4 to calculate Qslag s from Qpowd
t .
2.2 Powder Consumption (Q) 33

(a) (b)

Powder Consumption, Q c corr (kg m 2 )

0.8
Billets
0.7 Blooms
0.6 Slabs
Round
0.5 High Speed Billet
: Q s =0.44e
[4] -0.04R

0.4 :Qs=2/(R-5)
0.3 Thin Slabs

0.2

0.1
0
0 10 20 30 40 50
Surface Area to Volume Ratio, R (m-1)

Fig. 2.9 Powder consumption, Qs, as a function of the parameter, R* a used in moulds with
different dimensions; (w
x t in m) A (1.5 0.22) B (1.0 0.1); C (1.3 0.050);
D (0.16 0.16); E (0.13 0.13) [46] and b ◊ = billets; o = blooms; D = slabs; = thin
slabs; + = Rounds; X = high-speed billets; dotted line = Eq. 2.8 [46]; solid line = (Qs = 2/
(R – 5)) [45, 47] (permission granted), ISS/AIST, [46] and b UNESID, [45]

Qs ¼ 0:44 exp0:44R ð2:8Þ

R ¼ 2ðw þ tÞ=w
t ð2:9Þ

Subsequently, a similar study based on a much larger database and found that the
data shown in Fig. 2.9b could be represented better by Eq. 2.10 [45, 47].


Qs kg m2 ¼ 2=ðR  5Þ ð2:10Þ

Typical R* values are for Slabs 9–15; for Blooms 10–18; for Billets 22–30, Thin
slabs 40. It can be seen from Fig. 2.9b that the required powder consumption is
much greater for slabs than that for billets. Ogibayashi [34] pointed out that frictional
forces on the broad face tended to increase as the distance from the corner increased.
Consequently, Fox [2] derived an alternative relation for powder consumption,
(Qt f*), in terms of distance from the corner, Dcorn (Fig. 2.10 and Eq. 2.11,); note
(Qt f*) was calculated to avoid using R* in the conversion of Qpowd t to Qslag
s .

ðQt f  Þ ¼ 0:615 D0:586

corn ð2:11Þ

The overall conclusion is that mould dimensions have a marked influence on the
required powder consumption, regardless of whether the parameters R* or Dcorn is
used. However, with thin slabs the high value of R* suggests thin slabs require little
powder consumption whereas the fact the Dcorn is quite large suggests that a high
powder consumption is needed. It is also true that powder consumption values tend
to be low in thin-slab casting. The shell in thin-slab casting tends to be very thin
34 2 Slag Infiltration, Lubrication and Frictional Forces

Fig. 2.10 Schematic diagram

showing friction increasing
with increasing distance of
corner to centre; X = friction
at centre line for 3 mould
geometries; solid line = billet;
dotted line = bloom and solid
line = slab; (courtesy of AB
Fox [2])

because of the short residence time in the mould and thus shell shrinkage and the
size of the gap tend to be low also. Some support for this view is that 400 mm thick
slabs at Dillinger (where the shrinkage is high) tend to have high powder con-
sumption (Figs. 2.9b and 2.10).

2.2.6.2 Effect of Casting Speed

The effect of casting speed on powder consumption has been investigated by a large
number of workers [1, 3, 4, 20, 21, 23, 25, 32, 42, 46–52].
In Sect. 1.4 above, it was pointed out that Wolf [41, 42] proposed that optimum
casting was obtained when ηV2c = 5 ± 3 dPas (m min−1)2 and that Ogibayashi et al.
[43] found the fluctuations were at a minimum, (for both slag film thickness and
heat flux) when ηVc = 2 ± 1 dPas (m min−1). Subsequently, Wolf [42] converted
these relations into optimum powder consumption equations, i.e. Eqs. 2.12 and
2.13, respectively.

Wolf:

s req ¼ 0:7=g
Qpowd
Vc ð2:12Þ
0:5

Ogibayashi:

sreq ¼ 0:6=g
Vc
Qpowd ð2:13Þ

An investigation using a much larger database of Qs values, subsequently

indicated a modified version of Eq. 2.12 (given in Eq. 2.14) provided a better fit
with plant data [1, 48].

sreq ¼ 0:55=g
Qslag
Vc ð2:14Þ
0:5

These equations all indicate that powder consumption decreases as the casting
speed increases.
2.2 Powder Consumption (Q) 35

2.2.6.3 Slag Viscosity

The effect of viscosity on powder consumption has been studied by a number of

investigators [1, 15, 20, 21, 25, 31–33, 38, 45, 49–54].
It can be seen from Eqs. 2.12 to 2.14 that powder consumption increases with
decreasing viscosity. The powder consumption values obtained on plant with
high-viscosity fluxes used in high-speed billet casting are significantly lower than
required Qs values calculated with Eqs. 2.12–2.14. This is discussed in
Sect. 2.2.6.10. Low consumption values are also recorded on plants casting
Ti-stabilised stainless and LC steels. This is thought to be due to the blocking off of
slag infiltration by accretions of Ti(CN) or perovskite (CaO
TiO2) at the mouth of
the shell/mould channel (see Sect. 2.2.6.9). The minimum powder consumption
providing acceptable levels of lubrication has been discussed [42]. Recommended
values of Qs for casting sticker sensitive steel grade are 0.3 and 0.4 kg m−2 for
round billets and HC steel grades, and for crack-sensitive grades, values of 0.25 and
0.4 for round billets and heavy plate, respectively [42]. For high-speed casting,
minimum values of Qs = 0.1 kg m−2 have been recommended [55]. The minimum
levels of powder consumption required to avoid solid friction (which causes frac-
ture of the slag film) has also been discussed [19].
Values of Qs for non-Newtonian mould slags have been compared with those for
conventional slags (Qs conv < Qs non-Newt) since the high shear forces in the infil-
tration region result in a reduction of their viscosity [56].

2.2.6.4 Oscillation Parameters

The effects of the various oscillation parameters have been investigated by a

number of workers [1, 4, 20, 21, 23, 31, 32, 42, 49, 51–53, 57–61]. Various
equations have been reported for the calculation of powder consumption; these
involve a number of variables (e.g. casting speed, frequency, etc.). A number of the
reported equations are given in Table 2.3; the effect of an increase in the variable on
the powder consumption is also given. Inspection of Table 2.3 indicates:
• There is reasonable agreement that increases in casting speed, viscosity and
break temperature (Tbr) all result in a decrease in powder consumption.
• There is little agreement on the effect of other variables controlling powder
consumption
• There is no agreement on the effects of the oscillation characteristics; it should
be pointed out that some workers considered slag infiltration to occur during
negative strip time tn, (i.e. when the mould is descending faster than the shell)
which results in increased pressure produced by a descending slag rim. In the
other school of thought, slag inflow is considered to be difficult in tn (since the
channel gets blocked by the bending of the shell in this period) and thus slag
infiltration largely occurs in positive strip time, tp, (where tp + tn = tcycle). It has
been reported that Qs exhibited a stronger correlation with tp than with tn [25].
36 2 Slag Infiltration, Lubrication and Frictional Forces

• Some of the experimental studies indicate that Qs decreases with increasing

frequency (Fig. 2.4) but as can be seen from Table 2.3 other studies show the
reverse relation with frequency (i.e. Qs" as f")
• Wolf [42] examined plant data where all casting variables were kept constant
except the stroke length which was altered (i.e. tn and tp will change) and
found that as the stroke increased Qs increased (Qs" as s") as shown in
Fig. 2.11. This is supported by plant data from Oxelösund Steelworks which
exhibited a strong relationship between stroke length and mould powder
consumption. However, a statistical analysis of plant data showed the reverse
trend (Qs" as s#) and a parametric study suggested that increasing stroke only
had a slight effect on Qs [25].

2.2.6.5 Non-sinusoidal Oscillation

Several workers have reported that non-sinusoidal oscillation increases powder

consumption [25, 49, 53, 60, 64, 71]. It can be seen from Fig. 2.12 that the use of
non-sinusoidal oscillation leads to a 10% increase in powder consumption.

2.2.6.6 Solidification (or Break) Temperature (Tbr)

The effect of break (or solidification) temperature on powder consumption has been
studied by a number of investigators [4, 16, 20, 51–53, 64].
It can be seen from Table 2.1 that those equations which involve Tbr (or Tsol, or
Tliq) all agree that an increase in Tbr results in a decrease in powder consumption
(Qs#). This can be seen in Fig. 2.13. This is consistent with intuition since it can be

Fig. 2.11 Powder 0.5

consumption as a function of
casting speed showing Qs
increasing with increasing 0.4
stroke length; , s = 8 mm;
— = 6 mm;
Qs, kg m-2

= experimental data
(permission granted, 0.3
VDEh/Stahl Eisen, [42])

0.2

0.1
1 2 3 4 5 6
-1
Vc , m min
2.2 Powder Consumption (Q) 37

0.5

0.4
Qs, kg m -2

0.3

0.2
1.2 1.6 2 2.4 2.8
Vc , m min-1

Fig. 2.12 Powder consumption as a function of casting speed showing non-sinusoidal oscillation
) provides a 10% higher Qs than sinusoidal oscillation ( ) (permission granted, ISIJ, [49])

Fig. 2.13 Powder 0.5

consumption, Qs, as a
function of casting speed
showing Qs decreasing as Tsol 0.4
or Tbr increases as Tsol or Tbr
increases; = experimental
Qs, kg m -2

values; Tbr; , = 952 °C;

0.3
— = 970 °C, - - - = 1027 °C;
(permission granted, VDEh/
Stahl Eisen, [42], re-drawn)
0.2

0.1
1 2 3 4 5 6

Vc , m min-1

seen from Fig. 2.14 that an increase in Tbr leads to a decrease in thickness of the
liquid film (dl#) and an increase in solid film thickness (ds").
The mathematical model reported by Ramirez–Lopez [1, 21, 54] indicated that
the horizontal heat flux increased gradually in the period of the oscillation cycle
where the mould was descending. The increased heat flux, in turn, caused solid slag
to melt, resulting in an increase in liquid film thickness (dl") and a decrease in the
solid film thickness (ds#). This process was reversed when the mould ascended.
Thus, the liquid layer increased at the expense of the solid layer when the heat flux
was increasing and vice versa when the heat flux was decreasing.
38 2 Slag Infiltration, Lubrication and Frictional Forces

(a) (b)

Fig. 2.14 Schematic drawing showing effect of increasing Tbr on the thicknesses of solid and
liquid films, where Tbr(b) > Tbr(a) (note relative thickness values are not to scale)

2.2.6.7 Melting Rate

It was pointed out in Sect. 2.2.1 that powder consumption (in kg min−1 or kg s−1) in
Eq. 2.5 (Qslag
MR ) can be viewed as the melting rate. The powder consumption has been
found to increase with increasing slag pool depth (dpool) [72] and dpool is affected by
the melting rate. In practice, the melting rate is controlled by the amount of free
carbon and, to a lesser extent, by the size of the carbon particles (see Sect. 4.3).
Consequently, it is important that powder consumption is close to the required Qs for
the given casting conditions (mould dimensions, casting speed, etc.), i.e. it is not
restricted by an excessively high carbon content of the mould powder. A few casting
powders have a carbon content which is too high and thus leads to a restricted slag
infiltration.

2.2.6.8 Superheat (DT)

The effect of superheat on powder consumption has been studied by several

investigators [1, 3, 16]. Increasing superheat results in increased powder con-
sumption. On the basis of Eq. 2.12 or Eq. 2.13 increases in superheat will lead,
sequentially, to an increase in slag temperature, a lower viscosity and a higher value
of Qs.

2.2.6.9 Argon Flow

It has been reported that powder consumption increases with increasing argon flow
rate [73] (Fig. 2.15). One possible reason for this behaviour is that an increased Ar
flow rate causes more convection and hence, a higher vertical heat flux which, in
turn, increases the melting rate. A high Ar flow rate is known to affect the metal
flow patterns in the mould.
2.2 Powder Consumption (Q) 39

Fig. 2.15 Plant 0.8

measurements of powder
consumption, Qt, as a

, kg tonne -1
function of Ar flow rate;
(courtesy of Fox [2]) 0.6

powd
0.4

Qt
0.2
0 1 2 3 4 5
Ar flow rate, l min-1

2.2.6.10 Continuous Casting of Steels Containing Ti

It has been observed that powder consumption is frequently lower than predicted
when casting steel grades containing Ti. This is thought to be due to the formation
of TiN or Ti(C, N) which has a low solubility in the slag pool and thus, tends to
exist as solid particles [74, 75]. These particles agglomerate through turbulent
collisions and the agglomerates restrict the slag flow when they are sited in the
mouth of slag/mould channel (Fig. 2.16) [74, 76]. The solid particles also increase
the slag viscosity and thus, decrease Qs. Alternatively, TiO2 particles can form
perovskite (CaO.TiO2) which has a high melting temperature and thus reduces both
the thickness of the liquid slag film (dl) and Qs. It is necessary to keep the basicity,
(C/S) < 1.0 to avoid perovskite formation [77].

2.2.6.11 High-Viscosity Powders

The powder consumption data for most powders follow the empirical rules based on
the viscosity and casting speed [2, 42, 43] the only exceptions are the high-viscosity
(η1300 = 10–30 dPas) powders used for high-speed billet casting [2] and those used
in casting steel grades containing Ti. In high-speed billet casting, considerable
turbulence is generated which results in significant levels of slag entrapment. One
way of reducing slag entrapment is to increase the slag viscosity. However, the
reduction in slag entrapment levels comes at the expense of a significant decrease in
powder consumption. Fortunately, the powder consumption required for billets is
low (since the distance from centre line to corner is low and, furthermore, R* has
values > 22) which means this practice is widely used to minimise slag entrapment.
It has been suggested that these high-viscosity slags will form super-cooled
liquids (scl) rather than crystallites during cooling. The scl, although viscous, will
move in response to any stress applied by the ferro static pressure, the downward
movement of the shell or the oscillating motion of the mould; hence, the slag (scl)
40 2 Slag Infiltration, Lubrication and Frictional Forces

Fig. 2.16 Schematic drawing showing formation of agglomerates of Ti(C, N) when casting steels
containing Ti; (permission granted, ISS/AIST [74])

will supply some lubrication to the shell. The importance of retaining some glass
phase in the slag film has been pointed out by Hanao et al. [78].

2.2.6.12 Electromagnetic Braking (EMBr) and Casting (EMC)

Electromagnetic devices are reported to increase powder consumption. The appli-

cation of electromagnetic braking (EMBr) results in a 30% decrease in vertical heat
transfer [79] from the steel, resulting in a ca. 10 °C increase in meniscus temper-
ature [80–83]. This increased meniscus temperature results in a lower slag viscosity
and higher powder consumption. EMBr is widely used in high-speed, thin-slab
casting where powder consumption tends to be low [84].
Increases of 20% in Qs, have been reported when using pulsative,
Electromagnetic casting (EMC) [85]; it is known that the pinch force in EMC
reduces the horizontal heat transfer, which, in turn, results in increases in both
meniscus temperature and Qs.
2.2 Powder Consumption (Q) 41

2.2.6.13 Liquid Slag Feeding to the Mould

Some high Al-and Mn-steels have low-melting temperatures, so the vertical heat
flux is insufficient to melt the mould powder. Consequently, liquid slag feeding
technology has been developed to provide liquid slag to the steel surface. It is
reported that powder consumption is increased with liquid slag feeding [86].

2.3 Slag Infiltration During the Oscillation Cycle

A number of empirical rules have been proposed to calculate the powder con-
sumption. These are given in Table 2.3 and are based on plant observations and
physical modelling results. It was mentioned above that there is general agreement
that the powder consumption increases as the casting speed (Vc), slag viscosity (η)
and break temperature (Tbr) all decrease. However, there is no consensus as to
(i) which of the various oscillation parameters affect Qs or (ii) the way in which they
affect Qs. Furthermore, there are two schools of thought concerning the period of
the oscillation cycle where slag infiltration occurs;
• The first identifies tn as the primary period of slag infiltration, where the
descending slag rim increases the pressure on the molten slag which responds by
infiltrating into the shell/mould channel and
• The second considers tp as the principal period of infiltration, since infiltration is
restricted in tn because the bending-back of the shell is considered to block off
the slag flow during this period; thus slag infiltration is restricted to tp where the
shell does not interfere with the infiltration.
The infiltration mechanism has been studied in (i) plant trials [3, 55, 87–89] (ii) cold
modelling studies [5, 6, 12, 58, 90] (iii) hot modelling studies [7, 53, 91] and
(iv) mathematical modelling of the heat and fluid flow [1, 12, 16, 21, 22, 25, 36, 92, 93].
Mathematical models based on Navier–Stokes equations do provide a reasonable
description of the effects of casting speed, slag viscosity and Tbr, but, for the most
part, they also predict that Qs increases with increasing frequency which disagrees
with most experimental observations, e.g. Fig. 2.4. In an attempt to explain these
discrepancies, mathematical models have been developed to explore in which part
of the oscillation cycle the slag infiltration takes place [1, 21, 22, 25, 54].
It is customary to characterise oscillation in terms of negative and positive strip
times (tn and tp, respectively) where tn represents the time when the mould is
descending faster than the shell and tp constitutes the remainder of the cycle (i.e.
tn + tp = tcycle = 60/f where f is in cpm). However, the oscillation cycle can also be
characterised in terms of the position of the mould. The mould and the slag rim will
be at their highest position in late tp (denoted tlate
p ). It descends throughout tn and
reaches its lowest position in early tp (tearly
p ). The mould will then ascend steadily
through tp.
42 2 Slag Infiltration, Lubrication and Frictional Forces

The findings of the two studies due to Ojeda [22] and Ramirez–Lopez [1, 21, 54]
are in good agreement with both proposing that slag infiltration occurs during the
descent of the mould/rim covering the period (tn–tearlyp ) and that there is little slag
infiltration during the ascent of the mould (in tp). This can be seen in Fig. 2.17c and it
was also noted that the rate of slag flow into the shell/ mould channel is at its highest
when the mould/slag rim was at its lowest position (i.e. between tlate n and tp
early
).
The directions of flow in the slag pool at different periods of the oscillation cycle
have also been studied in the several investigations [1, 21, 22, 54] and are shown in
Fig. 2.18. It can be seen when the mould and slag rim are at their highest position
(Fig. 2.18a) that the slag flow into the mouth of the channel is radially outward and
upward. As the mould/rim descends the slag direction changes to downward and
there is evidence of a vortex in the region of the mouth (Fig. 2.18b).When the
mould and slag rim reach their lowest positions (in tearly p ) the flow is strongly
downward into the channel (Fig. 2.18c). Finally, halfway through tp the flow
changes direction to radially—outward and upward (Fig. 2.18d). Thus the direction
of the flow in the slag pool plays a significant role in the slag infiltration into the
shell/ mould channel and this is affected by the direction of movement of the mould
and slag rim.
The slag rim acts like a piston and helps to inject slag into the channel. However,
the movement of the mould alone will cause some downward flow of slag but the
slag rim certainly accentuates the downward slag flow.
Just after the mould reaches its highest position there is a tide—change in the
slag flow (radially outward and upward to downward) which results in a period of
“confused flow” (the remnants of which are shown in Fig. 2.18b). There is a similar
period of confused flow after the mould and rim reach their lowest points. There is
very little slag infiltration during these periods of confused flow. It has been sug-
gested that the lack of slag infiltration in the periods following a tide-change is
responsible for the failure of the models based on the Navier–Stokes equation to
predict the correct Qs dependency on frequency (namely, Qs# as f") observed on
plant [1]. For example, if f = 60 cpm and is increased to 120 cpm, there will twice
as many tide-changes per unit time. Since little powder consumption occurs during
these tide-changes, the increased number of tide-changes per minute will result in
an overall decrease in Qs. A parametric study showed [25] indicated that a 60%
increase in frequency resulted in only 2% change in Qs (where both increases and
decreases in Qs were recorded for different stroke lengths). However, it was found
Qcycle (kg m−1cycle−1) decreased by 35% with a 60% increase in f.
In summary, slag infiltration occurs through the downward flow of slag resulting
from the downward movement of the mould; the slag rim accentuates this slag flow.
The size of the slag rim is dependent upon the steel grade being cast, with high
basicity (C/S) slags (used for MC steels) forming large rims and low-(C/S) slags
forming smaller rims (see Fig. 1.4).
2.3 Slag Infiltration During the Oscillation Cycle 43

Fig. 2.17 Mathematical model predictions of a profile of strand surface, b heat flux, c powder
consumption in kg s−1, d liquid film thickness dl, e solid slag film thickness, ds and f pressure
during five oscillation cycles [1, 21, 54] the dotted, vertical lines indicate the onset (left) and end
(right) of negative strip periods of each cycle (0–5). Note that peaks in Q and dl lie in early tp;
(permission granted, ISIJ, [21])
44 2 Slag Infiltration, Lubrication and Frictional Forces

Fig. 2.18 Schematic diagram showing the direction of slag flow at different parts of the
oscillation cycle a at highest position of mould in late tp, b halfway through tn, c at lowest position
of mould in early tp and d midway through tp [1, 21, 54] (permission granted, ISIJ, [21])

2.4 Empirical Equations for Calculating Powder

Consumption

A number of empirical rules have been proposed to calculate the powder con-
sumption; the proposed equations are given in Table 2.3. Fox [2] carried out an
evaluation of the various empirical equations. This evaluation compared predictions
with plant measurements contained in an extensive database of powder consump-
tion and casting variables for a large number of trials carried out at different
steelworks casting slabs, blooms, billets and thin slabs. It should be pointed out that
the database contained powder consumption for high-viscosity billet powders used
to minimise slag entrapment at high casting speeds; Qs data for these slags are much
lower than for other powders and tend to distort the fit. The performance was
judged from DRMS which is calculated from Eqs. 2.15 and 2.16 where N = the
number of mould slags. The best performing models for this database were found to
be in the hierarchy, Ogibayashi [42, 43] > Kobayashi [69] > modified Wolf [1,
2] > Maeda [49]


d ¼ 100 Qmeas  Qpred =Qmeas ð2:15Þ

0:5
DRMS ¼ R d21 þ d22 þ d23 þ


=N ð2:16Þ

Analysis of plant data for powder consumption for slab-, bloom- and billet-
casting indicated that Qslag
t ( = f*
Qpowd
t ) is reasonably constant at 0.48 kg (tonne
−1
steel) except for the “high-viscosity billet powders” mentioned above in
Sect. 2.2.6.10.
2.4 Empirical Equations for Calculating Powder Consumption 45

2.4.1 Frictional Forces

Frictional forces (F) acting on the shell contain contributions from the liquid
frictional (Fl) and solid frictional (Fs) forces (Eq. 2.17).

F ¼ Fl þ Fs ð2:17Þ

It was pointed out in Eq. 2.1 that the liquid frictional force was inversely,
dependent on the thickness of the liquid layer (dl). Since dl = (Qs/ql) Eq. 2.1 can be
re-written as
Fl ¼ A
gðVm  Vc Þql =Qs : ð2:18Þ

It can be seen that that the liquid friction force is inversely dependent upon the
powder consumption and directly related to the slag viscosity and the difference
between the velocities of the mould and the strand.
The frictional forces tend to be highest when casting MC steels because the
corrugated shell formed increases both A and Fl and dl tends to low because high
Tbr slags are used to cast these steels. The friction forces measured when casting
with mould powders are lower than those for oil casting [94]. Longitudinal cracking
has been correlated with high frictional forces [95]. Frictional forces tend to
decrease as the cast proceeds [95].
Solid friction (between the shell and the solid slag) tends to occur in the lower half
of the mould. The formation of star and spongy cracks (Sect. 11.7) is associated
with solid/solid friction and the consequent spalling of the solid slag film which can
even result in the pick-up of copper by the strand. However, solid friction may also
occur in the upper mould in the corner regions if the corners are overcooled [4].
Since solid friction can occur in the bottom of the mould it is important to ensure
that all of the mould receives liquid lubrication [96]. A lubrication index (LI) was
proposed by Billany et al. [96] which is a measure of the fraction of the mould
enjoying liquid lubrication; this index is defined in Eq. 2.19. Ideally, the parameter,
LI, should have a value of 1.0.

LI ¼ðDistance from meniscus to point where T ¼ Tbr Þ=

ð2:19Þ
ðDistance  meniscus to mould exitÞ

If solid friction is a problem in the lower half of the mould, probably the best
measure is to increase the casting speed. In theory, a decrease in the flow rate of the
cooling water would also be beneficial but in practice, the remedial effect is small.
Sorimachi et al. [97] have pointed out that:
• The frictional forces refer to the entire mould wall and not to the local frictional
forces in the meniscus region which is of key importance in the formation of
sticker breakouts.
• The measured friction is that acting on the mould wall and is not that acting on
the shell.
46 2 Slag Infiltration, Lubrication and Frictional Forces

2.4.1.1 Measurement of Frictional Forces

Measurements on frictional forces have been derived using plant trials, simulation
experiments and mathematical modelling of the frictional forces.
Plant Measurements of Friction
In the past, a number of investigators have measured frictional forces by using load
cells attached to the oscillating mechanism and then applying Fourier analysis of
the signals produced [27, 71, 98]. Alternatively, frictional forces can be measured
using the MLTEKTOR system [99]. These devices have been used for the detection
of defects and longitudinal cracks [44, 100, 101].
Friction Measurements in Simulation Tests
Several tests have been devised to simulate the frictional forces acting on the shell
when using different mould fluxes. Short descriptions of these tests are given below.
Rotating Cold Finger Test
In this test, a water-cooled, copper finger (representing the mould) is rotated in a
steel crucible (representing the strand) containing the molten mould flux [98].
The copper finger becomes covered with a solid slag film of ca. 3 mm thickness and
a thin liquid layer. The frictional forces are measured by determining the torque
developed on the steel crucible as the finger is rotated at constant velocity
(10–50 rpm). The apparatus is shown in Fig. 2.19a.
Oscillating Cold Finger Test
This test resembles the rotating cold finger test but the cold finger is oscillated
vertically instead of being rotated [9, 11, 53, 102].

Fig. 2.19 Schematic drawings showing the apparatus used in a the rotating cold finger test [98] and
b the oscillating pad method [27, 103] (permission granted, Europ. Comm. Sci. and Tech. Publ. [27])
2.4 Empirical Equations for Calculating Powder Consumption 47

Oscillating Pad Test

In these tests, a water-cooled copper pad (mould), which can be oscillated at
different frequencies, is lowered onto a heated steel block (strand) covered with
molten mould slag (Fig. 2.19b). The block is then withdrawn at a fixed speed and
the frictional force exerted by the pad is measured as it bears down on the strand by
using a load cell mounted on the oscillation arm. The thickness of the molten slag
layer (ca. 0.3 mm) was monitored using a displacement transducer [27, 103].
Rotating and Oscillating Pads Test
A simulation test was reported by Sorimachi et al. [97] and is shown in Fig. 2.20. In
this test a graphite disc (representing the strand is rotated unidirectionally).
A second, lower, graphite disc (representing the mould) contains a 2 mm deep
liquid, mould slag of known viscosity; and this disc is oscillated sinusoidally. The
torque is measured continuously.
Miniature Continuous Caster
Friction measurements have been carried out in a miniature continuous caster [91].
In order to view the solidification process of the shell the steel was replaced by
Sn–5%Pb, the mould slag by stearic acid with Al2O3 particles to act as tracers and
one side of the oscillating, 50 mm2, Cu mould was replaced by silica to facilitate
viewing [91]. The friction between mould and shell was measured by load cells sited

Fig. 2.20 Schematic drawing showing the apparatus used by Sorimachi et al. [97] (permission
granted, ISS/AIST [97])
48 2 Slag Infiltration, Lubrication and Frictional Forces

below the mould. The frictional force per unit area between mould and shell was
taken as DF/A = (Fmax – Fmin)/A where A = surface area and the subscripts max and
min represent the maximum and minimum load in any one cycle, respectively.
Values of DF/A were found to decrease as liquid film thickness (dl) increased and
(Vm − Vc)/ dl decreased (i.e. DF/A# as dl" and as {(Vm − Vc)/ dl}#) [91]. Friction
measurements can be made in a similar manner in mould simulators [9–11].

2.4.1.2 Mathematical Modelling of Friction in Mould

Mathematical modelling of the frictional forces in the mould has been reported
[12, 18, 53, 97]. Schwerdtfeger and Tacke [18] derived a relation for the shear stress
in the liquid slag based on computations of the velocity. The frictional force was
calculated by multiplying the calculated stress by the area wetted by the slag.

2.4.2 Factors Affecting Frictional Forces in the Mould

It can be seen from Eq. 2.18 (Fl = Aη (Vm − Vc) ql/Qs) that the liquid friction force
(Fl) is inversely dependent upon the powder consumption (Qs); thus it follows Fl
will increase as Qs decreases (Fl" as Qs#).However, Qs is dependent upon other
factors, e.g. casting speed; the effect of the various parameters are given below.
However, high friction measurements recorded on plant (i.e. 10–20 kPa) have
been attributed to (i) movement of the solid slag layer (ii) excessive taper and
(iii) mould misalignment [19]. At low casting speeds the critical consumption is
high so variations in consumption, Qs, can lead to slag film fracture and high, solid
friction forces. In contrast, at high casting speeds the principal causes of high
frictional forces are excessive taper (see Fig. 1.48) and mould misalignment [19].

2.4.2.1 Casting Speed (Vc)

The powder consumption, Qs, is dependent upon (Vc)−1 (Eqs. 2.12–2.14). Using the
modified Wolf relation (Qs = 0.55/η0.5 Vc) to demonstrate the effect of casting
speed, it can be seen that Eq. 2.18 can be re-written as

Fl ¼ AgðVm  Vc Þql g0:5 Vc =0:55 ð2:20Þ

Similar relations could be derived with other relationships for Qs. It can be seen
from Eq. 2.20 that an increase in casting speed causes a decrease in the (Vm − Vc)
term, in addition, to the increase in the Vc term. These conflicting responses to a Vc
increase, result in a minimum (Vminc ) in the Fl–Vc plot shown in Fig. 2.21 reported
by D’Haeyer [99]. It can be seen that the (Vm − Vc) is dominant at low speeds and
the Vc term tends to dominate at higher casting speeds.
2.4 Empirical Equations for Calculating Powder Consumption 49

Fig. 2.21 Frictional force as (a) 100

a function of casting speed
a friction signal from ML 80
Tektor, b DFmax and

Signal, %
c frictional force using two 60
mould powders,
Q (η1300 = 0.9 dPas) = , ; R 40
(η1300 = 0.8 dPas) = ) [99];
(permission granted, 20
Europ. Comm. Sci. and Tech.
Publ., [99]) 0
0 0.5 1 1.5 2

Vc, m min-1

(b) 250

ΔFmax, g. cm-2 200

150

100

50

0
0 0.5 1 1.5 2
Vc, m min-1

(c) 200

180
Fric on force. kNm-2

160

140

120

100
1.5 1.7 1.9 2.1 2.3 2.5

Vc, m min-1

Tsutsumi et al. [91] carried out simulation experiments and reported that Fl
decreased:
• As casting speed increased (Fl# as Vc") indicating, Vc < Vmin c in their
experiments.
• As the liquid slag film thickness (dl) increased (Fl# as dl").
• As the velocity gradient ((Vm − Vc)/ dl) decreases.
• With the introduction of non-sinusoidal oscillation.
50 2 Slag Infiltration, Lubrication and Frictional Forces

2.4.2.2 Viscosity (η)

According to Eq. 2.20 the liquid frictional force is a function of (η1.5) and the
Ogibayashi [42, 43] relation (Eq. 2.13) leads to Fl exhibiting a dependence on (η2).
Thus the frictional forces increase with increasing viscosity. Wolf [42] reported a
minimum in the Fl- (η0.5 Vc) plot (Fig. 2.6) at 5 ± 2 (dPas)0.5 ⋅ m min−1; the
equivalent plot for the Ogibayashi relation for Qs leads to a minimum at
2 ± 1 dPas. m min−1 as shown in Fig. 2.7c [44].
Hering et al. [104] reported that the liquid frictional force increased with
increasing Al2O3 content in the casting slag (Fig. 2.22a). This is presumably due to
the increase in viscosity with increasing Al2O3 content. However, Hering et al.
[104] found that this was not always the case, (as can be seen from Fig. 2.22a) and
proposed that the friction was affected the nature of the mineralogical phase formed.
It is possible that with the wollastonite/gehlenite curve in Fig. 2.22b could be
explained by a lowering of Tbr with increasing Al2O3 which offsets the effect of
increasing viscosity.

Fig. 2.22 Frictional force as 30

function of a Al2O3 content
25
Fric on force. kNm-2

and b viscosity (cited in Pas,

thus multiply by 10 for dPas 20
[104]); (permission granted,
Stahl Eisen, [104], re-drawn) 15

10

0
0 4 8 12
Al 2O 3, mass %
2.4 Empirical Equations for Calculating Powder Consumption 51

Fig. 2.23 The apparent 0.6

frictional coefficient as a

Apparent fric on coeff.,

function of the unevenness (or 0.5
irregularity) of the shell [105]
0.4
for Vc = 2.0 (—) and
4.0 m min−1 (- - -) 0.3
(permission granted,
ISS/AIST, [105]) 0.2

0.1

0
0 10 20 30 40
Shell irregularity, %

2.4.2.3 Mould Dimensions and Surface Area (A)

It can be seen from Eq. 2.1 and Fig. 2.20 that the liquid frictional force (Fl) increases
as the surface area of the mould (or shell) increases. Ogibayashi et al. [105] pointed
out that the shrinkage of the steel will be greatest at the centreline, the point where
the shell is at its thinnest. Ogibayashi et al. [105] also pointed out that in peritectic,
MC steels the shell (in the meniscus region) tends to become uneven or corrugated;
this unevenness increases the surface area of the shell. Consequently, the friction
coefficient tends to increase as the unevenness of the shell increases (Fig. 2.23).

2.4.2.4 Break (or Solidification) Temperature (Tbr)

Increases in break temperature would be expected to reduce the thickness of the

liquid slag film (dl), as shown in Fig. 2.14; this would result in higher frictional
forces (Fl" as Tbr"). Furthermore, increasing Tbr will also enhance the amount of
solid friction. Thus on both counts the friction will tend to increase as Tbr increases.
Measurements of friction and friction coefficient are shown in Fig. 2.24a, b,
respectively. These figures show that there is a sharp increase in friction at a
temperature slightly below the break temperature, Tbr; this may imply that the
cooling rates in the friction experiments were slightly higher than those used in the
viscosity experiments, since Tbr decreases with increasing cooling rate. These fig-
ures show that relatively small amounts of solid friction can have a significant effect
on the total friction. The rate of friction rise was much greater in some cases,
denoted. Type A, e.g. Powder J with sharp Tbr temperatures) than in others (Type B
i.e. more “glassy” slags, e.g. Powder A).

2.4.2.5 Frequency (f)

Since most plant observations indicate that powder consumption Qs decreases as

f increases, it is expected that an increase in frequency would increase the liquid
52 2 Slag Infiltration, Lubrication and Frictional Forces

Fig. 2.24 a Frictional force [98] and b Coefficient of friction [27, 103] as functions of temperature
derived in rotating cold finger and oscillating pad tests, respectively (permission granted,
Europ. Comm. Sci. and Tech. Publ. [27, 98])

friction (Fl" as f"). This relationship (Fl" as f") has been confirmed by several
investigators [53, 64, 106]. It should be noted that an increase in frequency also
increases the velocity of the mould (Vm) and it can be seen from Eq. 2.18 that an
increase in Vm will result in an increase in friction.

2.4.2.6 Stroke Length (S)

It can be seen from Table 2.3 that there is no consensus on the effect of the stroke
length (s) on the powder consumption, Qs. The statistical analysis of plant data due
to Saraswat et al. [4] indicates that Qs decreases as the stroke increases (Qs# as s");
on this basis, Fl would be expected to increase (i.e. Fl" as s"). Other workers have
reported that Qs increases with increasing stroke (Qs" as s") [53, 64] which would
result in (Fl" as s#).However, Qs (or dl) is not the only factor affecting friction
forces and an increased stroke would lead to increased values for Vm and (Vm − Vc)
and Fl. Thus no relation between Fl and s can be recommended at this stage.

2.4.2.7 Negative and Positive Strip Time (Tn and Tp)

As mentioned above, there has been considerable debate as to whether powder

consumption occurs in negative strip time or positive strip time. Mathematical
models [1, 21, 22] indicate that slag infiltration occurs predominantly in the period
when the mould (plus rim) are descending, with the infiltration rate being at its
highest in late tn and early tp. Thus it may be concluded that increased negative strip
would increase Qs and hence decrease Fl.
2.4 Empirical Equations for Calculating Powder Consumption 53

However, Tsutsumi et al. [53, 64] reported that in their simulation experiments
that increased positive strip resulted in a decrease in Fl. It has also been reported
that the correlation of powder consumption with tp is stronger than that with tn [25].

2.4.2.8 Steel Temperature

It has been reported that a decrease in steel temperature results in increased friction
measurements, presumably due to the effect of the slag viscosities (which increase
at lower temperatures) on the friction.

2.4.2.9 Non-sinusoidal Oscillation

Non-sinusoidal oscillation has been reported to reduce liquid friction [60, 71, 99,
106–108], however, the reverse relation (Fl" as NSO") has been found in trials in
Sweden [109]. Mizukami et al. [60] reported that non- sinusoidal oscillation can
result in a 40% decrease in liquid friction compared with conventional, sinusoidal
oscillation (Fig. 2.25a) but only resulted in a 5% decrease in solid friction (Fs).

Fig. 2.25 Frictional force, Fl, (a) 16

as a function of casting speed
a showing differences 14
between sinusoidal ( ) 12
and two modes of
Fl max ,Nm-2

10
non-sinusoidal oscillation
8
( and ) and
b Fl in relation to tensile 6
strength (rB, rsutf, in 4
N mm−2) denoted in curves, 2
for both the average shell
0
temperature ( ) and the
1.8 2 2.2 2.4 2.6 2.8 3
surface temperature (o) [60]
(permission granted, ISIJ, Vc , m min-1
[60] re-drawn)
(b) 1

0.8

0.6
σ,

0.4

0.2

0
0 2 4 6 8 10 12
Vc , m min-1
54 2 Slag Infiltration, Lubrication and Frictional Forces

Non-sinusoidal oscillation results in a decrease in (Vm − Vc) which leads to a

concomitant decrease in the liquid friction (Fl) as shown in Eqs. 2.18 and 2.20. The
frictional forces were compared with the tensile strength of the steel (Fig. 2.25a)
from which it was concluded that the upper limit for the casting speed when casting
with an oscillating mould, lay between 5 and 8 m min−1 [60].

2.5 Summary

The following observations were made concerning the lubrication of the shell and
the frictional forces acting on it:
(i) Inadequate lubrication of the shell leads to various defects in the steel
product, such as, longitudinal cracks, sticker breakouts and star cracks.
(ii) The lubrication is supplied by the liquid slag infiltrating into the channel
between mould and shell; this occurs principally in the period where the
mould (and slag rim) are descending and Qs increases gradually through this
period with maximum infiltration corresponding with the lowest position of
the mould.
(iii) Changes in mould direction are accompanied by periods of confused flow
where little slag infiltration occurs.
(iv) The powder consumption, Qs (in kg m−2) provides a good measure of the
lubrication supplied and is related to the thickness of the liquid slag layer in
the slag film (Qreq
s = q
dl) where q is the density of the liquid slag.
(v) Several powder consumption terms are used and these terms are interre-
lated; the melting (QMR) rate must match the required powder consumption.
(vi) Since mould powders contain carbon and volatile materials it is necessary to
distinguish between powder and slag (Qslag s = f* Qpowd
s ) where f* is the
mass fraction of the powder forming slag
(vii) Analyses of plant data for powder consumption revealed that Qreq s increases:

– with increasing mould surface area, (Qreq

s = 2/ (R* –5))
– with decreasing casting speed and slag viscosity
– with decreasing oscillation frequency and stroke (although these rela-
tions are disputed by some workers); these effects are smaller than those
above.
– with increasing Argon flow rate.
(viii) The required values of viscosity, break temperature and Qreq
s can be calculated
for the given casting conditions using empirical rules
(ix) High frictional forces in the mould arise from (i) fracture of slag films at low
casting speeds and (ii) excessive taper and mould misalignment at higher
casting speeds
(x) Liquid friction increases with the:
2.5 Summary 55

– Increasing surface area of the shell (including any shell corrugations)

– Increasing viscosity, casting speed and (Vm − Vc); the plot of Fl versus
Vc exhibits a minimum due to the conflicting responses to increasing Vc
on (Vm − Vc) and Fl with (Vm − Vc) dominant at low casting speeds and
the Vc effect being dominant at high speeds.
– Increasing oscillation frequency but there is no consensus regarding the
effect of stroke length.
– With decreasing non-sinusoidal oscillation.

References

1. P.E. Ramirez-Lopez, K.C. Mills, P.D. Lee, B. Santilanna, Met. Mater. Trans. 43, 109,
(2011).
2. A.B. Fox, PhD Thesis, Mould fluxes their properties and performance. Dept. of Materials,
Imperial College, London, (2003).
3. Y Nuri, T Ohashi, N Miyasaka, K Shima, Y Uchida, Trans. ISIJ, 20, B170, (1980).
4. R. Saraswat, A.B. Fox, K.C. Mills, P.D. Lee, B. Deo, Scand. J. Met., 33, 85, (2004).
5. T. Kajitani, K. Okazawa, W. Yamada, H. Nakamura, ISIJ Intl. 46, 250, (2006).
6. M.S. Jenkins, PhD Thesis “Heat transfer in the continuous casting mould”, Monash Univ.,
Clayton, Vic., Australia, (1999).
7. Y. Itoh, S Nebeshima, K Sorimachi, Proc. 6th Intl. Conf. Molten slags, fluxes and salts,
Stockholm, Paper 152 (2000) see also S Nebeshima, Y Itoh, H Tozawa, H Nakato,
K Sorimachi, Proc. 4th Intl. Conf. Solidification Processing, Sheffield, 1997 (Sheffield Univ.,
2997) p 10.
8. K. Tsutsumi, Tetsu- to Hagane, 84, 617, (1998).
9. A. Badri, B.T. Natarajan, CC Snyder, K.D. Powers, F.J. Byrne, M. Byrne, A.W. Cramb,
Met. Mater. Trans. B, 36B, 355, (2005).
10. A. Badri, B.T. Natarajan, C.C. Snyder, K.D. Powers, F.J. Byrne, M. Byrne, A.W. Cramb,
Met. Mater. Trans B, 36B, 373, (2005).
11. EY Ko, J Choi, JY Park, I Sohn, Met. Mater. Intl. 20, 141 and 1103 (2014).
12. E Anzai, T Shigezumi, T Nakano, T Ando, M Ikeda, Nippon Steel Technical Report, 34, 35,
(1987).
13. C. Niggel, F. Felder. Lubrication by slag of continuous casting of steel. Report 9339, ECSC,
(Europ. Comm. Sci and Tech. Publ., Luxembourg, 1985).
14. J. Kor, “An analysis of the fluid flow of liquid mould flux into space between continuous
casting mold and steel shell” US Steel Report.
15. Y. Nuri, T. Ohashi, Trans. ISIJ, 20, B172, (1980).
16. S. Ogibayashi, Proc. 85th Steelmaking Conf. (2002) (ISS/AIME, Warrendale PA.) p. 175.
17. K. Okazawa,T. Kajitani, W. Yamada, H. Nakamura, ISIJ Intl., 46, 226 and 234 (2006).
18. K. Schwerdtfeger, K.H. Tacke, Fundamental study of behaviour of casting powders;
Report EUR 9560, (Europ. Comm.Sci. and Tech Publ., Luxembourg, 1985).
19. YA Meng, BG Thomas, Met. Mater. Trans. B, 34B, 707, (2003).
20. S. Itoyama, CAMP- ISIJ, 14, 893, (2001).
21. P.E. Ramirez-Lopez, K.C. Mills, P.D. Lee, B. Santilanna, ISIJ Intl., 30, 1797, (2010).
22. C. Ojeda, J. Sengupta, B.G. Thomas, J. Barco, J.L. Aruna, Proc. AIST Tech., 2006 vol 1
(ISS, Warrendale, PA) p. 1017.
23. H.J. Shin, S.H. Kim, B.G. Thomas, G.G. Lee, J.M. Park, J. Sengupta, ISIJ Intl., 46, 1635,
(2006).
56 2 Slag Infiltration, Lubrication and Frictional Forces

24. XN Meng, MY Zhu, Can. Metall. Q, 50, 45, (2011).

25. ASM Jonayat, BG Thomas, Met. Mater. Trans. B, 45B, 1862, (2014).
26. T.H. Billany, K.C. Mills, Mould flux performance during continuous casting. Final
Report ECSC Contract 7210 CA 820, 1987, (Europ. Comm. Sci. and Tech. Publ.
Luxembourg, 1987).
27. R.J. Gray, Behaviour of mould fluxes during continuous casting. Report EUR 9495 EN
1985) (Europ. Comm. Sci. and Tech. Publ., Luxembourg.1985.
28. T. Okazaki et al., Tetsu-to Hagane, 65 (10), 265, (1985).
29. JM Hill, YH Wu, B Witwatanapataphee, J Eng. Math., 36, 311, (1999).
30. S Ogibayashi, CAMP- ISIJ, 18, 126 and 127 (2003).
31. T Emi, H Nakato, K Suzuki, Y Iida, T Ueda, Tetsu- to Hagane, 60 (7), 981, (1974) Henry
Brutcher Translation HB9357.
32. T Fastner, C Furst, HP Narzt, G Xia, G Zuba, Proc. 3rd Europ. Conf. Continuous Casting
Madrid, (UNESID, Madrid,1998) p. 791.
33. Y Fukuda, H Kawai, M Okimori, M Hojo, S Tanaka, Proc. 5th Intl. Conf. Slags, Fluxes and
molten salts, Sydney,1997, (ISS, Warrendale, PA, 1997) p. 791.
34. S Ogibayashi, T Mizoguchi, T Kajatani, Intl. Workshop on Thermophys. Data for the
Development of Mathematical models of solidification, Gifu City, Japan (1995).
35. BG Thomas, A Moitra, R McDavid, Iron and Steelmaker, 23 (4), 51 (1996).
36. Y. Meng, B.G. Thomas, Proc. ISS Tech., Indianapolis, 2003 (ISS, Warrendale, PA, 2003)
p. 589.
37. MS Jenkins Proc. 78th Steelmaking Conf.,1995, (ISS, Warrendale, PA, 1995) p. 669.
38. MS Jenkins, BG Thomas, WC Chen, RB Mahapatra, Proc. 77th Steelmaking Conf., 1994,
(ISS, Warrendale, PA 1994) p. 337.
39. JW Kim, S. K. Kim, D. S. Kim, Y. D. Lee, J. I. Eum, E. S. Lee., Proc. 78th Steelmaking
Conf. 1995, (ISS, Warrendale, PA,1994) p. 333.
40. K Suzuki, C Matsumura, H Yamamoto, Y Kanrda, Proc. 73rd Steelmaking Conf., 1990,
(ISS, Warrendale, PA, 1990) p 197.
41. M. Wolf, AIME Elect. Furn. Proc., 40, 335, (1982).
42. M. Wolf, Proc. 2nd Europ. Conf. Continuous Casting, Dusseldorf, 1994 (VDEh, Dusseldorf,
1994) vol 1, p 78.
43. S. Ogibayashi, K Yamaguchi, T Mukat, T Takahashi, Y Mimura, K Koyama. Y Nagano,
T Nagano. Nippon Steel Technical Report, 34, 1, (1987).
44. Y. Nakamori, Y Fujikake, K Tokiwa, T Kataoka, S Tsuneoka, H Misumi, Proc. 10th Conf.
IMEKO TC3 on Measurement and Mass held Kobe, Japan Sept. (1984) and Tetsu-to
Hagane, 70(8),1282, (1984).
45. S. Sridhar, K.C. Mills, V. Ludlow, S.T. Mallaband, Proc. 3rd Europ. Conf. Continuous
Casting, Madrid, 1998, (UNESID, Madrid,1998) p. 807.
46. F Neumann, J Neal, MA Pedroza, AH Castillejos, FA Acosta, Proc. 79th Steelmaking Conf.
1996. (ISS, Warrendale, PA,1996) p. 249.
47. K.C. Mills, S. Sridhar, A.S. Normanton, S.T. Mallaband, Proc. Brimacombe Conf.,
Vancouver, BC, 2000,p 781.
48. K.C.Mills, A.B. Fox, ISIJ Intl.,43, 1479, (2003).
49. H. Maeda, T. Hirose, CAMP-ISIJ, 6, 280, (1993).
50. K Koyama, K Nagano, Y Nagano, T Nakano, Nippon Steel Technical. Report, 34, 41,
(1987).
51. OD Kwon, J Choi, IR Lee, JW Kim, KH Moon, YK Shin, Proc.74th Steelmaking
Conf.,1991, (ISS, Warrendale, PA,1991) p. 561.
52. K. Nakajima, S Hiraki, T Kanazawa, T Murakami, CAMP-ISIJ, 5, 1221, (1992).
53. K. Tsutsumi, H Murakami, S Nishioka, M Tada, M Nakada, M Komatsu, Tetsu- to- Hagane,
84, 617, (1998).
54. P E Ramirez-Lopez, P.D. Lee, K.C. Mills, ISIJ Intl., 50 (3), 425, (2010).
55. M Kawamoto, T Mizukami, M Hanao, H Kikikuchi, T Watanabe, Ironmaking and
Steelmaking, 29,199, (2002).
References 57

56. K Watanabe, K Tsutsumi, M Suzuki, H Fujita, S Hatori, T Omoto, ISIJ Intl., 54, 865, (2014).
57. T. Kitagawa, M. Ishiguro, Proc. 4th Japan-Germany Seminar, (ISIJ, Tokyo, 1980) p. 249.
58. T. Kajitani, K. Okazawa W. Yamada, H. Yamamura, ISIJ Intl., 46, 250 and 1432 (2006).
59. IR Lee, JW Kim, J Choi, D Kwon, YK Shin, Proc. Conf. on Continuous casting in
developing countries, Beijing, 1993, (SEAISI, Singapore, 1993) p. 814.
60. T. Mizukami, K Kawakami, T Kitagawa, M Suzuki, S Uchida, Y Komasu, Trans. ISIJ, 26,
B164, (1986).
61. M Suzuki, H Mizukami, T Kitagawa, K Kawakami, S Uchida, Y Komatsu, ISIJ Intl., 31,
254, (1991).
62. M. Wolf, “Effects of mould oscillation” presented Discussion Group on Continuous casting
of mould fluxes, Inst. of Metals, London (1984).
63. M. Wolf, Proc. Conf. Continuous casting of steel in developing countries, Beijing, China
(1994) p. 69.
64. K. Tsutsumi, T. Nagasaka, M. Hino, ISIJ Intl., 39, 1150, (1999).
65. T. Emi, H Nakato, K Suzuki, Y Iida, Proc. NOH- BOS Conf. (1978) p. 350.
66. H. Nakato, I. Muchi, Tetsu-to- Hagane, 66, 33, (1980).
67. H. Nakato T Sakuraya, T Nozaki, T Emi, H Nikoshawa, Mould fluxes for continuous casting
and bottom pour teeming (ISS, Warrendale, PA, 1987) p. 23.
68. K. Noguchi, K. Sawamura, Proc. 4th Intl. Conf. Cont. Casting, Brussels (1988)
(CRM/VDEh) p. 65.
69. Y. Kobayashi, S. Maruhashi,” Effects of operational on oscillation mark of continuously
cast, stainless steel slabs” Proc. 4th Japan-CSSR Seminar, Ostrava. (1983) p. 249.
70. S. Shimizu, Y. Imada et al., Proc. 6th Intl. Iron and Steel Congress (1990) p. 487.
71. M. Suzuki, S Miyahara, T Kitagawa, S Uchida, K Okimoto, Tetsu-to Hagane, 78, 113,
(1992).
72. M Ikeda, K Asano, T Nakano, M Fuji, S Mizoguchi, H Mizumi, Trans. ISIJ, 21, B 511,
(1981).
73. T. Mallaband, Metallugica. UK, private communication cited in AB Fox thesis [1].
74. T. Mukongo, C Pistorius, A Garbers-Craig, Ironmaking Steelmaking, 31,135, (2004).
75. Q. Wang, Y. Lu, S. He, K.C. Mills, Z.S. Li, Ironmaking and Steelmaking, 38, 297, (2011).
76. H Lei, Y Zhao, DQ Geng, ISIJ Intl., 54, 1629, (2014).
77. T Kishi, H Takeuchi, M Yamamiya, H Tsuboi, T Nakano, T Ando, Nippon Steel Tech.
Report, 34, 11, (1987).
78. M Hanao, Y Tsukaguchi, M Kawamoto, Proc. 4th Intl. Congress Science and Technol.,
2008, Gifu, Japan (ISI J, Tokyo, 2008), p. 694.
79. R. Koldwein, Unpublished Corus Internal Rept (2007) cited in KC Mills, J Kromhout,
A Hamoen, R Boom: Proc. Admet Conf, Dnipropetrovsk, 2007(Natl. Metall. Acad. Ukr.,
Dnipropetrovsk, 2007) vol 2 p 174.
80. M Washio, M Sugizawa, S Moriwaki, K Kariyaa, S Idogawa, S Takeuchi, Revue de
Metallurgie, CIT, 90 (April), 507, (1993).
81. D W van der Plas, C Platvoet, B Diesesme, JP Radot, JM Galpin, Proc. 2nd Europ. Conf.
Continuous casting, Dusseldorf, 1994, Metec Congress’94 (VDEh, Dusseldorf, 1994)
p. 109.
82. MY Ha, SG Lee, SH Seong, J. Mater. Processing Technol. 133. 322, (2003).
83. G Bocher, U Hoffman, P Muller, Proc. 2nd Europ. Conf. Continuous casting, Dusseldorf,
1994, Metec Congress’94 (VDEh, Dusseldorf, 1994)p. 102.
84. J Kromhout, RS Schimmel, Proc. 8th Europ. Conf. Continuous casting, Graz, Austria, 2014
(Austrian Met Mater. Soc., Vienna, 2014).
85. M Tani, T Toh, K Umetsu, K Tanaka, M Zeze, K Tsunenari, K Hayashi, S Fukunaga,
Nippon Steel Technical Report, 104, 62, (2013).
86. JK Park, JW Cho, KH Moon, SH Lee, KH Kim, HS Jeong, Proc. 7th Intl. Conf. Clean Steel,
Balatonfured, Hungary, 2007, (Hung. Min. Metall. Soc, Budapest, 2007) p. 264.
87. M. Wolf, Proc. 2nd Europ. Conf. Continuous Casting, METEC Congress’94 held
Dusseldorf, 1994 VDEh, Dusseldorf, 1994), vol 1, p 78.
58 2 Slag Infiltration, Lubrication and Frictional Forces

88. H Uchiyama, Proc. AISI Technical Committee on Strand casting, 1995 p.

89. J Sardemann, H Screwe, Stahl u Eisen, 111, (11), 39, (1991).
90. H Yamamura, T Kajitani, J Nakashima, M Yamasaki, S Mineta, Nippon Steel Technical
Report, 104, 54, (2013).
91. K Tsutsumi, J Ohtake, M Hino, ISIJ Intl., 40, 601, (2000).
92. H Steinruch, C Rudischer, W Schneider, Non-linear Analysis, Theory, Methods and
Applications, 30 (8), 4915, (1997) see also BHM 141 (1996)(9) 399.
93. H Steinruch, C Rudischer, W Schneider Proc. Conf. Modelling of Casting Welding and
Advanced Solidification processes VIII (MCWASP)(Minerals, Metals and Materials Soc.
1998).
94. PP Sahoo, S Basu, ISIJ Intl., 46, 219, (2006).
95. M Wolf, Trans ISIJ, 22, B204, (1982).
96. TJ Billany AS Normanton, KC Mills, P Grieveson, Ironmaking and Steelmaking 18, 403,
(1991).
97. K. Sorimachi, Proc. 5th Intl. Conf. Molten slags, fluxes and salts, Sydney,1997, (ISS,
Warrendale, PA, 1997) p. 781.
98. PV Riboud, Y Roux, Fundamental study of the behaviour of casting powders. Report EUR
9560,1985 (Eur. Comm. Sci and Tech. Publ., Luxembourg, 1985).
99. R.D.’Haeyer, Influence of chemical composition of continuous casting powders Report EUR
10326 EN (1987) (Eur. Comm. Sci and Tech. Publ., Luxembourg, 1987).
100. Y Nakamori et al, Nippon Steel Tech. Report, 34, 53, (1987).
101. B Mairy, D Ramelot, M Dutrieux, Proc, Technol. Conf., Measurement and Control
Instrumentation in the Iron and steel Industry, Detroit, 1985 (ISS, Warrendale, 1985) p. 101.
102. G. Saucedo et al, Proc.74th Steelmaking Conf. (1991) (ISS, Warendale, PA, 1991) p 79.
103. D. Bowen: Proc. Seminar on Mould powders for continuous casting, held British Steel
Teesside Laboratories, Sept (1989) Paper 8.
104. L. Hering, HP Heller, HW Fenske., Stahl u Eisen, 17, 61, (1992).
105. S. Ogibayashi et al, Proc.78th Steelmaking Conf., Nashville, TN, 1995, (ISS, Warrendale,
PA,1995) p. 451.
106. H Mizukami, M Komatsu, T Kitagawa, K Kawakami, Trans ISIJ, 24, B 181, (1984).
107. H Mizukami, K Kawakami, S Miyahara, M Suzuki, T Kitagawa, O Terada, Trans. ISIJ, 25,
B 300, (1985).
108. H Mizukami, A Ozeki, A Kurabayashi, N Hsebe, S Uchida, T Kitagawa, Trans ISIJ, 25,
B 301, (1985).
109. T Sohlgren, private communication, Sweden, 2015.
Chapter 3
Heat Transfer in the Mould and Shell
Solidification

Abstract The condition of the shell is paramount in continuous casting. The heat
transfer from the shell is important because it determines how thick the shell is and,
consequently, how strong the shell is. Four aspects of the heat transfer are con-
sidered here, namely, (i) horizontal heat transfer, (ii) shell solidification, (iii) verti-
cal heat transfer and (iv) the variability in heat transfer. The horizontal heat transfer
occurs across the slag film separating the shell from the mould. Heat is transferred
by two mechanisms, lattice conduction and radiation conduction. The latter is
usually controlled by (i) manipulation of the amounts of glassy (fgl) and crystalline
phases (fcrys) in the slag film and (ii) to a lesser extent, by incorporating transition
metal oxides into the mould powder to absorb the IR radiation. The key properties
of the slag film are (i) the thickness of solid slag film (which is dependent on the
solidification or break temperature) and (ii) the fraction of crystalline phase (fcrys)
formed in the film (which (a) reflects the IR radiation and (b) creates an interfacial
resistance, (RCu/sl) during crystallisation). In practice, all of these factors increase
with increasing basicity (C/S) of the mould slag. Other factors like the effect of
casting conditions (e.g. casting speed, metal flow pattern) on the heat flux (qhor) are
discussed. The factors affecting shell thickness (dshell) are discussed below. There
are two regimes controlling shell growth (i) a period of slower growth for the initial
period t
0.05 s and (ii) a subsequent period of linear growth of dshell which
exhibits a linear relation with t0.5. Vertical heat transfer is controlled through (i) the
depth of the bed, (ii) use of exothermic mould powders, (iii) to a less amount, by the
particle size of the powder and (iv) by use of electromagnetic braking (EMBr) in
the mould which reduces the efficiency of vertical heat transfer. Gaseous convection
is shown to be a major contributor to the vertical heat flux (qvert) and the perme-
ability of the powder bed is a key factor affecting qvert. Local variations in heat flux
are an issue because an uneven shell can lead to longitudinal cracking. The various
causes of local variations in shell thickness are discussed.

Symbols, Abbreviations and units

a Thermal diffusivity (m2 s−1)
dbed Depth of bed (m)

© Springer International Publishing AG 2017 59

K.C. Mills and C.-Å. Däcker, The Casting Powders Book,
DOI 10.1007/978-3-319-53616-3_3
60 3 Heat Transfer in the Mould and Shell Solidification

dl Thickness of liquid slag film (m)

dmenis Distance from meniscus (m)
ds Thickness of solid slag film (m)
dshell Thickness of steel shell (mm)
k Thermal conductivity (W m−1 K−1)
n Refractive index
q Heat flux density (W m−2 = J s−1 m−2)
qhor Horizontal heat flux (W m−2)
qvert Vertical heat flux (W m−2)
qR Heat flux by radiation conduction
qtotal Integral (total) heat flux (J)
R Thermal resistance (m2 K W−1)
RCu/sl Interfacial thermal resistance
T Temperature (oC)
Tbr Break temperature (oC)
t Time (s)
tmould Dwell (residence) time (s)
Vc Casting speed (m min−1)
a* Absorption coefficient (m−1)
η Viscosity (d Pas)
HC High-carbon steel
Hi Al High-Al steel
IR Infrared
LC Low-carbon steel
MC Medium-carbon steel
SEN Submerged entry nozzle

Subscripts, superscripts
Cu/sl Interface between Cu and slag
film Slag film
l Liquid
s Solid

3.1 Introduction

The various aspects of heat transfer and solidification are discussed in this chapter.
The control of the heat extraction from the newly formed shell is very important in
the continuous casting process. Its importance stems from the effect it has on the
shell. It is essential when casting peritectic, steel grades to reduce the horizontal
heat flux to a satisfactory level to create a thin, uniform shell, in order to avoid
3.1 Introduction 61

longitudinal cracking. However, to avoid sticker breakouts in HC steels, where the

shell is weak, it is necessary to increase the horizontal heat flux to create a thicker,
stronger shell. Thus, control of the heat extraction from the shell is essential in
obtaining good surface quality and in minimising process problems. It will be
shown below that this is principally achieved by controlling the nature of the slag
film formed between the steel shell and the mould.
The continual flow of molten steel into the water-cooled mould provides the
energy required for the process. For solidification to occur, heat must be lost. It can
be seen from the meniscus region of Fig. 3.1 that the solidified shell near the
meniscus has a curved shape, thus heat losses follow a radial path [1, 2]. However,
for simplicity and convenience, heat losses are usually divided into two contribu-
tions, namely, (i) horizontal heat flux (through the solid and liquid slag films to the
mould) and (ii) the vertical heat flux (through the slag pool and powder bed to the
atmosphere). The horizontal and vertical heat fluxes are related to one another
because they have a common source of heat (i.e. the flow of molten steel).
The importance of the horizontal heat transfer is due to the fact that it determines
the thickness of the solidified shell formed (especially in the meniscus region). The
thickness of the shell (dshell) can be calculated approximately by Eq. 3.1 where d is
the distance below the meniscus and K is a solidification constant; thus dshell
decreases with increasing casting speed (Vc) [3]:

dshell ¼ Kðd=Vc Þ0:5 : ð3:1Þ

Fig. 3.1 Schematic diagram showing the radial nature of the heat losses in the meniscus region,
the horizontal and vertical heat fluxes and the “sausage-shaped shell” resulting from variations in
heat flux (and shell solidification) during the oscillation cycle (permission granted, ISS/AIST, [2])
62 3 Heat Transfer in the Mould and Shell Solidification

3.1.1 Heat Flux

The magnitude of heat transfer is usually expressed in terms of the heat flux (q). For
steady-state conditions (where the temperature gradient (dT/dx) is constant), the
heat flux between two points (in the x direction) is usually given by Fourier’s first
law (Eq. 3.2) where k is the thermal conductivity and A is the area:

q ¼ kAðdT=dxÞ: ð3:2Þ

Heat flow is often considered in terms of electrical analogues and the heat flux
can be considered to be the equivalent of current (and (dT/dx) as potential).
For transient conditions [where the temperature gradient is changing with time


(t)], the rate of heat extracted (or supplied) d2 T=dx2 is usually given by Fourier’s
second law shown in Eqs. 3.3–3.5 for 1-, 2- and 3-dimensions, respectively, where
a is the thermal diffusivity.

1-dim:

ðd2 T=dx2 Þ ¼ ð1=aÞðdT=dtÞ ð3:3Þ

2-dim:

ðd2 T=dx2 Þ + ðd2 T=dy2 Þ ¼ ð1=aÞðdT=dtÞ ð3:4Þ

3-dim:

ðd2 T=dx2 Þ þ ðd2 T=dy2 Þ þ ðd2 T=dz2 Þ ¼ ð1=aÞðdT=dtÞ: ð3:5Þ

The total amount of heat supplied is usually calculated from Eq. 3.6 where
m; Cp ; V; q and (T2 − T1) are mass, heat capacity, volume, density and temperature
rise, respectively:

qtotal ¼ m  Cp ðT2  T1 Þ ¼ qV Cp ðT2  T1 Þ: ð3:6Þ

3.2 Horizontal Heat Flux

The horizontal heat flux is generally regarded as being more important than the
vertical heat flux since it determines the thickness of the steel shell formed in the
meniscus region. There are two issues here (which affect the surface quality of the
steel product) which are linked but are treated separately here, namely,
(i) The magnitude of the horizontal heat flux (described in Sects. 3.2 and 3.3) and
(ii) The variability of the horizontal heat flux (described in Sect. 3.4).
3.2 Horizontal Heat Flux 63

3.2.1 Heat Transfer Mechanisms Involved in Horizontal

Heat Transfer

The horizontal heat transfer refers to the transport of heat from the shell to the
water-cooled mould. Initially, oil was used to prevent the shell from sticking; the oil
evaporated to form a gaseous film which formed a thermal barrier to heat flow [4].
The replacement of oil by mould powders was found to improve the control of heat
extraction from the shell. This control is provided by the slag film formed between
the shell and the mould. The slag film consists of a mixture of crystalline and glassy
phases and the control of the horizontal heat flux is obtained from manipulation of
(i) the ratio of glass and crystalline phases and (ii) the thickness of the slag film.
The mechanism of heat transfer between the shell and the mould is complex
since it involves at least three heat transfer mechanisms, namely (i) lattice con-
duction through the slag film, (ii) radiation conduction and (iii) convection in the
liquid layer of the slag film.
The solid slag film formed in the first moments of casting is predominantly
glassy because of the high cooling rates involved when it solidifies in the
mould/strand gap. Heat transfer can occur by two mechanisms in solid glasses, i.e.
(i) normal, phonon or lattice conduction and (ii) radiation conduction. The latter
mechanism was first detected when measured thermal conductivities of glasses
were observed to increase with increasing sample thickness until a critical point was
reached [5] (Fig. 3.2a). The sample is denoted as “optically thick” at this critical
point, beyond which the measured thermal conductivity remained constant
(Fig. 3.2a) [5].
Consider a glass sample consisting of a number of layers. Radiated heat falls on
the surface layer of the sample and this causes the temperature of layer 1 to
increase. This layer now is at a higher temperature than layer-2 and so layer-1 emits
radiation to layer-2; the radiant energy is absorbed and thus the temperature of
layer-2 rises (Fig. 3.2b). The temperature of layer-2 is now higher than that for

Fig. 3.2 Schematic drawings (a)

showing a effective thermal
conductivity (keff) of a glassy
slag as a function of
temperature and sample
thickness (d which increases
from - - - to ) [5] and
b schematic drawing showing
the mechanism of radiation
conduction

(b)
64 3 Heat Transfer in the Mould and Shell Solidification

layer-3 so layer-2 radiates to layer-3 and so on. Thus, radiation conduction is a

process of absorption and re-emission (Fig. 3.2b).
A sample is usually considered to be optically thick when a*d > 3 where a* is
the absorption coefficient and d is the thickness of the sample. For optically thick
samples (a*d > 3) the radiation conductivity (kR) can be calculated from Eq. 3.7
where r = Stefan–Boltzmann constant, and n = refractive index which usually has
a value of ca. 1.58 for mould slags [6, 7].

kR ¼ 16 r n2 T 3 =3a ð3:7Þ

Since kR is a function of T3, kR tends to increase sharply with increasing tem-

perature (Fig. 3.2a). However, the values of kR are reduced by
(i) The presence of transition metal oxides (e.g. FeO, NiO, CrOx) which absorb
radiation [6, 8] and
(ii) Crystalline phases (e.g. cuspidine) which scatter or reflect the IR radiation
(Fig. 3.3) at the interface of the slag film with the liquid [8].
Usually, the heat transfer through the slag film is controlled by the reflection
mechanism but the transition metal oxides have been used to control the heat flux in
the casting of round billets [10]. However, additions of FeO, MnO, etc. tend to
increase the amount of glassy phase and thereby reduce the reflection of IR radi-
ation [11–13]. The scattering and reflection have been reported to increase with
increasing crystal size [12].
If the sample is glassy, it is possible to calculate the magnitude of kR from
Eq. 3.7, providing the value of the absorption coefficient of the slag is known and
the sample is optically thick. However, it is much more difficult to calculate the
effect of crystalline phases on kR. Some workers have replaced the absorption
coefficient (a*) in Eq. 3.7 by the extinction coefficient (E = a* + s), where s is the
scattering coefficient. However, the process of radiation scattering is different to that
of radiation absorption. A parameter known as the albedo (or reflection coefficient)
is needed for the calculation and there are no values for the albedo of casting slags
[7]. It has been shown that for slag films containing crystalline phases that radiation

Fig. 3.3 Schematic diagram

showing the scattering of IR
radiation by crystallites
3.2 Horizontal Heat Flux 65

is increasingly reflected (i.e. kR decreases) with (i) increasing crystalline fraction,

(ii) increasing grain size of the crystals, (iii) increasing film thickness and (iv) de-
creasing transition metal oxide (e.g. FeOx) content (this is a consequence of their
tendency to increase fglass) [11, 12, 14, 16]. Susa et al. [12] calculated that the heat
flux arising from radiation conduction, qR = 0.29 M W m−2 for a 1-mm-thick slag
film with mean crystal grain size of 2.5 lm. This compares with a heat flux via
conduction, qc = of 1.45 M W m−2, i.e. 20% of the heat was transferred by radi-
ation conduction. Similar values for the relative contribution of qR (=qR/qtotal) have
been obtained by other workers (see [14, 16, 17]) for crystalline slag films, e.g.
<10% (5–19%) [15] with (qR/qtot) values decreasing with increasing fraction of
crystalline phase (fcrys) in the slag film. Hanao et al. [17] cite results for (qR) which
indicate they constitute ca. 20% of qtot for crystalline slag films but ca. 60% of that
for purely glassy films. Reported values of (qR/qtot) for the liquid film are much
larger, e.g. 37–50% [14] but the liquid layer is probably optically thin.
The thermal conductivity (k) of slags in the crystalline state is about twice that of
the glassy phase (i.e. kcrys = 2 kgl) [18–20]. Despite this, the most common method
for reducing the horizontal heat flux is to increase the crystallinity of the slag film
(to reduce qR); this underlines the importance of crystallinity (in the slag film) in
reducing the magnitude of the radiation conductivity.

3.2.2 Interfacial Thermal Resistance (RCu/Sl)

An interfacial thermal resistance (RCu/sl) is created by the crystallisation of the

glassy solid in the slag film (formed by the infiltration of molten slag between the
shell and the copper mould). Some workers have suggested that this interfacial
resistance arises from the contraction of the shell on solidification. This is not
correct since any gap formed by shrinkage of the shell will automatically be filled
with molten slag, except in exceptional circumstances (e.g. overcooled corners and
where there is insufficient, liquid slag flow, e.g. in the lower mould) [21, 22].
Crystalline phases have higher densities than those of the glass and thus crys-
tallisation results in shrinkage. This shrinkage causes (i) the creation of pores near
the crystallites (Figs. 3.4b and 3.5) and (ii) some wrinkling of the slag surface (often
referred to as “surface roughness”) resulting in a “gas gap” at the interface between
the slag and the copper mould (Fig. 3.4b). The magnitude of the interfacial resis-
tance (RCu/sl) is related to the width of this gas gap (“surface roughness”) as can be
seen from the trend shown in Fig. 3.6. Since the interfacial resistance RCu/sl is
related to the shrinkage, RCu/sl increases with increasing proportion of crystalline
phase (fcrys) and increasing slag film thickness (ds) [16, 23–26]. Thus, ds increases
with increasing break temperature (Tbr or Tsol). The interfacial resistance (RCu/sl)
also tends to increase with increasing Tbr (or Tsol) [3, 23, 27–29].
Tsutsumi et al. [27] used an infrared image furnace to obtain cooling rates
similar to those encountered in the mould. They also used a confocal microscope to
66 3 Heat Transfer in the Mould and Shell Solidification

(a) Tbr2 (b) C

Cu Solid Liquid Shell

Cu Solid Liquid Shell slag slag
slag
Air gap

Fig. 3.4 Schematic diagrams showing a glassy slag film b partially crystallised slag film showing
pores and the formation of an “air gap” or “surface roughness”

Fig. 3.5 Photograph of pores

(shown by the arrow) formed
around crystalline phase in a
slag film taken from the
mould and the surface
roughness on the mould (left)
interface (courtesy of
Carboox)

Fig. 3.6 The interfacial

resistance (RCu/sl) as a
function of the thickness of
the air gap;
● = measurements; −−D,
□ = trendlines [27];
(permission granted, ISIJ,
[27], re-drawn)
3.2 Horizontal Heat Flux 67

Fig. 3.7 Schematic diagrams showing a surface roughness as a function of the normalised
cooling rate (= actual cooling rate/critical cooling rate); symbols: open = glass; closed = crys-
talline; ■ = flux for LC steels (C/S  1) ▲, ● = fluxes for MC steels (C/S  1.45) and b gas gap
width (dair) and the thermal resistances of the mould slag (Rf) and interface (Rint = RCu/sl) as a
function of slag film thickness [27]; Note: Slag films taken from the mould are usually about 2 mm
thick (permission granted, ISIJ, [27])

measure the surface roughness of the quenched sample. It was reported [27] that the
surface roughness increased as:
(i) The fraction of crystalline phase increased (Fig. 3.7),
(ii) The cooling rate decreased (Fig. 3.7) and;
(iii) The basicity (C/S) of the mould flux increased (Fig. 3.7).
Tsutsumi et al. [27] also calculated the width of gas gap (dair) from reported
values of RCu/sl and reported the surprising result that dair decreased as the thickness
of solid slag film increased ðdair # as ds "Þ which is contradictory to the findings of
Cho et al. [23] and others [30–33]. However, the reported values of dair were found
to be similar to those recorded for the surface roughness (ca. 30 lm).
There has been some discussion on the relative importance of the two mecha-
nisms responsible for reducing heat flux in the slag film, i.e. interfacial resistance
(RCu/sl) and the reflection by crystals in the slag film [3, 17, 28, 34–36]. However,
Hanao and Kawamoto [3, 17] carried out elegant, pilot plant measurements of the
heat flux and extracted the slag film from the mould. A value for the interfacial
resistance was obtained by analysis of these heat flux measurements. It was found
that the interfacial resistance value was smaller than values reported by other
workers using simulation experiments (Table 3.1). It was concluded that the fer-
rostatic pressure on the shell effectively reduced the thickness of the “gas gap” and
thereby reduced the RCu/sl value. Pilot plant trials reflect continuous casting con-
ditions more closely than simulation experiments, so preference is given to the pilot
plant data.
 68

Table 3.1 Values of interfacial resistance, RCu/sl (10−3 m2 K W−1) reported by various investigators [3, 17, 27, 31, 33, 35, 37–40]; values in parenthesis = ds
(mm). *indicates liquid phase value
Reference Watanabe Inoue Yamauchi Kawamoto Nakato Cho Stone Yamamura Hanao
[31] [33] [37] [38] [39] [23, 28] [40] [35] [3, 17]
RCu/sl 0.4–0.6 0.8 0.7 0.7 0.5 0.6–1 1.5 0.2*–0.7 0.4
ds (0.2) (0.25) (0.3) (0.1) (0.1) (0.7) (1.0) (1.0) (1.0)
3 Heat Transfer in the Mould and Shell Solidification
3.2 Horizontal Heat Flux 69

Fig. 3.8 Schematic

representation of the thermal
resistance of the various
layers between the shell and
the mould (permission
granted, ISIJ, [14])

Susa et al. [11–13] considered that much of the reduction in horizontal heat flux
came from the reflection and scattering of the IR radiation back towards the shell.
They reported that the reflection increased as the (i) fraction of crystalline phase
increased and (ii) grain size of the crystallites increased (i.e. qhor # as Dcrys ").
It is possible to represent the thermal resistance (Rtotal) between shell and mould
as a series of resistances (i.e. using Ohm’s Law) where R is the thermal resistance
and RCu/sl is the interfacial resistance of the “gas gap”, d is the thickness of the slag
film layer, and subscripts Cu, sl, gl and crys represent the mould, slag, glass and
crystalline layers, respectively (Fig. 3.8) [14, 41]. The terms (d/k)shell and the
contact resistance at the shell (Rshell) are negligible and (d/k)liq is small because d is
small (ca. 0.1 mm) and the effective thermal conductivity is high:

Rtotal ¼ ðd=kÞshell þ ðd=kÞliq þ ðd=kÞgl þ ðd=kÞcryst þ RCu=sl þ Rshell : ð3:8Þ

Radiation conduction can be taken into account using a parallel circuit in the
Ohm’s law circuit diagram [14, 15, 41].
In practice, the glass and crystalline phases are partially intermixed.
Consequently, the terms ðd=kÞgl and ðd=kÞcryst can be replaced by ðð1  fcrys Þ 


dsolid =kgl Þ and fcrys  dsolid =kcrys , respectively, where fcrys = fraction of crystalline
phase present in the slag film. Thus Eq. 3.8 can be rewritten




Rtotal ¼ ðd=kÞliq þ 1  fcrys  dsolid =kgl þ fcrys  dsolid =kcrys þ RCu=sl : ð3:9Þ

In an analysis of heat transfer measurements during
castings on pilot plant,

 Rtotal
was considered to be composed of two terms Rtotal ¼ RCu=sl þ Rfilm , where
ðRfilm ¼ ðds =kÞ þ RR Þ and RR is the thermal resistance associated with the scattering
and reflection of IR radiation [17]. In this study [17, 42] it was found that
70 3 Heat Transfer in the Mould and Shell Solidification

Fig. 3.9 Thermal resistance Rtotal as a function of a cooling rate, b Rfilm, RCu/sl and keff as
functions of basicity (C/S), c Rfilm and RCu/sl as functions of fcrys (represented by the intensity of the
cuspidine diffraction peak), d Rfilm and RCu/sl as functions of casting speed [17, 42]. Note values
cited in (c, d) refer to thickness of air gap (lm) (permission granted, ISIJ, [17, 42])

(i) Rtotal increased as the cooling rate increased (Fig. 3.9a);

(ii) Rfilm and RCu/sl both increased with increasing basicity (C/S) (Fig. 3.9b);
Rfilm was found to increase only slightly with increasing fcrys whereas RCu/sl
increased with increasing fcrys (Fig. 3.9c);
(iii) Rfilm was found to be essentially independent of casting speed whereas RCu/sl
decreased with increasing casting speed (Fig. 3.9d); and
(iv) Rfilm was greater in magnitude than RCu/sl (Fig. 3.9d).
In summary, the interfacial resistance can be controlled by (i) the amount of
crystalline phase formed (fcrys) and (ii) the slag film thickness (ds which is
dependent upon Tbr). In conventional (F-containing) mould slags these parameters
are determined largely by the basicity (C/S) and the F content of the mould slag
(Fig. 3.8b, c). Thus, increasing basicity (and hence fcrys) results in both increased
scattering of IR radiation and increasing RCu/sl.
3.2 Horizontal Heat Flux 71

3.2.3 Factors Affecting the Horizontal Heat Flux

There are a significant number of factors affecting the horizontal heat flux. These are
described below in order of grouping rather than in order of importance.

3.2.3.1 Mould Position

The magnitude of the heat flux varies with position, in both the horizontal and
vertical directions. The cooling is greater in the corners than in the centre of the
mould; a typical horizontal profile of heat flux obtained during high-speed casting is
shown in Fig. 3.10a [30]. The heat flux density is not uniform around the mould;
qhor is greatest in the regions near the narrow faces and decreases gradually to the
region around the centre of the wide face. The increased shrinkage in the corner
regions results in an increase in slag film thickness (ds) and, subsequently, causes a
decrease in qhor at the narrow faces. The mould flux ratio [43], defined in Eq. 3.10,
is used to measure the uniformity of the heat flux in the mould:



Heat Flux Ratio ¼ qnarrow
hor =average qwide
hor : ð3:10Þ

The taper must compensate for the shrinkage, which, in turn, is related to the
heat flux (see Sect. 3.2.3.5 (iii) below). Thus, the required taper changes according
to the mould slag used. An increase in taper from 1.1 to 1.4% was observed to
increase the heat flux ratio from 0.6 to 0.8 [44].
In the vertical direction, the heat flux is at a maximum at the point where the
metal flow impacts with the shell; this occurs around 45 mm below the meniscus
[42]. Heat fluxes and mould temperatures decrease as the distance from the
meniscus (dmenis) increases; typical results are shown in Fig. 3.10b [42].

Fig. 3.10 Local heat flux in the mould as a in the horizontal direction (wide face) [30]
● = MCsteel with cracks; D = LCAK steel, and b in the vertical direction, [42] for different
casting speeds [42] (permission granted, JSPS and M Kawamoto [30] (b) ISIJ, [42])
72 3 Heat Transfer in the Mould and Shell Solidification

3.2.3.2 Shell Conditions

Oscillation marks are formed in the early stages of shell solidification [2]. The
oscillation marks are filled, or partially filled, with slag and this reduces the heat
flux in the deeper parts of the mark. Similar effects are to be expected when
transverse depressions are formed. Thomas et al. [45] studied the effects of both
oscillation marks and transverse depressions on the horizontal heat flux.

3.2.3.3 Casting Conditions

(i) Casting speed (Vc)

The effect of casting speed on the horizontal heat flux has been studied by
several investigators [30, 46–51]. The effect of casting speed (Vc) can be quite
confusing since an increase in Vc results in
(i) An increase in the rate of heat flow (qhor, with units of W m−2 = J m−2 s−1,
sometimes referred to as heat flux density) as can be seen from Fig. 3.11a [46]
and
(ii) A decrease in total heat flow (qtotal ¼ A  qhor  Lmould =Vc with units of J)
shown in Fig. 3.11b [47]; this is a consequence of the shorter residence time of
the steel in the mould and results in a decrease in qtotal [43].
The effect of increased
 casting speed is best  viewed through its effect on the
0:5
thickness of the shell dshell ¼ K ðdmenis =Vc Þ ; where K is a constant and dmenis is
the distance to the meniscus. Thus, an increase in casting speed results in a thinner
shell ðdshell # as Vc "Þ; i.e. since ðdshell # as qtotal #Þ it follows that ðqtotal # as Vc "Þ
and thus the total heat flux is the key factor affecting shell thickness. Since shell
thickness is affected by residence time in the mould (tmould) it follows that the length

(a) 14 (b)
70
12
q,108. kJm-2

qtot108. kJm-3

10 60
8
50
6

4 40
0 2 4 6 0.6 1 1.4 1.8
Vc, m min-1 Vc, m min-1

Fig. 3.11 The effect of casting speed (Vc) on a the heat flux density (q) (courtesy of ISIJ) and
b total heat flux, qtot total heat flux, qtot for three mould powders for casting MC steels (▲, ▲, ▲)
[42, 46] and LC steels (●, ●, ●); critical heat flux for MC = dashed line; LC = solid [47]
(permission granted, ISS/AIST, [47])
3.2 Horizontal Heat Flux 73

Fig. 3.12 Local heat flux at a

point 45 mm from the
meniscus as a function of
casting speed [42]
(permission granted, ISIJ,
[42])

of the mould should also be taken into account (see Sect. 3.2.3.5 (iv) below); longer
moulds are required for high-speed casting [52].
The heat flux density at a point 45 mm below the meniscus (i.e. maximum value
of qhor where the metal flow impacts the shell) can be calculated by Eq. 3.11 (see
Fig. 3.12) for casting speeds between 1 and 5 m min−1 [42]:


q45 mm M W m2 ¼ 0:649 þ 0:828 Vc  8:38x102 Vc2 : ð3:11Þ

(ii) Superheat
Superheat tends to increase the heat flux since it increases the thermal gradient
between molten steel and mould [50, 51]. However, superheat will also tend to
delay solidification to a lower position in the mould.
A statistical analysis of plant data [48] established that increasing superheat
caused a significant increase in horizontal heat flux for both the narrow and wide
faces of the mould.
(iii) Oscillation characteristics (stroke length, frequency)
A statistical analysis of plant heat flux measurements showed that the horizontal
heat flux increased with increasing stroke length for both the narrow and wide faces
of the mould [48]. A possible explanation is given below in Sect. 3.4.1. The pre-
diction of a mathematical model [1, 2] indicates that as the mould starts descending
a downward, “Arctic” slag flow is produced which causes increases in both heat
flux and shell solidification. Thus an increased stroke length will cause both an
increased period where this flow is in operation and bring the slag flow closer to the
shell tip; both of these factors will increase the horizontal heat flux.
Preliminary measurements obtained with a mould simulator [53] indicated that
(i) Increase in frequency (2.14 to 2.43 Hz) resulted in a increase in mould tem-
perature ðf "! tn #! Tmould "Þ and
(ii) Increase in stroke (3.4 to 6.8 mm) resulted in a decrease in mould temperature
ðs "! Tmould #Þ.
74 3 Heat Transfer in the Mould and Shell Solidification

These results are consistent with mould temperature increasing as the negative
strip time (tn) decreases ðTmould " as tn #Þ. Non-sinusoidal oscillation was found to
give a 10% decrease in qhor [54].

3.2.3.4 Slag Characteristics

(i) Thickness of solid slag film (ds)

It can be seen from Eq. 3.9 that the thermal resistance of the slag film increases
as the thickness of the solid layer (ds) increases. The principal factor affecting ds is
the break (or solidification) temperature (Tbr); it can be seen from Figs. 3.13 and
3.14b that an increase in Tbr results in an increase in ds. The effect of break
temperature on the heat flux density can be clearly seen in Fig. 3.14a ðq # as Tbr "Þ;
an increase in Tbr of 150 °C has been found to result in a 10–20% decrease in qhor
[17, 32, 55]. It has been reported that the slag film thickness (ds) remains constant
for all mould positions (dmenis) below dmenis = (35–45) mm (i.e. dmenis > 35–
45 mm) [34].
(ii) Interfacial Resistance
The interfacial resistance (RCu/sl) has been discussed above in Sect. 3.2.2. In
summary, the magnitude of RCu/sl increases with (i) increasing fcrys [14, 16],
(ii) increasing solid film thickness, ds [14, 16] and (iii) increasing gas gap thickness
(dgas) (Fig. 3.6). There is still some dispute about (i) the magnitude of RCu/sl values
obtained in simulation experiments and in pilot plant trials and (ii) the relative
magnitudes of RCu/sl and Rfilm. The pilot plant data are preferred here because they
simulate the conditions in the mould more closely than those in simulation
experiments; thus (i) a value of RCu/sl = 0.4  10−3 m2 K W−1 [3, 17] and

(a) (b)

Fig. 3.13 Schematic drawings showing the dependence of relative thicknesses of solid (grey) and
liquid (yellow) layers on break temperature (Tbr) shown as vertical dashed line. a With a low Tbr
and b high Tbr with larger solid layer thickness
3.2 Horizontal Heat Flux 75

Fig. 3.14 The effect of (a) 2.2

solidification (or break)
temperature on a heat flux 2
density and b fsolid = 1.8

q ,MW s-1
ds/(ds + dl) in slag
(permission granted, ISIJ) 1.6

1.4

1.2

1
1000 1100 1200 1300
Tbr or Tsol, oC
(b) 1

0.8
ds/(ds+dl)
0.6

0.4

0.2

0
1000 1100 1200 1300
Tbr or Tsol, oC

(ii) RCu/sl < Rfilm (where Rfilm = (ds/k) + RR) are preferred. The ratio of the two
resistances (Rfilm/RCu/sl) was found to increase with increasing casting speeds as a
result of the decrease in RCu/sl with increasing casting speed (Fig. 3.9d) [17].
Cho et al. [23] reported that the interfacial resistance was slightly lower when the
copper mould was coated with Ni–Fe coating.
(iii) Crystallisation and porosity of slag film
The principal way to control the horizontal heat flux is via the control of the
amount of crystallinity in the slag film. This is usually done by selecting the
appropriate chemical composition (or (%CaO)/(%SiO2) or (C/S) ratio) of the
casting powder. Cuspidine (3CaO2SiO2CaF2 or C3S2Fl) is the principal crystalline
phase formed. Other phases are formed; the phase formed depends upon the
composition. A phase diagram has been reported to identify these phases for
powders with a C/S ratio of ca. 1 [56]. The basicity of casting powders varies
between 0.6 and 1.3, for instance, (C/S) ratios of around 1.3; (0.7–0.9) and 1.0,
respectively, are used when casting, peritectic (MC), sticker-sensitive (HC) and
other steels (LC, ULC) [57]. Environmental concerns have led to a drive to reduce F
levels in casting powders; this has led to a search for a crystalline phase to replace
cuspidine for powders used to cast slabs. One possible candidate is perovskite,
CaTiO3 [58, 59] but it has been reported that TiO2 reacts in the mould to form Ti
(CN) which results in sticker breakouts [60].
76 3 Heat Transfer in the Mould and Shell Solidification

Crystallisation of the glassy phase of a slag film results in

(i) Reflection of the IR radiation by the crystals with a resultant decrease in qR;
(ii) The formation of porosity and an interfacial thermal resistance, RCu/sl,
(caused by the resultant shrinkage) which both cause a reduction in kR and
qR; and
(iii) Substantial increases in the lattice thermal conductivity (klat).
The overall effect of crystallisation is a significant decrease in the horizontal heat
flux.
Further studies [11–13] showed the following:
(i) Crystals tended to increase the reflectivity and decrease the transmissivity,
whereas FeO additions tended to increase the absorption of IR radiation and
decrease the reflectivity; FeO additions promote glass formation, which
enhances qR and thereby opposes the reduction in qR resulting from reflec-
tion but simultaneously reduce qR through increased IR absorption [11];
(ii) For FeO-free fluxes, the reflectivity of IR radiation at higher wavelengths
tended to increase with increasing cuspidine grain size (i.e. kR # as dgrain ")
but the effect was less marked at for slags containing FeO [12]; and
(iii) The IR radiation tends to be reflected at the liquid/ crystal interface [9].
Crystallisation is not the only cause of porosity. Casting powder with a high
moisture content (or water leaks) can lead to hydrogen pores in the slag film.
Despite the fact that kH2 [ kair , the increased porosity leads to a decrease in heat
flux (Fig. 3.15) and to sticker breakouts [61]. Blowholes containing CO (g) can be
formed from reaction of C and FeO in the slag.

3.2.3.5 Slag Viscosity

Hanao et al. [17] have reported a slight increase in slag film thickness as the mould
slag viscosity was increased (Fig. 3.15); this results in a decrease in qhor. Further
evidence for this proposition comes from a statistical analysis of plant data which
indicated that the heat flux decreased with increasing slag viscosity (q# as η") [48]
and from other sources [1, 50]. It was suggested that the thickness of the solid layer
increases with increasing viscosity which would result in a decreased heat flux.
Another possible explanation for this opposition is that increased viscosity would

Fig. 3.15 Mould heat 90

removal, q, as a function of
hydrogen in the slag film (a
80
q , kJ kg-1

measure of the porosity) for

melts which were degassed
(●) and where no degassing 70
was carried out (▲) [60];
Vc = 1 m min−1 (permission
granted, EDP Sciences, [61]) 60
0 2 4 6 8 10
H2, ppm
3.2 Horizontal Heat Flux 77

Fig. 3.16 Slag film thickness 1.5

as a function of slag viscosity
at 1300 °C for three mould
slags [17] (permission
granted, ISIJ, [17])

dfilm ,mm
1

0.5
0 0.5 1 1.5 2
η1300, dPas

oppose the cold downward slag flow in the pool when the mould is descending (see
Sect. 3.4.1) and would hence reduce the heat flux (Fig. 3.16).

3.2.3.6 Shell Condition

The solidification of the shell occurs during the period when the mould is
descending (i.e. during negative strip time) but there is little solidification when the
mould is ascending. Consequently, the variation in solidification results in a shell
with variable thickness. In MC peritectic grades, the shell is wrinkled and this too
leads to variability of the heat flux. Furthermore, oscillation marks will tend to
increase the variability of the heat flux [45].

3.2.3.7 Mould Conditions

(i) Position in the mould

The heat flux varies in both the horizontal and vertical directions (see
Sect. 3.2.3.1 and Fig. 3.10a). The heat flux is higher in the corners than in the
mid-face; this results in a thicker film in the narrow faces cf., the broad face and
lower value of q; this affects the heat flux ratio (defined in Eq. 3.10). In the vertical
direction the maximum mould temperature occurs at ca. 45 mm below the meniscus
and decreases with increasing distance from the meniscus (Fig. 3.10b).
(ii) Water flow rate
One practice used to obtain “mild cooling” (in order to reduce longitudinal
cracking) is to reduce the water cooling rates to reduce the horizontal heat flux [50,
62, 63]. However, Lu et al. [62] found that reduction of water cooling rates made
only a small reduction in shell thickness (Fig. 3.17).
78 3 Heat Transfer in the Mould and Shell Solidification

Fig. 3.17 The effect of water 18

cooling flow rate on the
thickness of the shell for two
different superheats (20
16
°C = blue and 60 °C = red)

dshell, mm
during casting of stainless
steel 304 (- - -) and CS 1026
(──) [62]; 14
Vc = 0.9 m min−1
(permission granted,
ISS/AIST [62])
12
2.2 2.6 3 3.4 3.8
Water cooling rate, 103 litre min-1

(iii) Mould Taper

When the steel solidifies, the shell tends to shrink away from the mould.
Consequently, the narrow faces are tapered, so that they can provide support to the
shell. If the taper is insufficient, a depression is usually formed 50–100 mm from
the edge of the slab which gives rise to cracks. When the taper is insufficient to
compensate for shrinkage, a “gas gap” is formed, providing there is insufficient
liquid slag to flow into the gap; this is a particular problem in both the corners
(where overcooled corners can lead to transverse cracks [64]) and in the lower
mould. It is important that the taper should match the shrinkage (see Fig. 11.48),
which, in turn, is related to the heat flux Tapers are dependent on the steel being
cast (e.g. Taper MC < Taper HC due to higher qhor [65]). Mathematical models have
been developed to calculate the appropriate taper for the steel and casting condi-
tions [64, 65].
A statistical analysis of plant measurements, relating cracking and taper, indi-
cated that increased taper produced a thicker and more uniform shell and, conse-
quently, resulted in fewer cracks [48].
(iv) Mould length
The mould length (Lmould) is an important factor when minimising the incidents
of sticker breakout. Mould lengths vary from 0.7 to 1.5 m [52]. The residence (or
dwell) time (tmould) determines the shell thickness at the mould exit and is calcu-
lated by (tmould = (Lmould − 0.10)/Vc) where (Lmould − 0.10) in m is the active
mould length. The heat flux density (q) decreases with increasing tmould and the
relation (qtotal (M W m−2) = 7.3/t0.5mould where tmould is in seconds) was obtained
[52]; however, the integral or total heat flux increases with increasing viscosity. It
was found that the heat flux q was ca. 30% higher for high-speed casting (qto-
0.5
tal = 9.5/tmould) than for conventional slab casting [52]. This was attributed to a
thinner slag film thickness (ds) resulting from lower slag infiltration during
high-speed casting [52]. High-speed casting requires a longer mould.
3.2 Horizontal Heat Flux 79

(v) Low conductivity coatings and refractories

Coating the mould [27, 66–68] with high-melting metal (such as nickel, chro-
mium or molybdenum) tends to reduce the magnitude of the horizontal heat flux
because the conductivities (k25) of these metals (= 83, 90, 139 W m−1 K−1,
respectively) are much lower than that of copper (= 402 W m−1 K−1). The thermal
resistance of the coating is given by (d/k) where d is the thickness of the coating and
k is the conductivity. Thus the thermal resistance of a Ni coating is 5X that of
copper. Coating the mould with these metals also improves mould life [68]; coat-
ings of WC were found to be particularly effective in improving mould life but were
expensive [69]. However, it has been reported that coating the mould with Fe/Ni
caused a slight decrease in the interfacial resistance (RCu/sl) [23].
(vi) Meniscus free casting
The term “Meniscus free casting” is applied to the practice of lining the mould at
the meniscus level with a refractory material (usually boron nitride) [61, 70, 71].
Meniscus free casting reduces the horizontal heat transfer but it can only be used for
billets and blooms because the refractory is not available in pieces long enough for
use with slabs. The reduction of horizontal heat flux in Meniscus free casting causes
an increase in metal temperature at the meniscus level. This causes solidification to
occur lower down the mould (away from the turbulent meniscus region) and has
been reported to give the following benefits:
(i) Shallow oscillation marks and
(ii) A reduction in inclusions trapped by the shell at meniscus level.

(vii) Grooved moulds

In the quest to produce a thin, uniform shell, to minimise any unevenness in the
shell, (especially when casting MC peritectic steels) several workers have used
machined grooves in the meniscus region of the copper mould to reduce the heat
flux in this region [72–79]. The aim is to (i) thicken the slag film and (ii) produce
gas gaps between the slag film and the mould surface which reduce the horizontal
heat transfer and decrease the unevenness in shell thickness (Fig. 3.18). Grooved
moulds give a ca. 10% decrease in heat flux [73, 80]. Sugitani et al. [73] have
discussed the best designs for grooved moulds.
Cho et al. [74] studied the effect of grooves on the slag film using vertical
grooves in the central portion of the mould for distance 0 to 150 mm below the
meniscus. The rapidly cooled slag in the groove tended to form a glassy phase.
They found that
(i) Both the total film and glass film thicknesses decreased exponentially with
increasing distance from the meniscus (Fig. 3.19);
(ii) The slag infiltration into the groove tended to be semi-circular and thus did
not fill the whole groove;
80 3 Heat Transfer in the Mould and Shell Solidification

8
100

Water
(a) S S
Uneven growth
δ -γ

360
Water cooled plate transformation
+

250
Solidified shell thermal contraction
Fine crock

(b)
(A) (B) (c)
S Flat Longitudinal Lattice
Even growth Groove Groove

Fig. 3.18 Schematic diagrams showing a growth of a shell formed during the quench on a flat
plate and that formed on a lattice grooved plate and b various plates used in the quench and the
arrangements of the grooves (D = depth of groove, W = width of groove (0.5 mm) (0.3, 0.5 and
1.0 mm) and s = pitch of grooves (3, 5, 10, 15, 30 mm) [75] (permission granted, ISIJ, [75])

Fig. 3.19 Photographs of the slag film showing thicknesses of the total slag film and the glassy
film as functions of distance from the meniscus [74] (permission granted, ISS/AIST, [74])

(iii) The calculated infiltration ratio (dslag/dgroove) increased as (a) the surface
tension of the slag decreased and (b) groove width increased and
(iv) The calculated infiltration ratio increased from ca. 80 to 95% for positions
50–150 mm below the meniscus.

3.2.4 Measurement and Calculation of Heat Fluxes

3.2.4.1 Plant Trials

The total heat flux (qtotal) in the mould can be determined by monitoring the
temperatures of the cooling water at the entrance and exit of the mould (Tin and Tout,
3.2 Horizontal Heat Flux 81

respectively), the volume flow rate (dV/dT) of the cooling water and using Eq. 3.12
where Cp and qT are the heat capacity and the density of the water at the mean
temperature (= 0.5 (Tout + Tin)) and t is the duration of the flow [3, 17, 43, 49, 81]:

qtotal ¼ ðdV=dT Þt  qT Cp ðTout  Tin Þ: ð3:12Þ

Mould thermal monitoring (MTM) is widely used for detection of potential

sticker breakouts [3, 17, 82–84]. An array of thermocouples provides heat flux
measurements at different locations in the mould. Thus, local values of heat flux are
available with these instrumented moulds. Hanao et al. [3, 17] carried out a thor-
ough investigation of the horizontal heat flux using a pilot caster with an instru-
mented mould in which the slag film was removed and characterised. The MTM
technique is widely used and the heat flux of mould powders has been used to
compare heat fluxes for CS- and CA-based fluxes for casting high Al (Trip) steels
(qCS > qCA) [84].

3.2.4.2 Simulation Experiments

A number of simulation experiments have been carried out to gain an understanding

of the various factors affecting the horizontal heat transfer. The experiments have
tended to become more sophisticated with time. The thickness of the slag specimen
is important because of its effect on (i) thermal resistance of the slag sample (ds/k)
and (ii) optical thickness (a*ds > 3). Consequently, it is important that ds for the
simulation experiment should be similar to that in the slag film; there are some
indications that slag films may be optically thin [85]. Some of the simulation
experiments described below involve both molten steel and slag, whereas others
involve only slag [58, 81]. It has been proposed that the low values of interfacial
resistance values recorded in pilot plant trials [3, 17] are due to the ferrostatic
pressure exerted on the shell; this will be absent in experiments which exclude
molten iron.
(i) “Cold finger” experiments [35, 86–89]
In these experiments a water-cooled, copper mould is lowered, mechanically,
through a slag pool into a bath of molten iron. The passage of the cold finger
through the slag layer results in the formation of a frozen slag film around the finger
(Fig. 3.20a). The slag film was found to be of similar thickness to that formed in the
continuous casting mould [86–88]. Heat fluxes were determined by measuring the
water temperatures at the entrance and exit of the finger and the water flow rate. The
instrument has been modified to allow the effect of oscillation to be identified [88].
The tests which involve only molten slag have been used on various types of slags
including those used for casting HiAl steels [59, 81, 90] and F-free slags containing
TiO2 [91].
82 3 Heat Transfer in the Mould and Shell Solidification

Water in Water Water

in out

Infiltrated slag
1cm Mould
Water Solidified metal
out
Molten metal Copper or Tm
Copper stainless steel
finger Molten slag T1ΔT
Crucible Insulation

Steel plate ds T1
H.F. coil Tst Casting
powder

1.
2.
3.
4.

7. 5. 6.
Tm1 Tm2
J Tw2
H2 O
J T1 /T 2
H2 O
Tw1

Fig. 3.20 Schematic drawings of a cold finger [86], b parallel plate [92] and c vertical plate
experiments [94] J = water in and out (permission granted, a IOM/Taylor & Francis [86] b Wiley
[92] c Verlag Stahleisen GmbH [94])

(ii) Parallel plate experiments [17, 92, 93]

The thickness of the slag film can be accurately controlled in this experiment.
A steel plate (representing the shell) is machined to form a well (of known thick-
ness) to contain the mould slag when it is introduced. It is then heated to melt the
slag (Fig. 3.20b). A water-cooled copper plate (i.e. the mould) is then lowered onto
the surface of the molten slag. The heat flux is determined from the temperature rise
of the water and the water flow rate. An equation relating the thermal conductivity
of the slag film to the composition was obtained with this technique (Eq. 3.13)
where %CaO* = R % (CaO + MgO + MnO + Na2O).


q1200  C Wm1 K1 ¼ 2:030:459ðf%CaO =%SiO2 g þ %B2 O3 Þ
 0:1695ð%FeOÞ  0:034ð%Al2 O3 Þ: ð3:13Þ
In these experiments it was determined that horizontal heat fluxes were consider-
ably greater for glassy slags than those for crystalline or partially crystalline slags [92].
The steel plate has been replaced by both an AlN plate and by a tungsten plate
[17] and the mould by stainless steel [93] in similar tests.
3.2 Horizontal Heat Flux 83

(iii) Vertical plate experiments [40, 94]

In these experiments the slag film thickness is controlled and sited vertically by
pouring molten slag into a vertical rectangular gap (Fig. 3.20c). The plate is heated
on one side and water-cooled on the other as shown in Fig. 3.20c [94]. Another
variation of this technique was developed by Stone [40].
(iv) Bottom-filling simulators [95–97]
These bottom-filling methods have been used mostly in studies of the formation
of oscillation marks [95] and ripple marks [97] and have not been used for heat flux
measurements.
(v) Mould simulators [53, 98–101]
Mould simulators [98, 53, 99–101] are essentially small-scale models of a
continuous caster. The mould consists of two water-cooled, copper plates which
contain a number of thermocouples embedded in the copper to measure local heat
fluxes. A cover plate (connected to the extractor system) separates the two copper
plates. The mould can be oscillated, and the assembly is lowered into a bath of
molten steel (at the target temperature) which is covered with mould powder
(Fig. 3.21a). The cover plate is then lowered and thus the copper plate is exposed to
the molten steel and a steel shell is formed (Fig. 3.21b). The cover plate and shell
are then lowered a further (60–100 mm) which results in further growth of the shell
(Fig. 3.21c). At the end of the experiment the assembly is raised and removed from
the steel and the shell is allowed to cool. The simulator is fitted with a device to

Fig. 3.21 Schematic diagram showing the various stages in the casting process using a mould
simulator [53] a mould is lowered into steel, b cover plate in place (no shell solidification) and
c cover plate is lowered and shell solidification occurs [53]. (courtesy of Prof. Il Sohn)
84 3 Heat Transfer in the Mould and Shell Solidification

maintain a constant mould level. Mould simulators have been used to study (i) heat
flux variations during the oscillation cycle [99, 100] and (ii) the effect of casting
speed and oscillation characteristics on the heat flux [53].

3.2.4.3 Physical Property Measurements

Measurements of the physical properties of mould slags proved very useful in

understanding the heat transfer mechanisms involved in the heat transfer between
the shell and mould. These property measurements are covered in detail in Chap. 9.
The key properties affecting the horizontal heat transfer are as follows:
(i) The thermal conductivity/diffusivity of the slag and the slag film [7, 8, 14, 15,
19, 20, 102–107];
(ii) The optical properties of the slag (particularly the absorption coefficient,
reflectivity and the refractive index) since they affect the magnitude of both
the radiation conduction [6–8, 14] and the optical thickness; the absorption
coefficients (a*) increase with increasing % transition metal oxide, e.g. FeO,
[6, 13, 104, 108–110];
(iii) The fraction of crystalline phase (fcrys) in the slag film since this affects the
magnitude of (a) the thermal conductivity (kcrys  2kglass), (b) the radiation
conductivity (kR, with kR " as fcrys #) and (c) the porosity and surface
roughness (or interfacial resistance,’ RCu/sl) of the slag film which tend to
decrease the horizontal heat flux; and
(iv) The solidification or break temperature (Tbr) since this affects the thickness
of the solid slag film (ds) [111].
These data have been used to gain both an understanding of the mechanisms of
heat transfer occurring in the slag film and to model their effect on the heat transfer
[11, 12, 112]. Values of keff = 2.3 to 3 W m−1 K−1 were obtained from an analysis
of heat flux data recorded in pilot plant trials [17] on three mould slags with
basicities of 0.8, 1.7 and 1.8. The results of this study indicate that kR contributes
ca. 20% to keff for highly crystallised slags but ca. 60% in glassy slags.

3.2.4.4 Mathematical Modelling of Heat Transfer in the Mould

Mathematical modelling of the heat transfer in the mould has evolved over time [1, 2,
16, 32, 113–120]. The more recent models couple shell solidification and heat transfer
for the fluid flow in the steel and slag phases. Furthermore, no assumptions are made
[1, 2] concerning (i) the formation of a slag film and its thickness and (ii) the shape of
the meniscus. It can be seen from Fig. 3.22 that the predicted heat fluxes are in
excellent agreement with those obtained in plant trials [1, 120].
3.3 Shell Solidification and Growth 85

Fig. 3.22 The peak heat flux 4

(i.e. at 45 mm below
meniscus) as a function of
casting speed; ● = plant

qpeak. MWJm-2
3
measurements [121, 125] and
solid line = predicted values
[1, 2, 120] (permission
granted, ISIJ [121] re-drawn) 2

1
0 2 4 6
Vc, m min-1

3.3 Shell Solidification and Growth

Slag films, mould fluxes are important because they affect the nature of the shell;
the condition of the shell is all-important in continuous casting and slag film per-
formance should be regarded in terms of how it affects the shell. The heat flux is
important because it determines the thickness (and hence the strength) of the
solidified shell. Shell solidification occurs principally in the period when the mould
is descending and grows little during the rest of the cycle. Thus the shell does not
have uniform thickness but tends to have a “sausage-shaped” profile (Fig. 3.1).
Unevenness in the shell can be harmful in certain steel grades, particularly those
prone to longitudinal cracking.
Two methods are commonly used to measure shell thickness (dshell). In the first
method the steel is doped with FeS alloy at a specific time and the solidified steel
product at that time is identified and retrieved. The sample is then cut length-wise
and a sulphur print of the section is prepared. The width of the white section near
the surface is taken to be the thickness of the solidified shell and is accurately
measured [42].
In the second method, the growth rate of the shell (d {dshell/dt}) is calculated
from the cooling rate (dT/dt) at specific locations. The cooling rate is determined by
measuring either the arm spacings of the primary or the secondary dendrites [17, 34,
42]. Samples are taken from regions close to the centreline of the slab. The samples
are taken parallel to the surface for measurements of primary arm spacings [44]. In
contrast, samples are taken perpendicular to the casting direction where measure-
ments were made of the secondary arm spacings [42]. The samples were polished
and etched and the secondary arm spacings were measured on photographs of the
optical micrographs [42]. The cooling rate (dT/dt) is related to the temperature
gradient (dT/dx) and the rate of solidification (d {dshell/dt}) as shown in Eq. 3.14
where x is distance. The cooling rate at different distances can be determined in this
way:
86 3 Heat Transfer in the Mould and Shell Solidification

Fig. 3.23 Thickness of the

Thickness of solidified shell, dshell (mm)

6
solidified shell (45 mm below 4.0 m/min
meniscus); o, as a function of 5 dshell=30.0 √ t-1.76
5.0 m/min
square root of solidification 45mm below meniscus level
time [42]; o, □ = 4 and 5 m 4 3.0m/min
1.2 ≠ Vc
4.0m/min
min-1, respectively; solid line 3 5.0m/min
above t0.5 > 0.06 is dshell value
calculated by 30t0.5 − 1.76 2
(permission granted, ISIJ,
[42]) 1 dshell=69.7t

0 0.05 0.10 0.15 0.20 0.25

1/2 (min 1/2)
Square root of solidification time, t

ðdT=dtÞ ¼ ðdT=dxÞðd dshell =dtÞ: ð3:14Þ

q ¼ kðdT=dxÞ ð3:15Þ

ðdT=dtÞ ¼ ðq=kÞðddshell =dtÞ: ð3:16Þ

Figure 3.23 a shows the results reported by Hanao et al. [42]. The solidification
rate is given by the gradient of the plot (d dshell/d√t).
An alternative method to calculate the cooling rate is by the use of the relation
between it and the secondary dendrite arm spacing (dSDA) reported by Suzuki
[122, 123]:

dSDA ¼ 710ðdT=dtÞ0:39 : ð3:17Þ

It is customary to express the thickness of the shell in the form of Eq. 3.18 (see
[124]), where K is the solidification coefficient and a′ is a constant (usually with a
negative value):
pffi
dshell ¼ K t þ a0 : ð3:18Þ

Relationships relevant to shell thicknesses have been reported by various

investigators and the results are summarised in Table 3.2. Most of the results show
that the initial growth rate of the shell is slow and then at a certain point (around
0.07 min−0.5) the growth rate suddenly increases as shown in Fig. 3.23 [42]. Values
for (dshell/dt0.5) = 69.7t for 0 < t
0.005 and (dshell/dt0.5) = 30√t − 1.76) for t >
0.0055. Most of the results of the studies reported in Table 3.2 show similar
behaviour [17, 42, 122, 125–130]. One possible reason for the slow initial growth is
that the initial shell has a curved profile (see Fig. 3.1) and the early shell is further
from the mould wall than later but this separation distance will decrease as the shell
moves downward and reaches the gap. If this is correct, the delay time would be
3.3 Shell Solidification and Growth 87

Table 3.2 Reported relationships for the thickness of the shell; *denotes mean values; A, B,
C = 3 different mould powders with (C/S) = 0.8, 1.8.1.7, respectively
Reference Steel Casting speed Dimensions Super dshell (mm) K
(m min−1) w  t (m) heat (oC) (m min−0.5)
a′ (mm)
Nagaoka [126] 0.15C 0.5–0.65 1.6  0.20 ¼19.1√t − 2.4
Suzuki [122] HC 45–55 ¼k√t − a′ k = 34.7 to
38
a′ = 2.7 to
3.2
Okano [127] MC 0.65 0.11  0.11 ¼18.4√t − 1.5
Mori [128] 1.6  0.23 ¼23.9√t − 0.86
Fujii [129] LC 0.65 2.1  0.25 20 ¼28.3√t − 8.6
Narita [130] LC 0.65 1.6  0.23 ¼27.5√t − 2.8
M Suzuki LC 2.0–2.5 1.55  0.22 25–30 ¼23√t − 2.72
[125]
Hanao [17, 42] MC 1.1–1.6 0.9*  0.1* 12–36 ¼30√t − 1.76
MC 1.9*  0.25* 80–89 A: = 24.8√t − 1.47
0.8  0.1 80–89 B: = 20.6√t − 1.14
C: = 18.7√t − 0.99

expected to decrease with increasing superheat since solidification will be delayed

to a point further down the mould. The shell thickness increases with increasing
time (Eq 3.18) in the mould and thus increases as the distance from the meniscus
(dmenis) increases. An empirical relation, Eq. 3.19 was derived from an analysis of
plant data [52] for the thickness of the shell at the mould exit (dexit
shell in mm) where K
′ is a constant with a value of 7.3 for conventional slab casting and 9.5 for
high-speed casting:

dshellexit ¼ K 0 tmould
0:5
=3:6: ð3:19Þ

The effect of increased casting speed on shell thickness can be clearly seen in
Fig. 3.10 [42] where increased casting speed results in a thinner shell at
t0.5 = 0.07 (min) −0.5−or t = ca. 5  10−4 min because of the shorter residence time
(tmould).
The nature of the mould slag was found to affect the dendritic structure of the
solidified shell with a more crystalline slag film giving an increase in primary arm
spacing [34].

3.4 Variability in Heat Flux

Variations in heat flux can cause longitudinal cracking, longitudinal corner cracking
and star cracking. These variations in heat flux cover those arising from thermal
gradients (dT/dx) and from variations over time (dT/dt). Some variations are
88 3 Heat Transfer in the Mould and Shell Solidification

inevitable (e.g. those occurring through an oscillation cycle) whereas other sources
of variability can be minimised or partially controlled. The various sources of
variability in heat flux in the mould are discussed below. Heat flux variations result
in (i) variations in shell thickness and longitudinal cracking in the upper mould
[131, 132] and (ii) star cracking, in the lower mould [133].

3.4.1 Variations in Heat Flux (qHor) During the Oscillation

Cycle

The heat flux varies through the oscillation cycle [99, 100]. The predictions of a
recent mathematical model [1, 2] are shown in Fig. 3.24. The oscillation charac-
teristics are usually defined in terms of negative (tn) and positive strip time (tp)
shown as a solid line in Fig. 3.24b but can also be defined in terms of position in the
mould (shown as a dotted line in Fig. 3.24b). It can be seen from Fig. 3.24c that the
heat flux (shown in red) is at its lowest value when the mould is at its highest
position (in late tp). The heat flux continues to increase as the mould descends
(throughout tn) and reaches its highest value when the mould is at its lowest
position (in early tp). The heat flux continues to decrease as the mould ascends.
Shell solidification follows the heat flux since a high heat flux results in a thick
shell. Thus the variation of heat flux through the oscillation cycle results in a shell
of varying thickness which is responsible for the corrugated or “sausage-shaped”
appearance of the shell.
These variations in heat flux during the oscillation cycle result from changes in
the direction of the flow in the slag pool. It can be seen from Fig. 3.25a that when the
mould is ascending the slag flow is outward (towards the narrow face) and upward;
such a flow is warm (tropical) because it moves over the surface of the molten steel.
In contrast when the mould is descending the direction of flow in the slag pool is
downward (Fig. 3.25b) [1, 2]. This flow will be cold (Arctic) because it originates in
the colder parts of the slag pool. This downward, arctic flow is responsible for
increasing heat flux and shell solidification and these will achieve a maximum value
when the mould is at its lowest position (in early tp) shown in Fig. 3.25c.

3.4.2 Thermal Gradient Variations Arising from Metal

Flow and Other Causes

The variations in metal flow are initiated in the SEN. The first metal flowing
through the SEN strikes the baseplate and rebounds; this results in the formation of
a vortex in the sump of the SEN as shown in Fig. 3.26a [134]. This vortex affects
the flow of metal through the ports of the SEN, sometimes favouring the flow
through one port and at other times favouring the flow through the other port.
3.4 Variability in Heat Flux 89

Fig. 3.24 Schematic drawings showing a the shell and slag rim, b the relation between mould
velocity (solid curve) and position (dotted curve during an oscillation cycle) and c mathematical
model predictions [1, 2] from the top of (i) profile of strand surface, (ii) heat flux, (iii) powder
consumption in kg s−1, (iv) liquid film thickness, dl, (v) solid slag film thickness, ds and
(vi) pressure during five oscillation cycles (permission granted, ISIJ, [1])

Recently, the predictions of a mathematical model of the metal flow in the SEN
indicated that the vortex creates a swirl in the metal which favours the flow towards
the loose side [135] as shown in Fig. 3.26b. Thus the metal flow in the mould in the
mould is continually fluctuating. Furthermore, the metal flow rebounds from the
narrow face wall and this causes the formation of waves and in some cases, vortices
in the meniscus region. These waves and vortices tend to move either side of the
90 3 Heat Transfer in the Mould and Shell Solidification

Fig. 3.25 Schematic diagram showing the direction of slag flow at different parts of the
oscillation cycle a at highest position of mould in late tp, b halfway through tn, c at lowest position
of mould in early tp and d midway through tp [1, 2] (permission granted, ISIJ, [1])

Fig. 3.26 Schematic drawings showing a the fluid flow in the SEN and the formation of vortex in
the sump of the SEN [134] and model predictions showing, b the bias in metal flow even with
EMBr [135] and c low-frequency variations of heat flux and solid slag film thickness over a
number of oscillations [136] (permission granted, a ISIJ [134], b Verlag Stahleisen GmbH [135],
c IOM/Taylor and Francis, [136])

SEN in response to which flow is the more dominant at the time. In fact, the metal
flows and heat flows are continually fluctuating and interacting with other param-
eters so the situation is one of continual changes in response to the existing con-
ditions. These transient fluctuations increase with increasing casting speed. In
extreme cases, when metal flow variations combine with other variations (e.g.
mould level variations) and all the effects are acting in collusion, they can lead to a
calamitous event such as a breakout. This has been cited as an example of “the
3.4 Variability in Heat Flux 91

butterfly effect” [136]. Such variations through one oscillation are often referred to
as high-frequency variations and exemplified by the variations in oscillation mark
depth shown in Fig. 3.24c (i). However, model predictions indicate that there are
also low-frequency variations going on in the mould which occur over a period of
ca. 30 min; these can be seen in Fig. 3.26c [1].
Transient thermal fluctuations (i.e. dT/dt) can also occur in the lower mould
when the liquid slag film has become too thin and the resulting frictional forces
cause “spalling” and fracture of the solid slag film. This, in turn, results in vari-
ability in the horizontal heat flux (as shown in Fig. 3.27) and in star cracking
[133, 137].
Thermal gradients (dT/dx) in the mould can occur locally in response to
(i) Overcooled corners in billets and at “corners” in both beam-casting moulds
and football-shaped moulds used in thin-slab casting (see Chap. 11;
Figs. 11.13, 11.14 and 11.15);
(ii) Conduction via high thermal conductivity SENs causing local “cold spots” in
the mould [132] and
(iii) The impact of the metal flow on the shell resulting in both a very thin shell
and a large thermal gradient in the corner (see Fig. 3.31) which can lead to
longitudinal corner cracking [138].

3.4.3 Mould Level Variations

Model level variations can also cause variations in the heat flux. Level variations
can arise through (i) poor meniscus level control or (ii) in response to bulging and

Fig. 3.27 Heat flux as a function of time when mould powder L11 is replaced by powder L9
resulting in large fluctuations in heat flux and then the re-establishment of a stable heat flux after
the replacement of powder L9 by powder L11; incidences of star cracking coincided with large
heat flux variations [133, 137] (permission granted, IOM/Taylor & Francis, [133])
92 3 Heat Transfer in the Mould and Shell Solidification

squeezing at the first rolls [139] when the shell at the mould exit is too thin (usually
when casting MC steels). Large mould level variations can result in the capture of
the slag rim which causes a rapid local decrease in horizontal heat flux as it passes
down the mould (see Sect. 11.7.2).

3.4.4 Carbon Content of Steel

When molten steel solidifies d-Fe (ferrite) is formed and this phase then transforms
to austenite (c–Fe) via the peritectic reaction:

Liquid Fe þ d  FeðferriteÞ ¼ cFeðausteniteÞ: ð3:20Þ

The peritectic reaction (d ! c) is particularly important in steels containing

0.08–0.17%C (although the range is affected by the presence of other elements).
The packing in the FCC, austenite phase is tighter than that in the bcc, d-ferrite, so
the transformation is accompanied by a volume decrease of 0.4–0.6%. The
shrinkage produces stresses which cause distortion in the shell (see Fig. 3.28) [74]
which can result in longitudinal cracking [140]. These distortions cause unevenness
in the shell (Fig. 3.28a) and this unevenness is maintained by the strength of the
austenite phase.

Fig. 3.28 Photographs showing a wrinkled shell resulting from the peritectic reaction [141] and
b unevenness of shell formed by (i) a steel with %C = 0.057 quenched onto copper, (ii) steel with
%C = 0.122 and quenched onto copper, (iii) steel with %C = 0.133 and quenched onto SUS 304
stainless steel (slower cooling rate) and (iv) as in c but SUS304 heated to 200 °C [141] (permission
granted, ISS/AIST, [141])
3.4 Variability in Heat Flux 93

Fig. 3.29 a Schematic diagram showing the measurements made to derive the unevenness
parameter (Dd/l)mean and b values for the parameter (Dd/l)mean as a function of the C-content of the
steel [75] (courtesy of ISIJ)

Murikami et al. [75] carried out dip tests in which water-cooled plates were
dipped into molten steels and the resulting shells were examined. The unevenness
of the shells produced was quantified in terms of the average value of the parameter
(Dd/l)mean (= R (d1 − d2/l) see Fig. 3.29a). It can be seen from Fig. 3.29b that the
parameter (Dd/l)mean represents the unevenness of the shell and is much greater for
steels in the peritectic range.
The unevenness in the shell also appeared to increase with increasing cooling
rate. Hanao et al. [17, 34] used a different measure of unevenness, the Uneveness
Index (= d dshell) where d dshell is the standard deviation of differences in dshell. This
index was found to (i) become significant when dshell 1 mm, (ii) was greater for
glassy slags than for predominantly crystalline slags (since both qhor and the cooling
rate were higher) and (iii) increased gradually with increasing dshell [17, 34].
This unevenness in the shell causes variability in the mould thermal monitoring
(MTM) traces (Fig. 3.30) [82]; since longitudinal cracking is associated with thick,

Fig. 3.30 Mould temperature

as a function of elapsed time
in mould thermal monitoring
(MTM) traces [82]
(permission granted,
IOM/Taylor & Francis, [82])
94 3 Heat Transfer in the Mould and Shell Solidification

(a)

2mm

(b)

Region of solidification delay

Corner
Crack
Wide side
Narrow side

“White band”
Interface of liquid and solid at some moment

Fig. 3.31 a Photograph of sulphur print showing shell profile and the location of the longitudinal
corner cracking, b schematic drawing showing the “shell thinning” caused by the metal flow [138]
(permission granted, UNESID, [138])

uneven shells, it has been proposed that variability in MTM traces provides a way
of monitoring for outbreaks of longitudinal cracking. It would appear that when
shells (usually for MC steels) develop a certain thickness, the unevenness in the
shell is sufficient to cause variability in the MTM trace and this signals the pos-
sibility of longitudinal cracking.

3.4.5 Thermal Gradients in the Mould

Some forms of cracking are related to variations in the local heat flux along the
shell. These heat flux variations can arise from a variety of conditions:
(i) The presence of a high conductivity, SEN (containing carbon) can extract
heat and if the SEN is close to the mould, it can create a cold spot in the
mould, which, in turn, can cause large, local differences in shell thickness
3.4 Variability in Heat Flux 95

Fig. 3.32 The effect of EMS 1

Longitudinal crack index

stirring velocity on the
incidence of longitudinal 0.8
cracking [142] (permission
granted, Nippon Steel 0.6
Sumitomo Metal
Corp. NSSMC [142]) 0.4

0.2

0
0 20 40 60 80 100
EM s rring velocity

which create stresses and lead to longitudinal cracking (especially in MC

steels, see Chap. 11) [80].
(ii) Excessive cooling in the corners (especially in billet casting) can result in
variations in shell thickness which can lead to longitudinal corner cracking.
(iii) In slab casting, variations in the shell thickness can result from the flow of
hot metal into the corner region which causes “shell thinning” in these corner
regions and the resultant variations in shell thickness can lead to longitudinal
corner cracking [138].
(iv) It has been reported that mould coatings (e.g. Ni) tend to reduce thermal
gradients in the mould [68].
(v) EMS and EMBr help to reduce thermal gradients in the mould and thus help
to minimise longitudinal cracking caused by steep differences in shell
thickness. The beneficial effects of EMS in reducing local thermal gradients
can be clearly seen in Fig. 3.32 where the incidence of longitudinal cracking
is decreased with increasing stirring velocity [142].

3.4.6 Fracture of Slag Films

Some variability in the heat flux is caused by the fracture of the slag film [143, 144]:
These fractures can occur in both the upper and lower parts of the mould; they are
treated separately below.
Fractures can occur in the border region between slag rim and slag film [143,
144]. Following the fracture, the space between rim and film is filled by liquid slag
which results in a glassy, repaired region [143, 144] of the slag film. These fractures
are characterised by increases in heat flux (due to the high radiation conduction
contribution through the new glassy slag film) followed by a gradual decrease in
heat flux (due to the crystallisation of the slag film and the decrease in kR) as shown
in Fig. 3.33 [144, 145].
96 3 Heat Transfer in the Mould and Shell Solidification

Fig. 3.33 Mould temperature 250

229 mm FROM MOLD TOP
as a function of time showing POWDER A
periodic increases in mould
temperature caused by

TEMPERATURE, C
fracture of the slag film 200
followed by a gradual
decrease in temperature
during the repair period [145]
(permission granted, 150
ISS/AIST [145])
WIDE WALL - QUARTER PLANE
EAST NARROW WALL - MID. PLANE

100
0 2 4 6 8 10 12 14 16 18 20
TIME, MINS,

Fracture in the lower mould occurs when enhanced heat extraction in the upper
part of the mould results in the loss of liquid lubrication in the lower mould (i.e.
liquid slag film is too thin). This results in increased friction and in “spalling” and
fracture of the solid slag film [133, 137] which causes marked variations in heat flux
(see Fig. 3.27). The repair of the slag film can take a significant time (e.g. 20 min)
because of the poor supply of liquid slag. The variations in heat flux result in “star
cracking” and “spongy cracking” [133, 137] (Chap. 11, Sect. 11.8).

3.5 Vertical Heat Flux

The upward transfer of heat from the steel meniscus through the liquid pool,
sintered and powder layers is referred to as vertical heat transfer. The formation of
the bed is an important step in the process. The powder is usually dispensed
automatically and various types of dispensers are available; the strengths and
weaknesses of various dispensers have been reviewed [146]. A device for moni-
toring the thickness of the bed has also been reported [147].

3.5.1 Heat Transfer Mechanisms Involved in Vertical Heat

Transfer

Several heat transfer mechanisms are simultaneously operative in both the liquid
and solid layers of the bed (shown in Fig. 3.34). In the liquid slag pool, convection,
lattice conduction and radiation conduction occur, simultaneously. In the powder
bed (consisting of sintered and powder layers) lattice conduction and gaseous
conduction are the principal mechanisms involved.
3.5 Vertical Heat Flux 97

Fig. 3.34 Schematic drawing showing the various layers formed in the powder bed

3.5.2 Factors Affecting Vertical Heat Transfer

The vertical heat flux is linked to the horizontal heat flux since they share a
common source of heat, i.e. the molten metal discharging from the SEN. The
vertical heat flux is determined by several factors as follows:
(i) The source of heat, i.e. the flow rate of the molten steel, the metal flow
pattern and the degree of superheat;
(ii) The efficiency of heat transfer; and
(iii) The thermal insulation provided by the bed.
These factors are themselves affected by other parameters, e.g. thermal insulation
is affected by (a) the temperature gradients in the beds and (b) the average thermal
conductivities of the sinter and powder beds.
The various factors affecting the vertical heat transfer are detailed below.

3.5.2.1 Steel Flow Rate and Superheat

It is self-evident that the vertical heat flux density (qvert with units of W m−2), like
the horizontal heat transfer, will increase with increasing casting speed. It should be
noted the total horizontal heat extracted (J m−2) in the mould tends to decrease with
increasing casting speed because of the shorter residence time [47].
The heat flux increases with increasing temperature gradient (dT/dx see Eq. 3.2)
and superheat increases the temperature gradient. Thus increased superheat will
result in an increase in heat flux density (qvert).
The metal flow pattern is important too. The “double roll” pattern is preferred
usually to the “single roll” pattern since the latter brings cold metal to the surface
which can result in freezing of the metal (i.e. skull formation) [148]. The metal flow
pattern is determined by (i) steel flow rate, (ii) SEN immersion depth, (iii) SEN port
98 3 Heat Transfer in the Mould and Shell Solidification

design and (iv) argon flow rate and these factors are interactive. Recent work has
shown that small changes in Ar flow rate can cause changes in the metal flow
pattern (see Sect. 3.5.2.7).

3.5.2.2 Efficiency of Heat Transfer

Heat transfer through turbulent flow is more efficient than that with laminar flow.
Thus vertical heat transfer is expected to increase with casting speed increase since
the metal flow will become more turbulent and this, in turn, will induce more
turbulent flow in the molten slag pool. Convection in the slag pool makes a sig-
nificant contribution to the vertical heat flux; convection in the slag pool is inversely
dependent upon the slag viscosity. Some mathematical models assume convection
of the pool is a multiple of the conductivity of the liquid slag (kconv = C klat) and
reported values for the constant C vary between 10 and 100 with an optimum value
of 15 [149].
Electromagnetic braking (EMBr) is used to reduce both the metal flow velocities
and the turbulence in the mould when using high-casting speeds. EMBr reduces
(i) the turbulent flow at the interface and (ii) the penetration depth of the metal flow;
the latter results in a 5–10 °C increase in the steel meniscus temperature. It has been
calculated that EMBr causes a 30% decrease in vertical heat flux [150]. However,
this will be partially offset by increased convection in the slag pool resulting from
the lower slag viscosity arising from the increased meniscus temperature [151]
(Fig. 3.35).

3.5.2.3 Thermal Insulation of Beds

If we consider the bed to consist of two layers (i.e. powder and sintered slag), the
thicknesses of these two layers and the liquid slag pool (dpool) have been studied by
a number of investigators [152–158]. The vertical heat flux for the bed is given by

Fig. 3.35 Difference -10

between tundish and mould
temperatures as a function of
(Tmould- T tundish) , oC

position in the mould

-20
(distance from centre) [151];
● = with EMBr; o = no
EMBr, (re-drawn after [151])
-30

-40
0 200 400 600 800
Posi on in mould, mm
3.5 Vertical Heat Flux 99


qvert;bed ¼ kbed Tbed=pool  Tair=bed =dbed ; ð3:21Þ

where kbed = mean thermal conductivity for the two layers, Tbed/pool = the tem-
perature at the interface between the pool and the bed, Tair/bed is the temperature at
the air/bed interface (usually between 200 and 400 °C) [154]) and dbed is the
distance between the interfaces of the bed with the air and the molten pool. It is
apparent that thermal insulation of the bed (1/kbed) increases as the depth of the bed
increases (see Fig. 4.60 [152–158]) as kbed decreases and the temperature gradient
decreases.
The same finding would be obtained if we applied the analysis to the heat flux to
any specific layer. In practice, the thickness of these layers is determined by the heat
flux entering and leaving each layer.
The vertical heat transfer has been studied using (i) simulation experiments [29,
159, 160] (see Sect. 9.8.4 and Fig. 9.52) and (ii) mathematical modelling [155–
158]. In the latter, the temperature gradients in the various layers of the bed are
affected by both the occurrence and the site of the various endothermic and
exothermic reactions [152, 155, 157, 158, 161]. The exothermic, carbon combus-
tion is affected by the size of the carbon particles (DC) with larger particles taking
longer to combust. The kinetics of carbon combustion has been studied by both
experiments [157, 158, 162] and through mathematical models of the bed [155–
158, 163] in which the permeability of the bed is taken into account [157, 158]. The
rate of sintering of the oxide particles in the sinter layer has also been studied
through experiments [152, 162, 164] and mathematical modelling [163, 164].
Thermal conductivity measurements have been carried out on casting powders
by several workers [102, 103, 165–167]; reported values are shown in Fig. 3.36.

0.8
kλ, W m-1K-1

0.4

0
200 400 600 800 1000 1200 1400 1600
T, K

Fig. 3.36 Thermal conductivity of casting powders as a function of temperature (in K); dotted and
solid lines, = Andersson values [167]; +, X; ♦ = k THW due to Macho [166]; dashed line = Nagata
values [101] □, D, ○ = values due to Taylor [102]; ▲, ◊ = Neumann [165] granules and powder,
respectively (permission granted, Taylor & Francis, [167])
100 3 Heat Transfer in the Mould and Shell Solidification

Fig. 3.37 Thermal 0.16

conductivity of powders at
20 °C as a function of bulk 0.15
density [167] (permission
0.14

k20, Wm-1K-1
granted, Taylor & Francis,
[167]) 0.13

0.12

0.11

0.1
500 600 700 800 900
ρbulk , kg.m-3

With normal packing, the thermal conductivities of the powders (kpowd ) at 25 °C

have values around 0.12–0.15 W m−1 K−1 [167]. Values of kpowd increase with
increasing temperature (dk dT = 9  10−5 W m−1 K−2). Thus approximate values
for kpowd for temperatures in the range (25–750 °C) are given by Eq. 3.22:


kpowd T Wm1 K1 ¼ 0:13 þ 9x 105 ðT  25  CÞ: ð3:22Þ

Values of kpowd also increase with increasing powder density ðqbulk Þ due to the
improved packing density and improved contact between particles (Fig. 3.37). The
following equation was obtained by Andersson [167]:


kpowd 25 Wm1 K1 ¼ 0:01 þ 1:69x 104 qbulk : ð3:23Þ

The thermal conductivities of the sintered layer will, in general, be higher than
those for the powder bed. The thermal conductivity of the sinter layer will decrease
with increasing porosity (or decreasing density). The porosity in the sinter layer
may decrease (or thermal conductivity increase) with increasing local temperature
in the sinter bed. Thermal conductivity values for the sintered layer will lie between
those for powders (ca. 0.15 W m−1 K−1) and values reported for sintered slags with
higher densities (1–2 W m−1 K−1); in the absence of measured data a value of ca.
0.5 W m−1 K−1 is suggested for ksint.
However, thermal insulation tests give thermal conductivity values for the bed in
the range kbed = 1.0 (for powders) to 1.3 W m−1 K−1 (for granules) [160] which
are almost an order of magnitude higher than the values for powders given in
Fig. 3.36. This discrepancy is probably due to the contribution from gaseous
convection which would be expected to be large in tests where the powder is heated
from below and where there is a large temperature gradient between the lower and
upper surfaces (as there is in the thermal insulation tests). The question arises which
value should be used to calculate the vertical heat flux. It is suggested that
3.5 Vertical Heat Flux 101

(i) The higher kbed value be used if gaseous convection is not being calculated
separately and
(ii) The lower kbed value be used if gaseous convection is being calculated
separately.

3.5.2.4 Depth of Bed

It can be seen from Eq. 3.19 that the heat flux decreases (i.e. thermal insulation
improved) as the depth of the powder bed increases (see Fig. 4.6) [153–156, 158].
Thus using a deeper powder bed is one way of decreasing the vertical heat flux and
can be used to shorten the size of the solidified meniscus in order to reduce the
entrapment of inclusions and bubbles and the depth of oscillation marks [165].

3.5.2.5 Powder/ Granule Size and Packing Density

The thermal insulation of the bed has been reported to increase as the mean
powder/granule size (Dbed) decreased [160]. This is due to the fact that smaller
grained powders pack more densely and thereby reduce the permeability of the bed
to gases passing through it. This, in turn, reduces the effective thermal conductivity
of the sample. Thus thermal insulation is better for powder mould fluxes than for
granulated fluxes since powder particles have smaller diameters.

3.5.2.6 Powders Containing Exothermic Agents

It can be seen from Eq. 3.19 that the heat flux increases with increasing temperature
gradient (dT/dx). Consider two locations in the bed, with site 2 close to the interface
with the liquid pool and site 1 higher up the bed; the temperature gradient is given
by {(T2 − T1)/(d2 − d1)} where d is the distance from the powder/air interface.
Exothermic powders usually contain silicon in the form of silicides (e.g. Fe/Si).
At a certain temperature these silicides react with oxygen to form SiO2. This
reaction is very exothermic and the heat given out causes the local temperature to
rise and thereby reduces the temperature gradient and hence, the vertical heat flux.
The combustion of carbon is also exothermic and incidences of skull formation
have been reported when free carbon contents have been lowered too much in the
search for faster melting rates.
By the same token, the decomposition of carbonates and the melting of slag are
both endothermic and thus these will cause decrease in local temperature increase in
temperature gradient thereby increasing the vertical heat flux.
102 3 Heat Transfer in the Mould and Shell Solidification

Fig. 3.38 The powder 0.8

consumption, Qt, as a
function of Ar flow rate [170]

-1
, kg tonne
(courtesy of AB Fox) 0.6

powd
0.4

Qt
0.2
0 1 2 3 4 5
Ar flow rate, l min-1

3.5.2.7 Argon Blowing

Argon blowing increases the vertical heat flux due to the promotion of gaseous
convection. It is known that bigger bubbles emerge close to the SEN and smaller
bubbles emerge near the mould walls (where they may be incorporated into the
metal flow [168]). It would thus be expected that the vertical heat flux due to Ar
flow would be greater near the SEN than at the edges of the mould. However, Ar
bubbling also affects the metal flow in the mould. The Ar bubbles tend to “cushion”
the metal flow. However, it has been reported that an increase in Ar flow rate from 4
to 5 L min−1 causes the metal flow pattern to change from a “double roll” to “single
roll” pattern [135, 169]; this is accompanied by a decrease of the steel temperature
(at the slag/steel interface) and has been reported to lead to skull formation on
occasions.
Since gaseous convection contributes to the vertical heat flux, it is necessary to
consider the permeability of the powder bed since this too affects the magnitude of
the contribution. It is known that (i) the thermal insulation of a bed derived using
powders is greater than that for granules and (ii) thermal insulation improves as the
diameter (D) of the particle or granule decreases (kbed # as D #). This might help to
explain the increased powder consumption with increased Ar flow rate (Fig. 3.38)
[170]. Increased vertical heat flux (i.e. gaseous convection) leaving the molten slag
pool would be expected to increase the melting rate of the powder in the bed and
thus lead to higher powder consumption.

3.6 Summary

Control of the heat transfer from shell to mould is essential in order to minimise
incidences of longitudinal cracking and sticker breakout. The heat transfer is usu-
ally divided into horizontal and vertical heat transfer with the horizontal heat flux
being the more important. The factors affecting four aspects of heat transfer in the
mould are considered here, namely (i) horizontal heat transfer, (ii) shell solidifi-
cation, (iii) vertical heat transfer and (iv) the variability in heat transfer.
3.6 Summary 103

Although many factors affect horizontal heat transfer, control is usually exerted
through manipulation of the slag film formed between the steel shell and the mould.
The principal factors involved are (i) the thickness of solid slag film (which is
dependent on the solidification or break temperature) and (ii) the fraction of crys-
talline phase (fcryst) formed in the film (which reduces the radiation conduction and
creates an interfacial resistance, RCu/sl). In practice, the heat flux is usually reduced
by increasing the basicity of the mould powder.
The metal flow is also an important factor but usually the factors controlling
metal flow (such as SEN immersion depth) are optimised to provide a satisfactory
“double roll” pattern and so metal flow tends not to be an issue. However, SEN
blockages and clogging can lead to unsatisfactory metal flow patterns and excessive
Ar flow rates can lead to the formation of “single roll” patterns.
Vertical heat transfer can be controlled through (i) the depth of the bed, (ii) use of
exothermic mould powders, (iii) to a less amount by the particle size of the powder
and (iv) by the use of EMBr which reduces the efficiency of heat transfer by
retarding the metal flow velocity. Gaseous convection is a contributor to the vertical
heat flux, which, in turn, is affected by the Ar flow rate, the carbonate content of the
mould powder and the permeability of the powder bed

References

1. PE Ramirez-Lopez, KC Mills, PD Lee, B Santillana, ISIJ Intl., 50, 1797, (2010).

2. PE Ramirez-Lopez, KC Mills, PD Lee, B Santillana, Met. Mater. Trans. 43, 109, (2011).
3. M Hanao, M Kawamoto, ISIJ Intl., 48, 180, (2008).
4. H Jacobi, Arch. Eisenhüttenwesen, 47, 441, (1976).
5. R. Gardon, Proc. 2nd Intl. Conf. Thermal Conductivity, Ottawa, 1962, p. 167 and J. Amer.
Ceram. Soc. 44, 305, (1961).
6. M Susa, K Nagata, KC Mills, Ironmaking and Steelmaking, 20, 312, (1993).
7. S Ozawa, M Susa, T Goto, R Kojima, KC Mills, ISIJ Intl., 46, 413, (2006).
8. M. Susa, KC Mills, MJ Richardson, R Taylor, D Stewart, Ironmaking and Steelmaking, 21,
279, (1994).
9. M Hayashi, K Matsuo, K Nagata, H Nakada, Proc. 8th Intl. Conf. Molten Slags, fluxes and
salts, Santiago, Chile, 2009, ( GECAMIN, Santiago, Chile, 2009) p. 1091.
10. M. Kawamoto, Paper presented at Intl. Workshop on “Thermophysical data for the
development of mathematical models of solidification”, Gifu City, Japan, Oct. (1995) p. 10.
11. M Susa, A Kushimoto, H Toyoto, M Hayashi, R Endo, Y Kobayashi, ISIJ Intl., 49, 1722,
(2009).
12. M Susa, A Kushimoto, R Endo, Y Kobayashi, ISIJ Intl., 51, 1587, (2011).
13. Y Kobayashi, R Maehashi, R Endo, M Susa, ISIJ Intl., 53, 1725, (2013).
14. JW Cho, T Emi, H Shibata, M Suzuki, ISIJ Intl., 38, 268, and 440 (1998).
15. H Shibata, K Kondo, M Suzuki, T Emi, ISIJ Intl., 36, S 179, (1996).
16. MS Jenkins, Proc. 78th Steelmaking Conf. (1995) p 669.
17. M Hanao, M Kawamoto, A Yamanaka, ISIJ Intl., 52, 1310, (2012).
18. K Nishioka, T Maeda, M Shimizu, ISIJ Intl., 46, 427, (2006).
19. P. Andersson, C Eggertson, Ironmaking and Steelmaking, 42, No 6, (2015), p. 456.
20. P Andersson, Thermal conductivity of powders used in continuous casting of slabs and
billets. To be published.
104 3 Heat Transfer in the Mould and Shell Solidification

21. R. Saraswat, A.B. Fox, K.C. Mills, P.D. Lee, B. Deo, Scand., J Met., 33, 85, (2004).
22. IV Samarasekera, JK Brimacombe, Intl. Metals Review, 286, (1978).
23. J. W. Cho, H. Shibata, T. Emi, M. Suzuki, ISIJ Int., 38 (5), 440 (1998),
24. K Watanabe et al. Proc.79th Steelmaking Conf.,1996, (ISS, Warrendale, PA, 1996) p. 265.
25. H Shibata, JW Cho, T Emi, M Suzuki, Proc. 5th Intl. Conf. Molten slags, fluxes and salts,
Sydney, 1997, (ISI –AIME, Warrendale, PA, 1997) p. 771.
26. H Nakada, K Nagata, ISIJ Intl., 46, 441, (2006).
27. K Tsutsumi, T Nagasaka, M Hino, ISIJ Intl., 39, 1150, (1999).
28. JW Cho, T Emi, H Shibata, M Suzuki, ISIJ Intl., 38, 834 (1998).
29. KH Spitzer, JF Holzhauser, FU Bruchner, B Siera, HJ Gretthe, K Schwerdtfeger, Stahl u
Eisen, 108 (9), 71, (1988).
30. M Ichihashi, M Kawamoto, Y Tsukaguchi, N Nishida, T Kanazawa, S Hiraki, Jap. Soc.
Promotion of Science (1995).
31. K Watanabe, M Suzuki, K Murakami, Tetsu-to-Hagane, 83, 115, (1997).
32. A Yamauchi, K Sorimachi, T Yamauchi, Ironmaking and Steelmaking, 29, 203, (2002).
33. T Inoue, K Noro, Y Akita, I Katano, Nippon Steel Technical Rept. 12, 86, (1978) 86/96 see
also Seitetsu-Kenkyu: 294, (1973) 12412.
34. M Hanao, M Kawamoto, A Yamanaka, Proc. 8th Europ. Conf. Continuous Casting, Graz,
2014, (Austrian Metals Mater. Soc., 2014).
35. H Yamamura, T Kajitani, J Nakashima, M Yamasaki, S Mineta, Nippon Steel Technical
Report, 104, 54, (2013).
36. J Kromhout, RS Schimmel, Proc. 8th Europ. Conf on Continuous Casting, Graz, 2014,
(Austrian Metals Soc., 2014).
37. A Yamauchi, K Sorimachi, T Sakuraya, T Fujii, ISIJ Intl., 33, 140, (1993).
38. M Kawamoto, A Hakari, T Watanabe, CAMP-ISIJ, 10, 890, (1997).
39. H Nakato, M Ozawa, K Kinoshita, Tetsu-to Hagane, 67, 1200, (1981).
40. DT Stone, BG Thomas, Canad. Met. Quart., 38 (5), 363, (1999).
41. A Yamauchi, T Yamauchi, S Seetharaman, Proc. 6th Intl. Conf. Molten slags, fluxes and
salts, Stockholm, 2000, paper 148.
42. M Hanao, M Kawamoto, ISIJ Intl., 49, 365, (2009).
43. G Volk, K Wunnenberg, Klepzig Fachberichte, 80 (10), 491, (1972).
44. JA Kromhout, ER Dekker, M Kawamoto, R Boom, Ironmaking and Steelmaking, 40, 206,
(2013).
45. BG Thomas, B Ho, G Li, Proc. McLean Symp., Toronto, (1998) (ISS, Warrendale, PA,
1998) p 177.
46. K Nakajima, S Hiraki, T Kanazawa, T Murakami, CAMP-ISIJ, 5, 1221, (1992).
47. R Bommaraju, Proc. 74th Steelmaking Conf. (1991) p. 131.
48. CA Dacker, T Sohlgren, Steel Research Intl., 81, 899, (2010).
49. R Alberny, A Leclercq, D Amoury, M Lahousse, Revue de Metall., 73 (July/Aug), 545,
(1976).
50. HP Narzt, R Martinelli, C Furst, G Xia, H Pressinger, Proc. 3rd Europ. Conf. Continuous
Casting, Madrid, 1998 (UNESID, Madrid, 1998) p 747.
51. S. Ogibayashi, Proc. 85th Steelmaking Conf. (2002) (ISS/AIME Warrendale, PA, 2002)
p 175.
52. MM Wolf, Iron and Steelmaker, 23 (2), 47, (1996).
53. EY Ko, J Choi, JY Park, I Sohn, Met. Mater. Intl. 20, 141 and 1103 (2014).
54. M Suzuki, S Miyahara, T Kitagawa, S Uchida, K Okimoto, Tetsu-to- Hagane, 78, 113,
(1992).
55. WL Wang, AW Cramb, ISIJ Intl., 45, 1864, ( 2005).
56. P Grieveson, S Bagha, N Machingawuta, K Liddell, KC Mills, Ironmaking and Steelmaking,
15, 181, (1988).
57. KC Mills, AB Fox, ISI J Intl., 43, 1479, (2003).
58. GH Wen, S Sridhar, P Tang, X Qi, TQ Liu, ISIJ Intl., 47, 1117 (2007).
59. H Nakada, K Nagata, ISIJ Intl., 46, 441, (2006).
References 105

60. Q Wang, YJ Lu, SP He, KC Mills, Z Li, Ironmaking and Steelmaking, 38, 297, (2011).
61. JN Pontoire, JP Radot, V Delvaux, E Dehaussy, Revue Metall., CIT, 93, 1237, (1996).
62. MJ Lu, JH Chen, DH Tsai, CH Huang, Proc. 78th Steelmaking Conf., 1995, (ISS,
Warrendale, PA, 1995) p. 359.
63. J Nagai, M Ohnishi, T Yamamoto, K Hirayama, S Ohto, T Fujiyama, Tetsu-to- Hagane, 69,
S158, (1983).
64. CS Li, BG Thomas, Proc. ISS Tech Conf. (2003), Indianapolis, (ISS, Warrendale, PA, 2003)
p. 685.
65. BG Thomas, C Ojeda, Proc. ISS Tech. Conf., Indianapolis, 2003, (ISS, Warrendale, PA,
2003) p. 295.
66. A Yamauchi, S Itoyama, Y Kishimoto, H Tozawa, K Sorimachi, ISIJ Intl., 42, 1094, (2002).
67. T Watanabe, CAMP- ISIJ, 14, 153, (2001).
68. A Cristallini, R Langella, A Polk, Proc. 3rd Europ. Conf. Continuous Casting, Madrid, Spain,
1998, (UNESID, Madrid, 1998) p. 589.
69. S Riaz, Private communication on “Reducing horizontal heat flux and increasing mould life
by coatings on mould”. Tata Steel Teesside Research Laboratories, Middlesborough (2008).
70. C Bertolletti, P Courbe, JM Jolivet, PP Naveau, A Oper, E Perrin, C Salaris, AL Spierings, E
Wiesseldinger, Proc. 3rd Europ. Conf Continuous Casting, Madrid, 1998, (UNESID,
Madrid, 1998) p. 65.
71. M M’Hamdi, J Lessoult, E Perrin, JM Jolivet, ISIJ Intl., 36, S197, (1996).
72. K Nakai T Sakashita, M Hashio, M Kawasaki, K Nakajima, Y Sugitani, Tetsu-to-Hagane, 73
(3), 498, (1987).
73. Y Sugitani, M Nakamura, M Okuda, M Kawasaki, K Nakajima, Trans. ISIJ, 25, B91,
(1985).
74. JW Cho, HT Jeong, Met. Mater. Trans. B, 44B, 146, (2013).
75. H Murukami, M Suzuki, T Kitagawa, S Miyahawa, Tetsu-to- Hagane, 78, 105, (1992).
76. T Mochida, S Kohriyama, S Itoyama, T Yamashita, M Suzuki, Y Kishimoto, CAMP-ISIJ,
20, 851, (2007).
77. K Ichikawa, A Morita, Y Kawabe, Shinagawa Tech. Report, 36, 283, (1993).
78. K Watanabe, M Suzuki, K Murakumi, H Kondo, A Miyamoto, T Shiomi, Tetsu-to-Hagane,
83, 115, (1997).
79. M Hanao, M Kawamoto, M Hara, T Murakami, H Kikuchi, K Hanazaki, Tetsu-to-Hagane,
88, 23, (2002).
80. V Guyot, JF Martin, A Ruelle, A d’Anselme, JP Radot, M Bobadilla, JV Lavant, JN
Pontoire, ISIJ Intl., 36, S 227, (1996).
81. GH Wen, P Tang, B Yang, XB Zhu, ISIJ Intl., 52, 1179, (2012).
82. AS Normanton, PN Hewitt, NS Hunter, B Harris, D Scoones, Proc. 4th Europ. Conf.
Continuous Casting, Birmingham, 2002, (IOM, London, 2002) p. 561.
83. M Hanao, M Kawamoto, M Hara, T Murakami, H Kikuchi, A Yamanaka, Proc. 5th
Europ. Conf. Cont, Casting, Nice, France (2005) see also Tetsu-to Hagane, 88 (1), 23,
(2002).
84. CB Shi, MD Seo, JW Cho, SH Kim, Proc. 8th Europ. Conf. Continuous Casting, Graz, 2014,
(Austrian Metals Soc. 2014).
85. KC Mills, ISIJ Intl., 56 (1), 1 and 14, (2016).
86. S Bagha, N Machingawuta, P Grieveson, Proc. 3rd Intl. Conf. Molten slags and fluxes,
Glasgow, 1988, (Inst. of Metals, London, 1988) p. 235.
87. N Machingawuta, S Bagha, P Grieveson, Proc. 74th Steelmaking Conf.,1991, (ISS-AIME,
Warrendale, PA, 1991) p. 163.
88. MS Jenkins, PhD thesis, Heat transfer in the continuous casting mould. Monash Univ.,
Australia, (1999).
89. D Bouchard, FC Hamel, JP Nadeau, S Bellemare, F Dreneau, DA Tremblay, D Simad, Met.
Mater. Trans. B, 32B, 111, (2001).
90. X Yu, GH Wen, P Tang, H Wang, Ironmaking and Steelmaking, 36, 623, (2009).
91. X Qi, GH Wen, P Tang, J Non- Cryst. Solids, 354, (52/54), 5444, (2008).
106 3 Heat Transfer in the Mould and Shell Solidification

92. JF Holzhauzer, KH Spitzer, K Schwerdtfeger, Steel Research, 70, 252, (1999).

93. A Yamauchi, K Sorimachi, T Sakuraya, Proc. 4th Intl. Conf. Molten slags and fluxes, Sendai,
Japan, 1992 (ISIJ, Tokyo, 1992) p. 415.
94. R Scheel, W Korte, Metall. Plant and Technol., 1987 (6), 22, (1987).
95. H Tomono, H Ackermann, W Kurz, W Heinemann, Solidification Technology in the foundry
and Cast house (Metals Soc., Warwick, 1983) Book 273; p. 524.
96. PJ Wray, Met. Trans B, 12B, 167, (1981).
97. DK Stemple, EC Zuluta, MC Flemings, Met. Trans B, 13B, 503, (1982).
98. I Saucedo, Initial Solidification and strand surface quality of pertectic steels. Continuous
Casting. Vol. 9, (ISS-AIME Warrendale, PA, 1997) p. 131.
99. A. Badri, T.T. Natarajan, C.C. Snyder, K.D. Powers, F.J. Mannion, A.W. Cramb: Met.
Mater. Trans. B, 36B, 355, (2005).
100. A. Badri, BT Natarajan, CC Snyder, KD Powers, FJ Byrne, M Byrne, AW Cramb, Met.
Mater. Trans B, 36B, 373, (2005).
101. I Sohn, TT Natarajan, TJ Piccone, KD Powers, CC Snyder, Proc. 6th. Europ Conf.
Continuous Casting, Riccone, Italy, 2008.
102. K Nagata, Proc. 2nd Intl. Symp. Metallurgical slags and fluxes, Lake Tahoe, NV, 1984, ed.
HA Fine and DR Gaskell, (Met. Soc. AIME, Warrendale, PA. 1984) p. 875.
103. R Taylor, KC Mills, Ironmaking and Steelmaking, 15, 187, (1988).
104. H Ohta, K Masuda, K Watanabe, K Nakajima, H Shibata, Y Waseda, Tetsu to Hagane, 80
(6), 33, (1994).
105. Y Waseda, K Masuda, K Watanabe, H Shibata, H Ohta, K Nakajima, High Temp. Materials
& Processes, 13, 267, (1994).
106. M Gonerup, M Hayashi, CA Dacker, S Seetharaman, Proc. 7th Intl. Conf. Molten slags,
fluxes and salts, Capetown, 2004, (SAIMM, Johannesburg, 2004) p. 745.
107. BJ Monaghan, RF Brooks, Ironmaking and Steelmaking, 29, 115, (2002).
108. J Diao, B Xie, J Xiao, C Ji, ISIJ Intl., 49, 1710, (2009).
109. J Chen, J Xie, F He, E Wan, Z Xu, C Yang, G Liu, Ironmaking and Steelmaking, 42, 128,
(2015).
110. R Endo, Y Kono, Y Kobayashi, M Susa, S Mineta, H Yamamura, Tetsu- to Hagane, 100,
571, (2014).
111. S Sridhar; KC Mills; ODC Afrange, HP Lorz, R Carli, Ironmaking and Steelmaking, 27,
238, (2000).
112. H Nakada, M Susa, Y Seko, M Hayashi, K Nagata, ISIJ Intl., 48, 446, (2008).
113. S Ogibayashi, Proc. 85th Steelmaking Conf. 2002, (ISS, Warrendale, PA, 2002) p. 175.
114. S Ogibayashi, CAMP-ISIJ, 18, 126, (2003).
115. R Saraswat, DM Maijer, PD Lee, KC Mills, ISIJ Intl., 47. 96, (2007).
116. BG Thomas, Proc. McLean Symp., Toronto, 1998, (ISS, Warrendale, PA) p. 177.
117. Y Meng, BG Thomas, Met. Mater. Trans., B, 34B, 681 and 707, (2003).
118. BG Thomas, Proc. ISS Tech Conf., 2003, (ISS, Warrendale, PA, 2003) p. 295.
119. C Li, BG Thomas, Met. Mater. Trans. B, 35B, 1151, (2004).
120. KC Mills, PE Ramirez- Lopez, PD Lee, High Temp. Mater, and Processes, 31, 221, (2012).
121. M Hanao, K Kawamoto, ISIJ Intl. 48, 180, (2008).
122. A Suzuki, T Suzuki, Y Nozaki, Tetsu-to Hagane. 55, S 110, (1969) and 56, S 427, (1970).
123. A Suzuki, T Suzuki, Y Nagaoka and Y Iwata: J Jap. Inst. Metals, 32, 1301, (1968).
124. DR Poirier, EJ Poirier, Heat transfer fundamentals for metal casting. (TMS, Warrendale,
PA, 1994).
125. M Suzuki, S Miyahara, T Kitagawa, S Uchida, K Okimoto, Tetsu-to- Hagane, 78, 113,
(1992).
126. N Nagaoka, K Iwamoto, G Usui, S Nemota, Y Ohkawa, Tetsu- to- Hagane, 55, S 109,
(1969).
127. S Okano, T Nishimura, H Ooi, T Chino, Tetsu- to- Hagane, 61, 2982, (1975).
128. T Mori, K Ayata, M Fujimaki, T Soejima, M Kawahara, Tetsu- to- Hagane, 62, S 132,
(1976).
References 107

129. S Fujii, T Ohashi, Tetsu- to- Hagane, 64, S 646, (1978).

130. K Narita, T Mori, K Ayata, J Miyazaki, M Fujimaki, T Shiomi, Tetsu- to- Hagane, 64, S
659, (1978).
131. H Nakato, Y Habu, H Kitaoka, K Kinoshita, T Emi, Trans. ISIJ, 21, B 393, (1981).
132. JA Moore, C Cimeno, S Diehl, RJ Phillips, D Piwinski, Proc. 79th Steelmaking Conf., 1996,
(ISS, Warrendale, PA, 1996) p. 259.
133. TJ Billany AS Normanton, KC Mills, P Grieveson, Ironmaking and Steelmaking, 18, 403,
(1991).
134. C Real, R Miranda, G Vulchis, M Barron, L Hoyes, J Gonzalez ISIJ Intl., 46, 1183, (2006).
135. E van Vliet, DW van der Plas; SP Carless, A A Kamperman, AE Westendorp, Proc. 7th
Europ. Conf. Cont. Casting, Dusseldorf, (2011) Session 4.
136. PD Lee, PE Ramirez-Lopez, KC Mills, B Santillana, Ironmaking and Steelmaking, 39 (4),
244, (2012).
137. RJ O’ Malley, J Neal, Proc. Conf. on new developments in Metall. Proc. Technol.,
Dusseldorf,1999, METEC Congress, p. 73.
138. H Tai, M Morashita, T Miyake, Proc. 3rd Europ. Conf. Continuous Casting, Madrid, 1998,
(UNESID, Madrid,1998) p. 447.
139. T Matsumiya, ISIJ Intl., 46, 1800, (2006).
140. SN Singh, KE Blazek, J Metals, 26 (10), 17, (1974).
141. M Suzuki, JW Cho, H Sato, H Shibata, T Emi, Proc 81st Steelmaking Conf., 1998, (ISS,
Warrendale, PA., 1998) p. 165.
142. S Ogibayashi, K Yamaguchi, T Mukat, T Takahashi, Y Mimura, K Koyama. Y Nagano, T
Nagano., Nippon Steel Technical Report, 34, 1, (1987).
143. J Kromhout, PhD Thesis, “Mould powders for the high speed continuous casting of steel”.
Univ. of Delft, (2011) p. 165.
144. C-A Dacker, P Andersson, C Eggertson, The evaluation of the mould slag film during
continuous casting of steel, Proc. ISIJ-VDEh-Jernkontoret Joint Symposium, Osaka (2013),
p. 182.
145. MR Ozgu, B Kocatulum, Iron and Steelmaker, 21 (5), 77, (1994).
146. F Mantovani, S Spagnul, M Padovan, A Bianco, Proc 8th Europ. Conf Continuous casting,
Graz, 2014 (Austrian Metals Soc., 2014).
147. T Lamp, M Tamminga, H Kochner, S Schiewe, D Kirsch, Proc. 8th Europ. Conf. Continuous
Casting, Graz, 2014 (Austrian Metals Soc., 2014).
148. NS Hunter, JD Madill, PN Hewitt, AS Normanton, Proc. 3rd Europ. Conf. Continuous
casting, Madrid, 1998 (UNESID, Madrid, 1998) p. 289.
149. R Saraswat, PhD Thesis, “Modelling the effect of mould flux on steel shell formation during
continuous casting”, Imperial College, London (2006).
150. R Koldewijn Unpublished Corus Internal Rept (2007) cited in KC Mills, J Kromhout, A
Hamoen, R Boom, Proc. of Admet Conf., Dnipropetrovsk, Ukraine, 2007, (Natl. Metall.
Acad. Ukr., Dnipropetrovsk,2007) vol. 2, p. 174.
151. H Take, H Osanai, J Hasanuma, T Yamamoto, H Bada, H Tozaka, Proc.. Conf. Quality
Improvement Technol for ordinary steel in Iron and steelmaking process, Bangkok, 1994
Session 3 Paper 1.
152. K Schwerdtfeger, “Giessen und Erstarren von Stahl III”. Final Report EUR 8569 (1981)
Research contract to ECSC, 7210. CA/112 9 (Europ. Comm. Sci. & Tech. Publ.,
Luxembourg, 1981).
153. T Sakuraya, T Emi, T Imai, K Emoto, M Kodama, Tetsu-to- Hagane, 67, 1220, (1981).
154. R Shah, JG Williams, G Hecko, Proc. ISS Tech. Conf., 2003, (ISS, Warrendale, PA, 2003)
p. 555.
155. NN Viswanathan, S Sridhar, KC Mills, S Du, Scand. J Metall., 31, 191, (2002).
156. RM McDavid, BG Thomas, Met. Mater. Trans B, 27B, 672, (1996).
157. H Nakato, S Takeuchi, T Fujii, T Nozaki, N Washio, Proc. 74th Steelmaking Conf.,
Washington, DC, 1991 (ISS, Warrendale, PA, 1991) p. 639.
108 3 Heat Transfer in the Mould and Shell Solidification

158. H Nakato, T Sakuraya, T Nozaki, T Emi, H Nikoshawa, Proc. 69th Steelmaking Conf.,
Washington, DC, 1986, (ISS, Warrendale, PA, 1986) p. 137.
159. KH Spitzer, K Schwerdtfeger: Investigation of the isolating properties and melting
behaviour of mould powders for continuous casting. Tech Univ. Clausthal Report 7249 (Dec
1989).
160. S Diehl, JA Moore, RJ Phillips, Proc. 78th Steelmaking Conf., Nashville, TN,1995, ( ISS,
Warrendale, PA, 1986) p. 351.
161. Y Matsushita, T Takahashi, Kobe Research & Development, 43 (2), 123 (1993).
162. M Supradist, AW Cramb, K Schwerdtfeger, ISIJ Intl., 44, 817, (2004).
163. MB Goldschmidt, JC Gonzalez, EN Dvorkin, Ironmaking and Steelmaking 20, 379, (1993).
164. M Kawamoto, K Nakajima, CAMP-ISIJ, 5, 1297, (1992).
165. F Neumann, J Neal, MA Pedroza, AH Castiliejos E., FA Acosta: Proc.79th. Steelmaking
Conf., 1996, (ISS, Warrendale, PA, 1996) p. 249.
166. JJ Macho, G Hecko, B Golinmowski, M Frazee: Development of a new generation of no free
carbon continuous casting fluxes. Preprints 33rd McMaster Symp. On Iron and Steelmaking,
Hamilton, Ont., Canada, (2005) (McMaster Univ. Hamilton, CN, 2005) p. 131.
167. P Andersson, Ironmaking and Steelmaking 42, 6 (2015), p. 465.
168. K Pericleous, private communication, University of Greenwich, UK, (2009).
169. PE Ramirez- Lopez, PN Jalali, J Bjorkvall, U Sjostrom, C Nilsson, Proc. 8th Europ. Conf
Continuous Casting, Graz, Austria, 2014 (ASMET, Vienna, 2014).
170. T Mallaband, Unpublished results from Metallurgica, UK, cited in A Fox: PhD Thesis, Dept.
of Materials, Imperial College, London (2003).
Chapter 4
How to Manipulate Slag Behaviour
in the Mould

Abstract The mould slag plays a key part in the continuous casting process. The
mould slag carries out a series of tasks and it is essential that the slags perform each
of these tasks efficiently. This chapter looks at these tasks and analyses the key
factors affecting performance and suggests ways in which the slag can be manip-
ulated to perform efficiently. The following processes are examined: (1) Thermal
insulation of the powder bed and vertical heat transfer (2) Melting rate of the mould
powder (3) Formation of a slag pool (4) Control of powder consumption and
lubrication of the shell (5) Control of Horizontal heat transfer and the effect of the
thickness of the solid slag film (6) Control of crystalline and glass phases in the slag
film and (7) Control of both horizontal and vertical heat transfer in order to delay
solidification of the shell. The proposed actions are summarised in a table.

Symbols, Abbreviations and Units

D diameter (m)
d depth, thickness (m)
fcrys fraction crystalline
k Thermal conductivity (Wm−1K−1)
QMR Melting Rate (kg min−1 or kgs−1)
Qs Powder consumption (kgslag m−2)
q Heat flux density (Wm−2)
Rth Thermal resistance (m2KW−1)
R* (Surface area/volume) of mould (m−1)
T Temperature (oC)
t thickness of mould (m) or time (s)
Vc Casting speed (m min−1)
w width of mould (m)
AC Alternating current
DC Direct current
EMBr Electro-magnetic braking
EMC Electro-magnetic casting
SEN Submerged entry nozzle

© Springer International Publishing AG 2017 109

K.C. Mills and C.-Å. Däcker, The Casting Powders Book,
DOI 10.1007/978-3-319-53616-3_4
110 4 How to Manipulate Slag Behaviour in the Mould

Subscripts and Superscripts

gran granule
hor horizontal
lat lattice
pool slag pool
powd powder (layer or pulverised)
req required
sint sintered layer
slag slag cf. powder
vert vertical

Chemical Formula for Slags and Minerals

A Al2O3
B B2O3
C CaO
F FeO
Fl CaF2
K K2O
L Li2O
M MgO
Mn MnO
N Na2O
S SiO2
T TiO2

4.1 Introduction

There are many factors affecting the performance of the mould powder in the
continuous casting mould. Furthermore, these factors tend to be interactive. When
certain defects occur in continuous casting it is necessary to take remedial action
(e.g. if the steel product has a large inclusion content, the problem can be partially
solved by reducing the length of the steel meniscus and this can be achieved by
reducing the vertical heat flux). The necessary, remedial actions, such as the
reduction of vertical heat flux, are discussed in this chapter and the ways in which
the mould powder can be manipulated to achieve this and other targets, are
described. Some of these topics have been discussed in other chapters; the aim here
has been to place all remedial actions together in one location. The possible
remedial actions are summarised in a single table at the end of the chapter to
provide the reader with quick access to the different possible remedial actions which
are available. In this approach, only the immediate effects of the remedial action are
described, the effects of collateral interactions are largely ignored.
4.1 Introduction 111

The various topics are arranged in terms of their location in the mould, starting at
the top of the mould and working downwards.

4.2 Vertical Heat Flux and Thermal Insulation of Bed

These two topics are intimately linked since the thermal insulation supplied by the
powder bed is used to control the vertical heat flux. The resistance to heat flow is
that provided by the slag pool, the sinter layer and the powder bed (Fig. 4.1).

4.2.1 Vertical Heat Flux

There are several mechanisms involved in the vertical heat transfer. In the liquid
slag pool, heat transfer occurs by (i) lattice conduction (ii) thermal convection and
(iii) radiation conduction; the latter two mechanisms dominate in the liquid pool. In
the sinter and powder layers, the principal mechanisms are thermal conduction and
gaseous convection.
The thermal insulation provided by the mould powder can be considered as that
provided by the three layers (powder, sinter and liquid) of thickness, d. The vertical
heat flux density (qvert) from the surface of the steel to the powder/air interface
(Fig. 4.1) can be calculated by Eq. 4.1 (assuming 1-dim heat transfer) where
dbed = (dpool + dsint + dpowd) and kbed is the mean value for the three layer system
and A = area = width
thickness of the mould.

Fig. 4.1 Schematic drawing showing the various slag layers formed in the powder bed and the
slag film; Note (i) glassy phase and liquid shown in white and crystalline phase as grey (ii) high
cooling rates promote glass formation and so glass forms initially on the mould side of slag film
but cuspidine is also precipitated at high temperature on mould side and, subsequently, crystal
growth occurs in the high temperature regions of the slag film, over time
112 4 How to Manipulate Slag Behaviour in the Mould

qvert ¼ kbed A Tsteel=slag  Tpowd=air =dbed ð4:1Þ

The thermal resistance, Rth = (dlayer/klayer) of the three layers in the bed (Eq. 4.2)
can be obtained by assuming a series resistance for Ohm’s Law (in a similar way to
that applied to the slag film [1–3]) and where k for the various layers refers to the
effective thermal conductivity (i.e. containing contributions from conduction,
convection and sometimes, radiation). The thermal resistance for each layer
increases with increasing thickness (d) and decreasing effective thermal conduc-
tivity, k.

Rth ¼ ðd=kÞpool þ ðd=k Þsint þ ðd=k Þpowd ð4:2Þ

The depth of the slag pool and the thicknesses of the sintered and powder layers
are affected by the differences between the heat entering and the heat leaving that
layer (Eq. 4.3). The depth of the slag pool decreases with decreasing qin vert, as
witnessed by the lower pool depths associated with the application of EMBr. The
depth of any layer increases with increasing, Dqvert; thus, the depth of the layer
increases with increasing qin out
vert and decreasing qvert

Dqvert ¼ qin
vert  qvert :
out
ð4:3Þ

In the case of the sintered and powder layer, porosity will have conflicting effects
on the vertical heat flux since increased porosity enhances gaseous convection and
decreases lattice thermal conduction.

4.2.1.1 Importance of Vertical Heat Flux

Control of the vertical heat flux is essential in order to ensure that:

(i) The steel surface does not solidify (due to an excessive qvert).
(ii) The slag pool has sufficient depth to (a) cover the surface waves formed in
the steel meniscus (b) provide good slag consumption and (c) avoid
C-pick-up by the shell [4–6].
(iii) The solid shell/meniscus is not excessively long since a long meniscus can
lead to (a) inclusion and bubble entrapment (b) deep oscillation marks (and
transverse cracks) [7].
4.2 Vertical Heat Flux and Thermal Insulation of Bed 113

4.2.1.2 Control of Vertical Heat Flux

The vertical heat flux is largely controlled through the thermal insulation of the bed.
It can be seen from Eq. 4.2 that thermal insulation is largely determined by the
thickness and effective thermal conductivity of the bed.
The effective thermal conductivity of the liquid pool is determined by the
contribution from convection (some models assume keff = N klat where N is given
values between 10 and 100 but usually optimises at ca. 15 [8]). The convective
contribution will decrease with increasing viscosity of the slag pool and radiation
contributions will decrease with additions of transition metal oxides [9, 10].
Values of keff for the solid layers contain contributions from the lattice con-
duction and gaseous convection; the latter is affected by the permeability of the
layer and so is affected by the packing density (or bulk density, qbulk, the powder
size (Dpowd)) and the thermal gradient across the layer. The results of thermal
insulation tests [11] suggest that gaseous convection is the dominant mechanism in
the solid layers (see Sects. 4.2.3 and 9.4.5.8); thus, qvert decreases as both per-
meability and Dpowd decrease and as qbulk increases.
Exothermic reactions cause local increases in bed temperature, which reduce the
temperature gradient and thereby reduce qvert (i.e. improve the thermal insulation)
and increase the depth of the molten slag pool. Endothermic reactions, such as
carbonate decomposition, will increase the heat losses by (a) decreasing the local
temperature gradient and thereby increasing the thermal gradient and (b) creating
more porosity and (c) by adding to the gaseous convection.
The vertical heat flux is also affected by the following casting conditions:
Steel throughput and Casting speed
Increases in steel throughput are manifested as casting speed increases which result
in an increase in the heat flux density (qvert). Since both the enthalpy demand (DH,
defined in Eq. 4.4) and cross-sectional area of the mould remain constant, the ratio
(qvert. A/ DH) must increase and so qvert must increase, in response to an increase in
casting speed.
Steel Temperature
It can be seen from Eq. 4.1 that the heat flux is dependent upon the temperature
gradient (Tsteel/slag − Tpowd/air). Thus, a decrease in steel temperature will reduce this
gradient and hence, the heat flux. Steel liquidus temperatures vary from (ca. 1510 °C
for LC steels) to 1440 °C (for austenitic stainless steels); values of qvert also increase
with increasing superheat. Good thermal insulation is essential when casting steels
with low Tliq or when casting at low speeds to ensure that that the steel meniscus
does not freeze.
Metal Flow
The vertical heat flux is affected by both the flow pattern and the turbulence
developed in the metal flow. Freezing of the steel meniscus is more common with a
single roll pattern than with a double roll. The values of qvert increase with
114 4 How to Manipulate Slag Behaviour in the Mould

increasing turbulence in the metal flow, since heat transfer for turbulent flow is
more efficient than that for laminar flow.
Electromagnetic Braking (EMBr)
Electromagnetic braking is used in thin slab- and high-speed casting to suppress
turbulent flow in the molten metal; it has been calculated to cause a 30% reduction
in qvert in thin slab-casting [12] due to the fact that laminar flow is a less-efficient
form of heat transfer than turbulent flow. Thus, the application of EMBr would be
expected to reduce both qvert and the melting rate.
Ar Flow Rate
The Ar flow contributes to the vertical heat flux by gaseous convection; it increases
the vertical heat flux leaving the metal but also increases the heat flux leaving the
pool (or that leaving the sinter layer) due to increased gaseous conduction. Argon
flow rate has also been reported to alter flow patterns in the metal phase when it
exceeds a critical value (ca 5 L min−1) [13, 14].

4.2.1.3 Ways of Controlling the Vertical Heat Flux

The vertical heat flux leaving the metal (and entering the slag pool) can be
decreased by:
• Decreasing the casting speed and superheat.
• Decreasing the Ar flow velocity.
• Using EMBr to reduce turbulence in the metal flow.
The vertical heat flux leaving the slag pool can also be reduced by improving the
thermal insulation of the bed (see Sect. 4.2.2.2) by:
• Increasing the depth of the bed.
• Using granules of smaller diameters to increase packing density and reduce
gaseous permeability.
• Adding exothermic agents to the powder to reduce the thermal gradients for qout
and by minimising the carbonate content of the powder.

4.2.2 Thermal Insulation of the Bed

It can be seen from Eq. 4.2 that the thermal resistance (or thermal insulation) of
each layer increases as the ratio (d/k) increases. It follows that thermal insulation
will increase if the depths of the powder and sintered layers are increased and the
values of the thermal conductivity are low. The only actions open for the liquid pool
are to reduce the convection by using a high-viscosity slag or to reduce the radiation
by adding FeO or MnO to the slag.
4.2 Vertical Heat Flux and Thermal Insulation of Bed 115

In order to heat the mould powder from room temperature to the steel temper-
ature, the endothermic energy must be sufficient to provide for the following:
– To heat the slag components from 25 °C to the steel temperature (DHendo).
– The enthalpy of fusion of the slag formed (DHfus).
– The heat of vaporisation of the moisture (DHvap).
– The enthalpy of sublimation for the carbonates (DHsub).
– To heat any O2 and N2 present in the bed from 25 °C to the temperature of
reaction (DHgas).
However, the following exothermic reactions (DHexo) reduce the energy budget:
– The oxidation of the free carbon to form CO(g) and CO2(g).
– The oxidation of any exothermic agents (e.g. Ca, Si, Fe) present.
Thus, the energy demand (DH) is given by:

DH ¼ DHendo þ DH fus þ DH vap þ DH sub þ DHgas  DHexo : ð4:4Þ

Once the powder bed has achieved steady state, the vertical heat flux (qvert) must
meet this energy demand (denoted DH) and thus, DH = qvert . A where
A = cross-sectional area of the mould. The ratio (qvert . A/DH) has units of kg s−1,
i.e. identical to that of the melting rate (QMR).
The thicknesses of the various layers in the powder bed in Fig. 4.1 represent the
balance between the required energy and the vertical heat flux at steady state.
Ar blowing will tend to increase the gaseous conduction and thereby, increase
the heat flow into the sinter layer (qinvert) but it will also increase the heat loss from
the sinter to the powder layer (qout
vert).

4.2.2.1 Importance of Thermal Insulation

See that for vertical heat transfer–Sect. 4.2.1.1.

4.2.2.2 The Effects of Powder Characteristics on Thermal Insulation

In addition to the casting conditions, the vertical heat flux is also affected by the
powder bed characteristics. Thus, the heat flux density leaving the molten slag pool
is reduced by increasing the thermal insulation of the sinter and powder layers. The
thermal insulation is improved (i.e. qvert decreases) by the actions summarised
above in Sect. 4.2.1.3, namely, (i): increased bed depth [15] (ii) using smaller
particles with lower permeability (iii) using powders containing exothermic agents
(iv) minimising the carbonate content since their decomposition is endothermic and
(v) reduce convection and radiation in the pool by increasing the viscosity or the
FeO, MnO contents, respectively.
116 4 How to Manipulate Slag Behaviour in the Mould

4.2.3 Measurements of Thermal Insulation of Powders

Measurements methods tend to fall into two classes, namely:

(i) Measurements of the thermal conductivity of powders using the transient hot
wire (THW) [7, 16–20] or the laser pulse methods [21] (See Sect. 9.4.5.4).
(ii) Hot plate test (HPT) measurements [11, 22–24], where the temperatures are
measured at different positions in the powder bed lying between the hot plate
and powder surface (See Sect. 9.8.4).
The values of the effective thermal conductivity (keff) obtained in the two
experiments show large differences [18] (e.g. keff (Wm−1 K−1) values of 0.1–0.2
(in THW studies) cf. 1–1.3 (for HPT) [11]. These differences are due to the gaseous
convection contribution which is large in the HPT studies and much smaller in
THW and LP (Laser Pulse) studies where the temperature gradient across the
specimen is small (in contrast to HPT where it is ca. 1000 °C. The HPT method
provides a closer simulation of keff for the bed but the THW studies give a more
accurate value for klat of the powder (See Sect. 9.4.5.8).
Increased bulk density of the sample results in:
(i) An increase in kpowd = (0.010 + 1.69
10−4 qbulk) [18].
(ii) A decrease in gaseous convection; the gaseous conduction appears to domi-
nate. Bulk density increases with decreasing powder size (Dpowd).

4.2.4 Ways of Improving the Thermal Insulation of the Bed

The various ways of improving thermal insulation are identical to those given for
reducing vertical heat flux in Sect. 4.2.1.3.

4.3 Melting Rate of the Powder (QMR)

The melting rate is an important factor since:

(i) The melting rate should match the required powder consumption (Qreq
s ) to
ensure that good lubrication is supplied to the newly formed shell
(Qreq
s  QMR).
(ii) The melting rate also affects the slag pool depth.
The melting rate (QMR) is dependent upon the following factors:
• Properties of the mould powder.
• Vertical heat flux, which, in turn, is affected by a variety of factors including,
casting speed, steel temperature, electromagnetic braking and Argon flow rates.
4.3 Melting Rate of the Powder (QMR) 117

A number of tests have been developed to provide a value, or a ranking, of the

melting rate for various mould powders. These are listed below:
(i) Combustion boat tests in which 1.5 g of the mould powder is placed in a
combustion boat with the flat-end cut off to provide easy viewing of the
sample. The sample is placed in a muffle furnace at a known temperature; the
time taken for the sample to melt is inversely related to the melting rate.
(ii) Molten slag drip tests [25–29] using the apparatus shown in Fig. 4.2a, where
the graphite block is heated to 1500 °C. The slag melts and drips through the
graphite block and then is collected and weighed at different time intervals.
The melting rate results derived may be affected by the fluidity (1/η) of the
slag.
(iii) Vitrification rate tests [30] in which 25 g of the mould powder is placed in a
graphite crucible (30 mm
40 mm) and heated with a uni-directional heat
source for 7 min and the amount of powder melted determined. The mea-
sured rate is denoted the vitrification rate [30].
(iv) Sumitomo tests [31] are carried out using 1 tonne of steel (Fig. 4.2b); the
steel surface is divided by refractory boards to allow different powders to be
tested at the same time. A known weight of powder is added and the amount
of slag formed in a certain time is determined using dip tests with steel and
Cu wires.
(v) Pradhan test [29] is a small scale, Sumitimo test, where a mould powder is
held within a refractory ring on the surface of the molten steel and the
refractory ring is oscillated to allow the effect of mould oscillation to be
explored.
(vi) Däcker test [32] in which a constant depth (20 mm) of powder is maintained
in a graphite crucible, the latter contains machined slots in the base to allow
the liquid slag to drip off. The crucible base is maintained at 1500 °C and a
regulated flow of air is directed on to the powder surface to aid the

Fig. 4.2 Schematic drawings showing (a) the molten slag drip test [25] (courtesy of
Swerea/Kimab) and (b) the Sumitomo test [31] (permission granted, ISIJ [31])
118 4 How to Manipulate Slag Behaviour in the Mould

combustion of carbon. A siphon system is used to maintain the height of the

slag at 10 mm and the mould powder stock is weighed continuously which
allows the consumption to be measured as a function of time.
(vii) Kromhout method [33] contains two tests. In the softening test, a pressed
cylinder of mould powder is put in a furnace at a known temperature and an
alumina rod is placed on the top surface of the sample; the time taken for a
20% displacement of the rod is measured. In the melting test, the mould
powder is held in a steel crucible with a base made of Al foil. The crucible is
placed on liquid steel at a known, controlled temperature. The time for the
sample to melt was observed using a camera mounted above the crucible.
The two tests were used in combination [33].

4.3.1 The Effect of Mould Powder Properties on Melting

Rate

The effect of the powder properties on the melting rate has been studied by a
number of investigators [34–45].
Several properties of the powder affect the melting rate, namely:
(i) The free carbon content of the flux (Cfree%) with QMR increasing as %Cfree
decreases [35].
(ii) The mean diameter of the carbon particles (DC) with QMR, to a less extent,
increasing as DC increases [35].
(iii) The bulk density (qbulk) with QMR increasing as qbulk decreases which is
probably due to the increasing gaseous conduction resulting from an
increasingly permeable powder bed. (Sect. 4.2.1.2) [35].
(iv) The carbonate content (%CO32−) with QMR increasing as (%CO32−) content
increases which again may be associated with increased gaseous conduction
and the formation of a more permeable powder bed [35].
(v) The liquidus temperature has been reported to have a slight effect on the
melting rate with QMR increasing as Tliq decreases [29].
(vi) The contact angle (h, of slag on carbon) and energy of adhesion have been
reported to affect the melting rate, with QMR increasing as h decreases and
Eadhes increases [37].
Carbon particles are non-wetting to molten slag globules and hinder the
agglomeration of the globules (Fig. 4.3). However, at high temperatures the carbon
particles oxidise and gradually disappear. Smaller carbon particles, obviously,
provide more separation per unit mass of carbon. However, smaller particles also
have a much larger (surface area/ mass) ratio and consequently oxidise more
rapidly. The fact that melting rate decreases with decreasing size, indicates the
separation provided by carbon particles is more important than the faster oxidation
rate. Physical and mathematical modelling of the kinetics of carbon oxidation has
4.3 Melting Rate of the Powder (QMR) 119

Fig. 4.3 Schematic drawing

showing non-wetting carbon
particles (●) preventing the
agglomeration of slag
globules ( )

been carried out [44]. The average size of the globules is a linear function of the
melting rate [28].

4.3.2 The Effect of Casting Conditions on Melting Rate

The principal factor affecting the melting rate is the vertical heat flux (qvert); so any
variable which increases the heat coming into the pool (qin vert) or reduces the heat
leaving (qout
vert) will increase the melting rate. The qvert is increased by:

(i) Increasing casting speed (increases (qin

vert).
(ii) Turbulent heat transfer (cf. laminar heat transfer) increases qin
vert (EMBr
reduces qin
vert ).
(iii) Improved thermal insulation of powder bed (e.g. a deeper bed or more dense
packing reduces qout
vert).

The effect of Ar flow is more uncertain since it increases both qin out
vert and qvert but it
probably enhances the melting rate overall.

4.3.3 Ways of Increasing Melting Rate

The principal ways of increasing the melting rate are:

• To decrease the %Cfree in the mould powder but care must be taken since C is an
exothermic agent and large decreases in %Cfree may lead to freezing of the steel
meniscus.
• To use carbon particles with a larger mean diameter (DC").
• To decrease the bulk density of the mould powder; this can be achieved by
increasing the particle size of the granules (Dgran").
120 4 How to Manipulate Slag Behaviour in the Mould

• To increase the carbonate content of the powder.

• Increase the vertical heat flux (e.g. Increase casting speed but this will cause
other collateral changes).

4.4 Depth of Molten Slag Pool

4.4.1 Molten Slag Pool

The molten slag pool acts as a reservoir of molten slag which is, subsequently,
pumped into the channel between shell and mould during the period of oscillation
cycle when the mould is descending [46].
It should be noted that the slag pool does not have a uniform depth (dpool);
reported values of dpool indicate that [47, 48]: dpool (SEN) > dpool (midway) > dpool
(large face) > dpool (narrow face) (Fig. 4.4). This has been attributed to the flow
separation, which occurs ca. 200 mm from the narrow face and results in a cold
spot [48]. The application of EMBr has been found to produce a more uniform slag

Fig. 4.4 Schematic drawing (a)

showing the depth of the
molten slag pool as a function
of position in the mould [11]
(a) General view of liquid
slag layer (shown in light
grey) and (b) experimental
measurements (●) of slag
pool depth and predicted
values (▬) versus distance
from narrow face [48]
((permission granted,
(a) ISS/AIST [11])
(b) ISS/AIST [48]) (b) 30
Slag pool depth, mm

20

10

0
0 100 200 300 400 500 600 700
Distance from narrow face, mm
4.4 Depth of Molten Slag Pool 121

layer [49] presumably, by reducing the size of the surface waves on the steel
meniscus.
The pool depth, at steady state, represents the balance between (i) the heat
coming in and leaving the liquid zone and (ii) the fluid flow in and out of the pool
(i.e. the melting rate and slag consumption, respectively); thus an increase in
melting rate or a decrease in Qs will result in a deeper pool.
Several methods have been proposed for measuring the depth of the slag pool
(dpool). The following methods have been used.

4.4.1.1 Dip Tests

These usually contain two or three wires (of steel, Cu and Al which are poked
through the bed into the steel for a few seconds and then removed (Fig. 4.5a). The
melting point of Cu (1085 °C) is close to the Tliq of most mould slags, so the
distance between the Cu and steel wires represents the depth of the slag pool. When
Al wire is used its height represents the 660 °C temperature contour.

(a) (b)

660o 1080o 1500o

(c)

Fig. 4.5 (a) Schematic diagram and photographs showing (a) before and after the dip test
(b) plate-dip test and (c) photographs of plates removed from the mould, showing slag attached to
the plate; note that the steel surface is not flat (courtesy of F Shahbazian, Kimab, Swerea [119])
122 4 How to Manipulate Slag Behaviour in the Mould

4.4.1.2 Plate-Dip Tests

In these tests a thin steel plate (ca. 1 mm thick) is pushed through the powder bed
and held for ca. 1 s and then removed. The slag sticks to the steel plate (see
Fig. 4.5b) and its depth can be measured (Fig. 4.5b).

4.4.1.3 Nakamori Test

A device has been reported that uses two eddy current sensors, the low frequency
sensor reveals the position of the steel surface and the high frequency sensor locates
the upper surface of the slag pool [50]. More recently, the slag pool thickness has
been determined using two sensors, an EM sensor to detect the steel level and a
sensor to detect radioactive signals for the slag level [51].

4.4.2 Importance of Depth of Molten Slag Pool

Manipulation of slag pool depth is needed to:

(i) Ensure good powder consumption since it is known that a shallow slag
pool leads to poor powder consumption (a value of dpool = 10 mm is often
cited as being necessary but >20 mm is cited as being necessary for
high-speed casting).
(ii) Minimise carbon pick-up by the shell and steel surface in order to separate
the layer of carbon particles floating at the top of the pool from the solidi-
fying shell.
(iii) Dissolve inclusions, since a large slag volume is needed when casting
(a) steels containing Ti which form Ti(CN) or Perovskite (CaOTiO2) which
has low saturated solubility in slag and (b) High-Al steels where large
amounts of Al2O3 are produced.

4.4.3 Factors Affecting Slag Pool Depth

The principal factors affecting the depth of the molten pool (dpool) are the
magnitudes of the heat flux and the fluid flow entering and leaving the liquid pool.
The heat transfer involves both the thermal insulation properties of the bed and the
casting conditions. The fluid flow involves the melting rate (QMR, coming in) and
the powder consumption (flowing out). Obviously, the depth of the slag pool will
tend to increase as the liquidus temperature of the mould slag deceases [48].
Consider the molten slag pool layer:
4.4 Depth of Molten Slag Pool 123

Fig. 4.6 Schematic drawings showing the effect of powder depth on the thicknesses of various
layers in the powder bed for two mould powders; black Powder; white Sintered; dotted layer
molten; larger dots indicate mushy zone (permission granted, ISIJ [15]; re-drawn Swerea/Kimab)

(i) heat is supplied to the pool by the metal (qinvert)

(ii) heat is lost principally through the sinter layer and powder bed (qout
vert).

Obviously, an increase in qin vert will result in a higher melting rate (see
Sect. 4.2.2.2) which will result in an increase in the slag pool depth (dpool). The
application of EMBr has been calculated to cause a 30% decrease in qin vert, [12]
which, thus, results in a decrease in pool depth (dpool).
An improvement in thermal insulation will cause a decrease in qout vert and this too
will result in an increase in dpool. This can be clearly seen in Fig. 4.6 [18] where the
improved thermal insulation, resulting from a deeper powder bed [52,48], causes an
increase in dpool. In addition, it can be seen from Fig. 4.6 that improving the thermal
insulation of the bed also increases the thicknesses of the sinter and powder layers.
A similar argument can be applied to the sinter and powder layers where qin vert refers
to the layer below and qout vert to the layer above.
Several different types of mathematical models have been developed to predict
the depth of the molten pool [37, 48, 52, 53]. One model considers the bed to be a
packed bed and solves the heat and mass transfer of the gaseous and condensed
phases [52] to calculate the temperature contours in the bed; the pool depth can be
calculated from the these contours. A coupled heat and fluid flow model has been
used to determine the depth of the pool in other studies [48, 53]. A mathematical
model for the slag pool depth formed in thin slab-casting has been reported [54].
124 4 How to Manipulate Slag Behaviour in the Mould

Fig. 4.7 Time-transients for the depth of slag pool and the casting speed, showing the effects of
(a) an increase in Vc [47] and (b) a decrease in casting speed (solid line) and dpool = dashed line
[55] (permission granted, (a) Europ. Sci. Tech. Publ. [47] and (b) ISS/AIST [55])

4.4.4 The Effect of Casting Speed and Oscillation

Characteristics

When the casting speed is increased abruptly, it causes increases in heat flux
density, i.e. for both qvert and qhor. An increased qvert needs time to work its way
through to the sinter and powder layers; ultimately, an increased melting rate results
in an increase in dpool. This process takes time and there is a time lag (of 2–5 min
(see Fig. 4.7) before it responds [47, 55]. The same applies to qhor and the changes
in slag film thickness. When steady state has been reached an increase in casting
speed results in a deeper slag pool [38, 47, 56].
It can be seen from Fig. 4.7 that as the casting speed increases the slag pool
depth initially decreases in response to the higher demand for slag infiltration [47,
55]. However, after a while the increased qhor, results, sequentially, in a thinner
solid slag layer (ds), thicker liquid layer (dl) and a higher melting rate. Thus, dpool,
after its initial fall, starts to increase and reaches an enhanced value when
steady-state conditions are attained. Similarly, a decrease in casting speed results in
an initial increase in dpool, which is followed by a decrease in dpool when the
decreased qvert starts to take effect.
It has been found that oscillation caused a ca. 20% increase in slag pool depth
recorded in simulation tests [29].
It has been reported that the pool depth increases with increasing % negative
strip [53, 57].
4.4 Depth of Molten Slag Pool 125

4.4.5 The Effect of Thermal Insulation of Bed on Pool

Depth

As mentioned above, the depth of any layer (pool, sinter or powder) will increase if
qin out
vert increases or qvert decreases; improved thermal insulation affects the latter. The
following actions lead to improved thermal insulation:
• Using a deeper powder bed.
• Replacing granules with powder or powders with a smaller diameter (Dpowd#)
which reduces the gaseous conduction.
• Adding exothermic agents to the powder to reduce the temperature gradient in
the sinter layer.
• Using, a slag with a high Tliq and viscosity since these will tend to reduce the
contribution from convection to the vertical heat transfer.

4.4.6 Ways of Increasing the Melting Rate

An increased melting rate will increase the depth of pool; this can be achieved by:
• Reducing the free carbon (%Cfree) but it should be noted that C combustion is
exothermic and a reduction of C would lead to an increase in the temperature
gradient and qout
vert.
• Increasing the size of the C particles (DC").
• Decreasing the bulk density (by increasing the particle size of the granules,
Dgran").
• Increase the carbonate content of the powder.
• Increase the vertical heat flux by increasing the casting speed, the superheat or
the Ar flow rate (providing it remains below 5 L min−1.

4.5 Powder Consumption (Q) and Liquid Film

Thickness (dl)

Molten slag from the slag pool infiltrates into the shell/mould channel during the
downward movement of the mould and the slag rim. The molten slag entering the
channel, subsequently, partially solidifies to give a slag film with a solid layer (ca.
2 mm thick) and a liquid layer (ca. 0.1 mm thick).The liquid slag film moves
downward during the descent of the mould and is the major contributor to the
powder consumption, although some downward movement of the solid slag layer
may also occur [58].
126 4 How to Manipulate Slag Behaviour in the Mould

Most slag films consist of a crystalline and a glass phase. It should be noted that
glass phases form a super-cooled liquid (scl) at temperatures above Tg, the glass
transition temperature. These scl’s will provide some lubrication, despite their high
viscosities, and some glassy, high-viscosity slags are used in billet casting.
A description of the methods used to determine powder consumption are given
in Chap. 2 (Sects. 2.1.2 and 2.5.1, respectively).

4.5.1 Reasons for Controlling Powder Consumption

Powder consumption provides a measure of the lubrication supplied by the molten

slag to the newly formed shell. It can be seen from Eq. 4.5 that the liquid frictional
force applied to the shell (F) decreases with increasing liquid slag film thickness (dl)
and decreasing viscosity (η), where A is the area of shell and Vc and Vm are the
casting speed and mould velocity, respectively.

F ¼ AðVm  Vc Þg=dl ð4:5Þ

Inadequate powder consumption can lead to longitudinal and transverse cracking

and to star cracking in the lower half of the mould.
It should also be noted that not all of the mould powder forms slag and it is the
slag which provides the lubrication; thus it is necessary to differentiate between slag
and powder. There are several powder consumption terms in common use which
are defined below in Eqs. (4.6)–(4.9). The term, Qpowd t is the quantity usually
measured on plant, Qslag
s is the best measure of the lubrication supplied to the shell
and this can be related to the melting rate Qslag
MR via Eq. 4.9. It should be noted that
the liquid friction becomes dependent on η1.5 when empirical relations (e.g.
Qs = 0.55/η0.5Vc) are coupled with Eqs. 4.5 and 4.8.

Qslag
t ¼ f  Qpowd
t ; ð4:6Þ

where

f  ¼ ð 100  %Cfree  %H2 OÞ  fð44=12Þð%Ctotal  %Cfree Þg=100

¼ ð100  %LOIÞ=100 ð4:7Þ


Qslag
s kg m2 ¼ f  7:6 Qpowd
t =R ¼ q dl  2550  dl : ð4:8Þ

1


Qslag
MR kg s ¼ 2ðw þ tÞQs Vc =60 ð4:9Þ

The required powder consumption is that which is necessary to ensure good

lubrication of the shell. Wolf [59, 60] introduced the concept of minimum powder
consumption and recommended values of 0.25 [59] and 0.4 [60] kg m−2,
4.5 Powder Consumption (Q) and Liquid Film Thickness (dl) 127

respectively, for casting round billets and heavy plates. There are no reliable values
for Qmin
s but one possible rule would be Qmin s = 0.5 Qreq
s .

4.5.2 Factors Affecting Powder Consumption

There are a number of factors affecting both the required powder consumption for
the casting conditions and the actual powder consumption achieved in practice; in
practice, powder consumption can be constrained by factors like the free carbon
content of the powder.

4.5.2.1 Mould Dimensions

The slag consumption per tonne of steel is approximately constant at

Qslag
t = 0.48 kg tonne steel−1 [61]. However, the required powder consumption per
m of mould (Qs in kg (flux) m−2) increases with increasing surface area of mould
2

(or decreasing surface area/volume, denoted as R* [62] as shown in Fig. 4.8 and
Eqs. 4.10 and 4.11. It is important that the free carbon content of the casting

0.8
Billets
0.7
Powder Consumption, Q ccorr (kg m 2)

Blooms
Slabs
0.6
Round
0.5 High Speed Billet
[4] -0.04R
: Qs =0.44e
0.4
: Qs=2/(R-5)
Thin Slabs
0.3

0.2

0.1

0
0 10 20 30 40 50
-1
Surface Area to Volume Ratio, R (m )

Fig. 4.8 Powder consumption, Qslag s req as a function of R*, the ratio of (surface area/volume) of the
mould, diamond billets; ● = blooms; D = slabs; ▬ = thin slabs; + = rounds; + =high-speed billets;
solid line Eq. 4.10 [62] dotted line [7] (permission granted, UNESID [62])
128 4 How to Manipulate Slag Behaviour in the Mould

powder should be at a level where it allows the required, powder consumption to be

MR  QMR
attained, i.e. Qslag slag req
.

s req ¼ 2=ðR  5Þ
Qslag ð4:10Þ

R ¼ 2ðw þ tÞ=wt ð4:11Þ

4.5.2.2 Casting Speed (Vc) and Viscosity (η)

The powder consumption, Qs, is affected by several casting variables (as can be
seen from Eqs. 4.12 to 4.16). However, the most influential terms are the casting
speed and the viscosity. Several empirical rules have been proposed for the required
powder consumption involving these two parameters, [59–64] such as those shown
in Eqs. 4.12 and 4.13.

s req ¼ 0:55=g
Qslag  Vc ð4:12Þ
0:5

s req ¼ 0:6=g  Vc
Qpowd ð4:13Þ

The required viscosity (ηslag

req ) tends to decrease as the casting speed is increased
(rough values of ηslag
req can be calculated from the following rules: ηreq = 2/Vc
slag
2
or = (5/Vc )).
As mentioned above, the combination of Eqs. 4.5 and 4.8 give Eq. 4.14,
showing that the frictional forces acting on the shell are proportional to η1.5.

F ¼ AgðVm  Vc Þql g0:5 Vc =0:55 ð4:14Þ

The viscosity of non-Newtonian slags decrease with increasing shear rate. Thus,
they provide a high viscosity in the low-shear rate (meniscus) region (and thereby
reduce slag entrapment) and a lower viscosity in the high-shear rate, region (at the
mouth of the shell/mould channel) which reduces friction acting on the shell.

4.5.2.3 Solidification (or Break) Temperature (Tbr)

If the solidification (or break, Tbr) temperature (Tsol) of the solid slag film is
increased, the thickness of the solid slag film (ds) is increased and the thickness of
the liquid layer (dl) is reduced (ds" and dl# as Tbr (or Tsol)"). Since Qs = qldl and ql
is reasonably constant (at 2550–2600 kg m−3) it would be expected that Qs is
inversely related to Tbr. Equation 4.15 was derived from physical modelling studies
and shows an inverse relationship [64] between Qs and Tsol. Equation 4.16 was
obtained from a statistical analysis of plant data and also indicates a similar inverse
relation between Qs and Tbr.
4.5 Powder Consumption (Q) and Liquid Film Thickness (dl) 129

 
kb s0:4 1 1000Vc
Qs ¼ 1:6 0:5 cos ; ð4:15Þ
Tsol g Vc 2pfs





Qs ¼ 1=gVc0:46 : 1= f 0:49 1=s1:37 1=Tbr3:48 exp28:81 : ð4:16Þ

4.5.2.4 Oscillation Characteristics

There is no agreement on the effect of stroke length (s) on the powder consumption
but a statistical analysis of plant data indicated that Qs decreases with increasing
stroke length. Mathematical models of powder consumption based on Navier–
Stokes indicate that Qs increases with increasing oscillation frequency (f). However,
most plant data show the reverse trend, i.e. Qs decreases as frequency increases (e.g.
Eq. 4.16). It has been suggested that “tide-changes” of the flow in the slag pool are
followed by periods of “confused flow” where little slag infiltration occurs and the
models fail to account for this [46]; an increase in frequency will increase the
number of these periods resulting in less time for slag infiltration.
There is no agreement on the effects of negative and positive strip times (tn, tp).
A recent mathematical model of slag infiltration indicated that it occurred in the
period when the mould was descending [46]; thus a higher tn value might thus be
expected to increase Qs.
Powder consumption increases of 10% have been reported when using
non-sinusoidal oscillation [65–68].

4.5.2.5 Slag Pool Depth and Vertical Heat Flux

It might be expected that powder consumption, Qs, will increase with increasing
slag pool depth (dpool) as slag is pushed into the shell/mould channel during the
descent of the slag rim. However, there is no direct evidence for such a
relationship. It is known, however, that values of dpool and Qs both tend to be low in
thin slab- casting; this provides some circumstantial evidence for the view that Qs
increases with increasing dpool.
The slag pool depth tends to increase if Dqvert = (qin
vert − qvert) increases and thus
out

increasing vertical heat flux might be expected to increase the powder consumption.
The effect of vertical heat transfer on powder consumption is not clear. Increased
qvert will cause increases in melting rate and dpool. However, increased powder
consumption has been reported to increase when EMBr is applied which has been
attributed to the 10 °C increase in meniscus temperature and its effect on slag
viscosity but the reduced qvert will also result in a shallow dpool. If the reduced
viscosity is the cause of enhanced Qs then superheat would be expected to produce
the same effect. This problem needs resolving.
130 4 How to Manipulate Slag Behaviour in the Mould

Fig. 4.9 Powder 0.8

consumption, Qpowd
t as a
function of Ar flow rate
(courtesy of A. B Fox [69])

, kg tonne-1
0.6

powd
0.4

Qt 0.2
0 1 2 3 4 5
-1
Ar flow rate, l min

4.5.2.6 Argon Flow Rates

It has been reported that powder consumption increases with increasing Ar flow rate
(Fig. 4.9) [69]. This is probably associated with the increase of gaseous conduction
in the molten metal phase which will enhance the vertical heat flux. This will
increase the heat entering the slag pool (qin vert). However, the Ar flow rate will also
increase the heat leaving the molten slag pool (qout vert) to the sinter layer. If
Dqvert = (qin
vert − q out
vert ) increases with increasing Ar flow rate, then the higher heat
flux will result sequentially in a higher melting rate, a deeper pool and increased
powder consumption. However, recent work has shown that the Ar flow interacts
with the metal flow and can alter the metal flow pattern when the Ar flow rate
exceeds 5 L min−1 [13] which could lead to a higher qvert.

4.5.2.7 Electromagnetic Casting (EMC)

It has been reported that the application of EMC results in a 20% increase in powder
consumption which has been attributed to an increase the temperature of ca. 10 °C
at the steel/slag interface and change in meniscus shape [70]. One possible
explanation of the enhanced powder consumption is that the enhanced meniscus
temperature reduces the slag viscosity slightly.

4.5.2.8 Blockage to Slag Infiltration

It has been observed that powder consumption is frequently lower than predicted
when casting steel grades containing Ti (ULC and stainless steels). This is thought
to be due to the formation of TiN or Ti(CN) or Perovskite (CaTiO3). TiN has a low
solubility in the slag pool and CaTiO3 has a high melting point, so a large number
of solid particles are present in the slag pool [71, 72]. These particles agglomerate
4.5 Powder Consumption (Q) and Liquid Film Thickness (dl) 131

through turbulent collisions. If these agglomerates collect in the mouth of

slag/mould channel they will restrict the slag flow in the channel (see Fig. 2.16). It
has also been suggested that Ti(CN) can be formed in the mould [73] through the
reaction of N2 in F-free slags, where CaF2 has been replaced with 6% TiO2 [73, 74].
A similar mechanism probably applies when casting TRIP steels with >1% Al,
where large amounts of Al2O3 are formed which only partially dissolve and,
subsequently, impede the flow of molten slag [75] into the shell/mould channel.
In these cases, the formation of a deep slag pool will be beneficial since it will:
• Dissolve more TiN or Al2O3 resulting in fewer agglomerates to impede the flow
of molten slag into the mould/shell channel.
• Seal off the metal from gases in the sinter layer more efficiently.

4.5.3 Ways of Controlling the Powder Consumption

The powder consumption can be increased by the following:

• Decreasing slag viscosity (η).
• Decreasing the casting speed (Vc).
• To a less extent, by decreasing frequency (f) and stroke length (s).
• Decreasing the Solidification (or Break) temperature (Tbr).
• Increased slag pool depth (dpool) when casting Ti- steels and TRIP steels to
dissolve Ti(CN) or Al2O3, respectively, which tend to block slag infiltration.
• Increasing the Argon flow rate (providing it does not exceed 5 L min−1).
• Use of non-sinusoidal oscillation.
• Use of EMBr and EMC.

4.6 Solid Slag Film and Horizontal Heat Flux

The slag film formed between the shell and the mould is probably initially glassy but
crystallises over time. The mixture of glassy and crystalline phases is essential to
control of both the horizontal heat flux and the provision of some lubrication to the
shell. The horizontal heat flux between shell and mould involves two heat transfer
mechanisms, namely, phonon (or lattice) conduction and radiation conduction (see
Sect. 3.2). The lattice conduction can be treated by assuming the various layers in
the slag film (liquid, crystalline and glass) are viewed as thermal resistances in series.
In addition, there is also an interfacial resistance (RCu/sl) created at the copper/slag
interface as a result of the shrinkage caused by the transformation of glass to crys-
talline phase; this too is treated as a series resistance [1–3]. The radiation conduction
has been accounted for by assuming it to be a parallel resistance [3]. The thermal
resistance of the slag film is calculated by Eq. 4.17.
132 4 How to Manipulate Slag Behaviour in the Mould

Rth ¼ ðd=k Þliq þ ðd=kÞglass þ ðd=k Þcryst þ RCu=sl ð4:17Þ

The glass and crystalline phases are intermixed in the slag film with highest
concentrations of crystals located on the shell (high temperature) side and the glass
on the mould side [76] because high temperatures favour crystal growth.

4.6.1 Reasons for Control of Slag Film Thickness

and Horizontal Heat Flux

It is important to control the horizontal heat transfer across the slag film to ensure
that the newly formed shell has the optimum thickness for the type of steel being
cast and the casting conditions. Thus for the casting of the following steel grades:
(i) In MC (Medium Carbon), peritectic steels, a thin, even shell is needed to
avoid longitudinal and other surface cracking and this is obtained with a low
horizontal heat flux density (qhor).
(ii) In HC (High Carbon) steels the shell is relatively weak and a thick shell is
required to provide mechanical strength; failure to do so can result in sticker
breakouts and hot tears.
(iii) In steels where the primarily solidification results in the formation of ferrite
(e.g. ULC steels (Ultra Low carbon)) the shell is soft and a strong cooling is
needed to create a strong shell to minimise the risk of sticker breakouts.
The horizontal heat flux is also involved in delaying solidification to reduce both
the inclusion content and depths of oscillation marks (Sect. 4.8).

4.6.2 Factors Affecting of Slag Film Thickness

and Horizontal Heat Flux

The horizontal heat flux between the shell and the water-cooled copper mould is
affected by the following factors.

4.6.2.1 Thickness of the Solid Slag Film (ds)

It can be seen from Eq. 4.17 that the thermal resistance of the slag film increases as
the ratio (d/k) increases for both the glass and crystalline phases. Since the thickness
of the solid slag (ds) is composed of both glass and crystalline phases, qhor will
decrease as (k/ds) decreases (qhor# as ds" as k#). The slag film thickness is usually
between 1 and 3 mm [58, 76–79] but a value of ds = 0.3 mm has been reported for
thin slabs [80]. The thickness, ds, can be increased by increasing the break or
4.6 Solid Slag Film and Horizontal Heat Flux 133

solidification temperature (Tbr). In practice, this is usually achieved by increasing

the basicity (C/S) of the powder; frequently, powders with a C/S ratio in the range,
1.1–1.35, are used to cast steels prone to longitudinal and other forms of surface
cracking.

4.6.2.2 The Fraction of Crystalline Phase (fcrys)

The presence of crystalline phases in the slag film has two effects on the horizontal
heat flux:
(i) Crystals in the slag film scatter the IR radiation and thus reduce kR.
(ii) Crystallisation of a glassy slag film increases the interfacial resistance (RCu/sl)
and creates some porosity in the slag film (klat#) due to the shrinkage (since
qcrys > qglass); the interfacial resistance and the associated surface roughness
of the Cu/slag interface increase with increasing fcrys [81].
Although, kcryst  2 kglass, the value of qhor is significantly lower for slag films
containing crystals; this demonstrates the importance of reducing kR. It has been
estimated that the kR contribution to the total heat flux for partially crystalline
mould fluxes is 10–20% [10]. The fraction crystallinity, fcrys, is usually increased by
increasing both basicity (C/S) and the CaF2 content but large pick-ups of Al2O3
when casting high Al steels are also reported to increase fcrys [82].

4.6.2.3 Porosity of the Slag Film

Some porosity of the slag film results from the crystallisation of an initially glassy
slag film [83]; porosity reduces both the thermal conductivity and qhor by lowering
the lattice conductivity and by the scattering of IR radiation by the pores. However,
gas pores can also be formed in the slag film when the mould powder has high
moisture content or there is a water-leak in the process. Sticker breakouts
(accompanied by reduced qhor) have been reported to occur when hydrogen con-
tents in the slag film are high [84–86].

4.6.2.4 Viscosity of the Mould Slag

There is inconsistency in the reported effects of slag viscosity on the thickness of

the slag film (ds), and hence the horizontal heat flux (i.e. qhor) has been reported to
increase with increasing viscosity [87] and also with decreasing viscosity [88, 89].
134 4 How to Manipulate Slag Behaviour in the Mould

4.6.2.5 Casting Speed and Superheat (DT)

An increase in casting speed results in a higher rate of heat flux (Js−1 m−2) but the
total heat loss (Jm−2) is lower because of the shorter residence time. The effect of Vc
is best viewed through its effect on the shell, i.e. an increase in Vc produces a
thinner shell (dexit 0.5
shell = K′ tmould/3.6) where K′ = constant with values of 7.3 for slab
casting and 9.5 for high-speed casting [90]. Thus a high Vc will result in a lower
value for tmould and lead to a thinner shell. The heat flux also increases with
increasing superheat and leads to thinner shell due to the delay in solidification [91].

4.6.2.6 Non-sinusoidal Oscillation

The use of non-sinusoidal oscillation has been reported to result in a 10% decrease
in qhor [92].

4.6.2.7 Taper and Uniformity of Mould Temperature

The heat flux density is not uniform around the mould; qhor is greatest in the regions
near the narrow faces and decreases gradually to the region around the centre of the
wide face (see Fig. 3.10a). The heat transfer is highest near the corners of the
mould. Thus, the shrinkage of the shell is greater near the narrow faces than in
centre of the wide face. The increased shrinkage results in increased slag film
thickness (ds) and, subsequently, results in a decrease in qhor at the narrow faces.
The mould flux ratio, defined in Eq. 4.18 is used to measure the uniformity of the
heat flux in the mould.



Heat Flux Ratio = qnarrow
hor =average qwide
hor ð4:18Þ

Fig. 4.10 The effect of water 18

cooling flow rate on the
thickness of the shell for
two different superheats
16
d shell , mm

(20 °C = blue and 60 °

C = red) during casting of
stainless steel 304 (- - -) and
CS 1026(──) [94]; 14
Vc = 0.9 m min−1.
(permission granted,
ISS/AIST [94] re-drawn)
12
2. 2 2.6 3 3.4 3.8
Water cooling rate, 10 3 litre min-1
4.6 Solid Slag Film and Horizontal Heat Flux 135

The taper must compensate for the shrinkage, which in turn is related to the heat
flux and will change according to the mould slag used. An increase in taper from
1.1 to 1.4% was observed to increase the heat flux ratio from 0.6 to 0.8 [93].

4.6.2.8 The Mould Water-flow Rate

The water-flow rate has a small effect on both qhor and the shell thickness, as can be
seen from Fig. 4.10 [94]. The flow rate must also be high enough to avoid the
formation of steam bubbles which are detrimental to the cooling and can cause
excessive temperatures in the copper mould.

4.6.2.9 Coating Moulds, Grooved Moulds

Copper has a very high-thermal conductivity, so the horizontal heat flux can be
reduced by coating the mould with a metal or material with lower conductivity
(e.g. Nickel) [95]. In Meniscus-free casting (MFC) a refractory is placed at the top
of the mould to reduce qhor and delay solidification to a site further down the
mould [96, 97].
Grooved moulds have also been used to reduce qhor in the meniscus region
[98–103].

4.6.3 Measurement of Horizontal Heat Flux

A number of techniques have been used to study the horizontal heat flux across the
slag film, these are described in Sect. 3.2.4.
In summary; the horizontal heat flux can be reduced by:
• Increasing the slag film thickness (by increasing the Tbr).
• Increasing the fraction of crystalline phase (fcrys) by increasing the C/S ratio and
%CaF2.
• Decreasing the water-flow rate (small effect).
• Reducing the casting speed.
• Applying coatings to the mould and machining grooves in the mould.

4.7 Crystallinity in Slag Film

The slag film formed in the shell/mould channel usually contains a mixture of glass
and crystalline phases. The crystal phase aids the control of the heat transfer and the
glass phase aids the lubrication of the shell. Glass forms a super-cooled liquid (scl)
136 4 How to Manipulate Slag Behaviour in the Mould

at temperatures above Tg (ca. 630 °C) and the scl, despite its high viscosity, pro-
vides some measure of lubrication to the shell. This is a particular issue when
casting high Al steels where the large Al2O3 pick-up results in a fully crystalline
slag which provides little lubrication. Slag films with a high fcrys tend to be fragile
and break along a line of pores (created during the crystallisation process) but
glassy slag films are less prone to fracture [104].
The first slag film formed in the shell/mould channel is probably glassy because
of the high cooling rates involved. However, the slag film crystallises over time
until fcrys in the slag film reaches a steady state.
In conventional, (F–containing) slags, cuspidine (3CaO2SiO2CaF2) is precip-
itated during primary solidification and other phases form during secondary
solidification [83]. The cuspidine is precipitated at high temperature (on the mould
side of the slag film) and secondary precipitation occurs in hotter regions (on shell
side) of the slag film [76] since high temperatures promote crystal growth.
Crystallisation involves two mechanisms, namely, nucleation and growth.
Temperature has different effects on these two processes; Nucleation is promoted by
high undercooling (lower temperatures) whereas growth is enhanced by high
temperatures. Thus, crystals formed at high temperatures (just below Tliq) tend to be
few but large (due to strong growth) whereas many finer crystals are formed at
lower temperature due to the high undercooling and slow growth. Some oxide
particles (e.g. TiO2 and ZrO2) have been reported to aid nucleation [105].
Agitation has been found to affect primary solidification (presumably, due to the
removal of local undercooling) but has little effect on secondary solidification
[106].
The application of an electrical potential to a slag has been reported to change
fcrys in the sample [107, 108]; it was reported that when using DC, increasing
voltage resulted in increases in both fcrys and Dcrys, whereas with AC it caused
increasing fcrys and decreasing Dcrys [107].

4.7.1 Importance of Crystallinity to the Casting Process

The horizontal heat flux from the shell to mould is largely controlled by (i) the
thickness of the slag film (ds) (ii) the fraction crystalline phase (fcrys) and the size of
the crystallites (Dcrys) in the slag film and (iii) the interfacial resistance (RCu/sl) and
porosity formed by the shrinkage of slag during crystallisation. The thickness of the
solid slag is a linear function of the break (Tbr, or solidification) temperature and the
horizontal heat transfer can be controlled through Tbr. The principal issues are
(i) controlling fcrys and (ii) replacing cuspidine (C3S2F1) with a suitable alternative,
crystalline phase. Increasing basicity (C/S) causes both ds and fcrys to increase.
However, the replacement of cuspidine has proved difficult. Mould slags have been
divided into three categories in Sect. 4.7.2, and crystallisation is considered indi-
vidually, for each type of slag.
4.7 Crystallinity in Slag Film 137

4.7.2 Factors Affecting fcrys

The required values of the powder consumption (Qreq req

s ), viscosity (η ) and the
req
break temperature (Tbr ) are determined by the mould dimensions, the casting speed
and the steel grade being cast; these required values apply to all three types of slag,
discussed below. The mould slag must possess these required properties plus an
appropriate value of fcrys.

4.7.2.1 Conventional (F-Containing) Mould Slags

Cuspidine (3CaO2SiO2CaF2 or C3S2F1) is precipitated during primary crystalli-

sation in these slags. Other phases are crystallised later during secondary crys-
tallisation [109]. The compositions of most mould slags lie outside the cuspidine
phase field and moving the slag composition into this field results in a reduction in
the horizontal heat flux [110].
Both fcrys and the crystal size (Dcrys) increase with increasing basicity (C/S) for
the slag film. However, additions of other oxides promote glass formation (which
tends to increase the horizontal heat flux). Consequently, the various slag con-
stituents are divided into cuspidine promoters and glass promoters:
(i) Cuspidine promoters (i) CaO (ii) SiO2 (when <CaO) and (iii) CaF2 and
(iv) large amounts of Al2O3 [111].
(ii) Glass Promoters: (i) B2O3 (ii) Na2O, (iii) Li2O (iv) FeO [112] and (v) >4%
MnO and >7% MgO [113].
Although increasing basicity (C/S) is the main tool in controlling both fcrys and
qhor, care must be taken since a high (C/S) can lead to an increased Tliq which can
result in the formation of large slag rims. High (C/S) slags have the advantage of
possessing low oxygen potential.
The glass promoters also tend to slow down the crystallisation process (see
Sect. 9.3.4) but Na2O is thought to reduce the incubation time (i.e. accelerate crys-
tallisation [114]). Additions of >5% TiO2 have been reported to result in the for-
mation of CaOTiO2 which, subsequently, suppresses cuspidine precipitation [115].
Melilite slags have a high basicity (C/S) but have been reported to maintain some
glassy phase, which proved useful (by providing some lubrication) when casting
high Al steels where the large amounts of Al2O3 formed result in a highly crys-
talline slag film [111].

4.7.2.2 F-Free (FF) Mould Slags

The replacement of cuspidine has proved difficult. The phases, perovskite

(CT) [74] CST [114] and NC2S3 and melilite [111] have been proposed as
replacements for cuspidine, in order to reduce qhor. The problem with perovskite is
138 4 How to Manipulate Slag Behaviour in the Mould

that it has a high melting point and in the casting of Ti- stabilised steels, the
basicity must be kept below (C/S) = 1 to ensure that CT is not formed. Melilite
slags have been used for casting round billets (see Sect. 5.2.9) and high Al steels
(see Sect. 6.2.12). The formation of a glassy phase is an advantage in both cases.
Additions of Na2O tend to decrease incubation times for crystallisation [116, 117].

4.7.2.3 Calcium Aluminate (CA) Mould Slags

These slags have been developed for the casting of high Al steels to minimise the
formation of Al2O3. There is a eutectic in the CaO–Al2O3 system at ca. 50% CaO
(close to C12A7) and <10% SiO2 and fluxes are added to the eutectic to reduce
the melting temperature. These CA-type slags have proved to be effective in
casting high Al steels but still need some development to address the following
problems:
(i) The fluxes are restricted to CaF2, Na2O, Li2O and K2O because Al in the steel
reacts with other potential fluxes (e.g. B2O3) to form Al2O3.
(ii) The slag films are crystalline, fragile and provide little lubrication to the shell;
thus it is necessary to find a way of introducing some glassy phase.
The thickness of the slag film can be kept to a minimum by ensuring (C/A) ratio
lies between 0.9 and 2 [118]. The incubation time for crystallisation was found to
be at a maximum using a slag with (C/A) = 1.5 [118] and had the weakest crys-
tallisation tendency.
An equation has been reported for calculating the value of fcrys in the slag film
from the chemical composition (in terms of the modified (NBO/T) of the mould
powder (Eq. 4.19) [113].The modified (NBO/T)* is defined in Eq. 4.20, where the
contents in the curly bracket {}, are included in the numerator if MgO is <7% and
MnO is <4% and in the denominator when they exceed these values:

% crys ¼ 100 fcrys ¼ 284 þ 141:1ðNBO=TÞ ð4:19Þ


ModifiedðNBO/TÞ ¼ 2ðXCaO þ XFeO þ XNa2 O þ XK2 O þ XCaF2 þ XMgO þ XMnO


 XAl2 O3 Þ= XSiO2 þ 2XAl2 O3 þ fXMgO þ XMnO g
ð4:20Þ

According to Eq. 4.19, the slag film will be totally glassy (fcrys = 0) when
(NBO/T)*  2 and totally crystalline (fcrys = 1) when (NBO/T)*  2.75.
4.7 Crystallinity in Slag Film 139

4.7.3 Ways of Increasing Crystallinity in Slag Film

In conventional, F-containing powders the usual method of increasing the frac-

tion of crystalline phase, fcrys, is to increase the {%CaO/%SiO2} ratio from about
1.0 for casting LC grades to 1.3 (up to 1.6) for casting MC, peritectic steel
grades.

4.8 Delaying Solidification and Shortening the Length

of Shell

There is considerable turbulence in the meniscus region which results in inclusion

and bubble entrapment. Delaying the solidification of the shell, to a position lower
in the mould, provides significant benefits since it results in a shorter shell (like that
shown in Fig. 4.11). The shorter shell/meniscus has the following advantages:
• It results in fewer inclusions being trapped in the surface of the steel product.
• It results in shallow oscillation marks (since delayed solidification results in an
increase in (slag rim-shell tip distance) and consequently, results in shallow
oscillation marks.
Delaying solidification involves both vertical and horizontal heat fluxes.

Fig. 4.11 Schematic figure

showing that a shorter
meniscus shell leads to less
slag entrapment (permission
granted, ISS/AIST [7])
140 4 How to Manipulate Slag Behaviour in the Mould

4.8.1 Factors Affecting Shell Length

The following methods can be used to delay solidification and to shorten the length
of the meniscus tip.

4.8.1.1 Reducing the Vertical Heat Transfer

This can be achieved by improving the thermal insulation of the bed (Sects. 4.2.1.3
and 4.2.4) by:
(i) Using a deeper powder bed.
(ii) Using smaller granules and lower Ar flow rates to reduce gaseous
conduction.
(iii) Using exothermic agents in the powder to reduce the thermal gradient in bed.
(iv) Using EMBr to reduce turbulent flow which also decreases qvert.

4.8.1.2 Reducing the Horizontal Heat Transfer

The horizontal heat flux can be reduced by using:

• A thick, crystalline slag (i.e. with a high Tbr and fcrys) which also creates a thick
slag rim.
• Meniscus-free casting (where a refractory lining is added to the meniscus region
of the mould [96, 97]—see Sect. 3.2.3.3)
• Electromagnetic casting (EMC) where the “pinch effect” causes a reduction in
qhor.

4.8.1.3 Superheat

Solidification can also be delayed by increasing the superheat.

4.8.2 Ways of Controlling the Length of Meniscus/ Shell

Solidification can be delayed by:

(i) Reducing vertical heat transfer.
(ii) Reducing horizontal heat transfer in the meniscus region.
(iii) Increasing superheat.
4.8 Delaying Solidification and Shortening the Length of Shell 141

Table 4.1 Summary of the remedial actions that can be undertaken to bring about certain tasks or
objectives
Task Variable Remedial action Repercussions, Comments
Vertical heat flux qvert# (a) From metal by (i) #Vc and Gaseous convection
and thermal Insulation # superheat (ii) # Ar flow rate probably dominant
insulation of the " (iii) use EMBr ! # mechanism in bed
powder bed turbulence and qvert
(b) From slag pool by
improving thermal insulation
(i) Increase depth of bed
(ii) Use smaller granules to #
gaseous convection
(iii) Incorporate exothermic
agents into powder
(iv) Reduce CO32− content
Melting rate QMR" (i) Decrease %Cfree (a) Also decreases
(MR) (ii) Increase C particle size exothermic heat—may cause
(DC") T gradient in powder bed to
(iii) Decrease bulk density increase ! steel meniscus
(qbulk#) may solidify
(iv) Increase carbonate
content
(v) Increase qvert into slag
pool (qin vert")
See vertical heat flux
above-note EMBr ! qvert#
Depth of slag dpool" increasing Dqvert = (qin vert − (a) Vc" ! turbulence
pool qout
vert) by "; ! tn#; ! Qs#;
(i) Increasing qin vert by (a) Vc" (b) dpool tends to be low for
(b) Superheat " thin slabs
(ii) Improving bed thermal
insulation by (a) "dbed
(b) Dgran# (c) Using
exothermic agents
Powder Qs" (i) Decrease viscosity (η#) (a) (η#) Slag entrapment
consumption and (ii) Decrease Vc (Vc#) danger
liquid slag film (iii) f# s# and Tbr# (b) Keep flow rate <5 L
(iv) Increase Ar flow rate min−1
(v) Use Non-sinusoidal osc’n (c) Qs low for thin slabs
(vi) Use EMBr, EMC
(vii) Large amounts of Ti
(CN) + Al2O3 ! form
blockages ! use " dpool to
dissolve more solid
Horizontal heat qhor# By: (i) Tbr" (via. C/S") (a) For Thin slabs of MC
flux and solid slag (ii) fcrys" (via. C/S" + CaF2% steel- danger of bulging
film ")
(iii) Use Non-sinusoidal
oscillation
(iv) Coating mould and MFC
(v) Use EMC
(vi) Use of grooved mould
(continued)
142 4 How to Manipulate Slag Behaviour in the Mould

Table 4.1 (continued)

Task Variable Remedial action Repercussions, Comments
Crystallisation of fcrys" (a) Cuspidine- (i) use high (a) Shrinkage occur during
slag film Dcrys" (C/S) and CaF2 crystallisation ! Porosity"
(ii) Maximise cuspidine (b) Important slag film keeps
formation;% some glass phase for
Na2O" ! Tincubn# i.e. lubrication
accelerates crystallisation Glass promoted by SiO2,
(iii) Large amounts of B2O3; >4%MnO,
Al2O3 ! " fcryst FeO, >75MgO
(b) FF- fluxes- TiO2 (c) Slag films with high
suppresses Cuspidine fcrys—fragile and prone to
(c) CA fluxes- have high fcrys cracking
(e) Dcrys high at high T Dcrys
fine at lower T
Delayed qvert# (i) Reducing qvert by (a) dbed" leads to Nincl# and dOM#
solidification and qhor# (b) Dgran# (c) Use exothermic
shortening shell agents (d) EMBr
tip (ii) Reducing qhor by (a) Tbr"
(b) fcrys" (c) Ar flow#
(d) using MFC and EMC
(iii) Increasing superheat
Note in right-hand column (i) etc. refers to (i) in column to left; MFC Meniscus-free casting; "
increase; # decrease

4.9 Summary

The remedial actions which can be taken to achieve certain tasks or objectives are
summarised in Table 4.1.

References

1. JW Cho T Emi, H Shibata, M Suzuki, ISI J Intl. 38, 440, (1998).

2. JW Cho H Shibata, T Emi, M Suzuki, ISI J Intl. 38, 268 (1998).
3. A Yamauchi, T Emi, S Seetharaman, ISIJ Intl., 42, 1084, (2002).
4. C Lefevre, JP Radot, JN Pontoire, Y Roux, Revue de Metall. CIT, 94, 489 (1997).
5. Y Tanizawa, CAMP-ISIJ, 3, 257, (1991).
6. K Yamaguchi, CAMP-ISIJ, 4 , 181 (1990).
7. F Neumann, J Neal, MA Pedroza, AH Castillejos, FA Acosta, Proc. 79th Steelmaking Conf.
(ISS ,Warrendale, PA, 1996), p. 249.
8. R Saraswat, PhD Thesis,“ Modelling the effect of mould flux on steel shell formation during
continuous casting”, Imperial College, London , 2006.
9. M. Susa, K Nagata, KC Mills , Ironmaking and Steelmaking, 20, 372, (1993).
10. M. Susa, KC Mills, MJ Richardson, R Taylor, D Stewart, Ironmaking and Steelmaking, 21,
279, (1994).
References 143

11. S Diehl, JA Moore, RJ Phillips, Proc. 78th Steelmaking Conf. ( ISS , Warrendale, PA, 1995)
p. 351.
12. R Koldewijn Unpublished Corus Internal Rept (2007) cited in KC Mills, J Kromhout, A
Hamoen and R Boom: Proc. Admet Conf held Dnipetrovsk, Ukraine, 2007.
13. PE Ramirez- Lopez, PN Jalali, J Bjorkvall, U Sjostrom, C Nilsson, ISIJ Intl. 54, 342.
(2014) and Proc. 8th Europ. Conf Continuous Casting , Graz, Austria, 2014 (ASMET,
Vienna, 2014).
14. E van Vliet, DW van der Plas, SP Carless, A A Kamperman, AE Westendorp: Proc. 7th
Europ Conf. Cont. Casting, Dusseldorf, 2011 (VDEh, Dusseldorf, 2011) Session 4.
15. T Sakuraya, T Emi. T Imai, K Emoto, M Kodama, Tetsu-to- Hagane, 67,1220, (1981).
16. K Nagata, KS Goto, Proc. 2nd Intl. Conf. Metallurgical slags and fluxes, Lake Tahoe, NV,
(1984) ed. by HA Fine and DR Gaskell (TMS-AIME, Warrendale, PA, 1984) p. 875.
17. KS Goto, HW Gudenau, K Nagata, KH Lindner, Stahl u Eisen, 105, 1387, (1985).
18. SP Andersson, Ironmaking Steelmaking, 42, 465, (2015).
19. P Andersson, to be published in Ironmaking Steelmaking.
20. JJ Macho, G Hecko, B Golinmowski, M Frazee, Preprints 33rd McMaster Symp. on Iron
and Steelmaking,. McMaster Univ., Hamilton, CN, June (2005).
21. R Taylor, KC Mills: Ironmaking and Steelmaking, 15, 187, (1988).
22. KH Spitzer , K Schwerdtfeger, Investigation of the isolation properties and melting
behaviour of mould powders used in the continuous casting of steel. Report 7249,Tech.
Univ. Clausthal, DE, (1989).
23. KH Spitzer, K Schwerdtfeger, Stahl u Eisen, 108, 441, (1988).
24. M Supradist , AW Cramb , K Schwerdtfeger, ISIJ Intl., 44 , 817, (2004).
25. P Hasselström, “On the formation of a lubrication film from mould powders used in
continuous casting.” Swedish Inst. Metals Research Report IM-1467, (1980).
26. H Lidefeldt , P Hasselström, Proc. Continuous Casting; 4th Intl. Iron and Steel Congress,
(The Metals Society, London , 1982) p. 10.
27. H Litterscheid, “Untersuchung des Verhalten beim Strangeisen, Giesen und erstarren von
stahl III” Report EUR 8569, Contract 7210- CA/112 (EC Sci. Tech. Publ.,
Luxembourg,1984) p. 34.
28. JW Kim, SK Kim, YD Lee, Proc. 4th Europ. Conf. on Continuous Casting, Birmingham,
UK, 2002, ( IOM, London, 2002) p. 371.
29. N Pradhan, M Ghosh, DS Basu, S Mazumdar, ISIJ Intl., 39, 804, (1999).
30. K Koyama, K Nagano, Y Nagano, T Nakano. Nippon Steel Technical Report, 34, 41,
(1987).
31. M Kawamoto, K Nakajima, T Kanazawa, K Nakai, ISIJ Intl., 34, 593, (1994).
32. CÅ Däcker, C Eggertsson, J Lonnqvist, VIII Intl. Conf. Molten slags, fluxes & salts,
Santiago, Chile, Jan., 2009 (GECAMIN, Santiago, 2009) p. 1111.
33. JA Kromhout, DW Plas, Ironmaking and Steelmaking, 29, 303, (2002).
34. T Emi, H Nakato, K Suzuki, Y Iida, T Ueda, Tetsu- to Hagane, 60 (7), 981, (1974) ( Henry
Brutcher Translation, HB9357).
35. M Kawamoto, K Nakajima, T Kanazawa, K Nakai, ISIJ Intl., 34, 593, (1994). see also Proc.
75th Steelmaking Conf., Toronto, (1992) (ISS, Warrendale, PA, 1992) p. 389.
36. T Kishi, Y Nagano, T Nagano, Trans. ISIJ, 25, B 302, (1985).
37. T Nakano, M Fujii, K Nakano, T Matsuyama, Nippon Steel Tech Rept., 34, 21, (1987).
38. J Sardemann, H Screwe, Stahl u Eisen, 111 (11), 39, (1991).
39. R Sato, Bull. Jap. Inst. Metals, 12 (6), 391, (1973).
40. D Singh, P Bhardwaj, Y Dong, A McLean, M Hasegawa, M Iwase, Proc. 8th Intl. Conf.
Molten Slags, fluxes and salts, Santiago, Chile, 2009 (GECAMIN, Santiago, Chile, 2009)
p. 1073.
41. EF Wang, YD Yang, CL Feng, ID Sommerville, A McLean, J Iron Steel Res. Intl. 13 (2), 22,
(2006).
42. QC Wei, YQ Ding, KD Peng, J Chongqing Univ., 4, 110, (1995).
144 4 How to Manipulate Slag Behaviour in the Mould

43. B Xie, J Wu, Y Gan, Proc. 74th Steelmaking Conf., Washington, DC, 1991 ( ISS,
Warrendale, PA, 1991) p. 647.
44. K Schwerdtfeger, A Jablonka, Steel Research Intl., 64, 77, (1993).
45. K Schwerdtfeger, J Sardemann, HJ Grethe, Stahl u Eisen, 114, 57, (1994).
46. PE Ramirez-Lopez, KC Mills, PD Lee, B Santillana, Met. Mater. Trans. B, 66, 109 (2011).
47. P.V. Riboud ,Y. Roux: Fundamental study of the behaviour of casting powders Report 9560,
(1985) (Europ. Comm. Sci. Tech. Publ., Luxembourg, 1985).
48. RM McDavid , BG Thomas, Metall. Trans. B, 27B, 672, (1996).
49. M Washio, M Sugizawa, S Moriwaki, K Kariyaa, S Idogawa, S Takeuchi, Revue de
Metallurgie, CIT, 90 (April), 507, (1993).
50. Y Nakamori et al, Nippon Steel Tech Report, 34, 53, (1987).
51. F Mantovani, S Spagnul, M Padovan, A Bianco, Proc. 8th Europ. Conf. Continuous casting ,
Graz, Austria, (2014) (Austrian Soc. Metals, 2014).
52. N Viswanathan, AB Fox , KC Mills, Proc. Mills Symp., London, (2002) ed. RE Aune and S
Sridhar, (IOM, London 2002) p. 513.
53. B Zhao, SP Vanka, BG Thomas, Intl. J Heat and Fluid Flow, 26, 105, (2005).
54. E Maclas, AH Castillejos, FA Acosta, GM Herrera, F Neumann, Ironmaking and
Steelmaking, 29, 347, (2002).
55. R Bommaraju , Proc. 74th Steelmaking Conf., Washington, DC, 1991, ( ISS, Warrendale,
PA, 1991) p. 95.
56. JA Moore, C Camino, S Diehl, RJ Phillips , D Piwinsky, Proc. 79th Steelmaking Conf.,
Pittsburgh, PA , 1996, (ISS, Warrendale, PA, 1996) p. 259.
57. T Nakano, M Fuji, K Nakano, S Mizoguchi, T Yamamoto, K Asano, Trans. ISI J, 21, B 307,
(1981).
58. C-Å Däcker, A Salwén, P Andersson, C Eggertsson, Proc. 7th Europ. Continuous Casting
Conf., Düsseldorf, (2011) Session 12, pp. 1–8.
59. M Wolf, AIME Electrical Furnace Proc., 40, 335, (1982).
60. M Wolf, Proc. 2nd Europ. Conf. on Continuous casting, Dusseldorf, 1994, p. 335.
61. AB Fox, PhD Thesis, Mould fluxes, their properties and performance. Imperial College,
London (2003).
62. S. Sridhar, K.C. Mills, V. Ludlow, S.T. Mallaband, Proc. 3rd Europ. Conf. Continuous
Casting, Madrid, (UNESID, Madrid, 1998) p. 807.
63. S Ogibayashi, K Yamaguchi, T Mukai, T Takahashi, Y Mimura, K Koyama. Y Nagano, T
Nagano, Nippon Steel Technical Report, 34, 1, (1987).
64. K. Tsutsumi, H Murakami, S Nishioka, M Tada, M Nakada, M Komatsu, Tetsu- to Hagane,
84 , 617, (1998).
65. H Maeda, T Hirose et al, CAMP-ISIJ, 6, 280, (1993).
66. T Mizukami, K Kawakami, T Kitagawa, M Suzuki, S Uchida, Y Komasu, Trans ISIJ, 26, B
164, (1986).
67. R D’Haeyer, Influence of powder consumption on physical properties of continuous casting
powder, Report EUR 10326 (Europ. Comm.Sci. Tech. Publ, Luxembourg, 1987).
68. M Suzuki, S Miyahara, T Kitagawa, S Uchida, K Okimoto, Tetsu-to Hagane, 78, 113,
(1992).
69. T Mallaband: Unpublished results from (Metallurgica, UK) cited in A Fox: PhD Thesis,
Imperial College, London. (2003).
70. M Tani, T Toh, K Umetsu, K Tanaka, M Zeze, K Tsunenari, K Hayashi, S Fukunaga,
Nippon Steel Technical Report, 104, 62, (2013).
71. T. Mukongo, MEng Thesis , “Effect of titanium pick-up on mould flux viscosity in continuous
casting of titanium-stabilised stainless steel.”, Faculty of Eng., Univ. Pretoria, South Africa,
(2003).
72. RC Nunnington, N Sutcliffe, Proc. 59th Electric Furn. Conf., Pheonix, AZ, Nov (2001).
73. Q Wang, YJ Lu, SP He, KC Mills, Z Li, Ironmaking and Steelmaking, 38, 297, (2011).
74. GH Wen, S Sridhar, P Tang, X Qi, TQ Liu, ISIJ Intl., 47, 1117, (2007).
75. JW Cho, KE Blazek, MJ Frazee, HB Yin, JE Park , SW Moon, ISIJ Intl., 53, 62, (2013).
References 145

76. M Hanao, M Kawamoto, ISIJ Intl., 48, 180, (2008).

77. MV Fonseca, ODC Afrange, A Lavinas, AA Ramos, CA Valadares, Proc. 5th Intl. Conf.
Molten Slags, fluxes and salts , Sydney, Australia,1997 (ISS, Warrendale, PA, 1997) p. 851.
78. ODC Afrange, MVA Fonseca, H Polivanov, « Differential thermal analysis applied in
design, quality control and establishment of prefused and mixed mould powders operational
parameters. Intl. Congress Metall. Technol. Materials, 1994, p 623.
79. MCC Bezerra, ODC Afrange, AC Valadares, Proc. Mills Symp., London, (2002) ed. RE
Aune and S Sridhar, (IOM, London, 2002 ) p. 293.
80. J Kromhout, RS Schimmel, Proc. 8th Europ. Conf on Continuous casting, Graz, Austria,
June, 2014 (Austrian Soc. Metals, 2014).
81. K Tsutsumi, R Nakamura, J Ohtake, T Nagasaka, M Hino, ASIA Steel International
Conference, 2000 (ASIA Steel’ 2000), (Chinese Society for Metals, Beijing, 2000) Vol. C,
(2000), p. 366.
82. HB Yin, K Blazek, J Macino, J Cotrell, Proc. 8th Europ. Conf. Continuous casting, Graz,
Austria, 2014, (Austrian Soc. Metals, 2014).
83. M Hanao M Kawamoto, M Hara, T Murakami, H Kikuchi, K Hanazaki, Tetsu-to-Hagane,
88, 23, (2002).
84. JN Pontoire, JP Radot,V Delvaux, E Dehaussy, Revue de Metallurgie, CIT. 93, 1237,
(1996).
85. Y Uemiya, T Mizoguchi, T Kajitani, Proc. 9th Intl. Conf. Molten Slags, fluxes and salts
(Molten 12) Beijing, 2012 (Chinese Soc. Metals, 2012).
86. H Yamamura, T Kajitani, J Nakashima, M Yamasaki, S Mineta, Nippon Steel Technical
Report, 104, 54, (2013).
87. HP Narzt, R Martinelli, C Furst, G Xia, H Pressinger, Proc 3rd Europ. Conf. Continuous
casting, Madrid, Spain, 1998, (UNESID, Madrid, 1998) p. 747.
88. L Hering, HP Heller, HW Fenske. Stahl u Eisen 17, 61 (1992).
89. P E Ramirez-Lopez, P.D. Lee, K.C. Mills, ISIJ Intl., 50 (3), 425, (2010).
90. M Wolf, Iron and Steelmaker, 23, 47, (1995).
91. M Hanao, M Kawamoto, A Yamanaka, ISIJ Intl., 52, 1310, (2012).
92. M Suzuki, S Miyahara, T Kitagawa, S Uchida, K Okimoto, Tetsu-to- Hagane, 78, 113,
(1992).
93. JA Kromhout, ER Dekker, M Kawamoto, R Boom, Ironmaking and Steelmaking, 40, 206,
(2013).
94. MJ Lu, JH Chen, DH Tsai, CH Huang, Proc. 78th Steelmaking Conf. (ISS, Warrendale, PA.,
1995) p.359.
95. A Cristallini, R Langella, A Polk, Proc. 3rd Europ. Conf. Continuous casting, Madrid, Spain
1998, (UNESID, Madrid, 1998) p 581.
96. C Bertoletti, P Courbe, JM Jolivet, PP Naveau, A Oper, E Perrin, C Salaris, AL Spierings , E
Weisseldinger, Proc. 3rd Europ. Conf. Continuous Casting, Madrid, 1998 (UNESID,
Madrid, 1998) p. 65.
97. P Courbe, JM Jolivet, G Lesoult, E Perrin, P Nilles, Steel Research, 70 (8/9), 338, (1999).
98. K Nakai T Sakashita, M Hashio, M Kawasaki, K Nakajima, Y Sugitani, Tetsu-to- Hagane,
73 (3), 498, (1987).
99. Y Sugitani, M Nakamura, M Okuda, M Kawasaki,K Nakajima, Trans. ISIJ, 25, B 91,
(1985).
100. JW Cho, HT Jeong, Met. Mater. Trans. B, 44B, 146, (2013).
101. H Murukami, M Suzuki, T Kitagawa, S Miyahawa, Tetsu-to- Hagane, 78, 105, (1992).
102. T Mochida, S Kohriyama, S Itoyama, T Yamashita, M Suzuki, Y Kishimoto, CAMP-ISIJ,
20, 851, (2007).
103. K Ichikawa, A Morita, Y Kawabe, Shinagawa Tech. Report, 36, 283, (1993).
104. J Kromhout, PhD thesis Mould powders for high speed continuous casting of steel. Univ of
Delft, NL (2011) Chapter 5.
105. P Rocabois, JN Pontoire, J Lehmann, H Gaye, J Non- Cryst. Solids, 283, 98, (2001).
146 4 How to Manipulate Slag Behaviour in the Mould

106. Y Harada, K Kusada, S Sukenaga, H Yamamura, Y Ueshima, T Mizoguchi, N Saito, K

Nakashima, ISIJ Intl., 54, 2071, (2014).
107. Y Wang, Yu Wang, L Fan , B Xie, XP Liang, X Zeng, ZW Wei, Proc, 9th Intl. Conf. Molten
slags, fluxes and salts , Beijing, China (2012) ((Chinese Metals Soc., 2012).
108. S Riaz, Ironmaking and Steelmaking, 39, 409, (2012).
109. P Grieveson, S Bagha, N Machingawuta, K Liddell, KC Mills, Ironmaking Steelmaking, 15,
181, (1988).
110. M Hanao, M Kawamoto, T Watanabe, ISIJ Intl., 44, 827, (2004).
111. M Hanao, Y Tsukaguchi, M Kawamoto, Proc. ICS 2008, Gifu, Japan (2008).
112. M Susa, A Kushimoto, H Toyota, M Hayashi, R Endo ,Y Kobayashi, ISIJ Intl., 49, 1722,
(2009).
113. Z Li, R Thackray, KC Mills, Proc. 7th Intl. Conf. Metallurgical slags, fluxes and salts ,
Capetown, 2004 (SAIMM, Johannesburg, 2004) p. 813.
114. H Nakada, K Nagata, ISIJ Intl., 46, 441, (2006).
115. Z Wang, QF Shu, XM Hou, KC Chou, Ironmaking and Steelmaking, 39, 210, (2012).
116. JL Klug, R Hagermann, NC Heck, JFC Viela , PR Scheller, ICS Conf. 2012 , Dresden, DE,
2012.
117. JL Klug, R Hagermann, NC Heck, JFC Viela, PR Scheller, Proc. 9th Intl. Conf. Molten slags,
fluxes and salts, Beijing, China (2012) (Chinese Metals Soc.).
118. XJ Fu, H Wen, P Tang, Q Liu, ZY Zhou, Ironmaking and Steelmaking, 41, 342, (2014).
119. F. Shahbazian, private communication concerning plate dip tests, SWEREA, KIMAB,
Stockholm, Oct. (2016).
Chapter 5
Effect of Casting Variables on Mould
Flux Performance

Abstract Continuous casting is a complex process affected by a number of factors

and it is necessary to optimise these factors (e.g. shell lubrication and heat
extraction) to minimise product defects and process problems. The mould powder is
expected to alleviate these problems. However, the casting conditions have a
pronounced effect on factors like slag infiltration and heat transfer and this requires
careful selection of mould slag properties to provide optimum casting conditions.
The effect of individual, casting variables (e.g., casting speed) on the casting pro-
cess is discussed in this chapter and the following casting variables are examined
and discussed: (i) mould characteristics (e.g., dimensions, taper, coatings)
(ii) oscillation characteristics (iii) casting speed (iv) metal flow characteristics
(v) metal-level variations (vi) fluctuations (in metal flow, mould level, etc. (vii) the
application of electromagnetic devices (viii) steel grade being cast (ix) water flow
rate. The effects of these individual factors are discussed and how the mould
powder can be used to optimise the casting process.

Symbols, Abbreviations and Units

ds Thickness of solid slag layer (m)
ds Thickness of liquid slag layer (m)
f Frequency (cpm or Hz)
fcrys Fraction crystalline phase
Qs Powder consumption (kg m−2)
q Heat flux density (Wm−2)
R* Mould (surface area/volume) (m−1)
s Stroke length (m)
T Temperature (°C)
Tbr = Tsol Break, Solidification temperature
t Time (s) or thickness of mould (m)
tn Negative strip time (s)
VC Casting speed (m min−1 or ms−1)
Vm Velocity of mould (ms−1)
w Width (m)

© Springer International Publishing AG 2017 147

K.C. Mills and C.-Å. Däcker, The Casting Powders Book,
DOI 10.1007/978-3-319-53616-3_5
148 5 Effect of Casting Variables on Mould Flux Performance

η Slag viscosity (dPas)

EMBr Electromagnetic braking
EMC Electromagnetic casting
EMS Electromagnetic stirring
HC High-carbon steel
LMF Level magnetic field
MC Medium carbon steel

5.1 Introduction

There are many factors affecting the control of the continuous casting process.
Frequently, the mould flux is asked to alleviate the problems arising from the
casting conditions (e.g. those due to metal flow turbulence). Consequently, it is
essential to have knowledge of how the various casting conditions affect (i) the heat
transfer, powder consumption, metal flow etc. (ii) the mechanisms responsible for
process control problems and defect formation and (iii) to identify the limitations of
any proposed solution involving mould flux use. It should be noted that many of the
factors itemised below are interactive and steps taken to solve one problem may
have an adverse effect in triggering other problems or defect mechanisms. In this
chapter, the effects of individual casting variables on the casting process are
examined and how mould fluxes can be used to alleviate these problems or defects.

5.2 Mould Characteristics

5.2.1 Mould Dimensions

It is customary to classify the cast steel products according to their mould geome-
tries, e.g. into (i) slabs (ii) blooms (iii) billets and (iv) thin slabs. The liquid friction
forces acting on a specific face of the shell tend to increase with increasing distance
from the corner (Fig. 5.1) [1, 2]. Consequently, the amount of lubrication required
(i.e. required powder consumption, Qreq s ) increases with increasing width (w) of the
wide face. Since the required lubrication is usually expressed in terms of the required
powder consumption Qreq req
s the hierarchy (for Qs ) is slabs > blooms > billets.
Although the distance from the corner has been used to represent the effect of
mould geometry on the required powder consumption, Qreq s , [2] it is customary to
use the parameter R*, (defined in Eq. 5.1) [3]; R* represents the ratio (surface
area/volume) of the mould. Typical values for R* values for slabs are, slabs ca. 10,
blooms (10–18), billets (22–30), thin slabs ca. 40. Values of R* for round billets
can be calculated using R* = 4/D where D is the diameter of the billet.
5.2 Mould Characteristics 149

Fig. 5.1 Diagram showing

the centre-line friction
increasing with increasing
distance from the corner;
x friction at centre line for 3
mould geometries [1, 2]; solid
line billet; dotted line bloom
and solid line slab (courtesy
of AB Fox [2])

R
ðm1 Þ ¼ 2ðw þ tÞ=w  t ð5:1Þ

Since slabs have a large surface area (cf. those for blooms and billets), it follows
that a higher powder consumption [in kg m2 (of mould)] is required to lubricate
slabs than for other mould dimensions. The relationship between Qs and R* shown
in Fig. 5.2 can be described by the following equation [4, 5].

0.8
Billets
Powder Consumption, Q ccorr (kg m 2)

0.7
Blooms

Slabs
0.6
Round

0.5 High Speed Billet

[4] -0.04R
: Qs =0.44e
0.4
: Qs=2/(R-5)

Thin Slabs
0.3

0.2

0.1

0
0 10 20 30 40 50
-1
Surface Area to Volume Ratio, R (m )

Fig. 5.2 Required powder consumption, Qreq s , as a function of the parameter, R* for different
mould geometries; ◊ = billets; o = blooms; D = slabs; ▬ = thin slabs; + = Rounds; x = high
speed billets; dotted line (Qs = 0.44 exp−0.44R*) [3]; solid line (Qs = 2/(R − 5) [4, 5]; (permission
granted, UNESID [5])
150 5 Effect of Casting Variables on Mould Flux Performance

s req ¼ 2=ðR  5Þ
Qslag ð5:2Þ

Thus, the required powder consumption is principally dependent upon the mould
dimensions.

5.2.2 Mould Length (Lmould)

It is important that (i) the shell is lubricated by liquid slag throughout the mould and
(ii) the steel shell at the mould exit is sufficiently thick to ensure that no bulging
occurs (which leads to mould level variations) [6]. The mould length affects both of
these issues. Wolf [7] has discussed the importance of mould length in the con-
tinuous casting process; the required Lmould tends to increase with increasing casting
speed.

5.2.3 Mould Taper (Lmould)

Mould taper is important since an insufficient mould taper for the steel being cast
leads to (i) hot spots at off-corner sites and (ii) to sub-mould bulging [6, 8]. In
contrast, an excessive taper leads to overcooled corners and transverse cracking [8]
(see Sect. 11.1.6 and Fig. 11.48). The ideal taper for the steel is dependent upon the
steel composition, the casting speed and the mould length; a parabolic taper is
recommended for the meniscus region [8].

5.2.4 Mould Coatings

Mould coatings (such as Mo, Cr or Ni) are usually applied to the upper part of the
mould in order to extend the life of a mould. However, these coatings also help to
reduce the horizontal heat flux since the conductivity of these metals is significantly
lower than that of copper; coatings are therefore beneficial in reducing longitudinal
cracking in MC steel grades (see Sect. 3.2.3.7(v)).

5.3 Casting Speed (Vc)

The casting speed (Vc) has a marked effect on the casting process. Increased pro-
ductivity is attained by using higher casting speeds. However, an increase in casting
speed results, simultaneously, in the following changes:
5.3 Casting Speed (Vc) 151

• A decrease in powder consumption, i.e. less lubrication for the shell.

• An increase in the rate of heat transfer (qhor, in Wm−2 or Js−1 m−2).
• A shorter residence time in the mould (resulting in decrease in the total heat
flux) and which leads to a thinner shell (despite the increase in qhor) and which
could result in improved lubrication in the lower mould.
• Increased metal flow turbulence and higher metal flow velocities (which could
result in slag entrapment and/or thinning of shell in the corner leading to lon-
gitudinal corner cracking).
• A decrease in negative strip time (tn) which will reduce the depth of the
oscillation marks formed.
It is apparent that casting speed changes affect mould performance in a variety of
ways and these are discussed in more detail below.

5.3.1 Effect of Casting Speed on Powder Consumption

The effect of casting speed on powder consumption can be seen from the empirical
rules reported for the required powder consumption, Qslag
sreq [9, 10], e.g. Eq. 5.3 (or
the alternative relation, Qpowd
s = 0.60/η Vc [9–11]). Thus, the required powder
consumption decreases with increasing casting speed and slag viscosity.

s req ¼ 0:55=g
Qslag  Vc ð5:3Þ
0:5

Thus the slag viscosity must be selected to satisfy both the mould dimensions
and casting speed used, e.g. by using Eqs. 5.2 and 5.3.

5.3.2 Effect of Casting Speed on Heat Transfer

Increasing the casting speed results in an increased heat flux density (q), but leads to
a decrease in the total heat flux (qtotal) because of the shorter residence time in the
mould ðq " and qtotal # as Vc "Þ. Increased casting speed results in a thinner shell
ðdshell # as Vc "Þ.
The horizontal heat flux is principally controlled by the nature of the solid slag
film formed between the shell and the mould; the heat flux decreases as (i) the solid
film thickness (ds) increases [which, in turn, increases as Tsol (or Tbr) increases (i.e.
ds" as Tbr")] and (ii) the fraction of the slag film forming crystalline phases (fcrys)
increases.
However, the response of the slag film to a change in heat flux is not immediate
since it takes time for the slag film thickness to adjust to these changes (Fig. 5.3)
152 5 Effect of Casting Variables on Mould Flux Performance

Fig. 5.3 Mould temperatures (related to heat flux) as functions of casting speed for a cycle
(a) where casting speed, Vc, is increased then decreased and (b) where Vc is decreased then
increased (permission granted, Nippon Steel Sumitomo Metal (NSSM) Corp. [11])

[11] and in some cases this has led to sticker breakouts [11]. This is particularly so
in the lower mould where fracture of the slag film takes 20–30 min to repair since
the rate of liquid slag supply is low in the lower mould [12].

5.3.3 Effect of Casting Speed on Metal Flow Turbulence

Increased casting speed results in increased metal flow turbulence which, in turn,
produces:
• Waves on the surface of the steel with heights of up to 30 mm in slabs and
50 mm in thin slabs; these waves(result from the impact of the metal flow on the
mould wall) [13].
• Vortices (shown in Fig. 5.4a) [13].
• Significant fluctuations in the metal flow.
Slag entrapment tends to become an ever-increasing problem as casting speeds
are increased. Electromagnetic devices (e.g, EMBr, see Sect. 5.8) have proved
useful in moderating metal flow velocities. High metal flow velocities can also lead
to thinning of the shell in the corner regions (Fig. 5.4b) leading to longitudinal
corner cracking [14].

5.3.4 Effect of Casting Speed on Negative Strip Time

The negative strip time (tn) is calculated using Eq. 5.4 where f = frequency and s is
the stroke length [15].
5.3 Casting Speed (Vc) 153

Fig. 5.4 Schematic diagrams showing (a) Karman vortex formed in steel [13]) and (b) thinning of
the shell in corner regions caused by excessive metal flow [14]; (permission granted (a) ISIJ [13]
and (b) UNESID [14])

   
60 vc
tn ¼  arc cos ð5:4Þ
pf psf

The equation contains two terms, with the first term being more important and
the second term acting so as to reduce the first term. Consequently, increases in the
second (arc cos) term result in a decrease in tn. Thus, an increase in casting speed
results in a decrease in tn. The depth of the oscillation marks (dOM) increases with
increasing tn [16]. Transverse cracking tends to increase with increasing dOM.

5.4 Oscillation Characteristics

The mould is oscillated to prevent the steel shell from sticking to the mould. The
oscillation characteristics affect:
• The powder consumption (i.e. the lubrication supplied to the shell).
• The heat transfer.
• The depth of the oscillation marks (dOM).
The oscillation characteristics are usually defined in terms of the positive (tp) and
negative strip times (tn) [15].The negative strip time is the fraction of the oscillation
cycle where the mould is descending faster than the mould (i.e. Vm > Vc) and is
calculated by Eq. 5.4. The positive strip time constitutes the remainder of the
oscillation cycle (Eq. 5.5).

tcycle ¼ tn þ tp ð5:5Þ

The predictions of a mathematical model [13, 16, 17] have shown that the slag
flow in the region of the mouth of the shell/mould gap is affected by the movement
of the mould. When the mould is ascending the slag flow is hot (“tropical”) and
154 5 Effect of Casting Variables on Mould Flux Performance

Fig. 5.5 Schematic diagram showing the direction of slag flow at different parts of the oscillation
cycle (a) at highest position of mould in late tp (b) halfway through tn (c) at lowest position of
mould in early tp and (d) midway through tp [13, 16, 17]; (Permission granted, ISIJ [13])

radially outward as shown in Fig. 5.5a. However, when the mould and the slag rim
reach their highest positions (in late tp) there is a tide change and the slag flow
becomes downward. Since this flow originates from the cooler parts of slag pool,
the resulting flow is cold or “arctic” (Fig. 5.5b, c). This arctic flow persists until the
mould reaches its lowest point in early tp. Tide changes (which occur at the highest
and lowest positions of the mould) are not instantaneous and there is a period of
confused flow before the new flow is established.
It has been reported that the distance between the slag rim and the liquid steel
adjacent to the shell tip (drim/tip) is an important factor affecting both the depth of
oscillation marks (since dOM increases as drim/tip decreases) and in delaying solid-
ification of the shell [13].
The oscillation characteristics affect the powder consumption, the heat flux and
depth of the oscillation marks (dOM). The results of a parametric study using a
mathematical model [16–18] are summarised in Table 5.1.

5.4.1 Effect of Oscillation Characteristics on Powder

Consumption

Mathematical models based on Navier–Stokes calculations consistently predict that

powder consumption (Qs) increases with increasing frequency (f) [19–22].
However, plant observations indicate the reverse relationship (i.e. Qs" as f#) [23].
This discrepancy between predicted and actual dependence of Qs on frequency has
been explained in terms of the “periods of confused flow” following the tide change
[16]. It has been proposed that slag infiltration is at a low level during these periods
(following a tide change). Furthermore, increasing the frequency effectively
5.4 Oscillation Characteristics 155

Table 5.1 Predicted cause and effect relationships during casting

Parameter !
vc ƒ! ! ! ! !s cmsl !
g ƒ!
DT tn tp f Tbr
# dOM ¼ " " # " " # " # #
" Qs ¼ # " " " # " " # #
# qpeak ¼ # # " # " # # " "
" = increase; # = decrease in values, e.g. dOM# as Vc" [16–18]; DT = superheat; cmsl = Interfacial
tension between liquid metal and liquid slag; η = slag viscosity

increases the number of these periods of low slag infiltration per second.
Consequently, increased frequency will lead to increase in the time where slag
infiltration is low and so will cause a decrease in powder consumption. The failure
to take into account the increased importance of these periods of confused flow
(accompanied by low slag flow rates) with increasing frequency may well be the
reason why slag infiltration models [19–22] give erroneous predictions of the effect
of frequency on powder consumption.
The results of the parametric study given in Table 5.1 are in agreement with
plant observations [23] with the exception of the stroke where a statistical analysis
of plant observations indicated that Qs increases as s decreases. This discrepancy
could possibly be due the fact the stroke affects Qs through its effect on tn, so that an
increase in stroke would lead to an increase in tn which causes Qs to increase (i.e.
s "! tn "! Qs ").
The use of non-sinusoidal oscillation has been reported to bring about an
increase in the liquid slag layer (i.e. Qs) [24, 25].

5.4.2 Effect of Oscillation Characteristics on Heat Flux

Shell solidification occurs in the period of the oscillation cycle when the mould is
descending [16, 17]. The heat flux increases throughout the negative strip period
and peaks in early moments of positive strip (tearly
p ) when the mould and slag rim
attain their lowest positions. Thus, a long stroke may be expected to reveal that the
heat flux, qhor, increases with increasing stroke length; a statistical analysis of plant
data has confirmed such a relationship (i.e. qhor" as s") [26].
The results of a parametric study using a mathematical model are also in
agreement with this finding (qhor# as s#) [17, 18]. The results of this study also
indicated that (qhor# as f") and (qhor# as tn").
The use of non-sinusoidal oscillation has been reported to bring about a decrease
in both the heat flux and the thickness of the solidified shell [25].
It has been reported that variability in qhor tends to increase with increasing
depth of oscillation mark [27].
156 5 Effect of Casting Variables on Mould Flux Performance

5.4.3 Effect of Oscillation Characteristics on Oscillation

Mark Depth (DOM)

On the basis of the mechanism proposed by Ramirez et al. [16] the depth of the
oscillation mark would be expected to increase as:
(i) The stroke length increases (i.e. dOM " as s").
(ii) The distance between rim and the shell tip (drim/tip) decreases (i.e.
dOM " as drim=tip #).
The depth of oscillation mark (dOM) has been reported to increase with
increasing negative strip time (i.e. dOM " as tn ") [16] and that the effect of other
oscillation variables on dOM can all be predicted from their effect on tn (i.e. dOM "
as tn " or as Vc #; s " f #). This is consistent with the trends reported in Table 5.1
obtained in a parametric study using a mathematical model [16, 18].

5.5 Mould-Level Control

Good mould-level control is essential to the process, since poor mould-level control
leads to a variety of problems. These include:
(i) decreases in mould temperature which lead to the formation of large slag
rims [28, 29].
(ii) increased incidences of sticker breakouts [30] which are probably associated
with the fracture of large slag rims.
(iii) the formation of longitudinal and transverse depressions which are also
associated with the fracture and capture of the slag rim [31–33].
(iv) Increased concentrations of inclusions and bubbles (pinholes) in the steel
[34, 35] and slag entrapment [36].
(v) Changes in the depth of oscillation marks [37, 38].
(vi) Increases in longitudinal cracking [29, 39].
(vii) The formation of bleeds [40].
There are two types of mould-level controllers in common use (i) eddy current
and (ii) radioactive controllers. It is reputed that the former provide tighter control
of the mould level. Mould-level fluctuations tend to be higher at the onset of casting
especially when the removal of the dummy bar results in a sharp drop in the metal
volume. Mould-level fluctuations can lead to variations in heat transfer (and to a
corrugated shell with large differences in thickness). This, in turn, results in lon-
gitudinal cracking and the formation of depressions. The predictions of a mathe-
matical model [37] indicate that a decrease in steel-level results in a thinner shell
with a smaller pitch between oscillation marks.
Mould-level variations can also result from “bulging” of the strand which occurs
when the shell (at the mould exit) is too thin to withstand the pressure exerted by
5.5 Mould-Level Control 157

(a)

(b)
3.5

Mould level variation,mm

3

2.5

1.5
1.5 2 2.5
-2
Heat flux,MWm

Fig. 5.6 a Schematic diagrams showing (a) mechanism for mould-level variations [6] when the
shell leaving the mould is too thin and (b) mould-level variations as a function of the horizontal
heat flux [41]; (permission granted (a) ISIJ [6] (b) drawn from data in [41])

the rolls [6]. It is considered good practice when casting, peritectic, medium-carbon
steel grades to reduce the horizontal heat transfer in order to minimise longitudinal
cracking. However, this low-heat flux practice can result in a very thin shell at the
mould exit; this, in turn, leads to bulging and mould-level fluctuations (Fig. 5.6a)
[6]. This mechanism is supported by the plant data for mould-level variations and
horizontal heat flux obtained on a thin-slab caster using different mould fluxes [41].
It can be seen from Fig. 5.6b that a high heat flux is needed to minimise the effect of
mould-level fluctuations arising from bulging of the shell [41]. In such cases, best
practice requires a compromise. Some work has been carried out on developing a
mould powder which produces a slag film which minimises the heat flux in the
meniscus region but encourages higher heat fluxes further down the mould to
provide a thicker shell at the mould exit [41].
Mould-level variations contain various frequency waves (f1 = < 0.1 Hz;
f2 = 0.1–1 Hz; and possibly f3 > 1 Hz) [42]. The long period waves are related to
the surface velocity of the molten steel [36]; the f1 frequency is linked to bulging
and f2 is possibly tied into surface flow and is affected by the mould dimensions
[42]. The mould-level fluctuations can be minimised through control of SEN design
and other casting variables [36]. These fluctuations can also be reduced using
electromagnetic devices (e.g. EMBr [43–46] and EMS [35, 47]). Control of the
upper poles of the EMBr provides the reduction in mould-level variation whereas
158 5 Effect of Casting Variables on Mould Flux Performance

the lower poles reduce the metal flow velocity [35, 42, 43] and the penetration into
the mould which results in a reduction of slag entrapment and pinholes in the cast
steel.

5.6 Metal Flow

The metal flow is very important in the control of the continuous casting process.
The drive for improved productivity has led to the use of high casting speeds. It has
been found that there are several metal flow patterns established in the mould. It is
generally accepted that the pattern exhibiting a “double roll” provides the best
conditions for continuous casting (Fig. 5.7a). The Single-roll flow pattern
(Fig. 5.7b) has a tendency to lead to skull formation (i.e. a frozen steel meniscus)
and does not provide sufficient time for the flotation of inclusions and gas bubbles
to reach the molten slag pool. Asymmetric flows are very undesirable and are
caused principally by clogging of the SEN (Fig. 5.7c) [48, 49].
High casting speeds lead to both turbulence and to considerable fluctuations in
the metal flow velocities; these are discussed in Sect. 5.4. These fluctuations start in
the SEN (Figs. 5.8, 5.9a and 5.10a). The impact of the metal flow on the SEN base
plate results in the formation of vortices (see, for instance, Figs. 5.9a and 5.10a) and
these vortices interfere, intermittently, with the flows passing through the SEN
ports. Thus, the metal flow leaving the SEN is continually fluctuating. However,
further turbulence is caused by the impact of the metal flow on the mould wall. This
results in the formation of surface waves and vortices (Fig. 5.4a); the height of the
surface wave (20–30 mm for slabs and up to 50 mm for thin slabs) increasing with
increasing metal velocity.
The metal flow usually impacts against the shell around 40–50 mm below the
metal surface and the hot metal flow can cause some thinning of the shell at this
point (Fig. 5.4b) [14]. The principal factors affecting metal flows are:

Fig. 5.7 Schematic diagrams showing various metal flow patterns (a) double roll (promoted by
low Ar flow rate, deep SEN immersion and high Vc) (a) single-roll (promoted by high Ar flow rate,
shallow SEN immersion and low Vc) and (c) asymmetric flow (promoted by SEN clogging) [48];
(permission granted, UNESID [48])
5.6 Metal Flow 159

• The casting speed (turbulence increases as Vc increases).

• The immersion depth of the SEN (turbulence increasing at lower immersion
depths) but it is customary to vary the immersion depth to spread the wear on the
SEN.
• The SEN port-design, [14, 50].
• Argon flow rate (VAr) [51, 52] since the Ar flow tends to “cushion” the metal
flow and the flow pattern has been found to transform from double roll to
single-roll when VAr > 5 L min−1 [51, 52].
Electromagnetic braking (EMBr) can be used to reduce metal flow turbulence.
However, it is necessary to optimise the settings of the EMBR. It is possible to
optimise the settings for the conditions using a reliable, validated mathematical
model. Figure 5.8 is made up of frames taken from videos of the model predictions
[53]. These frames show the flow patterns formed with (a) no EMBR (b) EMBR
setting 91 mm (c) EMBR setting 121 mm [53]. The flow systems shown in
Fig. 5.8a indicates the pattern obtained with no EMBR, the surface flow is a very
slow, single roll which means there is a possibility of the metal surface freezing (i.e.
skull formation). Figure 5.8b shows that the application of EMBR across the nozzle
ports does create a double-roll flow, but in this case, the flow is very unsteady with
occasional lapses into “asymmetric flow”, which is undesirable [53]. Figure 5.8c
shows a much more satisfactory and stable “double-roll” flow with the EMBR
positioned lower, below the nozzle ports [53]. Thus, mathematical models can aid
the setting of EMBr to provide a suitable metal flow.

Fig. 5.8 Frames from video of model predictions for (a) no EMBR (b) EMBR setting 92 mm
(across nozzle) (c) EMBR setting 121 mm (below nozzle) [53]; (permission granted, ISS/AIST
[53])
160 5 Effect of Casting Variables on Mould Flux Performance

Fig. 5.9 Predicted metal flow patterns showing (a) the vortex formed in the sump of the SEN
(b) single-roll pattern formed with no FC (=no EMBR) (c) double roll pattern formed with FC and
the flow bias towards the loose, wide side of the mould [52]; (permission granted, VDEh, Verlag
Stahleisen [52])

Mathematical modelling has also proved useful in optimising the effects of

EMBr on the metal flow pattern in other studies (Fig. 5.9). It was shown that:
• EMBr improved the metal flow pattern (Fig. 5.9b, c) by transforming a single
roll into a double-roll flow [52].
• The swirl in the sump of the SEN (Fig. 5.9a) was found to bias the flow,
preferentially, towards the loose, wide face of the mould [52].
It has been reported that swirling flow SEN can reduce the turbulence of the
metal flow leaving the SEN [54].

5.7 Fluctuations in Processes

The metal flow in the mould is in a constant state of fluctuation. As mentioned

above, these fluctuations originate in the SEN, where the first metal flowing into the
SEN strikes the bottom of the SEN and the subsequent backwash results in the
formation of a vortex (Fig. 5.10a) [55, 56]. This vortex, subsequently, interferes,
intermittently, with the metal flow through each port. This interference causes both
fluctuations and asymmetry in the flows leaving the SEN [57, 58]. The flow in the
mould is continually reacting to the current situation at any one time. It has been
5.7 Fluctuations in Processes 161

(a) (b) (c)

Fig. 5.10 Schematic drawings showing (a) the fluid flow in the SEN and the formation of a
vortex in the sump of the SEN [55] (b) frequencies of oscillations [57] and (c) formation of
periodic backflows in the mould during billet-casting resulting in periodic bouts of slag entrapment
[58]; (permission granted, ISIJ (a) [55] (b) [17])

suggested that a catastrophic event (such as a breakout) can occur when all the
factors are aligned together (i.e. “the butterfly effect” [59]).These fluctuations in
metal flow velocity are also responsible for the movements of both surface waves
and the vortices (like that shown in Fig. 5.4a).
In billet casting, the metal flows freely from an open SEN but it has been shown
that fluctuations and vortex-formation also occur, intermittently, in this system and
cause periodic “backflows” as shown in Fig. 5.10c which results in periodic bouts
of slag entrapment [58].
When the metal flow hits the narrow face of the mould, the impact causes a wave
to be formed on the surface of the liquid metal (a small wave can be seen in
Fig. 5.11a); the position and the height of the waves fluctuate continually [60].
Vortices can also be formed as a result of fluctuating, asymmetric metal flows, an
example can be seen in Fig. 5.4a and these vortices are continually changing
position [60]. Vortex-formation and turbulent metal flow are major causes of the
slag entrapment [61, 62]. Furthermore, turbulent flow plays an important part in
(i) SEN erosion and (ii) Carbon pick-up (especially when casting low-carbon steel
grades).
In the mould, the flows in the metal are in a continual state of fluctuation and
these flows, at any instant, are responding to the flows in the previous time period.
The effect of these fluctuations can be clearly seen in the varying depths of the
oscillation marks shown near the top of Fig. 5.11b [13, 16]. Although, the aero-
dynamic drag forces from the flow in the metal affect the flow in the slag phase, the
slag flow is dominated by the movement of the mould (and slag rim) as shown in
Fig. 5.5. When the mould is descending the slag flow is warm (tropical) and
outward and upward and when the mould is descending the flow is cold (arctic) and
downward.
162 5 Effect of Casting Variables on Mould Flux Performance

Fig. 5.11 Schematic diagrams showing (a) formation of waves on the surface of the steel and
(b) predicted oscillation mark depth [13] (permission granted, ISIJ [13])

5.8 Application of Electromagnetic Devices

There are several electromagnetic devices used in continuous casting to improve

process control and product quality and these have been described in a number of
publications [35, 43–45, 62, 64–71].

5.8.1 Electromagnetic Stirring (EMS)

Electromagnetic stirring (shown in Fig. 5.12) is carried out with AC, usually at a
frequency of 10–50 Hz, with the electromagnetic effect penetrating through the
volume. A travelling magnetic field (i.e. the steel) induces an electric current in the
molten steel. The interaction between this current and the travelling magnetic field

Fig. 5.12 Schematic diagram

showing EMS applied to
mould [64] (permission
granted, Nippon Steel,
Sumitomo Metal (NSSM)
Corp. [64])
5.8 Application of Electromagnetic Devices 163

Fig. 5.13 Schematic diagrams showing the principle of EMC a conventional casting without
EMC b EMC [67] (permission granted, Nippon Steel, Sumitomo Metal (NSSM) Corp. [67])

produces a Lorentz force which drives the stirring of the steel [6, 63–65]. The loss
of magnetic flux density can be minimised by using (i) a frequency of 10 Hz and
(ii) using a low-conductivity, stainless steel plate attached to the copper plate [62,
64]. EMS has proved effective in providing (i) a reduction in CO blowholes [62, 64]
(ii) a fivefold reduction in inclusions and gas bubbles [6, 62, 64] and (iii) a threefold
reduction in dOM [64] (iv) homogenisation of the melt composition and temperature
(leading to decrease in longitudinal cracking [64]) and (iv) refining of the
microstructure (Fig. 5.13).
EMS of liquid metal causes Saffman forces to be exerted on the inclusions, the
magnitude of the force increasing with increasing inclusion size [6]. Thus EMS aids
the removal of non-metallic inclusions and minimises slag entrapment. Inclusion
separation will occur if the Saffman velocity exceeds the velocity of the solidifi-
cation front. EMS is successful in removing inclusions (and bubbles) with particle
diameters >100 lm [6].

5.8.2 Level Magnetic Field (LMF)

This technique is also used in the removal of inclusions and bubbles. It involves
applying a static, electromagnetic field (in the thickness direction) over the mould
width. The magnetic field induces a current in the conducting fluid (steel) which
interacts with the magnetic field so as to oppose the movement. Initially, LMF was
used to reduce the velocity of the metal flowing out of the SEN. However, when
LMF is applied lower down the mould it was found to reduce the penetration of the
metal flow into the mould (see Fig. 5.14a). This reduced penetration with LMF
promotes the flotation of inclusions and bubbles to the liquid slag pool where they
are absorbed [6, 35, 66]. LMF is used in EMBr [6, 62–64]. The reduction in the
164 5 Effect of Casting Variables on Mould Flux Performance

Fig. 5.14 Schematic diagrams showing arrangements for (a) Original EMBr [63] (b) Level
magnetic field [63] (c) Conventional EMBr [44] (d) FC control [44] (e) Mould fitted with Flow
control [44]; (permission granted, (a, b) JOM [63] (c, d, e) SEAISI [44])

penetration of the metal stream is also responsible for the ca. 10 °C increase in steel
meniscus temperature [43, 45, 69, 71] which results in an increase in powder
consumption.
LMF technology has been used to suppress mixing of two steels in the mould in
the casting of, say a LC steel with a coating of stainless steel [62–64]. The two
steels are held in separate tundishes and are fed through two SEN’s and LMF is
then used to separate the two layers of different steels in the mould.

5.8.3 Electromagnetic Casting (EMC)

EMC consists of an external, solenoid coil wrapped around a core at meniscus level,
to which a high-frequency AC is applied (Fig. 5.13) [67]. This induces both a
magnetic field and an induced current. The interaction of the electromagnetic field
and the induced current produces a “pinch force” which pushes the shell away from
the mould and thus increases the distance between shell and mould (i.e. it opposes
the ferro-static pressure).The increased gap results in, sequentially, a thicker slag
film and a reduced, horizontal heat flux which tends to delay the solidification of the
steel. This delayed solidification takes the shell tip away from the turbulent inter-
face which results in fewer entrapped inclusions and shallower oscillation marks
(due to the increase in drim/tip).
However, the steel flow, induced by the electromagnetic field in EMC, is so
rapid that it can cause the steel meniscus to become unstable and irregular [67]. In
5.8 Application of Electromagnetic Devices 165

order to combat this problem, pulsative EMC was introduced [67].This entails
passing an AC current through the solenoid, intermittently, at a frequency, of say,
10–30 Hz [67]. This allows the metal velocity (induced by the electromagnetic
field) to be controlled. The induced electromagnetic field interferes with the eddy
current sensors, so the mould levels have to be sampled in the periods when no
current is flowing in the solenoid coil [67].
EMC has been largely used in billet casting but pulsative EMC has been suc-
cessfully developed for slab casting [67]. It provides the following benefits (i) a
fourfold decrease in defects [67] (ii) a threefold decrease in surface-roughness (i.e.
dOM) [67] (iii) a tenfold decrease in surface inclusions [67] (iv) a 20% increase in
powder consumption [67] and (vi) a reduction in longitudinal cracking [6, 67].

5.8.4 Electromagnetic Braking (EMBr)

Electromagnetic braking uses DC and promotes resistance to the movement of the

metal flow. When an electromagnetic field is applied to a moving, electrically
conducting fluid, an electrical current is induced. The components of the induced
current, which cross the magnetic field at right angles, exert a braking effect on the
steel flow. EMBr was initially used to decelerate the steel stream discharging from
the SEN (Fig. 5.14a). Although this was effective in reducing the metal flow tur-
bulence and slag and gas bubble entrapment, problems related to the stability of the
braking effect (and the associated metallurgical benefits) were encountered [62].
Consequently, Level Magnetic field (LMF) technology (Fig. 5.14b) was developed
to achieve these targets. Further work resulted in the development of flow control
EMBr (FC) [43, 62, 64, 68] and other designs (e.g. 3-pole arrangement [70]).
Flow control (FC) is shown in Figs. 5.14d, e. The upper magnetic field (pole) in
Fig. 11.14e acts to: (a) decrease the metal velocity at the meniscus (thereby min-
imising slag entrapment in the steel below) and (b) stabilise the meniscus (thereby
reducing mould-level variations, slag entrapment by the shell, and transverse
cracking [43, 67].The lower magnetic field (pole) acts to decrease downward
penetration of the metal flow which (c) increases the meniscus temperature by ca.
10 °C [43, 45, 69, 71] (thereby increasing powder consumption) and (d) promotes
the flotation of inclusions and gas bubbles [43, 67]. It should be noted that EMBr
results in a significant decrease in vertical heat transfer which can have effects on
the depth of the molten pool and powder consumption [72].
EMBr has been reported to provide the following benefits (i) lower inclusion and
pinhole concentrations in the cast steel [35, 43, 44, 65, 66, 68] (ii) reduced levels of
longitudinal cracking [43, 44, 65] but also results in (iii) higher levels of SEN
erosion [44].
166 5 Effect of Casting Variables on Mould Flux Performance

5.9 Steel Grade

The behaviour of the mould slag during continuous casting differs according to the
steel grade being cast. For the most part, the mould slag is tailored to provide the
appropriate treatment for the transformation of d-ferrite into austenite in various
steel grades. However, recently new powders have been designed and developed to
cast high-Al steels in order to minimise the amount of Al2O3 formed during casting.

5.9.1 Peritectic Steels

Inspection of the Fe–C phase diagram (given in Fig. 5.15) shows that for steels with
compositions in the range (C% = 0.06–0.17) solidification of liquid steel results in
the formation of a d-Fe shell, which subsequently, undergoes a peritectic phase
transition to c-Fe (austenite) as shown in Eq. 5.6. Austenite shells are stronger but
less ductile than ferrite shells.

FeðliqÞ þ d  Fe ! c  Fe ð5:6Þ

Liquid

L+δ

L+γ

Fig. 5.15 The Fe–C phase diagram for the peritectic range (0.08 < wt% C < 0.55) calculated
using ThermoCalc [75]; (drawn Kimab, Swerea)
5.9 Steel Grade 167

The transition (L + d ! c) is accompanied by a volume change of (0.4–0.6%)

due to the tighter packing in the austenite (cf. ferrite) and there is also a 4%
mismatch between the thermal shrinkage coefficients of d-Fe and c-Fe [73].
Austenite is stronger and has better creep resistance than ferrite because of the
better-packing of atoms in the FCC structure. However, ferrite (bcc-structure) is
more ductile and exhibits less-micro-segregation than austenite [74].
The mismatch in the thermal shrinkage coefficients of d-Fe and c-Fe in peritectic,
MC steels results in stresses. These stresses are much larger in thicker regions of the
shell than in the thin regions and result in a hoop stress around the mould. This hoop
stress is relieved by longitudinal cracking. The solidification process occurs during
the period of the oscillation cycle when the mould is descending (in tn and early tp);
little solidification occurs in the remaining portion of the cycle when the mould is
ascending. This results in a corrugated “sausage-like” shell and the resulting vari-
ations in shell thickness lead to stresses. The usual strategy used to minimise lon-
gitudinal cracking, is to produce a shell which is both as thin and as uniform as
possible. This is usually achieved by reducing the heat flux and this, in turn, is
derived by using a powder which creates a thick and crystalline slag film [10]. The
slag film thickness increases with increasing solidification temperature (or Tbr).
In contrast, longitudinal cracking is much less prevalent in steels which do not
undergo the peritectic reaction (shown in Eq. 5.6). Consequently, it is necessary to
differentiate between those steels undergoing the peritectic reaction and those that
do not.
Wolf [76, 77] introduced the term, “ferrite potential” as a measure of the amount of
peritectic reaction occurring in the steel. Although, the ferrite potential is affected by
the carbon content, it is also affected by other alloying elements; some of these
elements stabilise the ferrite (Cr, W, Mo, Al and Si) whilst others stabilise the
austenite (Ni, Mn, Co, N and Cu). For low-alloy steels, the ferrite potential (FP) can be
calculated by Eq. 5.7 where CP is the carbon potential which is defined in Eq. 5.8.

Low alloy:
FP ¼ 2:5ð0:5  CPÞ ð5:7Þ

CP ¼ ðwt%CÞ þ 0:04ð%MnÞ þ 0:1ð%NiÞ þ 0:7ð%NÞ  0:14ð%SiÞ  0:04ð%CrÞ

 0:1ð%MoÞ  0:24ð%TiÞ
ð5:8Þ

For stainless steels the FP is given by Eq. 5.9 where “Ni “and “Cr” are defined
in Eqs. 5.10 and 5.11, respectively.
Stainless steel:


FP ¼ 5:26 0:74  00 Ni00 =00 Cr00 ð5:9Þ
00
Ni00 ¼ ðwt %NiÞ þ 0:31ð%MnÞ þ 22ð%CÞ þ 17:5ð%NÞ ð5:10Þ
168 5 Effect of Casting Variables on Mould Flux Performance

Fig. 5.16 The tendencies of

steels with regard to cracking
(denoted Depression) and
sticker breakouts as functions
of the ferrite potential [76, 77]
(permission granted,
ISS/AIST [76] and re-drawn
by Kimab, Swerea)

00
Cr00 ¼ ðwt%CrÞ þ 1:5ð%SiÞ þ 1:65ð%MoÞ þ 2ð%NbÞ3ð%TiÞ ð5:11Þ

The peritectic reaction is at a maximum at FP = 1.0 and the range, FP = (0.8–

1.05) corresponds to the crack-sensitive range (Fig. 5.16).
The properties of various steels, as characterised by their ferrite potentials, are
given in Table 5.2.
In addition to the Wolf routine, there are several other methods available for
determining the amount of peritectic reaction occurring in specific steels. These
various methods used to calculate the amounts of peritectic reaction have been
reviewed by Santillana [78]. These methods tend to follow the Wolf routine but use
different coefficients to calculate the carbon potential; these coefficients are given in
Table 5.3 [78]. However, thermodynamic software has also been used to identify
those steels where the peritectic reaction is likely to occur [78–80].
It was mentioned above, that peritectic (MC) steels are prone to longitudinal
cracking and are often denoted as “crack-sensitive”. For these MC steels, stresses
are minimised by keeping the shell as thin and as uniform as possible. This is
achieved by reducing the horizontal heat flux (using a flux which gives a thick (i.e.
using a slag with a high Tbr) and crystalline slag film).
In contrast, the shells formed by HC steels tend to be weak and are prone to
“sticker breakouts”. The strategy used here is to create a thick, strong shell by
increasing the horizontal heat flux through the creation of a thin (i.e. low Tbr)
glassy, slag film.
A plot of viscosity versus Tbr for mould fluxes (Fig. 5.17) is effectively a plot of
(inverse lubrication) versus (inverse heat transfer) [4, 5, 10]. Figure 5.17 was
derived from data for the fluxes used in a large number of casters to cast MC and
HC steel grades [5]. It can be clearly seen that for a given viscosity the Tbr values

Table 5.2 Characteristics of shells formed for different bands of Ferrite Potential (FP)
FP Phase Properties of shell
<0.8 Austenite Strong and more ductile with
increasing FP
0.8–1.05 Initially forms ferrite, then transforms to Irregular, strong shell
austenite
>1.05 ferrite Weak, ductile shell
 5.9 Steel Grade

Table 5.3 Coefficients for selected equations to calculate Carbon potentials, after Santillana [78]
Ref. Mn Si N P S V Ti Cu Cr Ni Mo Others
Wolf 0.01 −0.1 −0.7 −0.04 0.04 −0.1
Trico 0.02 0.009 0.05 0.008 0.17 0.009 0.007 0.007 0.003 0.02 −0.007 Alt0.05
Nb 0.04
Sn0.0006
B 1.32
Ca-0.24
BSSTC 0.043 −0.14 1.06 0.029 0.11 −0.13 −0.024 0.037 −0.083 0.1 −0.003
SMS 0.14 −0.037 −0.04 0.222 0.003 0.023 −0.004
Howe 0.04 −0.14 0.7 −0.24 −0.04 0.1 −0.1
Alt = total Al
169
170 5 Effect of Casting Variables on Mould Flux Performance

Fig. 5.17 The break

temperature (Tbr) as a function
of slag viscosity for flux used
in the casting of slabs, blooms
and billets [5]; (permission
granted, UNESID [5])

for crack-sensitive MC grades are 100 °C higher than those used to cast HC grades
[4, 5]. This indicates that there has been a hidden logic underlying the development
of mould fluxes. The Tbr values of mould fluxes used to cast other steel grades tend
to fall between these two curves. Thus Tbr values for three categories of steels can
be expressed as a function of viscosity (in dPas) shown in Eqs. 5.12 to 5.14.
Thus, it is possible to differentiate various categories of steels in terms of their
carbon potentials or ferrite potentials [2, 76, 77].

Crack-sensitive:

CP ¼ 0:06  0:18 Tbr ð CÞ ¼ 1157 þ 60 ln g ð5:12Þ

Sticker-sensitive:

CP [ 0:4 Tbr ð CÞ ¼ 1051 þ 76:4 ln g ð5:13Þ

Others (intermediate):

CP 0  0:06 and 0:18  0:4 Tbr ð CÞ ¼ 1103 þ 68:5 ln g ð5:14Þ

5.9.2 High-Al Steels

High-Al steels combine good strength (from the martensite formed during plastic
deformation of retained austenite) with high ductility (from the retained austenite).
However, these steels have proved difficult to cast (giving rise to sticker breakouts,
formation of depressions, erratic powder melting and to false sticker-breakout
alarms). All of these casting problems are traceable to the large amount of Al2O3
formed via reaction 5.15 where the underline indicates in the steel.
5.9 Steel Grade 171

2Al þ 3SiO2 slag ¼ 3Si þ Al2 O3slag ð5:15Þ

The alumina pick-up by the slag is typically 4% when casting normal (low-Al)
steels and about 2% arises from the reaction shown in Eq. 5.15 and ca. 2% from
steelmaking reactions [81]. However, for steels containing 1% Al, the Al2O3
pick-up can be greater than 30% due to the reaction of Al and SiO2 in the slag. The
reaction of Al is not confined to SiO2 since similar reactions occur with most slag
constituents (e.g. FeO, MnO, B2O3, etc.) lying above the Al/Al2O3 curve in the
Ellingham diagram (see Fig. 6.21). Large amounts of Al2O3 in the slag:
• Cause an increase in both viscosity (thereby decreasing the lubrication supplied)
and the liquidus temperature of the slag.
• Are slow to dissolve and leave solid particles in the slag which agglomerate and
lead to large slag rims which fracture and, subsequently, cause depressions,
sticker breakouts and false alarms on sticker-detector systems.
• Tend to produce highly crystalline slag films which provide little lubrication to
the shell and furthermore, tend to fracture (where the fractured slag films/rims
can lead to blockages, depressions, and sticker breakouts).
It has been reported that conventional fluxes can be used for casting steels with
 0.6% Al [41]. However, for casting steels with  0.6 Al %, it is necessary to use
some of the specialised powders which have been developed in recent years (see
Sect. 6.3.12). The most promising powders appear to be those based on calcium
aluminates with added fluxes (CaF2, Na2O) and with the SiO2 content limited to ca.
<10% to minimise formation of Al2O3 by the reaction shown in Eq. 5.15.

5.10 Water Flow Rate

In the past, the rate of water cooling has been reduced to produce mild cooling
conditions in a quest to reduce longitudinal cracking. However, more recent studies
have shown that the effect of reducing water flow rate tends to be small, as can be
seen from the plant measurements [82] shown in Fig. 5.18.

Fig. 5.18 The effect of water 18

cooling flow rate on the
thickness of the shell for two
different superheats,
(20 °C = blue and 60 16
dshell, mm

°C = red) during casting of

stainless steel 304 (- - -)
and CS 1026(──); 14
Vc = 0.9 m min−1; (permission
granted ISS/AIST [82];
re-drawn)
12
2.2 2.6 3 3.4 3.8
Water cooling rate, 10 3 litre min-1
172 5 Effect of Casting Variables on Mould Flux Performance

Fig. 5.19 Schematic diagram 0.8

showing the effect of Argon
flow rate on the powder

-1
consumption (Qt) [84]

, kg tonne
(courtesy of AB Fox [84]) 0.6

powd
0.4

Qt
0.2
0 1 2 3 4 5
-1
Ar flow rate, l min

5.11 Argon Flow Rate

Argon is introduced into the SEN to minimise clogging of the SEN and to aid the
removal of inclusions from the steel. It has been reported [83] that the point of exit
for the bubbles lies close to the SEN for large bubbles and moves further to the edge
of the mould as the bubble size decreases. There is a possibility that small bubbles
could be swept with the metal flow.
The argon flow rate has also been reported to affect the metal flow in the mould.
Argon flow has been reported to have a “cushioning effect” on the metal flow and
thereby retards the flow [83]. Recent studies [51, 52] have indicated that when
Argon flow rates were increased from 4 to 5 litre min−1, the metal flow changed
from an inward surface flow (i.e. mould wall ! SEN) to an outward surface flow,
which is less desirable.
It has also been reported that powder consumption increased with increasing
argon flow rate (Fig. 5.19) [84]. One possible explanation is that the argon bubbles
enhance vertical heat transfer from the metal to the slag pool which results in a
higher temperature for the slag pool and a reduced slag viscosity.

References

1. S Ogibayashi, T Mizoguchi, T Kajatani, Intl. Workshop on Thermophys. Data for the

Development of Mathematical models of solidification, Gifu City, Japan, (1995).
2. AB Fox, PhD Thesis “Mould fluxes- their properties and performance” Imperial College,
London, (2003).
3. F Neumann, J Neal, MA Pedroza, AH Castiliejos, E FA Acosta G, Proc.. 79th Steelmaking
Conf., 1996, (ISS, Warrendale, PA, 1996) p. 249.
4. KC Mills, S Sridhar, AS Normanton, ST Mallaband, Proc. Brimacombe Conf., Vancouver,
2000, (Can. Inst. Min. Met. & Petrol., 2000), p. 781.
5. S Sridhar, KC Mills, V Ludlow, ST Mallaband, Proc. 3rd Europ. Conf. Continuous Casting,
Madrid, 1998, (UNESID, Madrid,1998) p. 807.
References 173

6. T Matsumiya, ISIJ Intl., 46, 1800, (2006).

7. MM Wolf, Iron and Steelmaker, 23, (2), 47, (1996).
8. CS Li, BG Thomas, Proc. ISS Tech Conf., Indianapolis, 2003, (ISS, Warrendale, PA, 2003)
p. 685, see also Met. Mater. Trans., 35B, 1151, (2004).
9. M. Wolf, Proc. 2nd Europ. Conf. Continuous Casting, Dusseldorf, 1994, METEC
Congress’94, (VDEh, Dusseldorf, 1994) vol. 1 p. 78,
10. KC Mills, AB Fox, ISIJ Intl., 43, 1479, (2003).
11. S Ogibayashi, K Yamaguchi, T Mukai, T Takahashi, Y Mimura, K Koyama, Y Nagano,
T Nagano, Nippon Steel Technical Report, 34, 1, (1987).
12. RJ O’Malley, J Neal, Proc. Intl. Conf. on New Developments in Metallurgical Process
Technol., Dusseldorf, METEC Congress (VDEh, Dusseldorf, 1999) p. 73.
13. PE Ramirez-Lopez, PD Lee, KC Mills, B Santillana, ISIJ Intl., 50, 1797, (2010).
14. H Tai, M Morashita, T Miyake, Proc 3rd Conf. Continuous Casting, Madrid, 1998, (UNESID,
Madrid, 1998) p. 447 see also T Miyake, K Nakayama, M Moroshita and H Tai, Kobelco
Technol. Review, 21, (April), 7, (1998).
15. ES Szerkeres, Iron and Steel Engineer, 73, (7), 29, (1997).
16. PE Ramirez-Lopez, KC Mills, PD Lee, B Santillana, Met. Mater. Trans., 66, 109, (2011).
17. PE Ramirez-Lopez, KC Mills, PD Lee, B Santillana, ISIJ Intl., 50, 425, (2010).
18. P E. Ramirez-Lopez, P D. Lee, KC. Mills, C Puncreobutr, D Farrugia, B Santillana,
U, Sjöström, Proc 7th Europ. Conf Continuous Casting, Dusseldorf, (2012) (VDEh,
Dusseldorf).
19. Y Meng, BG Thomas, Proc. ISS Technol., Indianapolis, (2003) p. 589 see also Met. Mater.
Trans., 34B, 685, (2003).
20. K Schwerdtfeger, KH Tacke, Report EUR 9339 (Europ. Comm. Sci. Tech. Publ.,
Luxembourg, 1985).
21. S Ogibayashi, CAMP-ISIJ, 18, 126, (2003) and Proc. 75th Steelmaking Conf. (2002) p. 175.
22. JM Hill, YH Wu, B Witwatanapataphee, J Eng. Math., 36, 311, (1999).
23. R Saraswat, AB Fox, KC Mills, PD Lee, B Deo, Scand. J Metal., 33, 85, (2004).
24. H Maeda, CAMP-ISIJ, 6, 280, (1993).
25. PE Ramirez- Lopez, J Bjorkvall, U Sjostrom, PN Jalali, PD Lee, KC Mills, B Jonsson, J Janis,
M Petajajarvi, J Pirinen: Proc. 9th Intl Conf. on CFD on Minerals and Process Industries,
CSIRO, Melbourne, 2012. paper available www.cfd.com.au/cfd_conf12.
26. CA Dacker, T Sohlgren, Steel Research Intl., 81, 899, (2010).
27. BG Thomas, B Ho, G Li, Proc. McLean Symp., Toronto, 1998 (ISS, Warrendale, PA, 1998)
p. 177.
28. JA Kromhout, C Liebske, S Meltzer, AA Kamperman, R Boom, Ironmaking and
Steelmaking, 36, 291, (2009).
29. J Kromhout, RS Schimmel, Proc 8th Europ. Conf Continuous casting, Graz, 2014 (Austrian
Met. Mater. Soc., 2014).
30. Y Liu, XD Wang, M Yao, AB Zhang, H Ma, Z Wang, JC Ma, X Wang, GQ Shi, Ironmaking
and Steelmaking, 41, 748, (2014).
31. M Jenkins, Proc. 77th Steelmaking Conf., (1994), (ISS, Warrendale, PA, 1994) p. 337.
32. M S Jenkins, BG Thomas, Proc. 80th Steelmaking Conf. (1997) (ISS, Warrendale, PA, 1995)
p. 285.
33. JW Kim, SK. Kim, D. S. Kim, Y. D. Lee, J. I. Eum, E. S. Lee, Proc. 78th Steelmaking Conf.
(ISS, Warrendale, PA, 1995) p. 333.
34. S Chakraborty, W Hill, Proc. 77th Steelmaking Conf., 1994, (ISS, Warrendale, PA, 1994)
p. 389.
35. K Miyazawa, Sci. & Technol. Advanced Materials, 2, 59, (2001).
36. T Teshima, J Kubota, M Suzuki, K Ozawa, T Masoka, S Miyahara, Tetsu-to- Hagane, 79 (5),
576, (1993).
37. R Aigner, H Steinruck, EUROTHERM 82, Numerical Heat Transfer Gliwice-Cracow,
Poland, 2005, ed. A Nowak, RA Bialecki: (Silesian Univ. Technol.) p. 1076.
174 5 Effect of Casting Variables on Mould Flux Performance

38. G.G. Lee, H. Shin, S.-H. Kim, S.-K. Kim, W.-Y. Choi, B.G. Thomas, Ironmaking and
Steelmaking, 36, (1), 40, (2009).
39. V Guyot, JF Martin, A Ruelle, A d’Anselme, JP Radot, M Bobadilla, JV Lavant, JN Pontoire,
ISIJ Intl., 36, S 227, (1996).
40. H Takeuchi, H Mori, T Nishida, T Yanai, K Mukunashi, Trans. ISIJ, 19, 274, (1979).
41. J Kromhout, PhD Thesis, “Mould powders for high speed continuous casting of steel”
University of Delft, NL, (2011).
42. M Cervantes, H Gustavsson, Proc. 3rd Europ Conf. Continuous casting, Madrid, 1998,
(UNESID, Madrid, 1998) p. 202.
43. M Washio, M Sugizawa, S Moriwaki, K Kariya, S Idogawa, S Takeuchi, Revue de
Metallurgie, CIT, 90, (April), 507, (1993).
44. H Take, H Osanai, J Hasunuma, T Yamamoto, H Bada, H Tozawa, Proc. Conf. “Quality of
ordinary steel in Iron- and steel-making”, Bangkok, (1994) Session 3, Paper 1.
45. D W van der Plas, C Platvoet, B Diesesme, JP Radot, JM Galpin, Proc. 2nd Europ. Conf.
Continuous casting, Dusseldorf, 1994, Metec Congress’94 (VDEh, Dusseldorf, 1994) p. 109.
46. R Singh, BG Thomas, SP Vanka, Met. Mater. Trans. B, 44 B, 1201, (2013).
47. M Morishita, CAMP- ISIJ, 20, 867, (2007).
48. D Gotthelf, P Andrezjewski, E Julius, H Haubrich, Proc. 3rd Europ. Conf. Continuous
Casting, Madrid, 1998, (UNESID, Madrid, 1998) vol. 2 p. 825.
49. BG Thomas, LF Zhang, ISIJ Intl., 41, 1181, (2001).
50. QK Robinson, DM Gerstl, Proc. 3rd Europ. Conf. Continuous Casting, Madrid, 1998,
(UNESID, Madrid, (1998) p. 1050.
51. PE Ramirez- Lopez, PN Jalali, J Bjorkvall, U Sjostrom, C Nilsson, ISIJ Intl. 54, 342.
(2014) and Proc. 8th Europ. Conf. Continuous Casting, Graz, Austria, 2014 (ASMET, 2014).
52. E van Vliet, DW van der Plas, SP Carless, A A Kamperman, AE Westendorp, Proc.. 7th
Europ. Conf. Cont. Casting, Dusseldorf, (2011), Session 4.
53. R Choudhary, BG Thomas, SP Vanka, Met. Mater. Trans. B, 43B, 532, (2012).
54. S Yokoya, S Tagaki, M Iguchi, K Marukawa, S Hata, Proc. 4th Europ. Continuous casting
Conf., Birmingham, 2002 (IOM, London, 2002) vol 1, p. 149.
55. C Real, R Miranda, G Vulchis, M Barron, L Hoyes, J Gonzalez, ISIJ Intl. 46, 1183, (2006).
56. A Ramos-Banderas, R Sánchez-Pérez, RD Morales, J Palafox-Ramos, L Demedices-García,
M Díaz-Cruz, Met.. Mater. Trans. B, 35B, 449, (2004).
57. PE Ramirez-Lopez, PD Lee, KC Mills, ISIJ Intl., 50, 425, (2010).
58. E Torres-Alonso, RD Morales, S Garcia-Hernandez, A Najera- Bastida, A Sandoval- Ramos,
Met. Mater. Trans. B, 39B, 840, (2008).
59. PD Lee, PE Ramirez-Lopez, KC Mills, B Santillana, Ironmaking and Steelmaking, 39, 244,
(2012).
60. KC Mills, PE Ramirez-Lopez, PD Lee, High Temp. Materials & Processes, 31, 221, (2012).
61. H Nakato et al, Continuous Casting, vol 6, (ISS, Warrendale, PA, 1992) p. 193.
62. S Feldbauer, I Jimbo, A Sharan, K Shimizu, W King, J Stepanek, J Harman, AW Cramb,
Proc. 78th Steelmaking Conf., 1995, (ISS, Warrendale, PA, 1995) p. 655.
63. E Takeuchi, JOM, 1995 (May), 42, (1995).
64. E Takeuchi, H Harada, H Tanaka, T Ishii, T Toh, M Zeze, M Hojo, K Shigematsu, Nippon
Steel Technical Report, 61, 29, (1994).
65. E Favre, S Kunstreich, MC Nove, D Rotelec, W Courths, E Korte, Proc. 3rd Intl Conf.
Continuous Casting, Madrid, 1998, (UNESID, Madrid, 1998) p. 595.
66. P Gardin, JM Galpin, MC Regnier, JP Radot, IEEE Trans. on Magnetics, 31, 2088, (1995).
67. M Tani, T Toh, K Umetsu, K Tanaka, M Zeze, K Tsunenari, K Hayashi, S Fukunaga, Nippon
Steel Technical Report, 104, 62, (2013).
68. SG Kollberg, HR Hackl, PJ Hanley, Iron and Steel Engineer, 73, (7), 24, (1996).
69. G Bocher, U Hoffman, P Muller, Proc. 2nd Europ. Conf. Continuous casting, Dusseldorf,
1994, Metec Congress’94 (VDEh, Dusseldorf, 1994) p. 103.
70. TWJ Peeters, R Koldwijn, JA Kromhout, AA Kampermann, Proc. 5th Conf Continuous
Casting, Nice, 2005, (La Revue Metall., Paris, 2005) Vol 2, p. 515.
References 175

71. MY Ha, SG Lee, SH Seong, J. Mater. Processing Technol., 133, 322, (2003).
72. R Koldewijn, Unpublished Corus Internal Rept. (2007) cited in KC Mills, J Kromhout,
A Hamoen, R Boom, Proc. Admet Conf., Dnipetrovsk, Ukraine, May 2007, (Natl. Metall.
Acad. Ukr., Dnipropetrovsk, 2007) vol 2, p. 174.
73. SN Singh, KE Blazek, J Metals, 26, (10), 17, (1974).
74. E Schmidtmann, L Pluegel, Archiv. fur Eisenhuttenw., 51, 49, (1980).
75. Thermocalc, www.thermocalc.com.
76. M Wolf, W Kurz, Met. Trans., 12, 85, (1981).
77. M Wolf, Proc. 81st Steelmaking Conf.,1998, (ISS, Warrendale, PA, 1998) p. 53.
78. MB Santillana, PhD Thesis: Thermo-mechanical properties and cracking during solidifica-
tion of thin slab cast steel. Univ. Delft, NL (2013) Chapter 3.
79. KE Blazek, O Lanzi, P Gano, D Kellog, Proc. AIST Tech. Conf., 2007, PR-351-141 – 2007.
80. AA Howe, PhD Thesis, Micro-segregation in multi-component steels involving the peritectic
reaction, Univ. Sheffield, (1993).
81. G Skoczylas, Proc. 79th Steelmaking Conf., 1996, (ISS Warrendale, PA) p. 269.
82. MJ Lu, JH Chen, DH Tsai, CH Huang, Proc. 78th Steelmaking Conf., 1995,(ISS, Warrendale,
PA) p. 359.
83. K Pericleous –private communication Univ. Greenwich, UK (2008).
84. T Mallaband (Metallurgica, UK) Unpublished results from cited by AB Fox, ref. 5–02.
Chapter 6
Different Types of Mould Powders

Abstract The mould powder must carry out a series of tasks. Arguably, the most
important of these tasks is the formation of a slag film which provides the optimum
level of lubrication and heat extraction from the shell. These properties are deter-
mined by the (i) mould dimensions (ii) the casting conditions and (iii) the steel
grade being cast. Empirical rules have been developed to express the properties
providing good casting performance; these include the required values for the
powder consumption, slag viscosity and the break temperature but the fraction of
crystal phase in the slag film (fcrys) should also be cited. The various types of mould
powders are described, e.g. starter, exothermic and prefused powders. Most mould
powders in use are denoted as Conventional; the properties of these powders would
be expected to be (i) consistent with those derived from empirical rules and (ii) to
contain fluorides to form cuspidine in the slag film. However, there are a number of
specialist powders which have been developed to carry out specific tasks or to cover
special steel grades or casting conditions. The specialist mould powders include the
following: F-free powders; C-free powders, Non-Newtonian powders, and fluxes
used to cast high–speed, thin slabs and round billets and powders used to cast steels
with high Al and rare earth contents and stainless steels. Although the properties of
all these fluxes are consistent with those predicted by the empirical rules, devel-
opments have been made to deal with special features for each type of powder.

Symbols, Abbreviations and Units

D Diameter (m)
d Thickness (m)
k Thermal conductivity (Wm−1K−1)
QMR Melting rate (kg s−1 or kg min−1)
Qs Powder consumption (kg m2(mould))
q Heat flux density (Wm−2)
t Thickness of mould (m)
w Width of mould (m)
c Surface tension (mNm−1)
η Viscosity (dPas)

© Springer International Publishing AG 2017 177

K.C. Mills and C.-Å. Däcker, The Casting Powders Book,
DOI 10.1007/978-3-319-53616-3_6
178 6 Different Types of Mould Powders

T Temperature (°C)
Tbr Break temperature (°C)
Tliq Liquidus temperature (°C)
Tsol Solidification temperature (°C)
Vc Casting speed (m min−1)
C/S %CaO/%SiO2 = basicity
EMBR Electromagnetic Braking
HC High carbon (steel)
IR Infrared
LC Low carbon (steel)
MC Medium carbon (steel)
SEN Submerged entry nozzle
ULC Ultra-low carbon (steel)

In chemical formulae for slags and minerals, the following abbreviations are
used:
A Al2O3
B B2O3
C CaO
F FeO
Fl CaF2
K K2O
L Li2O
M MgO
Mn MnO
N Na2O
S SiO2
T TiO2

6.1 Introduction

Mould powders were first used successfully in bottom-pouring castings in Belgium

in 1958. These powders were based on fly ash (a waste product in power genera-
tion) and they were applied to continuous casting in 1963. It was found that the use
of these powders reduced the heat losses significantly (cf. oil casting) and thus, the
use of powders led to a significant decrease in the superheat of the steel.
In this chapter, we discuss the different types of mould powders which are
available. These include conventional (F-containing) powders used for casting
slabs, blooms and billets as well as specialist powders developed for casting specific
steels or for very different casting conditions. The mould powders used to cast steels
under normal conditions are denoted as “Conventional” in the following text; such
6.1 Introduction 179

powders usually contain CaF2 to form the phase cuspidine in the slag film. Other
powders have been developed to deal with specific casting conditions or for
casting-specific steel grades; these will also be discussed below.
First, it is necessary to establish the functions and criteria which conventional
fluxes must fulfil. Then, the properties and compositions will be examined to
identify those providing a satisfactory performance of these tasks. When this has
been successfully applied to conventional mould powders, a similar analysis can
then be carried out for mould powders for the specialist casting.

6.1.1 Functions Carried Out by Mould Powder

The principal functions of the mould powder (discussed below with the aid of
Fig. 6.1) are
(i) To prevent the steel meniscus from freezing (i.e. reducing vertical heat
transfer).
(ii) To seal the liquid steel from the atmosphere to prevent oxidisation of the
metal.
(iii) To lubricate the newly formed shell.
(iv) To control the horizontal heat transfer from the shell to the mould.
(v) To absorb inclusions from the steel and to minimise slag entrapment.
Although these are the primary functions, there are other secondary tasks which
would be very welcome, if provided by the mould flux, namely
(vi) To minimise the pick-up of carbon from the casting powder.
(vii) To minimise SEN erosion rates.

Fig. 6.1 Schematic drawing

showing various layers
formed in the powder bed and
the slag film with liquid
phases and glassy portion of
slag film depicted in white
and crystalline portion in grey
(in practice, glassy and
crystalline phases are mixed
in the slag film) [1]
180 6 Different Types of Mould Powders

6.1.2 Criteria Affecting Selection of Mould Powders

Mould powders are selected to provide the various functions listed above. When
selecting a mould powder there are several criteria which must be addressed. Some
compromises are needed in this selection process; for instance, a high-slag viscosity
helps to reduce slag entrapment but will reduce the powder consumption (i.e.
lubrication supplied to the shell). The various criteria affecting the selection process
are summarised below (with particular emphasis on those criteria influenced by
mould powder properties).

6.1.2.1 Thermal Insulation Provided by the Powder Bed

In order to avoid the freezing of the steel (i.e. to prevent skull formation) it is
necessary to ensure that the bed provides the necessary thermal insulation. The
thermal resistance (Rth) of the bed (containing, powder, sinter and liquid layers) is a
measure of the thermal insulation and can be calculated from Eq. 6.1 where d is the
depth; or thickness, and k is the thermal conductivity. The temperature at the mould
powder surface (Tair/bed) usually lies between 200 and 400 °C and tends to increase
with decreasing bed depth [2, 3].

Rth ¼ ðd=kÞpowder þ ðd=kÞsinter þ ðd=kÞpool ¼ ðTsteel=pool  Tair=bed Þ=qvert ð6:1Þ

Vertical heat transfer from the surface of the steel is complex and involves a
number of heat transfer mechanisms, e.g. (i) convection, radiation and conduction
in the molten slag pool and (ii) conduction and gaseous convection in the powder
bed and sinter. Consequently, there are a number of factors affecting the vertical
heat flux and the thermal insulation, namely
(i) The depth of the bed (dbed) (with qvert # as dbed ").
(ii) The particle size of the mould powder (D) affects the packing of the bed and
the bulk density (qbulk) with qbulk increasing as D decreases. Smaller
particles have (a) higher conductivity (kpowder (Wm−1K−1) = 0.10 +
0.169  10−3qbulk [4, 5]) and (b) lower permeability (P) to gaseous con-
vection than bigger particles [6]. Thus, decreasing D results in an increase in
qlat and a decrease in qconv, i.e. they have opposing effects. Since tests show
that thermal insulation increases with decreasing particle size, gaseous con-
vection must be the dominant mechanism in the powder bed.
(iii) The reaction of O2 with carbon particles in the mould powder is exothermic.
The heat given out causes increases in local temperatures and causes a
decrease in the thermal gradient in the bed and in qvert. There is ca. 3  more
energy given out in the formation of CO2 cf. conversion to CO(g).
(iv) The introduction of exothermic agents (e.g. Ca/Si or Fe/Si) into the powder
has the same effect as carbon, in that, heat generated by combustion, results
6.1 Introduction 181

Fig. 6.2 DTA trace showing the various thermal events during heating of a mould powder [8]

in a local increase in temperature which reduces the temperature gradient and

hence, the vertical heat flux.
(v) Heat transfer in the liquid pool is dominated by convection and radiation (in
some models, keff = C klat is adopted with values of C varying between 10
and 100 and with a recommended value of 15 [7]. The vertical heat transfer
can be reduced by increasing (i) the slag viscosity (qconv #) or (ii) the FeO or
MnO % in the slag (qR #).
It should be noted that that there are a number of thermal events which occur at
various levels in the powder bed (Fig. 6.2, [8]) namely
(i) The endothermic decomposition of carbonates in the range, 300–400 °C.
(ii) The exothermic oxidation of carbon particles to CO(g) and CO2(g) between
500 and 700 °C and any exothermic agents present.
(iii) The endothermic sintering of particles at ca. 800 °C (this includes the
enhanced Cp values for the supercooled liquid).
(iv) The endothermic melting of the powder at 1100–1200 °C.
Exothermic events will lead to a local heating resulting in decreases in both the
local temperature gradient and hence qvert. Endothermic events cause local cooling
and increase in both the local temperature gradient and qvert.

6.1.2.2 Formation of Molten Slag Pool

The molten slag pool must protect the steel from reactions with O2, N2 and CO2(g).
Thus, it is important that the pool should have sufficient depth to prevent the
standing wave in the steel meniscus (resulting from turbulent metal flow) from
poking into the powder bed (Fig. 6.3) which contains these gases.
The principal factors affecting the depth of the pool are (i) the vertical heat flux
(dpool " as qvert ") and (ii) the melting rate (dpool " as QMR "). The vertical heat flux,
in turn, is affected by the depth of the powder bed, the particle size, and the
presence of exothermic agents in the mould flux. It is also affected by factors like
182 6 Different Types of Mould Powders

Fig. 6.3 Schematic drawing showing standing waves in the steel meniscus close to contact with
the sintered layer (permission granted, ISS/AIST [9])

electromagnetic braking (EMBr) which affects the efficiency of vertical heat

transfer. The principal properties of the mould powder affecting the melting rate are
the C% and to a lesser extent, the size of the carbon particles (QMR" as %C # and
DC ") the bulk density of the powder and the efficiency of heat transfer.

6.1.2.3 Slag Infiltration and Powder Consumption

The infiltration of liquid slag into the shell/mould gap provides the essential
lubrication of the shell. The powder consumption, Qs (in kg m−2 [of mould]) is
equal to (q.dl) where q is the density and dl the thickness of the liquid layer of the
slag film (see Chap. 2). Empirical rules have been developed to calculate a value for
Qs which provides a satisfactory level of lubrication [10, 11]. This required powder
consumption (Qreqs ) is very sensitive to mould dimensions [width, (w) and thick-
ness, (t)] but can be calculated from Eq. 6.2, where R* is defined in Eq. 6.3 [12].


Qreq
s kgm2 ¼ 2=ðR  5Þ ð6:2Þ

R ¼ 2ðw þ tÞ=w t ð6:3Þ

This calculated value of powder consumption can be related to the casting speed
(Vc) and the slag viscosity at 1300 °C (η) through the empirical, Eq. 6.4 [12],
thereby, providing the required slag viscosity for the given casting conditions (w, t
and Vc).

s req ¼ 0:55=g
Qslag  Vc ð6:4Þ
0:5
6.1 Introduction 183

Low powder consumption can arise when casting steels containing Ti, or when
using mould fluxes containing significant concentrations of TiO2, because solid TiN
(or TiCN) particles can be formed (which have a low solubility in the slag) and the
solid particles block the infiltration of molten slag [13, 14] and increase the slag
viscosity. This problem is discussed [12] below in Sects. 6.2.6 and 6.2.13 where the
superscript slag refers to the liquid slag {and Qslag = f* Qpowd where f* is the
fraction of powder forming liquid slag (see Chap. 2)}. Thus, it is possible to specify
the slag viscosity needed for the casting conditions [12].
It is important that Qslag
s req should not be constrained by the rate of the melting
rate, QMR which is linked to Qs via Eq. 6.5.

1
Qslag
MR ðkg s Þ ¼ 2ðw þ tÞQs Vc =60 ð6:5Þ

As seen above, the melting rate is principally controlled by (i) the % free carbon
in the powder (ii) to a less extent, by the particle size of the carbon (DC), i.e. (QMR #
as (% Cfree) " and DC #) and (iii) the vertical heat flux (QMR # as qvert #).
Consequently, it is important to ensure that the Cfree % is low enough to ensure that
the actual QMR is greater than the value of Qslag MR derived from Eq. 6.5.

6.1.2.4 Control of the Horizontal Heat Flux

Although the horizontal heat flux is affected by many factors, it is principally

controlled by the slag film formed between the shell and the mould (Chap. 3).
The following properties are used to control the heat flux:
(i) the solidification or break temperature (Tbr) of the slag which affects the
thickness of the solid layer (ds) of the slag film (qhor # as ds ").
(ii) the fraction of crystalline phase (fcrys) formed in slag film since high values of
fcrys result in increases in both the interfacial thermal resistance (RCu/sl " as
fcrys ") and reflection and scattering of radiation conduction (qR # as fcrys ").
In practice, qR and qhor, are reduced by increasing the basicity (C/S) of the
powder since both the thickness of solid slag layer (ds) and the fraction of crys-
talline phase of the slag film (fcrys) increase with increasing (C/S) ratio.
The appropriate level of horizontal heat flux (qhor) to avoid longitudinal cracking
varies according to the type of steel being cast [16, 17]; this is shown in Fig. 6.4 in
which the lowest curve for the heat flux pertains to MC, peritectic steels and the
higher curve for qhor refers to casting LC steels which have a weak shell. The
critical values of the required qhor for HC steel grades lie at higher qhor values.

6.1.2.5 Absorption of Inclusions and Minimising Slag Entrapment

Non-metallic inclusions reduce the mechanical strength of the steel. Thus, it is

important to minimise the inclusion content of the steel. Inclusions arise from
184 6 Different Types of Mould Powders

Fig. 6.4 Index of

longitudinal cracking as a
function of heat flux (qhor),
curve at lower qhor value and
D refer to MC peritectic steels
and curve for higher qhor
values and • are for a LC
steel; (permission granted,
ISS/AIST [16])

(i) oxidation reactions in steelmaking and casting (e.g. formation of Al2O3) [18] and
(ii) entrapped slag. The objective here is to transfer as many of the inclusions as
possible from the steel to the slag phase, where they can dissolve.
There principal strategy for reducing entrapment of inclusion particles (like
Al2O3) is to minimise the length of the solidified meniscus (Fig. 6.5a, b). This
reduces the probability of capture by the shell [19] and allows more time for the
inclusion to migrate to the slag phase (Fig. 6.5). A short solidified meniscus can be
achieved by reducing the vertical heat transfer by increasing the depth of the
powder bed or by using exothermic agents in the powders. However, it has also
been reported [20] that the number of slag inclusions trapped near the surface
decreases with increasing basicity; this was attributed to the increase in interfacial
tension (Fig. 6.5c) since cCaO  2cSiO2 where c is the surface tension [21].

Fig. 6.5 Schematic drawings showing a short solidified meniscus and b long meniscus and
showing the effect of meniscus length on the capture of inclusions and bubbles [19] c number of
slag inclusions as a function of basicity (C/S) of mould powder [20]; (permission granted,
a, b ISS/AIST [19] c ISIJ [20])
6.1 Introduction 185

Some inclusions like TiN (or Ti(CN)) and ZrO2 are not very soluble in the
molten slag (ca. 0.5 and 2%, respectively) and consequently, solid particles tend to
agglomerate through collisions in the slag pool. These agglomerates can block the
entrance to the mould/shell channel and thereby reduce the powder consumption
[13, 14]. In this case, the problem can be alleviated by increasing the volume of the
molten slag pool (by increasing bed depth or using exothermic powders). This
strategy is also used in casting TRIP steels where large amounts of Al2O3 are
formed and must be encouraged to dissolve.
The amount of entrapped slag can be reduced by either increasing the slag
viscosity (but this also decreases the powder consumption) or by increasing the
interfacial tension. The latter can be achieved by reducing the concentrations of
B2O3, K2O, Na2O, and SiO2 which have low surface tensions [21].

6.1.2.6 Carbon Pick-up by the Shell

Carbon pick-up by the steel is a particular issue when casting LC and ULC steel
grades. The principal sources of carbon pick-up are (i) the mould powder (ii) the
carbon floating near the surface of the slag pool and (iii) the slag rim which contains
glassy or amorphous carbon [22]. Carbon pick-up can be minimised by the
following:
(i) Maintaining a deep slag pool.
(ii) Reducing the carbon content of the powder.
(iii) Replacing the carbon with nitrides [22–24].

6.1.2.7 Erosion of SEN

The principal causes of SEN erosion are [25, 26]

(i) High flow velocities in the metal and slag phases which impact on the SEN
thus erosion rates decrease with increasing slag viscosity [25, 26].
(ii) The concentration difference (Csat − Cslag) is the driving force for dissolution
of the SEN refractory (e.g. ZrO2).
(iii) High CaF2 contents in the slag increase erosion rates (due to the attack of
oxides stabilising ZrO2); SEN erosion rates are lower in F-free slags [27] and
are affected by the ZrO2-stabiliser; erosion rates MgO > CaO > Y2O3 [26].
(iv) High O potentials for the slag increase erosion rates, so it is necessary to keep
FeO, MnO in the slag to a minimum.
In some cases, where SEN erosion is an important issue, about 2% ZrO2 can be
added to the mould powder to bring the concentration, CZrO2 , close to Csat; this
minimises SEN dissolution but also promotes the nucleation of crystalline phases
and increases Tsol.
186 6 Different Types of Mould Powders

6.1.2.8 Effect on Scale Formation

Certain mould slags adhere very closely to the surface of the strand and are not
removed in the secondary cooling zone. Pockets of adhering slag become the sites
where Fe2SiO4 is formed [28–38]; the Fe2SiO4 has a low melting temperature and
readily penetrates the grain boundaries of the steel thereby providing a “handhold”
for the scale to grip (see Sect. 11.11). Good adhesion between slag and steel is
promoted by powder constituents with low surface tension (particularly, K2O and
Na2O) which lower both the slag surface tension and interfacial tension [21]. Some
mould powders, which fully meet the criteria for lubrication and heat transfer, have
been rejected in thin-slab casting because of their effect on scale formation.

6.2 Selection of Mould Fluxes

Most mould flux compositions are selected to meet the criteria of

(i) Optimum powder consumption (i.e. lubrication of shell) for the given mould
dimensions and casting conditions.
(ii) Optimum level of horizontal, heat flux extraction for the steel being cast.
It has been shown above, that the required powder consumption can be
expressed as a function of the required viscosity at 1300 °C (ηreq). Figure 6.6 shows
the relation between ηreq and the required break temperature (T req br ) for the MC
(crack-sensitive) steels and HC (sticker sensitive) steels. The X-axis of this figure
represents the reciprocal lubrication and the Y-axis the reciprocal of horizontal

Fig. 6.6 Required break temperature as a function of required viscosity for mould slags used in
casting slabs, blooms and billets; dotted curves: upper MC steels, lower HC steels; (permission
granted, ISIJ [17])
6.2 Selection of Mould Fluxes 187

heat transfer (i.e. thermal resistance) since the thickness of the solid layer increases
with increasing Tbr. The Eqs. 6.6–6.9 were obtained to calculate the optimum T req br
for the mould slag from the ηreq value.
The various steel grades can be classified in terms of the carbon potential (CP)
which is defined in Eq. 6.6 [39] and then values T req br can be calculated for the
appropriate steel grade (Eqs. 6.7–6.9).

CP ¼ %C þ 10  2ð2%Mn þ 4%Ni  10Si%:Þ ð6:6Þ

MC-steels CPð0:06  0:18Þ : Tbrreq ¼ 1157 þ 60 ln g ð6:7Þ

HC steels CPð [ 0:4Þ : Tbrreq ¼ 1051 þ 76:4 ln g ð6:8Þ

Other steels : Tbrreq ¼ 1103 þ 68:5 ln g ð6:9Þ

As mentioned above, although most mould slag compositions are selected to

br and η . However, sometimes a compromise is
give the appropriate values of T req req

required, for instance, if slag entrapment is an issue, the slag viscosity may be
increased to reduce the number of entrapped particles but this comes at the expense
of a lower powder consumption.
The different types of powder available are discussed below. They include the
powders used for regular casting and those powders used in the casting of specialist
steel grades or used to combat certain casting problems or defects.

6.2.1 Conventional Mould Powders

As mentioned above, the mould powder composition is determined principally in

terms of providing (i) the required mould slag viscosity for the given casting
conditions and mould geometry and (ii) the optimum level of horizontal heat flux
(or thermal resistance of the slag film) which can be represented by the break
temperature of the slag. These conditions can be seen as minimising the chances of
obtaining longitudinal cracks and sticker break-outs. Thus the composition of the
powder is selected to give the required values of viscosity and Tbr. The fraction of
crystalline phase in the slag film should also be taken into account but the actions
taken to increase fcrys are identical to those for increasing Tbr.
The various mineral constituents can be classified into three groups, namely
(i) Network formers which includes SiO2 and Al2O3
(ii) Network breakers which includes CaO, MgO, FeO, MnO and occasionally
BaO
(iii) Fluxes to reduce Tliq and viscosity: Na2O, K2O, Li2O, CaF2 and B2O3.
The effects of individual constituents on the properties and the formation of glass
or cuspidine phases (C3S2Fl) are summarised in Table 6.1.
188 6 Different Types of Mould Powders

Table 6.1 The effect of increases in individual slag constituents on their properties and structure
Oxide/fluoride Role Promotes Effect on
properties
η Tbr fcrys
SiO2 Network former Glass formation—also C3S2Fl " # #
Al2O3 Network former Glass formation—high " # small
Al2O3 ! crystalline #
large
"
CaO Network C3S2Fl formation # " "
breaker
MgO Network >7%—glass formation # # "?
breaker
Na2O, K2O, Li2O Flux & breaker >4%—glass formation # # #
CaF2 Flux Cuspidine, C3S2Fl # # "
B2O3 Flux Glass formation # # #
FeO, MnO Network Glassy phases but also reduces kR # # #
breaker
ZrO2 Nucleates crystals " " "

Mould powders tend to have the following characteristics:

(i) The basicity, (denoted here as (C/S), i.e. (= %CaO*/%SiO2) where CaO* is
the total Ca content expressed as CaO) is often cited since it provides a
measure of both the structure of the slag and the viscosity. Values of (C/S) for
mould slags tend to fall within the range (0.6–1.3) but recently, experimental
fluxes with C/S ratios as high as 1.7 have been used. The (C/S) ratio of the
powder tends to be determined by the type of steel and casting speed, e.g.
for (i) C/S = 0.6 for low-castings speeds and HC steels (ii) C/S = 0.9–1.0
for LC, ULC steels and casting speeds around 1–1.5 m min−1 and
(iii) C/S = 1.3–1.7 for MC steels and high casting speeds.
(ii) The fluxes are added to reduce both the liquidus (Tliq) and the solidification
(or “break”, Tbr) temperatures of the slag. However, they also reduce the
viscosity of the liquid slag.
(iii) The FeO contents are usually low since they increase the oxygen potential
of the slag but both FeO and MnO are used to reduce radiation conduction
when it is not possible to use a crystalline slag film (e.g. when casting round
billets) [40, 41]. The pick-up of FeO by the slag is typically around 1% but
MnO pick-up can be appreciable when casting some Mn-containing steel
grades but pick-up decreases with increasing basicity [41].
(iv) The B2O3 contents rarely exceed 5% because of the problems associated
with B-pick-up by the steel.
6.2 Selection of Mould Fluxes 189

(v) Other components, such as TiO2, are usually present as impurities; TiO2
might be expected to behave as a network former but, in practice, it reduces
the viscosity, so TiO2 is usually treated individually in the modelling of slag
viscosity. TiO2 has been used to form crystalline phases in fluorine-free
mould fluxes (i.e. as a replacement for cuspidine). However, Ti tends to
form high melting phases, like perovskite, (CaTiO3) and it is necessary to
keep (C/S) < 0.9 to ensure that CaTiO3 is not formed [42]; Ti can also form
Ti(CN) which tends to lead to sticker break-outs [15]. ZrO2 is sometimes
added (ca. 2–3%) to the slag to minimise SEN erosion rates but, both ZrO2
and TiO2 tend to increase the break temperature and consequently, may
reduce the lubrication supplied to the shell.
(vi) In mould slags, Fluorine tends to bond exclusively with Ca2+ and Mg2+ ions
and once these Ca–F bonds are formed, the CaF2 has little effect on the
structure or on structure-related properties, like viscosity. It is for this rea-
son, the parameters, NBO/T and Q (measures of the degree of depoly-
merisation and polymerisation, respectively) are usually calculated by
ignoring the CaF2 in the slag.
(vii) Carbon has a fairly low solubility in liquid mould slags [43] and it is
non-wetting to molten slag (Fig. 6.7). Thus, molten slag globules are unable
to agglomerate until the carbon particles have been oxidised. Thus the
melting rate decreases as (i) the % carbon in the mould flux increases and
(ii) the carbon particle size decreases [44–46]. Carbon is usually added to
regulate the melting rate (QMR in kg s−1 or kg min−1). So it matches
demand (which is determined by the mould dimensions (w, t) and casting
speed (Vc) as can be seen from the combination of Eqs. 6.2, 6.3 and 6.5.
Carbon was, fortuitously, present in the original fly ash. Powders for slab
casting usually contain ca. 4% Cfree but powders for billet casting can
contain up to 25% Cfree. The free carbon can exist in various forms (e.g.
graphite, carbon black, coke-breeze, etc.). Different forms of carbon tend to

Fig. 6.7 Schematic drawing showing separation of molten slag globules ( ) by graphite particles ( )
190 6 Different Types of Mould Powders

have different particle sizes and mould powders tend to contain a mixture of
different carbon types (and different particle sizes). As mentioned above, the
particle size of the carbon (DC) is important with the melting rate decreasing
as DC decreases but smaller particles have a high (surface area/mass) ratio
and will thus oxidise more rapidly than bigger particles. Since smaller
carbon particles retard the melting rate, it is obvious that the enhanced
separation is more significant than the rate of oxidation. The carbon also
helps to maintain a reducing atmosphere in the mould.
(viii) Some mineral constituents are present as carbonates, these decompose
around 400 °C to form oxides and CO2(g). The carbon contents are fre-
quently cited as %Ctotal and %Cfree; the carbonate content %CCO2 can be
calculated from the equation

%CCO2 ¼ f44=12gð%Ctotal  %Cfree Þ ð6:10Þ

(ix) The moisture content (% H2O) in the as-received powder is usually around
0.5%; it is important to ensure % H2O  1% since this can lead to hydrogen
evolution in the slag film which can lead to sticker break-outs [47–49]. The
hydration of calcium aluminate mould powders used in the casting of TRIP
steels has been reported to cause flaring in the mould and it was found nec-
essary to bake the powder at 250 °C prior to its use in order to eradicate the
hydration [50].
(x) Zirconia, ZrO2 additions (2–3%) are used (a) to nucleate crystals and increase
Tbr and (b) to minimise SEN erosion.

6.2.1.1 Mineralogical Constituents

The first mould powders to be produced were based on fly ash, to which fluxes,
such as, Na2O and B2O3 were added. These powders contained some carbon which,
fortuitously, regulated the melting rate. In due course, more efficient power plants
were introduced and these produced fly ash with lower carbon contents. This meant
that fly ashes from different sources had to be carefully blended and it became more
difficult to ensure consistent powder composition and quality.
This stimulated the development of Synthetic mould powders in which fly ash
was replaced by silica, lime and bauxite and fluxing agents. These synthetic
powders were made of mixtures of minerals and it is possible to achieve similar
powder compositions from different mixtures of minerals. It should be noted that
the slag pool and the slag film should be unaffected by the nature of the minerals
used providing the composition is the same. For identical compositions produced
with different minerals, the only differences expected would occur during the
heating in the powder bed but once the minerals have been melted they should all
behave in an identical manner. However, there are some general rules which are
6.2 Selection of Mould Fluxes 191

used in the manufacture of synthetic mould powders, namely, the powder should
contain
(i) Mineral constituents with similar melting points (e.g. lithium feldspars,
sodium feldspars, wollastonite) to provide uniform melting.
(ii) The minimum number of minerals to achieve the target composition in order
to keep the recipe simple and to simplify quality assurance.
(iii) A minimum of mineral constituents which constitute a potential health
hazard (e.g. silica).
Fly-ash powders are still available and are relatively cheap and are used
extensively in ingot casting. However, in continuous casting, fly-ash powders have
been largely replaced by synthetic fluxes where the chemical compositions are
much better controlled.

6.2.1.2 Crystalline Phases Formed in Slag Film

When the molten slag solidifies in the shell/mould gap the first crystalline phase
formed in slags containing CaF2 is cuspidine, 3CaO2SiO2. CaF2 (denoted here as
C3S2Fl). The phase diagram containing the cuspidine phase has been reported [51].
Cuspidine is the first phase to solidify and it is formed on the mould side, but
growth of large crystals takes place at high temperature and thus large crystals tend
to be found on the shell side of the slag film [52, 53]. It can be seen from Fig. 6.8
that the slag film exhibits dendritic form on the shell side but much finer cubic
crystals are formed on the mould side [53].
It can be seen from Fig. 6.9 that most mould slag compositions tend to lie
outside the main crystallisation field for cuspidine. Thus, the powder composition

Fig. 6.8 Photograph of a

section of slag film showing
dendritic growth of crystals
on shell, high-temperature
side and fine cubic crystals on
the mould, low-temperature,
side [53]; (permission
granted, ISIJ [53])
192 6 Different Types of Mould Powders

Fig. 6.9 Ternary diagram showing cuspidine phase field; it can be seen that most conventional
mould slag compositions (o) lie outside the cuspidine phase field (solid line surrounding ▄);
dotted lines C/S values of 1 and 2 [20]; (permission granted, ISIJ [20])

was moved into the cuspidine field [20]; the powder was used in high-speed casting
(Vc  5 m min−1) and the slag was found to give lower horizontal heat fluxes (and
less longitudinal cracking) than the original powder [20].
The resultant slag film contained cuspidine dendrites through the entire section,
in contrast to the original powder which showed a glassy phase in addition to
cuspidine (Fig. 6.10) [20].
Hayashi et al. [54] studied the effect of Na2O on the formation of cuspidine
(C3S2Fl) crystals using FMAS-NMR, DTA and XRD and found that (i) F− ions
tend to bond with Ca2+ ions (ii) the maximum amount of cuspidine, after annealing,
was obtained with a sample containing 4% Na2O and this was attributed to the
effect of Na2O on nucleation and growth rates (via a lower viscosity) (iii) Tg and
Tcrys were decreased with increasing % Na2O.
Other mineralogical phases are formed during secondary crystallisation; a phase
diagram (Fig. 6.11) of the various phases formed has been reported for mould slags
with a basicity of ca. 1 [55].

6.2.1.3 Pick-up of Oxides by Mould Slag

The mould slag picks up various oxides from steelmaking reactions and reactions
between the steel and slag phases. Mould powders usually contain about 5% Al2O3
but there is typically, a 4% pick-up of alumina by the slag during the cast; the
6.2 Selection of Mould Fluxes 193

Fig. 6.10 Photographs of sections of slag films derived in high-speed casting with a conventional
powder (outside cuspidine field) showing a large glassy layer and b the developed powder (inside
cuspidine phase field) showing large crystalline layer; Note mould on left and shell on right side;
(permission granted, ISIJ [20])

Fig. 6.11 The mineralogical phases formed in slags after cuspidine has been formed for mould
powders with %CaO/%SiO2 ratios of ca. 1.0; (permission granted, IOM/Taylor and Francis [55]
re-drawn, Swerea/Kimab)

pick-up of inclusions arises, almost equally, [18] from (i) the steelmaking processes
and (ii) from the reaction between Al in the steel (denoted by an underline in
Eq. 6.11) and the slag pool. Slag viscosity increases with increasing Al2O3 content.
194 6 Different Types of Mould Powders

2SiO2 ðslagÞ þ 6Al ¼ 2Al2 O3 ðslag) þ 4Si: ð6:11Þ

Aluminium reacts in a similar manner with both MnO and FeO in the slag.
There is also pick-up of FeO and MnO by the slag. The pick-up of Mn when
casting Mn-steel grades occurs by the reaction shown in Eq. 6.12.

FeO (slag) þ Mn ¼ MnO ðslag) þ Fe ð6:12Þ

Thus, the composition of the slag pool and the liquid (which, subsequently,
forms the slag film) is slightly different from that determined from the mould
powder after allowing for the loss of carbon and carbonates. It is the slag pool
composition which should be used (i.e. including pick-up of Al2O3, MnO and FeO)
when calculating the properties of the slag film.
It should be noted that Al2O3 pick-up by the reaction shown in Eq. 6.11
becomes a major problem when casting TRIP and TWIP steels containing ca. 1%
Al; this is discussed below in Sect. 6.2.12.

6.2.1.4 Nature of the Powder

Mould fluxes are usually supplied either in the form of powders or granules.
Powders are produced by grinding, mixing and homogenisation of the various
minerals. Granules are usually prepared by mixing the flux ingredients into a slurry
which is then either extruded or spray-dried. In the latter process the slag is sprayed
as droplets (ca. 1 mm) which form granules or spheroids on drying. Extruded
powders are produced by a similar process but are extruded instead of being sprayed.
Expanding granules contain an expanding agent which expands on heating to pro-
duce irregular-shaped granules which decrease the flowability of the granules [6].
Flowability is a particularly important feature in ingot-casting powders where the
powder must flow easily in order to prevent areas developing where there is no slag
cover (i.e. no “open eyes” formed—see Sect. 7.4).
The advantages and disadvantages of powders and granules are summarised in
Table 6.2.
The performances of the various forms of powders (mixed, sintered, prefused)
have been compared [56, 57].

6.2.2 Pre-melted Fluxes

These are available but tend to be more expensive than granulated powders (or
spherodised) fluxes because of the higher energy costs involved in melting the
6.2 Selection of Mould Fluxes 195

Table 6.2 Advantages and disadvantages of powders and granules

Property Powders Granules
Advantages Disadvantages Advantages Disadvantages
Cost Cheaper More expensive
Thermal insulation Better More dust, More suitable Not so good
for automated
feeding
Homogeneity Worse-smaller Better
particles gravitate to
bottom of bag
Automatic feeding Worse Better
Health and Safety Dusts ! Hazard Less
hazardous
Flowability in Lower High—can
mould expose metal in
standing wave

mixture of oxides, and fluorides. However, pre-melted powders do reduce the rate
of fluorine emissions on the steel plant since the chemical activity of CaF2 in the
fused powder is lower than that of CaF2 particles present in the conventional
powder.

6.2.3 Starter Powders

Starter powders, as their name suggests, are used at the beginning of the cast. Starter
powders must melt rapidly and provide a pool of molten slag in order to (i) protect
the surface of the steel from oxidation and (ii) encourage slag infiltration into the
shell/mould gap. They have the following characteristics:
(i) They contain large concentrations of low-melting materials such as Na2O,
fluorides and borates.
(ii) The carbon contents are low (<1%) to ensure a high melting rate and min-
imise carburisation of the steel shell.
(iii) They sometimes contain exothermic agents to provide additional heat.
(iv) They have a tendency to form slag rims since low-melting materials (like
Na2O) tend to act like glue and incorporate solid materials into the rim;
consequently, starter powders should not be used for longer than necessary.
The first slag film is formed from the slag produced by the starter powder but this
will be modified in time by (i) any downward movement of the slag film which will
encourage new slag film formation (ii) any fracture and movement of slag film
which will encourage slag infiltration from the slag pool and (iii) gradual reaction
with the liquid slag film.
196 6 Different Types of Mould Powders

(a) (b)

Sinter

Pool

Fig. 6.12 Schematic diagram showing a powder bed and temperature gradient for a conventional
powder and b powder containing exothermic agents—note how exothermic agents reduce both the
temperature gradient and the vertical heat flux and increases the pool depth

6.2.4 Exothermic Fluxes

These contain exothermic agents such as Ca/Si or Fe/Si which give out heat when
reacting with oxygen (Eqs. 6.13 and 6.14):

Si þ O2 ðgÞ ¼ SiO2 DH25 C ¼ 911 kJ mol1 ð6:13Þ

Ca þ 0:5 O2 ðgÞ ¼ CaO DH25 C ¼ 635 kJ mol1 ð6:14Þ

This heat increases the local temperature in the bed and thereby reduces the
thermal gradient (DT = Tpool − Tlocal), and subsequently decreases the total heat
flux in the mould (qvert = w.t.kbed DT) where kbed is the effective thermal conduc-
tivity of the bed and w and t are the width and thickness of the mould, respectively
(Fig. 6.12).
Powders containing exothermic agents [23, 58, 59] are used to
(i) Increase the depth of the molten pool (e.g. to reduce C-pick-up by the shell
and to improve both the absorption of inclusions and powder consumption).
(ii) Reduce the length of the solidified meniscus by reducing the vertical heat flux
and thereby allowing the inclusions and gas bubbles to be transported to the
slag layer (e.g. Fig. 6.5a, b) [58, 59].
It should also be remembered that the free carbon is also an exothermic agent
and that conversion to CO2(g) provides nearly 3 more heat than conversion to
CO (g) (see Eqs. 6.15 and 6.16). In contrast, the decomposition of carbonates is
endothermic and will lower the local temperature which will stimulate a higher
heat flux in that locality of the bed at ca. 500 °C.
6.2 Selection of Mould Fluxes 197

C þ 0:5 O2 ¼ COðgÞ : DH298 ¼ 110 kJ mol1 ð6:15Þ

C þ O2 ¼ CO2 ðgÞ : DH298 ¼ 393 kJ mol1 ð6:16Þ

Liquid slag Feeding

The possibility of pouring liquid mould slag onto the surface of the steel has been
explored [60–62]. Such a practice might be expected to provide the following
benefits:
(i) Provision of extra heat to help maintain a high meniscus temperature.
(ii) Quicker establishment of steady state conditions.
(iii) The possibility of higher powder consumption and lower friction on the
shell.
(iv) The possibility of creating a deep pool to absorb inclusions (especially when
casting high-Al steels which produce large amounts of Al2O3).
A number of plant trials were carried out to cast blooms (0.38  0.225 m) at
Vc = 0.65–0.7 m min−1 [60]. The following procedures were used in casting
blooms:
(i) Enough mould powder was melted in a mobile furnace to produce 3 litres of
liquid.
(ii) 1.5 litre of the liquid mould slag was poured onto the steel surface (this is
equivalent to dpool = ca 2 mm) and the casting started.
(iii) The remaining 1.5 litre of liquid was subsequently poured into the mould.
(iv) The regular mould powder was then added at the appropriate time.
Several successful casts were carried out and it was reported [60] that
(i) Friction on the shell was low using this practice (good liquid lubrication).
(ii) The surface quality of the casts was not very good with a large number of
“cavities”; this was attributed to the freezing of mould slag to form “lumps”
which subsequently infiltrated into the shell/mould channel and left an
impression on the steel surface (Fig. 6.13).
(iii) Faint (very shallow) oscillation marks were observed but normal oscillation
marks were obtained when the switch to regular mould powder was made;
the faint oscillation marks suggests the slag pool depth was very low.
Liquid slag feeding casting of high-Al steels has also been reported [61, 62] and
was found to provide good powder consumption. Unfortunately, we were unable to
obtain these publications. However, more development of liquid slag feeding seems
inevitable, if these high-Al steels, with low-melting temperatures, are ever to be
continuously cast on a commercial scale.
198 6 Different Types of Mould Powders

Fig. 6.13 Photograph showing “cavities” in steel surface (marked) when using liquid mould slags
(permission granted, SWEREA [60])

6.2.5 Fluoride-Free Powders

Mould powders consist of a mixture of oxides and fluorides (referred to as an

oxy-fluoride system). Conventional powders tend to contain 7% F which is
usually added in the form of CaF2. Oxy-fluoride systems are basically unstable
since they react together to form gaseous fluorides (e.g. Eqs. 6.17–6.19) [63–65]. It
has been calculated that the partial pressure of NaF(g) exceeds 1 atm at tempera-
tures above 530 °C [64, 65]. The gaseous fluorides will also react with any moisture
to form HF(g). Mass-spectrographic studies showed that safety limits for HF(g)
(0.1 mg m−3) soluble fluorides (e.g. NaF, 0.2 mg m−3) and insoluble fluorides (e.g.
SiF4, 0.5 mg m−3) were breached on plant [63].

Na2 O þ CaF2 ¼ CaO þ 2NaFðgÞ ð6:17Þ

SiO2 þ 2CaF2 ¼ 2CaO þ 2SiF4 ðgÞ ð6:18Þ

Al2 O3 þ 3CaF2 ¼ 3CaO þ 2AlF3 ðgÞ ð6:19Þ

H2 O þ CaF2 ¼ CaO þ 2HFðgÞ ð6:20Þ

The HF emissions cause corrosion of plant equipment [63, 66] and constitute a
Health and Safety hazard to plant personnel [63]. Furthermore, the secondary
cooling water leaches fluoride out of the spent slag film (at the mould exit or slag
adhering to the steel strand) causing the cooling water to gradually increase in
acidity [67–69]. Consequently, the cooling water must be treated with alkaline
agents to reduce the acidity of the secondary cooling water [67].
6.2 Selection of Mould Fluxes 199

Environmental worries about fluoride emissions resulted in a search for suc-

cessful mould powders containing zero, or reduced, levels of CaF2. The challenge
has been to develop a F-free powder which could match the performance of the
extant, F-containing powder in current use with regard to the following properties:
(i) To provide the same level of powder consumption (i.e. primarily determined
by the required values of viscosity and Tbr).
(ii) To provide the same level of horizontal heat transfer (involving Tbr, RCu/sl
and fcrys).
(iii) To replace cuspidine with a suitable crystalline phase in order to meet the
required thermal resistance (this involves both RCu/sl and fcrys where RCu/sl
depends upon the density difference between the glass and the crystal phases
[qcryst − qgl]).
It is the third criterion which is the most difficult to meet.
Two different approaches have been adopted in the development of F-free
powders, namely
(i) To replace CaF2 with other fluxes (e.g. B2O3, Li2O) and then modify the
composition so that the properties of the candidate powder match the values of
the original powder for viscosity, Tbr and heat flux (qhor); one problem is that
fluxes like B2O3 tend to promote the glass phase resulting in an increase in qhor
[27, 70, 71].
(ii) To identify a suitable crystalline phase (e.g. perovskite [72, 73]) to replace
cuspidine, in order to reflect the IR radiation and create an interfacial thermal
resistance, RCu/sl (which is related to [qcrys–qgl]) to give a similar level of qhor
to that obtained with a slag film containing cuspidine.

6.2.5.1 Replacement Fluxes Giving Similar Property Values

Several groups have used the approach where CaF2 is replaced by other fluxes to
provide similar values of viscosity and Tbr to that of the F-containing slag, e.g.
using B2O3, [27, 70, 71, 75, 76]; Li2O [66, 77–79]; Na2O [69, 80]; BaO [79, 81,
82]; and TiO2 [72, 73, 76]. The effect of flux additions on the viscosity, Tliq and Tbr,
surface tension, heat flux and the ability to absorb inclusions has been studied in
several of these studies on both F-free and low F powders [70–73, 75, 78, 81]; the
results are summarised in Table 6.3. The effect of crystallisation on the viscosity,
Tliq, Tbr and fcrys of the remaining liquid for a F-free slag has also been determined
[86].
Powders for casting billets
F-free mould powders have been developed for casting billets of plain C steels [27,
70, 74, 85]. In designing F-free mould slags, it is important that the values of the
following properties should all match those of the original slag: (i) η1300°C (ii) Tbr
and (iii) degree of crystallinity. However, high-viscosity slags are frequently used in
200 6 Different Types of Mould Powders

Table 6.3 The effect of flux additions (in mass%) on melting temperature, Tbr; viscosity, surface
tension and heat flux measured in simulation experiments and effect on absorption of inclusions
Ref Compound dTliq/% (dη/%) qhor qhor Absorption Composition of
(dTbr/%) dPas MWm−2 MWm−2 base slag
°C %−1 %−1
Lu [81] BaO −15 −0.15b −6.2b " C/S = 0.8; 1.0;1.2;
A4;L3;F2.3;B2;
Ba1-5
Lu [78] Li2O −33c −0.35c −9c 0 " C/S = 0.8; 1.0;1.2;
Wen [72] −30 −1 A4;L2-10;F2.3
C/S = 0.95–1.15;
N5-8;B4-8;;L1.5;
Mn4; T4-7
Chen [73] Na2O −2.5 −0.3 −17* 0 ([8] C/S1.05;M7; N6;
Wen [72] %)e Mn5 Fl2.6
C/S = 0.95–1.15;
N5-8;B4-8;;L1.5;
Mn4; T4-7
Lu [83] K2O −59 −0.37 −4 " C/S = 0.8; 1.0;1.2;
A4; L3;F2.3;B2;
K1-5
Lu [75] B2O3 −8a −0.2a a
+0.04 0 C/S = 0.8; 1.0;1.2;
Fox [70] (−43) −0.2 A4;L2;F2.3;B2-10
Wen [72] −19 C/S = 0.95-1.15;
N5-8;B4-8;;L1.5;
Mn4;T4-7
Wen [72] MgO +5 −0.12 0 C/S = 0.95–1.15;
N5-8;B4-8;;L1.5;
Mn4;T4-7
Wen [72] MnO −7 0 −0.15 C/S = 0.95–1.15;
[4%]e N5-8;B4-8;;L1.5;
Mn4;T4-7
e
Wen [72] TiO2 +4.5 C/S = 0.95–1.15;
N5-8;B4-8;;L1.5;
Mn4;T4-7
Lu [84] NaF −7 d
−1.7 −0.1 " C/S = 0.8; 1.0;1.2;
[4%]e A4;L2;F2.3;B2
() indicates Tbr
*
cf-15 for CaF2
a
Varies with C/S ratio
b
Mean of three slags with C/S ratios of 0.8; 1; 1.2
c
Converted to Li2O (additions of Li2CO3)
d
Sharp drop from 3 to 6%, then steady
e
Unaffected to 6% MnO or NaF, then sharp increase
M MgO, N Na2O etc. in slag composition
6.2 Selection of Mould Fluxes 201

billet casting and these slags have no break temperature since they form a super-
cooled liquid slag and do not crystallise [70, 74]. This is due to the fact that in HC
steels a thick, strong steel is required and so restriction of the heat extraction from
the shell is not important. Thus, it is only necessary to match the viscosity (i.e. the
lubrication) when designing F-free mould powders for billets [27, 70, 74]. Thus, in
this case, 2.3% CaF2 was replaced by 2%Na2O and 1.1%B2O3. In plant trials it was
found that [27, 70, 74]
(i) The F-free flux performed as well as the original F-containing powder.
(ii) There was no discernible pick-up of boron by the steel.
Powders for casting blooms and slabs
The criteria for designing F-free mould slags are much more stringent for casting
slabs and blooms than for billet casting.
Wang et al. [87] pointed out that for a F-free flux to perform as well as a
conventional slag, it must have similar values not only for (i) Tliq (ii) η1300 °C
(iii) Tbr but also must satisfy the conditions for (iv) the ratio of crystalline phase in
the slag film (Rp%, which is equivalent to 100 fcrys) (v) the crystallisation tem-
perature (Tcrys) and (vi) the nature of the η-T curve. They also carried out Raman
spectroscopy on F-containing and F-free mould slags and found that F-free slag
contained less Q0 and Q1 species and more Q2 and Q4 species (where superscript
indicates the number of bonding O’s, see Sect. 11.3.2) than the conventional slags
(i.e. they contained more-polymerised species). A F-free flux for casting slabs was
found to perform satisfactorily and they have also been used to cast blooms [88]
slabs [71] and thin slabs [79]. One benefit of using F-free slags is that it results in
significant decreases in SEN erosion rates [27, 87].

6.2.5.2 Replacement of Cuspidine with Other Crystalline Phase

Several investigators have designed F-free slags by replacing the cuspidine in the
slag film with another crystalline phase; these are classified below in terms of the
crystalline phase formed. The presence of crystalline phases in the slag film is
essential to reduce the radiation conduction. Crystallisation of an initially glassy
slag film causes (i) reflection and scattering of the IR radiation (thereby reducing the
radiation conduction considerably) and (ii) shrinkage in the slag film, resulting in
both an interfacial resistance (RCu/sl) and porosity in the slag film.
TiO2 or CaTiO3 (CT)
The replacement of cuspidine by adding TiO2 to the powder has been studied by
several workers [72, 73, 87, 89–91]; the CaO–SiO2–TiO2 (CST) system was
studied as candidates for F-free fluxes [89]. Time–temperature–transformation
(TTT) curves for slags with a (C/S) ratio of 0.8 and containing up to 17 mol% TiO2
were constructed and it was found [89] that:
202 6 Different Types of Mould Powders

(i) The phase, titanite (CST) formed rapidly.

(ii) The CST phase did not increase the viscosity and melting temperature.
(iii) The precipitation of CST at higher temperatures was lower than that of
cuspidine-forming (conventional) mould slags.
(iv) Na2O additions probably reduce the time taken to crystallise (i.e. incubation
time) as they did with cuspidine [92].
Wen et al. [72] studied the replacement of slags containing ca. 8%F and (C/S)
ratios of 0.8–1.0 with TiO2 additions of 1–9% (but typically 6%). They used blast
furnace slag as a source of TiO2 (and CaO, SiO2 and Al2O3). They studied the
effects of various oxide additions on Tliq; η1300 °C and qhor (in a simulation test) as
shown in Table 6.3. A F-free powder was developed (C/S = 0.95–1.15; 5–8%N; 4–
8% B; 2%L 4%Mn and 4–7% T). Plant trials were carried out on the casting of thin
slabs and regular slabs for plain C and low-alloy steels. It was found that
(i) The crystalline phase formed in the slag film was perovskite (CT) and
fcrys = 0.4.
(ii) The F-free slag had similar viscosities, melting temps and heat fluxes to the
original, F-containing slag.
(iii) The surface quality of plain C and low-alloy slabs obtained with the F-free
slag was satisfactory with fewer cracks produced than with the conventional
slag.
(iv) There was a marked increase in qhor when Na2O > 8% and MnO > 4%
(which may be due to the formation of more glassy phase).
It was subsequently reported that Na2O reduced the incubation time for the
crystallisation of perovskite and that the kinetics of crystallisation were similar to
those for industrial mould slags, containing F (Fig. 6.14). Furthermore, it was shown
that TiO2 reduced the activation energies for the transitions (glass ! crystal) and
(liquid ! crystal) [93].

Fig. 6.14 Transformation-temperature-time (TTT) diagram for slags from the CaO–SiO2–TiO2–
Na2O system and for an industrial mould slag ♦ = CST; ● = CSNTA-3; ■ = CSTNA-1;
D = CSTNA-4; ◊ = CSTNA-2; (permission granted, Wiley [85])
6.2 Selection of Mould Fluxes 203

However, it was, subsequently, reported that the flux developed by Wen et al.
[72] resulted in sticker break-outs when it was used on plant [15]. Thermodynamic
calculations and laboratory experiments showed it was possible to form TiC or Ti
(C/N) by the reaction of TiO2 with carbon floating in the slag pool or in the slag bed
(Eqs. 6.21 and 6.22)

TiO2 þ 3C ¼ TiC þ 2COðgÞ ð6:21Þ

2TiO2 þ 4C þ N2 ðgÞ ¼ 2TiN þ 4COðgÞ ð6:22Þ

Both TiN and TiC have low solubility in the molten slag and the undissolved
particles agglomerate and block the infiltration of slag into the mould/strand gap
(and also cause a marked increase in slag viscosity) [15]. These conditions lead to
sticker break-outs [15].
Subsequently, Chen et al. [73] developed low F-fluxes, in which 7.5% F was
reduced to 3%, by introducing 6%TiO2 to form perovskite; no problems were
reported when using these mould slags to cast peritectic, MC steel grades.
However, perovskite has a high melting temperature which will restrict the liquid
slag thickness and lubrication of the shell. These contradictions in casting behaviour
when using mould slags with 6%TiO2 need to be resolved.

6.2.5.3 NC2S3 (Na2O2CaO3SiO2)

The phase, NC2S3 was studied as a candidate to replace cuspidine as the crystalline
phase formed in the slag film [94]. The phase diagram of the NCS system (shown in
Fig. 6.15) contains both NC2S3 (Tliq = 1285 ± 5 °C) and N2C2S3 (Tliq > 1400 °C).

Fig. 6.15 Ternary phase

diagram for the
Na2O + CaO + SiO2 system;
(permission granted, ISIJ,
[94])
204 6 Different Types of Mould Powders

The effect of small additions of Li2O, MgO, MnO and Al2O3 were investigated in
order to reduce the melting temperature. The effects were quantified in terms of the
composition length (L = {R(Xi − Yi)2}0.5 where Xi and Yi represent the compositions
of the actual slag and NC2S3, respectively and i represents the individual con-
stituents) [94]. A F-free flux (C/S = 0.62; (Na2O + Li2O) = 22%; others 4%;
Tsol = 1100 °C and η1300 °C = 3.2 dPas) was used to cast a HC (0.47%) steel slabs at
Vc = 0.7 m min−1 [94]. The F-free flux performed as well as the original,
F-containing powder. However, restriction of qhor is not an issue when casting HC
steels and its use when casting MC steels would be a more rigorous test of this type
of mould slag.

6.2.5.4 Gehlenite (C2AS) and Wollastonite (CS)

A fluoride-free flux (where 3% CaF2 was replaced by 1.5%B2O3 + 2.5%

Na2O + 4%Al2O3 + 3%MgO) was used to cast HC billets [27]. The slag film
contained a mixture of gehlenite (C2AS) pyrope [M3AS3 and wollastonite (CS)]
[27]. It performed well in plant trials and led to reduced SEN erosion rates.

6.2.5.5 Melilite

Melilite slags cover the range of solid solution between end-members, akermanite
(C2MS2) and gehlenite (C2AS). It can be seen from Fig. 6.16 that melilite slags tend
to have higher viscosities than conventional mould slags (based on cuspidine). They
have a reasonably high basicity and have a low oxygen potential.
Hanao et al. [20] pointed out that melilite slags retain some glass phase and
thereby maintain some lubrication via the supercooled liquid formed above Tg
(at ca. 600 °C). Since mould slags for casting billets frequently have high vis-
cosities, melilite slags are suitable for casting billets and Hanao et al. [20] used these
slags in the successful casting of round billets. Melilite slags have also been used in
the casting of TRIP steels [20].

Fig. 6.16 Viscosity at

1300 °C as a function of
basicity (C/S) = slags
forming cuspidine; □ = slags
forming melilite; (permission
granted, ISIJ [20])
6.2 Selection of Mould Fluxes 205

6.2.6 Reduced F-Powders

The first response by the powder manufacturers to the environmental and health and
safety concerns posed by Fluorine emissions and the leaching of fluorides was to
reduce the fluorine contents in mould powders. The advantage of this approach is
that cuspidine is maintained as a crystalline phase, even if its concentration in the
slag film is reduced [87]. Physical properties, simulation tests and plant trials were
carried out on casting MC steels with a conventional powder (containing 7.5% F)
and two candidate powders (containing 3.5% F and 6% TiO2) with C/S ratios of
1.15 and 1.35 [73]. Some of the cuspidine in the slag film (formed with the con-
ventional powder) is replaced by perovskite (CaTiO3=CT) and Ca2SiO4 (C2S) in the
two candidate powders [73]. The plant trials showed that the candidate powder with
C/S ratio = 1.35 performed as well as the conventional powder with similar heat
fluxes and powder consumption but the slag film was thinner. No problems were
reported with the formation of Ti(CN) and the sticker break-outs or false alarms.

6.2.7 C-Free Powders

The free carbon content of the casting powder is the main factor controlling the
melting rate (QMR in kg min−1) of the powder. The melting rate should be sufficient
to match the required powder consumption. The carbon is non-wetting to the
molten slag and prevents the slag globules from agglomerating (Fig. 6.17).
However, the carbon is oxidised by any O2(g) and CO2(g) present in the bed. So the
amount of carbon content decreases continually down the mould. Smaller particles
of carbon provide more separation of the globules per unit mass of carbon. Thus the
melting rate increases (Fig. 6.17) with (i) decreasing free carbon content (%Cfree)

(a) 7 (b) 7
10-2.QMR. mm s-1

10-2.QMR. mm s-1

6 6

5 5

4 4

3 3
0 1 2 3 4 500 600 700 800 900
C, mass % ρbulk, kg m-3

Fig. 6.17 Melting rate as functions of a Carbon content, ▲ = C black; ● = Coke; note DC
black < Dcoke so C black has lower melting rate and b bulk density ○ = 1% C ; ● = 1.3% C;
(permission granted, ISIJ [44])
206 6 Different Types of Mould Powders

Fig. 6.18 Photograph

showing Carbon smears
on the surface of thin slabs
(courtesy of Stollberg GmbH
[22])

(ii) with increasing particle size of the carbon (DC) and (iii) with decreasing bulk
density (qbulk) of the powder as shown in Fig. 6.17b (i.e. QMR " as %Cfree# as DC "
and as qbulk #).
However, excess carbon tends to float near the surface of the slag pool (Fig. 6.1)
and forms a layer of “amorphous carbon” [22]. Excess carbon can lead to
(i) Carbon pick-up by the steel, this is a particular problem when casting LC and
ULC steels [95].
(ii) The formation of Carbon smears on the surface of thin slabs (Fig. 6.18 [22])
through contact between the shell and the amorphous carbon layer formed at
the top of the slag pool.
(iii) The formation of depressions on the surface of slabs (see Sects. 11.8.1 and
11.8.2).
(iv) The formation of large slag rims or ropes which is promoted by both
amorphous carbon and high Na2O contents [22] and is prevalent when large
amounts of Al2O3 are formed.
Carbon is not essential for the control of melting rate. Any material which (i) is
non-wetting to molten slag and (ii) reacts with O2 and CO2 until it is consumed
would be suitable for this task. Nitrides have been used for this purpose; BN has
been used but is expensive and also poses a threat of Boron pick-up by the steel [23,
96]. However, other nitrides (e.g. Si3N4) have been used successfully to totally, or
partially, replace the carbon in the casting powder [22–24, 96, 97].

6.2.8 Powders for High-Speed Casting

and Thin Slab Casting

Thin slab- and high-speed casting (up to 8 m min−1) pose difficult casting condi-
tions, i.e.
6.2 Selection of Mould Fluxes 207

(i) The high casting speed results in both a high heat flux and a short residence
time in the mould; these conditions will tend to result, respectively, in an
uneven shell in the meniscus region (prone to longitudinal cracking) and to a
relatively thin shell, overall, at the mould exit (which make it prone to
bulging and mould level variations).
(ii) The powder consumption tends to be low because of the high-casting speed
and a mould powder with a low viscosity is needed to provide a reasonable
Qs value.
(iii) The high metal velocities can cause meniscus instability, slag entrapment and
frequently EMBr is used to decelerate the flow velocity; EMBr tends to
reduce the vertical heat flux which results in concomitant decrease in melting
rate, slag pool depth and powder consumption.
(iv) In thin-slab casting there is little room to accommodate the SEN; conse-
quently, misalignment of a high-conductivity, SEN can create “cold spot” in
the mould resulting in variations in shell thickness (which can lead to lon-
gitudinal cracking).
Thus the principal problems are longitudinal cracking, poor powder consump-
tion, slag entrapment and those issues associated with mould level variations.
Several studies have been carried out to develop mould powders for these trying
conditions [98–103].
The first thin-slab-casting slags tended to have high Tbr values to control the
horizontal heat flux; Tbr values were ca. 100 °C higher than those calculated using
either Eqs. 6.6–6.9 for the appropriate steel grade or derived from Fig. 6.6.
However, more recent powders for thin-slab casting tend to have values for the
required viscosity (ηreq) and break temperature (Treq
br ) that are consistent with values
calculated from Eqs. 6.2–6.4 and 6.7–6.9, i.e. they follow the same empirical rules
as those for conventional slab casting [104].
Specially designed moulds, e.g. “football moulds” are frequently used to
accommodate the SEN. However, the sharp corners where the regular mould meets
the football can also lead to cold spots and sequentially, to shells of variable
thickness and to longitudinal cracking. Thus the tendency is to use a more crys-
talline slag with a high Tbr to alleviate such problems. However, sharp corners
should be avoided and the mould should be designed to reflect this.
The problems of low powder consumption values (about a half of the values of
Qreq
s calculated with Eqs. 6.2–6.4) and low slag pool depths can be addressed by
increasing the melting rate by reducing the % carbon and the carbon particle size.
However, care must be taken to avoid both the % carbon and the carbon particle
size being reduced simultaneously, since they have opposing effects on the melting
rate. Precautions should be taken to avoid reducing the % carbon to <1% since the
concomitant loss of energy from carbon consumption can lead to meniscus freezing
(i.e. skull formation).
Bulging is a particular problem when casting MC-peritectic steels which are
prone to longitudinal cracking where slags with high Tbr and fcrys are used to reduce
qhor. This and the short residence time result in a thin shell at the mould exit which
208 6 Different Types of Mould Powders

is prone to bulging and to mould level variations [105]. Mould level variations
cause increases in (i) inclusions (ii) pinholes [106] (iii) large slag rims [104, 107]
and an increased risk of sticker break-out [108]. In order to combat the problem of
bulging, it has been proposed that the slag film should restrict heat transfer in the
meniscus region and the upper mould but should promote heat transfer in the lower
mould to create a thicker shell [104 p. 150]. Further work must be done if this target
is to achieved.

6.2.9 Powders for Casting Round Billets

The major problem with casting round billets is that the shell shrinks during
solidification and it is important that molten slag should fill any gap and help to
support the billet. Slag films which crystallise develop “surface roughness” on the
surface in contact with the mould and thus provide poor support for the billet. In
order to provide the required support of the billet it is necessary to use a glassy slag
film. The challenge is to provide both control of the horizontal heat flux whilst
providing support to the billet. This was obtained in practice by using a powder
which:
(i) Produced a glassy slag film (giving the necessary support).
(ii) Reducing the horizontal heat flux by creating a thick solid layer, ds (via. a high
solidification temperature) and by incorporating transition metal oxides (e.g.
MnO) into the slag to absorb the IR radiation [40, 109, 110].
Subsequent work showed that slags forming melilite phase performed well [20].

6.2.10 Powders for Casting Beam Blanks

Longitudinal cracking is a problem in the casting of beam blanks (Fig. 6.19).

It was reported [111] that that the longitudinal cracking was worst
(i) At the regions denoted B and E in Fig. 6.19 where the horizontal heat flux is
at its highest.
(ii) For mould slags with low solidification temperatures (Tsol).
(iii) For steels with higher S content.
Powders for casting beam blanks should provide a thick, crystalline slag film to
minimise any temperature variations (i.e. shell thickness variations) which lead to
longitudinal cracking.
6.2 Selection of Mould Fluxes 209

Longitudinal crack length

12

10

0
A B C D E F A' B' C' D' E' F'
PosiƟon

Fig. 6.19 Schematic drawing showing a identification of different areas of beam blank and
b prevalence of different areas to longitudinal cracking; (re-drawn after [111])

6.2.11 Non-Newtonian Powders

The viscosities of most liquid slags are considered to be Newtonian since their
viscosity is unaffected by the shear rate. In non-Newtonian melts the viscosity
decreases with increasing shear rate. Non-Newtonian slags have been produced to
reduce slag entrapment which occurs in the region of low stress near the slag/metal
interface. The area of lubrication (the region where liquid slag infiltrates into the
mould/shell gap) is considered to be a region of high stress. Thus a non-Newtonian
slag will exhibit a high viscosity in the slag entrapment region and a lower viscosity
in the lubrication region which will increase powder consumption.
Non-Newtonian slags were produced by adding Si3N4 to a conventional slag
(with C/S = 0.9) at 1450 °C and holding it at this temperature for 1 h; the N content
of the final slag was 0.2% [112, 113]. The viscosities of this flux are shown in
Fig. 6.20a and it can be seen that the viscosity was 7 dPas at low shear rates but fell
to 5 dPas at high shear rates. The viscosity of the conventional slag was 4.8 dPas
for all shear rates. The N enters the silicate structure as a bonding N in the silicate
tetrahedron and is also bonded to a Ca2+ ion which is also connected to an O in
another tetrahedron (Fig. 6.20c); the N–Ca bond is weak and is severed as the shear
rate is increased.
It was found that the non-Newtonian slag exhibited a lower contact angle on a
plate of pure Fe than the conventional slag (Fig. 6.20b) and that the heat flux was
greater for the non-Newtonian slag than for the conventional slag (qconv non−N
hor < qhor )
[112, 113].
Trials were carried out on a miniature continuous caster and it was found that the
powder consumption (Qs) was higher for the non-Newtonian flux than that for a
Newtonian slag (η = 3.5 dPas) (Qs = 0.5 and 0.35 kgm−2, respectively) [112, 113].
210 6 Different Types of Mould Powders

Fig. 6.20 Schematic diagrams showing a the viscosity of slag (at 1300 °C) as a function of shear
rate o = Newtonian and ● = non-Newtonian behaviour b Contact angle as a function of
temperature and c shows the change to the structure with increasing shear rate; (permission
granted, ISIJ [113] re-drawn)

6.2.12 Powders for Casting TRIP and TWIP Steels

TRIP steels have both good ductility and high strength; they contain (i) retained
austenite with good ductility and (ii) martensite formed during plastic deformation
(of retained austenite) which provides the strength. TRIP stands for
Transformation-induced plasticity. These steels tend to contain higher levels of C
and Si and additions of Al (typically ca. 1%). The Al enhances the C content in the
austenite and thereby promotes the stability of austenite; a further advantage is that
it reduces the Si needed in the steel. Normally, steels with >0.1% Si cannot be
galvannealed but the substitution of Si by 0.5–1%Al allows the steel to be gal-
vannealed. A typical TRIP steel contains 0.14%C; 1.8%Mn; 0.6%Si; 1%Al.
TWIP steel stands for Twinning-induced plasticity steel. These steels have high
Mn contents and the Al increases the stacking fault energy and stabilises the
austenite.
6.2 Selection of Mould Fluxes 211

6.2.12.1 Problems with Casting TRIP and TWIP Steels

TRIP and TWIP steels are difficult to cast because of the large amounts of Al2O3
formed by the reaction of Al in steel (denoted by underline) with SiO2 in the slag.

2Al þ 3SiO2slag ¼ 3Si þ Al2 O3slag ð6:23Þ

In fact, the reaction of the Al in the steel is not confined to SiO2 since similar
reactions will occur with any slag constituents above Al in the Ellingham diagram,
e.g. TiO2, MnO, etc.
Figure 6.21 shows the pick-up of Al2O3 when casting normal steels is frequently
around 4% with ca 2% coming from steelmaking reactions and ca. 2% from the
reaction, as shown in Eq. 6.23. [18]. When casting steels containing (0.4%Al) and
(0.6–1%Al), the Al2O3 pick–up values increase to (14–18%) and (17–25%),
respectively. Casting steels with Al contents >0.6% is difficult and frequently
casting is limited to one ladle of steel. These large increases in the Al2O3 content
result in significant increases in Tliq, Tsol, fcrys and viscosity with consequent,
decreases in the powder consumption (Qslag s req = 0.55/η Vc or Qs req
0.5 powd
= 0.6/
ηVc) and the lubrication supplied. The changes in Tbr and the crystal fraction in the
slag film affect the horizontal heat transfer. Given these significant changes, it is not
surprising that a high pick-up of Al2O3 by the slag is reported to result, sequentially,
in erratic melting, fluctuating mould temperatures and false alarms on the sticker
detection system [20, 114, 115]. It has been suggested that the Al2O3 formed tends

Fig. 6.21 Ellingham diagram

showing Gibbs free energy of -100
various oxides per mole of O2 Cu2O
as a function of temperature
(TK) (Note the temperatures -300 FeO
are in K.)

-500
-1
ΔG, kJ mol

Mn
-700

-900 Si Al 2O 3 MgO
CaO
-1100

-1300
200 600 1000 1400 1800
Temperature, T/K
212 6 Different Types of Mould Powders

Fig. 6.22 Crystallisation

index as a function of Al2O3
content of mould slag
(permission granted, ISIJ
[20])

to promote the formation of more crystalline phase at the expense of the glassy
phase (Fig. 6.22) [20] resulting in a loss of lubrication [20, 115].
The large amounts of Al2O3 generated in casting high-Al steels are not easily
dissolved in the slag pool and the particles tend to agglomerate and hinder the
infiltration of molten slag. Furthermore, these agglomerated particles can be cap-
tured by the shell (forming depressions or grooves in the steel). Furthermore, the
creation of large amounts of Al2O3 tend to result in large slag rims which have a
tendency to fracture, which, in some cases, block slag infiltration and thereby, cause
sticker break-outs.

6.2.12.2 Strategies to Minimise Al2O3 Pick-up

Several strategies have been adopted in designing powders for casting TRIP steels,
namely
(i) To select a slag composition which remains in a low-melting region even
with significant levels of Al2O3 pick-up.
(ii) To increase the FeO and MnO contents in order to minimise the pick-up of Si
by the metal since the reaction (3 MOslag + 2Al = 3M + Al2O3slag where
M=Fe or Mn) will occur in preference to Eq. 6.23.
(iii) To increase the volume of the slag pool to improve the dissolution of Al2O3
(this is usually carried out by including exothermic agents in the powder but
the use of a thicker powder bed would also be beneficial).
(iv) To increase the driving force for Al2O3 dissolution (i.e. Csat–Co) by keeping
the Al2O3 content in the powder (Co) to low levels in slags based on calcium
silicates.
(v) To minimise the formation of Al2O3 by keeping the SiO2 content of the
powder at a low level.
It is not possible to use all of these strategies simultaneously but some of them
are incorporated into each of the approaches described below.
6.2 Selection of Mould Fluxes 213

6.2.12.3 Different Approaches to Developing Powders for TRIP Steels

There have been two approaches taken in the development of powders for casting
high-Al steels, namely
(i) To modify conventional casting powders.
(ii) To develop new powders based on calcium aluminates in order to minimise
both the SiO2 content of the flux and amount of Al2O3 formed by Eq. 6.23.
Conventional powders
Steels containing less than 0.6%Al can be cast using conventional powders forming
slag films based on cuspidine. The slag pool is capable of dissolving the Al2O3
formed. However, difficulties are experienced when casting steels with higher Al
contents.
The composition range of conventional “cuspidine-forming slags” is shown in
Fig. 6.23 and the changes in composition (i.e. the gain in Al2O3 and losses in SiO2,
FeO and MnO) are shown by the direction of the arrow. The Al pick-up by the slag
is cited as % Al2O3 on the basis of chemical analysis but some of the Al could be
present as AlN [116].
Low-basicity powders [116–121]
At first sight, it would seem illogical to use low-basicity powders because this
entails high SiO2 contents which would promote the reaction Al and SiO2 shown in
Eq 6.23. However, careful examination of Fig. 6.24 indicates that the slag remains
longer in the molten range when using a slag with C/S ratio of ca. 0.6.
Plant trials performed on a TRIP (1%Al) steel with a low-basicity (C/S = 0.55;
N10%; F10%; Mn 6.5%) powder and the results were compared with those
obtained when casting LC steels; it was found that [117]:

Fig. 6.23 Ternary diagram SiO2

showing the approximate
compositional bounds for
“cuspidine-forming”
powders; CaO = %
CaO + %MgO + %Na2O;
= range of cuspidine-
forming fluxes; lines represent
C/S ratios, from top, of 0.667;
1.0 and 1.5; arrow shows the
change in composition
through reaction of Al and
oxides
CaO CaF2 +Al2O3
214 6 Different Types of Mould Powders

SiO2

1400oC

0.667

1.5

CaO CaF2+Al2O3

Fig. 6.24 Ternary diagram CaO + SiO2 + (CaF2 + Al2O3) where CaO = %CaO + %MgO + %
Na2O; the thin lines represent (C/S) ratios of 0.667, 1 and 1.5; the irregular shape is the
low-melting region and the arrows show the change in composition for (C/S) ratios of 0.6 (blue)
1.0 (red) and 1.3 (green)

(i) The Al2O3 pick-up was 25% with losses of 16% SiO2 and 5% MnO in the
slag.
(ii) Different crystalline phases were formed in the TRIP steel casting
(CaF2 + NaCaAlSi2O7 + Al2O3) cf. cuspidine in LC steel casting.
(iii) There was little change in the horizontal heat flux.
High basicity powders
High-basicity slags would appear to be a more logical selection for casting steels
containing Al because they contain (i) high concentrations of CaO to react with the
Al2O3 formed and (ii) lower concentrations of SiO2 to react with the Al in the
metal. However, it can be seen from Fig. 6.25 that the Al2O3 generated will cause

SiO2
(a) (b)

0.667
1

1.5

CaO+ Na2O CaF2 +Al 2 O3

Fig. 6.25 Comparison of cuspidine and mellite—forming slags in a Ternary diagram showing
positions of cuspidine—forming slags ( ) and melilite-forming slags ( ); lines represent different
C/S ratios and b as a quaternary diagram melilite on left; cuspidine on right; (permission granted
for b ISIJ [20])
6.2 Selection of Mould Fluxes 215

the melting temperature of the slag to increase, so high concentrations of fluxes will
be needed to lower the melting temperature.
Melilite—forming powders are high-basicity fluxes [20, 120]. Melilite consists
of a solid solution ranging between end-members akermanite (C2M S2) and
gehlenite (C2AS). It can be seen that gehlenite contains more Al2O3 than aker-
manite, so these slags offer a way of accommodating the Al2O3 into a solid solution.
Melilite has a melting temperature of 1280 °C which can be reduced by additions of
fluxes. Hanao et al. [20] pointed out that melilite slags retain some glassy phase
despite the accommodation of Al2O3 and this provides some measure of lubrication.
Plant trials were carried out using a melilite slag when casting steels containing Al
and gave no problems with either sticker break-outs or false sticker-alarms and
provided improved surface quality [20]. Two examples of melilite-type mould slag
have (C/S) ratios of 1.3 and 1.8 with additions of 8%Na2O and MnO and F 8–13%.
Plant trials carried out on low-basicity (C/S = 0.55; N = 11%; L = 2%;
Mn = 6%; F = 10%) powders resulted in increases in Al2O3 of 5–20% and SiO2
losses of 10% [115]. It was found that the surface of the cast product contained
[115]
(i) No “drag marks” indicating that there was sufficient lubrication provided by
the slag.
(ii) Many horizontal and vertical depressions which contained open cracks (due to
high heat flux in the meniscus region).
Powders based on calcium aluminates [114, 115, 122–124]
Most mould powders are based on the CaO–SiO2 system with added fluxes. An
alternative approach is to base the mould powders on the CaO–Al2O3 system with
added fluxes. The CaO–Al2O3 (CA) system contains a eutectic at a composition
corresponding to C12A7 (ca. 50% CaO) with Tliq = 1413 °C to which fluxes (e.g.
Na2O, Li2O, CaF2) are added to reduce Tliq further. The replacement of SiO2 by
Al2O3 results in a large reduction in the Al2O3 formed. In practice, these CA slags
tend to contain 5–10% SiO2 so some Al2O3 is still formed by reactions of Al with
SiO2, FeO, MnO and B2O3 but it is at a manageable level.
It has been reported that Al2O3 formed during the cast tends to get incorporated
into the AlO4 networks rather than the SiO4 networks via Eq. 6.24 [123].

fSi  O  Sig þ fAl  NBOg ¼ fAl  O  Sig þ fSi  NBOg ð6:24Þ

With regard to the fluxes, CaF2 additions reduce the viscosity but do not break
any Al–O bonds, so this viscosity decrease has been attributed to the liberation of
silicate units (Q0Si) from the aluminate network [123]. It was suggested that CaF2
additions had little effect on the overall polymerisation (i.e. the distribution of Qn
species) or on the number of BOAl units but CaF2 does tend to increase the number
of depolymerised (Q0Al) units [123]. Additions of TiO2 reduce the viscosity and
have been reported to reduce the degree of polymerisation in the AlO4 network
[125, 126].
216 6 Different Types of Mould Powders

The evolution of crystal phases formed in these CA-type slags has been
investigated [127, 128] it was found [127] that the size of crystals increased with
increasing basicity, (C/S) and that B2O3 additions reduced crystallisation (it tends to
promote glass formation).
The principal problems in casting these steels are (i) a loss of glassy phase
leading to lubrication problems (ii) erratic melting with its knock-on effects of
fluctuating mould temperatures and false alarms on the sticker detection system and
(iii) the formation of large slag rims which are fragile and which can fracture and
cause sticker break-outs.
Plant trials carried out on several candidate mould powders (in %; 30C;26A; 2–
10S;10 N; 10–15Fl) indicated that the slab surface contained [115];
(i) Very few longitudinal and transverse depressions and surface cracks (cf.
low-basicity powders) due to the lower horizontal heat flux.
(ii) Drag marks due to inadequate lubrication.
(iii) Less distinct oscillation marks (cf. those obtained with a low-basicity
powder).
Consequently three further candidate compositions were developed [115] [(in %;
38–42C;12A; 9–12S;9–9N; 5–6L;9F; 10–16B) by optimising the following (i) the
CaO/Al2O3 ratio (ii) the Na2O substituted by Li2O and (iii) the %B2O3 (which will
be reduced by Al). The (crystal/glass) ratio and fcrys are reduced with increasing
B2O3 since it promotes glass formation (and lubrication) and Na2O and Li2O which
retard crystallisation; these oxides therefore affect the horizontal heat flux through
their influence on crystallisation. It was found from the subsequent plant trials that
[115]:
(i) The reduction in %SiO2 was <3% but %Al2O3 was greater than expected due
to the reduction of B2O3.
(ii) Drag marks were obtained when using slags providing insufficient glass (and
lubrication) and which led to slag film breakage in the lower mould.
(iii) Depressions and cracks occurred in slags containing too much glass and too
few crystals to reduce the heat flux to an acceptable level.
Overall, the surface quality of TRIP steel slabs was better when using calcium
aluminate slags cf. high basicity calcium silicate slags [115] but further work is
required to optimise the (C/A) ratio and the levels of B2O3, Na2O and Li2O to
provide sufficient glass for good lubrication and sufficient crystalline content to
reduce the heat flux.
Plant trials were also carried out on a calcium aluminate mould powder
[(C/A) = 1.5; in mass % 10S; 13.5N; 6M; 2L; 3Mn; 14 B] and it was found [124]
that
(i) “Flaring” occurred in the mould which was attributed to hydration of the
powder in the spray drying process (CA + xH2O ! CAxH2O); it was found
that baking at 250 °C before use, eliminated this problem.
6.2 Selection of Mould Fluxes 217

(ii) Large slag rims were formed and the properties differed from the expected
values, e.g. η1300°C = 3.5 dPas and ds = 3.6 mm).
(iii) Most of the B2O3 reacted with the Al in the metal (B2O3 sl + 3Al = Al2O3 sl
+3B) with B2O3 in the slag decreasing from 14% to <4%.
Although the use of calcium aluminate slags has produced some encouraging
results, there is still room for further improvement in the selection of the fluxing
agents.
In summary, CA-type slags have proved to be reasonably effective but some
development of the mould powders is still needed to address the poor lubrication
provided by existing powders.

6.2.13 Powders for Casting Stainless Steels

The principal difficulties in casting stainless steels arise because of

(i) The low liquidus temperature of austenitic stainless steels which results,
sequentially, in lower values for the vertical heat flux (qvert) the melting rate,
pool depth (dpool) and lower powder consumption (Qs).
(ii) Reactions occur between Ti in Ti-stabilised steels and the more reducible
oxides (e.g. SiO2) in the slag (Eqs. 6.24 and 6.25).
(iii) High melting oxides (e.g. CaTiO3) can be formed by the reaction of Ti in the
steel with high basicity mould slags and it is necessary to keep the basicity
below 0.9 to avoid the formation of CaTiO3 [42].
(iv) TiN (or Ti(CN)) can be formed in the mould [13–15], in these Ti-stabilised
steels, and since TiN has a very low solubility (<0.5%) in mould slag,
particles do not dissolve significantly, so that TiN particles tend to
agglomerate (cluster) over time to form “lumps” and “clogs” [14, 129]. These
agglomerates hinder the infiltration of liquid slag into the shell/mould
channel and thus, reduce powder consumption; in extreme conditions, they
can cause sticker break-outs.

Ti þ SiO2sl ¼ Si þ TiO2sl ð6:25Þ

2TiO2 þ 4C þ N2 ðgÞ ¼ 2TiN þ 4COðgÞ ð6:26Þ

In mould slags with high basicity, the above reactions can result in the formation
of skin laminations (of CaTiO3 or TiN see Fig. 11–83b) which have high melting
points [42]. The replacement of SiO2 by TiO2 (up to 10%) was found to reduce the
viscosity but increase Tbr and Tliq and the CaTiO3 formed tended to suppress the
formation of cuspidine [130]. Solid particles also increase the slag viscosity.
218 6 Different Types of Mould Powders

6.2.14 Powders for Casting Steels with Rare Earths

Some pig iron emanating from China contains low concentrations of rare earths
(RE’s, e.g. La, Ce, etc.). It has been mooted to incorporate these low concentrations
of rare earths into steels in order to de-oxidise and desulphurise the metal. The
presence of RE’s in the steel would have a similar effect to that of Al, described in
Sect. 6.2.12 above.

2RE þ 1:5 SiO2 ¼ RE2 O3 þ 1:5 Si ð6:27Þ

Similar reactions could be expected with FeO and MnO leading to a build-up in
RE2O3 in the mould slag and Si in the steel (Eq. 6.28 where M = Mn or Fe)

2RE þ 3MO ¼ RE2 O3 þ 3M ð6:28Þ

Additions of RE2O3 to MgO–SiO2 slags resulted in (i) a decrease in viscosity

and (ii) an increase in surface tension [131] because cSiO2 is low cf. cRE2 O3 .
Additions of Re2O3 to mould slags with basicities (C/S) of 0.5, 0.9 and 1.2 were
found to reduce the viscosity but increase Tliq and Tbr [132]. This suggests that
RE2O3 acts as a network breaker in the silicate network. Thus the build-up of
RE2O3 in casting slags should not prove a problem with regard to the viscosity and
lubrication. (Qs = 0.6/ηVc).

6.3 Summary

The various types of mould powders used in continuous casting have been
described from the viewpoint of the functions to be carried out and the underlying
principles in their design and development. The underlying principles for the
selection of chemical composition and properties have been established for most
casting powders. There are two types of mould powders where development is still
active, namely, (i) F-free and F-reduced powders and (ii) powders for casting TRIP
and TWIP steel grades.
For F-free fluxes the primary difficulty encountered lies in replacing cuspidine in
the slag film. The two principal candidates are perovskite (CaTiO3) and melilite and
there have been claims that perovskite forms Ti(C/N) which leads to sticker
break-outs and counterclaims that it does not result in sticker break-outs. This must
be sorted out. Further work is also needed on the suitability of melilite-forming slags.
There are three types of powder currently used for casting steels with high-Al
contents, namely (i) high basicity calcium-silicates slags (ii) low-basicity calcium–
silicate slags and (iii) calcium aluminates slags. At the present time it seems that
there are fewer casting problems when using calcium aluminate slags but further
work is still required to determine the optimum mixture of fluxes in these powders.
References 219

References

1. KC Mills, A review of ECSC –funded research. EUR 13177, (1990) (Europ. Comm, Sci.
Tech. Publ., Luxembourg, 1991).
2. R Shah, JG Williams, G Hecko, Proc. ISS Tech., 2003, Conf., (ISS, Warrendale, PA, 2003)
p. 555.
3. T Sakuraya, T Emi. T Imai, K Emoto, M Kodama, Tetsu-to- Hagane, 67, 1220, (1981).
4. P Andersson. Ironmaking and Steelmaking, 42, 465, (2015).
5. P Andersson, Thermal conductivity of powders used in continuous casting of slabs and
billets. Ironmaking Steelmaking in press
6. S Diehl, JA Moore, RJ Phillips, Proc. 78th Steelmaking Conf., Nashville, TN, 1995, (ISS,
Warrendale, PA, 1995) p. 351.
7. R Saraswat, PhD Thesis, “Modelling the effect of mould flux on steel shell formation during
continuous casting”,” Imperial College, London, (2006).
8. KC Mills, NPL Report DMA (D) 405, Fundamental Study of the behaviour of casting
powders, Final Report EUR 9560, ECSC contract 7210 CA/131/ 311/ 810 (Europ. Sci.
Tech. Publ., Luxembourg, 1985).
9. PE Ramirez-Lopez, KC Mills, PD Lee, B Santillana, Met. Mater. Trans. B, 43B, (1), 109,
(2012).
10. M. Wolf, Proc. METEC Congress’94, 2nd Europ. Conf. Continuous Casting, Dusseldorf,
1994, (VDEh, Dusseldorf, 1994) vol. 1, p. 78.
11. S Ogibayashi, K Yamaguchi, T Mukai, T Takahashi, Y Mimura, K Koyama. Y Nagano,
T Nagano, Nippon Steel Technical Report, 34, 1, (1987).
12. AB Fox, PhD Thesis, Mould fluxes their properties and performance. Dept. of Materials,
Imperial College, London, (2003).
13. T Mukongo, C Pistorius, A Garbers-Craig, Ironmaking Steelmaking, 31, 135, (2004).
14. RC Nunnington, N Sutcliffe, Proc. 59th Electric Furnace Conf., Pheonix, AZ, (2001), p. 361.
15. Q Wang, YJ Lu, SP He, KC Mills, Z Li, Ironmaking and Steelmaking, 38, 297, (2011).
16. S Hiraki, K Nakajima, T Murakami, T Kanazawa, Proc. 77th Steelmaking Conf., Chicago,
1994 (ISS, Warrendale, PA, 1994) p. 397.
17. KC Mills, AB Fox, ISIJ Intl., 43, 1479, (2003).
18. G Skoczylas, Proc. 79th Steelmaking Conf., 1996, (ISS, Warrendale, PA, 1996) p. 269.
19. F Neumann, J Neal, MA Pedroza, AH Castiliejos E, FA Acosta G, Proc. 79th Steelmaking
Conf., 1996, (ISS, Warrendale, PA, 1996) p. 249.
20. M Hanao, Y Tsukaguchi, M Kawamoto, Proc. 4th Intl Cong. Sci. Technol, Gifu, Japan,
2008, p. 694.
21. H Nakato, S Takeuchi, T Fujii, T Nozaki, N Washio, Proc. 74th Steelmaking Conf., 1991,
(ISS, Warrendale, PA, 1991), p. 639.
22. J Macho, G Hecko, B Golomowski, M Frazee, McMaster Iron & Steelmaking
Symp. “Thinner slab casting”, McMaster University, Hamilton, Ont. (2005) p. 131.
23. H Takeuchi, H Mori, T Nishida, T Yanai, K Mukunashi, Trans. ISIJ, 19, 274, (1979).
24. C Lefebvre, JP Radot, JN Pontoire, Y Roux, Revue de Metallurgie, CIT, 94, 489, (1997).
25. Y Nakamura,T Ando, K Kurata, M Ikeda, Trans. ISIJ, 26, 1052, (1986).
26. N Tsukamoto, Taikabutsu (Refractories), 44 (5), 270, (1992).
27. MC Bezerra, CA Valadares, IP Rocha, JP Bulota, MC Carboni, IL Scripnic, CR Santos,
K Mills, D Lever, Proc. 37th Steelmaking Seminar, Porto Allegre, RS- Brazil, 2005, (ABM,
Sao Paulo, 2005) p. 190. http//www.carboox.com/pdf/fluxante_2007.pdf.
28. MM Wolf, Ironmaker & Steelmaker, 2000, (Feb), 65, (2000).
29. MM Wolf, Ironmaker & Steelmaker, 2000, (Mar), 69, (2000).
30. MM Wolf, Ironmaker & Steelmaker, 2000, (Apr), 58, (2000).
31. MM Wolf, Ironmaker & Steelmaker, 2000, (May), 78, (2000).
32. MM Wolf, Ironmaker & Steelmaker, 2000, (June), 22, (2000).
33. MM Wolf, Ironmaker & Steelmaker, 2000, (July), 63, (2000).
220 6 Different Types of Mould Powders

34. MM Wolf, Ironmaker & Steelmaker, 2000, (Aug), 75, (2000).

35. MM Wolf, Ironmaker & Steelmaker, 2000, (Sept.), 90, (2000).
36. MM Wolf, Ironmaker & Steelmaker, 2000, (Oct), 114, (2000).
37. MM Wolf, Ironmaker & Steelmaker, 2000, (Nov), 67, (2000).
38. MM Wolf, Ironmaker & Steelmaker, 2000, (Dec), 45, (2000).
39. MM Wolf, Proc. 81st Steelmaking Conf., 1998, (ISS, Warrendale, PA, 1998) p. 53
40. M. Kawamoto: Paper presented at Intl. Workshop on “Thermophysical data for the
development of mathematical models of solidification”, Gifu City, Japan, Oct. (1995).
41. R Carli, V Ghilardi, Iron and Steelmaker, 1998, (June), 43, (1998).
42. T Kishi, H Takeuchi, M Yamamiya, H Tsuboi, T Nakano, T Ando, Nippon Steel Tech.
Report, 34, 11, (1987).
43. JY Park, SM Jung, I Sohn, Met. Mater. Trans., B, 45B, 329, (2014).
44. M Kawamoto, K Nakajima, T Kanazawa, K Nakai. ISIJ Intl., 34, 593, (1994).
45. K Schwerdtfeger, A Jablonka, Steel Research Intl., 64, 77, (1993).
46. SY Kim, J Choi, Proc. 4th Intl. Conf. Continuous Casting, Birmingham, 2002, (IOM,
London, 2002) p. 594.
47. JN Pontoire, JP Radot, V Delvaux, E Dehaussy, Revue de Metallurgie, CIT. 93, 1237,
(1996).
48. Y Ueshima, T Mizoguchi, T Kajitani, Proc. 9th Intl. Conf. Molten slags, fluxes and salts
(Molten 12), Beijing, 2012, (Chinese Metals Soc., 2012)
49. H Yamamura, T Kajitani, J Nakashima, M Yamasaki, S Mineta, Nippon Steel Technical
Report, 104, 54, (2013).
50. Q Liu, GH Wen, JZ Li, XJ Fu, P Tang, W Li, Ironmaking and Steelmaking, 41, 292, (2014).
51. T Watanabe, H Fukuyama, M Susa, K Nagata, Met. Mater. Trans. B, 31B, 1273 (2000) see
also ISIJ Intl., 42, 489, (2002).
52. M Hanao, M Kawamoto, M Hara, T Murakami, H Kikuchi, Tetsu-to Hagane, 88 (1), 23,
(2002).
53. H Nakada, K Nagata, ISIJ Intl., 46, 441, (2006).
54. M Hayashi, T Watanabe, H Nakada, K Nagata, ISIJ Intl., 46, 1805, (2006).
55. P Grieveson, S Bagha, N Machingawuta, K Liddell, KC Mills, Ironmaking and Steelmaking,
15, 181, (1988).
56. JH Park, JS Park, JY Park, Steel Times Intl., 1998, (Sept), CC14 (1998).
57. PP Sahoo, A Dey, Proc. 9th Intl. Conf. Molten slags, fluxes and salts, Beijing, 2012,
(Chinese Metals Soc. 2012).
58. SH Chang, IJ Lee, MR Kim, SM Yang, J Choi, JS Park, Proc. Conf. on Continuous Casting
in developing countries, Beijing, 1993, p. 832.
59. K Tsukaguchi, S Ura, A Shiraishi, Y Hitomi, T Nagahata, Steel Technol. Intl., 1995/6, 175,
(1995/6).
60. LE From: “Optimisation of mould powder performance in casting long products- Tech Rept
No. 6”. MEFOS Report: MEF99024 K: P3568: Konto 130410 (1999).
61. JW Cho, JK Park, KH Moon, SH Lee, KH Kim, HS Jeong, “ Characteristics of molten mold
flux feeding technology. Proc. AISTech. 2007, Indianapolis, IN, 2007 (ISS Warrendale, PA,
2007).
62. JK Park, JW Cho, KH Moon, SH Lee, KH Kim, HS Jeong, “Study on the initial
solidification behaviour under the new process of molten mold flux feeding technology in the
continuous casting mould.” Proc. 7th Intl. Conf on Clean Steel, Balatonfured, Hungary,
2007, (Hung. Min. Metall. Soc., Budapest, 2007), p. 264.
63. AI Zaitsev, AV Leites, AD Litvina, B Mogutov, Steel Research, 65, 368, (1994).
64. F Shimizu, F Tokunaga, N Saito, K Nakashima, ISIJ Intl., 46, 385, (2006).
65. K Shimizu, AW Cramb, Iron and Steel Maker, 2002, (6), 43, (2002).
66. H Heimbach, K Schulz, J Markardt, HJ Ehrenberg, Stahl u Eisen, 117, 105, (1997).
67. DE Sturgill, Proc. 79th Steelmaking Conf., Pittsburgh, 1996, (ISS, Warrendale, PA, 1996)
p. 265.
References 221

68. RE Hargrave, DW Reichgott, Proc.78th Steelmaking Conf., 1995, (ISS, Warrendale, PA,
1996) p. 385.
69. H Abratis, F Hofer, M Junemann, H Sardmann, H Stoffel, Stahl u Eisen, 116, 85, (1996).
70. AB Fox, KC Mills, D Lever, C Bezerra, C Valadares, I Unamono, J Laraudogoitia, J Gisby,
ISIJ Intl., 45, 1051, (2005).
71. C Zhang, DX Cai, Proc. 8th Europ, Conf. Continuous Casting, Graz, 2014, (Austrian Metals
Soc. 2014).
72. GH Wen, S. Sridhar, P Tang, X Qi, YQ Liu, ISIJ Intl., 47, 1117, (2007).
73. LY Chen, GH Wen, CL Yang, F Mei, CY Shi, P Tang, Ironmaking and Steelmaking, 42,
105, (2015).
74. AB Fox, K Mills, D Lever, M C Bezerra, CA Valadares, I Unamono, J Laraudogoitia,
J Gisby, Proc. 36th Steelmaking Seminar, Vittoria, ES, Brazil, (2005) p. 222 available on line
at http:/www.carboox.com/pdf/paper2005sfs.pdf.
75. Y Lu, GD Zhang, MF Jiang, Advanced Mater. Res., 233/235, 805, (2011).
76. Z Wang, Q Shu, KC Chou, Steel Research, 84, 768, (2013).
77. PS Kharlashin, Azovetal 1999 (4), 34, (1999).
78. Y Lu, GD Zhang, Mater. Sci. Forum, 675/677, 877, (2011).
79. P Tang, GH Wen, ZB Wang, SB Fan, ZB Zhang, SC Li, H Li, Proc. ISS Tech. 2003 Conf.,
(ISS, Warrendale, PA, 2003) p. 567.
80. SY Choi, DH Lee, DW Shin, JW Cho, JM Park, J Non-Cryst. Solids, 15, (8), 157, (2004).
81. Y Lu, X Fang, GD Zhang, Advanced. Mater. Res., 287/290, 1866, (2011).
82. QC Wei, YQ Ding, KD Peng, J Chongqing Univ., 4, 110, (1995).
83. Y Lu, GD Zhang, X Yu, Appl. Mech. and Mater., 71/78, 2899, (2011).
84. Y Lu, X Fang, XF Yu, Advanced Mater. Res., 455/456, 134, (2012).
85. JL Klug, DR Silva, SL Freitas, MMSM Pereira, NC Heck, ACF Viela, D Jung, Steel Res.
Intl., 83, 791, (2012).
86. SP He, X Long, JF Xu, T Wu, Q Wang, Ironmaking and Steelmaking, 39, 593, (2012).
87. Q. Wang, SP He, KC Mills: Proc. 4th Intl. Conf. Cont. Casting Steel in Developing
Countries, Beijing, China (2008). (Chinese Soc. Metals, Beijing, 2008), p. 715.
88. YK Jung, SK Jung, JW Cho, JH Park, “Commercial plant trial of Fluorine-free mold flux for
continuous casting of blooms”. Proc. SEAISI Conf., SEAISI, Singapore, (2014)
89. H Nakada, K Nagata, ISIJ Intl., 46, 441, (2006).
90. JL Klug, R Hagermann, NC Heck, JFC Viela, PR Scheller. Proc. 9th Intl. Conf. Molten slags,
fluxes and salts, Beijing, 2012 (Chinese Metals Soc., 2012)
91. J Klug, N Heck, A Faria, Proc. 8th Intl Conf. Molten slags, fluxes and salts, Santiago, Chile,
2009 (GECAMIN, Santiago, Chile, 2009), p. 1053.
92. H Hashimoto, T Watanabe, K Nagata, CAMP-ISIJ, 17, 849, (2004).
93. X Qi, GH Wen, P Tang, J Non Cryst. Solids, 354 (No 52/54), 5444, (2008).
94. N Takahira, M Hanao, Y Tsukaguchi, ISIJ Intl. 53, p. 818, (2013)
95. A Kusano, N Sato, M Okimori, S Fukunaga, K Nishihara, M Sato, Y Minigawa, Proc. 74th
Steelmaking Conf., 1991, (ISS, Warrendale, PA, 1991) p. 111.
96. S Terada, S Kaneo, T Ishikawa, Y Yoshida, Proc. 74th Steelmaking Conf.,Washington, DC,
1991 (ISS, Warrendale, PA, 1991) p. 635.
97. B. Debiesme, J Radot, D Coulombet, C Lefebre, Y Roux, C Demarval, US Patent No 5 876,
482 (1999) and US Patent No 6, 328, 781, (2001).
98. T Kanazawa, T Marukawa, K Nakai, T Yamada, K Nakajima, Steel Times Supplement,
222 (6), 16, (1994).
99. M Kawamoto, T Kanazaka, S Hiraki, S Kumakura, Proc. 5th Intl. Conf. Molten slags, fluxes
and salts, Sydney, 1997, (ISS, Warrendale, PA, (1997) p. 777).
100. M Kawamoto, K Nakajima, T Kanazawa, K Nakai, ISIJ Intl., 34, 593, (1994).
101. M Hanao, M Kawamoto, M Hara, T Murakami H Kikuchi, A Yamanaka,. Proc. 5th
Europ. Conf. Cont, Casting, Nice,2005, (La Revue Metall., Paris, 2005) see also Tetsu-to
Hagane, 88 (1), 23 (2002)
222 6 Different Types of Mould Powders

102. JA Kromhout, S Meltzer, RW Zinngrebe, AA Kamperman, R Boom, Proc. 9th Intl. Conf.
Molten slags, fluxes and salts, Beijing, 2012, (Chinese Metals Soc.)
103. JA Kromhout, M Kawamoto, M Hanao, Y Tsukaguchi, E Dekker, R Boom, Steel Research
Intl., 80, 575, (2009).
104. JA Kromhout, PhD Thesis, “Mould powders for the high speed continuous casting of steel”,
Univ. Delft, (2011), p. 165.
105. T Matsumiya, ISIJ Intl., 46, 1800, (2006).
106. S Chakraborty, W Hill, Proc. 77th Steelmaking Conf., 1994, (ISS, Warrendale, PA, 1994)
p. 389.
107. J Kromhout, RS Schimmel, Proc 8th Europ. Conf. Continuous Casting, Graz, 2014,
(Austrian Metals Soc.)
108. Y Liu, XD Wang, M Yao, AB Zhang, H Ma, Z Wang, JC Ma, X Wang, GQ Shi, Ironmaking
and Steelmaking, 41, 748, (2014).
109. M Kawamoto, CAMP-ISIJ, 6, 1176, (1993).
110. S Umeda, Y Tsukaguchi, H Mikai, Y Hitomi, M Kawamoto, CAMP- ISIJ, 7, 302, (1994).
111. R Nishimachi, Y Ogura, SEAISI Quarterly, 25, (Oct), 44, (1996).
112. K Tsutsumi, K Watanabe, J Kubota, S Hatori, Y Miki, T Suzuki, T Omoto, Proc.
7th Europ. Cont. Casting Conf., Dusseldorf, 2011, (VDEh, Dusseldorf, 2011) Session p. 1.
113. K Watanabe, K Tsutsumi, M Suzuki, H Fujita, S Hatori, T Omoto, ISIJ Intl., 54, 865, (2014).
114. K Blazek, H Yin, G Skoczylas, M McClymonds, MJ Frazee, AIST Tech. 2011 (AIST,
Warrendale, PA, 2011) also Iron & Steel Technol., 8 (3), 232, (2011).
115. JW Cho, KE Blazek, MJ Frazee, HB Yin, JE Park, SW Moon, ISIJ Intl., 53, 62, (2013).
116. H Yin, G Skoczylas, Proc. ATS Tech. Conf., Indianapolis, 2006 (AIST-ISS Warrendale, PA,
2006) p. 753.
117. WL Wang, K Blazek, AW Cramb, Met. Mater. Trans. B, 39 B, 66, (2008).
118. X Yu, GH Wen, P Tang, H Wang, Ironmaking and Steelmaking, 36, 623, (2009).
119. IH Jung, MA van Ende, Proc. 8th Europ. Conf. Continuous Casting, Graz, 2014, (Austrian
Metals Soc. 2014).
120. Q Wang, JH Chi, B Xie, Y He, B Zhu, W Chen, B Xie, JH Chi, Proc. Conf. Continuous
Casting in Developing Countries, Beijing, 1993 (SEAISI, Singapore, 1993) p 842.
121. Y Tsukaguchi, CAMP- ISIJ, 21, 826, (2008).
122. K Blazek, HB Yin, G Skoczylas, M McClymonds, M Frazee, Iron and Steel Technology,
8, (3), 232, (2011).
123. TS Kim, JH Park, ISIJ Intl., 54, 2031, (2014).
124. Q Liu, GH Wen, JZ Li, XJ Fu, P Tang, W Li, Ironmaking and Steelmaking, 41, 292, (2014).
125. JL Li, QF Shu, KC Chou, Can. Metall. Quart., 54, 85, (2015).
126. QF Shu, Z Wang, J L Klug, K Chou, PR. Scheller, Steel Research, 84, 1138 (2013).
127. MD Seo, CB Shi, JW Cho, SY Kim,. Proc. 8th Europ. Conf. Continuous Casting, Graz,
2014, (Austrian Metals Soc., 2014).
128. JW Cho, SY Kim, SC Moon, H Shibata, Proc. 6th Intl Conf. Molten Slags, Fluxes and Salts,
Stockholm, Helsinki, June 2000, Paper 154, Filename 341.pdf
129. P Misra, S Sridhar, AW Cramb, Met. Mater. Trans. B, 32B, 963, (2001).
130. Z Wang, QF Shu, XM Hou, KC Chou, Ironmaking and Steelmaking, 39, 210, (2012).
131. F Shimizu, F Tokunaga, N Saito, K Nakashima, ISIJ Intl., 46, 385, (2006).
132. DY Wang, M Jiang, CJ Liu, PY Shi, J North Eastern Univ. (Natural Sci.), 26, (11), 1082,
(2005).
Chapter 7
Fluxes for Ingot Casting

Abstract Mould powders for ingot casting have, in many ways, the same function
as mould powders for continuous casting. However, in some aspects the properties
and compositions must meet requirements other than those for continuous casting.
After the introduction of continuous casting on a bigger scale and adaptation of
mould powder for this technology, the development of ingot casting powder has
mainly been based on experience derived from the development of CC mould
powders. Thus, there has been little development in casting powder research and
ingot casting technology for more than 30 years. The remaining steel grades that are
cast by the ingot process route today tend to be high-grade, high-quality steels and
there is a new, re-awakened interest in the development of both ingot casting
technology and mould powders for ingot casting. Today most ingot casting is
carried out with killed steel cast into bottom end up (BEU) ingots. Thus, the most
important properties of the mould powders are good flowability and good thermal
insulation. The creation of a mould powder slag is important but the thickness and
composition of this must be designed according to the steel grade being cast and the
process conditions which require careful control of the composition and melting
speed. Most ingot powders are still produced with fly ash as a base but the use of
synthetic mould powders is growing because of its greater versatility in composi-
tion. Granulated powders for ingot casting are also attracting a growing interest,
since they are more environmentally friendly, causing much less dust during
casting. They also give better flowability, compared to loose synthetic powders, and
open up the opportunity of continuous feeding during casting.

Symbols, Abbreviations and Units

a* Capillary constant
Bi Basicity index
g Acceleration due to gravity (ms−2)
k Thermal conductivity (Wm−1 K−1)
a* Absorption coefficient
qFe Density of liquid steel (kg m−3)
qsl Density of liquid mould slag (kg m−3)

© Springer International Publishing AG 2017 223

K.C. Mills and C.-Å. Däcker, The Casting Powders Book,
DOI 10.1007/978-3-319-53616-3_7
224 7 Fluxes for Ingot Casting

cm Surface tension between liquid steel and gas above it (mNm−1)

cmsl Interfacial tension between liquid steel and slag (mNm−1)
Tbr Break temperature (°C)
Tliq Melting temperature (°C)

Abbreviations, Subscripts and Superscripts

CC Continuous casting

7.1 The Ingot Casting Process

The dominant casting process today is continuous casting but still around 8% of the
World’s annual steel output consists of ingot cast steels. There are a number of
different motives underlying the use of ingot casting (instead of continuous casting),
these are:
– Lack of investment capital to build equipment for continuous casting which is
much more expensive compared to ingot casting.
– Large products such as rolls, shafts, etc., which can be produced by hot forging.
– Product programme with high demands on reduction rate, such as thick heavy
plate.
– A product programme with a great variety of different dimensions.
– Steel grades which are very sensitive to surface cracks.
Ingot casting is carried out in cast iron moulds with square, round or polygon
cross-sections. Ingots with square cross section are used for rolling into billets, rails
and other structural sections or for hot forging to their final shape. Ingots with
rectangular cross section (also known as slabs), are mostly used for rolling into flat
products but also for forging of flat products. Round ingots are used for tube
making. Polygon ingots are used in the production of tyres, wheels, etc. Typically,
ingots weigh 5–20 tons for rolling, but range from 100 up to 300 tons when used
for forging.
Cast iron is used to fabricate the mould which consists of grey iron, ductile iron
or malleable iron. The moulds are essentially of two types;
– Wide-end-up moulds are used to produce ingots of killed carbon or alloyed
steels.
– Narrow-end-up moulds are commonly used to produce rimming and semi-killed
steel ingots.
The casting from the steel ladle can be made in two ways, either by top pouring
or bottom pouring. In top pouring (or downhill teeming) the mould is simply filled
directly from the ladle into the individual ingot. In bottom pouring (or uphill
7.1 The Ingot Casting Process 225

Ladle

Argon protection

Trumpet

Hot top cover

Hot top side liner

Iron mould
Mould powder

Bottom plate

Runner bricks

Fig. 7.1 Schematic diagram showing the bottom pouring of ingots (courtesy of Uddeholms AB)

teeming) one or (mostly) more ingots are placed on a bottom plate (Fig. 7.1). The
steel is poured via a sliding gate from the ladle into a trumpet where the open stream
mostly is protected with an argon shield. The trumpet, made of iron, is thermally
protected by a tube formed from bricks. This connects to a distributor at the bottom
which distributes the steel stream to the mould inlet via the rectangular runner
bricks. The steel is covered with mould powder which normally is added from
paper sacks hanging about 30 cm from the bottom which open spontaneously when
the paper burns in response to the radiation emitted from the incoming steel.
The thermal conductivity and thickness of the mould and bottom plate are
sufficient to keep the surface temperature at a secure low temperature, and thereby
avoid sticking between the steel and mould/bottom plate. So almost immediately a
steel shell is formed when the steel comes in contact with the mould/bottom plate.

7.1.1 Classification of Ingot Cast Steels

The ingot casting procedure is classified according to the deoxidisation of the steel
melt; thus, there are three groups, namely; rimmed, semi-killed, and killed steels.
Rimmed steel is also known as drawing quality steel, and has little, or no,
deoxidising agent added to it during steelmaking which causes carbon monoxide to
evolve rapidly from the ingot during solidification. This causes small blow holes to
be formed at a certain distance from the steel surface which are later closed up
during the hot rolling process. Another consequence is the segregation of elements
where most of the carbon, phosphorus, and sulphur move to the centre of the ingot,
leaving an almost perfect “rim” of pure steel on the outside of the ingot. This gives
the ingot an excellent surface finish because of the steel rim, but also results in a
226 7 Fluxes for Ingot Casting

strongly segregated ingot. Most rimmed steels have a carbon content below
0.25 wt-%, a manganese content below 0.6 wt-%, and are not alloyed with alu-
minium, silicon and titanium.
These types of steels are commonly used for sheet and strip metal because of their
excellent surface condition and are commonly used for cold-bending, cold-forming,
cold-heading and, as the name implies, drawing. Due to the non-uniformity of
alloying elements they are not recommended for hot-working applications. The yield
of rimmed steel is slightly better than that for semi-killed steel.
Semi-killed steel is incompletely deoxidised (with silicon) and contains a suf-
ficient level of dissolved oxygen to react with the carbon to form carbon monoxide
which creates blowhole-type porosity distributed throughout the ingot. The porosity
eliminates the pipe found in killed steel and increases the yield to approximately
90% by weight. Semi-killed steel is commonly used for structural steel with carbon
contents between 0.15 and 0.25 wt-% of carbon. Since pipe cavities are minimised,
semi-killed steels are usually cast in big-end-down moulds without hot-tops. The
quality of semi-killed steels is not as good as killed steels because of the large
amount of manganese–silicates formed during steel-making. These do not readily
separate from the metal and consequently, they tend to remain in the final ingot.
Killed steel is completely deoxidised by the addition of an agent in the ladle
before casting; consequently, the level of oxygen is so low there is practically no
evolution of gas during solidification. These steels are characterised by a high
degree of chemical homogeneity and freedom from larger gas porosities. The steel
is said to be “killed” because it will quietly solidify in the mould, with no gas
bubbling. Common deoxidising agents include aluminium, ferrosilicon and man-
ganese. Aluminium reacts with the dissolved oxygen to form aluminium oxide.
Aluminium also has the added benefit of forming pin grain boundaries, which
prevent grain growth during heat treatments. Killed steels suffer from deep pipe
shrinkage defects. Thus, in order to minimise the amount of metal that must be
discarded because of the shrinkage, the upper part of the ingot walls and top
surface, must be provided with a so-called “hot top” for thermal insulation.
A summary of the different steel types is made in Table 7.1.
Cross sections of ingots with different states of deoxidisation are schematically
shown in Fig. 7.2 where it must be pointed out that the killed steel in (a) is cast
without hot top insulation. The dots in the figures for the rimmed and semi-killed
steels are carbon monoxide blow holes.
The introduction and expansion of the continuous casting process has, to a large
extent, put an end to the casting of rimmed and semi-killed steels. This is exem-
plified by Sumitomo [1] who reduced the amount of rimmed steel for cold-rolled
sheet production to 2% by the mid-80s and completely stopped the production in
1995 (Fig. 7.3). The high segregation in rimmed steels and the high amount of
manganese–silicate inclusions in semi-killed steels have led to the situation where
almost all ingot casting today is carried out using killed steels.
7.1 The Ingot Casting Process 227

Table 7.1 Summary of steel types with different deoxidation treatment

Rimming; no de-O Semi-killed; de-O by Si Killed: de-O by Al, Fe/Si,
Mn
Steel C < 0.25%; C = 0.15– 0.25% Low alloyed and high
composition Mn < 0.6% alloyed carbon steels
Steel use Sheet, strip Structural Tool steel, rolls, shafts,
ball bearing etc.
Gas CO(g) ! blowholes; O % sufficient ! CO(g) None
evolution closed in rolling blowholes through ingot,
closed in rolling
Type of Top pouring; without Top pouring; Bottom pouring for most
casting hot top insulation big-end-down; without cases
hot top insulation Top pouring for very large
ingots with multiple
pouring
Hot top insulation
Defects Porosity in steel: S,P, C Blowholes Deep pipe shrinkage-
segregates to ingot Mn silicate inclusions Minimized with hot top,
centre surface cracks, macro
segregates
Yield Greater than >90% 80–85%
semi-killed
Steel Inferior to killed steel
quality

Fig. 7.2 Schematic figures showing cross sections of steel ingots for a killed steel, b semi-killed
steel, c rimmed steel
228 7 Fluxes for Ingot Casting

Fig. 7.3 Schematic diagram showing changes in steel production of cold-rolled sheet [1]

7.1.2 Ingot Casting of Killed Steels

As mentioned above, when casting killed steels, a very deep pipe is formed (see
Fig. 7.2a) in the upper part of the mould due to the solidification shrinkage caused
by the density differences between liquid and solid steel. The density of ordinary
carbon steel at room temperature is: q = 7850 kg/m3 and as liquid at the melting
temperature: q = 7010 kg/m3. In order to eliminate the deep pipe, a technique to
insulate the upper part of the mould has been developed, the so called “hot top”.
This consists of hot-top side liners of ceramic material, with good insulating
properties, in the upper part of the mould. These are mounted in the mould in one
piece for small ingots or in four or more sections for larger ingots. The top part of
the ingot must also be thermally insulated, this is achieved with mould powder
during the casting and normally, with exothermic powder, or board, at the end of
casting. The hot top cover contains aluminium which makes it exothermic and also
special graphite, which makes it expand and thereby improves its insulating
properties. For larger ingots, with long solidification times, it is also common to
further increase the insulation capability by adding insulating material, such as
perlite or rice ash, to the top of the exothermic powder/board after it has burned
through. A picture from a steel plant with hot top boards, where the ingot on the
front right has just burned through, is shown in Fig. 7.4.
7.1 The Ingot Casting Process 229

Fig. 7.4 Photograph showing hot top with exothermic boards; note that the board has just burnt
through for ingot at bottom right

With a good insulated hot top, a reservoir of molten steel is created at the top of
the ingot which feeds the ingot during its final solidification and shrinkage. This is
shown in Fig. 7.5 from a simulation made with the software THERCAST® from
Transvalor.
A properly installed hot top will eliminate the pipe and produce an almost flat
top surface as shown in Fig. 7.6.
The yield of killed steels is normally around 80–85% and is mainly restricted in
the upper part because of heavy macrosegregation especially for large ingots with a
long solidification time. This is exemplified in Fig. 7.7 for a 65 ton steel ingot with
a carbon content of 0.22 wt-% [2].
Traditionally, all ingots were top poured directly from the ladle into the mould,
(so-called downhill casting) which is the cheapest way to produce ingots. When
new, more advanced, steel grades were introduced, there was a need for better ingot
quality which was achieved by the introduction of bottom pouring or uphill casting.
The pros (+) and cons (−) of uphill casting (cf. downhill casting) are:
+ Better surface quality due to low amount of splashes during casting and the use of
mould powders.
+ Better steel quality and fewer inclusions since there is less re-oxidation (no open
steel stream).
+ Higher casting rate, and simultaneous casting of many ingots.
+ Lower wear of moulds and bottom plates.
− Additional costs for extra material such as trumpets and refractory material.
− Extra working steps for preparation of casting and cleaning operation after
casting.
230 7 Fluxes for Ingot Casting

Fig. 7.5 Schematic drawing from a simulation of ingot solidification using THERCAST® from
Transvalor, the colour scale represents the liquid fraction where red is fully liquid

Fig. 7.6 Photograph

showing top surfaces from
casting of killed steel with a
well-functioning hot top
7.1 The Ingot Casting Process 231

Equiaxed cr ystals

Positive segregation

Columnar crystals

Equiaxed globular crystals

Negative segregation

Fig. 7.7 Schematic drawing showing structure pattern (left) and a map of final average mass
fraction of carbon (on right side of figure) [2]

Today’s demand for high-quality steel products has forced almost all ingot
casting steel makers to use the uphill casting method.

7.2 Aspects of Importance for Ingot Casting Quality

More information on the ingot casting process is needed before any recommen-
dation can be made on which mould powders to use. Thus, a more detailed
description of the ingot casting process and its limitations are presented below to
provide a better understanding of the use and composition of mould powders.

7.2.1 Surface Quality

To understand the function of mould powders and their effect on the surface quality,
it is important to have an idea of the situation at the first stage of solidification in the
meniscus area.
232 7 Fluxes for Ingot Casting

Fig. 7.8 Schematic drawing

showing the liquid steel
meniscus in the mould [3]

This part is very well described by Tomono [3] who made in situ studies of
solidification on both an organic melt (with analogous solidification behaviour to
steel) and on steels (in where the meniscus was observed through a quartz window).
When the bottom of a mould is filled with liquid steel, a meniscus is formed in
contact with the mould wall, as shown in Fig. 7.8.
For a straight wall, the shape of the meniscus can be calculated [4] according to
the following equation (note this equation is not valid for corners or round format):
pffiffiffi
2 a a  2 þ ð2a2  y2 Þ0:5
X ¼ 2ða  y Þ 2 0:5
þ pffiffiffi ln þ 0:3768 a ; ð7:1Þ
2 y

where x is the horizontal and y the vertical distance from meniscus.

a* is the capillary constant defined as:

2cm
a ¼ ð7:2Þ
g ðqFe Þ

cm = surface tension between liquid steel and the gas above it

qFe = density of liquid steel
g = acceleration due to gravity
7.2 Aspects of Importance for Ingot Casting Quality 233

If the steel surface is covered by a mould powder slag, the capillary constant can
be derived from the following expression:

2cmsl
a2 ¼ ð7:3Þ
g ðqFe  qsl Þ

cmsl = interfacial tension between liquid steel and slag

qsl = density of liquid mould powder slag.
Increasing interfacial tension leads to increases in both the radius of curvature
and the total height of the meniscus.
If the meniscus shape is simplified to that of a quarter of a circle, the following
simplified calculation of the radius can be derived [5]:

r ¼ 1:7  ½cm = ðqFe  qSl Þ  gÞ0:5 : ð7:4Þ

Most mould powders used in uphill ingot casting tend to have a high carbon
content which results in a low melting rate during the teeming operation. They
function mostly as a heat insulating cover but also produce a reducing atmosphere
to protect the steel surface from oxidation. The surface tension of liquid steel in
contact with this type of gas (CO) is rather high, (1500–1850 mN/m) due to the low
soluble oxygen content but is sensitive to the sulphur content.
During teeming the meniscus partially solidifies and forms a shell. This shell
grows slowly in the meniscus region because of the surface/interfacial tension,
which holds the shell away from direct contact with the mould. During the teeming
process the first shell continues to increase in thickness, where the shell thickness
achieved is dependent on the following variables:
• Fluid flow. Where a strong fluid flow leads to a thin solid shell.
• Teeming speed. Where a quick teeming leads to a thin solid shell.
• Superheat. Where high superheats also lead to a thin solid shell.
• Interfacial tension. A high interfacial tension leads to a long meniscus and a
long, but weak, solid shell.
• Steel grade. This is dependent on both chemical composition (which determines
the way of solidification) and the amount of micro segregating elements, which
reduce the high temperature strength and thus, determine the strength of the
steel.
The steel meniscus grows upward as the steel pours into the mould. This is
different to continuous casting where the meniscus (i.e. shell) is both withdrawn
downwards and subject to mould oscillation. This difference is reflected in the need
for lubrication of the shell since lubrication is not important for gradual, upward
movement in ingot casting but is essential in handling the stresses generated by the
vigorous movement in the CC mould.
234 7 Fluxes for Ingot Casting

When the steel meniscus is covered by a thin pool of liquid slag, gravity feeding
would be expected to cause some infiltration of liquid slag into the shell/mould
gap. However, the high viscosity of the slag of most ingot casting mould powders
would tend to restrict the slag infiltration.
Another difference between ingot and continuous casting lies in the horizontal
heat transfer. In CC the thermal gradients across the slag film are large and the
cooling rates are large because of a water-cooled copper mould. In ingot casting
both the thermal gradients and the cooling rates are much smaller in the cast iron
mould and hence the need to control heat extraction from the shell is much less
demanding than in CC. However, in ingot casting control of the vertical heat flux is
important to stop the steel meniscus from freezing during casting.
The shell forms periodic marks (so-called ripple marks) during the teeming. Two
different types of ripple mark formations can be observed depending on the cir-
cumstances [3].

7.2.1.1 Ripple Mark Formation Due to Folding

This mechanism is schematically described in Fig. 7.9.

In this case, the steel shell formed at the meniscus is so weak that the “meniscus
shell” is bent towards the mould surface by increasing static pressure due to the
increase in the height of the bath (cycles 2–5). As the meniscus is pressed against
the mould (cycle 6) the shell starts to solidify immediately. The next cycle of ripple
mark starts to form when the shell has sufficient strength to withstand the ferrostatic
pressure.

Fig. 7.9 Schematic diagram showing typical cycle of ripple mark formation due to folding [3]
7.2 Aspects of Importance for Ingot Casting Quality 235

7.2.1.2 Ripple Mark Formation Due to Overflow

In this case, the solidified shell in the meniscus region is so strong that only the top
part is bent towards the mould surface with increasing static pressure. If the bath
level rises to a point which is higher than a certain critical position, the surface
tension can no longer support the pressure and liquid steel overflows the solidified
meniscus shell and fills the gap between mould surface and meniscus shell, thereby
forming a “hook”. This case is typical for casting at too low speed, low superheat or
a steel meniscus with insufficient insulation or protection against oxidation.
(Fig. 7.10).
In Tomono’s [3] in situ studies of uphill-teemed steel, shell growth in the
meniscus was varied by changing (i) the gas atmosphere (thus changing, simulta-
neously, both the surface tension and the heat transfer between steel and mould),
(ii) mould material (thus changing the heat transfer) and (iii) the superheat of the
melt. These changes affected both the pitch and the depth of the ripple mark, as
shown in Fig. 7.11.
The experimental results can be summarised as:
• Increasing solidification in the meniscus area resulted in increases in both the
depth and the pitch of the ripple mark.
• Both depth and pitch of folding marks (extent of solidification) increased with a
lower surface temperature of the mould. In the above example, the mould
temperatures were 190 °C for Copper (CU), 342 °C for Cast Iron and 615 °C
for Stainless Steel (SUS).
• The depth and pitch decrease when the superheat in the mould is high (42 °C).
• Changing the atmosphere above the surface from air to hydrogen (H2) caused
two changes, which both influence the depth and pitch of the ripple mark:

Fig. 7.10 Schematic drawing showing the formation of ripple marks by the overflow mechanism
[3]
236 7 Fluxes for Ingot Casting

Fig. 7.11 Schematic diagram

showing the relation between
depth and pitch of folding
marks. SUS Stainless steel
mould, CU copper mould, H2
and AIR is the gas
atmosphere respectively,
Dt superheat [3]

– First and probably the most dominant variable is that the thermal conduction
is changed by a factor of ca. 7 from a low value in air to a high value in
hydrogen (see Table 7.2).
– Secondly the interfacial tension is changed from approx. 500–800 mN/m in
air (dependent on the extent of absorption of oxygen and the formation of an
oxide layer) to approximately 1500 mN/m for hydrogen; this will result in an
elongation of the meniscus.
Similar results have been reported in another investigation [6] in which steel was
uphill teemed in a mould consisting of three sides of ceramic material and one side
of water-cooled copper. The mould was installed in a vacuum furnace, in which,
vacuum or different gas atmospheres could be chosen. In Fig. 7.12 the surface
temperature of the solidifying shell on the copper side is shown for casting trials
with low carbon steel.

Table 7.2 Thermal Thermal conductivity (°C), mWm−1K−1

conductivity of some gases
from 27–337 °C at 1 bar Gas 27 °C 127 °C 227 °C 337°C
pressure Ar 17.9 22.6 26.8 30.6
O2 26.3 32.3 39.2 45.7
H2O (steam) 18.7 27.1 35.7 47.1
H2 186.9 230.4
H2S 14.6 20.5 26.4 32.4
Air 26.2 33.3 39.7 45.7
CO 25.0 32.3 39.2 45.7
7.2 Aspects of Importance for Ingot Casting Quality 237

Comments:
• The casting trials were made without mould powder. This means that the heat
flux from steel to the copper mould consists only of radiation and conduction.
• In the upper curve in which the casting is made in vacuum, the heat flux results
solely from radiation.
• The other curves clearly reflect the differences in thermal conductivity for their
respective gases in which they were cast according to Table 7.2.
In Fig. 7.12 it can be seen that the surface temperature decreases more rapidly
for the PH2 O curve than for the PO2 curve despite the fact that kO2 [ kH2 O . This
anomaly is due to the oxidation of the steel which then provides a thermal resis-
tance; the oxidation and thermal resistance will be greater for O2 than H2O thus the
temperature decrease will be more rapid for H2O than O2.
From some of the castings, photographs were taken from the surface (shown in
Fig. 7.13). The same pattern as that shown in Fig. 7.11 is also visible here. Casting
in a H2 atmosphere creates a large and strong solidification at the meniscus due to
high values of the surface tension and thermal conductivity which, in this case,
leads to deep ripple marks of the overflow type.
Casting in an oxygen atmosphere both lowers the surface tension but, above all,
reduces the heat flux compared with that for a casting in a H2 atmosphere. This
results in shallow ripple marks probably of the folding type.

Fig. 7.12 Schematic diagram

showing surface temperatures
as a function of solidification
time [6]
238 7 Fluxes for Ingot Casting

Fig. 7.13 Surface appearance for castings in different gas atmospheres: a H2 b H2S c O2 [6]

7.2.1.3 Uphill Teeming with Mould Slag

If a molten slag is put on top of the steel bath, a number of things change:
• The interfacial tension is reduced by an amount, dependent on the chemical
composition of the slag. Interfacial tension (cmsl) values for mould powder slags
lie in the range 900–1300 mN/m but are very dependent upon the S content of
the steel and can be lowered by chemical reactions between steel and slag [7].
The shape of the meniscus will also change as a consequence of the density of
the mould powder slag.
• If the mould powder has a high melting rate and/or a low viscosity, the mould
slag can partially fill the gap between mould and ingot. The thermal conductivity
of an amorphous mould slag is around 1.1 W/m, K at 300 °C compared to a
value of 0.044 W/m, K for air or carbon monoxide at the same temperature. This
will increase the thermal conductivity by a factor of ca. 25. In practice, the heat
flux will be lower since some of the heat flux by radiation will be absorbed by
the high FeO content in the mould slag [8]. This effect has an impact on the
initial solidification when the distance between the steel shell and mould is very
small and the gas conductivity has a major role and convection is suppressed by
the thin gap. When the air gap (or gas gap) has developed, the contributions
from radiation (along with some convection) to the heat flux (between steel
surface and ingot) will be dominant. This is well illustrated in Fig. 7.14 [9] from
7.2 Aspects of Importance for Ingot Casting Quality 239

1000
900 Air gap formation
800
700

h, W/m2,K
600
500
400
300
200
100
0
0 5 10 15 20
Heat transfer between ingot surface and mould

Fig. 7.14 Heat transfer between ingot and mould as function of solidification time. ● = measured
values,  = calculated heat transfer by radiation, ---- = calculated heat transfer by gas
conduction with a gas gap of 1 mm. - - - - - = calculated heat transfer by gas conduction with a gas
gap of 0.1 mm. Re-drawn from [9]

measurements taken on a wide side of a mould where it can be concluded that,

once a gas gap has developed, radiation is the dominant heat transfer mecha-
nism, aided, to some extent, by gas convection.
The overall effect is that meniscus solidification, the pitch and depth of the ripple
marks all increase (Fig. 7.15) with high rates of heat transfer.

Fig. 7.15 Example of an ingot with strong ripple marks due to mould slag entrainment between
mould and ingot [9]
240 7 Fluxes for Ingot Casting

7.2.2 Inner Quality

The mechanical properties of high alloyed steel, produced by the ingot route, are
controlled to a large extent, by the volume fraction, distribution, composition and
morphology of inclusions and precipitates. Inclusions act as stress raisers with
larger inclusions being much more detrimental than small ones. Ingot-cast steel can
contain a great number of different types of inclusions; it is customary in steel
terminology to classify inclusions by their origin [9, 11].

7.2.2.1 Exogenous

This type of inclusion arises from:

• Entrained slag (mould slag, slag from re-oxidation).
• Erosion products from ceramic lining (ladle, trumpet, runner).
• Entrapped hot top materials.
• Re-oxidation products.
• Entrained mould powder.
• Sand from the preparation of the runner bricks.

7.2.2.2 Endogenous

Inclusions of this type arise from:

• A-segregates, which arise due to the flow of solute-rich, interdendritic fluid via
thermosolutal convection. They are characterised in the final solidified
microstructure as channels of enriched solid, often with near-eutectic
composition.
• V-segregates are found in the centre of the casting during the final stage of ingot
solidification which is usually occupied by a network of loosely connected
equiaxed grains. It is thought that V-segregates arise due to the fissuring of such
networks under the combined actions of a metallostatic head (i.e. the weight of
the material above) and the solidification shrinkage; this leads to the formation
of open shear planes that can fill with any remaining liquid. This remaining
liquid will have become enriched throughout solidification.
• Positive segregation.
• Negative segregation.
• Oxidation products formed in the micro segregates.
The mould powder does not affect all factors affecting the internal quality (e.g.
V-segregates) but does influence the following:
– Entrained mould slag.
– Re-oxidation of mould slag.
7.2 Aspects of Importance for Ingot Casting Quality 241

Fig. 7.16 Illustration of the

flotation of Al2O3 inclusions
to the steel/slag interface [12]

– Ability to absorb non-metallic inclusions of all origins that will rise to the steel
surface during casting and solidification.
The principal non-metallic inclusion in ingot casting of aluminium-killed steel is
alumina (Al2O3) which is formed by the de-oxidation of steel with Aluminium. It
could also form as the result of the reaction between dissolved Al in the steel and
reducible oxides (such as FeO, MnO and SiO2) in the mould slag or by contact with
the air from an unprotected steel surface.
For Al2O3 inclusions to be absorbed by the melted mould powder, the inclusions
have to reach the steel/slag interface as shown in Fig. 7.16 [12].
The dissolution process starts once the inclusions reach the interface and have
been absorbed by the slag. The absorption of Al2O3 inclusions in the mould slag
phase results in property changes of the slag, as follows [12, 14]:
– The activity of Al2O3 will increase with absorption.
– Solid absorbed Al2O3 clusters will be dissolved and then diluted into the slag
which will increase its viscosity.
– The accumulation of undissolved alumina clusters, as a second solid phase in
the slag will increase its viscosity and will result in non-Newtonian behaviour.
– Once the Al2O3 has reached the saturation concentration in the slag, the ther-
modynamic driving force for the absorption will be zero.
– The heterogeneous Al2O3-saturated slag will prevent further dissolution of
inclusions into the slag.
From a metallurgical point of view, an ideal mould powder slag with good
absorption ability should have the following composition [12]:
– The basicity index, Bi (based on the oxygen-ion attractive forces and defined in
Eq. 7.5 [13]) should be in the range of 1.5– 2.0.
– A viscosity in the range of 2–7 dPas at 1300 °C.
– The initial content of Al2O3 in the slag should be less than 15 wt-%.
242 7 Fluxes for Ingot Casting

– For Al-killed steels (especially when Si-content is low) a slag with a low SiO2
content and a very low FeO content is preferred.

1:53ðCaO þ MgO þ CaF2 Þ þ 1:94Na2 O þ 3:55Li2 O

Bi ¼ ð7:5Þ
1:4SiO2 þ 0:1Al2 O3

7.2.3 Macro Segregation (Hot Top Insulation)

As shown in Fig. 7.2 the yield (from ingot to final product) for the top part of the
cast is much restricted by the thermal efficiency of the hot top and the extent of the
macro segregation. The mould powder influences the thermal efficiency of the hot
top in two ways:
• Endothermic reactions of the remaining mould powder on the top of the ingot
will increase the heat losses.
• Absorption of radiation improves the thermal insulation of the mould slag but is
dependent upon its iron oxide content.
The heating, melting and the decomposition of carbonates in the mould powder
are all endothermic reactions. The combustion of carbon is exothermic with the
formation of CO2(g) providing almost 3 times the heat supplied by the formation of
CO(g). Although the mould powder contains a large amount of carbon, most of the
carbon forms a reducing atmosphere of CO(g), (C + O ! CO) in the mould
powder bed and it is only later that some of the CO is oxidised above the ingot
surface to form CO2. Thus the heat generation is limited and the carbon combustion
is inefficient. Even with a high-carbon, fly-ash powder the overall effect of the
powder is endothermic. This is clearly shown in an investigation where the thermal
insulation of the exothermic hot top covering powders were studied [15] using the
experimental equipment shown in Fig. 7.17. In these experiments a plate of SiC
was heated to 1425 °C using super Kanthal elements while the plate was insulated
with ceramic fibre. When the target temperature was reached and had stabilised, the

Fig. 7.17 Schematic drawing

of the equipment for testing of Hearth
thermal insulation of covering
powders [15] Thermocouples

Insulation,
ceramic
fibre

Super Kanthal elements

7.2 Aspects of Importance for Ingot Casting Quality 243

5000
4500 (a)
(b)
4000 (c)
(d)
3500
3000
Joule/cm 2

2500
2000
1500
1000
500
0
0.00 10.00 20.00 30.00 40.00 50.00 60.00
Minutes

Fig. 7.18 Schematic diagram showing the accumulated heat losses from tests with exothermic
covering powder. a 10 kg exothermic powder, b 20 kg exothermic powder, c 5 kg mould powder
+20 kg exothermic powder, d 10 kg mould powder +20 kg exothermic powder. Re-drawn from
[15]

insulation was taken off and the exothermic covering powder was applied. The
SiC-plate, provided with thermocouples, was regulated to maintain the temperature.
The additional energy required to maintain temperature was measured to monitor
the thermal insulation provided by the powder.
The test equipment was designed to enable one to test a combination of mould
powder and exothermic powder and the result from such a test is shown in Fig. 7.18
where the mould powder consisted of fly-ash powder.
The results given in Fig. 7.18 indicate the following:
• For the powder containing 10 kg exothermic powder, the exothermic reaction
provides insufficient heat to compensate for the energy needed to heat the
powder from room temperature to the test temperature, 1425 °C.
• Larger amounts of exothermic powder (20 kg) decrease the heat losses and have
negative heat losses, during the period when the exothermic reactions occur.
• The combination of the fly-ash mould powder with exothermic powder
increases the heat losses, especially during the first 40 min of the test.
• Increasing amounts of mould powder result in increasing heat losses.
When the mould powder finally melts, the molten slag cover affects the sub-
sequent heat losses from the steel surface since the heat flux from the steel surface is
influenced by the properties of the molten mould slag. Heat transfer through the
slag involves three mechanisms, namely:
• Conduction.
• Convection.
• Radiation.
244 7 Fluxes for Ingot Casting

The conduction is rather low for mould powder slags so the largest contribution
to the heat flux consists of convection and radiation. The convection contains two
contributions (i) convection in the molten slag cover and (ii) gaseous conduction
through the layer of any unmelted powder above the liquid pool. The convection in
the liquid slag is dependent on the gradient (Tsteel/slag−Tslag/powder)/dslag) and the slag
viscosity, where (Tsteel/slag−Tslag/powder) represents the temperature difference
between the interfaces of the liquid slag with the liquid steel and the unmelted
powder and dslag represents the depth of the liquid slag layer. The gaseous con-
duction in the unmelted powder layer will be affected by both the temperature
difference across the slag layers, the depth of the powder layer and the packing
density (or permeability) of the layer.
Radiation conduction (kR) across the liquid layer is high because it is propor-
tional to TK3, where TK is in K and is close to the liquidus temperature of the steel.
The magnitude of kR is inversely related to the absorption coefficient (a*) of the
mould slag which increases significantly with % transition metal (e.g. FeO), as
shown in Fig. 7.19 [16, 17]. The figure gives important information showing that
the absorption of Fe2O3 is not so high but will increase strongly if the slag contains

Fig. 7.19 Absorption

coefficients for CaO–SiO2–
Al2O3 slag containing iron
oxides obtained in studies
from Refs. [16, 17]
7.2 Aspects of Importance for Ingot Casting Quality 245

FeO. The thickness of the liquid slag is also important. Values of kR will be lower in
the unmelted powder layer than in the liquid layer; this is due to the lower tem-
peratures, the TK3 relation and the much higher absorption.
Fly-ash powders normally have a high FeO content. An exchange from a
high-viscosity, fly-ash powder to high-viscosity mould powder with a low content
of Fe2O3 (which will react during melting with the carbon in the powders:
Fe2O3 + C ! 2FeO + CO) will therefore result in higher heat losses from the steel
surface; this is due to less absorption of the radiation. If, on the other hand an
exchange is made to a high basicity slag with low iron content this will crystallise
on cooling leading to significant reflection and scattering of the irradiation.

7.3 History of the Development of Mould Powders

for Ingot Casting (and CC)

In research and development it is important to know where it all started and the
history of the first development; in the case of mould powder development, the first
work was carried out for ingot casting, which is why a thorough description of that
work is made in this chapter.
Traditionally ingot casting was made by top pouring of various (rimmed,
semi-killed) steels and there were no need for mould powders. Splashing and bad
surfaces on the bottom part of the ingot were general problems which were found to
be reduced by surface coating of the mould surfaces with tar or other carbon
additives. However, one attempt to add mould powder in connection with top
pouring has been reported [18].
At the introduction of bottom pouring, surface cracks were a great problem
because of the oxidation and the cooling of the unprotected steel surface (see
Fig. 7.20). A laboratory scale, investigation from 1957 [19] revealed that surface
dressings improved the surface quality of steel, especially if the dressings were
volatile, since:
– They prevent re-oxidation of the steel meniscus.
– They keep surface films or slag away from the interface.
– They prevent excessive decarburisation of the ingot surface.
– They prevent sticking of splashed steel to the mould wall.
– They promote a smooth metal rise in the mould.
It was concluded from this study that bottom pouring required a steel surface
covered with a non-oxidising atmosphere.
In the beginning of the 1960s more plants carried out tests on bottom pouring of
killed steels and it was found that some sort of insulation was needed to protect the
steel surface from freezing. From the inspired work of Thornton [19] it was obvious
that this must contain re-oxidation material. However, the amount of steel being
246 7 Fluxes for Ingot Casting

Fig. 7.20 Schematic illustration of a correct action of an active mould dressing b action of an
active mould dressing on a crazed mould [20]

cast by bottom pouring in the mid-1960 was very low and only three steel plants in
USA practised this technique.
In England they suffered from bad ingot surfaces with major defects such as
hanger cracks, corner cracks and scabs. Thus, surface planing with an oxygen lance,
or grinding, was necessary and could result in a yield loss of around 5%.
For many years, active dressings were applied in the manufacture of killed steels,
to reduce these problems. Dehydrated gasworks tar was an effective dressing but it
was difficult to apply an even coating of this material. Several problems were
encountered as a result of its high viscosity at ordinary temperatures, sprays
blockages and maintenance proved difficult. This led to the development of alter-
native lacquers of less viscous, hydrocarbon derivatives.
The study by Thornton [19] clearly showed that a correctly applied volatile
hydrocarbon burns in advance of the steel meniscus, producing a slightly reducing
atmosphere above the liquid steel surface. Furthermore, the current of hot gases
prevented the oxide film from reaching and adhering to the mould (Fig. 7.20a).
This technique improved the surface quality of bottom-poured, killed steel but
caused problems when older moulds with crazed cracks were used. Application of a
fluid to a crazed mould wall resulted in the dressing being absorbed, by capillary
attraction, into the fissures. Under these circumstances, the dressing is shielded
from the heat of the rising metal and does not volatilise until the metal comes
adjacent to the mould defect. The sudden evolution of gas caused turbulence within
the mould and raised the hydrocarbon content of the rapidly solidifying, surface
layer. This led to the formation of subcutaneous blowholes [20] as shown in
Fig. 7.20b.
When production started of grain-size-controlled, killed steels containing alu-
minium, for all types of alloyed steels, the active dressings were unable to produce a
7.3 History of the Development of Mould Powders for Ingot Casting (and CC) 247

sufficiently reducing atmosphere to prevent oxide layer formation. This layer

eventually became entrapped, causing laps with associated hot tears, blowholes and
subcutaneous refractory inclusions.
Thus, there was a need for the improvement and development of better surface
dressings with the following requirements [20]:
• The dressing must prevent atmospheric oxidation of the steel and its alloys.
• It should contain no readily volatile matter which might get entrapped below the
rising liquid level thereby causing blowholes.
• It should prevent teeming scum from becoming entrapped in the ingot surface.
In addition, the dressing should act as a good thermal insulator of the steel bath
surface.
Rippon [20] was the first to describe the use of mould powders which were both
insulating and which would partially flux at a temperature just below that of liquid
steel. He commented that:
• The method, using mould powder, is simple to control, is reproducible and
produces an ingot with fine surface quality when correct teeming rates are used
on bottom-poured material.
• Perhaps the most remarkable effect of these powders, however, is the production
of a good ingot surface when using crazed moulds; this is illustrated in
Fig. 7.21.

Fig. 7.21 Schematic illustration when a mould powder is used during teeming into a mould with
a crazed surface [20]
248 7 Fluxes for Ingot Casting

The mould powder described by Rippon was based on fly ash, a by-product from
coal combustion. This was revealed in an article by Alexander [21] where he also
explained the problems that could be met by using this product. “Fly ashes vary
considerably in composition, form, particle size and impurities, according to the
fuel mix burnt, the efficiency and type of combustion furnace, the methods of
separation from the flue gases by cyclone, electrostatic or washers and lagoons, and
thus of 50 examined, only two or three are possible starters for this application”.
Alexander [21] was very enthusiastic regarding the use of mould powders and
put forward the following advantages in using this new technique:
(1) Elimination, or minimising, of subcutaneous cavities and inclusions.
(2) Elimination of cold laps and the improvement of the general surface
smoothness.
(3) Minimisation of heat losses with the additional advantage that low tapping and
pouring temperatures can be used whilst still producing first-class, quality ingots.
It should be noted that in the early days of using these powders, it was common
practice to leave an opening in the centre of the steel surface but subsequent work
showed the importance of maintaining a complete cover of the steel.
In the USSR pure graphite powder were tested and used on a regular basis which
was reported from a steel plant [22] with the following result:
– Treating killed steel with graphite powder while it was being bottom-cast in
large plate, ingot moulds hinders the development of oxide films on the metal
meniscus; this improves the surfaces of the ingots, reducing the amount of
teeming laps and gas blowholes, also (to a lesser degree) the amount of cracks.
– The use of graphite during the teeming of killed steel at the Orsk-Khalilovo
combine reduced the proportion of slabs and plates rejected and also the number
of products with incorrect dimensions. Providing the amount of graphite does
not exceed 250–350 g/ton of steel, the metal is not carburised.
Probably inspired by the work carried out in England, tests were also made with
fly-ash powder in USSR [23]. It was observed that a normal fly ash, with low
carbon content, fused rapidly during the casting process. This had a markedly
adverse effect on the heat-insulating properties and gave rise to defects on the ingot
surface.
To avoid this problem, they produced a mixture of fly-ash material and graphite
with a ratio of ash (75–85%) to graphite (25–15%) controlled so as to produce a
carbon content in the mixture of not less than 24% which was found to be optimum.
If the carbon content was lower than 24%, it was difficult to obtain good flowability
and spreading of the ash–graphite mixture; this made it difficult to ensure thermal
insulation of the entire surface of the steel which rises in the big-end-up ingot
moulds. A rim of molten metal incidentally appeared on the mould walls which
could adversely affect the ingot surface quality.
Tests were carried out where a number of differently killed steel grades (con-
ventional grades, low-alloyed steel, chromium and chromium–nickel steels) were
7.3 History of the Development of Mould Powders for Ingot Casting (and CC) 249

cast with the ash-graphite powder. The results were then compared with those
obtained using the old technique employing a mould dressed with tar. The fol-
lowing advantages were identified from two years’ experience of using fly-ash–
graphite powders:
– There was a great reduction of billet rejects due to surface defects.
– There was a reduction of top discard of 5–10 kg/ton steel.
– There was a reduction of metal rejected at customers’ factories.
– There was an improvement in mould life by 1.5 kg/ton of steel.
Fly ash was accordingly established as a good material for mould powder
production and is still the dominant raw material used for ingot-casting mould
powders. It was soon recognised [23] that the fly ash could only be satisfactorily
used in regular production if it was used as a base material into which other
materials could be mixed to obtain the desired properties [24, 25]. Fly ash consists
mainly of Al2O3, SiO2 and FeO to which both carbon and sodium carbonate are
added; the latter decreases the viscosity and increases the melting speed. It was
found desirable to have a thin slag cover on the steel surface so as to avoid
oxidation if open areas occurred, without powder cover, during the teeming oper-
ation. The slag layer counteracts carburisation of the steel by eliminating direct
contact between the carbon-rich powder and the steel surface.
The reason for fly ash being so successful, as a base for mould powder pro-
duction, lies not only in its low price [25] but also because it is a binary phase
having the right chemicals and carbon structure. Adequate grain size, combined
with the particle form, gave both good insulation and flowability. Only a few
fluxing agents are necessary to achieve the right melting point.
Although, the first articles regarding mould powders were written in UK and
USSR, the real pioneer in their development was Hans Joachim Eitel, the founder of
Metallurgica. His PhD Thesis from 1990 [26] is a detailed description of the
development of mould powders both for ingot and continuous casting and a brief
summary of his work is worth bringing out into the light.
Eitel states that in the development of bottom-pouring, ingot casting, the
introduction of mould dressing was not optimum because it introduced restrictions
regarding the pouring rate. For a good casting, a high pouring rate was needed but
that increased the risk for longitudinal cracks. If a low pouring rate was used the
steel surface tended to freeze during the casting which lead to internal quality
problems in the top section causing reduction in the yield.
To overcome this problem a number of different insulation materials, to be put
on top of the steel surface, were tested such as:
– Wood shavings impregnated with water glass.
– Boards made of wood.
– Strips of corrugated cardboard.
– Rice ash.
250 7 Fluxes for Ingot Casting

None of these materials gave the desired result, so in 1957 the first trial was
made with a pure fly-ash powder at the Belgium steel plant Cockerill Seraing. This
showed good results and it was shown that it was possible to cast at a very low
casting speed, which eliminated longitudinal cracks, without getting freezing of the
top surface, due to the very good thermal insulation properties of the fly ash.
At the beginning, pure fly ash was used but at an early stage of the tests it was
recognised that it was desirable to produce a molten slag layer to avoid fly-ash
powder coming in direct contact with the steel surface (and thereby carburising it).
As mentioned above, fly ash consists mainly of SiO2, Al2O3 and FeO to which soda
ash (Na2CO3) is added to lower the melting point of the powder, creating a high
viscosity slag.
The first production of mould powder for ingot casting was started in February
1958 with fly ash from Kraftwerk Osterfeld and later also Kraftwerk Wehlheim and
Schlägel & Eisen in Recklinghausen. Around 10–15 wt-% fine grained soda ash
was added to the fly ash and then mixed. This was delivered to those steel plants in
Germany using bottom pouring, i.e. Dillinger Hüttenwerke, Stahlwerk
Phoenix-Rhein-Ruhr, Stahlwerk Südwestfalen and Edelstahlwerke Buderus.
The production of mould powder increased during the following years and it was
found that the powder was not only a good product for ingot casting but could also
be used as surface dressings for the ladle and tundish.
Different recipes were developed for mixtures of fly ash, soda ash and carbon
additions (such as coke breeze) to meet the needs of different steel grades and
casting practice. The need to add coke breeze was identified because of the great
variations in carbon content in the fly ash. As the power plants became more
efficient the carbon content fell and carbon additions became essential.

7.3.1 Development of Mould Powders for Continuous

Casting

The development of mould powder for ingot casting was also the basis for the
development of powders for the continuous casting process. This started on a small
scale for billet casting in the 1950s and in 1961 mould powders were used for the
first time in the slab machine at Dillinger Hüttenwerke. The introduction of the
continuous casting technique was slow in the beginning due to technical problems,
large investment costs and the need for higher education of both operators and
maintenance staff. The percentage of world steel production being continuously cast
rose sharply in the 1970s and 80s from 14.2% in 1975, to 30% in 1980 and 54.8%
in 1987. However, in 1987 the CC-cast proportion in Germany was 87.9%, in
France 93.1% and in Japan 93.3% [27].
When Dillinger Hüttenwerke started to operate their slab machine they needed
lubricant to prevent sticking of the solidifying shell in the mould; this, traditionally,
7.3 History of the Development of Mould Powders for Ingot Casting (and CC) 251

was performed by using vegetable oil. This technique, for slab casting had a
number of drawbacks [38] such as:
– Unstable distribution.
– Hydrogen and carbon pickup.
– Large amount of breakouts.
– High depression frequency.
The much calmer and wider steel surface in a slab machine probably increased
the problem (cf. billet casting) with its greater need for thermal insulation of the
steel surface.
Due to these problems, as an initiative from Dillinger Hütte, a mould powder
was developed in 1963 dedicated for continuous casting which was based on the
experience obtained from ingot-casting powder development. However, in addition
to providing good thermal insulation properties, the mould powder slag was
required to act as a lubricant (i.e. to replace the vegetable oil).
This put new demands on both the development and quality assurance of the
mould powder, since a poorly performing mould powder had a much greater
financial impact, especially, if it resulted in a breakout. In the beginning, only
silicon-killed steels (with low addition of aluminium) were continuously cast and it
worked fairly well for these steels. A standard mould powder for ingot casting,
containing a mixture of fly ash and soda was used. When they started to cast
Al-killed, deep-drawing steels the situation changed dramatically and led to the
introduction of the first, low-viscosity, mould powder for continuous casting which
contained a mixture of; 10% fluorspar, 7.5% soda, 30% Portland cement and the
rest fly ash.
The rapid spreading of the continuous casting technology, especially in Europe
and Japan, went hand in hand with the development of mould powders for CC and
became the prime interest. Further development of new ingot casting powders was,
to a large extent, based on the experience derived from the development of powders
for continuous casting.

7.3.2 Development of Synthetic Mould Powders

In the beginning it was considered that fly-ash powder contained all the necessary
components needed for a mould powder, i.e. SiO2, Al2O3, CaO, Fe2O3 and C.
However, it was realised that this restricted range of components limited better
design of the mould powders (e.g. the high FeO content of fly ash was undesirable
when casting Al-killed steels). The increasing demand for better properties of CC
powders led to the development of mould powders which excluded fly ash. These
powders were a mixture of a number of different raw materials and are frequently
referred to as “synthetic mould powders” to distinguish them from the fly ash-based
252 7 Fluxes for Ingot Casting

powders. The raw materials and production technology of synthetic mould powders
are described in Chap. 8.

7.3.3 Development of Granulated Powders

In 1976 the Rautaruukki steel plant in Finland required a CC mould powder which
should be dust-free and thus was needed in an agglomerated form. To start this
development, (led by Hans Joachim Eitel [26]), a list of demands were drawn up
(Table 7.3).
With this background, different techniques for granulation were studied and two
options were identified: i.e. roll granulation or spray-granulation.
Spray granulation was found to be the better method for two reasons:
– It resulted in a product with lower density.
– Lower water content in the granules and no need for subsequent drying.
At that time, spray granulation had been developed for the chemical- and food
industry, so it seemed to offer a reasonably quick route in the development of mould
powder granules (although it was an untried technology with no documented
activities). With this background Eitel [26] in 1978 received a patent for granulation
of mould powders with the title “Verfahren zur Herstellung von Giebpulver”. Eitel
was looking for a suitable factory which could be rented for trials with mould
powder and found the company Dorst in Kochel am See which produced granules
for the ceramic industry and in November 1978 the first trial was made to produce
mould powder granules.
With optimisation of the spray technique regarding nozzle design, composition
of the slurry and spray pressure, a reasonably good product could be manufactured.
In the beginning, around 150 ton/month was produced for the first customers for the
granules; i.e. the steel plants of Rautaruukki (Finland), Sollac (France) and BSC
(UK).
The advantages of granules over loose powder were soon identified and the word
was quickly spread among the continuous casting community. In 1983 the pro-
duction was 1300 ton/month. With this success of the new product, a factory
dedicated to the production of mould powder granules was built with a capacity of
1500 ton/month which was started in 1983. The increased demand for this product

Table 7.3 Demands for a new granulated powder generation

Mould powder Mould slag
Good thermal insulation properties Optimum viscosity depending on steel grade and
due to low density process parameters of the casting.
Dust free Protection of the steel surface from re-oxidation
Good flowability Good absorption ability for non-metallic inclusions
Low water content Good lubricating ability between steel shell and mould
Good homogeneity Good and even heat transfer from steel surface to mould
Suitable melting properties
7.3 History of the Development of Mould Powders for Ingot Casting (and CC) 253

lead to an increase in production and an investment in a second spray dryer in 1990.

It also included sales of licenses of the technology to companies in Japan and USA.

7.3.4 Today’s Situation Regarding Mould Powders

for Ingot Casting

After the rapid increase in continuous casting and the development of mould
powders specifically adapted for this technology, the development of ingot casting
powder has mainly been based on experience from CC mould powder development.
The literature of this subject has been sparse compared to the corresponding lit-
erature of continuous casting and were mostly published during the 1970s and 80s
[24, 25, 28–37] with concerns to find the optimum mould powders for ingot casting.
After that, for a period of more than 20 years, almost no research was published on
mould powders for ingot casting; the only exceptions being [38, 39].
The remaining steel grades that are cast via the ingot process route today are
high grade, high quality steels. This has sparked a new, awakened interest in the
development of both ingot casting technology and mould powders for ingot casting.
The need for more research was identified by Stahlinstitut, VDEh in Germany;
consequently, they arranged the first international conference on ingot casting,
rolling and forging (ICRF) 2012 in Aachen which was followed up by ICRF 2014
in Milan, Italy. These conferences resulted in a number of new proceedings con-
cerning mould powders for ingot casting [12, 40–43].

7.4 Selection of Mould Powders for Ingot Casting

To understand the influence mould powders have on the quality of ingot cast steels
one must distinguish between the two different materials involved.
• Mould powder.
• Mould slag.

7.4.1 Important Properties of the Mould Powder

The three key physical properties for ingot powders are:

– Melting behaviour.
– Flowability.
– Thermal insulation.
254 7 Fluxes for Ingot Casting

7.4.1.1 Melting Behaviour

A number of variables are known to influence the melting behaviour (see Chap. 4)
such as:
Steel; Steel temperature, steel flow (pouring rate), ingot geometry.
Mould powder; Mineralogical composition, carbon content, carbonate content
and effective thermal conductivity.
Addition of free carbon to the mould powder is the most efficient way to control
the melting speed. While mould powders are composed of a mixture of mineral
components with high melting points and fluxing agents, such as CaF2 and Na2O,
carbon particles delay the contact and the sintering of oxide particles. The carbon
also delays the agglomeration of molten slag droplets. The melting speed is con-
trolled by both the amount and the size of the particles (where a finer carbon is
much more effective (per unit mass of C) in reducing melting speed).
Mould slag irradiative absorptivity; This is the ability of the slag to absorb the
IR radiation emitted by the steel shell through the molten mould slag to the cov-
ering, hot-top powder. The heat flux is influenced by the thickness of the slag and
the concentration of transition metal oxides (e.g. FeO) in the slag. The slag viscosity
also affects heat transfer from the steel bath to the powder by convection in the
liquid slag since convection is inversely proportional to the slag viscosity.

7.4.1.2 Flowability

The ability of the mould powder to spread out over the steel surface is very
important in order to ensure that the entire surface is covered with mould powder
throughout the casting. It is important to prevent both the formation of an “open
eye” and subsequent re-oxidation of the steel. Flowability is not so critical for
smaller ingots but is important for larger and, especially, big-end-up (BEU) ingots
where the negative ingot taper requires increased spreading of the powder as the
pour proceeds. High-carbon, fly-ash, mould powders normally show a good
flowability, whereas synthetic powders tend to exhibit poorer flowability [36].
Granulated mould powders normally have a good flowability because of their
narrow particle size distribution.

7.4.1.3 Thermal Insulation Properties

Heat is lost from the molten steel by radiation, conduction and by liquid and
gaseous convection. Radiation is reduced by ensuring the FeO content of the slag is
high to absorb the infra-red radiation; alternatively, a high-basicity, crystalline slag
could be used to reflect and scatter the radiation where high FeO contents are
unacceptable (e.g. in Al-killed steels). Liquid convection is reduced by using a
high-viscosity, fly-ash slag. Gaseous convection through the powder and sinter
decreases with increasing bulk density; the latter, in turn, is increased by using
7.4 Selection of Mould Powders for Ingot Casting 255

Fig. 7.22 Thermal 0.35

conductivity of a granulated
0.3
mould powder as function of
temperature measured by the 0.25

k, W/m 2,K
transient hot wire method [40]
0.2

0.15

0.1

0.05

0
0 200 400 600 800 1000 1200 1400 1600
Temperature, ºC

granules of smaller size. The insulation properties of mould powders are normally
very good as witnessed by the low thermal conductivity (Fig. 7.22 [40]) for a
granulated mould powder. It must be noted that the thermal conductivity in reality
is somewhat higher due to convection from the gases evolved during heating
(described in more detail in Sect. 9.6.4).

7.4.2 Important Properties of the Mould Powder Slag

The most important properties for the mould powder slag are the following:
– Viscosity.
– Ability to absorb inclusions.
– Irradiative absorption properties.
– Interfacial tension.

7.4.2.1 Viscosity

The viscosity determines the infiltration of liquid slag into the gap between the solid
steel and the mould and there is a large difference in viscosity for different oxide
systems. The viscosity is very dependent upon the degree of polymerisation of the
silicate or alumina–silicate network and upon temperature. A typical viscosity curve
for a high viscosity, commercial soda-lime silica glass composed of 75 wt-% SiO2,
15 wt-% Na2O and 10 wt-% CaO is shown in Fig. 7.23 [44].
The viscosity curve is very steep and the dependence on temperature is strong
due to the strong network bonding of SiO44− tetrahedra in the glass.
Experimental, viscosity curves from high-basicity, blast furnace slags are shown
as a comparison in Fig. 7.24 [45]. The compositions of these slags were 36–47 wt-
% SiO2, 32–43 wt-% CaO, 13.0 wt-% MgO and 8.0 wt-% Al2O3 and the silicate
256 7 Fluxes for Ingot Casting

Fig. 7.23 Typical curve for

viscosity as function of
temperature for a
soda-lime-silica melt [44]
showing characteristic points
such as the glass transition
temperature (Tg).

Viscosity, Poise
(Reproduced by permission of
The Royal Society of
Chemistry)

network of the slag is much less polymerised than that of the glass shown in
Fig. 7.23.
The temperature dependence is also much lower for the blast furnace slag
compared to the glass because the melt is much more depolymerised due to a high
concentration of CaO and MgO. The marked increase in viscosity at around 1375 °
C is due to precipitation of crystals in the slags which is defined as the break
temperature (Tbr).
Due to a high amount of SiO2 and Al2O3 in most fly-ash-based mould powders,
the temperature dependency of the slag viscosity is similar to that shown for glassy
slags in Fig. 7.23. In contrast, the viscosity–temperature relation for a typical
high-basicity, synthetic mould slag has the same appearance as that for the blast
furnace slag but has a higher degree of de-polymerisation which leads to:
– The viscosity is lower than the blast-furnace slag.
– The temperature dependence of viscosity is very low until the crystallisation (or
break) temperature is reached.
– The break temperature is lower than that for the blast-furnace slag.

7.4.2.2 Ability to Absorb Inclusions

For clean steel production the mould slag can act as a flux to dissolve high-melting
oxides and inclusions, (such as Al2O3, de-oxidation and re-oxidation products); this
prevents these oxides from coalescing and possibly becoming trapped in the ingot
surface. Emulsification and slag entrainment become easier with lower interfacial
tension, lower slag viscosity and higher slag density [11].
7.4 Selection of Mould Powders for Ingot Casting 257

Fig. 7.24 Temperature

dependence of viscosity (in
dPas) of various blast furnace
slags [45]

7.4.2.3 Irradiative Absorption Properties

A molten mould slag has a rather low thermal conductivity of around 0.4 W/m, K
[8]. Of more importance is the ability to absorb radiation which is very much related
to the content of FeO. Heat flux by radiation is mainly transferred within the
infrared region 0.7–3.0 µm. The absorption coefficient of a mould slag at a
wavelength of 2 µm can be calculated by the following formula, [8]:

a ¼ 30 þ 910 ð%FeOÞ þ 5 ð%MnOÞ þ 410 ð%NiOÞ þ 390 ð%Cr2 O3 Þ þ 370 ð%Cr2 O3 Þ:

ð7:6Þ

7.4.2.4 Interfacial Tension

The interfacial tension between mould powder slag and the liquid steel influences
two important factors:
• The size of the meniscus, which, in turn, affects the size of ripple marks. A low
interfacial tension results in a shorter meniscus, leading to shallow ripple marks,
which normally is preferred and even to their disappearance.
• The emulsification of the mould slag. If the interfacial tension is low there is a
risk that slag drops get dragged from the mould slag surface and can be trapped
in the surface of the steel shell as macro inclusions. The risk for this to happen is
greater when the fluid flow is high (as at the start of casting and for higher
258 7 Fluxes for Ingot Casting

pouring rates). On the other hand a mould slag with a low interfacial tension has
a greater ability to absorb inclusions coming from the steel.

7.4.3 Selection of Mould Powders in Regard to Steel Grade

The principal tasks which the ingot powder is expected to perform are:
– To protect the steel surface from oxidation.
– To provide thermal insulation from the surface to prevent the steel from freezing
during casting.
– To absorb inclusions from the steel during and after casting.
Unlike mould powders for continuous casting, ingot powders are not expected to:
– Lubricate the ingot/shell surface—so powder consumption is not an issue here
except consumption must be sufficiently low to ensure that the thermal insula-
tion remains intact for the entire cast.
– To provide controlled, horizontal heat transfer and solidification.
So what properties are desirable in an ingot mould powder?
– Tliq; ought to be low but not at the expense of producing a high melting rate.
– Viscosity; should be low for good Al2O3 absorption but this will possibly lead to
high consumption and a risk of mould slag floating in between the mould and
ingot causing both increased heat transfer and slag patches on the ingot surfaces.
So there is a balance and the viscosity must be chosen according to the steel
grade being cast.
– FeO-content; a high content will decrease thermal radiation during and after
casting but means an increased risk of Al2O3 formation in Al-killed steel, so the
trend today is to lower the FeO-content in mould powders for ingot casting.
– Interfacial tension; a low interfacial tension provides better surface finish (less
deep ripple marks) assisted by high Na2O and a high S content in the steel.
However, low interfacial tension leads to more slag entrapment and to scale
formation during the rolling operation.
– Melting speed; in most cases a low melting speed is desirable; this ensures that
there is a good insulation of the steel surface during casting without much mould
slag penetrating between the mould and the ingot.
Today there are two types of mould powders on the market, fly-ash-based and
synthetic powders. Most powders are still, fly-ash, loose powders but granulated
powders are being used to a higher extent mainly because:
– They are more environmentally-friendly with much lower emissions of dust
during casting.
7.4 Selection of Mould Powders for Ingot Casting 259

– The flowability of synthetic powders are normally inferior to the fly-ash pow-
ders but are improved with granulation, due to the narrower fraction distribution.
The selection of mould powders is to a large extent dependent upon the steel
grade [33] but still fly-ash-based mould powders with high carbon content tend to
dominate sales for the following reasons:
– Low price.
– Very good insulation properties.
– Low consumption rate which will secure a thick insulation blanket throughout
the whole casting.
– High melting temperature which reduces the risk of slag entrapment during the
early, turbulent, stage of the casting due to low slag formation at this stage.
– High irradiative absorption of the molten slag which reduces the heat losses
from the top of the mould during solidification thus making the hot top more
effective.
A typical composition of a fly-ash-based, high carbon, mould powder is shown
in Table 7.4 and its properties are listed in Table 7.5.

Table 7.4 Chemical SiO2 (Wt-%) 30.0–40.0

composition of a typical
CaO+MgO (Wt-%) 2.0–8.0
fly-ash mould powder
Al2O3 (Wt-%) 15.5–21.5
Na2O+K2O (Wt-%) 7.5–10.5
FeO (Wt-%) 4.0–10.0
Cfree (Wt-%) 19.0–21.0
CO2 (Wt-%) 3.0–5.0
Ctotal (Wt-%) 20.0–22.0
H2O600° (Wt-%) <0.8
Basicity <0.2

Table 7.5 Properties of a Softening point, °C 1080 ± 30

typical fly-ash mould powder;
Melting point, °C 1260 ± 30
* denotes calculated value
Fluidity point, °C 1390 ± 20
Viscosity1300 °C, dPas* Approx. 600
Absorption. Coeff., m−1* Approx. 5000–10,000
Sieve analysis, 500 lm <1.0
Sieve analysis, 125 lm 12.0–18.0
Sieve analysis, 63 lm 26.0–36.0
Bulk density, kg/m3 600–700
260 7 Fluxes for Ingot Casting

The drawbacks with the fly-ash powder are its low basicity and the high amount
of iron oxide. The latter leads to a risk of alumina formation during the casting by
the reaction:

3FeO þ 2Al ! Al2 O3 þ 3Fe ð7:7Þ

The thickness of the liquid slag on top of the ingot will become very shallow
because of the high melting point of the powder and the low melting rate (due to the
high carbon content). It may be expected that the reaction rate for alumina gen-
eration will be reduced by the high slag viscosity; however, the ability to absorb
inclusions will also be very low for the very same reasons.
The recommendations for selection of mould powders for ingot casting related to
steel grades are as follows.

7.4.3.1 Low to Medium Carbon, Low Alloyed Steels

The primary functions of the mould powders are to protect the steel meniscus from
re-oxidation and heat loss, and to provide a good cover during the whole teeming
process. Fly-ash, high carbon, mould powders are capable of performing these tasks
satisfactorily.

7.4.3.2 High Carbon (HC) Steels

It is possible to use a similar powder for high carbon steels, (>0.8 wt-%) to that
used for low carbon, low alloyed, steels with one exception. Since HC steels have a
much lower liquidus temperature and hence, a low pouring temperature, it is
important to ensure that a liquid slag layer is formed to cover the steels. Thus, the
melting properties must be improved by adding more fluidisers (i.e. soda or
fluorspar) for casting HC steels.

7.4.3.3 Tool Steels

The demands on tool steel are very high, especially with regards to non-metallic
inclusions; since the mechanical properties of the steel deteriorate with increases in
both the number and size of the inclusions. Thus, a mould powder forming a mould
slag with good capability to absorb inclusions is desirable. This calls for a
high-basicity powder with a low FeO content.
The result is a low-viscosity mould powder slag and a mould powder with higher
melting speed. The higher melting rate of the powder can lead to some drawbacks
7.4 Selection of Mould Powders for Ingot Casting 261

which must be taken into consideration when designing the powder in terms of
carbon content and the magnitude of mould powder additions.
– A high consumption of mould powders can result in open steel areas without
powder cover in the middle of the ingot (“open eyes”) which poses a greater risk
of creating alumina.
– Excessive melting can result in a very deep slag pool with slag entrainment
between mould and ingot which can be detrimental to the surface quality and
can also lead to problems in the heating furnace.
– A low-viscosity slag has greater ability to absorb non-metallic inclusions but it
is also more reactive and if its sulphur content is too high there is a risk of
increased sulphur pick-up by the steel and more sulphides in the final product.
– A mould powder slag with low iron content will have lower thermal absorp-
tivity, thus, heat losses by radiation from the top surface during solidification
will be larger which will have an impact on the hot top efficiency.

7.4.3.4 Ultralow Carbon Steels

For ultra-low carbon, (<0.02 wt-%,) steels, it is important to avoid carbon pickup
during casting and total carbon of the mould powder should not exceed 1 wt-%.
This, in turn, will increase the melting rate of the mould powder and ensure that a
thick slag layer is formed, which will minimise the contact between mould powder
and steel. To avoid excessive melting of the powder, with the risk of slag patches on
the surface, the basicity and the amount of fluidiser ought to be lower than that used
in the mould powder for tool steels.

7.4.3.5 Stainless and High-Alloy Steel

During casting of stainless and high-alloy steel grades, it is normal to generate a

high concentration of refractory oxides. Absorption of non-metallic inclusions is
therefore very important to avoid severe surface defects. A high-basicity powder is
recommended with good melting properties entailing the addition of fluidisers such
as Na2O and CaF2 in order to ensure a thick molten pool on the steel surface during
casting. The carbon content should be low to avoid carbon pickup and to increase
the melting rate.
An example of recommended mould powder selection from a large producer of
mould powder for ingot casting is shown in Table 7.6.
262 7 Fluxes for Ingot Casting

Table 7.6 Typical properties of mould powders for ingot casting (courtesy of ALSICAL
Hüttenwerkstechnik GmbH)
Brand name PC10–PC27* PC2 or PA1 PA3 PEx1
GC10–GC27** GC2
Range of application
Steel grade Unalloyed and Tool Stainless ELC-steels Heating
alloyed carbon steels steels steels conductor
Steels with material
Al and Ti
Ingot size All 0.5– All All 0.5–2.0
(ton) 2.0
Consumption 1.5–1.8 2.0 2.5 2.0 6.0
(kg/ton)
Chemical composition, % (approx.)
Cfree 10–27 18 <4 <0.3 <0.5
SiO2 33 23 43 50 25
CaO/MgO 8 30 30 40 25
Al2O3 18 9 9 4 20
Na2O/K2O 8 9 6 2 4
Fe2O3 2–5 3 2 1 15
H2O <0.6 <0.6 <0.6 <0.4 <0.4
Physical parameters
Density 0.6 0.7 0.6 0.4 1.0
(kg/dm3)
Sintering 1060 1030 1150 1100 Exothermic
point (°C)
Melting point 1260 1150 1200 1300 Exothermic
(°C)
Fluidity point 1320 1170 1220 1320 Exothermic
(°C)
Remarks *Carbon content depends on ingot size and steel grade. **P powder, G granule

7.5 Application Techniques for Mould Powders

Several techniques are available to apply the mould powders to the steel surface.
Normally, mould powders are hung in paper bags around 300 mm from the bottom
plate. This is shown in Fig. 7.25 where four 5 kg bags are hung from the hot-top
tile wedges in a 10 ton ingot mould. When the steel enters the mould, the strong
radiation from the steel surface sets the paper bags on fire and the mould powder
pours out from the bags. It subsequently spreads over the steel surface.
Another way of applying the mould powder [25], is shown in Fig. 7.26 where
the mould powder bag is placed on an expanded cardboard shape. This will
eventually burn, which ensures that the mould powder reaches the steel surface
when a certain amount of steel has been filled in the ingot.
7.5 Application Techniques for Mould Powders 263

Fig. 7.25 Photograph showing 10 ton ingot mould prepared with mould powders

Fig. 7.26 Diagram showing

mould powder on an
expanded cardboard shape
[25]

Attempts have also been made to add the mould powder as a preformed board as
shown in Fig. 7.27 [25]. This method has not become widespread probably because
most ingots today are big end up (BEU) and this means that the board will not cover
264 7 Fluxes for Ingot Casting

Fig. 7.27 Schematic drawing

showing a pre-formed board
of flux material [25]

the full meniscus during the casting, resulting in unprotected areas on one, or more,
sides of the ingot.
This problem was clearly experienced from work within a RFCS project
regarding ingot casting, IPTINGOT (RFSR-CT-2011-00006) where pre-formed,
mould powder boards were tested. From the plant trials it was shown that mould
powder boards can have another positive effect when it comes to minimise the
addition of mould powder.
In order to avoid “open eyes” with the increased risk of alumina creation, it is
common practice to over-compensate and add more mould powder than necessary
to ensure that the centre of the mould is covered throughout the whole teeming
operation, especially for BEU ingots. This has a negative impact on the hot-top
function because of the endothermic nature of the melting of mould powder (see
Fig. 7.18). In some cases this problem is solved by manually adding small mould
powder bags during the casting to cover the “open eye” but in many cases this is
impossible because there is no manual access to the mould.
In these cases a pre-formed mould powder board could be a solution, since these
can be hung over the middle of the mould before casting. An example of such a test,
from the IPTINGOT project, is shown in Fig. 7.28.
If the board has a chemical composition with a high ability to absorb
non-metallic inclusions, and has low sulphur content, this would be a way of
achieving cleaner steel.
Another way to add mould powder is by continuous additions from a bin
through a steel pipe where the addition rate can be regulated by a ball valve. This
7.5 Application Techniques for Mould Powders 265

Fig. 7.28 Photographs showing. a An “open eye” in a 6 ton ingot. b Closing the “open eye” with
a mould powder board

technique is suitable for granulated powders but loose powders cannot be applied in
this way. Using this technique it is possible to avoid the splashes which normally
emerge in the beginning of the casting, when the mould powder bags open; this
results in a better surface in the bottom part of the ingot. The overcompensation of
mould powder to avoid “open eye” can also be reduced by the continuous feeding
in the centre of the ingot. An example of such equipment is shown in Fig. 7.29.

Fig. 7.29 Photograph of the equipment for continuous feeding of granulated mould powder
266 7 Fluxes for Ingot Casting

7.6 Use of Mould Powders to Minimise Defects

and Process Problems

There are few publications providing practical advice on how to avoid defects in
ingot casting and how mould powders can influence defect-formation. The cata-
logue of ingot defects, published by Stahlinstitut VDEh [10], provide some infor-
mation. However, the best advice, so far, is that reported by McCauley and Paul
[33] who describe the different defects that can emerge during ingot casting, and
how to avoid them, these include:
– Laps and ripple marks.
– Entrapped oxides.
– Slag patches.
– Porosity.
– Cracks.
– Bottom-end defects.
The reader is recommended to study their paper [33] but a short summary is
given in this chapter.

7.6.1 Laps and Ripple Marks

Laps form in two ways; through oxidation of the meniscus and by rapid solidifi-
cation of the meniscus. If the meniscus is not protected from the atmosphere, a
coherent oxide layer can form on the curved surface of the liquid steel. The oxide
layer prevents the liquid steel from rising near the mould until a sufficient, ferro-
static head has been generated and the liquid metal can overflow the oxide layer.
Similarly, if the heat loss from the meniscus is too rapid, a solidified shell of metal
can form on the curved meniscus near the mould, having the same effect as an oxide
layer.
To produce a lap-and ripple-free surface, a combination of higher teeming
temperature and faster teeming with an insulating flux cover is desirable. However,
excessive superheats can cause refractory erosion problems and surface crackings
so the desired teeming rate will depend strongly on the steel grade being cast.

7.6.2 Entrapped Oxides

Entrapped oxide patches are caused by the coalescence of high-melting oxides, such
as Al2O3, CeO2, Cr2O3 and MnTiO3. These can be trapped in the advancing, solid-
ifying shell. The mould powder slag fluxes these high-melting oxides so they can
7.6 Use of Mould Powders to Minimise Defects and Process Problems 267

be retained in a liquid flux pool on the meniscus or be consumed by infiltration

between the mould and advancing steel shell.
Entrapped oxide-type defects can also be caused by the flux itself, if the melting
characteristics of the flux are not properly controlled. If the flux sinters at the steel
surface, a sintered mass of the flux can become trapped in the solidifying steel in the
same way as high melting oxides. The lower viscosity fluxes used for stainless and
high alloy steels are consumed rapidly; consequently, the flux layer must be
allowed to melt faster to maintain a liquid flux pool on the meniscus.

7.6.3 Slag Patches

Intermittent slag patches are associated with a high-fluidity flux, or a flux which
melts too fast, or unevenly. If the fused flux layer on the meniscus is too fluid, or the
thickness increases rapidly, the molten flux will infiltrate between the mould and
ingot shell. These patches are simply layers on the ingot surface and are not
entrapped in the surface. Slag patches usually fall off during handling of the ingot or
can be easily removed from the ingot leaving a surface which is smooth, shiny and
free from oxidation.

7.6.4 Porosity

Surface porosity in bottom poured ingots is sometimes a problem. However, the

variation in the nature and the inconsistent appearance of the porosity makes it
difficult to pinpoint the origin and to find a solution to the problem.
In many cases, the source of porosity is the ingot mould. Wet moulds, for
example, will increase the incidence of porosity. Also, dirty and cracked, or crazed,
moulds provide places where moisture and gases are trapped and then absorbed by
the steel. A routine of mould cleaning and repair will solve most porosity problems.
Arranging production so that moulds are reused hot, rather than allowing them to
cool to ambient temperature will also reduce porosity, caused by moisture on the
mould.
The flux cannot be eliminated as a source of porosity because of the presence of
CO and CO2 from the oxidation of carbon; it can also arise from the decomposition
of carbonates contained in the flux, and the small amount of surface moisture
present with all materials. Results from steel grades exhibiting severe surface
porosity problems have shown that a non-gas producing flux will significantly
reduce the incidence of porosity.
268 7 Fluxes for Ingot Casting

7.6.5 Cracks

Longitudinal surface cracks are a problem associated with teeming conditions and
steel grade. High superheats and teeming rates enhance the risk of cracking in
crack-sensitive grades, especially, those containing 0.10–0.16 wt-% C and
re-sulfurised steel grades. In the 1010-type and similar steels, the peritectic trans-
formation causes major volume changes in the rapidly-solidifying, ingot shell. If
solidification is too rapid, cracking can easily occur. Also, if the pouring rate is too
fast, the build-up of ferrostatic pressure can cause cracking of a weak, insufficiently
thick, ingot shell.
Mould flux has little direct influence on the cracking except to reduce the heat
losses from the meniscus.

7.6.6 Bottom-End Defects

Bottom-end defects are the result of the initial addition of flux powder to the mould
after teeming has begun. The rapid development of a large amount of flux and the
turbulence associated with the initial pouring stream causes some of the flux
powder to get trapped between the mould and ingot. Under these conditions there is
insufficient time for the flux to establish equilibrium of the fused flux and the
powder flux layers: the flux trapped between the mould and steel does not have time
to melt. Bottom-end defects, therefore, are characterised by entrapped, sintered flux
powder and roughness caused by any gas generated by the flux between the mould
and semi-solid steel.
Multiple ingates and flared ingate geometry will reduce the incoming velocity
and turbulence; this minimises the occurrence of bottom-end defects.
The logistics of a bottom-pour setup precludes direct operator involvement; hand
additions at the start of teeming are not an acceptable solution. Manual additions of
flux during the pour have caused the formation of severe ripple marks, suspending
the bags of the flux too close to the bottom plate (i.e. <15 cm) and near the mould
wall increase the probability of bottom-end defects. Lowering the flux melting point
has proven successful in some situations.

References

1. A. Okamoto, Sumitomo Search, 59, pp. 3–11, (1997)

2. H. Combeau et al., 1st International Conference on Ingot Casting, Rolling and Forging,
Aachen, 2012, (Steel Institute VDEh, 2012)
3. H. Tomono, PhD Thesis No. 330, Département des Materiaux Ecole polytechnique Federale
de Lausanne (1979)
References 269

4. J.J. Bikerman, Physical Surfaces (Academic Press, New York and London, 1970), pp. 10–15
5. T.G. Gammal, U. Schoneberg, Stahl u. Eisen, 112 (1), (1992)
6. H. Jacobi, Arch. Eisenhüttenwesen, 47, 441, (1976)
7. H. Gaye, L.D. Lucas, M. Olette, P.V. Riboud, Can. Metall. Quart., 23(2), 179, (1984)
8. M. Susa, K. Nagata, K.C. Mills, Ironmaking and Steelmaking, 20(5), 372, (1993)
9. A.F. Diener, A. Drastik, W. Haumann, Archiv. F. Eisenhüttenwesen,43, 1, (1972)
10. Blockfehlerkatalog, Stahlinstitut VDEh (Verlag Stahleisen GmbH, Düsseldorf, 2003)
11. L. Zhang, B.G. Thomas, Met.Mater. Trans. B, 37B, 733, (2006)
12. L. Hallgren, C-Å Däcker, L. Teng, Proc. 2nd Intl. Conf. Ingot Casting, Rolling and Forging,
Milan, 2014, (Associazione Italiana di Metallurgia, 2014)
13. H. Nakato, S. Omiya, Y. Habu, T. Emi, K. Hamagami, T. Koshikawa, J. of Metals 1984, 44,
(1984)
14. M. Valdez, G. S. Shannon, S. Sridhar, ISIJ Intl., 46 (3), (2006)
15. B. Sjöholm, report from Swerea MEFOS, MF79012 (1979)
16. M.Susa, K Nagata, K.C. Mills, Ironmaking and Steelmaking, 20 (5), 372, (1993)
17. H.A. Fine, T. Engh, J. F. Elliott, Met. Trans. B, 7B, 277, (1976)
18. V.A. Efimov, V.I. Legenchuk, G.V. Sivtsov, I.M. Konovalov, G.D. Bykov, A.G.
Tatyanshchikov, Stal, 12, 927 (1962)
19. D.R. Thornton, JISI July, 300, (1956)
20. J.R. Rippon, STEEL TIMES, 12, 369, (1965)
21. W.O. Alexander, STEEL TIMES, 25, (1966), p. 251
22. G.A. Sedach, N.P. Izotov, G.N. Kamyshev, Stal, 1967 (Feb),124 (1967)
23. G.A. Panev, A.M. Ofengenden, G.G. Zhitnik, M.L. Plepler, L.I. Krupman, Steel in the USSR,
1972, (Jan), 16, (1972)
24. R. Sato, Bull. Jap, Inst. Metals, 12 (6), 48, (1973)
25. R.H. Hammerton, A. Hettler, Proc. 62nd Steelmaking Conf., 1979 (ISS, Warrendale, PA,
1979), p. 68
26. H.J. Eitel, PhD Dissertation Fakultät für Bergbau, Hüttenwesen und Geowissenschaften,
RWTH, Aachen, (1990)
27. Jahrbuch Stahl, Band 1, (Verlag Stahleisen 1988), p. 114
28. S.P. Bakumenko, V.M. Shatov, E.V. Verkhovtsev, N.A. Ponomarev, B.B. Gulyaev, Stal, 12,
669, (1976)
29. R. Gray, H. Marston, Iron and Steel Society Reference Text 101 (1979), p. 93
30. R.L. Harvey, A.P. Banks, Foseco INC. Seminar, Cleveland, Ohio, 1983
31. M.W. Nichols, A.P. Lingras, D. Apelian, Proc. 2nd Intl. Symp. Metallurgical Slags and
Fluxes, Lake Tahoe, 1984 ed. DR Gaskell and HA Fine, (TMS-AIME, Warrendale, PA,
1984), p. 235
32. R.W. Paul, W.L. Mc Cauley, Proc. 2nd Intl. Symp. Metallurgical Slags and Fluxes,Lake
Tahoe, 1984 ed. DR Gaskell and HA Fine, (TMS-AIME, Warrendale, PA, 1984), p. 253
33. W.L. Mc Cauley, R.W Paul, Iron and Steel Engineer, 1985 (Sept.), 36, (1985)
34. K. Sumitomo, M. Hashio, T. Kishida, A. Kawami, Iron and Steel Engineer, 1985, (Mar), 54,
(1985)
35. J.G. Bartholomew, R.L. Harvey, D.J., Hurtuk, Proc. 69th Steelmaking Conf., 1986 (ISS,
Warrendale, PA,1986), p. 121
36. C.M. Loane, K.H. Hashimoto, Proc. 69th Steelmaking Conf., 1986 (ISS, Warrendale, PA,
1986), p. 129
37. H. Marston, Proc. 69th Steelmaking Conf., 1986, (ISS, Warrendale, PA,1986), p. 107
38. R. Carli, A. Del Moro, C. Righi, La Metallurgia Italiana, 2008, (May), 13, (2008)
39. B.V.R. Raja, Steel World, 2008 (Dec.), 34, (2008)
40. J. Lönnqvist, C.-Å. Däcker, P. Andersson, 1st Intl Conf. Ingot Casting, Rolling and Forging,
Aachen, 2012, (Steel Institute VDEh, 2012)
41. M. Alloni, R. Carli, A. Del Moro, 1st Intl Conf. Ingot Casting, Rolling and Forging, Aachen,
2012, (Steel Institute VDEh 2012)
270 7 Fluxes for Ingot Casting

42. C.-Å. Däcker, 1st Intl Conf. Ingot Casting, Rolling and Forging, Aachen, 2012, (Steel
Institute VDEh 2012)
43. M. Svensson, E. Sjöqvist-Persson, L. Hallberg, C-Å Däcker, 2nd Intl. Conf. Ingot Casting,
Rolling and Forging, Milan,2014, (Associazione Italiana di Metallurgia, 2014)
44. J.E. Shelby, Introduction to Glass Science and Technology, (Royal Society of Chemistry,
New York 1997)
45. Slag Atlas, 2nd edition, (Verlag Stahleisen GmbH, Dusseldorf, 1995)
Chapter 8
Manufacture of Mould Fluxes

Abstract It is difficult to get a complete knowledge of the composition of mould

powders because the source and mixture of the raw materials is considered by the
manufacturers to be confidential. Most raw materials are open-pit-mined minerals
and can vary considerably in composition from batch to batch. For that reason, the
mould powder is composed of a number of different raw materials where price
optimisation is also an important factor. This chapter provides information on how
mould powders are produced, its constituents, and how the quality and properties
are controlled by the steel plants in order to get a better understanding of the mould
powder and how it affects the casting process. The data sheet supplied by the
manufacturer gives little information of value when it comes to the performance of
the powder in the mould and when the given chemical composition of the mould
powder can be achieved in an endless number of combinations of raw materials.

8.1 Introduction

Mould powder in a liquid state (mould powder slag) consists of a mixture of oxides
(mainly of silicon, alkali, alkali-earths) and fluorides. The liquid slags are ionic in
nature. The building block of most mould powders is the SiO44−-tetrahedron which
has the ability to create a three-dimensional network, providing the SiO2 content is
high enough. The network is broken down with the addition of cations, such as
Ca2+, Mg2+, Na+, K+ and Li+. Alumina forms anions and increases the polymeri-
sation of the silicate network, resulting in increased viscosity. Fluorine decreases
the binding force and consequently reduces the viscosity of the slag system.

© Springer International Publishing AG 2017 271

K.C. Mills and C.-Å. Däcker, The Casting Powders Book,
DOI 10.1007/978-3-319-53616-3_8
272 8 Manufacture of Mould Fluxes

8.2 Raw Materials

It is apparent that the desired properties of a mould powder can be achieved with
several different combinations of chemical composition. Table 8.1 shows the ranges
of different combinations that are commonly used in powders for continuous
casting and in synthetic, ingot-casting powders [1].
Older generations of mould fluxes contained Fe2O3 or FeO as fluidizers and as
wetting agents. However, these oxides are easily reduced by Al-killed steels and
change the properties of the mould powder slag during casting and create alumina
inclusions. For this reason, mould powders used in casting of these steels tend to
have a low level of iron oxides.
However, FeO has the ability to absorb radiative heat and this property is used in
high-viscosity, mould slags for billet casting when “soft cooling” is desired.
It is preferable to restrict the fluorine content. High amount of fluorine causes
excessive erosion of the SEN and fluoride emissions are detrimental to the envi-
ronment and pose a health hazard.
The raw material selection, for the most part, is driven by costs and environ-
mental concerns. With the exception of a few synthetic products, such as cement
and prefused materials, most raw materials are open-pit mined minerals and thus are
subject to considerable variations in composition from batch to batch. For that
reason, it is preferable that mould powders are composed of a number of different
raw materials. In Table 8.2 some common raw material sources for mould powder
production are listed.
Fly ash was the main component of mould powders in the past. However, the
wide variations in composition and the high amount of iron oxide in fly ash have
resulted in it being mostly replaced by other raw materials today; the principal
exception is in ingot casting where it is still the dominant raw material for most
mould powders [2] (see Chap. 7).

Table 8.1 Typical Network (glass) formers SiO2 17–56

composition ranges for
Al2O3 0–13
mould powders (wt%) [1]
TiO2 0–5
B2O3 0–19
Network breakers, basic oxides CaO 22–45
MgO 0–10
BaO 0–10
Network breakers, alkali Na2O 0–20
K2O 0–5
Li2O 0–5
Network breakers, fluidizers F 2–15
MnO 0–5
Fe2O3 0–10
Melting speed control C 2–25
8.2 Raw Materials 273

Table 8.2 Raw materials for mould powder production

SiO2 Feldspar, wollastonite, perlite, spodumene, cullet (crushed glass), fly ash, quartz
dust
CaO Lime, limestone, calcite, wollastonite, cement, slags, gypsum, calcium silicate.
Al2O3 Mullite, corundum
Na2O/K2O Natrite, potash, feldspar, sodium silicate, cullet (crushed glass)
F Fluorspar, sodium fluoride, cryolite, sodium hexafluorosilicate
Li2O Spodumene, lithium carbonate
C Carbon black, coke dust, fine graphite, fly ash

Fig. 8.1 Photograph of fly

ash, spherical, smooth
particles which mainly consist
of a SiO2–Al2O3 glass

Fly ash is a mineral product which originates from energy production of elec-
tricity and heat in carbon-fuelled power and heating plants. It consists of the
non-combustible part of carbon particles which are burned in the combustion
chamber of the energy plant. Each carbon particle becomes a fly ash particle
(Fig. 8.1) which is separated in an electrostatic filter before the flue gas is emitted
from the plant’s chimney. Fly ash contains some carbon, the amount of which
depends upon the efficiency of the combustion in the power plant.
It must be pointed out that most mould powder producers consider the choice of
raw materials as confidential and information of how this is made is not normally
provided. However, the chemical composition and the results of DSC–TG analysis
provide a reasonable idea of how the mould powder has been composed. It is
possible to get further understanding of the raw materials used in the formulation of
the mould powder from X-ray analysis and some geological expertise.
It is considered good practice to use
– A combination of minerals with similar melting points to promote homogeneous
melting of the mould powder and
– A minimum number of different mineral sources (i.e. to keep the recipe simple)
which helps to provide good quality control.
274 8 Manufacture of Mould Fluxes

8.2.1 Selection of Carbon Additions to Mould Powders

Addition of free carbon to the mould powder is the most efficient way to control the
melting speed. Whilst mould powders are composed of a mixture of mineral
components (with high melting points) and fluxing agents (for example CaF2 and
Na2CO3), the carbon particles delay the contact and sintering of these components.
The carbon also delays the agglomeration of the slag droplets. The effect of carbon
on melting speed is determined by both the amount and the size of the carbon
particles, where finer carbon particles cause a greater reduction of the melting rate
per unit mass of carbon. Thus melting speed is controlled by both the amount and
particle size of the carbon. The particle size is dependent on carbon source which is
illustrated in Table 8.3. As shown in Fig. 8.2 it can be seen that carbon black,
because of its high surface area and high reactivity, has the lowest ignition tem-
perature and graphite the highest.
Carbon black has the finest particle size and is believed to coat each particle of
the mould powder. Sakuraya et al. [3] have shown that a minimum of 0.4 wt%
carbon black is needed to coat a typical mould powder with a monolayer of carbon
black.

Table 8.3 Typical particle Type of carbon Particle size (lm) Burn temperature
size and burn temperatures for (°C)
different carbon types [1]
Carbon black 0.028 386–522
Metallurgic coke 20 511–718
Graphite 74 613–897
Fine graphite 1.36 613–897

Fig. 8.2 Diagram showing

thermal analysis curves for
carbon black, metallurgical
coke dust and graphite
(heating rate 1 °C/min) [1]
8.2 Raw Materials 275

Carbon particles are non-wetting to molten slag and thus separate the molten
globules and prevent them from agglomerating. For a given mass of carbon, the
finer particle sizes will provide more separation than coarser particles. However, the
carbon particles react with oxygen in the mould and gradually disappear; the
oxygen sources are gaseous diffusion through the bed, CO2 emitted during
decomposition of carbonates and oxides like FeO in the powder. Finer particles
have a greater (surface area/volume) and so will tend to combust more readily. The
fact that the melting rate decreases with decreasing particle size indicates that the
separation of globules determines the melting rate rather than the carbon com-
bustion rate.

8.2.2 Reactions During Melting and Cooling

of Mould Powders

A large amount of information regarding the mould powder can be gained from a
combined DSC (Differential Scanning Calorimetry) and TG (Thermal Gravity)
measurement. An example from a measurement from a commercial mould powder
for continuous casting of peritectic slabs is shown in Fig. 8.3 [4].

Fig. 8.3 Diagram showing DTA/TG examination of a mould powder used for peritectic steels [4].
The temperature at any specific time is given from the curve shown in bold; red weight loss in %;
blue DSC (mW/mg)
276 8 Manufacture of Mould Fluxes

During the heating and cooling process, possibly, the following endothermic/
exothermic reactions occur [4]:
I. Gas desorption (possibly) at around 95 °C. This is an endothermic reaction
accompanied by a small weight loss.
II. Water evaporation at around 130 °C. This is an endothermic reaction
accompanied by weight loss.
III. Ignition of organic substance (binding agents for granulation of powders) at
around 225 °C. This is an exothermic reaction accompanied by weight loss.
IV. Ignition of sulphur at around 360 °C. This is an exothermic reaction
accompanied by weight loss. The sulphur can come from the coke in the
powders.
V. Combustion of carbon black at around 472 °C. This is an exothermic
reaction accompanied by weight loss.
VI. Combustion of coke at around 554 °C. This is an exothermic reaction
accompanied by weight loss.
VII. Combustion of graphite at around 722 °C. This is an exothermic reaction
accompanied by weight loss. There is a small endothermic event before this
peak at around 700 °C, which may be caused by the decomposition of
sodium carbonates. These two phenomena are so close that the endothermic
reaction is almost overlapped by the big exothermic peak.
VIII. Around 950 °C also a small endothermic peak is shown which reflects the
decomposition of calcium carbonates (Na2CO3, K2CO3). The following
peak at 1143 °C relates to the melting of the powder. This is an endothermic
reaction; there was no significant weight change.
IX. Evaporation of Na2O at around 1216 °C: this is an exothermic reaction
accompanied by weight loss of ca. 1.5%.
X. Crystallisation of the slags: this is an exothermic reaction without any
significant weight change. Normally, crystallisation occurs in two steps: the
first stage at high temperature and second stage at a lower temperature,
during which some latent heat due to crystallisation is released. The two
stages correspond to the two peaks appearing in the DSC signal.

8.3 Manufacturing

The manufacturing process is shown schematically in Fig. 8.4.

The production steps are as follows:
• A large amount of different raw materials are used for the production of different
mould powders; these arrive at the plant in a fine-grained condition (max.
70 mesh) either as bulk in a silo transport or in big bags.
• A rigid control of the incoming raw materials is exerted using chemical analysis.
8.3 Manufacturing 277

raw material silos air mixer

raw material

S S S
moving scale

mech. mixer

storage bin
raw materials
S pneumatic
conveying
vessel
S
bags silo truck finished product silo
granulating plant

S = sample taking
bags bulk bag

Fig. 8.4 Diagram showing material flow for production of loose mould powder (courtesy of
IMERYS)

• The material is blended according to the desired formula. This is done either in a
mechanical mixer or in an air mixer. The carbon materials are added at this
stage.
• After mixing, the non-granulated (loose) powders are ready for delivery either in
big bags or in small plastic bags for direct use.
For granulated powders, the mixed powder undergoes a second production
process shown schematically in Fig. 8.5.
The production steps are as follows:
• Batches of ca. 10 tonnes powder are mixed with approx. 4 tonnes of water in a
turbo mixer.
• The slurry is transported to a second tank where the binding agent is added.
Previously, water glass was frequently used as binder but this has gradually
changed to an organic material (for example, maize starch) which gives better
melting properties for the powder.
• The slurry is transported through a tube filter to a spray dryer.
• In the spray dryer, hot air, heated with gas or oil to a temperature of around
600 °C, circulates through the dryer. The slurry is injected into the hot air
through a nozzle consisting of a large number of small holes.
• When the slurry is pressed out of the nozzle, a mist of droplets is formed. The
surfaces of the droplets dry quickly (due to the heat and the large amount of
278 8 Manufacture of Mould Fluxes

from powder section o

600 C

finished product silo

storage bin swinging
screen
burner
heated tank ( gas / oil )
dissolving
with agitator
vessel

slurry piston
S pumps
spray dryer

tank with
agitator S
swinging
tube filter screen

bags bulk bag

cooling section
S = sample taking

Fig. 8.5 Schematic diagram showing material flow for production of granulated mould powder
(courtesy of IMERYS)

Fig. 8.6 Schematic diagrams

showing the hollow
appearance of granulated
mould powder

(a) A solid shell has (b) A cavity has formed

appeared on the granule in the granule by the steam
explosion

convection coming from the hot air) forming a shell of solid, dried material
(Fig. 8.6a). With continued drying an overpressure is formed within the droplets
because of its large water content. Finally, the granule explodes leaving a
hollow granule as shown in Fig. 8.6b.
• The granule lands on the bottom of the dryer from where they are collected,
transported, screened and packed into small bags or big bags.
Due to the cavity and a narrow particle size distribution, granulated mould
powders possess a low density. A photograph of a granulated mould powder at high
magnification is shown in Fig. 8.7.
8.3 Manufacturing 279

Hole in cavity

Fig. 8.7 Photograph of mould powder granules, magnification 40X

100 70

Cumulative distribution
90 60

deensity distribution q3/g(X) / %

cumulative distribution Q3(X) / %

80
50
70
40
60
30
50
20
40

30 10
Density distribution

20 0
0.5 1 5 10 50 100 500 1000
particle size / µm

Fig. 8.8 Diagram showing particle size and cumulative distribution of standard, non-granulated,
mould powder

The particle size distribution varies significantly between non-granulated and

granulated powders as can be seen in Figs. 8.8 and 8.9, respectively.
Granulation of mould powders increases the price of the mould powders but has
a number of advantages [5], such as
• The composition of the mould powder is uniform throughout the mould powder
bed which makes its process behaviour both stable and reliable.
• The narrow particle distribution, with a low amount of fine particles, results in a
low amount of dust during handling and casting; this makes them much more
environmentally friendly to use.
280 8 Manufacture of Mould Fluxes

100 300
Cumulative distribution
90

density distribution q3/g(X) / %

cumulative distribution Q3(X) / %

250
80
70
200
60
50 150
40
100
30
20
50
10 Density distribution
0 0
1 5 10 50 100 500 1000
particle size / µm

Fig. 8.9 Diagram showing particle size and cumulative distribution of standard, granulated,
mould powder

• The granules will have a good flowability that makes it possible to use an
automatic feeding system. This, in turn, leads to a secure feeding and control of
the height of mould powder in the mould. The mould pool depth will be more
even due to more uniform melting properties; this, in turn, ensures more
homogeneous feeding of molten slag into the gap between steel shell and mould.

8.4 Quality Control at the Manufacturer

The raw materials for mould powders can vary a great deal in composition.
Consequently, it is important that the manufacturer carries out regular control
checks on both of the incoming raw material as well as the batches of mould
powders before dispatch to the steel plants.

8.5 Information Provided by the Manufacturer

An example of a typical data sheet for mould powders from a manufacturer is

shown in Table 8.4.
The following information is given:
– Chemical composition of the mould powder;
– Melting range as determined by DIN-standard 51730;
– Calculated viscosity using Riboud model [6] which is based on multiple sta-
tistical analyses from a large number of synthetic and commercial mould
powder slags;
– Sieve analysis and
– Bulk density.
8.5 Information Provided by the Manufacturer 281

Table 8.4 Typical data sheet Batch number: 11/100 Packing date: 30-May-11
for a granulated mould
powder Chemical analysis (%) Melting Range (°C)
LOI 10.62 Softening point 1155
C tot. 5.81 Melting point 1220
C free 4.05 Flowing point 1250
SiO2 34.82
Fe2O3 0.85
Al2O3 3.85
CaO 39.10 Viscosity (dPa s)
MgO 1.90 (IRSID Math.
Method)
Na2O + K2O 4.82
F- 7.22 1300 °C 1.7
MnO 0.00
S 0.08
H2O 0.35 Sieve analysis
(120 °C)
Mesh size (mm) Residue
(%wt)
0.400 24.0
CaO/SiO2 1.12 0.125 91.0
0.063 98.5
Bulk Density 0.72 (kg/dm3)

Compared to most other products, the information is very sparse and does not
give much information on how the mould powder will function in the steel plant.
The missing information includes the following:
– Some information regarding the raw materials used to compose the mould
powder;
– Measured viscosity with the rotary spindle method. Actually there is an ASTM
standard for mould powders [7];
– Melting rate;
– Break or solidification temperature (determined with a viscometer with a small
distance between crucible and spindle);
– Flowability, which is an important property especially for ingot casting of big
end up, moulds (BEU). There is also an ASTM standard available [8] for this
property;
– Interfacial tension between mould slag and steel;
– Mould slag irradiative absorptivity and
– Crystal phase (fcrys) in the slag film.
282 8 Manufacture of Mould Fluxes

8.6 Delivery Control by the Steel Makers

Large steelmakers have their own production of mould powders and thereby exert
full control of the product, including all necessary information on its raw materials.
Another way of maintaining full control of the raw materials is to use 100%
prefused mould powder but this is very expensive and non-realistic but does reduce
the fluorine emissions on the caster.
Most steel plants have to live with the sparse information provided by the
manufacturer. In order to get more necessary information, ambitious steelmakers
exert their own delivery control of the mould powder batches received from the
manufacturer.
One example of such a delivery test is shown in Table 8.5. The following tests
are performed by a Swedish steelmaker:
– Chemical composition;
– DTA to get a “fingerprint” of the mould powder behaviour and an idea of its
components, and important information on the crystallisation temperatures;

Table 8.5 Information from Mould powder A B

a delivery control of mould
powders A and B SiO2 (wt%) 32.9 32.7
CaO (wt%) 39.1 28.8
MgO (wt%) 0.37 1.77
Al2O3 (wt%) 6.5 4.7
TiO2 (wt%) 0.03 0.11
Fe2O3 (wt%) 0.47 1.24
MnO (wt%) <0.10 <0.10
Na2O (wt%) 7.1 11.3
K2O (wt%) 0.17 0.31
F (wt%) 8.3 9.4
Ctotal (wt%) 5.5 6.6
Loss on ignition (wt%) 10.1 14.1
H2O (wt%) 0.3 0.3
Basicity (CaO/SiO2) 1.19 0.88
Basicity (CaO + MgO/SiO2 + Al2O3) 1.0 0.82
Melting time (s) 169 115
Viscosity 1300 °C (Poise) 2.72 2.51
Break temp. (°C) 1267 1165
DTA melting temp. 1 °C 1079 1223
DTA melting temp. 2 °C 1231 1146
DTA solidification temp. 1 °C 1250 1247
DTA solidification temp. 2 °C – –
Density (g/cm3) 0.74 0.54
8.6 Delivery Control by the Steel Makers 283

– Viscosity, to obtain both an experimental value of the viscosity and the break
temperature (or solidification temperature). Observe that the determination of
the break temperature is made possible by the use of a rotary viscometer with a
narrow gap between rotor and crucible;
– Melting time which is determined from the time a certain amount of mould
powder is melted in a tube-type furnace at 1500 °C and
– Bulk density, which is determined by weighing a certain volume of mould
powder.

References

1. R. Bommaraju, Proc. 72nd Steelmaking Conf., 1991, (ISS, Warrendale, PA, 1991) p. 131.
2. M. Alloni, R. Carli, A. del Moro, 1st Intl. Conf. Ingot Casting, Rolling and Forging, Aachen,
2012 (Steel Institute VDEh 2012).
3. T. Sakuraya, T. Emi, K. Emoto, T. Koshikawa, Proc. 2nd Process Technology Conf., Chicago,
1981, (ISS, Warrendale, PA, 1981) p. 141.
4. T. Sun, X. Hu, L. Hallgren, L. Teng, Jernkontoret TO24-192 (2012).
5. H.J. Eitel, PhD Dissertation, Fakultät für Bergbau, Hüttenwesen und Geowissenschaften,
RWTH, Aachen, 1990.
6. P.V. Riboud, Y. Roux, L-D. Lucas, H. Gaye, Fachberichte Hüttenpraxis
Metallweiterverarbeitung, 19, 859, (1981).
7. ASTM Designation: C 1276 – 94, Standard Test Method for Measuring the Viscosity of Mold
Powders Above Their Melting Point Using a Rotational Viscometer.
8. ASTM Designation: C 1444 – 99, Standard Test Method for Measuring the Angle of Repose of
Free-Flowing Mold Powders.
Chapter 9
Properties of Mould Fluxes and Slag Films

Abstract In this chapter, the properties and the structure of the slags used in
continuous casting are collated, analysed and discussed. The required property
values of mould slags for successful casting are determined by the mould dimen-
sions, the casting conditions and the steel grade being cast; these can be calculated
using empirical rules. Required values of the powder consumption (Qreq s ) viscosity
br ) and the fraction crystal in the slag film (fcrys) are the
(ηreq), break temperature (Treq
key properties. However, other properties such as the interfacial tension are
important when dealing with slag entrapment, thus a large amount of property data
for mould slags are needed. Mould slags have been classified here into two groups
(i) Conventional slags used in continuous casting and (ii) Specialised slags, which
includes F-free slags, non-Newtonian slags, calcium aluminate (CA) type slags, etc.
Some properties are very dependent on the structure of the slags. Thus, published
data on slag structures for all types of slag have been collated and analysed.
Routines, developed to estimate the various properties, are outlined and reviewed.
The data for the following properties of the various mould slag families have been
reviewed (i) viscosity (ii) Liquidus (Tliq), Break (Tbr) and Glass transition (Tg)
temperatures (iii) Thermal conductivities (iv) Surface and Interfacial tensions
(v) Densities and thermal expansion coefficients (vi) Heat capacity and enthalpy
and (vii) Optical properties. In addition, the crystallisation process in mould slags
and the ability of the liquid slag to dissolve inclusions are reviewed and discussed.
The requirements and the properties of slags used in continuous casting and ingot
casting are compared and discussed.

Symbols, Abbreviations and Units

A Surface area (m2)
d Thickness of slag (m)
Cp Heat capacity (JK−1 kg−1)

The following abbreviations to chemical formulae have been used; A = Al2O3; B = BaO; C = CaO;
F = FeO; Fl = CaF2; L = Li2O; M = MgO; Mn = MnO; N = Na2O; S = SiO2; T = TiO2; e.g.
40C + 40S + 20A = 40% CaO + 40%SiO2 + 20% Al2O3 or C3S2Fl = 3CaO
2SiO2
CaF2.

© Springer International Publishing AG 2017 285

K.C. Mills and C.-Å. Däcker, The Casting Powders Book,
DOI 10.1007/978-3-319-53616-3_9
286 9 Properties of Mould Fluxes and Slag Films

D Diameter (m)
f Oscillation frequency (Hz, cycle min−1)
f* Fraction of flux-forming slag
k Thermal conductivity (Wm−1K−1)
Q Powder consumption (various)
q Heat flux (Js−1K−1 = WK−1)
R Thermal resistance (m2KW−1)
R* (Surface area/volume) of mould
s Stroke (m)
T Temperature (oC or K)
Tbr Break temperature (oC or K)
t Time or thickness of mould (m)
tn Negative strip time (s)
tp Positive strip time (s)
Vc Casting speed (m min−1 or m s−1)
Vm Velocity of mould (m s−1)
w Width of mould (m)
a Thermal expansion coefficient (oC, K)−1
a* Absorption coefficient
η Viscosity (d Pas)
q Density (kg m−3)
CA Calcium aluminate slags
CC Continuous casting
CP Carbon potential
(C/S) Basicity
DSC Differential scanning calorimetry
DTA Differential thermal analysis
EMBr Electromagnetic braking
EMS Electromagnetic stirring
FF Fluoride-free slags
FP Ferrite Potential
HC High-carbon steels
HiAl High-Al steels (TRIP)
IC Ingot casting
IR Infrared
LC Low-carbon steels
MC Medium-carbon steels
NN Non-Newtonian slags
OM Oscillation mark
SEN Submerged entry nozzle
9 Properties of Mould Fluxes and Slag Films 287

Subscripts, Superscripts
crys Crystalline phase
gl Glass phase
l Liquid
m Metal
s Solid
sl Slag

9.1 Introduction

The structure and the thermophysical properties of continuous-casting slags are dis-
cussed in this chapter. It will be seen that the values of the properties of conventional slags
for successful continuous casting (such as viscosity and Tbr) are determined by the mould
dimensions, casting conditions and the steel grade being cast. These required property
values can be calculated by empirical rules. However, in recent years, special mould slags
have been developed to cast specific steel grades or for certain casting conditions. These
include: (i) fluoride-free slags (ii) carbon-free powders (iii) non-Newtonian slags and
slags for casting (iv) high-Al steels (v) stainless steels (vi) round billets and
(vii) high-speed thin slabs. Although, the required properties for these types of slags tend
to conform with the values derived from empirical rules they do have to deal with other
problems which must be met. Consequently, powders and slags are classified below as
either “Conventional” or “for Specialist casting” (see Chap. 6). The structure and
property data for all types of mould slag are collated and discussed below.

9.2 Structure of Slags

The performance of slags in the continuous-casting process depends largely on their

properties (e.g. viscosity) and the properties are, in turn, very dependent upon the
structure of the mould slag [1–3].

9.2.1 Effect of Individual Slag Components on Structure

Casting powders are essentially calcium silicate slags with added fluxes. The
properties of slags are principally determined by the structure of the slag. It should
be noted that slags are ionic in nature and that reactions, such as that shown in
Eq. 9.1, are ionic (i.e. involve the transfer of electrons); thus they are best described
in the form of Eq. 9.2 where the underline indicates an element in the metal phase:
288 9 Properties of Mould Fluxes and Slag Films

4Alsteel þ 3 SiO2slag ¼ 3Sisteel þ 2Al2 O3slag : ð9:1Þ

4Al þ 3 Si4 þ ¼ 3Si þ 4Al3 þ ð9:2Þ

Pure SiO2 has a three-dimensional structure based on covalent Si–O bonds. Each
Si4+ ion is surrounded by four O− ions in the form of a tetrahedron (Fig. 9.1a) and
each O2− connects two Si4+ tetrahedra. For pure SiO2, this results in a
three-dimensional polymerized network (Fig. 9.1b) [1, 2]. Thus, in mould slags,
SiO2 is considered to be a network former.
When cations, such as Na+ or Ca2+, are added (in the form of Na2O or CaO) to SiO2,
they break some of the Si4+–O− bonds and replace them with an ionic Na+–O− bond
(for Ca2+ the cation is bonded to two O− bonds). Further addition of Na+ ions results in
progressively- more de-polymerisation of the melt; the cations (Na+, Ca2+) are
referred to as “network breakers”. Thus silicate slags contain both covalent and ionic
bonds (Fig. 9.2) and the thermophysical properties are very dependent upon the level
of polymerisation in the slag.
The different types of bonds are frequently classified in terms of the O bonds
formed, namely, (i) Bridging O’s (e.g. Si–O) denoted both as BO and O° (ii) Non-
bridging O’s (denoted NBO or O−) and (iii) Free O’s (i.e. bonded to cations but not
to Si) which are denoted as FO or O2−.
When Al2O3 is introduced into the silicate network the Al3+ ions can be
accommodated into the silicate structure (i.e. Al3+ adopts 4–coordination like Si4+).
However, a Na+ ion must be sited near the Al3+ to provide electrical charge balance
[i.e. forms (NaAl4+)] [1–3] (Fig. 9.3). If M2+ (e.g. Ca2+) ions do the charge bal-
ancing, the M2+ ions must be placed between two Al3+ ions. It is considered that a
Na+ ion acting on charge-balancing duty cannot act as a network breaker. Thus,
Al2O3 additions act principally as network formers but when large amounts of
Al2O3 are added to the slag (e.g. during the casting of high-Al, Trip steels) the Al3+
ions can also act as network breakers (i.e. exhibit six––(sometimes denoted VI)––
fold coordination and for this reason, Al2O3 is often referred to as “amphoteric”.
For most mould slags:

Fig. 9.1 Schematic diagrams showing a Si4+–O− bonds arranged in form of tetrahedron and b the
creation of a network through the connection of each O2− to two Si4+ ions (permission granted,
VDEh, Stahleisen, [3])
9.2 Structure of Slags 289

Fig. 9.2 Schematic drawings of silicate chain with Bridging O (O°) shown in dark blue,
non-bridging O (O−) in pink and free O2− (shown as inset with free Oxygens)

Al

+
Na

Fig. 9.3 Schematic drawing showing Al3+ incorporated into silicate (Si4+) chain which requires
cation (shown here as Na+) to maintain charge balance

(i) Al2O3 contents are low (<7%) and thus Al3+ ions tend to be in tetrahedral (4 or
IV) coordination.
(ii) Charge-balancing duties tend to be carried out by K+ or Na+ ions since they
have lower field strength cf. Ca2+ ions. The effect of replacing Ca2+ by K+ as
charge-balancing cations and its effect on viscosity have been studied [4, 5].
Large amounts of Al2O3 are produced when casting high-Al (TRIP) steels. For
low (C/S) slags used to cast these steels, increasing the (XA/XS) ratio increased the
viscosity (presumably due to more cations being needed for charge balancing
leaving fewer for network breaking) [6]. Calcium aluminate (CA) slags are also
used to cast these steels. There is a eutectic in the CaO–Al2O3 system close to the
C12A7 composition (corresponding to 48% CaO) to which fluxes are added to
290 9 Properties of Mould Fluxes and Slag Films

reduce the melting temperature (Tliq). Commercial CA slags usually contain (5–
10%) SiO2. The effects of (C/S) and (C/A) ratios on the viscosity indicated that
(i) the viscosity was more sensitive to (C/A) than to (C/S) (ii) SiO4 units tended to
be placed at the ends of the AlO4 chains and (iii) there was a tendency to segregate
into QnSi + QnAl units [7]. The viscosity was found to decrease when K+ replaced
Ca2+ in cation–NBO bonds [5].
It has been reported that Al–O bonds exhibit four-, five- and sixfold coordination
(i.e. form AlO4, AlO5 and AlO6, respectively) [8, 9]. The AlO4 units are the
dominant species for XCaO > 0.5, i.e. when there are sufficient Ca2+ ions to charge
balance the AlO45− ions; any excess Ca2+ cations are then free to depolymerise the
AlO4 network and reduce the viscosity. Thus viscosity decreases with increasing
CaO (or CaO/Al2O3 ratio).
In the casting of high-Al steels, the Al2O3 concentration increases steadily as the
cast proceeds and there may come a point where there are insufficient cations to
charge balance the AlO4 anions; under these circumstances some of the Al ions
must take on the charge-balancing duties. In these conditions, additions of Al2O3 to
the slag result in decreases in the number of AlO4 units and to gradual increases in
both AlO5 and AlO6 species [7–9].
Mould powders contain various fluxes, the charge-balancing duties will be done
by cations in the hierarchy of decreasing cation field strength (i.e.
K+ > Na+ > Li+ > Ca2+ > Mg2+). Recently, it has been shown that that the addition
of K2O to a CaO–Al2O3–SiO2 slag resulted in K+ ions taking over the
charge-balancing duties performed by Ca2+ ions with the released Ca2+ ions
gradually forming more NBOs [5]. This was found to be accompanied by a slight
increase in viscosity followed by a decrease in viscosity when (XK/XA) > 0.9 [5].
This behaviour was attributed to the competition between the average bond strength
of the NBOs (i.e. O–Ca vs. O–K) and the number of NBOs [5].
Conventional mould powders contain calcium fluoride (CaF2) since it fluxes the
slag and aids the formation of cuspidine (3CaO
2 SiO2
CaF2 or C3S2Fl) crystals in
the slag film.
When F− ions are added to a slag, few Si–F bonds are created and although some
Al–F bonds may be formed, most of the F− ions bond with Ca2+ and Mg2+ cations
[10]. Recent work [11–14] has indicated that when Fluorine is added to a silicate
melt, the F− ions bond almost exclusively with Ca2+ ions to form CaF2 which then
acts principally as a diluent to the silicate structure. It has been proposed that Ca2+
and F− ions take up preferred positions; this allows them to form CaF2 clusters [15].
The CaF2 formed does not interfere with the concentrations of the various silicate
ions formed (i.e. the various Qn units remain unaltered). Thus the CaF2 acts merely
as a diluent or “thinner”.
Fluorine additions can be made by either replacing CaO with CaF2 or by adding
CaF2 to the slag. For the case, where CaO is replaced by CaF2, it has been sug-
gested that the F− additions promote polymerisation of the melt [11–14, 16]. In
contrast, CaF2 additions serve only to dilute the polymerised slag.
The mechanism accounting for how CaF2 modifies the network has been studied
by a number of investigators [1, 2, 11, 12, 14, 16–18]. Basically, there models fall
9.2 Structure of Slags 291

Fig. 9.4 Proposed

mechanism for the addition of
CaF2 to silicate slag
(permission granted, ISIJ,
[16])

into two types (the Baak-Bills and Sasaki models [11, 19, 20]. The mechanism
shown in Fig. 9.4 is similar to the Sasaki mechanism [11].
TiO2 additions would be expected to behave like SiO2 and further polymerise
the melt. However, TiO2 has been shown to decrease the viscosities of slags [21–
24, 26]; this suggests that TiO2 may act as a weak network breaker. There are three
possible ways that TiO2 can modify the silicate structure, namely. (i) Ti4+ is
incorporated into the SiO4 network (ii) Ti acts as a network breaker (by adopting 5-
or 6- coordination) and (iii) by forming TiO2-like structures (clusters) [25]. There is
some dispute which is the dominant mechanism in the geological literature and
recent publications on slags are equally inconclusive. The results reported by Li
et al. [26] for CA-based slags indicate the Ti prefers to form TiO2-like clusters and
there is a small increase in polymerisation (the decrease in viscosity arises because
the Ti–O bond is weaker than the Si–O bond). Alternatively, it has been suggested
that that TiO2 acts as a “thinner” [27]. However, the spectroscopic results reported
by Zhang et al. [25] support the formation of TiO6 octahedra (i.e., Ti acts as a
network breaker).
There use of B2O3 in mould powders as a flux has been limited by the fear that it
could lead to B-pick-up by the steel. However, it has been used quite widely used as
a flux in slags usd to cast high-Al steels. B2O3 is considered to be a network former
in slags and is known to form complex borosilicate structures. However, B2O3
additions frequently result in a decrease in viscosity [28–31]. In pure borax (B2O3)
the structure is made up of 3-coordinated BO3 units arranged in the form of 2-dim.
rings. However, additions of CaO or Na2O result in the formation of structural units
with both 3- and 4-coordination (i.e. three-dimensional BO4 units) respectively [29]
as shown in Eqs. 9.3 and 9.4. It has been reported that the BO3 ring species are
dominant [29] in calcium aluminate mould slags. It has also been pointed out [30]
that the viscosity is determined by the strength of the bond of the flow unit; since
the B–O bond is weaker than the Si–O bond, B2O3 additions result in a decrease in
viscosity.

2BO3 þ CaO ¼ 2 BO

4 þ Ca
2þ
ð9:3Þ

2BO3 þ CaO ¼ 2 BO

3 þ Ca
2þ
ð9:4Þ

It might also be anticipated that Fe2O3 and Cr2O3 would act like Al2O3 in slags
and result in the accomodation of Fe3+ and Cr3+ into the silicate network. This is
292 9 Properties of Mould Fluxes and Slag Films

largely true in the case of Cr2O3 (except when the slag contains a significant Al2O3
content) but Fe2O3 appears to act as both network breaker and network former [32,
33]. However, the reducing conditions in the mould would appear to rule out the
formation of both Fe2O3 and Cr2O3 in the mould slag.
In summary, the various constituents of mould fluxes can be classified as
follows:
Network formers: SiO2, Al2O3
Network breakers: CaO, MgO, BaO; FeO, MnO, CrO, Na2O, Li2O and K2O
Fluxes: CaF2 and B2O3.

9.2.2 Parameters to Represent the Structure of Slags

It is important to have a parameter to represent the structure of the silicate slag.

Several parameters have been used and are outlined below (Sect. 9.2.2).

9.2.2.1 Basicity

Various basicity indices (e.g. B = %CaO/ %SiO2) have been used to represent slag
structure in the calculation of viscosities [34]. The major problem with this
approach lies in dealing with oxides which can act as both a networker former and
network breaker (e.g. Fe2O3, Al2O3).

9.2.2.2 NBO/T and Q

The ratio of (non-bridging O/Tetragonal O) ratio, which is denoted (NBO/T) is a

measure of the de-polymerisation of the melt [1, 2]. It is the ratio of the mole fraction of
available network-breaking oxides (where available indicates corrected for charge
balancing) divided by the mole fraction of the network-forming oxides. Thus
(NBO/T) ratio is a kind of basicity index which allows for corrections for cations
acting on charge-balancing duties [33]. There has been a long-standing problem on
how to deal with the CaF2 when calculating (NBO/T) for mould fluxes. It was sug-
gested [3] that CaF2 acts as (i) a network breaker in polymerised slags and (ii) a diluent
in basic slags. Recent structural studies indicate that when fluorine is added the F− ions
bond to Ca2+ ions and thereupon, remain apart from the slag [11–14, 16]. This would
suggest that mould fluxes can be regarded as (remaining slag + CaF2). Thus, the
CaF2 concentration has been ignored when calculating the NBO/T ratio for
the remaining slag (Eq. 9.5 where X is the mole fraction and where
XMO = XMgO + XCaO + XBaO + XFeO + XMnO +.. and XM2 O ¼ XLi2 O þ XNa2 O þ XK2 O ).
9.2 Structure of Slags 293

NBO=T ¼ 2ðRXMO þ RXM2 O  XAl2 O3 Þ=ðXSiO2 þ 2XAl2 O3 Þ ð9:5Þ

Since it is not known whether TiO2 should be regarded as a network breaker or

network former it is also suggested that TiO2 and ZrO2 concentrations should be
ignored when calculating the (NBO/T) for the slag component and to determine any
effects on property values as a function of XTiO2 or XZrO2 .
Many people find it easier to visualise polymerisation than de-polymerisation,
and thus, the parameter, Q, which is a measure of the polymerisation of the melt is
preferred here; it can be calculated from (NBO/T) using Eq. 9.6.

Q ¼ 4  ðNBO=T Þ ð9:6Þ

Typical, “remaining slag”, Q values range from Q = ca. 2 (for high-speed

casting with η1300 °C = 0.6 dPas) to Q = ca. 3 (for high-viscosity billet fluxes,
η1300 °C = >15 d Pas). (Note Q-values for the remaining slag are calculated by
ignoring XCaF2 , XTiO2 and XZrO2 ).

9.2.2.3 Optical Basicity (K)

The major problem when using (NBO/T) or Q, lies in the fact that it does not
differentiate between different cations e.g. Na+ or Ca2+. The optical basicity was
introduced to partially resolve this problem [35–37]. The optical basicity (K) is a
measure of the electron donor power of different ions relative to that of CaO [38,
39]. Measurements were derived from the shift in the frequency absorption band of
the UV region (associated with 6s ! 6p transition) which is related to the basicity
of the slag. It was subsequently shown that it could be calculated from Pauling
electron negativity values [39].
It was also used as a measure of the de-polymerisation of the melt [35–37] and
can be calculated by Eq. 9.7 where m is the number of O atoms, e.g. 1 for CaO and
2 for SiO2 and K1 is the optical basicity value for oxide 1.

K ¼ RðX1 m1 K1 þ X2 m2 K2 þ X3 m3 K3 þ . . .Þ=RðX1 m1 þ X2 m2 þ X3 m3 þ : :Þ ð9:7Þ

The composition can be corrected to allow for the cations on charge-balancing

duties [35] by subtracting 2XAl2 O3 from R (X1m1K1) + R (X2m2K2) where 1 and 2
refer to MO and M2O, respectively; this can be done by assuming that
charge-balancing duties are done (i) preferentially by the largest cations or (ii) ac-
cording to the concentrations (i.e. randomly). Optical basicities for the various
constituents of casting powders are given in Table 9.1.
The principal disadvantage to the use of optical basicity results from:
• Uncertainties in the K values for some components, e.g. transition metal oxides.
FeO, and, more importantly for mould fluxes, CaF2.
 294

Table 9.1 Values of optical basicity cited in Slag Atlas [3]

CaO SiO2 Al2O3 MgO BaO Na2O Li2O K2O FeO MnO CaF2 B2O3 TiO2
UV shift 1.0 0.48 0.605 0.78 1.15 1.15 1.0 0.42
Pauling e. n 1.0 0.60 0.78 1.15 1.15 1.0 1.4 0.51 0.59 0.43, 0.67 0.42 0.61
Electron densities 1.0 0.47 0.66 0.92 1.08 1.11 1.06 1.16 0.94 0.95 0.42 0.65
Values derived from (a) measurements of UV shift (b) Pauling electron negativities (b) and (c) Electron densities [38, 39]
9 Properties of Mould Fluxes and Slag Films
9.2 Structure of Slags 295

• The optical basicity is a reasonable measure of the M–O bond strength [35] but
does not differentiate between the size of cations which is important in electrical
resistivity and diffusion.
The corrected optical basicity has been used to estimate the following properties:
viscosity [36, 41], density [42], electrical conductivity and resistivity [43] and
thermal conductivity of molten slags [35] but has not been widely used for mould
fluxes because of the uncertainty in the value for KCaF2 .

9.2.2.4 Concentrations of O°; O− and O2−

The structure can be represented in terms of the concentrations in the slag of

bridging O’s (O°) non-bridging O’s (O−) and free O’s (O2−). The principal limi-
tation to this approach has been that it requires a mathematical model [44–46] to
determine these concentrations. However, recently a method has been developed to
calculate these concentrations without the aid of a mathematical model [40, 47–49]
and this would appear to be exceedingly useful in predicting properties since it can
differentiate between Si–O and Al–O bonds and their effects on specific properties.
Recently, the method has been applied to the prediction of viscosities of slags
containing CaF2 [49] but the model, at the present time, has not been extended to
slags containing B2O3 which is used as a flux, particularly in calcium aluminate
mould slags used for casting high-Al steels.

9.2.3 Effect of Cations

Although the degree of polymerisation in the silicate network is the dominant factor
affecting the properties of slags, the nature of the cations (e.g. Ca2+ or Na+) can also
affect the properties [50, 51]. Cations can affect properties of slags (i) directly (e.g.
the effect of cation size on electrical conductivity) or (ii) indirectly by affecting the
slag structure. The properties are affected by the following.

9.2.3.1 The Field Strength of the Ionic M+–O− Bond

Which is frequently represented by either the parameter (z/r2) where z is the charge
and r is radius of the cation (M+) or by the optical basicity, K (Fig. 9.5). The bond
strength increases as (z/r2) increases and as K decreases; The optical basicity (K)
has been reported to have a linear and inverse relationship with (z/r2) [35].
296 9 Properties of Mould Fluxes and Slag Films

Fig. 9.5 Schematic diagram showing relative strengths (denoted by thickness of bond) of M+–O−
bonds for Li+, Na+ and K+ [50]

9.2.3.2 The Fraction of Ionic Bonding in the M–O Bond

M–O bonds are mostly ionic but also contain some covalence (e.g. the fraction of
ionic bonding (expressed as %) is 78, 48 and 59% for CaO, FeO and MnO,
respectively [40]); oxides with significant levels of covalent bonding tend to pro-
mote the formation of the glass phase.

9.2.3.3 The Size of the Cations

Since larger cations tend to (i) hinder the motion of one layer of molecules over
another layer and thereby increase viscosity and (ii) find it difficult to pass through
the “holes” in the silicate network which thereby cause a decrease in electrical
conductivity and diffusion (Fig. 9.6). The field strength (z/r2) is also affected by
cation size and large cations (e.g. K+) form weaker NBOs and thus are more easily
broken than those formed with Na+ and Li+ ions; this should aid the electrical
conduction and diffusion. However, any advantage for the K+ must be offset against
the more sluggish movement of these large ions.

Fig. 9.6 Schematic diagram showing hindrance caused by large cations in (top) viscosity and
(bottom) electrical conductivity [50]
9.2 Structure of Slags 297

9.2.3.4 The Bridging of Cations

Bridging occurs when divalent cations (M2+, e.g. Ca2+) carry out network breaking
or charge-balancing duties; this effect is absent with monovalent cations (M+, e.g.
Na+) and could lead to increases in viscosity and electrical resistivity (Fig. 9.7).

9.2.3.5 The Number of Cations

CaO gives rise to one Ca2+ ion, whereas, Na2O gives two Na+ ions; the number of
cations available to move is an important parameter in electrical conductivity and
diffusion [50, 51].

9.2.3.6 The Mixed Alkali Effect

The mixed alkali effect occurs when the slag contains two (or more) cations with
very different radii (e.g. when K2O is added to Li2O–SiO2, the decrease in electrical
conductivity is much greater than that predicted from linear mixing rules). It has
been attributed to the cations occupying foreign sites [52].

9.2.4 Effect of Temperature on Properties

Slag properties are very sensitive to temperature and this sensitivity increases with
increasing polymerisation of the melt. Increasing temperature tends to loosen the
silicate structure (akin to the de-polymerisation caused by addition of cations). Thus
the effect of increasing temperature on various properties can be predicted since it is
equivalent to de-polymerisation (i.e. it causes decreasing viscosity, electrical
resistivity and thermal conductivity).

Fig. 9.7 Schematic

representation of “bridging”
with Ca2+ ions: note bridging
is absent with Na+ ions [50]
298 9 Properties of Mould Fluxes and Slag Films

9.2.4.1 Solid Slags

Most slag films consist of a mixture of glass and crystalline phases. The ratio of
glass and crystalline phases in the slag film is dependent upon both composition
(glasses are promoted by high Q value) and the thermal history of the sample.
Most mould slags can be produced in the form of a glass by rapid quenching of
the molten slag. If the heat capacity of this glassy mould slag is measured during
heating, the following sequence of events occurs:
• There is a smooth increase in Cp until the temperature exceeds the glass tran-
sition temperature (Tg)
• At Tg there is a “step increase” in Cp (Figs. 9.8a and 9.9 [53]) and a threefold
increase in thermal expansion coefficient (a) (Fig. 9.8b); these events are
associated with the transition of the glass to a supercooled liquid (scl).

(a) (b)
3000 1500 40
2500
Density ρ, kgm-3
HT-H298 , kJkg-1

CP , Jkg K

30
2000 1000

10 α, K
2400

6
20
-1

1000 500 2300

-1
10
-1

0 0 2200 0
300 600 900 1200 1500 1800 300 600 900 1200 1500 1800
T, K T, K

Fig. 9.8 Schematic diagrams showing the effects of the glass transition on a Cp (right axis) and
enthalpy (HT–H298) (left axis); Cp : - .-. -, and * = scl and liquid phases; o and solid line = glass
and crystalline phase and (HT–H298): solid line = glass, crystalline and liquid phases; D = scl; and
b density and thermal expansion coefficient, a (right axis) Density (left axis), □ = crystalline phase
o; glass and scl and liquid phases; a; ▲ = glass and scl phases

Fig. 9.9 Heat capacity of a

mould slag as a function of
temperature showing both the
step-increase in Cp at Tg for
the glass and the slag film and
the apparent decrease in Cp
accompanying crystallisation
(solid line) (permission
granted, IOM/Taylor and
Francis [53])
9.2 Structure of Slags 299

• In the case of some mould slags, an inverted peak in Cp is observed to start at

about 100 °C above Tg; this is due to partial crystallisation of the mould slag
(Fig. 9.9). (Note crystallisation is a first-order, exothermic transition and in DSC
measurements the exothermic, enthalpy of transition manifests itself as an ap-
parent decrease in Cp).
• The mixture of (glass+crystalline) phases undergoes an endothermic transition
at the solidus temperature of the slag (Tsol) when the crystal fraction starts to
melt and this process (crystal ! liquid) is complete when the temperature
reaches Tliq; note the scl undergoes a smooth transition to liquid (Fig. 9.8a).
The (crystal ! liquid) transition is accompanied by an enthalpy of fusion and a
change in density (or volume), whereas the equivalent values for the (scl ! liquid)
transition are very low or are zero.

9.2.4.2 Liquid Slags

When a molten slag is cooled below Tliq it forms either crystallites or a supercooled
liquid (scl). If viscosities are measured in a cooling cycle, when solid particles
(crystallites) are precipitated (Fig. 9.10a), there is a sharp increase in viscosity and
the slag exhibits non-Newtonian behaviour (i.e. measured viscosities differ for
different rotation speeds). In contrast, when a liquid slag transforms to a scl it is
accompanied by a smooth increase in viscosity (Fig. 9.10b) until the temperature
reaches Tg (log10 η (dPas) = 13.4 at Tg).

9.3 Crystallisation in Mould Fluxes

9.3.1 Importance of Crystallisation to the Process

The first slag to form in the (shell/mould) channel is probably glassy because of the
high cooling rates involved. However, crystallisation of the slag film takes place as the
cast proceeds. The slag film controls the heat transfer from the shell to the mould.
Crystallisation of the slag film is very important to the continuous-casting pro-
cess. The three principal factors affecting the heat transfer across the slag film are
(i) the slag film thickness (which is related to the break temperature, ds" as Tbr")

Fig. 9.10 Plots of log10

viscosity as functions of
reciprocal temperature (T−1
K )
for a a slag where crystallites
are precipitated (Q = 2) and
b glassy slag forming a scl
(Q > 2.5)
300 9 Properties of Mould Fluxes and Slag Films

(ii) the degree of crystallisation of the slag film (or fcrys) and (iii) the size of the
crystallites (Dcrys). Increased crystallisation in the slag film results in:
• Both scattering and reflection of the IR radiation (thereby, decreasing the
effective thermal conductivity) (qhor # as fcrys ") [54, 55]; the reflection and
scattering of IR radiation increases with increasing crystal size (Dcrys) with the
extinction coefficient of the slag, increasing with increasing Dcrys [56, 57].
• The creation of both porosity and an interfacial resistance (RCu/sl) due to the
formation of a “gas gap” (or “surface roughness”) at the mould/slag interface
[58] caused by the shrinkage resulting from the (glass ! crystal) transition
(qcrys > qglass); the porosity and RCu/sl both reduce the horizontal heat flux (qhor
# as fcrys ") [59].
• An increase in lattice thermal conductivity (klat with klat crys  2klat gl) [60–62];
thus (qhor " as fcrys ").
Since the overall effect of crystallisation is to reduce the horizontal heat transfer,
it follows that the combined effects of the reduction in radiation conduction along
with the creation of the interfacial resistance (RCu/sl) are more important than the
increase in lattice thermal conductivity resulting from crystallisation. There has
been some dispute concerning the relative magnitudes of the decrease in heat flux
due to scattering of radiant energy and the interfacial resistance. However, the
analysis of heat flux data obtained in plant trials [63, 64] identified the scattering of
radiant energy as the more important process. As mentioned above, scattering of IR
radiation is also affected by the size of the crystals in the slag film with scattering
(i) increasing with increasing size and (ii) being at an optimum level for Dcrys = 2–
3 lm [56]. The crystal size has been reported to increase with increasing slag
basicity (C/S) [65].
The formation of a slag film with a well-developed crystalline fraction is essential
in minimising longitudinal cracking in peritectic MC steels. In contrast, for HC steels
a thin, slag film, composed of glass (with no crystalline phase) is important to
promote horizontal heat transfer to create a thick and stronger shell to avoid sticker
breakouts. This is especially important for steel grades which result in a weak shell,
e.g. HC steels, in ULC steels where ferrite is formed and in ferritic stainless steels.
The soft, weak shell needs strong cooling to avoid sticker breakouts.

9.3.2 Crystalline Phases Formed in Slag Films

In conventional mould slags, the primary crystallising phase is cuspidine. However,

other primary crystalline phases are formed in other, specialised, mould slags (e.g.
F-free and CA-type slags used to cast high-Al steels); these include the phases, CT
(perovskite) [66] CST [67], NC2S3 [68] melilite [69]. Consequently, the phases
formed in conventional and fluoride-free mould slags are discussed individually
below.
9.3 Crystallisation in Mould Fluxes 301

9.3.2.1 Conventional (F-Containing) Powders

Grieveson et al. [70] produced a phase diagram for F-containing, mould slags with
(%CaO/%SiO2)  1.0. It was found that cuspidine was always formed, but the
nature of the other phases (both glassy and crystalline) formed was affected by their
chemical composition. The phase diagram for the 3CaO
2SiO2
CaF2 – CaF2
system has been investigated [57, 70] and the effects of Na2O on the cuspidine
phase relations have also been reported [69, 72].
Certain oxides are reported to promote the formation of crystalline phases,
whereas others promote the formation of glassy phases.
Crystalline phase formation is promoted by increases in concentrations of the
following:
(i) CaF2 (for cuspidine formation);
(ii) High basicity (C/S) [73] or decreasing SiO2 content; both the amount of
cuspidine and its crystal size (Dcrys) in the slag film increase with increases in
basicity (C/S) [65] from 0.8 to >1 [65, 74].
(iii) Li2O
(iv) Particles of ZrO2 [77] TiO2 [75] and Ti(CN) [25, 27, 75, 76] are reported to
act as nucleating agents and evidence has been obtained for cuspidine
nucleating on ZrO2 particles [77].
(v) The application of an electric field to the slag affects crystallisation [78, 79]
with both AC and DC increasing fcrys with crystal size (Dcrys) increasing with
DC and decreasing with AC [79].
Glass phase formation is reported to be promoted by additions of the following
oxides:
(i) SiO2
(ii) B2O3 [79, 80]
(iii) Na2O [81]; MnO [82, 83] (with >4%MnO and >7%MgO) and FeO [56]
which all tend to retard the crystallisation process [82, 83];
In F-containing powders (denoted here as conventional powders) the principal
crystalline phase is cuspidine (3CaO
2SiO2
CaF2 which is sometimes abbreviated
to C3S2Fl).
Cuspidine is usually the first phase to crystallise; consequently, cuspidine pre-
cipitation is usually referred to as primary crystallisation. Since it is formed at an
early stage, cuspidine is found as small crystals on the mould-side of the slag film
(i.e. the cold region) [84, 85]. The Gibbs free energy of formation of cuspidine has
been determined as DGf (kJmol−1) = −5188 + 0.82 TK [86] and as DGf = −RTK

log pSiF4 = −5253 + 0.864 TK [87] where TK is in K.
The kinetics of crystallisation are usually carried out by determining
Transformation–time–temperature (TTT) curves [88, 89] for the mould slags (see
Sect. 9.2.4). It was found that crystallisation times increased with increasing SiO2
content in the slag.
302 9 Properties of Mould Fluxes and Slag Films

Fig. 9.11 Phases formed on solidification by molten casting fluxes (cuspidine is not shown but is
always present) (permission granted, IOM/Taylor and Francis [70]; re-drawn SWEREA/KIMAB)

Secondary crystallisation consists in the precipitation of other phases (e.g.

nepheline or gehlenite) and tends to be found towards the shell side of the slag film.
The mineralogical nature of these late-forming phases for powders with a C/S ratio
close to 1.0 is shown as a phase diagram in Fig. 9.11 [70].
Agitation was found to have significant affect on primary crystallisation but had
a much smaller effect on secondary crystallisation [90].
A study of the effect of Na2O on the formation of cuspidine (C3S2Fl) crystals found
that (i) F− ions tend to bond with Ca2+ ions (ii) the maximum amount of cuspidine after
annealing was obtained with a sample containing 4% Na2O [72]. The latter finding
was attributed to the effect of Na2O on both the nucleation and growth rate (via a lower
viscosity). Most mould slag compositions lie outside the phase field for cuspidine and
it has been reported that moving the slag chemical composition into the cuspidine-
field provides slag films with reduced, horizontal, heat fluxes [69].
Low-basicity (C/S) mould slags have been used to cast high-Al steels and CaF2
was identified in the mould slag [91]. More recently, thermodynamic databases
have been used to identify the crystalline phases formed in these low (C/S) mould
slags [92].

9.3.2.2 Fluoride-Free Fluxes

In recent years, there has been a drive to reduce the magnitude of fluorine emissions
in steelmaking operations. In F-free fluxes, the absence of F means that no cus-
pidine can be formed to scatter the IR radiation and thereby reduce the horizontal
heat flux.
9.3 Crystallisation in Mould Fluxes 303

Successful F-free powders have been developed for billet casting but the
development of a successful mould powder for slab casting remains an ongoing
challenge at the present time. Investigations into possible replacements for CaF2
tend to fall into one of two classes:
(i) Those where CaF2 is replaced by other fluxes, e.g. Li2O [93–95] Na2O [96,
97]:K2O [98] BaO [99, 100] B2O3 [28, 101, 102] TiO2 [66, 67, 102, 103].
(ii) Those where cuspidine is replaced by specific, crystalline phases (e.g. per-
ovskite, CaTiO3 or CT) [66, 103] or melilite (solid solution ranging from
akermanite (C2MS2) to gehlenite (C2AS) [69] or the phase NC2S3 [68].
It should be noted that the nature of the crystalline phase is not an issue, pro-
viding (i) the crystalline phase brings about the necessary reduction in horizontal
heat flux and (ii) it causes no collateral problems. One example of the latter powders
where cuspidine is replaced by CaTiO3 [66] reputably, resulted in a high incidence
of sticker breakouts; this was attributed to the formation of Ti(CN) (formed by
reactions of the TiO2 and C and N2 in the mould) [104]. However, this view has
been disputed recently [102].

9.3.2.3 Calcium Aluminate Slags

The effect of the (C/A) ratio on the phases formed in CA-type slags has been
studied [105]. When the (C/A) ratio was less than 0.9, CaF2 and lithium aluminate,
LA, were the principal phases formed. With (C/A) = 1, no CaF2 remained and the
phases, LA, C3A, C11NA4 and C3-xNxA but when (C/A) = 2.3 the LA disappeared
and was replaced by CaO [105]. The LA phase formed at high alumina comcen-
trations has a high melting temperature. In CA slags containing B2O3 the phases
(C9B3Fl and CB+ a large amount of C3S2Fl) were detected [106].

9.3.3 Crystallisation Process

It has been reported above that primary crystallisation results in the formation of
cuspidine and other phases (e.g. carnegieite, NAS2) are formed in secondary
crystallisation [70]. Homogeneous nucleation tends to be diffusion-controlled [77]
with outward diffusion of MgO, Al2O3 and Na2O and inward diffusion of CaO,
SiO2 and CaF2 [84]. However, heterogeneous nucleation has been observed to
occur on ZrO2 particles [77]. It has also been suggested that TiO2 and ZrO2 par-
ticles [77] promote crystallisation principally by acting as nucleating agents, (ZrO2
has low solubility, ca. 2%, in mould slags). It is known that ZrO2 and TiO2 tend to
increase the break temperature and thereby create thicker slag films.
304 9 Properties of Mould Fluxes and Slag Films

Scheller [111] observed tiny (5 lm) droplets of metals (e.g. Na) resulting from
redox reactions in the mould slag; it was suggested that these droplets could also act
as nucleation sites for crystallisation.
It has also been suggested [107] that pores in the slag film can act as nucleating
agents (Fig. 9.12a, b). However, this is a “Chicken and egg situation” and it is
debatable whether the pores are located near the crystals because of their nucleation
of crystals or, alternatively, the pores are the result of the shrinkage accompanying
the crystallisation process. Figure 9.12c shows the number of pores as a function of
pore diameter and it can be seen that there is a large number of fine pores [107]. It
was suggested that in the liquid layer, bubbles can move, collide and grow and thus
larger bubbles are found in the liquid layer. However, once the crystallisation is
initiated there is rapid dendritic growth and the bubbles cannot escape into the
liquid and consequently, get trapped in the slag film [107].
Slag components which promote the glassy phase (e.g. B2O3; Na2O, Li2O [108,
109] FeO [110] and high amounts of MnO and MgO [83] reduce the fcrys in the slag
film. However, components which tend to promote cuspidine formation (e.g. CaF2
and the (C/S) ratio [69] also increase fcrys. The absorption of large amounts of
alumina results in increased crystallisation [69]. Components, such as TiO2, result
in the formation of perovskite (CaTiO3) and thereby increase fcrys.
The application of an electrical potential has also been found to affect the
crystallisation [78, 79] with increasing voltage causing increases in fcrys with both
AC and DC and an increase in Dcrys for DC and a decrease in Dcrys for AC.
It has been reported that the crystal grain size for the slag film in a simulation test
was greater in the lower mould than in the upper mould.

Fig. 9.12 a, b Photographs (a) (b)

of slag films A (fcrys = 0.9)
and B (fcrys = 0.3) with mould
on left and shell on the right;
showing 1 = fine crystals
2 = dendritic crystals and
3 = glassy (former liquid)
[107] and c number of pores
as a function of pore diameter,
A = ●, ♦; B = ▲, ■; [107]
(courtesy of, Li and Carboox
Number of pores per mm film

[107]) 100
(c)
10

0.1
1 10 100 1000
Pore diameter, μm
9.3 Crystallisation in Mould Fluxes 305

Fig. 9.13 a, b Schematic drawing showing the development of the crystalline nature of the slag
film during the casting showing dendritic growth in the warmer parts of the slag film and fine
crystals formed in cold regions of slag film where there is significant undercooling (courtesy of Li
and Carboox [107])

9.3.4 Crystallisation Kinetics

The first slag to penetrate into the mould/shell gap probably forms a glassy slag film
because of the high cooling rate involved. This glassy, slag film crystallises over
time. The rate of crystallisation of the slag film is dependent upon both the time and
temperature experienced. The kinetics of crystallisation is usually represented in the
form of a Time–Temperature–Transformation (T–T–T diagram which plots time
and temperature and shows contours to reach a certain % of transformation.
A typical example is shown in Fig. 9.14a. Such diagrams are usually constructed
from measurements from the double hot thermocouple (DHTC) method [88, 89,
111] or other methods such as DTA [112] high-temperature optical microscopy or
by quenching in sealed capsules [77, 113].
The kinetics of crystallisation involves two processes, nucleation and growth.
Temperature has opposing effects on these processes; nucleation is promoted by
undercooling (i.e. decreasing temperature) whereas growth increases with
increasing temperature. Take a case where the TTT curves exhibit a single nose
indicating the precipitation of a single phase. There are two temperature regimes
[114]; at high temperatures (just below Tliq) there is no undercooling but as the melt
cools as the undercooling/nucleation increases and once a crystal is nucleated, the
high temperatures promote rapid crystal growth.
At lower temperatures, nucleation is easy but the growth process is slow (due to
the high melt viscosity) and so fine crystals are formed.

9.3.4.1 Conventional Fluxes

It is generally assumed that the first slag film formed is glassy because of the high
cooling rates involved. Over time, this glassy slag film can crystallise, the extent to
which it crystallises is dependent upon its composition (especially the C/S ratio) the
annealing time and the local temperature. Longitudinal cracking tends to be more
prevalent in the early part of the casting. In these early parts of the cast the
306 9 Properties of Mould Fluxes and Slag Films

Fig. 9.14 a T–T–T diagram

for a mould slag with
composition 45%CaO + 41%
SiO2 + 7%Al2O3 + 7% Na2O
showing the onset of
crystallisation (curve and
points) and b resultant
morphologies; A = Equiaxed
crystals formed at high
temperature; B = Columnar
crystals; C = faceted crystals
and D = very fine crystals
formed at high undercooling
and c Photograph of section
of slag film showing dendritic
growth of crystals on shell
side and fine cubic crystals on
mould side (permission
granted, (a, b) ISS, re-drawn
SWEREA/KIMAB after [88]
(c) ISIJ [67])

crystalline layer of the slag film is not fully developed (Fig. 9.13a) and this results
in high heat fluxes and to longitudinal cracking. However, the cracking will tend to
decrease as the crystalline layer of the slag film (Fig. 9.13b) develops and this layer
reflects and scatters the IR radiation.
9.3 Crystallisation in Mould Fluxes 307

The primary crystallisation (of cuspidine) occurs by inward diffusion of CaF2,

CaO and SiO2 and by the outward diffusion of Na2O, Al2O3 and MgO [84]; it
would appear that MnO aids this process [82] whereas TiO2 additions result in the
formation of perovskite (CT) which suppresses the formation of cuspidine [30].
A typical example of transformation–time–temperature (T–T–T) plot is shown in
Fig. 9.14a [88]. The various smooth curves represent different cooling rates where
the two irregular curves show the contours for 0 and 50% transformation. It can be
seen from Fig. 9.14a that:
• The cooling rates increase as we move from right to left and the sample tends to
become more glassy, as we move from right to left.
• When annealing at 900 °C, the sample remains totally glassy for just over 10 s
and is 50% crystalline at 50 s.
• The rate of crystallisation decreases at higher and lower temperatures (e.g.
800 and 1200 °C, respectively).
It can be seen from Fig. 9.14b that large equiaxed crystals are formed at high
temperature and that fine crystals are produced at lower temperature; thus, crystal
size tends to increase with increasing local temperature in the slag film. These
features can also be seen in the slag film shown in Fig. 9.14c [67]. The time taken
for crystallisation to occur is referred to as the “incubation time”. It has been
reported that Na2O reduces the incubation time and is thus beneficial for faster
crystallisation of the slag film [109].
Rocabois et al. [77] examined the theory regarding the temperature of the “nose”
in Fig. 9.14a and suggested a value of Tnose = 0.77 Tliq for most slag systems,
However, they found values for mould slags tended to follow Tnose = 0.86 (±0.06)
Tliq. They also looked at the critical cooling rate (dT/dt)crit = (Tliq − Tnose)/tnose
which is the cooling rate necessary to produce a glass.
Inspection of TTT curves for various mould fluxes shows the transformation
contours tend to move to right with increasing SiO2 content (or increasing poly-
merisation, Q), i.e. they require longer crystallisation times. Some mould fluxes
with high SiO2 contents may not crystallise at all.
It can be seen from Fig. 9.14a that the morphologies of the crystals are also
dependent on time and temperature with large dendritic crystals being formed in
high-temperature regions and progressively finer crystals being formed with
increasing supercoiling (i.e. lower temperatures) [88].
The effects of the following oxides on the crystallisation kinetics have been
investigated: Na2O [57, 67, 72], Li2O [57, 65], K2O [57], MnO [82], B2O3 [65].

9.3.4.2 F-Free Fluxes

TTT diagrams have also been developed for F-free fluxes containing TiO2 [81] and it
was found that TiO2 reduced the activation energy for crystallisation and that Na2O
also reduced the incubation time. However, it was also reported that the horizontal
heat flux increased when Na2O > 7% and MnO > 4% [81] to these oxides promoting
308 9 Properties of Mould Fluxes and Slag Films

the glass phase. This latter finding is consistent with observations of Susa et al. [56,
110] who observed that FeO (another transition metal oxide) additions (up to 2%)
reduced the apparent reflectivity of the crystals. This is a surprising result since FeO,
MnO, NiO, etc. are added to increase the absorption of IR radiation in mould fluxes
where no crystalline phases are formed (i.e. qhor# as % FeO "). Thus for slag films
where crystals are formed, the total heat flux increases with increasing FeO since the
FeO reduces both the reflectivity (R′) of the crystals [110] and the grain size (Dcrys) of
the crystals [56] (i.e. qhor" as % FeO " since R′# and Dcrys #). The reflectivity was
found to increase with increasing grain size and it has been calculated that control of
grain size could give a 7–8% decrease in the horizontal heat flux [56] and manifests
itself as an increase in Extinction coefficient [57].

9.3.4.3 Calcium Aluminate Slags

The effect of (C/A) ratio on the incubation times of CA-type slags has been
investigated [105]. It was found that the incubation times were longest (i.e. crys-
tallisation was slowest) when (C/A) = 1.5 [105].

9.3.5 Effects of Crystallisation

Crystallisation causes changes in the properties of mould slags; these are detailed
below.

9.3.5.1 Density

The crystalline phase has a greater density than the glass (qcrys > qglass) because of
the better packing in the crystal; thus, shrinkage occurs during crystallisation. Slag
films contain some porosity and this porosity lowers the density of the slag film. At
temperatures above Tg, the supercooled liquid (scl) formed from the glass has a
much higher thermal expansion coefficient than that of the crystalline phase.

9.3.5.2 Transport of Heat

It was pointed out that the shrinkage accompanying crystallisation in the slag film
has the following effects on the horizontal heat flux (qhor).
(i) The tiny gas bubbles in the crystallised region (Figs. 9.12 and 9.15) reduce both
the effective thermal conductivity of the slag film and the heat flux (qhor) [107].
(ii) The creation of an interfacial resistance (RCu/sl) or “gas gap” at the
mould/slag interface also reduces qhor [115].
9.3 Crystallisation in Mould Fluxes 309

(iii) The crystallites scatter the IR radiation, reducing both kR and qhor [56, 110].
(iv) The lattice thermal conductivity is increased (i.e. kcryst  2kglass) [60, 61].
The overall effect of crystallisation is to reduce qhor.

9.3.5.3 Hydrogen in Slag Film

Significant levels of porosity have been reported to occur when the slag film
contains high levels of H2 and this has been linked to incidences of sticker breakout
[116, 117].
Presumably, the porous slag film has a low thermal conductivity which reduces
the horizontal heat flux and results in a thin, weak shell. Possible sources of H2 are
(i) high moisture levels in casting powders and (ii) water leaks in the
continuous-casting process.

9.3.5.4 Heat Capacity

The heat capacity of the scl formed from the glass phase at T > Tg is higher than
that of the crystalline phase for the temperature range between Tg and the solidus
temperature of the slag, as can be seen in Fig. 9.9.

9.3.5.5 Friction Forces and Lubrication of the Shell

Hering et al. [118] reported that the liquid frictional forces are dependent upon the
mineralogical phases formed (Figs. 9.15 and 9.16); the friction force for nepheline
being greater than those where combeite; wollastonite and gehlenite are formed
which may be related to the Tliq values of the phases.

Fig. 9.15 Photograph

showing fine pores associated
with crystallisation
(left = mould; right = shell)
(courtesy of Carboox [107])
310 9 Properties of Mould Fluxes and Slag Films

Fig. 9.16 Friction force as a 30

function of slag viscosity (in vG : 0.9 - 1.0 m/min
Pas = 10 d Pas) showing the
effect of mineralogical phase 25
(11)
on the friction (permission
granted, re-drawn 20

Friction ,kNm-2
SWEREA/KIMAB after Nepheline
[118])
Combeite
15
(5) Gehlenite
(3)
10 (10) (15)

Wollastonite
5
( ) %Al2O3

0
0 0.03 0.06 0.09 0.12 0.15 0.18 0.21
Viscosity (Pas = 10 mPaS)

There is little, or no, crystallisation in some high-viscosity billet slags. These slags
form a supercooled liquid (scl) at temperatures above Tg. Although, the slag film has
a high viscosity, it will tend to move when subjected to stress from the moving shell.
In thermal expansion measurements at T > Tg, the slag sample cannot support its
own weight, and so the sample collapses due to the pressure of the probes, (see
Fig. 9.41a). Thus, a glassy slag film (scl) will provide some measure of lubrication to
the shell. In contrast, crystalline phases tend to be more rigid and are prone to fracture
[119]. The fracture of the slag film can occur near the bottom of the mould and results
in variable heat transfer and in “star and spongy cracking” [120, 121].

9.3.6 Methods of Determining Fraction of Crystalline Phase

in Slag Films

The fraction of crystalline phase (fcrys) in the slag film is an important process
variable. Yet, hitherto, it has not been treated with such reverence, as say, the slag
viscosity or the break temperature. A possible reason for this is that it fcrys has
proved difficult to quantify and the measurements obtained tend to show consid-
erable variation according to the method used.
Experimental Methods

Several methods have been proposed to determine the fraction of crystalline phases
(fcrys) formed in the slag film under steady state conditions. The various techniques
used are summarised below. There is an urgent need for an inter-comparison of the
results for fcrys obtained with these various methods. The development of a standard
test and recommended procedures for determining fcrys are urgently needed.
9.3 Crystallisation in Mould Fluxes 311

9.3.6.1 Metallographic Method

In this method, the slag film is mounted so that its cross section can be examined
and then polished to 1 µm and etched in 2.5% aqueous HF solution [83, 107]. The
sample is then examined in an optical microscope and the relative amounts of
crystalline and glassy phase are determined. The method is slow and time con-
suming and since the observation area is small and a large number of observations
are needed to obtain a representative value for the whole slag film [107, 122].

9.3.6.2 X-Ray Diffraction Methods

The slag film is ground to a powder and then subjected to X-ray diffraction (XRD).
A typical XRD pattern is shown in Fig. 9.17; the crystalline phases are charac-
terised by sharp peaks and the glassy phases appear as a diffuse “hump”, as shown
in the lowest pattern in Fig. 9.17. Measurements can be made of (i) the intensities
(I) of specific peaks (corresponding to crystallographic plane) [54, 55, 123] or (ii) to
the area under the peaks [61, 62, 123].

9.3.6.3 Measurements of Intensities

These experiments are usually comparative methods where the intensities of a

specific peak are compared with those of standard samples with known fcrys values.
The intensities of peaks from the slag film have been compared with the following:
(i) Cuspidine/a-Al2O3 [54, 55].
(ii) Cuspidine/glass mixtures [123].

Fig. 9.17 Typical X-ray 2500

diffraction (XRD) patterns of Cuspidine
two slag films (A (fcryst = 0.9) Fluorite
and B (fcrys = 0.3)) showing 2000
sharp peaks associated with
crystalline phases and the 1500 B-mould
“hump” due to glassy phase
Counts

(bottom pattern) (courtesy of

S Riaz [123]) 1000 A-mould

500 A-shell

B-shell
0
10 20 30 40 50 60 70
2θ
312 9 Properties of Mould Fluxes and Slag Films

In these experiments, the ground slag film is mixed with known weights of the
Al2O3 or glass and the samples homogenised. The intensities of specific peaks in
the XRD pattern are then compared with those from standard samples (e.g. samples
of 100/0, 75/25, 50/50, 25/75, 0/100, respectively, of cuspidine and a-Al2O3 [54,
55]. The XRD patterns of these various mixtures are then measured and the ratio of
integrated intensities (e.g. Icuspidine (222)/IAl2 O3 (012)) determined (where () denotes the
crystallographic plane of the peak [54, 55]. In these experiments the slag film is
ground and mixed (in the case of the Al2O3 experiments) with a known amount of
Al2O3 and the intensity ratio determined by XRD. The intensities are then com-
pared with those for the standard mixtures. Since the samples are small at least five
measurements should be made to obtain a representative value. One major problem
with this method is that pure cuspidine is extremely difficult to obtain.

9.3.6.4 Measurements of Areas

It can be seen from Fig. 9.17 that the glass appears as a “hump” and the crystalline
phases as sharp peaks. Thus the total area under the peaks (Atotal) represents (glass +
crystalline) and the area under the hump (Ahump) represents the glass phase. The
crystal phase is represented by (Atotal – Ahump). Standard mixtures of glass and
cuspidine were made up and it was observed that fgl and fcrys could be determined
from Eq. 9.8 [123].


fgl ¼ 1  fcrys ¼ Ahump =Atotal ð9:8Þ

A similar method has been used by Andersson [61, 62]. Since the sample size is
small, a number of determinations should be made to ensure the fcrys is represen-
tative of the whole slag film.

9.3.6.5 Measurement of Physical Properties

Measurements of heat capacity (Cp) and thermal expansion coefficient can be used to
measure fcrys. When a glass is heated through the glass transition temperature
(Tg, which occurs around 600 °C, for most mould slags) it transforms from a solid into
a supercooled liquid (scl). This transition is accompanied by (i) a large-step change in
heat capacity (DCTg
p ) and (ii) a threefold increase in thermal expansion coefficient (a)
[124, 125]. The method consists in determining the property changes at Tg for (a) the
glass prepared from the slag film and (b) the slag film [124, 125]. Values of fcrys can
then be calculated by Eq. 9.9. It was found difficult to use measurements of the
thermal expansion to determine fcrys since the sample (scl) collapses about 650 °C
which severely restricts the temperature range of the measurements [124].
9.3 Crystallisation in Mould Fluxes 313

fgl ¼ ð1  fcrys Þ ¼ ðDCpTg Þslagfilm =ðDCpTg Þglass : ð9:9Þ

Measurements of thermal conductivity/diffusivity of the slag film at room

temperature would appear to offer considerable potential. Measurements on both a
slag film and a glass (made from the slag film) can provide a sensitive scale for
measuring fcrys since kcryst  2kglass.

9.3.7 Tests to Simulate fcrys Formed in Slag Film

Several tests have been reported to simulate the fcrys formed in the slag film with the
de-carburised mould powder. These tests are briefly described in Table 9.2.

9.3.8 Empirical Rules to Calculate the Crystal Fraction

in Slag Films

The fraction of crystalline material (fcrys) will be dependent upon both the com-
position and the thermal history of the slag film (time and temperature). Equations
correlating fcrys with compositional parameters have been reported by several
investigators [80, 83, 108]. In the relation proposed by Sakai [108] fcrys is shown as
a graphical function of (XCaO + XMgO + XCaF2 ). In the Sankaranarayanan relation
[80] fcrys is represented as function of the de-polymerisation index (DI) but does not
contain the component CaF2 which is essential for cupidine formation in mould
slags.

Table 9.2 Summary of published tests using mould fluxes to determine the fraction crystalline
phase (fcrys) developed in slag films
Test/reference Details of method
Lidefelt [126] Triangular—(side 30  30 mm)  45 mm long hole machined in refractory
brick and Cu plate inserted. Molten flux at 1500 °C poured into brick. Cut
down the middle and examined metallographically to determine fcrys
Sakai [108] Hole (30  30  5 mm wide) machined in stainless steel block. Molten
mould flux poured into mould which was held at 300 °C. fcrys determined
metallographically
Stollberg 25 g molten mould flux poured into a copper mould held in cold water. The
[127] sample was then removed and examined metallographically to determine fcrys
Li [107] The following procedure was found to provide best match to fcrys for slag films.
20 g of molten decarburised mould powder is heated to 1300 °C and held there
for 20 min. It is then poured into a stainless steel crucible at 610 °C and then
held at 610 °C for 20 min. The fcrys of the sample was determined
metallographically
314 9 Properties of Mould Fluxes and Slag Films

Fig. 9.18 Fraction crystalline

phase (fcrys expressed as %) as
a function of modified
(NBO/T)* ratio (permission
granted, SAIMM, [83])

Li et al. [83] measured the fraction of crystalline phase (fcrys) in slag films taken
from the mould and tested various parameters (i.e. modified O/Si ratio, modified
mole ratio and modified NBO/T* ratio and maximum cuspidine formed) to provide
a fit with the experimental fcrys values (Fig. 9.18). It was found that the modified
(NBO/T)* provided the best fit (Eqs. 9.10 and 9.11) where X is the mole fraction of
a component (e.g. CaO). Note CaF2 is treated as a network breaker in the calcu-
lation of the modified (NBO/T)* in Eq. 9.9. This equation presumably pertains to
the time when the crystallisation has achieved a steady state.

%crys ¼ 100 fcrys ¼ 284 þ 141:1ðNBO=TÞÞ ð9:10Þ

Modified ðNBO=TÞ ¼ 2fXCaO þ XFeO þ XNa2 O þ XK2 O

  
þ XCaF2 þ XMgO þ XMnO  XAl2 O3 ð9:11Þ

 
= XSiO2 þ 2XAl2 O3 þ XMgO þ XMnO

Where the contents in the curly bracket {}, are included in the numerator if MgO
is <7% and MnO is <4% and in the denominator when they exceed these values. It
can be seen from Eq. 9.8 and Fig. 9.18 that fcrys = 0 (i.e. slag film is totally glassy)
for mould slags with (NBO/T)* values of 2.0 or less and fcrys = 1 (i.e. slag film is
totally crystalline) for slags with (NBO/T)* > 2.75.

9.3.9 Data for fcrys

Measurements of fcrys have been carried out by several workers [54, 61, 62, 83, 110,
123, 125, 128, 129], with fcrys increasing as (i) Li2O decreased [108] (ii) as (C/S) or
NBO/T increases [130]. The evolution of fcrys with time has also been investigated
[57, 71].
9.4 Physical Properties of Mould Slags 315

9.4 Physical Properties of Mould Slags

The performance of the mould slag in the continuous-casting process is greatly

affected by its physical properties. The various physical properties are discussed
below. In the following text, mould slags used for continuous casting are divided
into two categories, namely, (i) conventional slags (containing CaF2) and (ii) mould
slags for specialist casting (see Chap. 6) which may, or may not, contain CaF2.
Although the required property values are largely determined by the mould
dimensions, casting conditions, steel grade, etc., the compositions of the various
slags must be adjusted to meet the required values.
Some of the required tasks carried out by continuous-casting mould powders are
identical to those performed by powders in Ingot casting, whereas others vary in
importance. Powders for ingot casting are described in detail in Chap. 7 but the
properties of powders for continuous casting and ingot casting are compared in
Sect. 9.7.

9.4.1 Thermodynamic Properties and Liquidus

Temperatures (Tliq)

The Gibbs free energy for cuspidine formation has been determined in two studies
and the DGf values were found to be in good agreement [86, 87].
A thermodynamic database covering some mould powder compositions has been
developed by Jung and van Ende [131] and has been used to predict the pick-up of
Al2O3 when casting a high-Al steel with a low basicity mould slag containing CaF2.

9.4.1.1 Importance to the Process

The liquidus temperature (Tliq) is the temperature where the casting powder
becomes completely molten. Knowledge of the Tliq of the casting powder helps us
to define the upper limit of the molten slag pool. A decrease in the liquidus tem-
perature would be expected to increase the depth of the molten slag pool.

9.4.1.2 Measurement Methods

The melting range of a casting flux can be measured by a variety of techniques

outlined in Table 9.3.
316 9 Properties of Mould Fluxes and Slag Films

Table 9.3 Methods used to determine liquidus temperature

Method Details of method Comments
Boat tests Known mass of decarburised sample placed in
combustion boats and placed in muffle furnace
(for say, 7 min) to come to thermal equilibrium at
various set temperatures and then quenched.
Examined metallographically for evidence of
melting
Quench tests As above, but mould powder sealed into Pt tubes
[113]
DTA/DSC There are 2 pans, one holding the sample and the
[132] other a reference material. The pans sit in a metal
block and are heated at a set heating rate and the
temperature difference between the two pans is
monitored continuously (Fig. 9.19a). When
sample undergoes an exothermic or endothermic
event there is departure from the baseline. For
accurate temperatures, measurements should be
carried out with (i) calibrants to check the
temperature scale and (ii) corrected for heating
rate. A typical DTA/DSC output is shown in
Fig. 9.19b
Hot stage The melted sample is placed in a loop in the Tendency for
microscope thermocouple (which also acts as heating element) undercooling
SHTC and allowed to solidify. The sample is then heated
DHTC at 5 °C min−1 and the specimen observed through
[133, 134, a microscope for melting. The sample is then
136] cooled at 15 °C min−1 and monitored for signs of
precipitation. Double hot thermocouple technique
contains 2 thermocouples and a thermal gradient
can be imposed across specimen. The temperature
field in DHTC has been modelled [136]
Leitz The shape of a pressed cylinder of de-carburised Tliq values tend to differ
microscope powder is monitored as it is heated at a set heating from Tliq values measured
test [135] rate. The temperatures where the sample takes up by DTA/DSC
DIN 51740 a set shape is determined Tsoftening, Themisphere and
[135] Tfluidity are measured to define the melting range
(Fig. 9.20)

9.4.1.3 Liquidus Temperature Data

Values of the liquidus temperatures have been reported by several groups [17, 66,
95, 98–100, 102, 113, 132]. The effects of various constituents on the liquidus
temperatures for various types of mould slags have also been reported: for con-
ventional fluxes; Al2O3 [138, 139], Li2O [95, 139], B2O3 [100], MnO [139, 140],
9.4 Physical Properties of Mould Slags 317

Fig. 9.19 Schematic diagram showing a DTA apparatus and b Typical DTA/DSC trace for a
casting powder during heating (permission granted, Europ. Sci. Tech. Publ. [137])

Fig. 9.20 Schematic drawings a–c showing typical shapes corresponding to Tsoftening, Themisphere
and Tfluidity, respectively

Na2O on cuspidine phase [69, 141]: F-Free:Na2O [68]:Hi-Al low (C/S) [142];
CA-based: B2O3 [29]. The Tliq and Tbr of the remaining liquid have been deter-
mined after precipitation of the crystallising phase [128].

9.4.1.4 Methods to Calculate Tliq

Commercial thermodynamic software (e.g. Factsage, MTDATA, Thermocalc) are

capable of providing reliable values for Tliq of casting slags.
Less accurate, values of Tliq for most mould slags can be calculated to ±30 °C
using Eqs. 9.12 and 9.13 [83, 143]; these were derived using numerical “best fit”
analysis of measured Tliq values and compositions of mould fluxes. This method
identifies the changes in Tliq resulting from the addition of specific oxides
(or fluorides) on the slag and provides a mean value (or constant) to represent this
trend. When this method is applied to slags in general, Tliq values are subject to
much greater uncertainty (>100 °C) due to the effect of the phase field on Tliq [144].
The smaller uncertainty value associated with mould flux data is due to the fact that
mould fluxes tend to have similar compositions and tend to lie in the same phase
field.
318 9 Properties of Mould Fluxes and Slag Films

Tliq ð CÞ ¼ 1191 þ 11:4%SiO2  11%CaO þ 4:2% Al2 O3 þ 5:7%MgO

ð9:12Þ
 10:1%Na2 O  15:8%K2 O þ 1:9%F þ 8:3%Fe2 O3 þ 11:6%MnO

Tliq ð CÞ ¼ 1200  1:518%SiO2 þ 2:59%CaO þ 1:56%Al2 O3  17:1%MgO

 9:06%Na2 O  %K2 O þ %Li2 O þ 4:8% F  9:87%FeO  2:12%MnO
ð9:13Þ

9.4.2 Break Temperature (Tbr)

The break (or solidification) temperature is the temperature where solids are pro-
duced on cooling. The term, Break temperature, Tbr, is defined as the temperature
where the viscosity increases dramatically during cooling. Values of Tbr tend to
decrease with increasing cooling rate and the cooling rates in the mould are vey
high. It should be noted that Tbr is not an equilibrium value like Tliq and that Tbr
decreases with increasing cooling rate and thus most cited values relate to a
specified cooling rate (usually 10 K min−1). However, at high cooling rates it has
been observed the Tbr tends to come to a fixed value [145].

9.4.2.1 Importance to the Process

The break temperature is important because it determines the boundary between

solid and liquid layers in the slag film and thus the magnitude of the solid and liquid
film thicknesses (i.e. ds and dl). Since the horizontal heat flux, qhor, is inversely
dependent upon ds, any relation between qhor and ds (which is difficult to measure)
can be represented in terms of qhor and Tbr. The friction forces exerted on the shell
are dependent upon (dl)−1; thus, Tbr also affects the lubrication supplied to the shell
by the mould slag.

9.4.2.2 Measurement Methods

The break temperature is derived from viscosity measurements on liquid mould

slags when cooling at a set rate. It is detected by a sudden departure from the
baseline (Fig. 9.21) and the appearance of non-Newtonian behaviour (due to
presence of solid particles). It should be noted that it is difficult to detect Tbr if the
distance separating the two concentric cylinders is too large (say 10 mm) and a
value, of say, <6 mm is needed to detect Tbr.
9.4 Physical Properties of Mould Slags 319

Fig. 9.21 Schematic diagram

of log10 η as a function of
reciprocal temperature (K−1)
showing the break
temperature

Fig. 9.22 Plot of measured 1400.0

values versus calculated
values for Tbr (permission
granted, ISS/AIST, re-drawn
Tcalc oC

from [74])
1200.0

1000.0
1000 1200 1400
Tmeas (oC)

Kawamoto et al. [74] defined Tbr as the point where the fraction solid in the
(solid+liquid) exceeds a value of 0.49 and calculated this temperature using a
thermodynamic software package. The calculated and measured Tbr showed rea-
sonable agreement [74] (Fig. 9.22).
The solidification temperature is measured by DTA/DSC or by hot thermo-
couple method when cooled at a set rate.
Some workers use the term, crystallisation temperature. Feldbauer and Cramb
[146] have described this term as confusing since Tcrys will depend upon on TTT
curves and thus will be dependent upon the method and procedures used.

9.4.2.3 Data for Tbr, Tsol

Values of Tbr have been obtained from viscosity measurements e.g. [147–150] and
the effect of individual components on the Tbr or Tsol has been studied by a number of
workers: Al2O3 [151], ZrO2 [151, 152], TiO2 [27, 76, 109, 151, 153], B2O3 [28, 29],
MnO [140], RE O [154], Na2O [130], BaO, Li2O [155], Basicity [130].
320 9 Properties of Mould Fluxes and Slag Films

9.4.2.4 Methods to Calculate Tbr

Numerical analysis of experimental values of break temperature derived during

viscosity measurements yielded Eq. 9.14 [147, 156].

Tcrys ¼ Tbr ð C) ¼ 1241:6  0:0215XMgO  0:0141XAl2 O3

 0:0449XNa2 O  0:0855XCaF2 ð9:14Þ
 0:0641XLi2 O  0:1528XB2 O3 :

Values of Tbr were also estimated using the following relation [157] which was
derived from regression analysis (i.e. “best fit”) of experimental data for Tbr and
mould slag compositions. The uncertainty associated with this equation is ±25 °C.

Tbr ð CÞ ¼ 1120  3:3%SiO2  8:43% Al2 O3 þ 8:65%CaO  13:86%MgO

 3:3%Na2 O  18:4%FeO  3:2%MnO  2:2%K2 O  6:6%Li2 O
 6:47%F
ð9:15Þ

The performances of some of the above equations have been checked against Tbr
values obtained during viscosity measurements [148].

9.4.3 Glass Transition Temperatures (Tg)

The glass transition temperature (Tg) is the temperature, on heating, where a glass
is transformed into a supercooled liquid (scl) and this transition is accompanied by a
marked change in some properties (e.g. in Cp and thermal expansion coefficient).
Values of Tg [53, 132, 158] for most mould slags lie close to 600 °C (i.e. fall in the
range, 570–630 °C). A regression analysis was carried out on Tg data obtained in
DSC and DTA studies which yielded Eq. 9.16 [143]; the uncertainty in the cal-
culated Tg values is ±30 °C.

Tg ðKÞ ¼ 906  330:5XSiO2 þ 190XCaO þ 440XAl2 O3

 449XN2 O þ K2 O  11XMgO þ 154XCaF2 ð9:16Þ
 309XMnO  1391XFeO
9.4 Physical Properties of Mould Slags 321

9.4.4 Viscosities (η)

9.4.4.1 Importance to the Process

The slag viscosity is important to the continuous-casting performance because it

affects:
(i) The lubrication supplied to the mould (which correlates with 1/η).
(ii) The SEN erosion rate which is proportional to (1/η) and
(iii) slag entrapment can be reduced by increasing the viscosity of the mould slag
(but at the expense of the powder consumption [159]).

9.4.4.2 Factors Affecting Viscosities of Mould Slags

The viscosity is very dependent upon the degree of polymerisation of the slag. In
fact, the viscosity has been used by some workers to represent the slag structure.
The viscosity increases with increasing concentrations of (i) network formers
(e.g. SiO2, Al2O3) (ii) decreasing concentrations of network breakers (e.g. CaO,
MgO, MnO and FeO) and (iii) fluxes (e.g. Na2O, K2O, Li2O, CaF2, B2O3).
It is customary to represent the temperature dependence of viscosity by using
either the Arrhenius or the Weymann relations (denoted as subscripts A and W in
Eqs. 9.17 and 9.18, respectively)

g ¼ AA exp ðBA =TK Þ ð9:17Þ

g=TK ¼ AW exp ðBW =TK Þ ð9:18Þ

where η is in dPas, TK is in K and AA and AW are the pre-exponential terms and BA

and BW represent the activation energy (E) term (B = E/8.314).The parameters,
BA and BW increase with increasing polymerisation (i.e. increasing Q).
It is customary to cite the viscosity at 1300 °C (1573 K) since this approximates
to the mean temperature in the liquid layer of the slag film.

9.4.4.3 Measurement Methods

Several methods have been used to measure slag viscosities, these are detailed
below. Some of the classical methods of measuring viscosity (e.g. capillary vis-
cometry) are rarely used because of the difficulty in obtaining a uniform temper-
ature zone of sufficient length at the required temperatures.
Concentric Cylinder Method
These viscometers consist of two concentric cylinders namely, the crucible (holding
the molten sample) and the bob (Fig. 9.23). When one of these cylinders is rotated
322 9 Properties of Mould Fluxes and Slag Films

Brookfield head

articulated wire
sliding seal

thermocouple upper furnace

tube seal
furnace
crucible
lifting
bob
mechanism

alumina
support tube
support
tube seal

Fig. 9.23 Schematic diagram showing the rotating bob (or cylinder) method for measuring
viscosity (courtesy of RF Brooks, NPL, Teddington, UK)

at constant speed, the viscosity can be determined from measurements of the torque
developed [3, 159, 160]. It is difficult to obtain the accurate alignment of the two
cylinders but the viscosity of mould fluxes is usually high enough to ensure
self-centring of the bob; it is for this reason (i.e. ease of alignment) that the rotating
bob (or cylinder) method (denoted RCyl) has become the more popular method.

Fig. 9.24 Schematic drawing

showing the oscillating plate
method [162, 163] (courtesy
of RF Brooks, NPL,
Teddington, UK)
9.4 Physical Properties of Mould Slags 323

The various sources of uncertainty in the rotating cylinder method have been
analysed by Wu et al. [161]. Ideally, the distance between the cylinders should be
small to ensure the stress gradient is linear.
Oscillating Plate Method
The oscillating plate method has been used to measure the viscosities of mould
fluxes [162, 163]. A plate is immersed in the melt and undergoes vertical oscilla-
tions in the fluid. When steady state has been established the amplitudes of the
oscillations are determined in both the fluid (u) and in the gaseous atmosphere
(uatm) (Fig. 9.24). Care must be taken to ensure “backwash” from the crucible walls
does not affect the results. The method gives values for the product (η
q) and so
values of the slag density (q) are needed to calculate the viscosity (η).

g
q ¼ Gfðuatm =uÞ  1g ð9:19Þ

where G is a cell constant which is determined in calibration experiments with

liquids of known viscosity.
Falling Ball Method
In this method, the time for a sphere to travel through a known distance in the slag
is measured; this can be done by either letting a ball fall or by dragging a wire
attached to the sphere vertically through the melt [3]. The principal problem lies in
the difficulty of obtaining a zone of uniform temperature of sufficient length. The
viscosity is calculated from the following equation where g is the gravitational
constant, rsph and vsph are the radius and velocity of the sphere, q is the viscosity
and subscript, fl refers to the liquid slag sample.


g ¼ 2 g rsph
2
qsph  qfl =vsph : ð9:20Þ

More Recent Methods

The draining crucible (DC) method [164] and the SLLS (Surface Laser Light
Scattering) methods have not been applied yet to slag systems but would appear to
be viable methods for these systems.
In the draining crucible technique, the sample is allowed to drain through an
orifice of known diameter in the base of the crucible and the rate of mass loss is
recorded. The viscosity, surface tension and density can be derived from the
hydrodynamic analysis of the data [164].
In the Surface laser light scattering (SLLS) method “ripplons” are monitored
[165–167]. The surface of a melt may appear smooth but it is being continually
deformed by thermal fluctuations of the molecules. Capillary waves (ripplons) have
small amplitudes (ca. 1 nm) and a wavelength of ca. 100 lm which is dependent
upon the frequency. Ripplon action relies upon surface tension for restoration and
the kinematic viscosity (m = η/q) for oscillation damping. The spectrum of the
324 9 Properties of Mould Fluxes and Slag Films

Fig. 9.25 The length of the

slag ribbon (L) as a function
of the fluidity (1/η) for
Inclined Plane Test; R1, R2,
R3, refer to reference materials
with known viscosity
(permission granted,
ISS/AIST, [168])

ripplons is derived using a Fourier spectrum analyser; this allows the surface ten-
sion and the viscosity to be determined, simultaneously. The method has been
successfully used for measurements on liquid silicon [166] and LiNbO3 up to
1480 °C. It has been estimated that it can be used for liquids with viscosities in the
range 0.005–10 dPas [165] which covers most of the mould flux range. However, a
value of the density is required to calculate the dynamic viscosity (η).
The advantages of these methods are that they also provide simultaneous values
for both viscosity and other properties (density and surface tension with DC method
and surface tension with SLLS method).
Inclined Plane Method
This simple test has been used for quality assurance tests on mould powder sup-
plies. In this test 10 g of de-carburised mould flux is weighed into a carbon crucible
and this is then placed in a muffle furnace at the required temperature for about
15 min to allow temperature equilibration [168]. It is then removed from the fur-
nace and poured onto a V-shaped inclined plane (usually an incline of 9° is
adopted) where it forms a slag ribbon. The length (L) of the solidified slag ribbon is
then measured and is related to (1/η). It is possible to obtain viscosities of rea-
sonable accuracy with this method by taking suitable precautions and using
improved procedures [168] (Fig. 9.25).

9.4.4.4 Viscosity Data for Mould Slags

There are a large number of investigations of viscosities of mould fluxes; for con-
venience, these have been divided into two tables covering conventional mould fluxes
(i.e. containing CaF2) in Table 9.4a and mould slags for specialist casting in
9.4 Physical Properties of Mould Slags 325

Table 9.4a Summary of viscosity measurements on mould fluxes

References Method Viscosity, Slag Comments
η1300 dPas type
Esaulov [17] OscV 0.5–100 Conv 10 mould slags
McCauley RCyl 1–14 Svnth 13 synthetic mould slags; effect of
[176] (C/S) = 0.8–1.2; Slags 1–6 similar
composition, η varies ±25% around mean:
B2O3 slag
Riboud RCyl 1–12 Synth 16 synthetic slags from CSNFl system
[176, 177]
Zhuk [178] RCyl Conv Effect of B2O3 addition
Lanyi [179] RCyl 0.3–10 Conv 18 slags; Effect of Al2O3 B2O3 additions;
Activation energies
Kim [147] RCyl 0.4–3 Conv 35 slags: Effect of F, Li2O, Na2O, B2O3 and
Al2O3 on η
Chang RCyl 3–20 Conv (32–37)C +(39–46)S + 12A + (1–5)N + (0–
[180, 181] 12)Fl
Mills RCyl 2–6 Conv 4+3slags; post-measurement composition
[132, 168]
Kawamoto O Plate 1 Conv Effect of Al2O3, SiO2, B2O3, MgO, BaO,
[182] CaO, K2O, CaF2, Na2O, Li2O on η
Lopez [183] RCyl 8.5–12.4 Conv Slags for billet casting
Elahipanah RCyl 1–3 Conv 9 mould slags;
[148]
Park [14, 184] RCyl Conv Effect of CaF2 on viscosity
Wang [154] 1–2 Conv Effect of RE oxides η# as %RE "; also B2O3;
Li2O added
Lu [95, 98, RCyl 1.5–4 Conv No temps given: 1300 °C? (C/S) = 0.8, 1 and
100, 102, 185, 1.2; Effect of Na3AlF6; B2O3; BaO; K2O;
186] NaF; Li2O on η of slag;
Persson [149] RCyl 0.7–4.3 Conv 7 slags + additions of Al2O3;
post-measurement slag composition
Marschall RCyl 0.6–1.1 Conv 2slags; Effect of Al2O3 and TiO2 on viscosity
[150]
Shahbazian RCyl Synth CaO + SiO2 + (13–40)CaF2
[187, 188]
In chemical formulae note: A Al2O3, Fl CaF2, etc. and 5M = 5% MgO; Mould powders are mostly
for conventional slag, Conv Conventional; Synth Synthetic; RCyl Rotating cylinder method; OscV
Oscillating Viscometer

Table 9.4b. The effects of individual components on the viscosities of various mould
slags has been studied by a number of workers: e.g. Al2O3; [5, 96, 168, 170]: TiO2 [30,
76, 82, 133, 168–172]; ZrO2 [96, 152] MnO [140, 147]. The effect of precipitates on
the slag viscosities have been reported after crystallisation [128, 173] and after TiC
precipitation [142]. The effect of applying an electric potential on the measured
viscosity has been studied and it was found that increasing DC voltage caused a
326 9 Properties of Mould Fluxes and Slag Films

Table 9.4b Viscosities for specialist mould powders: ηT is for 1300 °C unless stated otherwise
References Method ηT dPas Slag Comments
type
FF-flux
Fox [28, RCyl 1 FF*- Effect of B2O3 on η
101] flux
Wen [66] RCyl 2–9 FF-CT F-free powders; Effect of (C/S); TiO2; MnO;
Na2O; Li2O; MgO; B2O3 on η (36C, 38S,
6A, 11N, 9%F)
Wu [161] RCyl 2–6 FF-CT Effect of TiO2, Na2O and Al2O3 on η
Chen [102] RCyl 0.7–1 FF-CT CaF2 replaced by TiO2
Wang RCyl FF-flux Effect of B2O3 on η
[189]
Klug [190] RCyl 17 FF*- Effect of Na2O on η of slags for billet casting
flux
Wang RCyl 4–7 FF-CT Effect of B2O3; TiO2 on η
[172]
Zhang
[191]
NN
Tsutsumi RCyl 4 NN Non-Newtonian flux; effect of different shear
[192, 193] rates on η
HiAl
Li [6] RCyl η1500, low Effect of (A/S) on η: 25C + (20–45)S + (35–
¼2–3 (C/S) 10)A + 20Fl; (1350–1500 °C); ηT at
¼6–8 minimum at ca. (A/S) = 1.0
Zhang RCyl 3–14 low Effect of (A/S) on η
[168] (C/S)
Yu [170] RCyl 1–5 low Effect of (A/S) on η
(C/S)
Higo [5] RCyl η1500, low Effect of (A/S) on η
24–36 (C/S)
Hanao [69] Osc P 3–5 cf 2 Melilite Melilite slag cf conventional slag; Also used
in round billets
CA
Kim’ Park RCyl η1500 = 1–2 CA CA + (7–12S) +5 M + (0–15)Fl; (1500–
[7] 1600 °C); Qn values
ηT# as (C/S)" as (C/A)" and CaF2"
Li, Shu RCyl η1500 = 0.7– CA CA + 5M + 5L + 8N + 10Fl + 0 – 10T;
[26] 1 (1450–1700 °C) ηT # as %T"
Huang [29] RCyl η1673 = 3– CA Effect of B2O3 on η
27
Q Liu RCyl CA
[194]
SS-Ti st
(continued)
9.4 Physical Properties of Mould Slags 327

Table 9.4b (continued)

References Method ηT dPas Slag Comments
type
Kusano RCyl Effect of TiO2 on η
[195]
Wang [30] RCyl Conv 30C + (28–38)S +
6A + 1.5M + 10N + 14Fl + T (0–10)
Replace SiO2 by TiO2 causes# η
Zhang RCyl η1500 = 1.5 Conv Effect of TiC and TiN particles on η;
[196, 197] non-Newtonian
Mukongo RCyl 1.0 Conv Effect of TiO2 on η
[27, 76]
Kishi [133] RCyl 1–2.5 Conv TiO2 lowers—when (C/S) < 1
Chen [102] RCyl 0.7–1 CaF2 mostly replaced by TiO2
HSTS Vc up to 8 m min−1
Kromhout RCyl Viscosity fits with empirical rules for slag
[198] selection
Kawamoto Osc P
[199]
FF Fluoride free; NN non-Newtonian; fluxes for casting steels; Hi Al High-Al slag (3 types,
denoted as (i) low (C/S) (ii) Melilite and (iii) CA = calcium aluminate slag); HSTS High-speed thin
slabs; SS Stainless steel; *for casting billets: Osc P Oscillating plate method

decrease in viscosity [174, 175] due to the decrease in metal/slag contact angle which,
in turn increased “slip” between the slag and the rotor (and the crucible).

9.4.4.5 Methods to Calculate Viscosity

Several models have been reported to estimate viscosities of mould fluxes from
their chemical compositions and these are summarised in Table 9.5 where these
estimated values are, for the most part, subject to uncertainties of ca. ±25%. This
uncertainty is high because the uncertainty associated with the experimental vis-
cosities in the databases is also high (±10–25%).

9.4.5 Thermal Conductivities

9.4.5.1 Importance to the Process

The occurrence of both longitudinal cracking and sticker breakouts are associated
with the control of horizontal heat transfer; this involves both the thermal con-
ductivity and thickness of the slag film.
328 9 Properties of Mould Fluxes and Slag Films

Table 9.5 Summary of models used to calculate viscosities of mould slags from chemical
composition
Model/references Details of model Uncertainty
Riboud [200] AW; BW functions of 5 groups ±30%
Modified const “CaO” + “SiO2” + “Al2O3” + “CaF2” + “Na2O” where
[6, 29, 168, 170] X“CaO” contains XCaO +XMgO + XFeO + XMnO etc.; ln
A = −19.81 +1.73 XCaO +3.58 XCaF2 + 7.02 XNa2O – 35.76
XAl2 O3 B = 31140−23896 XCaO – 46356 XCaF2 − 39159 XNa2 O
−68833 XAl2 O3
For calculation of η for fluxes for High-Al steels-low(C/S) [6,
168] CA–type [6, 170] +FF [29]
Iida [34] η(Pas) = Aη0 exp (E/Bi) where η0 = viscosity of hypothetical ±25%
network-forming melt and Bi = basicity index E = 11.11 −
3.65  10−3 T and A = 1.745 − 1.962  10−3T + 7  10−7
T2 and Bi = R(ai %i)B/R(ai %I)A where A = acid oxides and
B = basic oxides or fluorides: η = 1.8  10−7 (M = i Tm)0.5
exp (Hi/RTm)/V0.667
m exp (Hi/RTm) where Hi = 5.1 Tm and
values of a given
Kim [147] ln A = −2.307 −0.046 XSiO2 −0.07 XCaO − 0.041 XMgO − ±45%
0.185 XAl2 O3 + 0.035 XCaF2 − 0.095 XB2 O3
B = 6807 + 70.7 XSiO2 + 32.58 XCaO + 312.7 XAl2 O3 − 34.8
XNa2 O − 176.1 XCaF2 −167.4 XLi2 O + 59.7 XB2 O3 where X in
mole %
Hanao [201] Neural network model; available commercially
Zhang [44–46] Calculates BOs, NBOs and FOs and distinguishes between
Al–O and Si–O bonds
Gupta [37] 3 groups “Acidic”: Yx = (%SiO2/60) + a(%Al2O3/102); ±35%
“Basic” YO = (%CaO/56) + (%MgO/40) + (% Na2O/63) +.
(%FeO/72. “Fluorides” YF = (%F/19): NF = YF/
(YX + YO + YF): NX = YX/(YX + YO + YF): NO = YO/
(YX + YO + YF); Na+/Ca2+affected η: Values of a0 to a5 given
for 1573 and 1673 K; η = a0 + a1(M+/M2+) + a2NX + a3Nx
(M+/M2) + a4(M+/M2)2 + a5N2X
Du, Seetha η = hNAq/M exp (DGη/RT) where h = Planck constant,
raman [202] NA = Avogadro No Structure accounted for—
Hayashi [203] thermodynamics:DG*η = RDG*η (oxides) + DGmix η + 3R*T
X1X2:DGmixη for interactions of cations only
Works well for synthetic slags (±15%)
Mills [33, 144, Two different models to calculate viscosity of wide
204] compositional range of slags
GH Zhang Calculates BOs, NBOs and FOs and distinguishes between
[47–49] Al–O and Si–O bonds
Mills [143] For cont. casting slags only; Values of XCaF2 and XTiO2
calculated; Qrem calculated for remaining slag. Note T in K:
ln η1573 K, ref = 0.1227 exp (1.1454 Qrem) − 9 XCaF2 ;
Bη = 795830.8 + 8.476  10−6exp (Q/0.2611) +795841.4
exp (Q/211587.4); ln ηT = ln η1573 + (B/T) – (B/1573)
(continued)
9.4 Physical Properties of Mould Slags 329

Table 9.5 (continued)

Model/references Details of model Uncertainty
Elahipanah [148, Reviews performances of several models
205]
Shu [205] Arrhenius equn; CMAS system; takes charge balance into ±20%
account
Subscripts: A Arrhenius; W Weymann equations

9.4.5.2 Factors Affecting Thermal Conductivities of Slag Films

Horizontal heat transfer from the shell to the mould involves several heat transfer
mechanisms, namely:
• Lattice (or phonon) thermal conduction.
• Radiation conduction.
• Convection in the liquid layer (aided by movement of the shell and mould
oscillation).
As mentioned above, the effective thermal conductivity of the slag film (keff) has
a significant effect on the horizontal heat flux (qhor) and keff is largely determined by
the physical state of the slag film. The magnitude of keff is affected by:
• The fraction of crystalline phase (fcrys) present in the slag film since (a) kcrys  2
kgl and (b) crystalline grains reflect IR radiation back towards the shell [56, 110]
and thereby reduces radiation conduction (kR) and (c) the shrinkage associated
with crystallisation results in the formation of both pores and an interfacial
resistance (RCu/sl). The overall effect is that (keff)crys < (keff)gl and for most slag
films it is thought the kR contribution to keff is less than 20% of the total [115].
• The presence of gas pores in the slag film reduces keff (see Sect. 9.3.3).
• The presence of transition metal oxides in the slag film (e.g. FeO, NiO) reduces
kR through the absorption of infrared radiation [53, 54, 206] but this must be
offset against a reduction in reflected heat transfer associated with the promotion
of more glass phase by FeO, MnO, etc.

9.4.5.3 Measurement Problems

There are two major problems encountered when measuring thermal conductivities
of slags, namely, how to deal with both convection and radiation conduction.
With convection, the usual approach taken is to determine a value of the thermal
conductivity of the liquid slag (free from convection) and then to calculate the
convective contribution using turbulence models (e.g. k-e model). The main diffi-
culty lies in determining the value of the thermal conductivity of liquid slag, free
330 9 Properties of Mould Fluxes and Slag Films

from convective contributions. This is usually achieved by using a transient method

in which the measurements are completed before convection has time to develop
(typically around 1 s). Steady state methods are unsuitable for these measurements.
The two transient techniques in common use are the Laser Pulse (LP) and the
Transient Hot Wire (THW, sometimes referred to as the line source) method.
Slags are semi-transparent media and thermal conductivity measurements on
both solid and liquid slags at high temperatures are known to contain contributions
from both lattice (klat) and radiation conduction (kR, described in Chap. 3,
Sect. 3.2.1) as shown in Eq. 9.21.

keff ¼ klat þ kR ð9:21Þ

Values of kR have been found to increase with increasing sample thickness, d,

[207] until a critical point is reached, beyond which, kR remains constant. Samples
below this critical point are described as optically thin and those above, as optically
thick. Optical thickness occurs when a * d  3, where a* is the absorption
coefficient and d is the sample thickness. Values of kR can be calculated for opti-
cally thick samples by use of the following equation where r = Stefan–Boltzman
constant, n = refractive index and TK is the temperature in K. Contributions from kR
increase dramatically with increasing temperature (since k is proportional to T3K)

kR ¼ 16
r
n2 TK3 =3a ð9:22Þ

In partially crystalline samples, IR radiation can be both absorbed and reflected and
it is not possible to replace a* by the Extinction coefficient (E) in Eq. 9.22 without
knowledge of the albedo (or reflectivity coefficient). There is little information on
the albedo but a method for calculating this parameter has been reported [208].

9.4.5.4 Measurement Methods

Laser Pulse (LP) Method [209–213]

Fig. 9.26 Laser pulse

method used to determine
thermal diffusivities of solids
a specimen assembly and
b typical temperature transient
(permission granted, Verlag
Stahleisen GmbH [3])
9.4 Physical Properties of Mould Slags 331

The laser pulse provides values for the thermal diffusivity (a); the thermal con-
ductivity (k) is derived from the thermal diffusivity by using the following relation,
where Cp and q are the heat capacity and the density of the slag, respectively.

keff ¼ aeff
Cp ; q ð9:23Þ

The sample is usually in the form of a disc (10–15 mm diam.  1–2 mm thick)
which may be coated on both planar surfaces. Experiments have been carried out in
two ways. In the first method (Fig. 9.26a), a laser beam is directed onto the front
face of the specimen and the temperature of the back-face is monitored continu-
ously. The thermal diffusivity is calculated using the relation, a = 1.37 d2/p2 t0.5

Fig. 9.27 Laser pulse method used to determine thermal diffusivities of liquid and solids
a specimen assembly and b typical temperature transient (permission granted, Verlag Stahleisen
GmbH [3])

Fig. 9.28 Transient hot wire method a schematic drawing of probe and b temperature transient
(permission granted, Verlag Stahleisen GmbH, [3])
332 9 Properties of Mould Fluxes and Slag Films

where t0.5 is the time required to reach 0.5DTmax (see Fig. 9.26b) and d is the
sample thickness. Corrections are usually made to account for radiation losses. This
method is used only for measurements on solid samples.
The second method can be used for solid and liquid samples (Fig. 9.27). The
sample is in good contact with a Pt disc (Fig. 9.27a) and the laser pulse is directed
onto the Pt disc and the temperature decay of the Pt disc is monitored continuously.
Numerical methods have been used to disentangle kR and kc contributions [214, 215].

Transient Hot Wire (THW) Method [54, 55, 61, 62, 216, 217].
In this method the sample is placed in a crucible and melted. Then a long, thin wire
(typically, 0.1 mm diam.) is immersed in the molten slag. The thin wire acts as both
a heating element and as a temperature sensor (Fig. 9.28a). When thermal equi-
librium is attained, a current is applied for a short duration (ca. 1 s). The temper-
ature of the wire is monitored continuously. The thermal conductivity (k) can be
calculated from the following equation where q = heat input per unit length of wire,
r = radius of wire, DT = temperature rise and C = exp c* where c* = Euler’s
constant.


DT ¼ ðq=4pkÞ ln at=r 2 C ð9:24Þ

However, k is usually determined from the slope (=1/k) of the plot of the tem-
perature rise (DT) as a function of ln time (Fig. 9.28b). The use of graphite com-
ponents in the THW method is not recommended since reactions between FeO,
MnO, etc. (in the mould flux) and graphite produce CO (g) bubbles. The bubbles
increase porosity in the sample which lowers the thermal conductivity [62].
The hot wire method can also be used to determine the effective thermal con-
ductivity of powders [61, 218, 219].

9.4.5.5 Thermal Conductivity Data for Mould Slags

The various measurements of thermal conductivity (k) and diffusivity (a) recorded
for mould slags and powders are summarised in Table 9.6; the entries are grouped
according to specimen type. The various factors affecting heat transfer have been
discussed above. Although klat for crystals is approximately twice that of for the
glass phase, the total thermal heat flux (qeff) is lower in the crystalline phase than in
the glass. This is due to the scattering of IR radiation by the crystals and the
subsequent reduction in kR. The kR contributions can also be reduced by introducing
transition metal oxides (e.g. FeO, NiO) into the glass which absorb the IR radiation
[206].
Most slag films contain both glassy and crystalline phases. It should be noted that
crystallisation results in some porosity in the slag film which will tend to reduce keff.
Table 9.6 Summary of thermal conductivity (k) and diffusivity (a) on mould fluxes
Specimen Reference Method Values (dk/dT) or Comments
{da/dt}
Liquid Bagha [220] Sim k = ca. 0.3 Analysis of heat flux data; Cu cold finger
Nagata [216] THW kT = ca. 0.2
Powell [221] THW kT = ca. 0.2
Shibata [211] LP 107a = (4–6.5) Several liquid slags
Cho [59]
Hayashi [222] LP 107a = (4.65)
Gornerup [223] LP 107a = (4)
Monaghan [224] LP 107a = (4) Equivalent to k  1.5 Wm−1K−1
9.4 Physical Properties of Mould Slags

Waseda [210] LP 107a = (4) {0} (1100–1300 °C) Studied effects of TiO2, ZrO2, HfO2 on kR contribution
Ohta [212, 225]
Andersson [61] THW kT = ca. 0.2 Powder sample heated into liquid phase
Kim [226] THW k1500 = 0.2–0.3 Liquid for CSB system
Slag film
Susa [53, 227] LP 107a = (6–8) 5 slag films
Andersson [62] THW k25 = 0.75–1.2 3 slag films from C-steel billet casting; Slag films exhibit increasing
porosity with crystallis’ n
Crystal
Shibata [211] LP 107a = (4–5) 3 part crystalline slags; 25–800 °C
El Gammal [228] RHF k = 1.2–1.7 Probably contains kR contributions
Hayashi [222] LP Effect of fcrys
Ozawa [54] THW k25 = 1.5–1.8 (2  10−4) k increased with increasing fcrys but (dk/dT) changed from positive to
negative
Andersson THW k25 = 1.2–1.8 Partially cryst. samples; keff, 25 = 1.07 + 0.7 fcrys:
[61, 62] k767 = 1.65; (dk/dT) calculated from k25 and k767
(continued)
333
Table 9.6 (continued)
334

Specimen Reference Method Values (dk/dT) or Comments

{da/dt}
Glass
Taylor [209] LP 107a = (4–5) {ca. 0} 10 different samples (200–900 °C) 107 a = 4.5 ± 0.5
Shibata [211] LP 107a = (4–5) 2 glassy mould slags (25–800 °C)
Hayashi [222] LP 107a = (4.6)
Ozawa [54] THW k25 = 1.1 ± 0.03
Andersson [61, THW k25 = 1.07 ± 0.03 (0.78  10−3) (25–767 °C); kgl T = 1.07 + 0.78  10−3(T–25 °C)
62] kT decreases sharply at 767 ± 10 °C
Bed
Powder Taylor [209] LP k200 = 0.17–0.35 (3  10−4) (200–700 °C) in vacuo k calculated from a values
Nagata [216] THW kT = 0.2–0.4 Goes through maximum at ca 1000 °C
Goto [219] THW kT = 0.12 4 MPs values very similar
Neumann [229] THW k25 = 0.12–0.22 (7  10−5*) * = powder; G = granules
(11  10−5)G
Andersson [61] THW k25 = 0.12–0.16 (9  10−5) k25 = 0.010 + 1.69  10−4 qbulk
Bratchikov [230] k100 = ca. 0.3* Compressed powder qbulk = 1555 kg m−3; 10–1000 °C
Ichihashi [231]
Powder Macho [218] THW k25 = 0.06–0.1 (94  10−5)
bed tests
Diehl [232] Hot-plate k25 = 1*–1.4** Probably contains contributions from gas convection
Ichihashi [231] Sim k20 = 0.5 Tcrit = 1030 °C
Sinter Taylor [209] LP k increases as qbulk# Suggests kconv contribution
Andersson [61] THW 0.4 For powder sample which sintered at high temp.
Note k values in Wm−1K−1; 107 a values are denoted by parenthesis e.g. (4) = 4  10−7 m2s−1; Sim Simulation experiments; THW Transient Hot Wire, LP
Lase Pulse; *powder; G Granules
9 Properties of Mould Fluxes and Slag Films
9.4 Physical Properties of Mould Slags 335

Fig. 9.29 The thermal 2

conductivity of glassy, mould
slags as a function of 1.5
temperature; o, D = [61];

k, Wm-1K-1
solid line = Eq. 9.25; dashed
line = values above Tcrit [61]; 1
●▲ = Ozawa [54]; □,
■ = k values calculated from 0.5
107a = 4.5 and 4 m2s−1,
respectively, for glass and
liquid; vertical dashed 0
0 200 400 600 800 1000 1200 1400
line = Tcrit
Temperature, ToC

9.4.5.6 Glassy Slags

It can be seen from Table 9.6 and Fig. 9.29 that thermal conductivity values have
been measured using the THW method for glassy, mould slags [54, 61, 62].
Measurements have also been reported for a powder sample heated into the liquid
which then formed a glass on cooling [216]. The thermal conductivities at 25 °C for
glassy mould slags were found to lie between 1.04 and 1.1 Wm−1K−1 [54, 61, 62].
When a glass is heated, it undergoes a transition from a frozen glass to a
supercooled liquid at the glass transition temperature (Tg); Tg occurs at ca. 600 °C
for most mould slags. However, the slag crystallises at a temperature about 50 °C
above Tg (see Fig. 9.9); this results in increases in both the fraction of crystalline
phase (fcrys) and in kT.
Andersson [61, 62] found that kT for all glassy samples increased with increasing
temperature up to a critical point (Tcrit or the deformation temperature) beyond
which, there was a marked decrease in kT with increasing temperature (Fig. 9.29).

5.5

5
107.α, m2s-1

4.5

3.5
0 200 400 600 800 1000 1200
T ,oC

Fig. 9.30 The thermal diffusivity (107.a) as function of temperature for glass and liquid mould
slags; o = mean of 3 samples studied by Shibata [211]; dashed lines = high and low values of 10
slags [211]; dotted line = sample THA [211] showing sharp drop near Tcrit; += Values obtained by
TPS method [61]; x, ◊, □ = [222]; Liquid ▲ = Ohta [212, 225] values; ■ = Gonerup [223].;
vertical dotted line = Tcrit
336 9 Properties of Mould Fluxes and Slag Films

Thus the temperature dependence of kgl for glassy mould slags can be described by
Eq. 9.25. Values of Tcrit were found to occur at 767 ± 10 °C for mould slags [61]
and measured values for kgl at Tcrit had values of 1.65 ± 0.05 Wm−1K−1
(Fig. 9.29). It should be noted that Tcrit is ca. 50–100 °C above the point where
crystallisation is usually initiated but on occasion, crystallisation can occur below
Tg (for instance see Fig. 9.9).


ð25  767  CÞ k gl Wm1 K1 ¼ 1:07 þ 0:78  103 ðT  25  CÞ ð9:25Þ

Measurements of absorption coefficients of glassy mould slags indicate that a

2-mm thick, initially glassy slag film would be optically thin.
Thermal diffusivity values have also been reported for glassy mould slags by a
number of investigators, most of which were obtained with the LP method [209,
211, 222]; these values are shown in Fig. 9.30. The results show that the thermal
diffusivity of all the glassy mould slags were found to lie between 4 and 5  10−7
m2s−1 and values were virtually independent of temperature (Fig. 9.30). These
values of aLP were converted into kLP values using estimated Cp and density values
and are compared with the kTHW values in Fig. 9.30. It can be seen that kTHW
values are in good agreement with kLP values for temperatures up to Tcrit but for
T > Tcrit, kTHW values decrease markedly with temperature, whereas kLP values,
shown as squares, increase slightly with temperature (Figs. 9.29 and 9.30). Similar
behaviour has been identified [204] with kLP and kTHW measurements for CaO–
Na2O–SiO2 slags of similar composition [233, 234]. Three propositions have been
put forward to account for the divergence:
1. kLP values contain significant contributions from kR since the surface area (A) of
the heat source is 10 greater in the LP (cf. THW) experiments
(ALP = 10ATHW).
2. kTHW values are low because of electrical leakage from the hot wire into the
melt.
3. Crystallisation results in shrinkage of the slag and the formation of an air gap at
the slag/heater interface.

Fig. 9.31 Thermal

conductivity of mould slags 1.6
as a function of temperature;
Glass: denoted D and faint
k, W m -1 K -1

1.2
line [61]; Partially crystalline;
4 samples denoted by x, +, 0.8
bold line, dotted lines [61];
□ = values for a melted 0.4
powder on cooling [216];
vertical dotted line = Tcrit 0
0 200 400 600 800 1000 1200
T, oC
9.4 Physical Properties of Mould Slags 337

At this point of time, the problem is unresolved and no values can be recom-
mended for k for temperatures above 760 °C. However, it has been suggested that
the rapid decrease in kTHW above Tcrit is due to the loss in rigidity of the silicate
network between Tsoft and Tflow [50] and that Tcrit corresponds with the temperature
where the viscosity had a value of 106 dPas [204] i.e. Tcrit is equivalent to the
deformation temperature of the slag.

9.4.5.7 Slag Films and Partially Crystalline Samples

It was noted that when glass samples were heated to high temperatures that the
values recorded on the cooling cycle were significantly higher than those measured
for the glass in the heating cycle, as shown in Fig. 9.31 [61, 62]. This enhancement
of k values is due to the crystallisation of the sample at higher temperatures and
these high k values are maintained, or increased, during subsequent measurement
cycles. Thus the thermal conductivity increases with increasing crystallisation.
Thermal conductivity and diffusivity values for crystalline samples of mould
fluxes are significantly higher than those for glassy samples [61, 209, 211]; this is in
line with measurements on other slags where kcrys  2kglass [60]. The effect of the
fraction crystalline phase (fcrys) on the thermal conductivity can be clearly seen in
Fig. 9.32a, b [54, 55, 61]. Equation 9.26 was obtained from the data given in
Fig. 9.32b.

k25 ðWm1 K1 Þ ¼ 1:07 þ 0:7 fcrys : ð9:26Þ

Fig. 9.32 Thermal (a) 1.8

conductivity of mould slag
a as function of fraction 1.6
k,W m-1K-1

crystalline phase (fcrys)

expressed as % and b k25 as a 1.4
function of fraction crystalline
1.2
phase; ● [61]; ▲, D [54, 55]
for fluxes 1 and 3, 1
respectively; (permission 0 10 20 30
granted, Taylor and Francis 100. fcrys, %
(a) re-drawn after [54]
(b) [61])
(b) 2
k295, Wm-1K-1

1.5

0.5
0 0.2 0.4 0.6 0.8 1
fcryst
338 9 Properties of Mould Fluxes and Slag Films

(a)
Thermal diffusivity,10-7m2s-1 (b)

7
8

107.α, m2 s-1
7

6
5
5

3 3
0 200 400 600 800 1000 0 200 400 600 800 1000 1200
Temperature, ToC T, oC

Fig. 9.33 Comparison of thermal diffusivity values for slag films, partially crystalline samples
and liquid; a for partially crystalline (●▲♦) and glassy (o) mould slags with C/S values of
1.07 = ●; 1.16 = ▲ and 1.29 = ♦ [211] b slag films = various curves [53]; ● = partially
crystalline (C/S = 1.29) and o = glass of same slag; liquid, ▲ = Ohta [212, 225]; ♦ = Gonerup
[223] (b permission granted, Taylor and Francis, [61])

Since values of k25 > 1.8 Wm−1K−1 have been recorded and the partially
crystallised samples contain some porosity, Eq. 9.26 may provide an underestimate
of k25 for fully—crystalline samples.
It can be seen from Fig. 9.32 that the kTHW values for various partially crys-
talline mould slags (like the glass phase) attain a value of kTHW = 1.65 ± 0.05
Wm−1K−1 at Tcrit [61]. It is interesting to note that kTHW values for partially
crystalline samples show the same rapid decrease for temperatures above Tcrit that
the glassy phase samples display. This may be due to the rapid deterioration in kgl
for T > Tcrit for the glassy phase.
The temperature coefficient (dk/dT) for temperatures between 25 and 767 °C for
partially crystalline samples can be positive or negative depending on whether
k25 < kTcrit or k25 > kTcrit, respectively.
Thermal diffusivity values for partially crystalline samples and slag films are
shown in Fig. 9.33. It can be seen that (i) the values for the slag films have similar
values to those of the partially crystalline samples (ii) the values are significantly
higher than those of the glass and (iii) there is some variation in values, presumably
due to differences in fcrys. It is also interesting to note the aLP value recorded for one
slag film shows a drop near Tcrit.
It has been calculated that in partially crystalline samples, kR contributions
constitute <20% of the effective thermal conductivity (keff). This suggests that keff
values derived on slag films with a significant crystalline fraction are unlikely to
contain large contributions from kR. The kTHW values for partially crystalline
samples indicate a decrease in k for T > 767 °C; this could indicate that fcrys is not
high enough to provide sufficient rigidity in the sample. However, it is difficult to
make recommendations for k values for T > 600 °C until the discrepancy between
kLP and kTHW is resolved.
9.4 Physical Properties of Mould Slags 339

9.4.5.8 Powder Bed

The effective thermal conductivity (keff) of mould powders and the powder bed are
affected by a number of factors:
(i) The porosity of the sample with keff increasing as the porosity decreases.
(ii) The magnitude of keff increases as the number of points of contact between
particles (Ncontact) increases (Note small particles have a much greater
number of points of contact than large particles per unit mass, thus keff
increases with decreasing mean particle size (Dmean). It should also be noted
that compression of the powder increases Ncontact and thus keff increases with
increasing bulk density (qbulk).
(iii) Contributions to keff from gaseous convection (kgasconv) with kgasconv
increasing as the permeability (P) of the powder bed increases (Note
P increases as Dmean increases and thus kgasconv increases as Dmean increases).
(iv) Exothermic or endothermic reactions (e.g. carbonate decomposition, sinter-
ing, carbon combustion) will affect the local temperature gradients in the bed
and hence the local heat flux.
(v) The thermal conductivity of the powders increases with increasing temper-
ature; the mean temperature coefficient (dk/dT) is +9  10−5 Wm−1K−2 [61].
It should be noted that kgasconv will be affected by the nature of the experiment; it
will be large for experiments where there is a large temperature gradient across the
sample and where the sample is heated from below. Furthermore, decomposition of
carbonates, the oxidation of carbon particles and argon injection will all lead to
enhanced gaseous convection in the bed.
Powders have a smaller diameter and higher bulk density than granules and
consequently, have lower porosity and more points of contact, so kpowd > kgran. The
following relation has been reported [61] and the results are shown in Fig. 9.34.


k25 Wm1 K1 ¼ 0:010 þ 1:69  104 qbulk ð9:27Þ

Fig. 9.34 Effective thermal 0.2

conductivity of mould 0.18
-1
k, Wm K

powders as a function of bulk 0.16

-1

density [61, 62] obtained 0.14

using the THW method 0.12
0.1
0.08
500 700 900
Bulk density,kgm-3,
340 9 Properties of Mould Fluxes and Slag Films

Fig. 9.35 Thermal 0.8

conductivity of powders; solid
bold line; dotted line, +,

k, Wm-1K-1
X = Andersson [62]; ♦ = k
THW due to Macho [218]; 0.4
dashed line = Nagata [216]
values; □, D, ○ = values due
to Taylor [209]; ▲,
■ = Neumann [229] granules
and powder, respectively; 0
0 200 400 600 800 1000 1200
(permission granted, Taylor
and Francis, [61]) T,oC

The results reported for the thermal conductivity of powders are shown in
Fig. 9.35 and Table 9.6. Inspection of these results shows that measurements of keff
fall into three bands:
(i) keff = ca 0.12 Wm−1K−1 with the values reported by Macho [218] Neumann
[229] and Anderson [61] in good agreement.
(ii) keff = 0.1–0.35 Wm−1K−1 and 0.3–0.5 Wm−1K−1 at 200 and 800 °C,
respectively, [209, 216]; the higher keff values probably resulted from higher
bulk densities.
(iii) keff = ca 1–1.4 Wm−1K−1 [232] derived in thermal insulation tests where the
enhanced keff values probably arise from large kgasconv contributions.
The divergence in these values (0.1 to >1.0 Wm−1K−1) raises the question as to
which value to use. When the gaseous convection contribution is calculated inde-
pendently, the value of kpowd  ca 0.12 Wm−1K−1 would be appropriate. However,
the value keff = ca 1–1.4 Wm−1K−1 obtained in thermal insulation tests contains
contributions from both kpowd and kgasconv and should be used (cautiously) when
kgasconv is not calculated independently. It should be noted that these findings
indicate that kgasconv  10 kpowd in these experiments which aim to simulate the
heat transfer in the powder bed.
Values have also been measured for the sinter portion of the bed. A value of
k = ca. 0.4 Wm−1K−1 was derived for a powder sample cooled after heating to high
temperature; it was suggested that this may constitute a representative value for the
sinter layer [61]. Much higher values of aLP (ca 7  10−7 m2s−1 for a fully dense
sample) and kLP (1–2 Wm−1K−1) were derived [209] but these were for dense
(compressed) de-carburised powder samples.
9.4 Physical Properties of Mould Slags 341

Table 9.7 Summary of published models to calculate the thermal conductivity of slag films and
liquid mould fluxes
Model/Reference Details of model
Mills [144] Calculates, ln k at 25 °C and Tg; ln k = −a + b  10−5exp (Q/c) + dXLi2 O :
ln k25{ln kTg}; a = 0.424{0.435}, b = 2 {5}, c = 0.299{0.332}; d = 3.2
{3.0}
Mills [204] Accepts kTHW data: k (Wm−1K−1) = a +b exp (Q/c) + d exp (Q/e) + f(X
(z/r2)); k25:{kTcrit} a = 7.94 {−0.85}; b = 7.22 {0.987}; c = 199.3 {1.006};
d = 0.109 {−0.8}, e = 3.634 {0.965}. f = 0.61 {1.62}
Fluegel [235] Glass samples-only covers low levels of F. Numerical fit of thermal
conductivity data for glasses at ambient temperatures
Mills [143] Accepts kTHW data Glass; k25 = 1.07 Wm−1K−1. Partially crys: Calculate
fcrys from composition. k25 = 1.07 + 0.7 fcrys. k767 = 1.65; kTliq = 0.0278
exp (1.0242Q)
crys crystalline

9.4.5.9 Liquid Slags

There is a large discrepancy in the values measured with the THW and LP methods
with kLP  1.5 Wm−1K−1 and kTHW  0.2 Wm−1K−1 [216, 221] and it is difficult
to recommend a value at this stage.

9.4.5.10 Calculation of Thermal Conductivity

Few models have been reported to calculate the thermal conductivity of mould
fluxes as can be seen from Table 9.7.

9.4.6 Interfacial Tension (cmsl) and Surface Tension (cs)

9.4.6.1 Importance of Interfacial Tension to the Process

The interfacial tension, cmsl, (between the metal and slag) affects the following:
• Slag entrapment with entrapment decreasing as cmsl increases [146, 236].
• The shape of the steel meniscus is affected by by cmsl.
• It has also been proposed that the adhesion of scale and scum to the steel surface
is dependent on interfacial tension [237].
Values of the surface tension of the slag (csl) are needed to calculate the inter-
facial tension (cmsl) as can be seen from Eq. 9.28.
342 9 Properties of Mould Fluxes and Slag Films

9.4.6.2 Factors Affecting Surface and Interfacial Tensions

of Mould Slags

The interfacial tension (cmsl) is given by the Good–Girifalco equation [238] where
cm and csl are the surface tensions of the metal and slag phases, respectively, and u
is an interaction coefficient.
cmsl ¼ cm þ csl  2uðcm csl Þ0:5 ð9:28Þ

The parameter, u, was found to have a value of 0.5 for slags free of FeO but
increased with FeO additions [239]. It was proposed that the following equation
could be applied

Fig. 9.36 The effect of S on (a) 2000

a surface tension (cm) of steels
Surface tension, mNm-1

(re-drawn from [241]) and

b interfacial tension;
(permission granted, re-drawn 1800
(a) IOM/Taylor and Francis
[241] (b) from ISS/AIST
[146]) 1600

1400
0 100 200 300
S content, ppm

(b) 1200
Interfacial tension, mNm-1

1000

800

600
0 0.2 0.4 0.6 0.8
AcƟvity S
9.4 Physical Properties of Mould Slags 343

Fig. 9.37 Transient sessile drop profiles during metal/slag reactions a at start of experiment b,
c during the experiment and d at end of the experiment, when slag/metal reactions are complete
(permission granted, IOM/Taylor and Francis) [244]

/ ¼ 0:5 þ 0:3XFeO ð9:29Þ

Alternatively, Chung [238, 240] suggested that the interaction coefficient (/)
could be calculated from values of the Gibbs energies (DGi) of formation the
various liquid components of the mould slag.
The surface tensions of both steels and slags are dependent upon the concen-
trations of surfactants present. Surfactants tend to be materials with low surface
tension and the surface layer of a liquid contains a high concentration of surfactants.
That is why ppm levels of surfactants can have a significant effect on surface tension.
In steels, the principal surfactants are soluble sulphur and oxygen. In contrast, the
principal surfactants in slags are B2O3, K2O and Na2O and CaF2; however, their
effect on csl is much less dramatic than the effect of S and O on cm of iron.
Surface tensions of the steel (cm) tend to have values between 1500 and
1850 mNm−1, whereas the surface tension (csl) of most slags have values in the
range, 300–450 mNm−1; thus cm  4 csl. Consequently, the most important term in
Eq. 9.28 is cm and this is very dependent upon the soluble S- and O-contents of the
steel [241] (Fig. 9.36a). In most steels, the Al content is sufficiently high to hold the
soluble O content down to ca. 4 ppm but Al has little effect on the soluble

Fig. 9.38 Transient measurements of a Interfacial tension and b mass transfer rate of Al (via
Eqs. 9.1 and 9.2) (permission granted, Taylor and Francis [243])
344 9 Properties of Mould Fluxes and Slag Films

S content. Thus, cm is largely determined by the S content of the steel and it can be
seen from Fig. 9.36a that 50 ppm S causes a 25% decrease in cm. However, FeO in
the mould slag and metal/slag reactions will increase the soluble O in the steel and it
is probably the latter which causes the apparent increase in the interaction coeffi-
cient (/) (Eq. 9.29) and a decrease in cmsl.
It has been reported [242–245] that reactions between slag and metal (e.g.
S-transfer or Al transfer, Eqs. 9.1 and 9.2), the vigorous mass transfer (Figs. 9.37
and 9.38) apparently results in a dramatic reduction in the interfacial tension
(Fig. 9.38a, b [243]). The interfacial tension remains low while there is vigorous
mass transfer but increases sharply when most of the slag/metal reaction is complete
(Fig. 9.38). It has been suggested (on the basis of observations on organic systems)
that this effect is related to the large differences in volume of the slag and metal
(Vsl > Vm) and that there would be a smooth change in interfacial tension if the
metal and slag phases had similar volumes [246] i.e. Vsl/Vm 1.
The interfacial tension has been reported to increase as (i) Al2O3 increases [245]
and (ii) the contents of Na2O, CaF2, FeO and MnO decreased in the mould flux
[245]. The interfacial tension is usually increased by reducing the Na2O content of
the flux [237].
Measured values of cmsl tend to be in the range 1100–1400 mNm−1 and values
of the surface tensions of mould fluxes tend to be in the range 300–450 mNm−1.

9.4.6.3 Measurement Methods

Surface tension (csl)

There are several methods available to measure the surface tensions of molten
mould fluxes, these are detailed below.

Fig. 9.39 Schematic drawings showing a sessile drop and b big drop (BD) methods (permission
granted, Verlag Stahleisen GmbH, [3])
9.4 Physical Properties of Mould Slags 345

Maximum bubble pressure (MBP) method

In this method a gas/slag interface (i.e. a gas bubble) is formed at the tip of a metal
capillary by gradually increasing the gas pressure. A transducer records the gas
pressure continuously. The maximum bubble pressure (MBP or PMBP) corresponds
to the point where the bubble attains a hemispherical profile. The surface tension of
the slag is calculated from the Laplace equation where h is the depth of immersion
of the capillary in the slag and q is the density of the slag. It is customary to carry
out measurements at various depths to derive the density of the slag [3]. It is also
customary to chamfer the capillary to produce a knife-edge tip and to use a slow
rate of bubble evolution [3].

PMBP ¼ 2=r þ qgh ð9:30Þ

Sessile drop (SD) and Big drop (BD) methods

The shape adopted by a sessile drop (Fig. 9.39a) represents the balance of surface
and gravitational forces. These forces involve the surface tension and the density,
respectively, and both properties can be derived by this method. A molten slag
droplet is sited on a metallic plaque and the dimensions of the drop are measured
accurately; several methods have been used to derive the surface tension and
density from the dimensions [3]. In recent years, the accuracy of surface tension and
density values has been improved by using software which calculates “best fit”
values for the parameters affecting the drop profile.
The Big Drop (BD) method is a variant of the sessile drop technique (Fig. 9.39b)
where the slag is held in a crucible and the surface profile of the drop is determined.
Pendent drop (PD) and Drop weight (DW) method
A pendent drop also represents the balance of surface and gravitational forces [3].
Frequently, when this method is applied at high temperatures, the sample is in rod
form and the tip of the rod is heated by electron beam or laser heating. Thus surface
tension values can only be determined for the liquidus temperature. Software
similar to that used in the sessile drop method is used to provide a “best fit” of the
measured drop profile.
The pendent drop method is often used in combination with the drop weight
method; since the pendent drop eventually falls and forms a spherical drop which
can be collected. The solidified drops are weighed and the mass (Wideal) determined.
The surface tension is calculated from the following equation where r = the radius
of the tube or rod.

Wideal ¼ 2pcr ð9:31Þ

346 9 Properties of Mould Fluxes and Slag Films

However, the measured weight (Wmeas) of the drops is less than Wideal and so a
correction term (fc) is usually applied and fc is determined in preliminary experi-
ments using liquids of known surface tension [3, 247].

Wideal ¼ 2pcrfc ð9:32Þ

Detachment methods
In this technique, a tube (or rod, plate or ring) is located just below the surface of
the liquid. The probe is attached to a balance and then the mass is measured
continuously as the probe is slowly pulled from the melt [3]. The force acting on the
probe is measured; this force is the resultant of surface tension and mass. This force
goes through a maximum at the point where the liquid film at the meniscus is about
to break and thus, the measured weight also exhibits a maximum (Wmax). The
surface tension is calculated by Eq. 9.33 where r = radius of tube or probe and G is
a correction factor

c ¼ ðWmax gÞG=4pr ð9:33Þ

Other methods
As mentioned above (Sect. 9.4.4.3) the draining crucible method is also capable of
measuring the surface tension. A crucible has a small orifice machined in its base
and the orifice is blocked with a stopper rod. Slag is placed in the crucible, melted
and allowed to drain and the mass flow rate is monitored. The surface tension is
derived from a hydrodynamic analysis of the data [164].
In the Surface laser light scattering (SLLS) method “ripplons” are monitored
[165–167]. Ripplons are Capillary waves caused by thermal fluctuations of the
surface. Ripplon action is initiated by surface tension and is damped by the kine-
matic viscosity. A Fourier spectrum analyser is used to determine the spectrum of
the ripplons and the surface tension is derived from the data.

Fig. 9.40 Schematic drawing Ar gas

showing the X-ray sessile
drop method [244]
Monitor

X-ray Detector
source

Mould slag
Steel
9.4 Physical Properties of Mould Slags 347

Table 9.8 Summary of surface and interfacial tension studies carried out on mould fluxes
Property Reference Method c Values Comments
mNm−1
Surface Yavoisky 310–320
tension, csl [248]
Yakushev 200–320
[249]
Krusina 360–380 Slag:40C + 45S +10A + 5N + (0–15)
[250] Fl replacing S or C
Elfsburg SD 430–550 Continual gas evolution from drop
[251]
Kusano Det* 250–320 Det* = Detachmant method
[252]
Lu [185] RD 320–340 No temps given; (C/S) = 0.8; 1 and 1.2;
3–15% Na3Al F6 added (dc/d% Na3Al
F6) = −2.5
Lu [102] RD 305–355 No temps given; (C/S) = 0.8; 1 and 1.2;
2–10%B2O3 added (dc/d%B2O3) = −6.5
Lu [100] RD 320–370 No temps given; (C/S) = 0.8; 1 and 1.2;
1–5%BaO added (dc/d% BaO) = −4
Lu [98] RD 320–355 No temps given; (C/S) = 0.8; 1 and 1.2;
1–5% K2O added (dc/d% K2O) = −5
Lu [186] RD 300–340 No temps given; (C/S) = 0.8; 1 1.2; 3–
15%NaF added (dc/d% NaF) = −2.5
Lu [95] RD 300–340 No temps given; (C/S) = 0.8; 1 and 1.2;
2–10%Li2O added (dc/d% Li2O) = −3
Monaghan MBP 320 (1200–1400 °C); dc/dT = ca. 0
[224]
Cheng RD c1400 = 400 c1400 = 400 + 560 (C/S−1.0);
[253] c1400 = 415 + 8(%Al2O3);
c1400 = 400 – 15%F; c1400 = 420−10%
Na2O
Duberstein MBP c1400 = 370 (dc1400/d%Na2O) = −6 mNm−1 %−1
[254] −310
Konovalov MBP c1550 = 300 (C/S) = 0.5−2.28
[255] −350
Gonerup XRSD 320−500
[223]
Contact Feldbauer h1300 °C = 55 Mould slag on LC steel: h1300 °C = 55°
angle (h) [146] h1300 °C = 75 Slag on steel:
Nakato slag on solid Fe: h1300 °C: Conv h = 75°
[237] NNslag; h1300 °C = 65°
Tsutsumi
[192, 193]
Interfacial Gornerup XRSD A- 1150
tension cms [223] B- 910
(continued)
348 9 Properties of Mould Fluxes and Slag Films

Table 9.8 (continued)

Property Reference Method c Values Comments
mNm−1
2 mould slags:
A = 37C + 29S + 7A + 7 N
+6Fl + 0.8F
Slag B – 29C + 33S + 5ª + 11N + 9Fl
+1.2F
Elfsberg XRSD 800–1400 4 steels; 4 mould slags;
[251]
Cramb XRSD 1000–1350 cmsl decreases#- as soluble O and S in
[245] steel"; -as slag- FeO"; as Al2O3#, as
Feldbauer Na2O"; as CaF2#
[146]
El Gammal 1400–800 Effect of TiO2 and S on cmsl at 1550 °C
[227] of 40C + 40S + 20A
Konovalov 1000–1100 Sessile drop values for steel and MBP
[255] for slag
Hagemann Drop 820–1060 Low (C/S) slag to cast highAl steel: 2
[256] wt steels with 100 and 500 ppm S: "Al2O3
and MgO ! cms": "TiO2 ! cms#
c and cmsl are mNm−1 units.; RD Ring detachment method; Drop wt Drop weight method; Det*
Detachment

9.4.6.4 Methods to Measure Interfacial Tension (cmsl)

Two methods have been used to measure metal/slag interfacial tension.

X-ray sessile drop method
A cylinder of steel is placed in a MgO crucible and then covered with de-carburised
mould flux and is then heated in a tube furnace under an inert atmosphere [3]. When
both slag and metal have melted, a sessile drop of steel in molten mould flux is
formed and observed by passing X-rays through the crucible (Fig. 9.40 [244]). The
slag is transparent to X-rays whereas the steel is opaque and thus the image of the
sessile drop can be observed. The contact angle and the dimensions of the drop are
measured and the interfacial tension is measured using software packages based on
the Laplace equation. Most of the published data have been obtained with this
method.
Lens method
In this method mould slag is placed on the meniscus of molten steel. The slag melts
and forms a lens-shaped drop [3]. The dimensions and contact angle are observed
and measured.
9.4 Physical Properties of Mould Slags 349

Table 9.9 Summary of models available to calculate the surface tension of mould slags and
interfacial tension
Model/references Details of model Uncertainties
Surface tension
Mills [257] Two different components; (i) Regular, e.g. CaO, SiO2, ±10%
Al2O3 and (ii) Surfactants B2O3, K2O, CaF2, Na2O, Li2O;
Regular: c = R X1 c1 + X2 c2 + X3 c3 +; Surfactants- 2
regimes (a) Xi < 0.12;.c2i = a + bX + cX2 and
(b) Xi > 0.12; by X2 c2 = a′ +b′X; where, i = surfactant
species
Tends to over-emphasise effect of surfactant
Mills [143] Above model modified; (1) surface active element are
considered as regular components once surface active
concentration (Xsurf. > 0.12) (2) dc/dT = R(dc/dT)i for slag
components
Nakamoto [258] Thermodynamic model involving cation ionic radii; ternary
systems C + A + Fl; C + F+B2O3; and C + S+N
Hanao [259] Based on Tanaka model [239] using molar volumes of ±17.5%
components oxides. Extended to
0.4C + 0.4S + 0.2Fl + additions of Al2O3 + MgO + Na2O
Nakamoto [260] Neural network model applied to ternary systems;
Interfacial
tension
Chung [238, Cites values for DG: of constituents: DGDi . = = DGFeO- R Xi
240] DGi u = 0.89 − 1.5  10−3DGD i
Tanaka [239] cmsl = cm + csl + 2 / (cm. csl)0.5; / = 0.5 + 0.3XFeO
Mills [143, 261] csl; calculated as above; cm = 1880−0.41(T−1530 °C)
−0.109 ln (1 + e{(194111/T)−4.6849} x %S
cmsl = cm + csl + 2 (cm. csl)0.5;

9.4.6.5 Surface and Interfacial Tension Data

The studies of the surface and interfacial tension involving mould slags are sum-
marised in Table 9.8.

9.4.6.6 Methods to Calculate Surface (csl) and Interfacial Tension (cms)

A few models have been reported to calculate the interfacial tension; these are
summarised in Table 9.9.
350 9 Properties of Mould Fluxes and Slag Films

(a) (b)
8000 80
Length change,ppm

Thermal expansion
70
6000 60
50
4000 40
30
20
2000 10
0
0 -10
0 200 400 600 0 200 400 600 800 1000
Temperature T/oC Temperature (C)

Fig. 9.41 Change in length as a function of temperature (oC) for a a glassy mould slag; dashed
vertical line = Tg showing a sharp decrease at deformation temperature; b From top at 700 °C; slag
film (gray) crystalline (bold) and glass (faint line) of same mould slag; (permission granted, Taylor
and Francis, [124])

9.4.7 Density (q) and Thermal Expansion Coefficient (a)

9.4.7.1 Importance of Density to the Process

The density of the mould slag is one of the factors affecting the rate of slag flow into
the mould/strand channel (i.e. powder consumption). Density data are needed for
the mathematical modelling of powder consumption.

9.4.7.2 Factors Affecting Density and Thermal Expansion

The molar volume (V) is only slightly affected by the polymerisation of the slag and
this can be accommodated by expressing VSiO2 (and VAl2 O3 ) as polynomials of
(XSiO2 VSiO2 ) where X is the mole fraction. The oxides with higher molecular weight
(M) tend to have lower molar volumes (or higher densities) e.g.
VK2 O \ VNa2 O \ VLi2 O .

Vq ¼ M ð9:34Þ

The thermal expansion coefficient can be cited as either as the linear thermal
expansion (a) or the volume thermal expansion (b) which are defined in Eqs. 9.35
and 9.36, respectively, where DT = (T – Tref) and Tref = reference temperature (e.g.
20 °C) a and b values pertain to the mean temperature, i.e. 0.5 (T + Tref).

a ¼ ðLT  Lref Þ=Lref ðT  Tref Þ ð9:35Þ

b ¼ ðVT  Vref Þ=Vref ðT  Tref Þ ð9:36Þ

9.4 Physical Properties of Mould Slags 351

V ¼ V0 ð1 þ aDT Þ3 ð9:37Þ

The principal factors affecting the thermal expansion are (i) the degree of
polymerisation (i.e. Q with a and b decreasing as Q increases, (a # as Q")) and
(ii) the bond strength of the M–O bonds with a decreasing as field strength (z/r2) of
the cation increases, where z = cation charge (e.g. 1 for Na+ and 2 for Ca2+).
The thermal expansion coefficient of a glass undergoes a threefold increase for
T > Tg when it transforms from a glass to a supercooled liquid (scl); this can be
seen as an abrupt change of slope for the lowest curve in Fig. 9.41b. The sample
collapses at the deformation temperature (which is ca. 50 °C above Tg) and no
measurements are possible above this temperature. Crystalline samples do not
undergo these changes but, instead, exhibit a sharp volume increase on fusion
between Tsol and Tliq. Thus, the densities in the range between Tg and Tsol vary
according to the degree of crystallisation developed in the sample (Fig. 9.8b).

9.4.7.3 Measurement Methods

Solid slags
Density
For solid slags at room temperature the density is usually measured by the
Archimedian method. The slag sample, of known weight and volume, is suspended
on a wire from a balance and the weight of the sample is determined in air (Watm)
and then in water (Wfluid) (or a more dense liquid, e.g. bromoform or acetylene
tetrabromide, q25 = 2887 and 2953 kgm−3, respectively). The density is then cal-
culated from Eq. 9.38.

q ¼ ðWatm  Wfluid Þ=V ð9:38Þ

Densities at room temperature can also be determined by the Sink/Float method

in which the sample either floats or sinks in standard liquids (like bromoform or
acetylene tetrabromide) of known density.
Thermal expansion—Dilatometry
Thermal expansions of slags are usually measured by machining two parallel faces
on the sample (usually ca. 30 mm long) and measuring the changes in length as the
specimen is heated at a known heating rate. For glassy samples, the specimen
collapses at the “softening point” which is ca. 50–100 °C above the glass transition
temperature (Tg) and no further measurements can be made.
Liquid slags
The following methods are frequently used to measure the density of molten mould
slags and fluxes; the experimental uncertainty is ca. 2–5%.
352 9 Properties of Mould Fluxes and Slag Films

Archimedian (or buoyancy) method

A bob of known volume is suspended from a balance. The weight, before
immersion (Watm) and then after full immersion (Wfluid) in the molten slag, is
measured. It is necessary to correct the density value for the effect of the surface
tension forces acting on the wire [3].
Pyknometry (or Weighing) method
The slag is melted in a specially designed vessel of known volume and the mass of
the slag is determined. Corrections must be made for the thermal expansion of the
vessel.
Maximum bubble pressure (MBP) method [3, 254]
The MBP method has been described above (Sect. 9.4.6.3). The MBP (pMBP) is
determined at different depths of immersion in the molten slag (h1, h2, etc.) and the
density is calculated by Eq. 9.39 or from the gradient of the plot of pMBP as a
function of h [3, 254].

q ¼ ðpMBP2  pMBP1 Þ=ðh2  h1 Þ ð9:39Þ

Sessile drop method

The volume of a sessile drop of known mass is determined through measurements
of the drop dimensions [3] Recent improvements in software to determine the
profile of the drop have improved the accuracy of this method.
Draining crucible
This technique has not been used to date, on molten slags but would appear to be
viable for the measurement of mould slag densities. The rate of drainage of slag
from an orifice in the crucible is determined by continuous weighing of the crucible.
The density is derived through hydrodynamic analysis of the measurements [164].

9.4.7.4 Density Data for Mould Slags

The reported density data for solid and liquid mould slags are given in Table 9.10
and for the thermal expansion coefficient in Table 9.11. For liquid mould fluxes, the
density values lie mostly in the range q = (2500–2700) kgm−3. The reported
density and thermal expansion coefficient measurements are shown in Fig. 9.8b.
There are few measurements of density and thermal expansion. This is quite sur-
prising, given that these measurements are relatively easy to perform and are
subject to low levels of experimental uncertainty.
It should be noted that Olivares et al. [262] annealed glass samples at 600 °C and
found that the density values of annealed samples were lower than those for
Table 9.10 Density values reported for mould fluxes
Specimen Reference Method Values (dq/dT) Comments
Bed Taylor [209] V&m q25: 600–1000
powder Andersson V&m qbulk = 700–1100 qbulk increases with decreasing particle diameter
[61, 62]
Glass
Taylor [209] Arch q25 = 2750 ± 50
Olivares [262] SF q25 = 2820–2920 2 mould slags + syn slags 39−46C + 38−48S + 6−16A + 6−16N + 4−16 Fl;
25 °C; Annealed 600 °C
Slag film
9.4 Physical Properties of Mould Slags

Susa [53, 227] Arch q25 = 2730– 5 samples

2940 ± 50
Liquid
Monaghan MBP q1300 = 2580 ± 50 –0.2 (1200–1450 °C)
[224]
Matsushita LD-ESL q1300 = 2580 −0.298 Supercooled (1150–900 °C)
[263]
Matsushita SD q1200 = 2850–2700 Seems high; Gas evolution- reaction with C plaque?
[263]
Quested [264] MBP q1300 = 2650 ± 100
Konovalov MBP q1550 = 2200–2400 (C/S) = 0.5–2.28
[255]
Duberstein MBP q1400 = 2740–2700 (dq1400/d%Na2O) = −8
[254]
V & m Volume and mass; Arch Archimedian; SF Sink of float; LD Levitated drop; SD Sessile drop
353
354 9 Properties of Mould Fluxes and Slag Films

Table 9.11 Experimental values reported for the thermal expansion of mould fluxes
Specimen Reference Method 107 a; Comments
K−1
(25–
500 °C)
Glass
Mills Dilatom 9–13 13 glassy mould slags
[132]
Courtney Dilatom 6.5 Glassy mould slag
[124]
Slag film
Courtney Dilatom 8.5 and {(dL/dT)T<Tg/(dL/dT)
[124] 11 T>Tg} = 27/12 = 2.3
Part-crystalline
Mills Dilatom a is ca. 10% higher than that for
[132] glass
Courtney Dilatom 8.5 1 crystallised sample annealed
[124] 15 h at 900 °C
Dilatom. Dilatometry

un-annealed samples; this may be due to densification but could also be due to
crystallisation of the sample. It should also be noted that crystallisation results in
some porosity and thus slag films will have density values below that of the
theoretical density.

9.4.7.5 Calculation of Densities and Thermal Expansion Coefficients

of Mould Slags

The models reported for the calculation of density and thermal expansion from
chemical composition are given in Table 9.12.

9.4.8 Heat Capacity (Cp) and Enthalpy (HT–H298)

9.4.8.1 Importance of Cp and Enthalpy to the Process

Heat capacity data are needed for carrying out heat balance calculations in the
mould.
9.4 Physical Properties of Mould Slags 355

Table 9.12 Outlines of models to calculate density and thermal expansion coefficient of mould
slags
Model/references Details of model Uncertainties
Density
Mills [143, 257] Partial molar approach, V1773 = R X1V1 + X2V2 + X3V3 + ±2%
… at 1500 °C, V (10−6m3mol−1) values for: CaO = 20.7;
FeO = 15.8; Fe2O3 = 38.4: MnO = 15.6; MgO = 16.1;
Na2O = 33; K2O = 51.8: TiO2 = 24; P2O5 = 65.7;
SiO2 = (19.55 + 7.97 XSiO2 ) Al2O3 = (28.3 + 32XAl2 O3 −
2
31.45XAl 2 O3
)
(dV/dT) = + 0.01%K−1
Olivares [262] Unannealed : q25 (kgm−3) = 2930 +4.1%C − 4.37%S −
2.99%A − 4.75%N + 29.9%Fl
Annealed: q25 (kgm−3) = 2930 +3.7%C − 3.4%S − 2.2%A
− 3.5%N + 29%Fl
Bottinga [257, Partial molar deals with slags with high Al2O3 contents
265]
Persson [266, Relates q to thermodynamic enthalpy of mixtures; Applied
267] to binary systems and then to ternary and higher systems.
CaF2 not included.
Fluegel [268, Numerical analysis of glass q25 database; only low ±0.5–3%
269] concentrations of Fluorine covered
Fluegel [270] Numerical analysis of qT database for liquid phase glasses,
1200–1400 °C; no F % included
Priven [271] Based on MDL SciGlass database for glasses-does not
include F%.
Tokuda [163] Data given to calculate V and (dV/dT) and hence qT
Thermal expans.
coefficient
Mills 15 [143] Solid 106a (20 − Tg) = 10; Liquid; (1/a′) = 0.25 Q + 0.4
(z/r2) where a′ = 106 a
Fluegel [269] Numerical analysis of TEC data at 210 °C for glasses; no
CaF2 included.
Fluegel [270] Numerical analysis of TEC database for liquid phase
glasses, 1200−1400 °C; no F % included
Stebbins [272] Partial molar approach: (dV/dT) for liquid
Units q: kgm−3: a: K−1

9.4.8.2 Factors Affecting Cp and Thermal Enthalpy

The Cp of glassy samples of mould fluxes exhibit a “step-like” increase at Tg which

is associated with the transformation of a frozen glass to a supercooled liquid (scl);
this can be seen in the dashed line in Fig. 9.42a [53]. The twin valleys in the
apparent Cp values in Fig. 9.42b indicate that further crystallisation of the sample
occurs above 500 °C in this sample. The decrease in Cp is not a true effect but is
due to the enthalpy of crystallisation (DHcrys) which is exothermic but which
356 9 Properties of Mould Fluxes and Slag Films

Fig. 9.42 Apparent heat capacity as a function of temperature for slag film and glassy specimen
derived from slag film; ──slag film; -
-
-
- = glassy specimen; - - - - = extrapolation for slag film;
● = estimate a slag 4 showing step-increase in Cp of, glass at Tg; and b slag 3 showing an apparent
decrease in Cp resulting from crystallisation (permission granted, Taylor and Francis, [53])

manifests itself as an apparent decrease in Cp; Values for DHcrys can be derived by
integrating the area under the curve.
These effects result in different Cp–T curves for glasses and crystalline speci-
mens in the region between Tg and Tsol and the actual Cp and enthalpy values will
fall between the two curves and will depend upon the fraction of the slag which has
crystallised (see Fig. 9.8a).

9.4.8.3 Measurement Methods

Values of Cp and enthalpy can be determined using a variety of calorimeters but

many of the traditional, calorimetric methods have been replaced by more rapid
techniques, especially for measurements at high temperatures. However, in practice,
the principal methods employed are differential scanning calorimetry and drop
calorimetry [273], or a combination of both. Experimental uncertainty is about
±2% with these techniques.
Differential Scanning Calorimetry [132, 274]
The principles underlying DSC are similar to those given above (in Fig. 9.19 [132,
274]). The sample (in the form of a disc or loose powder) is placed in one crucible
and the other crucible is kept empty. The instrument is then heated at a set heating
rate. The sample requires more energy than the empty crucible and thus the tem-
perature of the sample crucible lags behind that of the empty crucible. This tem-
perature difference is monitored. Three sets of measurement runs are made with the
sample crucible (i) empty (ii) filled with a known mass of sample (msample) and
(iii) with a sapphire (Al2 O3 ) disc of known mass (mAl2 O3 ). The Cp sample can be
calculated from the difference in signals (Ssample′–Sempty′) and for (SAl2 O3 ′–Sempty′)
9.4 Physical Properties of Mould Slags 357

Table 9.13 Details of investigations to measure Cp and enthalpies of mould fluxes

Specimen Reference Method T oC Cp Values Comments
JK−1 kg−1
Glass Mills DPSC 25 (800–880) 10 samples (25–700 °C)
[132] Tg {1050–1150}
Susa [53] DPSC Tg {1100–1200} 5 samples made from slag films
Mills DSC Tg {1250} 1 sample (500−720°C)
[227]
Courtney DSC Tg {1100} 1 sample (450–720 °C)
[124]
Slag film Susa [53] DSC 25 (800–850) ca. 5 slag films
Tg {1100}
Courtney DSC Tg {1100} Cp(scl) = ca. 1.5 JK−1 kg−1
[124]
Mills DSC Tg {1250} 1 slag film (500–720 °C);
[125] Cp(scl) = ca. 1.5 JK−1 kg−1
Liquid No data available
() = Cp25 °C. {} = CpTg

and msample and mAl2 O3 [274]. This type of DSC is usually denoted as differential
temperature scanning calorimetry (DTSC). However, there is a second type of DSC
in which the power required to keep the two crucibles at the same temperature is
monitored (differential power scanning calorimetry, DPSC) [273].
Drop calorimetry [273]
The sample, of known weight, is placed in a Pt crucible and heated to the required
temperature and held there until the temperature has equilibrated. The gate sepa-
rating the furnace from the calorimeter is then opened and the sample is dropped
into a massive, silver (or copper) calorimeter held at constant temperature. The
temperature rise (<5 °C) is then measured accurately and converted into enthalpy
using the mass and heat capacitance of the calorimeter. A series of blank experi-
ments are carried out with the empty crucible and the enthalpy of the blank for the
measuring temperature is then subtracted from that for the (crucible+sample).
Values of the Cp of the sample at 25 °C are needed to differentiate the enthalpy
values to obtain Cp values; the most accurate data are obtained by combining the
drop-calorimetry data with Cp values obtained by DSC. Rapid quenching produces
a glassy specimen which is in a metastable state; corrections are needed to correct
the measured values, so they pertain to the reference (crystalline) state.

9.4.8.4 Cp and Enthalpy Data for Mould Slags

Values of the Cp have been reported for mould slags at temperatures between (25–
720 °C); these are given in Table 9.13 and details of the investigations are given in
Table 9.14. The Cp values for a range of mould slags show little variation; Cp
358 9 Properties of Mould Fluxes and Slag Films

Table 9.14 Details of published models to calculate Cp and enthalpies (HT–H25) of mould fluxes
Model/reference Details of model Uncertainties
Partial molar model Cp = a′ + b′ T + c′/T2. where, a′, b′ and c′ are ±2%
mills [33, 143, 144] constants and are calculated by
a′ = R (XCaO a′CaO) + (XSiO2 a′ SiO2 ) + (XAl2 O3 a′
Al2 O3 ) + (XCaF2 a′ CaF2 ) +
Similarly with b′ and c′.…
Cp of liquid by
Cp liq = R (XCaO Cp liqCaO) + (XSiO2
CpliqSiO2 ) + (XAl2 O3 Cp liq Al2 O3 ) + (XCaF2
Cp liq CaF2 ) + ….
(HT–H298) = T298 R Cp dT = a(T–298) + 0.5bT2 − 0.5b
(298)2 + (c/T) − (c/298)
Enthalpy of fusion (DHfus) calculated from entropy of
fusion (DSfus) where T is in K
DSfus = R(XCaO DSfus CaO) + (XSiO2 DSfus
….
SiO2 ) + (XAl2 O3 DS Al2 O3 ) + (XCaF2 DS
fus fus
CaF2 ) +
DH = TliqDs
fus fus

Stebbins [272] Partial molar approach to Cp and DSfus; Based on

mineral and glass database; no CaF2.
Thermodynamic Commercial thermodynamic models, such as
models [275–277] Factsage; MTDATA and Thermocalc, provide
accurate thermodynamic data for slag systems

values at 25 °C and Tg (ca. 600 °C) are typically Cp 25 °C = 840 ± 40 and

CpTg = 1150 ± 50 JK−1 kg−1. The Cp–T relationships are complex for tempera-
tures above 550 °C since (i) the sample transforms to a supercooled liquid (scl) at
Tg which usually occurs at 600 ± 30 °C and (ii) crystallisation occurs, usually at
about 50 °C above Tg but which can occur as low as 550 °C for some slag com-
positions. Crystallisation is accompanied by an apparent sharp decrease in Cp which
is not a true Cp value since the enthalpy of crystallisation manifests itself as an
apparent decrease in Cp; the enthalpy of crystallisation can be obtained by inte-
grating the area below the Cp–T baseline. No values were found for the liquid
phase. However, estimated values for the solid phase were found to be in good
agreement with the measured values for the solid phase and it is known that Kopp’s
Law provides good estimates for Cp of slags, in general.

9.4.8.5 Calculation of Cp and Enthalpy

The models reported for estimating heat capacity (Cp) and enthalpy (HT–H25) are
given in Table 9.14. The predicted Cp values are compared against experimental
values in Fig. 9.42.
9.4 Physical Properties of Mould Slags 359

Table 9.15 Summary of the investigations of optical properties for mould fluxes and slag films
Property/reference Phase Method Values Comments
Refractive index Glass n
(n)
Susa [53] Abbe R 1.57–
1.60
Ozawa [55] Abbe R 1.55– 3 slags; 25 °C; k = 633 nm; n# with
1.57 "% CaF2 and (dn/dT) is negative
Liquid
Firoz [278] 42C + 42S + 6A + 5N + (0–12.5%
CaF2) (1127–1400 °C) (dn/dT) = ca.
−0.0008; n# as % CaF2"
Slag film
Susa [206] Abbe R 1.57–
1.61
Susa [53] Abbe R 1.57– 5 slag films; k = 0.546 lm
1.60
Holzhauser [279] 1.60 Slag film used in simulation experiment
Yamauchi [282]
Absorption coeff. Glass a*
(a*)
Mills [132] Spect 100– 7 glassy slags; a*at 25 °C;
500
Susa [206] Spect 550– 2 slag films; (0.5–5 lm) Effect of
1800 various oxides; Da* (m−1) = k%; k
(FeO) = 910; k(NiO) = 410; k
(MnO) = 5;
Ohta [212] Spect 200– At 250C; k = (2–5 lm) Effect of TiO2,
Waseda [210] 2000 ZriO2 HfiO2, FeO
Diao [280] Effect on a*; of FeO; MnO: TiO2 [280];
Chen [281] Co2O3 [281]
Shibata [211] Spect 500– 7 slag samples, (25–580 °C); k = (2–
Cho [59] 1000 5 lm); a* increases as Temperature
increases
Ozawa [55] Spect 200– 3 slag samples; (25 °C); k = (0–
500 2500 lm)
Extinction coefff Partially E
(E) crystalline
Susa [53] Spect. 2000– 5 slag films and glasses; Effect of Fe2O3
15000 on E & a*
Susa [206] 2000–
5000
Susa [56] k = (0–2.5 lm) A′ unaffected by D crys,
but A′# with D crys" for D crys > 2 lm
Shibata [211] Spect 15000– 7 slags; 25–580 °C; k = (2–5 lm); E i"
35000 as T"
(continued)
360 9 Properties of Mould Fluxes and Slag Films

Table 9.15 (continued)

Property/reference Phase Method Values Comments
Ozawa [55] Spect 20000– 3slags: k = (0–2500 lm); annealed at
40000 varius temps
Reflectivity (R′) Partially R′
crystalline
Susa [56] Spect-IS k = (0–2.5 lm) Effect of crystal diam
(Dcrys); T′# as Dcrys"; Dcrys = 1.2–2.2l
m
Susa [110] Spect-IS k = (0–2500 lm); effect of 0–2%
Fe2O3 and fcrys on R′
R’" as fcrys" and Fe2O3#; Effect" with
k"
Transmissivity (T Partially T′
′) crystalline
Susa [56] Spect-IS k = (0–2.5 lm) Effect of crystal diam
(Dcrys) T′# as Dcrys"; Dcrys = 1.2–
2.2 lm
Susa [110] Spect-IS k = (0-2500 lm); effect of 0-2% Fe2O3
and fcrys on T’
T′# as fcrys" and Fe2O3#; Effect" with
k"
Emissivity (e) Liquid e
Keene in [132] cfBB 0.89 1 slag (1280–1450 °C)
Solid
Keene in [132] cfBB 0.91 2 slags (1080–1260 °C)
Seko [283] Spect-IS 0.38– Crystalline slag
0.60
k wavelength; Abbe R Abbe refractometer; cfBB Comparison with black body; Elip Elipsometry;
IS Integrating sphere; Spect Spectrophotometer

9.5 Optical Properties of Mould Slags

Radiation conduction is a major contributor to the heat flux at high temperatures.

The radiation conductivity, kR can be calculated for optical thick conditions
(a*d > 3, where d = sample thickness, a* = absorption coefficient (m−1),
r = Stefan–Boltzmann coefficient and n = refractive index and TK is in K).

kR ¼ 16 r n2 TK3 = 3a  ð9:40Þ

Values for the various optical constants have been determined by various
investigators and details of these studies are given in Table 9.15.
9.5 Optical Properties of Mould Slags 361

Fig. 9.43 Values of a Absorption coefficient and b Extinction coefficient at 25 °C and 500 °C,
respectively, of various mould slags as function of wavelength; LC Low C; MC Medium C and
ULC Ultra low C steels (permission granted, ISIJ [59])

9.5.1 Refractive Indices (n) [53, 55, 206, 278, 279]

Values of the refractive index have been reported for the glass, slag film and the
liquid phases of mould slags (Table 9.15). The reported values range from 1.55 to
1.61 and a mean value, n = 1.58 ± 0.03 is recommended.

9.5.2 Absorption Coefficients (a*) [53, 55, 56, 59, 110, 206,
211, 212, 280, 281]

The magnitude of kR can be reduced in slag films by introducing either transition

metal oxides to absorb, or crystals to scatter, IR radiation. Values of the absorption
coefficient, a*, of glasses and glassy mould slags are significantly increased by the
presence of transition metal oxides in the slag (a* (m−1) = a0 + K%MO where
a0 = absorption coefficient with 0% transition metal oxide (MO) where K has
values of 910, 5, 410 m−1 for FeO, MnO and NiO, respectively, and KCr2 O3 > KFeO
[53, 206]. The effects of other transition metal oxides on a* for mould slags has also
been studied, FeO, MnO and TiO2 [212] and Co2O3 [281].
The absorption coefficient pertains to liquid and glassy phases but for crystalline
or partially crystalline materials it is necessary to use the extinction coefficient,
E (=a* + s, where s = scattering coefficient). It is not possible to use Eq. 9.40 for
partially crystalline materials since the scattering mechanism is different to that for
absorption. In order to use Eq. 9.40 for crystalline slags it is necessary to have
values for the albedo (or reflection coefficient) which is used to link scattering to
absorption. Unfortunately, we have little data available for the albedo.
362 9 Properties of Mould Fluxes and Slag Films

When IR radiation strikes a partially crystalline slag film, only a fraction of the
energy is transmitted, since energy is also absorbed and reflected; Eq. 9.41 applies
where A′, R′ and T′ are absorptivity, reflectivity and transmissivity, respectively.

A0 þ R0 þ T 0 ¼ 1 ð9:41Þ

From the Lambert–Beer Law, the transmissivity is given by the following

equation, where d = sample thickness.

T 0 ¼ expfða þ sÞd g ð9:42Þ

Susa et al. [56, 110] measured A′, R′ and T′ for two mould slags and reported
that:
• Transmissivity decreased and reflectivity increased as the fraction of crystalline
phase (fcrys) increased [110] (R′ "and T′# as fcrys ").
• The addition of 1–2% FeOx causes A′ to increase but R′ to decrease [110] due to
the promotion of the glassy phase by FeOx.
• R′ increases and T′ decreases with increasing grain size [56] (R′ "and T′# as
Dcrys "); a grain size of 2–3 µm is recommended for the lowest qhor [56].
• For casting MC, peritectic steels, the use of a mould flux free of FeOx is
recommended since it leads to a lower qhor [56].
A summary of investigations of optical properties for mould fluxes and slag
films is given in Table 9.15.
Typical absorption coefficients and extinction coefficients are shown in
Fig. 9.43a, b, respectively, for mould slags used in the casting of different steel
grades [53, 211]; it can be seen that the Extinction coefficients:
(i) Are considerably higher than the absorption coefficients.
(ii) Are particularly high for the mould slags used for casting medium carbon
(MC) steels which tend to have a high basicity and which results in high
values of fcrys.
(iii) Calculations using the absorption coefficients and assuming a slag film of
2 mm, indicate that the initial glassy slag film would be optically thin [284]
i.e. the kR value would be lower than that calculated from Eq. 9.40.

9.5.3 Reflectivity, Transmissivity and Emissivity

The published investigations for these properties are summarised in Table 9.16. It
can be seen that the reflectivity (R′) increases with:
(i) increasing fraction crystalline (fcrys) and increasing crystal diameter (Dcrys)
(ii) decreasing FeO content, since FeO promotes glass formation which tends to
offset any gains from FeO increasing the absorption coefficient.
9.5 Optical Properties of Mould Slags 363

Table 9.16 Models and routines to calculate optical properties

Property/reference Details of model Uncert
Ref index, n
Susa [56] {(n−1)/10−3 q} = 0.001 k%; k values; CaO = 2.25; MgO = 2.0; ±1%
K2O = 1.89; Na2O = 1.81; FeO = 1.87; MnO = 2.24;
Cr2O3 = 0.22; Al2O3 = 2.14; SiO2 = 2.07; TiO2 = 3.97
Priven [271] Model for glasses
Matsushita [2, Review of models for calculating optical properties
287]
Abs, coeff. a*
Susa [56] Da* (m−1) = k %; k values: FeO = 910; NiO = 410; MnO = 5;
Da* = 390(%Cr2O3) + 370(%Cr2O3)2

Increased reflectivity reduces the Transmissivity.

Several models, or routines, have been reported to estimate the optical properties
of mould slags; these are summarised in Table 9.16.

9.6 Thermomechanical Properties of Mould Slags

Slag rims and slag films can crack [198] and they must be repaired or replaced by
the infiltration and solidification of molten slag. Slag films tend to crack in the
lower half of the mould when liquid lubrication is low. The replacement process
can take up to 20 min because of the low flow rates of slag infiltration. Slag films
taken from the mould were found to show signs of fracture “along a line of bub-
bles” in the crystalline regions of the slag film [198]. Since crystallisation is
accompanied by the creation of pores, it seems reasonable to assume that the “line
of bubbles” was formed during crystallisation. Glassy slag films are less likely to
fracture than crystalline slag films.

9.6.1 Thermomechanical Tests

The thermomechanical properties of molten slag have been measured in a test,

shown in Fig. 9.44, where the mould powder is sandwiched between two crucibles
and the displacement is measured as the upper crucible is lowered [285]. On
heating, the rod was displaced gradually with increasing temperature; the dis-
placement was observed to increase more rapidly when the temperature exceeded
the deformation temperature (or softening point) temperature and increased sharply
when T > Tliq.
364 9 Properties of Mould Fluxes and Slag Films

Fig. 9.44 Schematic diagram

of the thermo-mechanical
analyser (Permission granted,
ISIJ [285])

A finite-element model has been developed to predict (a) shear stress (b) friction
(c) slip and (d) fracture of slag film [286]; the crystalline phase was predicted to
fracture in the meniscus region and the glass phase near the mould exit [286].

9.6.2 Stress Relaxation

The molten slag in the mouth of the infiltration channels is subjected to com-
pression as a result of the ferrostatic pressure. This could affect the lubricating
properties. Consequently, the stress relaxation of four mould slags were investigated
[287] in which a sessile drop of liquid mould slag was compressed between two Pt
plates, in an Instron machine, to 50% compression and the force measured. The
compression was then terminated and the slag allowed to undergo stress relaxation.
The stress attenuated until a constant stress value was attained; the time taken to
reach this point, the relaxation time, was measured. It was found that the relaxation
time increased with increasing polymerisation (or viscosity) of the slag [287].

9.7 Dissolution of Oxides, Nitrides and Carbides in Mould

Slags

Non-metallic inclusions (such as Al2O3 or TiO2) have a detrimental effect on the

high-temperature, mechanical properties of steels. This loss in mechanical strength
increases with increasing concentration and size of the inclusions. Thus, steel-
makers go to considerable length to minimise inclusion concentrations; the
continuous-casting mould provides one last chance to remove inclusions.
9.7 Dissolution of Oxides, Nitrides and Carbides in Mould Slags 365

9.7.1 Origin of Inclusions

There are two types of inclusions found in steel:

• Oxides, sulphides, nitrides and carbides formed in steelmaking and from
metal/slag reactions.
• Entrapped mould slag resulting from metal flow turbulence (Sect. 11.11).
The oxides and nitrides formed in steelmaking occur through reactions such as:

2Al þ 3O ¼ Al2 O3 ðsÞ ð9:43Þ

Ti þ N ¼ TiNðsÞ ð9:44Þ

Mn þ S ¼ MnSðsÞ ð9:45Þ

where the underline indicates it is dissolved in the steel.

Oxides can also be formed by reactions between metal and slag (in the slag
pool), e.g.

4Al þ 3 SiO2sl ¼ 3Si þ 2 Al2 O3sl ð9:46Þ

where the subscript sl indicates that it is in the slag phase. The above reaction is not
restricted to SiO2, since similar reactions occur with FeO, MnO, NiO, CrO, B2O3 etc.
Ehrenberg [288] proposed the following reaction:

3CaOsl þ 3S þ 2Al ¼ 3CaS(s) þ Al2 O3 ðsÞ ð9:47Þ

Typically, the pick-up of Al2O3 is 4%. It can be seen from Fig. 9.45 that the
pick-up arises from both steelmaking reactions and metal/slag reactions [289].
However, much higher Al2O3 pick-up occurs (up to 35%) when casting High-Al
(Trip and Twip) steels containing ca. 1% Al. It can be seen that the Al2O3 pick- up
due to slag/metal reactions is variable; this is due to variations in (i) the Al content

Fig. 9.45 Alumina pick-up 5

(mass%) resulting from
absorption of steelmaking 4
reactions (light grey) and
metal/slag reactions (dark) for 3
four mould powders
2
(permission granted,
ISS/AIST re-drawn after 1
[289])
0
A B C D
366 9 Properties of Mould Fluxes and Slag Films

of the steel and (ii) the concentrations of the more easily reduced oxides in the
mould slag (e.g. FeO, MnO, SiO2).
Kiyose et al. [290] developed a mathematical model which calculated the
compositional changes associated with metal/slag reactions (Fig. 9.46) for steels
containing Ti which reacts with SiO2 in slag; they claim that these changes can
result in the formation of cracks.

9.7.2 Mechanism of Inclusion Removal

For an inclusion to be successfully dissolved by the slag phase, several stages are
involved:
1. Transport of inclusion to the slag/metal interface.
2. It must satisfy the interfacial requirements for passage into the slag phase.
3. Dissolve in the slag phase.
4. It must be transported away from dissolution site.

9.7.3 Transport of Inclusions to the Slag/Metal Interface

Most inclusions (e.g. Al2O3) have lower densities than molten steel and so can be
removed by flotation. The velocity of the inclusion, (VI) as it moves towards the
slag pool, is given by Stokes Law (Eq. 9.48) where r = inclusion radius,
g = gravitational constant, q = density and η = viscosity (in Pas) and subscripts,
m and I denote the metal and inclusions, respectively.

VI ¼ 2 rI2 ðqI  qm Þ=9gm : ð9:48Þ

Fig. 9.46 a, b Predicted and measured changes in mould slag composition over time [290] due to
reactions between Ti-containing steel and mould slag and c showing the effect of basicity on the
the magnitude of TiO2 pick-up (Permission granted, ISIJ [290])
9.7 Dissolution of Oxides, Nitrides and Carbides in Mould Slags 367

Fig. 9.47 Schematic

diagrams showing metal flow
in the mould a double-roll
and b single-roll patterns;
(permission granted, UNESID
[291])

However, as we have seen in Sect. 5.3, the metal flow in the mould can be very
vigorous and can move inclusions towards, or away, from the metal/slag interface.
Furthermore, the flow pattern of metal flow is important; it can be seen that the
“single roll” will initially tend to bring the inclusions towards the metal/slag
interface before sweeping down the mould whereas the “double roll” (favoured in
many continuous-casting operations) will tend to drag some of the inclusions away
from the interface (Fig. 9.47).
The transport of inclusions to the interface is promoted by Argon bubbling and
by the use of electromagnetic devices in the mould.

9.7.3.1 Argon Bubbling

Argon stirring in the mould is used to remove inclusions and to minimise clogging
in the SEN. The attachment of inclusions to a bubble involves the interfacial
properties of inclusion, metal and gas (denoted by subscripts I, m and G, respect-
fully). The following equation is derived from a balance of forces (Fig. 9.48) where
h is the contact angle.

cIm þ cmG cos h  cIG ¼ 0 ð9:49Þ

Good flotation with the bubble occurs if the flotation coefficient (D, defined in
Eq. 9.50) is positive and has a high value. Thus good inclusion removal occurs
when (i) h has a high value (i.e. is non-wetting) and (ii) the surface tension of the

(a) (b)

Fig. 9.48 Schematic diagram showing the balance of interfacial forces for a a wetting system and
b a non-wetting system
368 9 Properties of Mould Fluxes and Slag Films

metal (cmG) has a high value, which corresponds to steels with low sulphur
contents.

D ¼ cIm þ cmG  cIG ð9:50Þ

The contact angles (h) for steel on most inclusions tend to be high (e.g. Al2O3,
h = ca. 130°; TiO2, h = ca. 80°; ZrO2, h = ca. 120°; TiN, h = ca. 120°; [3, 292]
Fig. 9.48). However, any reaction between the inclusions and the steel will cause a
decrease in the contact angle (i.e. reaction wetting) [293].

9.7.3.2 Electromagnetic Devices

Several electromagnetic devices are used in continuous casting; these devices have
different effects on steel cleanliness (i.e. inclusions) [294].
Electromagnetic stirring (EMS using AC ca. 50 Hz)
Inclusions and gas bubbles can be removed from the steel by using electromagnetic
stirring (EMS). EMS of the steel is achieved by the interaction of a magnetic field
(produced by a static induction coil) with the electrically conducting liquid metal.
The magnetic field can be applied either to the mould (M-EMS) or to the strand
(S-EMS). The principal benefits of EMS are that it provides (i) homogenisation of
the melt (ii) a refined microstructure (iii) inclusion removal [294, 295] and (iv) has
been reported to reduce the number of longitudinal cracks. The inclusions are
removed because small particles, in a shear field, experience a (Saffman) lift force
perpendicular to the direction of the flow. The inclusions are removed if the
Saffman velocity exceeds the velocity of the solidification front [296]. The Saffman
force increases as the size of the inclusion increases. When EMS is applied, the
Saffman forces generated were found to successfully remove inclusions and gas
bubbles with diameters greater than 100 lm [294, 295, 297].
Electromagnetic braking (EMBr)-DC
Electromagnetic braking (EMBr) exists in several forms. The original EMBr was
developed to reduce the velocity of the metal flow (Vmetal) in the SEN region by
using a DC static magnetic field [294, 295, 298]. Initially EMBR was applied
around the SEN region of the mould to reduce the velocity of the metal flow leaving
the SEN; a reduction of 40% in Vmetal has been reported [299].
9.7 Dissolution of Oxides, Nitrides and Carbides in Mould Slags 369

Level magnetic field (LMF)-DC

However, another form of EMBr has been developed in which the DC magnetic
field is applied across the whole width of the mould. This is known as level
magnetic field (LMF) and it is effective in reducing metal flow penetration in the
mould and in suppressing the mixing of different steel grades during ladle changes
[294, 295]. The reduction in metal velocity and penetration promotes the flotation
of inclusions and bubbles.
Flow control (FC)
EMBr has been developed to produce two magnetic fields across the full width of
the mould where (i) the upper field reduces the meniscus turbulence and (ii) the
lower field decelerates the metal flow, promotes Ar bubble and inclusion flotation
[298] and increases the meniscus steel temperature by ca. 10 °C [300, 301]. This
form of EMBr is often referred to as Flow control (FC) [298]. The reduction in
metal flow velocity results in a 30% decrease in vertical heat flux and tends to
produce shallow slag pools [302]. The velocity of the slag flow is also reduced as a
consequence of the decrease in metal flow and this leads to a decrease in slag
entrapment (see Chap. 11, Sect. 1.11). The use of EMBr has been reported to
reduce (i) the number of inclusions [295, 297, 298, 301, 303–305] (ii) pinholes
[298, 301, 304, 305] (iii) mould-level variations [298, 301] and (iv) the size of the
slag rim [306].
Electromagnetic casting (EMC-AC high frequency)
In EMC, an AC current is passed through a solenoid coil which surrounds the
mould. The resulting Lorenz force creates a horizontal, circulating secondary field
which produces an inward-acting force [294, 307]. This pinch force pushes the melt
away from the wall and results in reduced horizontal heat transfer. This, in turn,
causes delayed solidification of the shell and results in a short meniscus; both of
these effects result in a decrease in capture of inclusions by the shell (see Chap. 11–
Sect. 1.11). However, in EMC there is a tendency for unstable metal flow in the
meniscus region. Pulsative EMC (P-EMC) was introduced to control the metal flow
induced by the magnetic field [307]. The application of P-EMC was found to
significantly decrease the number of inclusions (Nincl) the depth of the the oscil-
lation marks (dOM) and the number of longitudinal cracks (NLcr) [307, 294].

9.7.4 Transport Through Slag/Metal Interface

It has been suggested that inclusions need a rest period before they can penetrate the
metal/slag interface but it was observed that larger particles, with greater
momentum, were capable of piercing the interface directly [292].
370 9 Properties of Mould Fluxes and Slag Films

9.7.5 Dissolution of Inclusions

One of the principal tasks of the mould slag is to dissolve inclusions. The ability of
the slag to dissolve inclusions (usually Al2O3) has been reported by several workers
[95, 96, 98, 100, 102, 185, 186, 308–310].

9.7.5.1 Factors Affecting Dissolution

(i) Saturated concentration of inclusions

The driving force for the dissolution of inclusions, (e.g. Al2O3) in the slag is the
Al2O3 concentration difference (Csat–C), i.e. the difference between saturated and
actual concentrations of Al2O3 in the slag. There is a wide variation in the reported
values for Csat for different inclusions (e.g. Al2O3: ca. 40% and ca. 50% at 1400
and 1500 °C, respectively, [311, 312]; TiO2 ca 10% [153] ZrO2 ca. 2% [313] and
TiN ca 0.5% [76].

9.7.5.2 Inclusion Blockages

There is a tendency for undissolved particles to accumulate and to agglomerate,

especially where (Csat–C) is low (e.g. TiC, TiN, ZrO2); the agglomeration (or
clustering) occurs through collisions of particles [314]. Collisions increase with
increasing metal flow velocities. The agglomerates tend to congregate in the mouth
of the infiltration channel and close down the space for slag infiltration and hinder
the flow of liquid slag into the mould/strand channel, resulting in a decrease in
lubrication. For instance, powder consumption tends to be low when casting
Ti-stabilised stainless steels due to the accumulation of TiN, Ti(CN) or perovskite
(CaTiO3) accretions in the infiltration channel. It has been suggested that the
amount of CaTiO3 can be minimised by restricting the amount of CaO, i.e. use a
mould slag with low basicity [133]. In extreme cases, the accumulations could
cause sticker breakouts. The presence of TiC and TiN particles also results in an
increase in viscosity [196, 197]. A mathematical model for the clustering of par-
ticles has been reported [315].

9.7.5.3 Chemical Composition of the Mould Slag

The dissolution of inclusions is also affected by the mould slag composition; the
dissolution of Al2O3 in mould slag is promoted by:
• Additions of those oxides which increase the basicity index (Eq. 9.51), e.g. BaO
[100].
• Low Al2O3 concentrations (C) in the mould slag, i.e. high (Csat–C) [316, 317].
9.7 Dissolution of Oxides, Nitrides and Carbides in Mould Slags 371

• High fluoride, Li2O and Na2O contents in the mould slag [98, 182, 186, 317].
• B2O3 was found to have little effect on the slag’s ability to dissolve Al2O3 [102].
• Low slag viscosity, the rate of dissolution increasing with decreasing viscosity
[318, 319].

BI ¼f1:53%CaO þ %MgO þ %CaF2 þ 1:94%Na2 O þ 3:55%Li2 O g

ð9:51Þ
=ð1:4%SiO2 þ 0:1%Al2 O3 Þ

9.7.5.4 Kinetics of Inclusion Dissolution

The kinetics of inclusion dissolution has attracted attention in recent years with the
introduction of high-Al steel casting where large amounts of Al2O3 are formed;
these must be dissolved quickly to avoid casting problems.
The kinetics of Al2O3 dissolution have been studied by several workers [308,
311, 312, 318–320]. In these experiments an Alumina disc was rotated in molten
mould fluxes for a known time and the rate of Al2O3 dissolution was determined by
re-weighing the disc. Confocal laser microscopy [321, 322] and thermodynamic
calculations [131] have also been used to determine the rate of Al2O3 dissolution in
mould slag. Values for the diffusion constant (D) of (2–8) and (4–17)  10−8 m2s−1
were recorded at 1400 and 1500 °C, respectively [311]. Alumina dissolution is
promoted by increasing mass flow rates and agitation. The mass flow rate (J) in the
rotating disc experiment is given by the following equation where m = kinematic
viscosity (=η/q) of the liquid and x = angular velocity (in s−1).

J ¼ 0:62D2=3 m1=6 ðCsat  CÞ
x0:5 ð9:52Þ

A mathematical model has been developed to calculate the dissolution time for
alumina particles [311]. However, it has been reported that Marangoni flows
enhance the effective diffusion coefficients [323, 324]. Metal/ slag reactions are
ionic and the dissolution of an Al2O3 particle at the steel/slag interface causes local
changes in the charge on the slag anions (Eq. 9.53) and gives rise to local, circu-
latory (Marangoni) flows which enhance mass flow rates and the dissolution of
inclusions [323, 324].

4Al þ 3Si4 þ ¼ 3Si þ 4Al3 þ ð9:53Þ

Scheller [323, 324] studied TiO2 transfer from metal to slag and found
Deff  20D where Deff is the effective diffusion coefficient. It is customary to use
the dimensionless Sherwood number (Sh) to represent mass transfer (in this case,
DTiO2/Ti). It was also noted that the ratio (DTiO2/Ti) increased with increasing
pool depth (dpool) for pool depths of 0–6 mm but remained constant for dpool > 6
mm [323, 324]. Consequently, Scheller proposed that the modified Bodenstein
372 9 Properties of Mould Fluxes and Slag Films

No. (Bo*) replace the Sherwood No. (Sh) [323, 324] for the description of mass
transfer. The Bodenstein No. (Bo*) is given by Bo* = (total mass transferred)/(total
mass flux density from diffusion of various species).


Bo ¼ dpool
msl =Di Fqsl Þ ðDCRi =DCD Þ ð9:54Þ

where m is the mass flux density, C = concentration, Fqsl = inertia force for slag, the
subscript, i = species, e.g. Ti or Al and R = total (i.e. by convection + diffusion).
Values of log Bo* were correlated with log (Ra. Re2/We) where Ra = Rayleigh No
(represents convective forces) Re = Reynolds No. (=balance of inertial and viscous
forces) and We = Weber No. [=balance of inertial and interfacial (or surface) forces].
The dissolution process of Al2O3, MgO and MgAl2O4 inclusions in F-free mould
slag has also been explored using Confocal laser microscopy; it was found to be
diffusion-controlled [321]; typical dissolution times of ca. 200 s were found for
150 lm inclusions. A similar conclusion was also found in rotating disc studies in
CA-type mould slags (used to cast high-Al steels) where the rate-determining step
for the dissolution of Al2O3 was found to be the diffusion of Al2O3 in the slag
boundary layer; the intermediate phase, CA2 was formed during the dissolution
[319]. The rates of dissolution of MgO, Al2O3 and MgO
Al2O3 in slag were found
to be ca. 4 times higher than that of ZrO2 [313].

9.8 Other Tests Used on Mould Powders

A variety of other laboratory tests are used to characterise both the behaviour of
mould powders and the performance of the powder bed [232, 325–328]. Some of
these methods have also been covered in the individual chapters, e.g. melting rate
tests and methods for measuring pool depths are described in Chap. 4, Sects. 4.4
and 4.7, respectively.

9.8.1 Bulk Density

The thermal insulation provided by the powder bed is essential to prevent solid,
steel skulls from forming on the surface of the metal. Thermal insulation increases
with increasing bulk density of the casting powder and with increased powder bed
depth.
Bulk density is usually measured by pouring the casting powder into a beaker of
known volume (V) and then determining the mass of the powder added [i.e.
DW = (mass of filled crucible)––(mass of empty crucible)]. The bulk density is
calculated from (DW/V). The bulk density tends to be in the following hierarchy
(powder > extruded > granulated) as can be seen from Fig. 9.49a [232] since
smaller particles (i.e. in powders) tend to pack more compactly.
9.8 Other Tests Used on Mould Powders 373

(a) (b)
1 40
0.8 30
0.6
20
0.4
0.2 10
0
0
Powd Extr Gran Exp-gr
Powd Extr Gran Exp-gr
Bulk density, kgm-3
Angle of repose, degrees

Fig. 9.49 Representations of a bulk density and b angle of repose as a function of powder type;
Powd Powder; Extr Extruded granules; Gran Granules; Exp-gr Expanding granules (permission
granted; ISS/AIST, re-drawn after [232])

9.8.2 Flowability

Spherical granules flow easily and when the metal flow turbulence causes a
standing wave to be formed (Fig. 9.50) the spherical granules can flow downhill
from the peak and reduce the thermal insulation in the area around the peak.
Flowability is particularly important in ingot casting to minimise the formation of
“open eyes”.
A measure of the flowability can derived using the angle of repose test [325,
326] in which a known volume of material is allowed to flow into a pile formed on
a flat plate and the angle of inclination is measured. Powders with a greater
flowability have a smaller angle of repose.

Fig. 9.50 Schematic

drawings showing (a) the
formation of a standing wave
resulting from turbulent metal
flow (permission granted,
ISS/AIST, [232])
374 9 Properties of Mould Fluxes and Slag Films

Fig. 9.51 Permeability Index 8

for different powder types,
Powd Powder; Extr Extruded 6
granules; Gran Granules;
4
Exp-gr Expanding granules
(Permission granted, 2
ISS/AIST, re-drawn after
[232]) 0
Powd Extr Gran Exp-gr

Permeability,

9.8.3 Permeability Index

Vertical heat transfer through the bed contains contributions from gaseous heat
transfer (i.e. convection of Ar, CO, CO2, N2) passing up through the bed. The
permeability test is designed to provide a measure of gaseous conduction through
the sintered and powder layers of the bed [232, 237]. The permeability index is
determined by measuring the time taken for 1 litre of air to pass through similar
volumes of powders. The air flow is regulated by a semi-permeable membrane
positioned in front of the sample. It can be seen from Fig. 9.51 that spheroids and
granules have much greater permeability than powders because the voids between
particles are much greater than those for the larger granules.

9.8.4 Thermal Insulation

The thermal insulation provided by the powder bed is important in controlling the
depth of the molten pool (See Sect. 4.2.1). There is a test for determining the
insulating properties of various casting powders. The heat flux passing through a
known depth of powder is measured in this test (Fig. 9.52a) [232]. The sample is
placed on the heat source (e.g. a hot SiC plate) which ensures that the heat transfer is
unidirectional, as it is in the powder bed. Typical results are shown in Fig. 9.53. The
temperature gradient (DT/Dd) in the bed is calculated by monitoring temperature
readings from two thermocouples immersed at known distances (d) in the powder
bed (Eq. 9.55). The heat flux can also be determined using a water-cooled plate
(Fig. 9.52b) by measuring the temperature increase (DT) for a known mass flow rate
of water (mH2 O /dT) (Eq. 9.56) [329, 330]. Another way of measuring the vertical
heat flux is to measure the sample surface with infrared detection sensors. Spitzer
et al. [329, 330] used the data generated with their apparatus to develop a mathe-
matical model to predict the thicknesses of the various layers in the powder bed.
9.8 Other Tests Used on Mould Powders 375

SAMPLE UNDER TEST COVER Thermocouples

STEEL TUBING Cooling water Heat flux
SILICONE CARBIDE prooe
Al2O3-Sheaths Steel ring
PLATE
HEATING Thermocouples
ELEMENTS Al2O3-Rod
Graphite tube

OPTICAL CONTROLLER Casting powder

SIGHTING HOLE

Electrical
connection

Graphic plate
at 1400oC

Fig. 9.52 Schematic diagrams showing apparatus used to determine the thermal insulation
provided by mould powders a due to Diehl [232] and b Spitzer [329, 330] (permission granted,
(a) ISS/AIST [232] (b) Verlag Stahleisen [329])

Fig. 9.53 Effective thermal 1.6

conductivity (keff) of various
types of powders; Powd 1.2
Powder; Extr Extruded
granule; Gran Spherodised 0.8
granule; Exp-gr Spherodised
granule containing an 0.4
expanding agent like starch
(permission granted, 0
ISS/AIST, re-drawn after Powd Extr Gran Exp-gr
[232])
keff, Wm-1K-1

DT=Dd ¼ ðT1  T2 Þ=ðd1  d2 Þ ¼ q=kbed : ð9:55Þ

q ¼ ðmH2 O =dT ÞðTexit  Tentrance Þ ð9:56Þ

It can be seen from the effective thermal conductivity values given in Fig. 9.53
that:
(i) Thermal conductivities of powders are significantly lower than those of
granules and spheroids.
(ii) The values obtained in these insulation tests [232] are ca.10 higher than the
values obtained using the Transient Hot Wire (THW) method on powders [61,
62, 229] shown in Fig. 9.34.
This is probably due to the lower contributions from gaseous convection in the
more densely packed (lower permeability) powders than in granules (see
Sect. 9.3.5, Powder bed). It also indicates that gaseous convection makes
376 9 Properties of Mould Fluxes and Slag Films

(a) (b)

Fig. 9.54 Schematic drawings showing the apparatus used to determine the hydrogen content of
slag [116]

significant contributions to keff in these insulation tests (where there is an appre-

ciable temperature gradient across the sample and where the specimen is heated
from the bottom). These conditions are not dissimilar to those in the mould.

9.8.5 Measurement of Moisture and Hydrogen

If the moisture content of a casting powder is high it can result in the creation of
slag films with pores containing hydrogen. These slag films containing pores,
reduce the horizontal heat flux significantly and can result in sticker breakouts
because the resulting shell is too thin and too weak [116] [117].
Free moisture contents of casting powders are usually measured from the
recorded mass loss when a known mass of powder is heated at 110 °C for a set
time.
Crystalline moisture is usually measured by the Karl Fischer method.
The amount of hydrogen present in a known mass of slag film has been mea-
sured with the apparatus shown in Fig. 9.54 [116]. It should be noted that this
method requires 65 g of slag film which is a large amount to retrieve from the
mould. However, Japanese workers have used solid H1 Nuclear Magnetic reso-
nance equipment to measure the hydrogen content [117, 296] which presumably
requires much smaller samples.

9.9 Comparison of Properties of Powders Used

in Ingot- (IC) and Continuous Casting (CC)

Since ingot casting is mostly carried out by uphill teeming, it has been assumed
below that ingot casting is carried out in this manner. It will be seen below that
there are some similarities between mould powders used for continuous casting
(CC) and ingot casting (IC) but there are also some significant differences between
them.
9.9 Comparison of Properties of Powders Used in Ingot- (IC) … 377

9.9.1 Differences in Properties of Mould Powders Used

in CC and IC

The principal difference between powders used in CC and IC lies in their powder
consumption or powder requirement. In continuous casting, the mould powder is
fed continuously into the mould, where it forms a slag pool which infiltrates into the
shell/mould channel; this provides the newly formed shell with the necessary
lubrication. The powder consumption (Qs) in CC should be appropriate for the
mould dimensions and casting conditions and Qslag t ¼ f  Qpowd
t has a value of ca.
−1
0.48 kg (slag) tonne (steel).
In ingot casting, the slag pool is necessary to seal the metal from the gaseous
atmosphere and to provide some thermal insulation. Lubrication is not so important
in IC because the steel moves in a slow and steady manner. The powder require-
ment ðQIC Þ is the amount of powder added at the start of casting to ensure the
formation of a protective slag pool and to maintain thermal insulation throughout
the cast. Consequently, powder requirement (Qpowd t IC ) is somewhat different to
powder consumption (Qpowd t CC ) in CC. The powder consumption (Qpowd t ) is the
slag
weight of powder and assuming f* = 0.75, typical values of Qt IC are ca. 3 times
those in CC. This is because in IC, the powder requirement, Qpowdt IC , should ensure
that that no “open eyes” are formed and the steel meniscus does not freeze. It should
be noted that the Qpowd
t IC value given above is a mean value, whereas the value for
fly-ash powders is 15% lower since little molten slag is formed but is 15% higher
for low C powders where more liquid slag is formed.
The formation of a slag film is an important function in CC slags. A slag film is
probably formed in ingot casting but is restricted because of the high viscosity
(especially for fly-ash slags).
Thus in CC, powder consumption should be preferably high enough to provide
good lubrication. In IC, powder requirements are already high, so efforts are made
to ensure it is not excessive. Powder consumption/requirement values increase with
reductions in Tliq, viscosity and casting speed.
Although the principal difference between CC and IC lies in the powder
consumption/requirement, there are other differences in CC and IC; these are
summarised below.

9.9.1.1 Thermal Insulation and Vertical Heat Flux (qvert)

The thermal insulation supplied by the powder bed is important to both IC and CC
since it controls the depth of the liquid pool. However, it is probably more important
in IC since the powder is added to the metal surface in the early moments of casting
and must provide thermal insulation throughout the cast, whereas the powder is fed
continuously in CC. Thus, thermal insulation is more critical in IC for casts of <10
tonnes and greater care must be taken with flowability; melting rate, etc.
378 9 Properties of Mould Fluxes and Slag Films

9.9.1.2 Slag/Pool Composition

Slag/metal reactions (e.g. 3FeOsl + 2Al = Al2O3 sl + 3Fe) change the composition
of the slag pool. In CC the slag is continuously replenished, so compositional
changes tend to be moderate. In IC the powder is added as a batch at the start;
consequently, changes in slag compositions tend to be much larger. The loss of FeO
will cause a large increase in qR but this is offset by a decrease in convection due to
the increase in viscosity resulting from Al2O3 pick-up.

9.9.1.3 Steel Movement

In IC there is a gradual, upward movement of the steel whereas in CC the steel is

withdrawn downwards rapidly and oscillation of the strand is imposed.

9.9.1.4 Lubrication

This is very important in CC (especially in high-speed thin slab casting) but appears
to be less important in IC, presumably, because the shell is moving at a much lower
speed and there is no oscillation of the mould.

9.9.1.5 Mould Geometry

Slabs are usually continuously cast and slabs have a considerably smaller ratio for
R* (=surface area/volume) of the mould compared with billets and consequently,
require considerably more lubrication than billets.

9.9.1.6 Heat Extraction

Cast iron moulds are used in IC, whereas water-cooled copper moulds are used in
CC; consequently, the cooling rates and horizontal heat fluxes are considerably
larger in CC than in IC (Note kCu/kFe = ca. 10).

9.9.1.7 Formation of Slag Film

In CC the formation of a slag film is essential to the control of the heat extraction
rate from the shell (qhor) and the process is aided by the oscillation of the mould.
Once a slag pool is formed in IC some slag infiltration should occur through the
effect of gravity; the flow rate will be dependent on the depth of the slag pool and
the fluidity of the slag (i.e. 1/η) the magnitude of qhor does not appear to be an
important issue in IC.
9.9 Comparison of Properties of Powders Used in Ingot- (IC) … 379

9.9.1.8 Sticking of Shell to Mould

The sticking of the shell to the mould is a serious issue for CC. It is tackled by
(i) oscillating the mould and (ii) creating a slag film with a liquid layer next to the
shell. In general, the sticking of the shell to the mould does not appear to be an issue
in IC but can occur in big-end-up moulds where it results in vertical cracks.

9.9.2 Tasks Carried Out by Powders Used

in Continuous- and Ingot Casting

The powders used in CC and IC must carry out specific tasks, these are summarised
in Table 9.17; It can be seen from Table 9.17 that some tasks are equally important
in IC and CC but other tasks are more important in CC than in IC (e.g. control of
lubrication and horizontal heat transfer are paramount in CC, whereas, the prime
tasks in IC are to provide cover against meniscus freezing and steel oxidation).

9.9.3 Properties and Characteristics of Powders Used

in Continuous and Ingot Casting

The properties and casting parameters for mould powders used in continuous
casting are compared against those for two different types of powders used in ingot
casting in Table 9.18. The latter are (i) fly-ash-based powder and (ii) high basicity
mould fluxes for casting Al-killed steels (which contain little FeO).

9.9.4 Conclusions from Comparison of CC and IC Mould

Powders

1. The powder consumption/requirement is the property showing largest differ-

t  3Qt
ences with QIC CC
due to QIC
t being applied as a batch which must provide
cover through the entire cast whereas QCC t is fed continuously.
2. Since powder consumption, QIC t , is already high, IC powders tend to be
designed to minimise excessive consumption (e.g. high viscosity, low melting
rate, etc.).
3. In CC powders, the prime functions are associated with the formation of the slag
film, i.e. provision of lubrication and heat extraction from the shell which
involves optimisation of properties (e.g. viscosity, Tbr and fcrys) according to the
casting conditions and steel grade being cast; lubrication and heat transfer (qhor)
do not appear to be important in ingot casting.
380 9 Properties of Mould Fluxes and Slag Films

Table 9.17 Summary of the various tasks to be carried out by powders for CC and IC
Task Continuous casting (CC) Ingot casting (IC)
Seal from Important:pool formed at an Important: pool formed fairly early
atmosphere-Slag early stage in cast
pool High (C/S) leads to low O High FeO, low(C/S) leads to high O
potential potential
Deep pool helps consumption Low η slag leads to “open eye” and
Deep pool minimises C pick-up low dpool
by steel Deep pool minimises C pick- up by
steel
Thermal insulation Important in forming slag pool: Very Important: qvert#
(top) Exothermic reactions reduce T IR absorption by FeO reduces
gradients radiation
Deeper powder bed reduces qvert. High pool viscosity reduces
Low viscosity tends to increased convection
convection Exothermic reactions reduce T
EMBr reduces qvert. by ca. 30% gradients:
High packing density helps
Dissolve inclusions Important: Important:
Aided by high (C/S); %Na2O; % Retarded by high viscosity and low
CaF2; low viscosity (C/S)
High FeO promotes Al2O3, SiO2
formation
Lubricate shell Very Important: friction# as dl" Seems less important:
Control via dl" as Tbr# Glasses form scl above Tg- move
Glassy phases in slag film form under stress
scl at T > Tg High viscosity lowers consumption
Consumption increases with QIC needs to be low to last entire
decreasing η cast
Qs increases as parameter, R*
increases
Control heat Very Important: Seems less important
extraction (qhor) Tbr controls ds: Lower qhor and cooling rates
qR reduced by increasing fcrys and qR reduced by FeO content
Dcrys
Sticking of Very Important: Seems less important—possibly
shell/mould Formation of liquid layer in slag due to lower shell speeds
Avoid blockages to slag flow, e.g.
broken rim
Powder requirement in IC is based on total amount of powder added to maintain cover of the steel
surface for entire cast and is different to Qs, Qt etc. in CC

4. Thermal insulation and the formation of a slag pool seem to be important in both
processes but are probably more important in IC because of the need to maintain
cover through the entire cast to avoid “open eye” formation and the resulting
re-oxidation of the steel surface.
9.9 Comparison of Properties of Powders Used in Ingot- (IC) … 381

Table 9.18 Comparison of important casting parameters and physical properties between
(i) mould powders for continuous casting (CC) and (ii) two types of mould powder used in Ingot
casting (IC), i.e. fly-ash-based and high basicity powders; {} = Typical values
Property CC Mould powder/slag IC Fly-ash powder IC high (C/S) powder
Slag pool Promoted by low Tliq, Seals steel from air Seals steel from air
depth deep bed, exothermic Deep pool-low C-pick-up Deep pool-low
powders Promotes consumption C-pick-up
-provide good Promotes
consumption consumption
Deep pool-low Low free carbon
Cpick-up promotes slag
-fewer inclusions. Low melting
free carbon content
Powder Qs = Appropriate for Qt IC should be low- aided Qt IC should be low-
consumption mould dimensions (R*), by high viscosity not helped by low
Vc; f, s, High C%, low DC -to keep viscosity
- Affected by melting melting rate low. Very little High C%, low DC -to
rate melting. keep melting rate low
{Qslag
t = 0.48 kg {Qpowd
t = 1.5–2. kg {Qslag
t = 2 kg
tonne−1} tonne−1} tonne−1}
Viscosity, η = appropriate for Vc; High η minimises Qt IC Low η minimises Qt
dPas R* -slag entrapment reduced IC
High η slag entrapment -reduces convection, -promotes slag
reduced -slows Al2O3 dissolution entrapment
{η1300 = 0.5–2} -{η1300 = 300–400} -increases
convection,
improves Al2O3
dissolution
{η1300 = 3–7}
Tliq, Tbr Tbr-appropriate-steel High Tliq-shallow pool
grade High Tliq-reduces Qt IC
Tbr-controls qhor via ds Tliq {1150 °C}
Tbr-controls lubrication
via dl
Tliq {1150 °C}
Dissolution - Promoted by high Retarded as cast proceeds - Better than
of inclusions (C/S); % Na2O; %CaF2; lower (Csat–Co)-higher η- fly-ash-aided by high
low η; low cmsl: high high Al2O3 by slag/metal (C/S) lower η–No
(Csat- C0) reaction FeO-Less Al2O3
Slag has high O potential Slag-Low O potential
Interfacial Sensitive to S%; slag Low cmsl due to FeO % cmsl Sensitive to S
tension FeO% !Slag entrapment content-cmsl lower
(cmsl) Low cmsl ! Slag -Aids Al2O3 dissolution than fly-ash
entrapment !Small Ripple marks
Thermal Promoted-deep powder Good thermal insulation Moderate insulation
insulation bed-Exothermic High FeO−low qvert. High due to- no IR
powders high packing η-low convection absorption, higher qR
density Good powder structure Higher convection-
low η
Poorer powder
structure
382 9 Properties of Mould Fluxes and Slag Films

5. The high FeO content of fly ash is both a blessing and a curse since its absorbs
large amounts of radiated heat during solidification but is a continual source of
O for Al2O3 formation.

9.10 Summary

It is obvious that the performance of a mould slag relies heavily on its physical
properties. The lubrication the liquid slag supplied to the shell is directly related to
the fluidity (i.e. reciprocal viscosity) and the horizontal heat transfer is largely
determined by the solidification temperature (Tbr) the fraction of crystalline phase
(fcrys) and the thermal conductivity of the slag film. However, other properties can
affect defects, e.g. slag/metal interfacial tension affects slag entrapment and scale
formation on the steel. Thus, virtually all the thermophysical properties are
important. However, some slag properties can be measured accurately (e.g. density
can usually be measured to ±2%) whereas others (e.g. viscosity) are subject to
much greater uncertainty (±25%). Thermal conductivity is particularly vulnerable
to uncertainty because (i) it is difficult to quantify the individual contributions from
both convection and radiation conduction in the measurements and (ii) there is
currently an unresolved dispute over which of two sets of thermal conductivity
values should be used for the slag for T > 700 °C. The situation is further com-
plicated by the fact that the radiation conductivity is dependent on specimen
thickness and there is some indications that the initial glassy slag film is optically
thin. This poses the questions for simulation experiments, e.g. is the sample of
similar thickness to the slag film formed in the mould and how representative is it of
the property value for the slag film?
It is surprising that the property database for mould slags is far from complete.
For instance, there are no Cp data for mould slags for T > 750 °C and there are few
density data available. The lack of property data is particularly severe for the
specialist mould slags (e.g. F-free or CA-type slags).
Mathematical models (of heat and fluid flow) are now capable of predicting
product defects but these models require property data for the mould slag and the
steel, as an input. There are a large number of mould slags in use and it is
impractical to measure all the required properties since performing measurements is
both time consuming and expensive. Thus, it is important to develop models which
calculate the required property data from chemical composition, since this is usually
available for both mould slag and steel. Some models are available to calculate
properties and some of these are referenced in this chapter. However, the uncer-
tainty in the predicted values will always be greater than the uncertainties for the
experimental property values. Another problem affecting the quality of property
predictions is related to mould powders changing composition with time in the
9.10 Summary 383

experiment, e.g. CO2, some F and Na2O are evolved and Al2O3 is picked up during
a casting. In order to improve property- composition models, it is necessary to have
post-measurement compositions of the slag available; unfortunately, these data are
rarely available.

References

1. BO Mysen, Structure and properties of silicate melts, (Elsevier, Amsterdam, 1988).

2. BO Mysen, Earth Science Reviews, 27, 281, (1990).
3. Slag Atlas, (Stahl u Eisen,VDEh, Dusseldorf,1995).
4. BN Roy, A Navrotsky, J Amer. Ceram. Soc., 67, 606, (1985).
5. T Higo, S Sukenaga, K Kanehashi, H Shibata, T Osugi, ISIJ Intl., 54, 2039, (2014).
6. JL Li, QF Shu, KC Chou, Ironmaking and Steelmakng, 42, 154, (2015).
7. TS Kim, JH Park, ISIJ Intl., 54, 2031, (2014).
8. BT Poe, PF McMillan, J Amer. Ceram. Soc., 77, 1832, (1994).
9. PF McMillan, WT Petuskey, B Cote, D Massiot, C Landron, JP Coutres, J Non-Cryst Solids,
195, 261, (1996).
10. TJF Stebbins, Q Zeng, J Non-Cryst. Solids, 262, 1, (2000).
11. Y Sasaki, M Iguchi, M Hino, ISIJ Intl., 47, 346, (2007).
12. Y Sasaki, M Iguchi, M Hino, ISIJ Intl., 47, 638, and 1370 (2007).
13. T Asado, Y Yamada, K Ito, ISIJ Intl., 48, 120, (2008).
14. JH Park, ISIJ Intl., 47, 1368, (2007).
15. S Hayakawa, A Nakao, C Ohtsuki, A Osaka, S Matsumoto, Y Miura, J Mater. Res., 13, 729,
(1998).
16. M Hayashi, N Nabeshima, H Fukuyama, K Nagata, ISIJ Intl., 42, 352, (2002).
17. VS Esaulov, GF Konovalov, SI Popel, VI Sokolov, Steel USSR, 6, 311, (1976).
18. M Susa, S Kubota, M Hayashi, KC Mills, Ironmaking and Steelmaking, 28, 390, (2001).
19. L Baak, A Olander, Acta Chem. Scand., 9, 1350, (1955).
20. PM Bills: J Iron Steel Inst., 201, 133, (1963).
21. A van Bernst, C Delaunois, Verres Defract, 20, 435 (1966).
22. T Nakamura, K Morinaga, T Yanagase, Nippon Kinzoku Gakkaishi, 41, 1300, (1977).
23. A Shankah, PhD thesis, Studies of high- alumina blast-furnace slags, KTH, Stockholm
(2007).
24. S Yagi, K Mizoguchi, Y Suginahara, Bull. Kyushu Inst. Technol. Sci. Technol., 40, 33,
(1980).
25. SF Zhang, X Zhang, H Peng, LY Wen, GB Qui, ML Hu, CG Bai, ISIJ Intl., 54, 734, (2014).
26. JF Li, Q Shu, KC Chou, Canad. Met. Quart., 54, 85, (2015).
27. T Mukongo, C Pistorius, A Garbers-Craig, Ironmaking Steelmaking, 31,135, (2004).
28. AB Fox, KC Mills, D Lever, C Bezerra, C Valadares, I Unamono, J Laraudogoitia, J Gisby,
ISIJ Intl., 45, 1051, (2005).
29. H. Huang, J. L. Liao, K. Zheng, H. H. Hu, F. M. Wang, Z. T. Zhang, Ironmaking and
Steelmaking, 41, 67, (2014).
30. Z Wang, QF Shu, XM Hou, KC Chou, Ironmaking and Steelmaking, 39, 210, (2012).
31. QF Shu, Z Wang, J L.Klug, K Chou, P R. Scheller, Steel Res. Intl., 84, 1138, (2013).
32. G Urbain, Steel Research, 58, 111, (1987).
33. KC Mills, L Yuan, Z Li: Estimation of the thermo-physical properties of slags Part 1
Viscosity to be published.
34. T Iida, H Sakai, Y Kita, K Murakami, High Temp. Mater. Processes, 19, 153, (2000).
35. KC Mills, ISIJ Intl., 33, 148, (1993).
36. KC Mills, S Sridhar, Ironmaking and Steelmaking, 26, 262, (1999).
384 9 Properties of Mould Fluxes and Slag Films

37. VK Gupta, SP Sinha, B Raj, Steel India, 21, 22 (1998) and 17, 74, (1994).
38. J Duffy, MD Ingram, J Non-Cryst. Solids, 21, 373, (1976).
39. T Nakamura, Y Ueda, JM Toguri, J Jap. Inst. Metals, 50 (5), 456, (1986).
40. GH Zhang, KC Chou, G Xue, KC Mills, Met. Mater. Trans. B, 43B, 64, (2012).
41. QF Shu, KJ Hu, BJ Yan, JY Zhang, KC Chou, Ironmaking and Steelmaking, 37, 387,
(2010).
42. Q Shu, KC Chou, Ironmaking and Steelmaking, 40, 571, (2013).
43. GH Zhang, KC Chou, Met. Mater. Trans. B, 41B, 131, (2010)..
44. H Gaye, J Welfinger, Proc. 2nd Intl. Symp. Metall. Slags and glasses, Lake Tahoe, NV ed.
by HA Fine and DR Gaskell (TMS-AIME, Warrendale, PA, 1984) p. 357.
45. L Zhang, S Sun, S Jahanshai, Met. Trans. B, 29B, 177, (1998).
46. L Zhang, S Sun, S Jahanshai: Met. Trans B, 29B, 187, (1998).
47. GH Zhang, KC Chou, KC Mills, ISIJ Intl. 52, 355, (2012).
48. GH Zhang, KC Chou, J Min. Metall. B 48, 1/ 10, (2012).
49. GH Zhang, KC Chou, 40, 376, (2013).
50. KC Mills, L. Yuan; Z Li, GH Zhang, KC Chou, Proc. 9th Intl. Conf. Molten slags, fluxes
and salts (Molten 12), Beijing, China,2012 (Chinese Met. Soc., 2012)
51. KC Mills, L. Yuan, Z Li, GH Zhang, KC Chou, High Temp. Mater. Processes, 31, 301,
(2012).
52. A Fluegel, Electrical resistivity of silicate glass melts based on SciGlass Database.
http//glassproperties.com. 2007 p. 21.
53. M.Susa, KC Mills, MJ Richardson, R Taylor, D Stewart, Ironmaking and Steelmaking, 21,
279, (1994).
54. S Ozawa, M Susa, Ironmaking and Steelmaking, 32, 487, (2005).
55. S Ozawa, M Susa, T Goto, R Kojima, KC Mills, ISIJ Intl., 46, 413, (2006).
56. M Susa, A Kushimoto, R Endo, Y Kobayashi, ISIJ Intl., 51, 1587, (2011).
57. T Watanabe, M Hayashi, K Nagata, Paper presented at ISIJ –VDEh-Jernkontoret, Joint
Symposium, Osaka, Japan, April, 2013, p. 192.
58. K Tsutsumi, R Nakamura, J Ohtake, T Nagasaka, M Hino, ASIA Steel Intl. Conf., 2000,
(ASIA Steel’ 2000), The Chinese Society for Metals, Beijing, Vol. C, (2000), p. 366
59. JW Cho, H Shibata, T Emi, M Suzuki, ISIJ Intl., 38, 268, (1998).
60. K Nishioka, T Maeda, M Shimizu, ISIJ Intl., 46, 427, (2006).
61. P. Andersson, C Eggertson, Ironmaking and Steelmaking, 42, 456, (2015) and 465.
62. P Andersson, Thermal conductivity of powders used in continuous casting of slabs and
billets, in press.
63. M Hanao, M Kawamoto, ISIJ Intl., 48, 180 (2008) see also Tetsu-to Hagane 92 (1), 13,
(2006).
64. M Hanao, M Kawamoto, A Yamanaka, ISIJ Intl., 52, 1310, (2012).
65. MD Seo, CB Shi, JW Cho, SY Kim: Proc. 8th Europ. Conf. Continuous Casting, Graz,
Austria, 2014 (Austrian Met. Mater. Soc., 2014).
66. GH Wen, S.Sridhar, P Tang, X Qi, YQ Liu, ISIJ Intl., 47, 1117, (2007).
67. H Nakada, K Nagata, ISIJ Intl., 46, 441, (2006).
68. N Takahira, N Hanao, Y Tsukaguchi, ISIJ Intl., 53, 818, (2013).
69. M Hanao, Y Tsukaguchi, M Kawamoto, Proc. 4th Intl. Congress Science and Technol 2008,
Gifu, Japan, 2008, (ISIJ, Tokyo, 2008) p. 694.
70. P Grieveson, S Bagha, N Machingawuta, K Liddell, KC Mills, Ironmaking and Steelmaking,
15, 181, (1988).
71. T Watanabe, H Fukuyama, K Nagata, ISIJ Intl., 42, 489, (2002).
72. M Hayashi, T Watanabe, H Nakada, K Nagata, ISIJ Intl., 46, 1805, (2006).
73. K. Sorimachi, Fachber. Huttenpraxis Metallweitver., 20 (4), 244, (1982).
74. M Kawamoto, T Kanazaka, S Hiraki, S Kumakura, Proc. 5th Intl. Conf. Molten slags, fluxes
and salts, Sydney (1997) (ISS, Warrendale, PA, 1997) p. 777.
75. B Tarrant, G Brooks: Iron and Steelmaker, 2003 (5), 52, (2003).
References 385

76. T. Mukongo, MEng Thesis, Effect of titanium pick up on mould flux viscosity in continuous
casting of Ti-stabilised steel. Dept Mater. Sci. & Metall. Eng., Univ. Pretoria, 2003.
77. P Rocabois, JN Pontoire, J Lehmann, H Gaye, J Non- Cryst. Solids, 283, 98, (2001).
78. S Riaz, Ironmaking and Steelmaking, 39, 409, (2012).
79. Y Wang, Yu Wang, L Fan, B Xie, XP Liang, X Zeng, ZW Wei, Proc. 9th Intl. Conf Molten
slags, fluxes and salts (Molten 2012), Beijing, China, 2012 (Chinese Met. Soc., 2012).
80. S. Sankararanarynan, Proc. 75th Steelmaking Conf., 1992, (ISS, Warrendale, PA, 1992)
p 607.
81. JL Klug, R Hagermann, NC Heck, JFC Viela, PR Scheller, Proc. 5th Intl. Congress Science
and Technol. Steelmaking, Dresden, (2012) http//dx.doi.org/10.4322/tmm.2012.005.
82. C Righi, R Carli, V Ghilardi, Proc. 6th Intl. Conf. Metallurgical slags and fluxes, Stockholm,
Sweden, paper 102 (2000).
83. Z Li, R Thackray, KC Mills, Proc. 7th Intl. Conf. Molten slags, fluxes and salts, Capetown,
South Africa, 2004 (SAIMM, Johannesburg, 2004) p. 813.
84. M Hanao M Kawamoto, M Hara, T Murakami, H Kikuchi, K Hanazaki, Tetsu-to-Hagane
88, 23, (2002).
85. MV Fonseca, ODC Afrange, A Lavinas, AA Ramos, CA Valadares, Proc. 5th Intl. Conf.
Molten slags, fluxes and salts, Sydney, 1997 (ISS, Warrendale, PA, 1997) p 851.
86. H Fukuyama, T Watanabe, H Takahashi, H Tabata, T Oshima, K Nagata, Metallurgical and
Materials Processing: Principles and Technologies; vol 1, Materials processing fundamen-
tals and new technologies, ed. by F Kongoli, K Ingoki, C Yamauchi and HY Sohn (TMS
(2003) p. 937.
87. R Kinoshita, CAMP- ISIJ, 20, 781, (2007).
88. C Orrling, AW Cramb, Iron and Steelmaker, (2000), 53, (2000).
89. S Lachmann S., Scheller P. R., Proc. 6th Europ. Conf. on Continuous Casting, Riccone, Italy,
2008.
90. Y Harada, K Kusada, S Sukenaga, H Yamamura, Y Ueshima, T Mizoguchi, N Saito, K
Nakashima, ISIJ Intl., 54, 2071, (2014).
91. WL Wang, K Blazek, AW Cramb, Met. Mater. Trans. B, 39B, 66, (2008).
92. J Rudnizki, R Shepherd, E Balichev, S Karrasch, F Kruger, Proc 8th Europ. Conf.
Continuous casting, Graz, Austria, 2014 (Austrian Met. Mater. Soc., 2014).
93. PS Kharlashin. Azovetal, 1999 (4), 34, (1999).
94. H Heimbach, K Schulz, J Markardt, HJ Ehrenberg, Stahl u Eisen, 117, 105 (1997).
95. Y Lu, GD Zhang: Mater. Sci. Forum, 676/677, 877 (2011).
96. H Abratis, F Hofer, M Junemann, H Sardmann, H Stoffel, Stahl u Eisen, 116, 85, (1996).
97. SY Choi, DH Lee, DW Shin, JW Cho, JM Park, J Non-Cryst. Solids, 157 (8), 15, (2004).
98. Y Lu, GD Zhang, X Yu, Appl. Mech and Mater., 71/78, 2899, (2011).
99. QC Wei, YQ Ding, KD Peng, J Chonqing Univ., 4, 110, (1995).
100. Y Lu, X Fang, GD Zhang, Adv. Mater. Res., 287/290, 805, (2011).
101. AB Fox, K Mills, D Lever, M C Bezerra, CA Valadares, Unamono, J Laraudogoitia, J
Gisby: Proc. 36th Steelmaking Seminar, Vittoria, ES, Brazil, May (2005) available on line at
http://www.carboox.com/pdf/paper2005sfs.pdf.
102. Y Lu, GD Zhang, MF Jiang, Advanced Mater. Res., 233/235, 805, (2011).
103. LY Chen, GH Wen, CL Yang, F Mei, CY Shi, P Tang, Ironmaking and Steelmaking, 42,
105, (2015).
104. Q Wang, YJ Lu, SP He, KC Mills, Z Li, Ironmaking and Steelmaking, 38, 297, (2011).
105. X. J. Fu, G. H. Wen, P. Tang, Q. Liu, Z. Y. Zhou, Ironmaking and Steelmaking, 41,342,
(2014).
106. CB Shi, MD Seo, JW Cho, SH Kim, Proc. 8th Europ. Conf. Continuous Casting, Graz,
Austria, 2014, (Austrian Met. Mater. Soc., 2014).
107. Z Li, KC Mills, MC Bezerra, Proc. XXXV Semin. De Fusao Refino e Solidifacao Metals,
Salvador, Brazil (2004) p. 281.
108. H Sakai, T Kawashima, T Shiomi, K Watanabe,T Iida: Proc. 5th Intl. Conf. Molten slags,
fluxes and salts, Sydney (1997) (ISS, Warrendale, PA, 1997) p. 787.
386 9 Properties of Mould Fluxes and Slag Films

109. H Hashimoto, T Watanabe, K Nagata, CAMP-ISIJ, 17, 849, (2004).

110. M Susa, A Kushimoto, H Toyota, M Hayashi, R Endo, Y Kobayashi, ISIJ Intl., 49, 1722,
(2009).
111. P R Scheller, Proc. 6th Intl. Conf on Slags, fluxes and salts, Stockholm, 2000, Session 6,
Filename 405 pdf.
112. YG Moldana, FA Acosta, AH Castillejos, BG Thomas, Iron and Steel Technology, 2013,
(July),1, (2013).
113. P.V. Riboud, Y. Roux: Fundamental study of the behaviour of casting powders, Report EUR
9560 (Europ. Comm. Sci. and Tech. Publ., Luxembourg, 1985).
114. PV Riboud, C Gatellier, H Gaye, JN Pontoire, P Rocabois, ISIJ Intl., 36, 522, (1996).
115. K Tsutsumi, T Nagasaka, M Hino, ISIJ Intl., 39, 1150, (1999).
116. JN Pontoire, Revue de Metallurgie, CIT, 93, 1237, (1996) see also Proc. 3rd Europ. Conf.
Continuous Casting,Madrid, Spain (1998) p. 973.
117. Y Uemiya, T Mizoguchi,T Kajitani, Proc. 9th Intl. Conf Molten slags, fluxes and salts
(Molten 12) Beijing (2012) (Chinese Met. Soc, 2012).
118. L Hering, HP Heller, HW Fenske, Stahl u Eisen, 17, 61, (1992).
119. Y Meng, BG Thomas, Proc. ISS Tech., Indianapolis, (2003) see also Met.. Mater. Trans.
34B, 685 and 707, (2003).
120. TJ Billany, AS Normanton, KC Mills, P Grieveson, Ironmaking and Steelmaking, 18, 403,
(1991).
121. R J O’Malley, J Neal, Proc. Intl. Conf. New Developments in Metallurgical Process
Technol., Dusseldorf, METEC Congress (1999) p. 73.
122. MCC Bezerra, ODC Afrange, AC Valadares, Proc. Mills Symp., London (2002). (Inst of
Materials, London, 2002) p. 293.
123. S Riaz, KC Mills, K Nagata, AS Normanton,. Proc. Mills Symp., London (2002) (IOM,
London, 2002) p. 465.
124. L Courtney, S Nuortie-Perkkio, C Valadares, MJ Richardson, KC Mills: Ironmaking and
Steelmaking, 28 (5), 412, (2001).
125. KC Mills L Courtney, AB Fox, B Harris, Z Idoyaga, MJ Richardson, Thermochica Acta,
391, 175, (2002).
126. H Lidefelt, Swedish Inst. Metals Research Report JK D45; IM 1755, (1983) Stockholm.
127. SH Chang, Proc. Conf. Continuous casting of steel in Developing countries, Beijing (1993)
p. 832.
128. SP He, X Long, JF Xu, T Wu, Q Wang, Ironmaking and Steelmaking, 39, 593, (2012).
129. A, Lopez, Z Idoyaga, F Plazaola, S Riaz, KC Mills, Proc. 4th Europ. Conf. Continuous
casting, Birmingham, UK, 2002 (IOM Commun., London, 2002) vol 1, p. 360.
130. Q Wang, Y Wang, Jh Chi, B Xie, YM He, B Zhu, WM Chen, J Chongqing Univ –Eng. Ed,
5, 65, (2004).
131. IH Jung, MA van Ende, Proc. 8th Europ. Conf. Continuous casting. Graz, Austria, 2014
(Austrian Met. Mater. Soc., 2014).
132. KC Mills, A Olusanya, RF Brooks, R Morrell, S Bagha, Ironmaking and Steelmaking, 15,
257, (1988) see also KC Mills, Funamental study of the behaviour of casting powders.
ECSC Final Report 7210 CA 131/810 (1983) (Europ. Comm. Sci. and Tech Publ.
Luxembourg).
133. T Kishi, H Tsubio, H Takeuchi, T Nakano, M Yamamiya, T Ando, Nippon Steel Tech.
Report, 34, 11, (1987) see also T Kishi, Y Nagano T Nagano, Trans. ISIJ, 25, B302, (1985).
134. C Orrling, Y Fan, N Phinichka, S Sridhar, AW Cramb, JOM, 51 (7)(1999). http://www.tms.
org/pubs/journals/jom/9907/orrling/orrling-9907.html.
135. German Standard DIN 51740.
136. LJ Zhou, WL Wang, R Liu, BG Thomas, Met. Mater. Trans. B, 44B, 1284, (2013).
137. KC Mills, PV Riboud, K Schwerdtfeger, Report on ECSC Contract 7210 CA/131/311/810
“Fundamental study of the behaviour of casting powders” (1983).
138. P Artelt, Stahl u Eisen, 97 (23), 1171, (1977).
References 387

139. JA Moore, RJ Phillips, TR Gibbs, Proc. 74th Steelmaking Conf., Washington, DC,1991,
(ISS, Warrendale, PA, 1991) p 615.
140. V Ghilardi, R Carli, GL Faravelli, Proc. 3rd Europ. Conf. Continuous casting, Madrid, 1998,
p 971.
141. M Hanao, M Kawamoto, T Watanabe, ISIJ Intl., 44, 827, (2004).
142. J Chen, J Xie, F He, E Wan, Z Xu, C Yang, G Liu, Ironmaking and Steelmaking, 42, 126,
(2015).
143. KC Mills, S Karagadde, L Yuan, PD Lee, F Shahbazian, ISIJ Intl., 56, 264, (2016)
144. KC Mills, L Yuan, Z Li, RT Jones, J South Afr. Inst. Min. Metall., 111, 649, (2011).
145. S Riaz, Private communication Tata Steel Teesside Research Laboratories, Middlesbrough,
UK. (2008).
146. S Feldbauer, I Jimbo, A Sharan, K Shimizu, W King, J Stepanek, J Harman, AW Cramb,
Proc. 78 th Steelmaking Conf., 1995, (ISS, Warrendale, PA, 1995) p. 655.
147. JW Kim, J Choi, OD Kwon, IR Lee YK Shin, JS Park, Proc. 4th Intl. Conf. Molten slags and
fluxes, Sendai (1992) (ISIJ, Tokyo, 1992) p. 468.
148. Z Elahipanah, MSc Thesis “Thermo-physical properties of mould flux slags for the
continuous casting of steel.” KTH, Stockholm, (2012) available on line.
149. M Persson, M Gornerup, S Seetharaman, ISIJ Intl., 47, 1533, (2007).
150. I Marschall, N Kohl, H Harmuth, GM Xia, Proc. of 9th Intl. Conf. Molten slags, fluxes and
salts (Molten 12), Beijing, China, 2012 (Chinese Met. Soc., 2012).
151. V Kircher, I Marschall, N Kolbl, H Harmuth, Proc. 8th Europ. Conf. Continuous Casting,
Graz, Austria, 2014 (Austrian Met. Mater. Soc., 2014).
152. RJ Phillips, WF Salem, WR Emling, HD Baker, Proc. 73rd Steelmaking Conf. (1990) (ISS,
Warrendale, PA, 1990) p 247.
153. PW Johnstone, G Brooks, Proc. 5th Intl. Conf. Molten slags, fluxes and salts, Sydney (1997)
(ISS, Warrendale, PA, 1997) p. 845.
154. D Wang, M Jiang, CJ Liu, PY Shi, J North Eastern Univ. (Natural Sci.) 26 (11), 1082,
(2005).
155. P Tang, GH Wen, ZB Wang, SB Fan, ZB Zhang, SC Li, H Li, Proc. ISS Tech Conf.
(2003) (ISS, Warrendale, PA, 2003) p. 567.
156. S Sridhar, KC Mills; ODC Afrange; HP Lorz, R Carli, Ironmaking and Steelmaking, 27,
238, (2000).
157. KC Mills, AB Fox, R P Thackray, Z Li, Proc. 7th Intl. Conf Molten slags, fluxes and salts,
Capetown, (2004) (SAIMM, Johannesburg, 2004) p. 713.
158. JV Dubrawski, JM Camplin, J Thermal Analysis, 40, 329, (1993).
159. KC Mills, AB Fox, ISIJ Intl., 43, 1479, (2003).
160. JD Mackenzie, Physico-chemical measurements at high temperatures ed. by JOM Bockris,
White and JD Mackenzie, (Butterworths, London 1959) Chapter 15, p. 313.
161. LJ Wu, P Heller, XG Bi, PR Scheller, Proc. 9th Intl. Conf. Molten slags, fluxes and salts
(Molten 12) Beijing, China, 2012 (Chinese Met. Soc., 2012).
162. T Iida, T Tanaka, Proc. 4th Intl. Conf. Metallurgical slags and fluxes, Sendai (1992). (ISIJ,
Tokyo, 1992) p. 444.
163. M Tokuda, K Nakajima, T Yamamoto, H Nakajima, K Kawaguchi, The Sumitomo Search,
44, 368, (1990).
164. SJ Roach, H Henein, Canad. Met. Quart., 42, 175, (2003).
165. AP Froba, A Leipertz, Proc. (Abstracts) of 16th Europ. Conf. on Thermophysical properties,
London, (2002), p. 59.
166. N Kawasaki, K Watanabe, Y Nagasaka, High Temp.-High Press., 30, 91, (1998).
167. Y Kobayashi, Y Nagasaka, Proc. (Abstracts) 16th Europ. Conf. on Thermophysical
properties, London, Sept. (2002) p. 103.
168. KC Mills, M Halali, HP Lorz, A Kinder, R Pomfret, BA Walker, Proc. 5th Intl. Conf.
Metallurgical slags and fluxes, Sydney, Australia, 1997 (ISS, Warrendale, PA.) p. 535.
169. Z Zhang, G Wen, P Tang, S Sridhar, ISIJ Intl., 48, 739, (2008).
170. X Yu, GH Wen, P Tang, H Wang, Ironmaking and Steelmaking, 36, 623, (2009).
388 9 Properties of Mould Fluxes and Slag Films

171. JL Li, QF Shu, YA Liu, KC Chou, Ironmaking and Steelmaking, 41, 732, (2014).
172. Z Wang, Q Shu, KC Chou, Steel Research, 84, 768, (2013).
173. S Yoshimura, CAMP-ISIJ, 20, 837, (2007).
174. K Okazawa: CAMP-ISIJ, 16, 125, (2003).
175. S Riaz, Ironmaking and Steelmaking, 39, 409 (2012).
176. W McCauley, D Apelian, Canad. Met. Quart., 20, 247, (1981).
177. PV Riboud, M Olette,H Leclerc, W Pollak, Proc. 61st NOH- BOS Conf., Chicago,1978,
(ISS, Warrendale, PA, 1978) p. 411.
178. VL Zhuk, SV Timofeeva,VN Moshenskii, Steel USSR 21 (6), 255, (1991).
179. M Lanyi, CF Rosa. Metall. Trans., 12B, 387, (1981).
180. HY Chang, TF Lee, YC Ko, ISS Trans., 4, 27 (1984).
181. HY Chang, TF Lee, T Ejima, Trans. ISIJ, 27, 797, (1987).
182. M Kawamoto, K Nakajima, T Kanazawa, K Nakai, Tetsu-to- Hagane, 80, (3), 219, (1994).
183. A, Lopez, Z Idoyaga, F Plazaola, S Riaz, KC Mills, Proc. 4th Europ. Conf. Continuous
casting, Birmingham, UK, 2002 (IOM Commun.,London 2002), vol 1 p. 360.
184. JH Park, DJ Min, HS Song, YD Lee,. Proc. 85th Steelmaking Conf., Nashville, TN, 2002
(ISS, Warrendale, PA, 2002) p. 775.
185. T Lu, X Fang, G Zhang, Proc. Conf. Mater. For Renewable Energy and Environment
(2011)- p. 869 (2011).
186. Y Lu, X Fang, XF Yu, Advanced Mater. Res., 455, 805, (2012).
187. F Shahbazian, S Du, KC Mills, S Seetharaman, Ironmaking Steelmaking, 26, 193, (1999).
188. F Shahbazian, S Du, S Seetharaman, ISIJ Intl., 42, 155, (2002).
189. WL Wang, K Blazek, AW Cramb, Met. Mater. Trans. B, 39B, 66, (2008).
190. JL Klug, DR Silva, SL Freitas, MMSM Pereira, NC Heck, ACF Viela, D Jung, Steel Res.
Intl., 83, 791, (2012).
191. C Zhang, DX Cai, Proc. 8th Europ, Conf. Continuous Casting, Graz, Austria, 2014 (Austrian
Soc. Metall. Materials, 2014).
192. K Tsutsumi, K Watanabe, J Kubota, S Hatori, Y Miki, T Suzuki, T Omoto, Proc. 7th
Europ. Cont. Casting Conf., Dusseldorf, 2011, (VDEh, Dusseldorf, 2011) Session pp 1 / 4.
193. K Watanabe, K Tsutsumi, M Suzuki, H Fujita, S Hatori T Suzuki, T Omoto, ISIJ Intl.54,
868, (2014).
194. Q Liu, GH Wen, JZ Li, XJ Fu, P Tang, W Li, Ironmaking Steelmaking, 41, 292 (2014).
195. A Kusano, N Sato, M Okimori, S Fukunaga,K Nishihara, M Sato, Y Minigawa, Proc. 74th
Steelmaking Conf. (1991) (ISS, Warrendale, PA, 1991) p. 111.
196. YL Chen, GH Zhang, KC Chou, Met. Mater. Trans. B, 46B, 155, (2015).
197. YL Zhen, GH Zhang, KC Chou, KC Mills, Canad Met. Quart., 54, 340, (2015).
198. JA Kromhout: PhD Thesis “Mould powders for the high speed continuous casting of steel”.
Univ of Delft, NL, (2011) p. 165.
199. M Kawamoto, CAMP-ISIJ, 6, 1176, (1993).
200. PV Riboud, Y Roux, LD Lucas, H Gaye: Fachber. Huttenprax. Metallweiterverarb., 19, 859,
(1981).
201. M Hanao, M Kawamoto, T Tanaka, M Nakamoto, ISIJ Intl., 46, 346, (2006).
202. S Du, J Bygden, S Seetharaman, Met. Trans. B, 25B, 519 and 589. (1994).
203. M Hayashi, S Seetharaman, CAMP- ISIJ, 16, 860, (2003).
204. KC Mills, L Yuan, Z Li, GH Zhang, High Temp.- High Press., 42, 232, (2013).
205. QF Shu, Steel Research Intl., 80, 107, (2009).
206. M. Susa, K Nagata, KC Mills: Ironmaking and Steelmaking, 20, 312, (1993).
207. R Gardon, Proc. 2nd Intl. Conf. Thermal Conductivity, Ottawa (1962) p. 167 and J. Amer.
Ceram. Soc. 44, 305, (1961).
208. H Ohta, H Shibata, T Emi, Y Waseda, Tetsu-to Hagane, 61 (4), 350, (1997).
209. R Taylor, KC Mills, Ironmaking and Steelmaking, 15, 187, (1988).
210. Y Waseda, M Masuda, K Watanabe, H Shibata, H Ohta, K Nakajima, High Temp. Mater.
Processes, 13, 267, (1994).
References 389

211. H Shibata, JW Cho, T Emi, M Suzuki, Proc.5th Intl. Conf. Molten slags, fluxes and salts,
Sydney, (1997) (ISS, Warrendale, PA., 1997) p. 771.
212. H Ohta, M Masuda, K Watanabe, K Nakajima, H Shibata,Y Waseda, Tetsu-to- Hagane, 80,
463, (1994).
213. H Ohta, H Shibata, H Hasegawa, Y Kawakri, Y Shinoki, S Kitamura, Y Waseda, Proc. 9th
Intl. Conf. Molten slags, fluxes and salts (Molten 12), Beijing, China, 2012 (Chinese Met.
Soc., 2012).
214. H Ohta, Y Shiraishi, Proc. 2nd Intl. Conf. Metallurgical slags and fluxes, Lake Tahoe, NV,
USA, 1984, ed. HA Fine and DR Gaskell, (Met. Soc. AIME, Warrendale, PA, 1984) p 863.
215. M Kishimoto, M Maeda, K Mori, Y Kawai, Proc. 2nd Intl. Conf. Metallurgical slags and
fluxes, Lake Tahoe, NV, USA, ed. HA Fine and DR Gaskell, (Met. Soc., AIME,
Warrendale, PA, 1984) p. 891.
216. K Nagata, KS Goto, Proc. 2nd Intl. Conf. Metallurgical slags and fluxes, Lake Tahoe, NV,
(1984), ed. HA Fine and DR Gaskell, (TMS-AIME, Warrendale, PA, 1984) p. 875.
217. M Hayashi, H Matsui, Unpublished thermal conductivity measurements, Tokyo Inst. Tech.,
Tokyo March, 2015.
218. JJ Macho, G Hecko, B Golomowski, M Frazee, Proc. Conf. on Thinner Slab Casting,
Hamilton, ONT, Canada (McMaster Univ, Hamilton, 2005) p. 131.
219. KS Goto, HW Gudenau, K Nagata, KH Lindner, Stahl u Eisen, 105, 1387, (1985).
220. S Bagha, N Machingawuta, P Grieveson, Proc. 3rd Intl. Conf. Molten slags and fluxes;
Glasgow, 1988, (Inst. Of Metals, London, 1988) p 235.
221. JS Powell Unpublished data from NPL Teddington, cited in Slag Atlas [3] p 598.
222. M Hayashi, AA Riad, S Seetharaman, ISIJ Intl., 44 (5), 691, (2004).
223. M Gornerup, M Hayashi, CA Dacker, S Seetharaman, Proc. 7th Intl. Conf. Molten slags,
fluxes and salts, Capetown, South Africa, 2004 (SAIMM, Johannesburg, 2004) p. 745.
224. BJ Monaghan, RF Brooks, Ironmaking and Steelmaking, 29, 115, (2002).
225. H Ohta, K Nakajima, M Masuda, Y Waseda, Proc 4th Intl Conf Molten slags and fluxes,
Sendai, 1992 (ISIJ, Tokyo, 1992) p. 421.
226. Y Kim, K Morita, ISIJ Intl., 53, 2077, (2013).
227. KC Mills, M Susa, V Ludlow, Proc. 13th Proc. Tech. Conf., Nashville, TN, 1995 (ISS,
Warrendale, PA, 1995) p. 157.
228. T El Gammal, U Schoneberg, Stahl u Eisen, 112 (1), 45, (1992).
229. F Neumann, J Neal, MA Pedroza, AH Castillejos, FA Acosra, Proc 79th Steelmaking Conf.
(1996) (ISS, Warrendale, PA., 1996) p 249.
230. SG Bratchikov, GF Konvalov, MV Frolov, Teplotekh. Protsessov Vplavski Stahli Splavov,
1977 (5), 130, (1977).
231. M Ichihashi, M Kawamoto, Y Tsukaguchi, N Nishida, T Kanazawa, S Hiraki, Tetsu to
Hagane, 81, (10), 1132, (1995).
232. S Diehl, JA Moore, RJ Phillips, Proc. 78th Steelmaking Conf., Nashville, TN, (1995) (ISS,
Warrendale, PA., 1995) p. 351.
233. S Ozawa, R Endo, M Susa, Tetsu-to Hagane, 93, 8, (2007).
234. H Hasegawa; T Kowasari, Y Shiroki, H Shibata, H Ohta, Y Waseda, Met. Mater. Trans. B,
43 B, 1411, (2012) see also H Ohta H Shibata: J Manufact. Sci and Production, 13, 115
(2013).
235. A.Fluegel: http//glassproperties.com/thermal –conductivity/ (2011).
236. K Tsutsumi, KWatanabe, M Suzuki, M Nakada, T Shiomi, Proc.7th Intl. Conf. Molten slags,
fluxes and salts, Capetown, South Africa, 2004 (SAIMM, Johannesburg, 2004) p. 803.
237. H Nakato, S Takeuchi, T Fujii, T Nozaki, M Washio, Proc 74th Steelmaking Conf.,
Washington, DC (1991) (ISS, Warrendale, PA, 1991) p. 639.
238. Yongsug Chung, PhD Thesis, “Surface and interfacial phenomena in iron-alloys” Carnegie-
Mellon Univ., Pittsburgh, PA (1999).
239. T Tanaka, M Nakamoto, J Lee, Proc. Conf. Metal Separation Technology, Copper
Mountain, CO. (2004) ed. by R E Aune and M Kekkonen (Helsinki Univ. Technol, 2004).
p. 135.
390 9 Properties of Mould Fluxes and Slag Films

240. Y Chung, AW Cramb, Met. Mater. Trans. B, 31B, 957, (2000).

241. KC Mills, BJ Keene, RF Brooks, A Shirali, Phil. Trans. Roy. Soc, London, A 356, 911,
(1998).
242. H Gaye, D Lucas, M Olette, PV Riboud, Canad. Met. Quart., 23, 179, (1984).
243. PV Riboud, LD Lucas,: Canad. Met. Quart., 20, 199, (1981).
244. A Jakobsson, M Nasu, H Manguiro, KC Mills, S Seethraman, Proc. Marangoni and
interfacial phenomena in Materials Processing, Royal Soc. London, ed. by ED Hondros, M
McLean and KC Mills, (IOM, London, 1998) p. 181.
245. AW Cramb, Proc.5th Intl. Conf. Molten slags, fluxes and salts, Sydney, 1997, (ISS
Warrendale, PA, 1997) p. 35.
246. M Ferrari, L Liggieri, F Ravena, C Amodio, A Passerone R Miller, J Colloid Interface Sci.,
186, (1997) 40/45 and 46/52.
247. M Wegener, L Muhmood, S. Sun, AV Deev: Met. Mater. Trans B,46B, 318, (2015).
248. VI Yavoisky, YM Nechkin, IV Zinkovsky, VG Padalka, Sb. Mosk Inst Stali Splavov, 74,
87, (1973).
249. AM Yakushev, VM Romashin, NV Ivanova, Steel USSR, 15, 425, (1985)..
250. J Krusina, T Mylsevic, Kovove Mater., 16 (4), 510, (1978).
251. J Elfsberg, T Matsushita, Steel Res. Intl., 82 (4) p. 404.
252. A Kusano, N Sato, M Okimori, S Fukunaga, K Nishihara, M Sato, Y Minigawa. Proc. 74th
Steelmaking Conf. (1991) (ISS, Warrendale, PA, 1991) p. 111
253. Y Cheng, Y Wang, D Li, H Miao, Continuous Casting, 2008 (4). 42,
254. T Duberstein HP Heller, P R Scheller: Determination of the thermophysical properties of
molten slags by maximum bubble pressure. www.pyrometallurgy.co.za/MoltenSlags2012/
W180pdf.
255. GF Konovalov, Izv. VUZ Chern Met., 1974 (2), 15, (1974).
256. R Hagemann, HP Heller. S Lachmann, S Seetharaman, PR Scheller, Ironmaking and
Steelmaking, 39, 508, (2014).
257. KC Mills, BJ Keene, Intl. Materials Rev., 22, 1, (1987) and Proc. Symp. Mineral matter and
ash in coal, Philadelphia, PA (1984) ACS Monograph Series 301 (1986) p. 197
258. M Nakamoto, A Kiyose, T Tanaka, L Holappa, M Hamalainen, ISIJ Intl., 47,38, (2007).
259. M Hanao, T Tanaka, M Kawamoto, K Takatani, ISIJ Intl., 47, 935, (2007).
260. M Nakamoto, T Tanaka, L Holappa, M Hamalainen, ISIJ Intl., 47, 211, (2007).
261. KC Mills, S Karagadde, L Yuan, PD Lee, F Shahbazian, ISIJ Intl., 56, 264, (2016).
262. R Olivares, MP Brungs, H Liang, Met. Trans. B, 22B, 305, (1991).
263. T Matsushita, T Ishikawa, PF Paradis, K Mukai, S Seetharaman, ISIJ Intl., 46, 606, (2006).
264. P Quested, RF Brooks, AP Day, KC Mills, Proc. Conf. on Fluid flow phenomena in
Materials Processing, San Diego, CA, 1999.
265. Y Bottinga, D Weill, Amer. J Sci., 272, 438, (1972).
266. M Persson, T Matsushita, J Zhang, S Seetharaman, Steel Res. Intl., 78, 102, (2007).
267. M Persson, J Zhang, S Seetharaman, Steel Res. Intl., 78, 290, (2007).
268. A Fluegel, J Amer. Glass Technol., 90 (8), 2622, (2007).
269. A Fluegel, Thermal expansion calculation on silicate glasses at 210 °C based on systematic
analysis of global databases., http//glassproperties.com.
270. A Fluegel, DA Earl, AK Varshneya, TP Seward III, Phys. and Chem. Glasses,
Europ. J Glass Sci. and Technol., Part B, 49 (5), 245, (2008).
271. I K Priven: Glass Technol., 45 (6), 244, (2004).
272. JF Stebbins, ISE Carmichael, IK Moret, Contrib. Mineralogy and Petrology, 86, 131, (1984).
273. DA Ditmars, in Compendium of Thermophysical property measurement methods; Volume 1:
Survey of Measurement techniques, ed. by KD Maglic, A Cezairliyan and VE Peletsky,
(Plenum Press, New York, 1984) Chapter 13, p 527.
274. MJ Richardson, in Compendium of Thermophysical property measurement methods; Volume
1 Survey of Measurement techniques ed. by KD Maglic, A Cezairliyan and VE Peletsky,
(Plenum Press, New York, 1984) Chapter 17, p. 669.
275. Factsage: website www.factsage.com: for use see for example - reference [131]
References 391

276. MTDATA; website www.ntech.npl.co.uk- see J Robinson, AT Chapman, Pn Quested, BJ

Monaghan, J Gisby and KC Mills: Proc. 2nd Intl. Conf. Science and Technol. of
Steelmaking, Swansea, 2001 (1OM, London, 2001) p. 149.
277. Thermocalc: website www, thermocalc.com
278. SH Firoz, R Endo, M Susa, Ironmaking Steelmaking, 34, 437, (2007).
279. JF Holzhauser, KH Spitzer, K Schwerdtfeger, Final Rept. ECSC Contract 7210 CA/ 137
(1988)(Europ. Comm. Sci. Publ., Luxemborg, 1988).
280. J Diao, B Xie, JP Xiao, Ironmaking and Steelmaking, 36, 610, (2009).
281. J Chen, J Xie, F He, E Wan, Z Xu, C Yang, G Liu, Ironmaking and Steelmaking, 42, 126,
(2015).
282. A Yamauchi, K Sorimachi, T Sakuraya, T Fujii, ISIJ Intl., 33, 140, (1993). See also Tetsu-to
Hagane, 79 (2), 167, (1993).
283. Y Seko: BSc Thesis Tokyo Inst Technol (2006).
284. KC Mills: ISIJ Intl. 56, 1, (2016).
285. M Kawamoto, T Watanabe, T Ikeda, Tetsu-to Hagane, 81(12), 1132, (1995).
286. Y. Meng, B.G. Thomas, Proc. ISS Tech. Conf. (2003), Indianapolis, see also Met. Mater.
Trans., 34B, 685 and 707 (2003).
287. T Matsushita, T Watanabe, M Hayashi, K Mukai, S Seetharaman; Intl. Mater. Rev., 56, 287,
(2011).
288. H Ehrenburg, Proc. 3rd Europ. Conf. Continuous casting, Madrid, 1998 (UNESID, Madrid,
1998) p. 645.
289. G. Skoczylas, Proc. 79th Steelmaking Conf., 1996, (ISS, Warrendale, PA,1996) p. 269.
290. A Kiyose, K Miyazawa, W Yamada, K Watanabe, H Takahashi, ISIJ Intl., 36, S 155, (1996).
291. D Gotthelf, P Andrezjewski, E Julius, H Haubrich, Proc. 3rd Europ. Conf. Continuous
Casting, Madrid (1998) (UNESID, Madrid, 1998) vol 2, p. 825.
292. H Sharan, AW Cramb, Met. Mater. Trans. B, 26B, 318, (1995) see also., ISS Trans., 15, 95
(1995).
293. Era Kapilsharami, V Sahajwalla, S Seetharaman, Ironmaking and Steelmaking, 31,509,
(2004).
294. T Matsumiya, ISIJ Intl., 46, 1800, (2006).
295. K Miyazawa, Sci. Technol. Adv. Mater., 2, 59, (2001).
296. H Yamamura, T Kajitani, J Nakashima, M Yamasaki, S Mineta, Nippon Steel Technical
Report, 104, 54, (2013).
297. E Favre, S Kunstreich, MC Nove, D Rotelec, W Courths, E Korte, Proc 3rd Intl Conf.
Continuous Casting, Madrid, 1998 ((UNESID, Madrid,1998) p. 595.
298. H Take, H Osanai, J Hasunuma, T Yamamoto, H Bada, H Tozawa, Proc. Conf. on Quality of
ordinary steel in Iron- and steel-making, Bangkok, (1994) Session 3, Paper 1.
299. P Gardin, JM Galpin, MC Regnier, JP Radot, IEEE Trans. On Magnetics, 31, 2088, (1995).
300. D W van der Plas, C Platvoet, B Diesesme, JP Radot, JM Galpin: Proc. 2nd Europ. Conf.
Continuous Casting, Dussseldorf, 1994 (Stahl u Eisen, Dusseldorf, 1994), p. 109.
301. M Washio, M Sugizawa, S Moriwaki, K Kariyaa, S Idogawa, S Takeuchi, Revue de
Metallurgie, CIT, 90 (April), 507, (1993).
302. R Koldewijn Unpublished Corus Internal Rept (2007) cited in KC Mills, J Kromhout, A
Hamoen, R Boom, Proc. of Admet Conf., Dnipetrovsk, Ukraine, May 2007(Natl. Metall.
Acad. Ukr., Dnipropetrovsk, 2007), vol 2 p. 174.
303. SG Kollberg, HR Hackl, PJ Hanley, Iron and Steel Engineer, 73 (7), 24, (1996).
304. E Takeuchi, JOM, 1995 (May), 42, (1995).
305. E Takeuchi, H Harada, H Tanaka, T Ishii, T Toh, M Zeze, M Hojo, K Shigematsu, Nippon
Steel Tech. Report, 61, 29, (1994).
306. J Kromhout, RS Schimmel, Proc. 8th Europ. Conf. Continuous casting, Graz, Austria, June,
2014. (Austrian Met. Mater. Soc., 2014).
307. M Tani, T Toh, K Umetsu, K Tanaka, M Zeze, K Tsunenari, K Hayashi, S Fukunaga:
Nippon Steel Tech. Report, 104, 62, (2013).
308. P Rocabois, J Lehmann, C Gatellier, JP Teres, Ironmaking and Steelmaking, 30, 95, (2003).
392 9 Properties of Mould Fluxes and Slag Films

309. PV Riboud M Olette, J Leclerc, W Pollak, Proc. 61st Conf NOH- BOS Conf., Chicago, IL,
1978, (ISS, Warrendale, PA, 1978) p. 411.
310. T Emi, Proc. 74th Steelmaking Conf., Washington, DC, 1991, (ISS Warrendale, PA, 1991)
p. 623.
311. M Bruhl, J Sardemamm, F Oeters. Proc. Scanjet VII part II 7th Intl Conf. Refining Processes
(1996) p. 171.
312. X Yu, RJ Pomfret, KS Coley, Met. Trans B, 28B, 275, (1997).
313. AB Fox, ME Valdez, J Gisby, RC Attwood, PD Lee,S Sridhar, ISIJ Intl., 44, 836, (2004).
314. P Misra, S Sridhar, AW Cramb, Met. Mater. Trans. B, 32B, 963, (2001).
315. H Lei, Y Zhao, DQ Geng, ISIJ Intl., 54, 1629, (2014).
316. K Koyama, K Nagano, Y Nagano, T Nakano, Nippon Steel Tech. Report, 34, 41, (1987).
317. T Araki et al. Proc. 5th Japan-USSR Joint Symp. On physical chemistry of metallurgical
processes. Publ. ISIJ Special Report No 22, 263, (1975).
318. YQ Liu, LJ Wang, KC Chou, ISIJ Intl., 54, 728, (2014).
319. JL Li, QF Shu, YA Liu, KC Chou, Ironmaking and Steelmaking, 41, 732, (2014).
320. VS Esaulov, GF Konovalov, SI Popel, Izv VUZ Chern Met., 1976 (8), 36, (1976).
321. M Valdez, K Prapacom, AW Cramb, S Sridhar,. Steel Research, 72, 291, (2000).
322. L Teng: KTH Report “Inclusion absorption capacity of mould flux slag for Ingot casting”
(IPTINGOT) Dept. Materials Science and Eng. KTH Stockholm (2013).
323. P Scheller, Proc. Mills Symp., London (2002) (IOM, London, 2002) p. 477.
324. P Scheller, Proc. 3rd Europ. Conf. Cont. Casting, Madrid,1998 (UNESID, Madrid,1998)
p. 797.
325. RE Fash, WF Salem, DC Evans, Proc. 74th Steelmaking Conf., Washington, DC, 1991. (ISS,
Warrendale, PA, 1991).
326. R Branion, DA Dukelow, GD Lawson, J Schade, M Schmidt, HT Tsai, Proc. 78th
Steelmaknig Conf. (1995) (ISS, Warrendale, PA,1995) p. 647..
327. MJ Frazee, Proc. 78th Steelmaking Conf. (1995) (ISS, Warrendale, PA, 1995) p. 639.
328. NT Mills, BN Bhat, Iron and Steelmaker, 1978 (Oct), 18, (1978).
329. KH Spitzer, et al., Stahl u Eisen, 108, 441 (1988).
330. KH Spitzer, K Schwerdtfeger: “Investigation of the isolating properties and melting
behaviour of mould powders for continuous casting”. Tech. Univ. Clausthal Report 7249
(Dec 1989).
Chapter 10
Selection of Mould Fluxes and Special
Mould Fluxes for Continuous Casting

Abstract The various factors influencing the selection of mould powder

compositions are discussed in this chapter. Mould powders contain two types of
component (i) the minerals which form the slag film and (ii) Carbon which controls
the melting rate. Both are vital to the successful performance of the mould powder.
The composition of the mould powder is determined by the mould dimensions, the
casting conditions and the steel grade being cast. Empirical rules have been
developed to provide the required values of the powder consumption (Qreq s ), the
viscosity (ηreq) and the break temperature (Treqbr ) for satisfactory casting (i.e. free
from longitudinal cracking and sticker break-outs). More than 85% of mould
powders conform to these empirical rules. However, other factors can also influence
the powder selection process. These include

(i) Increasing the slag viscosity or the interfacial tension to reduce slag
entrapment.
(ii) Increasing the interfacial tension to minimise scaling of the steel surface.
(iii) Minimising carbon pick-up (especially for LC and ULC steels).
(iv) Providing both support in the mould and optimal heat transfer when casting
round billets.
(v) Providing reasonable heat transfer in mould configurations with corners
acting as heat sinks.
(vi) Handling the large amounts of Al2O3 formed when casting high-Al steels
whilst maintaining good lubrication throughout the casting.
(vii) Handling Ti-stabilised steels which form either TiN or Ti(CN) in the slag
pool (which has low solubility in mould slag) or CaO
TiO2 (which has a
high Tliq).
(viii) Reducing the fluoride content of slag to minimise environmental and health
concerns.
All of these problems are discussed in detail and possible solutions are outlined.

© Springer International Publishing AG 2017 393

K.C. Mills and C.-Å. Däcker, The Casting Powders Book,
DOI 10.1007/978-3-319-53616-3_10
394 10 Selection of Mould Fluxes and Special Mould Fluxes …

Symbols, Abbreviations and Units

DC Diameter of carbon particles (m)
f Frequency (Hz or cpm)
f* fraction of powder forming slag
Qs Powder consumption (kgm−2)
q Heat flux density (Wm−2)
R* (Surface area/Volume)—mould (m−1)
s stroke (m)
T Temperature (°C)
Tbr Break or Solidification temperature
t Thickness of mould (m) or time (s)
Vc Casting speed (m min−1)
w Width of mould (m)
η Slag viscosity (dPas)
q Density (kg m−3)
EMBr Electromagnetic braking
IR Infrared radiation
LC Low-carbon steel
MC Medium-carbon steel
SEN Submerged entry nozzle
ULC Ultra-low-carbon steel

In chemical formulae
A Al2O3
C CaO
F FeO
Fl CaF2
M MgO
S SiO2
T TiO2

10.1 Introduction

It is obviously important to select the best mould powder for the given casting
conditions in order to obtain problem-free casting and the minimum of defects in
the steel product. Mould powder components can be divided into (i) the mineral
components (oxides and fluorides) and (ii) carbon particles. The mineral compo-
nents are the most important since they form the mould slag and, subsequently, the
slag film which largely determines the performance of the mould powder. The
carbon particles play a minor role, i.e. in controlling the melting rate of the powder;
10.1 Introduction 395

however, it is absolutely essential that the melting rate matches the demand for
liquid slag. Consequently, the selection of the carbon content is also important.
It is shown below that selection of the mineral components is dependent upon
the casting conditions and the steel grade being cast. Much of our present
knowledge is based on “ad hoc” research where compositional changes were made
and the results observed. Mould powders giving a good performance were obtained
in this way and it was only later that the logic underlying the selection of mould
powder composition was discovered. Certain empirical rules were deduced for the
selection of mould powders. However, with our expanding knowledge of how
mould slags work, there is evidence that scientific knowledge is being used to
develop efficient mould powders (such as in the development of slags to cast both
high-Al steels and round billets which are described below).
Using the empirical rules developed to predict slag properties required for the
given casting conditions and steel grade, it was found that 85–90% of the experi-
mental values were consistent with the predicted slag properties. The other 10–15%
usually does not fit because they have been modified to carry out further duties, e.g.
reduce either slag entrapment or SEN erosion rates. It has always been the case that
mould powders are expected to be “flexible” or “forgiving”, i.e. will help to alleviate
problems caused by the casting conditions imposed. These special cases are iden-
tified below along with the special slags developed for certain conditions.

10.2 Selection of Mineral Compositions of Mould Powder

for Given Casting Conditions

The molten mould slag must perform the following functions satisfactorily:
• It must form a liquid slag pool which seals off the molten steel from the gaseous
atmosphere and thereby prevents oxidation and nitration of the steel.
• The powder bed must provide thermal insulation to prevent freezing of the steel
meniscus (i.e. skull formation).
• The liquid slag pool should absorb inclusions and gas bubbles from the steel.
• The liquid layer of the slag film formed between the shell and the mould must
be sufficiently thick to provide good lubrication of the newly formed shell.
• The solid layer of the slag film must reduce the horizontal heat flux to a suitable
level for the steel grade being cast.
Although all of these five tasks are important, the selection of mould powder
compositions is mainly determined by the latter two conditions. It should always be
remembered that mould powders are selected for the effect that they have on the
newly formed shell. The principal factors affecting the selection of mould powder
compositions and properties are
396 10 Selection of Mould Fluxes and Special Mould Fluxes …

• The mould geometries selected for the casting.

• The casting conditions used in the casting.
• The type of steel being cast.

10.2.1 Effect of Mould Geometry on Mould Powder

Selection

The mould geometry affects the level of lubrication required by the newly formed
shell.
It is customary to classify the mould dimensions into four main classes
– Billets.
– Blooms.
– Slabs.
– Thin slabs.
In addition, there are (i) round billets (ii) beam blanks and (iii) hollow billets (for
pipe production) [1] which all have their own specific demands.
The liquid friction forces (Fl) acting on a specific face of the shell tend to
increase with increasing distance from the corner (Fig. 10.1) [1–3]. Thus the fric-
tional forces acting at the centre-line of slabs tend to increase with increasing slab
width (w) [2, 3]. The powder consumption, Qs (in kgm−2) is frequently used as a
measure of the lubrication supplied to the shell. Since the frictional forces increase
with increasing width, it follows that more lubrication must be supplied as the width
increases. Consequently, the required powder consumption, Qreq s increases with

Fig. 10.1 Diagram showing

centre-line friction increasing
with increasing distance from
the corner; X = friction at
centre-line for 3 mould
geometries; solid line billet;
dotted line bloom and solid
line slab; (courtesy AB Fox
[3])
10.2 Selection of Mineral Compositions of Mould Powder … 397

0.8
Billets

Powder Consumption, Q ccorr (kg m 2)

0.7
Blooms

Slabs
0.6
Round
0.5 High Speed Billet
[4] -0.04R
: Qs =0.44e
0.4
: Qs=2/(R-5)

0.3 Thin Slabs

0.2

0.1

0
0 10 20 30 40 50
-1
Surface Area to Volume Ratio, R (m )

Fig. 10.2 Required powder consumption, Qreq s , as a function of the parameter, R* for different
mould geometries; ◊ = billets; o = blooms; D = slabs; ▬ = thin slabs; + = Rounds; X high-speed
billets; dotted line equation, Qs = 0.44 exp(−0.04R*) [6] solid line (Qs = 2/(R* − 5)) [4, 5].
Typical ranges of R* values; slabs (8–15) blooms (12–16) billets (20–30) thin slabs (25–40);
(permission granted, UNESID [5])

increasing width (w) of the wide face. Furthermore, Qreq s values are in the hierarchy
slabs > blooms > billets as can be seen in Fig. 10.2 [4, 5].
Although relationships have been derived for Qreq s as a function of the distance
from the centre-line and corner (dcent-corner) [3], it is customary to use the parameter,
R* rather than (dcent-corner) [6]. The parameter, R* represents the ratio {surface
area/volume} of the mould and is calculated by Eq. 10.1 where w and t are the
width and thickness of the mould. For round billets, R* = 2/r, where r is the radius
of the mould.

R ¼ 2ðw þ tÞ= ðw:tÞ ð10:1Þ

The mould powder contains both carbon [which is oxidised to CO (g) or CO2 (g)]
and carbonates (which form CO2 (g) during the heating process in the powder bed).
Consequently, not all the powder forms slag and it must be remembered that it is the
slag which provides the lubrication of the shell. Consequently, it is necessary to
calculate the fraction of the powder forming slag (f*) using Eq. 10.2 where %Cfree,
%Ctotal are free and total C contents and LOI is the loss on ignition. Thus, it is
necessary to differentiate between the powder and the slag which are denoted by
superscripts powd and slag, respectively
398 10 Selection of Mould Fluxes and Special Mould Fluxes …

f  ¼ 0:01 f100  %Cfree  ð44=12Þ ð%Ctotal  %Cfree Þg ¼ ð100  %LOIÞ=100

ð10:2Þ

Powder consumption can be expressed in various ways, the parameter Qt is the

powder consumption measured on plant with units of kg (powder) tonne (steel)−1
and represents the amount of mould powder used per tonne of steel. Another
powder consumption term is the melting rate, QMR (in kgs−1 or kg min−1) which is
a measure of the rate of slag supply. It is essential that the actual melting rate
should match the required powder consumption, QreqMR. These various terms (Qt, Qs
and QMR) can be calculated by Eqs. 10.3–10.4, respectively where q is the density
and dl is the thickness of the molten slag layer.

Qslag
s ¼ f  7:6
Qt = R ¼ qdl ð10:3Þ

1


Qslag
MR kgs ¼ 2 ðw þ tÞ
Qs : Vc = 60 ð10:4Þ

It can be seen from Fig. 10.2 that Qslag

s is a function of the parameter R* and that
wide slabs (with R* = ca.10) require considerably more powder consumption (i.e.
lubrication) than billets (with R* = ca.30). The relationship between Qslag
s and R* is
shown in Fig. 10.2 and can be represented by Eq. 10.5 [4, 5]; it is also compared
with an earlier relationship reported by Neumann et al. [6] (i.e. Qslags = 0.44 exp
(−0.04R*) in this figure.

Qslag
s ¼ 2=ðR  5Þ ð10:5Þ

Thus it is possible to calculate the required powder consumption (Qslag

s ) for the
given mould dimensions.

10.2.2 Effect of Casting Conditions on Mould Powder

Selection

The powder consumption, Qslag s , is also affected, by factors other than the mould
dimensions, principally by the casting speed and the viscosity (η) and, to a lesser
extent, by the frequency (f) stroke (s) and the break temperature (Tbr).
Empirical relationships have been reported for the required powder consump-
tion, Qreq
s as functions of casting speed and viscosity, such as those shown in
Eqs. 10.6–10.8 [3, 7–9].

Wolf [7]:

s req ¼ 0:7=g
Qpowd
Vc ð10:6Þ
0:5
10.2 Selection of Mineral Compositions of Mould Powder … 399

Ogibayashi [8]:

s req ¼ 0:6=g
Vc
Qpowd ð10:7Þ

Subsequently, Eq. 10.5 was modified to provide a better fit with plant data [3, 9].

s req ¼ 0:55 =g
Qslag
Vc ð10:8Þ
0:5

The effect of other casting variables on the powder consumption has been
reported by a number of investigators; the reported relationships vary considerably
on the effect of the variable on the powder consumption [3]. A typical equation
involving f, s and Tbr is shown in Eq. 10.6 [10].
 
kb s0:4 1 1000Vc
Qs ¼ 1:6 0:5 cos ð10:9Þ
Tsol g Vc 2pfs

The effects of f, s and Tbr tend to be relatively small and thus empirical rules
involving only Vc and η work reasonably well. Thus, the required powder con-
s req) can be expressed in terms of a required viscosity, η
sumption (Qslag req
(at 1300 °C)
for the given casting speed, etc. using Eqs. 10.6–10.8. It should be noted that
mathematical models of the slag infiltration into the shell/mould channel indicate Qs
increases with increasing frequency whereas plant trials indicate the reverse
relationship. It has been suggested that increasing frequency increases the number
of periods of confused flow accompanying the changes in mould direction (little
slag infiltration occurs in these periods) and thus, increased frequency lowers
powder consumption [11]. It should also be noted that the amount of slag required
per tonne of steel is reasonably constant (f*Qt = 0.48 kg tonne−1 [3]) for most
casting but is not valid for high-viscosity, billet slags and cases where slag infil-
tration is restricted (e.g. casting of Ti-stabilised stainless steels).

10.2.3 Effect of Steel Grade on Mould Powder Selection

The type of steel being cast also affects the selection of the mould powder com-
position. When molten steels with C contents in the range 0.06–0.17% (denoted
MC steels) solidify, they form a d-Fe shell, which, subsequently, undergoes a
peritectic phase transition to c-Fe (austenite) (Fig. 10.3).

Fe ðliqÞ þ d  Fe ! c  Fe ð10:10Þ

The transition (L + d ! c) is accompanied by a volume change of (0.4–0.6%)

due to the tighter packing in the austenite (cf. ferrite) and there is also a 4%
mismatch between the thermal shrinkage coefficients of d-Fe and c-Fe [12]. These
changes give rise to wrinkled shells with considerable variations in shell thickness.
400 10 Selection of Mould Fluxes and Special Mould Fluxes …

Fig. 10.3 The Fe–C phase

diagram for the peritectic
range (0.08 < wt%C < 0.55)
[13]; drawn by
SWEREA/KIMAB

The mismatch in shrinkage coefficient, coupled with an uneven shell, creates an

accumulation of stresses which are relieved by longitudinal cracking. The strategy
used to combat longitudinal cracking involves the creation of a thin, uniform shell
and this is achieved using a thick, crystalline, slag film. A thick slag film is obtained
by producing a mould slag with a high solidification (or break) temperature (Tsol or
Tbr) and this is achieved by increasing the slag basicity (C/S) up to 1.3, if necessary.
Increased basicity also promotes further crystallisation in the slag film.
High-carbon (HC) shells tend to be weak and thus thin shells are prone to sticker
break-outs and hot tears. Consequently, thick shells are needed to provide the
necessary strength. This is achieved by promoting high heat transfer between the
shell and mould by use of a glassy slag film with a low Tbr; i.e. the reverse of the
practice used for MC steels prone to longitudinal cracking. Slags used to cast HC
steels tend to have a low basicity.
Consequently, when selecting a mould powder it is necessary to differentiate
between MC, HC and other steels on the basis of steel composition.

10.2.4 Routines to Differentiate Between Steel Grades

Several routines have been explored to differentiate those steels undergoing the
peritectic reaction [14–16]. In plain carbon steels the peritectic range lies between
0.06 and 0.17%.
Wolf [17, 18] introduced the term “ferrite potential” as a measure of the amount
of peritectic reaction occurring in the steel. Although the ferrite potential is affected
by the carbon content, it is also affected by other alloying elements; some of these
10.2 Selection of Mineral Compositions of Mould Powder … 401

elements stabilise the ferrite (Cr, W, Mo, Al and Si) whilst others stabilise the
austenite (Ni, Mn, Co, N and Cu). For low-alloy steels, the ferrite potential (FP) can
be calculated by Eq. 10.11 where CP is the carbon potential which is defined in
Eq. 10.12.

Low alloy

FP ¼ 2:5ð0:5  CPÞ ð10:11Þ

CP ¼ ðwt%CÞ þ 0:04ð%MnÞ þ 0:1ð%NiÞ þ 0:7ð%NÞ  0:14ð%SiÞ

ð10:12Þ
 0:04ð%CrÞ  0:1ð%MoÞ  0:24ð%TiÞ

For stainless steels the FP is given by Eq. 10.13 where “Ni” and “Cr” are
defined in Eqs. 10.14 and 10.15, respectively, [17, 18].

Stainless steel
 
FP ¼ 5:26 0:74 “ Ni” =“ Cr ” ð10:13Þ

“
Ni” ¼ ðwt%NiÞ þ 0:31ð%MnÞ þ 22ð%CÞ þ 17:5ð%NÞ ð10:14Þ
“
Cr ” ¼ ðwt%CrÞ þ 1:5 ð%SiÞ þ 1:65 ð%MoÞ þ 2ð%NbÞ þ 3ð%TiÞ ð10:15Þ

The peritectic reaction is at a maximum at FP = 1.0 and the range,

FP = (0.8–1.05) corresponds to the crack-sensitive range (Fig. 10.4).
Other investigators have reported coefficients which differ from those given in
Eq. 10.12, these are summarised in [14–16] (Table 10.1).

Fig. 10.4 The tendencies of steels towards longitudinal cracking (depression) and sticker
break-outs as functions of the ferrite potential [17, 18] (permission granted, VDEh, Verlag
Stahleisen, [18]; re-drawn Swerea/Kimab)
 402

Table 10.1 Coefficients for selected equations to calculate Carbon potentials (after [14]); Alt = total Al
Ref. Mn Si N P S V Ti Cu Cr Ni Mo Others
10

Wolf 0.01 −0.1 −0.7 −0.04 0.04 −0.1

Trico 0.02 0.009 0.05 0.008 0.17 0.009 0.007 0.007 0.003 0.02 −0.007 Alt 0.05
Nb 0.04
Sn 0.0006
B 1.32
Ca −0.24
BSSTC 0.043 −0.14 1.06 0.029 0.11 −0.13 −0.024 0.037 −0.083 0.1 −0.003
SMS 0.14 −0.037 −0.04 0.222 0.003 0.023 −0.004
Howe 0.04 −0.14 0.7 −0.24 −0.04 0.1 −0.1
Selection of Mould Fluxes and Special Mould Fluxes …
10.2 Selection of Mineral Compositions of Mould Powder … 403

10.2.5 Plots of Tbr as a Function of Slag Viscosity

Figure 10.5 shows a plot of Tbr as a function of slag viscosity for a number of
mould slags used to cast MC and HC steels along with low and intermediate C
steels (LC and IC) [5, 9]. It can be seen that slags used for MC, crack-sensitive
steels fall on the upper curve, those used with HC steels on the lower curve and
those for LC and IC steels tend to fall between the two curves. Such plots represent
inverse lubrication (viscosity) as a function of inverse heat transfer (Tbr) and thus,
indicate the optimum properties for casting. It should be noted that the heat transfer
is also dependent upon the fraction of crystalline phase (fcrys) in the slag films. This
type of plot works because slag basicity affects both Tbr and fcrys in the same way
(i.e. the heat flux decreases).
The data shown in Fig. 10.5 were expressed as three curves (Fig. 10.6) and
Eqs. (10.16–10.18); these can be used to calculate the required break temperature
(Treq req
br ) from the required slag viscosity (η ) for three different steel grades.

Crack-sensitive

CP ¼ 0:06  0:18 Tbr ð CÞ ¼ 1157 þ 60 ln greq ð10:16Þ

Sticker-sensitive

CP [ 0:4 Tbr ð CÞ ¼ 1051 þ 76:4 ln greq ð10:17Þ

Others (intermediate)

CP 0  0:06 and 0:18  0:4 Tbr ð CÞ ¼ 1103 þ 68:5 ln greq ð10:18Þ

Fig. 10.5 The break temperature (Tbr) as a function of slag viscosity for slags used in the casting
of slabs, blooms and billets; (permission granted, UNESID [5])
404 10 Selection of Mould Fluxes and Special Mould Fluxes …

Fig. 10.6 The required break 1300

temperature (Treqbr ) as a
Thin Slabs Crack
function of required viscosity

Break Temperature (oC)

(ηreq) for different steel 1200 Middle
grades; upper curve
crack-sensitive (MC) steels;
lowest curve sticker-sensitive 1100
(HC) steels; middle curve
other steels; (courtesy of AB Sticker
Fox [3]) 1000

900
0 0.5 1 1.5 2 2.5 3 3.5 4
Viscosity at 1300C (dPas)

Thus, the required powder consumption can be expressed in terms of the

required values for Qs, viscosity and break temperature. As mentioned above, it was
found that 85–90% of the plant data conformed with the calculated values of Qreqs ,
ηreq and Treq
br [9].

10.2.6 Other Casting Conditions Affecting Powder

Consumption

There are reasons why the above empirical rules do not apply to 10–15% of the
mould powders. These exceptions fall into two classes
(i) Where slag compositions are deliberately altered in order to combat other
problems, e.g. by increasing viscosity (thereby reducing Qs) to tackle high
levels of slag entrapment or SEN erosion.
(ii) Where other conditions (e.g. metal and Ar flows) affect the powder con-
sumption, etc.
The first class of exceptions is discussed in Sect. 10.3. With regard to the second
group, there are a number of conditions which affect powder consumption, these are
Steels containing Titanium
TThe Ti in these steels reacts with nitrogen and carbon in the bed to form Ti(CN)
particles which have a low solubility limit in the slag and thus remain largely as Ti
(CN) particles in the slag pool. These particles agglomerate and block the entrance
to the mould/shell channel (see Fig. 2.16) and thereby reduce the powder con-
sumption [19] and also increase the slag viscosity; thus, Qs values tend to be low
when casting these steels.
Metal flow
Electromagnetic braking (EMBr) is frequently used to deal with the effects of
turbulent metal flow in the mould. One consequence of applying EMBr is that it
10.2 Selection of Mineral Compositions of Mould Powder … 405

Fig. 10.7 Powder 0.8

consumption, Qt (in kg
tonne−1) as a function of Ar

-1
, kg tonne
flow rate (courtesy of AB Fox 0.6
[21])

powd
0.4

Qt
0.2
0 1 2 3 4 5
Ar flow rate, l min-1

reduces the vertical heat transfer [20] which leads to shallow slag pool depths and
low powder consumption. There are a number of possible remedies for dealing with
these conditions, e.g. (i) using a deeper powder bed (ii) incorporating exothermic
agents into the mould powder (iii) reducing the C% in the powder.
Argon flow
It can be seen from Fig. 10.7 that the powder consumption, Qt is affected by the
argon flow rate [21]. This is thought to be due to the increased vertical heat transfer
resulting from the gaseous convection which promotes a deeper pool. However, at
higher flow rates (>5 litre min−1) the Ar flow causes a switch in the direction of the
metal flow (double roll to single roll) and higher vertical heat flux values [22, 23].

10.3 Selection of Carbon Components of Mould Powders

As mentioned above, the principal function of the carbon particles is to control the
melting rate. Carbon is non-wetting to molten slag and thus hinders the agglom-
eration of molten slag globules and the formation of a molten slag pool (Fig. 10.8).
However, the carbon particles are gradually oxidised by both O2 (g) and CO2
(g) present in the bed and eventually, the liquid slag forms a molten slag pool.

Fig. 10.8 Schematic diagram

showing non-wetting carbon
particles (•) preventing the
agglomeration of liquid slag
globules (O)
406 10 Selection of Mould Fluxes and Special Mould Fluxes …

The melting rate (QMR) must match the required powder consumption (Qreq s ).
Several factors affect the melting rate with QMR increasing (QMR ") as (i) %C
decreases (ii) the mean size of Carbon particles increases (DC ") (iii) the bulk
density of the powder decreases (qbulk #) and (iv) as the carbonate content increases
(%CO32− ") [24]. The melting rate is usually controlled by the carbon content and,
to a lesser extent, by the size of the carbon particles. Different types of carbon (e.g.
lamp black or coke breeze tend to have small and large diameters, respectively) and
powder manufacturers obtain the appropriate mean particle size by blending the
various forms of carbon.
Since the carbon is dependent upon both C% and DC and manufacturers rarely
provide detailed data for carbon size distribution, thus it is difficult to derive
empirical rules in a similar way to that for the mineral content. Plots correlating
melting rate (QMR) with just one carbon parameter (either Cfree% or Ctotal%) were
found to show significant scatter [3].
However, care must be taken when reducing the C% since carbon is a fuel
(exothermic) and if the reduction of C% is too stringent the steel meniscus can
freeze and form a skull. Carbon-free powders have been developed and are avail-
able commercially [25–28].

10.4 Mould Powder Selection for Special Conditions

It was reported above that 85–90% of the mould powders in an extensive database
of plant data fitted the empirical rules used to cast described above [9], the 10–15%
of powders which do not conform with these empirical rules arise because the
mould powder is used to alleviate special conditions arising in the casting. These
special conditions are described below.
Reducing slag entrapment
The drive for higher production rates has resulted in increases in casting speed,
High casting speeds, in turn, have led to high metal flow velocities and turbulent
flow in the mould. Slag entrapment is principally caused by these turbulent flows.
There are several mechanisms which are responsible for slag entrapment [29–32].
The most effective way of reducing entrapment is by reducing the effect of the
turbulence, for instance, by introducing EMBr or by adjusting SEN immersion
depths to minimise the turbulence. However, some workers have chosen to change
the mould slag properties to minimise slag entrapment. Slag entrapment can be
reduced by increasing (i) the slag viscosity and (ii) the interfacial tension (cmsl)
between the metal and slag [29, 30].
Viscosity
A number of workers have reduced slag entrapment using mould powders with
higher viscosities. This practice does reduce slag entrapment but also reduces
powder consumption and consequently, should be used with caution. The slag
10.4 Mould Powder Selection for Special Conditions 407

Fig. 10.9 Viscosity as a

function of shear rate for
Newtonian (○; dashed line)
and non-Newtonian slags (●;
solid line) (permission
granted, ISIJ [34])

viscosity can be raised by (i) increasing SiO2 and Al2O3 content or by (ii) reducing
the network-breaking oxides (e.g. CaO) or the flux content (e.g. Na2O).
Tsutsumi et al. [33, 34] used an interesting approach involving non-Newtonian
slags. The viscosity of a non-Newtonian slag decreases at high shear rate
(Fig. 10.9). These workers argued that the shear rate was low at meniscus level
where slag entrapment occurs but was much higher in the region where slag
infiltration occurs. Thus, viscosity is high (7 dPas) at meniscus level but lower in
the infiltration region (5 dPas). Non-Newtonian slags contain about 0.2% Si3N4 and
form some Si–N bonds which are readily broken at high shear rates, thereby,
causing a decrease in viscosity.
Interfacial tension
High interfacial tension between the metal and slag tends provides some resistance
to slag entrapment. The interfacial tension (cmsl) can be calculated by Eq. 10.19
where u is an interaction coefficient, which frequently adopts a value of ca. 0.5.

cmsl ¼ cm þ csl  2uðcm
csl Þ0:5 : ð10:19Þ

The principal way of increasing the surface tension of the slag is to reduce the
content of the surface-active components of the slag (B2O3; K2O; Na2O; CaF2).
However, the surface tension of metal (cm) is the dominant term in Eq. 10.19
(cm = ca 4 csl). Consequently, the most effective way of increasing cm is to reduce
the S content of the steel but the reduction of B2O3, K2O, Na2O and CaF2 in the
slag will also help.
408 10 Selection of Mould Fluxes and Special Mould Fluxes …

10.4.1 Thin-Slab Casting

The parameter, R* given in Eq. 10.1 has a high value for thin slabs (e.g. 30–40)
which suggests that the required powder consumption (Qreq s ) is low like that for
casting billets. However, the frictional forces acting on the centre regions of the
wide faces of the thin slabs are almost certainly higher than the friction forces
exerted on billets and consequently, require more lubrication (i.e. powder con-
sumption) than that calculated from Eq. 10.5. In thin-slab casting, frequently the
depth of the slag pool is low and the powder consumption is well below the
calculated Qreq
s value (Fig. 10.10) [35]). In thin-slab casting, speeds are frequently
very high in order to maintain high production rates, Electromagnetic braking
(EMBr) is frequently used to minimise metal flow turbulence in the mould. The use
of EMBr results in a decrease in vertical heat flux [20] which causes a reduction in
slag pool depth; so frequently, slag pool depths are very low in thin-slab casting.
Consequently, the mould flux may be modified to (i) increase powder consumption
and (ii) increase the depth of the slag pool.
The powder consumption can be increased by decreasing the slag viscosity; but
in practice, the slag viscosities are already very low (because of the high casting
speeds used).
The depth of the slag pool can be increased by (i) improving the thermal
insulation of the mould and (ii) increasing the melting rate of the powder. The
thermal insulation is improved by using (i) deeper powder beds and (ii) exothermic
agents (e.g. Ca/Si or Fe/Si) in the powder.
The melting rate can be increased by (i) reducing the C% in the powder (%C #)
(ii) increasing the average particle size of the carbon particles (DC ") (iii) increasing
the carbonate content (%CO2 ") and (iv) reducing the bulk density of the powder
(qbulk #) [24]. The C content tends to be the dominant factor and a reduction in C%
does result in a deeper pool. However, some workers have reduced both C% and

Fig. 10.10 Powder 0.15

consumption as a function of
casting speed; •, solid line
measurements on thin slab
caster; dashed line values
10-2.QMR. mm s-1

0.1
predicted b equation
Qslag
s req = 0.55/η
0.5

Vc
(permission granted, Tata
Steel, re-drawn from [35]);
Tata Steel point out that the 0.05
results were obtained with an
experimental powder, thus,
the results shown may not be
typical for a regular thin-slab
powder 0
3.5 4.5 5.5
Vc, m min-1
10.4 Mould Powder Selection for Special Conditions 409

DC, simultaneously; the latter tends to work against any improvement obtained with
the reduction in %C. Furthermore, care must be taken to avoid reducing the carbon
content too low since Carbon is an exothermic agent and freezing of the steel
surface (“skull formation”) can occur if the C% is too low. In summary, a deeper
pool can be obtained by (i) using a deeper powder bed (ii) reducing the C% in the
powder.

10.4.2 Round Billets

It has been pointed out that when casting round billets, it is necessary to provide
good support to the shell all around the mould [36]. It is also necessary to control
the heat extraction rate in order to minimise longitudinal cracking; this is especially
so when casting peritectic, MC steels. This is usually achieved by using a powder
with a (C/S) ratio >1 to create a slag film containing crystalline phases. The
crystallisation causes a reduction in heat flux by (i) scattering of IR radiation and
(ii) the formation of an interfacial resistance (RCu/sl) and (iii) porosity in the slag
film resulting from the shrinkage accompanying crystallisation. The interfacial
resistance at the copper/slag interface is often referred to as “surface roughness” or
as a “gas gap”. The surface roughness between mould and slag results in poor
support for the round billet. Consequently, a different type of mould powder was
developed which produced a glassy slag film (i.e. containing no crystalline phases)
with no surface roughness [36]. The heat flux was reduced by adding transition
metal oxides (e.g. MnO) to the powder to absorb the IR radiation [36]. This powder
was reported to be successful in casting round billets [36]. Hanao et al. [37]
reported that a slag forming melilite (a solid solution ranging from akermanite
(C2MS2) to gehlenite C2AS) had proved successful in casting round billets; melilite
slags maintain a high fraction of glass in the slag film and tend to have higher
viscosities higher than those of conventional slags.

10.4.3 Mould Powder Selection for Moulds with Large

“Corner” Regions

Some mould geometries contain extensive corner regions, such as moulds used for
casting beam blanks [38] or where the “football” meets the regular section in funnel
—thin-slab moulds. These corner regions promote high heat transfer rates. Thus the
resulting shell is much thicker in these regions and shells exhibit considerable
variation in shell thickness around these “corner” regions. The variations in shell
thickness lead to stresses during the solidification process, especially when casting
peritectic, MC steels which, in turn, lead to longitudinal cracking in the corner
regions.
410 10 Selection of Mould Fluxes and Special Mould Fluxes …

Fig. 10.11 Schematic

diagram showing the thinning
of the shell in the corner
regions caused by strong
metal flows (Permission
granted, UNESID [39])

Most of the usual remedial actions do not involve mould powder selection (e.g.
reducing the sharpness of the corners, coatings, variable water-cooling for different
regions, etc.). However, the formation of a thick, crystalline slag film minimises the
variations in shell thickness and result in less longitudinal cracking. This can be
achieved by using a powder with a high break temperature which is usually
achieved by increasing the basicity of the mould powder.
A similar approach can be used (i.e. to increase Tbr of the slag) to combat
longitudinal corner cracking caused by thinning of the shell in the corner due to a
high metal flow velocity (Fig. 10.11) [39]; such conditions frequently occur after
reducing the width dimension of the mould with a concomitant increase in casting
speed to maintain throughput. The use of a powder with higher Tbr would alleviate
the severity of this problem.

10.4.4 Casting High-Al (Trip, Twip) Steel Grades

Steels containing high Al (1% or more) have excellent mechanical properties

combining high strength with good ductility. However, these steels have proved
difficult to cast with conventional mould powders because of the huge amounts of
Al2O3 (>30%) produced by reaction shown in Eq. 10.20 (see Sect. 6.2.12).

2Al þ 3SiO2 slag ¼ 3Si þ Al2 O3 slag ð10:20Þ

The Al2O3 pick-up results in increased viscosity and liquidus temperatures (Tliq)
and causes increased crystallisation of the slag film. Alumina particles remain
undissolved in the slag and collide to form agglomerates which can
• Block off liquid slag infiltration into shell/mould channel and
• Promote the formation of large slag rims, which are crystalline and tend to
crack; cracked slag rims can result in a blockage to slag infiltration in the
shell/mould channel and cause depressions.
10.4 Mould Powder Selection for Special Conditions 411

Both of these events can result in sticker break-outs. The cracked slag rims also
cause depressions and false sticker-break-out alarms.
Similar reactions to Eq. 10.19 occur between Al and other slag components like
FeO, MnO, B2O3, etc. Conventional mould powders can be used to cast steels with
up to 0.6% Al but a different family of casting powders is needed to cast steels with
>1% Al. Three types of mould powder are used to cast these steels (with 1% Al),
(i) low basicity slags (ii) high basicity powders (including those forming melilite)
and (iii) calcium aluminate (CA) type powders (see Sect. 6.2.12). Most of the
recent work has focused on casting powders which consist of (calcium alumi-
nates + fluxing agents) [40–43]. The (%CaO/%Al2O3 (or C/A) ratio) is usually kept
around 1 to hold Tliq down and SiO2 contents are kept around 5–10% to minimise
Al2O3 formation via Eq. 10.20. Recent work has shown that FeO, MnO and B2O3
are not very satisfactory fluxing agents since they (like SiO2) are reduced and form
Al2O3 [43]. This leaves CaF2, Na2O, K2O and Li2O as the prime candidates to flux
the calcium aluminate slag; more work is still needed to derive satisfactory mixture
of fluxes to cast these steels.
Another factor which is important in the selection of powders for casting these
steels is that they should create a deep slag pool to dissolve as much Al2O3 as
possible. Deep pool formation is promoted by mould powders containing
(i) exothermic agents (ii) smaller granule size and (iii) low-carbonate content.
High Al, Mn steels have low Tliq values and the vertical heat flux is insufficient
to melt the mould powder. Consequently, liquid slag feeding [44] will have to be
developed to satisfactorily cast these steels.

10.4.5 Fluoride-Free Powders

There has been a drive to reduce the use of fluorides in slags used in all steelmaking
operations. Fluorides pose a health hazard to plant personnel, a threat to the envi-
ronment and cause plant corrosion since they (i) emit HF(g) by reaction with moisture
and (ii) fluorides are leached from slag wastes. The addition of fluorides results in
• A decrease in both liquidus temperature and viscosity of the slag.
• The formation of crystalline cuspidine (3CaO
2SiO2
CaF2) in the slag film
which controls the heat transfer between the shell and the mould.
The slag composition can readily be adjusted to accommodate the replacement
of CaF2 to provide similar Tbr and viscosity values (e.g. by using alternative fluxes
like Na2O) but the replacement of both CaF2 and cuspidine for control of the
horizontal heat flux has proved more difficult. Several crystalline compounds have
been suggested as a substitute for cuspidine, e.g. perovskite (CaTiO3) [42, 45, 46]
melilite [37] and for peritectic MC steels. Na2O2
CaO
3SiO2 [47]. However, it has
been claimed the TiO2 additions to the mould powder result in the formation of Ti
(CN) particles which subsequently, result in sticker break-outs [48]; this allegation
has been disputed [46].
412 10 Selection of Mould Fluxes and Special Mould Fluxes …

The current situation is that satisfactory F-free mould powders have been
developed to replace F-containing (conventional) mould powders for casting billets
[49–52]. The challenge is to develop a satisfactory F-free powder for slab casting,
especially for MC peritectic steels. Another approach is to reduce the amount of
CaF2 added and rely on other crystalline phases to augment the scattering of IR
radiation provided by the remaining cuspidine [46].

10.4.6 Reducing SEN Erosion Rates

SEN erosion rates are affected by several factors such as metal and slag flow
velocities (ii) dissolution of the refractory by the mould slag and (iii) the dissolution
of any carbon in the refractory by the steel. There are three mould slag-related
factors
• The slag viscosity (with SEN erosion rates decreasing as the viscosity increases)
since viscous resistance reduces the flow velocity of the slag phase [53].
• The concentration (C) difference (Csat − C0) where subscripts sat and 0 represent,
respectively, the saturated limit and the actual concentrations of the refractory
(e.g. ZrO2) which affect the dissolution rate.
• The CaF2 content of the slag (since ZrO2 undergoes a high-temperature phase
transition and CaO, MgO or Y2O3 are used to stabilise the ZrO2) due to the
attack of CaF2 on the stabilising oxide [54].
Thus the mould slag can be adjusted in the following ways to reduce SEN
erosion rates:
• By increasing the slag viscosity (via increases in SiO2, Al2O3 or decreasing
Na2O, CaO, etc.) but this will have the effect of decreasing the powder con-
sumption [53].
• By adding to 2% ZrO2 to the powder since Csat occurs at ca. 2% ZrO2 in mould
slags and this reduces the driving force for dissolution (Csat − C0) to zero.
• By reducing the CaF2 content since it reduces the attack on the oxide stabilising
the ZrO2 [49, 53–55].

10.4.7 Minimising Carbon Pick-up

Carbon pick-up by the steel is a significant problem when casting ULC and LC
steels. The mould powder is the main source of the carbon pick-up. The carbon is
added to control the melting rate. In order to minimise C-pick-up powder
manufacturers have developed powders where carbon has been replaced by other
minerals [25–28]. The molten mould slag is non-wetting to the carbon and thus
carbon particles prevent the slag globules from agglomerating and retard the
formation of a molten slag pool (Fig. 10.8). The carbon particles are continuously
10.4 Mould Powder Selection for Special Conditions 413

oxidised by any O2 (g) or CO2 (g) present in the powder bed. Smaller carbon
particles have higher ratios of (surface area/mass) than large particles. Thus smaller
particles provide more non-wetting separation but oxidise more rapidly. The fact
that the melting rate decreases with decreasing mean carbon particle diameter (DC)
indicated the non-wetting separation is the more important of these two processes.
Thus, any substitute for carbon must (i) be non-wetting to molten slag globules
and (ii) react with O2 and CO2 in the powder bed. Various nitrides (e.g. BN, Si3N4)
have been used to replace carbon and some success has been reported [25–28] but
these powders have not been widely adopted because they tend to increase the size
of slag rims.
The best way of reducing C-pick-up by the steel is to operate with a deep slag
pool to keep the shell far away from both the C particles floating in the slag pool
and the slag rim. This entails operating with (i) deep powder bed (ii) exothermic
agents in the powder and (iii) a lower C% in powder to promote melting rate.

10.4.8 Minimising Scale Formation

Some mould powders, which are perfectly satisfactory in every other way, have
been rejected because they promote the formation of scales on the surface of the
steel. The retention of the mould slag (on the steel surface) in the secondary cooling
is key since it provides the necessary SiO2 to form low-melting FeO
SiO2 which
penetrates into the grain boundaries [56–67]. The retention of slag is promoted by a
low interfacial tension between steel and slag (cmsl) [68]. The principal components
causing low interfacial tension are K2O, Na2O and Li2O but it might be expected
that B2O3 would also promote low cmsl. Thus, where scaling is a problem, the
powder should be reformulated to provide similar Qreq s ,η
req
and Treq
br values but with
lower contents of Na2O, etc.

10.5 Summary

Mould powders contain two types of material (i) mineral components which form a
liquid which provides the slag film and (ii) the carbon particles which regulate the
melting rate of the powder. Both are vital to the success of the casting process.
Satisfactory mould slags (i.e. mineral compositions) can be derived for the
specified mould dimensions, casting conditions and steel grade using empirical
rules based on plant data where no problems or defects were experienced [4, 5].
These empirical rules provide the required values of powder consumption (Qreq s );
viscosity (ηreq) and break temperature (Treq
br ) for these conditions. It has been esti-
mated that 85% of the powders in the database conformed with the required values
calculated using these empirical rules [3]. The 15% with property values which did
not conform with those derived with the empirical rules did so because
414 10 Selection of Mould Fluxes and Special Mould Fluxes …

• The viscosity of the mould slag has been deliberately increased to minimise slag
entrapment or SEN erosion rates; this is particularly prevalent in billet casting
where the required powder consumption is relatively low.
• For powders used in casting steels containing Ti, the latter reacts with the mould
powder to form Ti(CN) which has a low solubility in the slag and forms
agglomerated, Ti(CN) particles; these block the entrance to the shell/mould
channel [19] and thereby reduce powder consumption to a low level [3].
• The required break temperature and fcrys are deliberately increased to minimise
variations in shell thickness (which lead to longitudinal corner cracking) which
can arise in the casting of beam blanks and where there are sharp corners (e.g. in
funnel moulds) and also where the metal flow causes thinning of the shell in the
corner [39].
• Certain in-mould conditions affect powder consumption, e.g. Qs increases with
increasing Ar flow rate and tends to decrease when EMBr is applied (because of
the reduction in vertical heat transfer).
The carbon, in addition to the mineral content, of the mould powder must also be
carefully controlled. This involves selection of both the carbon content and the
mean carbon particle size and these must be selected to ensure that the melting rate
of the powder matches the demand for liquid slag (i.e. Qreq s ). Some workers have
decreased the carbon content of the powder in order to either increase the depth of
the molten slag pool or to reduce C-pick-up by the steel. However, care must be
taken with the reduction of C% since carbon is a fuel (exothermic) and if the
reduction of C% is too stringent the steel meniscus can freeze and form a skull.
Carbon-free powders have been developed and are available commercially. The
best way of minimising C-pick-up is operate with a deep slag pool.
The elimination or reduction of F-emissions has led to the development of F-free
powders in which cuspidine (formed in the slag film) is fully or partially replaced
by other crystalline phases. This has proved a difficult task in the case of slab
casting but some progress has been made. These F-free powders result in lower
SEN erosion rates.
High-Al steels (with >0.6% Al) have proved difficult to cast with conventional
mould powders. Consequently, a new family of slags based on calcium aluminates
has been developed to maintain Al2O3 at a manageable level. Although these have
proved promising, further developments are still required to maintain some glassy
phase (to provide lubrication and minimise fracture of crystalline, slag rims) [41].
High-Al steels have low liquidus temperatures and liquid mould slag feeding will
be necessary to cast steels with higher Al and Mn contents.

References

1. H Harada, E Anzai, E Takeuchi, Canad. Met. Quart., 39, (3), 307, (2000).
2. S Ogibayashi, T Mizoguchi,T Kajatani, Intl. Workshop on Thermophys. Data for the
Development of Mathematical models of solidification. Gifu City, Japan (1995).
References 415

3. AB Fox, PhD Thesis, “Mould fluxes- their properties and performance” Imperial College,
London, (2003).
4. KC Mills, S Sridhar, AS Normanton, ST Mallaband, Proc. Brimacombe Conf., Vancouver,
(2000) p. 781.
5. S Sridhar, KC Mills, V Ludlow, ST Mallaband, Proc. 3rd Europ. Conf. Continuous Casting,
Madrid, 1998 (UNESID, Madrid, 1998) p. 807.
6. F Neumann, J Neal, MA Pedroza, AH Castiliejos E, FA Acosta G, Proc. 79th. Steelmaking
Conf., 1996 (ISS, Warrendale, PA) p. 249.
7. M Wolf, Proc. 2nd Europ. Conf. Continuous Casting, Dusseldorf, 1994, (VDEh, Dusseldorf,
1994) p. 78.
8. S Ogibayashi, Y Yamaguchi, T Mukai, T Takahashi, Y Mimura, K Koyama, Y Nagano, T
Nakano, Nippon Steel Technical Report, 34, 1, (1987).
9. KC Mills, AB Fox, ISIJ Intl., 43, 1479, (2003).
10. K Tsutsumi, S Murakami, S Nishioka, Tetsu-to-Hagane, 84, 617, (1998).
11. PE Ramirez-Lopez, KC Mills, PD Lee, B Santillana, Met. Mater. Trans., B, 43B, 109, (2012).
12. SN Singh, KE Blazek, J Metals, 26 (10), 17, (1974).
13. Thermocalc see www.thermocalc.com.
14. MB Santillana, PhD Thesis, Thermo-mechanical properties and cracking during solidifica-
tion of thin slab cast steel. Univ. Delft, NL (2013) Chapter 3.
15. J Miettinen, AA Howe, Ironmaking and Steelmaking, 27, 212, (2000).
16. KE Blazek, O Lanzi, P Gano, D Kellogg, Proc. AISTech., 2007, (AIST, Warrendale, PA,
2007) PR-351–141 – 2007.
17. M Wolf, W Kurz, Met. Mater. Trans., 12, 85, (1981).
18. M Wolf, Proc. METEC Congress’94;. 2nd Europ. Conf. Continuous Casting, Dusseldorf,
1994, (VDEh, Dusseldorf, 1994) vol. 1, pp 78.
19. T Mukongo, C Pistorius, A Garbers-Craig, Ironmaking and Steelmaking, 31, 135, (2004).
20. R Koldewijn, Unpublished Corus Internal Rept. (2007) cited in KC Mills, J Kromhout,
A Hamoen, R Boom, Proc. Admet Conf., Dnipropetrovsk, 2007, (Natl. Metall. Acad. Ukr.,
Dnipropetrovsk, 2007) vol. 2, p. 174.
21. T Mallaband (Metallurgica, UK) Unpublished results cited by AB Fox, ref. [3].
22. PE Ramirez- Lopez, PN Jalali, J Bjorkvall, U Sjostrom, C Nilsson, ISIJ Intl., 54, 342, (2014)
and Proc. 8th Europ. Conf. Continuous Casting, Graz, Austria, 2014 (ASMET, Vienna, 2014).
23. E van Vliet, DW van der Plas, SP Carless, A A Kamperman, AE Westendorp, Proc. 7th
Europ. Conf. Cont. Casting, Dusseldorf, 2011, (VDEh, Dusseldorf, 2011) Session 4.
24. M Kawamoto, K Nakajima, T Kanazawa, K Nakai, ISIJ Intl., 34, 593, (1994).
25. C Lefebvre, JP Radot, JN Pontoire, Y Roux, Revue de Metallurgie, CIT, 94, 489, (1997).
26. B. Debiesme, J Radot, D Coulombet, C Lefebvre, Y Roux, C Demarval, US Patent No 5 876,
482 (1999) and US Patent No 6, 328, 781 (2001).
27. S Terada, S Kaneo, T Ishikawa, Y Yoshida, Proc. 74th Steelmaking Conf., Washington, DC,
1991, (ISS, Warrendale, PA, 1991) p. 635.
28. J Macho, G Hecko, B Golomowski, M Frazee, McMaster Iron & Steelmaking
Symp. “Thinner slab casting”, 2005, (McMaster Univ., Hamilton, Ont., 2005) p. 131.
29. LC Hibbeler, BG Thomas, Proc. AISTech 2013, (AIST, Warrendale, PA, 2013) p. 1215.
30. S Feldbauer, AW Cramb, Proc. 79th Steelmaking Conf., 1996 (ISS, Warrendale, PA, 1996)
p. 595.
31. K Tsutsumi, K Watanabe, M Suzuki, M Nakada, T Shiomi, Proc. 7th Intl. Conf. Molten slags,
fluxes and salts, Cape Town, 2004, (SAIMM, Johannesburg, 2004) p. 803.
32. J Yoshida,T Ohmi, M Iguchi, ISIJ Intl., 45, 1160, (2005).
33. K Tsutsumi, K Watanabe, J Kubota, S Hatori, Y Miki, T Suzuki, T Omoto, Proc. 7th
Europ. Cont. Casting Conf., Dusseldorf, 2011, (VDEH, Dusseldorf, 2011) Session p. 1.
34. K Watanabe, K Tsutsumi, M Suzuki, H Fujita, S Hatori, T Omoto, ISIJ Intl., 54, 865, (2014).
35. J Kromhout, PhD thesis, Mould powders for high speed continuous casting of steel. Univ. of
Delft, NL (2011), p. 99.
416 10 Selection of Mould Fluxes and Special Mould Fluxes …

36. M Kawamoto, Paper presented at International Workshop on “Thermo-physical data for the
development of mathematical models of solidification” Gifu City, Japan, (1995).
37. M Hanao, Y Tsukaguchi, M Kawamoto, Proc. 4th Intl. Congress Science and Technol., Gifu,
2008, (ISIJ, Tokyo, 2008) p. 694.
38. R Nishimachi, Y Ogura, SEAISI Quarterly, 25, (4), 44, (1996).
39. H Tai, M Morashita, T Miyake, Proc. 3rd Europ. Conf. Continuous Casting, Madrid, 1998,
(UNESID, Madrid, 1998) p. 447.
40. K Blazek, HB Yin, G Skoczylas, M McClymonds, M Frazee, Proc. EEEC-METEC
Dusseldorf, 2011 (VDEh, Dusseldorf, 2011).
41. JW Cho, KE Blazek, MJ Frazee, HB Yin, JE Park, SW Moon, ISIJ Intl., 53, 62, (2013).
42. GH Wen, S Sridhar, P Tang, X Qi, YQ Liu, ISIJ Intl., 47, 1117, (2007).
43. Q Liu, GH Wen, JZ Li, XJ Fu, P Tang, W Li, Ironmaking and Steelmaking, 41, 292, (2014).
44. JK Park, JW Cho, KH Moon, SH Lee, KH Kim, HS Jeong, Proc. 7th Intl. Conf. Clean Steel,
Balatonfured, Hungary, 2007, (Hung. Min. Metall. Soc., Budapest 2007), p. 264.
45. H Nakada, K Nagata, ISIJ Intl., 46, 441, (2006).
46. LY Chen, GH Wen, CL Yang, F Mei, CY Shi, P Tang, Ironmaking and Steelmaking, 42, 105,
(2015).
47. N Takahira, N Hanao, Y Tsukaguchi, ISIJ Intl., 53, 818, (2013).
48. Q Wang, YJ Lu, SP He, KC Mills, Z Li, Ironmaking and Steelmaking, 38, 297, (2011).
49. MC Bezerra, CA Valadares, IP Rocha, JP Bulota, MC Carboni, IL Scripnic, CR Santos,
K Mills, D Lever, Proc. 36th Steelmaking Seminar, Porto Allegre, RS- Brazil, 2005, (ABM,
Sao Paulo, 2005). p. 190 available on line at www.carboox.com/pdf/fluxante_2007.pdf.
50. AB Fox, KC Mills, D Lever, C Bezerra, C Valadares, I Unamono, J Laraudogoitia, J Gisby,
ISIJ Intl., 45, 1051, (2005).
51. AB Fox, K Mills, D Lever, M C Bezerra, CA Valadares, I Unamono, J Laraudogoitia, J
Gisby, Proc. 36th Steelmaking Seminar, Vittoria ES- Brazil, 2005, (ABM, Sao Paulo, 2005)
p. 222, available on line at www.carboox.com/pdf/paper2005sfs.pdf,
52. Z Li, K Mills, MC-Bezerra, Proc. 35th Seminar de Fusao Refino e Solidificao Metals,
Salvador, Brazil, 2004 (ABM, Sao Paulo, 2004) p. 281.
53. Y Nakamura, T Ando, K Kurata, M Ikeda, Trans. ISIJ, 26, 1052, (1986).
54. DW Bruce, NS Hunter, Proc. 2nd Intl. Conf. Continuous Casting, Dusseldorf, 1994, (VDEh,
Dusseldorf, 1994) p. 156.
55. Q.Wang, SP He, KC Mills, Proc. 4th Intl. Conf. Cont. Casting Steel in Developing Countries,
Beijing, China (2008). (Chinese Soc. Metals, Beijing, 2008), p. 715.
56. MM Wolf, Ironmaker & Steelmaker, 2000, (Jan), 22, (2000).
57. MM Wolf, Ironmaker & Steelmaker, 2000, (Feb.), 65, (2000).
58. MM Wolf, Ironmaker & Steelmaker, 2000, (March), 69, (2000).
59. MM Wolf, Ironmaker & Steelmaker, 2000, (April), 58, (2000).
60. MM Wolf, Ironmaker & Steelmaker, 2000, (May), 78, (2000).
61. MM Wolf, Ironmaker & Steelmaker, 2000, (June), 22, (2000).
62. MM Wolf, Ironmaker & Steelmaker, 2000, (July), 63, (2000).
63. MM Wolf, Ironmaker & Steelmaker, 2000, (Aug), 75, (2000).
64. MM Wolf, Ironmaker & Steelmaker, 2000, (Sept), 90, (2000).
65. MM Wolf, Ironmaker & Steelmaker, 2000, (Oct), 114, (2000).
66. MM Wolf, Ironmaker & Steelmaker, 2000, (Nov), 67, (2000).
67. MM Wolf, Ironmaker & Steelmaker, 2000, (Dec), 45, (2000).
68. H Nakato, S Takeuchi, T Fujii, T Nozaki, N Washio, Proc. 74th Steelmaking Conf., 1991,
(ISS, Warrendale, PA, 1991) p. 639.
Chapter 11
Using Mould Fluxes to Minimise Defects
and Process Problems

Abstract The quality of the cast product is affected by any defects present. The
causes of each individual defect are discussed, along with the factors affecting their
formation and proposed treatment to minimise the severity of the defect. The fol-
lowing defects and process problems are covered in this chapter: (i) longitudinal
cracks, (ii) longitudinal corner cracks, (iii) sticker breakouts, (iv) oscillation marks,
(v) transverse cracks, (vi) star cracks, (vii) depressions, (viii) overflows, (ix) en-
trapment of slag, gas and inclusions, (x) formation of scales on the surface of the
steel, (xi) Carbon pick-up by the steel, (xii) SEN erosion and (xiii) fluoride emis-
sions. These defects are discussed from the viewpoint of how the mould powder can
help to alleviate the problem.

Symbols, Abbreviations and Units

d Thickness or depth (m)
f Frequency (Hz or cpm)
h Height or depth (m)
L Length or pitch (m)
Q Powder consumption (kg (slag) m−2)
q Heat flux (W m−2)
s Stroke length (m)
T Temperature (oC)
Tbr Break temperature (oC)
t Thickness of mould (m) or time (s)
Vc Casting speed (m min−1)
Vm Velocity of mould (m s−1)
w Width of mould (m)
C/S = %CaO/%SiO2 Basicity
EMBr Electro-magnetic braking
EMC Electromagnetic casting
EMS Electro-magnetic stirring
HC High carbon steel
LC Low carbon steel
LCAK Low carbon Al-killed steel

© Springer International Publishing AG 2017 417

K.C. Mills and C.-Å. Däcker, The Casting Powders Book,
DOI 10.1007/978-3-319-53616-3_11
418 11 Using Mould Fluxes to Minimise Defects and Process Problems

MC Medium carbon steel

MTM Mould thermal monitoring
OM’s Oscillation marks
ULC Ultra-low carbon steel

11.1 Introduction

Continuous casting is a highly successful process. Nevertheless, it is affected by

both process problems and defects which affect the quality of the cast product. In
this chapter, data concerning these various problems and defects are examined and
analysed to determine the underlying causes of the problem and the mechanisms
involved. From this knowledge, ways of dealing with the various problems will be
proposed. Frequently, the underlying problem originates from changes in the
casting conditions (e.g. changes in casting speed), but the mould slag is expected to
minimise the impact of the problem. So, one focus of this chapter will be to
highlight what changes in the mould fluxes will help to alleviate the problem. The
various problems and defects are discussed individually below.

11.2 Longitudinal Cracking

Longitudinal cracking occurs on the surface and the sub-surface of the cast steel
product. The cracking is a major problem since cracks are usually removed by
grinding (or scarfing) which is time-consuming and leads to a loss in yield.
Longitudinal cracks can be classified into two types (Fig. 11.1) [1, 2]:

Fig. 11.1 Photographs showing a Gross crack b Small sub-surface cracks (permission granted,
ISS/AIST, [2])
11.2 Longitudinal Cracking 419

Fig. 11.2 Schematic diagram

showing incidence of surface
cracks as a function of wt%C
in steel (permission granted,
ISIJ, [3])

(i) Gross cracks (up to 400 mm long) usually accompanied by longitudinal

depressions; this form of crack is usually associated with severe casting
problems (such as poor mould level control).
(ii) Sub-surface, shallow cracks found when casting MC, peritectic, steel grades;
these cracks (typically, 30–40 mm) are much smaller than the gross cracks.
There are several factors affecting the formation of longitudinal cracks such as
• The type of steel.
• The rate of heat extraction from the shell (i.e. horizontal heat flux, qhor) in the
meniscus region.
• The lubrication and powder consumption supplied to the shell.
• The metal flow and the use of electromagnetic appliances to reduce turbulence.

11.2.1 Type of Steel

Longitudinal cracking is particularly prevalent in peritectic, MC steels

(C% = 0.085–0.145) as shown in Fig. 11.2 [3]. On solidification, the steel forms
d-Fe (ferrite) which at lower temperatures, transforms to c-Fe (austenite). Austenite
shells are stronger but less ductile than ferrite shells (Fig. 11.3) [4].
The transition (L + d!c) is accompanied by a volume change of (0.4–0.6%)
due to the tighter packing in the austenite, and there is also a 4% mismatch between
the thermal shrinkage coefficients of d-Fe and c-Fe. This mismatch results in
stresses. These stresses are much larger in thicker regions of the shell than in thinner
regions and result in a hoop stress around the mould. This hoop stress is relieved by
longitudinal cracking.
Wolf [5] introduced a parameter (the ferrite potential, FP) to describe the amount
of peritectic reaction taking place in the steel. In this system, the maximum amount
of peritectic reaction occurs at FP = 1.0. The calculation of ferrite potential for
various steels requires knowledge of other parameters, namely, the carbon equiv-
alent for peritectic reaction (CP) (Eq. 11.1), and for stainless steels the “Ni” and
420 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.3 Ultimate tensile

strength at 1400 °C at high
strain rate as function of %C
in the steel (permission
granted, Verlag Stahleisen,
re-drawn after [4])

“Cr” terms (given in Eqs. 11.2 and 11.3, respectively) where all contents of ele-
ments in steel are in mass%.

CP ¼ %C þ 0:04 Mn þ 0:10 Ni þ 0:7 N  0:14Si  0:04 Cr  0:1 Mo  0:24Ti

ð11:1Þ
00
Ni00 ¼ %Ni þ 0:31 Mn þ 22 C þ 17:5 N ð11:2Þ
00
Cr00 ¼ %Cr þ 1:5 Si þ 1:65 Mo þ 2 Nb þ 3 Ti ð11:3Þ

The ferrite potential can then be calculated by Eq. 11.4 or 11.5.

Low alloy steel:

FP ¼ 2:5 ð0:5  CPÞ ð11:4Þ

Stainless steel:

FP ¼ 5:26f0:74ð00 Ni00 =00 Cr00 Þg ð11:5Þ

The characteristics of the shells formed for steels with various ferrite potentials
are summarised in Table 11.1 and are depicted in Fig. 11.4.

Table 11.1 Characteristics of shells formed for different bands of ferrite potential (FP)
FP Phase Properties of shell
<0.0 Austenite Strong and more ductile
with increasing FP
0.0–1.0 Initially forms ferrite, Irregular, strong shell
then transforms to austenite
>1.0 ferrite Weak, ductile shell
11.2 Longitudinal Cracking 421

Fig. 11.4 Schematic diagram showing tendencies to form longitudinal cracks and sticker
breakouts as functions of ferrite potential (FP). It can be seen that longitudinal cracking is low in
the d-Fe and c-Fe regions but high in two-phase region (permission granted, VDEh, Verlag
Stahleisen [5]; re-drawn SWEREA/ KIMAB)

According to Fig. 11.4, the region with largest risk of depressions lies between
FP = 0.8 and 1.1.

11.2.2 Heat Flux

The longitudinal cracks are thought to be initiated in the meniscus region of the
shell. Longitudinal cracking appears to be related to
(i) The magnitude of the heat extraction rate from the shell (i.e. horizontal heat
flux, qhor).
(ii) Local variations in heat flux (e.g. where a hot spot is formed in the mould)
which results in a shell of variable thickness.
These are studied separately below. The strategy usually adopted is to minimise
longitudinal cracking by producing a shell which is both as thin and as uniform as
possible.

11.2.2.1 Horizontal Heat Flux

The relation between longitudinal cracking and horizontal hear flux (qhor) has been
studied by a number of investigators [6–17].
Hiraki et al. [6] established that longitudinal cracking occurred when the hori-
zontal heat flux in the meniscus region exceeded a certain critical value (qcrit) and
that the level of cracking increased sharply with further increases in heat flux above
qcrit. Inspection of Fig. 11.5 shows that qcrit has values of 1.5 M W m−2 for ULC
(hypo-peritectic) steels and 2.5 M W m−2 for LC steels. Thus, a much lower heat
flux is required to avoid longitudinal cracking in a ULC, hypo-peritectic steel
422 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.5 The incidence of

longitudinal surface cracks as
a function of horizontal heat
flux for LC (●) and
hypo-peritectic steel (D);
index of longitudinal
cracks = R(surface cracks
>5 mm/cast length); qcrit
occurs at 1.5 and
2.5 M W m−2 for
hypo-peritectic and LC steels,
respectively (permission
granted, ISS/AIST, [6])

(FP > 1.0) than for a LC steel (1.0 < FP < 1.2). It can be seen in Fig. 11.5 that
longitudinal cracking increases sharply when qcrit is exceeded. Similar findings
have been reported by other workers [7, 9, 11, 14].
The following methods have been used to reduce horizontal heat flux:
• Decreasing the water flow rate [18–20].
• Using Meniscus-free casting (in which a refractory material is placed inside the
mould in the meniscus region [21–23].
• Machining grooves in the copper mould around the meniscus level to form gas
gaps and create a thicker slag film [24–31]; grooved moulds provide a 10%
decrease in horizontal heat flux [25, 31].
• Coating the mould with a material with a lower conductivity than copper
[20, 32–37].
• By using a mould powder which provides a thick, crystalline, slag film [38],
the thickness of solid slag film is related to the break temperature (Fig. 11.6a, b)

Fig. 11.6 Diagrams showing the effect of break (or solidification) temperature, Tbr, of mould
slags on a longitudinal cracking and b horizontal heat flux, qhor, when casting hypo-, peritectic
steel; a ▬ and b ▲low-C steels, a solid line and b ●, respectively (permission granted, ISS/AIST
[6], re-drawn)
11.2 Longitudinal Cracking 423

and a high-crystalline fraction (fcrys) provides (a) an interfacial resistance (RCu/sl,

or gas gap) caused by the shrinkage accompanying crystallisation of the slag
film and (b) the reflection of the radiant energy emitted by the shell by crys-
talline phases in the slag film.
• By adding transition metal oxides (e.g. MnO, FeO, NiO,) to the casting powder
to create a slag film which absorbs (and hence reduces) the radiant energy
emitted by the shell [39, 40].
• The use of non-sinusoidal oscillation has been reported to decrease the heat flux
by 10% [41] but this claim has been disputed.
It has been reported that longitudinal cracking increases as negative strip time
(tn) increases [42]. The effect of slag basicity (C/S) and negative strip time on
longitudinal cracks for a micro-alloyed, peritectic, medium-carbon steel was
explored in a study covering 27 heats at SSAB, Oxelösund [43]. It can be seen from
Fig. 11.7 that longitudinal cracking (i) decreases with increasing basicity (since
heat flux, qhor, tends to decrease with increasing basicity) but (ii) exhibits a max-
imum at ca. tn = 0.16. It has been suggested that the quadratic relationship in
Fig. 11.7 is related to the fact that longitudinal cracks tend to initiate in the base of
oscillation marks and occur more frequently as the depth (dOM) increases. Values of
dOM tend to decrease with lower tn and qhor (i.e. with high basicity). However, at
higher tn values the higher slag infiltration tends to lead to provide more even
cooling [43].

Fig. 11.7 Longitudinal crack length (m/m) as function of slag basicity and negative strip time
(permission granted, [43])
424 11 Using Mould Fluxes to Minimise Defects and Process Problems

11.2.2.2 Local Variations in Heat Flux

Solidification (and shell growth) occurs mostly in the period of the oscillation cycle
when the mould is descending, and there is little steel growth in the period when the
mould is ascending. This leads to a shell of varying thickness (“sausage-like”).
Thus, there are variations in shell thickness arising naturally from the mould
oscillation. However, further local variations in shell thickness can arise if a “cold
spot or a hot spot” develops in the mould. Such a cold (or hot) spot will create even
greater differences in shell thickness. The accumulations of stress (due to the
mismatch in shrinkage coefficients of d- and c-Fe phases) are much greater in the
thick regions of the shell than in the thinner regions and result in an increased risk
of longitudinal cracking.
“Cold spots” can be created (i) when high-conductivity, SENs are too close to the
mould (this is especially important in thin slab casting), (ii) by overcooled corners
(see Sect. 3.4.5). Hot spots can be formed where the flow of metal impinges on the
shell, which causes melt-back and thus a thin region in the shell.
Longitudinal cracking tends to be prevalent at the beginning of the cast; this can
be seen in Fig. 11.8 where the incidence of longitudinal cracking is high in the first
metres of the strand but gradually decreases as the thermal conditions and the slag
film stabilise [44]. It can also be seen that the use of casting powders containing
exothermic agents help to stabilise the thermal conditions and thus reduce longi-
tudinal cracking [44]. This may be due to the exothermic agents reducing,
sequentially, the vertical heat flux, the length of the shell and the depth of the
oscillation mark.

Fig. 11.8 Longitudinal crack

index as a function of cast
length of the strand; note
cracking levels are reduced as
the cast proceeds and by using
powders containing
exothermic agents (-─ ▲ -)
cf. no exothermic agents
(─●─); re-drawn after [44]
11.2 Longitudinal Cracking 425

11.2.3 Lubrication and Powder Consumption

Ogibayashi et al. [45] pointed out that frictional forces acting on the shell increase
as the distance from the corner increases (Fig. 11.9) and they are highest at the
narrow sides due to the large taper. Consequently, it is necessary to provide good
lubrication in the centre of the strand where the shell is thinnest. They attributed
longitudinal cracking to (i) poor lubrication and (ii) to irregular shell growth at low
casting speeds [45] but most experience shows that higher casting speed leads to
fewer longitudinal cracks due to a thinner steel shell.

11.2.4 Metal Flow, Use of EMBr, EMC and EMS

Longitudinal cracking has been correlated with the amplitude of the waves running
along the steel surface (Fig. 11.10a). These waves are formed by the backwash
resulting from the impact of the metal flow on the mould wall. The velocity of the
metal flow (Vmetal) is linearly related to the casting speed (Fig. 11.10b).
Various electromagnetic applications are used in continuous casting (see Sect. 5.8).
Elecromagnetic stirring (EMS) is used to homogenise both the steel composition and
the temperature of the steel in the mould. EMS has been reported to reduce both
longitudinal cracking and entrapped inclusion content (Figs. 11.10a [52] and
Fig. 11.11, respectively). Presumably, reduced longitudinal cracking arose from the
minimisation of temperature differences and thus, differences in shell thickness [45,
46]. Electromagnetic braking (EMBr) is used to reduce (i) the metal level variations
(i.e. waves) and (ii) both Vmetal (Fig. 11.10b, c) and the penetration of the metal flow
into the mould. EMBr has been reported to reduce longitudinal cracking [46–49].
In Electromagnetic casting (EMC), the applied field creates a “pinch force”
which pushes the shell back and allows more slag infiltration, thereby forming a
thicker slag film. This delays solidification and produces a short and thin shell
which results in shallow oscillation marks and decreased longitudinal cracking [50].

Fig. 11.9 Schematic diagrams showing (upper) the frictional forces across the mould and (lower)
Thickness of the steel shell across the mould [45]
426 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.10 Diagrams (a)

showing longitudinal crack 1.6

Total crack length, mm

length as function of a wave
amplitude [52] (determined 1.2
by plate-dip tests) upper
curve = without EMBr and 0.8
lower curve = with EMBr
b metal flow velocity as 0.4
functions of casting speed
[52] and c metal flow velocity 0
as a function of EMBr
0 2 4 6 8 10
intensity [52]; Vc (m min−1),
● = 2; ▲ = 3; ■ = 3.5; Amplitude surface wave, mm
♦ = 5 (permission granted,
ISIJ, re-drawn after [52]) (b)

(c)

Increased levels of longitudinal cracking were found to occur when asymmetric

flow patterns developed as a result of SEN clogging [53]; presumably, this is due to
either the metal flow creating hot spots where the metal flow impacts or to the
creation of large surface waves (Fig. 11.10a).

11.2.5 Causes and Mechanisms

The initial solidification of the shell in the meniscus region of the mould is not a
smooth process. Heat extraction is affected by the oscillation cycle with rapid shell
11.2 Longitudinal Cracking 427

(a) (b)
1

Index longitudinal cracks

Longitudinal crack index

18
0.8 14
0.6 10
0.4 6
0.2 2
0 -2
0 20 40 60 80 100 0.95 1.35 1.75
EM srring velocity C % x100

Fig. 11.11 Diagrams showing Index of longitudinal cracking reduced a with EMS stirring
velocity [46] and b by using EMBr [46] (with EMBR = ; without EMBr = ) showing
longitudinal cracks prevalent in peritectic range and the reductions obtained with EMBR
(permission granted, NSSM Corp.; re-drawn after [46])

growth occurring when the mould is descending (i.e. in negative strip periods) and
very little solidification occurring during the ascent of the shell (i.e. in positive strip
periods). The maximum solidification (and heat flux) occurring in the period
between late in and early tp. These cyclical periods of growth of the shell during the
oscillation cycle tend to result in a corrugated shell [53, 55].
Longitudinal cracking results from a sequence of events. It was mentioned above
that there is a 4% difference in the thermal shrinkage coefficients of the d-Fe and
c-Fe phases (Fig. 11.12) which results in stresses [56]. The accumulation of stress is
much greater in the thick regions of the shell than in the thinner regions. These local
stress gradients result in hoop stresses which are subsequently relieved by

5
10 4 x Shrinkage coeff, oC-1

0
0 0.1 0.2 0.3 0.4 0.5 0.6
% C steel

Fig. 11.12 Average shrinkage coefficient as a function of carbon content of the steel; note
enhanced shrinkage coefficients in the peritectic range, C% = 0.06–0.15 (permission granted,
re-drawn after [56])
428 11 Using Mould Fluxes to Minimise Defects and Process Problems

longitudinal cracking. Thus, large variations in shell thickness lead to enhanced

longitudinal cracking. Bouts of longitudinal cracking can be detected from vari-
ability in the heat flux.

11.2.6 Ways of Dealing with Longitudinal Cracking

Longitudinal cracking is principally caused by local differences in shell thickness,

so, remedial treatments are based on providing a thinner, more uniform shell. This
can be achieved in the following ways.

11.2.6.1 By Decreasing the Horizontal Heat Flux in the Meniscus

Region (“Mild Cooling”)

This can be achieved by

• Modifying the casting powder so that it forms a thick, crystalline slag film with a
high Tbr [6, 38], this is usually obtained by increasing the (%CaO/%SiO2) ratio
to ca. 1.3 and having a sufficient amount of Fluorine for the formation of
cuspidine, or by making additions of 2–3% ZrO2 (which acts as a nucleant for
crystallisation).
• Using a glassy mould slag with high IR absorption ability with a high content of
FeO, MnO or NiO [39] could be a solution for special cases such as round-billet
casting. In this case, the solid slag film must provide a good support for the
strand and the creation of a gas gap (resulting from slag crystallisation) must be
avoided.
• Using mould coatings with lower thermal conductivities than copper, preferably
Nickel [37].
• Machining grooves in the meniscus region of the mould [24–27].
• Reducing the water flow rate but this has only a small effect [18–20].
• Using non-sinusoidal mould oscillation [41].
• It is thought that some longitudinal cracks are initiated at the base of oscillation
marks [57]; consequently, taking steps to reduce dOM should also be beneficial
in reducing longitudinal cracking, these include (i) reducing the vertical heat
flux and the length of the shell and (ii) by reducing tn and stroke (and any other
parameters which decrease tn, such as increases in Vc and f).

11.2.6.2 By Reducing the Variations in Shell Thickness

The variations in shell thickness can be achieved by

11.2 Longitudinal Cracking 429

• Proper design of the SEN to ensure an optimum and even steel flow in the
mould.
• Using EMBr and EMS to smooth out any temperature gradients.
• Avoiding asymmetric flows arising from SEN clogging.
• Ensuring sufficient melting of the mould powders
• Enhanced levels of longitudinal cracking tend to occur in the early moments of
the cast (Fig. 11.8), but using casting powders containing exothermic agents
helps to establish steady state conditions quicker and thereby, reduce longitu-
dinal cracking.

11.2.6.3 Monitoring to Avoid Longitudinal Cracking

Readings from moulds fitted with thermocouples, for example, breakout detection
systems, can be used to detect longitudinal cracking. Presumably, the thermocou-
ples detect the presence of a “buckled shell” on the heat flux which would be
expected to result in (i) variability of the heat flux (i.e. temperature response) and
(ii) a decrease in heat flux. It has been reported that periods of longitudinal cracking
can be detected via
• Temperature variability (Fig. 11.13a) [1, 2, 58].
• A sudden decrease in temperature which successfully detected 80% of gross
cracks and 40% of sub-surface cracks (Fig. 11.13b) [52].

Fig. 11.13 Mould thermal monitoring (MTM) of temperature showing a variability in

temperature associated with longitudinal cracking [1, 2, 58] and b conventional trace (upper)
and drop in temperature associated with longitudinal cracking (lower) [52] (permission granted,
a 10 M/Taylor and Francis, [58] b EPD Sci. [52]
430 11 Using Mould Fluxes to Minimise Defects and Process Problems

11.2.6.4 Improving Powder Consumption

It has been suggested that longitudinal cracking can arise from insufficient lubri-
cation and the use of an incorrect (low) taper [42, 45].
Another way of successfully avoiding longitudinal, midway, cracks for slab
casting is by increasing the superheat [15]. This leads to delayed solidification of
the shell and leads to improved powder consumption.

11.3 Longitudinal Corner Cracking

11.3.1 Published Information

The causes of longitudinal corner cracking are similar to those responsible for
longitudinal (midway) cracking. Narzt et al. [12] reported that the incidence of
longitudinal corner cracking could be reduced by lowering the total heat flux
(Fig. 11.14) i.e. by reducing horizontal heat transfer by using a casting powder with
a high basicity which produces a thick, crystalline slag film.
Significant decreases in the incidence of longitudinal corner cracking have been
recorded after applying EMBr to reduce both the metal flow velocity and surface
instabilities [13].
It has also been reported that longitudinal corner cracking increases with
(i) increasing S content of the steel [59] and (ii) increasing negative strip time,
tn [42] (i.e. similar relations to those for longitudinal cracking).

11.3.2 Causes, Mechanisms

The causes of longitudinal corner cracking are similar to those for longitudinal
cracking, namely, sharp differences in shell thickness, which, for certain steel

Fig. 11.14 Incidence of 0.5

Longitudinal crack index, %

longitudinal corner cracking

as a function of total 0.4
(horizontal) heat flux
(permission granted, 0.3
UNESID, re-drawn from [12])
0.2

0.1

0
1 1.1 1.2 1.3 1.4
Integral heat flux, qtot, MWm-2
11.3 Longitudinal Corner Cracking 431

grades (especially medium-carbon steels) lead to stresses resulting from the mis-
match of thermal shrinkage coefficients of the d-Fe and c-Fe. These stresses result
in strains which are relieved by cracking. There are two different types of longi-
tudinal corner cracking:
• The type caused by overcooled corners which is prevalent in billet casting and
some other mould geometries [60]; a low taper leads to an increase in longi-
tudinal cracking [42] due to the formation of an air gap and hence a shell of
variable thickness.
• The type caused by “thinning of the shell in the corner” resulting from the local
impact of the metal flow [61–63]; this type is prevalent in slab casting but can
also be a result of improper narrow-side geometry (i.e. corner radius [60] and
taper [42, 60, 64].

11.3.2.1 Longitudinal Corner Cracking Arising from Overcooled

Corners

It can be seen in Fig. 11.15 that the heat in the corner regions has a much greater
area to flow into than that in the centre. Consequently, the heat flux is much higher
in the corners than in the centre of the mould. The high heat flux results in a thick
shell in the corner, and during the subsequent contraction, slag normally flows in to
fill the gap. However, in some cases, excessive contraction can lead to the for-
mation of a gas gap which results in a sharp decrease in heat flux. Large variations
in shell thickness lead to longitudinal corner cracking. The distance between corner
and centre is much smaller in billets and leads to large, thermal gradients in the
horizontal direction. These gradients lead to sharp changes in shell thickness and
are particularly large in the corner region which results, sequentially, in stresses and
then cracking.
“Football” moulds
There is very little room for the SEN in thin slab casting moulds and the “football
mould” is one possible solution to this problem (Fig. 11.16). However, it can be
seen that four “corners” are created around the SEN. There is enhanced heat flow
through these corners which creates a thicker shell in these regions. Longitudinal
corner cracking tends to occur in these regions.

Fig. 11.15 Schematic

diagram showing the path for mould
heat extraction in the mould;
note that there is a much shell
greater area for heat removal
in the corners than in the
central regions of the mould
432 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.16 Schematic drawing showing “football mould” used in thin slab casting and the
creation of four “corners” with enhanced heat flow (see arrows) which are potential sites for
longitudinal corner cracking; the SEN is shown in blue

Longitudinal crack length

12

10

0
A B C D E F A' B' C' D' E' F'
Posion

Fig. 11.17 Regions of longitudinal corner cracking in blank beam casting where it can be seen
that cracking is particularly severe at positions B and E, i.e. internal corners (Re-drawn from [65])

Beam blank casting

Beam blanks can be cast in the desired shape. However, these moulds contain
“corners” which can lead to enhanced cooling and a thicker shell in these regions.
The consequent variation in shell thickness can lead to longitudinal corner cracking.
It has been reported [65] that the cracking
• was worse for the internal corners (B and E in Fig. 11.17).
• was worse when using mould slags with low Tbr (or Tsol).
• tended to increase with increasing sulphur content of the steel.

11.3.2.2 Longitudinal Corner Cracking Resulting from Thinning

of Shell by the Metal Flow

A typical example of this type of longitudinal cracking is shown in Fig. 11.18a) [61, 62]
which clearly shows a “white band” resulting from negative segregation. The white
11.3 Longitudinal Corner Cracking 433

Fig. 11.18 a Photograph showing occurrence of “white band” and b drawing outlining shell
thickness and white band and location of corner cracks (permission granted, UNESID, [61])

band reveals the location of the instantaneous solid/liquid interface (Fig. 11.18b) and it
can be seen that the shell is thinner in the vicinity (on both sides) of the corner. This was
attributed to direct impingement of the molten steel flow on the shell which causes
melt-back and thinning of the shell and thus, a shell of variable thickness.
These casting conditions can occur when the mould dimensions are changed (i.e.
the slab width is reduced and the casting speed increased in order to maintain the
steel throughput at its current level). Both of these changes (in mould width and
casting speed) result in increased metal flow velocity, which causes thinning of the
shell in the impingement region and a shell of variable thickness. Significant
decreases in the incidences of longitudinal corner cracking have been recorded after
applying EMBr to reduce the metal flow velocity [13].

11.3.3 Ways of Dealing with Longitudinal Corner Cracking

The normal strategy to deal with this problem is to reduce the overall heat flux and
create a more uniform, thinner shell. Obviously, the use of casting powders which
434 11 Using Mould Fluxes to Minimise Defects and Process Problems

produce a high-melting (i.e. high Tbr) thick, crystalline slag film will be beneficial in
most cases.

11.3.3.1 Longitudinal Corner Cracking Due to Overcooled Corners

Longitudinal corner cracking of this type can possibly be reduced by

• Replacing sharp corners with smooth chamfered corners [64, 66].
• Reduce water flow rates in the corner regions (cf. those for themed-face) [18, 19]
but it has been reported that reduced water flow rates have little effect on the
total heat flux [20], shutting off the water flow in the corners has been found to
reduce longitudinal corner cracking [19].
• Apply coatings to the corner region of the mould to reduce heat transfer.

11.3.3.2 Longitudinal Corner Cracking Due to Shell Thinning

In this case, it is beneficial to reduce the melt-back of the shell due to the impact of
the metal flow. This can be carried out by
• Redesigning the SEN port so as to give “more diffusive flow” into the corner
regions of the mould [61, 62].
• Reducing the superheat.
• Reducing the casting speed.

11.4 Sticker Breakouts

Sticker breakouts are a major problem and result in hot metal streaming from the
mould. They pose a safety hazard to the operating staff and result in loss of pro-
duction; they have been estimated to cost between 0.1 and 1 million $ per incident.
There are probably several causes for sticker breakouts but all involve the formation
of a thin shell at some location and a loss of lubrication at that site. The solidified
shell retrieved from the mould has the following distinctive characteristics:
• The “stuck shell” fans out from the sticking point, marked as d) in Fig. 11.19
and in detail in Fig. 11.20.
• The regular oscillation marks disappear and (in the stuck shell) are replaced by
ripple marks with a characteristic V-shape fanning out from the breakout point
(Fig. 11.19b).
• The shell is at its thinnest at the sticking point with the stuck shell increasing in
thickness as we travel upwards (Fig. 11.19a).
11.4 Sticker Breakouts 435

(i) (ii)

Fig. 11.19 Schematic drawings of the appearance of the strand after sticker breakout where the
breakout occurs in (i) the broad face and (ii) the corner; a denotes regions of regular oscillation
marks, b denotes ripple marks on stuck shell, c denotes the line of constrained shell and d denotes
the sticking (or breakout) point (or position) (re-drawn by Swerea/Kimab)

Fig. 11.20 Schematic diagrams showing (a) the shell is thinnest at the sticking point and (b) the
appearance of the stuck region of the shell (permission granted, ISS/AIST [67])

11.4.1 Factors Affecting Sticker Breakouts

There is probably more than one mechanism involved in the creation of sticker
breakouts, but all are thought to be linked to the formation of a shell which is too
thin and too weak. A number of factors play a part in the creation of a sticker
breakout, these are discussed below.
436 11 Using Mould Fluxes to Minimise Defects and Process Problems

11.4.1.1 Steel Grades

Sticker breakouts are prevalent whilst casting high-Carbon (C > 0.4%) steel grades.
The shells formed when casting these steels have poor mechanical strength due to
micro-segregation between the austenite grains. Wolf [5] proposed that sticker
breakouts were prevalent when the ferrite potential of the steel (FP) was less than
0.6 (where FP is given by Eq. 11.4 and CP is the carbon potential of the steel which
can be calculated by Eq. 11.1).

11.4.1.2 Mould Dimensions and in Mould Conditions

A statistical analysis of plant data indicated that sticker breakouts tended to increase
with (i) increasing mould width, w [68], (ii) decreasing mould length, Lmould [68,
69], (iii) increased casting speed (leading to a thinner shell) and (iv) increased
mould level variations [68]. The effects of mould width and length are consistent
with the formation of a thinner shell. Mould level variations will cause (i) carbon
pick-up by the steel in contact with slag rim and the floating carbon particles (see
Sect. 11.4.1.5 below) and (ii) the slag rim to fracture (see Sect. 11.4.1.3 below).

11.4.1.3 Steelmaking Conditions Creating Large Amounts of Al2O3

or TiN or ZrO2

Sticker breakouts are thought to occur when an agglomerate gets trapped in the
shell/ mould gap and blocks off the flow of liquid slag in this region. These
agglomerates can arise from (i) the fracture of slag rim (ii) collisions of solid
particles which lie undissolved in the liquid slag and (iii) liberated accretions ripped
from the tundish stopper and SEN ports [70]. It might be expected that large mould
level variations would also aid the fracture of slag films [68, 71].
There is a considerable amount of anecdotal evidence that sticker breakouts tend
to occur when the steelmaking conditions are creating more Al2O3 than usual. It has
been suggested that turbulent collisions of the Al2O3 particles causes the particles to
agglomerate. Some Al2O3 accretions can also accumulate on the SEN and stopper
rod refractories. However, Al2O3 pick-up by the slag is known to increase the
crystallinity of slag films and rims [72] and which tend to fracture easily [73, 74].
Furthermore, large amounts of Al2O3 formation also lead to the formation of large,
slag rims [74]. Thus, when large amounts of Al2O3 are formed the blockages to the
liquid slag supply may come from either agglomerated Al2O3 particles or from
fractured slag rims. In the casting of high-Al steels, the Al2O3 pick-up (>30% for
1%Al steels) results in frequent sticker breakouts.
Steels containing Ti (e.g. ULC and Ti-stabilised stainless steels) have also been
reported to be prone to sticker breakouts [75]. Large amounts of TiN or Ti(CN) tend
to form and these have low, saturated solubilities in mould slag (Csat = ca. 0.5%)
and these undissolved particles form agglomerates through collisions [76–78].
11.4 Sticker Breakouts 437

A high incidence of breakouts has been reported when using F-free mould slags
containing TiO2 [75].
It has also been reported for Ti-stabilised stainless steels, that the Ti in the steel
reacts with SiO2 in the mould slag to form perovskite (CaOTiO2) [75, 79] which
has a high-melting point. Consequently, it precipitates into the molten slag where it
(i) increases the viscosity of the slag and (ii) agglomerates and blocks off the supply
of molten slag to the shell. It is necessary to keep the basicity (C/S) of the slag < 1
to avoid perovskite precipitation [79].
ZrO2 also has a low, saturated solubility in mould slag (Csat = ca. 2%) [80]. Slag
films attached to the stuck portion of the shell were found to contain high-ZrO2
contents (up to 17%) [81]. Consequently, ZrO2 probably behaves like TiN and
forms agglomerated particles which shut off the flow of liquid slag in local sections
of the mould. However, ZrO2 will (i) increase Tbr for the mould slag by 20–40 °C
[80] and (ii) particles will act as nucleants for the precipitation of cuspidine and
other crystalline phases [81]. These will result in a thicker and more crystalline slag
film slag film, respectively, and thus will reduce the horizontal heat flux and shell
thickness and thereby increase the incidences of sticker breakout.

11.4.1.4 Loss of Lubrication and Frictional Forces

The local loss of lubrication caused by a blockage to the liquid slag flow will
increase the frictional forces acting on the shell. The shear stress and ferrostatic
forces tend to increase as the shell moves further down the mould [67]. When the
conditions result in a flaw in the solidified shell, breakout will occur when the shear
stresses exceed the yield stress. The frictional forces calculated for these conditions
were found to be very similar to those recorded under normal conditions [82].
A mathematical model has been developed to calculate the frictional forces, the
shear stress and the high temperature, strength of the shell through the oscillation
cycle [83, 84]. Tsuneoka et al. [67] developed a steady state, heat transfer model to
simulate the proposed mechanism for sticker breakouts and obtained good agree-
ment between predicted and measured temperature transitions.
Loss of lubrication can also occur when the melting rate of the mould powder
fails to meet the required powder consumption. This can occur when large standing
waves are formed and the granules run downhill from the peak leaving the peak
with little mould slag. This can cause high friction and, in the worst case, breakouts;
this is exemplified in Fig. 11.21.

11.4.1.5 Carbon Pick-up by Liquid Steel Near the Sticking Point

A metallographic examination of the stuck shell in the mould showed that the shell
near the sticking point had a carburised structure, and there were some molten metal
droplets with a carburised structure and an enhanced C content (ca. 4%) [85]. These
droplets also contained cavities thought to have been formed by CO (g) during
438 11 Using Mould Fluxes to Minimise Defects and Process Problems

(a) (b)

Sintered mould
slag patches

Break-out region

Fig. 11.21 Photographs showing a breakout slab with a region without mould slag (only sintered
slag) b the breakout region (courtesy CA Dacker)

solidification. Thus, it was concluded that the carburised structures that there was
local Carbon pick-up by the steel [85]. There are several sources of carbon:
– Unmelted casting powder in the bed [85] especially when there are large mould
level variations (see also Fig. 11.21).
– The carbon particles floating near the top of the slag pool.
– Slag rims undergoing large mould level variations and fractured slag rims.
– Refractories (e.g. SEN, stopper rods) [70].
Pick-up of Carbon by the steel will reduce its melting point and steel will tend
not to solidify when the mould is descending (in negative strip time) which results
in a flaw in the shell.

11.4.1.6 Low Heat Transfer Across the Slag Film

Slag films producing a very thin solidified shell can also lead to sticker breakouts.
The overly thin shells can result from several events and lead to sticker breakouts:
• Excess crystallisation in the slag film.
• Pick-up of ZrO2 by the slag film which (i) increases Tbr and ds (ii) nucleates
cuspidine and other crystalline phases; both of these events decrease qhor and
reduce dshell [80, 81].
• Hydrogen in steel which results in a porous slag film with low thermal con-
ductivity [86–88].
• Cyclical changes in casting speed which affect the thickness of the solid slag
film [89].
11.4 Sticker Breakouts 439

Fig. 11.22 The effect of 0.3

crystallisation index on the

Breakout frequency, %
incidence of sticker breakouts
(permission granted, ISIJ, 0.2
re-drawn from [90])

0.1

0
0 1 2 3 4
Crystallisa on Index

These various cases are described individually below.

Crystallisation in Slag Film

It can be seen from Fig. 11.22 that the incidence of sticker breakouts was found to
increase when using casting powders with a high-crystallisation index (which were
determined by metallographic examination [90]). The horizontal heat flux decreases
with increasing crystallisation and thus the resulting steel shell will be both thinner
and weaker [90, 91]. Thus, it would appear that incidences of sticker breakout can
be reduced by using mould powders forming glassy slag films which lead to
thicker, stronger steel shells. This is especially important when casting low carbon
steel with an initial solidification of d-ferrite which is very soft.

ZrO2 Pick-up in Slag Film

The pick-up of up to 17% ZrO2 in slag films attached to the stuck portion of the
shell [81] has been discussed above (Sect. 11.3.1.3). It was pointed out that ZrO2
additions will reduce the horizontal heat flux and thus, result in a thin shell. The
most probable sources of ZrO2 are the refractories used for the SEN and stopper
rods [80]. The pick-up of ZrO2 provides some measure of support for the mecha-
nism involving the ripping off of accretions attached to the refractories by the metal
flow [70].

Hydrogen in Slag Film

Incidences of sticker breakout have been correlated with high concentrations of

hydrogen in the slag film [86–88]. Slag films with high-hydrogen contents contain
many gas pores; the porous slag film has a low thermal conductivity. Consequently,
the horizontal heat flux is low and a very thin shell is formed which is prone to
440 11 Using Mould Fluxes to Minimise Defects and Process Problems

(b)
3
(a)

Breakout frequency, %
90

2
q , kJ kg-1

80

70 1

60 0
0 2 4 6 8 10
0 2 4 6 8 10 12 14
H2, ppm
H2 Content, ppm

Fig. 11.23 a Heat extraction from the mould as a function of H2 content of steel showing the effect
of degassing (● = de-gassed; ▲ = not de-gassed), and b frequency of sticker breakout as a function
of hydrogen concentrations in the slag film (permission granted, EDP Sci., re-drawn from [86])

sticker breakout. The effect of H2 concentration in the steel can be seen in

Fig. 11.23 where degassing of molten steel results, sequentially, in lower H2
contents, lower porosity in slag film, higher horizontal heat flux and thicker shells
[86] The incidence of sticker breakouts was found to increase linearly with the
measured H2 content in the slag film [86].
The principal sources of hydrogen in the metal are (i) moisture in the mould
powders and refractories and (ii) water/steam leaks in the steel processing. It has
been proposed [87, 88] that moisture in the mould powder results in the formation
of hydroxyl ions (OH−) formed by the reaction:

H2 O(g) þ ðO2 Þslag ¼ 2ðOH Þslag ð11:6Þ

These hydroxyl ions then react with Al or Si in the metal (denoted Al)

2ðAlÞ þ 6ðOH Þslag ¼ 6½H þ 4ðO2 Þslag ð11:7Þ

ðSiÞ þ 4ðOH Þslag ¼ 4½H þ 2ðO2 Þslag ð11:8Þ

Ueshima et al. [87] pointed out that sticker breakouts can occur in de-gassed
Si-killed steels, whereas, they rarely occur in de-gassed, Al-killed steels. The slag
film in contact with the Si-killed de-gassed steel was found to contain micro-pores
(Fig. 11.24b) whereas the slag film in contact with the de-gassed, Al-killed steel
was found to contain little porosity (Fig. 11.24a) [87, 88]. In Al-killed, de-gassed
steels, the strong reducing power of Al results in the consumption of the hydroxyl
ions and there is no accumulation of OH− ions [87, 88]. In contrast, in Si-killed,
de-gassed steels the reductive power of Si is not sufficient to consume enough of the
hydroxyl ions and there is an accumulation of OH− ions in the slag. In the latter
case, when mould slag infiltrates into the mould/strand channel, the slag is in
contact with the cold shell and thus becomes supersaturated with OH− ions which,
subsequently, form H2O (g) in the slag film by the reaction:
11.4 Sticker Breakouts 441

Fig. 11.24 Cross sections of slag film taken from below meniscus in the mould a with Al-killed,
de-gassed steel and b Si-killed de-gassed steel (permission granted, NSSM Corp. [88])

2ðOH Þslag ¼ H2 O(gÞ þ ðO2 Þslag : ð11:9Þ

It has been reported that sticker breakouts occur when the steel contains >7 ppm
hydrogen [87, 88, 92, 93]. The partial pressure, pH2, can exceed 1 atm. during steel
solidification and then passes into the molten slag where it forms a porous slag film.

Cyclical Changes in Casting Speed

Sticker breakouts occur when the total heat flux is low since this leads to a thin shell
which is too weak to withstand the ferrostatic pressure. Sticker breakouts can occur
as a result of cycling the casting speed. The following changes all occur when the
casting speed is increased ðVc "Þ:
(i) The mould temperature (Tmould) increases (Fig. 11.25a, b).
(ii) The horizontal heat flux density (qhor in W m−2) increases but the total heat
flux qtot (in J) decreases because of the reduced residence time (see Sect. 3.3).
(iii) A thinner shell is produced ðdshell #Þ (see Sect. 3.3).
(iv) The increased qhor results in partial melt-back of the solid slag film (i.e. ds #).
This sequence of events can be summarised as ðVc "! Tmould "! qhor "! ds #Þ
but ðqtot #! dshell #Þ:
Inspection of Fig. 11.25a, b indicates (i) dshell is thin when Tmould is high and
(ii) that the changes to the slag film (ds) and hence (dshell), are not instantaneous but
take time to adjust to the changes in qhor. The danger of sticker breakout is greatest
when (i) dshell is thin (i.e. when Tmould is high in Fig. 11.25) and (ii) when ds is
eqm
thicker than the equilibrium (or steady state) value (i.e. dst [ dseqm or Tmould
t
[ Tmould .
The periods of the cycle where there is a danger of sticker breakout in Fig. 11.25a, b
are marked by A.
442 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.25 Mould temperatures (related to heat flux) as functions of casting speed for a cycle
where a casting speed is first increased then decreased and b where Vc is first decreased and then
increased (permission granted, NSSM Corp. [89])

11.4.2 Causes, Mechanisms

11.4.2.1 Formation of a Pseudo-Meniscus

It has been proposed that the pseudo-meniscus is formed following the occurrence
of a “constraint” [67]. It was also proposed that this pseudo-meniscus separated
from the sound shell during periods of negative strip (i.e. when the strand is
descending faster than the mould) but the shell is “healed” in periods of positive
strip. The shear stress increases as the pseudo-meniscus moves down the mould.
Breakout finally occurs when the shear stress exceeds the yield stress of the steel.
This mechanism of rupture and repair is depicted in Fig. 11.26. An analysis of the
various forces acting on the shell showed that the shell does not rupture under
normal conditions but when a constraint occurs, the rupture will propagate down-
wards once the shear stresses exceed the yield stress. A 3-dim., non-steady state,
heat transfer model was made to simulate the proposed mechanism; this obtained
good agreement between predicted and measured temperature transitions [67].

Fig. 11.26 Schematic drawing showing the mechanism for the evolution of sticker breakouts [67]
A = Stuck shell: B = normal shell before constraint: C = newly solidified shell: D = ripple marks.
Note the drawing does not include a slag film (permission granted, ISS/AIST, [67])
11.4 Sticker Breakouts 443

(a) (b)

Fig. 11.27 Schematic description of the start of a breakout, a downward mould movement
b upward mould movement pushing up the slag rim (purple); (courtesy of CD Dacker)

In some cases, the step-by-step withdrawal of the damaged steel shell can be
observed in the mould and a breakout can be avoided by employing a sharp
reduction in the casting speed. If the mould powder used is designed for soft
cooling (i.e. by using a high-basicity powder), a slag rim will be created in the
meniscus region. When the friction is too high the upper part of the steel shell will
get stuck to the mould and the shell is ripped off when the lower part follows the
strand (Fig. 11.27a). The upper, stuck part will oscillate with the mould but cannot
move freely. The ripped area will be filled with steel that freezes to form a thin
shell.
The effect is that the ripped shell is pushed up when the mould is ascending,
which, in turn, pushes the slag rim into the powder layer (Fig. 11.27b). In this way,
sticker breakouts can be avoided by observant operators, especially in slab casting,
where the signs are more obvious.

11.4.3 Ways of Dealing with Sticker Breakout

11.4.3.1 Use Sticker Detection Systems

It is customary to minimise the number of sticker breakouts by using a sticker

detection system [1, 2, 52, 58]. These systems consist of an instrumented mould
containing several rows of thermocouples). During potential breakouts, the mould
444 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.28 Schematic drawing showing the evolution of mould temperature readings during a
sticker breakout (permission granted, IOM/Taylor & Francis [58])

temperatures show the characteristic “crossover” in readings, (as shown in

Fig. 11.28) associated with the formation of a pseudo-meniscus [58, 94]. This
behaviour triggers an alarm allowing the plant personnel to take remedial action.
The major problem, to date, has been the large number of false alarms and con-
siderable effort has been devoted to reducing the incidence of false alarms. Success
rates, where 80% of the alarms are genuine breakouts, have been claimed [58, 94].
Monitoring of the mould friction has also been proposed for the detection of sticker
breakouts [95].
When a breakout alarm is sounded, remedial action must be taken. The first
action is to reduce the casting speed which allows the shell to thicken and
strengthen. However, in thin slab casting involving casting speeds of >4 m min−1,
there is very little time to take the remedial action [96]. Furthermore, the reduction
in casting speed requires additional responses, i.e.
• The reduction of water flow rates in the secondary cooling zone;
• Adjustments to the mould level control since if the metal inflow is not reduced,
an overflow is developed which is followed by a period of poor mould level
control [96].
POSCO has installed a sticker-recovery system which is outlined in Fig. 11.29
which has been estimated to save 6 million $ p.a. [96].
Siemens VAI has developed a breakout prevention system called SIMETAL
Mould Expert. The system works with two or three rows of thermocouple and
sophisticated software which determines when there is an obvious risk of break out
and seeks to avoid false alarms. When the system has identified a breakout situation
the system automatically reduces the casting speed to a low level. After a healing
period, the casting speed is manually increased, step-by-step back, to normal
production.
11.4 Sticker Breakouts 445

(a)
Mould level Fuzzy SEN stopper
controller logic Position
(b)
Mould level
sensor
System (BOPS)
Vc control

MTM

Sec. cool.
control

Fig. 11.29 Schematic diagrams showing a the steps taken in the sticker-recovery procedure and
b traces of casting speed and mould level during a period of sticker recovery using the
sticker-recovery procedure; BOPS Breakout Prediction System MTM Mould Thermal monitoring
(permission granted, ISIJ [96])

Another advantage, of using a breakout system based on instrumented moulds, is

that it can be used as a detailed tool for evaluation of plant trials with, for example,
new mould powders.

11.4.3.2 Use Casting Powders Which Help Form a Thicker,

Stronger Shell

(i) Use mould powders forming a thin glassy slag film with a low Tbr; this
results in a thicker (and hence) stronger shell. This is especially important for
steels which solidify to form a soft, ferritic, low-strength, shell.
(ii) Avoid pick-up of ZrO2 by the slag which results in a high Tbr, a low heat flux
and a thin shell.
(iii) Degas the liquid steel to reduce the hydrogen content since pores of H2 and
H2O in the slag film impart a low thermal conductivity which leads to a thin
shell.
(iv) Minimise H2 levels by (a) storing mould fluxes in dry conditions (b) pre-
heating the casting powder to reduce moisture contents (iii) inspect steel-
making process for possible water/steam leaks
(v) Surprisingly, Ueshima et al. [87] reported that the incidence of sticker
breakouts associated with high H2 contents could be reduced by using a
mould powder with a very high basicity (%CaO/%SiO2) = 1.8 (this would
be expected to produce a crystalline slag film with high Tbr, resulting in a thin
shell) and the powder would be vulnerable to water absorption.
446 11 Using Mould Fluxes to Minimise Defects and Process Problems

11.4.3.3 Minimise the Numbers of TiN and ZrO2 Particles in the Steel

TiN and ZrO2 have low solubilities in molten mould slag so they remain largely
undissolved and tend to agglomerate in the metal phase. Consequently, it is possible
that sticker breakouts in Ti-containing steels can be reduced by the following:
(i) Maintaining a deep slag pool which will dissolve more of the TiN and will
keep TiN particles away from the carbon (in the sinter bed or floating at the
top of the slag pool).
(ii) Minimising the TiO2 content in the casting powder (TiO2 has been used to
produce crystalline phases in F-free fluxes as an alternative to cuspidine) [75]
and where TiO2 is used in the powder, it is necessary to have a basicity of <1
to prevent perovskite formation [79].
(iii) Use a tundish two-layer cover with a melting slag on top of the steel followed
by an insulation powder. The molten slag layer will help to reduce the
pick-up of nitrogen from the atmosphere.
The same steps can be taken with ZrO2 particles (some mould fluxes contain 2%
ZrO2 to minimise SEN erosion).

11.5 Oscillation Marks (OM’s)

11.5.1 Characteristics of Oscillation Marks

Oscillation marks are regular, transverse indentations formed on the steel surface
(Fig. 11.30). Oscillation marks are considered to be defects. Although oscillation
marks do not impose much threat in themselves, cracks tend to form in the base of
the mark due to micro-segregation. The severity of these cracks tends to increase as
the depth of the mark (dOM) increases. The segregation occurs as low-melting metal
is squeezed out during negative strip time (when the mould is descending faster
than the shell) [97, 98].
Oscillation marks are a natural occurrence and are related to the oscillation
characteristics. They have also been found on the surface of steel cast with rapeseed
oil. However, these marks were found to be shallower than those observed when
using mould fluxes. However, marks at regular spacing can also be observed in
static (ingot casting); these are known as “ripple marks” and are probably formed
by an overflow of the newly formed shell by molten steel (similar to the overflow
mechanism described below) or as folding marks when the shell formation is slower
due to higher casting speed [99, 100]. Ripple marks have also been identified for
static casting in the predictions of a mathematical model which coupled the fluid
flow, heat transfer and shell solidification in the mould [101].
11.5 Oscillation Marks (OM’s) 447

Fig. 11.30 Photographs of oscillation marks on narrow face of stainless steel slabs showing
a regular pattern of marks and b an irregular pattern of marks (courtesy of CA Dacker [102])

The pitch between oscillation marks (LOM) is also determined by the oscillation
characteristics where the theoretical value for the pitch is given in Eq. 11.10 where
s = stroke and f = frequency.

LOM ¼ Vc = f ð11:10Þ

It can be seen from Fig. 11.30 that oscillation marks vary in depth and pitch. In
some cases, these oscillation marks have a regular pattern (Fig. 11.30a), whereas, in
other cases the OM pattern is more irregular Fig. 11.30b [102]. These irregular
marks occurred more commonly when casting peritectic steel grades where the
newly formed shell is both buckled and strong. The variations in depth and pitch
reflect the fluctuations in flow occurring in the mould; when all the factors come
together, a deep OM is obtained but when they cancel one another out, a shallow
OM is formed [103].
The interior of the shell also exhibits oscillation marks (Fig. 11.31) [103–105].
Thus, the internal thickness mirrors the external thickness. The variation in shell
thickness is due to the variations in shell growth during the oscillation cycle (i.e.
rapid growth occurring between (tmid n and tearly
p ) and the limited, or zero, growth
mid early
between (tp and tn ).
448 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.31 Breakout shell showing internal depressions matching external depressions (permis-
sion granted, Taylor & Francis, [103])

11.5.2 Mould Oscillation

The mould is oscillated to prevent the steel shell from sticking to the mould. The
oscillation characteristics are usually defined in terms of the positive (tp) and
negative strip times (tn) [106] as shown in Eq. 11.11. The negative strip time is the
time in the oscillation cycle when the mould is descending faster than the mould
(Vm > Vc) and is calculated by Eq. 11.12 where f = frequency and s = stroke
length. The positive strip time is the remainder of the oscillation cycle, i.e.
tp ¼ ðtcycle  tn Þ. In terms of the position of the mould and slag rim, the mould is at
its highest position in late tp and its lowest position in early tp. Thus, the mould is
descending throughout tn and is ascending for the major part of tp:

tp þ tn ¼ tcycle ð11:11Þ
11.5 Oscillation Marks (OM’s) 449

   
60 Vc
tn ¼  arc cos : ð11:12Þ
pf psf

It can be seen from Eq. 11.12 that there are two terms involving the frequency:
the first term tends to be dominant and the second term tends to partially cancel the
effect of the first term. Thus, tn increases as f decreases as Vc decreases and s
increases (tn " as f#, s ", Vc#).

11.5.3 Factors Affecting Depth of OM’s (DOM)

The factors affecting the depth of oscillation marks (dOM) have been studied by a
number of investigators using plant observations, hot physical models and math-
ematical modelling studies. The various factors, reported to affect the depth of
oscillation marks, can be divided into the following categories.

11.5.3.1 Oscillation and Casting Variables

Several of casting variables have been shown to affect the depth of the oscillation
mark. There is general agreement on the effect of some parameters affecting dOM
but disagreement in the published literature over the effect of other variables.
(i) There is general agreement that the depth of the oscillation mark (dOM)
decreases as the negative strip time decreases (dOM #as tn #) [43, 107–111,
113, 114] (see Fig. 11.32a).
(ii) It has been reported that the effect of frequency, stroke and casting speed on
dOM can be predicted through their effect on tn, (as given in Eq. 11.12), i.e.
dOM # as tn #as f, # s#, Vc ") [55].
(iii) There is also general agreement on the relationship between dOM and stroke
length, s, where the relation (dOM # as s#) has been proposed in [106–109,
113].
(iv) However, there is disagreement on the reported relations between dOM and
casting speed where (dOM # as Vc ") is supported by [55, 73, 106, 113–117,
119] and (dOM # as Vc #) by [106, 111, 118].
(v) Similarly, there are differences in the reported relationships between dOM
and frequency, f, with (dOM # as, f #) supported by [55, 73] and (dOM # as,
f ") by [46, 107, 108, 113, 114, 119, 120, 121]; however, it has been
reported that the dOM − f relation goes through a maximum at ca. 100 cpm
[111] and this may be due to f having opposing effects on the two terms in
Eq. 11.12. Values of dOM have been reported to decrease as the parameter
(s/f0.5) decreased [108].
450 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.32 Depth of (a)

oscillation mark as a functions
of a negative strip time and
basicity and b frequency;
● = with Ar blow;
O = without Ar blow
(permission granted, a EPD
Sci. [43] and b ISIJ [108],
re-drawn)

(b) 1.4

1.2

1
dOM ,mm

0.8

0.6

0.4

0.2

0
140 160 180 200
Frequency, cpm

(vi) The depth of an oscillation mark has been reported to decrease with
decreasing superheat, DT, (i.e. dOM # as DT #) [108, 118]; this a surprising
result since increased superheat would be expected to delay shell solidifi-
cation (see Sect. 11.5.3.3 below) and hence, lead to an increase in the
separation distance between the slag rim and the shell tip which would lead
to shallower oscillation marks (i.e. dOM # as DT") [55].
(vii) It has been suggested that the separation distance between the slag rim and
the shell tip (drim to tip) is an important variable affecting dOM (with dOM # as
drim to tip ") [55]; note a short stroke would increase drim to tip and thus
decrease dOM.
(viii) A regression analysis of plant data from 27 trials indicated that dOM
decreased when the basicity (C/S) of the mould powder increased (dOM " as
(C/S)") [43] (Fig. 11.32a) and the relation, dOM (mm) = 0.49 + 1.11 tn −
0.30 (C/S) [43]; this was attributed to an increase in both Tbr and slag film
11.5 Oscillation Marks (OM’s) 451

thickness with increasing (C/S) which resulted in a shorter shell and delayed
solidification (see Sect. 11.5.3.3) [43].
(ix) The use of non-sinusoidal oscillation has been reported to decrease dOM
[51, 108]; this is consistent with a decrease in negative strip time (tn #).
(x) Values of dOM were greater when Ar flow was used (cf. when no Ar flow
was applied in the SEN) (Fig. 11.32b); this is consistent with the Ar flow
boosting the vertical heat flux and thereby, increasing the length of the shell
(see Sect. 11.5.3.2). Ar flow has been reported to increase powder con-
sumption (see Fig. 2.15) and thus this finding is consistent with dOM
increasing with increasing powder consumption.
(xi) Oscillation mark depths have been found to increase with mould level
variations [105, 118]; both dOM and the pitch, LOM, decreased for low
positions of the (slag/steel interface) in the mould (Fig. 11.33) [105].
(xii) Reductions (of 50%) in dOM and in the variability of dOM have been reported
when using Electromagnetic devices fitted to the mould (see Sect. 5.8); the
effects of using EMS [50] EMC [50, 51] and EMBr [46] on dOM are shown in
Fig. 11.34a, b and c, respectively. This was attributed to a shorter shell and
delayed solidification resulting from (a) the homogenisation of the steel
temperature with EMS (b) the “pinch force” generated in EMC and (c) the
increased meniscus temperature and the reduced qvert in EMBr.

Fig. 11.33 Schematic diagrams showing the predicted effect of mould level change on depth and
pitch of oscillation marks with dOM and LOM increasing as metal level increases (permission
granted, Eurotherm and H Steinruck [105])
452 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.34 The depth of the oscillation mark as functions of a the molten steel velocity in EMS
[46], b the application of EMC [51] and c the electromagnetic flux density applied in EMBr [46]
(permission granted NSSM Corp.; a [46], b [51], c [46])

11.5.3.2 Reducing the Length of the Solidified Meniscus

It has been suggested that the depth of the oscillation mark can be reduced by
reducing the length of the solidified shell [122]; it can be seen that dOM would be
smaller in the case of Fig. 11.35a than for Fig. 11.35b. The length of the solidified
shell can be reduced by reducing either (i) the vertical heat flux (qvert, by increasing
the depth of the powder bed, or incorporating exothermic agents) (ii) reducing the
horizontal heat flux (qhor, e.g. by using EMBr, EMC or creating a slag film with
high Tbr and fcrys). A shorter, solidified meniscus will also reduce inclusion and
bubble entrapment [122]. Reduced Ar flow rates will tend to decrease oscillation
mark depth [108] (Fig. 11.32b) by reducing qvert (since gaseous convection
increases vertical heat transfer and hence, increases the length of the shell).
11.5 Oscillation Marks (OM’s) 453

Fig. 11.35 Schematic

drawings showing the
formation of a a short
solidified shell and b a long
solidified shell (permission
granted, ISS/AIST [122])

11.5.3.3 Delaying Solidification Further Down the Mould

It was pointed out above, that the distance between slag rim and steel tip (drim/tip) is
one of the key factors affecting the depth of oscillation marks with dOM # as drim/tip ".
Thus reducing the length of the solidified shell can be seen as one way of increasing
drim/tip and thereby reducing dOM. Another way of increasing drim/tip is to initiate
solidification further down the mould. This achieved in ways which are similar to
those used to reduce the length of the shell.

Increasing the Superheat or the Steel Meniscus Temperature

It has been reported that (dOM # as DT") from the results obtained in a sensitivity
study (Table 11.2) [55] but the reverse relation (dOM # as DT#) has been reported in
two studies [108, 118]. The latter finding could be attributed to the effect of powder
consumption on dOM (dOM "as Qs ") However, the application of EMBr has been
reported to increase the steel meniscus temperature by ca. 10 °C [47, 123] and also
to cause a reduction in dOM which provides support for (dOM # as DT"). Applying
EMBr also reduces qvert [124] and leads to a decreased slag pool depth. The
confusion over the effect of superheat probably arises because of the effect of these
competing effects of Qs and qvert.

Reducing Horizontal Heat Flux (qhor)

The horizontal heat flux in the meniscus region can be reduced by using (a) a mould
powder with high basicity (C/S) and Tbr (Figs. 11.6 and 11.7) (b) Meniscus-Free
casting (in which a refractory feeder-head is added to the top of the mould) to
reduce the horizontal heat transfer in the meniscus region [21–23] or
(c) Electromagnetic casting (EMC) where the pinch-effect reduces qhor in the
meniscus region [50, 125].
454 11 Using Mould Fluxes to Minimise Defects and Process Problems

Table 11.2 Summary of the change needed in the various factors to bring about a decrease in the
depth of the oscillation mark (dOM) [55]
Vc DT tn tp f s c η Tbr
dOM # as " " # " " # " # #

Delayed solidification has the benefit of taking the shell away from the turbulent,
metal/ slag interface which results in less slag, inclusion and bubble entrapment.

11.5.3.4 Changing Mould Slag Properties

It has been reported that mould slag properties can affect the depth of the oscillation
marks.

Viscosity (η)

Most information supports the view that viscosity affects dOM through its effect on
powder consumption (Qs); since Qs decreases with increasing viscosity. Thus, dOM
would be expected to decrease as η increases (dOM # as η ") [110, 115, 127].
Additional support for this relation comes from the slag flow mechanism
(Sect. 11.4.4.3) [55] where increased viscosity resists the downward flow velocity
in the slag pool and reduces the depth of the “dimple”. However, the reverse
relation (i.e. dOM # as η #) has been reported in Table 11.2 [55] and in [106, 126].
This confusion may arise because dOM is dependent upon both qhor and Qs and the
following relations apply (dOM # as Qs # ! η " ! (C/S) #) and (dOM # as qhor
# ! (C/S) " ! η #), respectively; thus, the viscosity has opposing effects on these
two factors affecting dOM..

Interfacial Tension (cm/Sl)

It has been proposed that dOM decreases as interfacial tension increases (dOM # as
cm/sl ") since a high cm/sl results in a smaller radius of curvature and hence a smaller
distance between metal and mould [53, 101]. It has also been proposed that cm/sl
influences the pitch of the oscillation marks (LOM) through its effect on the radius of
curvature [128]; it was proposed that LOM was greater for the flow mechanism than
for the overflow mechanism [128] (see Sect. 11.5.4).

Break Temperature (Tbr)

It can be seen from Fig. 11.32a) that dOM decreases with increasing slag basicity
(C/S), (dOM = 0.49 + 1.11tn − 0.30(C/S)). Since the break temperature and fcrys
11.5 Oscillation Marks (OM’s) 455

tend to increase with increasing basicity, it would be expected that dOM would
decrease with increasing Tbr (dOM #as (C/S) "as Tbr "). It might be argued that a
higher Tbr leads to a lower powder consumption and hence a decreased dOM.
Conversely, a sensitivity study using a mathematical model found the reverse
relation with Tbr, i.e. dOM decreased as Tbr decreased (dOM # as Tbr #) as shown in
Table 11.2.

11.5.3.5 Powder Consumption

Values of dOM have been reported to decrease with decreasing powder consumption
(dOM #as Q #) [110, 115, 127, 129]. It has also been reported that oscillation marks
became indistinct and irregular when the slag pool becomes very shallow [130].
Jenkins et al. [131] noted that oscillation marks disappeared when slag infiltration
was cut off. This behaviour could be interpreted as evidence that dOM is affected by
the slag flow (i.e. powder consumption, Q). Jenkins [131] also observed that there
was a pressure increase as the rim approached the slag/steel interface.
Itoyama [108] considered that the shell was deformed by the positive pressure
exerted by the liquid slag during negative strip time. They modified an experimental
caster so that, in addition to the vertical sinusoidal or non-sinusoidal oscillation, a
horizontal oscillation could be simultaneously applied to the narrow faces. It was
found that horizontal oscillation was beneficial in reducing dOM especially using
mode B (i.e. when the distance between the faces was widened following the mould
descent and reduced at the beginning of the descent).

11.5.3.6 Increasing the Strength of the Solidified Shell

Fukuda et al. [127] reported that a stronger shell would result in shallower oscil-
lation mark (Fig. 11.36). This was attributed to the formation of a thicker shell (by
increasing horizontal heat transfer). However, a shell with a finer grain size might

Fig. 11.36 Schematic diagrams showing that a a thin shell results in deeper oscillation marks and
b a thicker shell leads to shallower oscillation marks (permission granted, ISS/AIST, [127])
456 11 Using Mould Fluxes to Minimise Defects and Process Problems

be expected to give the same effect. This finding is in conflict with the view that
dOM can be reduced by decreasing qhor.

11.5.3.7 Effect of Metal Flow on dOM

Eung et al. [132] reported that the depths of hooks (dhook) could be reduced by
improving the metal flow in the mould; this was achieved by modification of the
SEN port design. There is a linear relation between dOM and dhook [113].

11.5.4 Causes, Mechanisms

Two mechanisms have been proposed for the formation of oscillation marks; these
are known as the “Hook or Overflow” type and the “Depression or Folded” type
[104, 133–138]. Recently, another mechanism has been proposed based on the
predicted flows of a mathematical model (Slag flow mechanism) [53, 55].
The curved meniscus (between steel and molten slag) solidifies to form a hooked
shell tip (Fig. 11.57a). The ferrostatic pressure subjects this shell to deformation.
The type of oscillation mark formed (Overflow or Folded) is determined by the
ability of the shell to resist the deformation and this, in turn, is largely determined
by the thickness of the shell.

11.5.4.1 Hook or Overflow OM Mechanism

Hooks are formed when there is insufficient heat supplied to the meniscus region,
especially (i) at lower casting speeds and (ii) in regions near the narrow faces,
where temperatures are at their lowest and (iii) for steels with low liquidus tem-
peratures (e.g. ferritic stainless steels and high-Al steels). If the shell is strong
enough to avoid deformation, the molten steel overflows the steel tip and a new
meniscus is formed which solidifies to give a new oscillation mark (Fig. 11.37a). In
some cases, the overflowing liquid remelts the meniscus tip (Fig. 11.37b) and the
hook, or nail, disappears but in other cases the hook, or nail, survives (Fig. 11.37a).
The critical difference lies in the thickness of the shell. In the first case, the shell is
strong enough to avoid deformation, causing the steel meniscus to overflow the tip
(i.e. overflow mechanism shown in Fig. 11.38a; such a case is shown in Fig. 11.39a
[139]. Several workers have discussed the mechanisms responsible for the forma-
tion of hooks [55, 140–142]. The mean, hook depth (dhook) and thickness (thook)
were found to decrease as (i) dhook # as tn# as Vc " as f " and (ii) thook # as Vc "as as
f "and (iii) the length of hook, lhook # as DT " [140]. Thus, it can be concluded that
dhook and lhook are linearly related to dOM and that dhook and lhook are linked [140].
The number of hooks has been reported to increase with increasing stroke length
[50, 116, 118, 140, 141].
11.5 Oscillation Marks (OM’s) 457

Fig. 11.37 Schematic diagram showing types of oscillation mark formed by a Overflow
mechanism with the formation of nail or hook b Overflow mechanism with remelting of the hook
and c Folding mechanism where the shell is bent back during positive strip time (permission
granted, ISS/AIST [138])

Fig. 11.38 Schematic diagrams showing the evolution of oscillation marks during positive strip
time by a Overflow mechanism and b Folding mechanism (permission granted, IOM/Taylor and
Francis [136])

11.5.4.2 Depression or Folded OM Mechanism

If the shell is unable to resist the deformation, it is folded back during positive strip
time as shown in Figs. 11.37c and 11.38b. The pitch of oscillation marks (LOM) has
been reported to vary according to whether it is an overflow or folded type [128]
OM \LOM and LOM [ LOM
with Lmean : An example of a folded oscillation mark is
folded mean overflow

shown in Fig. 11.39b.

458 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.39 Micrograhs showing a surface structure with a hook (marked with an arrow) in base
of OM in a peritectic, carbon steel [139]; casting direction right to left and b surface structure for a
folding mark at the base of an OM (permission granted, Europ. Comm. Sci.Tech. Publ. [139])

11.5.4.3 Slag Flow Mechanism for OMs

This mechanism was proposed as a result of the analysis of the predicted fluid and
heat flows and shell solidification in the mould [53, 55, 142]. It was observed that
there were changes in the direction of the flow in the slag pool at different parts of
the oscillation cycle. This is particularly important in the region of the mouth of the
channel between shell and mould. These changes can be seen in Fig. 11.40 where
the following stages occur:

Fig. 11.40 Schematic diagram showing the direction of slag flow at different parts of the
oscillation cycle a midway through, tp b halfway through tn c at lowest position of mould in early
tp d at highest position of mould in late tp (permission granted, ISS/AIST, [55])
11.5 Oscillation Marks (OM’s) 459

(i) When the mould is ascending, the flow in the slag pool is hot (tropical) and
radially outward and there is little, or zero, shell growth at this time
(Fig. 11.40a).
(ii) When the mould and slag rim reach their highest position (in tplate ) there is a
tide change and, subsequently, the slag flow becomes downward and since
this flow originates in the cooler parts of slag pool it is cold, or arctic, flow
(Fig. 11.40b, c); this cold convective flow causes increases in both the heat
flux and shell growth which increase gradually as the mould descends further
during tn (Fig. 11.40a, d).
(iii) The mould descends throughout tn and reaches its lowest point in early tp and
during the period (tlate
n to tp
early
) the pressure of the cold downward flow in the
slag pool causes the formation of a “dimple” in the liquid adjacent to the shell
tip; this dimple solidifies rapidly because of the cold (arctic) convective flow
and the dimple is the initial oscillation mark which deepens during further
solidification (Fig. 11.40b, c).
Thus, in summary, oscillation marks are formed because of (i) tide changes in
the direction of flow in the slag pool and the gradual development of a cold,
convective flow when the mould is descending and (ii) the pressure associated with
the cold downward flow creates a “dimple” in the liquid adjacent to the shell tip and
the dimple rapidly solidifies to form the initial mark.
It was concluded that (i) the distance between the slag rim and shell tip (drim/tip)
is an important factor (since dOM " as drim/tip #) and (ii) the effect of casting
parameters can be predicted from their effect on tn [55]. Table 11.2 was derived
from modelling predictions. The mechanism proposed for the formation of hooks is
shown in Fig. 11.41 [55].

11.5.5 Ways of Dealing with Deep OMs

The depths of oscillation marks can be reduced by decreasing (i) negative strip time,
tn (ii) the vertical heat flux, qvert (iii) the horizontal heat flux, qhor and (iv) powder
consumption, Qs; these are discussed below. However, some variables affect these
parameters in different (and sometimes, in opposing) ways. An example of this is
that high, slag viscosity causes (i) (η " ! Qs # ! dOM #) and (ii) involves low
basicity (C/S) slags giving (η " ! (C/S#) ! qhor "- ! dOM "). A second example
is increasing superheat, DT, which delays solidification (dOM #) but reduces slag
viscosity (η # ! Qs" ! dOM ").

11.5.5.1 Reduce Negative Strip Time (tn)

It can be seen from Eq. 11.12 above that the negative strip time (tn) can be reduced
by decreasing the stroke length (s #) or by increasing the casting speed (Vc ").
460 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.41 Schematic diagrams depicting the evolution of hook formation during the oscillation
cycle a at tstart mid late
n , b tn , c at tn (near lowest point of mould) and d at tmid
p (permission granted,
ISS/AIST, [55])

The effect of frequency on tn is more complicated since Eq. 11.12 contains two
terms involving the frequency, with the first term being the dominant term and the
second term tending to cancel out the effect of the first term. Overall, tn tends to
decrease as frequency increases (tn # as f ").

11.5.5.2 Increase the Distance Between Base of the Slag Rim and Steel
Tip (drim/tip)

The argument here is that if drim/tip can be increased, it will reduce the depth of the
dimple formed next to the shell tip. This could be achieved in the following ways:
(i) Decreasing the stroke length (s #).
(ii) Increasing mould slag viscosity (η") resists the downward, flow velocity and
hence decreases the depth of the dimple and thus, reduces dOM but, as explained
above, glassy, low basicity slags also cause increases in both qhor and dOM.
(iii) Delaying solidification to a position further down the mould and which
should also help to reduce inclusion levels; this can be achieved by
– increasing superheat or meniscus temperature.
– using Meniscus-free casting which uses a refractory to reduce the hori-
zontal heat flux in the meniscus region of the mould [21–23].
– Using EMC which reduces the horizontal heat flux in the meniscus region.
11.5 Oscillation Marks (OM’s) 461

11.5.5.3 Reduce the Length of the Shell Tip

As stated above, this too may be regarded as increasing the distance between slag
rim and shell tip (drim/tip). Reducing the length of the shell tip should also help to
decrease slag, inclusion and bubble capture; this can be achieved by reducing the
vertical heat transfer by
(i) Using a deeper powder bed.
(ii) Reducing the thermal gradient in the powder bed by using casting powders
containing exothermic agents [13].
(iii) Using Electromagnetic braking [125] to slow down the metal flow and
decrease the metal penetration into the mould; EMBr has been reported to
cause a 30% decrease in vertical heat transfer [124].
(iv) Reducing the argon flow rate [108].
(v) Increasing slag viscosity will reduce convective heat transfer in the slag pool
but will also reduce powder consumption.

11.5.5.4 Adjusting Mould Slag Properties

There is some uncertainty about the effects of changing mould slag properties, but it
has been reported that dOM can be decreased by increasing the interfacial tension
(cmsl); this can best be achieved by lowering the S content of the steel but reducing
the Na2O, K2O and Li2O contents of the casting powder should also be helpful. The
result presented in Fig. 11.32a [43] indicated that dOM decreased with increasing
basicity; however, increased basicity would also be expected to promote the
sequence ((C/S)" ! η# ! Qs" ! dOM") The results in Fig. 11.32a suggest that
the delayed solidification resulting from the decrease in qhor (and increase in drim/tip)
is greater than the opposing effect of increased powder consumption).

11.5.5.5 Use of Mathematical Models Covering OM Formation

Mathematical models have reached the stage of development where they can
provide an insight into the processes and mechanisms occurring in the mould.
Furthermore, it is shown above that there are sometimes opposing effects taking
place which affect dOM. Thus, mathematical models may prove the best way of
predicting the overall effect. Mathematical models covering oscillation mark for-
mation have been reported by several groups [53, 55, 106, 113, 119, 140, 142–144].

11.6 Transverse and Corner Cracking

Transverse and corner cracks occur on both the surface (Fig. 11.42) [139] and in the
interior of steels (Fig. 11.43), [139, 145]. Transverse cracks result from
462 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.42 Micrograph showing a transverse crack on the surface of a micro-alloyed, low carbon
steel (permission granted, Europ. Comm. Sci. Tech. Publ. [139])

• Cracks formed in the base of deep oscillation marks as a result of stress con-
centrations arising from micro-segregation [97], (Fig. 11.42[139]).
• Surface strains occur during bending in a low ductility, temperature region for
crack sensitive steel grades (especially, peritectic, micro-alloyed, low carbon
steels); these regions occur, in the upper part of the strand for a straight mould
machine and, alternatively, at the straightening unit at the end of the strand. The
crack occurs at the site (surface or interior) which is experiencing the ductility
trough (Fig. 11.42 or 11.43, respectively).
• Improper taper in the mould.
• Improper process parameters and the use of an unsuitable mould powder.

Fig. 11.43 Micrograhs showing transverse cracks in the interior of a micro-alloyed peritectic,
carbon steel (permission granted, Europ. Comm. Sci. Tech. Publ. [139, 145])
11.6 Transverse and Corner Cracking 463

11.6.1 Factors Affecting Transverse Cracking

One form of transverse cracking is associated with the bending and straightening
operation in continuous casting where the surface of the steel experiences a tensile
stress. There is loss of ductility in the steel between 900 and 700 °C and a “ductility
trough” is encountered in this temperature range (Figs. 11.44 and 11.44).
Transverse cracking occurs when the bending and straightening processes coincides
with the ductility trough. The usual strategy adopted is to prevent the steel from
falling into this temperature range during the straightening process. Further details
on this problem can be obtained from the review carried out by Wolf [146].
Transverse cracking on the upper surface of slabs has proved a serious problem
when casting micro-alloyed, low carbon, peritectic steels (e.g. pipe line steels X70)
[154]. The problem has been attributed to “abnormal grain growth of the austenitic
grains, in combination with precipitation of AlN and NbC, in the grain boundaries”.
During straightening and bending of the strand, these grains are subjected to a very
slow deformation speed which causes creep and, subsequently, transverse cracking.
This finding (Fig. 11.45) has been corroborated in a number hot ductility tests using
a Gleeble machine [145, 155].
It can be seen from Fig. 11.44 that the ductility drops dramatically at low strain
rates; this explains the problems encountered at the straightening unit where the
deformation rate is low. It also shows that the variations in cooling on the strand
surface (resulting from contact with the support rolls and water cooling) tend to

Fig. 11.44 Schematic

drawing showing ductility of
micro-alloyed, peritectic steel
as function of temperature,
strain rate and cooling curves
(permission granted, [145])
464 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.45 Schematic drawing showing the micro mechanism of hot ductility in micro-alloyed
steels, after [147]

expand the low-ductility area. The origin of the low-ductility trough is explained in
more detail in Fig. 11.45 [147].
For the explanation, the ductility behaviour has been divided into four regions:
(1) HDH (High Ductility, High temperature): in this region the austenite grains
have a high ductility since no precipitation takes place.
(2) Trough in austenite: Precipitation of micro alloying elements (e.g. Nb, V, Al,
N) takes place (mainly in the grain boundaries) where the slow tensile strain
results in creep and the formation of voids; this causes a marked reduction in
ductility. The effect is accentuated by abnormal grain growth of the austenite
grains.
(3) Trough in ferrite: The pro-eutectoid formation starts in the grain boundaries
as a very thin film with micro alloy precipitates. The ferrite is softer than the
austenite, so the strains concentrate here and this also becomes a low-ductility
region.
(4) HDL (High Ductility, Low temperature). At lower temperatures the ferrite
fraction increases leading to a strong increase in ductility.
The abnormal grain growth of austenite grains can be seen in Fig. 11.46a and
the micro-alloyed precipitates can be seen in the FEG-SEM image (Fig. 11.46b).
Micro-segregation leads to the surface enrichment of residual (or tramp) ele-
ments (Cu, Sn, Sb and As) and also leads to brittleness. Face brittleness (or hot
shortness) is associated with the presence of Cu-rich phases in the surface region
and the application of stresses. This arises because the solubility limit of Cu in
11.6 Transverse and Corner Cracking 465

Fig. 11.46 Micrographs of a surface fracture of tensile test samples a Surface from fracture,
x = 26, b SEM image of fracture surface, x = 50.000 showing precipitates (permission granted,
SWEREA [148])

austenite is <0.2%. If the surface concentration of Cu exceeds this value, it will

form a Cu-rich liquid, which is preferentially oxidised and the resulting liquid
penetrates into the grain boundaries. Other elements, such as Sn and Sb, also
promote the formation of the oxidised liquid phase. Nickel additions are usually
made to alleviate this problem, since Ni increases the solubility limit of Cu in
austenite and favours the occlusion of Cu from the scale [149]. The Copper
equivalent (Cueq) is used as a measure of the propensity to form these liquid phases
which penetrate the grain boundaries (Cueq = %Cu + n(%Sn + %Sb)- %Ni) where
n has a value between 6 and 8). The transverse cracking was found to increase with
increasing copper equivalent (Fig. 11.47) [149, 150].
The effects of mould taper and casting speed have been investigated by
Ogibayashi [45]. The results, shown in Fig. 11.48, show the minimum powder

Fig. 11.47 Schematic 25

drawing showing the
Max crack length,mm

Transverse crack length as a 20

function of the copper
equivalent (Re-drawn from 15
[149])
10

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Cu equivalent
466 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.48 The minimum 0.7

powder consumption to avoid
transverse cracking (Qmin
s ) as
a function of mould taper on 0.6
narrow face cracks are formed

Qs, kg.m -2
if powder consumption
lies below curve 0.5
( for Vc = 1.8 and No cracks at Vc = 1.8 m min-1 but cracks at 2.2 m min-1
( = 2.2 m min−1) but no
0.4
cracks where consumption
above curve (Re-drawn [45]) cracks at Vc =1.8 m min-1
0.3
1 1.2 1.4 1.6 1.8
Taper, %,

consumption, Qmin
s (or liquid slag film thickness, dlmin ) required to avoid cracking at
two casting speeds, Vc = 1.8 and 2.2 m min−1. It can be seen that in their study:
• Qmin
s is reasonably constant (at both speeds) for tapers of <1.4% but increases
steadily for tapers >1.4%.
• Qmin
s increases with increasing casting speed; thus, higher casting speeds will
require a higher powder consumption to avoid cracking (Note that Qmin s
increases with increasing Vc and the situation is further complicated by the fact
that Qs tends to decrease with increasing Vc (e.g. Qslag
s ¼ 0:55=g0:5 Vc ).
A statistical evaluation of 36 heats involving transverse and corner cracks in
micro-alloyed, low carbon steel showed that cracking decreased with decreasing
negative strip time (due to its effect on dOM) [43]. It was also found that it was
important to use a mould with corner shrinkage compensation and a mould powder
with high basicity (to reduce qhor). The transverse crack length (TcrL = crack length
(m/m)) was found to show linear relations (Eq. 11.13) with (i) tn (ii) the parameter,
c* = (narrow side concavity + taper) and (iii) basicity (C/S) [43]. It was also
reported that transverse cracking decreased with increasing slag viscosity [43].

TcrL ¼ 26:8 þ 12:4  tn  1:6  c  5:2  ðC=SÞ ð11:13Þ

The reduction in qhor was obtained using “soft cooling” mould slags (i.e. with
high basicity, Tbr and fcrys) [43]. However, high-Tbr values result in a thin liquid slag
film and a slag film with high fcrys but make the slag film brittle and fragile. Thus,
slag films with high basicity create high frictional forces (which act on the shell)
and, furthermore, tend to fracture easily. Such behaviour can be clearly seen in
Fig. 11.49 [151] where the sharp increase in mould temperature corresponds to the
fracture and loss of the slag film which is gradually replaced (over ca. 8 min) with a
new slag film.
11.6 Transverse and Corner Cracking 467

Fig. 11.49 Mould

temperature (229 mm from
mould top) as a function of
time when casting with a
mould powder with C/S =
1.62; solid line = wide face,
1/4 plane; dashed line =
narrow face midway; the “step
decreases then increases” in
mould temperature correspond
to the influx of molten slag
and fracture of slag film and
the formation of a new slag
film (permission granted,
ISS/AIST, [151])

11.6.2 Ways of Dealing with Transverse and Corner

Cracking

11.6.2.1 Bending at the Low-Ductility Trough for Crack Sensitive

Steel Grades

This problem is particularly severe when casting of micro-alloyed, low carbon

steels and requires significant oxygen scarfing and grinding to remove the surface
cracks. The problem has been minimised by the development of new casting
machines which seek to avoid the low-ductility trough. This can be done in two
ways:
(i) Straightening at temperatures above the low-ductility trough. This is achieved
using proper secondary cooling and a high casting speed, since it is necessary
to maintain a liquid core to attain a reasonably-high temperature [145].
However, the resultant thin shell is vulnerable to bulging between the rolls and
to internal cracking; this has led machine builders to provide better roll support
(using smaller rolls).
(ii) Straightening at temperatures below the low-ductility trough. This procedure
is used for thick-slab casting. For this case, there is no need for a close
distance between the rolls because the strand is fully solidified and has a
strong surface.

11.6.2.2 Improper Taper in the Mould

The taper should match the shrinkage of the shell as closely as possible. An
excessive taper results, sequentially, in overcooled corners, high-mould friction and
corner cracks [64]. An insufficient-taper results in loss of contact between mould
and shell with a consequent reheating of the shell (the “rebound effect”).
468 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.50 Schematic

diagram showing the Segregation
sequence of events leading to Internal Crack initiation
transverse cracking in
secondary cooling zone Propagation-segregation

Propagation, uneven
cooling + shell bulging

Crack growth on straightening

This causes abnormal grain growth and an extra tension in the surface when the
shell and enters the secondary cooling zone (Fig. 11.50). A parabolic taper is
required for the meniscus region but the optimal taper is dependent upon casting
speed, mould length and steel grade [64]; a mathematical model to predict the
optimal taper has been reported [64]. However, in practice, casters tend to handle a
large number of steel grades and use a variety of casting speeds.
The loss of contact in the corners is a problem which can be solved by using a
mould design which compensates for the corner shrinkage [43].

11.6.2.3 Optimisation of Process Parameters and Use of a Suitable

Mould Powder

A number of actions must be taken to minimise the risk for transverse and corner
cracking:
1. Reduce the depth of the oscillation marks by
• Decreasing stroke length, negative strip time (tn) or increasing frequency.
• Increasing the rim to shell tip distance by decreasing both stroke length and
the vertical heat transfer (with a thick bed of mould powder) and the hori-
zontal heat flux by using a “soft cooling” mould powder with a high-break
temperature.
2. Increase the strand temperature to avoid the low-ductility range of the steel (ca.
820 °C) in the straightening region (i.e. reduce heat lost by the strand) by
• Use a casting speed that is high enough to ensure that the surface temperature
is above the critical temperature range.
• Use a mould flux with high values of Tbr and fcrys.
• Reduce the velocity of the water-sprays, or by shutting off the sprays to the
corner of the strand [152].
3. Minimise the concentration of Cu, Sn and Sb in the steel or add Nickel to reduce
the copper equivalent.
11.6 Transverse and Corner Cracking 469

11.6.2.4 Use of Electromagnetic Braking

It has been reported that EMBr reduces the number of transverse, corner cracks; this
was ascribed to the reduction of the surface waves [153]. It was concluded that the
strength of the applied magnetic field has to be adjusted according to the steel grade
being used. [153]. However, the beneficial effect of EMBr may arise from its
reduction of dOM due to delayed solidification resulting from the 10 °C increase in
meniscus temperature.

11.7 Star Cracking

Star (or Spongy) cracking tends to occur in the lower half of the mould.
A photograph of a star crack is shown in Fig. 11.51 [156].

11.7.1 Factors Affecting Star Cracking

Star cracking was, in the past, usually associated with the presence of copper
droplets on the strand [157] due to contact of the shell with the mould. This resulted

Fig. 11.51 Photograph

showing a star crack
(permission granted, Taylor &
Francis. [156])
470 11 Using Mould Fluxes to Minimise Defects and Process Problems

in the copper melting to form a droplet which, subsequently, penetrated into the
grain boundaries where it oxidised. The usual procedure adopted to avoid star and
spongy cracking was to coat the mould with high-melting metals such as Ni, Cr,
Mo, etc. However, Billany et al. [158] observed that star and spongy cracking could
occur in Ni-coated moulds. It was also observed that cracking levels (Fig. 11.52):
• did not correlate with overall heat flux values.
• but did correlate with periods where there were large variations in heat flux
(Fig. 11.52).
It was, subsequently, proposed that star cracking resulted from heat flux varia-
tions caused by a lack of lubrication (i.e. the liquid slag film is too shallow) which
results in stresses which cause the fracture of the solid slag film which requires time
to repair the (see Fig. 11.49 [151]). O’Malley et al. [159] also observed cracking of
the slag film in the bottom half of the mould; this was accompanied by periods of
heat flux variability until the slag film was fully restored. Crystalline slag films
(with high basicity) are prone to fracture.
Meng and Thomas [73] developed a model to account for the fracture of the slag
film in the lower half of the mould, from which they concluded:
• If powder consumption decreases to a critical value (Qcrit, i.e. dl decreases) then
an axial stress is produced which can exceed the fracture strength of the slag
film; this, subsequently, results in film fracture and movement of the slag film.
• The critical powder consumption value (Qcrit) was determined from the viscosity
–T curve.
• A thick, crystalline slag film will result in a hotter shell and a thicker liquid film,
i.e. it will provide more lubrication.

Fig. 11.52 Heat flux variations during casting using two casting powders, L11 (cast 1 periods A–
C and cast 2 periods D–F) and L9 (cast 1 D–F and cast 2A–C); star cracking occurred with powder
L9 and was accompanied by heat flux variability (permission granted, Taylor & Francis [158])
11.7 Star Cracking 471

• Glassy slag films tend to fracture near the mould exit but any advantage is offset
by the fact that (Qcrit) for a glass is higher than that for crystalline slag film.
• A crystalline, slag film will fracture higher up the mould because of the
high-break temperature.

11.7.2 Ways of Dealing with Star Cracking

• Use a mould powder which provides a liquid slag film (of sufficient thickness to
lubricate the shell) throughout the mould. However, this is not a straightforward
task. We can select a slag with a lower Tbr which would be expected to increase
the liquid slag film thickness (dl). However, the resulting, thinner, solid slag film
will lead to a high-heat flux and more rapid solidification of the slag. The latter
would tend to cancel any gains obtained with a thicker liquid film.
• Reducing the overall heat loss in the mould should help to reduce star cracking;
this can be achieved by using a low-conductivity, coating on the mould or by
applying EMBr to reduce heat losses.
• Possibly the most effective treatment is to increase the casting speed since this
will reduce the amount of heat lost in the mould, although it would lead to a
high-heat flux density, the shorter residence time will more than offset this
effect.
• Reduced water flow rates in the mould would also be beneficial.
• Increased superheat could also be beneficial to combat star cracking.

11.8 Depressions

Under certain casting conditions, depressions are formed on the surface of the
strand. Typical examples are shown in Fig. 11.53a, b.

11.8.1 Longitudinal Depressions

11.8.1.1 Factors Affecting the Formation of Longitudinal Depressions

There have been two in-depth studies on the formation of longitudinal depressions
[129, 131]. These depressions had the following dimensions (50–500 mm long)
(1–2 mm wide)  (0.3–1 mm deep) [129] and 10 mm long  2 mm deep,
respectively [131]. The depressions were found to occur with and without entrap-
ped scums [129]. The following observations were made:
472 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.53 Photographs of Longitudinal depressions (indicated by arrows) with cracks on a

stainless bloom (steel 411), a depression with a crack, b depression with a crack penetrated with
liquid steel (courtesy of CA Dacker, SWEREA)

• The depressions corresponded to periods of mould level instability [129, 131]

(Fig. 11.54a)
• The depressions tended to occur at regular intervals (0.5–1 m apart) in regions
where no oscillation marks could be seen (Fig. 11.54b) [131].
• Depressions occurred when the mould level was at its peak (Fig. 11.54b) [131];
slag infiltration in the meniscus region will be blocked, if fluctuations in mould
level exceed (slag pool depth + mould level amplitude) [129].
• Depressions corresponded with low values of the horizontal heat flux [131].
• Cracks were frequently associated with the depressions.
• Metallographic studies of the steel showed that depressions corresponded to
lower cooling rates than the norm [131].
• It was concluded that depressions were caused by irregular, and insufficient, slag
infiltration [129].
• It was suggested that the mould level fluctuations were caused by the erosion of
the refractory used as the stopper rod in the tundish [129].
• Depressions occurred when the SEN immersion (dSEN) depth was too deep; these
workers highlighted the importance of optimising the SEN immersion depth
(dopt opt opt
SEN ) since, if dSEN [ dSEN this leads to depressions and if dSEN \dSEN this will
give turbulent metal flow and slag entrapment. Kim et al. [129] proposed that the
opt
mould level turbulence index (It) was helpful in determining dSEN (Eq. 11.14
where w and t are the mould dimensions and Rc is the casting speed in tonne
min−1. As It increases dSEN should be increased.
11.8 Depressions 473

(a)
4
(b)
Depression Index, %

0
0 1 2 3 4
Mould level fluctua on, ± mm

Fig. 11.54 a Depression index as a function of mould level fluctuations when casting 304
stainless steel at Vc = 0.8–1.0 m min−1 (re-drawn after [129]) and b the appearance of the slab;
depressions (─) occur when the oscillation marks (│) disappear and also show that these periods
occur when mould level is at its highest point [131] (permission granted, ISS/AIST [129, 131])

It ¼ 0:33ðw=tÞRc Vc ð11:14Þ

• Kim et al. [129] concluded that as depressions appeared to be associated with

insufficient slag infiltration. They replaced mould powder A, with powders B
and C (B provided higher powder consumption through an increase in the
(CaO/SiO2) ratio and the Na2O content); Powder B was adopted since Powder C
led to deeper oscillation marks (Fig. 11.55) possibly because of the high-powder
consumption.
It should be pointed out that if insufficient slag infiltration is responsible for
depressions, this could arise because the carbon levels in the powder are too high.
This would cause
• A reduction in the melting rate and powder consumption.
• Incorporation of excess carbon into the slag rim.

11.8.1.2 Causes, Mechanisms

There seem to be general agreement that longitudinal depressions are associated

with
• Large fluctuations in mould level.
• Insufficient slag infiltration at some periods.
Jenkins [131] proposed the following mechanism:
474 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.55 Effect of mould (a) 1

powder consumption (Qt) on

Depression Index,
a depression index and 0.8
b depth of oscillation mark,
using three mould powders 0.6
(permission granted,
ISS/AIST re-drawn from 0.4
[129]
0.2

0
0.3 0.4 0.5 0.6
Q t , kg tonne-1

(b) 0.4

0.3
dOM, mm

0.2

0.1
0.3 0.4 0.5 0.6
Q t , kg tonne-1

(1) A rising mould level causes deeper oscillation marks (Fig. 11.56a) (in line with
Aigner’s findings [105] see Fig. 11.33)
(2) Further rises in mould level result in narrowing of the rim/shell gap
(Fig. 11.56b) and the formation of “glaciation marks”.
(3) Even further rises in mould level cause freezing of the shell, above the slag rim,
resulting in blocking off of slag infiltration (Fig. 11.56c).
(4) Then the shell captures the slag rim and drags it down the mould as the mould
level drops (Fig. 11.56d); there is little, or no, slag infiltration during this period
and glaciated regions (i.e. with no OMs) are formed.
(5) As the mould level continues to drop, a new slag film is formed and normal
oscillation marks reappear on the strand (Fig. 11.56e).
Jenkins [131] was uncertain as to whether the captured rim was held in place
during its descent in the mould or whether it escaped leaving an imprint on the shell
filled with gas and slag. In both cases, the depression would result in lower hori-
zontal heat transfer to the mould and this is consistent with the observations.
11.8 Depressions 475

Fig. 11.56 Proposed mechanism for the formation of longitudinal depressions (permission
granted, ISS/AIST [131])

11.8.1.3 Ways of Dealing with Longitudinal Depressions

1. Mould level variations—the action taken will depend upon what is causing the
mould level fluctuations;
– If it is due to strand bulging, the use of a mould powder with a lower Tbr
which produces a lower crystalline fraction in the slag film would help (but
maybe at the expense of more longitudinal cracking)
– Kim et al. [129] attributed mould level variations to erosion in the tundish
stopper rod and decreased the incidence of depressions by improving the
corrosion resistance of the stopper rod.
– Jenkins [131] obtained better mould level control by “detuning” the
controller.
2. Improving slag infiltration (i.e. powder consumption) by reducing the slag
viscosity but this may cause deeper oscillation marks.
3. In some cases, the carbon content of the powder may be too high resulting in a
low-melting rate and a low powder consumption value.
4. “Soft cooling” with a high basicity mould powder has also proven to be
effective.
476 11 Using Mould Fluxes to Minimise Defects and Process Problems

11.8.2 Transverse Depressions

Jenkins [160] observed transverse depressions (up to 10 mm deep) were formed in

the first few metres of the cast when casting billets of high-C steel (Fig. 11.57).

11.8.2.1 Published Information

Jenkins made the following observations concerning the casts exhibiting transverse
depressions:
• Depressions tended to be prevalent in the first few metres of the cast.
• In these casts, the mould temperatures tended to be low for up to 60 min of
casting.
• Depressions correlated with mould level fluctuations and tended to occur when
the mould level was rising [160].
• Depressions did not correlate with the movement of the strand.
• Depressions were assumed to be associated with the movement of slag rim/film.

11.8.2.2 Causes, Mechanisms

Since it was associated with mould level variations, it was assumed that the
transverse depression mechanism was similar to that for longitudinal depressions
and involved the capture of the slag rim [160] (see Fig. 11.58). During large mould
level changes, the rising shell envelopes the slag rim and cuts off slag infiltration; it
then captures the slag rim and drags it down the mould. The captured slag rim is a
good thermal insulator and local mould temperatures drop suddenly as it passes
down the mould.

Fig. 11.57 Photograph

showing transverse
depressions formed in HC
steel billets (permission
granted, ISS/AIST, [160])
11.8 Depressions 477

Fig. 11.58 Proposed mechanism for the formation of transverse depressions, where mould level
changes allow the rising shell to capture the slag rim (permission granted, ISS/AIST, [160])

11.8.2.3 Ways of Dealing with Transverse Depressions

Jenkins [160] found that the incidence of transverse depressions could be signifi-
cantly reduced by allowing an additional 30 s before adding the operating powder;
this allowed the starter powder to form a glassy slag film and restricted the size of
the rim (or rope) formed.
It must be emphasised that this type of problem tends to be rare today because of
the development of more sophisticated, mould level control systems.

11.8.3 Off-Corner Depressions

Off-corner depressions or “gutters” are depicted in Fig. 11.59a. They have been
observed in the casting of stainless steel and plain carbon (0.12%C) steels [161].
The depressions run in the longitudinal direction and may contain sub-surface
cracks.

11.8.3.1 Published Information

Thomas et al. [161] carried out an investigation of off-corner depressions which

included finite-element modelling studies.
478 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.59 Schematic drawings showing a the appearance of off-corner depressions [161] and
b how the thin shell is buckled by excessive mould taper (permission granted, ISS/AIST, [161])

11.8.3.2 Causes, Mechanisms

Thomas et al. [161] concluded that the depression was caused by:
• A hot, thin shell in the corner region.
• Cyclic bending of the shell due to bulging between the rolls, resulting in the
bending of the thin, off-corner region of the shell.
• Improper mould geometry, no corner-shrinkage, compensation of the narrow
face and no compensation for parabolic solidification.
Thomas et al. [161] suggested that the hot, thin shell could be formed by
inadequate liquid slag feeding. However, it is also possible that the thin hot shell
could be caused by the nature of the metal flow, as shown in Fig. 11.18 in
Sect. 11.3.2.2. An excessive taper will then bring the narrow face close to the shell
and which will then cause buckling of the shell in the corner region of the wide face
(Fig. 11.59b).
Li and Thomas [64] pointed out that the shell in the corners of billets can be bent
out by ferrostatic pressure which induces stresses perpendicular to the direction of
dendritic growth; this leads to sub-surface cracks located 10–15 mm from the
corner. These stresses increase with increasing casting speed and can lead to hot
tears in extreme cases. Producing cold corners can help to minimise this problem.

11.8.3.3 Ways of Dealing with Off-Corner Depressions

Thomas et al. [161] suggested that this problem could be treated in the following
ways:
• Minimise the formation of a hot, thin shell in the corner region by
– Improving the metal flow pattern.
– Reducing the superheat.
11.8 Depressions 479

– Ensuring uniform spray intensities on shell perimeter.

– Increasing powder consumption.
• Ensuring taper on the narrow face is correct for the steel being cast.
• Reducing bulging in both wide faces by maintaining roll alignment.
• Compensating for corner shrinkage by better design of the narrow face mould
[43].

11.9 Overflows

11.9.1 Factors Affecting Overflows

When overflows (or C-type effects) occur, the overflowing steel forms a protruding,
double-skin (typically, with a thickness of 0.3 mm); the overflow has a higher
carbon content than that of the strand [162]. These C-type effects have a “tounge-
like” appearance. It can be seen from Fig. 11.60 that the overflow corresponds to a
period where the oscillation marks disappear.

Fig. 11.60 Schematic diagrams of C-type effects showing a general appearance b steel surface
profile of overflow for X–X section in (a) and c mould level variations corresponding to Fig. 11.60
(a) (permission granted, ISS/AIST, [162])
480 11 Using Mould Fluxes to Minimise Defects and Process Problems

11.9.2 Causes, Mechanisms

These overflows result from sudden changes in mould level. If the slag pool has
insufficient depth, the sudden rise in mould level results in the following sequence
of events [162]:
• It brings molten steel into contact with the carbon-rich, sinter layer; this results
in carbon pick-up by the steel and the formation of a low-melting steel.
• The molten steel cuts off the flow of molten slag into the mould/steel channel in
a similar way to that shown in 11.58d, e; this results in the disappearance of
oscillation marks.
• When the mould level drops, the higher C, low-melting steel flows over the
previously formed shell producing a second skin of steel.

11.9.3 Ways of Dealing with C-Type Effects

11.9.3.1 Increase the Depth of the Slag Pool

The slag pool depth can be increased by

• Reducing the vertical heat flux by increasing the bed thickness, or incorporating
exothermic agents into the casting powder.
• Increasing the melting rate; this can be achieved reducing the carbon content of
the mould powder.

11.9.3.2 Reduce Carbon Content of the Mould Powder

Reducing the carbon content of the mould powder results in

• A reduction of the carbon pick-up by the steel.
• An increase in melting rate of the mould powder which, in turn, will assist the
formation of a deeper slag pool.

11.10 Slag, Gas Entrapment and Sliver Formation

Entrapment (of slag, or gas) is also referred to as, entrainment, engulfment and
emulsification. The entrapment of slag, oxides and gas bubbles remains a serious
problem, especially as casting speeds are raised to improve productivity. Mould
slag, non-metallic inclusions (e.g. Al2O3) and gas bubbles get trapped by the newly
11.10 Slag, Gas Entrapment and Sliver Formation 481

Fig. 11.61 Micrograph

showing slab surface with
porosity due to argon
injection (permission granted,
Europ. Comm. Sci. Tech.
Publ. [139])

formed shell in the meniscus region. An example of gas entrapment is given in

Fig. 11.61. These trapped inclusions and gas pores reduce the mechanical strength
of the cast steel. In the case of inclusions, their thermal expansion coefficients are
much lower than those of the steel matrix. Thus, when the steel is heated, stresses
build up around the inclusion, which result in local cracks, with consequent loss in
mechanical strength of the steel. Slag entrapment has become much more important
in recent years with the use of increased casting speeds to meet productivity
demands. The velocity and turbulence of the metal flow in the mould is a key factor
in all three forms of entrapment. Entrapment can occur by a variety of mechanisms,
[163, 164]; these various mechanisms are discussed first and then the slag, gas and
inclusion entrapment are discussed, individually, in Sects. 11.10.1, 11.10.2 and
11.11.3, respectively.
Since entrapment is so closely tied to metal flow in the mould, it is not surprising
that much of our knowledge of the mechanisms of entrapment have been derived
from water-modelling and mathematical-modelling studies. These room tempera-
ture, physical models use the Froude and Reynolds numbers to represent the metal
flow pattern and meniscus level instabilities and the Weber number to represent the
effects of interfacial tension [163, 164]. However, the similarity criteria are not well
satisfied (especially in the case of the Weber No.) in these water models [163, 164];
thus, some of the information obtained tends not to be quantitative.

11.10.1 Metal Flow Conditions Leading to Entrapment

The metal flow leaving the SEN collides against the mould wall; this creates a
reverse flow across the metal surface. The reverse flow is accompanied by surface
waves, vortex formation and turbulence. The high, interface velocities shear the
slag into filaments which, subsequently, detach and are released into the metal
stream. Slag entrapment increases with increasing meniscus fluctuations [163].
482 11 Using Mould Fluxes to Minimise Defects and Process Problems

Entrapment can occur by a variety of mechanisms. Hibbeler and Thomas [163,

164] have published an excellent review of both the various mechanisms in play
and the conditions promoting individual mechanisms [164, 165]. Nine different
mechanisms are identified [163, 164] and summarised below but for more details
the reader should consult the original articles.

11.10.1.1 Mould Level Fluctuations

This mechanism is shown in Fig. 11.62 [163, 164]. It occurs when there is a sudden
drop in mould level and the mould movement, causes shearing of the slag, (in
contact with the shell tip) resulting in “necking and detachment” of the slag filament
formed [163]. It has been proposed that mould level variations associated with
mould oscillation are unlikely to trigger entrapment but larger level fluctuations,
due to transient changes in metal flow pattern, are more likely to cause of
entrapment by this mechanism [163].
Mould level fluctuations (and hence, entrainment) have been reported to increase
[163] with
(i) Increasing casting speed (Vc ") [165, 166]; however, it is also affected by
increases in the Ar flow rate [165, 167, 168].
(iii) Increasing bore of SEN (DSEN ") [167].
(iv) Increasing SEN port angle [165, 167].
(v) Decreasing SEN immersion depth (himmers#) [165].
(vi) Decreasing mould width (w#) [163, 164].
(vii) Increasing Ar flow rate (VAr ") [165, 167, 168] and high, SEN/Ar, contact
angle (h) with increasing h (i.e. decreasing wettability) leading to increased
fluctuations [168] but increased casting speed reduces the effect of the Ar
flow rate [165].
Meniscus level fluctuations can be suppressed by optimising the above factors for
the selected casting speed and mould width. It should also be noted that meniscus

Fig. 11.62 Schematic drawings showing slag entrapment caused by mould level fluctuations;
a contact of metal with slag pool during metal level descent, b mould level rise and necking of a
slag filament during ascent, c detachment of slag droplet and entrainment during ascent and
d entrapment in the shell dendrites during the subsequent descent (permission granted, ISS/AIST,
[163, 164])
11.10 Slag, Gas Entrapment and Sliver Formation 483

fluctuation behaviour is different in thin slab casters to that experienced in thick-slab

casting, [163, 164]. Meniscus level variations due to bulging can be reduced by
promoting heat transfer (qhor) in the mould to create a thicker shell. The magnitude of
surface waves can also be reduced using EMBr [13, 153] (Sect. 5.8.4).
The meniscus fluctuations exhibit, at least, two frequencies, i.e. ca. 0.2 and 2 Hz
[169–172]. The 2 Hz variations are associated with the mould oscillation, and the
0.2 Hz frequency has been variously ascribed to bulging [169] and to fluctuations at
the nozzle ports [171]. Direct observation of the meniscus showed that the 2 Hz
fluctuations have an amplitude of ca. 70% of that for oscillation but tended to
increase with increasing casting speed [170].
The number of defects tends to increase with increasing metal flow velocity
(Vmetal), consequently, equations have been derived to characterise the strength of
the metal flow up the narrow face. The “F’ value” is a measure of this flow. It is
calculated from Eq. 11.15 [165] where Qmetal = steel throughput, Vcoll = velocity
of metal at point of collision on the narrow face and hcoll and hcoll are defined in
Fig. 11.63 [163, 164].

F 0 ¼ qQmetal  Vcoll ð1  sin hcoll Þ=4=hcoll : ð11:15Þ

The F′ factor has also been linked (Eq. 11.16) to both the amplitude of the
fluctuations (Afluct) and the velocity of the metal flow along the steel surface, Vsurf,
(which is associated with several types of entrapment)


Afluct ðmmÞ ¼ 3F 0 ¼ 35 Vsurf ms1 : ð11:16Þ

Thus, using the critical F′ limits to avoid defects
(3–5 N m−1) given above, it
crit
can be shown that the critical surface velocity Vsurf corresponds to 0.2–0.4 m s−1
[166]. However, critical surface velocities for other slag entrainment mechanisms
tend to vary depending upon the mechanism involved.

Fig. 11.63 Schematic

drawing showing definitions
of terms used in calculating
the F′ factor (permission
granted, ISS/AIST, [163])
484 11 Using Mould Fluxes to Minimise Defects and Process Problems

11.10.1.2 Meniscus Freezing and Hook Formation

Hooks are formed when there is insufficient heat supplied to the meniscus region,
especially, at lower casting speeds and in regions near the narrow faces where
temperatures are at their lowest. Such conditions also pertain when there is no Ar
flow or when casting low-melting steels (e.g. ferritic stainless steels). The hooks
trap slag, inclusions (e.g. Al2O3) and gas bubbles (Fig. 11.64a–c). Furthermore, any
overflow of the frozen meniscus by the molten metal can cause slag to be trapped
against the hook.
The formation of hooks occurs when the SEN immersion depth (himmers) is too
shallow since this creates a downward, single roll, flow which promotes low ver-
tical heat transfer and a danger of meniscus freezing. It has also been reported that
when the Ar flow velocity exceeds a critical value (5 L min−1) a double roll flow
system is transformed to a single roll flow [173, 174].
Hook formation can be minimised by
(i) Reducing the number of hooks by increasing the heat supplied to the
meniscus region; this can be done by increasing (a) superheat, (DT ")
(b) redesign of SEN port angles (c) flow velocity when casting at low casting
speed (Vsurf = > 0.2 m min−1) [166] and by decreasing the depth of
immersion [163, 164].
(ii) Decreasing the length of the hooks can be achieved by decreasing the vertical
heat flux in the bed by (a) increasing the depth of the powder bed (b) using
powders containing exothermic agents [13] and (c) increasing qvert (via gas
convection) by increasing Ar gas flow (but must be <5 L min−1)
(iii) Using electromagnetic devices since (a) EMS homogenises the meniscus
temperature and minimises freezing near the cooler, narrow face and
(b) EMBr provides a 10 °C increase in meniscus temperature which reduces
hook formation (however, the reduction of the metal flow velocity reduces
heat losses (qvert) which leads to a shorter meniscus shell (hook)).

Fig. 11.64 Schematic representation of a entrapment by hooks [163], b a particle rising towards a
hook and c its subsequent capture by the hook (permission granted, ISS/AIST a [163], b, c [164])
11.10 Slag, Gas Entrapment and Sliver Formation 485

Slag entrapment by hook formation can be reduced by

(a) Reducing both the number and size of the hooks (see (i) and (ii) above).
(b) Using EMS and EMBr (as in (iii) above).
(c) Using mould slags with high viscosity (but this will reduce the powder con-
sumption) and high interfacial tension (best achieved by reducing the S content
of the steel).

11.10.1.3 Argon Bubble Interactions

Argon bubbling is used to reduce nozzle clogging and is also beneficial in reducing
the hook formation by increasing the vertical flux via gas convection. However, it
can lead to both slag and gas entrapment since slag tends to coat the gas bubbles.
Entrapment can occur by the two mechanisms shown in Figs. 11.65 and 11.66. The
mechanism shown in Fig. 11.65 involves the formation of a foam in the slag pool.
The interfacial forces favour the covering of the SEN with slag and this results in
slag crawling down the SEN carrying the foam with it (Fig. 11.65a). When it
reaches the point where the foam and the metal stream come in contact
(Fig. 11.65b) gas and slag are released into the metal stream. This mechanism tends
to be prevalent when the bubbles are small with low inertia; small bubbles are
formed when the steel throughput is high.
Slag and gas entrapment by this mechanism increases with
(a) decreasing bubble diameters which have low inertia (promoted by increased
steel throughput),
(b) increasing slab width,
(c) increasing argon injection rates,
(d) increasing slag viscosity, but this may be a consequence of high SiO2 pro-
ducing high η and low cmsl which aids crawling,

Fig. 11.65 Schematic diagrams of slag foam being carried down the SEN by slag crawling down
the SEN to a a sub-critical depth (no entrainment) and b a critical depth where entrainment occurs
(permission granted, ISS/AIST, [163, 164])
486 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.66 Schematic drawings showing a gas bubble rising, b bubble covered with steel,
pushing into slag pool, c bubble penetrating the slag and powder layers and d draining of metal
carrying slag droplets (permission granted, ISS/AIST, [164])

(e) decreasing slag density, and

(f) decreasing interfacial tension (slag/ metal).
The critical gas injection rate (i.e. above which entrapment occurs) decreases
with (i) increasing casting speed, (ii) decreasing slag layer viscosity, (iii) decreasing
nozzle port angles, and (iv) decreasing wettability between the slag and SEN
material [168].
Larger bubbles have enough inertia to penetrate both the slag pool and the
powder layer and escape into the atmosphere. A thin layer of steel covers the top
surface of the bubble (Fig. 11.66b) and when the bubble collapses and escapes, this
steel drains back carrying slag with it (Fig. 11.66d). [175–178].
The only physical property found to affect this mechanism is the interfacial
tension (cmsl) with entrapment increasing as cmsl decreases but entrapment has been
reported to increase with increasing throughput.

11.10.1.4 Slag Crawling

Interfacial tension forces favour the covering of the SEN with slag; this results in
slag crawling down the SEN. However, Hibbeler and Thomas [163, 164] point out
slag crawling is aided by the pressure differences arising from asymmetric flow.
Asymmetric flow (like that shown in Fig. 11.67a results in both a pressure increase
and meniscus elevation (shown on the right side of the SEN and a pressure decrease
and meniscus drop shown on the left). Low pressure (on the left side of the SEN in
Fig. 11.66a) promotes slag crawling down this left side. Asymmetric flows can
arise from several sources, (i) nozzle clogging [179, 180] (ii) turbulent flow (em-
anating from SEN ports) (iii) stopper rod misalignment [181] and (iv) increasing
mould width since this leads to higher throughput [163, 164]. When the slag
reaches the SEN port, it is dragged into the steel by the flow travelling down the
SEN (Fig. 11.67b).

11.10.1.5 von Karman Vortex Formation

The SEN tends to block part of the metal flow and vortices are formed in the wake
of a metal flow. These vortices are referred to as von Karman vortices and they
11.10 Slag, Gas Entrapment and Sliver Formation 487

Fig. 11.67 Schematic representation of entrapment by slag crawling a high pressure and
meniscus elevation shown here on the right side of the SEN and decreased pressure and a meniscus
drop on the left side of the SEN, b entrapment of slag droplets in the metal stream (permission
granted, ISS/AIST, [164])

occur on the low-pressure side of the SEN (on the left of the SEN in Figs. 11.68
and 11.69b) [182–184]. Karman vortices can cause slag entrapment (i) between
dendrites of the shell on the wide face near the SEN or (ii) by creating a funnel deep
enough to be captured by the metal flow leaving the SEN [182–184]. Asymmetric
flow in the mould can also cause entrapment by the latter mechanism. The vortex
formation is affected by the SEN geometry and positioning. The vortex depth
increases with increasing meniscus velocity [185, 186] and the vortex diameter
increases with increasing SEN misalignment [183, 184], The frequency of vortex
formation (fvort) increases with (i) increasing casting speed [184–187] (ii) increasing
slab width [184, 187, 188] (iii) increasing port angle [183, 184] (iv) increasing SEN
misalignment [182, 183] and (v) decreasing SEN immersion depth [184, 187, 188].
The relationship fvort ¼ f3=2F 0 g  10 was derived in water model studies with a
mean value of fvort ¼ 1=2F 0 [164, 167]. The location of the vortex is affected by the
type of flow; for a double roll system the vortex appears about 30 mm from the

Fig. 11.68 Schematic diagrams showing a flow in the mould with low and high asymmetry,
b vortex formation where vortex is unaffected by metal flow and c where vortex is sufficiently deep
to cause slag entrapment (permission granted, ISS/AIST, [164])
488 11 Using Mould Fluxes to Minimise Defects and Process Problems

SEN and moves towards the narrow face [184], whereas, for a single roll the vortex
appears near the narrow face [188].
Quantitative equations have been derived by several investigators from physical
modelling studies; these have been reviewed by Hibbeler and Thomas [163, 164].

11.10.1.6 Meniscus Standing Wave Instability

Surface waves are formed when there is flow beneath a free surface. Thus, standing
waves are formed on the steel surface. The standing wave is formed near the narrow
face for a double roll flow system. If these standing waves become unstable they
can turn over, entraining liquid slag in the process. It is customary to determine the
critical wave height (hwave = difference between crest and trough of the wave)
above which entrainment occurs. The stability criteria, shown in Eq. 11.17, [191],
is an example (where k = wavelength and is defined as distance between outside of
the SEN and the narrow face [192]).

ðhwave =kÞcrit ¼ 0:21 þ 0:14ðqsl =qm Þ2 ð11:17Þ

The critical height has been determined in various physical modelling studies in
terms of the velocity of the metal leaving the SEN and the diameter of the SEN port
(Dport) [171, 188, 192]. The various factors affecting the wave height are defined in
Fig. 11.69. The derived equations have been critically reviewed by Hibbeler and
Thomas [163, 164]. However, the wave height (hwave) was found to increase as
(i) casting speed increased (Vc ") [193–195], (ii) SEN port diameters increased
(Dport "), (iii) SEN immersion depth decreased (himmers #), (iv) SEN port angles
were aimed more upward, (v) decreasing mould widths (w #) [172, 188, 190–195]
and (vi) decreasing mould thickness (t #) [196].

Fig. 11.69 Definition of the

various casting parameters
affecting height of the
standing wave (permission
granted, ISS/AIST, [164])
11.10 Slag, Gas Entrapment and Sliver Formation 489

11.10.1.7 Shear Layer (Kelvin–Helmholtz) Instability

This movement of the molten steel at the meniscus creates an aerodynamic drag
force which causes the liquid slag to move in the same direction as the metal.
However, the velocity in the slag phase is much smaller in the slag than that in the
metal because the viscosity of the slag is much higher than that of the steel. The
velocity difference between the two liquid layers results in shearing of the slag layer
and slag filaments are created in the metal. Subsequently, necking and detachment
of the filament occurs and slag droplets are swept along in the metal stream, as
shown in Fig. 11.70. This phenomenon is known as the Kelvin–Helmholtz
instability.
Various quantitative equations have been derived for the critical velocity (Vcrit),
i.e. the velocity above which slag entrainment occurs; these equations have been
reviewed by Hibbeler and Thomas [163, 164]. An equation involving most the
major terms is given in Eq. 11.18 [163, 164, 191]. Thus, the critical velocity
increases (and slag entrapment decreases) with (i) increasing interfacial tension
(cmsl) (ii) increasing density difference (qm − qsl) and increasing slag viscosity (η).
Other equations indicate the critical velocity increases as the depth of the slag pool
increases.

Vcrit ¼ ð4g ðqm  qsl Þcmsl fgm þ gsl Þ4 =ðg2m  qsl þ g2sl  qm Þ2 g0:25 ð11:18Þ

11.10.1.8 Upward Flow Impinging on Meniscus

In double roll flow patterns, the upward movement of the metal results in filament
formation and subsequently, necking of the filaments, as shown in Fig. 11.71.
The detachment of the slag film can occur by dragging of the slag filament [197]

Fig. 11.70 Schematic

drawing showing necking and
detachment of slag filaments
by shear layer (Kelvin–
Helnhotz) instability;
(permission granted,
ISS/AIST, [163, 164])
490 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.71 Schematic drawings showing the sequence for slag entrapment by Kelvin–Helmholtz
instability (upper) for the dragging mode and (lower) for the cutting mode(permission granted,
ISS/AIST, [163, 164])

(Fig. 11.71-upper) or by cutting of the slag filament [198] (Fig. 11.71 lower). The
released slag droplets are, subsequently, trapped by the shell.
A number of quantitative equations for the critical surface velocity (Vcrit) have
been derived from physical modelling studies; these have been reviewed by
Hibbeler and Thomas [163, 164]. The critical velocity (Vcrit) decreases and
entrapment increases as (i) viscosity decreases, (ii) the interfacial tension decreases,
(iii) the density difference decreases and (iv) the depth of the slag pool increases.

11.10.1.9 Meniscus “Balding”

A rising metal flow (in a double roll system) with sufficient momentum can push
the slag layer away and can expose the metal to both the powder and to the
atmosphere (Fig. 11.72a). This mechanism is known as “meniscus balding”. The
bald patch (or eye) appears near the narrow faces and causes reoxidation of the steel
and the formation of inclusions (e.g. Al2O3 and entrained powder); this process is
more severe when the “eye” coincides with the trough of the standing wave.
Meniscus balding can also occur from excessive Argon flows which lead,
sequentially, to a single roll flow system and to bald patches close to SEN
(Fig. 11.72b).
From Sect. 11.10.1.6, it was noted that wave height (hwave) increases as
(i) casting speed increases (Vc "), (ii) SEN port diameters increases (Dport "),
(iii) SEN immersion depth decreases (himmers #), (iv) SEN port angles are aimed
more upward and (v) decreasing mould widths (w #) and thickness (t #). Slag
entrapment by meniscus balding can be minimised by creating a slag layer which is
deeper than the size of the meniscus wave.

11.10.1.10 Summary of the Factors Affecting Entrapment

The information on the factors affecting the different entrapment mechanisms is

summarised in Table 11.3. This covers the factors included in the various equations
11.10 Slag, Gas Entrapment and Sliver Formation 491

Fig. 11.72 Schematic drawings showing “meniscus balding” caused by a upward spouts in a
double roll flow system and b excessive Ar flow leading to a single roll flow system; the upper
figures show the location of the bald patches (permission granted, ISS/AIST, [164])

Table 11.3 Summary of the effect that casting variables have on the various slag entrapment
mechanisms; E = entrapment, lhook = length of hook, N = number, Qmetal = steel throughput,
VAr = Ar flow rate; DT = superheat; () = Intuitive deduction, " = increase, # = decrease
Mechanism Factors affecting entrapment
1. Mould level Fluctuations and E " as (i) Vc "(ii) DSEN "(iii) himmers# (iv) port angle
fluctuations " (v) w# (vi) VAr "a (vii) SEN /Ar contact angle "(viii) (η #)
2. Hook formation Nhooks " and E " as (i) Vc #(ii) DT #(iii) low himmers (iv) port angle "
(v) qvert #
lhook " and E " as (i) qvert bed "
3. Ar bubble (a) Foam formation and E " as (i) Dbubble # asVc " (ii) w" (iii) VAr"
interactions (iv) η" (v) cmsl # (vi) qsl # and (vii) Qmetal "(i.e., " Vc, w, t)
(a) Slag foam (b) E " as cmsl # (η #)
(b) Bubble collapse
4. Slag crawling (i) hpen < himmers and hpen "as qsl" Vsl" (ii) (η #) (iii) Asymmetric flow
"
5. Von Karman E" as (i) cmsl # (ii) (ηsl #) (iii) Asymmetric flow "
vortices
6. Standing wave hwave "and E"as "(i) Vc " (ii) Dport " (iii) himmers # (iv) w # (v) t #
instability (vi) SEN port angle more upward (vii) (η #) (cmsl #)
7. Shear layer Vcrit # and E" as (i) cmsl #(ii) ηsl # (iii) dpool # (iv) Dq#
(KH) instability
8. Upward flow on Vcrit # and E" as (i) cmsl #(ii) ηsl # (iii) dpool " (iv) qsl"
meniscus
9. Meniscus balding E " as (i) dpool #; (ii) (η #) (iii) (cmsl #)
a
Interaction between Vc and VAr; increasing Vc tends to decrease effect of VAr and increased VAr
tends to reduce effect of Vc; when VAr > 5 L min−1 double roll transforms to single roll
492 11 Using Mould Fluxes to Minimise Defects and Process Problems

derived in physical modelling studies [163, 164]. It also includes some intuitive
deductions, e.g. increased slag viscosity would be expected to resist meniscus
balding and standing wave instability but reduce slag pool convection and qvert;
these deductions are denoted by parenthesis.
The effect of increases in the various casting variables on the entrapment (de-
noted E) is given in Table 11.4. It can be seen form this table that
• Increases in the casting variables have contrasting effects on the different
mechanisms; only the slag/metal interfacial tension has a consistent effect (i.e.
decreasing entrapment with increasing cmsl).
• Increases in casting speed and slag fluidity (i.e. reciprocal viscosity, 1/ η) tend to
increase entrapment, the only exceptions being the hook and foam formation
mechanisms, respectively.
• Even the effect of qvert on entrapment is contradictory since increased heat flux
to the surface will reduce the number of hooks formed (decreasing entrapment)
but increased heat transfer through the bed will increase the length of the hooks
(increasing entrapment).
• The metal flow system formed (Fig. 11.72) is very important in slag entrapment,
especially if an asymmetric flow system is established.
• The metal flow system is dependent upon the casting speed, the SEN immersion
depth, the Argon flow rate, etc.
• The casting speed (Vc) and Ar flow rate (VAr) tend to oppose each other, so that
increasing VAr tends to cushion the metal flow (and hence, its effect on

Table 11.4 The effect of increases in the various, casting parameters on the different mechanisms
of entrapment; E" = increase in entrapment; E# = decreased entrapment; 1, 2, 3, etc. refer to
mechanisms listed in Table 11.3; () = by intuitive deduction
Ariable /mechanism 1 2 3a 3b 4 5 6 7 8 9
Casting speed (Vc ") E" E# E" (E ") (E ") E" E" E" (E ")
SEN immers depth E# (E ") E" E#
(himmers ")
Ar flow rate (VAr ") E" (E #) E"
Mould width (w") E# E" E#
Thickness (t ") E#
Superheat (DT ") E#
SEN Bore (Dbore ") E"
Slag pool depth (dpool") E# E" (E#)
Slag viscosity (η") E# (E#) E" E# E# E# E# E# E# E#
Interfacial tension (cmsl ") E# E# E# (E#) E# E# (E#)
Slag density diiff E# E" E"
(qmetal − qsl ")
Vertical heat flux (qvert ") E#
11.10 Slag, Gas Entrapment and Sliver Formation 493

entrapment) and can even cause a change from double roll system to single roll
when VAr > 5 L min−1. Similarly, increased Vc tends to oppose the effects of
VAr on entrapment.
The various flow systems are shown in Fig. 11.73 and the conditions leading to
various forms of entrapment are given in Table 11.5.
The equations derived in physical modelling studies may not always provide
quantitative values for entrapment by specific mechanisms for the actual caster,
because of differences in the similarity criteria for water and steel; this is especially
true where the interfacial tension is involved. Nevertheless, the trends summarised
in Tables 11.4 and 11.5 are valid for the casting mould. It should be remembered
that gas bubbles are coated with liquid slag and are used to remove inclusions from
the steel, so mechanisms involving gas bubbles apply to slag, gas and inclusion
entrapment.

11.10.2 Slag Entrapment

Published data on slag entrapment have been classified below according to the
mechanism involved.
It is apparent from the previous section that entrapment is mainly affected by the
metal flow velocity and the flow pattern. However, slag entrapment can be reduced
by using a mould slag with a high-interfacial tension (i.e. reducing S content of
steel) and high viscosity (but only at the expense of decreased powder
consumption).

Fig. 11.73 Schematic drawings showing various metal flow patterns a Single roll, b Double roll
and c Asymmetric meniscus roll (permission granted, UNESID, [53])
494 11 Using Mould Fluxes to Minimise Defects and Process Problems

Table 11.5 Operating conditions leading to different modes of entrainment [207]

Mode Conditions Findings and outcomes
1 Low Vc; low SEN Downward recirculation which tore slag particles away
immersion depth
2 Higher Vc; higher Recirculation flow at narrow face causes necking and
SEN immersion detachment of slag
3 Excessive argon flow When (i) Ar flow <3 L min−1; decreased entrainment
(ii) >3 L min−1; increased entrainment
4 Foam formation When foam reaches SEN port it is swept away and results
around SEN in pencil pipe defects. Lower Ar flow rate
Important at high Vc

11.10.2.1 Mould Level Fluctuations

The normal mould level variations in continuous casting are unlikely to cause slag
entrapment by the mechanism depicted in Fig. 11.62 [163]. However, periodic flow
oscillations occur continually and tend to magnify as casting speed is increased
[193]. As mentioned in Sect. 11.10.1.1, the velocity of the flow running up the
narrow wall can be characterised by the F′ factor (defined in Eq. 11.15 and
Fig. 11.63) [165]. It has been found in a large number of plant trials, that good
surface quality was obtained when the F′ factor had values between 3 and 5 N/m
(grade not stipulated) [165], or between 2 and 3 N/m for LCAK steels [166].

11.10.2.2 Hook Formation and Meniscus Freezing

Hooks and frozen skulls, (formed on the steel surface) can trap slag droplets, gas
bubbles and inclusions. They are formed when insufficient heat is supplied to the
meniscus region. These conditions arise when casting with (i) low casting speeds,
(ii) low superheat, (iii) steels with low-melting temperatures, (iv) poor thermal
insulation in the powder bed and (v) when a single roll flow system occurs. The
problem can be solved by
(i) Reducing the number of hooks formed (by improving the heat transfer to the
meniscus); this can be achieved by increasing, casting speed, superheat and Ar
flow rate (<5 L min−1). Alternatively, the establishment of a double roll flow
system or by using EMS to smooth out temperature differences across the
mould or EMBr to increase the meniscus temperature.
(j) Reducing the size of the hook (the underlying principle is explained in
Fig. 11.35 [122]), this can be achieved by improving the thermal insulation of
the powder bed (by increasing the powder bed thickness, or adding exothermic
agents to the mould powder [44]) or by using EMBr or EMC to delay solidi-
fication. Note increased slag viscosity will reduce the vertical heat transfer by
convection in the slag pool but will reduce powder consumption also.
11.10 Slag, Gas Entrapment and Sliver Formation 495

11.10.2.3 Argon Bubble Interactions

Slag coats the surface of gas bubbles and there are two mechanisms in which slag is
released and, subsequently, captured by the shell. These involve (i) slag foam
moving down the SEN (Fig. 11.65) and (ii) the penetration and collapse of larger
bubbles (Fig. 11.66). Smaller bubbles have less inertia than bigger bubbles so they
find it difficult to penetrate the slag pool so tend to create a foam in the slag pool.
Bubble diameters decrease with increased steel throughput (= qmVcwt) so foam
formation is promoted by higher casting speeds and mould dimensions. High values
of viscosity and interfacial tension will resist the escape of the bubbles and thereby,
promote foam formation.
Larger bubbles have more inertia and can penetrate the liquid slag pool. When
the bubble collapses the slag coating forms droplets which drain with the help of the
draining metal. High values of slag viscosity and interfacial tension will resist the
movement of bubbles.

11.10.2.4 Slag Crawling

The slag crawling mechanism is shown in Fig. 11.67 [164]. Slag crawling is aided
by the pressure difference induced across the SEN resulting from the interference of
the SEN to the reversing flows (travelling from the narrow faces towards the SEN)
[200]. The slag penetration depth (hpen) increases with increases in the Blockage
factor which is usually taken to be (DSEN/t) where DSEN is the diameter of the SEN
and t is the thickness of the mould. The magnitude of the slag penetration depth,
hpen, is dependent on the interfacial velocities and densities in the metal and slag
phases [200]. The densities of most steels and mould slag do not vary much. Slag
entrapment takes place when hpen > himmers and so entrapment increases for low
SEN immersion depths. It has been proposed that this mechanism is the most
important contributor to the slag entrapment [200].

11.10.2.5 von Karman Vortices

Some workers consider this mechanism to be the most potent source of slag
entrapment in continuously cast steel [189, 201]. The slag droplets held in the tube
of the vortex escape when the vortex becomes unstable. Entrapped mould slag
droplets in the metal can be seen in the specimens of hot physical modelling
studies where Karman vortices were induced (Fig. 11.74a) [189, 201]. These
Karman vortices have also been identified in the predicted flows at the slag/metal
interface using a model coupling heat and fluid flow and shell solidification
(Fig. 11.74b) [53].
Hot and cold modelling studies involving Karman vortices showed that the
amount of entrapped slag decreased with increasing slag viscosity (Fig. 11.75a) and
inceasing interfacial tension (Fig. 11.75b). An empirical relation, given in
496 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.74 a Photograph showing slag entrapped by von Karman votices during hot modelling
experiments [189] and b Formation of a Karman vortex close to the SEN predicted by a
mathematical model (permission granted, a SAIMM, [189], b ISIJ, [53])

Fig. 11.75 The mass of entrapped casting slag as a function of a viscosity in Pas and b interfacial
tension in m N m−1for 2 powders A, B (permission granted, ISIJ, [201])

Eq. 11.19 was obtained (note the units are viscosity, ηsl in Pas, and interfacial
tension, cmsl, in m N m−1) [189, 201]. Increased viscosity can be obtained by
reducing slag basicity and increased interfacial tension can be best achieved by
lowering the S and O contents of the metal but decreasing the B2O3, K2O and CaF2
contents in the slag would also help.

mentrap ¼ 1:06  107 ðgsl Þ0:255  ðcmsl Þ2:18 ð11:19Þ

11.10.2.6 Standing Wave Instability

Standing waves are formed on the steel meniscus and if they become unstable, slag
can be trapped in the overturning wave. The wave instability arises either, because
of the turnover of the wave, or by a shearing mechanism. It has been pointed out
11.10 Slag, Gas Entrapment and Sliver Formation 497

that the former mechanism is unlikely to be a major cause of slag entrapment since
the critical surface velocity (Vcrit) for this mechanism is higher than that for the
shearing mechanism [163, 164]. Surface velocities have been reported to increase in
thin slab casting. The wave height (hwave) was found to increase with increases in
(i) casting speed (Vc "), (ii) SEN port diameters increased (Dport ") and with
decreases in, (iii) SEN immersion depth (himmers #), (iv) mould widths (w #) and
(v) as SEN port angles were aimed more upward (with the jet impingement
occurring 25–50 mm lower for every 0.2 m min−1 increase in casting speed [187].

11.10.2.7 Shear Layer (Kelvin–Hemholtz) Instability

This is a natural phenomenon which occurs when two liquid layers are travelling at
different speeds (DVmsl). The reversing flow (formed near the narrow-face wall)
shears the slag to form a slag filament (depicted in Fig. 11.70). The shear force
causes necking of the slag filament. Then Kelvin–Helmholtz instabilities cause the
slag filament to break up resulting in detachment of the slag droplet. The predic-
tions of a mathematical model of heat and fluid flow (shown in Fig. 11.76 [54, 55,
202]) show cases of the necking and detachment mechanism. This mechanism is
considered to be a major source of slag entrapment.
Slag entrapment occurs when the velocity difference between slag and metal
phases exceeds the critical velocity difference (i.e. when DVmsl > DVcrit). It has
been reported that the critical velocity difference (DVcrit) increases (i.e. entrapment
decreases) with increasing interfacial tension, viscosity, density difference
(qmetal − qsl) and slag pool depth (DVcrit ", as cmsl" [203], as ηsl " [203],
(qmetal − qsl") [203] and as dpool ") [204]. Thus, slag entrapment by this mechanism
can be minimised by (i) increasing interfacial tension (by minimising the S content
of the steel or the FeO, Na2O, B2O3 contents of the slag), (ii) by using low basicity
(C/S) slags with higher viscosity and (iii) by using a deep slag pool; note,
(qmetal − qsl") does not vary much, so has little effect.

Fig. 11.76 The predictions of a mathematical model showing “necking and detachment” of slag
(light blue) in steel (dark blue) (permission granted, [202])
498 11 Using Mould Fluxes to Minimise Defects and Process Problems

Magnetic fields applied in a direction perpendicular to the flow do not affect the
interface but a magnetic field applied parallel to the flow does stabilise the interface
[205, 206]; its effect is akin to that of increasing the interfacial tension but it does
not affect DVcrit.

11.10.2.8 Upward Flow on Meniscus

With this mechanism, the metal flow impacts against the mould wall and creates an
upward flow which in turn, creates a vortex (Fig. 11.77a); the shearing force asso-
ciated with the vortex produces a slag filament (Fig. 11.77b). Subsequently, the
shearing force causes necking and detachment of the filament to occur. Since the
vortex is produced near the narrow face, the slag droplets are deposited near the edges
of the slab (Fig. 11.77c). It can be seen from Fig. 11.71 that there are two flow modes
causing necking and detachment, namely, i.e. the dragging and cutting modes.
Water-modelling studies have indicated that the velocity at the interface (Vsurf),
decreases (and entrapment decreases) when the interfacial tension, slag viscosity and
density difference (Dq = qm − qsl) all increase (i.e. Vsurf # as cmsl" as η" as Dq ")
and when the depth of the slag pool decreases (dpool #) as shown in Fig. 11.78 [198,
207]. In practice, (i) the densities of slags and metals vary little and (ii) the interfacial
tension is principally determined by the Sulphur content o the steel (cmsl"as S
(ppm) #) however, high FeO and MnO concentration in the slag also reduce inter-
facial tension and (c) slag viscosity is largely determined by the basicity (C/S, with
η" as (C/S) #). Although entrapment can be decreased by reducing the depth of the
slag pool, this would have a deleterious effect on powder consumption and
C-pick-up by the shell.
A special case of the upward flow mechanism occurs in billet casting where
downward pouring is used. For high SEN immersion depths it has been reported
that downward flows in the mould result, periodically, in the formation of vortices
which, in turn, result in periodic backflows (Fig. 11.80) [208]. These backflows

Fig. 11.77 Schematic drawings showing a the reversing flow formed in the mould, b the
sequence of events during slag necking and detachment (b) location of entrapped slag on slab
(Courtesy of RF Brooks, National Physical Laboratory)
11.10 Slag, Gas Entrapment and Sliver Formation 499

(a) 15 (b) 60
50

Velocity,cms -1
Velocity,cms -1
10 40
30

5 20
10

0 0
0 10 20 30 40 50 0 20 40 60 80 100 120
-1 l, ,dPas
msl, ,mNm

(c) 20 (d) 20

18
Velocity, cms -1

15

Velocity, cms-1
16
10
14

12 5

10 0
0 100 200 300 400 500 0 1 2 3 4 5
( m- sl ) kgm-3 dpool, mm

Fig. 11.78 Velocity at interface as a functions of a interfacial tension, b Slag viscosity,

c differential density (Dq = = qm − qsl) for slag pool depths of □ = 3 mm; ◊ = 8 mm;
o = 1.3 mm; d slag pool depth; with Dq = qm − qsl kg m−3 has the following values;
● = 1300; ▲ = 1200 ■ = 1100; ♦ = 1000 (permission granted, ISS/AIST, [198, 207])

result, sequentially, in meniscus instability, high vorticity in the flow and a rotating
flow around the billet which causes slag entrapment [208]. Thus, billets were found
to contain clusters of slag inclusions at periodic distances on the inner radius [208]
(Fig. 11.79).

11.10.2.9 Meniscus Balding

This mechanism of slag entrapment occurs when an upward flow causes the
“balding” of the mould powder and the creation of an “eye” in the powder layer
which leads to reoxidation of the steel. A double flow system results in a bald patch
being produced near the SEN and a single roll system in bald pitches forming near
the narrow face. Entrapment by this mechanism can be minimised by using
(i) a deep powder bed so dslag þ powd [ hwave ; the resulting decrease in qvert will
also help to reduce the size of the hook.
(ii) expanding mould powders reduce the tendency of spherical granules in the
powder bed to run downhill [209].
(iii) EMBr to stabilise the meniscus and reduce the height of the standing wave
[48].
500 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.79 Schematic

drawings showing periodic
formation of vortices and the
formation of a back flow
resulting in slag entrapment in
billet casting. mould = brown
blue = slag

11.10.2.10 Summary of Slag Entrapment Mechanisms

The following conclusions can be drawn from the review of slag entrapment
mechanisms:
(i) Slag entrapment occurs by a number of different mechanisms.
(ii) The most potent mechanisms resulting in slag entrapment appear to be
(a) slag crawling (aided by pressure differences across the SEN), (b) Karman
vortices, (c) Kelvin–Helmholtz instabilities and (d) upward flows.
(iii) Most of the casting variables have contradictory effects on the various
mechanisms but, overall, the slag entrapment can be reduced by (a) reducing
the casting speed, (b) increasing interfacial tension and (c) increasing slag
viscosity.
(iv) Slag entrapment can be reduced by using EMBr, EMS and EMC
(v) Slag entrapment can also be reduced by improving the thermal insulation of
the powder bed (i.e. decreases qvert) since this will have beneficial effect on
slag entrapment associated with hook formation and meniscus balding and
will help to reduce the length of the shell; however, it will also increase dpool
which tends to increase entrainment via the upward flow mechanism.

11.10.3 Gas Entrapment

Argon gas is fed through the SEN to minimise nozzle clogging, but it also helps to
reduce the formation of hooks by increasing the vertical heat flux by increasing gas
11.10 Slag, Gas Entrapment and Sliver Formation 501

convection. Most of the Argon finishes up in the slag pool and powder layer but
some of bubbles get incorporated into the metal flow [210]. Argon bubbling also
aids the removal of inclusions from the metal by transporting them to the molten
slag pool. However, Argon bubbling is the principal source of gas entrapment in the
steel. Entrapped gas bubbles cause two types of defects in the cast product
(i) “pinholes” and “blow holes” consisting of trapped gas bubbles like that shown in
Fig. 11.61 and (ii) the defect known as “pencil pipe” (sometimes referred to as
“pencil blister”) which arises when gas bubbles carrying inclusions are captured
(Fig. 11.81b). Pinholes consist of small argon bubbles. Blowholes usually contain
CO (g) [46]; they are formed by dissolved oxygen content (O) in the liquid steel
gradually increasing as solidification proceeds and CO (g) bubbles are formed when
the O concentration exceeds a certain, critical value. Atmospheric O2 and N2 can be
drawn into the steel via an overturning wave in the Standing Wave instability and
Meniscus balding mechanisms, and during periods of high mould level variations
[211]. Pencil pipe defects are typically 1–2 mm wide and 12–300 mm long [212]
and exhibit a raised surface (“pencil”) after annealing (Fig. 11.80b).
Gas entrapment occurs principally through the formation of gas foams
(Fig. 11.65), but bubbles can be swept into the metal flow where, subsequently,
they are captured by the shell (Fig. 11.80a). Larger bubbles have more inertia than
smaller bubbles and tend to emerge into the slag pool near the SEN, whereas,
smaller bubbles emerge nearer the narrow-face walls. Consequently, there is a much
greater risk of small bubbles being swept into the metal stream. Increased casting
speeds have been reported to increase the number of small bubbles. Furthermore,
there is a much greater risk of bubble entrainment with a single roll flow system
since the double roll system gives the gas bubbles time to float to the slag pool
whereas, the single roll system sweeps the bubbles straight down the mould
(Fig. 11.80a). High Ar flow rates can cause a double roll system to transform ro a
single roll system [173, 174]. The models used for determining bubble movement
in the mould have been reviewed [210].

Fig. 11.80 Schematic drawings showing a how bubbles are trapped in steel [212] and b the
formation of “Pencil pipe” defects [198] (permission granted, ISS/AIST, [198, 212])
502 11 Using Mould Fluxes to Minimise Defects and Process Problems

11.10.3.1 Pinholes

The following factors affect the incidence of pinholes.

Argon Bubbling

The incidence of pinholes tends to increase with increasing Argon flow rate.
However, excessive flow rates (>5 L min−1) cause double roll flow system to
transform to a single roll system [172–174, 207]. The aim is to keep the bubbles
near the slag pool where they can be absorbed but single roll flows tend to move the
bubble down the mould and away from the slag pool and should be avoided.

Reducing the Length of the Meniscus and Delayed Solidification

It has been proposed that bubble entrapment can be reduced by reducing the size of
the solidified shell [122]. The principle of the proposed treatment is shown in
Fig. 11.35 [122]. A shorter meniscus or shell can be obtained by (i) reducing the
vertical heat flux or (ii) increasing the superheat (Fig. 11.81a). The vertical heat flux
(qvert) can be reduced by
• Increasing the depth of powder bed and decreasing the granule size.
• Using exothermic agents which reduce the thermal gradient in the bed
(Fig. 11.81a).
• Decreasing Ar flow rates.
Delayed solidification, in which the shell solidifies further down the mould,
results in a shorter solidified meniscus but also removes the shell away from the
turbulent interface. Delayed solidification can be achieved with (i) meniscus-free
casting [21, 22] (ii) Electromagnetic casting (EMC) [51] and (iii) a “soft cooling”
mould powder with high crystallinity and a high-melting temperature which will
create an insulating slag film in the meniscus region [38].

Use of Electromagnetic Devices

The pinch force generated in Pulsative Electromagnetic casting (EMC, for more
details see Sect. 5.8.3) results, sequentially, in a thicker slag film, reduced hori-
zontal heat flux and delayed shell solidification. This has been reported to lead to a
5–10-fold decrease in defects (inclusions and pinholes) [50, 51] and also provides
the extra benefit of producing shallow oscillation marks.
Electromagnetic braking (EMBr or Flow control, for more details see Sect. 5.8.4)
reduces (i) the velocity of the metal flow flowing from the SEN and (ii) the pene-
tration depth of the metal flow into the mould which results in an increase in
meniscus temperature [47, 48]. In addition, it has been reported to cause a 30%
reduction in qvert [124] leading to the formation of a short shell. It can be seen from
11.10 Slag, Gas Entrapment and Sliver Formation 503

Fig. 11.81b that the use of Flow Control (EMBr) leads to significant reduction of
pinholes [47, 48].
Electromagnetic stirring (EMS, for more details see Sect. 5.8.2) of liquid metal
causes Saffman forces to be exerted on the gas bubbles and inclusions, the

Fig. 11.81 Schematic (a) 6

drawings showing a effect of
superheat and exothermic

Pinhole index
agents in the mould powder
4
on the pinhole index [44];
conventional mould powder
(─ ● ─ ─) and containing
exothermic agents (─O─), 2
b effect of Flow control
(EMBr) on the pinhole index
[47] and c Critical dissolved 0
O concentration in steel 10 20 30 40
versus C content of steel Superheat, o C
showing increased stirring
velocity (- - = no EMS;
▬ = 0.4 ms−1;
(b) 1
dash-dot = 0.8 ms−1) in EMS 0.8
reduces CO blowhole
Pinhole Index

formation (●, ■ = CO 0.6

blowholes, o, □ = no CO
blowholes) [46] (permission 0.4
granted, a re-drawn [44],
b EPD Sci., re-drawn [47], 0.2
c NSSM Corp.)
0
0 400 800 1200
Current, A
(c)
504 11 Using Mould Fluxes to Minimise Defects and Process Problems

magnitude of the force increasing with increasing bubble/inclusion size [50, 213].
Bubble/inclusion separation occurs when the Saffman velocity exceeds the velocity
of the solidification front. EMS is successful in removing bubbles and inclusions
with particle diameters, >100 lm) [50]. Saffman forces easily remove bubbles with
Dbubble > 1 mm and will even remove bubbles with Dbubble < 0.4 mm with steels
with lower S contents (higher interfacial tension) [213]. EMS also reduces the level
of CO blowholes formed when casting Al-killed steels [46]. The metal flow created
by EMS circulates the liquid metal at the solidification front which reduces the
build up of the O concentration in the liquid steel in this region, thereby, sup-
pressing CO blowhole formation (Fig. 11.81c).

Metal Flow Velocity and Bubble Size

Hanao et al. [214] found that the index of pinholes decreased as the molten steel
flow velocity increased and tended to be slightly higher on the inner radius side than
on the outer radius side (Fig. 11.82a). This is a surprising result since increased
casting speeds tend to provide smaller bubbles. Hanao et al. [214] calculated Dc
(given in Eq. 11.20) and found that it became increasingly negative as the casting
speed increased. Negative Dc values indicate that the bubble trapped by the shell is
unstable. This can be viewed as higher velocities liberate trapped gas bubbles. Their
calculations also indicated that larger bubbles trapped by the shell were more
unstable than smaller bubbles, i.e. bubble entrapment increases with decreasing
bubble diameter.

Surface and Interface Tension

Hanao et al. [214] examined the various tensions acting on a gas bubble trapped
against the solid shell and reported Eq. 11.20, where Dc = stability index of bubble

Fig. 11.82 Schematic diagrams showing a Index of pinholes as a function of metal flow velocity,
b the tensions acting on a bubble trapped by the shell and c the pinhole index as a function of
stability index, Dc (permission granted, ISS/AIST [214])
11.10 Slag, Gas Entrapment and Sliver Formation 505

on the solid interface; cLS = interfacial tension between solid and liquid (steel),
cL = surface tension of liquid (steel); cS = surface tension of solid (steel); a, b
angles (°) made by tensions; s = shear stress exerted by metal flow; d = thickness
of trapped part of bubble and Dbub = diameter of bubble and A/D = area of circular
boundary line (Fig. 11.82b)

Dc ¼ cLS þ cL cos a  cS cosb  sðA=Dbub Þ: ð11:20Þ

It can be seen from Fig. 11.82c that that the index of pinholes increases as the
stability index, Dc, increases [214].

Summary of Ways to Reduce Pinholes

The incidence of pinholes can be reduced by

• Optimising the casting speed and Ar flow to ensure that a double roll, flow
system is established.
• Decreasing the vertical heat flux by (i) increasing depth of powder bed
(ii) incorporating exothermic agents into powder bed (iii) increasing metal
superheat.
• Delaying solidification by using (a) Meniscus-free casting (b) EMC or (c) using
a high basicity mould slag (with high values of fcrys and Tbr) to reduce the
horizontal heat flux.
• Using EMBr to increase the meniscus temperature and decrease the penetration
depth of the metal flow or, EMS to remove pinholes and EMC to delay
solidification.
• Increasing the casting speed which renders bubbles, captured by the shell, more
unstable.
• Reducing the S content of the steel decreases the number of gas bubbles in the
steel [213].
• Increasing the vertical section of the caster by 2.5 m which allows more time for
pinholes to float out [215].
• Minimising mould level variations [211].

11.10.3.2 Pencil Pipe

The source of pencil pipe defects are small gas bubbles or gas foam swept up in the
metal flow (Fig. 11.80a); a typical gas bubble is shown in Fig. 11.61. Gas bubbles
(ca. 0.5 mm diam.) get captured at the solidification front, particularly on the inner
radius (loose side) (Fig. 11.80a). The bubbles elongate on rolling where the
inclusions (attached to the bubble) prevent them from being welded shut during
rolling [212]. These defects are usually prevalent in steels with low yield strength
(i.e. Ti- or Nb-stabilised LC and ULC grades) [212]; this allows the bubble to
506 11 Using Mould Fluxes to Minimise Defects and Process Problems

expand during annealing, especially if carried out at high temperature [198]. Pencil
pipe defects tend to concentrated in a band around the 1/8th point and are prevalent
in wide slabs (>1.8 m).
The effect of casting variables (e.g. casting speed, mould width, biased flow,
metal level variations, etc.) on the incidence of pencil pipe defects, revealed that
there was a strong correlation between the defects and steel throughput, the defects
increasing with increasing throughput [212]. All other factors had little, if any,
effect on the incidence of pencil pipe defects. Thus, the best way to reduce pencil
pipe defects is to reduce the steel throughput [212] but this was found to result in a
slightly higher incidence of slivers [212].

11.10.4 Inclusion Capture, Sliver Formation

There are two principal types of inclusions, namely, (i) those arising from mould
slag entrapment and (ii) those (like Al2O3 or TiN) which originate from steelmaking
reactions or from slag/metal reactions (e.g. Eq. 11.21); these occur in the tundish
and the casting mould.

3 SiO2 sl þ 4Al ¼ 2 Al2 O3 sl þ 3Si ð11:21Þ

Non-metallic inclusions, such as Al2O3 or TiN, reduce the mechanical strength

of steels [216]. Furthermore, the mechanical strength decreases with increasing
inclusion size [216]. Consequently, considerable efforts are made to reduce the
number of inclusions in the steel. Inclusions tend to agglomerate through turbulent
collisions and these agglomerated particles can restrict the flow of molten slag into
the channel between shell and the solid slag film (adhering to the mould)
(Fig. 11.83a [76, 77]). Large inclusions have a higher probability of forming
clusters by collisions [217]. Agglomeration is encouraged by the fact that the
solubilities of TiN and ZrO2 in mould slag are both low. It can be seen from
Fig. 11.83a that these agglomerates contain both solidified d-ferrite and some gas
bubbles; these give rise to slivers and scums in the product. Slivers are surface
laminations and in thin strip products, the slivers consist of a tiny metal film which
has been, either completely, or partially, torn away from the strip surface [218].
Skin laminations can be seen in Fig. 11.83b; inclusions in laminations were iden-
tified as TiN, mould powder and reaction products of mould slag and solute ele-
ments in steel [78]. Slivers originate from several sources, (inclusions) but also in
association with bubbles, surface cracks and internal cracks [213].
Cracks are not an issue in ULC steels because of the good ductility at high
temperatures [213]. Slivers associated with FeO have been reported to occur
(i) from FeO carried over from the ladle [211] and (ii) from scale formed on the
surface of the steel [219].
11.10 Slag, Gas Entrapment and Sliver Formation 507

Fig. 11.83 Schematic diagrams showing a the formation of agglomerates (of TiN in Ti-stabilised
stainless steel) [76–78] and b photograph of a skin lamination on the surface of Ti-stabilised
stainless steel [78] (permission granted, ISS/AIST, [78])

11.10.4.1 Reducing Inclusion Levels in Steel

Great effort is made to reduce the number and size of inclusions in the steel. This is
usually achieved by transporting the inclusions to the liquid slag phase (of the ladle,
tundish and casting mould) where they can dissolve. This removal to the slag phase
is achieved by the use of
(i) Flotation of the inclusions which is related to the density difference
(qsteel − qincl); see Sect. 11.10.3.2 below for further details.
(ii) Attachment to gas bubbles [210]; see Sect. 11.10.3.2.
(iii) Saffman forces (generated in EMS) which is effective in removing inclusions
sized >100 lm [50, 215].
In addition, inclusion levels are reduced by the following:
(i) Reducing the size of the solidified meniscus shell (or hook) as shown in
Fig. 11.35 [122] and it can be seen from Fig. 11.85 a that the inclusion
levels decrease with reducing hook size [48]; this is achieved by (a) in-
creasing superheat or meniscus temperature (b) by reducing the vertical heat
flux by creating a deeper powder bed and using a powder containing
exothermic agents and a small granule size.
(ii) Maximising the time for flotation by (a) adjusting the SEN immersion depth,
port angle and Ar flow rate to ensure that a double roll flow system
(Fig. 11.73) is established which keeps the inclusion close to the slag layer,
in contrast to single and asymmetric flows which tend to carry inclusions
away from the slag pool and (b) increase the vertical section of the caster
(by 2.5 m) to aid flotation of inclusions [215].
508 11 Using Mould Fluxes to Minimise Defects and Process Problems

(iii) Delaying solidification to a position further down the mould which can be
achieved using meniscus-free casting, EMC or by forming a slag film with
high crystallinity and Tbr.
(iv) Increasing the slag/metal interfacial tension has been reported to reduce the
number of inclusions in the steel [10, 220, 221]; the number of slivers,
associated with gas bubbles, was found to decrease with decreasing S
content of the steel (i.e. increasing cmsl) [213].
(v) Increasing slag viscosity was found to reduce the number of mould-
slag-related inclusions by ca. 20% [126].
(vi) Casting speeds were found to have little effect on the number of inclusions
but abrupt changes in casting speed were found to increase the number of
inclusions [222].
(vii) The amount of inclusions trapped increases as (i) the size of the inclusion
increases (Dincl ") and (ii) the size of the bubble decreases (Dbubble #) [221].
(viii) Using mould slags with high basicity since this reduces the concentration of
more reducible oxides (e.g. SiO2) which can react with Al and Ti in the steel
[223, 224].

11.10.4.2 Flotation of Inclusions

The velocity of an inclusion (VI) rising in liquid steel is given by Stokes’ Law
(Eq. 11.22) where g is the gravitational constant, r is the radius of the inclusion (in
m), ηM is the viscosity of the metal (in Pas) and the subscripts I, M, and G represent
the inclusion, metal and gas, respectively.

VI ¼ 2 ðqI  qM Þ  g  rI2 =9gM ð11:22Þ

Thus, large particles will float quicker than small particles and agglomeration of
inclusions will assist flotation.

11.10.4.3 Flotation by Attachment to Gas Bubbles

Consider a gas bubble approaching an inclusion in the liquid metal. For an inclu-
sion to be taken up by the bubble, new interfaces must be formed. Work must be
done to account for the change to the interfacial tension. This involves the eradi-
cation of the inclusion/metal (IM) interface and the creation of two new interfaces
[i.e. metal/ gas (MG) and inclusion/gas (IG)].
The interfacial relationships are given by Young’s Equation (Eq. 11.23) which
was derived from a balance of forces (Fig. 11.84) where h is the contact angle
between inclusion and gas.
0 ¼ cIM þ cMG  coshcIG : ð11:23Þ
11.10 Slag, Gas Entrapment and Sliver Formation 509

Fig. 11.84 Schematic drawings showing a, b bubble contacting inclusion for a wetting and
b non-wetting conditions and c, d sessile drop experiments for, c wetting and d non-wetting
conditions (permission granted, Verlag Stahleisen [225])

The work of adhesion (WA) represents the work done and is defined in
Eq. 11.24.

WA ¼ cMG þ cIG cIM ¼ cMG ð1 þ cos hÞ ð11:24Þ

For good flotation, it is necessary that the flotation coefficient (D, defined in
Eq. 11.25) should be both positive and have a high value.

D ¼ cMG þ cIM cIG ¼ cMG ð1  coshÞ ð11:25Þ

Thus, flotation is promoted by high values of both cMG (i.e. low S content of
steel) and the contact angle, h. At high temperatures, poor wettability (h > 90°) is
frequently associated with low reactivity; thus, a high contact angle is obtained
when there is little reactivity between inclusion and metal.
Once the gas bubble has reached the metal/slag interface it is necessary for it to
travel through the interface and on into the slag phase. The spreading coefficient
(S*) is the measure of the ability of the liquid (metal or slag) to spread across the
solid; it is defined by Eq. 11.26.

S ¼ cIG  cMG cIG ¼ cMG ðcos h  1Þ ð11:26Þ

The spreading increases as S* becomes more positive and is favoured by low

values of (cMG) and h, i.e. when the S content of the metal is high and there is
significant reactivity between inclusion and metal.
It is also important that the inclusion should not be entrained and it has been
shown [226] and this is favoured when DG in Eq. 11.27 is negative. These condi-
tions are favoured by (i) high values of the inclusion surface tension (cIG) and metal
surface tension, (cMG) (i.e. low S content) and (ii) low interfacial tension, cMI, (i.e.
no adsorption layer or chemical reaction at the interface). Emergence of the inclusion
is favourable when cMG > cIM, this is usually the case for oxide inclusions in steels

DG ¼ cIG þ cMG cIM ð11:27Þ

510 11 Using Mould Fluxes to Minimise Defects and Process Problems

11.10.4.4 Electromagnetic Devices Used to Reduce Inclusion Levels

EMS, EMBr and EMC are all effective in reducing the inclusion levels in steel.
However, each method achieves inclusion reduction in its own individual way.
The Saffman forces induced on the inclusion (by EMS) suppress its entrapment;
if the velocity of the Saffmann force exceeds the velocity of the solidification front
then the inclusion will be washed away. Saffmann forces increase with increasing
inclusion size, so EMS is very effective in removing inclusions >100 lm [50]. The
application of EMS has been reported to give two–threefold [215, 227] reductions
in alumina clusters (Fig. 11.85b). It can be seen from Fig. 11.86 that for a solidi-
fication velocity of 1.2 mm s−1, inclusions of >100 lm will be removed when the
steel velocity exceeds 0.3 m s−1 [50]. Thus, EMS is an effective method for
removing larger inclusions.
Flow control (i.e. EMBr) causes (i) a 5–10 °C increase in meniscus temperature,
(ii) a large decrease in vertical heat transfer and (iii) the suppression of turbulence at
the slag/metal interface. All of these lead to a shorter meniscus shell and

Fig. 11.85 a Trap (a) 0.8

ratio = (number of inclusions
Accumulated trap ra o

trapped under hook/total

0.6
number of inclusions) as a
function of depth of meniscus
hook [48], b The ratio (NEM/ 0.4
N0−EM), i.e. the ratio of
(inclusions where EM devices
0.2
used divided by inclusions
where no EM was used) from
the left, EMS ( ), EMBr 0
( ) and EMC ( ); 0 0.5 1 1.5
a re-drawn after [48] d hook , mm

(b)
0.6
N EM / N0-EM

0.4

0.2

0
EMS, EMBr 1; EMBr 2; EMBr 3;EMC
11.10 Slag, Gas Entrapment and Sliver Formation 511

Fig. 11.86 Maximum

diameter of inclusion cluster
as a function of metal flow
velocity for three different
solidification rates (with units
of 10−3 m s−1) solid line 1.2;
dotted line 2.4; dash-dot line
0.6 [50] (permission granted,
ISIJ, [50])

subsequently, to lower levels of inclusion entrapment. The use of EMBr has been
reported to bring about significant reductions in inclusion levels, (e.g. reductions of
fivefold [47], 2–3 fold [48] and twofold [13]) shown in Fig. 11.85b.
In EMC, the pinch force generated results in a thicker slag film with a conse-
quent decrease in horizontal heat flux (qhor). This, in turn, results in a short
meniscus shell which is displaced a distance from the turbulent, interface area. The
use of EMC has been reported to bring fivefold [50, 51] reduction in the inclusion
levels (Fig. 11.85b).

11.11 Formation of Scales

Scales are formed by the oxidation of the steel surface [228–239]. Scaling tends to
occur below the secondary cooling zone and results in the following:
• Yield loss.
• Reduction of equipment lifetime.
• Entrapped scales or scums which form slivers on rolling.
Scales can be formed at various stages of the process and scale formation is
usually classified as follows:
• Primary scale formation occurs in the secondary cooling zone and scales (up to
2 mm thick) are usually removed [236].
• Secondary scale formation occurs during the re-heating and rolling stages and
the scale has a thickness of 100 lm [237].
• Tertiary scale formation occurs, immediately before, and during, the final strip
rolling process and can lead to wear in the rolls.
512 11 Using Mould Fluxes to Minimise Defects and Process Problems

11.11.1 Factors Affecting Scale Formation

Si-containing steels are difficult to descale because low-melting, fayalite is formed

on the steel surface and, subsequently, penetrates into the grain boundaries
The principal factors affecting scale formation are as follows:
• Surface temperature of the strand.
• Partial pressure of oxygen (pO2) in the atmosphere (containing H2O, air, furnace
gas).
• The copper content of the steel, when this is >0.2%, it exceeds the Cu solubility
level in austenite and is preferentially oxidised to form a low-melting, liquid
phases which penetrate into the grain boundaries and which lead to cracks [149].
• Nickel additions are usually made to minimise the effect of copper since Ni
increases the solubility of Cu in austenite and favours Cu-occlusion in the scale
[149],
• Other elements (Sn and Sb), in addition to Cu, promote the formation of the
oxidised, low-melting liquid; the Cu equivalent is used as a measure of the
propensity to form liquid (Cueq ¼ %Cu þ nð%Sn þ %SbÞ  %Ni), where n has
a value between 6 and 8). [149].
• The carbon content of the steel (scaling decreases as %C in the steel increases).
• Certain elements (e.g. Al and Cr) form dense oxides which lower the diffusion
rates.
There is anecdotal evidence that certain mould slags tend to produce more scale
than other slags. This is thought to be due to the fact that certain slag constituents
(e.g. Na2O) reduce the surface tension and cause the slag to wet the steel and hence,
adhere strongly to the steel. Wetting is defined by the contact angle (h) which is the
angle between cGM and cIM (in Figs. 11.84c, d) and a liquid is considered to wet the
solid when h < 90° (Fig. 11.84c) and is non-wetting when h > 90° (Fig. 11.84d).
Nakato et al. [219] calculated the adhesion energy (cad) from the contact angle
(h), with cad increasing as h decreased (i.e. as wetting of steel (by slag) improves);
experiments were carried out with two mould fluxes, A and B, (Table 11.6) with
three different steels (Table 11.7). It can be seen from Fig. 11.87a, b that:
• cA
ad [ cad and cad decreases as the S content of the steel decreases (cm ").
B

• Mould slag A (with high Na2O%) wets the steel more than mould slag B (with
low Na2O%), i.e. hA < hB, which suggests h decreases with increasing Na2O%.
An analysis of the number and size of sliver defects in cold-rolled sheets showed
that the number of macro inclusions was greater for slag A than for slag B
(Fig. 11.88) (Table 11.8).
The constituents with lowest surface tension in mould slag are
B2O3 < K2O < Na2O < Li2O and these oxides would be expected to promote
wetting. The effect of Na2O on the interfacial tension (cmsl) can be clearly seen in
Fig. 11.89.
11.11 Formation of Scales 513

Table 11.6 Chemical compositions for mould powders A and B [219]

SiO2 CaO Al2O3 Na2O C η1300 dPas
A 30 30 7.5 15.2 2.9 1.1
B 25.4 31.1 5.4 0.7 5.0 1.5

Table 11.7 Chemical compositions (mass%)for steels [219]

Steel C Mn P S Al Ti
LCAK 0.041 0.23 0.018 0.017 0.024 –
ULCAK1 0.0019 0.21 0.014 0.013 0.045 0.032
ULCAK2 0.0023 0.20 0.015 0.010 0.065 0.063

Fig. 11.87 Values of the (a) 900

a adhesion energy and b the
Adhesion energy,mNm-1

contact angle for two mould 850

fluxes used to cast LCAK and
two ULC steels; powder A 800
(=●) and B( ) showing the
effect of Na2O% (15.2 and 750
0.7%, respectively) in slag
and S content of steel on on 700
these properties (Re-drawn
from [219]) 650
0.5 1 1.5 2
-2
Ssteel content , 10 %

(b)
60
Contact angle, θ o

40

20

0
0.5 1 1.5 2
-2
Ssteel content , 10 %

11.11.2 Causes, Mechanisms

Oxidation of the steel surface by oxygen and water vapour in the atmosphere causes
the formation of a surface coating of iron oxides. In Si-steels, the diffusion of Fe2+
in the Fe1−x O phase (Fig. 11.90), results in the formation of Fe2SiO4 (fayalite) with
514 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.88 Diagram showing 12

effect of mould powders A
and B on the numbers and 10
sizes of macro inclusions in

Frequency
8
steel when casting with mould
slags A ( ) and B ( ); 6
Re-drawn from [219]
4

0
1 2 3 4 5 6 7 8
Number of macro-inclusions, m-2

Table 11.8 Index of sliver With conditioning Without conditioning

defects in steels cast with
slags A and B [219] A 40 100
B 0.08 0.09

Fig. 11.89 The effect of 1600

Interfacial tension,,mNm-1

additions of Al2O3 (●) and

Na2O ( ) on the interfacial 1400
tension (cmsl) between molten 1200
steel and mould slag
(permission granted, 1000
ISS/AIST, re-drawn from
[198, 207]) 800

600

400
0 5 10 15 20
Al2O3 or Na2O , addi on , %

Fig. 11.90 Schematic Fe2O3

diagram showing the location
of various iron oxides and the
Fe3O4
penetration of fayalite
(Fe2SiO4 = purple) down the
grain boundaries FeO

Fe2SiO4

steel
11.11 Formation of Scales 515

a low-melting point, 1177 °C. Molten fayalite penetrates into the steel via the grain
boundaries and, on cooling, provides “handholds” for the scale to adhere to the
steel, making it difficult to remove. Mould fluxes, adhering to the steel surface, are
another source of SiO2 and promote the formation of Fe2SiO4.

11.11.3 Ways of Dealing with Scaling

1. Use an inert atmosphere.

2. Ensure that the copper and tin contents of steel are low or treat with same mass
of Ni to remove the copper from the grain boundaries.
3. Use mould slags with lower K2O and Na2O contents to reduce adhesion to the
steel surface.

11.12 Carbon Pick-up

11.12.1 Factors Affecting Carbon Pick-up

Carbon pick-up by the steel is a serious problem, especially when casting ULC steel
grades [126, 240–243]. There are several potential sources of carbon, namely:
• The bed of powder in the mould and core samples taken from the mould
frequently exhibit a carbon-rich layer at the top of the liquid pool which is due to
unreacted carbon particles floating at the top of the liquid pool.
• The slag rim which is a mixture of unreacted mould powder and casting slag
which results in the formation of “amorphous graphite” in the slag rim [243].
• Any broken pieces of Carbon-oxide refractory which is swept into the slag pool.
Carbon pick-up has been found to increase as
• The carbon content of the powder increases (as shown in Fig. 11.91a)
[126, 241, 242].
• Mould level variations increase [162].
• The depth of the slag pool decreases (Fig. 11.91b) [126, 219]; these results
could also explain Fig. 11.91a because the slag depth is connected to the
melting rate, which, in turn, is inversely related to the carbon content in the
mould powder (C% " ! QMR # ! dpool #).
• The particle size of the carbon particles present, since smaller particles (with a
high surface area to mass) tend to combust quicker than larger particles; this
results in lower carbon concentrations in the lower bed and the slag pool [244].
516 11 Using Mould Fluxes to Minimise Defects and Process Problems

(a) 12 (b) 12
10 10
C pick- up, ppm

C pick- up, ppm

8 8

6 6

4 4

2 2

0 0
0 1 2 3 4 0 4 8 12 16 20
Ctot, mass % Mean dpool, mm

Fig. 11.91 Carbon pick-up by the steel as functions of a the initial carbon content of the casting
powder and b the depth of the slag pool (permission granted, ISS/AIST, re-drawn from [126])

Note: smaller carbon particles also slow the melting rate and thus, less carbon is
required to obtain the required melting rate.

11.12.2 Causes, Mechanisms

Two mechanisms have been proposed for the carbon pick-up:

• The Painting mechanism in which the slag rim is considered to be the major
source of carbon and where the carbon in the rim “paints” carbon onto the shell
as the rim moves up and down as shown in Fig. 11.92a, b [242, 245].
• In the alternative mechanism, the metal dissolves carbon because the flow
creates a standing wave which allows contact with either the powder bed or the
carbon-rich layer of the slag pool and this carbon dissolves in the metal
(Fig. 11.92c) [242, 246]. Thus, carbon pick-up by this mechanism would be
expected to increase with increasing casting speed since this would lead to taller,
standing waves.
High casting speeds lead to the formation of both standing waves and Karmann
vortices (Figs. 11.69 and 11.68, respectively). It is probable that these vortices
could lead to carbon pick-up by a similar mechanism to that responsible for slag
entrapment.
It is apparent that higher carbon contents in the powder will tend to increase
(i) the amount of carbon in the slag rim (Painting mechanism), (ii) the amount of
carbon in the lower section of the powder bed and (iii) the amount of carbon
11.12 Carbon Pick-up 517

Fig. 11.92 Mechanisms

proposed for the
recarburisation of the steel in
the continuous casting mould
[242] a, b the painting
mechanism [242], c the
alternative mechanism arising
from the standing wave [242]
(Permission granted, EPD Sci.
[242])

floating in the slag pool (alternative mechanism). It has been reported that the slag
rim is built up by the painting mechanism [247].
The slag pool depth affects both mechanisms. A deep slag pool reduces the area
of contact between carbon and the shell in both mechanisms.

11.12.3 Ways of Dealing with Carbon Pick-up

11.12.3.1 Reducing the Carbon Concentration in the Lower Bed

and the Slag Pool

This can be achieved by

• decreasing the overall, carbon content of the casting powder; this should be
carried out by reducing the carbon content whilst maintaining the melting rate at
the same level (by using smaller carbon particle sizes, e.g. carbon black).
However, it should be noted that small carbon particles also combust more
rapidly.
• by using fluxes containing MnO to promote the oxidation of carbon particles in
the lower mould (but elements such as Al and Ti in steel will react with the MnO
to give more oxide inclusions).
518 11 Using Mould Fluxes to Minimise Defects and Process Problems

11.12.3.2 Increase the Depth of the Slag Pool

By using the methods summarised in Sect. 4.4, namely:

Increase the Melting Rate of the Powder

This can be achieved by (i) reducing the carbon content of the powder, (ii) in-
creasing the particle size of carbon particles in the powder and (iii) increasing the
vertical heat transfer (note introducing EMBr would tend to reduce both the metal
flow turbulence and the vertical heat transfer).

Increase the Thermal Insulation of the Powder Bed

This can be achieved by (i) increasing the depth of the powder bed, (ii) by using
casting powders with smaller granule size or (iii) by introducing exothermic agents
into the casting powder.

11.12.3.3 Reduce the Stroke Length and Improve Mould Level

Control

This seeks to minimise the area of contact between the carbon-rich zones of the
powder bed and slag pool and the shell (Painting mechanism).

11.12.3.4 Reduce Metal Flow Rate to Decrease the Height

of the Standing Wave

The height of the standing wave increases with increasing metal flow rate (i.e.
increasing turbulence).Where high production (or casting) rates are needed the use
of EMBr is usually introduced to minimise the effects of metal flow turbulence.

11.12.3.5 Replace Carbon in Powders with Carbides or Nitrides

The carbon in the mould powder is there to control the melting rate. This task can
be carried out by other materials providing they are non-wetting to molten casting
powder and combust with air. Casting powders, where the carbon has been replaced
with boron nitride, silicon nitride or silicon carbide, have been produced com-
mercially [242, 243, 248]. Most of these powders have proved sensitive to sintering
and resulted in the formation of large rims and consequently, have not been widely
used.
11.12 Carbon Pick-up 519

It should be noted that some of the factors above appear to be acting in a

contradictory manner, for instance, decreasing the carbon particle size will reduce,
sequentially, the melting rate, the pool depth and increase the carbon pick-up.
However, decreasing particle size will also lead to faster combustion and produce
lower carbon contents in the bed and pool and thus, reduced carbon pick-up. It
would appear that the latter effect is dominant [244]. In the same way, EMBr leads
to reduced pool depth and greater carbon pick-up but would result in a lower
standing wave and hence a lower carbon pick-up.

11.13 SEN Erosion

Most SENs are fabricated from oxide/graphite refractories (such as Al2O3/C,

ZrO2/C, etc.) with inserts of ZrO2 or ZrO2/C in the region subjected to extensive
wear which is usually referred to as the Z-band. The principal area of erosion lies
between the region of the SEN in contact with the slag pool and to the impact zone
of the circulating metal flow. The eroded region consists of a deep groove in the
SEN. It is usual practice to alter the immersion depth of the SEN during casting to
even out the wear in the SEN. Wear can also occur in both the interior of the SEN
[49] and in the Z–band.
Erosion rate studies are usually carried out in simulation experiments in which
the refractory sample is either immersed or rotated in the liquid (slag and steel) for a
known time. The refractory sample is sometimes covered by a mould powder to
minimise oxidation of the graphite (Fig. 11.93a). It is important to exclude air
during these experiments since oxidation of the carbon in the SEN can have a
serious effect on the erosion rates. The erosion rate is measured from either the
average depth of the groove or by estimating the volume of the eroded area (i.e. the
groove) (Fig. 11.93b) in a known time period.
There are a number of factors affecting SEN erosion:
• The slag phase will dissolve the oxide particles and the molten steel will dis-
solve the graphite particles.
• The driving force for both of these processes is the concentration difference
(Csat − C0) where C is the concentration and the subscripts, sat and 0, denote
the saturation limits and the current value, respectively.
• The saturation limits for various oxides in mould slags vary considerably (e.g.
Al2O3 is ca. 40%, TiO2 is ca. 10% and ZrO2 ca. 2%).
• These saturation limits also affect the kinetics of dissolution, for example in low
C steels (where Csat − C0 is large) the slowest (rate-determining) step is the
dissolution of oxide whereas, in high-C steels (where Csat − C0 is small) the
rate-determining step is the dissolution of graphite by the steel [250, 251].
520 11 Using Mould Fluxes to Minimise Defects and Process Problems

Fig. 11.93 Schematic drawings showing a dipping test and b measurement of volume of sample
eroded (permission granted, ISIJ, [249])

• The velocities of the flow of liquid metal and the concomitant flow of slag on
the SEN would also be expected to be significant factors affecting the erosion
rate.
• The SEN erosion rate was found to be greater when using EMBr that when not
using EMBr [48]; this was attributed to the better mould level control obtained
with EMBr which thus reduced the erosion zone and hence increased the depth
of erosion [48]. The increased meniscus temperature obtained with EMBr may
also be factor. EMS has also been reported to increase SEN erosion rates but this
was not attributed to the swirling flow produced by EMS but to slag crawling
and Chemical attack of the SEN by the entrained slag [49].

11.13.1 Factors Affecting SEN Erosion Rates

Nakamura et al. [249] and Tsukamoto [252] studied the erosion of ZrO2/C and
Al2O3/C refractories, respectively, in simulation tests and found that the erosion rate
increased as
• The mould slag viscosity decreased (or as the fluidity increased) and shown in
Fig. 11.94a [249, 252]. Plant data also indicate that erosion rates increase as the
mould slag viscosity decreases [81].
11.13 SEN Erosion 521

(a) 12 (b) 12

Erosion rate, mm hr -1
Erosion rate, mm hr -1 10 10

8 8

6 6

4 4

2 2

0 0
-0.5 0 0.5 1 1.5 2 2.5 0 2 4 6 8 10 12
in 1300 (dPas) F content, mass %

(c) 10 (d)
14

Erosion rate, mm hr -1
Erosion rate, mm hr -1

12
8
10

6 8
6
4 4
2
2 0
5 10 15 20 25 0 5 10 15 20 25
Na2O content, mass % O content in steel, 10-2 %

Fig. 11.94 SEN erosion rates of ZrO2/C refractory as a function of a ln η1300°C (d Pas) using four
mould powders [249], b F content of mould slag [249] and c Na2O content of mould slag showing
the effect of the ZrO2 stabiliser (● = stabiliser CaO; square = baddelyite; ▲ = MgO;
diamond = Y2O3) on the refractory [252] and d as a function of O content of the steel [252]
(permission granted, ISIJ, [249, 252]; re-drawn)

• The fluoride content of the mould slag increased (Fig. 11.94b) [81, 249, 252].
Several workers have reported that SEN erosion rates were halved when the
conventional powder was replaced with a F-free powder [253, 254].
• In the hierarchy of erosion rates for different SEN materials,
Baddleyite > MgO-stabilised ZrO2 > CaO-stabilised ZrO2 > Yttria-stabilised
ZrO2 [252] (Fig. 11.94c).
• The basicity, CaO/SiO2 ratio, of the mould slag increased (i.e. increasing
fluidity) [249, 252].
• The C content of the refractory increased (or the Al2O3 or ZrO2 content
decreased) [252].
• As the soluble Oxygen in the steel increases [252] (Fig. 11.94d).
• As MnO content of the slag increases [252].
• As Na2O content of the slag increases [252] (Fig. 11.94c).
522 11 Using Mould Fluxes to Minimise Defects and Process Problems

It should be noted that ZrO2 undergoes a phase transition at high temperature

and consequently, ZrO2 is usually stabilised by addition of CaO, MgO orY2O3. It
has been reported that fluorine and SiO2 in the mould slag tend to leach out the
stabiliser (e.g. CaO) and hence destabilise the ZrO2 [81]. SEN erosion has also been
found to increase with increasing metal flow velocity.

11.13.2 Causes, Mechanisms

The erosion of refractories is common to many high-temperature processes and is

frequently concentrated in the interfacial region in contact with both slag and metal.
Such erosion is often referred to as “slag-line attack”. In traditional oxide refrac-
tories, slag wets the oxide refractory and forms a meniscus (Fig. 11.95, where the
refractory is SiO2) and the slag, subsequently, dissolves the oxide. If the refractory
oxide causes a reduction in surface tension, the surface tension of the slag at the
point A will be lower than that at point B (i.e. cA < cB). This surface tension
gradient results in a Marangoni flow (which is always in the direction of low c to
high c, i.e. along liquid surface from A to B). This creates a clockwise vortex which
continually erodes the refractory. If the dissolution of oxide increases the surface
tension of the slag then cA > cB and the Marangoni flow will be in the direction of
B to A resulting in the formation of a vortex moving with an anticlockwise motion.
Several investigators have suggested that Marangoni flow is involved in
refractory erosion [250, 251, 255]. Hauck and Potschke [255] proposed that the
erosion takes place in two steps:
• Dissolution of carbon in the molten steel at the metal surface,
• Dissolution of the carbon in the slag (this would seem improbable since the
saturated limit for C in most slags is ca. 0.2%).
It was suggested that these steps resulted in the changes in surface tension, and
the Marangoni forces caused the formation of two vortices, at the metal /slag and
slag/air interfaces. It was postulated that these two erosion zones gradually
expanded with time and eventually, merged.

Fig. 11.95 Marangoni convection of slag film in the local corrosion zone of a SiO2(s)–(FeO–
SiO2) slag system showing typical “slag-line” attack (permission granted, IOM/Taylor & Francis,
[250])
11.13 SEN Erosion 523

In an alternative mechanism based on erosion rate measurements with Al2O3/C

and ZrO2/C refractories [256], it was proposed that oxide dissolution into the slag
was the rate-determining step; C-pick-up by the metal was considered as a quali-
tative measure of refractory attack [256]. Their results supported the view that ZrO2
increased the viscosity of the slag and created an inert protective layer [256].
Mukai [250, 251] proposed another mechanism to explain the erosion of oxide/C
refractories. The molten metal dissolves carbon and the slag dissolves oxides. At
high temperatures, most reactions (like dissolution) are accompanied by “reactive
wetting”. Mukai [250, 251] proposed that refractory erosion of oxide/C refractories
involved a cyclical two-stage process. Starting with the situation where the disso-
lution of carbon by the metal has left the refractory surface predominantly covered
by oxide (Fig. 11.96a). Since oxide is non-wetting to most metals but reacts with
slags, the interfacial conditions favour the covering of the surface by slag, so the
metal retreats and the refractory surface is covered by slag. The slag then dissolves
the oxide particles in the refractory until graphite particles mainly occupy the
surface sites (Fig. 11.96b). Since slag is non-wetting on graphite, the interfacial
conditions now favour the covering of the refractory by metal, so the slag retreats
and metal covers the refractory. Then graphite dissolves in the metal until the
surface is predominantly covered by oxide and the whole process starts over again.
Thus, dissolution is a cyclical process.

Fig. 11.96 Schematic diagrams illustrating the proposed mechanism for oxide/C refractories
where refractory surface is principally covered by a Oxide (white) and b Graphite (black)
(permission granted, IOM/Taylor & Francis, [250])
524 11 Using Mould Fluxes to Minimise Defects and Process Problems

The above mechanisms apply to steady state conditions. However, in continuous

casting the metal flow is both vigorous and fluctuating and the metal flow causes
drag forces in the slag pool. The velocities of these flows would be expected to
make a significant contribution to the erosion process.

11.13.3 Ways of Dealing with SEN Erosion

11.13.3.1 Reduce the Metal Flow Viscosity and the Concomitant Flow
in the Slag Pool

This can be achieved in the following ways:

• By optimising the SEN immersion depth and port design to minimise the metal
flow velocity and turbulence for a given casting speed [49].
• By applying EMBr or EMS to reduce the metal flow velocity; however, plant
data seem to indicate that EMBr and EMS seem to lead to slightly higher
erosion rates.
• Use a high viscosity mould powder to restrict the flow velocity in the slag pool
(although this will also reduce powder consumption which is not desirable).

11.13.3.2 Reduce the Driving Force for Dissolution of Refractory

Oxide

Use a mould powder containing 2% ZrO2 to minimise the driving force for ZrO2
dissolution (Csat − C0) since Csat for most mould slags is about 2% ZrO2 and thus
(Csa − C0)  0.

11.13.3.3 Reduce the Fluoride Content of the Mould Flux

It was reported that the replacement of a conventional powder with a fluoride-free

powder reduced the rate of SEN erosion [253] to half of that for a conventional,
F-containing, slag.

11.13.3.4 Increasing the Oxide Content of the SEN Refractory

Increasing the ZrO2 content of the refractory reduces the rate of SEN erosion [249,
252, 255, 256].
11.14 Fluorine Emissions 525

11.14 Fluorine Emissions

Casting powders consist of various oxides and most contain fluorides. These
oxyfluorides are basically unstable on heating since there are reactions between
oxides and fluorides at higher temperatures producing gaseous fluorides or HF (g).
These fluoride emissions pose several problems since
• They present a potential health hazard to plant operators [257].
• They change the composition of the mould powder (note these changes tend to
be much larger in laboratory experiments than in plant operation since the ratio
[surface area/mass (or volume)] is much larger in the laboratory experiments).
• The acidic emissions can cause corrosion to the plant.
• The F-emissions acidify the secondary cooling water which too can cause plant
corrosion [258, 259] and alkaline agents are added to the cooling water.

11.14.1 Factors Affecting Fluoride Emissions

When mould powders are heated they undergo a series of reactions occurring
between the components of the powders producing, for example [257]:

Na2 O þ CaF2 ¼ CaO þ 2NaF ðgÞ ð11:28Þ

SiO2 þ 2CaF2 ¼ 2CaO þ SiF4 ðgÞ ð11:29Þ

Al2 O3 þ 3 CaF2 ¼ 3CaO þ 2AlF3 ðgÞ: ð11:30Þ

These gaseous fluorides react with any moisture present in the atmosphere to
form
HF (g) for example:

H2 O ðgÞ þ 2NaF ðgÞ ¼ CaO þ 2HF ðgÞ: ð11:33Þ

The various gaseous fluorides emitted have been investigated by several workers
[257–261]. It was noted that NaF(g) and KF(g) were emitted at temperatures above
600 °C and the pressure of these fluorides increases with increasing temperature
[207]. Thermodynamic calculations indicate that the vapour pressure of SiF4(g) at
high temperature can exceed 0.5 atm [207, 260–262]. The kinetics of these fluoride
emissions has also been studied [263].
526 11 Using Mould Fluxes to Minimise Defects and Process Problems

It has also been observed that fluoride emissions increased as

• The fluoride content of the mould slag increased. [257, 264].
• The CaO and MgO content decreased. [257, 264].
• The SiO2, Na2O and TiO2 contents increased [264].
• As CaF2 was replaced with cryolite (Na3AlF6) (causing a 10-fold increase in
emissions) [257].
• As the basicity decreased [257].
All of the above findings are consistent with recent observations on the effect of
CaF2 on slag structure, namely, that F− ions tend to bond preferentially with cations
of high-field strength (z/r2) like Mg2+ and Ca2+ and do not tend to bond with other
components in the slag, e.g. Si4+. High pSiF4 values arise because of the interaction
of SiO2 (which is present in large amounts) with the fluorides present. Prefusing the
slag reduces the chemical activities of the various fluorides in the mould slag and
thereby reduces the partial pressures of the gaseous fluorides.

11.14.2 Ways of Dealing with Fluoride Emissions

Fluoride emissions can be reduced by

• Reducing the amount of fluorine in the slag.
• Using F-free slags which would have the additional benefit of reducing
SEN-wear [253, 254].
• Using a pre-fused casting powder which reduces the chemical activity of the
CaF2 (aCaF2) and hence, reduces the vapour pressures of fluoride species [257].
• Use an alkaline additive to the cooling water to reduce its acidification [258].

11.15 Summary

The causes of the various defects and process problems, the strategies adopted and
suggested remedial treatments are all summarised in Table 11.9 for each type of
defect.
11.15 Summary 527

Table 11.9 Summary of the causes, strategies and treatments used to overcome defects and
process problems in continuous casting. TSL thermal shrinkage coefficient
Defect/problem Cause Strategy Treatment
Longitudinal 4% difference in TSL for Produce thin uniform Reduce qhor by
cracking [6–10, ∂-Fe and c-Fe shell by reducing qhor creating a thick,
12–16] especially, MC steels! in meniscus region, i.e. crystalline slag film
Variations in shell create a thick (with –use flux with high
thickness ! “hoop high Tbr) crystalline Tbr and (C/S)  1.3
stresses” ! relieved by slag film
cracking
Longitudinal Local variations in shell Produce thin uniform 1. Use flux with high
corner cracking thickness shell Tbr and forms
[42, 60–64] 1. Billets: over-cooled crystalline slag film
corners 2. Redesign SEN
2. Slabs: Metal flow ports use lower
causes shell thinning in casting speed
corners [50]
Sticker Various causes all Produce thick stronger Use a glassy slag with
breakouts [67, involve forming of a shell by increasing qhor low Tbr
68, 70, 72, 73, thin, weak shell with (C/S) 0.9
81, 83, 85, 86, 1. Large agglomerates Ensure %
90] blocking slag infiltration H2O < 0.5% and no
plus Carbon water leaks
pick-up ! low-melting
shell—does not freeze in
tp
2. H2 in slag film [60]
Oscillation Natural event due to tide Decrease dOM by 1. 1. Reduce tn by "f;
marks changes in slag flow. Reducing tn; "Vc:#s
(OMs) [43, 55, (Sect. 5.3) 2. increase drim/tip Increase slag viscosity
103, 107, 109, 2. Delay solidification
111, 112] [105, by
113] "superheat
Transverse 1. Due to incorrect taper Ensure good powder Ensure correct taper
cracks [45, 139, for steel consumption [29] for steel [29]
145–147, 149] 2. Associated with deep Reducing qhor in Reduce dOM ! #tn # s
OM’s meniscus region, i.e.
create a thick (with
high Tbr) crystalline
slag film
Star cracking Loss of liquid slag in Create liquid slag film 1. Increase casting
[156–158] lower mould ! spalling through mould speed and superheat
of solid slag film and 2. Decrease water
variable heat flux [51] flow rate
Depressions Poor mould level Allow longer time Improve mould level
1. Longitudinal control + formation of before replacing starter control and use a
[129–131] latge slag flux with conventional lower viscosity flux
2. Transverse rim ! overflow and flux [111]
[160] fracture and capture of
rim
(continued)
528 11 Using Mould Fluxes to Minimise Defects and Process Problems

Table 11.9 (continued)

Defect/problem Cause Strategy Treatment
Slag entrapment 1. Slag entrapment a. Decrease metal flow Optimise SEN depth,
[163–166, 179, Turbulence in metal and velocity (e.g. by port design and Ar
183, 191, 197, slag flows EMBr) flow rate
198] Al2O3 formed by b. Increase slag Use EMBr
Slivers reaction of Al in steel viscosity Minimise slag
[78, 211, 213, with FeO in slag-trapped Minimise slag carry-over and
219–221] by shell carry-over from ladle decrease throughput
Gas entrapment Gas trapped by Reduce the length of Thicker bed,
Pinholes (LC, meniscus-metal flow meniscus hook [28] exothermic
ULC steels) [46, takes bubble too far Optimise metal flow agents-Optimise SEN
50, 51, 213, 214] down-inclusion attaches Delay solidification depth, port design and
“Pencil pipe” to bubble and is not Ar flow rate, EMBr
[198, 212] welded shut in rolling
SEN erosion ZrO2 has 2% solubility Possible to: Reduce metal and
[249, 250, 252, in slag; C dissolves in Saturate slag with 2% slag flows
253] steel ZrO2 —Use F-free powders
Metal (& slag) flow Increase viscosity of
impact on SEN slag
Carbon pick-up 1. Contaminated by C in 1. Minimise standing 1. Reduce C in flux
Especially in slag rim wave 2. Replace C partially
LC, ULC grades 2. Steel poking through 2. Keep a deep slag with SiN
[242–244, 247, into powder bed pool
248]
Fluoride Reactions of oxides and 1. Reduce F-in powder Add alkaline additive
emissions [207, fluorides at higher 2. Use F-free powders to cooling water
257, 258, 263, temperatures 3. Use pefused
264] powders

References

1. NS Hunter, JD Madill, DJ Coones, PN Hewitt, D Stewart, Steel Technol. Intl., 1996, 171,
(1996).
2. D Stewart, PN Hewitt, L Peeters, Proc. 79th Steelmaking Conf. 1996, (ISS, Warrendale, PA,
1996) p. 207.
3. Y Maehara, K Yasumoto, Y Sugitani, K Gunji, Trans. ISIJ, 25, 1045, (1985).
4. E Schmidtmann, L Pluegel, Archiv. f. Eisenhuttenw., 51, 49, (1980).
5. M Wolf, Proc. METEC Congress ’94, 2nd Europ. Conf. Continuous Casting, Dusseldorf,
1994, (VDEh, Dusseldorf, 1994) vol. 1 p. 78 see also Proc. 81st Steelmaking Conf., 1998,
(ISS, Warrendale, PA, 1998) p. 53.
6. S Hiraki, K Nakajima, T Murakami, T Kanazawa, Proc. 77th Steelmaking Conf., Chicago,
1994, (ISS, Warrendale, PA, 1994) p. 397.
7. M Kawamoto, T Kanazaka, S Hiraki, S Kumakura, Proc. 5th Intl. Conf. Molten slags, salts
and fluxes, Sydney, 1997, (ISS, Warrendale, PA, 1997) p. 777.
8. D Springorum, Steel Times, 1969 (Nov), 727, (1969).
9. K. Nakajima, S Hiraki, T Kanazawa, T Murakami, CAMP-ISIJ, 5, 1221, (1992).
References 529

10. KC Mills, AB Fox, R P Thackray, Z Li, Proc. 7th Intl. Conf. Molten slags, fluxes and salts,
Capetown, 2004, (SAIMM, Johannesburg, 2004) p. 713.
11. H Uchiyama, Paper prepared for AISI Technical Committee on Strand casting, Sept., 1985.
12. HP Narzt, R Martinelli, C Furst, G Xia, H Pressinger, Proc. 3rd Europ. Conf. Continuous
casting, Madrid, 1998, (UNESID, Madrid, 1998) p. 747.
13. SG Kollberg, HR Hackl, PJ Hanley, Iron and Steel Engineer, 73, (7), 24, (1996).
14. S Umeda, Y Tsukaguchi, H Mikai, Y Hitomi, M Kawamoto, CAMP- ISIJ, 7, 302, (1994).
15. CA Dacker, T Sohlgren, Steel Research Intl., 81, 899, (2010).
16. K Brimacombe, F Weinberg, EB Hawbolt, Met. Trans. B, 10B, 279, (1979).
17. RJ Gray, A Perkins, B Walker, Proc. Intl. Conf. on Soldification, Sheffield, 1977, (Metals
Soc., London, 1977) p. 300.
18. T Narita, T Nozaki, T Mori, H Yasunaka, T Fujimoto, T Ohnishi, Trans. ISIJ, 21, B 159,
(1981).
19. Y Ito, I Shimoda, M Suzuki, Y Miki, Proc. 4th Intl. Congress Science, Technol. Steelmaking,
Gifu City, 2008, (ISIJ, Tokyo, 2008) p. 698.
20. MJ Lu, JH Chen, DH Tsai, CH Huang, Proc. 78th Steelmaking Conf., 1995, (ISS,
Warrendale, PA, 1995) p. 359.
21. C Bertolletti, P Courbe, JM Jolivet, PP Naveau, A Oper, E Perrin, C Salaris, AL Spierings, E
Wiesseldinger, Proc. 3rd Europ. Conf. Continuous Casting, Madrid, 1998, (UNESID,
Madrid, 1998) p. 65.
22. P Courbe, JM Jolivet, G Lesoult, E Perrin, P Nilles, Steel Research, 70 (8/9), 338, (1999).
23. M M’Hamdi. G Lesoult, E Perrin, JM Jolivet, ISIJ Intl., 36, S 197, (1996).
24. K Nakai, T Sakashita, M Hashio, M Kawasaki, K Nakajima, Y Sugitani, Tetsu-to-Hagane,
73, (3), 498, (1987).
25. Y Sugitani, M Nakamura, M Okuda, M Kawasaki, K Nakajima,Trans. ISIJ, 25, B 91,
(1985).
26. JW Cho, HT Jeong, Met. Mater. Trans. B, 44B, 146, (2013).
27. H Murukami, M Suzuki, T Kitagawa, S Miyahawa, Tetsu-to- Hagane, 78, 105, (1992).
28. T Mochida, S Kohriyama, S Itoyama, T Yamashita, M Suzuki, Y Kishimoto, CAMP-ISIJ,
20, 851, (2007).
29. K Ichikawa, A Morita, Y Kawabe, Shinagawa Tech. Report, 36, 283, (1993).
30. K Watanabe, M Suzuki, K Murakumi, H Kondo, A Miyamoto, T Shiomi, Tetsu-to-Hagane,
83, 115, (1997).
31. V Guyot, JF Martin, A Ruelle, A d’Anselme, JP Radot, M Bobadilla, JV Lavant, JN
Pontoire, ISIJ Intl., 36, S 227, (1996).
32. A Yamauchi, S Itoyama, Y Kishimoto, H Tozawa, K Sorimachi, ISIJ Intl., 42, 1094, (2002).
33. T Watanabe, CAMP- ISIJ, 14, 153, (2001).
34. A Cristallini, R Langella, A Polk, Proc. 3rd Europ. Conf. Continuous Casting, Madrid, 1998,
(UNESID, Madrid, 1998) p. 589.
35. A Delhalle, M Larrecq, J Petegnief, JP Radot, 4th Intl Conf. Continuous Casting, Brussels,
1988, (VDEh, Dusseldorf, 1988) Preprints 1, p. 38.
36. T Tsai, H Yasunaka, M Kogita, T Inoue, K Kan, CAMP-ISIJ, 7, 1149, (1994).
37. M Ozgu, MB Perrini GP Buckley, Proc. 79th Steelmaking Conf., 1996, (ISS Warrendale,
PA, 1996) p. 195.
38. C-Å Däcker, A Salwén, P Andersson, C Eggertsson, Proc. 7th Europ. Continuous Casting
Conf., Düsseldorf, 2011, (VDEh, Dusseldorf, 2011). Session 12.
39. M Kawamoto, T. M u r a k a m i, M. H a n a o, H. K i k u c h i, T. Wa t a n a b e, Proc. 6th
Intl. Conf. Molten slags, fluxes and salts, Stockholm, paper 146, (2000).
40. M Kawamoto, Paper presented at Intl. workshop on “ Thermophysical Data for development
of mathematical models of solidification”, Gifu City, Japan, (1995).
41. M Suzuki, S Miyahara, T Kitagawa, S Uchida, K Okim, Tetsu- to Hagane, 78, 113, (1992).
42. K Sorimachi M Kuga, K Marumoto, T Koshikawa, K Hamagami, H Kitaoka, Trans. ISIJ, 22
(2), B 17, (1982).
43. C-Å Däcker, T. Sohlgren, Proc. 5th Europ. Cont. Casting Conf., Nice, 2005, pp. 10–17.
530 11 Using Mould Fluxes to Minimise Defects and Process Problems

44. K Tsukaguchi, S Ura, A Shiraishi, Y Hitomi, T Nagahata, Steel Technol. Intl., 6, 175,
(1995/6).
45. S Ogibayashi, Paper Intl. Workshop on “ Thermophysical Data for development of
mathematical models of solidification”, Gifu City, Japan, (1995).
46. E Takeuchi, H Harada, H Tanaka, T Ishii, T Toh, M Zeze, M Hojo, K Shigematsu, Nippon
Steel Technical Report, 61, 29, (1994).
47. M Washio, M Sugizawa, S Moriwaki, K Kariya, S Idogawa, S Takeuchi, Revue de
Metallurgie, CIT, 90, (April), 507, (1993).
48. H Take, H Osanai, J Hasunuma, T Yamamoto, H Bada, H Tozawa, Proc. Conf. “Quality of
ordinary steel in Iron- and steel-making”, Bangkok, (1994) Session 3, Paper 1.
49. E Favre, S Kunstreich, MC Nove, D Rotelec, W Courths, E Korte, Proc. 3rd Intl Conf.
Continuous Casting, Madrid, 1998, (UNESID, Madrid, 1998) p. 595.
50. T Matsumiya, ISIJ Intl., 46, 1800, (2006).
51. M Tani, T Toh, K Umetsu, K Tanaka, M Zeze, K Tsunenari, K Hayashi, S Fukunaga,
Nippon Steel Technical Report, 104, 62, (2013).
52. M Hanao, M Kawamoto, M Hara, T Murakami, H Kikuchi, A Yamanaka, Proc. 5th
Europ. Conf. Cont, Casting, Nice, 2005 (La Revue Metall., Paris, 2005), see also Tetsu-to
Hagane, 88, (1), 23, (2002).
53. D Gottlelf, P Andrezjewski, E Julius, H Haubrich, Proc. 3rd Europ. Conf. Continuous
Casting, Madrid, 1998, (UNESID, Madrid, 1998) p. 825.
54. P E Ramirez-Lopez, P.D. Lee, K.C. Mills, ISIJ Intl., 50, (3), 425, (2010).
55. PE Ramirez-Lopez, KC Mills, PD Lee, B Santillana, Met. Mater. Trans., B, 43B, 109
(2012).
56. SN Singh, KE Blazek, J. Metals, 26, (10), 17, (1974).
57. C-Å Däcker, T Sohlgren, Steel Research Intl., 10, 899, (2010).
58. AS Normanton, PN Hewitt, NS Hunter, B Harris, D Scoones, Proc. 4th Europ. Conf.
Continuous casting, Birmingham, 2002, (IOM, London, 2002) p. 561.
59. T Emi, Proc. 74th Steelmaking Conf., Washington, DC, 1991 (ISS, Warrendale, PA, 1991),
p. 623.
60. JK Park, BG Thomas, IV Samarasekera, Ironmaking and Steelmaking, 29, 359, (2002).
61. H Tai, M Morashita, T Miyake, Proc. 3rd Europ. Conf. Continuous Casting, Madrid, 1998,
(UNESID, Madrid, 1998) p. 447.
62. T Miyake, K Nakayama, M Moroshita, H Tai, Kobelco Technol. Review, 21, (Apr), 7,
(1998).
63. E Kivela, Revue de Metallurgie, CIT, 91, (Jan), 115, (1994).
64. CS Li, BG Thomas, Proc. ISS Tech Conf., Indianapolis, 2003, (ISS, Warrendale, PA, 2003)
p. 685 see also Met. Mater. Trans. 35B, 1151, (2004).
65. R Nishimachi, Y Ogura, SEAISI Quarterly, 25, (Oct.), 44, (1996).
66. M Miyazaki, CAMP- ISIJ, 20, 866, (2007).
67. A. Tsuneoka, W. Ohashi, S. Ishitobi, Proc. 68th Steelmaking Conf., 1985, (ISS, Warrendale,
PA, 1985) p. 3.
68. Y Liu, XD Wang, M Yao, AB Zhang, H Ma, Z Wang, JC Ma, X Wang, GQ Shi, Ironmaking
and Steelmaking, 41, 748, (2014).
69. MM Wolf, Iron and Steelmaker, 23, (2), 47, (1996).
70. KC Mills, TJ Billany, AS Normanton, B Walker, P Grieveson, Ironmaking and Steelmaking,
18, 253, (1991).
71. J Kromhout, RS Schimmel, Proc. 8th Europ. Conf. Continuous casting, Graz, Austria, 2014,
(Austrian Metals Soc., 2004).
72. M Hanao, Y Tsukaguchi, M Kawamoto, Proc. 4th Intl. Cong. Sci. Technol. Steelmaking,
Gifu, p. 694 (ISIJ, Tokyo. 2008).
73. YA Meng, BG Thomas, Met. Mater. Trans. B, 34B, 707, (2003).
74. JA Kromhout, PhD Thesis “Mould powders for the high speed continuous casting of steel”
Univ. of Delft, (2011).
75. Q Wang Y. Lu, S. He, K.C. Mills, Z.S. Li, Ironmaking and Steelmaking, 38, 297, (2011).
References 531

76. T Mukongo, Effect of titanium pick up on mould flux viscosity in continuous casting of
Ti-stabilised steel. MSc Thesis, Univ. Pretoria: (2003) see also ” Proc. ISSTech. Conf., 2003,
(ISS, Warrendale, PA, 2003) p. 675.
77. T Mukongo, C Pistorius, A Garbers-Craig, Ironmaking Steelmaking, 31, 135, (2004).
78. RC Nunnington, N Sutcliffe, Proc. 59th Electric Furn. Conf., Pheonix, AZ, Nov. (2001),
(AIST, Warrendale, PA) p. 361.
79. T Kishi, H Takeuchi, M Yamamiya, H Tsuboi, T Nakano, T Ando, Nippon Steel Technical
Report, 34, 11, (1987).
80. RJ Phillips, WF Salem, WR Emling, HD Baker, Proc. 73rd Steelmaking Conf., 1990, (ISS,
Warrendale, PA, 1990) p. 247.
81. DW Bruce, NS Hunter, Proc. 2nd Europ. Conf. Continuous Casting, Dusseldorf, 1994
(VDEh, Dusseldorf, 1994) p. 156.
82. K Tokiwa et al., Tetsu-to-Hagane, 69, S 1034, (1983).
83. T. Mizukami, K Kawakami, S Miyahara, M Suzuki, T Kitagawa, O Terada, Trans. ISIJ, 25,
B 300, (1985).
84. T. Mizukami, A Ozeki, A Kurabayashi, N Hsebe, S Uchida, T Kitagawa, Trans ISIJ, 25, B
301, (1985).
85. M Mukai, K Yamaguchi, S Ogibayashi, Trans ISIJ, 26, (4), B 163, (1986).
86. JN Pontoire, JP Radot, V Delvaux, E Dehaussy, Revue de Metallurgie, CIT, 93, 1237,
(1996).
87. Y Ueshima, T Mizoguchi, T Kajitani: Proc. 9th Intl. Conf. Molten slags, fluxes and salts
(Molten 12), Beijing, 2012, (Chinese Metals Soc., Beijing, 2012).
88. H Yamamura, T Kajitani, J Nakashima, M Yamasaki, S Mineta, Nippon Steel Technical
Report, 104, 54, (2013).
89. S Ogibayashi, K Yamaguchi, T Mukat, T Takahashi, Y Mimura, K Koyama, Y Nagano, T
Nagano., Nippon Steel Technical Rept, 34, 1, (1987).
90. K Sorimachi, M Kuga, H Nishikawa, K Marumoto, H Nakato, Trans ISIJ, 22 (4), B 435,
(1982).
91. H Nakato, T Sakuraya, T Nozaki, T Emi, H Nikoshawa, Proc. 69 th Steelmaking Conf.,
Washington, DC, 1986, p. 23.
92. H Kyoden, T Diohara, O Nomura, Mould Powders for Continuous Casting and Bottom Pour
Teeming (ISS, Warrendale, PA, 1987) p. 45.
93. P Zasawoski, DJ Sosinski, Proc. 73rd Steelmaking Conf., 1990, (ISS, Warrendale, PA, 1990)
p. 253.
94. SY Kim, J Choi, Proc. 4th Europ. Conf. Continuous casting, Birmingham, 2002, (IOM,
London, 2002) p. 594.
95. XD Wang, M Yao, X Chen, ISIJ Intl., 48, 1047, (2008).
96. CH Moon, DM Lee, SC Moon, HD Park, ISIJ Intl., 48, 48, (2008).
97. S Tanaka, H Misumi, S Mizoguchi, H Horiguchi, Trans. ISIJ, 21, B 350, (1981).
98. H Takeuchi, S Matsumara, R Hidaka, Y Nagano, Y Suzuki, Trans. ISIJ, 22, B 16, (1982).
99. H Jacobi, Arch. Eisenhüttenwesen, 47, 441, (1976).
100. M Tomono, W Kurz, W Heinemann, Met. Trans. B, 12B, 409, (1981).
101. K.C. Mills, P.E. Ramirez-Lopez, P.D. Lee, Sano Symp., Tokyo, 2008, (Univ. Tokyo, 2008)
p. 158.
102. C-Å Däcker, Research Report “Case study-mould powder” IM -2003-534, Swerea KIMAB,
Stockholm, 2003.
103. PD Lee, R Ramirez-Lopez, KC Mills, B Santillana: Ironmaking and Steelmaking, 39, 244,
(2012).
104. E Takeuchi, J K Brimacombe, Met. Trans. B, 15B, 493, (1984).
105. R. Aigner, H. Steinruck, Eurotherm 82, Numerical Heat Transfer, ed. R.A. Białecki and A.
Nowak, Gliwice–Cracow, Poland, 2005.
106. JM Hill, YH Wu, B Witwatanapataphee, J. Eng. Math., 36, 311, (1999).
107. S Itoyama, CAMP-ISIJ, 5, 1225, (1992).
108. S Itoyama, Curr. Adv. Mater. Proc. (Japan), CAMP-ISIJ, 5, 1225, (1992).
532 11 Using Mould Fluxes to Minimise Defects and Process Problems

109. S Itoyama, H Tozawa, K Sorimachi, Proc. 77 th Steelmaking Conf. (ISS, Warrendale, PA,
1994) p. 365.
110. T Fastner, C Furst, HP Narzt, G Xia, G Zuba, Proc. 3 rd Intl. Conf. Continuous Casting,
Madrid, 1998, (UNESID, Madrid,1998) p. 791.
111. K Schwerdtfeger, Hong Sha, Met. Trans. B, 31B, 813, (2000).
112. S Nebeshima, Y Itoh, H Tozawa, H Nakato, K Sorimachi, Proc. 4th Intl. Conf. Solidification
Processing, Sheffield, 1997 (Sheffield Univ., 1997) p. 10.
113. H.J. Shin, S.H. Kim, B.G. Thomas, G.G. Lee, J.M. Park, J. Sengupta, ISIJ Intl., 46, 1635,
(2006).
114. K Shimizu, AW Cramb, Iron and Steel Maker, 2002, (6), 43, (2002).
115. Y Murunaka et al, 102nd ISIJ Meeting, November 1981, S 905.
116. OD Kwon, J Choi, IR Lee, JW Kim, KH Moon, YK Shin, Proc. 74th Steelmaking Conf.,
1991, (ISS, Warrendale, PA, 1991) p. 561.
117. RB Mahapatra, K Brimacombe, IV Samarasekera. Met. Trans. B, 22 B, 875, (1991).
118. G.G. Lee, H. Shin, S.-H. Kim, S.-K. Kim, W.-Y. Choi, B.G. Thomas, Ironmaking
Steelmaking, 36 (1), 40, (2009).
119. H Steinruch, C Rudischer, W Schneider, Proc. Conf. Modelling Casting, Welding and
Advanced Solidification processes VIII (MCWASP) (Minerals, Metals and Materials Soc.,
1998).
120. H Yasunaka, T Mori, H Nakata, T Mori, F Kamai, S Narada, Trans. ISIJ, 25, B90, (1985).
121. M Suzuki, T Kitagawa, S Uchida, K Ozawa, T Mori, Trans. ISIJ, 26, B164, (1986).
122. F Neumann, J Neal, MA Pedroza, AH Castillejo, FA Acostra, Proc. 79th Steelmaking Conf.
(ISS, Warrendale, PA., 1996) p. 249.
123. D W van der Plas, C Platvoet, B Diesesme, JP Radot, JM Galpin, Proc. 2nd Europ. Conf.
Continuous casting, Dusseldorf, 1994, Metec Congress’94 (VDEh, Dusseldorf, 1994)
p. 109.
124. R Koldewijn Unpublished Corus Internal Rept (2007) cited in KC Mills, J Kromhout, A
Hamoen, R Boom, Proc. Admet Conf, Dnipetrovsk, 2007, (Natl. Metall. Acad. Ukr.,
Dnipropetrovsk) vol 2, p. 174.
125. T. Toh, E. Takeuchi, M. Hojo, H. Kawai, S. Matsumura, ISIJ Intl, 37, 1112, (1997).
126. G. Skoczylas, Proc. 79th Steelmaking Conf., 1996, (ISS, Warrendale, PA, 1996) p. 269.
127. Y Fukuda, H Kawai, M Okimori, M Hojo, S Tanaka, Proc. 5th Intl. Conf. Slags, Fluxes and
molten slags, Sydney, 1997, (ISS, Warrendale, PA., 1997) p. 791.
128. H Fredrikson, J Elfsberg, J Scand. Metall., 31, 292, (2002).
129. JW Kim, S. K. Kim, D. S. Kim, Y. D. Lee, J. I. Eum, E. S. Lee, Proc. 78th Steelmaking
Conf.,1995, (ISS, Warrendale, PA, 1995) p. 333.
130. J Madill, Proc. Workshop on Steelmaking, Steel cleanliness and mould thermal monitoring,
Brussels, 1993, (EC Sci. Tech. Comm. Publ., Luxembourg, 1994) p. 104.
131. M Jenkins, BG Thomas, WC Chen, RB Mahapatra, Proc. 77th Steelmaking Conf., 1994,
(ISS, Warrendale, PA. 1994) p. 337.
132. Jae Eung. CAMP-ISIJ, 8, 218, (2003).
133. P. Ackermann, Ph.D. Dissertation, Ecole Polytechnique Federale de Lausanne, Lausanne,
(1983).
134. T Emi, H Nakato, Y Iida, K Emoto, R Tachibana, T Imai, H Bada, Proc. 61st NOH-BOS
Conf., 1978, (ISS, Warrendale, PA, 1978) p. 350.
135. H. Tomono, Ph.D. Dissertation, Ecole Poly. Federale, d’Lausanne, Lausanne, 1979.
136. H Tomono, H Ackermann, W Kurz, W Heinemann, Solidification Technology in the
foundry and Cast house, Warwick, 1983 (The Metals Soc., London, 1983) Book 273 p. 524.
137. M Nakato, Tetsu-to Hagane, 61 (12), 131, (1975).
138. P Riboud and M Larrecq, Proc. 62nd NOH-BOS Conf., Detroit, 1979, (ISS, Warrendale,
PA., 1979) p. 78.
139. A.S. Normanton, V. Ludlow, M. Hecht, B. Harris, C-Å. Däcker, A. Di. Donato, T. Sohlgren,
ECSC-Report 7210.PR/273, 2005, (Europ. Comm. Sci. Tech. Publ., Luxembourg, 2005).
References 533

140. J Sengupta, BG Thomas, HJ Shin, GG Lee, SJ Kim, Met. Mater. Trans. A, 37A, 1597,
(2006).
141. J. Sengupta, B.G. Thomas, JOM, 58, 16, (2006).
142. PE Ramirez-Lopez, KC Mills, PD Lee, B Santillana, Proc. 7th Europ., Conf. Continuous
Casting, 2011, Metec Congress, Dusseldorf, 2014 (VDEh, Dusseldorf, 2011).
143. BG Thomas, J Sengupta, C Ojeda, Proc. 2nd Baosteel Biennial Conf., 2006, Shanghai, vol. 1,
(2006), p. 112.
144. C. Ojeda, J. Sengupta, B.G. Thomas, J. Barco, J.L. Aruna, Proc. AIST Tech., 2006, vol.
1 (ISS Warrendale, PA) p. 1017.
145. C. Offerman, C-Å. Däcker, C. Enström, Scand. J Metallurgy, 10, 115, (1981).
146. M.M. Wolf, Continuous Casting, Volume 9, Iron & Steel Society, Warrendale, PA, p. 285
(1997).
147. A.M. El-Wazri, E.Es-Sadiqi, S. Yue, Proc. 39th MCWSP Conf., ISS, XXXV, 1998, p. 721.
148. H. Bruce, Swerea KIMAB Report, IM-2002-530 (2002).
149. Z Buzek, E Mazancova, Z Jonsta, F Mazanek, www.metal2013.com/files/proceedings.
150. C Houpert, V Lautari, JM Jolivet, M Guttmann, JP Birat, M Jallon, M Contente, Revue. de
Metall., 94, 1369, (1997).
151. M R Ozgu, B Kocatulum, Iron and Steelmaker, 21, (5), 77, (1994).
152. J Nieto, S Morales, R Garcia, G Campos, Proc. 3rd Intl. Conf. Cont. Casting, Madrid, 1998,
(UNESID, Madrid, 1998) p 485.
153. G Bocher, U Hoffmann, P Muller, Proc. 2nd Europ. Conf. Continuous casting, Dusseldorf,
1994, Metec Congress’94 (VDEh, Dusseldorf, 1994) p. 103.
154. L. Schmidt, A. Josefsson, Scand. J. Metallurgy, 10, (1981).
155. B. Mintz, R. S. Yue, J.J. Jonas, Intl. Materials Review, 36, 5, (1991).
156. A Wyckaert, Ironmaking and Steelmaking, 4, (6), 340, (1977).
157. K Relander, Jernkonorets Ann., 155, (9), 565, (1971).
158. TJ Billany, AS Normanton, KC Mills, P Grieveson, Ironmaking and Steelmaking, 18, 403,
(1991).
159. RJ O’Malley, J Neal, Proc. Intl. Conf. New Developments in Metallurgical Process
Technol., Dusseldorf, METEC Congress, 1999, (VDEh, Dusseldorf, 1999) p. 73.
160. M S Jenkins, BG Thomas, Proc. 80th Steelmaking Conf., Chicago, IL, 1997, (ISS,
Warrendale, PA, 1997) p. 285.
161. BG Thomas, A Moitra, R McDavid. Process Technol. Conf., 13, 1996, (ISS, Warrendale,
PA, 1996) p. 143.
162. A Kusano, N Sato, M Okimori, S Fukunaga, K Nishihara, M Sato, Y Minigawa, Proc. 74th
Steelmaking Conf., 1991, (ISS, Warrendale, PA, 1991) p. 111.
163. LC Hibbeler, BG Thomas, Proc. AISTech Conf. 2010, Pittsburgh (ISS, Warrendale. PA,
2010) p. 1215.
164. LC Hibbeler, BG Thomas, Iron and Steel Technology, 2013, (Oct), 121, (2013).
165. T. Teshima, M. Osame, K. Okimoto, Y. Nimura, Proc. 71st Steelmaking Conf., 1988, (ISS,
Warrendale, PA, 1988), p. 111.
166. J. Kubota, K. Okimoto, A. Shirayama, H. Murakami, Proc. 74th Steelmaking Conf., 1991,
(ISS, Warrendale, PA, 1991) p. 233.
167. H. Nakato, K. Saito, Y. Oguchi, N. Namura, K. Sorimachi, Proc. 70th Steelmaking Conf.,
1987, (ISS, Warrendale, PA, 1987) p. 427.
168. Z. Wang, K. Mukai, Z. Ma, M. Nishi, H. Tsukamoto, F. Shi, ISIJ Intl, 39, 795, (1999).
169. M Cervantes, H Gustavsson, Proc. 3rd Europ Conf. Continuous casting, Madrid, 1998,
(UNESID, Madrid, 1998) p. 202.
170. A. Matsushita, K. Isogami, M. Temma, T. Ninomiya, K. Tsutsumi, Trans. ISIJ, 28, 531,
(1988).
171. B.S. Moghaddam, E. Steinmetz, P. R. Scheller, Proc. 3rd Intl. Congress Sci. Technol.
Steelmaking, 2005, (AIST, Warrendale, PA, 2005) p. 911.
172. G.A. Panaras, A. Theodorakakos, G. Bergeles, Met. Mater. Trans, 29B, 1117, (1998).
534 11 Using Mould Fluxes to Minimise Defects and Process Problems

173. PE Ramirez- Lopez, PN Jalali, J Bjorkvall, U Sjostrom, C Nilsson, ISIJ Intl. 54, 342.
(2014) and Proc. 8th Europ. Conf. Continuous Casting, Graz, Austria, 2014 (ASMET,
Vienna, 2014).
174. E van Vliet, DW van der Plas, SP Carless, A A Kamperman, AE Westendorp: Proc. 7th
Europ Conf. Cont. Casting, Dusseldorf, 2011, (VDEh, Dusseldorf, 2011) Session 4.
175. S. Yamashita, M. Iguchi, ISIJ Intl., 41, 1529, (2001).
176. T. Watanabe, M. Iguchi, ISIJ Intl., 49, 182, (2009).
177. R.M. McDavid, B.G. Thomas, Met. Mater. Trans., 27B, 672, (1996).
178. B Zhao, SP Vanka, BG Thomas, Intl. J Heat Fluid Flow, 26, 105, (2005).
179. Q. He, G. Evans, R. Serje, T. Jaques, Proc. AISTech, 2009, (AIST, Warrendale, PA, 2009),
No. 2, p. 57.
180. H. Bai, B.G. Thomas, Met. Mater. Trans, 32B, 707, (2001),
181. G.-G. Lee, H.-J. Shin, B.G. Thomas, S.-H. Kim, Proc. AISTech, 2008, (AIST, Warrendale,
PA, 2008), No. 2, p. 63.
182. B. Li, T. Okane, T. Umeda, Met. Mater. Trans.; 32B, 1053, (2001).
183. B. Li, F. Tsukihashi, ISIJ Intl., 45, 30, (2005).
184. N. Kasai, M. Iguchi, ISIJ Intl., 47, 982, (2007).
185. B. Li, T. Okane, T. Umeda, Met. Mater. Trans, 32B, 1053, (2001).
186. M. Gebhard, Q.L. He, J. Herbertson, Proc. 76th Steelmaking Conf., (ISS, Warrendale, PA,
1993), p. 441.
187. Y.H. Wang, Proc. 73rd Steelmaking Conf., 1990, (ISS, Warrendale, PA, 1990), p. 473.
188. D. Gupta, A.K. Lahiri, Met. Mater. Trans, 25B, 227, (1994).
189. K Tsutsumi, K Watanabe, M Suzuki, M Nakada, T Shiomi, Proc. 7th Intl. Conf. Molten
slags, fluxes and salts, Cape Town, 2004, (SAIMM, Johannesburg, 2004) p. 803.
190. Q. He, ISIJ Intl., 33, 343, (1993).
191. A. Theodorakakos, G. Bergeles, Met. Mater. Trans, 29B, 1321, (1998).
192. D. Gupta, A.K. Lahiri, Met. Mater. Trans, 27B, 695, (1996).
193. T. Honeyands, J. Herbertson, Steel Research Intl, 66, 287, (1995).
194. J. Anagnostopoulos, G. Bergeles, Met. Mater. Trans, 30B, 1095, (1999).
195. M. Kamal, Y. Sahai, Steel Research Intl., 76, 44, (2005).
196. R. Chaudhary, B.T. Rietow, B.G. Thomas, Proc. 2009 Mater. Sci. and Technol. Conf.,
(AIST/TMS, Warrendale, PA., 2009), p. 1090.
197. M. Harman, A.W. Cramb, Proc. 79th Steelmaking Conf. 1996, (ISS, Warrendale, PA, 1996),
p. 773.
198. W Emling, T A Waugaman, SL Feldbauer, AW Cramb, Proc. 77th Steelmaking Conf. (ISS,
Warrendale, PA., 1994) p. 371.
199. .K Tsukaguchi, S Ura, A Shiraishi, Y Hitomi T Nagahata, Steel Technol. Intl., 16, 175,
(1995/6).
200. J Yoshida, T Ohmi, M Iguchi, ISIJ Intl., 45, 1160, (2005).
201. K Watanabe, K Tsutsumi, M Suzuki, M Nakada, T Shiomi, ISIJ Intl,49, 1161, (2009).
202. K.C. Mills, P.E. Ramirez-Lopez, P.D. Lee, Sano Symposium, Tokyo, 2008, (Japan,
International Research Center for Sustainable Materials, University of Tokyo, 2008).p 158.
203. T. Funada, D.D. Joseph, J. Fluid Mechanics, 445, 263, (2001).
204. L.M. Milne-Thomson, Theoretical Hydrodynamics, Macmillan Press, London, (1968).
205. S. Chandrasekhar, Hydrodynamic and Hydromagnetic Stability, (Clarendon Press,
Oxford,,1961), p. 481.
206. P. Cha, J. Yoon, Met. Mater. Trans. B, 31B, 317, (2000).
207. S Feldbauer, I Jimbo, A Sharan, K Shimizu, W King, J Stepanek, J Harman, AW Cramb,
Proc. 78 th Steelmaking Conf., 1995, (ISS, Warrendale, PA, 1995), p. 655.
208. E Torres- Alonso, RD Morales, S Hernandez-Garcia, J Palafox- Ramos, Met. Mater. Trans.
B, 39B, 840, (2008).
209. S Diehl, JA Moore, RJ Phillips, Proc. 78th Steelmaking Conf., Nashville, TN, 1995 (ISS
Warrendale, PA, 1995), p. 351.
210. BG Thomas, LF Zhang, ISIJ Intl., 41, 1181, (2001).
References 535

211. S Chakraborty, W Hill, Proc. 77th Steelmaking Conf., 1994, (ISS, Warrendale, PA, 1994)
p. 389.
212. J Knoepke, M Hubbard, J Kelly, R Kittridge, J Lucas, Proc. 77th Steelmaking Conf., 1994,
(ISS, Warrendale, PA., 1994) p. 381.
213. T Miyake, M Morishita, H Nakata, M Kokita, ISIJ Intl., 46, 1817, (2006).
214. M Hanao, M Kawamoto, H Mizukami, K Hanazaki, Ironmaker Steelmaker, 27, (Nov.), 55,
(2000).
215. S Shima, Y Maruki, T Toh, K Tsunenari, Proc. ISS Tech. 2003 Conf. (ISS, Warrendale, PA,
2003) p. 721.
216. L Zhang, BG Thomas, Proc. 24th National Steelmaking Conf., Morelia, Mich. Mexico,
p. 138.
217. H Lei, Y Zhao, DQ Geng, ISIJ Intl., 54, 1629, (2014).
218. PV Riboud, C Gatellier, H Gaye, JN Pontoire, P Rocabois, ISIJ Intl., 36, 522, (1996).
219. H Nakato, S Takeuchi, T Fujii, T Nozaki, N Washio, Proc. 74th Steelmaking Conf.,
Washington, DC,1991, (ISS, Warrendale, PA, 1991) p. 639.
220. M Hanao, M Kawamoto, ISIJ Intl., 48, 1210, (2008).
221. P Wei, K Uemura, S Koyama, Tetsu-to- Hagane, 78, 1361, (1992).
222. QY Zhang, L Wang, XH Wang, ISIJ Intl., 46, 1421, (2006).
223. M Hanao, CAMP- ISIJ, 19, 803 (2006).
224. M Hanao, CAMP-ISIJ, 21, 822, (2008).
225. Slag Atlas, (Stahl u Eisen, VDEh, Dusseldorf, 1995).
226. P Kozakevitch, LD Lucas, Revue Metallurgie, 68a, 589, (1968).
227. E Takeuchi, JOM, 1995, (May), 42, (1995).
228. MM Wolf, Ironmaker & Steelmaker, 2000, (Jan), 22, (2000).
229. MM Wolf, Ironmaker & Steelmaker, 2000, (Feb), 65, (2000).
230. MM Wolf, Ironmaker & Steelmaker, 2000, (March), 69, (2000).
231. MM Wolf, Ironmaker & Steelmaker, 2000, (April), 58, (2000).
232. MM Wolf, Ironmaker & Steelmaker, 2000, (May), 78, (2000).
233. MM Wolf, Ironmaker & Steelmaker, 2000, (June), 22, (2000).
234. MM Wolf, Ironmaker & Steelmaker, 2000, (July), 63, (2000).
235. MM Wolf, Ironmaker & Steelmaker, 2000, (Aug.), 75, (2000).
236. MM Wolf, Ironmaker & Steelmaker, 2000, (Sept.), 90, (2000).
237. MM Wolf, Ironmaker & Steelmaker, 2000, (Oct.), 114, (2000).
238. MM Wolf, Ironmaker & Steelmaker, 2000, (Nov.), 67, (2000).
239. MM Wolf, Ironmaker & Steelmaker, 2000, (Dec.). 45, (2000).
240. S Suresh, RC Sinha, JB Singh, A Narayan, MS Sadhu, Tata Search, 2006, 291, (2006).
241. C Knocke, R Berger, W Pluschkell, Stahl u Eisen, 112, 81, (1992).
242. C Lefebvre, JP Radot, JN Pontoire, Y Roux, Revue de Metallurgie, CIT, 94, 489, (1997).
243. JJ Macho, G Hecko, B Golinmowski, M Frazee, Preprints 33rd McMaster Symp. Iron and
Steelmaking, Hamilton, Ont., 2005, (McMaster Univ. Hamilton, 2005) p. 131.
244. P Valentin, C Brucke, K Harste, J Potschke, Steel Res. Intl., 74, 139, (2003).
245. Y Tanizawa, CAMP-ISIJ, 4, 257, (1991).
246. K Yamaguchi, CAMP-ISIJ, 3, 181, (1990).
247. C Perrot, JN Pontoire, C Marchionni, MR Ridolfi, LF Sancho, Proc. 5th Europ. Conf.
Continuous Casting. Nice, 2005 (La Revue Metall., Paris, 2005) p. 36.
248. B. Debiesme, J Radot, D Coulombet, C Lefevbre, Y Roux, C Demarval, US Patent No 5,
876,482 (1999) and US Patent No 6, 328, 781 (2001).
249. Y Nakamura, T Ando, K Kurata, M Ikeda, Trans. ISIJ, 26, 1052, (1986).
250. K Mukai, Marangoni and Interfacial Phenomena in Materials Processing ed. ED Hondros,
M McLean and KC Mills, (IOM, London, 1998) p. 201.
251. K Mukai et al, Taikabutsu (Refractories), 42, 710, (1990).
252. N Tsukamoto, Taikabutsu (Refractories), 44, (5), 270, (1992).
536 11 Using Mould Fluxes to Minimise Defects and Process Problems

253. MC Bezerra, CA Valadares, IP Rocha, JP Bulota, MC Carboni, IL Scripnic, CR Santos, K

Mills, D Lever, Proc. 37th Steelmaking Seminar, Porto Allegre, RS, Brazil, 2005) (ABM,
Sao Paulo, 2005) p. 190. http//www.carboox.com/pdf/fluxante_2007.pdf.
254. Q Wang, JH Chi, B Xie, Y He, B Zhu, W Chen, Proc. Conf. Continuous Casting in
Developing Countries, Beijing, 1993 (SEAISI, 1993) p. 842.
255. F Hauck, J Potschke, Arch f. Eisenfuttenw. 53, 133, (1982).
256. AF Dick, S Zarrug, RF Pomfret, KS Coley. Proc. 4th Intl Conf. Molten slags and fluxes,
Sendai, (1992) (ISIJ, Tokyo, 1992) p. 380.
257. AI Zaitsev, AV Leites, AD Litvina, B Mogutov, Steel Research, 65, 368, (1994).
258. DE Sturgill. Proc. 79th Steelmaking Conf., Pittsbugh, PA, 1996, (ISS Warrendale, PA)
p. 265.
259. IR Lee, JW Kim, J Choi, D Kwon, YK Shin, Proc. Conf. on Continuous casting in
Developing countries, Beijing, China, (1993), p. 814.
260. M Shinmei, T Machida, Met. Trans., 4, 1996, (1973).
261. K Shimizu, T Suzuki, I Jimbo, AW Cramb, Proc. 79th Steelmaking Conf., Pittsburgh, 1996,
(ISS, Warrendale, PA, 1996) p. 723.
262. NN Viswanathan, F Shahbazian, S Du, S Seetharaman, Steel Research, 70, 53, (1999).
263. M Persson, S Seetharaman, S Seetharaman, ISIJ Intl., 47, 1711, (2007).
264. H Heimbach, K Schulz, J Markardt, HJ Ehrenberg, Stahl u Eisen, 117, 105, (1997).