DESIGN

alae
OF
THERMAL
SYSTEMS

R.RBOEHM /
 DESIGN ANALYSIS OF
THERMAL SYSTEMS

sf
eeJo

Robert F. Boehm

University of Utah

ISMN ERMOAD DUSTRIAL DE SANTANDER Ne. Clasificecian

| Convo bo Tacumeptacion y Biblingralla oat 4 Z

i.
BIBLIOTECA Ret
f-

0 4 6 | 0° 3G sition » fecha Recibo

? Ye * FaPrecia Dpto. Solicitante |

* etn . e i@ ME &Q WIA eS .

John Wiley & Sons

New York - Chichester : Brisbane - Toronto - Singapore

To Marcia, Debbie, Chris, and the Corvairs, who keep my life full
and my wallet empty.

Copyright © 1987, by John Wiley & Sons, Inc.

Alt rights reserved. Published simultaneously in Canada.

Reproduction or translation of any part of

this work beyond that permitted by Sections
107 and 108 of the 1976 United States Copyright
Act without permission of the copyright
owner is unlawful. Requests for permission
or further information should be addressed to
the Permissions Department, John Wiley & Sons,

Library of Congress Cataloging-in-Publication Data:

Boehm, R. F.

Design analysis of thermal systems.

Includes bibliographies and index.

1. Heat engineering. 2. Engineering
design. |. Title.
TJ260.B57 1987 621.402 86-32609
ISBN 0-471-83204-9

Printed in the United States of America

1098765432
 PREFACE

This book is an outgrowth of my teaching of a senior-level thermal

systems design course over the last several years at the University of Utah.
Generally, the courses in design are difficult for many people to teach, and
this one was that way for me. There were two complicating factors. First,
| started putting this course together as a fairly green assistant professor.
| had had some industrial experience with a few years at the General
Electric Atomic Power Equipment Department (later called Nuclear Energy
Division) and a number of summer engineering jobs. But it was still a
difficult transition from being a doctoral student to being a professor who
is supposed to know all things about all things.
Second, there were new currents moving in the field. Computers
were proving their value as an engineering tool in everything from
characterizing materials to guiding us to the moon. But it seemed that the
field of thermal systems was among the last to reap the benefits of this
technology. Certainly there were exceptions. Finite difference and, later,
finite element techniques were proving to be key tools in many heat
transfer analyses. Modeling of fluid flow in a variety of fields was given a
considerable boost from a number of breakthroughs during the space race
and the associated technological binge the world was experiencing. Even
the modeling of control systems seemed to move along rapidly. Power
plants, however, were still designed by-"hand.” Even today, cooling towers
continue to be designed by many companies from graphical techniques
developed before most of us can remember.
Hence, to aid my own education on thermal systems, and to see where
ihe field might go if computer-aided design techniques were applied, |
began studying this field with more than a little interest. It finally got to
the point where the file was overflowing with material, and | had to make
thie decision to do something with it. In 1984-1985 | went on sabbatical
leave and began this book.
A general concept of the course and this book developed quite
naturally. In this concept, it seemed that the book should draw heavily on
the basic analysis courses that students have in the sophomore, junior, and
senior years of their engineering curricula. Second, the text should give
insights to some of the practical aspects of design that too often are not

Vv
 Preface

found in the analysis courses Overall, the course should generally deal
with topics in a computer-aided context. Finally, the student should be
introduced to the thermal design literature so that the full power of the
various techniques can be studied in detail, if desired.
From my course work in thermal systems design, | had ideas about
what specific topics | wanted to include in the text. Of course, there
should be a discussion of the practical aspects of equipment selection. Of
particular concern were some keys to choices between various types of
available heat exchangers. For example, too many students leave school
without knowing the difference between a floating-head heat exchanger and
a U-tube heat exchanger. 1! addressed questions of this sort with
information on selection criteria and costs in a concise format. In this
regard, | appreciated Gael Ulrich (of the University of New Hampshire)
allowing me to borrow some of his excellent material.
| felt there should be some information on the old, but newly
"rediscovered," topic of availability analysis. Workers around the world
were showing the value of this concept for the analysis of complex
systems where design choices had to be made between energy in various
forms. Of course, the cogeneration example is one that illustrates the
concept of economic trade-offs between heat and work, and this is used in
the text. I think availability will become more of a generally applied tool
as we all better understand the concepts and their power in the solution of
a wide range of problems.
Flowsheeting is a topic that is given little thought in the systems
analysis performed in an introductory thermodynamics course. However, as
the system becomes larger and more complicated, the definition of an
appropriate flowsheet can be an extremely critical step in the analysis of a
system. Many of the questions involved in setting up an appropriate
flowsheet have been addressed in the development of powerful codes like
ASPEN, and this is an area where research continues. An introduction to
flowsheeting ideas is also included.
One topic that is not often covered in basic courses is costs and
their effects on the appropriate design selections. Cost data are very
difficult to encapsulate accurately, both at a given point in time and in a
few pages of a book. The approach used here is the simplified power
function form used in the chemical engineering literature. One of the
problems with this approach is that the student may view the limited data
to be both precise and all inclusive and thus not be aware of the vast
differences in costs that can result from various materials of
construction, surface finish, or all of the other aspects that can have
profound effects on costs. In spite of all of the admitted limitations, the
method presented here is chosen as a good "first-cut" approach to the
problem of trying to attach costs to components shown on a block diagram.
Even the question of which components' prices should be included is a
perplexing one. A large number of types of equipment are given in Appendix
D, but key items are almost certainly missing. In spite of this, the problem
of gathering cost informatioto
n perform preliminary design analysis with

vi
 Preface

this information should be greatly simplified compared with starting from

“scratch.”
Estimates of future costs of energy are likewise very difficult to
predict, but this topic is also important to the choices of design options.
The data given here are the best estimates of people who make the
prediction of prices in their business. Those of us who have watched
energy prices through the 1970s and 1980s know how unpredictable these
data can be. As the energy cost estimates included here become more
dated, users of the book should seek more current information.
A brief chapter on general economic analysis is included. At many
schools, these ideas are covered much more thoroughly in an engineering
economy course than is the case in this text.
Numerical analysis topics are covered in the text. Included are curve
fitting and equation solving in Chapter 4 and optimization techniques in
Chapter 9. These chapters share with all of the others the fact that
complete texts have been written on subtopics of each chapter. I have
attempted to distill the information to give a qualitative feel for
approaches represented by the various techniques. Not all techniques are
necessarily included. Some that may be very important to various
instructors undoubtedly do not appear. Hence, in these topics, as well as in
virtually all others, the instructor should supplement the information given
here to emphasize important points to him/her.
Curve fits or other correlation information to be used for the
prediction of properties are given in Appendix B. At one time or another,
we have all curve fit thermophysical properties. In the tabulations
included here, | have attempted to make available published data to
simplify this thrust.
From the initial stages of the drafting of this book, the plan was to
make it concise and qualitative in its approach to the variety of topics
covered. However, | wanted to give the student/worker insight to the rich
literature spanning both the mechanical and chemical engineering fields.
Hence, you will find that every chapter addresses topics that could be
covered in complete texts (most, in fact, are the topic of complete books
elsewhere), as well as a number of design related papers that may not have
found their way into the archival literature. On the other hand, each
references section has many more entries than most students will be
inclined to consult. Students will find that this book does not have all the
information they may want stated explicitly, but they will be able to find
it through the use of this text, if they so desire. Some may not want to be
bothered by supplementing this text with library materials, but to
understand a given topic fully this will usually have to be done.
Several items are not in the text because they have purposefully been
left out. Perhaps most obvious is the almost total lack of concepts covered
in engineering fundamentals courses. For example, do not look for
treatments of the LMTD or Effectiveness-NTU techniques of heat exchanger
analysis. They are not here. This is not to imply that there is no need for
this information in a design course, but rather it is a recognition that this

vii
 Preface

information is conveniently available elsewhere to students. In fact, |

usually review some of the basic material in this course. There are
problems in the most chapters that are of a review nature.
In my view, the most important aspect of the student experience in
design is the application of the various concepts in a project setting. The
book does not carry this aspect through to completion. Only on a
teacher-worker level can this work. Decisions may have to be made at each
step of the way regarding directions, approximations, desired results, and
soon. We may someday find thermodynamics and other courses taught by
programmed learning (through computer interaction or special text), but
the application of these learning concepts to the design of systems will be
among the last to make this transition. There is no replacing the professor
or experienced engineer in the day-to-day development of a design. This
text addresses topics that should facilitate the process, but these cannot:
replace the process.
[ want to acknowledge people who were helpful in the development of
this text, realizing that not all will be mentioned who should be. It all
started with John Wiley & Sons' willingness to take on the project, and |
want to thank them first. Engineering editors Bill Stenquist and, later,
Charity Robey were great facilitators in the overall process. Numerous
people at Sandia National Laboratories, Livermore, where | was on
sabbatical leave when the drafting began, were willing to take time from
their own to look over preliminary notes and make comments. Chuck
Hartwig, Jack Swearengen, Joe lannucci, Jim Dirks, and others are among
those in this group. Sandy Baum, an industrial engineering colleague of
mine, looked over the engineering economy section. He eliminated several
serious errors | had there. | only wish | could capture his enthusiastic
expositions on those topics. A number of students, including the brave ones
who took the ME 562 course or assisted me when the notes were
developing, are especially appreciated. | note particularly Turnhan Coban.
Most importantly, though, | acknowledge my prime proofreader, wife,
and friend, Marcia. She gave considerable amounts of both time and
encouragement thoughout the whole process. Without her, this book would
not have been completed.
My whole-career in general and this book in particular have benefited
from the guidance and examples set by several advisers and colleagues.
Included are Chang-Lin Tien, John Lienhard IV, Frank Kreith, and H. R. (Bob)
Jacobs. Thanks, guys.
A final note. Items that probably should have been included in the
text were not. Errors are present, but | do not know where they are now. |
will appreciate feedback on both aspects of the book.
May your design experiences be among your best.

January 1987 Robert F. Boehm

vill
A Note on the Production of this Text

This text, with the exception of Figures 3-10 and 6-1, was produced
entirely on an Apple Macintosh personal computer and printed on an Apple
Laserwriter. The computer was the 512 k.memory machine with dual
single-sided disk drives (affectionately referred to by some as the
"hummer"). Software applications used were MacWrite for word processing,
MacDraw for rendering the line diagrams, and Microsoft Chart for the
plotting done in Chapter 6. In all cases the application Switcher was used
to go between other applications and the word processer. Printing was done
entirely in Geneva 9, 10, and 12 point fonts with laser font substitution.
Final rendering was done on 8 1/2 by 14-in. sheets, which were used for
reproduction at Wiley. Appreciation is expressed to the University of Utah
for making this equipment available.
About the Author

Robert F. Boehm was born in Portland, Oregon in 1940 and was raised in the
state of Washington. He attended Washington State University, receiving a
B.S.M.E. degree in 1962 and an M.S.M.E. degree in 1964. He then joined the
General Electric Company, Atomic Power Equipment Department. He left GE
to pursue a Ph.D. at the University of California at Berkeley where he
received the degree in 1968. He then accepted a position at the University
of Utah where he is now Professor of Mechanical Engineering. During his
tenure at the University of Utah he has served as chairman of the Mechanical
and Industrial Engineering Department. The 1984-1985 academic year was
spent at Sandia National Laboratories, Livermore, California, where this
text was started. He is the author or coauthor of nearly 100 technical
articles, 2 other texts, and approximately 10 chapters in texts on heat
transfer and thermal systems. Research interests include experimental and
numerical heat transfer studies and analysis of thermal systems, with
particular emphasis in applications to energy conversion. He is a Fellow of
the American Society of Mechanical Engineers and serves as Technical
Editor of the ASME's Journal of Solar Energy Engineering. Dr. Boehm is a
registered professional engineer in the state of California.

xi
 CONTENTS

1 THE DESIGN ANALYSIS PROCESS

os
1.1 WHAT IS DESIGN?
1.2 CATEGORIES OF DESIGN

oonh.A
AR
1.2.1 Empirical versus Analytical
1.2.2 Nonfunctional, Functional, Satisfactory, and Optimal Designs
1.3 ELEMENTS INTHE DESIGN ANALYSIS OF THERMAL SYSTEMS
1.4 THE APPROACH OUTLINED IN THIS TEXT
1.4.1. Summary of Topics Covered Here
1.4.2 Steady-State Analysis

G WP Pp
1.4.3 Conclusion \

ak
REFERENCE
PROBLEMS

SELECTION OF FLUID FLOW EQUIPMENT 16

2.1 INTRODUCTION 16
2.2 PUMPS 17
2.3 FANS 24
2.4 COMPRESSORS 26
2.5 TURBINES AND OTHER EXPANDERS 27
2.6 STORAGE VESSELS 29
2.7 VALVES 30
2.8 PIPING 35
REFERENCES 37
PROBLEMS 38

3 HEAT EXCHANGE DESIGN OPTIONS 41

3.1 INTRODUCTION At
3.2 HEAT EXCHANGERS At
3.2.1 Overview At

xiii
 Contents

3.2.2 Shell-and-Tube Heat Exchangers 43

3.2.2.a General Comments 43
3.2.2.0 Return Bend (U-Tube) Type Shell-and-Tube Heat
Exchangers 46
3.2.2.¢ Fixed-Tubesheet Exchangers 47
3.2.2.4 Floating-Head Exchangers 48
3.2.3 Plate Heat Exchangers 49
3.2.4 Phase Change Heat Exchangers 31
3.2.5 Air-Cooled Heat Exchangers 52
3.2.6 Direct-Contact Heat Exchangers 54
3.2.6.a General Comments 54
3.2.6.0 Cooling Towers, Evaporative Coolers 56
3.2.6.c Deaerating Heaters 60
3.2.7 Other Types of Heat Exchangers 60 |
3.3 THERMAL INSULATION 61
REFERENCES 64
PROBLEMS 66

FITTING DATA AND SOLVING EQUATIONS 70

4.1 DATA FITTING 70

4.1.1 Introduction 70
4.1.2 Functions of One Independent Variable, “Exact Fit" 71
4.1.3 Least Squares Technique, "Best Fit" 74
4.1.4 Two or More Independent Variables 77
4.1.5 Interpolation 79
4.1.6 Relationships from Physical Concepts 82
4.2 SOLUTION OF ALGEBRAIC EQUATIONS 83
4.2.1 Introduction 83
4.2.2 Solution of Nonlinear Equations 83
4.2.3 Solution of Systems of Linear Algebraic Equations (Dense) 86
4.2.4 Sparse Systems of Linear Equations 92
REFERENCES 94
PROBLEMS 95

ECONOMIC EVALUATION 99

5.1 EVALUATING THE ECONOMICS OF A PRQUECT 99

5.1.1 Some Basic Ideas 99
5.1.2 Some Definitions 103
5.2 SIMPLIFIED OR PARTIAL METHODS OF ECONOMIC EVALUATION 104
5.2.1 Introduction 104
5.2.2 Return on Investment (ROI) Method 104
5.2.3 Payback Method 106
5.3 LIFE CYCLE COSTING 107

xiv
 Contents

5.4 THE EFFECTS OF INFLATION 112

5.4.1 Introduction 112
5.4.2 Analysis Principles 112
REFERENCES 114
PROBLEMS 114

PRELIMINARY COST ESTIMATION 118

6.1 INTRODUCTION 118

6.2 THE TECHNIQUE 120
6.2.1 Size Effects on Cost of Equipment 120
6.2.2 Updating Historical Data 123
6.2.3 The Factor Method 127
6.3 USING THE TABLES—EXAMPLES 129
6.4 PREDICTION OF ENERGY COSTS 131
REFERENCES 134
PROBLEMS 136

AVAILABILITY ANALYSIS 140

7.1 INTRODUCTION 140

7.2 BACKGROUND 141
7.2.1 Availability Definitions 141
7.2.2 Second Law Efficiencies 145
7.3 PROCESS APPLICATIONS 148
7.3.1 Introduction 148
7.3.2. Turbine Process 148
7.3.3 Heat Exchange Processes 150
7.3.3.a Some Basic Ideas 151
7.3.3.0 Heat Exchanger Concepts 152
7.3.3.c Forced Flow Convection 155
7.3.4 Psychrometric Processes 156
7.3.5 Combinations of Processes, Systems 158
7.4 ENERGY AND AVAILABILITY ACCOUNTING 160
REFERENCES 163
PROBLEMS 166

SYSTEM FLOWSHEETING 168

8.1 INTRODUCTION 168

8.2 THE NEED FOR A CONSISTENT NOTATION 170
8.3 MATHEMATICAL DESCRIPTION 172
8.4 TWO APPROACHES 173
8.4.1 The Sequential-Modular Approach 173

XV
 Contents

8.4.2 The Simultaneous-Solution Approach 176

8.4.3 Comparing the Sequential and Simultaneous Approaches 179
8.4.4 Applications Experience 180
8.5 EXAMPLES OF MODULAR FORMS 180
8.5.1 Pump Representation 181
8.5.2 Turbine Representation 182
8.5.3 Flow Splitter 182
8.5.4 Heat Exchanger 182
REFERENCES 184
PROBLEMS 184

9 INTRODUCTION TO OPTIMIZATION TECHNIQUES 187

9.1 INTRODUCTION 187

9.2 GENERAL PROBLEM FORMALISM AND BASIC IDEAS 188
9.3 UNCONSTRAINED FUNCTIONS OF ONE INDEPENDENT VARIABLE 192
9.3.1 Sequential or Exhaustive Search 192
9.3.2 Golden Section Search 193
9.3.3 Polynomial Approximations 195
9.4 CONSTRAINED MINIMIZATION OF FUNCTIONS OF A SINGLE VARIABLE 195
9.5 UNCONSTRAINED OPTIMIZATION OF MULTIVARIABLE FUNCTIONS — 197
9.6 MULTIVARIABLE OPTIMIZATION WITH CONSTRAINTS 203
9.6.1 Linear Programming 203
9.6.2 Nonlinear Systems with Constraints 206
9.6.2.a Introduction . 206
9.6.2.b Use of Barrier or Penalty Functions 206
9.6.2.c Techniques That Control Search Direction and Step Size 208
9.6.2.d Other Techniques 208
9.7 PRACTICAL CONSIDERATIONS IN OPTIMIZATION SOLUTIONS 208
9.8 SUMMARY COMMENTS 211,
REFERENCES 211
PROBLEMS 213

APPENDIX A: DESIGN PROJECT SUGGESTIONS 217

A.1 DESIGN OF A HEAT WHEEL REGENERATOR FOR HEAT RECOVERY 217

A.2. HEAT-DRIVEN HEAT PUMP 218 ©
A.3 WINTERICE FOR SUMMER AIR CONDITIONING 218
A4 SOLAR-DRIVEN IRRIGATION SYSTEM 219
AS POWER GENERATION HEAT RECOVERY UNITS 219
A6 RESIDENTIAL ENERGY CONSERVATION OPTIONS EVALUATION 219
A.7 STEAM-INJECTED GAS TURBINE 220
A8& PROPER AMOUNT OF PIPING INSULATION 220
AQ FLASHED-GEOTHERMAL POWER PLANT 221
A110 FLASH VERSUS BINARY GEOTHERMAL POWER PLANT 222

xvi
 Contents

A.i1 DISTRICT HEATING/COOLING EVALUATION 222

A.12 HILLTOP TO VALLEY FLOOR POWER PLANT 223
A.13. POWER PLANT DESIGN OPTIMIZATION 223
A.14 STEAM-POWERED SPEED CAR 224

APPENDIX B: PROPERTY EVALUATION 225

B.1. COMPUTER EVALUATION OF PROPERTIES 225

B.2 CORRELATIONS 225
B.2.1 Thermal Conductivity 225
B.2.2 Heat Capacity 227
B.2.3 Dynamic Viscosity 228
B.2.4 Latent Heat of Vaporization 229
B.2.5 Density 229
B.2.6 Other Properties 230
REFERENCES 230

APPENDIX C: CURVE FITTING CATALOG 231

C.1 INTRODUCTION 231

C.2 CURVE FIT FORMS FOR SINGLE-VARIABLE FUNCTIONS 231
C.2.1 The Straight Line 231
C.2.2 Hyperbola Forms 232
C.2.3 Forms Involving Exponential Functions. ° 233
C.2.4 Logarithmic Forms 234
C.2.5 Power Function Forms 235
C.2.6 Polynomial Representations 236
C.3 FUNCTIONS OF MORE THAN ONE VARIABLE 237
C.4 FURTHER INFORMATION 238

APPENDIX D: PRELIMINARY COST ESTIMATION DATA 238

D.1 INTRODUCTION 238

D.2 INDEX, DEFINITIONS, AND COST INFORMATION 238
REFERENCES 244

APPENDIX E: PERTINENT GOVERNMENT PUBLICATIONS 248

AUTHOR, SOURCE INDEX 253

SUBJECT INDEX 259

xvii
 CHAPTER 1
THE DESIGN ANALYSIS
~ PROCESS

1.1 WHAT IS DESIGN?

Much has been written about the topic of design. Design has always
been important in all engineering practice, but its relative importance as a
distinct field in engineering curricula has ebbed and flowed over the years.
Generally, the design process involves the application of concepts from
engineering science topics in a generally specified manner coupled with a
creative touch.
Successful design is a collection of several processes. First, insight
into the desired end result is necessary. This step might be called
conception. Second, the ways in which that end result might be
accomplished must be defined. The term synthesis might be applied to this
step. Finally, a significant amount of analysis is often needed to
supplement the first two steps to bring the design to reality. Analysis can
also supplement the synthesis function.
The insight noted in the first step depends very strongly on a
hard-to-define characteristic called creativity. Does a person have
creativity at birth, or is it something that can be learned? Since there are
numerous successful inventors (invention can be a form of design) who do
not have a formal education, it would seem that some people are born with
a natural ability in this area. There are varying opinions about whether or
not it can be learned. The second step, called synthesis, is obviously very
important. It requires both learned information and creative insight. A
person can come a long way in setting up a more efficient thermodynamic
system by studying the various kinds of processes possible and the factors
that influence their efficiency. From time to time, though, creative insight
will render a clear breakthrough in a given design. Finally, the third step
related to analysis is clearly something that can be learned. The analysis
function can find application in the synthesis step, causing a gray area in
the definition of these two terms.
Compare distinctions between mechanical design and thermal
process design. Like many, this division may be somewhat arbitrary, but it
is important to the thrust of this text. Some elaboration on this
comparison as well as the elements of the creative aspect in each category
 The Design: Analysis Process

are in order.
Suppose that a firm wants to manufacture and sell a new type of
mechanical can opener. Suppose further that in order to have a market
edge, the new device will have an operational advantage over existing
designs and will sell for a competitive price (this may actually be a higher
or lower price than the other openers on the market, depending upon the
perceived operational advantage of the new design). Creative work must
then occur in two areas. First, the operational advantage must be
conceived and reduced to practice. Some people think that the creative
procedures involved in this step are not, in general, easily categorized or
learned. Once a device is designed, the processes used in the manufacturing
steps must be determined. Although there are avenues to demonstrate
creative genius in these aspects, several dimensions can be developed into |
a specific technology and can be categorized and learned. If the product is
truly successful, hundreds of thousands might be manufactured and sold, so
the clever manufacturing of the device may be just as important as its
original invention.
In the example of the can opener, as well as other more involved
mechanical devices, creative aspects can be the most critical in the device
conception stage and can be of somewhat lesser importance in the other
facets of bringing the product to the consumer. The relative importance of
the upfront creativity is lessened in more complicated mechanical systems
and thermal systems where refinement of designs may make up a large
portion of the creative process. In these examples, the modeling of the
devices or systems may be extremely valuable, enabling the overall
improvement of function and performance.
Computer-aided engineering can be of great value in the solution of
mechanical design projects. Often this involves a graphical representation
of the part or device via the computer. See the left-hand side of Figure 1.1.
This can ultimately be used for assisting with the manufacture of the
device. In addition, as the design and analysis functions take place, the
graphical representations can be used for stress, electric/magnetic field,
and/or temperature analyses of the device.
On the other hand, the simulation of thermal systems often involves
the synthesis of components into an overall system. See the right-hand
side of Figure 1.1. This is a subset of the overall category of process
design. In contrast to mechanical design, the computer can be used here to
simulate processes. Both fields can share a need for optimization and
other types of numerical analysis, but there is a fundamental difference
between the two.
Consider applications to the design of thermal devices and systems.
A system may be very large and have a single application. An example of
this is a giant mine-mouth power plant in Wyoming. Alternatively, it could
be some system produced in large numbers, such as a new refrigeration unit
to be applied as an automobile air conditioner.
What are some points of contrast between most thermal systems and
the simple mechanical devices as illustrated above with the can opener

2
 The Design Analysis Process

example? In contrasting them, there is always the fear of overgener-

alizing, but certain aspects are more often true than not.

LEE TEES Y
LLL Le

Figure 1.1 Computer-aided engineering can play a vital part in design. In mechanical design
this usually involves a graphical representation of the part or assembly (left). Process
representations in the computer are important in thermal system design (right).

1. Thermal systems generally involve components that can be

categorized. For example, pumps, heat exchangers,
turbines, mixers, and other items are found in a large
number of thermal systems.

2. in fact, thermal systems often involve a large number of

components in one design.
3. The presence of a number of components means that there
is a large number (perhaps in the thousands) of parameters
that must be set, either arbitrarily or in relation to other
aspects of the design.
4. Parameters to be set in the design usually affect both
capital and operating costs, the latter being greatly
influenced by, and often totally dominated by, energy
costs.

5. The general system that is to be designed may be well

defined, but it can still involve a major design activity,
possibly including modeling, to satisfy the project
objectives.
6. Modeling of thermal systems can be systematized to a
large degree.
 The Design Analysis Process

The topic of thermal systems design owes much to W. Stoecker of the

University of Illinois. His initial treatise (Stoecker, 1971) on the topic
was a pioneering work in this area.

1.2 CATEGORIES OF DESIGN

1.2.1 Empirical versus Analytical

Thermal systems have been designed for hundreds of years. Their

development has been due, by and large, to empirical techniques. Perhaps a
blacksmith fashioned a firebox to a size that he felt would be satisfactory
for a given application. Or perhaps an engineer designed a boiler using
available tubing lengths and sheet metal sizes without much thought about
what else might be possible. The automobile engine is an excellent
example of how a fairly complex device can be developed almost entirely
from empirical means.
During the last half century, empirical means are gradually being
replaced by analytical techniques. This trend was aided in its initial
stages by the development of analytical methods in heat transfer, fluid
mechanics, chemical processes,. and thermodynamics, which, in turn,
evolved from empiricisms themselves. The analytical trend was given a
dramatic boost in the 1950s with the development of the high-speed digital
computer. More recently, the application of optimization techniques has
proved to be extremely valuable.

1.2.2 Nonfunctional, Functional, Satisfactory, and Optima!

Designs

Regardless of the techniques used for defining the design, the

synthesis process will result in either a functional or a nonfunctional
design. If a solar-driven refrigerator is designed, but it will not actually
cool, then the design is said to be nonfunctional. However, if the
refrigerator does perform in some manner as it should (i.e., it cools), the
design is termed functional. Usually this is not sufficient. While any
design is functional or nonfunctional, this distinction is usually obvious.
The engineer is concerned with a great deal more than simply a functional
design.
A satisfactory design is drawn from the class of functional designs.
The satisfactory design (there may actually be several in specific
situations) meets some kind of stipulated criteria. Usually the stipulation
is on some aspect of performance ("it must cool 6 m°/s of air from 30°C to
4°C" or "it must produce 4000 kWe"). This can give rise to a number of
designs that all meet the objective. Many times there could be cost factors
that dictate a satisfactory design ("it must cost less than $500,000") or,
as in the case of the space program, reliability factors, with cost factors

4
 The Design Analysis Process

seemingly taking a secondary role. (Actually, the manned space program

implicitly put a very high value on human life, a stipulation that led to
requirements for high reliability.)
The final, and often most desirable, category is that of the optimal
design. Here some very specific restrictions may be placed upon the final
product, and usually these are in the form of some stipulation on cost. For
example, a power unit that produces 4000 kWe may be desired, but an
optimal design that minimizes life cycle costs over an anticipated 20-year
life is also desired. There may be virtually thousands of satisfactory
designs (ones that produce 4000 kWe) but only one (or at most a few)
optimal designs.
The categories of designs can be shown on a solution space. See
Figure 1.2 for an example involving two independent design variables.

y* Rectangle represents all solutions

Nonfunctional solutions

Functional solutions

Satisfactory solutions
+ Optimal
solution

Figure 1.2 A hypothetical example showing various system solutions.

Too often, engineering practice is satisfied with simply a

satisfactory design (and sometimes just a functional one!) for a thermal
system. This has been due to a number of factors, including low energy
costs and the lack of design analysis tools.
The impact of energy costs is easily envisioned. If operating costs
are negligible, then first (capital) costs become the determining factor. In
this case, minimization of capital costs may minimize life cycle costs.
However, as energy and other operational costs rise, designs that simply
minimize first costs fall into the category of functional or, at best,
satisfactory design possibilities.
Design analysis tools have not been present historically. This has
changed with the advent of the computer. Included are both hardware and
software impacts. Both of these aspects have benefited from considerable
development over the last twenty years. Desktop computers are now more
powerful than were the largest of all the supercomputers less than fifteen
years ago. Software libraries become more extensive with each passing

5
 The Design Analysis Process

year. The coupling of the engineering science knowledge with the existing
computer software can make a powerful combination. It is to this end that
the present text is dedicated.

1.3 ELEMENTS IN THE DESIGN ANALYSIS OF THERMAL SYSTEMS

There are several steps in the design process for thermal systems.
In the discussion that follows, a system is assumed to be made up of more
than one process, so that process elements are the foundation of ‘the
synthesis procedure. As shown in Figure 1.3, one way of looking at the
general approach is to assume that it is made of three general steps.

Concept

| Process formulation |

Vv
| System synthesis —_|
d
| Optimization

Final design

Figure 1.3 A broad categorization of the three basic elements in the definition of an
optimal design of a thermal system: process formulation, system synthesis, and
optimization.

First, and by far the most important aspect, is the necessity of

having a good understanding of the physics of the various processes and
other elements that will be needed to make up the system. Without an
understanding of the basic ideas of heat transfer, thermodynamics,
chemical processes, and fluid mechanics, for example, it would be
improbable that a person would be very successful in designing a thermal
system. (More will be said about this point later in this chapter.) This will
be denoted here as process formulation.
Second, this knowledge of physics, including a great deal of
supplementary information of a more empirical nature, is also needed to
build the elements into the kind of system necessary to accomplish the
desired result. For our purposes, this will be termed system synthesis.
Finally, the system model should be exercised to yield the optimal,
or at least a nearly optimal, solution. This stage is defined here as the
optimization block.
For the most part, it is assumed that someone interested in the
design of thermal systems has already been thoroughly introduced to the
physics of process formulation and, to some degree, system synthesis. For

6
 The Design Analysis Process

example, consider the system design of a regenerated Brayton engine with a

bottoming steam cycle. It should be obvious that some knowledge is
necessary of what components constitute these kinds of cycles. For this, it
is assumed that knowledge of the LMTD and Effectiveness-NTU methods for
sizing of heat exchangers is reasonably understood also. Insight to the
general performance prediction theory of turbines, compressors, and pumps
is also necessary. .
The techniques related to the system design aspects can then be used
to formulate the optimal system. Stress will always be placed on the
optimal design. Although some systems may not have an optimal form, this
occurs very infrequently. Usually the problem is not that there is not an
optimal (or at least a "nearly optimal") solution. Instead, it may be that
the optimum condition is difficult to define fully and to find. The first of
these problems is often due to imperfect understanding of the physical
problem, whereas the second may be due to the mathematical complexities
of a solution technique.
A more accurate representation of the analysis needed to arrive at a
desired design is shown in Figure 1.4. The design concept consists of
defining what is to be done by the design and how it might be accomplished.
Not infrequently, there may be several approaches that may satisfy a given
input-to-output requirement. After analysis, one may be shown to be
preferable, but usually this insight is not known in advance.

Design concept
Design specifications
Component definition

burve
ting
method} Process deserters
Ay Spiel costs
[Cal
/ ais Life oyole costs
Optimizer functions

Costing

Optimizer
iterate
Final design

Figure 1.4 The analysis performed in pursuit of a final design of a thermal system can
involve a large number of steps. In this diagram, each block can represent many steps that
are fully defined here, but most of these are described in detail later in this text.

The design specifications step, in which the output is explicitly

defined, is especially critical. It is hard to achieve a goal that is not well

7
 The Design Analysis Process

specified. Not only is it important to note what is desired, but it is also

necessary to specify what is not wanted. A fin design for a tube that
maximizes heat transfer may also be the largest one that can be built.
Some concern about the size is probably also of value. Often, consideration
of component costs in the overall analysis will help control many of these
types of factors.
Component definition requires the application of concepts of the
physics of the various aspects of the overall system. The usual goal is to
determine the necessary mathematical relation(s) in order to define fully
each process component. While it is desired that this be done in closed
. form (i.e., based upon theoretical concepts) so that the model will cover the
range of operating parameters, in fact, it is often necessary to apply
empirical relations. Even in designs using heat exchangers, where the
performance of the latter is fairly well understood, the pressure drop and
heat transfer coefficient characteristics will undoubtedly be represented
empirically.
The flowsheet layout step is important for those design problems
where there is a large number of process components. If a system has only
one component, or at most only a very few, the formality represented by
this block on the diagram is not so critical. For complex plants with a
large number of interconnected components, this step is absolutely
necessary to allow analysis to proceed. Even in simple systems involving
only a small number of processes, it is desirable to lay out the flow
diagram to aid in the understanding of system-operating characteristics.
A large number of variations is represented by the block denoted
solution method. As with the flowsheet layout step, this block can
represent a very complex set of procedures when the system involves a
large number of processes. Typically, these procedures will involve a
step-by-step approach to cast the block diagram into a form that will yield
a mathematical solution.
At this point in the scheme laid out in Figure 1.4, the system has
been defined, the mathematical functions for all of the processes have been
found, and the general solution technique has been laid out. Now a need will
often exist for initial guesses for a large number of the problem variables
so that the solution can proceed. Obviously, some systems can be analyzed
directly from the given inputs, and this step is not necessary in these
cases. However, in the more complicated systems, this step can be very
important. In some instances, the ultimate success of the solution is
dependent upon reasonable choices of initial guesses.
The step denoted calculations is where the computer takes over.
Generally, what is represented here is a system model where variable
inputs, possibly including component sizes, some process inlet pressures,
temperatures, and so on, are then used in the calculation procedure to find
all of the system interactions. This latter category may include power
inputs and outputs, heat additions and rejections, and all other inlet and all
outlet mass flow stream definitions.
in many analyses, the solution will be carried out on a unit-mass

8
 The Design Analysis Process

basis. Certainly in these situations, but also in a number of other cases,

there is a point where the size of the components must be defined. How
many kilowatts must be produced by that generator? What is the total duty
of this heat exchanger? These kinds of questions must be answered at this
point if they have not been addressed earlier in the analysis. This function
is denoted on the diagram by the block entitled equipment sizing.
Costing is the step where a dollar value is attached to each of the
aspects of the plant. Usually, this not only will involve the assigning of
the costs to the various pieces of equipment, but it also should include the
estimation of the operating costs and all other factors that will give the
"bottom line" on the overall price of the plant. This step is always
important; and it can be extremely involved, particularly in the final design
process. In this text, only the first step ("preliminary cost estimation") is
emphasized. Capital costs are taken as the appropriate economic indicator
in this approach. While this simplification can yield misleading results in
some situations, this is generally a good first-step analysis for most
thermal systems. Operating and maintenance costs, where significantly
different for various design options, can often be considered in another
complete evaluation of the designs.
At this point on the diagram in Figure 1.4, a single system has been
defined and all of the costs evaluated. This system could be quite far from
the optimum form. Moving the solution toward the desired overall result is
the function of the optimizer.

1.4 THE APPROACH OUTLINED IN THIS TEXT

1.4.1 Summary of Topics Covered Here

In what follows, an attempt is made to summarize much of the

material that is needed to model thermal devices and systems. It is
assumed that the reader has a background in mathematics, fluid mechanics,
thermodynamics, heat transfer, chemical processes, engineering
economics, and computer programming even if this background is very
minimal. There are too many items that need to be discussed about other
aspects of steady-state system simulations to spend much time here
addressing these fundamental topics. The coverage of this book and how it
is assumed to fit with more familiar topics is depicted in Figure 1.5.
Chapter 2 covers aspects of selecting pumps, fans, compressors, and
other fluid flow equipment. While the theoretical aspects of fluid
mechanics as well as the thermodynamic analysis of devices like
compressors or turbines would have been introduced to the engineer or
student at this point, too often, practical distinctions between generic
types of equipment are not covered in introductory courses. For example,
what are the basic types of compressors; and when should one type be used
rather than another type? In many instances, “rules of thumb" are given to
help determine which general type of equipment will be most appropriate.

9
 The Design Analysis Process

These distinctions may impact some aspect of the physical analysis (e.g.,
one type of device may have a higher efficiency than another), and these
distinctions almost always have important cost considerations. Estimating
the cost of capital equipment is treated in Chapter 6, but a key to the
application of cost data is through the appropriate equipment selection
criteria. Example problems and problems at the end of the chapter can be
used to review basic methods of analysis covered in previous courses.

Physics > Fluid machinery selection

Heat transfer Thermal device selection
Fluid mechanics Second law analysis
Thermodynamics
Chemical processes .
ae x ,.Vesign
Mathematics Equation solution and ~ analysis
General math curve fitting
Calculus Process flowsheeting

lEconomics — Life cycle costing -

Discounting Cost estimating
End
Previous work Covered here result

Figure 1.5 The topics of this book (shown in the second column here) are assumed to
supplement previous knowledge to enable the reader to perform a design analysis.

Selection of heat exchange devices is the focus of Chapter 3. Of

course, this includes a general description of the various types of heat
exchangers that are commercially available, but also covered are
combinations of equipment items that are combined into subsystems, like
those used for heat rejection functions to the atmosphere. Wherever
possible, rules of thumb are given to give insight into when one type would
be used compared to another. While generalities of cost comparisons are
listed here, more specific costing information is given in Chapter 6. As is
true for fluid flow devices in Chapter 2, example problems and problems at
the end of the chapter can be used to review the methods of analysis of
heat exchange devices.
In Chapter 4, some important numerical analysis tools used often in
thermal system simulation are discussed. These include means of root
finding, curve fitting, interpolation, and solution of systems of equations.
As with virtually all of the chapters given in this book, the coverage of this
chapter could be the topics found in a complete text. The emphasis here is
on developing physical insights for application of a few techniques. A
coverage of the mathematical basis for the techniques, their absolute

10
 The Design Analysis Process

accuracy bounds, and alternative approaches will have to be sought out in

other literature. A key to additional information is given at the end of the
chapter.
Most design decisions are made on the basis of economics. While a
variety of techniques can be used for determining project costs, the
emphasis here is on simplified approaches. The ones addressed can be used
easily and applied widely to allow cost comparison between various
options. Topics that are often found to be included under the title
“Engineering Economics" are given in Chapter 5. Some simplified
techniques that can be used for determining the value of systems are given,
as these sometimes can give quick answers to questions related to
investments. Drawbacks to these shortcut approaches are also outlined.
Most of the emphasis in this chapter is given to the method of life cycle
cost analysis. Some comparisons to other techniques used to evaluate
project overall economic value are given also.
A very important part of performing an economic analysis of a
project is the cost of the major pieces of equipment used. This step in the
design analysis process might be called "Preliminary Cost Estimation,” and
it is covered in Chapter 6. A key to the technique is a tabulation of
pertinent cost data for a large number of equipment pieces in a variety of
sizes. In Appendix D, an extensive table of this kind of information is
given. Costs can become outdated quite quickly, and the techniques for
updating most historical data are also given in Chapter 6. References to
more detailed and accurate sources of costing information are given so that
"second cut" designs can be evaluated if desired. Also included in this
chapter are some professionally performed predictions of energy costs for
the next couple of decades. This is another key ingredient to the
determination of the overall economic viability of a project.
In Chapter 7, some concepts from applied thermodynamics are
presented. The focus here is on Second Law applications. There is a
growing emphasis on the use of the concept of thermodynamic availability
in the analysis of thermal systems. Although many readers may have
covered this topic before, the design implications are stressed here. The
value of this concept in making comparisons between work and heat in
systems, as well as evaluating maximum performance possibilities, is
emphasized.
As shown in Figure 1.4, generally the next step in the building of a
simulation model is the representation of a series of processes into a
mathematical description of the whole system. This has to be set up ina
way that analysis can follow to determine the performance of the system.
Chapter 8, "Process Flowsheeting," addresses some of the ways that this
can be done. Usually, the resulting system is one of two types: one where
the calculations can be performed in a sequential manner from the
beginning of the system to the end; or one where the calculations must be
done in a simultaneous manner, top to bottom. While the general design
process usually requires the latter approach, many times the sequential
technique can be used in an iterative way to determine the same end result.

11

I, GR? SPW, BE
 The Design Analysis Process

The latter approach is often much easier to set up for solution.

A final major topic of this text is an introduction to optimization
meihods. This is covered in Chapter 9. As is the case with each of the
chapters, the topic touched here fills volumes of other texts. Selected
aspects are included that illustrate the topic in a conceptual manner.
References are given to allow the reader to pursue other approaches of a
different type or of more complexity.
Additional material is included in the appendices of this book. A list
of project possibilities for applying the concepts given in this text is
included as Appendix A. Some information on correlations that can be used
for generating computer routines that determine thermodynamic and
thermophysical properties is given in Appendix B. References to sources of
computer routines for property evaluation are included also. Appendix C
gives a catalog of curve fitting functions to aid in the selection of
appropriate forms for specific sets of data. This material is supplement to
the information given in Chapter 4. As already mentioned, Appendix D gives
capital-cost data to be used with the ideas outlined in Chapter 6. Finally, a
list of a few (of literally thousands) government documents that give data
or show techniques pertinent to the design of thermal systems is given in
Appendix E.

1.4.2 Steady-State Analysis

One important aspect should be noted before ending this introductory

chapter. The emphasis of this text is on steady-state simulations. While
the point can certainly be made that virtually all systems have some
element of transient behavior and are constantly cycling between
off-design states in actual operation, the importance of mastering
steady-state simulations is fundamental. In spite of the presence of the
generally transient behavior of systems, the fact remains that practical
design is based on steady-state performance.
Much of what is discussed here is pertinent to the transient analysis
of systems. Often it is sufficient to equate a steady-state balance of flow
quantities (specie, mass, energy, or availability) to a time rate-of-change
of the pertinent system state quantity (specie, mass, energy, or
availability). First-order, ordinary-differential-equation systems often
result. A number of texts treat dynamic system analysis, and the reader
who is seeking information on the dynamic analysis should have no trouble
finding a great deal of background material on this topic.

1.4.3 Conclusion

In all, this text should go a long way in building the reader's

experiential base on thermal systems simulation. As in most fields, in the
design of thermal systems, there is no substitute for understanding the
analytical tools available and gaining experience in applying those tools.
Now is the time to start!

12
 The Design Analysis Process

REFERENCE

Stoecker, W., 1971, DESIGN OF THERMAL SYSTEMS, McGraw-Hill Book

Company, New York.

PROBLEMS

1.1 Consider a simple, ideal Rankine cycle operating between pressures of

0.1 and 50 atm. 100°C of superheat is present entering the turbine (i.e., if
the saturation temperature is Tap then this temperature is 100 + Tay) It
is desired to add a single stage of closed regeneration to the cycle.
Suppose the pressure at the extraction point is taken to be 20 atm.

(a) Is this a functional or a nonfunctional regeneration design?

(b) What is the thermal efficiency of this configuration?

(c) If the optimal design is taken to be the one that demonstrates

maximum cycle efficiency, find the appropriate extraction pressure.

(d) What is the cycle efficiency of the optimal design?

1.2 It is desired to heat 50,000 lbm/hr of a motor oil from 70 to 180°F

using steam condensing at 1 atm pressure. 3/4-in.-ID, 13/16-in.-OD copper
tubes 8 ft long are readily available.

(a) Specify a satisfactory design for a heat exchanger that would

accomplish this heating requirement.

(6) In practice, what stipulation(s) might be used to define an

optimal design? Speculate how you might accomplish this design.
(The optimal design is not required.)

1.3 A 400-ft-long pipe is to be run from a receiver tank on a water pump

to a valve that exhausts to atmospheric pressure. No more than a 2-psi
pressure drop is to be allowed in the line between the pump and the valve.
See Figure 1.6.

(a) Plot the maximum flow rate possible against the pipe diameter.

If a 200-gpm flow is desired, show on your plot the following:

(b) the functional design(s);

(c) the nonfunctional design(s);

13
 The Design Analysis Process

(d) the satisfactory design(s); and

(e) the optimal design(s).

(f) If one, or more, of the above categories does not have an answer
you can show, speculate what kinds of information might be given to
enable you to find the answer.

Purr
L=400 ft
Receiver 4P=2 psimax valve
Tank

Figure 1.6 Sketch of system discussed in Problem 1.3.

1.4 In a system related to that noted in Problem 1.3, water is pumped

through 400 ft of one diameter of pipe to a point where 200 gpm exits
through a valve. The pipeline, now constructed of a different diameter,
extends from that point another 400 ft to a second valve where a second
200 gpm is to exit at the end of the line. See Figure 1.7. If a total
maximum pressure drop of 10 psi from the pump to the second valve can be
tolerated, specify some possible diameters for:

(a) the functional design(s); and

({b) the satisfactory design(s).

(c) Assuming that the cost of the pipeline is directly proportional

to the diameter of the pipe used and that it is desired to have the
least cost pipeline, specify the optimal solution (i.e., give the
appropriate pipe diameters to yield a 10-psi pressure drop).

Pump }—_—_—____— AP=10 psi max 8

aC 4 a, OSs.
«— 1-400 ft —e] A140 t—>| 200 gpm
|\2oo gom
Figure 1.7 Diagram for system described in Problem 1.4.

1.5 Double-glazed windows are a common product used to decrease heat

loss through fenestrations. A particular manufacturer's products have
1/4-in. spacing between panes. Neglecting heat loss through the frames, is
this a good design assuming preference is given to minimizing heat loss?
Support your conclusion with calculations.

14
 The Design Analysis Process

4.6 Consider a horizontal, stainless steel pipe line of length L. The pipe
inside and outside diameters are D, and D,, respectively. Assuming
steam is condensing in the pipe at temperature T,,, due to heat loss to the
ambient at temperature T,, discuss the following points.

(a) What calculations would be needed to determine a functional

solution for decreasing the heat loss from the pipe? (Discuss and
illustrate with equations as appropriate.)

(b) What calculations would be needed to determine a satisfactory

solution for decreasing the heat loss from the pipe? [Again discuss
and illustrate with equations as appropriate. Specifically note
items that are different in this part compared with part (a).]

(c) Give a reasonable statement of the optimal solution for

decreasing. of heat loss from the pipe. Indicate what additional
information and calculations would be necessary to carry out this
portion of the solution.

1.7 A developer wishes to break the speed record for steam-powered

automobiles.

(a) Define several of the general considerations you should

stipulate in the design to yield a satisfactory solution. This should
include not only the end result(s) desired but also the technical
requirements for accomplishing the end result(s).

(b) is there an optimal solution here? What factors might be

involved in defining a solution of this type?

1.8 Consider combustion of methane, CH,, with air, taking place at

standard atmospheric pressure in a chamber. Define the amount of air
required for each of the following situations. In each case, assume that it
is desired to achieve a temperature leaving the chamber that is greater than
90% of the stoichiometric adiabatic flame temperature (in absolute degrees).

(a) The functional solution(s).

(b) The satisfactory solution(s).

(c) The optimal solution(s).

15
 CHAPTER 2
SELECTION OF
FLUID FLOW EQUIPMENT

2.1 INTRODUCTION

The beginning designer can be faced with a very confusing array of

equipment choices in formulating a system design. This problem may be
lessened by the engineering practices in many companies. In these
practices, the novice is assigned to a very small portion of a well-
established design line and is asked to focus on small modifications of a
particular kind. The inexperienced person thus actually serves a type of an
apprenticeship. This approach tends to ensure that the modified overall
design will not function too differently from the original. This is good in
that the old design presumably functioned to some degree in a desirable
manner. However, aspects of the device or system may be able to function
much better if the whole ts given a critical examination overall.
Sources of practical information about choosing equipment and the
corresponding standard engineering practice in industry are, too often, not
in the hands of the newly graduated engineer. Textbooks usually minimize
this type of information in favor of a focus on theoretical topics. Other
than whatever information may have been gathered in the given industrial
organization, perhaps both in the senior engineers’ experience and in formal
company documents, the easily accessible repositories for this important
knowledge are often difficult to find. More typically, available specialized
texts contain this type of information. Most frequently, however, the
information will surface in trade publications within a given field. Thus,
any engineer may find the frequent reading of these kinds of publications to
be of great value, and these publications can be of even more value to the
novice.
A magazine such as Heating, Ventilating, and Air Conditioning can be
of great value to people in the HVAC industry, while a magazine like Power
often has articles of great practical interest to engineers in the electrical
generation industry. Chemical Engineering is a similar type of publication
that focuses on the process industry. Numerous other publications,
including the magazines put out by the various technical societies like the
American Institute of Chemical Engineers (AIChE), American Society of
Mechanical Engineers (ASME), and American Society of Heating,

16
 Selection of Fluid Flow Equipment

Refrigerating, and Air Conditioning Engineers (ASHRAE) are available.

(Also, each of the technical societies usually publishes one or more
archival journals. Examples include the ASME's Journal of Solar Energy
Engineering and the AIChE Journal. These publications tend to focus on
more fundamental information, but they often contain practical design
information for the beginning engineer also.) While each of these
publications has a very distinct focus, engineers who design thermal
systems in any industry may find much of value in all of these types of
publications.
To illustrate in a small way the kinds of information found in these
publications, some insights about the selection of equipment are extracted
from a variety of typical publications and are summarized here and in
Chapter 3.
Some considerations that go into the correct choice of fluid flow
equipment are discussed in this chapter. Included here are factors that
should be contemplated when specifying pumps, fans, compressors,
turbines and other expanders, storage, valves, and piping. Proper choices
between types of devices in the first five of these categories are critical
in preliminary design studies of systems that handle fluids. On the other
hand, a need for any kind of detailed specification of valves and piping is
normally not present at the preliminary design stage. Information on these
latter two categories of devices is included to provide background for
people who are unfamiliar with these topics.
As with virtually all sections of this book, the background material
on the topics touched upon fills volumes. Hence, the descriptions given
here should be anticipated to be quite abbreviated. References to more
information are given at the end.

2.2 PUMPS

The number of different types of pumps available is very large. Each

one, in turn, has a number of applications that it can perform. Often, there
may be an optimal type of pump for a given situation. At other times,
several cost-effective solutions may exist. For the beginning system
designer, an introduction to the general categories is critically needed.
Traditionally, pumps have been classified into three general groups:
reciprocating, centrifugal, and rotary. Over the years, a large number of
new developments have taken place, and these distinctions have been
blurred. As a result, some have chosen to define two general categories of
pumps: kinetic energy forms and devices that operate on positive
displacement concepts.
In what follows, pumps will be described in terms of the categories
of centrifugal and positive displacement. Qualitative explanations of
performance and costs will be given. Many aspects relating to the full
description of the pumps, like actual performance curves, drive options,
and materials of construction, will not be included in this discussion.

17
 Selection of Fluid Flow Equipment

Much more complete descriptions of pump distinctions and

performance can be found in the literature. See, for example, the special
report given in Power magazine (O'Keefe, 1972). This excellent summary
gives examples of many of the pumps in existence today.
To illustrate a few of the differences between reciprocating,
centrifugal, and rotary pumps, consider Figure 2.1. Here the general
concepts of operation are shown. In the rotating form, elements
continually contain volumes of fluid and physically "push" those volumes
through the outlet. Thus, the rotary form is similar to the reciprocating
type of pump. The latter type uses the back-and-forth movement of an
elemeni to push the pumped fluid through the ouilet valve. An inlet valve
opens at appropriate points during the movement of the back-and-forth
element to allow lower pressure fluid to flow into the chamber. Both the
rotary and reciprocating pumps are positive displacement devices. Finally, .
the centrifugal type of pump accomplishes its pressure boost by imparting
kinetic energy to the fluid. In the situation shown in Figure 2.1, the lower
pressure fluid enters the pump assembly from the center of the rotor. It
flows in a generally radial direction, picking up velocity. As the fluid exits
the pump, a pressure rise due to centrifugal force is given. The
characteristics of the pump types just described could easily lead to the
definition of two categories of pumps: positive displacement (both the
rotary and reciprocating devices described here) and the centrifugal.

Rotary Reciprocating Centrifugal

Figure 2.1 One possible way of classifying pump types is shown.

Although there are many differences between pumps of various

types, including the materials of construction, which can have a significant
effect on the pump cost, the more important ones for the person performing
preliminary designs involve the pressure head range possible from the
pump, the fluid flow capacity range of the pump, and the functional
relationship between these two variables. Each of the thousands of types
of pumps that are available has different kinds of characteristics, so it is
impossible to give any specific summary here. However, keep in mind the
genera! differences between the positive displacement and the centrifugal
types. Although exceptions can certainly be found, some typical variations
are shown in Figure 2.2. Positive displacement types of pumps usually

18
 Selection of Fluid Flow Equipment

show a very steep characteristic head-versus-capacity curve, while

centrifugal pumps usually demonstrate a flatter curve. One exception is
the diaphragm pump type, which operates on a positive displacement
principle. In this device, a flexible diaphragm is moved back and forth,
either by some mechanically reciprocating drive or by the admission and
venting of a gas. This type of pump can have an output curve more similar
to the centrifugal than to the positive displacement types. :

| Centrifugal
Pressure
rise

Reciprocating

| | | Jt] |
Flow capacity

Figure 2.2 A generalization of the pressure-head-vs.-flow-capacity characteristics is

shown for pumps of the centrifugal and reciprocating types. The latter tend to be much
steeper, although exceptions can be found for both the variations shown here.

Gross simplications are normally made when comparing the

performance of the two types of pumps. (i) If high heads are needed,
usually a reciprocating pump will be necessary. (ii) If high flows are
needed, usually the design will require a centrifugal-type pump.
Centrifugal pumps can normally be valved-out (flow cut to zero), but such
is not considered good practice with a reciprocating type. In the -
applications of the latter, a bypass is normally incorporated if zero flows
are anticipated through the piping circuit.
Efficiency
Pressure
rise

Constant tip speed characteristics

{ l | | | | I |
Flow capacity
Figure 2.3. A typical pressure-head-vs.-flow-capacity curve for a centrifugal pump
operating at constant tip speed. The efficiency characteristics of the pump are
superimposed on the pressure head curve.

19
 Selection of Fluid Flow Equipment

Centrifugal pumps find the greatest number of applications in most

plant designs. For this reason, it is of value to indicate typical
characteristics for these devices. Figure 2.3 is a qualitative plot of
centrifugal pump head characteristics. In addition, another critical piece
of information is shown--the pump efficiency. This is obviously important
in determining the power required to perform a given pumping process.

!
, , Constant
rf 1» efficiency lines _]
Pressure
rise

Constant tip speed characteristics

{ | | | | | L |
Flow capacity
Figure 2.4 A schematic of the performance curves for a family of centrifugal pumps. The
Solid lines denote the pump head characteristics with flow at various impeller diameters.
Lines that are dashed show the map of constant efficiency. All lines on the curve are
assumed to be at the same impeller tip speed.

Figure 2.4 shows related curves to those in Figure 2.3. In Figure 2.4,
the performance for several impeller diameters is shown. Dashed lines of
constant pump efficiency are shown superimposed. Information of this
type is highly desirable in any simulation of a system requiring detailed
information on the performance of a centrifugal pump.
General comparisons of pumps may be of value beyond the high-flow,
low-head (centrifugal) and vice versa (positive displacement) observation
made earlier. Typically, the centrifugal is favored where initial cost is
very important, or where fluids carrying solids that can erode internal
valves are pumped. The centrifugal also demonstrates a fairly uniform
output pressure, while a positive displacement pump may give a fluctuating
output pressure in tempo wiih the cadence of the rotors, piston, or
whatever the internal mechanism.
Another important characteristic is the net positive suction head
(NPSH) required for a given pump and the available NPSH in the piping
circuit. Although it is critical that all actual installations operate with
the available amount of NPSH greater than that needed, this aspect is
usually not of critical concern in the preliminary design stage. Note: one
often encounters "condensate" or "hot-water" pumps. Normally, these are
pumps that have a very low NPSH requirement.
Another type of pump used for special service is the jet pump. This
device, and its close relative the ejector, which is used for moving gases,
works without moving parts. See Figure 2.5. A high-pressure fluid is used

20
 Selection of Fluid Flow Equipment

to move a lower pressure fluid by injecting the former through a combined

nozzle/converging-diverging assembly. Note that the high pressure of the
propelling stream must be generated by some source, and this will
undoubtedly require a separate pump of another type. The jet pump can be
used to move a low-pressure fluid, which may be very difficult to handle.
Devices of this sort have also been used in locations where high reliability
ig needed. One example of the latter is an application for moving the core
coolant totally within the containment vessel of a nuclear reactor.

| Low-pressure fluid
High-
pressure
fluid: —
—~— —_—_
Combined flow

Figure 2.5 Schematic diagram of a jet pump. An ejector is a similar device used for '
moving gases or vapors.
:

When designing systems where fluid movement must be 3

3
accomplished, devices that might not be thought of as "pumps" may have
application. Included here is the use of gas "liffs” where a gas
simultaneously flows through a vertical column of liquid and moves the
liquid. Another approach is to pressurize a tank holding the liquid with an
overlaying gas.
A summary of selection criteria for pumps is givenin Table 2.1. This
table was abridged from Ulrich (1984). Information is given there not only
on the flow/pressure rise ranges, but also on applicable temperature and
viscosity ranges. See the original source for similar information on a
greater number of pump types and additional criteria on materials of
construction.
Further information on the description and applications of pumps can
be found in a number of sources. (See, e.g., Hicks and Edwards, 1971;
O'Keefe, 1972; Walker, 1972; Neerken, 1974; Karassik et al., 1976; Peters
and Timmerhaus, 1980; Pollak, 1980; Stewart and Philbin, 1984; Ulrich,
1984; Warring, 1984a,b.)

EXAMPLE 2.1

Specify an appropriate type of pump driven by an electric motor and

calculate the power required to pump 8 gom of water through a pressure
rise of 700 psi at ambient temperature. Although specific costs are not to
be addressed, choose a lower cost option if more than one type is available.
Assume the electric motor and other driving gear have an overall efficiency
(1) OF 90%.

21
 Selection of Fluid Flow Equipment

Table 2.1
Data for the Preliminary Selection of Pumps@

< Centrifugal > < Rotary > < Reciprocating> <Jet>

Axial External Sliding

Flow Radial Turbine Gear Screw5 Vane Piston Diaphragm Jet

Max P (bars) 350 350 50 350 350 350 1000 350 350

Min TC) -240 -240 -30 -30 -30 -30 -30 -30 -240

Max T (°C) 500 500 250 400 370 270 370 270 500

AP/stage(bar) 2 20 35 200 20 150 1500 70 1

AP total (bar) 200

Max V (m/s) 5 10 1 0.1 0.1 0.1 0.03 0.006 1

Min pu (Pars) 0 0 0.0014 0.001 0.001 0.001 0.001

Max p (Pa's) 0.2 0.1 400 1000 100 400 100

Efficiency (%) 50-85© 50-85 20-40P 40-85 40-70 40-85 60-90 40-70 5-20
Relative costs
Purchase Low Low Mod Mod High Mod High Mod Low
Installation Low Low Mod Low Low Low High Mod Low
Maintenance Low Low Mod Low Mod . Low High Mod Low

Service compatibility”
Cavitation D E B B B B B B A
Corrosive c Cc c Cc Cc Cc Cc Cc A
Dry operation E E E D D E E B A
High flows A A E D D D D E E
High pressures X Cc B B B B A B x
High temperatures C Cc D Cc Cc Cc Cc Cc A
Low flows x D A A A A B A B
Variable flows A A B Cc c Cc Cc Cc A
Variable AP E D D Cc Cc Cc Cc Cc D

Potential problems®
Pulsations A A A B A B c,D A
Noise A A A A B B D B B
Reversibility
of flow x Xx xX Cc x x Cc Cc xX
Overpressure
protection A A D Cc Cc Cc Cc Cc A

Other
Disadvantages d,e d,e d,e d,e d-f 9g
Advantages h h

Footnotes: a. Modified from Ulrich (1984). Used with permission. b. Independent of viscosity up to
0.05 Pa‘s. c. Key: A=excellent, B=modest limitations, C=special units available at higher cost to
minimize problems, D=limited in this regard, E=severely limited in this regard, X=unacceptable. d.
Motor gear reducers are often necessary. e. Pressure relief protection necessary. f. Diaphragm
failure should be anticipated. g. Process fluid may be contaminated by motive fluid. h. Operated
conveniently with steam or compressed air.

22
 Selection of Fluid Flow Equipment

To solve this, first consider the various types of pumps listed in

Table 2.1. Since nothing specific is noted in the example statement here
about service, assume no special consideration is required. Next, note that
a AP of 700 psi is given. Converting this to SI, the pressure rise at 48.23
bar (4823 kPa) is too large for some of the pumps shown. While it may be
possible to install more than one pump in series and achieve the necessary
pressure differential, often this complicates the installation and
maintenance situation. Hence, eliminate all pump categories that do not
have a high enough single-stage pressure increase for the given situation.
This leaves the following candidates: external gear, sliding vane,
reciprocating piston, and reciprocating diaphragm. C
Next, consider the cost aspect. The reciprocating piston type is
preferred at very high AP's, but at this moderate value this type pump may) =
have a premium first cost. This is demonstrated by the "high" notation in
the purchase cost category. All of the other three have "moderate"
purchase costs. Since the diaphragm pump has a moderate maintenance
cost and the other two are "low," eliminate the diaphragm model.

a
In
checking the remaining two--the external gear and the sliding vane--it is
difficult to eliminate one relative to another. For this reason, either will
suffice for this exercise. Arbitrarily choosing an efficiency value from the
range given there (both types have 40%-85% given), assume that Np* 70%.
Now the power required can be calculated. f ge
From basic thermodynamics considerations, the power can be beer
'
calculated for a steady-state, steady-flow adiabatic process from the |
mass flow rate and an integral involving the pressure and the specific uo
volume. Assuming further that the specific volume does not change ©
significantly with pressure gives (Vis the volume flow rate)

W =m (-[v dP) =-mv AP =-m (AP)/p=-VAP

Incorporating both the pump and motor efficiencies will allow the
electrical power required to be found:

W = VAP/(n, Ny)

Substituting the appropriate numbers yields the following result:

-(8 gal/min)(0.003785 m°/gal)(700 psi)(6.894 kPa/psi)

W = ; = - 3.87 kW
( 60 min/s)(0.7)(0.9)

23
 Selection of Fluid Flow Equipment

2.3 FANS

Fans share some general characteristics with pumps. Three

categories of fans are normally defined. These are axial, propeller, and
centrifugal. Applications are found in numerous locations in an industrial
plant. Included are forced- or induced-draft cooling towers, blowers,
ventilators, air-conditioning fans, and other places where there is need to
move a gas, particularly air. Factors normally considered in applications
are volume handled versus pressure rise across the fan and the ability to
conirol the flow rate.
Axial and propeller fans are closely related in that both move air by
the angle of attack of the rotating blade. In the axial fan, the housing
around the blade has an important interaction with the gas, particularly in
control functions. In the propeller-type fan, the housing plays little or no
role in controlling the flow. Hence, propeller fans are not desirable for
control purposes. As in the centrifugal pump, the centrifugal fan imparts a
significant amount of kinetic energy to the gas during the latter's outward
flow through the fan. The centrifugal force acting at the perimeter of the
fan causes a pressure rise as the gas exits.

Backward inclined
/ (Shown at right)

( Backward curved

(- Airfoil Rotation
Single \/ direction
blade
— Aadial iN

Radial tip or
) Forward curved

\/ | Forward curved

Figure 2.6 Examples of some blade profiles found on fans.

Specific fan characteristics are, of course, important in determining

final designs. For preliminary designs, they are less critical. To-touch on
just a few aspects of fan specification, consider the examples shown in
Figure 2.6. Typically, the comments below apply to centrifugal fans, but
similar considerations are found in other types also. Backward-inclined
fans are generally less expensive than some of the other types available.
They do, however, demonstrate an instability over a range of flow rates.
This type of fan has the characteristic that it can never be overloaded
because of increases in load. The backward-curved fan shows similar
characteristics, but the instability region is less pronounced. Both of

24
 Selection of Fluid Flow Equipment

these fans are used at high speed so that service in streams with
suspended particles is not recommended due to possible erosion.

Table 2.2
Data for the Preliminary Selection of Fans@

< Centrifugal > < Axial >

Radial Backward Curved

(paddle wheel) (squirrel cage) Tube Vane

Absolute P (gauge, bars) Near 14 Near 1 Near 1 Near 1

APistage 15 kPa 10 kPa 1kPa 5 kPa

Max V (std m9/s) 300 500 300 300

Cificiency range (%) 65-70 75-80 60-65 60-70

Relative costs
Purchase Moderate Low Low Moderate
Installation Moderate Moderate Low Low
Maintenance Low Low Low Low

Compatibility
Corrosive gases Cc Cc Cc Cc
High-temperature gases Cc Cc D D
Abrasive gases (particles) A Cc D D
Vacuum service x Xx x Xx
Variable flows A A A A
Variable pressures E E E E

Performance problems?
Lubrication contamination A A A A
Flow pulsations A A A A
Noise B B D D
Vibration B B A A
Explosion hazards B B B B

Other Flow direction reverses easily

Footnotes: a. Modified from Ulrich (1984). Used with permission. b. Key: A-=excellent, B=modest
limitations, C=special units available at higher cost to minimize problems, D=limited in this regard,
E=severely limited in this regard, X=unacceptable. :

Use of the airfoil type of blades is a way of increasing fan

efficiency. This comes with an increased cost compared to that of flat-
blade devices. Usually, airfoil blades are applied to clean gas streams to
minimize blade erosion.
Radial blades usually form one of the least cost fan types. Blade
speed is slow, and the efficiency is low. High particulate loadings in the
gas stream is a situation where this fan is often used. Blades are most
easily replaced on open configurations (no shroud around the blades), and
these types are often found in highly erosive environments. Radial-tip
blades may be a more desirable configuration for erosive environments as

25
 Selection of Fluid Flow Equipment

operating costs are lower than the straight-blade form. The forward-
curved blades produce large volume flows for a given fan size. They are not
recommended for potentially erosive flows.
Fan performance characteristics are a function of the type of fan
used. Many types of fans yield characteristics that are not too different
than those shown for pumps in Figure 2.4. In applications, it is desirable to
consider control factors as well as whether or not the fan is self-limiting
in case of an unanticipated addition of load resistance.
Some criteria that should be considered when choosing fans are given
in Table 2.2. An excellent summary of most aspects that should be
considered in the application of many of the fan types available is found in
the literature (B&W, 1972; Singer, 1981; Reason, 1983; Thompson ana
Trickler, 1984).

24 COMPRESSORS

Compressors overlap with both pumps and fans in terms of physical

characteristics. Typically, compressors are taken to be devices that
perform a pressure increase process on gases where the total pressure
change is larger than that found from fans. The latter devices are normally
meant to cause the gas to flow without sizeable pressure increases.
Partially like some aspects of pumps and some aspects of fans,
compressors are found in reciprocating, centrifugal, and axial flow types.
Since the general characteristics of these types of operation have been
discussed above, they will not be repeated. Much of what does appear here
is adapted from Dimoplon (1979).

Inlet flow, cfm

10 102 10% 10% 10° 10°
TTT 10°
& 10°F
“< 5
_ Centritugal | 4 &
EG, << 10 Q

oc _ 4 SO o
Q i \ +10" &
2 { \ w
S10 \ S
®oS { ag 10" ©
8 10° L 2 2
% Reciprocating Axial flow ~| 1 3
“10 l | | 1 S
2
10° 10 1 10 «6°10
Inlet flow, m¥s

figure 2.7 Appproximate ranges of reciprocating, centrifugal, and axial flow compressors.
Adapted from Dimoplon (1979).

26
 Selection of Fluid Flow Equipment

The ranges of applicability of the various types are important.

Dimopion (1979) has given a set of ranges for these types of compressors.
See Figure 2.7.
As is the case with pumps, the reciprocating types of compressors
are favored for lower flow, higher pressure service compared to other
types of compressors. Usually, higher pressure service is accomplished in
stages, with maximum pressure ratios of 3 or 4 found in each stage, and
with a maximum pressure rise in each stage limited to about 1000 psi
(6900 kPa). Overall adiabatic efficiencies for these devices can range from
60% to 80%, but they are typically in the range of 75% to 80%.
Centrifugal compressors are found in a variety of applications. The
reasons for this involve both their operational characteristics and ease of

|
repair. Percent polytropic efficiencies of centrifugal compressors are
usually in the 70% to 80% range.
Axial flow compressors are found in a more limited range of output

a.
conditions, both in terms of flow rate and pressure rise, than are other
types of compressors. Axial flow types overlap somewhat with centrifugal
compressors in their overall operational ranges. These devices are favored
over the centrifugal types when higher operating efficiencies are desired.
Efficiencies of axial flow compressors are in the range of 80% to 85%. f
Selection data is given in Table 2.3, and additional information on
compressors is available in the literature. (See, e.g., Gulf, 1979: Bloch et
al., 1982; Brown, 1986.)

2.5 TURBINES AND OTHER EXPANDERS

There are many situations where turbines and other expanders may
be required in a design application. Some of these cases obviously include
power plant design. In addition, however, there may be occasions where a
pressure drop is taken in some other type of device. You as the engineer
have the option of designing in a totally irreversible pressure drop (i.e.,
across a valve) or going for a partially reversible pressure drop through
some type of work-producing expander.
Turbine selection for steam power plants is a highly refined
business. In practice, most of this work is performed in conjunction with a
power plant equipment vendor, primarily General Electric and Westinghouse
in this country. Company representatives have presented papers on various
aspects of design considerations over the last several decades. Reprints of
this information are available from the vendors. See, for example, Bailey
et al. (1967), which has been reprinted by General Electric.
Not so prevalent, compared with information on steam turbines, is
application on turbines for other fluids and other types of expanders.
Limited information on general expanders has appeared in the literature
over the last two decades. One example is ihe text by Bloch et al. (1982).
Use Table 2.4 to assist you in making selections between types. Note that
the term turndown ratio used in that table refers to the fraction of full

27
 Selection of Fluid Flow Equipment

flow rate that can still operate the turbine or expander.

Table 2.3
Data for the Preliminary Selection of Compressors4

<Centrifugal> < Axial > < Rotary > <Recip> <Other>

Twin- Sliding
Lobe Screw Vane Piston Ejector
Single& Single& Single& Single& Single&
Single Staged Single Staged Staged Staged Staged Staged Staged

Prange (atm) 1-2 1-700 1-2 1-14 3-2 1-10 1-10 .01-3E3 01-5

Max Stage Po/P; 1.4 1.2 b 1.4 2.0 4.0 4.0 4.0 -

Max Stages 8 15 1 1 1 8 5

Max V (std m/s) 80 200 300 20 15 0.8 1.5 -

Efficiency, % 70-80 70-80 80-85 80-85 60-80 60-80 60-80 60-80 25-30
50-70© 50-70© 50-70% 50-70© 40-60 40-60% 40-60°

Relative costs
Purchase Mod Mod Mod — High Mod Mod Mod High Very low
Installation Mod Mod Mod High Mod Mod Mod High Low
Maintenance Low Low Mod Mod Mod Mod Mod High Very low

Compatibility
Corrosive gases C Cc E E E D E D A
High-7 gases D D D D D D D E A
Abrasive gases c Cc E E E D E X A
Vacuum service CC c Cc Cc B B B A A
Variable Q's Cc Cc E E Cc c Cc D A
Variable P's D D E E A A A A A

Problems
Lube contamination A A A A Cc Cc Cc c D
Flow pulsations A A A A Cc B B Cc Cc
Noise D D D D B Cc B D B
Vibration D D D D A A A B A
Explosion hazards D D D D B B B E A

Footnotes: a. Modified from Ulrich (1984). Used with permission. b. Seldom used without staging. c.
Vacuum operation. d. Key: A=excellent, B=modest limitations, C=special units available at higher cost
to minimize problems, D=limited in this regard, E=severely limited in this regard, X=unacceptable.

Approximate efficiencies for all types of process expanders (e.g.,

steam turbines and gas and liquid expanders) is given as a function of
output power (P,,, in kW) in Equation 2.1. This equation is an approximate
curve fit of a recommended plot given by Ulrich (1984).

Ne = 0.45 ( P,.,/
out
100 )2-128 100 kW <P,,,<10,000kW (2.1)
Use this relationship for preliminary prediction of the efficiency of a given
turbine or other expander.

28
 Selection of Fluid Flow Equipment

Table 2.4

Data for the Preliminary Selection of Turbines and Expanders?

< Drives > < , Power Recovery Machines >

Steam Gas Expansion Turbines Liquid

Turbines Air Radial
(noncondensing) Expanders Axial Radial Expanders

Maximum capacity, Py 4 (kW) 15,000 - 5000 1000 1000

Normal feed temperature (°C) 400 25 <500 <550 25

Normal feed pressure (bar) 45 4 <175 <175 b

Normal exhaust temperature (°C) 150 100 Various 25

Normal exhaust pressure (bar) 4.5 11 Various Various 1

Turndown ratio . 0.6 0.5 0.8 0.7 0.6

Max liquid in discharge (%) <20 <20 <20

Efficiency (%) c c 75-88 50-60

Compatibility”
Corrosive fluids D A D B Cc
Explosive atmospheres B A B B A

Footnotes: a. Abridged from Ulrich (1984). Used with permission. b. Liquid expanders can tolerate
any pressure that can be contained by a centrifugal pump. c. See Equation 2.1. d. Key: A=excellent,
B=modest limitations, C=higher cost units available to minimize this problem, D=limited in this regard.

2.6 STORAGE VESSELS

Storage vessels are often incorporated into thermal systems to

buffer the flow mass or energy of one stream relative to the flow of
another. For example, coal may be delivered to a plant once a day, but the
coal might be needed on a regular basis throughout the day. The normal
solution is to incorporate storage between the train and the plant. This
particular storage may simply be a pile of coal in the yard, sometimes
denoted by the term open yard storage. We consider a solid here to be a
special case of fluids.
Another common type of storage is the household water heater.
Whether this is gas or electrically heated, there is still a need to allow the
domestic hot water sufficient time to become fully heated as it travels
from the water supply line to the end use. This is needed because there
may be times when the utilization of energy via the hot water occurs at a
faster rate than the addition of energy via the burner or electrical heater
 Selection of Fluid Flow Equipment

element, or vice versa. Keep in mind that storage can be used for mass or
energy accumulation purposes.

Floating roof a |)

Le
pl be oP
Fixed (conical) roof Gas holder Bin
Cylindrical (bullet) tank Q

Open yard (pile) rl. Spherical

Figure 2.8 Schematic diagrams of common types of storage vessels and forms.

Figure 2.8 shows a variety of types of storage vessels as well as the

open yard (pile) type of configuration. Table 2.5 presents a brief summary
of factors that may be of value to consider when designing with storage.
Of course, this figure and table do not cover all possible types of storage
methods and containers, but they do give a fair idea about some of the more
common applications.

2.7 VALVES

Valve selection is not normally thought to be of much concern in

preliminary design studies, and usually this is indeed the case. On the
other hand, valves are often very important in the final design of a system.
Although each valve has a small cost compared to mosi system costs, there
may be thousands of valves in a given design; and thus their costs are
important. The appropriate choice of a valve could have far-reaching
implications about the control of a system. Another point to note is that
the choice of a valve type could have important implications about the type
of load applied in a pump or compressor circuit.
As has been pointed out by Miller (1983), the use of valves dates to
2700 BC, when designs of leather and wood were used in a “flap"
configuration on the island of Crete. A form of a check valve was used by
the Romans fifteen centuries before the birth of Christ. Bronze valves of
the plug cock variety were used by the Greeks in the construction of
fountains at least as far back as 400 BC. Valves have played a very
important role in more recent industrialization, including the development
of the steam engine. It is imperative that any system designer has at least

30
 Selection of Fluid Flow Equipment

a basic understanding of the general kinds of valves available and their

corresponding application trade-offs.

Table 2.5
Data for the Preliminary Selection of Storage Vessels?

< Atmospheric Storage > «Pressurized Storage>

Fixed Open Cylindrical

(conical) Floating Gas Yard (bullet) = Spherical
Roof Roof Holder Bin (pile) Tank Tank

Typical Maximum Size Cd

Volume (m°) 100,000 100,000 20,000 4,000 200,000 1600 15,000 be
Height or length (m) 15 15 30 50 50 20 30 —_
Diameter or width {m) 90 90 30 10 120 10 30 =
Length/diameter <2 <2 1-2 2-5 0.4 2-5 1 be

Stored Medium
Solid x X
Liquid Xx x xX Xx
Gas x x Xx

Orientation
Axis vertical x x x x x xX -
Axis horizontal x -

Maximum pressure, fee.

(bars, gage) 0.2 0.2 0.2 - 0 17 14

Temperature range (°C) -201040 -20t040 -201040 -20to40 -20to40 b b

Common Construction Materials

Carbon steel x x x x x X
Concrete Xx x x x
Plastics or fiberglass x x
Alloys and coated
orcladsteel _ x x x Xx x

Footnotes: a. Modified from Ulrich (1984). Used with permission. b. For steel, the range is -20 to
600°C; for aluminum, -250 to 200°C; for stainless steel, -250 to 800°C; for nickel-based alloys, -200
to 700°C.

Valves and their performance have been described in numerous

articles previously. (See, e.g., O'Keefe, 1971; O'Keefe, 1976a,b; Miller,
1983.) Only some brief highlights regarding valve selection will be given
here,
Globe valves (see Figure 2.9) are often used for tight shut-off and
throttling situations. The active portion of the valve can be a circular plug
that enters a circular hole (the seat). In the most-used design, the fluid
must turn 90° to flow through the seat and then turn back again to the
original direction. As a result, the pressure drop across the valve is
relatively high. Angle valves have similar characteristics but are used in
place of a valve plus an elbow in a piping run. !n this manner, the overall
cost of the angle valve installation may be less than the cost of a

31
 Selection of Fluid Flow Equipment

straight-run valve plus the elbow. Both types of valves demonstrate good
throttling performance and seal tightly in the "off" position. If wear of the
seat does occur, repair is usually easily accomplished compared to many
other types of valves.

\
=

Figure 2.10 Schematic diagram of a gate valve.

Gate valves (see Figure 2.10) operate on a "slide-past" principle,

often using a tapered wedge to close off the flow. Open operation usually
demonstrates lower pressure drop than a corresponding globe valve.
However, the gate valve is not designed for throttling service, as
“wire-drawing" erosion can occur to. the seat. Frequent "off"-and-"on"

32
 Selection of Fluid Flow Equipment

service can cause similar problems. The valve is not “bubble-tight" in the
off position, and this characteristic is made worse with handling of dirty
fluids. Costs of gate valves are usually less than corresponding globe
valves.

SS NS
SSS

Figure 2.11 Schematic of the active portions of a butterfly valve (left) and a ball valve
(right). Arrows here show direction of closure.

For quick off-on operation, as well as low pressure drop in the on

position, the butterfly and ball valves are prime candidates. These devices
typically require only a 90° turn of the stem for complete off-on operation.
(In fact, the on-to-off action can take place so quickly, there should be
some concern shown for waterhammer possibilities.) A schematic of two
basic types of off-on valves is shown in Figure 2.11. In the first of these,
a 90° rotation of flap arrangement either seals against the shoulders or
opens fully to flow. The second of the schematics shows the ball-type
construction. This latter design seals better in the closed position and
demonstrates lower pressure drop in the open position than the butterfly
type, but there is usually a premium cost for the ball type. The plug valve
(not shown) is schematically similar to the ball valve, but the ball is
replaced with a more cylindrically shaped rotating element. Most of the
valves in this category have the characteristic that they may not be
suitable for high-temperature applications because of the seating
material's temperature limitations.
Check valves are used to ensure that the flow in a pipe is in one
direction only. This kind of valve is one of the few types that is truly
automatic in its operation. The modes of operation of these valves vary
from a flapper arrangement like that shown in the left-hand side of Figure
2.12 and a diaphragm concept like that shown in the right-hand side of the
figure to a ball stopper arrangement that is not shown here. Most of the
valves in this category are not normally considered to be able to react to
high-frequency variations in flow, such as might be found in the output of a
reciprocating pump. Another concern normally shown about these types of
valves is that they are not typically bubble-tight.
A large category of valves found in typical designs is denoted as

33
 Selection of Fluid Flow Equipment

control valves. Whereas many of the basic types of control valves are not
too different than some of those discussed above, the emphasis here is on
the ability to control the flow rate to desired values. Several possibilities
are available in a given design. Some valves open quickly (relative to their
total stem travel) while others open more slowly. This has implications on
the corresponding pressure change in the line downstream, which is
obviously related to the flow rate.

ASScE
Figure 2.12 The method of operation of two types of check valves is shown. The left-hand
side uses a flapper arrangement to constrain the flow to one direction, while the right-hand
side uses a diaphragm concept. The normal flow direction is as shown by the horizontal
arrows.

See Figure 2.13 for some typical examples of control valve action.
Although the overail control of flow depends upon both the effect of the
valve and the complete downstream resistance, the valve denoted in the
figure as "equal percentage" will often give an approximately linear system
characteristic. Most control valves demonstrate performance between this
type of variation and the one denoted as "linear" in the figure. Special
concern is given in the design of control valves (particularly the trim,
which is a name for the orifice formed between the gate or plug and the
seat) and the material specifications to result in a device with high
reliability.

Quick |
= [-apening
2
L§ Linear
g§
i

MS 6 f— —

Qe
Equal _|
percentage
| | |
Percent of stem travel

Figure 2.13 Plot of typical responses of various types of control valves

34
 Selection of Fluid Flow Equipment

Numerous aspects related to the specification of the many types of

valves have not been covered here. In applications, concern would have to
be given to the valve (and its various subcomponents’) materials, the bonnet
design, the body-bonnet connections, the body end connections (this is
related to the discussion of piping given in the next subsection), the
seating details, the packing, and, of course, the cost. To show the wide
variation in the latter, prices for one size of different types of valves are
shown in Table 2.6. It should be obvious that valves can vary widely in
price and can affect total plant cost. For an introduction to many of the
other aspects regarding valve specification, refer to the references at the
end of this chapter.

Table 2.6

Prices of Various 4-in. Valves@

Description Priceb

Class 125 cast iron gate valve bronze mounted $125

Class 125 cast iron check valve bronze mounted 110
Class 125 cast iron globe valve bronze mounted 250
Class 300 cast steel gate valve 13% chrome steel trim 410
Class 300 cast steel check valve 13% chrome steel trim 450
Class 300 cast steel globe valve 13% chrome steel trim 620
Class 150 ductile iron plug valve with nonlubricated plug 370
Class 300 ductile iron plug valve with nonlubricated plug 550
Class 300 cast steel p ug valve with lubricated plug 1050
Class 150 cast steel ball valve with Teflon trim 615
Class 300 cast steel ball valve with Teflon trim 715
Class 600 pipe line valve with cast steel high-temperature trim 7400
High performance butterfly valve with cast steel seats 500

a. From Miller (1983). b. Mid-1983 prices. Quantity purchased and country of origin will affect
prices.

2.8 PIPING

Precise specification of piping is almost never required in the

preliminary design analysis of systems. But there may be times when there
is a need to define some general aspects about piping. Three items should
be of special concern when this happens.
One of these has to do with the bore of the fluid channel. This is
normally found in a specification of the schedule. Schedules of 40, 80,
160, or XX pipe are common.
A second item to be specified relates to the mechanical strength of
the fittings. Here the normally found values are 150-, 3000-, and 6000-Ib
ratings. Usually, the 150-Ib components are either cast or forged, while
the higher rating counterparts are typically forged. Flanges (see additional
discussion below) are found in pressure classes of 150, 300, 400, 600,
900, 1500, and 2500 Ib.

35
 Selection of Fluid Flow Equipment

N Lars ENS Threaded

RES
Face detail Attachment Lap joint
(see below) detail (see
at right) ;
Slip on
IN BRASMS ABN N LAT SEN
Flat Raised Ring-type LNG Socket weld
face face face

Welding neck

Figure 2.14 Schematic details of the various face designs and attachment forms of common
flanges. See accompanying text for additional explanation.

The third item to address is the type of fitting connections to be

used. Of course, there are fittings like ells, tees, couplings, and so on, that
must be specified; and the correct incorporation of these components into a
piping configuration is assumed. On the other hand, some choices can be
made regarding the connections. In terms of connections, the normally
available options are: threaded, socket welded, and butt welded. Flanges
can be used or not used with any of these types of connections, and the
flanges are among the most expensive of any elements other than valves.
Both the facing of the flange and the general attachment mode are
necessary in the specifications. See Figure 2.14.
Some description of the standard attaching designs may be of value.
The threaded flange has a diameter that matches the pipe size. It is used
for low-pressure systems and where welding could be hazardous. The
threaded type is not normally used with larger diameter pipe. A /ap joint
type is bored slightly larger than the OD of the pipe, and the radius on the
bottom matches the radius on the stub end. The fitting is slipped over the
pipe, and the stub end is welded onto the pipe and bolted up. This design is
used for systems that need frequent cleaning or inspection.
The slip-on flange has a low hub and is bored slightly larger than the
OD of the pipe. This flange is welded on boih the inside and outside of the
flange face to prevent leakage. A welding-neck flange (see discussion that
follows) is used more frequently than this design in common application,
but the slip-on flange finds usage when cost or space may be a major
consideration.
Socket-weld flanges are bored to the ID of the pipe and counter-
bored slightly larger than the OD of the pipe to allow the pipe to be
inserted and welded in place. Normal application finds this type of flange
in smaller diameter and higher pressure systems.
By far the most common flange type is the welding neck. It is bored

36
 Selection of Fluid Flow Equipment

to the ID of the pipe and has a high neck. Use of this type of flange usually
results in the soundest welds.
Not shown is the blind flange. This is simply a solid circular piece
of metal that is drilled for flange bolts. It is used to blank off a piping run.
Special problems are often faced in specifying proper expansion
capabilities for piping runs. Two approaches that are considered are
piping loops and special expansion joints. As outlined by Broyles (1985),
there are numerous reasons why expansion joints are specified, including:
limitations on space available or piping diameter too large to make loops
practical; desire to minimize pressure drop; better ability to handle
abrasive fluids; and pipe diameter. Piping loops may also be favored for a
variety of reasons. Included here are: the ability to incorporate loops with
needed changes in direction of the piping run; impracticality of expansion
joints for very high pressure and/or corrosive service; and possible code
violations for the use of expansion joints.
Excellent descriptions of proper piping practices are available.
(Included are King 1979; Crane, 1979.)

REFERENCES
B&W, 1972, STEAM/ITS GENERATION AND USE, The Babcock & Wilcox
Company, New York.
Bailey, F., K. Cotton, and R. Spencer, 1967, "Predicting the Performance of
Large Steam Turbine-Generators Operating with Saturated and Low
Superheat Steam Conditions," American Power Conference Paper
(reprinted by the General Electric Company).
Bloch, H., J. Cameron, F. Danowski, R. James, J. Swearingen, and M.
Weightman, 1982, COMPRESSORS AND EXPANDERS, Marcel Dekker, New
York.
Brown, R., 1986, COMPRESSORS--SELECTION & SIZING, Gulf Publishing
Company, Houston.
Broyles, R., 1985, "Pipe Loop or Expansion Joint," Chemical Engineering,
October 14, pp. 103-106.
Crane, 1979, FLOW OF FLUID THROUGH VALVES, FITTINGS, AND PIPE, The
Crane Co., New York.
Dimoplon, W., 1979, "What Process Engineers Need to Know About
Compressors," in COMPRESSOR HANDBOOK FOR THE HYDROCARBON
PROCESSING INDUSTRIES, Gulf Publishing Co., Houston, pp. 1-8.
Gulf Publishing Co., 1979, COMPRESSOR HANDBOOK FOR THE HYDROCARBON
PROCESSING INDUSTRIES, Gulf Publishing Co., Houston.
Hicks, T., and T. Edwards, 1971, PUMP APPLICATION ENGINEERING, McGraw-
Hill, New York. :
Karassik, |., W. Krutzsch, W. Fraser, and J. Messina (Eds.), 1976, PUMP
HANDBOOK, Mc Graw-Hill, New York.
King, R., 1979 (Ed.), PIPING HANDBOOK, FIFTH ED., McGraw-Hill, New York.
Miller, R., 1983, "Valves: Selection, Specification, and Application,"
Heating/Piping/Air Conditioning, October, pp. 99-118.

37
 Selection of Fluid Flow Equipment

Neerken, R., 1974, "Pump Selection for the Chemical Process Industries,"
Chemical Engineering, February 18, p. 104.
O'Keefe, W., 1971, "Valves," Power, March, pp. S:1-S-16.
O'Keefe, W., 1972, "Pumps," Power, June, pp. S-1-S-24.
O'Keefe, W., 1976a, “Control Valves, Actuators, Regulators, Positioners,”
Power, April, pp. S°1-S°16.
O'Keefe, W., 1976b, "Check Valves," Power, August, pp. 25-36.
Peters, M., and K. Timmerhaus, 1980, PLANT DESIGN AND ECONOMICS FOR
CHEMICAL ENGINEERS, 3RD EDITION, McGraw-Hill, New York.
Pollak, F. (Ed.), 1980, PUMP USERS' HANDBOOK, Gulf Publishing Co., Houston.
Reason, J., 1983, "Special Report--Fans," Power, September, pp. S:1-S-24.
Singer, J. (Ed.), 1981, COMBUSTION POWER SYSTEMS, THIRD EDITION,
Combustion Engineering, Inc., Windsor, CT.
Stewart, H., and T. Philbin, 1984, PUMPS, The Bobbs-Merrill Co., New York.
Thompson, J. E., and C. J. Trickler, 1983, "Fans and Fan Systems," Chemical
Engineering, pp. 46-63, March 24.
Ulrich, G. E., 1984, A GUIDE TO CHEMICAL ENGINEERING PROCESS DESIGN AND
ECONOMICS, Wiley.
Walker, R., 1972, PUMP SELECTION, Ann Arbor Science Publishers, Ann
Arbor, Michigan.
Warring, R., 1984a, PUMPING MANUAL, SEVENTH ED., Gulf Publishing Co.,
Houston.
Warring, R., 1984b, PUMPS: SELECTION, SYSTEMS AND APPLICATIONS,
SECOND ED., Gulf Publishing Co., Houston.

PROBLEMS

2.1 Size and specify a pump for the following application. A process fluid
is to be pumped at a rate of 100 gpm through a pressure rise of 100 psi.
Your company lab measures the density and the viscosity and reports their
values as 52 lbm/ft? and 2000 centipoise, respectively. Cost, although not
a totally dominating factor, should be made no larger than necessary. Be
sure to state any considerations you make when selecting an appropriate
pump. This should include not only why particular pumps are rejected and
one selected, but it should also include any potential application problems
with the one selected (e.g., if flow reversal could be a problem with the
pump selected, so state). ‘*

2.2 A fan is to be selected for installation in your manufacturing facility.

In this facility, construction of heavy equipment is performed, including
cutting, welding, sand blasting, and use of large cranes. There are many
particulates in the air around the fan inlet. It is desired to move 60,000
cfm from the facility (atmospheric pressure is 12.9 psia, temperature is
usually around 80°F) through a pressure rise of 0.2 psi. Specify and size a
fan for this application. Be sure to state any considerations you have
through the selection process.

38
 Selection of Fluid Flow Equipment

2.3 The preliminary design of a small power plant is being performed. One
required aspect of this design is the specification of the fan for the boiler.
Design output of the plant is to be 100 kWe. The overall plant efficiency is
anticipated to be approximately 38%. Natural gas will be the fuel used. It
is assumed that a boiler can be configured such that it will have an
efficiency of 91% and a pressure drop at design conditions of approximately
4 kPa. Select and size a fan for this application. Make appropriate
assumptions if necessary.

2.4 A volume flow rate of 17 m°/s of air is required at 14 bars, and this
air is to be furnished by a new compressor from atmospheric conditions in
a plant (approximately 0.95 bars and 30°C). Determine a power requirement
for this compressor assuming an appropriate efficiency, find the output
temperature of the air, and specify an appropriate type of compressor for
this duty.

2.5 The feedstream of a fluid (specific heat of 2.1 kJ/kg’C, density of

1100 kg/m) for a process plant must be heated from 20 to 65°C. It is
anticipated that the fluid will flow at a rate of 20 kg/s for periods of 3
min and then no flow will occur for 10 min. Size a heater tank (much like a
home water heater) that uses electricity to accomplish the heating.
Specify the physical size of the tank and the electrical element rating
required such that the size of each is minimized. Add an “over-design
factor" to the specification (safety factor) by assuming that the flow rate
of the fluid is actually 30 kg/s.

2.6 Contact a pump vendor and secure information of the sort shown in
Figure 2.4 for a single product line of pumps. Plot the information if that
is not already done.

2.7 From information similar to that used in Problem 2.6, but for only a
single diameter, present appropriate curves of pressure rise versus flow
capacity for two of the pumps connected in series and in parallel.

2.8 Summarize the key trade-offs between radial and axial flow gas
expansion turbines.

2.9 Air is to be compressed from atmospheric conditions to an outlet

pressure of 3 bars (absolute). Make up a brief table that shows the inlet
flow ranges of the appropriate kinds of compressors that could be used.

2.10 CO, is available at one point in a process at 40 bars and 300°C. Itis
desired to estimate the power that may be available if this stream is
expanded through an appropriate work device to atmospheric pressure. A
flow rate of 100 kg/s of this gas is available. Select an appropriate
expander, estimate the power output, and determine the temperature of the
exhaust.

39
 Selection of Fluid Flow Equipment

2.11 Analyze a jet pump to move water at 200 kPa to a pressure of 300
kPa. A total of 40 kg/s of water at 40°C is needed. Plot the pressure
against the flow rate (both for the high-pressure stream) for this
situation. Make and state appropriate assumptions about the conversion of
the high-pressure stream in the pump.

2.12 Make a list of the important considerations that would go into a

choice between an axial flow compressor and a centrifugal compressor.

2.13 Specify an appropriate type of pump driven by an electric motor, and

calculate the power required to pump 800 gpm of water through a pressure
rise of 80 psi. Choose a lower cost option if more than one type is
available. Assume that the electric drive and associated hardware have an
overall power conversion efficiency of 89%.

2.14 By referring to some of the literature cited in this chapter or some

other appropriate information, make a list of factors that must be
specified for the actual purchase of a centrifugal pump.

2.15 List the considerations that must be given for the actual
specification of an axial flow compressor. You may consult a vendor,
reference material, or whatever. Be sure that your list is reasonably
complete.

40
 CHAPTER 3
HEAT EXCHANGE
DESIGN OPTIONS

3.1 INTRODUCTION

Consider now various design options in heat exchange. A major

element in this general topic is heat exchangers. These devices can be
found in so many configurations that a person who has been simply
introduced to the Log-Mean-Temperature Difference and Effectiveness-NTU
methods of analysis can be quite perplexed in trying to determine which of
the almost limitless types of heat exchangers available, many apparently
satisfying the required heat transfer duty, should be used.
Designs that incorporate tubes, for example, are only a subset of the
many heat exchangers available. In spite of being only a subset, there is an
organization that sets standards for tubular heat exchangers (TEMA).
[Actually several organizations deal with standards for heat exchangers
including Tubular Exchanger Manufacturers Association (TEMA, 1978),
American Society of Mechanical Engineers (ASME, 1980), Heat Exchange
Institute (HEI, 1978), and American Petroleum Institute (API, -).] Since
heat exchangers can be found in so many forms, distinctions are often left
out of discussions in beginning heat transfer texts. However, since heat
exchangers find broad applications in virtually all thermal systems, some
practical aspects are addressed here. As will be seen, "heat exchangers" is
a broad term involving both single devices as well as systems.
In a final section of this chapter, some aspects related to design
choices in thermal insulation are discussed. It is not always the case that
insulation selection is a critical element in preliminary design studies, but
some knowledge of this facet of plant engineering can be of value at that
point.

3.2 HEAT EXCHANGERS

3.2.1 Overview

Heat exchangers have been the focus of uncountable articles, papers,

and books. Most of these outline the numerous design analysis techniques
41
 Heat Exchange Design Options

that are available. (Included are Kern, 1950; Fraas and Ozisik, 1965; Kays
and London, 1964; Butterworth and Cousins, 1976; Feldman et al., 1976;
Butterworth, 1977; Bellotty and Stock, 1979; Karag et al., 1981; Karag et
al., 1983; Pettigrew et al., 1983; Shah, 1983a,b; Taborek and Hewitt, 1982;
Crane and Gregg, 1983; Boehm and Kreith, 1987.)
Analysis of most heat exchanger applications is done in a manner
very much like that discussed in introductory heat transfer texts, either
using the conventional modified Log-Mean-Temperature Difference (LMTD)
approach or the Effectiveness-NTU (e-NTU) technique. In general, the LMTD
approach may offer a more direct solution if all of the temperatures of a
given design are known and other quantities must be found. The e-NTU
technique is preferred when the outgoing temperatures are not known, but
this method can be used for any case. In practice, there is a wide variety
of methods being applied that range from simply entering the LMTD
approach on a programmable calculator (Crane and Gregg, 1983) to detailed
hydraulic analysis of the internal flow field in a heat exchanger (Pettigrew
et al., 1983).
Extensive tabulations of the functional forms of the e-NTU method
are given by Shah (1983b). He also gives the relationships between the
LMTD variables and the e-NTU variables.
Often, the most critical step in the analysis of a heat exchanger is
the determination of the overall heat transfer coefficient, U. This involves
the application of convection and/or phase change correlations to find the
surface coefficients, h, and use these with the areas, A, and Ap, and wall
resistance, R,,, to find the result of Equation 3.1.

1
UA, = UrA, (3.1)
Uh,A, + 1/Ry +1/hA,

The determination of pressure drop should be evaluated also as this

quantity is always important in design applications. Some heat exchangers
that perform very well thermally may have very high pumping power
requirements. Evaluating the relative economic values of heat and work is
the focus of a later chapter.
Short of performing detailed studies of various types of heat
exchangers to see which type is preferred for a particular application,
some shortcut methods have been given in the literature. One such method
is the “effectiveness index" technique described by Brown (1986). He
defines the effectiveness index as the overall heat transfer coefficient for
the given heat exchanger divided by the cost per unit area of the heat
exchanger. Of course, the higher the value of this index, all other factors
being constant, the better the buy. Average values typically range between
1 and 6 {Btu/hr’F$} for a variety of heat exchangers.
While much design information is given generally in the various

42
 Heat Exchange Design Options

monographs on heat exchangers, the beginning design engineer is rightfully

perplexed by the myriad of types of heat exchangers available. Which type
should be used in a specific application? If only the most appropriate type
could be selected, it might not be too difficult to analyze. To outline
general characteristics of certain types of heat exchangers, some
categorizations of these devices follow. Clearly, the list of types of heat
exchangers, as well as each type's advantages and disadvantages, is
necessarily limited in scope. An attempt will be made to touch only on
some of the more important aspects. A summary of much of the
information discussed in the following section is given in Table 3.1. You
may wish to refer to this to put the discussion into perspective.

3.2.2 Shell-and-Tube Heat Exchangers

3.2.2.a General Comments

By far the most widely applied heat exchangers are those

constructed of a "shell," which contains one of the fluids as well as the ro
tubes, and the "tubes," which contain the other fluid. The heat transfer * )

Classification and discussion of characteristics of these devices have been

given in numerous places. (See, e.g., Lord, et al. 1970; Fanaritis and |
Bevevino, 1976; Gutterman, 1980; Mehra, 1983, as well as the monographs ¢
noted above.) For purposes here, it is appropriate to subdivide the *
shell-and-tube category into three major subcategories: return bend «©
(sometimes called U-tube ), fixed tubesheet, and floating head (sometimes
called floating tubesheet.)
Before describing the generally encountered configurations, some |
information about applications will be given. me

® Tube Diameter--Make this as small as possible to increase the

surface area per unit volume of fluid. Limitations on pressure
drop and the ability to clean the outside of tubes may place a
lower limit on this parameter.

9 Tube Length--Generally make this as long as possible to decrease

costs. Consider the cost of heat exchange duty per unit length of
tubes to see whether or not longer tubes are, indeed, increasing
performance.

® Tube Pitch--Usually a triangular arrangement will decrease the

overall size. Other arrangements (e.g., square) might be needed to
decrease pressure drop and allow mechanical cleaning.

% Shell Design--Often the fluid on the shell side is in laminar flow.

Concern must be given that the shell-side fluid cannot shortcut the
desired path, and this is accomplished with well-designed baffles.

43
 Heat Exchange Design Options

To achieve desirable shell-side velocities, multiple shells in series

might be used for low flows, and high flows might require the use
of multiple shells in parallel.

Table 3.1

Data for the Preliminary Selection of Surface Heat Exchangers?

< Shell and Tube >< Plate ><Air Cooled>

Fixed Floating Flat

Tube Sheet U-Tube Bayonet Head? Plate Fin-Fan

Maximum surface area (m2) 800 800 100 1000 1500 2000

Typical number of passes (shell/tube) 1-2/1-4 1-2/2-4 2/i 1-2/1-4 WA 1/2

Maximum operating temperature (°C) 150 350 350 350 260 260

Typical maximum operating

pressures, bar (shell/tube) 140/140 140/140 140/140 140/40 20/20 -/140

Minimum practical A T approach (°C) 5 5 5 5 1 5

Maximum Flow Capacity

Liquid (m3/s, shell/tube) 0.7/0.7
Gas (std m/s, shell/tube} 15/15 15/15 2/2 15/15 AI5

Typical Mean Flow Velocity (m/s)

Liquid (shell/tube) 1-2/2-3 1-2/2-3 1-2/2-3 1-2/2-3 - -/2-3
Gas (shell/tube) 5-10/ 5-10/ 5-10/ 5-10/ - 3-6/
10-20 10-20 10-20 10-20 10-20

Compatibilit®
Fouling service (shell/tube) E/B D/D A/D B/B A/A -IA
Cleanability E/B D/D B/E B/B A/A -/A
In-service tube replacement A D A A A A
Differential thermal expansion Cc A A B A A
Thermal shock E A A D A A
Toxic or hazardous fluids (shell/tube) A/A A/A AIA X/A B/B JA
Condensing service (shell/tube) A/B B/B A/B A/B E/E -/B
Evaporative service (shell/tube) A/A A/D A/D AIA E/E -1E
Viscous liquids (shell/tube) E/B E/B D/D D/B B/B -/B
Maintenance B D B B A A
Alloy construction (shell/tube) D/C D/C D/C D/C CIC AC
Heat transfer efficiency B B D B A B

Relative cost (1 = low, 4 = high) 1 1 4 2 4 2

Pressure drop (bar)

Shell 0.2-0.6 0.2-0.6 0.2-0.6 0.2-0.6 0.5-1.5 0.0012
Tube 0.2-0.6 0.2-0.6 0.4-1.0 0.2-0.6 0.5-1.5 0.2-0.6

Footnotes: a. Modified from Ulrich (1984). Used with permission. b. Packed-tube sheet type. c.
Key: A=excellent or no limitations, B=modest limitations, C=special units available at higher cost to
minimize problems, D=limited in this regard, E=severely limited in this regard, X=unacceptable.

44
 Heat Exchange Design Options

© Shell-Side versus Tube-Side Applications of Fluids--Normally,

turbulence is more easily initiated on the shell side because of the
typically more complicated flow path there. Hence, apply higher
viscosity and lower flow rate fluids on the shell side, all other
factors being equal. Also, if a fluid use requires periodic cleaning
of the heat transfer surface or has other special needs (e.g., if it is
fouling, corrosive, toxic, or has a high pressure or a high
temperature), it is normally flowed through the tubes. Highest heat
transfer per unit of pressure drop is generally possidle in the
tubes. Condensing fluids are usually placed on the shell side.. A
need for specialty metals in contact with the condensing fluid may
necessitate tube-side condensation to decrease costs.

® Counterflow versus Parallel Flow--lt is well known that

counterflow offers the potential for the maximum temperature
change of a fluid stream when compared to parallel flow. Only in
special circumstances is counterflow not used, and many of these
circumstances are a result of the physical layout of the heat
exchanger. Also, if one of the fluids is only changing phase, there is
no distinction between counterflow and parallel-flow performance.

For simple shell-and-tube heat exchangers, the tubes represent

approximately 60%-75% of the total purchase cost (Dart and Whitbeck,
1980). Hence, the type of material used for the tube construction can have
a major effect on the cost of the heat exchanger. Relative costs of some of
the more common tube materials are given in Table 3.2.

Table 3.2

‘Relative Cost of Common Heat Exchanger Tube Materials

Approximate Relative
Material Material Cost

pow-carbon steel
r
SCODDDOUDOWN$Oo

Red brass
G9 PO NO = as a

Admiralty brass
90/10 copper-nickel
Aluminum
304 stainless steel
316 stainless steel
Nickel
Monel
ON 90 C1

Inconel
Titanium
=—

Hastelloy

@From Dart and Whitbeck (1980).

Operating pressures have profound implications on the cost of shell-

and-tube heat exchangers also. While the specific application will dictate

45
 Heat Exchange Design Options

the details of the costs, an example of pressure effects on shell cost will
be noted. This information is given Table 3.3. Similar kinds of cost effects
due to tube pressure are also present. Additional cost factors are given in
the Appendix.

Table 3.3

Effect of Shell-Side Design Pressure on Heat Exchanger Cost®

Design Pressure Heat Exchanger
MPa (psi) Relative Cost

2 (300 1.0
4 (600 1.3
5 (750 1.6
7 1200) 2.0
8 (1200 2.5

@From Dart and Whitbeck (1980).

3.2.2.6 Return Bend (U-Tube) Type Shell-and-Tube Heat Exchangers

One of the more common shell-and-tube heat exchangers in use in

industry is the return bend, or U-tube, type. A simplified drawing of one of
these devices is shown in Figure 3.1. In practice, many tubes would be used
where only one tube is shown in the figure. The vertical plates shown can
serve as supports for the tubes, baffles to change the flow direction of the
shell-side fluid, or both purposes.

Baffles or supports

Longitudinal baffle

Figure 3.1 A sketch of a simplified shell-and-tube heat exchanger. The shell-side fluid
flows as shown by the upward-pointing (gray) arrows, and the tube-side fluid flows as
shown by the black arrows. While only one tube is shown here, in practice, a large number of
tubes would be used.

This shell-and-tube configuration allows a great deal of flexibility in

design applications and, as a result, is frequently found in industrial

46
 Heat Exchange Design Options

systems. Among the large number of advantages attributable to this type

of heat exchanger are the following: (i) large variations in temperature of
the service can be tolerated because the tubes can readily expand or
contract, and no added expansion joints (as are found in some other types of
heat exchangers) are necessary; (ii) very high pressures can be applied on
the tube side; (ill) generally, this type of design is less expensive than
many of the other tube-and-shell types; (iv) the tube bundle can usually be
removed for cleaning (this can be a difficult task, however) or repair (not
all tubes may be individually accessible for removal); (v) no internal
gaskets or special packings are required; and (vi) in design configurations
that are slightly different than those shown in Figure 3.1, the shell-side
inlet can be moved to a point where the shell-side fluid will not impinge on
the tubes upon entry. The latter characteristic may be desirable in some
applications where the direct impact of the entering fluid could erode away
the surface of the tubes at the location of incidence. Alternatively, an
impingement plate can be located to protect the tubes from the incoming
shell-side fluid, but this is an added expense in the unit.
~~-- "This type of heat exchanger is not without its disadvantages in some
applications. Included are the following: (i) typically, service is res-
tricted to clean fluids because cleaning is difficult by mechanical means on
either the shell or tube side; and (ii) many of the tubes may be almost
impossible to replace if a failure occurs. (iii) U-tubes always result in
even numbers of tube passes. While this is a limitation, it is only minor.
Tube bundles similar to the tube section of a shell-and-tube heat
exchanger are often inserted into vessels to serve as stand-by heaters or
boilers. .This form is sometimes called a bayonet application, but this
name is also given to a modified form that employs straight tubes, sealed
at one end and fixed to a tubesheet at the other. In this latter form, the
bundle is inserted into a shell. A second bundle is formed from a second
series of smaller tubes, each open at both ends but fastened at only one end
in another sheet. The second bundle is inserted into the first bundle with
the tubesheets separated by a spool spacer. This latter form of bayonet
heater offers some operational and maintenance improvements over the
U-tube variety but at an increased cost. The straight-tube form shares
some common characteristics with fixed-tubesheet exchangers described
in the next section.

3.2.2.6 Fixed-Tubesheet Exchangers

A simplified sketch of a fixed-tubesheet heat exchanger is shown in

Figure 3.2. In contrast to the U-tube heat exchanger, here the tubes are
attached to both sides of the shell arrangement.
The fixed-tubesheet heat exchanger is similar to the U-tube in many
respects. One clearly distinguishing feature is that an end chamber is
found in this type that functions to return the tube-side flow through the
second set of tubes. Hence, the U-portion of the heat exchanger is replaced
with a plenum region in the fixed-tubesheet device. This leads to some

47
 Heat Exchange Design Options

differences in possible function compared to the U-tube heat exchanger.

Tubes
Shell Partition
- ale plate

‘a ubesheet
— (One at
Baffles each end)
Tie rods and

Figure 3.2 A simplified sketch of a typical fixed-tubesheet heat exchanger. Several tie
rods and spacers, as well as many pairs of tubes, would be found in an actual device.

Among the advantages of the fixed-tubesheet types of exchangers are

the following. (i) This type of heat exchanger can handle fouling fluids on
the tube side. The straight-through form of the tubes allows mechanical
cleaning inside. (ii) Fixed-tube exchangers generally allow configurations
with odd- (including one) or multiple-tube passes. As noted before, the odd
number of passes is not possible with U-tubes. (iii) A fixed-tube
arrangement has fewer joints than other types of straight-through
exchangers. (iv) Minimum capital expense is involved in the fixed-
tubesheet type when compared to other straight-through types. (v) Of all
straight-tube types of exchangers, the fixed-tubesheet type offers the
maximum protection against leakage of the shell-side fluid to the
environment. (vi) This configuration can result in the minimum shell
diameter of all shell-and-tube heat exchangers for a given heat transfer
surface, with the same diameter, length and number of tubes, and tube
passes (Mehra, 1983).
Of course, some disadvantages are also inherent in this type of
device. (i) Perhaps the most important concern is that thermal stresses
can become critical if the effects of temperature profiles in the tubes and
shell are not matched appropriately. Too much expansion in the tubes
compared to that of the shell, for example, will cause the tubes to be in
compression. This condition can have very involved implications because
all aspects of operation (including start-up, shutdown, normal operation,
and any unusual situations) must be considered. Expensive expansion joints
may be required. (ii) While the tube side can accommodate fouling fluids
because mechanical cleaning is easily accomplished there, the shell side is
restricted to clean fluids.

3.2.2.0 Floating-Head Exchangers

These heat exchangers are a variation of the fixed-tubesheet types,

designed to accommodate the movement of the tubes that might result from -

48
 Heat Exchange Design Options

thermal expansion and contraction. See Figure 3.3. For this reason, these
exchangers yield most of the advantages of the fixed type without the
drawback of concern for thermal stress effects. In fact, this seemingly
simple modification complicates the design and maintenance of the device.
As a result, both capital and operational costs are higher. Care must be
taken in the design of the way the floating head operates so that leakage
does not result. This implies many concerns including the need to make
sure that uneven thermal expansion of the tubes does not "tip" the head.

Shell Tubes

fe —

—b Me
ng
sealed in one of a varie
of ways against leakage

Figure 3.3 A simplified sketch of a floating-head-type heat exchanger. Note that this
device is very similar to the one_shown in Figure 3.2, except that this one has a movable
tubesheet on the left-hand side. The way that the movement of the tubes is accommodated
while not allowing the tube-side fluid to mix with the shell-side fluid varies considerably
in actual designs.

3.2.3 Plate Heat Exchangers

Plate heat exchangers are a relatively early development, with at

least one patent dating back to the late 1870s (Clark, 1974). These devices
are made from specially formed metal plates with grooves pressed in them
similar to the simple schematic shown in Figure 3.4 and the right-hand side
of Figure 3.5. The grooves serve two basic functions: aid the heat transfer
process and add mechanical rigidity to the overall assembly. As shown in
the right-hand side of Figure 3.5, actual plate heat exchangers differ from
the one shown in Figure 3.4 in that a given plate could incorporate several
hundred grooves and a complete exchange would be built out of several
plates stacked together. Heat transfer occurs between two streams across n
plates in the overall assembly. Two additional plates form the two outer
containers of the complete device. Usually, the plates are constructed from a
cold-worked metal, often a stainless steel with a thickness around 1 mm. Plate
heat exchangers are seldom constructed from mild steel. Seals, often referred
to as gaskets, are used to contain the fluids within the flow channels of the
assembly. The plates are held together in a layered form by a frame and
tightening bolts. Plate heat exchangers can be purchased with total heat
transfer areas down to a fraction of a square meter and up to over 1000 m2.

49
 Heat Exchange Design Options

5
Photographs of a commercial
St unit and a typical plate are shown in Figure

©
<¢ Top views Sections
Figure 3.4 Conceptual sketch of two plates from a simplified plate heat exchanger. Plate
A would be mounted in front of plate B, and one fluid (1) would flow in and out of one pair of
ports, and the other fluid (2) would flow through the second pair.

Figure 3.5 A commercial plate heat exchanger is shown on the left, while a typical plate
from such a unit is shown-on the right. Photos courtesy of Alfa-Laval, Inc.

Performance prediction techniques for plate heat exchangers can be

found in a number of papers and books. One such paper has recently
compared the prediction techniques available in the open literature
(Caciula and Rudy, 1983). In general, these devices have approximately the
same effectiveness as a true counterflow heat exchanger.

50
 Heat Exchange Design Options

lt is of value to contrast the plate heat exchanger to the

shell-and-tube configuration in terms of advantages and disadvantages. A
comprehensive comparison has been given (Cooper, 1974). The points made
there are summarized below.

9 The plate heat exchanger is best for liquid-liquid duty with (flow
rate)(specific heat) product nearly the same for the two fluids.
Flows with dissimilar products can be applied with some decrease
in overall effectiveness.

For high-pressure duty (above about 300 psi), a shell-and-tube heat

exchanger is preferred over a plate-type device. The basic
construction of plate heat exchangers makes them unable to contain
high pressures.

More area per unit volume of heat transfer surface is available in

the plate-type device compared to virtually all other closed-type
exchangers. When this characteristic is coupled with the ease of
manufacture of the plate device, the plate type usually has a lower
cost than any tube-type heat exchanger. However, this comparison
is made between exchangers constructed from special materials,
like stainless steel. Because of the typically larger surface areas
of plate heat exchangers compared to the shell-and-tube type, as
well as the nature of the flow in a plate device compared to tubes,
fouling applications are more easily handled in the plate exchanger.

If an application does not need a heat exchanger made of a special

material, a shell-and-tube device may be less expensive than a
plate type. This is because a plate exchanger is normally not
constructed with mild steel. The shell-and-tube heat exchanger is
definitely preferred when the device can be constructed from mild
steel and a close approach temperature is not required.

One of the more expensive components of the plate heat exchanger

is the gasket, and the materials of construction of this component
will normally set the temperature limit of the heat exchanger's
operation. Typically, this limit will be about 500°F (260°C).

3.2.4 Phase Change Heat Exchangers

A number of heat exchanger applications involve one or both streams

undergoing a change of phase. While this can include freezing/thawing, in
fact, the most important situations are those where boiling/condensing
takes place. Some of the heat exchangers described above can be used for
these applications. There may be special concern for the orientation of the
heat exchanger because gravity effects are often very important in the
overall heat transfer performance. For example, there is a need to remove

51
 Heat Exchange Design Options

the condensed liquid readily from a condenser surface so that the heat
exchanger can operate more efficiently.
Devices where boiling occurs offer particular challenges. A number
of factors contribute to this. For example, the boiler in a coal-fired steam
power plant must be able to transfer the heat from the hot combustion
gases to sections where preheating, boiling, and superheating are taking
place. Each one of these three sections offers special design challenges
(B&W, 1972; Singer, 1981). Since the design of these devices is so
specialized, the interested reader is referred to the literature just noted.
On the other hand, the engineer in the chemical process industry
(CP) faces a whole range of different kinds of problems in the design and
specification of evaporators. The CPI engineer usually designs for the
separation of two or more components in a stream, while the power plant
application generally has the more direct end goal of steam generation. The
power plant application generally uses highly treated water whereas the
CPI applications often use streams with high levels of "impurities." (This
term is used loosely because one engineer's impurities might be another's
end product.) As a result, continuous blowdown is usually required. Also,
when a mixture of substances is used, rather than a single substance like
water, the resulting mixture experiences a boiling point rise as the
evaporation takes place. Finally, it is often the case in the CPI! that the
desired product is the residue; while in the power plant application, the
goal is the manufacture of steam.
Evaporators are found commercially in a wide range of designs.
These range from simple batch-pan evaporators that have either external
steam jackets or internal heating coils to the much more complex agitated
thin-film evaporators that can handle extremely viscous, heat-sensitive,
crystallizing, and fouling materials. An excellent review of many of the
types of commonly used evaporators has recently been given by Mehra
(1983).

3.2.5 Air-Cooled Heat Exchangers

When heat is removed from a stream to a.temperature close to

ambient, air-cooled heat exchangers can be applied. While this is the most
frequent use of heat exchangers where one fluid is a gas, it is certainly not
the only one. The boiler, superheater, and reheat elements in a power plant
are obvious, though fairly specialized, examples of these kinds of devices.
Recuperators are often used where energy is recovered from a high-
temperature stream. Refrigeration units almost always have some kind of
device for heat transfer from a liquid and/or vapor to a gas.
The elements of design analysis for heat exchangers in a gas stream
are given in many texts. Usually, it is apparent that extended surfaces
(fins) are of value in decreasing the thermal resistance on the gas side of
the heat exchangers. The addition of fins can increase the external area of
a tube by factors up to 50 times that of a nonfinned tube (note Equation
3.1). Internally finned tubes are available also, but these are less typically

52
 Heat Exchange Design Options

applied. Extensive design and application data are given in Kays and London
(1964) and Shah (1983a) and in the HVAC literature. Design techniques
have even been published for programmable calculators (Shaikh, 1983).
Often something more than simply a heat exchanger will be needed
for transferring heat from a liquid to a gas stream or the ambient air, and
it is necessary to consider a heat transfer system. When removal of waste
heat to the ambient is desired, several approaches may be available. Four
of these options have been compared by Huber (1976), and these are shown
in Figure 3.6.

| Alr cooling
Surge
Lea fank

wr
Drain

Evaporative
L cooler
Heat
source

41-4 wr Water

Y Drain

Shell and tube Once-through

Heat y Drain
source Shell and tube
Surge f
tank | Drain "WT denotes a water treating process

A ver 3.6 Schematic diagrams of several types of heat rejection systems, after Huber

In all cases shown in Figure 3.6, it is assumed that a liquid medium

carries the heat from the source to the rejection device. Hence, a pump,
surge tank, and fluid circulation loop is shown in each option.
In the first option shown ("air cooling"), the circulating fluid flows
through a liquid/air heat exchanger denoted as an air coil in the figure. Use
of a cooling tower is shown in the second option. Aspects of the first two
options are shown combined in the third, called "evaporative cooler" in the

53
 Heat Exchange Design Options

figure. Finally a once-through system is shown in the fourth figure, where the
heat is transferred to water, which is then dumped to a drain. in the fourth
option, there may be the possibility in some instances to use the treated water
directly in removing the heat from the source. Normally, however, the practice
is to use the configuration shown.
Comparisons of these systems may help in many instances in
determining which to apply. In terms of initial costs, the air-cooling
option normally will often be the highest. This option also requires the
largest land area. The once-through water system is usually the least
expensive in all acquisition costs. In terms of operational costs, the water
system will need more raw water and sewer capacities, and these aspects
could account for sizable costs. Pollution control in any of the systems
that dump water must also be considered. Electrical consumption is nor-
mally higher on any of the systems that move air. Water treatment costs
must be considered on the second, third, and fourth options. Maintenance
and repair costs tend to favor the dry air systems in almost every respect.
One of the major points of comparison for the systems shown in
Figure 3.6 is the way in which the heat rejection duty is accomplished and
the implications of this. The air system depends upon the dry-bulb
temperature for its cooling capacity. Both the cooling tower and the
evaporative cooler achieve primary fluid temperatures that approach the
wet bulb. Of course, the once-through system may be independent of either
of these in certain circumstances. If freezing conditions exist during part
of the year, both the cooling tower and evaporative cooler may require
special operation. The evaporative cooling option may be able to operate in
the wet/dry modes, where spray water is not used in cold periods. With
appropriate design, this option may perform similarly to the air-cooling
option when the spray water is not used.

3.2.6 Direct-Contact Heat Exchangers

3.2.6.a General Comments

A very large number of configurations can be classified under the

category of direct-contact heat exchangers. Included are applications
where solids, liquids, or gases exchange heat with liquids or gases.
Devices of this type range from the often applied (e.g., evaporative coolers,
cooling towers, and open feedwater heaters) to farms that are much less
frequently used (e.g., coolers for air compressors or heat-reclaiming
devices for combustion processes).
General aspects of direct-contact heat exchangers have been
discussed recently (Boehm and Kreith, 1986; Boehm and Kreith, 1987). A
very large set of possibilities exists for this general category of devices.
Included are very high temperature solids/gas heat transfer applications to
very low temperature difference, heat transfer situations between
immiscible liquids. One example of the latter type of device is shown in
Figure 3.7.

54
 Heat Exchange Design Options O P

Continuous phase in

@8©eee0oeg0e2086006606068686
@eeeo0e60060606660666
eeeeoeoeeoeo6606666

@®eeeeeoeoeocogoeogoeeos
Droplets ("dispersed
Separation phase") of one fluid
zones °@
in bulk of second
8
6
8
fluid ("continuous
% phase")

Continuous E Dispersed
phase out E phase in

Figure 3.7 A schematic diagram of one of many varied types of direct-contact heat
exchangers. This type is patterned after mass transfer columns and the fluid combinations
can be liquid/liquid or liquid/gas.

There are a number of advantages to these types of devices. Direct-

contact exchangers have the inherent characteristic that heat can be
transferred between two streams without an intervening material surface.
This virtually eliminates the detrimental corrosion and fouling aspects
that can occur in conventional heat exchangers. Other benefits normally
include lower approach or pinchpoint temperature differences as well as
lower costs than those of closed-type exchangers. In summary, direct-
contact heat exchange can eliminate fouling and corrosion of heat exchange
surfaces, improve performance, and cut costs.
While this all sounds too good to be true (and many times it is true
and very good), there are some limitations to keep in mind. First, since the
two streams are in direct contact, they are necessarily at the same
pressure during the heat transfer process. Thus, high differentials of
pressure between the two streams are normally not possible. Also, it may
be the case that the two streams should be such that one does not go into
solution in the other or otherwise merge in a manner that makes their
separation difficult. (Sometimes it is of value to have this happen,
however. More on this follows.) Usually this does not cause the direct
contact application to be impossible, but this requirement may result in
the need for some creative engineering. Finally, and perhaps most

55
 Heat Exchange Design Options

importantly, the existing design techniques are less comprehensive and

straightforward than is the corresponding situation with closed
exchangers. While techniques have been established for the more commonly
applied devices like cooling towers, and some relatively simple methods do
exist for more general types (see Fair, 1972; Boehm and Kreith, 1987), this
absence of generalized design techniques is a hindrance to widespread
applications of these devices. Then, to summarize the limitations, both
streams have to be at the same pressure, a moderate to high solubility of
one of the fluids in the other can cause operational problems, and
simplified design techniques are lacking.
An important aspect to note in passing is the role of mass transfer
in direct-contact heat transfer devices. This aspect could appear on both
the benefits and drawbacks list given above. In some situations, mass
transfer is desired. For example, a wet cooling tower (discussed in the
next section) functions better when larger amounts of water are
evaporated into air. Another example, discussed in Section 3.2.6.c, where
mass transfer is desired is the deaerating heater. There are situations,
however, where direct-contact heat exchanger designs may benefit from
the minimization of mass transfer and significant amounts of mass transfer
could be considered undesirable. An energy-reclaiming heat exchanger on a
combustion exhaust gas stream may be one example of this.

3.2.6.6 Cooling Towers, Evaporative Coolers

Cooling towers and evaporative coolers are very important devices in

thermal systems for a variety of reasons. Two major distinctions can be
made between these two categories of equipment. (i) Normally the output
sought from cooling towers is cooled water, while the end result of an
evaporative cooler is cooled air. (ii) Cooling towers are found in more
types of configurations than evaporative coolers are. For example, seldom
is an evaporative cooler found in a natural-draft design. (See discussion of
this term that follows.) Both types of devices benefit from mass transfer
that accompanies the heat transfer.
What follows will focus on cooling towers. Keep in mind that many
of the comments are applicable io evaporative coolers also.
Empirically based design techniques for cooling towers have
developed over the years and are now being partially replaced with more
fundamentally based approaches. Typical design calculations have been
outlined in the literature (Campbell,1987). A national organization
oversees cooling-tower testing and certification (the Cooling Tower
institute in Houston, Texas), and this group has issued acceptance test
procedures and certification standards for commercial units.
There are a number of forms of cooling towers. Generally, they can
be divided into groups according to whether or not the air is forced through
by fans (sometimes called mechanical draft) or flows by free convection
(sometimes called natural draft). Subdivisions of the mechanical draft
types are made according to whether the fan is located where the air flow

56
 Heat Exchange Design Options

comes in (forced draft) or where the air exits (induced draft). Further
subdivisions of all types of cooling towers are based on the direction of the
air flow, usually being either across the water flow or counter to it. One
further distinction is made according to whether or not the water and the
air come in direct contact. The former.are termed wef and are the most
prevalent. The latter are denoted as dry. There are some types that can
function in either mode. Several schematics are shown in Figures 3.8 and
3.9 to illustrate some of the types of cooling towers.

fo
i

Wet, (mechanical) Wet, (mechanical)

induced draft, forced draft,
counter flow ‘ cross flow

key > Water Air

SSN
Fen Drift Tubes
eliminators

ve
Wet, (mechanical) Wet/ary, (mechanical)
induced draft, induced draft,
cross flow cross flow

Figure 3.8 Examples of three wet cooling towers and one combination wet and dry tower.
Evaporative coolers function very similarly to the wet types shown.

An excellent review of the various types of cooling towers available

and their characteristics has beén provided by Robertson (1980). Several
aspects of operational interest have been covered by Elliott (1985).
There are relatively clear design choices between the various types
of cooling towers. Although there can be exceptions, some of the
considerations are noted below (Robertson, 1980).

57
 Heat Exchange Design Options

Warm air out

(Not to scale)

Warm |
water in
Cool water
Cooled <@-—~ basin
water out Water flow detail

Figure 3.9 A sketch of a hyperbolic, natural-draft cooling tower.

® Natural-draft cooling towers are more influenced in their

operation by plant and environmental conditions than are their
forced-draft counterparts. This is because the natural-draft air
flow rate varies with a number of environmental factors.

® Water pumping costs are generally less in a mechanical draft

tower.

& Overall auxiliary power consumption is. less in a natural-draft

cooling tower because the decrease from not having to move air is
significantly greater than the increased pumping costs.

® Maintenance costs are appreciably greater for mechanical draft

cooling towers.

® More land area, site preparation, and piping costs are required for
mechanical draft cooling towers.

9 Natural-draft cooling towers usually suffer from less

environmental impacts, such as ground-level fog, and
recirculation problems because the moist exhaust air is rejected
at higher elevations than the corresponding mechanical draft type.

9 Capital cost comparisons reflect a clear advantage for the

mechanical draft type at small installation size; but a crossover

58
 Heat Exchange Design Options

occurs as size increases, and the natural-draft type usually

shows a clear advantage at larger sizes.

® The physical size and shape of the natural-draft type of tower

may be a major factor in whether or not this type is selected.

For power plant applications, there has been a trend toward the use
of hyperbolic towers (Elliott, 1985). A example of one of these is shown in
Figure 3.9. As these devices were making their impact on the market in the
1970s, it was believed that they would be used where the following
situations were applicable: low humidity and wet-bulb temperatures
existed; heavy winter loads were possible; and high inlet and exit water >
temperatures were present. More recent applications in humid regions have |
also proven to be satisfactory. Elliott (1985) notes that a 10-14°F
(5.5-7.8°C) approach to the summer wet-bulb temperature is not uncommon.
Dry cooling towers are simply a closed heat exchanger where the
water is circulated. Air is forced across the tubes to perform the cooling.
As would be expected, the approach temperature difference can be ~
considerably higher than that for a cooling tower using direct-contact L
processes; but this type of cooling eliminates water use. A compromise
that greatly improves performance at the cost of small water use is the &
dry/wet tower. See Figure 3.8. The capital costs for this type of device * :
are higher than are the costs of either the dry or the wet towers. Mitchell
and Henwood (1978) compared various cost factors for power plants using tes
wet, dry, or dry/wet towers for the cooling systems. Some aspects Of
their study are presented in Table 3.4. Note that the first line of the table ¢

er,
is indicative of the relative capital costs of the various types of cooling =~
systems (alone) in large sizes. The numbers in the other lines show the — .
effects of cooling system, and other items needed to boost the plant...
- capacity back to the original rating, on total plant cost and produced power. |

Table 3.4

Relative Cost? Comparisons between Power Plants

using Wet, Dry, and Dry/5% Wet Cooling Towers®

ltem Wet Dry Dry/5% Wet

Capital costs of cooling system only® 1 2.05 2.80

Capital costs: cooling and extra capacity®-4 1 3.02 2.78

Busbar energy costs©-4 1 1.12 1.07

a. Costs are given relative to the “wet" type of tower. b. Uses cost data
resented by Mitchell and Henwood (1978). ¢. Applies to a 1000-MWe (net)
ossil-fired power plant in the Central Valley of California. d. Includes only
those costs affected by type and/or design of cooling system.

59
 Heat Exchange Design Options

3.2.6.¢ Deaerating Heaters

Both the power industry and the chemical process industry use
deaerating heaters. These are simply direct-contact heat exchangers that
allow a venting of noncondensable gases. Normally, these heaters operate
at atmospheric pressure, but this may not be the case in special instances.

Vents
Water box
Nozzles
Spray chamber
Tray stack
Steam equalizers

Storage

To boiler or feedpump

Figure 3.10 Schematic diagram of a deaerating heater patterned after Drabnis (1986). In
the lower left a simplified Rankine cycle is denoted with the location of deaerating heater
shown. There the following holds: B=boiler, T=turbine, C=condenser, P=condensate pump (and
boiler feedpump in this situation), D=deaerating heater, M=makeup, and V=vent.

In power plant applications, the heater serves to heat condensate and

make up water with bleed steam. A simplified application example is
shown in Figure 3.10. Although the details of various designs differ, the
one shown sprays the water onto a series of trays. Steam flows up through
the trays, condenses on the water while transferring heat, and picks up
noncondensables that are liberated during the heating process. This brief
introduction to deaerating heaters is included to give another example
where mass transfer and heat transfer occur simultaneously to good
advantage in a direct-contact device. ,

3.2.7 Other Types of Heat Exchangers

The number of heat exchangers that will not be discussed here far
outnumber those already mentioned. A classification, as well as reference
to some of the important literature, of many types of heat exchangers used
in the chemical process industry has been given by Lord et al., (1970).
Included are double-pipe and cascade exchangers, coils and bayonet forms,
spiral tube, falling-film and jacketed units, scraped-surface exchangers,
thin-film heat exchangers (includes agitated film, wiped film, and
centrifugal film), and froth-contact heat exchangers.

60
 Heat Exchange Design Options

Usually, the needs dictated by new applications require special

approaches. Just one of many examples of this is the use of heat
exchangers in high-temperature power cycles. (See, e.g., Fraas, 1975.)
The importance of special materials of construction can have a large
impact on specifics of design. For example, the use of "unusual" (from a
heat exchanger point of view) materials like ceramics (e.g., Forster et al.,
4980) or graphite (e.g., Hills, 1974-5) may accomplish needed heat
exchanger applications with a nonconventional configuration.

3.3 THERMAL INSULATION

There are a number of purposes why thermal insulation is applied

within a thermal system. Included are to:

® Conserve energy.

& Control condensation.

® Prevent potential. injury to operating personnel by lowering

surface temperatures.

® Maintain process temperatures.

In addition, there may be a number of nonthermal motivations for using

insulation on high-temperature devices, such as decreasing emitted noise.
While precise specification of insulation is normally not required in the
preliminary design process, an appreciation of basic ideas of insulation
application is of value.
Almost all insulations are made of solid materials. While some
insulations rely on achieving a thermal conductivity nearly that of air or
some other gas, the gas is usually contained within some form of solid
insulation.
insulating materials can be classified into three major categories.
These are fibrous, cellular, and granular. All forms function to decrease
convection and radiation. While convection is quite easily decreased to the
conduction limit, the minimization of radiation normally requires some
additional aspects to the basic material. For example, a basic insulating
material may incorporate a low-emissivity cover to decrease radiation.
Fibrous insulations consist of thin organic or inorganic fibers
manufactured in a blanket configuration. This form of insulation can be
very soft and light or it can be rigid and hard.
Cellular materials are constructed so that they have small bubbles
integral with a base component. Often, the manufacturing of these
materials involves a "foaming" process.
Granular insulation is normally a loose-fill form, but it can be in a
molded configuration also. Voids found around the particles, as well as

61
 Heat Exchange Design Options

those often found within the material, are responsible for the normally
very good insulating characteristics.
Eliminate from consideration, for the moment, both extremely high
temperature (above about 1100 K) and extremely low temperature (below
approximately 100 K) applications. For the remaining range of applications
where most thermal system designs are found, the typically encountered
generic insulations are as shown in Table 3.5. Also given there are some
typical values of thermal conductivity variation with temperature, the.
temperature application range, and density. For selected materials, the
costs relative to polyisocyanurate are given for simple straight-run pipe
insulation applications.

Table 3.5

Insulations for Moderate Temperature Applications?

Temperature Relative Cost? for Pipe Insulation

Thermal Conductivity Application Application
Insulation Wim °C Range Density Diameter Single Layer Double Layer
Type @-75°C @95°C @260°C °C kg/m? in. 2in. Sin. 4in. 6in.

Fibrous

Glass 0.029 0.037 0.072 -30 to 455 85 6 1.53 2.20 2.91 4.95
12 252 35 417 7.08.

Mineral Wool - 0.05 0.075 15to 1040 128 6 1.21 159 245 3.52
12 186 2.38 3.62 4.97

Cellular

Glass 0.039 0.061 0.094 -180t0650 136

Polyiso- - c - -180to 150 32 6 1 1.53 2.22 4.08

cyanurate 12 1.67 253 3.36 5.49

Polystyrene 0.026 d - -40 to 135 32

Polyurethane 0.020 0.036 - -75 to 100 32

Granular

Calcium - 0.061 0.079 1510815 176to 6 145 1.92 3.06 4.46

silicate 240 12 226 292 4.49 6.84

Perlite - 0.068 0.084 50to 815 80 to

208

a. Modified from information given by Liss (1986). b- Installed costs of straight pipe with 0.016-in.
aluminum jacket with moisture barrier (when required). Numbers given are the cost of particular
insulation divided by the cost of 2 in. of polyisocyanurate insulation on a 6-in. pipe. Based upon a 1985
cost survey in the U.S. Gulf Coastal region. ¢. At 24°C the thermal conductivity is approximately
0.020. d. At 24°C the thermal conductivity is approximately 0.033.

62
 Heat Exchange Design Options

Some variation in thermal conductivity values is noted between the

various insulations shown in Table 3.5. If the variation is analyzed, it is
found that there is an approximately inverse correlation between these
values and the cost for a given thickness. To express this another way,
there is a relationship between cost and A value (the thermal resistance of
ihe insulation) such that the product is approximately constant among the
insulations. Other factors that bear upon choices that are more expensive
per unit R value involve flammability and moisture-absorption
considerations. For example, the cellular insulations tend to be
combustible, with glass being the single exception of those shown. On the
other hand, the cellular insulations tend to absorb very little moisture, and
this factor may be very important in some applications. Sometimes a less _
expensive insulation (on a per unit FR basis) is eliminated because of space —
limitations in the installed location.
High-temperature insulations are normally made of materials that
are more dense than the insulations noted in Table 3.5. As a result, the
higher temperature types normally have higher values of thermal
conductivity. These insulations have been found in a variety of forms over
the years. Sprayed asbestos was used quite extensively in previous eras,
but concerns about health effects of asbestos have virtually eliminated
this particular material from all applications. Other forms of sprayed
insulation have replaced asbestos. Firebrick is heavily used in high-
temperature applications. This material is made by mixing an organic
material with the refractory before molding and then by burning off the
organic material during firing. As a result, voids permeate the brick.
Another common form of high-temperature insulation consists of ceramic
fibers. Here various fibers are mixed with binders for castable
refractories. Even though plastics contain organics that are generally not
considered for high-temperature applications, do not overlook them for
transient heating applications. For example, one approach to the insulating |
of reentry vehicles from outside the earth's atmosphere is to use an...
ablating (burning) surface made of an organic material. Some of these
forms can protect against temperatures in excess of 8000°C for short
periods of time.
For purposes of comparison to the lower temperature insulations
shown in Table 3.5, Table 3.6 shows some similar data for refractories that
are normally manufactured in block form. As is the case with the lower
temperature insulations, the density of the material can have a profound
effect on the value of the thermal conductivity and the cost.
Insulations for very low temperatures offer special challenges. If
the temperatures are low enough, most gases that may be present could
liquify. This, of course, will minimize convection. Generally, radiation is
the dominant, and sometimes virtually the only, mechanism of heat
transfer. The classical approach to thermal design for low temperatures is
to use evacuated regions for insulation and to apply a very low emissivity
to the surfaces enclosing the vacuum space. Placement of radiation shields
in the evacuated space will further decrease the heat gain.

63
 Heat Exchange Design Options

Table 3.6
Selection Data for Block Form Refractories®

Thermal Conductivity, W/m°C Density, Relative® Cost

Material 260°C 538°C 816°C 1093°C kg/m3 Installed, Flat

Mineral wool block 0.066 0.109 - - 240 1.0

Superfex block 0.075 0.130 - - 240 1.04
" 0.090 0.1114 - - 384 1.14
Ceramic fibers
Fiber chrome 0.072 0.180 0.187 0.272 128 3.44
Cerafelt 0.072 0.115 0.187 0.272 128 2.36
0.072 0.101 0.153 0.215 192 3.22
Insulating firebrick 0.126 0.153 0.183 0.195 496 1.26
0.298 0.320 0.350 0.412 816 1.42

a. Data calculated from that given by Neal and Clark (1976). 6. Cost is relative to the cost
for mineral wool block.

REFERENCES

ASME, 1980, ASME BOILER AND PRESSURE VESSEL CODE, American Society
of Mechanical Engineers, New York.
API, -, HEAT EXCHANGERS FOR GENERAL REFINERY SERVICE, API Standard
660, American Petroleum Institute, Washington, DC.
B&W, 1972, STEAM/ITS GENERATION AND USE, The Babcock & Wilcox
Company, New York.
Bellotty, J., and D. Stock, 1979, "A Numerical Design Scheme for Concentric
Heat Exchangers," ASHRAE Transactions, Vol. 85, Part 2.
Boehm, R., and F. Kreith, 1986, "Direct Contact--a High Performance, Low
Cost Option in Heat Exchange," Mechanical Engineering, March, pp.
78-81.
Boehm, R., and F. Kreith (Eds.), 1987, DIRECT CONTACT HEAT EXCHANGERS,
Hemisphere Publishing, Washington, DC. (To appear.) ~
Brown, T. R., 1986, "Use These Guidelines for Quick Preliminary Selection
of Heat-Exchanger Type,” Chemical Engineering, February 3, pp. 107-
108.
Burke, P., 1982, "Compressor Intercoolers and Aftercoolers: Predicting
Off-Performance," Chemical Engineering, September 20.
Butterworth, D., 1977, "Developments in the Design of Shell-and-Tube
Condensers," ASME Paper 77-WA/HT-24.
Butterworth, D., and L. Cousins, 1976, "Use of Computer Programs in
Heat-Exchanger Design,” Chemical Engineering, July 5, pp. 72-76.
Caciula, L., and T. Rudy, 1983, "Prediction of Plate Heat Exchanger
Performance," AICHE SYMPOSIUM SERIES, HEAT TRANSFER--SEATTLE,
AIChE, New York, pp. 76-89.

64
 Heat Exchange Design Options

Campbell, J. C., 1987, "Thermal Design of Water-Cooling Towers,” in

DIRECT CONTACT HEAT EXCHANGERS (Boehm, R., and F. Kreith, Eds.),
Hemisphere Publishing, Washington, DC. (To appear.)
Clark, D., 1974, "Plate Heat Exchanger Design and Recent Development,” The
Chemical Engineer, May, pp. 275-279.
Cooper, A., 1974, "Recover More Heat with Plate Heat Exchangers," The
Chemical Engineer, May, pp. 280-285.
Crane, R., and R. Gregg, 1983, "Program for Evaluation of Shell-and-Tube
Heat Exchangers," Chemical Engineering, July 25, pp. 76-80.
Dart, R. H., and J. Whitbeck, 1980, "Brine Heat Exchangers," in SOURCEBOOK
ON THE PRODUCTION OF ELECTRICITY FROM GEOTHERMAL ENERGY UJ.
Kestin et al., Eds.), U. S. Department of Energy Publication, DOE/RA/
4051-1, pp. 379-413.
Drabnis, A., 1986, "Selecting a Deaerating Heater to Suit Your Needs,”
Power, May, pp. 42-45.
Elliott, T. C., 1985, "Cooling Towers--A Special Report," Power, December,
pp. S:1-S-16.
Fair, J.. 1972, "Designing Direct Contact Coolers/Condensers," Chemical
Engineering, June, pp. 91-100.
Fanaritis, J.. and J. Bevevino, 1976, "Designing Shell-and-Tube Heat
Exchangers: How to Select the Optimum Shell-and-Tube Heat
Exchanger," Chemical Engineering, July 5, pp. 62-71.
Feldman, K., D. Lu, and L. Cowley, 1976, "Computer Aided Design of a Shell
and Tube Heat Exchanger for Oil and Asphalt Heating," ASME paper ....
76-WA/HT-6. fe
Forster, S., et al., 1980, "Development of High Temperature Ceramic
Plate-Type Heat Exchanger-and Burner-Elements," ASME/AIChE Heat
Transfer Conference Paper.
Fraas, A. P., 1975, “Heat Exchangers for High Temperature Thermodynamic
Cycles," ASME Paper 75-WA/HT-102.
Fraas, A. P., and M. N. Ozisik, 1965, HEAT EXCHANGER DESIGN, Wiley, New
York.
Gutterman, G., 1980, "Specify the Right Heat Exchanger,” Hydrocarbon
Processing, April, pp. 161-163.
HEI, 1978, STANDARDS FOR STEAM SURFACE CONDENSERS, Heat Exchange
Institute, Cleveland, Ohio.
Hills, D., 1974-5, "Graphite Heat Exchangers--I," Chemical Engineering,
December 23, 1974, pp. 80-83. "Graphite Heat Exchangers--ll,"
Chemical Engineering, January 20, 1975, pp. 116-179.
Huber, F. V., 1976, "Basic Principles, Advantages, and Limitations of
Air-Cooled Heat Exchangers,” Plant Engineering, June 24.
Karag, S., A. Bergles, and F. Mayinger (Eds.), 1981, HEAT EXCHANGERS--
THERMAL HYDRAULIC FUNDAMENTALS AND DESIGN, Hemisphere
Publishing, Washington, DC.
Karag, S., R. Shah, and A. Bergles (Eds.), 1983, LOW REYNOLDS NUMBER FLOW
HEAT EXCHANGERS, Hemisphere Publishing, Washington, DC.
Kays, W., and A. London, 1964, COMPACT HEAT EXCHANGERS, SECOND

65
 Heat Exchange Design Options

EDITION, McGraw-Hill, New York.

Kern, D. Q., 1950, PROCESS HEAT TRANSFER, McGraw-Hill, New York.
Liss, V. M., 1986, "Selecting Thermal Insulation," Chemical Engineering, May
26, pp. 103-105.
Lord, R., P. Minton, and R. Slusser, 1970. "Design of Heat Exchangers,"
Chemical Engineering. January 26, pp. 96-118.
Mehra, D., 1983, "Shell-and-Tube Heat Exchangers," Chemical Engineering,
July 25, pp. 47-56.
Mitchell, R., and K. Henwood, 1978, "Cost of Conserving Water in Power
Plant Cooling," Preprint 3346, American Society of Civil Engineers.
Neal, J., and R. Clark, 1976, "The Effective Use of Insulating Refractories,"
Johns-Manville Application Information, Johns-Manville, Denver,
Colorado.
Pettigrew, M., L. Carlucci, P. Ko, G. Holloway, and A. Campagna, 1983,
“Computer Techniques to Analyse Shell-and-Tube Heat Exchangers,”
ASME Paper 83-HT-61.
Robertson, R. C., 1980, "Cooling Towers," in SOURCEBOOK ON THE
PRODUCTION OF ELECTRICITY FROM GEOTHERMAL ENERGY (J. Kestin etal,
Eds.), U. S. Department of Energy Publication, DOE/RA/4051-1, pp.
611-644.
Shah, R., 1983a, "Compact Heat Exchanger Surface Selection, Optimization,
and Computer-Aided Thermal Design," in LOW REYNOLDS NUMBER FLOW
HEAT EXCHANGERS (Ed. by S. Karag et al.), Hemisphere Publishing,
Washington, DC, pp. 845-874.
Shah, R., 1983b, "Heat Exchanger Basic Design Methods,” in LOW REYNOLDS
NUMBER FLOW HEAT EXCHANGERS (Ed. by S. Karag et al.), Hemisphere
Publishing, Washington, DC, pp. 21-72.
Shaikh, N., 1983, "Estimate Air-Cooler Size," Chemical Engineering,
December 12, pp. 65-68.
Singer, J. (Ed.), 1981, COMBUSTION POWER SYSTEMS, THIRD EDITION,
Combustion Engineering Inc., Windsor, CT.
Taborek, J., and G. Hewitt (Eds.), 1982, ADVANCEMENT IN HEAT EXCHANGERS,
Hemisphere Publishing, Washington, DC.
TEMA, 1978, STANDARDS OF TUBULAR EXCHANGER MANUFACTURERS
ASSOCIATION. SIXTH EDITION, Tubular Exchanger Manufacturers
Association, Tarrytown, New York.

PROBLEMS

Use texts on heat transfer, thermodynamics, and other topics as needed for
assistance in working the following problems. Some of the problems listed
here are meant to be a review of calculational procedures covered in basic
courses, while others apply specifically to information given here.

3.1 Develop a computer program that will make Effectiveness-NTU

calculations for either simple counterflow or simple parallel-flow

66
 Heat Exchange Design Options

shell-and-tube heat exchangers (i.e., the program, on an input cue, will

make either type of calculation). Assume that overall heat transfer
coefficient, U, as well as both mass-flow-ra te-specific-heat products are
read in as data and that one of the following two situations will always
hold. Show a flow chart and sample calculation.

(a) The area of the heat exchanger and both incoming temperatures
are given as data, and the outgoing temperatures are desired as
output.

(b) One outgoing and both incoming temperatures are given as data,
and the area of the heat exchanger is desired as output.

3,2 An evaporative cooler, similar to the device shown in the upper rignt-
hand section of Figure 3.8, is used to cool air. The ambient conditions
include P= 1 atm, T = 40°C, and relative humidity of 15%.

(a) What is the minimum temperature that can be achieved by the

air? What will be the amount of water required (per kilogram of
dry air) to accomplish this?

(b) if the air exits at 25°C and the fan flow is rated at 150
m/min, find the exiting relative humidity and the amount of water
used in a 6-h duty.

3.3 A wet, counterflow cooling tower has an inlet water flow rate of 40
kg/s at 40°C, and the cooled water exits at 18°C. The ambient air at 20°C
and 10% relative humidity flows in a direction counter to the water. If the
air/water mixture leaves the tower at a dry-bulb temperature of 32°C and
a wet-bulb temperature of 25°C, find the makeup water flow rate that is
required. Assume that the pressure is 1 atm throughout and that there is
no heat transfer, except between the air and water.

3.4 Water at 15°C is to be heated with an oil (thermal properties like

engine oil) at 110°C in a shell-and-tube heat exchanger. The heat
exchanger is a U-tube configuration with one shell pass and two tube
passes made of 1-cm-diameter thin-walled tubes. There is approximately
1 m2 of area. The amount of oil available is 4 kg/s. Show the resulting
water exit temperature as a function of the water mass flow rate. Assume
tubes 1 m long and in a triangular configuration.
(a) Without making any calculations, should the oil flow through
the tubes or the shell?
(b) Perform the calculations for the water flowing through the
tubes.
(c) Perform the calculations for the oil flowing through the tubes.
 Heat Exchange Design Options

3.5 It is desired to find the off-design performance of a counterflow

aftercooler on an air compressor (water cooling the exhaust air). At design
conditions, with the flow of air being 24.3 kg/min throughout the heat
exchanger and the water flow being 30.3 !/min at its 16°C entry, the air is
cooled from 121 to 38°C. At off-design, the air flow is increased to 28.2
kg/min and the entering water flow is 22.7 l/min at 27°C. If the air flow
enters at 135°C, find the corresponding air exit termperature (Burke, 1982).
Can this calculation be made as easily with the e-NTU and the modified
LMTD methods? If so, use both methods and compare the results. If not, use
the simpler method.

3.6 Compare the various cooling options shown in Figure 3.6. Assume the
following: all mass-flow-specific-heat products are the same throughout
all heat exchangers; all liquid/liquid heat exchangers have an effectiveness |
of 75% and all liquid/gas closed heat exchangers have an effectiveness of
60%; the heat source generates a 15°C temperature difference in the coolant
fluid after flowing through the heat source; the dry-buib temperature is 30°C;
and the relative humidity is 20%. Make whatever further assumptions are
necessary to rank the various concepts from most efficient to least efficient.
Define your criteria for efficiency. List other factors specifically that should be
considered in choosing between the various concepts.

3.7 Check with a vendor to determine the cost of heat exchangers with
approximately the following specifications (data may not be applicable to
all situations): 2m? area, 0.5-cm diameter tubes, 14-bar tube pressure
rating, 2-bar shell pressure rating, and temperature application range from
15 to 150°C. Do this for the following configurations:

(a) Shell and tube.

(i) Low carbon steel tubes.

(ii) Monel tubes.

(b) Floating head configuration of any specific type.

(c) Plate.

Where possible, compare your results to information given in Tables 3.1 and
3.2.

3.8 List all types of heat exchangers mentioned in this chapter that could
be used for the following duty: heating of large flow rates of water with
high flow rates of oil at temperatures between 30 and 300°C. For each of
the types listed, note any further considerations you may have to evaluate
before selecting that particular type.

68
 Heat Exchange Design Options

before selecting that particular type.

3.9 Assuming that 40% of the cost of installing 1 in. of cellular glass is
due to the basic cost of the material and 60% is due to labor, and further
assuming that the absolute cost of the latter does not change if thicker
material is used, plot a variable related to the total installed cost of the
cellular glass insulation against insulation thickness, the latter ranging
from 1 in. up to 12 in. thick. On the same plot show the total R value of the,
insulation over this same range of thicknesses. Finally, show on the same —
plot how the heat transfer varies through the insulated system. For this,
assume that there is an external AR equivalent to the value of 1 in. of this
insulation over this same range of thicknesses. Show on the same plot how
the heat transfer varies through the insulated system. Finally, try to illustrate
how the energy saved compares to the insulation total cost. For this insulation
and that a 100°C temperature difference exists across the whole system.
Assume further that this value of external R and temperature difference do
not change with insulation thickness.

3.10 Rank the following heat exchanger types for service in each of the
separate situations shown below. If you cannot distinguish between two
types for a given situation, or you need more information, so state. (a)
Shell and tube. (b) Fixed tubesheet. (c) Plate. (d) Direct contact.

(i) Heavy fouling propensity of one fluid.

(ii) One fluid is extremely viscous.

(iii) Large ranges of temperature difference are encountered

in short times.

{iv) High pressures are found in one stream.

(v) High pressures are found in both streams.

(vi) Very close approach temperatures are required.

3.11 A cooling tower is to be installed on a chemical processing plant in

the driest portion of Arizona. Indicate the options you have and the
information you would seek to determine which is the best type of cooling
tower to install. Do not be concerned with brand names; simply focus on
generic types.

69
 CHAPTER 4
FITTING DATA AND
SOLVING EQUATIONS

4.1 DATA FITTING

4.1.1. Introduction

Physical data surround us and are critically needed for the design
process. This information may result from data gathering on an
experiment, or it may be found in a manufacturer's product sheet, just to
cite two examples. Even the data required to perform an optimal design of
a complex thermal system could be almost limitless. Since the options for
most designs are so numerous, the designer must have capabilities for
generating mathematical formulations of physical data pertinent to the
task at hand.
Depending upon the situation, the required data may be readily
available within the designer's group. This is particularly true. in
established companies that have had a long history of design of specific
items. The internal combustion engine is a typical example of this. Most
heat exchanger manufacturers are another common possibility of this. They
may have performance and cost data information that could be readily
available for typical "new" designs.
There are instances, however, in even the most established
engineering design groups when some new data might be required.
Sometimes it may be a single piece of information, such as the price of 100
barrels of fuel oil delivered to Tobruk. Generally, though, the design
process will require a range of information, such as the flow and power
characteristics and price information on several models of blowers.
lf the design is to be carried out with the aid of a computer, as
assumed here, it will be desirable to reduce this data to one or more
mathematical correlation(s). This is termed curve fitting, or, as we shall
refer to it, fitting. While most computer systems (and many calculators!)
have some kind of software that is handily used for fitting, some basic
insights into this topic may be of value.
This section outlines some of the fundamental ideas of fitting. In
the following subsection, the curve shapes are discussed that are
represented by simple equations of one variable. A supplement to this

70
 Fitting Data and Solving Equations

information in the form of a brief catalog of function plots is given in

Appendix C. In the third subsection given here, attention is directed to the
fitting of equations when there is only enough data and when there is more
than enough data to determine the parameters of the equation. The final
subsection covers the fitting of data with more than one independent
variable.

4.1.2 Functions of One Independent Variable, "Exact Fit"

When only one independent variable is present, then Equation 4.1 is

applicable.

y=f(x) (4.1)
The fitting of engineering data to functions of this form has been described
by Daniel and Wood (1971) and Kolb (1982). The simplest situation is to
consider the /inear case, represented by Equation 4.2.

Y=ay + ajx (4.2)

Equation 4.2 has many more ramifications than might appear at first. This
is due to several reasons. First, the correspondence between much of the
data encountered is indeed linear. Second, if the data show a linear
correspondence between the independent and dependent variables, there is a
simple association of the variables that can be clearly seen. Third, there
are some situations when nonlinear relationships can be cast into the
linear form shown in Equation 4.2. As the old saying goes, "Be wise--
linearize." This approach to analysis can be quite helpful in many instances
to simplify what could be a much more complicated analysis.
Another form of linearity is also important. This applies to the
coefficients in the equation. Equation 4.2 is not only linear in x, but it is
also linear in the coefficients a, and a,. Because of this latter feature, the
equation will be denoted here as having linear coefficients. Linear-
coefficient equations yield to direct solution for the coefficients and to
the simple application of least-squares fitting, which is discussed later in
this section. The least-squares fitting is so important that it is of value
to reduce seemingly nonlinear-coefficient equations to linear-coefficient
forms wherever possible. An example of this is the simple power function
form found in the correlation of heat transfer data.

y = ax" (4.3)

While y clearly varies in a nonlinear fashion with x, the equation can be

cast in linear form by taking the logarithm! of both sides.

1 In this text "log" will be taken to be the logarithm to the base 10 and "In" will denote the
natural logarithm.

71
 Fitting Data and Solving Equations

Probably the single most used function for fitting of data is the
polynomial representation. A general form is given by Equation 4.4.

n
ys lax (4.4)
<0

While any number of terms can be applied, from a practical standpoint it is

almost never the case that an order greater than four (i.e., 7 > 4) is used. In
fact, most data will be fitted by no greater than a second-order form as
shown in Equation 4.5. Equation 4.5 shows that although there is a
nonlinear aspect between x and y that cannot be removed by a trans-
formation, this equation is linear in its coefficients as shown.

VY =A)
+ A,X + ayx* (4.5)

A number of functions, including the polynomial representation, are

discussed and plotted in Appendix C of this text. Use this appendix for
selecting appropriate types of functions to match the basic forms of data
you need. In all cases, the linear or nonlinear nature of the coefficients is
also mentioned.
Finally, it should be noted that sums of the various types of
equations can be used to represent hard-to-fit data. In fact, many of the
curve fitting programs have this capability. The problem is that intuition,
which can be beneficial in selecting which functional form should fit the
data, is of no value when the forms become complex. Excellent discussions
of the general topic of curve fitting are given in a number of places,
including works by Maron (1982) and Viswanathan (1984).
Consider now how the various equations discussed above are actually
fit to the data of interest. In this discussion, attention will be given only
to equations with the linear coefficient forms. The focus on this subset
will not be a severe restriction. As is shown in Appendix C, a number of
mathematical equations have linear coefficients that represent a large
variety of curve shapes.
To solidify the idea of linear coefficients, consider the relationship
shown in Equation 4.3. Taking the natural logarithm of both sides of this
equation yields Equation 4.6.

Iny=Iina+tninx (4.6)

if variables are defined such that Y=iIny, X =inx, and A=in a, Equation
4.7 results.

Y=A+nx (4.7)

In general, equations of one independent variable with linear

coefficients can be written in the form of Equation 4.8.

72
 Fitting Data and Solving Equations

y= a; f(x)= F(x) (4.8)

i=]

Thus, there are n unknowns in this general linear representation. To

determine these unknowns, / pairs of (x, Y) data must be given, where yis
used for the predicted value and Y; denotes the data value. The curve fit
then becomes a problem of finding the n values of a; from the known
information. In matrix form, this is given as

FA y(x4) fol4) essen FA(X4) Wea, 1 Ty, 7

Fy(Xo) folXp) vee F(X) ap Yo
desvsssesnennnenannsss nesses tenes =. (4.9)
Lty(X,) folXp) vee £(x,) IL a, lo Ly,J

In the remainder of this text, the following matrix format will be used to
represent this type of an equation in shorter notation:

[f] {4} = {y} (4.9a)

Brackets ([ ]) around a boldface letter will denote a square matrix, while
braces ({ }) around a boldface letter will denote a vector representation (a
one-column-wide matrix). |The unknown coefficients a; are found by
inverting the [#] matrix.

{a}=[fl'{y} (4.10)

EXAMPLE

Data are available for the cost of smail electrically driven fans. If
a 6-in.-diameter fan costs $13 and a 12-in.-diameter fan costs $18, find a
curve fit for predicting the price of these fans. It is known from the study
of fans that the cost of this model varies with the diameter of the fan in a
form similar to Equation 4.3.
The linear equations that must be solved use the alternate form
similar to that shown in Equation 4.4, but we will arbitrarily use the base
10 logarithm here. Using the given data, the numbers are as follows:

log 13 = 1.114
= loga + nlog6é a+0.778n

log 18=1.255 = loga + niog12 = a+1.079n

In matrix form, this is given as

73
 Fitting Data and Solving Equations

[1.0 o77ei al _ 11.114)

L 7.0 1079\|L n | 11.255|

This system is easily cast into a triangular form by subtracting the first -
equation from the second (this approach is discussed in greater detail when
the solution of systems of equations is treated later in this chapter). It is
then found that n = 0.468, log a= 0.750, and a= 5.623. Or

$ = 5.623d 9.468 (answer)

Checking this result, find that at d = 6, the curve fit yields a cost of
$13.006, while at d =72, the curve fit yields a cost of $18.000.

The general approach outlined above is used often. When this is done,
keep in mind what situation exists. The number of data pairs exactly
equals the number of constants to be determined in the equation. For the
curve fit to be accurate, the data pairs have to have a low uncertainty and
be fit by a curve that adequately describes the data over the range of
interest. Thus, the analyst must have insight into the proper form of the
mathematical relationship of the data at hand, and the data must be typical
and/or precise. This is denoted as the exact-fit case.

4.1.3 Least-Squares Technique, "Best Fit"

Many times quite a different situation than that described in the

"exacttit" case is encountered. There may be considerably more data pairs
available than the number of free parameters in the equation that is
appropriate. In this circumstance, it may be desired to achieve the "best
fit" that the data will allow. This is where the classical concept of /east-
squares fit comes into play.
The idea behind the concept of least squares is very simple. This
technique is a way of choosing the parameters in a curve fit such that the
differences between the predicted values and the initial data are
minimized. However, the criterion is not simply that the difference is
made as small as possible. If it were, then a line halfway between two
data points at the same value of the independent parameter would yield a
zero error. (This is the case since the difference between the prediction
and the high point would just exactly balance the negative value between
the prediction and the low point. An indication of zero error would result.)
Instead, the concept of least-squares fit uses the square of the differences
between the prediction and the actual data and minimizes these values.
To demonstrate the ideas involved in a least-squares fit, the concept
of a residual is important. The residual is the difference between the
prediction and the data point at the same value of the independent variable.
At the same value of the independent variable, denote the predicted values

74
 Fitting Data and Solving Equations

by the notation y; and the data values as the symbol Y; Thus, the residual
is the quantity given by the symbol r,in Equation 4.11.

The least-squares fit then requires that the sum of the squares
of the
residuals be minimized to achieve the best fit. Since any number
of
parameters must be determined in a given curve fitting equation (usually,
though, the number will be between 1 and 4), the sum of the squares
of the
residuals is minimized with respect to each of the parameters.
This is
best visualized by examining a specific case.
Consider a second-order polynomial as shown in Equation 4.5. At
each x; value, there is a predicted value, Yj» and an actual value, Y,; The
difference is the residual, as has been noted already. Squaring all residuals
and taking their sum, we have

n n

o= X72
= S(Y;- ay - a, Xj)-ayx,2)? (4.12)
is? i=1

The parameters to be determined are aq 4, and a,, and these

will be set to
make the sum a minimum value. This is accomplished by taking
the
derivative of Equation 4.12 with respect to the three parameters.

Wo) n
= 0=3 2 - ag: a, X- ay X;2)(-1) ' (4.13a)
day int
oc) n
= 0 =X 2V > ag - a)Xp AgX;2)(-x,) (4.13b)
O a, izt
ofc) n
= 0=3 QV - ag aX; AX; )(-x;2) (4.13c)
O a, fet

If the sum in each portion of Equation 4.13 is expanded, a series of three

linear equations in the three unknowns (ag, 2,, and a) results, as is shown
in Equation 4.14. This system of equations is easily solved to yield the
desired coefficients. Note that the (x, Y) pairs are the data values given in
the problem.

n n n
LY; = an + a, XX; + a, Xx? (4.14a)
i=f i=t i=

75 :
 Fitting Data and Solving Equations

n fi n n
x x) Yj; = Ag LX + ay LXF + a, x;3 (4.14b)
ixt inf i=1 i=1
n n n n

2x? Y; = ay 2x/* + 4, 2X + ag 2x" (4.14¢)

i= i= f= i=

Before ending this subsection, two observations should be made.

First, keep in mind that although the constants found as outlined above will
minimize the square of the prediction error, an appropriate form of the
equation must be known or the fit could be meaningless. A good check on
this aspect is to plot the residuals as a function of the independent
variable. If the fit is good, the residuals will show no particular trend.
Examples of curve fits (a) requiring a better equation to describe the data
and (b) not requiring a better equation are shown in Figure 4.1. In the left-
hand portion of the figure, the residuals show a correlation with x, which
indicates that another fit might match the data better. The right-hand
portion of the figure shows no particular correlation between the residuals
and the independent parameter. The magnitudes of the residuals overall in
the right-hand portion of the figure indicate the consistency of the data. If
the residuals were all very small, the curve fit would be very good.
+
+

- Residual
- Residual

v
©

Figure 4.1 Plotting the residuals of a given curve fit against the independent variable may
show whether or not the equation chosen is appropriate. The plot on the left shows a
correlation of the residual with x, indicating another form of the curve fitting equation could
improve the resulting prediction. Randomly distributed residuals as shown on the right
usually mean the form of the chosen equation is appropriate. :

Second, the emphasis given earlier in this chapter to curve fitting

equations with linear coefficients should again be noted. This
characteristic allows the direct solution for least squares shown in this
section. Equations with nonlinear coefficients can be fit with more effort

76
 Fitting Data and Solving Equations

(Daniel and Wood, 1971; Norris, 1981). Many computer systems have
software that will perform these types of operations. This topic will not
be covered here.

4.1.4 Two or More Independent Variables

When curve fits with more than one independent variable are needed,
the corresponding procedures are more complicated than the ones used in
subsections 4.1.3 and 4.1.4. Complete books have been written on this
topic. (See, e.g., Daniel and Wood, 1971.) In general, though, the basic ideas
for accomplishing fits to multivariable functions are related to the
concepts already discussed for functions of a single variable. Depending
upon the data available, either an “exact-fit" approach or a "best-fit"
approach may be of value.
Consider first an exact-fit situation shown in Figure 4.2. Here a
function z = f(x,y) is plotted on three separate graphs. It is assumed that
the data specified are limited and accurate (or all that are available), so
that the data can be fit with an equation that has the same number of
parameters as there are data points. In the case shown in the figure, it is
assumed that a total of nine data values are given and that one parameter
is held constant while a second is given for three points. ,

é (1) Z (2) z (3)

®
mn Y=V 5
® ®
@ ) @ ®
yey
e 2 e
x x x.

Figure 4.2 A function of two independent variables plotted parametrically. It is assumed

that nine data points are given, with the values of z occurring at the same values of x {.e., x7,
X9, X3) in each case.

For an "exact-fit" case when there are two or more independent

variables, start the same way as the single-variable case: plot the data.
Assume that this has been done as shown in Figure 4.2. Next, the form of
the equation to be applied must be chosen. If something about the physics
or some other basic understanding of the data can help in deciding the form
to use, by all means invoke that form. Otherwise, the situation is one of
choosing an equation that should fit the data over the range given.
Examination of the three curves in Figure 4.2 indicates that each shape is
quite different. While different forms of functions might be chosen to fit

77
 Fitting Data and Solving Equations

each of them, an all-purpose form such as a polynomial should be used.

Then each of the three correlations should have three parameters that can
be determined from the data.
For the purposes of being specific here, a polynomial is chosen to
accomplish the fit. Using a simultaneous-solution approach as was
described in Section 4.1.3, a curve fit is found for each of the three
portions. For example, the first of these could be represented as

Z(X, 4) = A,(¥4) + Dy y4) X + C1(V;) x2, Y=yy1 (4.15)

When all three representations like that shown in Equation 4.15 are found,
the final step is to fit the functions of y,, Yo, and yz. This is precisely
analogous to the situation shown in Equation 4.8. The specific equations
that must be fit in the second step include the following three:

a=a,t+b,y+e,y?

C =a,+DgV+C3y*

Similar equations for 4, a3, Oo, bz, Cy, and C, are determined in the second
set of solutions. [Keep in mind that when the equation shown above for ais
evaluated at y,, a;(y,) results.] The second solution set simply casts the
originally determined nine constants (from equations similar to 4.15) into
another set of nine constants (from equations similar to 4.16). Other ways
of fitting to functions of more than one variable are essentially
formalisms of the technique just outlined.
The "best-fit" analysis follows in a manner similar to that outlined
for a single independent variable in Section 4.1.3. Considering a function of
two variables, the generalized formulation can be written as shown in
Equation 4.17. More variables than two can be handled by generalization.

Z(X,y)
= 4) + a; f, (X,Y)
+ ap fo (X,Y) (4.17)

There is no restriction on the arbitrary functions except that they cannot

include undetermined constants. In other words, any constants must be in
the form of linear coefficients already discussed at the beginning of this
section. Thus, a permissible functional form could be the following:

z(X,y)
=a, + a, xX siny + ay
y* Inx (4.18)

The multivariable situation becomes considerably more complex than

its single-variable counterpart. In general, factors such as the possible
dependence between the independent variables becomes of concern. The

78
 Fitting Data and Solving Equations

spacing of the independent data, while important in the single-variable

situation, becomes even more critical here.
Readers who desire more information on this topic are referred to
the definitive text by Daniel and Wood (1971).

4.1.5 Interpolation

An important aspect of curve fitting is interpolation. Often data may

be available in tabular form and used that way, without the application of a
complete empirical fitting equation. When this happens, a means of
interpolation is usually needed. Interpolation is a means of inferring data
at points that fall between given data on the number scale. What kinds of
techniques are available to accomplish this?
A variety of techniques can be used for interpolation. Most texts on
numerical methods contain a great deal of theory on these techniques.
Included are forward difference, backward difference, and central
difference formulations of various types attributed to such mathematical
greats as Newton, Taylor, Stirling, Bessel, Lagrange, and Everett. Several
of the approaches are similar to curve fitting concepts discussed earlier in
this chapter. A limited number of techniques are dealt with here in the
interpolation context.
First, it is of value to state formally what is involved in
interpolation. If there is a function F(x), and values are known for it at
distinct points, say at x; then what are estimated values for the function
between the given values? °

F(x)

x4 Xo X3 X4 X5 xg

Figure 4.3 An example of a function F(x). Assume that discrete pairs of (x,F) values are
given at various locations xj. It is desired to find approximating functions f; to the spans of
the actual function F.

Consider the function shown in Figure 4.3. In specific, it is desired

to find approximating functions, denoted here as f(x), to the actual
function F(x), so that values of the actual function can be estimated
between any two given points, say x;and x;,,;. Denote the region of the
curve between any two points / and i+7 as the span of the curve in that
region. Of course, it is desirable to have the approximation function be as

79
 / Fitting Data and Solving Equations

close to the actual function as possible using only the discrete values of
the function. Hence, knowing the discrete points, what is an approximating
relationship for the spanning regions? (A related problem is the need to
estimate extreme values at either end of the given data sequence. For the
sake of brevity, the latter problem will not be addressed here.)
Perhaps the most familiar of all interpolation techniques is the one
that is often denoted by the name linear interpolation. Few designers have
not enountered this concept before. In this approach, the points are
assumed to be connected with straight lines. See Figure 4.4. The approx-
imating function for the ith span is given by Equation 4.19.

F(x)

‘5
5 5
5§ 4
4 5
4 ‘
‘ 4
4 4
4 4
4 4

xy Xo Xg X4 Xs Xe

Figure 4.4 The case where given data are spanned with straight lines is the well-known
linear interpolation method.

[ x-x,, 4] [ x-x; |
F(x) = F(x) + F(x.) (for
x,S x $X;,,) (4.19)
LXPXy1 Lx.7% J
For this situation, the value of the function is preserved across the given
point. Or:

f4(X) = Fx) (4.20)

Intuition, mathematics, and Figure 4.4 can all show, however, that the same
slope is not preserved on each side of the given points. To allow a better
fit, it is necessary to use a more complicated approximating function.
The linear relationship is a simple form of the general polynomial
interpolation (Maron, 1982). An nth-degree polynomial will have n
unknowns that can be used to match the value of the function and the
(n-7)st derivatives of the function across the point. In practice, it is
important that a compromise is made between ease of use and goodness of
fit. Third-order polynomials are an example of a very good compromise.
The application of these, called cubic splines, is now discussed.
Cubic splines are often used in interpolation applications. Typically,

80
 Fitting Data and Solving Equations

the form of these interpolating functions for each span is taken as follows:

i(x)=atrbx+ecexr +d (4.21)
Not only does this approach allow the values of the function to be made the
same across the given data points, as shown in Equation 4.20, but the first
and second derivatives of the functions are made the same across each
point. That is
fh. (X) = fi(x) (4.22)
for i=71 to n-1
fig" (X) = 5'(x) (4.23)
In addition, there is normally a stipulation that the second derivatives of
the span functions vanish as the two endpoints are approached. From
Equation 4.21, it can be seen that the second derivative of the function is
linear. A sketch of two spans of a possible set of cubic splines is shown in
Figure 4.5. The value and slope of both fand f' are preserved between
the
two spans (across the data point /). Only the value (not the slope)
of fis
the same.

f P'leiy f

i+? 1 I+]

Figure 4.5 A sketch of the variation of the approximating function, f(x), and its first
derivatives, f(x) and f(x), for two spans of data. The assumed linear two
behavior of f"is shown.

Taking the second derivatives of the cubic spline functions to be

linearly interpolated (Hang, 1976; Maron, 1982), a form like Equation
4.19
in terms of second derivatives is

PX,4-X7 [ x-x, ]
P'0d = f(x) + P(X), 4) (4.24)
Lx,.4-%; J Lx, 4-%; J

In this equation, all of the terms shown as f" are unknowns. Integrating
this equation twice, and applying the continuity of values and slopes, as
described above, a system of equations like that shown in Equations 4.25
results. Here the notation of f",= f"(x,) is used. The Greek symbols
represent groups of constants. Details of the development are left to the

81
 Fitting Data and Solving Equations

reader. (See the Problems section.)

Bp fy" + Yo f" = 69

Og fy"
+ Ba fy"
+ Yq £4" = 6,

vee cece ane vaste (4.25)

we is we _

Oy. 2 fg + Brg fe + Yp.2 fet ~ 6n.2

On Iy-2" + By ty 1 =9n4
Once the f," through f," are found (remember that Ff" and f," are assumed to
be zero), then the solutions for the approximating functions are easily
determined.

MRC
It has been discussed in the literature (e.g., see Hang, 1976) that

RRs ge Se
inaccuracies can arise from the use of cubic splines that can be lessened
with the use of higher order polynomials. Normally, the additional
complexity that is required does not result in that much better of a fit.
Splines are discussed in a number of sources in the numerical analysis
literature, and the interested reader should have no trouble finding a great
deal of additional information.

4.1.6 Relationships from Physical Concepts

As this section ends, an important point needs to be noted. Many

times some good physical basis can be used in determining relationships
between key variables in a given problem. Whenever facing a curve fit
situation, be sure to keep in mind any physics that may be at play. Use that
information to build better approximations.
For example, if relationships are to be found between variables in
any kind of heat exchange element, fluid flow device, or thermodynamic
process, try first to determine the theoretically based relationship(s).
Sometimes the complete relationship(s) can be found directly. At other
times, some empirically based factor enters into the data and modifies the
theoretical performance. For the latter situation, use the known
performance relationships from their theoretical basis and then determine
a functional relation for the empirical factor. An example of this is the
efficiency of a pump, compressor, fan, or turbine. The performance of
these devices is easily described theoretically and modified with a curve
fit efficiency.
This topic is not explored in any depth here. The reason for this is
that the basic texts in heat transfer, thermodynamics, fluid flow, and other
engineering fundamentals deal with the modeling of physical processes at
great length. Keep the concepts covered there clearly in mind during your
work with system simulations.

82
 Fitting Data and Solving Equations

4,2 SOLUTION OF ALGEBRAIC EQUATIONS

4.2.1 Introduction

Other sections of this text deal with how to set up equations to

model thermal systems. Everything from thermodynamics to costs are
addressed in the model building. At some point, a number of equations
must be solved. Depending upon the type of model developed, the number of
equations could range from one to a few thousand. The equation(s) may be
linear or nonlinear in the independent variable(s). There is a possibility
that the equations could be of differential form, particularly if the
transient response of a system is modeled, but this situation will not be
addressed here. Hence, the numerical solution of algebraic equations will
be of primary concern. Even with this restriction in topic, the field is very
broad.
Often the solutions that are needed are those for systems of linear
algebraic equations and those for single and systems of nonlinear
equations. As will be shown, the solution for systems of nonlinear
equations requires the solution for systems of linear equations. so the
latter topic is doubly important.
When solving systems of linear algebraic equations, several
characteristics about the systems are often present. First, the systems
can be very large. There may be several unknowns for each block in a
flowsheet, so not very many blocks are required to result in a large
equation system that must be solved. Second, the systems are often
sparse. This means that only a few of the variables appear in each of the
equations. Using conventional solution techniques can be very inefficient.
The focus here will be on aspects that relate to both of these facts.
Finally, it should be noted that more texts and other literature, as
well as existing software, are available to the reader on this topic than any
other topic covered in this book. This means that only limited, introductory
discussions are realistic here. It also means that additional information
and software are abundant and readily accessible. Use the material here as
an introduction and continue your work by referring to the numerous other
resources availabie.

4.2.2 Solution of Nonlinear Equations

Consider the solution of a single equation as shown in Equation 4.26,

which is assumed to be nonlinear. In other words, find the root(s) of this
equation.

Fix)=y for y=0 (4.26)

Assume that there is only one root in the region of interest. Solutions of
this type arise in many situations. An example is the determination of the
operating point for the fluid flow loop consisting of a pump and pressure

83