Synthesis Lectures on

Engineering, Science, and Technology

Emre Tokgoz

Six Sigma and

Quality Concepts
for Industrial
Engineers
Synthesis Lectures on Engineering, Science,
and Technology
The focus of this series is general topics, and applications about, and for, engineers
and scientists on a wide array of applications, methods and advances. Most titles cover
subjects such as professional development, education, and study skills, as well as basic
introductory undergraduate material and other topics appropriate for a broader and less
technical audience.
Emre Tokgoz

Six Sigma and Quality

Concepts for Industrial
Engineers
Emre Tokgoz
Farmingdale State College
The State University of New York
Farmingdale, NY, USA

ISSN 2690-0300 ISSN 2690-0327 (electronic)

Synthesis Lectures on Engineering, Science, and Technology
ISBN 978-3-031-55739-2 ISBN 978-3-031-55740-8 (eBook)
https://doi.org/10.1007/978-3-031-55740-8

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG
2024

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

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

Paper in this product is recyclable.

Preface

Six sigma and quality improvements are essential ingredients that industrial engineers can
utilize for advancement of systems, and employee training in organizations. An indus-
trial engineer typically learns optimization, simulation, and systems’ design components;
therefore, the value addition of six sigma conceptual knowledge to an industrial engi-
neer for waste reduction is very valuable. Even though an industrial engineer’s essential
training is towards learning about how to improve systems, six sigma provides an orga-
nized methodology of how to implement learned improvements; therefore, six sigma is
a guidance to industrial engineers as an improvement technique that systematically helps
to improve systems through observations, data collection, and analysis. Given that an
industrial engineer is expected to have a developed background in data analytics, simu-
lation, manufacturing, health care, facility layout, and material handling, this book can
be a very valuable resource to both industrial engineers individually and it can also help
companies that would be interested to train industrial engineers on six sigma concepts.
These concepts are particularly helpful in determination and implementation of improve-
ment techniques. Industrial engineers need to understand the nature of systems through
observations, collect data, and analyze the collected data during their duties; therefore,
they naturally take place in six sigma projects; six sigma approach gives these abilities
a structure. This book covers Define, Measure, Analyze, Improve, and Control (DMAIC)
steps of six sigma methodology along with some of statistics essential for advancement
and improvement of systems with the active involvement of industrial engineers. It is
authored based on the extensive project experience of the author during Black Belt Six
Sigma training of university students by working with industry partners that included but
not limited to Pratt & Whitney, Parker Hannifin, Nucor Stainless Steel, Yale New Haven
Hospital, Hartford Hospital, Turner Construction, and ITW Drawform in New York and
Connecticut areas in the United States. Conceptual understanding of six sigma cannot be

v
vi Preface

limited to this book; therefore, additional six sigma project experience is essential for fur-
ther successful completion of a six sigma project after understanding real-life examples
that are given in this textbook.

New York, USA Emre Tokgoz

About This Book

An industrial engineer who completes an undergraduate bachelor’s degree naturally has

a good theoretical foundation. Pairing such knowledge with the theoretical and applied
knowledge in six sigma can result in the ability for an individual to be able to improve
systems in a very effective manner. Industrial engineers are typically trained to collect
and analyze data, learn, and apply improvement techniques, and adopt changes to remove
waste and increase productivity. In addition, given that six sigma requires statistical and
technology knowledge for data collection, interpretation, and analysis, this textbook cov-
ers a part of the associated lean, six sigma, and technology-related concepts that an
industrial engineer can apply. Completion of this book content would help you to develop
an organizational six sigma methodology as an industrial engineer for understanding and
improving systems.

vii
Contents

1 Basic Statistics for Six Sigma Applications for Industrial Engineers . . . . . . . 1

1.1 Basic Probability and Distribution Applications . . . . . . . . . . . . . . . . . . . . . 2
1.2 Basic Statistics Measurement Tools for Six Sigma . . . . . . . . . . . . . . . . . . . 4
1.3 Statistical Distributions with a Discrete Nature . . . . . . . . . . . . . . . . . . . . . . 5
1.3.1 Distributions for Discrete Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.3.2 Normally Distributed Discrete Data . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3.3 Binomial Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.3.4 Poisson Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.3.5 Geometric Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.3.6 Hypergeometric Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.4 Distributions for Continuous Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.4.1 Normally Distributed Continuous Data . . . . . . . . . . . . . . . . . . . . . . 17
1.4.2 Uniform Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.4.3 Exponential Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.4.4 Gamma Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
1.4.5 Weibull Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.4.6 Lognormal Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
1.5 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2 Fundamentals of Six Sigma: Meanings of Waste, Value, and Key
Performance Indicators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.1 Identification of Waste and Definition of Quality . . . . . . . . . . . . . . . . . . . . 35
2.2 Management of Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.3 Meaning of Sigma Term and DMAIC Approach . . . . . . . . . . . . . . . . . . . . 41
2.4 A Methodological Approach Through Lean Principles . . . . . . . . . . . . . . . 45
2.5 Aiming to Achieve Business Excellence Through Measurable,
Attainable and Realistic Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.6 Short Period Improvement Opportunities Through Kaizen Event
Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.7 Structuring Strategy Through Mission and Vision . . . . . . . . . . . . . . . . . . . 50

ix
x Contents

2.8 An Industrial Engineer’s Role in Six Sigma Applications . . . . . . . . . . . . . 52

2.9 Waste Identification Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
2.10 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3 Defining the Six Sigma Project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
3.1 Fundamentals of Define Phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
3.2 Outlining Input, Process, and Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
3.3 Diagram for Supplier, Input, Process, Output, and Control . . . . . . . . . . . . 66
3.4 Structuring Diagrams for Designing Flow . . . . . . . . . . . . . . . . . . . . . . . . . . 70
3.5 Mapping Processes for Business Operations . . . . . . . . . . . . . . . . . . . . . . . . 72
3.6 Diagram of Operations Flow Driven by Business Functions . . . . . . . . . . 73
3.7 Structuring a Chater for the Project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
3.8 Mapping of the Movements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
3.9 Structuring a Tree for Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
3.10 Initial Phase of Value Stream Mapping for Operations . . . . . . . . . . . . . . . 81
3.11 Satisfaction, Dissatisfaction, and Delighters . . . . . . . . . . . . . . . . . . . . . . . . 87
3.12 An Example in Health Care . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
3.13 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
4 Measurements for the Defined Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
4.1 Quantifying Measurements Through Statistical Distributions . . . . . . . . . . 100
4.2 Sampling and Collecting Empirical Data . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
4.2.1 Designing Data Collection Through Non-Random Sampling . . . 101
4.2.2 Strategically Sampling Through Random Choices . . . . . . . . . . . . . 103
4.2.3 Collecting Survey Data and the Associated KPI
Determination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
4.2.4 Organizing and Planning the Data Gathering . . . . . . . . . . . . . . . . . 106
4.3 Designing Graphs and Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
4.3.1 Bar Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
4.3.2 Plotting Time-Dependent Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
4.3.3 Designing Meaningful Pie Charts . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
4.3.4 Plots by Using Box and Whisker . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
4.3.5 Designing Meaningful Pareto Charts . . . . . . . . . . . . . . . . . . . . . . . . 115
4.4 Calculating Number of Operators Needed for Task Completions
and Takt Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
4.5 Analysis of System Measurements and Gage R&R . . . . . . . . . . . . . . . . . . 117
4.5.1 Structuring Gage Repeatability and Reproducibility
Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
4.5.2 Meaning of Bias and Determining It . . . . . . . . . . . . . . . . . . . . . . . . 119
4.5.3 Tendency for Linear Behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
4.5.4 Stable Behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
4.5.5 Ability to Repeat Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Contents xi

4.5.6 Ability to Reproduce Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

4.5.7 Organizing Measurements into a Matrix Form . . . . . . . . . . . . . . . . 126
4.6 Measuring What is Performed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
4.6.1 Measuring Effectiveness of Equipment . . . . . . . . . . . . . . . . . . . . . . 128
4.6.2 Calculation of Defect Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
4.6.3 Successful Part Production Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
4.6.4 Calculation of Defects in a Million . . . . . . . . . . . . . . . . . . . . . . . . . 130
4.7 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
5 Analysis of the Measurements and Project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
5.1 Structuring “W” Questions and How . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
5.2 Structuring Fish-Bone Diagram with Causes and Effects . . . . . . . . . . . . . 136
5.3 Identification of Data and KPI Correlation . . . . . . . . . . . . . . . . . . . . . . . . . 137
5.3.1 Correlation Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
5.3.2 Correlation Calculations by Using the LSM (Least Squares
Method) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
5.3.3 Multivariate LSM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
5.4 Designing Tables for Interaction of Variables . . . . . . . . . . . . . . . . . . . . . . . 153
5.5 KPI Analysis Driven by Time-Dependent Variables . . . . . . . . . . . . . . . . . . 155
5.6 Testing Scenarios Through Hypothesis Generations . . . . . . . . . . . . . . . . . . 156
5.6.1 Designing Intervals to Identify Confidence Levels . . . . . . . . . . . . . 157
5.6.2 CI Formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
5.6.3 Choice of Sample Size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
5.6.4 Structuring Null Hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
5.7 Calculating Variation Using Analysis of Variance . . . . . . . . . . . . . . . . . . . 166
5.8 Structuring Matrix for Severity Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
5.9 Analysis of Efficiency Through Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
5.9.1 Analysis for Improvement of VSM . . . . . . . . . . . . . . . . . . . . . . . . . 169
5.9.2 Analysis Based on Lean Principles . . . . . . . . . . . . . . . . . . . . . . . . . . 171
5.10 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
6 Let’s Improve the System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
6.1 Preventing Mistakes Through Control Mechanisms . . . . . . . . . . . . . . . . . . 176
6.1.1 Control Mechanism Placement at Each Phase . . . . . . . . . . . . . . . . 176
6.1.2 Designing a Control Mechanism for Some Phases . . . . . . . . . . . . 178
6.1.3 Systems’ Without Control Mechanisms and Improvement . . . . . . 179
6.1.4 General Operational Control Mechanisms
and Improvements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
6.2 Improvements Through Sorting, Setting in Order, Shinning,
Standardizing, and Sustaining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
xii Contents

6.3 Restructuring the Design for Improvements . . . . . . . . . . . . . . . . . . . . . . . . . 183

6.3.1 Re-Designing Layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
6.3.2 Redesigning Transportation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
6.3.3 Designing Documents for Project Results . . . . . . . . . . . . . . . . . . . . 186
6.3.4 Training of Staff and Improvements . . . . . . . . . . . . . . . . . . . . . . . . . 187
6.4 Single Minute Exchange of Die (SMED) . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
6.5 Mistake Proofing (Poka Yoke) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
6.6 Time-Dependent Improvements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
6.7 Did the System Improve? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
6.8 Did the Improvements Work? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
6.9 Improvement on Value Stream Map . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
6.10 Improvements by Using Robotics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
6.11 Simulation of Operations and Layout Design with Improvements . . . . . . 195
6.12 Automation of Transportation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
6.13 Possible Applications of Artificial Intelligence (AI) and Machine
Learning (ML) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
6.14 Advanced Material Handling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
6.15 Advancements Through App Development . . . . . . . . . . . . . . . . . . . . . . . . . 203
6.16 Use of Augmented/Virtual Reality (AR/VR) . . . . . . . . . . . . . . . . . . . . . . . . 204
6.17 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
7 Structuring Control for Sustainability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
7.1 Data Collection After Improvements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
7.2 Structuring Charts for Control Observations . . . . . . . . . . . . . . . . . . . . . . . . 210
7.2.1 Control Chart Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
7.2.2 Structuring Continuous Data Control Charts . . . . . . . . . . . . . . . . . . 213
7.2.3 Designing Control Charts for Discrete Data . . . . . . . . . . . . . . . . . . 220
7.3 Out-Of-Control Conditions’ Handling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
7.4 Reporting Outcomes and Completing Project with Documentation . . . . . 224
7.5 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225

Epilogue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
About the Author

Emre Tokgoz Ph.D., is a Visiting Professor at the Farmingdale State College campus of
The State University of New York. Before that, he was the Director and an Associate Pro-
fessor of Industrial Engineering at Quinnipiac University. This book is authored based on
his extensive project experience during Six Sigma Black Belt training of university stu-
dents in conjunction with industry partners, including Pratt & Whitney, Parker Hannifin,
Nucor Stainless Steel, Yale New Haven Hospital, Hartford Hospital, Turner Construction,
Brook & Whittle Ltd., and ITW Drawform. Prof. Tokgoz completed a Ph.D. in mathemat-
ics, another Ph.D. in industrial engineering, an MS in computer science, and an MA in
mathematics at the University of Oklahoma. He also completed an MS and BA in mathe-
matics at the University of Ankara, Turkey. His most recent education is a Master’s degree
in Engineering Management with a concentration on Applied Biomedical Engineering
for Professionals. His research, publication, and teaching areas and interests included
pedagogy, optimization, biomedical engineering, robotics, game theory, network analy-
sis, financial engineering, facility allocation, inventory systems, queueing theory analysis,
supply chain, renewable energy sources, STEM education, machine and deep learning,
and Riemannian geometry.

xiii
Basic Statistics for Six Sigma Applications
for Industrial Engineers
1

Statistics is in the center of meaningful measurable six sigma results. The typical statistics-
related expectations remain at a fundamental level when compared to the typical calculus-
based statistics courses that are taught to industrial engineering students. The following
are the typical expected statistics-related expectations in a six sigma project that will be
covered in this chapter:

• Defining random variable that may have discrete or continuous nature.

• Determining a probability density function.
• Calculating the average, standard deviation, median, mode, data range, and confidence
intervals.
• A variety of distribution types such as normal, uniform, exponential, and Gaussian
with their distribution functions and probabilities.

An event that includes probabilistic outcomes with the involvement of uncertainty and
chance results in the introduction of a statistical variable that is called random variable.
For instance, picking up one manufactured item among a bucket of produced items (sup-
pose there are 1000 items produced) is driven by chance of getting equally likely (i.e.,
uniformly) outcomes that result in one a thousand which is 0.1% chance of getting the
one that we want, and all these items are equally likely chance to be picked up.

• Variables with random nature. These are the variables that represent inputs that have
a random nature.
• Continuous random variable. This is the type of variable that represents an input that
has a continuous nature with randomness.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 1

E. Tokgoz, Six Sigma and Quality Concepts for Industrial Engineers, Synthesis Lectures
on Engineering, Science, and Technology, https://doi.org/10.1007/978-3-031-55740-8_1
2 1 Basic Statistics for Six Sigma Applications for Industrial Engineers

• Discrete random variable. This is the type of variable that represents an input that
has a discrete nature with randomness.

Example 1.1 An example of a discrete random variable is the number of defective parts
collected during the production. This variable is discrete and its randomness depending on
the operation and machine it is collected from introduces uncertainty in the variability. The
rate of production over time, on the other hand, can be considered as a continuous variable as
a part of continuing production flow. The following are other examples of random variables:

• Manufacturing The number of defects in a production.

• Health care The number of nurses working in a hospital per day can be a random variable.
• Transportation service The time it takes to ship goods from location to location can be
a random variable.

Measuring, analyzing, improving, and controlling of a system require statistics to be

implemented with the associated probability, average, and standard deviation calculations.
Such calculations may dramatically depend on the project’s nature and the nature of the
variables to be covered next.

1.1 Basic Probability and Distribution Applications

Random variables (r.v.) require the use of probabilistic calculations for measuring the
likelihood of an event to happen. Given an event X' s with a random nature, the probability
p(x) is ranging from 0 to 100%, hence 0 ≤ p(x) ≤ 1.

• Event X would be impossible to happen if p(x) = 0.

• Event X would happen with certainty if p(x) = 1.
• As covered previously, the probability of a random variable can be discrete or
continuous with the associated distribution.

Distribution. All outcomes of an event X can be summarized as a graph by displaying

data points or by fitting a mathematical distribution depending on the nature of the event.
Continuous distribution. Given a continuous r.v. u, a continuous distribution representing
u is known as probability density function (pdf). The cumulative function that has a
cumulative nature of the pdf is known as the cumulative distribution function. Figure 1.1
shows two graphs with the left-hand side on containing examples of two pdf functions
while the two graphs on the right side are the corresponding cdf graphs to these pdf
functions.
1.1 Basic Probability and Distribution Applications 3

Fig. 1.1 Left image has two examples of pdf while the right two images are the corresponding cdf
to these two functions

Fig. 1.2 The graph of a discrete distribution with scatter points representing the data points and the
line representing the linear model fitting to this data

• Discrete distribution. Given a discrete random variable, a discrete distribution is the

distribution of the r.v. that forms a discrete function that we will call probability mass
function (pmf). The nature of a pmf’s graph would be structured in a way to fit a
discrete structure just like the image below. Figure 1.3 shows an example of scatter
plot representing a discrete data set with the straight line representing the linear model
fit to this data (Fig. 1.2).
4 1 Basic Statistics for Six Sigma Applications for Industrial Engineers

Fig. 1.3 Triangular distribution of rolling two dices

1.2 Basic Statistics Measurement Tools for Six Sigma

Basic statistics knowledge and understanding are needed at the most fundamental level
for understanding and applying six sigma concepts that include the following:

• Average (μ). Also known as the mean or expected value and it is the average value of
discrete data. For continuous data, the calculations require knowledge of integrals.
• Range of a data set. A data set’s range is the difference between the maximum and
minimum data points of the set.
• Mode value. This is the data point that exists the most in comparison to the other data
points.
• Median value. The median value is the middle value in a data set when the data set
values are ordered from the largest to smallest or vice versa.
• Standard deviation (σ). The standard deviation is one way to measure deviations from
the standard (or average value). Standard deviation allows us to understand the nature
of the data in comparison to the mean value. The closer the values are to the mean,
the smaller the standard deviation would be.
• Variance (σ2 ). Mathematically, variance is the square of standard deviation in which
case it helps to measure how much the data varies from the average.

These terms, particularly average and standard deviation, are frequently used in six sigma
applications.
1.3 Statistical Distributions with a Discrete Nature 5

1.3 Statistical Distributions with a Discrete Nature

We consider discrete distributions when the data is discrete. The distribution of such a
data set can be represented in different ways, including mathematical functions and visual
graphs, charts, etc.

Example 1.3 Production of a particular tool data collected from different machines is an
example of discrete data and a discrete random variable represents such data. The quality
of an item in this production can be categorized into good or bad.

Given two outcomes that can be success or failure, the probability of success for such
a discrete random variable is
Successful attempts
Prob (success) =
Number of attempts

Example 1.4. Suppose our goal is to determine the low-quality items produced by a machine
with the assigned number M6428 with the discrete random variable representing the number
of items produced by this machine. Let

p: Probability of a M6428 to produce low − quality products.

x: The number of low − quality items produced.

s: Number of low − quality items observed during one shift.

w : Number of good quality items observed during one shift.

It is possible to calculate the percentage of defective items in the overall production

by calculating the simple formula:
s
p=
s+w
The chance of getting a good quality item is
s w
1−p=1− =
s+w s+w

Definition of Probability Mass Function (pmf). A function representing a discrete ran-

dom variable X is called probability mass function. The discrete nature of the input is
6 1 Basic Statistics for Six Sigma Applications for Industrial Engineers

the main factor that differs from the continuous version. In the case for calculating the
probability of X = a given that we know the pmf to be f(x), f(a) = P(X = a).

Example 1.2 The following are examples of pmf for different applications:

• Manufacturing. In the case of determining three number of defects produced by a specific

machine, we define the random variable X to be the number of defects produced by this
specific machine and calculate the probability of having three defects to be f(3) = P(X
= 3) by using f(x) as the pmf.
• Health care. Suppose the number of patients treated by a specific doctor for a certain
type of illness is the random variable X. Finding the probability of three patients treated
would be f(3) = PX = 3) which requires the use of pmf.
• Transportation service. Suppose the number of trucks used for transportation of a certain
store is the random variable X. Finding the probability of three trucks to be used for a
certain shipment would be f(3) = PX = 3) which requires the use of pmf.

Cumulative Mass Function (cmf). A cumulative mass function is the accumulated total of
the probabilities of a discrete random variable X to determine the accumulated probability
up to a certain value of X. The following formula can be used for calculating cmf:

∑
k
P(X ≤ k) = P(X = i) = f (1) + f (2) + · · · + f (k)
i=1

Example 1.3 Suppose we are interested in identifying the probability of a machine pro-
ducing at most three defects based on a pre-determined pmf f(x). This probability can be
calculated as follows:

P(X ≤ 3) = P(X = 1) + P(X = 2) + P(X = 3) = f (1) + f (2) + f (3)

The following would be covered in this chapter for discrete distributions:

• Discrete distribution of a data set.

• Normal distribution for discrete data sets.
• Binomial distribution.
• Poisson distribution.
• Geometric distribution.
• Hypergeometric distribution.
1.3 Statistical Distributions with a Discrete Nature 7

1.3.1 Distributions for Discrete Data

A discrete data set S = {d1 , d2 , . . . , dm } would naturally have a discrete distribution with
statistical measurements used in real life for extracting meaningful outcomes from the
data. The following are some these measurements:

• Average or mean value (μ):

d1 + d 2 + . . . + d m
μ=
m

Excel Formula: = Average(S)

• Range (R):

R = dmax − dmin

Excel Formula: = Max(S) − Min(S)

where
dmax : Data set’s maximum value.
dmin : Data set’s minimum value.
• Mode. The most frequent data appearing in the data set S would be the mode of the
data.

Excel Formula: = Mode(S)

• Median. The sorted version of set S (either from largest to smallest or smallest to
largest) would have the middle value to be the median value of S.

Excel Formula: = Median(S)

• Standard Deviation (σ). Assuming that the average values in the data set S is μ, stan-
dard deviation σ is way to measure how much data points vary from μ. The standard
deviation formula for a collected data set is as follows:
/
(d1 − μ)2 + (d2 − μ)2 + . . . + (dm − μ)2
σ =
m−1

Excel Formula: = STDV .S(S)(when the data set S is sample)

= STDV.P(S)(when the data set is the population)

8 1 Basic Statistics for Six Sigma Applications for Industrial Engineers

• Variance (σ2 ). Variance is another measure used for calculating the deviation from
standard and it is the square of the standard deviation:

(d1 − μ)2 + (d2 − μ)2 + . . . + (dm − μ)2

σ2 =
m−1

Excel Formula: = VAR.S(S) (when the data set S is a sample)

= VAR.P(S) (when the data set S is the population)

Example 1.5 A data analyst collects the following data representing the number of mistakes
made in the collected data on the four data sets A, B, C, and D:

Data A B C D
Row 1 5 10 16 2
Row 2 15 250 120 240

Row 1 is the number of mistakes observed in the data by the analyst while row 2 is
the total number of observations. As an example, Row 1 value of A is 5 indicating that
there are 5 mistakes among the 15 observations listed on Row 2.
Some of the statistical measurements that can be applied on the data set includes the
following:

a. Average (μ).

5 + 10 + 16 + 2 33
μ= =
4 4

= 8.25 number of mistakes observed per data set

Excel Formula: = Average(A2 : D2)

The following is the interpretation of this calculation:

An average of 8.25 mistakes per data is observed among the four data sets observed.
The average can be calculated in other ways as well by using the total number of
observations.
1.3 Statistical Distributions with a Discrete Nature 9

5 + 10 + 16 + 2 33
μ= =
15 + 250 + 120 + 240 625

= 0.0528 number of mistakes per data collected

The interpretation corresponding to these calculations is as follows:

An average of 5.28% mistakes are determined per observation given in the four data
sets.

b. Range.

R = dmax − dmin = 16 − 2 = 14

Excel Formula: = Max(Row1) − Min(Row1)

c. Mode. This data set is a small data set with no repeating values among 5, 10, 16, and
2, therefore there is no mode of this data set entered in a blank cell would help to find
the mode easily.
d. Median. After ordering 5, 10, 16, and 2 from the largest to smallest (16, 10, 5, 2),
the data value in the middle can be either chosen as 5 or 10 since there is no middle
value. The median value in such a case is determined by averaging the two values in
the middle which would be 7.5. The following Excel formula entered in a blank cell
would help to find the median easily.

Excel Formula: = Median(A2 : D2)

e. Standard Deviation (σ).

/
(d1 − μ)2 + (d2 − μ)2 + . . . + (dm − μ)2
σ =
m−1
/
(5 − 8.25)2 + (10 − 8.25)2 + (16 − 8.25)2 + (2 − 8.25)2
=
4−1
= 6.1305 number of mistakes per data set

Excel Formula: = STDV .S(A2 : D2)

f. Variance (σ2 ).

σ 2 = (6.1305)2 = 37.58

Excel Formula: = VAR.S(A2 : D2); VAR.S is typically used for calculations

10 1 Basic Statistics for Six Sigma Applications for Industrial Engineers

1.3.2 Normally Distributed Discrete Data

A discrete data set’s distribution that has a bell-shaped structure would plot the data set’s
discrete probability density function.

Example 1.6 Suppose we want to find the probability distribution of two dices’ total. In
the following, dice total gives the total of two dices, and we divide it by two to normalize
to the value of six as the benchmark value to compare with the probability distribution of
three dices total in what follows next. The Prob column shown in Fig. 1.3 is the associated
probability value for the dice total and the right-most section displays the possible outcome(s)
corresponding to the row. As it can be seen in the figure below, the output displayed in the
y-axis is the probability of this distribution which has a triangular shape. This allows us to
place all 12 possible outcomes in compartments to be able to easily calculate the associated
probabilities.

According to the law of large numbers, we get closer and closer to the true mean value
as the number of data points increases. In addition, the central limit theorem indicates that
the increase in this sample data size allows to have an approximately normally distributed
data set. As an example, if we increase the data size from the total of two dices to three
dices and normalize it to the input range of six (just like in the previous example for two
dices’ total), then Fig. 1.4 is the data representation of the data set. In this case, the total
of three dices range between 1 and 18 on which the probabilities and the associated are
also listed.
If we convert this data into the graph format, we attain the distribution shown in
Fig. 1.5. As it can be seen on the graph, the triangular-shaped distribution changed to a
distribution that has somewhat similarity to normal distribution.

Fig. 1.4 The distribution of three rolled dice

1.3 Statistical Distributions with a Discrete Nature 11

n=3
0.14000
0.12000
0.10000
0.08000
0.06000
0.04000
0.02000
0.00000
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00

Fig. 1.5 The probability distribution of the data in Fig. 1.4

1.3.3 Binomial Distribution

Binomial distribution is one of the distributions that allows sampling with replacements.
Bernoulli trials are used in experimenting that is expected to satisfy the following:

• Experiment consists of n trials.

• Each trial has either success or failure as the outcome.
• The probability of success p is constant and does not change during any trial.
• The trials are independent.

A binomial random variable X consists of n independent Bernoulli random variables hav-

ing p as the parameter. The pmf, mean value, and standard deviation of X corresponding
to the Binomial distribution are as follows:
⎧( )
⎪
⎨ n px (1 − p)n−x if x = 0, 1 . . . , n
Pmf. f (x) = x
⎪
⎩
0 otherwise

Mean. μ = np

√
Standard Deviation. σ = np(1 − p)

where
(n) n!
=
x x!(n − x)!
12 1 Basic Statistics for Six Sigma Applications for Industrial Engineers

Fig. 1.6 The probability distribution of a data set

Example 1.7 An automated machine is determined to have 96% defect-free production

during a specific month. Assuming that the number of trials was five, we can determine the
probability of having four good products and one defect during the production as follows:
( )
5
P(X = 4) = 4 0.96 (1 − 0.96)
4 5−4
= 5 (0.96)4 (0.04)1 = 16.987%

Example 1.8 Given that real-life situations do not always allow a distribution to fit perfectly
to a data set, Fig. 1.6 shows an example of that data that may be data collected in real life
where a Binomial distribution with n = 20 and p = 0.5 can be a good fit to the data.

1.3.4 Poisson Distribution

Poisson distribution is used for modeling number of events that occur when there are
many opportunities for the event but a small probability that the event occurs on each
opportunity. The pmf, mean value, and standard deviation of X corresponding to the
Poisson distribution are as follows:
{ −λ x
e λ
if x = 0, 1, . . .
Probability mass function. f (x) = x!
0 otherwise

Average. μ = λ

√
Standard Deviation. σ = λ
1.3 Statistical Distributions with a Discrete Nature 13

Example 1.9 We can define the following:

X: The number of arrivals to a hospital during each 15-min period.

λ: Mean number of arrivals to the hospital every 15 min.
In addition, for instance, Poisson distribution can be used for modeling the number of
expected calls per hour for a call center to determine call center representatives to be
employed during specific hours.
The Poisson distribution can be used to approximate the binomial distribution when
the probability is small and n is large.

1.3.5 Geometric Distribution

Geometric distribution is used for modeling discrete variables when the number of trials
before the first success or failure are modeled by using Bernoulli distribution. The pmf,
mean value, and standard deviation of X corresponding to the geometric distribution are
as follows:
⎧( )
⎪
⎨ n px (1 − p)n−x if x = 0, 1 . . . , n
Pmf. f (x) = x
⎪
⎩
0 otherwise
1−p
Mean. μ = p

√
1−p
Standard Deviation. σ =
p

Example 1.10 Suppose we observe a machine to be able to identify the first failure of the
machine. If the probability of success is 0.5 then the probability of failure during the 11th
trial by using geometric distribution would be the following:

f (11) = 0.5(1 − 0.5)11 = 0.512 = 0.00024414

Please note that identification of the use of geometric distribution is essential prior to
using it to make sure that it is a sense-making approach for the case being considered.
14 1 Basic Statistics for Six Sigma Applications for Industrial Engineers

1.3.6 Hypergeometric Distribution

Sampling without replacement is used for hypergeometric distribution with the following
needed for the formulation:

• N. Size of the finite population.

• D. The number of items that fall into a group that is of interest.
• n. The size of a random sample drawn from the population without replacement.
• X. The number of items in the sample that belong to the class of interest; it is drawn
without replacement.

If X is a random variable respecting a hypergeometric distribution, then we have the

following pmf, mean value, and standard deviation for it:
⎧ ⎛ ⎞⎛ ⎞
⎪
⎪ D N − D
⎪ ⎝ ⎠⎝
⎪ ⎠
⎪
⎪
⎨ x ⎛ n⎞− x if x = 0, 1 . . . min(n, D)
Pmf. f (x) = N
⎪
⎪ ⎝ ⎠
⎪
⎪ n
⎪
⎪
⎩
0 otherwise

Mean. μ = nD
N

/
( )( N −n )
Standard Deviation. σ = nD
N 1− D
N N −1

where
( )
N N!
=
n n!(N − n)!

Next, we cover a variety of continuous distributions that can be used in six sigma
applications.

1.4 Distributions for Continuous Data

As the name itself states, any data that has a continuum is considered as continuous and
such data requires real number outcomes noting that the contrary would be requiring
discontinuity on the data set that forces the data have discrete nature. Simple examples
of continuous data are measurements in length, speed, energy, production rate, etc. There
are many continuous random variables in manufacturing and healthcare settings such as
1.4 Distributions for Continuous Data 15

height, pressure, weight, mass, temperature, density, volume, and distance. Some of the
continuous distributions used in research and real-life applications are as follows:

a. Normal distribution.
b. Uniform distribution.
c. Exponential distribution.
d. Gamma distribution.
e. Weibull distribution.
f. Lognormal distribution.

If f(x) is the pdf of a continuous random variable x, then we can calculate the following:
Probability Density Function (pdf) Definition. A function f(x) is pdf if the following
holds:

1. f(x) is non-negative (i.e., f(x) ≥ 0). This simply means that the graph of the function
needs to be either on or above the input axis.
2. The total area under the curve f(x) is 1 (or 100%). This can be represented
mathematically as follows:

∫∞
f (x)dx = 1
−∞

Cumulative Distribution Function (cdf). A cdf F(X) is the accumulated total of

probabilities by using the pdf f for a given continuous random variable X. Hence

F(X ) = P(X ≤ x) = f (X ≤ x)

Probability of a continuous random variable. If f(x) is a continuous distribution, then

the probability of x to be between two values of a and b (i.e., likelihood of x to be on the
interval [a,b]) can be calculated by using the following formula:

∫b
P(a ≤ x ≤ b) = f (x)dx
a

which is a number between 0 and 1 that represents the area between f(x) and the input
axis.
16 1 Basic Statistics for Six Sigma Applications for Industrial Engineers

• Mean/average value of a continuous random variable.

∫∞
Mean(μ) = xf (x)dx
−∞

• Standard deviation of a continuous random variable.

⎡
| ∫∞
|
|
Standard Deviation (σ ) = √ (x − μ)2 f (x)dx
−∞

• Variance.
∫∞
Variance (σ ) =
2
(x − μ)2 f (x)dx
−∞

Example 1.11 A jean manufacturer believes that the time it takes to manufacture 10 billion
jeans to its customers is given by the formula:
{ ( )
6 t − t 2 if 0 ≤ t ≤ 1
h(t) =
0 if t>1

where t is time measured in years. We first need to prove that h(t) is a pdf. The graph
of the function is shown in Fig. 1.7 that clearly shows that it is non-negative (which can
also be seen easily through algebraic calculations).
Next, we show that the area between h(t) and the t-axis is 1:

Fig. 1.7 An example of a probability distribution function

1.4 Distributions for Continuous Data 17

∫∞ ∫0 ∫1 ∫∞
h(t)dt = h(t)dt + h(t)dt + h(t)dt
−∞ −∞ 0 1
∫1 ∫1 ⎡ 2 ⎤1
( ) t t3
=0+ h(t)dt + 0 = 6 t − t 2 dt = 6 −
2 3 0
0 0
[ ]1
= 3t 2
− 2t 3 0 =1

Therefore h(t) is a pdf.

• Suppose now we want to determine the probability of the manufacturer to ship the
items within the first quarter of the year. Hence, 0 ≤ t ≤ 0.25 and we need to calculate:
0.25 ( )
P(0 ≤ t ≤ 0.25) = ∫ 6 t − t 2 dt
0
[ ]0.25
= 3t 2 − 2t 3 0 = 3(0.25)2 − 2(0.25)3 = 0.15625

This indicates that the probability of the manufacturer to ship the mentioned 10 million
items within the first 3 months of the year is 15.625%.

1.4.1 Normally Distributed Continuous Data

A normal distribution’s probability density function is structured like a bell as shown on

the figure below. The mean value μ and the standard deviation σ are the main drivers of
a normal distribution and the pdf corresponding to a normal distribution has the following
function form:
∫x (x−μ)2
1
f (x) = √ e 2σ 2 dx
σ 2π
−∞

An easy way to calculate the probability density function values is by using Excel
noting that f(x) has a complicated nature. There are other software packages that can be
used for calculating the associated probability and pdf calculations that is left to the reader
to find out and execute it. The Excel function that can be used for such calculations is as
follows:

Excel function. = NORM.DIST(x, mean, standard deviation, TRUE or FALSE)

with the associated explanations as follows:

18 1 Basic Statistics for Six Sigma Applications for Industrial Engineers

Fig. 1.8 Probability distribution function (on the left) and probability mass function (on the right)
of a normal distribution

• The syntax corresponding to the normal distribution is NORM.DIST.

• The distribution value is x used for function value calculations.
• STDV.S() function in Excel is used for sample data’s standard deviation calculations.
• TRUE or FALSE. The cumulative nature of the data depends on the phrases TRUE
and FALSE. If we want to find the probability value for x, then we need to enter
FALSE otherwise TRUE needs to be entered for cumulative calculations.

The image on the left below is an example of a normal pdf while the image on the
right side below is the cdf corresponding to this normal pdf (Fig. 1.8).
The pdf representation of a data set is a representative of a data set that can be strong or
weak. Clearly a strong fit to the data set can result in strong outcomes while the opposite
of this also holds. A pdf is useful for calculating probability of a number between two
values. One useful application of such pdf calculation is hypothesis testing.

Example 1.12 A manufacturer develops a part during production that is normally distributed
between 20 and 180 items per day on an average of 70 items used per day and standard
deviation of 10 items used per day. The probability distribution of the function is shown in
Fig. 1.9. The probability to be less than x items’ usage per day can be calculated as follows:

Step 1. The first column can consist of input values 10 through x in Excel while the
second column values can be the NORM.DIST(input) values. For instance, cell B3 = 20
has the output value C3 = NORM.DIST(B3, 80, 20, FALSE).
Step 2. Highlight all normal distribution values and plot the data to be able to view the
associated graph. For instance, if x = 160 then we get the following graph:
If we calculate all function values SUM(C3:C173) then we end up calculating cumulative
probability values corresponding to input values 20 through 180. We can compare this
1.4 Distributions for Continuous Data 19

Probability Distribution of Part Production

0.025

0.02
Probability Values

0.015

0.01

0.005

0
0 20 40 60 80 100 120 140 160 180

Part Production Qantities

Fig. 1.9 The probability distribution example of a part production

value with the actual cumulative probability by calculating NORM.DIST(180, 80, 20,
TRUE) in which case we should be getting a value that is very close to one for this
function. In this case, the NORM.DIST function approximates the normal distribution.
Given that normal distribution values range from -∞ to ∞, this result is not surprising.
One question that is essential to ask is whether the collected data has a normal dis-
tribution nature. There are a variety of normality tests for a statistical distribution that
are not covered here, however two of the well-known tests of normality are the Kol-
mogorov–Smirnov test and the Shapiro–Wilk test which are widely used methods to test
the normality of the data. In the case when data is collected, central limit theorem can be
used which indicates the possibility of having an approximately normally distributed data
set if the collected data set is sufficiently large under certain conditions. There are handy
software packages such as Minitab, SPSS, and MATLAB that can be used for normality
tests of data sets.

1.4.2 Uniform Distribution

A uniform distribution is the most basic continuous distribution. As the name suggests,
the probability of the random variable of a uniform distribution has an unchanging nature,
meaning the probability does not change in a certain interval when the input changes. The
following is the general formula used for a uniform distribution:
{
1
if a ≤ x ≤ b
f (x) = b−a
0 if otherwise
20 1 Basic Statistics for Six Sigma Applications for Industrial Engineers

Fig. 1.10 Uniform distribution example

For instance, if a = 2, b = 7 then we know the distribution’s function in which case

for any x between 2 and 7, the probability is 20% and this probability makes the pdf a
horizontal straight line as shown in Fig. 1.10.
{
1
if 2 ≤ x ≤ 7
f (x) = 7−2
0 if otherwise

Example 1.13 Suppose a machine produces between 10 and 100 car parts in an hour with
each production quantity having the same chance of occurrence.

The pdf for this statement has uniform distribution nature:

{ {
1
if 10 ≤ x ≤ 100 1
if 10 ≤ x ≤ 100
f (x) = 100−10 = 90
0 otherwise 0 otherwise

Now we can calculate a specific probability such as the likelihood of producing

anywhere between 35 and 55 car parts as follows:

∫∞ ∫55 [ x [55
1 55 35 20 ∼
f (x)dt = dx = = − = = 22.22%
90 90 35 90 90 90
−∞ 35

Therefore, that probability of producing anywhere between 35 and 55 car parts is

22.22%.

• Mean. We can also easily calculate the mean value as follows:

∫∞ ∫
100 ⎡ 2 ⎤100
x x 1002 102
xf (x)dt = dx = = − = 55 car parts per hour
90 180 10 180 180
−∞ 10
1.4 Distributions for Continuous Data 21

Hence, there will be an average of 55 car parts per hour to be produced by the specific
machine.

• Standard Deviation. We can also easily calculate the standard deviation value as
follows:

∫∞ ∫
100 ∫
100
(x − 55)2 1 ( )
(x − μ) f (x)dt =
2
dx = x2 − 110x + 3025 dx
90 90
−∞ 10 10
⎡ ⎤100
x3 110x2 3025x
= − + = 675 car parts per hour
270 180 90 10

Hence, there will be 25.98 car parts produced per hour as the deviation from the
standard by the specific machine.

1.4.3 Exponential Distribution

The following are expected in an exponential distribution:

• All outcomes are either positive or zero.

• Smaller outcomes are most likely to occur.
• The value of the largest possible outcome cannot be known.

The pdf of the exponential distribution has the following form:

{
ke−kx if 0 ≤ x
f (x) =
0 otherwise

where k = 1
μ is a constant.

Example 1.14 Suppose that the mean number of patients treated in a specific clinic is known
to be 5 patients per hour with an exponential distribution nature. We want to identify the
probability of treating 4 patients per hour.

• The pdf of the distribution is as follows:

{
1 − 51 x
5e if 0 ≤ x
f (x) =
0 if otherwise
22 1 Basic Statistics for Six Sigma Applications for Industrial Engineers

Fig. 1.11 Example of an exponential distribution

Figure 1.11 shows the corresponding graph of the pdf without the zero section for
negative values.
Hence, the probability of treating 4 patients can be calculated:

∫∞ ∫4 [ [4
1 −1x
e 5 dx = −e− 5 x = 1 − e− 5 ∼
1 4
f (x)dt = = 55.07%
5 0
−∞ 0

Hence, the probability of treating 4 patients within 1 h is 55.07%.

1.4.4 Gamma Distribution

Gamma distribution has been used for modeling several different systems including but
not limited to the following:

• Service times in queueing and service systems.

• Equipment lifetimes.
• Times between purchases of nondurable goods.
• Machine repair times.

The pdf of Gamma distribution is the following function:

⎧
⎨ xα−1 e −x
β
f (x) = β α ⎡(α) if 0<x
⎩ 0 otherwise

where α and β are what are so-called shape parameters that are known to be constant
values and ⎡(α) = (α − 1)! The shape parameters help to identify the shape of the
Gamma distribution that changes as these parameters change.
Excel function Gamma.Dist(x, α, β, True or False) can be used for calculating the
Gamma distribution values.
1.4 Distributions for Continuous Data 23

Gamma Distribution
0.0450

0.0400

0.0350

0.0300

0.0250

0.0200

0.0150

0.0100

0.0050

0.0000
0 1 2 3 4 5 6 7 8 9

Fig. 1.12 Example of a Gamma distribution

Example 1.15 Under certain conditions, the shape parameters of a Gamma distribution
can be determined based on the data collected. The following is an example of such a data
collection and the fitted distribution:

Input Gamma value Probability

x = 10 0.0421 4.21%
x = 11 0.0309 3.09%
x = 12 0.0223 2.23%
x = 13 0.0159 1.59%
x = 14 0.0112 1.12%
x = 15 0.0078 0.78%
x = 16 0.0054 0.54%
x = 17 0.0037 0.37%

In this data, for instance, the probability of x = 10 to happen is 4.12% when α = 3

and β = 2. The following is the Gamma distribution’s graph corresponding to the given
data above. Realistically, a perfect fit as shown in Fig. 1.12 is not expected.

1.4.5 Weibull Distribution

Weibull distribution is determined to fit lifetime data existing in reliability studies. The
pdf of Weibull distribution is the following function:
24 1 Basic Statistics for Six Sigma Applications for Industrial Engineers

⎧ ( )α
⎨ αxα−1 e −x
β
f (x) = β α ⎡(α) if 0<x
⎩ 0 otherwise

where α and β are what are so-called shape parameters that are known to be constant
values. The shape parameters help to identify the shape of the Weibull distribution that
changes as these parameters change just like the Gamma distribution.
Excel function Weibull.Dist(x, α, β, True or False) can be used for calculating the
Gamma distribution values.

Example 1.10 Suppose the following data is observed to be the lifetime decay data of a
machine:

Input (years) pdf output

x=2 0.135335283
x=3 0.049787068
x=4 0.018315639
x=5 0.006737947
x=6 0.002478752
x=7 0.000911882
x=8 0.000335463
x=9 0.00012341

where both shape parameter values are one. The pdf output x = 2 in this setting is the
probability of the machine to produce defective items during the second year. Hence,
Fig. 1.13 shows the graph of the distribution corresponding to the provided data.

1.4.6 Lognormal Distribution

If a random variable Y has a normal distribution, then X = eμ+σ Y has a lognormal

distribution. Lognormal distribution is frequently used in stock pricing changes and life-
time of equipment-related calculations. The pdf of lognormal distribution is the following
function:
⎧ 2
⎨ √1 x − (lnx−μ)
∫ e 2σ 2 dx if 0 < x
f (x) = xσ 2π −∞
⎩
0 otherwise
1.5 Questions 25

Weibull Distribution
0.16

0.14

0.12

0.1

0.08

0.06

0.04

0.02

0
0 1 2 3 4 5 6 7 8 9

Fig. 1.13 An example of a Weibull distribution

The Excel function Lognormal.Dist(x, μ, σ, TrueorFalse) can be used for calculating

lognormal distribution values.

1.5 Questions

Question 1.1 Suppose a nurse collected the data displayed in the table below representing
the patients that required to revisit a hospital treated by doctors A, B, C, and D. Row 1
represents the patients that were required to revisit the hospital among the total number of
patients treated displayed in row 2.

Data A B C D
1 5 10 15 20
2 52 55 112 88

Given this information, calculate the following statistical measurements:

a. Average.
b. Standard deviation.
c. Variance.
d. Median.
e. Mode.
f. Range.
26 1 Basic Statistics for Six Sigma Applications for Industrial Engineers

Question 1.2 The production of a particular part is determined to be normally distributed

between 30 and 260 items per month with an average and standard deviations of 90 and 60
items used per month, respectively. Calculate the probability that less than 165 items can be
produced by the manufacturer by using this part.

Question 1.3 Find an example of a continuous random variable that is not listed in this
chapter as an example of a continuous random variable and explain its application in real
life. Where is it used and how it is used? Please make sure to cite your references.

Question 1.4 Determine mathematically whether the following function is a pdf or not.
{
2t − t 2 if 0 ≤ t ≤ 2
h(t) =
0 if t > 2

Question 1.5 Find and explain an example of a normal distribution in real life. State its
pdf and calculate the associated mean and standard deviation values. Please make sure to
cite your references.

Question 1.6 Find and explain an example of a continuous normal distribution in real life.
State its pdf and calculate the associated mean and standard deviation values. Please make
sure to cite your references.

Question 1.7 Find and explain an example of a uniform distribution in real life. State its
pdf and calculate the associated mean and standard deviation values. Please make sure to
cite your references.

Question 1.8 Find and explain an example of an exponential distribution in health care
or manufacturing. State its pdf and calculate the associated mean and standard deviation
values. Please make sure to cite your references.

Question 1.9 Find and explain an example of a Gamma distribution in health care or
manufacturing. State its pdf and calculate the associated mean and standard deviation values.
Please make sure to cite your references.

Question 1.10 Find and explain an example of a Weibull distribution in health care or
manufacturing. State its pdf and calculate the associated mean and standard deviation values.
Please make sure to cite your references.

Question 1.11 Find and explain an example of a lognormal distribution in health care or
manufacturing. State its pdf and calculate the associated mean and standard deviation values.
Please make sure to cite your references.
1.5 Questions 27

Question 1.12 Find and explain to a Poisson distribution in health care or manufacturing.
State its pmf and calculate the associated mean and standard deviation values. Please make
sure to cite your references.

Question 1.13 Find and explain an application of binomial distribution in health care
or manufacturing. State its pmf and calculate the associated mean and standard deviation
values. Please make sure to cite your references.

Question 1.14 Find and explain to a geometric distribution in health care or manufacturing.
State its pmf and calculate the associated mean and standard deviation values. Please make
sure to cite your references.

Question 1.15 Find and explain hypergeometric distribution application in health care
or manufacturing. State its pmf and calculate the associated mean and standard deviation
values. Please make sure to cite your references.

Question 1.16 The following information is attained for a healthcare project:

Using the information on the graph, please answer the following:

a. Given the information on the graph, what information is missing to be able to

understand the graph? Please explain.
b. Do you think the graph is a good fit to the data? Please explain.
Fundamentals of Six Sigma: Meanings
of Waste, Value, and Key Performance
2
Indicators

Waste is one of the biggest outcomes of consumption in today’s world. Given that the
type of waste and what can be considered as waste is driven by the customer’s opinion
and additional standards that are required for production of goods and services, the main
aim of Six Sigma methodology is increasing customer satisfaction while making sure that
the improvements made are sustainable for the designed six sigma solution to not return
to its pre-improvement stages. Waste reduction along with quality improvement are two
of the main drivers of such an approach; some of the other waste reduction approaches
in six sigma include but not limited to delivery time reduction, and lowered costs. The
Six Sigma and Lean methodology explained in this book for industrial engineers focuses
on the typical education of an industrial undergraduate or graduate student. Some of the
fundamentals that are used throughout this book are the following:

• Input: Anything needed for completing products and services; These are the essential
ingredients that are fed into processes to complete such products and services.
• Output: The outcome of processed input(s). The preparation of the output is expected
to be driven by the customer expectations and the standards expected for fulfilling the
output requirements.
• Waste: A process step, product, or service may not be adding the desired specifications
that the customer was expecting therefore it becomes a waste due to the mismatch with
customer expectations. In such a case, the customer does not want to pay for waste.
• Value: Anything that matches with the customer’s expectations that is worth to invest
from customer’s perspective.
• Value-Added Activities: (VA): Activities that add value to the output of the product,
service, or process step that are not considered as waste.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 29

E. Tokgoz, Six Sigma and Quality Concepts for Industrial Engineers, Synthesis Lectures
on Engineering, Science, and Technology, https://doi.org/10.1007/978-3-031-55740-8_2
30 2 Fundamentals of Six Sigma: Meanings of Waste, Value, and Key …

Fig. 2.1 Value added, business value added, and non-value-added activities calculations in a process
[Ref Six Sigma Engineering [25]]

• Non-Value-Added Activities (NVA): Customers are not willing to pay for these
activities that are not adding value to the product or service.
• Business (or Enterprise)-Value-Added Activities (BVA or EVA): Actions that are
necessary to complete the expected product, service, or process step regardless of
customer’s willingness to pay for them.
• Process: Actions, steps, tasks, and operations taking place between the input and
output.
• Lead Time. It is the time for completion of production or services from the beginning
to the end.
• Critical to Quality (CTQ). There are certain quality specifications which are essential
to be fulfilled for attaining the desired output; Such specifications are typically driven
by the customer and quality standards expected on the output.

The six sigma foundations outlined in this work rely on the measured value added, non-
value added, and business value added activities therefore identifications of these activities
play a crucial role in the success of a six-sigma project. For instance, Fig. 2.1 below
demonstrates the VA, NVA, and EVA/BVA between the input and output as the value
analysis for the completion of a product or service as the waste walk by using the value
analysis with percentage calculations in Fig. 2.1 below [25].
The production of an item or service starts with the essential inputs needed for the
associated processes to complete tasks. In this case, the waste can initiate at any stage,
including the input, steps of the processes, and the output. A Waste Walk demonstrated in
Fig. 2.2 is one way to represent all the activities taking place in the scope of the project to
identify waste or NVA activities by breaking the process steps down into the NVA, BVA,
and VA. This approach relies on the judgement call of the six-sigma team to identify
2 Fundamentals of Six Sigma: Meanings of Waste, Value, and Key … 31

Fig. 2.2 Three stages of a Waste Walk with reduced waste in each step [25]

these activities which should be driven by customer provided and project driven informa-
tion. As shown in Fig. 2.2, three types of activities/work planned to be completed from
the beginning to the end of the processes comprehensively cover the entire waste walk.
On this figure, the first step of the value identification indicated the existence of three
VAs, three BVAs, and two NVAs from the beginning to the end throughout the lead time.
After determining and applying improvements during a six-sigma project, the line below
the first step shows the reduction on NVAs while EVAs and VAs remaining same in the
revised Waste Walk. The third Waste Walk on the bottom of Fig. 2.2 shows the eliminated
non-value-added activities throughout the lead time. This third stage of improvement is
faster in completion of the expected tasks; One essential ingredient during the elimina-
tion of waste is paying close attention to quality levels that should be either kept same
or improved when compared to the pre-improvement stage. In certain cases, changing the
nature of VA and BVA can also make a change in the attained quality outcomes even if
NVAs cannot be eliminated. Breaking out waste walk steps into processes and subpro-
cesses depend on the experiences of the six sigma group members and customer, therefore
a NVA determined by one person may not be identified by another person in a process as
a waste. A detail-oriented approach with six sigma project experience and knowledge of
improvement techniques plays crucial roles on the successes of a six sigma projects.
As the name itself suggests, a Key Performance Indicator (KPI) is a term often used for
outlining the key or essential elements that impact the performance of a product, service,
or system. This is a term used for outlining a single or a group of activities impacting
32 2 Fundamentals of Six Sigma: Meanings of Waste, Value, and Key …

the performance of the system; A six-sigma project’s KPIs are typically driven by the
customer’s expectations and needs from the project. One essential aspect is the measure-
ments of the KPIs to be able to quantitatively display the changes on the performance
of the system based on the associated KPI after improvements are applied; In particular,
quantitative outcomes are expected as a part of black belt six sigma projects by using
advanced statistical techniques to be able to track effectiveness of improvements.
List of expectations during a Six Sigma project from KPIs’ perspectives can be outlined
by using the following:

• Project Scope and Objective. The objective and scope of a project are essential drivers
of KPI choices. In particular, the goal of a project can be broken down to multiple
goals and the associated KPIs can be found accordingly. For instance, improving the
layout of a facility would not only require focusing on each department’s individ-
ual layout improvements but would also require focusing on the interactions between
the departments within the facility based on customer requirements. Given this fact
we have multiple KPIs in a project that would require maximizing the efficiency of
each department along with minimized distance traveled between these departments
depending on the frequency of traveling from one to another.
• Success from a financial perspective. The financial scope of a project is driven by the
customer just like quality expectations and on the extreme ends of possible projects, a
project may solely require focusing on reducing the costs of the output while another
project may solely ignore costs. The associated KPIs would depend on the project’s
scope with the financial spectrum in place. For instance, when an item would be pro-
duced, the material might be a specific metal and the design might have to be a specific
one therefore these specs of the project cannot be changed and these KPIs need to be
eliminated from financial improvement considerations. There are some costs that are
hidden therefore it is important to be detail oriented from a financial standpoint. For
instance, warehouse space is a cost that is not hindered in financial calculations.
• Project Performance Evaluation. Evaluation of performance during a project is KPI
dependent, and the measurable outcomes for a Six Sigma project may be attained by
using these KPIs. For instance, quality may be the main driver of the KPIs chosen
with less important performances for the output to be time, money, and quality as the
objectives of the customer.
• Project’s Output Evaluation. The design of the evaluation process has two depen-
dencies unless additional factors are incorporated into a project: Customer choices and
industry standards. Success of the output in a six-sigma project is expected to be mea-
surable based on the pre- and post-improvement differences. The data collected (which
should be driven by the KPIs) needs to reflect the output expected specifications. For
instance, 30 items are produced by using the mix of 10 pounds of titanium, 30 pounds
of solver, and 16 pounds of silver may result in low quality of output prior to the
beginning of a six-sigma project and the customer may employ a Six Sigma team to
2 Fundamentals of Six Sigma: Meanings of Waste, Value, and Key … 33

increase the strength of the manufactured devices by using the same different of metals
without changing the processes and item design with a minimal resizing. Suppose a
Six Sigma team is formed to re-design the device and produce 26 of the resized design
of the device. Evaluation of the project can be based on the quality of each unit based
on the material strength of the item. The cost of the item may increase per item’s out-
put due to an increase in the amount of titanium with a reduction on the silver used
while the expected quality can be observed to increase due to item’s strength testing.
• Improvement Approach/Strategy. The strategy followed for the six sigma steps
impacts both the project and the choices of the KPIs. Typically, there are multiple
ways to improve the outcomes of a six-sigma project and the results of a Six Sigma
project would be limited to the chosen strategy. The ideal approach would be to evalu-
ate multiple improvement strategies (3 or 4) that allow measurable outcome attainments
for the output and KPIs. If we consider the waste reduction that occur during trans-
portation within a facility as an example, then the strategy and the associated KPIs are
likely to vary:
– Improvement 1. Automating transportation by using advanced technologies such as
automated guided vehicles can speed up the processes and would reduce wasted
resources during transportation.
– Improvement 2. Redesigning the physical environment by changing the places of
departments and the associated storage areas, equipment, workstations etc. can
reduce the time wasted during transportation.
– Improvement 3. Changing the transportation schedule based on a better structured
operational design.
– Improvement 4. Redesigning the transportation route based on the operational struc-
ture. It is possible to combine one or more of these improvement strategies during
the implementation of a six-sigma project however constraints such as time and
budget may limit such an application hence the strategy needs to be formulated
carefully based on the project expectations.
• KPI to be measured. Meaningful measurement choices are essential for KPIs that
should have consistency during the entire project. For instance, if a certain time frame
is chosen for data collection from different places for a project, then the data would
have inconsistency. Such a variational nature of the data would cause problems later
without the appropriate choices of specifics of the data measurement technique. More
about measurements and measurement techniques will be covered in the measurement
section.
• Objective and KPI optimization. Constrained optimization is typically seen in real life
applications in which case the objective is to minimize or maximize one or more KPIs
that are constrained by pre-determined conditions. In some cases when the problem
to be solved has mathematical nature with too many KPIs (such as 4 or more) then
it may be ideal to structure a constrained optimization problem for it to be solved.
For instance, maximizing CTQ conditions by optimizing the KPIs driven by customer
34 2 Fundamentals of Six Sigma: Meanings of Waste, Value, and Key …

Fig. 2.3 An example of key performance indicator (KPI) related considerations in Industry 4.0
applications [26] in which case the scope of the project can be chosen based on the mentioned KPIs

expectations may be the essential target of a project. Constraints in such a case are
driven by either the quantities or qualitative statements outlined by the customer. As
an example, the focus may be to resize a tool by minimizing its 3 dimensions given
that the tool’s strength should fulfill certain expectations. It is possible to use oper-
ations research, data analytics, and simulation techniques for optimizing outcomes.
Improvement techniques are going to be covered later in the Improve section (Fig. 2.3).

Waste categories used in six sigma, Total Quality Management (TQM) that is practical for
quality improvement applications, Lean methodology, and Kaizen event are some of the
sections that will be covered in this chapter. The meaning of using six sigma in industrial
engineering applications would also be covered with an example at the end of the chapter.
2.1 Identification of Waste and Definition of Quality 35

2.1 Identification of Waste and Definition of Quality

The most essential aspect of waste is its definition completing depending on the customer.
For instance, a waste limit of 20% may be acceptable to a customer due to a system’s
limited capabilities while the zero-defect policy may be the objective of another cus-
tomer. Overall, anything that a customer is not willing to pay for can be defined as waste
[25]. The value proposition of a product or service defined by the customer that includes
industry standards and the customer’s customers can be some of the drivers of quality
expectations. Unnecessary costs on a product or service are typically used for defining
the scope of a six-sigma project. The starting point of a six-sigma project is particularly
important for a six-sigma team noting that misunderstood or miscommunicated initiation
dynamics of a six-sigma project can drive the project in a misleading path. Along with this
approach, making sure that the customer is communicated with the progress of the project
reduces the waste throughout the project’s completion. Identification of the desired out-
comes of the customer from a six-sigma project can be identified and measured in several
different ways:

• Applying Qualitative Techniques. Some of the information needs to be collected

qualitatively and verbal investigation may have to be incorporated into a project requir-
ing conversations and collecting non-numerical data. Several key considerations can
include outlining the product or service expectations of the customer (i.e., output),
expectations and specifications of the output within the associated industry, and exist-
ing capabilities of entities such as physical and software resources, tools etc. for
delivering the output.
• Investigation through quantification. Numerical data and the associated facts relat-
ing to the project need to be collected. Some of the essentials that may need to be
considered in numerical data collection include but not limited to the following:
– Data for finances. Some of the information related to finances such as sales
information along with the costs, profits, and revenues are essential to take into
consideration in the cases when they are applicable. Only the financial facts of the
project need to be considered including but not limited to operational costs and
materials costs.
– Time Driven Data. The expected deadlines and the additional timelines need to
be clearly outlined by the customer. Additional information on input, process, and
output related operation timelines would need to be shared by the customer.
– Data for quality. The empirical (i.e., numerical) quality expectations need to be
clearly outlined for anything expected from the six sigma project outcomes by
the customer. In the case it is a generic product (such as an on-shelf product
or a standard product that doesn’t require customer specifications used by gen-
eral population) then there are some specific considerations needed in the outcome
expectations:
36 2 Fundamentals of Six Sigma: Meanings of Waste, Value, and Key …

Expected standards of the industry (such as ISO 90001, 55,000, 55,001, and
55,002)
Data used for automation if it exists
Data on transportation if it exists
Methods used for handling materials and the associated data
Periodic service data

The objectives in a six-sigma project are just like the typical operations research prob-
lem with the focus on minimization of NVA activities and maximization of VA activities
under a set of constraints. In some cases, it is possible to formulate the problem mathe-
matically as an optimization problem for a six-sigma project while in many other cases
the preference may be to not use mathematical formulation due to the complexity of
mathematical programming and its solutions. The optimization problem formulation can
be applied in any stage of input, process, and output. The traditional waste categorization
suggests categorizing waste into seven categories while the eight category is added as a
talent by some others:

1. Defects. The outcomes of a work can result in a rework or rejection that is classified in
the defect category therefore a defect may occur at any stage of a work to be completed
even before starting to produce or service. In some cases, the existing materials or
received items can be defected that can cause inability to complete tasks. Reasons for
rejecting an item can include but are not limited to the following:
a. Supplied product does not meet the quality requirements to start production.
b. Internal customers reject items.
c. Inspected items are incompliant with expectations.
d. Critical to quality conditions are not fulfilled by the product.
2. Excessive Inventory. Items in the inventory that are not utilized and occupy unneces-
sary space are considered as inventory excess. The following are examples to some of
the wastes that may occur in different environments.

• Manufacturing. If 1000 items are purchased and kept in the inventory, and if only
900 of them are used over a span of long time, then 100 items can be considered
excessive inventory. Data collection and analysis can reveal how many items are
needed for a short period of time to eliminate such waste.
• Healthcare. Inventory access in the storage units of hospitals is not surprising. Such
excessive inventory causes expired items which is one of the major issues in health
care that increases cost of overall patient treatments.
• Transportation. Storage in transportation facilities is typically not surprising and
organization of such storage can be the main challenge since inventory typically
occupies a good amount of space.
2.1 Identification of Waste and Definition of Quality 37

3. Motion. The actions that do not add value by people or machines are considered
as motion waste. For instance, forklifts within an organization may be traveling much
more than needed and this is causing waste. One should be careful about distinguishing
an activity to be a BVA or a NVA is such a case.
4. Excessive Production (Overproduction). Producing much more than the amount
needed is the typical waste that occurs in many organizations. In such a case, produc-
ing more than the amount that is needed or sold by the internal or external customer
can cause overproduction. Internal and external excessive production may occur.
a. Internal Excessive Production. An excessive amount of item production that can
be used by internal customers within an organization is defined as internal over-
production. The following are examples of overproduction for various industrial
engineering applications:
• Manufacturing. Excessive production is not unusual when assemblies of units
are needed. One such example is the production of two items that need to be
assembled and one produced than the other one. In the case of operations that are
completed internally, this becomes an internal overproduction. In this scenario,
the internal customer can be one of the departments that operate internally within
the same organization.
• Health care. Suppose doctor A treats patients and sends all treated patients to
doctor B for further treatment. If doctor A treats patients much faster than doctor
B then this is an example of patient overproduction for doctor B in health care
which is called an internal overproduction. Doctor B would have a hard time
keeping up with the patients that are building in front of his/her office therefore
it would cause patient build up and this can result in patient dissatisfaction of the
health care service. In this case the patient is an external customer while Doctor
B is an internal customer.
• Transportation. In supply chain transportation, it is natural for organizations
such as UPS and FedEx to link several iterations of transportation from point to
point to be able to deliver to a customer. In such a setting, each point of trans-
portation taking place in the transaction of transportation from one to another
becomes an internal customer.
b. External Overproduction. The overproduction of items that an external customer
will not be willing to purchase is called external overproduction. The following are
examples of such overproduction:
• Manufacturing. A company may be producing products that can be sold to any-
one and overproduction of an item that won’t be sold due to the newer version of
the item can be considered as an external overproduction. This is one aspect that
companies pay attention to, noting that this overproduction would be inevitable
and therefore organizations pay special attention to such scenarios.
38 2 Fundamentals of Six Sigma: Meanings of Waste, Value, and Key …

• Health care. A clinic may be an external customer to a hospital if the hospital

sends patients to the clinic. If the hospital sends too many patients than the clinic
can handle, then it is an external overproduction.
• Transportation Service. External overproduction can occur if a customer of
UPS (such as Amazon) sends too many items for shipping to UPS.
5. Waiting. As the title suggests, the time wasted on scheduled work is this type of waste
category. We categorize waiting into four:
a. People. Delayed operations/services by people that require waiting to complete a
product or service. The following are examples of such waste.
• Manufacturing. Waiting for technicians to operate machines, waiting for forklift
operators to pick up items from locations, waiting for quality control to complete
a task.
• Health care. Waiting for a patient to be taken from one location to another for
treatment is one of the costs that occurs in health care.
• Transportation Service. Items waiting for people to transport goods from one
location to another is one of the major costs in transportation.
b. Machines. Waiting for a machine during production or service can be a waste if
there are no other alternatives for the completion of the work. The following are
examples of such waste.
• Manufacturing. Waiting for a machine to complete tasks while another machine
is waiting for items to be completed by this machine.
• Health care. Waiting for patients to go through x-ray is a waste.
• Transportation Service. Waiting for a machine to further progress transportation
of items would cause a waste in the system.
c. Material. Material waste can happen in different ways that include misuse and
mistakes made in operational use.
• Manufacturing. Misprocessing the materials can result in material waste that
may or may not be disposed of.
• Health care. Waiting for materials for treating patients such as medicine.
• Transportation Service. An example of material waste is the redundant use of
carton boxes for shipping while other materials that can be used for re-boxing
items.
d. Information. Preparation, use and structuring information along with waiting for
information in many products and services can cause wastes. The following are
some forms of information waste:
• Manufacturing. Misinterpreted information can result in mistakes in production.
• Health care. Waiting for x-ray results in the case when they are ready for sending
out to doctor.
• Transportation Service. Waiting for further transportation processing even
though the associated information is ready, but it is not communicated.
2.1 Identification of Waste and Definition of Quality 39

6. Transportation. Waste that occurs due to re-allocation of entities due to the NVA
nature of activities. One should be careful about distinguishing motion and transporta-
tion wastes. For instance, transportation services provided by many companies try to
eliminate waste of excessive movements from facility to facility. For instance, if the
patients are not sent to the right treatment location by a hospital chain, then this can
cause transportation waste as a part of patient treatment.
7. Process. Any work done between the input and the output of a product or service is
a process that may contain waste. NVA activities during the completion of such work
at any phase of the process is considered a waste. Examples of process waste include
unnecessary steps taken during the production of an item or treatment of a patient in
which cases there are additional process steps that exist without adding value to the
process and cost more to the customer.
8. Talent. Talent is not easy in today’s world and placing the right talent in the right
position is one of the challenges faced by organizations. Talent, defined as the skillset
of a person, can be underutilized. Placement of the right skillset within an organization
can be maximized by the right placement of staff in the right positions given their
known skillsets. This requires getting to know the staff and understanding their skills.
As an example, misplacement of a person with high levels of talent into a position that
utilizes 10% of the person’s skillset causes talent waste. Such a waste can be measured,
and it should be noted that the targeted waste reduction should not necessarily be 100%
as it may not be realistic. Training of staff to uplift staff knowledge and skillset can
be considered by an organization for reducing general talent waste.

All these eight wastes impact all parts of possible outcomes. In addition to these eight
wastes, a nineth waste category that makes sense to introduce in this work is the “Tech-
nology” waste (to be defined below) given that technology is a major part of today’s world
and there is a significant amount of waste that occurs in many areas of applications. One
example to Technology waste is the tremendous amount of satellite remaining that exist
in the Orbit of Earth. A typical application of technological waste is the application of
artificial intelligence (AI) in real life applications. While AI may help with reduction
of what may appear to be waste for some people, it is not recognized that the in some
cases outcomes of AI can be impacting outcomes negatively. Careful decisions on the AI
application needs to be made to understand the benefits or harm of the use of AI.

9. Technology waste: Waste technology may occur in several different ways:

a. Technology choices. The wrong choice of technology may result in significant
waste and such a waste may be essential to reduce before it occurs. One such exam-
ple is the purchasing of a brand-new machine that underperforms than expected
while a better machine may exist.
40 2 Fundamentals of Six Sigma: Meanings of Waste, Value, and Key …

b. Technological misuse. There are many environments where technology is misused,

and it is not realized. One such example is the misutilization of the machines in
manufacturing.
c. Technological quality. The quality of the input, process, and output of the tech-
nological choice may have the potential to generate waste and such a waste may
result in quality related issues.
d. Technology biproducts. The outcomes of technology are useful, however the
biproducts generate waste that is harmful to the environment. Given that environ-
mental consciousness should be the number 1 priority, handling wastes that arise as
biproducts of a technological outcome should be paid close attention.

2.2 Management of Quality

The customer is in the center of quality and Total Quality Management (TQM) allows
improvement of quality in an organizational manner that allows every member of an orga-
nization to be actively involved in improvements. After the 2nd World War, increase in
production volumes and cost reduction were needed for industrial growth which were
the highlights of continuous improvement ideas and initiatives. A high-volume and cost-
effective production was needed right away this period, and the quality component is
placed in production during the 1970s. Excellence of operations was needed and started
to take place for quality assurance and control around this time of operational work done.
The Japanese industry started to rapidly grow and invest more towards bettering quality
management in 1980s. In alignment with industrial growth, quality in health care opera-
tions also started to arise due to quality related issues’ increase during the treatment of
patients. Several Japanese ideologies and the associated principles were produced with
Juran’s principles of total quality management be one of such initiatives that followed the
following approach:

1. Solve problems by looking for improvement opportunities, setting up projects and

goals.
2. Reach goals by providing training and structuring organizational and approach.
3. Communicate and report progress after acknowledging improvements; Make sure that
the improvements are continuous.
4. Assign scores to improvements as a measurable approach for rewarding accomplish-
ments.

Juran’s principles are in alignment with Six Sigma’s DMAIC approach from a continuous
improvement perspective. It is aiming to increase organizational awareness at all levels
of the organization by active participation of the personnel at all levels in a continuous
fashion. This approach also removes the monotonicity of the daily work of employees
2.3 Meaning of Sigma Term and DMAIC Approach 41

if applied right. Hence, Juran’s principles taken from the Total Quality Management
principles helped Six Sigma with.

• DMAIC approach as a problem-solving technique,

• Project based approach with a trained team to solve a specific problem, and
• Use statistics for defect measurement.

2.3 Meaning of Sigma Term and DMAIC Approach

A successful business would typically want to reduce waste and improve operations in
order to reduce the costs and improve quality as much as possible. Motorola imple-
mented Six Sigma as a business strategy that was initiated by an engineer, Bill Smith, in
1985 through the adaptation of statistics for measuring, analyzing, and improving quality.
Between 1981 and 2001, General Electric (GE) implemented Six Sigma as CEO Jack
Welch’s initiative that was successfully implemented around 1995. There are well-known
organizations across the World today that apply Six Sigma such as Amazon, Google, and
Yahoo Inc. These organizations provide training and employ personnel to be able to suc-
cessfully implement operations at many different works completed. Health care, service,
and manufacturing industries continuously adopt Six Sigma and Kaizen projects daily.
The following definitions of Define, Measure, Analyze, Improve, and Control are very
simple to remember, however what matters the most in a Six Sigma project is the right
approaches followed with attention paid to details.

• Define: The problem needs to be introduced properly after its identification; This
section typically includes introductory information on the project information.
• Measure: The problem defined requires data collection and the associated measure-
ments by using appropriate techniques.
• Analyze: The measured qualitative and quantitative data needs to be Analyzed for
better understanding the system in place with the root-cause analysis of wastes to be
able to eliminate such waste.
• Improve: Improving the system in alignment with the defined problem is the key
approach in which case actions need to be taken to eliminate waste as much as possible.
• Control: Placement of right control systems to be able to not only measure how much
is a system under control but also designing mechanisms to keep an improved system
under control are essential to ensure that the improvements are sustainable, and this
is a key step in DMAIC approach. Eliminating control mechanisms in a system can
result in dramatic effects on the system that can cause a system to return to its unstable
design.
42 2 Fundamentals of Six Sigma: Meanings of Waste, Value, and Key …

The statistical term Sigma (σ) used in Six Sigma is a Greek letter used for variational
measurements of the data. This variation allows us to determine how much data points
deviate from the standard value known as the average (i.e., mean value.) The sigma term’s
high value indicates the deviation from the standard to be high which could have a pos-
itive or negative meaning depending on the context of the problem. Given a normally
distributed data set, the following sigma ranges are well-known for normal distribution:

• Two Sigma Range (2σ): This is the range with the mean value in the center covering
68.27% of the data.
• Four Sigma Range (4σ): This is the range with the mean value in the center covering
about 95.45% of the data.
• Six Sigma Range (6σ): This is the range of data with the mean value in the center
covering

In the case when the sigma value is low, the data will be closer to the central value
(i.e., average) therefore the graph of the normal distribution shrinks/narrows therefore the
graph of the normal distribution narrows. The typical expectation in six sigma low defects
with low standard deviation noting that highly varying operations are typically unreliable.
The typical goal of a Six Sigma team is to reduce process variability to minimize wastes’
occurrences. Therefore, the expectation is to reduce variability to a statistically signif-
icant level with the goal of having at most 3.4 defects per million productions. As an
example of low sigma levels, the production of health care products for health care ser-
vice can be considered; In manufacturing of medical tools, application of “zero defect”
is typically expected. In these cases of such applications, the 6σ excellence would kick
in noting that the defect remains zero in which case the variation remains to be zero. A
zero-defect approach is realistic and there are manufacturers that implement this approach
in their facilities. Automated inspection is typically helpful and used in such applica-
tions. Figure 2.4 outlines the percentages corresponding to 4σ, 8σ, and 12σ ranges from
the average. It is important to note that the above mentioned two, four, and six sigma
ranges assuming the ranges around the mean value while the ones outlined in Fig. 2.4 are
different.
DMAIC can be particularly successful when a process is broken down into subprocess.
In typical applications, the Define phase has the following as the starting point:

• Project Charter. Design a charter that outlines project specifications (Project Charter
will be covered in the Define phase)
• Specification of the Customer. The following needs to be identified for project’s
healthy initiation:
– Critical to Quality
– Key performance indicator outlined by the customer
– Deadlines and timelines of tasks
2.3 Meaning of Sigma Term and DMAIC Approach 43

Fig. 2.4 The sigma ranges of scores based on the frequency of occurrence [14]

– Input, process, and output specifications based on the definition of the problem
• Identifying the gaps. There can be issues that can be considered as gaps for continuous
and successful operations. These gaps lower the quality of operations and the expected
high-level industry expectations as well as the expected standards. Identification of
such gaps is essential in successful Six Sigma applications.

An example of a Six Sigma project is to improve health and safety conditions of

workers outlined for specific systems as outlined in Fig. 2.5. The steps of DMAIC are
summarized in [32]. There will be other examples given later for exemplifying other
real-life projects.
Steps of DMAIC flow typically follow Define, Measure, Analyze, Improve, and Con-
trol phases, hence the following order is the typical DMACI application in Six Sigma;
however, this order of application is not always the case.

De f i ne → M easur e → Anal yze → I mpr ove → C ont r ol

For instance, there may be measured data on an organization’s database and the company
may want to take advantage of such data therefore a part of the Measurement phase exists,
and this existing data may drive the definition of the project. In such a case, the data and
the project’s definition should be matching since otherwise there won’t be meaningful
outcomes attained.
44 2 Fundamentals of Six Sigma: Meanings of Waste, Value, and Key …

Fig. 2.5 An example of the DMAIC approach on a healthcare project with FES standing for Fuzzy
Expert System and ERI stands for Ergonomic Risk Indicator [32]

Six Sigma is a well-known training and certificate program offered through univer-
sities, consultants, hospitals, and other educational organizations. There are no specific
standards existing for a Six Sigma certificate program hence the understanding and the
strategy of application is very much driven by the organizational goals and intensity of the
expectations. There are several belts provided upon successful completion of Six Sigma
projects that include the following:

• White
• Yellow
• Green
• Black

White belt is the easiest in this ordered list of Six Sigma belts while the Black is the
most advanced belt. The level of significance between these belts arises from the level
of details and the associated statistics applications. A Black Belt Six Sigma project may
last anywhere between 6 months to a year while White Belt may be completed in one or
two weeks. Organizations typically prefer to invest in Green and Black Belts projects as
the level of outcomes are solid for further improvements and operations. Organizations
that include but not limited to Amazon, Google, Motorola, General Electric, American
Express, FedEx, and UPS employ six sigma for operational improvements on daily basis.
Six Sigma teams may require some other additional personnel that have specialized
knowledge in the case of advanced projects. For instance, the increasing complexity of
the problems due to large data sets in recent years started to give a rise to the need for
data scientists and engineers’ need in Six Sigma projects for further advanced analysis of
2.4 A Methodological Approach Through Lean Principles 45

operations. This also holds true for a variety of different settings. For instance, in hospitals
and health care clinics that employ six sigma may need the involvement of the following
personnel:

• Green Belts
• Black Belts
• Nurses
• Doctors
• Engineers
• Data Analysts
• Organizational Leaders and/or Sponsors

The financial support to initiate a Six Sigma project may depend on a variety of resources.
This funding comes through an organizational leader or a sponsor that indicates the neces-
sity of the project for the organization in the case of large projects that can be Black Belt
projects. In the case of small projects, a product manager or plant manager may show
interest in a small scale Six Sigma project that can be at Green Belt level. Master black
belts (managers of critical projects), black belts (experts to tackle critical projects for
improvement), and green belts (supporters for implementing changes) are examples of
Six Sigma personnel that can initiate a project and define the Six Sigma project’s scope.
A person with a black belt may not be necessary for a Six Sigma project noting that the
project’s scope drives the expectations from the project.
Next section’s focus will be on Lean methodology fundamentals and the variations of
it in applications.

2.4 A Methodological Approach Through Lean Principles

Lean methodology aims to structure an organized approach and principles based on the
structured approaches. Overall goals of Lean include but not limited to the following:

• Cost reduction
• Establishing high quality
• Cycle time shortening
• Waste identification and reduction relentlessly

The overarching goal is to reach a high-quality service or product through organized

methodologies. Lean principles are typically easy-to-implement techniques for a vari-
ety of groups so that the Lean improvements can be applied in different settings easily.
Lean motivates us to use organizational resources and supplies driven by customer needs.
46 2 Fundamentals of Six Sigma: Meanings of Waste, Value, and Key …

Increasing business success through improved processes is needed after Lean method-
ological applications. Examples of Lean methodologies applied on processes are the
following:

• High quality health care service provided to patients with reduced cost
• Processes completed for high quality items manufactured to customers with reduced
cost
• Transportation processes that are completed with high quality and reduced costs.

Lean methodology tools that we will cover in this book include the following:

• Mapping of processes through Process Mapping

• Structuring travel patterns by using Spaghetti diagrams
• High-level sketch of operations by designing a Value Stream Mapping
• Organized short period improvement applications through Kaizen events
• 5S technique (Sort, set in order, Standardize, sustain, and shine)

Additional Lean techniques include the following:

• Just-in-time
• Kanban
• Standardization
• Pull and Push logic driven systems

For instance, Lean Six Sigma in manufacturing is the combination of Lean applications in
manufacturing with Six Sigma steps taken for improvement. Lean methods in healthcare
are particularly important for excellence in health care service. DMAIC is one of the
approaches in Six Sigma that can be used for managing project improvements by data-
driven statistics-based measurements for process evaluation while Lean principles help
with organizing operations to achieve improvements in a uniform manner. Six Sigma
aims to reduce variability while integration of Lean helps with determination of the used
improvement techniques and sustainability of such improvements’ applications. Overall,
the expectation is to maximize VA activities while minimizing NVA activities.
The next section focused on the business aspects of the expected outcomes. Given
that there are three dimensions of six sigma team involvement (business, customer,
and employees), all these dimensions need to be kept in mind for further operational
improvements.
2.5 Aiming to Achieve Business Excellence Through Measurable, Attainable … 47

2.5 Aiming to Achieve Business Excellence Through

Measurable, Attainable and Realistic Outcomes

The initial step of a project’s scope is the ability to attain results. A project targeting
impossible or infeasible outcomes with the associated objectives or improvement oppor-
tunities would not be meaningful. While the goal may be to achieve business excellence
through improvements in a timely manner, a project’s approach needs to be specific, mea-
surable, attainable, and realistic. This takes us to introduce the idea to B.E. S.M.A.R.T.
for completing a project:

• Excellence Goal in Business. The typical goal in business is to achieve excellence and
the six-sigma team typically works along this line of achievement. The team should
keep in mind quality versus cost as a part of the excellence achievements. The decision
maker is the customer or the business, in which case excellence understanding changes
from customer to customer. The expected project outcomes should be high quality with
customer expectations fulfilled.
• Need to be Specific. A Six Sigma project has objectives and such objectives, and the
associated outcome attainment needs to be specific. Such specification clearly outlines
the expected details from a project for achievement needs. Such objectives need to be
specific to the CTQ and expected outcomes outlined.
• Outcomes need to be measurable. Any aspect of the project that is not measurable
can result in questionable attainment of goals unless improvements are clear. Such
outcomes’ measurements will be covered later throughout the book in several differ-
ent places. KPIs have important roles in making a connection between measurable
outcomes and project completion steps.
• Objectives’ Attainability. Targeting specific goals that can be attained during project’s
timeline and allocated budget for reducing ambiguity is needed. This also relates to
the need for a realistic approach during the project’s completion to be explained next.
Budget, timeline, goals, and objectives should be in harmony for attainable outcomes.
An experienced person typically needed in a Six Sigma project to be able to foresee
attainability of the outcomes.
• Project’s Realistic Nature. All steps of the project need to be realistic for attainable
outcomes. All project-related indicators such as available resources and other factors
would give rise to the realistic nature of such a project.
• Relating the Project to Time. Duration of phases and steps need to be clarified with
possible updates on the completion of a project.

The basics of B.E S.M.AR.T are explained in Fig. 2.6 below [25].
48 2 Fundamentals of Six Sigma: Meanings of Waste, Value, and Key …

Fig. 2.6 Business excellence steps through measurable, attainable, and realistic design [25]

2.6 Short Period Improvement Opportunities Through Kaizen

Event Approach

It is possible to improve a process as a part of product or service through a Kaizen event

as a focused effort. Such an event is best to use when the improvement is needed in
a specific small area that does not require extensive statistical analysis and it would be
possible to improve the specific area in a short period of time. Both the Kaizen event and
Six Sigma share the common goal of waste elimination by fulfilling a set of goals with
differences in approaches.
In addition to the specifications of a Kaizen event, Fig. 2.7 displays the commonalities
and differences between a Six Sigma project and a Kaizen event.
A Kaizen event can be grouped into three phases:
Stage 1: Current conditions of the system to be improved

a. Determine the improvement needed

b. Improvement goals should be outlined
c. Gather a Kaizen team with the set of capabilities to fulfill the expected tasks

Example 2.1 If the goal is to increase open space in a warehouse to be able to place more
products and machines then the associated team should be chosen based on the location of
the implementation. For such a project, Kaizen team members can include the plant manager,
quality specialist, industrial engineer, and safety specialist.
2.6 Short Period Improvement Opportunities Through Kaizen Event Approach 49

Fig. 2.7 Outline of Six Sigma and Kaizen event specifications

Example 2.2 The goal is to better space utilization in a hospital to increase the number of
patients treated and reduce wasted space. A Kaizen team member can be a senior doctor, a
senior nurse, quality specialist, industrial engineer, and facility manager.

Stage 2: Set of Actions to Be Taken

a. Training of Employees. Educating Kaizen team members for a short period of time
specific to the project.
b. Existing Structure of the Process. Explanation of the project’s scope, data collection
for solving the problem, and designing a process mapping for outlining operations.
c. Future Improvement Expectations. Identify the expected improvements by the stake-
holders for structuring a solution and waste elimination strategy to implement a
solution.
d. Ideas to be Tested. Using the identified improvement strategies, testing the system
by using these ideas and collecting data from the improved system would result in
choosing a strategy of improvement. Analysis of the data collected would help with
evaluation of an improved system.
50 2 Fundamentals of Six Sigma: Meanings of Waste, Value, and Key …

e. Documenting Event Details. Document the project details by outlining both the
improved and pre-improved systems.

Stage 3: Improvement Steps

a. Apply Kaizen outcomes. Apply the improvement actions on the processes.

b. Standardize Operations. Standardize the improvements in a procedural way.
c. Find and Extend Improvements to New Areas. Extend the Kaizen event improvements
to other areas that are either similar to the improved system that can be local or a
specific region, or to a global setting as the generalization of the work done. The
project to be extended may have to be a Six Sigma project if it is more extensive than
the Kaizen event.

Example 2.3 Designing a cell is a good example of a Kaizen event. Such a project can help
to reduce the costs of an organization if applied correctly in specific regions. This type of
Kaizen event can be implemented in other local areas within the organization. For instance,
cost reduction in manufacturing cell design can help with the cost reduction of each product.
Several improvements determined as a part of several Kaizen events in local areas can be
used as the benchmark for a Six Sigma project formulation.

Example 2.4 Rooms in a hospital are used for specific treatments and these rooms may be
under-utilized therefore it causes waste. Waste reduction of a specific treatment that requires
few rooms’ usage in a hospital setting can be a Kaizen event.

2.7 Structuring Strategy Through Mission and Vision

The vision and mission statements of a customer in a Six Sigma project may play a role
in customer’s expectations. Some of the organizations have six sigma projects on a rolling
basis to be able to reach their vision goals in alignment with their missions. Understanding
such statements for a Six Sigma project and incorporating ideas and works in alignment
with these statements would be critical. For instance, in some cases, an employee working
for a company may be the key personnel for a Six Sigma project and this person may be
expected to us mission and vision statements to be able to complete projects by working
with external Six Sigma team members; Such an expectation may be naturally occurring.
Vision and mission can be incorporated into strategy of a Six Sigma project by making a
connection between the Six Sigma project and these statements by using some or all of
the following:
2.7 Structuring Strategy Through Mission and Vision 51

• KPIs and the associated measurements

• Target goals
• Stating objectives
• Structuring initiatives

Some important specs to be kept in mind in alignment with these are for the following
categorizations:

• Looking from a customer’s perspective. In what ways can the customer experience
be improved?
• Educational needs and growth. What are the ways to improve the educational needs
of employees and staff to fulfill the progress of the project?
• Internal business. What should be the focus point of the Six Sigma project to be able
to improve the internal business experience?
• Financial outcomes. How can the expected financial outcomes be achieved during the
completion of the project?

Let.

• α : Budgeted Cost of Performed Work

• β : Actual Cost of Performed Work
• γ : Actual Cost of Scheduled Work

KPIs are essential to know for a Six Sigma project, and these indicators are determined
by the customer however the Six Sigma team should work with the customer to fine tune
the number of KPIs for a healthy completion of a project as the customer may not be able
to forecast the scope of the project and the associated timelines and costs. The following
are some of the key considerations when it comes to KPI design:

• Improvement of quality
• Existing cost and the expected change
• Expected engineering work’s effectiveness
• Cost/benefit ratio expectations
• Payback period identification of the improvement
• Expectations in changes of outsourcing input or operations when compared to the
existing operations.
• Improvement of the throughput
• Improvement of the cycle time
• Improvement of machine downtime
52 2 Fundamentals of Six Sigma: Meanings of Waste, Value, and Key …

• Improvement of planned maintenance

• Improvement of changeover time
• SIPOC steps
• Cost performance index (CPI) improvement expectations defined by the formula
α
CPI =
β
• Schedule Performance Index (SPI) can be calculated by using the formula
α
SPI =
γ
• The expectations of production goals can be calculated; The following formula is
used for determining percentage of product attainment based on the number of items
manufactured and the targeted production output levels.
# of manufactured items
Pr oduct Attainment = 100%
Production Output Targetted
• Operational downtime. Assessment of downtime in comparison to its actual operating
time for determining if any action is needed to make changes on the operations. This
includes the level of productivity. This concept can be applied to both machines and
staff based on the productivity analysis.
• Identification of Capacity Utilization. The ratio of the actual output to the maximum
possible output provides the overall system’s capacity utilization. This capacity utiliza-
tion depends on the work performed in a certain setting and the associated calculations.
The following simple formula can be used for such calculations:

Actual Output
Capacit y U tili zation = 100%
Maximum Possible Output
Typically, a Six Sigma team makes decisions in association with the customer to deter-
mine a combination of these KPIs structured by the scope of the project. There can be
other KPIs expected from a customer listed for a project.

2.8 An Industrial Engineer’s Role in Six Sigma Applications

An industrial engineer is trained to understand and engineer systems in a broad manner

by using advanced analysis and improvement techniques who has the necessary skills and
techniques developed by training. There are many concepts that an industrial engineer
learns for using as a part of Six Sigma projects’ data collection, analysis, improvement,
and control phases:
2.8 An Industrial Engineer’s Role in Six Sigma Applications 53

• Calculus based statistics

• Data Analytics
• Simulation
• Operations planning and control
• Design of experiments
• Human Factors
• Operations Research
• Lean systems
• Statistical process control
• Production Systems
• Facilities layout and material handling
• Advanced industrial concepts (such as Industry 4.0, robotics, IoT, AI etc.)

An industrial engineer typically participates in improvement projects in each organization

and Six Sigma techniques can be a guide for such improvements. In alignment with this
understanding, the following is typically needed to be able to advance systems:

• Development and research

• Designing appropriate experiments
• Determining right calculations
• Right data collection strategies
• Right choices of data analysis
• Using advanced technologies in the right time and place with the right costs and
abilities
• Choices of new equipment use and how to structure them
• Right communication skills

ABET (Accreditation Board for Engineering & Technology) is the accreditation body for
Industrial Engineering programs in the United States of America, and the accreditation
standards include both written and oral communication skills to be developed by indus-
trial engineering majors. An industrial engineer who cannot communicate properly may
have hard time in the completion of the associated projects and face challenges with the
employees and customer in the long run (as students of the author experienced over the
years during Six Sigma projects’ completion.)
Commercial Six Sigma books written with a general purpose include a broad perspec-
tive that do not necessarily relate to the education of an industrial engineer. A typical
industrial engineer has a theoretical statistical knowledge that utilizes integral calculus,
and this fact is not utilized in such books. This book contains some basic integral calculus-
based statistics and advanced technologies to a certain degree as these two areas are very
broad that cannot be covered in a single book. It is expected for industrial engineers to
54 2 Fundamentals of Six Sigma: Meanings of Waste, Value, and Key …

naturally take place in Six Sigma projects and take the lead for company-based improve-
ment ideas and opportunities therefore incorporating industrial engineers to six sigma
projects are important. This book is written with the aim of basic Lean methodology and
Six Sigma education of industrial engineers at a fundamental level.

2.9 Waste Identification Examples

Example 2.1 (Health Care & Supply Chain.). Linen quality and the associated wastes in
a hospital can be a cost issue for a hospital. The supply chain committee of the hospital
oversees and reviews the associated supplies and expenses to identify the opportunities
for waste reduction driven by cost calculations. The cost of linen processing impacts the
customer’s health care cost given that the cost to the hospital is a cost to both hospital
and customer which also impacts the economy of the country. Given that the hospital is
part of a hospital group in the area that owns a linen processing facility, the clean linens
are received from this facility at the receiving dock of the hospital every day based on the
orders placed the day before. The distribution of the linen is managed by the hospital’s
materials management group on a scheduled basis. Environmental Services gather the
soiled linen and ship them each morning when the clean linen is delivered. The soiled
linen is then delivered to the linen facility and cleaned. The following steps can be taken
for waste identification during the Define phase of Six Sigma to be covered in the next
section.

1. Understand the linen processing steps in detail.

2. Identify the step-by-step linen processing steps from the beginning to the end and list
them.
3. Determine the possible wastes associated with linen processing.

Example 2.2 (Flywheel Manufacturing.). A flywheel shown in Fig. 2.8 rotates the engine
and stores mechanical energy to balance the engine for it to continually have a good
performance. The flywheel works to balance the mechanical power by storing it therefore
manufacturing of it has important implications in the applications that it is used for. A
six-sigma project can be applied to the casting of flywheel [33].

2.10 Questions

Question 2.1 Choose a waste that depends on your industry waste (e.g., manufacturing,
health care, transportation, etc.), find and explain three examples of defects existing in the
industry and briefly explain why such defects may occur in such an industry.
2.10 Questions 55

Fig. 2.8 An example of cast flywheel during the manufacturing processes [33]

Question 2.2 Please explain the difference between CPI and SPI. From a supply chain
perspective, give an example and explain where they impact each other. In addition, find an
example where they won’t impact each other and explain your choice.

Question 2.3 Find a company that experienced transportation waste and applied the DMAIC
technique to eliminate the waste. Explain the type of waste, how it is determined, what was
done to eliminate it.

Question 2.4 Find an organization that is a part of Worldwide supply chain operations that
experienced a waste and applied A Six Sigma technique to eliminate the waste. Explain the
type of waste, how it is determined, what was done to eliminate it.

Question 2.5 Kaizen event is covered in this chapter in a brief manner. Find an example of
a Kaizen event applied in a manufacturing setting and explain the following:

• Who were the team members?

• What was the scope of the event?
• What was done to eliminate waste?
• What was the outcome of the event?
56 2 Fundamentals of Six Sigma: Meanings of Waste, Value, and Key …

Question 2.6 Kaizen event is covered in this chapter in a brief manner. Find an example of
a Kaizen event applied in a health care setting and explain the following:

• Who were the team members?

• What was the scope of the event?
• What was done to eliminate waste?
• What was the outcome of the event?

Question 2.7 The following costs are estimated initially for a work to be performed for a
Six Sigma project in the budget of a project:

• Step 1. Material Purchasing: 1500 USD

• Step 2. Labor: 2500 USD
• Step 3. Shipping: 4500 USD

The following actual costs are identified upon the completion of the tasks:

• Step 1. Material purchasing: 2750 USD

• Step 2. Labor: 4750 USD
• Step 3. Shipping: 5000 USD

Calculate the Cost Performance Index and interpret the result.

Question 2.8 A hospital can treat at most 1 thousand patients while it can treat 700 patients,
that is 150 patients below its actual target. Determine the capacity utilization of the hospital
and the target attainment.

Question 2.9 A work completed determined to have 150 h of value-added activity time,
135 h of business value added time, and a non-value-added time of 115 h. After one time
six-sigma application, the non-value-added time is reduced by 15 h while value-added time
increased by 10%. Following this first application, the second Six Sigma application resulted
in 20% of non-value-added time reduction. Calculate the following:

a. The number of value-added hours and its percentage after the second phase of the Six
Sigma application.
b. The number of non-value-added hours and its percentage after the second phase of the
Six Sigma application.
c. When the improvement percentage is calculated by finding the percentage of non-value-
added activities reduction, which round of Six Sigma application had the most waste
elimination and improvement?
2.10 Questions 57

Question 2.10 Find and explain a difference between the product attainment and capacity
utilization from a manufacturing perspective.

Question 2.11 Find the mission and vision statements of a company of your interest and
state them. How can these two statements be used for a Six Sigma project application.

Question 2.12 There are two possible Six Sigma statistical ranges explained in this chapter.
What is the difference between them? How can the application of the Six Sigma range
calculation impact the outcomes?

Question 2.13 Find an organization of your interest in an industry that applied Six Sigma
in operations. Explain how they applied Six Sigma and how they benefitted from such an
application.
Defining the Six Sigma Project
3

As has been made clear in previous sections, the customer is in the center of a six sigma
project. Defining a six sigma project also depends on the customer expectations. The
choice of the six sigma team is driven by the customer-defined problem and the associated
internal or external six sigma members should be chosen. In some cases, a company hires
an external organization for the completion of a six sigma project and this project is very
much driven by such an organization’s six sigma team. A customer may be an internal or
external customer to the organization. Potential member choices for a project can include
the following:

• A person in charge of the project (a stakeholder such as a plant manager).

• Master black belt or green belt six sigma certified person(s).
• Industrial (or another) engineer.
• Engineering technician(s).
• A person who understands the finances of a project and purchasing (if needed).
• Other additional employees of the company (if applicable).

Example 3.1 Suppose the scope of a project is to reduce the cost of solar panel placement
of customers within a region of New York state area with the goal of maximizing solar panel
utilization on the roofs of customer’s while minimizing the wasted space on the roofs.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 59

E. Tokgoz, Six Sigma and Quality Concepts for Industrial Engineers, Synthesis Lectures
on Engineering, Science, and Technology, https://doi.org/10.1007/978-3-031-55740-8_3
60 3 Defining the Six Sigma Project

Fig. 3.1 Supplied solar panels placed on roofs [25]

The following are some of the improvement opportunities that can be considered for
cost reduction:

• Transportation of panels from location to location.

• Space organization of the roof for proper panel placement.
• Cost/benefit calculations for a customer based on the panel type choice.
• Wasted space reduction based on government regulations.
• Operational waste reduction.
• Possible additional solar panel placement options (such as places other than roof).
• Design specifications of the solar panel.

Figure 3.1 shows an image of solar panel raw material that is placed on roofs [25].
As mentioned previously, the project’s definition is somehow clear, however it requires
some more details. The above-mentioned information is a high-level explanation of the
project. As we covered previously, cost reduction would be favored by many customers,
however the quality aspect of the project should also be noted to the customer based
on the choices made, because, otherwise, the customer may not be satisfied after the
placement of the panels, and this can result in the dissatisfaction of the customer after
starting to use the panels. The solar panel company that sells the solar panel system
must care about this aspect since, otherwise, the customer may express dissatisfaction in
web-based platforms to other potential customers of the company. The reputation of the
company can be challenged by such outcomes. The definition of the problem requires
some more details about the panel choices, and the following is one way to define such a
problem:
3 Defining the Six Sigma Project 61

Define There are three types of panel choices for consumers: 100-W, 300-W, and 400-W
pane placements on roofs. All these panels have pros and cons depending on their placements.
Possible sizes of roofs of consumers start from 300 (square feet) and increase by 50 up to
1000; the increment changes to 100 after 1000 square feet up to 5000. A company wants
to improve its customer service and minimize the wasted space on roofs. Therefore, the
problem can be defined as follows: How can the current solar panel placement process
change on customer roofs so that the customer makes the most out of the panel placement?

Goal Propose (if possible) multiple six sigma solutions to the customer for minimizing
wasted space on each roof while keeping the cost per roof as low as possible based on the
CTQ conditions.

Additional factors impacting the panel placement include the following:

• Roof placement operation details.

• Number of people placing roofs.
• Transportation of roof from supplier to customer.
• Panel size determination strategy.
• Cost per roof for panel plantation and the trade-off for different panel sizes.
• Existing wasted space on roofs.

Some further details need to be known when it comes to identification of process details
for solar panel placement:

• Layout Design Specifications.

– Inefficient layout design.
– External factors causing shading.
– Local weather.
– Shape of roof.
– Size of roof.
– Angle of roof.
– Orientation of roof.
– Material of roof.
• Installation Process.
– Panel installers’ inadequate training.
– Improper tool usage.
– Following instructions other than manufacturer’s instructions.
– Mounting equipment used.
– Type of solar panel used.
– Size of solar panel used.
62 3 Defining the Six Sigma Project

– Quality of solar panel used.

– Materials of solar panel used.
• Regulations.
– Codes of building to be followed.
– Safety requirements of the building.
– Zoning regulations.
– Restrictions on permit.
– City regulations for panel placement.
– State regulations for panel placement.

Process details can be supported by images at each step for ease of explanations and
understanding of operations. Such an approach can be particularly useful for defining
waste categories no matter what area these improvements are applied to. The images used
for each waste can help audience to visually help understanding the nature of the waste.
It is important to be careful description of images as the reader may associate a waste
category with an example given for it while the image may not be the only occurrence
(as a shape) for that category.

Example 3.2 A manufacturer wants to improve operations using the following set of KPI
to be measured by the management and engineering group with the data collected on the
following as a part of six sigma project:

• Operation type.
• Good and defected items production.
• Number of operational hours.
• Machine runtime for the operation.
• Product’s sale value.
• Downtime and uptime performances.

A careful strategy and planning are needed to be able to Define the project carefully
and collect data; however, even before starting to focus on a specific focus point needs
to be chosen. This information is typically obtained by verbal communication with the
customer and asking about the top priority focus. Another way of accomplishing this is
by using a Measurement Matrix that is covered in Chap. 4 (this concept is covered in
Chap. 4 due to finding out the details on the specific processes importance, however it
can be used here as well.) A combination of several KPIs may need to be used to be
able to identify the outcome needs. In this case, the overall good item production with
minimal waste is the expected output with minimal cost. Given this fact, we can use the
following formula:

Good qualit y pr oduction = T otal quantit y − W asted quantit y

3 Defining the Six Sigma Project 63

Hence, we can define the project as follows (without much known about the specific
of the product and the manufacturer; if these are known then they need to be outlined in
the defined problem as well):

Define The manufacturer aims to increase the good item production through reduced waste
and improved operational aspects. Hence, how can we change the operations related to
good quality and defected item productions, operational hours, machine runtimes with their
uptimes and downtimes, and overall operational hours so that the quality of the product
either remains same or improves with the product’s sale value to the customer improves?

Goal Use the operational data and the KPIs with the right six sigma approaches to collect
the right data and understand CTQ conditions (including measurable output expectations.)

It may be easy to see for some of the readers that the customer is cost conscious rather
than focusing on quality.
A “Kick-off meeting” is typically needed to understand the project’s scope and details
of the project from the customer’s perspective. This meeting is typically arranged after
identification of six sigma team to work on the project. This meeting is the first step of
team member’s interaction and getting to know each other while the brainstorming and
customer perspective is discussed. The group needs to complete certain tasks at this stage
of the project beginning.

• Project charter. This is a concept to be covered in Sect. 3.7. Given the importance of
documentation with the task-driven timelines, assigned team member tasks, customer
expectations, etc., a Project Charter not only outlines the project scope in its existing
form but also allows updates on the project’s scope as changes are likely to occur.
• Critical-to-quality expectations. Outline the customer’s quality expectations.
• Scope of the project. There are certain details to the project’s scope such as the number
of operations, machines, process steps, etc. that need to be known and the scope of the
project can be shaped based on these factors.
• DMAIC steps. The DMAIC steps to be taken given that there are certain aspects of
the project known.
• End goal. Given that there is at least one contact person for the team with the customer,
this person (or persons) should provide specifics of the customer’s perspectives.
• Timeline expectations. The overall timeline for the project’s completion and timing of
the specific project expectations. An adjustment to either the number of team members
or the project’s scope may be essential for the healthy progress of the project.
• Impact on project outcomes. In relation to the CTQ, what are the measurable impacts
expected by the customer and how is “satisfaction” defined by the customer?
64 3 Defining the Six Sigma Project

• Needs of the project. Additional project-impacting factors such as internal and exter-
nal factors that need to be factored into the project should be discussed. This should
also be assigned as a part of the team responsibilities. For instance, Example 3.2 had
regulations outlined that depend on the state and city rules. These may have to be
accomplished in consultation with a city planner.
• Additional resource needs. Current and additional resources needed for the project’s
completion should be clearly written down. This gap’s closure may have to be covered
from a variety of perspectives that include but not limited to quality, technology, staff,
etc. depending on the project’s scope and goals. For instance, if AI application is
needed to be incorporated during the Improve phase later, then the team may have to
either consult a company that can apply AI techniques for improvement on the project
and determine the cost–benefit trade-off or develop an AI technique for application if
it is a possibility.

3.1 Fundamentals of Define Phase

The initial step of DMAIC is to “Define” the problem on hand. The quotation used for
Define step is because a project’s scope is not verbally necessarily defined; it requires
a more comprehensive approach and more details to be shared about the nature of the
project and the problem on hand. These details can include several charts, diagrams,
company- and product/service-based information, and the customer shared additional
information. Typically, a six sigma team needs to visit the site of the operations to be able
to “have a feeling of” the operations (if possible and relevant) to be able to understand
the work done and shadow the persons who operate the project-related works.
Among this information, there are activities that the customer is not willing to pay
for (NVA), there are activities that the customer is willing to pay for (VA), and there
are also activities that the customer must pay for regardless of willingness. The project’s
scope needs to be defined clearly and carefully, and details of the associated works and
processes should be outlined by using visuals. Such diagrams can include waste, VA,
NVA, and BVA activities as they will be explained throughout this book.

Example 3.3 Suppose the motivation of a project is to reduce time wasted by patients in
each department as a part of healthcare operations that are likely to require operational
changes within a hospital. The healthcare operations may have few flows in certain depart-
ments that need attention by the six sigma team. For instance, a spaghetti diagram can be
designed to help lay out patient traveling frequencies between locations; such a diagram
can help to follow patient and staff movement flows in an organization with the volume of
traveling. Sketching the diagram would clearly show the inefficient movements in the work
environment; however, such a diagram can be challenging to structure given that it requires
extensive patient tracking.
3.1 Fundamentals of Define Phase 65

Highlights of the Define phase include the following:

• High-level explanations of the operations related to the project.

• Process steps associated with the six sigma project to be defined.
• Sketching of the operations by using icons with the associated details (such as a value
stream mapping).
• Some details of the specific operations need to be outlined as they relate to the problem
to be defined.
• Identify process weaknesses based on the waste categories.
• Activity categories (VA, BVA, and NVA) need to be determined.
• Document where, when, how, and by how much there can be improvement in the
process. One of the best ways to identify such improvement opportunities is to ask
personnel operating the works.
• Place the locations of project relevant places on appropriate maps.
• Identify the value proposition of the potential improvements based on their impact on
the project’s possible outcomes.

Some of the tools that can be used during the definition of the project are the following:

• Input-process-output diagram.
• VSM (value stream mapping).
• Supplier-input-process-output-customer diagram.
• Project charter.
• Critical-to-quality conditions and its tree structure.

Cost reduction and operational effectiveness are two of the top six sigma project goals
in the United States with adoption of automation and Lean principal applications in rise
as the focus of improvements. The selection of six sigma focus is driven by the leader
of the organization or sponsor of the project. During the Define phase it is important to
characterize the following for the project:

1. Purpose of the project.

2. Content coverage.
3. Project objectives.
4. Outcomes expected.
5. Money aspect (budget, costs, etc.).
6. Project resources (expected technology utilization, supplier info, etc.)
7. Timeline for project’s completion (lead time, etc.).

Strategies followed and the diagrams designed during the Define section of DMAIC have
important roles in the project outcomes.
66 3 Defining the Six Sigma Project

3.2 Outlining Input, Process, and Output

A visual representation of the work-in-progress can be simply structured by using an

Input-Process-Output (IPO) diagram. This basic diagram is useful for visually represent-
ing input that enters the process, the steps of the process, and the output of the process.
An IPO diagram can be as simple as outlining basic steps of the relevant information or
it can be as complicated as all steps included.
The sketching of an IPO diagram can be completed by following the steps:

• Step 1: A process needs to be known that outlines all the details between input and
output expected. As simple as the statement “choose a process,” as complicated as it
may be when it comes to making such a process choice. The ideal approach is to be
specific about the process itself such as “molding process” if possible.
• Step 2: The output(s) that needs to be the outcome of the process such as product,
service, etc. valued by the customer needs to be clearly stated. The outputs expected
from the process should be included in the output section. These outputs can be the
expected materials to be produced as well as the quality-related items that may be
retrieved from the process. For instance, in addition to the expected product of the
customer, the output can include the left-over materials from the production in the
output section.
• Step 3: Identify the input(s) that need to be used in the process that would result in
the output. Input is determined as the last component in the IPO diagram given that
the output is known along with the process.

The following diagram is a simple example of IPO for flywheel casting example explained
in Example 2.2 [33]. As can be seen on this example, quality flywheel is expected upon
the completion of casting. This IPO diagram is designed at a basic level without including
the details. For instance, a more detailed input would be the sand additives mentioned
one-by-one in the input with their associated inputs as well as the casting instruments.
More details can also be outlined later in a flow chart for the processing steps if it has a
complicated nature.
For some projects only input, process, and output are known. In the case of knowing
the supplier information and customer expectations, it would be ideal to design a SIPOC
diagram to be explained in the next section.

3.3 Diagram for Supplier, Input, Process, Output, and Control

If the input, process, and output are known for structuring IPO diagram, additional sup-
plier and customer information can be added if they are known to design a SIPOC
diagram. A SIPOC diagram should be treated as a cycle in which case the customer
3.3 Diagram for Supplier, Input, Process, Output, and Control 67

Fig. 3.2 A high-level IPO steps designed for flywheel casting [33]

feedback may require returning to any one of the S, I, P, and O steps. The same may not
apply for the IPO diagram as the output is typically known with certainty without any
changes expected. Information flow and workflow in a SIPOC diagram typically has the
structure shown in Fig. 3.3. In this flow, the information flow based on the CTQ condi-
tions flow in the reverse direction of C→O→P→I→S while the workflow has the regular
S→I→P→O→C steps.
Information flow and workflow driven by the critical-to-quality conditions helps with
the SIPOC cycle structure. The following steps outline the structure information flow, and
the workflow is left to the reader to configure:

• Step 1: Critical-to-quality information attained from the customer defines the output
requirements as the expected output would need to be known. The customer can also
provide process and input information in addition to the supplier information.
• Step 2: The output defines the processes that need to be completed.

Information Flow for CTQ

Supplier Input Process Output Customer

Workflow

Fig. 3.3 SIPOC diagram with work and information flows labeled
68 3 Defining the Six Sigma Project

• Step 3: The processes followed would define the input needs for such a work because
the process machines or works can alter the type of input used.
• Step 4: Process-defined input helps to identify the right supplier and materials to be
supplied.
• Step 5: Given that we know the materials to be supplied and the associated supplier,
the SIPOC diagram is completed. The workflow details can be understood from the
customer’s perspective in detail.
• Step 6: Customer feedback and new conditions for the output may bring in new con-
ditions to the project that may appear as new CTQ conditions that can take us back to
the 1–5 steps explained above.

As it has been noted, the customer satisfaction is important; however, the customer expec-
tations need to be clearly written down in an agreement signed by the customer and the
six sigma team as the customer expectations may change over and over throughout the
six sigma project which would not be reasonable and healthy for neither completion of
the project nor for the six sigma team’s success. The above-mentioned six steps of SIPOC
may be repeated in the case when the customer needs to provide additional information
on the product/service; it should not be new conditions that were not previously decided.
This simply takes us to the point of asking whether the customer is always right or not.
A customer would be right only if the agreed conditions for the project are unfulfilled by
the six sigma team; otherwise, the customer can change mind all the time. In some cases,
the customer may be the general population members and surveys drive decision-making
for the company that applies six sigma projects. Such outcomes of surveys can be bene-
ficial in defining a six sigma project. The customer also needs to be understanding when
it comes to the possible outcomes attained from such a six sigma project.
The following supplier and customer information are added to the IPO diagram to
complete the SIPOC diagram in [33]. These steps are typically expected to be more
comprehensive as more details are needed. However, such details cannot be shared with
public as these details require company-based private information to be shared and such
information cannot be shared publicly (FIg. 3.4).
Figure 3.5 contains an example of a SIPOC diagram that also illustrates the flow of the
work from the beginning to the end as the steps of SIPOC are taken [14]. The provided
example focuses on the painting job completion for the customers that visit a repair shop.
Such a project can be designed because of the increasing customer dissatisfaction with
the repair service in which case voice of the customer becomes a critical element of the
overall work to be completed by a six sigma team.
The details presented on a SIPOC mapping depend on the details shared by the six
sigma team. It is possible to incorporate the flow chart into the SIPOC that may require
a slightly different structure of SIPOC than the one presented in Fig. 3.6.
3.3 Diagram for Supplier, Input, Process, Output, and Control 69

Fig. 3.4 Supplier and customer information for the IPO diagram shared in Fig. 3.2 [27]

Fig. 3.5 Vehicle repair SIPOC diagram with step-by-step operations [14]
70 3 Defining the Six Sigma Project

Fig. 3.6 A basic flow chart example for a patient’s information entry in healthcare database

3.4 Structuring Diagrams for Designing Flow

The flow of any operation or service can be represented by a diagram called flow diagram
on which the necessary steps of the product/service flow can be sketched in a nutshell
by using standard icons from the beginning to the end of the product or service. The
following universal symbols are typically used for designing a flow chart:
3.4 Structuring Diagrams for Designing Flow 71

Ellipse Icon: Used as the icon for the beginning and the end of the
flow chart

Arrow Icon: Icon used as the pointer from one icon to another in the
flow diagram
Rectangle icon: Icon used for outlining a process step

Diamond Icon: Icon representing the decision-making in the process.

The typical responses to the question related to decision-making are
“Yes” and “No” with two arrows leaving this icon and pointing to two
different icons

Circle Icon: Connects parts of a flow diagram with a number letter in

the middle

A structured methodology for structuring a flow diagram is as follows:

• Identify the flow chart’s beginning and end based on the focus of the chart and use the
ellipse icon for these two stages.
• Summarize the major steps that take place between the beginning and the end of the
flow chart and summarize them.
• Use pointers between the flow chart icons to indicate the progress of the flow chart
from the beginning to the end.
• Use the rectangle icon in places where there are only process steps existing and use
the diamond icon with “yes” and “no” decisions launching from the diamond to two
other icons.
• All icons in the flow chart need to point to an icon, therefore there shouldn’t be any
icon hanging without a pointer. The icon representing the end of the process is the
place where the flow of operations eventually ends.
72 3 Defining the Six Sigma Project

As explained above, a flow diagram is a summary of the entire process flow from the
beginning to the end. Experience in the flow diagram development is impactful in the
decision-making steps’ specifications. Some of the flow chart completion essentials during
the development of a flow diagram for it to be a successful one is the following:

• Text explaining meaningful beginning and end of the process within the elliptic icon.
• Right phrases placed within the icons.
• Text reflecting the intended meaning within icons.
• There can be many icons placed that can overcrowd a flow chart. Right selection of
steps without too many details would be ideal.
• Right choices of decision-making steps.

Example 3.4 Suppose the goal of the six sigma project is to have a better understanding
and design of patient check-in process within a hospital. The following flow chart given in
Fig. 3.6 is a high-level sketch of a flow chart that does not contain too many details. This
chart can be easily improved and advanced with more details. On this chart, the check-in
process starts with the patient’s arrival at the hospital and ends with the patient waiting for
treatment. Everything in between is used for process steps and the decisions made. Some
more details such as including the database system entry details and the healthcare provider
involvement can be included on such a flow chart.

There are many different forms of flow charts that can include a different dimension
to the basic flow chart structure. A simple example is the functional flow diagram that
includes processing units/departments within an organization based on the functions of
the units and the shape of the flow chart changes. The map in Fig. 3.7 is designed for
Example 2.2 that is useful for understanding overall operations. It is essential to note here
that it would be ideal to use universal symbols (such as ellipse) for beginning and ending
nodes [33].

3.5 Mapping Processes for Business Operations

The drivers of a six sigma project may completely depend on the business processing in
which case the mapping process from the beginning to the end can be subdivided into
VA, BVA, and NVA activities. Such a mapping can be designed by categorizing into VA,
BVA, and NVA activities. Figure 3.8 shows an example of such a map from the beginning
to the end with all icons categorized into the three activities determined. There are some
more details that can be included in such a business process mapping as follows:

• The durations of subprocesses at each level.

• Processing costs per unit time at each level.
3.6 Diagram of Operations Flow Driven by Business Functions 73

Fig. 3.7 Flow mapping used for flywheel casting process [33]

• Entire production’s total cost.

• Process outcomes.
• Business units dealing with specific subprocess.

These details can be incorporated in Fig. 3.8.

3.6 Diagram of Operations Flow Driven by Business Functions

The function within a business defines the functionality of the department within the
organization. Each organization has its own infrastructure and defined departments to
fulfill tasks related to their products and services. For instance, a cell phone company
within the service industry may have the following departments:

• Marketing.
• Human resources.
74 3 Defining the Six Sigma Project

Fig. 3.8 VA, BVA, and NVA distribution of the steps of a flow

• Customer service.
• Technology group.
• Research and development (R&D).
• Sales.
• Financial planning and analysis.
• Accounting.
• Strategic information, etc.

One of the functions that any cell phone company has is customer service while some of
the other functions may change as the company may change. Given this fact, a six sigma
project centered around customer service can include but not limited to the following
functional groups:

• Technology group.
• Tellers working in the customer service.
• Managers of the customer service.
• Technology group members (such as network and security).

The problem’s definition and how it impacts the organization may increase or reduce the
number of functions involved in the company.
3.7 Structuring a Chater for the Project 75

Fig. 3.9 An example of a functional flow diagram with three functional units with the flow chart
components placed on the chart based on the functional units

Similar to the diagram shown in Fig. 3.8, the design of the functional flow diagram
takes advantage of the flow diagram. Figure 3.9 is an example of such functional flow
diagram design. The use of different colors in the design of a functional flow diagram can
help in viewing different departments with their functional units.

3.7 Structuring a Chater for the Project

Documentation of the project information is one of the most important aspects of a project
from the beginning of the project. Given that a six sigma project require a lot of infor-
mation. A project charter is used for documenting some of these details that include the
following information reported and updated several times:

• Project title. A meaningful title to the project needs to be assigned. It shouldn’t be

just called a “Six Sigma project.”
• Goals of the project. The goals that are expected to be fulfilled need to be outlined
for the project both from the customer’s and six sigma team’s perspectives.
• Date of project charter’s update. As mentioned previously, a project charter typically
requires updates as there are changes occurring in the project. The number of updates
and the date of update would be beneficial from the beginning of the project (even
before the Define phase if needed.)
• Business units’ involvement. The department(s) to take place in the project should be
written down.
76 3 Defining the Six Sigma Project

• Members of the six sigma team. Six sigma team members with their roles within the
organization (or in general) such as leader, sponsor, mentor, and members should be
outlined. Their skills can be summarized in this section.
• The Customer(s). The customer(s) of the six sigma project may be internal or external.
• Critical-to-quality conditions. The customer-driven quality conditions and expecta-
tions need to be clearly stated. This section should be as quantifiable as possible.
• Description of the six sigma project. The six sigma project’s brief description should
be written down.
• Scope of the six sigma project. Given the description, the project’s scope description
based on the goals of the project need to be written down.
• Costs/needed budget. Costs and the associated budget needs should be written down.
This part requires a careful structuring approach as changes in project costs can be
challenging later for the customer. Such costs do not necessarily include the costs that
may occur due to adoption of advanced technologies as such costs require additional
considerations; however, the customer should know this fact up front.
• Project’s beginning and ending dates. Dates of DMAIC steps with their objec-
tives’ completion can be helpful for the customer(s) to view the project’s progress
expectations.
• Improvements expected. The improvements that are expected upon the completion of
the project should be written down.

Figure 3.10 shows an example of a timeline sketched in relation to the project outlined in
Fig. 3.5 with the SIPOC steps.
There are certain specs of a Project Charter that makes its design to be a necessity for
a six sigma project:

• Allow communication between the six sigma team members as well as the customer
(if its design allows it to be shared with the customer).
• The changes in the project as it progresses.
• Overseeing the health progress of the project.
• Documenting the very beginning and progressive changes that can be useful in
implementation of the same project in similar areas within the organization.
• Timeline tracking changes and the changes in the information attained from the
customer.
• It serves as almost a formal diary of the project with specific details on the project
changes.

Changes in the project need to be tracked carefully as the customer may attempt to change
the direction of a project unintentionally. A project charter would clearly show this fact
as a historical document of the project expectations. The project charter update decisions
need to be made carefully as too many updates can be NVA. As a simple example, a
3.8 Mapping of the Movements 77

Fig. 3.10 The DMAIC steps and the associated goals and activities planned on weekly basis [14]

project with a budget of two million dollars may be required by the customer to be
completed within a month given predefined CTQ conditions. Updating the project charter
for minor reasons that don’t have much importance may not add value to the project and
the typical focus of changes needs to be centered around “must be completed” specs of
the project.

3.8 Mapping of the Movements

Mapping of movements can be structured by a spaghetti diagram with a sketching of the

operational movement flow. A variety of movements can be designed by using a spaghetti
diagram. Examples include employees’ and materials’ movement during the transportation
of items. The repetition of the operational movements is also reflected in a spaghetti
78 3 Defining the Six Sigma Project

diagram. Given that each “spaghetti piece” represents each movement, the volume of the
spaghetti pieces indicates the volume of the movements. The duration of the movements
needs to be included in the report itself. The documentation of the associated works is
needed for the overall operations which can be explained by using a VSM. One way to
complete a spaghetti diagram can be the completion of process cycle. In the case the
operations cannot be followed due to the sizes of the operations, movements can be video
recorded and sketched. Spaghetti diagram can also be used as a part of a Kaizen event
in addition to the six sigma projects. Examples of improvements through sketching of a
spaghetti diagram include the following:

• Materials’ (such as tools used for machines) flow within a location to track the tool
usage.
• Personnel flow within an environment.
• Document flow for tracking the signatures collected for completing a document
completion process.
• Forklifts traveling within an organization for effective facility movements.

In a transportation and/or manufacturing setting, the pathways for moving entities are
shown with the curves or lines in a spaghetti diagram between locations placed on the
diagram that include but not limited to the following:

• Trucks.
• Forklifts.
• Employees.
• Materials.
• Documents.

In a healthcare setting, the following can be the associated curves that can be sketched
during the operations:

• Doctors, nurses, interns, etc.

• Wheelchairs, moving trays, etc.
• Ambulances, etc.
• Document flow.
• Material flow.

Figure 3.11 shows an example of a spaghetti diagram in a healthcare lab setting with
different colors representing different personnels’ movement in the environment. On this
example, locations listed include the following:
3.8 Mapping of the Movements 79

Fig. 3.11 A spaghetti diagram designed by a six sigma team for displaying a lab’s worker flow

• Histology lab.
• Workstations.
• Chemical storage.
• Autopsy lab.
• Waste disposal room.
• Peloris machine.

In this setting, the two colors, red and blue, represent the two technicians working in the
environment and the spaghetti diagram is designed as a representation of the movements
of these technicians for chemical material handling. Such a graph needs to be designed if
it is necessary for the project to be viewed; otherwise, designing such a graph would be
NVA itself for the project. The number of “spaghetti pieces” on the graph indicates the
frequency of movements. As it is explained previously for other displays of operations,
spaghetti diagrams also require explanations. In some cases, it may be ideal to include
certain measurements on a spaghetti diagram such as the distances between units, number
of items carried, etc.
80 3 Defining the Six Sigma Project

3.9 Structuring a Tree for Quality

A way to structure Critical-to-Quality (CTQ) conditions is by using a tree structure and

incorporating the hierarchy involved based on the customer valued tree levels. General and
specific customer requirements can be included in a CTQ tree. The important drivers of
a CTQ tree depend on the outputs and inputs with the involved process. The levels of the
CTQ tree are variable and need to be decided by the project team and the customer. CTQ
tree can also be used in the Analysis phase of DMAIC approach as the data collected
from the customer would need to be analyzed and indicate what customer values the
most in the output. The highest level is the first level of the CTQ tree that reflects the
most general expectation of the customer from the output. The branching of the CTQ tree
levels’ structure is driven by customer expectations. Questions that can be answered to
structure a CTQ tree include the following:

• To be able to design Level 1 category, what is the most general expectation from the
output?
• To be able to design Level 2 category, answer the question “What are the 2nd level
expectations of the customer?”
• To be able to design Level 3 category, answer the question “What are the 3rd level
expectations of the customer?”
• To be able to design Level 4 category, answer the question “What are the 4th level
expectations of the customer?”

Example 3.5 A level-by-level approach for structuring a CTQ hierarchy tree starts with
Level 1 representing the highest level purpose for an item production. The two possible
cost reductions at the second level are machine and transportation cost reduction which
would be the two branches from Level 1 to design Level 2. Breaking down Level 2 into
several other subcategories would give Level 3 elements that can include “machine operator”
and “machine cleaning” improvements. Such details can be included to further detail the
example. The depth of the tree given in Fig. 3.12 is four which is the number of levels based
on the importance driven by the CTQ conditions.

A CTQ tree can be driven by the Voice of Customer (VoC) that helps with making a
connection between the critical-to-quality conditions and the customer expectations that
can be determined through surveying. A VoC is a good way for identifying the improve-
ment opportunities and understanding customer’s view of the outcomes based on the
“needs” analysis.
3.10 Initial Phase of Value Stream Mapping for Operations 81

Level 1: Highest Level Customer

Expectations' Summary

Level 2: Second level Level 2: Second level

expectations expectations

Level 3: Third Level 3: Third Level 3: Third level

level expectations level expectations expectations

Level 4. Fourth Level Level 4. Fourth Level 4. Fourth Level

Expectations Level Expectations Expectations

Fig. 3.12 A four-level CTQ tree structure that depends on the voice of the customer

3.10 Initial Phase of Value Stream Mapping for Operations

The layout of value stream map (VSM) requires using a set of universal symbols/icons to
be able to layout the essentials needed for the project on a single platform. This sketched
map outlines the operations in a nutshell, and it is driven by the six sigma project’s scope
for the defined problem. The initial phase of VSM for a manufacturing or healthcare
application typically has the following representations on the map:

• All essentials that need to be known for the supplier and customer:
– Method of communication between the customer and the organization.
– Shipment method used and its duration between customer and supplier.
– Design of information flow.
• Process and operational considerations:
– Process duration, expected quantities, and number of personnel.
– Structure of inventory.
– Design of interaction, communication, and shipment between upstream and down-
stream processes/operations.
– Processing and operational logic used.
– Quantities.
82 3 Defining the Six Sigma Project

• Time considerations:
– Process times.
– Duration of shipment between processes and downtimes.
– Lead time.
– VA time.
– NVA time.
– BVA time.
– Processing time.
– Setup time.
– Schedule.
– Downtime.
– Time per shift.
– Time to complete operations.
– Toal wasted time.
– Total value-added time.

A VSM is typically structured by using the following universal icons:

This icon represents a process step with its information. The

information that can be shown within this icon include the
following:
• The number of operators.
• Quantity of shifts.
• Unit used for quantification per shift.
Icon demonstrating personnel.

The rectangle icon is used for processes and the associated

information display

This is the icon used for entering information as a part of a

process that includes but not limited to the following:
• Operator number
• Cycle time.
• Size per batch.
• Setup time/uptime.
• Condition of work in process (WIP).
• Downtime.
• Defect percentage.
• Number of shifts.
• Lean six sigma process.
3.10 Initial Phase of Value Stream Mapping for Operations 83

A theoretical example to a process on a VSM is as follows:

This icon is used for customers and suppliers on VSM.

The truck image is used for truck shipment with its duration.

The ship image is used for cruise transportation with its

duration.

Mailing by using air transportation with its duration specified.

Inventory usage icon with the quantity of items used from the
inventory and the duration of item usage.

Finished goods’ directional movement from location to

location.

Pushing materials to the next step in processing.

84 3 Defining the Six Sigma Project

Electronic information flow.

Flow of information is manual.

Safety stock or buffer with duration of use.

Sign of a placed supermarket for inventory placement

allowing workers to store specific items and visit to shop
for these items as needed. Such supermarkets are typically
established between upstream and downstream.

Pull logic application by withdrawing items.

Item flow by using FIFO (first-in-first-out) logic.

Icon used for production leveling to determine the level of

production loads. This is different from batching.

Inspection icon used for visual examination.

Timeline icon with lead time derivation by using NVA,

VA, and BVA. The upper bar in the middle on the image
is used for the process’ duration while the two lower bars
before and after the middle bar are used for the time
elapsed before and after the process.
The icon for Kanban application for replenishment
purposes.
3.10 Initial Phase of Value Stream Mapping for Operations 85

Kanban sign used for withdrawing to be able to transfer

parts from location to location.

The Kanban icon used for production of upstream to

provide needed items/services.

The icon used for Kanban signaling replenishment (i.e., the

need for production or service) of items/service due to
reaching re-order point.

Sign used for improvement needed in a location on VSM.

A board can be used by the six sigma team to structure the map of the operation. Web-
based resources and some other software packages can also be used for designing VSM
by using the universal icons outlined above. The value steam map designed in Fig. 3.13
is generated during a real-life six sigma project without quantities entered on the graph.
Another example of a VSM shown on Fig. 3.14 is generated for energy industry in
[29]. This map is designed by incorporating the energy usage for each time-related VA

Fig. 3.13 Initial phase of VSM without numerical values displayed on the image
86 3 Defining the Six Sigma Project

Fig. 3.14 The VSM of energy industry manufacturing process improvement [29]

and NVA activities in manufacturing of the relevant items. The collected data is based
on each operation and the associated energy loss based on the four processes. During the
manufacturing of associated items, the output is the manufactured angles that are galva-
nized with cleats and holes stamped and notched. This process starts with the input of
raw material (i.e., angles) that goes through four processes. Raw materials (angles) first
undergo the straightening process, in which bend or distortion present in raw materials is
removed. Following this stage, the angles undergo CNC machining during which punch-
ing, stamping, and cutting operations are accomplished. What follows CNC machining is
notching of the angles that is followed by galvanization of the angles as the last stage.
A protective coating of zinc is placed on the angles and cleats during galvanization. This
completes the manufacturing process with the product to be ready for shipment.
Figure 3.15 shows the VSM after identification of improvements in the manufacturing
process outlined for Fig. 3.15.
The second phase of VSM includes outlining the locations where improvements can be
applied with sketching on the pre-improvement map. The third and last step of structuring
VSM is independent from the initial version and reflects the new business strategies with
the changes made on the initial VSM.
3.11 Satisfaction, Dissatisfaction, and Delighters 87

Fig. 3.15 The VSM after identification of improvements in Fig. 3.14 [29]

3.11 Satisfaction, Dissatisfaction, and Delighters

A Kano diagram is used for reflecting the customer’s satisfaction levels which can be
categorized into three:

• Satisfying. The customer can easily outline the expectations as a check list and the
satisfaction of these outcomes results in “satisfying” condition. The level of satisfaction
can be identified by the quantification of the outcomes.
• Dissatisfying. The CTQ conditions that are unfulfilled would result in customer dis-
satisfaction that would result in the dissatisfying condition. This outcome can also be
quantified depending on the associated operations and collected data.
• Delighters. A “delighter” is an improvement accomplished by the six sigma team that
would not be expected and this improvement results in better customer satisfaction.
Such an improvement is typically unplanned by the six sigma team.

There are certain action items that have higher priority than the others that’s driven by
customer’s CTW conditions.
88 3 Defining the Six Sigma Project

Fig. 3.16 Dimensions of a Kano model with various levels of satisfaction of the customer [25]

Figure 3.16 shows a way to design a Kano model theoretically by using collected data
[25]. Possible explanations and improvement dimensions need to be entered on the graph
and the associated degree of customer satisfaction as satisfied, dissatisfied, and delighter
can be determined.

3.12 An Example in Health Care

So far, we have given manufacturing-related examples in the Define phase. Now we will
provide an example related to healthcare operations.
Suppose the goal is to improve patient treatment operations in a clinic setting through
bettering room utilization. The following are some of the observations that are deter-
mined by a six sigma team of three industrial engineers upon visiting the facility with the
associated map:

• Small registration area.

• Three waiting areas.
• Sterilization and inventory, one room for each.
• Ten exam rooms (Fig. 3.17).
3.12 An Example in Health Care 89

Fig. 3.17 An example of a facility layout with 10 exam rooms and some other locations

The scope of the project was to optimize the process flow to fulfill the following:

• Increase the utilization of the exam rooms.

• Reduce the inefficiency in the clinic.
• Increase patient throughput and revenue yields.

The team structured the following SIPOC steps:

The associated flow chart has the structure shown in Fig. 3.19 (Fig. 3.18).

Fig. 3.18 The SIPOC diagram for the healthcare example

90 3 Defining the Six Sigma Project

Fig. 3.19 The flow chart designed for Fig. 3.18

The following data is decided to be collected during the progress of the project by the
team:

• Data collection: Procedure time includes 5 min of preparation time and 5 min of clean-
up time.
• Days of data collection: Tuesday, Thursday, and Fridays throughout 30 days.
• Information projected for Exam Room 10 is skewed noting that it is used mainly for
X-rays.
• Average revenue patient generates is average of the most general procedures done in
the clinic.
• 251 business days for the year.

The team further planned to measure the associated data, analyze the measured data,
suggest improvements to the clinic for further waste reduction, and establish a control
mechanism for making the improvements sustainable in the clinic. These will be explained
later throughout the associated MAIC chapters.
3.13 Questions 91

3.13 Questions

Question 3.1 The following flow chart is designed to be able to define a problem:
92 3 Defining the Six Sigma Project

Using this information, answer the following:

• What may be missing on this flow chart to make it meaningful?

• Are there any mistakes made on the design of this flow chart by considering the way
that a flow chart’s design covered in this chapter?
• Which of the concepts covered in this chapter can be structured by using this
information?
• Based on the explanations in the flow chart, what can be improved? Is it easy to
determine it by using this flow chart? If yes, please explain what can be improved. If
not, what additional information needs to be known?

Question 3.2 The following graph is designed for Question 3.1 to be presented to the
customer. What needs to be improved on this graph to be able to make it meaningful to be
able to follow the operations?

Question 3.3 The following wastes are determined by a six sigma team for the project
defined in Question 3.1. Given this information and the flow chart, what other Define phase
components can be covered for this project explained in this chapter?

Motion Non-value-added activity of a people and machine movements are considered as a

motion waste. The customer’s location is a big facility where forklifts are used frequently
between the aisles. Therefore, the motion of the employees, forklifts, plays a significant role.
3.13 Questions 93

If they are overused, it would be a waste. We observed that, even though they were using
wireless communication devices to communicate, an employee approached the wrapping
area to ask or talk about something to the employee by using the forklift for no reason. So,
the employee used the forklift out of its purpose of use. This is a waste of motion.

Transportation Waste that can happen while relocating the items. We observed that an
employee approached an aisle to pick up a pallet, then dropped the pallet into the next aisle
which he was not supposed to. Then he realized his mistake and went back to the aisle where
he dropped off the pallet, picked it back up again, and dropped off the correct aisle in the
back. This is a waste of transportation.

Overproduction There were too many pallets in the aisles that were supposed to be moved
there later. Because the pallets were moved there early, there is a chance that the goods in
the pallets might go bad or expire or get damaged. In a facility anything can happen. So, it
is very important to produce and do things on time.

Overprocessing f the employees were working at a high speed, and they were doing many
jobs in a short period of time. It is good that they were working hard, it is always good to
never forget the fact that working smart matters, not hard. Therefore, when they process
more than they were supposed to or needed, then it is overprocessing.

Unused Human Capital In the wrapping area, due to the delay on the arrival of pallets from
the aisles, the employee in the wrapping area was just standing there, doing nothing.

Waiting It causes a description of a workflow. Customers would never want to wait or face
a delay after they order. They would want to have their order on time. It is very important
to send the processed products on time to the customer to fulfill them with satisfaction. In
the customer’s facility, we observed that some pallets were waiting to be shipped in front
of the truck loading area. If they were loaded into trucks as soon as the truck arrived at the
facility, there would not be a waste of time. This is a waste of waiting.

Inventory It can be related to overproduction. Items in the inventory are not in the main
process. There were many pallets that were waiting to be shipped in the aisles. As is written
in the overproduction section because the goods in the pallets are consumable goods, there
is a chance that they can go bad, expire, or get damaged.
94 3 Defining the Six Sigma Project

Question 3.4 The following flow chart is designed by a six sigma team for explaining the
steps of an operation in healthcare facility.

a. Given the guidelines of a flow chart, what needs to change on the design of the graph and
what information would be needed?
b. Do you think the graph is clear to be able to explain the scope of the project? Please
explain if any additional information is needed.
Measurements for the Defined Problem
4

It was the definition of the problem that we focused on during the define phase. What
follows in the DMAIC approach is the measure phase where the focus of the project
shifts to the measurements needed for the defined problem to be analyzed later during the
analyze phase. The define phase provided us with information on several aspects of the
project focus:

• Customer’s interest and focus with the quality expectations and the associate KPIs.
• Six sigma team formation.
• Customer’s expectations from the project team with the SIPOC understood by the
team.
• Formal definition of the problem in addition to the materials that need to be included
for the define phase.
• A comprehensive report on the project observations for the define phase.

Data needs to be collected for the appropriate measurements that would help us to under-
stand waste and inefficiencies of any aspect of the project in place. Some organizations
already have data collected and this data may be requested to be used by the six sigma
team. The following are some of the scenarios that occur relating to the data collection
and the six sigma project scope:

• Existing Data and Project Scope Overlap. This is the case when the existing data
overlaps with the scope of the project and the collected data can be used as a part of
the measure phase. The six sigma team needs to figure out the conditions under which
the data is collected, the definitions of the performance indicators used, and how the

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 95

E. Tokgoz, Six Sigma and Quality Concepts for Industrial Engineers, Synthesis Lectures
on Engineering, Science, and Technology, https://doi.org/10.1007/978-3-031-55740-8_4
96 4 Measurements for the Defined Problem

relating operations work. The units used for measurements and the data collection
strategy also needs to be studied by the six sigma team
• Existing Data and Project Scope Somehow Correlates. In this case the existing data
may relate to the scope of the project somehow and therefore, it can be useful to the
six sigma team. The project team may benefit from the data by going over it and
incorporating this data into the additional new data that needs to be collected.
• Non-existing Data and Six Sigma Team’s Data Collection. There may not be any
existing data that relates to the project and the six sigma team may be able to collect
data depending on the complexity of the project. If the project does not overwhelming
data, then the team can collect.
• Non-existing Data and Organization’s Personnel Data Collection. There may not be
any existing data relating to the project and the six sigma team may not be able to
collect data because of the complexity of the project. This may result in either the
customer or the six sigma team designing a data collection strategy to be able to figure
out the data needed to fulfill the expectations for the defined problem.

Measure is the phase for the exploration and assessment of data. An industrial engineer
typically knows data analytics techniques to be able to implement data analysis after
implementing time studies. Data collection decisions made for completing a six sigma
project drive the analysis and the following improvements to be made; therefore, mea-
sure phase has critical importance when it comes to attaining sense-making outcomes. An
industrial engineer typically either knows or studies to understand a system’s innerwork-
ings to be able to improve it; stepping into a system’s setting with industrial engineering
knowledge, observing and understanding the big picture that relates to the project scope,
and collaborating with others are essential ingredients of a successful six sigma project.
There are several questions that you would need to answer for such an approach during
the measure phase:

• What is the system’s nature that the project takes place?

• What needs to be understood to be able to measure?
• What would be a good strategy for collecting data?
• How can the data be collected ?
• Is there data collected relating to the project and how does it relate?
• Does the customer approve of the data collection strategy (if you have an idea about how
to do it?)
• Does the team understand the operations relate to the data to be able to track the meaning
behind the data? If not, what needs to be done to be able to do it?

Regardless of whether the data exists or not prior to starting a six sigma project, the six
sigma team needs to understand the associated dynamics of the operations and design
several strategies for working out the details. Understanding what is needed and how to
4 Measurements for the Defined Problem 97

identify them is the initial step in such an approach. In some projects, you need to be a
part of the system and work out the steps of the work to be completed (if possible) to be
able to learn, measure, understand, and improve the system based on its special dynamics.
Developing a “gut feeling” of the way that system works would also help you build an
intuition for the entire project. There should be at least one six sigma team member that
can answer system related questions, meaning a person with conceptual knowledge of the
operations in place is needed in the team. It is ideal for each six sigma team member
to get “hands dirty” and spend time to understand the system in place and collect data
collection even if data exists!
The most ideal case is to have at least one person who is directly involved in the six
sigma project’s scope who knows most of the associated tasks. The scope of a project
may require the involvement of at least one person from several different departments
depending on the project-related stakeholders. A person operating a system would have
the experience and knowledge, but the data collection strategy may require the person to
be trained for it to be able to make sure that the data is collected right. Data collection
strategy overall requires careful steps to be taken:

• Customer communication is needed several times to structure the right data collection
strategy.
• Become a part of the project-related system to study, watch, learn, and document its
details.
• Identify key contacts that need to be known for asking questions and communication.
• Document the data collection details.

The first concept to be covered in this chapter is possible strategies of sampling. Once the
sampling strategy is known, there are several displays that can be used for summarizing
the collected data. Using statistical tools such as random variables may be needed in appli-
cations that are not covered in this chapter; however, random variables can be introduced
by fitting models to a data set. Using charts, diagrams, plotting techniques, and figures to
help understanding the structure of the measured data is also expected to be completed
in this chapter; therefore, the following are some of these techniques introduced in this
chapter:

• Box plot.
• Time Series plot.
• Bar chart (Histogram).
• Pie Chart.
• Pareto Chart.

Some of the key factors requiring attention during the data collection are listed in Fig. 4.1.
98 4 Measurements for the Defined Problem

Fig. 4.1 Data collection elements that can be used during a six sigma project

The following are the explanations of the factors outlined in Fig. 4.1 for collecting
data that would have a proper structure:

• Operational Definition. Understanding and defining the operational structure is the

first step to the data collection that needs to be completed for any project. All needed
information needs to be defined in the way they are used within the scope of the
project. One should not forget that such definitions can change even within the same
company.
• Satisfaction Factors. The satisfaction factors from stakeholders’ perspectives need to
be checked and adopted to the project.
• Measurement Choices. The measurements needed to be chosen in alignment with the
stakeholders’ interest and the customer is a part of (or the only) stakeholders.
• Data Type. Data type in this context is the nature of the data as a means of qual-
itative or quantitative nature of the data. For instance, the information needed may
be derived by using qualitative surveys while quantitative information may be derived
from numerical data collected or surveys. The following are some of the details for
data types that can be used:
4 Measurements for the Defined Problem 99

– Quantitative data. Data that has a numerical nature.

– Qualitative data. Verbal or non-numerical data that can be collected by
Interviews.
Surveys.
Verbal communication.
– Image data. Image data consists of images that respect a certain theme in the
application. This approach can be helpful when the associated physical data is not
possible to collect. For instance, the theme can be focusing on all possible defects
that exist on an item during production. The images of such defects would be the
image data for this application. It is ideal to collect image data as low as possible
as images occupy a large amount of memory.
– Tangible data. Collection of tangible data may be needed depending on the scope
of the project. For instance, a label producer may have labels manufactured for a
certain product and the associated tangible defect data may need to be collected for
raising different wastes that occur within the production machines. The quantity to
be collected needs to be determined carefully as it takes time and requires space
that can be not only costly but also overwhelming. Personnel that operate the system
can be the ideal person to collect data for the six sigma project.
• Sampling notes. Taking notes is an important aspect of the six sigma project and
it needs to be done for any project. The team should be able to collect appropriate
sampling notes during the project steps.

A systematic approach needs to be followed to be able to collect data. Such an approach

needs to be organized in a sense-making way. The CTQ tree approach covered in Chap. 3
is one of the ways to determine such conditions. The following is an example of such a
data collected and the decision made during a six sigma project.

Example 4.1 The data provided in Fig. 4.2 is an organization of qualitative and quantitative
data that a six sigma team utilized in [34]. The purpose of the project was to identify the
issues existing within a company on a certain machine and the time it takes to fix these
issues.

A tactical approach to select critical factors of machine breakdown in the center less
grinding (CLG) section in an automotive industry to make meaningful decisions. There
are six critical attributes displayed in Fig. 4.2 selected as a breakdown of KPIs. Meantime
between failure is determined to be the most important factor by the customer for the
center less grinding section.
100 4 Measurements for the Defined Problem

Fig. 4.2 An example of data gathered with both qualitative and quantitative information [34]

4.1 Quantifying Measurements Through Statistical

Distributions

Statistical distributions are commonly used for representing measurements. The following
example contains the display of such sample data.

Example 4.2 The following bar chart is an example of patient data collected based on the
age groups Fig. 4.3.

This graph represents a distribution of the data while it is possible to fit a statistical
distribution to the associated data.
4.2 Sampling and Collecting Empirical Data 101

Distribution of Patient Selection

16
14
FREQUENCY OF PATIENTS

12
10
8
6
4
2
0
9_15 16_22 23_29 30_36 37_43 44_50 51_58
PATIENTS GROUPED BY AGE

Fig. 4.3 An example of a patient distribution based on age grouping and the associated frequency

4.2 Sampling and Collecting Empirical Data

Sampling technique followed is the most important step in the measure step of DMAIC
approach. The target population chosen, and the collected sample data from it should be
in alignment with the six sigma project’s goals. The sampling size and strategy should be
in alignment with the population’s size and strategy. Two major categories for collecting
are the following that branch out to several subcategories:

• Non-random sampling.
• Random sampling.

These two sampling techniques and their subcategories are briefly explained in Fig. 4.4.
These sampling strategies will be explained in the next subsection.

4.2.1 Designing Data Collection Through Non-Random Sampling

Designing data collection through non-random sampling is used when a certain strategy
needs to be followed for data collection. The following strategies can be followed:
1. Non-Random Sampling from Subpopulation. Sampling from subpopulation requires
a specific sampling strategy that does not have a random nature.
102 4 Measurements for the Defined Problem

Fig. 4.4 Brief explanations of some of the sampling techniques

Example 4.3 In a hospital, suppose there are 20,000 patients seen every year and the hospital
administration wants to determine the quality of the patient treatment by using subpopu-
lation sampling. The administration believes that choosing 30% of the population which
corresponds to 6000 patients is sufficient to derive the needed outcomes and the sample
gathering is expected to be completed each month throughout the year. Given that 6000/12
= 500 patients per month, 5000 patient’s information needs to be collected each month. If
we break it down further to daily data collection, then we find approximately 25 patients to
be collected per day assuming that the hospital operates only 20 days of the month.

2. Non-Random Topic-Specific Sampling. In the case of non-random sampling, the data

collection may be needed based on the characteristics of a certain categorization in the
population that drives the sampling. These topic-specific categories are used for sampling
from the population with their respective ratios.

Example 4.4 In a hospital, suppose there are 20,000 patients seen every year and the hospital
administration wants to determine the quality of the patient treatment by using topic-specific
sampling in which case the topic is driven by the 20 treatment types of the patients. The
corresponding sampling strategy should be designed by the six sigma team based on the
data collected.
4.2 Sampling and Collecting Empirical Data 103

3. Non-random Systematic Sampling. A specific pattern is used for non-random

systematic sampling for data collection.

Example 4.5 Suppose there are 20,000 patients in a hospital seen every year and the hospital
administration wants to determine the quality of the patient treatment in general through a
systematic sampling approach by using 20% of the patient data. In this case, 4000 patients’
data will be collected and one way to do so is by recording every fifth patient’s treatment
information.

4.2.2 Strategically Sampling Through Random Choices

The randomness of sampling can be strategized into three categories:

1. Sampling Through Clustering.

2. Stratified sampling.
3. Simple sampling.

(1) Sampling Through Clustering. The initial step of cluster sampling relies on splitting
a population into groups first, and then sampling one of the subgroups randomly. The
criteria used for grouping needs to have a reasonable design aligning with the purpose of
the project.

Example 4.6 Suppose a hospital wants to find out the patient satisfaction levels, and the
problems faced by them after the treatment is completed in the hospital. There are 20,000
patients seen that year and the hospital decides the randomness to rely on a random group
of patients selected based on the monthly clusters of patients seen.

(2) Stratified Sampling. Stratification means dividing population into homogeneous sub-
groups, and this definition is the driver of stratified sampling; a certain percentage of each
subgroup can be chosen randomly for sampling in this sampling strategy.

Example 4.7 Let’s consider the World Health Organization (WHO) and suppose WHO
wants to collect data from each continent based on the treatment of a certain disease. The
data can be collected from the existing 7 different continents based on the number of existing
patients by calculating 10% of each continent’s patient numbers.
104 4 Measurements for the Defined Problem

(3) Simple sampling. As the sampling strategy suggests, sampling strategy is done in
a random fashion from the population. In this case, each population member has an
equally likely chance to be drawn as the choice is made randomly. The distribution of
this approach can be by using a uniform distribution.

Example 4.8 Suppose a hospital wants to select patients randomly to be able to identify
their opinions of the treatments they received from the hospital. If there are 20,000 patients
treated in the hospital and the hospital wants to collect 20% of this data, then randomly
chosen 4000 patients’ need to be picked to be able to apply a simple sampling strategy.

4.2.3 Collecting Survey Data and the Associated KPI

Determination

Collecting data through surveying is a technique that requires the design of a questionnaire
that targets the right questions to be asked in the proper manner and format. The structure
of the survey is project focused; therefore, the needs of the project need to be reflected in
the report by the six sigma team. After designing such a survey, it should be tested by the
team to be able to modify it before checking with the customer on its appropriateness. The
survey needs to change (if needed) after the customer reviews it with its appropriateness
for the associated purpose. The survey will be ready to hand to a focus group after it is
reviewed by the customer. This focus group can provide feedback on the design of the
survey which would then be distributed to the sample group for data collection.
In addition to the information above, your communication skills have critical impor-
tance particularly in this section. You need to make sure that everyone who fills out the
surveys knows the purpose of the survey and the associated expected outcomes. Some
of the people filling out the survey may underestimate its importance. The design of the
survey should be simplistic enough for any person to be able to fill it out easily without
additional explanations. The expectation is for you to reflect your strong understanding
of how data mechanism works through your observations on the survey. You should also
consider the consequences of the survey questions you ask and the analysis of the col-
lected data afterward. Asking 20 qualitative questions can put a six sigma team in trouble
as it could be very difficult to analyze and extract meaningful outcomes after collecting
such data!
The following is an example of a survey’s first draft prepared by a six sigma team for
collecting data from a clinic. This version of the survey was not tested by the six sigma
team on-site; therefore, it is open to questioning its appropriateness to be distributed for
data collection within the clinic. The goal was to collect data from nurses and doctors.
4.2 Sampling and Collecting Empirical Data 105

Brainstorming of the six sigma team members can be extremely helpful to the team
on a board just like it is shown in Fig. 4.5.
A data collection plan needs to be strategized by the six sigma team. A “divide-and-
conquer” approach can be followed: In such an approach, each subgroup can receive their

Fig. 4.5 The use of a white board for outlining survey needs, expectations, issues, and concerns
106 4 Measurements for the Defined Problem

own duty related questions after the survey group is separated into subgroups. The col-
lected responses need to be used for understanding each subgroup’s expectations, needs,
issues, etc. that are faced to be able to extract meaningful outcomes. The survey can be
helpful to figure out the best practices as well as issues handled for different cases. Sev-
eral iterations of surveys may have to be given under certain circumstances due to their
necessity of several iterations; however, right number of surveys should be given (please
keep in mind that many surveys may have to be given due to their importance; therefore,
a six sigma team should not try to reduce this number of surveys just because it is too
many.) Some of the basics of survey design include but not limited to the following:

• Survey design needs to be simple with the right number and type of questions.
• Short answer questions are ideal to ask with a Likert scale; possible options include
1–5, 1–10, and yes/no question types.
• The questions need to be concise and clear for anyone.
• Each survey should be administered to the right group and sample of the group with
the associated sample size determined carefully.
• The number of survey iterations needs to be identified based on the feedback attained
from the sampling group.
• Typically, each survey is expected to contain an open-ended question for entering
sampling group members’ opinions.

For statistical significance or to attain the expected number of data collected, a specific
amount of data needs to be collected. The project goals, customer expectations, indus-
try regulations, the system’s periodic behavior, the KPI measurements, and some other
technical area considerations should always be kept in mind during survey design.
Organization and planning of the data gathering will be covered next for understanding
the data collection planning methodology.

4.2.4 Organizing and Planning the Data Gathering

Gathering of the sample data requires initial planning with the right questions asked in
the right place. This planning needs to target the analyze and improvement phases to be
meaningful with the control structure to be also feasible; therefore, a good understanding
of the six sigma team is expected in the define phase while the measure phase is the first
key section to put all knowledge into works.
One of the organizational approaches, just like it is explained in Example 4.1, is to
divide-and-conquer (or a top-down work breakdown) a process into subprocesses to be
able to reduce waste in each subprocess that is expected to reduce the process waste. In
the case when a survey is given, the right questions need to be asked that target the end
goal of the project. As the six sigma team’s goal is to move onto an improved version
4.3 Designing Graphs and Figures 107

of the system, such a goal cannot be accomplished without understanding the existing
system dynamics and the opinions of the persons operating them with measurable KPIs.
Depending on the intensity of the data collection, splitting the general task into several
sub-tasks, and then collecting data for each task can be a meaningful approach.
For instance, suppose you want to improve the time a transportation system works for
a company. In the scenario below, the “customer” is the customer of the six sigma team
while the “consumers” are purchasers from the customer. The following are the steps that
can be followed to be able to identify the details of how data can be collected for the
customer and the consumer:

• Step 1. Find out the transportation types to categorize them.

• Step 2. Collect a certain period’s consumer data (i.e., six sigma team’s customers’ data)
to be able to identify the associated volume of transportation based on the types.
• Step 3. Use splitting of the data into the categories and further split the data into
subcategories depending on the importance of the KPIs.
• Step 4. Analyze the data to be able to extract meaningful outcomes.
• Step 5. Compare the analyzed outcomes with the customer and consumer expectations.

The splitting of the data into subcategories to be able to analyze it is where the divide-
and-conquer approach has the critical importance.

4.3 Designing Graphs and Figures

Designing graphs, charts, figures, etc. as the data displays is extremely useful for under-
standing and visualizing the sampled data as well as its analysis. Some of the frequently
used simple visuals that can be used for such displays are the following:

1. Box Plotting.
2. Fish-bone (Cause and Effect) Diagram.
3. Histogram (Bar Chart).
4. Pareto Chart.
5. Pie Chart.
6. Time Series Plot.

One should not forget that each designed graph needs to be explained within the text as
these graphs are not necessarily self-explanatory in many cases.
108 4 Measurements for the Defined Problem

4.3.1 Bar Chart

A set of data points can be represented by using a bar chart (histogram) that allows
the representation of the data in different formats. The data is grouped into “bars” to
display a meaningful outcome. There are different ways to design a bar chart and it is
not only useful for visual representation of a given massive data set but also helpful in
accomplishing statistical tasks. There are different ways to use a bar chart for displaying
data that include but not limited to the following:

• Displaying frequency of the data points with respect to their groups.

• Probability distributions.
• Statistical distributions with its mean and standard deviation values.
• Grouping data points to display progress.
• Demonstrating different KPIs with individual bars.

An example will be given next to display several different bar chart options to illustrate
different meanings that can be attained from such histograms.

Example 4.9 Let’s consider a data set with the disease information collected below as the
KPIs over the span of years 2000 to 2010.

• Infectious and parasitic diseases.

• Neoplasms.
• Endocrine; nutritional; and metabolic diseases and immunity disorders.
• Mental illness.
• Diseases of the nervous system and sense organs.
• Diseases of the circulatory system.
• Diseases of the respiratory system.
• Diseases of the digestive system.

During a six sigma project you are expected to use such a data to analyze such a data and
display the outcomes. An inexperienced six sigma team typically displays the collected
numbers (raw data) to be able to show the values of the data; however, this doesn’t
really serve a meaningful purpose. The audience cannot follow the data by reading the
numbers alone in a crowded data set! This takes us to the design of a histogram to be
able to display the data. One way to demonstrate such data is by displaying all diseases
together. Figure 4.6 demonstrates such a display per year on which each color-coded bar
demonstrates one of the disease types.
The six sigma team needs to extract meaningful outcomes from this graph. As men-
tioned above, there are several ways to display the data to be able to extract meaningful
4.3 Designing Graphs and Figures 109

Fig. 4.6 Disease data per year in bar chart format

outcomes. One other way to represent the same data is by displaying the cumulative
nature of the data by adding the color-coded disease information for each year on the
same bar demonstrating the year. Figure 4.7 is an example of such demonstration for this
example that displays the cumulative nature of all the diseases for each year that allows
us to extract different meaningful outcomes when compared to Fig. 4.6.

Fig. 4.7 Cumulative disease numbers per year

110 4 Measurements for the Defined Problem

Fig. 4.8 Cumulative bar chart categorized into disease type with the year representing the incre-
menting data

Another way to display the same data demonstrated in Figs. 4.7 and 4.8 is shown in
Fig. 4.8. This figure categorizes the data by disease type in which case the number of
years is displayed on the bar chart’s cumulative nature with the colors represented by the
years.
Other bar charts can also be produced to be able to demonstrate frequency distribution,
probability distribution, and statistical distribution of the data that is left to the reader to
practice.

4.3.2 Plotting Time-Dependent Data

Time-dependent data can be displayed by using time-series chart. This chart can also
be used for displaying data that progresses in the input in a meaningful fashion. This
visualization needs to be designed carefully to be able to make the most sense out of the
design.

Example 4.10 Referring to the information given in Fig. 4.9, the following time-series
plot displays the infectious and parasitic diseases over an 11-year period from years 2000
to 2010. This graph clearly shows the change in the associated numbers of this category
provided in the data set. Furthermore, detailed mathematical and statistical analysis can be
conducted to find out the details if needed.
4.3 Designing Graphs and Figures 111

Fig. 4.9 Infectious and parasitic disease data from 2000 to 2010

4.3.3 Designing Meaningful Pie Charts

A data set can be represented as a pie chart in a meaningful way if the number of cate-
gories is not too many and the numerical data to be displayed is not overwhelming. A pie
chart is a clear distribution of the data within its own categories in which case the entire
data adds up to 100%. The demonstration of the percentage for the number of categories/
items is ideal for a pie chart’s use.

Example 4.11 Structuring a pie chart for Example 4.1 would be sense making. Figure 4.10
is such a pie chart that displays the outcomes attained from the analysis of the outcomes.
The following are the outcomes attained from the project with their impacts.

• Green effect: 2%.

• Cost: 22%.
• Repair time: 15%.
• Maintenance: 8%.
• Safety: 4%.
• Mean Time Between Failures (MTBF): 49%.
112 4 Measurements for the Defined Problem

Fig. 4.10 The pie chart that represents the observations for Example 4.1 [34]

Another example of a pie chart in Fig. 4.11 that displays the pie chart of production
information with the focus on downtimes, speed, quality, and value-added time.

Manufacturing Data KPIs

Quality
14%
Speed
14%

Value Added Time

57%
Downtime
15%

Fig. 4.11 Manufacturing operations with the breakdown to quality, speed, VA time, and downtime
related issues
4.3 Designing Graphs and Figures 113

4.3.4 Plots by Using Box and Whisker

A Box and Whisker diagram also known as box plot by some of us is useful for data
distribution by considering five levels of the data and where the data falls based on these
five values. This data is categorized into quadrants The associated correspondence is the
following:

• Minimum value.
• Q1: Lower 25% percentile.
• Q2: Lower (or upper) 50% percentile (or median).
• Q3: Lower 75% percentile (or upper 25% percentile.
• Maximum value.

This diagram is particularly useful in not only visualizing the four sections based on the
five values stated above, but also useful for finding out the outliers. The Excel function
Quartile.Inc(array, quart) can be used for designing Box and Whisker diagram while the
data can also be highlighted in Excel with the choice of “Box and Whisker” diagram
choice.

Example 4.12 Table 4.1 contains the Box and Whisker values corresponding to the data
used for Example 4.1.

Table 4.1 Box and Whisker data corresponding to Example 4.1

Box and Whisker data Min Q1 Q2 (Medial Q3 Max
1 Infectious and parasitic 18.53 22.135 22.76 26.655 35.94
diseases
2 Neoplasms 64.52 80.375 107.12 130.115 142.69
3 Endocrine; nutritional: and 46.48 64.49 82.39 105.255 123.83
metabolic diseases and
immunity disorders
4 Mental illness 66.09 72.185 80.61 93.015 112.62
5 Diseases of the nervous 60.14 88.94 106.05 107.96 117.03
system and sense organs
6 Diseases of the circulatory 148.03 161.935 194.57 224.445 265.95
system
7 Diseases of the respiratory 73.44 81.49 88.19 96.665 117.14
system
8 Diseases of the digestive 49.61 69.02 92.83 100.665 111.57
system
114 4 Measurements for the Defined Problem

Fig. 4.12 Box and Whisker diagram for disease categorization given in Example 4.1

Figure 4.12 is the Box and Whisker diagram of Example 4.1 on which the series
correspond to the numbers assigned in Table 4.1
Figure 4.12 contains valuable information that can be further analyzed through graph
reading to extract meaningful outcomes that are not necessarily easy to recognize by just
viewing the plain data. Some of these meaningful outcomes are the following:

• Even the minimum of the circulatory system diseases category is greater than the
maximum of any of the other seven categories.
• Even the maximum of the observed nervous system and sense organ related diseases
is less than the minimum of any other disease categories.
• Other than the diseases corresponding to Series 1 and 6, the other Series have
overlapping amounts of data.
• Excluding the nervous system and sense organ disease data, the minimum occurs for
Series 3 and 8 data.

As mentioned previously, a box plot is helpful to detect outlier data. The outlier data
consists of extremely low or high values that appear as only points, not a part of the
single box plot structure. Such extreme points typically correspond to extreme cases, and
it may be ideal to view the data without them. These data points should be paid special
attention to see if there is any corrective action needed to remove or better them.
4.3 Designing Graphs and Figures 115

4.3.5 Designing Meaningful Pareto Charts

Designing meaningful Pareto Charts can be very helpful in understanding the cumulative
structure of the data with its graphical view.

Example 4.13 Suppose our goal is to design a Pareto Chart for production data attained in
a year with the input representing the length of the rod designed for manufacturing while
the output is the number of items sold in thousands. The line on the graph represents the
cumulative nature of the sale data based on the categories added from left to right. The per-
centage numbers on the right side of the graph Fig. 4.13 are the associated percentage as the
cumulative values increase. The data bins are structured in 10 units ([100,110], [110,120]…)
and the Pareto graph is designed in a way that data is structured from maximum to minimum
output from left to right. The bins on the input axis are not necessarily ordered in a chrono-
logical order. The red line is designed in a way that the output values are added to the line’s
increasing progress from left to right. The right side contains the percentage values for the
output that helps to read the cumulative values on the line. For instance, bin [110,120] has
the maximum value while there are three bins that share the maximum value.

There are several benefits of using a Pareto chart and it cannot be used for any type of
data. It is particularly useful under certain conditions that include but not limited to the
following:

• Displaying accumulating data for a certain KPI.

• Displaying data in categories from maximum to minimum.

Fig. 4.13 Categorized rod length versus number of the corresponding rods sold
116 4 Measurements for the Defined Problem

• Viewing the percentage of the data.

• Existing of frequency for categorized data.

4.4 Calculating Number of Operators Needed for Task

Completions and Takt Time

Takt time (Tt) is used for determining the operational time per unit that it takes to com-
plete a task to be able to deliver to the customer in relation to the quantity or service
demanded by the customer. It allows us to calculate the time it takes to complete one unit
of the customer’s expected item or service. It is particularly useful to measure product or
service pace per unit time. The formula of the takt time includes the following elements:

• Work Hours (tw ). The time it takes for the completion of the work after removing all
the time that was not used for completing the production or service.
• Demand of Customer (D). Product or service demand of the customer.

Hence, Takt time’s formula is

tw
Tt =
D
By using the Takt time, calculation of the needed number of operators (nop ) for
completing the production:
tw
nop =
D

Example 4.14 The head of the nurses in a hospital manages a certain number of nurses in
a hospital, and as an industrial engineer you need to determine the number of nurses needed
to treat a certain number of patients. The hospital management (customer) wants to identify
the number of nurses needed to treat 15,000 patients per year. The cycle time of patient
treatment is determined to be 30 min. The nurses to be hired in this hospital are expected to
have the following:

• Each shift would be 8 h.

• One 10-min break.
• A lunch break of 50 min.

The takt time would allow us to determine the number of hours needed to be able to treat
one patient without the takt time. We need to initially convert the time unit into hours
to be able to have a time unit in hours for common use. The hospital’s expected patient
4.5 Analysis of System Measurements and Gage R&R 117

number is 15000 patients per year, hence, D = 15,000 patients per year. Assuming that
the hospital is operating for 300 days, we have

15000 patients 15000 items

= = 6.25 patients per hour
1 year (300 days) ∗ (8 hours)

Hence, the hospital is expecting a total number of 6.25 patients to be treated per hour.

• Work Hours. The nurse works 8 h per day, has a total of 40 min of break, and 50-min
lunch break. Therefore, the hour amount that the nurse spends at work to treat patients
is

tw = 8 hours − [10 + 50] minutes = 7 hours per nurse

• Period = 0.5 h
• Takt time.
7
Tt = = 1.12
6.25
• Number of operators. The corresponding number of operators needed is the following:

1.12
nop = = 2.24
0.5

This number needs to be rounded up from 2.24 to 3 due to the need of more manpower.
It is essential to note that even though this example is a straightforward calculation, the
operations are more complicated and require a more detailed analysis of the operations.

4.5 Analysis of System Measurements and Gage R&R

System’s measurements require different analysis techniques and Gage Repeatability and
Reproducibility is one of such analysis. This is a good technique for evaluation of a data
set’s repeatability and reproducibility with the associated calculation formulas. By using a
series of collected measurements, Gage R&R helps to identify the data set’s significance in
similarity to ensure that the measurements collected under the same operating conditions
are not varying dramatically and it follows the expectations.
There are certain expectations for the application of Gage R&R. The correctness of
measurements and the associated results is needed to be able to analyze the data correctly.
There are standards used for Gage R&R application and such standards depend on the
industry of application. A set of six sigma’s measurements of systems analysis (MSA)
requires several stages of observations that include the following:
118 4 Measurements for the Defined Problem

Fig. 4.14 MSA elements’ breakdown

1. Bias.
2. Linearity.
3. Repeatability.
4. Reproducibility.
5. Stability.

Figure 4.14 is a summary of the relationship between these above-explained stages.

4.5.1 Structuring Gage Repeatability and Reproducibility Analysis

The scope of the project typically drives the Gage R&R procedure; however, the following
steps can be applied as a step-by-step approach during the data collection for controlling
quality as the accuracy of the data drives the accuracy of the results:

• Depending on the area of application (such as aerospace, health care, expert, customer,
etc.) identify the standard to be used.
• Determine the staff to collect the data and the associated entities to be examined. For
instance, 40 samples of the entities may need to be collected by two or more staff.
• Identify the type of parts to be selected for testing.
4.5 Analysis of System Measurements and Gage R&R 119

• Given the identified standard in the operations, the condition for pass and fail of the
entities should be clarified based on this standard. The collection of the data is expected
to be completed in a sequential manner.
• The inspection strategy (such as visual inspection) needs to be structured based on the
standard.
• Results attained from the outcomes such as good or not good, or the numerical values
need to be recorded in the data collection form.
• The quality inspection phase is done in a sequential manner with the randomization of
the parts to be inspected from the first operator to the second one and so on without
the loss of the original parts’ information attained by the first operator
• The procedure needs to be repeated without referencing to the prior results.
• As a part of the quality inspection accuracy, checking the matches between the
inspections and the expert assessment allows us to identify inspectors’ and expert’s
assessment overlap. The match is confirmed if the assessment result agrees with the
expert assessment. A good measurement system would be expected to be as close as
possible to 100% match.
• Given the CTQ conditions, good quality’s acceptance rate is expected to be determined
to be able to structure meaningful outcomes.
• Deviations from the expected threshold of quality should be determined upon the anal-
ysis of the parts with their respective reasons’ investigation for continuous improve-
ment purposes. In the case when the quality outcomes are below the determined
threshold, the following can be applied:
– Assess and modify the operational approach if there is a need for it.
– A training program needs to be required for the data collectors to complete it to be
able to improve their abilities and uplift their skills.
– Determine the factors that cause the challenges in the overall process and eliminate
these issues through corrective actions (if needed by using the DMAIC steps).
– Apply the procedural steps to recollect the data and reapply the above-mentioned
steps for comparison with the expert assessment.

4.5.2 Meaning of Bias and Determining It

The quality of an item’s production relies on the difference between the measured values
and the associated reference value used for item’s production. The difference arises due to
human-made errors or machine-related issues; therefore, the term “bias” arises from such
a difference as it represents how far the given measurements are away from the expected
standard. The standards or expectations are driven by either customer or industry standards
and the associated measurements can be scaled by an expert. The set of bias values is
determined by calculating the difference between the measured values and the reference
value; hence, the mathematical formula to calculate the bias term is
120 4 Measurements for the Defined Problem

Fig. 4.15 Bias calculation based on the measured and reference values

Bias Value = Measured Value − Reference Value

The bias calculation is the first step before further analysis of the data for waste reduc-
tion. Figure 4.15 is an example of bias calculation assuming that the collected bias data
respects a distribution, and the reference value is the central value of the distribution;
two of the bias values on this figure are shown as examples to demonstrate how such
calculations are accomplished.

4.5.3 Tendency for Linear Behavior

The percentage of variation in comparison to the overall variation is possible to calculate

by using Gage R&R analysis. The linearity displays the consistency of the gage measure-
ments across the reference values. For instance, if there is only one input in a system,
and we sample measurements from the occurrences such as 10 times at different time
periods, these measurements collected from the input can be compared with the standard
or expected value for the input in the system. The same can be repeated for another data
set to identify other variations within the same data set. The standard to be respected can
depend on a variety of reasons that include but not limited to the following:

• Value set by societies for certain applications based on experiences and existing data.
• Standard measurements needed to be applied in operations (e.g., aerospace).
• Customer’s CTQ conditions.
• Self-identified values may vary due to the system and observations.
4.5 Analysis of System Measurements and Gage R&R 121

The standards are typically driven by both the specific industry standards and/or the cus-
tomer expected standards. A standard procedure is expected to be followed for sample
data collection with minimum error during the data collection as well as minimal error on
the bias. The expected value and the standard deviation are the typical values calculated
for such data sets; as covered previously, the size of the data matters as it impacts the
accuracy of the results (Figs. 4.16 and 4.17).
Linearity calculations can be fulfilled by using the following information:

Fig. 4.16 A graph showing systematic measurements where Bias is not constant

Fig. 4.17 A graph with varying spread of the distribution with different mean values
122 4 Measurements for the Defined Problem

• Bias Values. Assuming that the data set is the measured values, bias is the differences
between the reference and measured values as covered previously.
• Linear model fitting to the bias data. Using the bias data, the linear model that fits to
the data; it is assumed here that this linear model has the form y = a0 + a1 x.
• Linear Model’s Slope. The a1 value in the fitting linear model’s formula.
• Linear Model’s y-intercept. The a0 value is determined in the trendline.
• Quantity of reference values (g). The quantity of the reference values used during the
data collection.
• Quantity of observations (m). Quantity of measurements in the data set for each
reference value.
• Taking the square value of bias term and adding them all.
• Bias terms’ total. The total of all bias terms needs to be calculated.
• Multiplication of each reference and bias value with each other and adding them all.
• Average of range values. All range values average.
• The difference between each range and average range value needs to be calculated and
each one of them needs to be squared first and then all these terms need to be summed
up.
• The quantity of counted data values need to be calculated which is so called Degree
of freedom.
• Standard Deviation (s). Deviation from the standard that we covered in Chap. 1 with
the application to this setting as follows:

Sum(Squared biases) − a0 ∗ (Sum(bias terms)) − a1 ∗ sum(bias ∗ ref values)
s=
g∗m−2

• Value of t-statistic.

|a1 | ∗ sumof (range − average range)2
t=
s

An example of these calculations is given in Example 4.15.

4.5.4 Stable Behavior

As the name itself suggests, the stability behavior of the system relies on its stable per-
formance. This can be measured by calculating the deviation from the standard in which
case we consider the standard to be the expected value. In the case when the system starts
producing non-standard points, the deviation from the average will be high indicating that
the out-of-control points started to exist. This is an indication of stability loss. The sys-
tem’s performance data needs to be collected over time and its stability is one of the
4.5 Analysis of System Measurements and Gage R&R 123

Fig. 4.18 Stability behavior of a system based on the observations during two different times

critical measurements and analysis in its success. As this observation is directly related
to the standard deviation, the 6 σ range of the data is directly related to it. Therefore, the
stability and 6 σ analysis used for the system relate to each other as the upper and lower
control limits of 6 σ relate to out-of-control points of the system that indicate unstable
behavior. Figure 4.18 reflects measurement stability with the change in bias over time.

4.5.5 Ability to Repeat Outcomes

A system’s ability to repeat the expected behavior is the ideal condition. As has been
mentioned so far, the ideal condition can be the reference value or the average. Assuming
that the reference value is the expected standard, repeatability, as the name itself suggests,
requires variation measurements from this standard based on the bias calculations. As a
simple example we can consider the use of different tools for producing a certain item
by using different attributes and every time a data point is collected, following the same
production technique by using each tool is expected to result in similar bias results and
repeatable results on their own. Figure 4.19 displays the repeatability of the system in
place with the ability of the system to produce the same output shown with the two-
sided arrow. Given that a data set has a reference value and a distribution, the expected
bias repeatability condition is needed to be as close as possible to the existing bias’s
distribution for Repeatability. This would require almost the same mean and standard
deviation value for different time frames for repeatability to hold.

4.5.6 Ability to Reproduce Results

The differences in observations reflect the ability to reproduce the results. In the case
when similar or the same results cannot be attained, the bias value makes a difference
124 4 Measurements for the Defined Problem

Fig. 4.19 The repeatability of the system in place with the ability of the system to produce the same
output shown with the two-sided arrow

Table 4.2 Patient ID numbers, reference values used, and the 10 observations
Patient ID 1120 1121 1122 1123 1124
Reference value 6 8 10 12 14
Observations 1 6.8 7.2 9.6 11.9 13.8
2 7 7.1 10.1 11.8 14.1
3 5.6 7.8 10.4 12.1 14.2
4 5.8 7.9 10 12 13.6
5 6 7.9 9.9 12.3 13.7
6 5.9 8.3 10.1 11.9 13.6
7 6.7 8.2 10.5 11.8 13.9
8 6.1 8.4 10.4 12.2 14.5
9 6.6 8 10.1 12.5 14.3
10 6.3 7.8 9.5 11.5 14.5

in outcomes. Such differences may relate to the differences in observers or the system’s
inability to produce the same results for similar products. Table 4.2 reflects healthcare data
collected for five different patient reference values collected. The associated 10 different
observations can be 10 different data collected for the same patient that is expected to be
close to the reference value without time dependence. Table 4.3 contains the bias values
produced based in the Table 4.2 data. The associated graph of the bias values and the
linear model fitting to the bias data is displayed in Fig. 4.20; in this figure the slope of
the linear model is very close to zero indicating the horizontal nature of the data.
4.5 Analysis of System Measurements and Gage R&R 125

Table 4.3 Bias data values for Table 4.2 observations

Patient ID 1120 1121 1122 1123 1124
Reference value 6 8 10 12 14
1 0.8 −0.8 −0.4 −0.1 −0.2
2 1 −0.9 0.1 −0.2 0.1
3 −0.4 −0.2 0.4 0.1 0.2
4 −0.2 −0.1 0 0 −0.4
5 0 −0.1 −0.1 0.3 −0.3
6 −0.1 0.3 0.1 −0.1 −0.4
7 0.7 0.2 0.5 −0.2 −0.1
8 0.7 0.4 0.4 0.2 0.5
9 0.6 0 0.1 0.5 0.3
10 0.3 −0.2 −0.5 −0.5 0.5

Fig. 4.20 The linear model fitting to the bias data

Example 4.15 Suppose the following data is collected:

We can easily calculate the following from Table 4.3 and the formulas introduced for
linearity:

• A total of 10 observations exist.

• There are five reference values for the five observations.
126 4 Measurements for the Defined Problem

• The linear model’s equation is y = −0.0229x + 0.37.

• The standard deviation of the bias is 0.398.
• Average range value is 10.
• T-distribution value is 0.2563.
• Primary reference value is 10.

We assume that the probability is not small if the t-distribution’s probability is less than
5%; therefore, the slope must not be zero. In this example, the probability 0.26 is not
small; hence, the slope is zero and bias does not change over the operating range of the
measurement system.

4.5.7 Organizing Measurements into a Matrix Form

Organization of measurements with the customer graded CTQ expectations and the
assigned values for operations help to gather all such information in a measurement matrix
that helps to determine the value proposition of each item considered for the six sigma
project’s completion. A summary of the input/output relationship is reflected operation
by operation for the entire operations to be completed. In each row one of the actions
can be placed for the task to be completed. The categories can be placed in columns just
like it is shown in Table 4.4. The values assigned to these rows and columns depend on
the weights assigned by the customer, the six sigma team, and/or industry standards and
requirements. The following can be used to design the measurement matrix:

Table 4.4 Example of a measurement matrix used for healthcare application

CTQ based weight 10 8 6 4 2 Value Total
Output requirements Quality Waiting Cost Travel Idle total percentage
time time
Needed actions Inputs Values assigned to actions
1 Check-in 1st Form 60 80 90 25 100 3000 42.86%
action filling (60*10) (80*10) (90*10) (25*4) (100*6)
2nd Data 90 60 70 25 100 1960 28%
action collected (90*6) (60*10) (70*6) (25*8) (100*2)
3rd Nurse visit 10 80 90 60 50 2040 29.14%
action (10*4) (80*10) (90*6) (60*6) (50*6)
2 Treatment 1st Nurse
action
2nd Doctor
action
3 Check-out 3rd Check-out
action
7000 100%
4.6 Measuring What is Performed 127

• CTQ Based Weight Assignment. As covered previously, CTQ reflects the customer’s
expectations, and measurement matrix requires such expectation in a measurable way.
A measurement matrix may have a fundamental but important role in the overall
outcomes to be attained from a six sigma project. Examples scales and the weight
assignments depend on the project, customer, and the six sigma team. Any scale can
be used; however, 5-, 10-, and 100-point scales are typically used in applications.
• Requirements for Output. Output requirements need to agree with the scope of the
project, SIPOC, and the overall assessment that is needed for conclusions. For instance,
quality, waiting time, cost, traveling needed, and idle time are such requirements listed
in Table 4.4 for a project and these outputs can be the customer’s choices.
• Needed Actions. The categories of actions can be broken down into sub-actions and
this information helps to see furthermore details. In Table 4.4, the check-in action has
three sub-actions that consist of form filling, data collected from the patient, and the
nurse visit.
• Inputs Needed for Actions. Inputs that are used for completing the needed actions
should be listed.
• Values Assigned to the Actions. The multiplication of the CTQ value and the scaled
value is placed in the table. The CTQ value can be any customer assigned value for
any input depending.
• Value of the Total. The total of the multiplied values in each row is added to calculate
the overall value assigned for the sub-action.
• Total Percentage. This is the percentage value attained by dividing total value of each
sub-action into the total of the total values attained.

Table 4.4 is an example of a measurement matrix:

4.6 Measuring What is Performed

Measuring the performed work or service is a task that may require you to follow a set
of different ways depending on the work completed or service provided. Some of these
techniques will be covered in this section for scaling the performance measurements.
Some of the fundamental methods include the following:

1. Overall Equipment Effectiveness (OEE).

2. Defects Per Million Opportunity (DPMO).
3. Parts Per Million (PPM).
4. Defects Per Unit (DPU).
128 4 Measurements for the Defined Problem

4.6.1 Measuring Effectiveness of Equipment

Measuring the effectiveness of the equipment is the overall equipment effectiveness (OEE)
process that helps to numerically determine how efficiently the equipment is used in
applications. Figure 4.21 is an example of how Overall Equipment Effectiveness can be
calculated in a simplistic way that displays the formula that can be used and an example
to show how it can be used.
OEE can be applied in a variety of industries that include but are not limited to produc-
tion and service. The formula of OEE is used for calculating the number of good items or
services into the total number of items or services. This ratio is useful for determining the
effectiveness of the used methodology as it requires determining the percentage of good
production. There are a variety of scenarios that this formula can be used that include but
not limited to the following:

• Breakdown. Machine breakdown calculations are one of the areas to determine the
number of good and bad reasons for machine breakdowns and the effectiveness of the
equipment.
• Setup. This is the time taken for setting up a machine (or a work in service) for usage.
OEE can help to determine the good percentage usage of the machine (or service).

Fig. 4.21 An example of Overall Equipment Effectiveness (OEE or OME) and the general formula
to calculate it
4.6 Measuring What is Performed 129

• Operational Break. The time that a machine or service may not be operated is the
time that the work is not completed for production or service. As it may be affecting
the pace of the work completed, OEE can help to determine how much it is affecting
the good number of items completion.
• Defects. Defect rate can be identified in relation to the good production for different
types of defects or wastes.

4.6.2 Calculation of Defect Rate

Defect rate formula is a basic formula that helps to calculate the rate of defect based on
the total number of items produced by using the formula

# of defects
DPU =
total number of units

Example 4.16 A construction worker determined 60 lumber pieces that cannot be used in
construction among the 200 items shipped by the supplier. After reworking these defects,
18 of them are corrected and are possible to use in construction. Calculate the DPU values
prior to the corrective actions and after reworking the defected units.
Before corrections, the defect rate is
60
DPU 1 = = 30%
200
Hence, 30% of the items were wasted. After finding out that 18 of the items can be
reworked, 18 units out of 60 units are corrected; hence, there are 42 defected units within
200 units. Hence, the new defect rate after corrective actions is
42
DPU 2 = = 21%
200

4.6.3 Successful Part Production Rate

Just like DPU, Parts Per Million (ppm) is calculated for determining the average number
of items in a million items. PPM can be applied in many different settings and can be
calculated by using the following formula:
Customer approved items 6
PPM = 10
Number of items produced
130 4 Measurements for the Defined Problem

Example 4.17 A medical part manufacturer determines 36 defects during the production
of 120 medical tools. 12 of these defects are possible to correct after rework; however, the
customer’s CTQ conditions for this tool allow only 10 of these to be corrected.
Out of 36 defects, only 14 can be corrected based on the CTQ conditions; therefore,
84 6
PPM = 10 = 700000
120
good tools are produced in one million tools. Out of 36 defective items, 12 tools can be
corrected, therefore,
96
DPU 1 = = 0.8 tools per unit
120

4.6.4 Calculation of Defects in a Million

The number rate of defective items in a million items produced based on the observed
opportunities can be calculated by using the Defects Per Million Opportunity (DPMO)
formula. Let.

• D: Number of defects in the production.

• N: Number of units produced.
• ED : Expected defect amount in the production.

The formula of DPMO is

D
DPMO = 106
nED

Example 4.18 Suppose a nurse fulfilled the following:

• Number of patients inspected = 2000.

• Total number of patients that needed to be retreated = 200.
• Number of opportunities per patient = 10.

Therefore,
200
DPMO = 106 = 10000
(2000)(10)
retreatments are expected when million patients are treated.
4.7 Questions 131

4.7 Questions

Question 4.1 Using the following data, answer the questions below.

Patient ID 1 2 3 4 5
Reference value 10 12 14 16 18
1 10.8 11.2 13.6 15.9 17.8
2 11 11.1 14.1 15.8 18.1
3 9.6 11.8 14.4 16.1 18.2
4 5.8 11.9 14 16 17.6
5 10 11.9 13.5 16.3 17.7
6 9.9 12.3 14.1 15.9 17.6
7 10.7 12.2 14.5 15.8 17.9
8 10.7 12.4 14.4 16.2 18.5
9 10.6 12 14.1 16.5 18.3
10 10.3 11.8 13.5 15.5 18.5

a. Apply the Gage R&R on the data and design a linear model for the bias data just like the
one designed in Fig. 4.20. What can we conclude from this data by undertaking such an
application?
b. What other techniques covered in this chapter can be applied to this data set? Choose
one and apply it on this data.
c. Apply linear regression on the data set? What can be concluded?
d. Calculate the mean and standard deviation values by using the data. What can be
concluded by using such information?
e. Box and Whisker diagram by using the information? Extract five meaningful outcomes
from the designed graph.
f. Apply Pareto Chart on the data. Extract five meaningful outcomes by using the chart.

Question 4.2 A nurse attained the following through observations:

• The number of patients observed is 2500.

• The total number of patients returned to the hospital due to mistreatment is 73.
• The number of opportunities per patient is 4 as the treatment.

Calculate the DPMO. Interpret the number you attained in the context of the problem.
132 4 Measurements for the Defined Problem

Question 4.3 A nurse manager identified 105 possible medical tools that may not be usable
during the inventory count of 425 items. After going through these tools in detail, 63 of them
are determined to be usable. After going over the tools one more time, 35 more tools are
corrected. Determine the average number of unusable tools after rework for both the first
and second rounds of corrections.
Analysis of the Measurements and Project
5

The phase following measure is the analyze phase in which the collected information
for the defined problem would be analyzed both qualitatively and quantitatively. The
quantitative results typically follow from statistical results with meaningful figures and
explanations. There are many more advanced analysis techniques that include artificial
intelligence and advanced data analytics concepts that cannot be covered here due to the
level of the textbook; however, such techniques can be also used in the analyze section
depending on the capabilities of the designed six sigma team and the scope of the project
along with the interest of the customer. In summary, the following have been fulfilled in
the D and M phases:

• Define:
– Provide project background information.
– Identify and state the problem.
– Specify the project scope and planning.
• Measure
– KPI driven data collection.
– IPO observations.
– Specification of collected data infrastructure.
– Designed charts and figures representing the measure data.

There are several techniques in analyze that may appear to be simple; however, they
require a sophisticated approach to make the most out of the technique and the next
section contains one of them. “5W and How” questioning to be covered in the next
section may look like a simple technique that requires asking six questions; however,

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 133
E. Tokgoz, Six Sigma and Quality Concepts for Industrial Engineers, Synthesis Lectures
on Engineering, Science, and Technology, https://doi.org/10.1007/978-3-031-55740-8_5
134 5 Analysis of the Measurements and Project

these questions need to be driven by the measurements’ outcomes that relate to the scope
of the project.

5.1 Structuring “W” Questions and How

“5W”s and “How” are the questions that are outlined in Fig. 5.1 [25]. The ideal application
of this questioning is to design a set of smart questions that need to be answered and these
questions should also be answered as a part of the project’s completion. It should be noted
that the 5W + How questioning should not necessarily be a one-time approach; therefore,
it can be applied multiple times.
One way to apply 5W and How questioning is as follows:

1. Determine the first phase where 5W can be applied in a meaningful way.

2. Design a set of sophisticated 5W + How questions that would be feasible to answer.
3. Answer all questions in alignment with the scope of the project.
4. Keep in mind that these questions may determine answers to parts of some other
sections in the DMAIC approach.

Fig. 5.1 Formulation of 5W and How questioning that also requires answering the formulated
questions
5.1 Structuring “W” Questions and How 135

Let’s consider the shopping example with the objective of Grocery shopping list prepa-
ration among the other objectives listed in Sect. 4.2.3. This subprocess deserves asking a
set of 5W and How questions by itself.

• What is the list of items to be purchased from the shopping center?

• Where in the supermarket would you need to purchase the items based on the generated
list? Do you have the right order to visit in the grocery store?
• Who are the people covered in the list for the grocery shopping to be done? Is everyone
covered?
• Why do we need such a list?
• When is the right time to go shopping during the week based on the generated list?
• How do you structure your grocery shopping path within the store?

Keeping in mind the grocery store’s isle layout and the list determined, you can optimize
the path you are traveling within the grocery store if you don’t like spending too much
time in it (like myself!) This would reduce the time wasted. This can be considered as a
strategic and non-random preparation of the list and the following steps to be taken. Given
that the list preparation is a subprocess of the general grocery store shopping, the other
subprocesses of overall grocery shopping can have additional 5W and How questions
designed separately.
Given that industrial engineering is a very broad, we can cover the lumber wasted in
the construction industry. A set of 5W and How questions can be designed for further
understanding of the lumber wasted:

• What type of lumber waste occurs in the order or greatest to least occurrences?
• Where does this lumber waste occur in the order of greatest to least?
• Who (or what employee position) causes the wastes in the order of greatest to lowest
waste occurrences?
• When does the waste occur in the order of greatest to lowest?
• How can such a waste be controlled for declining its occurrence?

The consideration of the data’s continuity nature is essential in the application. The typical
data collected can be discreet with even the time itself. The project team needs to decide
on the applicability of the analysis approach depending on the data collected even if the
variable has a continuous nature that can be used in a discrete manner. Such data needs
to be handled with caution noting that the data analysis used needs to reflect realistic
outcomes not for the data itself only but for the overall operations in general.
136 5 Analysis of the Measurements and Project

5.2 Structuring Fish-Bone Diagram with Causes and Effects

A structured way to design all causes and sub-causes is by using a cause-and-effect dia-
gram which is also called a fish-bone diagram. A step-by-step approach can be designed
as follows.

• First step: Identify the cause categories of the defined problem in relation to the data
collected.
• Second step: Identify sub-causes and their effects under the main categories.

Example 5.1 The following example given in Fig. 5.2 of the fish-bone diagram that displays
the effects of not reporting wrongdoing [28]. The consequences determined in this figure
are typically the outcomes of collected data and its analysis from the associated groups.

The following is a set of causes used in real-life applications:

• Environment.
• Manpower or Labor.
• Machine.
• Material.
• Measurement.
• Method.

Fig. 5.2 The causes listed for not reporting wrongdoing in the workplace [28]
5.3 Identification of Data and KPI Correlation 137

Fig. 5.3 An example of a fish-bone diagram designed for inefficient production [25]

Example 5.2 An example of a fish-bone diagram is inefficient production with its associated
assignable causes and effects listed in the graph below. It is the measurements and its analysis
that yield to the assigned causes and their effects on the graph. It is essential to note here that
this diagram can change dramatically for operations within the same facility for different
locations. Another important aspect to be mentioned here is the necessity of explanations
of all visuals displayed throughout the DMAIC approach by the team. The information
displayed on the graph is just a representation of the idea developed and such an idea needs
further explanations. As an example, “inefficient pull system” in the description is unclear
in the way it is outlined on the graph that needs furthermore explanations [25] (Fig. 5.3).

5.3 Identification of Data and KPI Correlation

The slope of the tangent line taught in calculus is essentially the one that determines the
slope (m) and the output-intercept (b) of a given data. The least squares method (LSM)
is the typical concept covered in operations research that can be used for determining
linear regression models that may have greater than 3 dimensions. Software packages
such as Minitab, SPSS, Excel, MATLAB, and R can be used for regression coefficients.
In a two-dimensional space, the model that fits to the given data set is a plane while a 3
or more-dimensional fitting is called hyperplane fitting. There are also programs written
for Texas Instrument (i.e., T.I.) calculators such as TI 83, 84, 89, and others that can
also be used for linear model fitting for low dimensions. In this chapter we will cover
138 5 Analysis of the Measurements and Project

different correlation types, LSM for two- and three-dimensional space data, and some of
the statistical formulas that are typically used for data correlation.

5.3.1 Correlation Types

Correlation is categorized into three types that consist of positive, neutral, and nega-
tive slopes that are solely depending on the mathematical nature of the fitting regression
model. An example to this correlation classification is shown in Fig. 5.4 where the rela-
tionship between the critical to quality and machine production is analyzed. The negative
correlation is an indicator of a decay in the critical-to-quality conditions as the machine
production quantities increase. The graph in the middle for positive correlation is an indi-
cator of the increase in the critical-to-quality conditions as the number of production
numbers increases. The neutral correlation indicates a no change in the critical-to-quality
conditions as machine production increases. There can be many different assignable
causes for these different scenarios. An example of such a CTQ measurement in health
care can be the number of patients that had to return to a hospital due to mistreatment.
The machine production can be replaced by the number of patients seen in the hospital by
a certain doctor. In this case the quality of the treatment by the doctor can be easily iden-
tified based on the daily (or shift driven) number of patient data collected. This approach
would result in data points that would specify the quality of treatment on the output axis
(that can be in numbers or percentages.)
Examples of critical to quality include but not limited to the following:

• Customer satisfaction based on deliveries.

• On-time deliveries of a transportation company.

Fig. 5.4 Examples of positive, negative, and neutral types of linear regression models fitting to the
given three different data sets
5.3 Identification of Data and KPI Correlation 139

• A nurse’s ability to successfully treat patients.

• Occurrences of production defects.
• Quality related issues’ occurrences.
• Item expiration in inventory.
• Customers waiting on the line for call services.

Correlation analysis is a critical element of a project to determine the outcomes success-

fully; such a data set needs to be analyzed carefully. In addition, the above-given examples
are driven by CTQ conditions; however, correlation analysis cannot be limited to CTQ
only. It can be observed for many different varying conditions and variables.
Fitting a linear model in Excel for two-dimensional data is rather simplistic given that
Excel has a feature called “trendline.” The following are the steps of trendline fitting in
Excel for Windows operating system users:

• Step 1. Given input data X and output data Y, enter these data sets in the column
format: Let X be the left most column and Y be the right most column.
• Step 2. Highlight the entered data that is desired to be graphed for fitting linear
regression.
• Step 3. Within the “Insert” Toolbar, choose “Scatter” plot in the “Charts” section to
graph the scatter plot of the data.
• Step 4. Choose the “Add chart element” to view the “Trendline” option on top left
corner in the Toolbar area. You can also access this feature through the use of the
“ + ” that will appear on the top left corner of the scatter plot if you click on the
graph. The “trendline” option would be available for you to choose from the shown
list. There would be an arrow to click on the left of the trendline to further choose
options. Figure 5.5 illustrates this application with the highlighted data set.
• Step 5. Once the trendline option is selected, it is important to choose “Linear” listed
under varying options to be able to view the plot shown in the Fig. 5.6. You would
need to also click on the option “Display Equation on Chart” as well as the R2 value
to be displayed.

The following are the observations based on the attained graph:

• The line has a positive slope; hence, the correlation between the two variables is posi-
tive noting that the graph has an increasing trend from left to right. It is important to
note that the meaning of positiveness here is an indicator of “one increases as the other
one increases.” The outcome of a positive correlation can be negative. For example,
the number of defects (Y variable) may be increasing as the number of items produced
is increasing (X variable).
• The slope of the line is an indicator of the strength of the correlation between X and
Y. When the slope is positive, the greater the number is, the stronger the correlation
140 5 Analysis of the Measurements and Project

Fig. 5.5 Designing a linear

regression by using the chart
element option in Excel

would be. The value that is approximately 0.045 shown in the figure above indicates a
certain level of strength to be covered later in this chapter.

Production can start with the items pulled from the Raw Material Inventory and then go
through five stages of the production line sequentially as a serial production line. In the
Fig 5.7, assuming → represents the production flow direction, there are five machines
processing the goods before the final product is taken to the warehouse.
If we consider the healthcare version of the above-given example, the patient treatment
is initiated in a hospital with the patient check in and the patient goes through sev-
eral healthcare treatment processing steps afterward before checking out with the forms
filled out. This healthcare example would be the driver of positive, negative, and neutral
correlations to be explained in the next subsections.

5.3.1.1 Positive Slope and Correlation

Negative slope of a linear regression (i.e., trendline for 2D data in Excel) is an indicator
of the negative correlation between the given two variables which is an indicator of a
5.3 Identification of Data and KPI Correlation 141

Fig. 5.6 Trendline option for linear regression design in Excel including the equation of the linear
regression

Fig. 5.7 Production that initiated with raw materials taken from the inventory, processed by five
machines M1 through M5, and finalized with the finished product taken to the Warehouse

negative relationship between them. The example we will cover in this section in health
care would have a negative correlation while the conceptual meaning would be a positive
outcome.

Example 5.3 Table 5.1 contains the failed measurement attempts of Nurse 1. This data is
collected to identify the nurse’s effectiveness in patient medical measurements’ collection.
In this scenario, the failure means the nurse’s attempt to collect the same data from the same
patient more than once due to failure of the initial attempt. Each number listed under “Nurse
142 5 Analysis of the Measurements and Project

Table 5.1 Nurse 1 collected

Nurse 1 seen patients Failed measurements (%)
data during seven shifts
120 6.7
140 4
135 4.5
128 4.6
136 3.8
122 6
137 3

1 Seen Patients” column is the number of patients seen by Nurse 1 during a 12-h shift. The
column “Failed Measurements” indicates the percentage of failed measurements by Nurse 1.
For instance, the first data point has 120 patients with 6.7% failure rate indicating that about
8 patients experienced more than one-time collected measurements by Nurse 1. In this case
we assume that the patient must revisit the hospital; hence, the second column also indicates
the number of patients who had to return to the hospital due to failed measurements.

Fundamental considerations in data analysis include the number of data points used,
how much the data reflects reality, and whether there are mistakes (particularly big mis-
takes) in the data set. The data set given in Table 5.1 has a very small sample size;
therefore, the outcomes of the analysis may not be reliable. The meaningful analysis out-
comes and the associated improvement ideas will be driven by the collected data and its
analysis; therefore, this section is very important to pay attention for.
The design of charts, figures, and graphs that have an important role during the pre-
sentation of the six sigma project to the customer; therefore, designing high quality and
meaningful figures would be important. The following is an example of such a figure that
is designed for displaying Nurse 1’s measurement failure rate:

• Titles. Axis titles need to be placed to show what they represent.

• Headline. The headline needs to capture the essence of the project in a nutshell.
• Legend. The data displayed on the graph has a certain color coding and structure and
a legend is designed to state the nature of this representation.
• Line Markers. In the following image the vertical and horizontal marking lines are
removed; however, they can be added by adjusting the settings in Excel.

Figure 5.8 can be interpreted as follows:

• The slope is a negative number (−0.1495); therefore, the two variables are negatively
correlated.
5.3 Identification of Data and KPI Correlation 143

Nurse 1 Measurements - Failure Rate

7

6.5
Percentage of Measurement Failure

6
y = -0.1495x + 24.268
5.5
R² = 0.8317
5

4.5

3.5

2.5

2
115 120 125 130 135 140 145
Number of Patients Seen by Nurse 1 per Shift
Series1 Linear (Series1)

Fig. 5.8 Failed measurement rate of Nurse 1 during the seven observed shifts

• As the number of patients treated by the nurse increases, the percentage of measure-
ment failures decreases indicating that the nurse is more successful as the number of
patients that the nurse sees increases. Furthermore, data collection and analysis appear
to be needed to determine the consistency of this outcome. In addition, a detailed
fish-bone diagram is likely to yield a better analysis.
• External factors should also be determined to see if there is any other factor impacting
the nurse’s success.
• Even though the correlation is negative, the outcome is positive as the failure rate is
considered.

5.3.1.2 Effect of Positive Correlation

A positive correlation between inputs X and Y in an indicator of positive relationship
(correlation) between the input and output values. Just like it was the case for the negative
correlation, the numbers may have positive correlation; however, the conceptual meaning
of the outcome can be negative.

Example 5.4 The following data is collected for doctor 1’s treatment success in which case
the doctor works with Nurse 1 and this doctor only treats the patients whose measurements
are collected by Nurse 1. The column “Doctor 1 Seen Patients” is the number of patients seen
during each shift of the doctor while the treatment success rate is the successful treatment
of the patients. The unsuccessful treatments of the doctor resulted in patients’ revisit to the
hospital (Table 5.2).
144 5 Analysis of the Measurements and Project

Table 5.2 The data collected

Doctor 1 seen patients Treatment success (%)
for doctor 1’s treatment success
110 91.5
130 96.3
120 92.8
112 91.8
121 92.9
106 84
129 96

The following is the scatter plot with the linear model fitting to the data.
The positive slope of the data and other information attained in Fig. 5.9 indicate the
following outcomes:

• The slope m = 0.385 > 0 indicates the positive correlation between the number of
patients seen by doctor #1 per shift and the associated percentage of patients’ treatment
success.
• The increase in the number of number of patients seen by doctor # 1 appears to have
a positive impact on the patient treatment success rate. Based on this result, Doctor #
1 is more productive when he/she sees more patients per shift. Even though the data

Doctor 1 Seen Patients - Success Rate

98

96
Percentage of Treatment Success

94 y = 0.385x + 46.651
R² = 0.7728
92

90

88

86

84

82 Series1 Linear (Series1)

80
105
Number of Patients Seen by Doctor 1 per Shift

Fig. 5.9 Treatment success display of doctor 1 for the treated patients right after Nurse 1
5.3 Identification of Data and KPI Correlation 145

indicates this result, furthermore analysis by collecting additional data is needed to

be able to structure a cause-effect diagram that reflects the reality. Additional basic
statistical analysis calculations such as mean and standard deviation can be helpful for
identifying the details.
• Doctor should be consulted on this outcome.

5.3.1.3 No Correlation for Neutrally Correlated Data

A neutral correlation of a data set is an indicator of the steady behavior of one of the
variables while the other one varies.

Example 5.5 Continuing to build on the previous example, the left most column on the
following data set is the number of patients that the doctor required for the patients to
undertake Test 1 with the other column representing the accuracy results attained for the
associated number of patients in percentage (Table 5.3).

The plot and the associated trendline are shown in the Fig. 5.10. We can conclude the
following from the displayed data and the fitting linear regression:

• The slope of the linear model is m = 0.0007 ≃ 0 (i.e., almost zero slope) ⇒ The
percentage of success does not change dramatically with the change in the number
of patients seen. This outcome can be easily supported by the standard deviation
calculation. The numbers displayed for “Accuracy” are a clear indication of it.
• The accuracy level of Test 1 does not change dramatically as the number of patients
is tested by using Test 1; however, there are still possible improvement opportunities
for making Test 1 better.
• It is typically ideal to reduce the variation of the operations in addition to increasing the
average accuracy of the outcome. In such a case, the best case would be for the quality
of the test accuracy to be almost 100% if 100% is impossible where the collected data
centers around the trendline.

Table 5.3 Machine # 5 data in

Test 1 Accuracy
a sequential production line of
5 machines 95 94.5
90 95
92 94.2
94 95
96 94.7
98 94.75
95 95
146 5 Analysis of the Measurements and Project

Accuracy of Test 1 Results

95.1
95
Percentage of Success

94.9
94.8
94.7
94.6
y = 0.0007x + 94.671
94.5
94.4
94.3
94.2
94.1
89 90 91 92 93 94 95 96 97 98 99
Number of Patients Tested by Using Test 1

Fig. 5.10 Accuracy results attained for Test 1 from the patients of doctor 1 and the fitting trendline
with its equation

• A cause-effect diagram may be essential to identify the real causes and effects of the
levels of accuracies.

The “no correlation” statement here does not mean that the two data sets are not impacting
each other, it means that the percentage of success has neither increasing nor increasing
trend.

5.3.1.4 Correlation Formula and Strength

The correlation level or strength between two random variables X and Y can be identified
by categorizing the correlation values to be explained below. The following are the values
to be used in the formulas below:

• The covariance between X and Y is represented by cov(X,Y).

• The correlation between X and Y are represented by corr(X,Y).

The correlation between X = {X i } and Y = {Yi } data sets with n points is shown with
the CORR(X,Y) function below (with its correspondence CORREL function in Excel)
and this is a well-known statistical function.
∑n
i=1 (X i − X )(Yi − Y )
C O R R(X , Y ) = /[ ][ ]
∑n 2 ∑n 2
i=1 (X i − X ) i=1 (Yi − Y )
5.3 Identification of Data and KPI Correlation 147

with the notation used X and Y as the average values of the data sets X and Y ,
respectively.
Covariance is another statistical method for measuring the relationship between two
data sets X = {X i } and Y = {Yi } by using the mathematical formula
( n )
1 ∑
C O V (X , Y ) = (X i − X )(Yi − Y )
n−1
i=1

There are degrees of correlation that indicate the strength of the relationship between
the data set points: Such a correlation assignment varies depending on the conditions of
application. The following is an example such correlation assignment:

• Almost Perfect Correlation. If the correlation value between two random variables
X and Y is varying between the values 0.999 and 1 then X and Y can be said to be
almost perfectly correlated.
• Strong Correlation. If the correlation value between two random variables X and Y
is varying between the values 0.75 and 0.999 then X and Y can be said to be strongly
correlated.
• Moderate Correlation. Two random variables X and Y can be considered as moder-
ately correlated if the correlation between the two variables is ranging between 0.5 and
0.75.
• Weak Correlation. Two random variables X and Y can be considered as weakly
correlated if the correlation between the two variables is ranging between 0.001 and
0.5.
• Almost No correlation. If the correlation between X and Y is either almost zero
or exactly zero, then X and Y have almost no correlation. Just like all the other
correlations explained above, this is subject to the conditions of correlation.

The Fig. 5.11 outlines examples of negative coefficient strengths and the mathematical
formula that outlines the relationship between correlation and covariance.

5.3.2 Correlation Calculations by Using the LSM (Least Squares

Method)

In this section we will cover least squares and slope-intercept methods. The least squares
method is useful for finding the best fitting model to the given data whereas slope-intercept
method only helps to determine a linear model that may not be necessarily the best fit
to the given data. Technology can be used for complicated data sets based on personal
preferences.
148 5 Analysis of the Measurements and Project

Fig. 5.11 Correlation and covariance correspondence formula and the correlation levels [25]

The least square method requires to follow the steps outlined below to be able to
determine the constant values m and b for the best fitting linear model y = m*x + b to
a data set given in the following table.

x (input) x1 x2 … xn
y (output) y1 y2 … yn

Step 1. Define the error term for each data point.

edata = m ∗ xdata + b − ydata for all data = 1, 2 . . . n

Step 2. Define the error function E(m, b) for the entire data set.

E(m, b) = e12 + e22 + e32

= (m + b − 55.3)2 + (2m + b − 57.4)2 + (3m + b − 58.2)2

Step 3. Calculate and solve the following differentials for the error function:
dE
=0
dm
5.3 Identification of Data and KPI Correlation 149

dE
=0
db
This would provide two equations with the two unknowns m and b values. Solving
this set of equations would result in determining the equation y = m ∗ x + b.

Example 5.6 Suppose you collected the following weight of a product for the first three
working days.

Days 1 2 3
Weight in kg 55.3 57.4 58.2

Determine a line that fits to the following data by using the least squares method.
Step 1. We use the error term for each data point.

e1 = m + b −55.3

e2 = 2m + b −57.4

e3 = 3m + b −58.2

Step 2. Define the error function E(m, b) for the entire data set.

E(m, b) = e12 + e22 + e32 = (m + b − 55.3)2 + (2m + b − 57.4)2 + (3m + b − 58.2)2

Step 3. Calculate and solve the following differentials for the error function:
dE
= 2(m + b − 55.3) + 4(2m + b − 57.4) + 6(3m + b − 58.2) = 0
dm

28m + 12b − 689.4 = 0

and
dE
= 2(m + b − 55.3) + 2(2m + b − 57.4) + 2(3m + b − 58.2) = 0
db

12m + 6b − 341.8 = 0

Hence, we need to determine m and b by solving the following two equations:

150 5 Analysis of the Measurements and Project

Fig. 5.12 The two-dimensional optimized linear regression model fitting to the given data set

28m + 12b = 689.4

12m + 6b = 341.8

An easy way to solve this set of equations is by variable elimination method. Solving
this set of equations would result in determining m = 1.45 and b = 54.067 and hence,
we find the equation y = 1.45*x + 54.067. The following graph sketched by using Excel
is a way to demonstrate such function. Figure 5.12 shows the da.

5.3.3 Multivariate LSM

The least squares method can be particularly useful for identifying multiple variables’
correlation simultaneously as it is typically not easy to determine it by using technology.
Given that we covered this method for one input and one output in the previous section,
we will generalize this case to k variables as follows:
Goal. Determine the constant values m1 , m2 …mk and b for the best fitting linear model.
y = m1 ∗ x1 + m2 ∗ x2 + . . . + mk ∗ xk + b to a data set given in the following table.

x1 (input 1) x11 x12 … x1n

x2 (input 2) x21 x22 … x2n
…
xk (input k) xk1 xk2 … xkn
(continued)
5.3 Identification of Data and KPI Correlation 151

(continued)
x1 (input 1) x11 x12 … x1n
y (output) y1 y2 … yn

Step 1. Define the error term for each data point.

edata = m1 ∗ x1,data + m2 ∗ x2,data + . . . + mk ∗ xk,data + b − ydata for all data

= 1, 2 . . . n

Step 2. Define the error function E(m, b) for the entire data set.

E(m, b) = e12 + e22 + · · · + en2

Step 3. Calculate and solve the following differentials for the error function:
dE dE dE dE
= 0, = 0, . . . , = 0 and =0
dm 1 dm 2 dm k db
This would provide m + 1 equations with the unknown values of mi and b. Assuming
that there are enough number of equations to attain a unique solution, solving this set of
equations would result in determining the equation y = m1 ∗ x1 + m2 ∗ x2 + . . . + mk ∗
xk + b.

Example 5.7 Suppose you collected the following defect data from two machines M1 and
M2 for three working days.

Observations M1 M2 Corrected items

Day 1 2 2 1
Day 2 1 3 2
Day 3 1 4 3

Determine a hyperplane that fits to the following data by using the least squares
method.
Step 1. We use the error term for each data point.

e1 = 2m1 + 2m2 + b −1

e2 = m1 + 3m2 + b −2
152 5 Analysis of the Measurements and Project

e3 = m1 + 4m2 + b −3

Step 2. Define the error function E(m, b) for the entire data set.

E(m, b) = e12 + e22 + e32

= (2m1 + 2m2 + b − 1)2 + (m1 + 3m2 + b − 2)2 + (m1 + 4m2 + b − 3)2

Step 3. Calculate and solve the following differentials for the error function:
dE
= 2(2m1 + 2m2 + b − 1) + (m1 + 3m2 + b − 2) + (m1 + 4m2 + b − 3) = 0
dm1

6m1 + 11m2 + 4b − 7 = 0

dE
= 2(2m1 + 2m2 + b − 1) + 3(m1 + 3m2 + b − 2) + 4(m1 + 4m2 + b − 3) = 0
dm2

11m1 + 29m2 + 9b − 20 = 0

and
dE
= (2m1 + 2m2 + b − 1) + (m1 + 3m2 + b − 2) + (m1 + 4m2 + b − 3) = 0
db

4m1 + 9m2 + 3b − 6 = 0

Hence, we need to determine m1 , m2 , and b by solving the following two equations:

6m1 + 11m2 + 4b = 7

11m1 + 29m2 + 9b = 20

4m1 + 9m2 + 3b = 6

Typically, in addition to linear algebra techniques such as Gauss-Jordan method of row

reduction, technology such as Excel and MATLAB can also be used for identifying the
parameters of the linear model. Solving this set of equations would result in determining
m1 = 0, m2 = 1, and b = − 1 and hence, we find the equation y = x2 − 1.
5.4 Designing Tables for Interaction of Variables 153

5.4 Designing Tables for Interaction of Variables

The association of two or more interactions between process or service components can
be outlined by using an interaction table. This table helps with understanding process and
service components’ impact on each other because of interaction. The designed system
with five machines shown in Fig. 5.7 can be displayed in an interaction table to demon-
strate the impact of machines on each other. Recall that the material flow in the process
is known to have the following trend:

I nventor y → M1 → M2 → M3 → M4 → M5 → Customer

Table 5.4 represents the number of wasted products determined during the three con-
secutive shifts. The flow of items follows the structure outlined in Fig. 5.7. Each column
number displayed for M1-M5 per observation is the number of scraps produced. Each
observation totals to 10,000 items’ production. The “Total Waste #” column displays the
total number of wastes collected from the five machines and the last column displays
“Waste Percentage” displays the corresponding percentage of wastes based on the total
production which is 10,000 items. For instance, the first observation has 10,000 items
with a total of 495 wasted items observed. There is quality control in some places and
there is none in other places. The scraps are assumed to be carried over from one machine
to another, in which case the reduced waste from one machine to another is due to the
quality control.
The interaction table helps to determine the interaction between table values for
operations by using the following signs:

↘ Sign representing the decrease in the number from one cell to another.
↗ Sign representing the increase in the number from one cell to another.
— Sign representing no change from one cell to another.

Table 5.5 is the interaction table corresponding to the data displayed in Table 5.4.
Table 5.5 is the interaction table representing the interactions between machines with
the associated changes in the scrap numbers reflecting Table 5.4 information. For example,
there is a decay in the number of wastes during the first observation from M1 to M2 as
120 is reduced to 10 which is represented by a decay sign.

Table 5.4 The waste quantities identified for five consecutive machines during three observations
Ml M2 M3 M4 M5
1st observation 120 10 125 120 120
2nd observation 110 115 120 112 112
3rd observation 140 148 160 155 155
154 5 Analysis of the Measurements and Project

Table 5.5 Interaction table reflecting Table 5.4 machine wastes

Ml & M2 M2 & M3 M3 & M4 M4 & M5
1st observation ↘ ↗ ↘ —
2nd observation ↗ ↗ ↘ —
3rd observation ↗ ↗ ↘ —

It may be ideal to apply 5W and How in this section. There are many more observations
and questions that can be asked here but the following is some of these questions and the
associated observations from the table:

• What type of quality inspection application is implemented from M1 to M2 during

observation 1?
– The answer to this question would provide an insight about the proper quality
application which would open a door for furthermore questioning:
Who was the quality controller?
Which strategy did the person use?
Why did this quality inspection make a difference compared to the other quality
inspections?
Where can this quality inspection be adopted for further waste reduction?
How can this quality inspection strategy be applied in different locations?
• The trend of increase from “M2 and M3” to “M3 and M4” is a decrease for all
three observations. What caused this decay? This also opens a door for 5W and How
questioning:
Who was the person that made a difference in all these three observations? Was it
only one person or was it more than one person?
Where were the quality inspections effective?
Which strategy was followed? Was it only one strategy or more than one?
How did the quality inspection change?

A smart approach in six sigma applications is to calculate the percentage of the changes on
the interaction table. The first observation in Table 5.6 is an example of such calculations.
The number of waste production changes from 120 on M1 to 10 on M2 which is a 1.1%
reduction as there is a total of 10,000 items produced.

Table 5.6 Interaction table with the percentage of defect change from one machine to another
Ml & M2 M2 & M3 M3 & M4 M4 & M5
1st observation ↘ (−1.1%) ↗ (+1.15%) ↘ (−0.05%) —
2nd observation ↗ ↗ ↘ —
3rd observation ↗ ↗ ↘ —
5.5 KPI Analysis Driven by Time-Dependent Variables 155

5.5 KPI Analysis Driven by Time-Dependent Variables

Analysis of actions requires specific focuses on certain aspects of time-dependent actions.

Time related changes need to be recorded with the associated outcomes. There can be
many different time related KPIs that can be analyzed; the following are some of these
that can be completed during the completion of a project:

• Customer Driven Dependencies. The customer should be communicated time-to-time

to be able to update the progress project. Examples of such customer dependencies
include general project updates, questions related to the project, collaboration with the
customer’s staff, etc.
• Time-driven Operations. Operations require time tracking in which case the collected
data can be analyzed to be able to attain time-driven outcomes. This analysis would be
required for both pre- and post-improvement analysis. Examples of such time-driven
data include time studies, time-driven production rate, etc. This analysis results in the
measured KPI outcomes. One important time consideration is the Takt time; statistical
distributions and control charts are also helpful in operations’ analysis.
• Movement of Items. Moving one item from one place to another requires timing of the
associated operations. Examples of such time related considerations include receiving
items to an inventory, and shipping of goods, and item movement from one machine
to another.
• Timing of Response. Timely response upon analyzed data to reduce waste is critical
in waste reduction. Correct analysis is essential in time-response analysis. Examples
of several time-driven considerations include the following:
– Operational times—What is the effectiveness of operations from KPI perspective?
– Time-sensitive issues that require quick response—How fast do we need to resolve
the time-sensitive issues?
– Machine downtimes—How can the downtime be reduced?
– Emergency response—An emergency in an operation can be a situation that requires
fast response. Such an issue can cause large amounts of waste very fast. The analysis
of the data would indicate such situations. The question then be “How fast can we
react if an emergency occurs?” A simple example is patient care in an emergency
care location where patients are treated slowly due to operational inefficiencies;
how can the operations be corrected to be able to make a difference in such an
environment that requires fast response?

Another time related tracking requirement is the timeline progress of the project for the
steps of DMAIC. The Project Charter explained in Chap. 3 with the demonstration in
Fig. 3.10 is an example of such a timeline requirement. Updating such a Project Charter
as timeline progresses with the updates on the completion of project steps with the due
dates of task completion needs to be analyzed. Such changes cannot be made easily given
that any time extension on the completion of a project is likely to result in additional
156 5 Analysis of the Measurements and Project

costs. In the case of a signed contract for the completion of a project with a deadline,
this contractual agreement would be effective. It is typically suggested to set up due dates
of project completion timelines ahead of time in case of lateness in project completion.
This aspect of early start also requires additional analysis. In the case when there is no
contract signed for the project, the customer should be made aware of the project progress
on possible lateness of the project; it is ideal to eliminate lateness of project completion
as much as possible as it reflects the professionalism of the six sigma project team. Such
a completion should be following the customer’s expectations.

5.6 Testing Scenarios Through Hypothesis Generations

Hypothesis testing is a well-known concept in STEM fields as an analysis method

for identifying improvement opportunities. Hypothesis in this setting means competing
claims. As a part of this approach, a suspected outcome is hypothesized, and tested for its
correctness. If the hypothesis fails, then the underlying mathematical reasoning needs to
be observed with its explanation within the context of the problem. Hypothesis testing is
a statistics-driven approach that tells u whether the suspected outcome is true or not with
the outcome limited to only the data used for its derivation. The larger the sample size,
the stronger the results would be. Objectives of this section include the following:

1. Explain the steps that can be taken for hypothesis testing.

2. Cover the role of confidence interval in hypothesis testing.
3. Apply hypotheses testing on different sample sizes.
4. Use z-score for calculating hypothesis testing outcomes.

By definition, a statistical hypothesis is a statement about the parameters of one or more

populations. Given that hypothesis testing requires choosing one option for testing, the
case when it doesn’t hold is the alternative option. Driven by a likelihood approach, how
likely that our hypothesis holds by using a data set would rely on the mean and standard
deviations of the nature of the data. As we covered previously, the outcomes attained
for normal distribution can be useful in such applications and it can help with hypothesis
testing. This approach can also be useful as fundamental calculations for such an approach
is well-known and already calculated; by using normally (or approximately normally)
distributed data, a confidence interval estimation will be also covered as a fundamental
concept for relevant calculations.
The use of confidence intervals in this chapter would rely on the following information:

• Structure confidence intervals for normally distributed data.

• Utilize variance information during the design of confidence intervals for normally
distributed data.
• Identify sample size based on the interval of confidence.
5.6 Testing Scenarios Through Hypothesis Generations 157

The goodness of estimation that indicates the strength of your hypothesis is needed. A
Confidence Interval (CI) is an interval that estimates such goodness of estimation. The
length of the CI is the driver of such estimation. An interval that is short is an indicator
of precise estimation. Another important interval estimation method is what is called
tolerance interval (TI). The CI approach will be used in this chapter and quality control
analysis with many other applications.
The central limit theorem is a well-known theorem in statistics and used for hypothesis
testing applications. According to this theorem, a sufficiently large data set would have an
approximately normally distributed nature under certain conditions. Having a sufficiently
large data set is the key for using this theorem, and we have the advantage of knowing
some facts about normal distribution in applications when compared to some other dis-
tributions. Given a normally distributed data set with average μ and standard deviation
σ that are not necessarily known, the 95% of the data set’s distribution falls within the
interval

(μ − 1.96σ, μ + 1.96σ )

As a general formula, one can use the following formula in which case X is the
average, k is a factor based on the data set’s percentage estimation for the corresponding
distribution, and s is the deviation from the standard:

(X − ks, X + ks)

There are applications in real life that use this normal distribution’s tolerance interval
approximation, but we are not going to cover the detail in this book.

5.6.1 Designing Intervals to Identify Confidence Levels

The formation of confidence interval includes the mean and standard deviation values.
A data set’s number distribution based on these average and standard deviations plays a
major role in understanding how the data points fit to the data set. A simple example is
understanding the role or meaning of a data set value 250,000 within a data set when we
don’t necessarily understand or know the entire data set’s values. This type of analysis is
typically seen in business and industrial settings as the entire data set cannot be necessarily
shared in a presentation. One way to understand the point’s significance in a data set
relies on structuring a random variable for the data set based on the distribution of the
data. What follows next is finding the mean and standard deviation values to be able to
determine how far the point is away from the average as a means of standard deviation.
If we are given a population, one other way to configure any point’s position based on
the average and standard deviation is by using the following z score formula that is well
known in statistics:
158 5 Analysis of the Measurements and Project

data point − average

z scor e o f a data point =
standar d deviation
The following steps can be followed to calculate a z-score of a data point within a
given sample:

• Given a population, suppose that the mean μ is unknown while variance σ2 is known.
• Let X1 , X2 …, Xn be a random sample from a normal distribution with unknown mean
μ and known variance σ2 .
• Let X be the mean value of the sample data that is normally distributed with a mean
value of μ and variance of σn .
2

• The z-score becomes the new random variable for calculations. Scaling by the square
root of n is due to the scaling need by the sample size n.

X −μ
Z=
√σ
n

• This corresponding formula for the population is the following:

X −μ
Z=
σ

The role of sample size n is significant in determining the size for accurate outcomes.
We will see later that the increase in sample size would result in better results from an
accuracy perspective.

5.6.2 CI Formula

Suppose that we have a data set with an average of X and size n drawn from a normally
distributed population with known variance σ2 . There is a close relationship between
the area under the curve of a normal distribution and the calculated confidence interval.
Figure 5.13 shows the upper and lower 0.5 α portions of the area between the distribution
and the input axis. This gives a total of α% of the entire area.
Using this information, we can make a connection between the area 1−α centered
around the mean value in the center and the CI by using the mean, standard deviation, and
the value of Z α2 (representing the upper 0.5α portion of the standard normal distribution)
by using the formula
σ σ
X − √ ∗ Z α2 ≤ μ ≤ X + √ ∗ Z α2
n n
5.6 Testing Scenarios Through Hypothesis Generations 159

Fig. 5.13 Given a normal distribution, the two tails of it in terms of α

Example 5.8 Suppose a normally distributed data set with 10 values that contains the
customer values provided by customers with a mean value of 7.3 and a standard deviation of
1.5. We can identify the spread of the data from a percentage perspective with the associated
hypothesis. We can identify the 95% confidence interval based on the average satisfaction
rate. Suppose we have.

• Data size: n = 10.

• Standard deviation: σ = 1.5.
• Sample average: X = 7.3.
• Z-value: 0.5 α = 100−95
2 % = 0.025% →; Z α2 = 1.96.

hence
1.5 1.5
7.3 − √ 1.96 ≤ μ ≤ 7.3 + √ 1.96
10 10

6.3703 ≤ μ ≤ 8.2297

Interpretation. The mean satisfaction of the customers is expected to be anywhere

between 6.37 and 8.23 based on the expected 95% data coverage.
The associated details of this example will be covered later in the hypothesis testing
section.
160 5 Analysis of the Measurements and Project

5.6.3 Choice of Sample Size

As you have been learning, we have been using two average values, X and μ that can be
confusing for new learners. The difference between the two gives rise to the error value

e = |X − μ|

In the case when X is as an estimation of the value of μ, our confidence can be

(1−α)100 percent confident that this error e would not exceed a specified amount of error
when the sample size is calculated by using the formula
( )2
σ ∗ Z α2
n=
e

Example 5.9 Continuing Example 5.8, suppose we would need to calculate the number of
data points needed for identifying 95% confidence interval using the error estimation value
of 40% of the confidence interval. Hence, we want to find n given e = 0.4, σ = 1.5, and Z α2
= 1.96. The required sample size is.

( )2
1.5 ∗ 1.96
n= = 54.023 ∼
= 55
0.4

Hence, 55 data points are needed. If we were going to lower our error estimation e to
20% then we would need about 217 data points.
( )2
1.5 ∗ 1.96
n= = 216.09 ∼
= 217
0.2

Further reducing e to 5% would result in a significant number of data points expected

to be collected:
( )
1.5 ∗ 1.96 2
n= = 3457.4 ∼ = 3458
0.05

The statistical techniques go beyond the scope of this book for calculating a variety
of confidence interval calculations for different distributions. Please refer to [13] if you
would be interested in learning more about CI.
5.6 Testing Scenarios Through Hypothesis Generations 161

5.6.4 Structuring Null Hypothesis

Structuring hypothesis testing relies on the initial guess that is called the null hypothesis.
Defining the null hypothesis is not a straightforward task; it relies heavily on the prior
observations that point to a certain outcome as the guess. For instance, data collected on
the rate of patients seen can be described as a probability distribution. This information
can be used to define null hypothesis for testing it. Once the null hypothesis is intro-
duced, the alternative option that assumes that null hypothesis does not hold is called the
alternative hypothesis.
As a simple example, we define the null hypothesis to be 2350 patients treated per
month given a data set that contains the number of patients treated each month for
10 years. In this case, the two-sided hypothesis can be structured with H0 representing
the null hypothesis while H1 represents the alternative hypothesis:

H0 : μ = 2350 patients per month

H1 : μ / = 2350 patients per month implying μ < 2350 and μ > 2350

It is also possible to design one sided hypothesis as follows:

H0 : μ = 2350 and H1 : μ < 2350

or

H0 : μ = 2350 and H1 : μ > 2350

Hypothesis testing relates to the population or the associated distribution. The main
idea behind it is to test a value’s statistical significance based on the data group that
it belongs to. Hence, test of a hypothesis is the hypothesis decision-making procedure
leading to a decision about the null hypothesis. There are certain outcomes expected from
testing a hypothesis that may result in several different outcomes. It may be possible that
the null hypothesis is correct, and we reject it while it may be false, and we may accept
it. This also leads us to the notion of making errors during hypothesis testing. Rejecting
the null hypothesis correctness leads to favoring the alternative. The region of the values
where the null hypothesis tested is accepted is called the acceptance region while the
complement of this region is called the critical region. The boundary values between the
critical and acceptance region are called the critical values.

Example 5.10 Suppose we have 10 data points with a null hypothesis based on x = 2350
patients seen per month, by assuming that the sample is a good estimation of the popu-
lation and hence, the sample’s average value represents an estimation of the population
162 5 Analysis of the Measurements and Project

to this extent. Let’s say our goal is to not reject the null hypothesis H0 :μ = 2350 when
2250 ≤ x ≤ 2450 (which is the acceptance region), and the region outside this interval is
called the critical region. In this case, 2250 and 2450 are the critical values as these serve as
the boundary values between the acceptance and critical regions for a two-sided hypothesis.

5.6.4.1 Making Type 1 Errors and Consequences

Rejection of the null hypothesis when it is true is called Type 1 error. This error’s
probability is calculated by using the formula

α = Prob(Type 1 error)
( )
= Prob Reject H0 while it is true

Significance level term is used alternatively to Type I error’s probability in addition to

calling it α-error and the size of the test.

Example 5.11 Assume that we are given the deviation from the standard to be 100 from
the average 2350 patients seen per month for 10 data points in a data set. Therefore, using
approximately normal distribution condition.

σ 100
√ = √ = 31.623
n 10
Type 1 error’s probability can be calculated by using this information:

α = P(x < 2250 when μ = 2350) + P(x > 2450 when μ = 2350)
= 0.0008 + 0.0008 = 0.0016 = 0.16%

Hence, only 0.16% of all random samples is expected to lead for rejecting the null
hypothesis μ = 2350 patients seen per month when the true mean rate is really 2350
patients seen per month.

Example 5.12 The z-values that correspond to the critical values 2150 and 2550 with the
average value of 2350 and the standard deviation of 400 would give the following:

x1 − μ 2150 − 2350
z1 = = = −1.5811
√σ 400
√
n 10
5.6 Testing Scenarios Through Hypothesis Generations 163

x2 − μ 2550 − 2350
z2 = = = 1.5811
√σ 400
√
n 10

Using at the z-score table values, the α-value is the following:

α = Pr ob(z < −1.5811) + Pr ob(z > 1.5811) = 2 ∗ 0.0571 = 0.1142 = 11.42%

Example 5.13 (Change of sample size) Now we will cover the impact of sample size
change on the calculations. Let’s triple the sample size from 10 to 30 in Example 5.12. This
change is expected to give us better accuracy for the hypothesis:

x1 − μ 2150 − 2350
z1 = = = −2.7386
√σ 400
√
n 30
x2 − μ 2550 − 2350
z2 = = = 2.7386
√σ 400
√
n 30

The z-score table provides the value P(z < −2.7386) ≃ 0.0031 hence

α = Pr ob(z < −2.7386) + Pr ob(z > 2.7386) = 2 ∗ 0.0031 = 0.0062 = 0.62%

Tripling the sample size resulted in making Type 1 error to be reduced from 11.42%
to 0.62%. One can also use the percentage to figure out the acceptance region or sample
size needed by using this formula. Excel and other programming languages can be used
for determining such values with basic calculations.

5.6.4.2 Making Type 2 Errors and Consequences

On the contrary to Type 1 error used for rejecting the null hypothesis, it is possible to
reject H0 when it is false is called Type 2 error. This error is also called β-error and its
probability is calculated by using the formula

β = Pr ob(T ype 2 err or )

= Pr ob(Rejecting H0 when it is f alse)

Example 5.14 We used a data set of sample size 10 in Example 5.12 to calculate Type 1
error and tested hypothesis for an average of 2150 with the testing relying on the sample to
be an estimation of the true population.

Let’s say we want to reject the H0 whenever the average is greater than 2550 or less
than 2150 patients seen per month. The probability of Type II error can be calculated by
164 5 Analysis of the Measurements and Project

using the values x = 2550 and x = 2150 given that the standard deviation is 400 with the
average of 2350. To do so, the calculated z-values in Example 5.12 do not change:

z 1 = −0.5811 = −z 2

Therefore, noting that Prob(z < 0.5811) = 0.7911 and Prob(z < −0.5811) = 0.281
probability of making a Type 2 error is

β = Pr ob (−0.5811 < z < 0.5811)

= Pr ob (z < 0.5811) − Pr ob (z < −0.5811)
= 0.7911 − 0.281 ∼
= 52%

This indicates the probability of making a Type 2 error to be approximately 52%.

Example 5.15 (Impact of sample size) Let’s say we double the sample size to 20 in
Example 5.14 by keeping the other values same. This is expected to provide us with better
accuracy.

x1 − μ 2150 − 2350
z1 = = = −2.2361
√σ 400
√
n 20
x2 − μ 2550 − 2350
z2 = = = 2.2361
√σ 400
√
n 20

We can calculate the following by using z-table:

β = P(−2.2361 < z < 2.2361)

= P(z < 2.2361) − P(z < −2.2361)
= 0.9873 − 0.0127 = 97.46%

The probability of rejecting H0 when it is false increased from 52% for a sample size
of 10 to 97.46% when the sample size is doubled to 20. This is a significant improvement
in accuracy as this increased probability tells us that the mistake to be made is reduced.
One other alternative that can be considered is the probability of rejecting the null
hypothesis when the alternative hypothesis is true, and this option is called the power of
a statistical test. The probability in this case is 1−β.

5.6.4.3 What Are the Differences Between the Two Error Types?
As you probably realized, Type 1 and Type 2 errors are related to each other. The
following are some of the facts about these two error types:
5.6 Testing Scenarios Through Hypothesis Generations 165

• An increase in the probability of one of the error type results in a decrease in the
probability of the other error assuming that the underlying conditions are the same.
• Assuming α is constant, an increase in sample size reduces β.
• An increase in the value of β happens as the true value of the parameter approaches
the value hypothesized in the null hypothesis when the null hypothesis is false.
• An increase in the difference of the true mean value and the hypothesized mean value
would cause the value of β to decrease.
• A Type 1 error is frequently used by using α = 0.05 for hypothesis testing; however,
this value may not be appropriate for all situations.

5.6.4.4 Additional Testing of Hypothesis Related Concepts

So far, our assumption was to have an (approximately) normally distributed data; how-
ever, this is one of the many different distributions. The complexity of mathematical
calculations may increase for different distributions for hypothesis testing which requires
a decision to be made in such a setting; interested readers can visit [13] for more advanced
hypothesis testing concepts that include but not limited to hypothesis testing for t- and
chi distributions’ related hypothesis testing analysis.
The following is one approach to use hypothesis testing in the DMAIC approach:

1. Parameters’ Identification. Identify where hypothesis testing with the associated KPIs
to be measured as there is data collected in the measurement section.
2. Right Data Selection. The size and structure of the data are important to pay attention
to in a six sigma project. If there are mistakes made during the data collection, then this
could cause significant problems in the analysis. The data corresponding to the KPIs
selected should be the focus in hypothesis testing with its impact on the outcomes.
This clearly applies to other variables if they are used.
3. Impact of Data Size. The sample size may have been pre-determined; therefore, the
completion of the measure phase clearly provides us with the sample size. If more data
points are needed to the need of a better accuracy and if it is also possible to collect
more data without biasing the nature of the data, then it would be ideal to collect
additional data.
4. Stating H0 . State the null hypothesis for each KPI that needs to be stated for.
5. Identify Test Statistic. A test statistic that is appropriate for the associated hypothesis
testing should be identified.
6. Error Type Decision. The conditions for rejection of the Null hypothesis should be
stated.
7. Calculations’ Completion. Calculate the statistical values that are needed to be found
determined for all hypothesis to be tested.
8. Sense Making Results. The attained numerical results need to be questioned based
on their meaning within the context of the problem. The interpretations should be
clearly written down in a down-to-earth manner for anyone who would need to read
166 5 Analysis of the Measurements and Project

and understand. If hypothesis testing has a critical importance for the project and if
the results are not sense making, then it is ideal to revisit the prior steps to do a better
analysis.
9. Drawing Conclusions. Supposing you have sense-making outcomes from the hypoth-
esis testing, the results should be interpreted using the problem statement in a way that
anyone can understand and follow the results.

It would be ideal to study hypothesis testing from a more comprehensive textbook if you
are interested in learning more about hypothesis testing.

5.7 Calculating Variation Using Analysis of Variance

Analysis of variance that is well-known as ANOVA, is a technique to analyze differences

in variances of determined KPIs for a comprehensive analysis of the data. The variances
of the KPIs can be attained from a single, or two or more data sets. There are three models
used for ANOVA applications:

• Random effects (Type I). The randomness of the data selection (due to withdrawal from
a population or using a random variable) impacts the outcomes and this randomness
requires additional assumptions to be stated to be able to extract meaningful outcomes.
• Fixed effects (Type II). In the case when the data set is fixed and cannot change, the
analysis of variance would have clear outcome indications.
• Mixed effects (Type III). This approach has both the random and fixed effect types in
the associated data.

A simple example in application is to be able to explain the complex reasoning between

multiple KPIs (or variables) in each data set. Suppose we are observing the effectiveness
of full-time technicians in a manufacturing setting. Given a data set with the following
features, the variation on the mean value can be analyzed based on several different factors
to find out the associated outcomes.

• The type of work completed by the technician.

• The department of the technician.
• The success rate of the technician depending on the department.
• Each technician completes a task in each department for a different job.

It is assumed that all technicians work in a rotating fashion in the manufacturing setting,
meaning they complete the task based on their specialization in different settings and
different jobs that are required to be completed in different departments. ANOVA would
result in meaningful outcomes of the overall work done by the operators and identify their
5.8 Structuring Matrix for Severity Analysis 167

success based on the mean number of jobs completed. A way to apply ANOVA for this
example is by grouping the data set components for the goal of the outcome:

• Group the data set based on the departments and identify the mean number of effective
production of operators based on their departments for each department.
• Group the data set based on the tasks completed by the technicians and identify the
mean number of effective production of operators based on their task completion.
• Group the data set based on the jobs completed by the technicians and identify the
mean number of jobs completed by the operators based on the job.
• Etc.

Given all these possible target-specific data set categorization, ANOVA can result in the
most effective production percentage to determine the operator with the best capability
for the associated job. Identification of this factor needs a structured approach to be able
to attain this result and this approach depends on the person who is analyzing the data in
a meaningful way.
ANOVA is meaningful to apply in hypothesis testing by following a set of basic rules:

• Structure the hypothesis to be tested.

• Identify the level of significance to be used.
• Select a test statistic.
• Design a decision rule for hypothesis to be tested.
• Compute the test statistic.
• Apply the appropriate data categorization based on KPIs.
• Identify the correctness of the hypothesis tested.

5.8 Structuring Matrix for Severity Analysis

Identification of the severity in relation to the corresponding probability is needed in

the severity of events and their occurrences. Analysis of the data from such a severity
of consequences approach with the associated probabilities can be visually presented by
using a threat matrix. The design of such a matrix strongly relies on the project and events
with the associated risks that depend on the risk considerations and the likelihood of their
occurrences with their impacts on the expected outcomes. The design of a threat matrix
helps with the strategies for mitigation of the risks. A threat matrix can be used for the
analysis of control mechanisms in addition to the measurement matrix that was introduced
in Chap. 4 (with Table 4.1) with the risk data collected.
Figure 5.14 displays displays one way of structuring a threat matrix [25]. In addition
to the probability of risk occurrences in the input, and the severity in the output, the color
coding of the boxes in red, yellow, and green are the indicators of the levels of severity
168 5 Analysis of the Measurements and Project

Fig. 5.14 Risks’ severity and probability can be displayed by a threat matrix [25]

and risk categorization. This severity or risk level can be assigned based on the impact of
the failure on the system. Scoring of such severity can be based on a variety of scaling
strategies (such as 10-point or 100-point scale).

5.9 Analysis of Efficiency Through Analysis

Process efficiency is a way to measure effectiveness of processes. It reflects the ratio of

maximized VA time in relation to take time as a means of percentage. Similarly, process
inefficiency can be calculated as the NVA time’s ratio to lead time. Process efficiency
ratio can be more meaningful than knowing the VA time given that knowing VA time
by itself does not indicate the efficiency level of the VA time. An additional inefficiency
calculation would also be meaningful to understand the percentage of inefficiency of
the processes/operations. Similarly, the percentage of BVA efficiency can be calculated
as well. Comparison of all the following efficiency related calculations can clarify the
process efficiency overall:
V A time
Pr ocess E f f iciency = ∗ 100%
Lead T ime
N V A time
I ne f f iciency = ∗ 100%
Lead T ime
5.9 Analysis of Efficiency Through Analysis 169

Fig. 5.15 Process efficiency formula and an example of process efficiency [25]

V A time
BV A E f f iciency = ∗ 100%
Lead T ime
Monitoring process efficiency (PE) before and after improvements can be a measure to
identify varying efficiency and inefficiency. If there are different improvement opportuni-
ties, then calculating the possible applications of these improvement ideas and calculating
the associated PE values would indicate the associated levels of improvements for differ-
ent applications. The efficiency of a process increases when the value-added time to the
process increases and gets closer to the lead time.
Figure 5.15 outlines the steps of doughnut manufacturing as an example to PE with
the associated lead time and VA time ratio general formula. Given that the process lead
time is the time it takes between customer’s online order and the time that the order is
delivered to the customer while the VA time is the time that it takes VA time during
the completion of all the steps explained during this processing time. For instance, if the
lead time prior to improvement is five hours with the VA time being two hours than PE
= 40%. In the case if the VA time is increased to three hours with the takt time being
reduced to 4 h would change PE to the 75%.

5.9.1 Analysis for Improvement of VSM

Recall that the first phase of value stream map is sketching of the operations in a nutshell
that we covered in Sect. 3.10. The next stage of VSM is to sketch the improvements on
the first phase of VSM by using the universal symbols that we covered in Chap. 3. This
170 5 Analysis of the Measurements and Project

Fig. 5.16 An example of a VSM with improvements marked on the map [25]

second phase would reflect analyzed data with the improvements placed in their respective
locations. Recall that the icon we used for improvement is the following:

• Improvement icon indicating location of improvement

An example of a basic sketching is shown in Fig. 5.16.

A VSM graph has many features that are covered previously in Chap. 3; we can easily
see that there are two locations with improvement opportunities: plate making and plate
mounting. Such a marking can be placed on different parts of the initial VSM that include
but not limited to the following components in Fig. 5.16:

• Shipment method.
• The choice of supplier.
• Printing process.
• Plate washing process.
• Storage of plates.
• Visual inspection.
• Inventory management.
• Electronic communication.
• The push logic used between processes.
5.10 Questions 171

After marking these improvements, the last phase of the VSM is the newly designed VSM
after improvements is applied and this step will be covered in Chap. 6.

5.9.2 Analysis Based on Lean Principles

Lean principles are techniques that can be used for improvements. The focus of the
analysis can be driven by a lean technique depending on how the six sigma project is
defined. There is a variety of lean techniques that can be applied with some of these
concepts covered in this book. A lean technique can be applied independently from a six
sigma project; however, lean methodologies can form a strong alliance with six sigma
approaches. Possible lean techniques that can be applied include but not limited to the
following:

• Mistake proofing (Poka Yoke).

• Design improvement.
• Facility layout improvement.
• Priority matrix.
• 5S.
• Failure mode and effect analysis (FMEA).
• Cell design and improvement.
• Heijunka.
• Single minute exchange of die (SMED).
• Kanban
– Production kanban
– Withdrawal kanban
– Three-bin kanban.
• Standard work.
• Mistake prevention.
• Design of experiment (DoE).
• Total productive maintenance.

5.10 Questions

Question 5.1 The following interaction table information is attained upon the completion
of a six sigma project applications for the scenario explained in Table 5.4.

Ml M2 M3 M4 M5
1st observation 95 55 45 35 35
(continued)
172 5 Analysis of the Measurements and Project

(continued)
Ml M2 M3 M4 M5
2nd observation 108 105 105 105 105
3rd observation 89 88 75 65 55

Answer the following questions by using this information.

a. Design an interaction table using this information.

b. What changed on these five machines based on the information provided?
c. Compared to Table 5.4 information and the interaction table analysis, which machines
got better, and which machines got worse? Please explain.
d. If you design a linear regression for each observation, what can be concluded from these
linear regression models in addition to the interaction table designed? Please explain.
e. Was the six sigma team successful in the application? Is there any specific machine
improved? Please explain.
f. Given this information, the six sigma project could be done better. How can it be
improved? Please explain your response.
g. Considering all the concepts covered so far, what other six sigma concepts covered can
help with the additional analysis for further improvement. Apply the one that you think
would help the most with the conclusions that you can draw.

Question 5.2 The following four subprocesses are observed during the completion of a six
sigma project by a manufacturer based on the lead times.

• Part 1 production (each batch is completed in 3 days and two hours).

• Part 2 production (each batch is completed in 1 day and 4 h).
• Assembly of Parts 1 and 2 (each batch is completed in 4 days and 22 h).

The following additional information is observed:

• There is a continuous flow of parts production; however, the parts are shipped between
operations in batches.
• Parts 1 and 2 are processed in parallel.
• One batch of each item is used on consecutive operations.

A six sigma project team decided to apply DMAIC approach on the operation and with the
initial process efficiencies determined to be 10%, 8%, and 15% for Part 1 production, Part
2 production, and both parts integration steps, respectively. After the completion of the six
sigma project, the process efficiencies are determined to be 15%, 12%, and 13% for Part 1
production, Part 2 production, and both parts integration steps, respectively.
5.10 Questions 173

a. Calculate the VA time of both the planning and completed stages of the six sigma project
when 10 batches are used during the parts’ production.
b. How many more parts can be produced by using the VA time after the six sigma project
is completed? Please explain your response.
c. How much did the process efficiency change? Please explain.

Question 5.3 Given your unique background in your discipline, explain two other analysis
techniques that can be applied that are not explained in this chapter.

Question 5.4 A nurse collected the following data during a six sigma project.

Number of items 5 10 15
Measurement 11.2 12.3 13.4

Apply the least squares method to identify a model that fits to this data set. Do you think
the table is recorded in the right way? Can you interpret the model by using the input and
output information. If yes, please interpret it; if not, please explain why.

Question 5.5 The nurse who collected the information for Question 5.4 collected the
following additional information for the measurement explained in the question.

Number of item 1 5 10 15
Number of item 2 12 14 11
Number of item 3 8 5 7
Measurement 11.2 12.3 13.4

These three items are used altogether to be able to collect the measurement explained in
the table. Apply multiple linear regression to design a linear model with the measurement
to be the output. Interpret your response.
Let’s Improve the System
6

Measure and analyze phases have prepared you for the improve stage and you may have
already determined improvement opportunities for the project. The major role of improve
phase in DMAIC is to identify measurable differences between the system that six sigma
project is defined and the improved (or potentially improved) system’s differences. The
ability to improve a system in place requires conceptual understanding of the inputs,
processes, and outputs in which case improvement can occur at any stage of the completed
work. Given that a six sigma project is driven by customer expectations, the improvements
are likely to be driven by the customer and therefore, the customer would need to be
consulted to be able to act upon the improvements. There are many approaches that can
be followed for improvements; however, there are certain aspects that should be kept in
mind when it comes to furthering the improvement ideas that include but not limited to
the following:

• Mistake prevention through control mechanisms.

• Answering 5W + How questions.
• Improving the design.
• Implementation of Kanban.
• Improving time aspect of project.
• Did we improve the system in place?
• Did the improvements work?
• Putting improvements in works.
• Use of technologies that can improve the systems.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 175
E. Tokgoz, Six Sigma and Quality Concepts for Industrial Engineers, Synthesis Lectures
on Engineering, Science, and Technology, https://doi.org/10.1007/978-3-031-55740-8_6
176 6 Let’s Improve the System

The improvements need to be structured in a way for them to be effective in a continuous

fashion rather than their effectiveness to be periodic (unless the operation is periodic.)
An improvement that lasts for a short period of time would cause the system in place to
return to its prior stage (or even worse) in which case the project and improvement efforts
would be meaningless. The continuity of the improvements needs to be incorporated into
the ideology of the design and the associated measurements and analysis need to be
completed afterwords. Continuous improvement would only be possible in the case when
the system does not return to its prior stages.

6.1 Preventing Mistakes Through Control Mechanisms

Prevention of mistakes before they occur or cause serious damage that cannot be fixed is
essential in the design of an improved system. Identification of defects’ root causes and
their effects on a system are critical for improvement steps to be taken. Designing a map
to detail process steps and defects along with cause-effect diagram would be valuable.
Mistakes can be prevented before they become defects and the associated work to be
undertaken completely depends on the system that is worked on and the dynamics of the
project.

6.1.1 Control Mechanism Placement at Each Phase

An improvement by controlling the system to be improved at each phase requires an

extensive investment in the control mechanism placement. In the case when the control
mechanism manual, the associated cost is reflected on the overall cost to the customer;
therefore, it is critical to pay attention to choosing control mechanism locations overall.
In some systems, placement of control mechanism at each phase of the operations is a
necessity due to the expected high-quality standards. The following steps are some of the
elements that can be utilized for this approach:

1. Attaining Feedback. A system that can generate feedback for pre-phases can be placed
for catching mistakes.
2. Mapping. A detailed mapping of the processes and their improvements can be
placed with the associated quality check points and control mechanisms. Value stream
mapping is one such example of mapping for the placement of structuring control
points.
3. Item Correction. In the case of the control mechanism application, a system for hand-
ing the defect, rework, and reject conditions needs to be known and the mapping of
6.1 Preventing Mistakes Through Control Mechanisms 177

such items needs to be placed accordingly. This approach requires a quality map-
ping with rework, reject, and rework handing depending on the type of quality issue
occurrence.

The following are some of the consequences of placement of control mechanisms

(regardless of automated or manual) within the process at each phase:

• If the system is highly reliable (mainly waste free) then the system quality check points
can be kept limited. Quality costs would be low for such systems.
• In the case when the system is highly unreliable for quality specs, the quality check
points need to increase throughout the operations; therefore, the associated quality
costs increase. In such systems the operations may produce too much waste that results
in extensive waste handling.

Waste reduction at each phase is particularly needed for high-quality production areas
and services where the mistake tolerance would be expected to be very low. High costing
low volume production lines with a high number of processing steps are a very good
example of an environment that controlling at each phase would be beneficial. In such
an environment, if there is a mistake made at an early stage of production that cannot be
fixed then the quality control at that step would stop the following processes to be kept
busy by such a production waste. If there wasn’t a quality control check point at a point
where the mistakes are not realized, the defected items would be carried over from one
process to another that would be causing NVA. Fixing such process components on time
is an important six sigma application that cannot be skipped.
The decision to place a control mechanism in each phase has a complex nature. There
are advantages and disadvantages to this approach to be explained below.
Advantages. Quick responses in the downstream operations can result in a variety of
benefits:

• Reduction of defects early on.

• Process VA time increase.
• If an item or service is correctable then the correction can be applied otherwise it may
become a waste with further processing.
• Unreliable systems with old machines and unreliable operations may require high-level
control for producing defects.

Disadvantages. There are disadvantages to this approach as well:

• A tide control mechanism can slow down operations.

• Placement of control in each location can be costly regardless of its automated or
manual control design.
178 6 Let’s Improve the System

• Too many unnecessary stops may occur interrupting the production and service. This
can result in another type of waste while trying to reduce one type of waste.

When is the best to use. It is ideal to place a control mechanism under conditions when
the quality expectations are high, and the operations are tight. Such quality control is
needed when the raw material can be costly in low volume production of items (such
as airplane engine metal frames). In addition, healthcare materials and similar nature
items may require almost zero-defect approach and this type of production or service
may require such quality mechanisms in each phase.

6.1.2 Designing a Control Mechanism for Some Phases

Control mechanisms placed in some phases in some of the systems for waste detection are
useful under certain conditions. The placement of such control mechanisms in appropriate
locations requires a detailed approach with measured and analyzed data. Locations that
happened to have unreliable processes can be identified after analysis of the collected data
and such places require control systems for quick responses in the cases when mistakes
occur that result in waste generation. This approach has also advantaged, disadvantages,
and conditions of usage:
Advantages. The main advantage of this approach is the ability to response waste
generation at the right spot on-time. Some other advantages include the following:

• More cost effective when compared to placement of control mechanism at each phase
as such a system is unnecessary due to the system’s nature.
• This approach allows you to have control over challenges of the operations where they
occur.
• Operations are not necessarily slow in the locations where mistakes are tolerated.

Disadvantages. The main disadvantage is its narrow focus of control while other loca-
tions may also have uncontrolled waste generation. Some other disadvantages include the
following:

• Cost can increase in the way that the customer may not be willing to pay for.
• Accumulation of defects can result in waste generation up until the control mechanism
locations.
• The locations that do not have control mechanisms may have serious control issues as
well if they are not realized.

Conditions of Control Mechanisms in Some Phases to be used. It is ideal to use this

approach under the conditions when the generated waste is quantitatively significant to
6.1 Preventing Mistakes Through Control Mechanisms 179

place control mechanisms in some locations. This approach requires a decision to be made
that is driven by the measured and analyzed data.

6.1.3 Systems’ Without Control Mechanisms and Improvement

Systems without a control mechanism placement is one other option as a sense-making

approach. Such systems are typically highly reliable and require hardly any control in the
production or service. One simple example is a fully automated automobile factory with
brand new machines and highly reliable structure. Data collected from such an environ-
ment would easily indicate the reliability of the system for production. A highly reliable
service can be the automated response mechanisms that take place in the customer ser-
vice lines for responding to customer inquiries. While these examples may sound highly
reliable, the system administrators’ success in setting up the associated conditions is the
main drivers of such success.
In the case when the system has waste, the occurrences of mistakes can be negligible
for them to be ignored; therefore, placement of control mechanisms may not be required.
There is also a decision-making process in this case just like any other control mechanism
case: The trade-off for cost and defect occurrence should be calculated. Typically, it is
only the output of the process that should be controlled to determine whether the product
matches the quality of what the customer is willing to pay for.

6.1.4 General Operational Control Mechanisms and Improvements

There are systems that require a control mechanism that does not necessarily rely on the
quality of the product but rather the preservation of quality. One such example is inven-
tory control by using bill of materials concepts in operations planning and control. The
control of the materials in an effective manner with the right amount and right time of
ordering from suppliers for replenishment purposes is one such example. In the case of
inventory management, as the product sits in the inventory, the quality of these items
degrades in addition to the cost to the company due to occupied space. Basic tracking of
information to understand the operations such as the following can be helpful in control-
ling the inventory through designed operations. Figure 6.1 is an outline of the operations
driven by Cost of Goods (COG) from a variety of perspectives [31]:

• Production overload.
• Logistics needs.
• Quality depreciation.
• Labor levels.
180 6 Let’s Improve the System

Fig. 6.1 Cost consideration that impact the quality on quarterly [31]

In such a setting, production levels impact the operations team members and the effec-
tiveness of lean strategies and performances. Each member of the organization (such
as managers, supervisors, planners, and controllers) needs to self-check operations for
effective outcomes, effectiveness, excess head count, and redundant activities.
There can be other applications that require specialized approaches, and we leave such
considerations for readers to investigate.

6.2 Improvements Through Sorting, Setting in Order,

Shinning, Standardizing, and Sustaining

Lean techniques help to develop organized approaches for businesses to improve sys-
tematic operations and apply corrective actions. Steps of the five S consist of sorting,
setting in order, shining, standardizing, and sustaining. The steps of application may vary
depending on the application in real life. Figure 6.2 contains the steps of 5S elements
with the explanation of these steps afterward.
The explanations of the 5S components are the following:

1. Sorting. A strategic approach to organizing an environment is through sorting of the

used items. In a typical workplace, both necessary and unnecessary items can be kept,
in which case unnecessary items that are not used overcrowded the environment and
do not add value to the work environment. In such an environment, the following are
some of the challenges that can be faced in relation to the wastes’ occurrences:
6.2 Improvements Through Sorting, Setting in Order, Shinning, … 181

Fig. 6.2 Steps of 5S

a. Time wasted to be able to find tools, machines, and paperwork within an

overcrowded environment.
b. Space wasted due be massy environment with unnecessary items placement in
locations where they shouldn’t be placed.
c. An unsorted place can cause motion or transportation waste due to the increased
distances in an overcrowded area.

Therefore, an ideal work environment can benefit from sorted entities, and this requires
identification of the needed and unneeded entities within the environment. A strate-
gic approach to such determination is by collecting and analyzing both qualitative and
quantitative data.

• Quantitative data and analysis. The number of entities used, their purposes, and the
goals of usage can be quantitatively collected. It is important to note that this usage
entities typically depend on the operation type that entities are used in the workplace;
therefore, it would be ideal to observe each operation separately. Sorting can be applied
based on the operation.
• Qualitative data and analysis. This analysis includes the information collection from
the operators (i.e., users) of the entities. Such data typically comes from surveying or
verbal communication with these operators. The type of data collected solely depends
on the project team and the scope of the project.
182 6 Let’s Improve the System

2. Setting in Order. The usage frequency of entities is an indicator of setting items in

order. Placing such entities in right locations not only speeds up the process in place
but also helps from a human factor perspective as the user of the entities would be
comforted with the operations of the entities. Just like sorting items, setting in order
has both qualitative and quantitative aspects when it comes to data collection and
analysis in application:
• Quantitative data and analysis. In addition to the frequency of usage that was men-
tioned in the Sorting phase of 5S, certain quantitative measurements may be needed
depending on the entities and observations. For instance, some items weights, some
other items duration of usage, and documents’ signature collection timing on a with
the takt time are examples of different types of quantitative data collection needed
depending on the conditions.
• Qualitative data and analysis. Just like the case in sorting, the operator’s perspec-
tive needs to be understood when it comes to operating such items for setting items
in order. For instance, the placement of items may have to be driven by the weights
of the items rather than the frequency of their use if the operator must lift these
items. A more detailed analysis of such situations would be needed in such a case
with the operator to be the decision maker.

The following are some of the highlights of setting in order application:

• Keep in mind the safety regulations and appropriateness of the items at the locations
that they are assigned.
• A visual map designed to place items within their respective places can be beneficial
for the operators as the workplace is a dynamic environment and such dynamic nature
may cause misplacement of items.
• Test whether the new “item map” is working well when it is implemented in the
workplace. The appropriate placement in the right location of such a map is needed.
• Ask the operator to evaluate the map to make sure that it serves the purpose that it
intended.

3. Shine. Making the items “shine” is a fundamental need particularly in manufacturing

and healthcare operations. Machines may tend to build up grease as operations continue
and the associated waste can increase due to this building dirt. The items’ need to shine
is a fundamental need in health care since otherwise there can be major health compli-
ance related issues. Hence, we can conclude that the “shine” aspect of 5S is a must in
some places while other workplaces can benefit from keeping items clean to make the
most out of them. Machines shining help them to operate smoothly; therefore, it helps
with waste reduction. The cleanness of tools and environment is important for clean-
ness related regulation compliances. A possible decrease in the number of machine
downtimes can increase production in return. A schedule for shining the essential
items, tools, machines, etc. can be designed as a 5S application:
6.3 Restructuring the Design for Improvements 183

• The frequency of the cleaning needs of the entities that need to be shining should
be determined.
• If applicable, a guideline for shining the entities is ideal to follow; such a guideline
can be a user manual, supplier, consultant, or some other resources.
• Design a chart for shining that includes the frequency of cleaning and how the
cleaning needs to be conducted with appropriate guidelines. Locate a place for
displaying this chart and provide a copy of it to the associated personnel.
4. Standardizing. The first 3S components’ implementation needs to become a habit and
the design of this habit is fulfilled as a part of standardizing this 3S in application. One
of the most challenging aspects of 5S is to consistently apply the 5S techniques with
potential improvements over time. The monotonicity of the daily work is one of the
reasons for 5S approach does not continue successfully and this aspect should be paid
close attention in standardizing this application.
5. Sustainability. Sustaining the 4S in implementation in a continuous manner is the
goal of Sustainability. Placement of the 5S developed materials, training of employees
in a periodic manner, and sustaining the standardized approach in a seamless way are
some of the approaches that can be followed.

6.3 Restructuring the Design for Improvements

Design improvement considerations include but not limited to.

1. Layouts,
2. Systems,
3. Operations, and
4. Services.

Design changes are expected to add value to the product or service in place with the
associated quality increase and/or cost decrease. The outcomes are expected to eliminate
NVA as much as possible; therefore, design considerations become a natural part of six
sigma projects.

6.3.1 Re-Designing Layout

Daily workspace related NVA activities are reflected on the completed works and these
activities are not recognized by workers due to the monotonicity of the daily work; such
works may require changes in the facility layout and the associated pathways used for
completing tasks. One way to determine this inefficiency is by using a Spaghetti Diagram
to follow item and staff flow. In addition to the observations by a six sigma team in
184 6 Let’s Improve the System

the workspace layout, the staff working within this workspace needs to be consulted and
they should be comfortable with the design of the workspace changes to be productive.
Experience has shown that the changes made without the staff buy-in of the six sigma
layout or operational changes may not be effective for improvement as the employees
may not apply them in the workplace because they didn’t like it. In such a case, if these
changes are required then appropriate training for the employees should be provided to
indicate the importance of the changes. Redesigning the locations of machines, materials,
office items, etc. can not only increase the productivity of the staff but also comfort them
during the tasks’ completion. The following is a set of steps that can be followed to design
an effective layout:

• Follow/observe the movements of the entities within the environment and record these
movements as a part of the Spaghetti diagram.
• Sketch the movements on a map with the associated measurements (by following units
such as time, height, width, and length) Apply a realistic scale of the layout on the
map. Such a mapping can also be accomplished by using cartons laid out in a room
with the actual lengths of the items. This scaled model can help with simulation of
items easily to observe the best design.
• The best possible design can be determined and implemented with the employees
within the environment to receive feedback from them. The staff should be comfortable
with the changes to be made.
• Use colored tapes to mark locations to point out the new locations of the items for staff
to follow these changes later. This way the staff would place items in their appropriate
places when they are dislocated.
• The new design’s printing should be kept on records and can be posted on a board.

An example of such a manual simulation approach is shown in Fig. 6.3 [1].

Another way to improve layout design is by using advanced technologies such as
simulation software that can also be helpful to visualize the operations. Figure 6.4 is
an example of a 3D simulation designed by using FlexSim. Simulation concepts will be
covered later in this chapter.

6.3.2 Redesigning Transportation

Transportation design of entities may also be the focus or a part of the improvements
needed. Transportation mode (such as transportation via air, water, and land), the method
used for transportation (such as train, truck, and air freight), and the nature of the goods
to be transported can change the improvement approach depending on the entities to be
transferred; therefore, possible transportation improvements can be a book by itself as a
6.3 Restructuring the Design for Improvements 185

Fig. 6.3 Placement of improved work environment tape markings on the ground [1]

Fig. 6.4 Manufacturing floor layout of a facility by using FlexSim

186 6 Let’s Improve the System

part of six sigma projects. For instance, transportation for healthcare items is typically dif-
ferent from transportation of manufactured items while the manufactured items can have a
different transportation nature depending on the perishability of the goods. Even with the
changing requirements of transportation several steps can be followed for improvement:

• The appropriate data should be measured and analyzed prior to improvement. The
mode and method of transportation should be initially questioned to understand if
these aspects may help with the improvements.
• Identify if the transportation route can be improved. Sometimes it is the order of goods
delivery that causes the waste which impacts the transportation route.
• Conduct a cost–benefit analysis to determine the best possible method. This cost–
benefit analysis is not always money driven.
• Design the best way of transportation with its needs; apply rerouting of the existing
transportation if needed.
• Design a map reflecting the rerouting of the transported goods.

6.3.3 Designing Documents for Project Results

The designs of documents’ improvements are often not realized in many workplaces. In
particular, the collection of signatures on a single document can be time-consuming and
it can be improved. In certain cases, the document may contain redundant information;
therefore, it may cause time to be wasted. A six sigma team needs to have a careful
approach to decisions to be made on redesigning documents as these changes can be very
tricky. For instance, a healthcare facility such as a hospital may ask the same questions
to the patient over and over and a six sigma team may think the collection of such data
may be redundant; however, the collection or checking the validity of such information
is BVA to be able to update the record or confirm the patients’ information. Automation
by using appropriate technology can be very beneficial when it comes to such data col-
lection or updates. The changes to the documents and their design may not have the best
possible structure that with NVA activities. A strategy that can be followed for document
improvement is the following:

• Understand the existing document needs and identify the best possible structure
possible from the customer’s perspective.
• Configure the documents’ strengths and weaknesses to observe the completion process
of the document and what it is designed for to be able to improve it.
• Redesign the document by taking out the unnecessary information from it.
• Provide special instructions for the completion of the document on it if needed.
• Test the improved document for approval by the customer (and/or end-users.)
6.3 Restructuring the Design for Improvements 187

The following are some of the document design details that one needs to keep in mind.

6.3.3.1 Document Complexity

The time it takes to complete a document is the duration of the document’s completion. It
is ideal to have a reduced complexity of documents to be able to improve the associated
tasks to be performed. One way to improve documentation by reducing its complexity is
to use visuals or a coded language that can be followed by employees.

6.3.3.2 Documents’ Language and Accessibility

The language used within a document is important for ease of understanding the insight
of what is explained in the document. An overcomplicated document not only causes con-
fusion for completing tasks but can also result in improper completion of the associated
tasks. Such an organization should be in a way that it has simplistic language for it to be
easily followed.
Ease of access to the documents by employees is also essential for reducing NVA
activities. Documents that are difficult to access may cause employees to disconnect from
the information of the documents. Electronic formats with online shared platforms are
typically ideal for such ease of access.

6.3.3.3 Conceptual Coverage of the Documents

Documents’ conceptual coverage requires special attention with priority to be given to the
most common issues that occur on the NVA activities related to the document. A weighted
approach with the maximum weight of corrective actions placed on the maximum number
of issues can be a measurable approach by collecting the corresponding data. The takt
time of document’s completion can also be a measure. A CTQ tree can be designed for
document improvements to be able to layout the improvement-need details. The branching
of the CTQ tree in such a case can reflect the levels of issues that need to be careful
about. Surveys collected from the end-users of the documents can help to identify such
information. This can also be accomplished by testing end-users or collecting data while
end-users are completing documents.

6.3.4 Training of Staff and Improvements

Mistakes that made resulting in NVA activities are typically due to human-made ones.
Some of the key factors that relate to human error include but not limited to the following:

• Misinterpreting or misunderstanding during communications,

• Inexperience,
• Unstructured standards,
• Missing training,
188 6 Let’s Improve the System

• Loss or reduced motivation,

• Forgetfulness,
• Misidentification of issues or tasks for completion.

Miscommunication is one of the top reasons for production of NVA activities when it
comes to production of NVA activities. It is typically ideal for organizations to train their
employees and invest in them to both improve their operations and increase their knowl-
edge base. This uplifting of the employee skillset would help the organization to have a
workforce with a healthy and valuable background. Employees’ credentials typically indi-
cate the type of training they would need based on their assigned tasks. The following are
some of the training specifications that employees can take advantage of:

• Safety. This is the top priority in almost all organizations for training and it is typically
mandatory. Additional safety training can be provided for improving issues for certain
employees to reduce NVA activities in such places.
• Quality. Just like safety, the training of quality is typically mandated by organizations
for employees to understand the meaning of quality from the organization’s perspec-
tive. Additional training may be needed for certain employees due to the importance
or additional quality standards of the specific industry.
• Use of Right Methods. This is standard training for any employee; however, the best
practices of some of the employees can be trained to other employees to make the best
of the completed products and services.
• Training in Technology. Providing training in technology to uplift employees’ tech-
nological knowledge driven by the needs of the works that need to be completed is an
essential aspect of employee training.
• Meanings of Waste and Their Place in Operational Activities. The meaning of waste
by explaining value added, non-value added, and business value added activities should
be clarified during training. Employers should make the employees aware of the qual-
ity expectations periodically and the customer perspective should be covered with
employees.
• Documentation. Documentation of the completed works with details is typically
expected as documentation is expected to be revisited in the future for similar oper-
ations. Management of such documentation and notifications are used for employee
training.

6.4 Single Minute Exchange of Die (SMED)

Single minute exchange of die is a technique for time reduction of equipment changeover.
This technique allows waste elimination during the changeover or setup time. On this set-
ting, changeover time is the time it takes to produce consecutive items. This approach can
6.5 Mistake Proofing (Poka Yoke) 189

also be adopted to different settings by defining its application. Production or service flow
needs to be monitored to be able to apply SMED within the system in place. Changeover
time reduction adds value to the process in place. Some of the essentials to keep in mind
for applying SMED to reduce NVA activities are the following:

• Internal Operations. Operations that occur while the machine is not functioning are
called internal operations. Such operations can be analyzed to identify NVA activities.
• External Operations. Operations that occur while the machine is functioning are
called external operations. Such operations can be analyzed to identify NVA activities
as well.
• Sizes of Batches. The sizes of items’ lots/collections/groupings need to be known as
such sizes are indicators of tasks completion. This can be used as the unit of completion
for NVA activities.
• Labor. Multiple or single operators may be working on the completion of the items
or service. Lack of teamwork or miscommunication can result in NVA activities based
on the reflection on the changeover time.
• Parts/Items used for operations. Appropriate choices of parts to reach the best
performance impact the changeover time. Design changes for not only items used
within machines but also the items used for operating them can also be the focus of
improvement.
• Completion of Operations’ Measurement. A good quantity of measurements of
completed operations are needed to be able to attain reliable outcomes.

External activities independence from machine run time allows them to be separated from
the functioning machine; therefore, these activities can be fulfilled independently from
the machine to reduce NVA. Internal activities do not allow operations to be separated
from the machine runtime as they occur while machine runs; therefore, internal activities
cannot be removed outside of the changeover time to reduce NVA, but they serve as BVA
activities.

6.5 Mistake Proofing (Poka Yoke)

The Japanese phrase Poka Yoke stands for mistake proofing as a technique with the goal of
reaching zero-defect. To reach such a strong outcome, usage of devices and/or procedures
effectively for error detection and elimination is the key for success. An error impacts
the economical outcomes within operations between its occurrence and realization and
this is the motivation and focus of Poka Yoke. Eliminating this error with the goal of
not repeating it would help to reach zero-defect. Identifying and applying the corrective
approach with when and how to react are the key steps to success. Poke Yoke design is
the design choice and Poka Yoke process is the detection mechanism on the process.
190 6 Let’s Improve the System

6.6 Time-Dependent Improvements

Lead time improvement relies on the improvement of all time-dependent processes. Such
a timing change is due to the changes in overall operations, and this can be seen on the
VSM at a high level. The changes on the VA, NVA, and BVA activities impact timings
of tasks’ completion. Time improvement aspects include but not limited to the following:

• Downtimes. Downtime is the duration that an entity is unexpectedly down for a certain
period. Such downtimes can be reduced by finding correct actions in several different
ways depending on the operations. Such corrective actions are typically variable.
• Setup Times. The timing of setting up processes, production, and service needs to
be designed carefully for making the best out of the operations otherwise the NVA
activities impact the setup time.
• Performance Timing. The entities used for performing operations need to be observed
with their timing. Such a timing depends on the performers’ abilities to complete tasks.
For instance, the performance of personnel is measured based on their operational abil-
ities. The same approach applies to machines that may not have the best performance
and timing expected from it. Several causes may include not being clean, old, broken,
etc. There can be a variety of ways to improve performance timing; for instance, the
timing of a machine can be improved by maintaining and controlling its performance
periodically.
• Time Management of Staff. Managing staff time related NVA activities include the
following:
– The right tasks need to be assigned to the right people to be able to complete tasks
with good quality and such tasks can be completed faster due to their skills for
fulfilling the expectations.
– Quality control of the employees’ completed work needs to be monitored to reduce
waste. The associated feedback can help with corrective actions.
– The scheduling of operations to personnel impacts the time management of the
operations and staff.
• Timing of Inventory Management. Time management of inventory impacts the
smoothness of operations and the associated VA, NVA, and BVA activities. For
instance, on-time ordering of items for the right inventory replenishment is important
for NVA activity reduction. Such a timing impacts different performance timing such
as staff time and production timeline. Important aspects of inventory time management
include but not limited to the following:
– Design of inventory utilization by using push and pull systems for improving the
timing of inventory utilization:
6.7 Did the System Improve? 191

Push System for Inventory. The items sent from the inventory out to other
locations are the “push” from inventory. Such a push mechanism is meaningful
to implement when NVA activities are eliminated. Such a push system can be
designed for any application.
Pull Systems for Inventory. The items pulled to the inventory from other loca-
tions form a “pull” system of inventory. Such a pull mechanism makes sense
to implement when NVA activities are eliminated. Such a pull system can be
designed for any application.
– Identify the expected inventory levels needed for producing an output of the process.
– Design a list of items needed to produce the output of the process.
– Local inventory markets to store and use items for successful production or service
eliminate NVA activities under certain conditions. Such an application can be useful
when it is close to the production line where it is used. Such markets can be used
for a variety of reasons including but not limited to defective item storage, items
overproduced, etc. Such an application can reduce motion and transportation waste.
• Timing of Transportation Management. Management of transportation with the right
timing of goods impacts NVA activities. Designing such transportation activities is a
research area by itself that many organizations and universities invest in for improve-
ment. There are many ways of transportation, and the associated time management
depends on the scope of the project:
– Small Transportation Operations. In the case when the transportation requires
small operations of item movement then the ways to improve the cost and the
time it takes to transport goods within the expectation of the company should be
analyzed. Corrective actions can only be taken driven by the analyzed data and
solution approaches.
– Big Transportation Operations. Big transportation operations require a large vol-
ume of items’ movement. Such improvements may require specialized personnel
with the associated knowledge for improvements.

6.7 Did the System Improve?

Questioning the system’s improvement is needed and it is an essential step of the six sigma
project. Six sigma team needs to discuss project outcomes and identify the significance
of improvements based on several different aspects:

• A detailed observation of every single step as much as possible from top to bottom is
needed.
• Pre- and post-improvement system’s KPI related data needs to be collected and
analyzed for identifying the significance of system’s evaluation.
• The method of data analysis should be questioned.
• VA, NVA, and BVA activities’ structure should be questioned.
192 6 Let’s Improve the System

• Measurable NVA activity outcomes need to be reported.

• Timing aspects of the improvement design need to be investigated.
• Staff members’ involvement in the improvements should be included.
• Document design improvements should be in place.
• Meeting the budget expectations should be questioned.
• All DMAIC steps’ comprehensiveness needs to be comprehensively covered.
• The continuous improvement aspect of the system with the improvements to last long
should be questioned.
• Meeting customer requirements and expectations should be questioned.
• Savings of the customer need to be checked.

Six sigma team members should discuss the project scope and development in a pro-
fessional manner. After discussing all these aspects (and more if needed) by the team
members, a project report needs to be produced by the team to explain project scopes,
aims, and outcomes with the materials developed to be shared with the customer’s staff.
Project outcomes need to be presented to the customer with structured charts, forms, etc.

6.8 Did the Improvements Work?

The improvements’ functionality needs to be questioned as the next step in the project
outcomes’ expectations to determine whether the improvements work effectively. One
way to identify the improvements’ effectiveness is the following (there can be other ways
to apply for identifying whether the improvements worked or not depending on the system
in place):

• Periodic application in a pilot region. The improvement technique determined upon

the six sigma project’s completion needs to be applied in a periodic manner in a pilot
(small) region. This application for a short period of time would indicate the effec-
tiveness of the applied improvements’ success. Measurements and their analysis need
to be completed within the pilot region. Statistical methods are particularly helpful in
comparing numerical results and attaining measurable outcomes.
• Full Implementation. If the periodic application in a pilot region and application
period is successful, then the improvement can be expanded from the pilot region
to a larger region.
• Similar applications. A successful improvement strategy can be applied in different
similar regions. Similar applications can be communicated by using the developed
documents and adoption of the improvements in these similar regions in application
with the necessary changes.
6.9 Improvement on Value Stream Map 193

Upon necessary changes implementation, the project team should prepare another report
outlining all the details of the project. Overall, the team needs to target answering how
much the customer expected improvements are made in the report.

6.9 Improvement on Value Stream Map

The value stream map concept is previously covered in Chaps. 3 and 5 of a process for
outlining the operations in detail from supplier to the customer by using special icons.
This map reflects VA, BVA, and NVA analysis and is helpful for a compact outline of the
operations. The improvement stage of a value stream map reflects the analyzed data for
the project with improvement opportunities marked on the associated locations. The six
sigma team should know the operational details to be able to design such a map. Figure 6.5
is an example of such a map designed based on the problem defined in Chap. 3 with the
associated initial stage VSM displayed in Fig. 3.15. As it can be seen in this figure,
some of the improvement remarks are general and well-known six sigma and lean related
concepts while some of these improvement ideas are specific to the project. The following
are the marked improvements on the VSM on this figure:

Fig. 6.5 The improvements marked on the VSM displayed in Fig. 3.15
194 6 Let’s Improve the System

• Rod stocking improvement (specific for the operation).

• Material waiting time reduction (general).
• Kanban to be designed (general).
• FIFO to be designed (general).
• Conveyor system placement (general).
• The need for proper toolbox use (general).

What follows next is to design these improvement strategies in collaboration with the
associated managers and workers to be able to make the most out of the project. The
last step is the application of the improvements and modification of the operations to
attain the newly designed system with updated quantities and icons on the VSM. In this
case, the measurements and analysis need to be repeated on the improved system iter-
atively to make the most out of the improvements and measurable outcomes. Such a
work typically requires six sigma professionals to be hired by companies as it takes time,
specialized knowledge, and energy. Many companies don’t realize the fact that such six
sigma improvement applications results in millions of dollars of savings while the hiring
of a six sigma professional will be negligible in comparison to such a cost.
Multiple iterations of DMAIC are not surprising to find the right tune of improved
operations and reduce waste for the finalized design of VSM. The six sigma team should
keep iterations of the VSMs and implement bookkeeping to view the changes on later
stages. The same idea applies for the reporting of the outcomes.

6.10 Improvements by Using Robotics

More and more companies are starting to show interest in advanced technologies and
automation by using robotics. Researching new robotics technologies and identifying new
robots that can be used as a part of six sigma projects can significantly impact the six
sigma project’s outcomes. In such a case, it is essential to conduct a cost–benefit analysis
of using such a technology in the system considered to be improved and checking with
the customer to see if the customer would be interested in adapting such technology prior
to conducting any analysis to not waste time. Clearly, the list of robotics technologies
and techniques cannot be covered in this section; however, there will be examples of such
improvements given below. Robotics can be particularly useful during several projects
that include but not limited to the following:

• Advanced vision systems (e.g., quality control).

• Reduction of human error when repetitive movements occur.
• A need for high precision occurs.
• Designing advanced error proofing systems.
• Tool trackers (e.g., error proofing).
6.11 Simulation of Operations and Layout Design with Improvements 195

• Lab work that requires repetitive movements.

• Automation of dangerous tasks that can be completed by a robot.

Examples of robots that can be used for automation purposes in manufacturing, health
care, and transportation include the following:

• ABB robotics [15].

• Fanuc robotics [21].
• Mitsubishi robotics [22].
• Kuka robotics [23].
• Universal robotics [24].

It may be essential to contract an external company or including a person with a knowl-

edge of such technology can be beneficial. One of the aspects that the customer would
likely want the six sigma team is to determine how much time it would take to recover
the cost of the technology if it is adapted.

Example 6.1 The following is an example of the use of robotics to be able to automate
work done by an operator originally by a manufacturer. The interest of this company was
to be able to utilize the talent of the operator better. The six sigma team that developed the
automation solution considered two possible options: IRB 120 and IRB 14,000robots that
are shown in the two images below. RobotStudio is a 3D robotics simulation software of
ABB that allows the simulation of IRB robots in a virtual environment [15]. There are also
industrial robotics programming languages such as Rapid programming used as a part of
ABB robots’ programming. The images displayed in Figs. 6.6 and 6.7 are images taken
from such simulation solutions.

One needs to conduct cost–benefit analysis two compare these two-robot options for
improvement. A benefit of IRB 120 is its low cost when compared to IRB 14,000 while
it requires the use of a shield due to safety reasons. The benefit of IRB 14,000 is its
collaborative nature and the two arms that allow the completion of twice as much work
of IRB 120. There are other features that need to be considered as a part of the comparison
and other robotic options can be considered.

6.11 Simulation of Operations and Layout Design

with Improvements

There are simulation software packages that are extremely useful to industrial engineers
and six sigma teams to be able to design models for physical locations and illustrate
how the operations work for current and improved phases of a six sigma project. In such
196 6 Let’s Improve the System

Fig. 6.6 The use of IRB 120 in the project that has one arm

Fig. 6.7 The use of IRB 14,000, a collaborative robot, in the project that has two arms [15]

simulation models it is easy to design an environment in which moving entities (such

as items produced, staff, operators, forklifts, and trays) along with the fixed objects that
complete tasks (such as machines, processors, and conveyor belts) can be designed to
view their interactions. Examples of such software packages that can be used for 3D
physical environments include but not limited to the following:
6.11 Simulation of Operations and Layout Design with Improvements 197

• FlexSim [16].
• Simio [17].
• AnyLogic [18].
• AnyLogistix [19].
• Arena [20].

It is essential that a person learns how to design models after choosing a simulation
software package. FlexSim and Simio are particularly useful for the simulation of 3D
environments with moving entities within the environment that include but not limited to
the following:

• Manufacturing operations.
• Healthcare operations.
• Road design and traffic flow.
• Warehouse layout design and simulation of operations.
• Facility layout and simulation of operations.

The use of simulation requires an in-depth knowledge and understanding of the physical
components along with the numerical data collected for information to be entered within
the software. Such information can be used in a table format that contains the real-life
data, or it can be in a statistical distribution format that requires the development of the
models for entering them in the simulation environment. Simulation concepts cannot be
covered extensively in this chapter as the associated contents are typically covered in
courses, and there are books written on simulation. We will illustrate several examples in
this section to give an idea about the usefulness of simulation to the reader for the learner
to understand how simulation can be employed in various settings.

Example 6.2 (Warehouse Simulation) The image in Fig. 6.8 is a warehouse layout design
by using FlexSim that exists in real life. The idea of improvement within this facility was
to improve the following:

• Improve item pickers’ ability to pick up items from shelves.

• Maximize space utilization by placing frequently used items together.
• Minimizing distance traveled.
• Maximizing the use of automation that would help with the delivery of items to the
customer.

Figure 6.9 contains the improvement ideas applied in Fig. 6.8 with the locations of
improvements highlighted in bright green.
198 6 Let’s Improve the System

Fig. 6.8 Design of a warehouse facility by using FlexSim prior to the six sigma project

Based on the improvement recommendations to the organizations by using simulation,

without providing the details, the following were the measurable outcomes attained from
the simulation only for the facility layout design of the facility:

• Expanded the existing dock by one gate for improving the loading and unloading of
items.
• Added a new dock for supporting the existing dock on a different location; this allowed
to reduce the high volume of shipment and receiving occurring on one side of the
facility and shifted some of the shipping operations to another side of the facility.
• Increased the number of racks by 19 after the reorganization of the office space.

Example 6.3 (Healthcare Simulation) The following image is a 5-floor simulation of

a clinic’s treatment operations located in New York. Patient entry to the clinic and their
treatment are entered in the simulation environment that includes an elevator. The clinic
received funding for the renovation of the entire building and the goal of the six sigma
project was to maximize the space utilization within the facility by redesigning the facility.
The following are developed during the completion of the project:
6.11 Simulation of Operations and Layout Design with Improvements 199

Fig. 6.9 Redesign of the warehouse displayed in Fig. 6.8 by using FlexSim after the improve phase
of six sigma with bright green areas marked for improvements

• Current layout’s model.

• Alternative layouts by incorporating improvements.
• Statistical results for layout and patient flow changes.
• Comparing current and potential improvement differences.
200 6 Let’s Improve the System

Alternative simulation models for other improvements are incorporated the following:

• Patient travel time and distance reduction.

• Reduce bottlenecks.
• Establish clear physical separation between departments.
• Increase and centralize storage spaces.
• Capacity increase of waiting rooms.
• Redesigned check-in areas.
• Reallocation of WIC to second floor and administrative offices to fifth floor.

Another idea of improvement was to focus of ergonomic improvements that included the
following:

• Color coded signage to direct patients to departments.

• Taping pathways by using color-coded tapes for organized patient flow throughout the
facility.

Figure 6.10 is a screenshot of the simulation developed for the project. Figure 6.11 illus-
trates the color coding of patient and staff flow with the associated operations planned
within one floor’s layout. Such a planning is necessary for developing a simulation solu-
tion to be able to enter the operations within the simulation environment. Color coding
of the operations and enumerating them are also essential for tracking the patient and
staff movement. These movements are then supported with the development of statistical
models to be entered in the simulation environment with the designated pathways labeled
within the simulation environment.

6.12 Automation of Transportation

Automation of transportation is possible by using robots or automated guided vehicles

(AGV). Such a method requires fulfillment of certain conditions for transportation as it
may not be possible to adapt such automation in every setting. Such automation can be
applied in several settings that include but not limited to the following:

• Health care. Trays used by nurses or transportation of healthcare items from one place
to another within a hospital.
• Manufacturing. Transportation of goods from one machine to another.
• Transportation service. Transportation of items by using AGVs from one location
within a warehouse to another location.
6.12 Automation of Transportation 201

Fig. 6.10 Five-Floor clinic redesign by using FlexSim simulation

Fig. 6.11 Color coding and enumeration of the operations on a single floor with the labeled path-
ways
202 6 Let’s Improve the System

Fig. 6.12 An AGV with

shelves placed on it for storage
and shipment of goods

For instance, Fig. 6.12 is an example of an AGV with trays placed on it. Such a robot
can be used in the above-mentioned settings.

6.13 Possible Applications of Artificial Intelligence (AI)

and Machine Learning (ML)

Identification and implementation of AI and ML related improvements are typically not

straightforward to realize as it requires specialized knowledge of such concepts. There
are several AI and ML applications that can be adapted in real-life settings; however,
implementation of such improvements in real-life settings requires personnel with the
associated knowledge. In addition, such improvements are not only about adapting them
in the workplace, but you also need to maintain such improvements and that requires
expensive troubleshooting requirements.
6.15 Advancements Through App Development 203

Example 6.4 (Manufacturing) Materials used, and operations of several machines are
expensive; therefore, there is a need for corrective actions to be implemented as production
continues. In such a case, given that there are aerospace standards, a system with AI capability
can implement automated defect detection that disagrees with the standards and calibrate
the machine operations to be able to make a correction on the on-going production. An
operator cannot necessarily realize the defective issues on the run as such measurements
require measurements.

Example 6.5 (Health care) During surgical procedures, there are robots used for accom-
plishing certain tasks that help with precision. Given that such a surgical environment is
dynamic, and it has a fast-moving pace, the integration of an AI system to detect robotic
movements and fixing erratic movements would be an application of AI in health care.

Example 6.6 (Transportation service) One of the most recent AI improvements planned
on transportation is driverless trucks. Such an application would have a tremendous impact
on the cost effectiveness of transportation of goods and services.

6.14 Advanced Material Handling

The use of advanced materials in the workplace requires searching for such advance-
ments. Web-based research can be the first option to be considered for identifying such
technologies. Other alternative approaches can be finding similar practices and identifying
the techniques or methods used by them. In addition, consultation can be an option to be
investigated for identifying possible advanced material handling technologies that vary as
the environment and business needs changes. It can also be possible to develop your own
material if it is a viable option based on your knowledge vase.

6.15 Advancements Through App Development

Application development is gaining more and more popularity as technological advance-

ment is increasing day by day. Adaption of cellular apps within the workplaces allows
ease of access to data and its analysis. The development of an app can have plenty of
benefits that include but not limited to the following:

• Tracking real-time production data.

• Tracking real-time analyzed production information.
204 6 Let’s Improve the System

• Viewing updates inventory and production quantities.

• Viewing defect data that can be collected and analyzed in real time.
• Fast communication.
• Fast reaction to out-of-control situations.

One needs to be careful about the development and use of such apps as there can be a
need for cybersecurity related concerns.

6.16 Use of Augmented/Virtual Reality (AR/VR)

There are various applications of augmented/virtual reality that depends on the improve-
ment opportunities considered. Several examples of such technologies include the
following:

• Visualization of facility layout and its redesign.

• Instructions displayed to operators to complete tasks by using AR/VR.
• Training of personnel on quality.
• Quality inspection.
• Inventory management.
• Systems’ design.

One needs to make sure that it is worthwhile to invest time, energy, and money in the
use of AR/VR as it can be expensive to develop such a technology. It is essential to
remember that incorporating such technologies not only increases the cost of the project
but also increases the troubleshooting costs that are likely to occur later.

6.17 Questions

Question 6.1 Discuss the trade-off between control mechanism options. Which one of the
following would be the most ideal for the following? Please explain your reasoning.

• Quality control of advanced vision systems.

• Medical implant manufacturing.
• Chocolate manufacturing.

Question 6.2 Given your unique background in engineering, explain a method that is not
explained in this chapter that can be used for improvement in a specific scenario for which
a six sigma project can be completed.
6.17 Questions 205

Question 6.3 The application of AI is rising in real life and the associated outcomes have a
valuable insight in many different settings. Find a project (that is not necessarily a six sigma
project) in which an AI improvement technique is used. Can this project be considered as a
six sigma project based on its description?
Structuring Control for Sustainability
7

The last stage of DMAIC is control that requires control over the operations with a sus-
tainable approach. The control mechanism in place can be designed in several different
ways that may be completely manual, hybrid, or fully automated depending on the nature
of the operations for the system to be improved. This control allows the improvements
to be sustainable with the system in place to be kept under control so that the system
does not return to its prior stages. A measurable way to identify the status of data points
“under-control” nature, it is ideal to design a control chart with sigma limits identifica-
tion. The design of a control chart requires determining an upper control limit (UCL), a
lower control limit (LCL), and the average (CL) as the center line of the chart for the
entire data. The following figure, Fig. 7.1, shows an animated structure of a control chart
with the UCL, LCL, and CL sketched as straight lines. In the case of data points above
the UCL, the data point is out of control. Similarly, the values below LCL also indicate
the behavior of the data either in a positive or negative way depending on the insight of
the data. The tendency to be close to the center line with minimum variation is typically
ideal on such control charts for the stability of a system in place [25].
One of the most important and useful aspects of a control chart is the ability to use
subsets of data to be able to determine the quality of items in batches. This batch size
can be one or more depending on the way the data is collected, and the size of the subset
can be constant or a varying number.
The designs of control charts for both pre- and post-improvement quality are needed
to be able to compare the outcomes. Pre-improvement out of control needs to be tracked
to see if there are improvements in the associated graphs.
Another aspect of control phase is the ability to identify and signal out-of-control
situations for the operations in place. Such control mechanisms can be determined as an

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 207
E. Tokgoz, Six Sigma and Quality Concepts for Industrial Engineers, Synthesis Lectures
on Engineering, Science, and Technology, https://doi.org/10.1007/978-3-031-55740-8_7
208 7 Structuring Control for Sustainability

Fig. 7.1 An animated control chart with UCL, LCL, and CL placed on the chart [25]

improvement of the system, or they can be placed for designing the sustainability of the
system for control purposes. Regardless of the purpose, such control mechanisms have a
tremendous impact on the operational quality. The following is the content that will be
covered in this section:

• Control conditions’ statistics.

• Control charts.
• Under-control and out-of-control conditions.
• Documenting outcomes for the sustainability of project outcomes.
• How do we finalize a six sigma project?

Just like in the other phases of DMAIC, quantification and analysis of a data set requires
calculations such as average and standard deviation. The significance of the results is
driven by the number of data points in a data set and such quantification is cared sig-
nificantly on the reliability of the results. Identifying and reacting to “out of control”
data points open a door for further improvement ideas and opportunities. Documenting
all stages of the project is a must for the project’s healthy reporting that will be also
explained in this section.
7.1 Data Collection After Improvements 209

7.1 Data Collection After Improvements

The quantification of the improved system right after improvements is necessary to iden-
tify the impact level of improvements. Even if the improvements are completed in a
theoretical manner (only improvement ideas implemented either by software or by a logi-
cal approach) without applying in real life) the real-life outcomes would reflect the actual
improvement results in application. There can be several unforeseen matters that may be
faced in real-life application that may be due to the lack of data recording, human errors,
machine malfunctions, etc. The same steps followed for pre-improvement can be applied
for post-improvement purposes.
Collection of data prior to improvements. Recall that the collected data followed
a certain methodology. The out-of-control points determined for this data set would be
useful for post-improvement analysis. We should already have cause-effect outcomes for
the out-of-control points from the analyze phase; therefore, these extraordinary points
need to be paid special attention (unless they are unusual points in the data set that are
not encountered in a normal fashion (i.e., outliers.)
Collection of data after improvements. Now that we have a pre-improvement control
chart with the out-of-control points identified, we need to have the post-improvement data
collection and the associated out-of-control points that follow the pre-improvement data
collection steps; this is when the project team pays attention to details and collect data:

• Explain and record the pre-improvement details of data collection.

• List the pre-improvement specifications such as the operator, the period the data
collected, and the date of collection for the data set.
• Pre- and post-improvement data collected need to respect the same underlying con-
ditions as much as possible because otherwise there will be bias on the outcomes
attained. If such an approach is not possible, the differences between the two need to
be clearly outlined and explained. For instance, if a nurse collects data every 30 min
from a patient, then the same data can be collected from the same patient every 30 min
after the improvement of the process with the same patient.
• All the specs of the data collected in both periods need to follow the DMAI step ele-
ments. Deviating from this pattern can result in unknown data formation and outcomes
unless the improvements require the SIPOC to change.
• The input and output should have the same nature IPO unless the improvements require
them to differ for pre- and post-improvements.
• The control charts should be in alignment with the goals of the project.
• Special periods impacting the data require special treatment. For instance, a company
with a special sales deal announced may experience a shortage in supplies; therefore,
the number of sales would be an outlier in the typical number of sales as well as the
handling of inventory levels for such sales as well as replenishment of these goods.
210 7 Structuring Control for Sustainability

• The data size is an important aspect of the collected data, and a substantial number of
samples need to be drawn from the population for accuracy of the data.

The next section will be focusing on the fundamentals of continuous and discrete data
control charts.

7.2 Structuring Charts for Control Observations

Everyday operations have natural control limits that can even be seen in a person’s life.
Such limits on operations are identified for drawing a limit on the boundaries of expecta-
tions and such expectations depend on a certain number of factors such as quality, cost,
reliability, and expected project completion. Quantification of such information is a nat-
ural need for supervision of the product or service. This approach depends on the nature
of the data that can be either one of the following as we covered previously:

• Continuous.
• Discrete.

The control chart ranges are typically formed in even multipliers of sigma such as 2σ, 4σ,
and 6σ . On some control charts only 6σ range is displayed while some others display
the even sigma ranges. The examples to be given in this section would have a variety of
ranges to be displayed below.

7.2.1 Control Chart Definitions

Formulas used for designing control charts will be given in this section The following
abbreviation will be used:

• UCL . Control limit designed for upper (maximum) quantification.

• LCL . Control limit designed for lower (minimum) quantification.

These control limits have a variety of applications based on the provided data:

1. The limits UCL and LCL are statistical quantification of the controls on quality or
similar conceptual considerations that allow such limits’ calculations.
2. The analysis associated with control charts depends on the average, standard deviation,
etc. of the data. Such information may depend on continuous distributions or discrete
data attained.
7.2 Structuring Charts for Control Observations 211

3. Meaningful explanations of “Out-of-control” points based on the procedural operations

need to be explained. These points and the rest of the data cannot be easily explained
by an outsider; therefore, close involvement with the operations may be a necessity for
the six sigma team to be able to reduce waste. This can also be accomplished by the
organizational connections within the system.

Quality control analysis requires careful data collection both before and after improve-
ments:

• Data Collection Prior to Improvements. You need to be able to answer a set of

questions to be able to fulfill the needs of data collection:
– We have the right data collection strategy?
– Does the collected data aim to answer the questions that we want to answer?
– Does the data collected have the right size and batch quantification?
– Have we worked with the right people for data collection?
– Do we have the right SIPOC components’ identification?
• Data Collection After Improvements Applied.
– The control chart after improvements indicates the outcomes of improvement from
a statistical significance and sigma range perspectives.
– Close attention needs to be paid to the areas where improvements are needed to
observe how changes impacted on the system.
– The post-improvement data may still have out-of-control points; however, this
doesn’t mean the system has unusual behavior; these out-of-control points would
be indicators of next phase improvement opportunities.

The following are the basics of control charts that we would use for formulation:

• Let the subset partition of a data set G to be represented by Gi with k number of

subsets collected:

G = {G 1 , G 2 , . . . , G k }

In this data set each subset may or may not have the same number of data points.

• G i (G-bar). Average of subset Gi for all i = 1,2…k

• G (G double bar). Average of the subsets average:

G1 + G2 + · · · + Gk
G=
k
212 7 Structuring Control for Sustainability

• Ranges. Given a data set, the range between the maximum and minimum values of
the data set can be calculated by using the formula

G = G max − G min

where G min and G max represent the minimum and maximum values of the data set G
simultaneously. The same formula applies to subset G i to be able to calculate the average
ri of each subset G i

ri = G i ,max − G i,min

where

G i ,max is the maximum value of Gi

G i,min is the minimum value Gi

The average of ranges r os subsets ri is the averages of all range averages:

r1 + r2 + · · · + rk
r=
k
Control charts require the use of certain values that are going to be used for both con-
tinuous and discrete distributions [13]. There are three control charts that can be designed
for a continuous random variable that we will cover:

• X-R Chart (Xbar-R chart). There are two graphs that need to be designed for this
chart: X graph that represents the subset data set average points and R graph repre-
senting the range of each subset. X -R chart can be used for a set with subsets of size
n = 4, 5, or 6 which can be at most 10.
• X-S Chart. There are two graphs that need to be designed for this chart: X graph that
represents the subset data average points and “S” graph of subsets’ standard deviations.
Sample sizes with a size greater than 10 or samples with a continuous variable can
benefit from this chart.
• IMR. The IMR graphs can be useful when the subset size is 1. This chart is essentially
used for a regular data set with no subset classification (Table 7.1).
7.2 Structuring Charts for Control Observations 213

Table 7.1 A table of control chart values

Sample Size n
2 3 4 5 6 7 8 9 10
A2 1.88 1.023 0.729 0.577 0.483 0.419 0.373 0.337 0.308
Constant X-bar_R D3 0 0 0 0 0 0.076 0.136 0.184 0.223
value chart D4 3.267 2.575 2.282 2.115 2.004 1.924 1.864 1.816 1.777
X-bar_S A3 2.659 1.954 1.628 1.427 1.287 1.182 1.099 1.032 0.975
chart B3 0 0 0 0 0.03 0.118 0.185 0.239 0.284
B4 3.267 2.568 2.266 2.089 1.97 1.882 1.815 1.761 1.716
C4 0.7979 0.8862 0.9213 0.94 0.9515 0.9594 0.965 0.9693 0.9727
IMR D3 0 0 0 0 0 0.076 0.136 0.184 0.223
diagram D4 3.267 2.574 2.282 2.114 2.004 1.924 1.864 1.816 1.777
d2 1.128 1.693 2.059 2.326 2.534 2.704 2.847 2.97 3.087

A more extensive table of sample values is shown in Table 7.2 for sample sizes larger
than 10 less than 25.

7.2.2 Structuring Continuous Data Control Charts

We will cover the three control chart types for continuous random variable data in this
section.

7.2.2.1 XBar-R Chart

As explained in the previous section, there are two graphs for an XBar-R chart:

a. X graph. Given a set A with subsets Ai , the XBar graph is the plot of all the average
values A of the data are placed as the data points, and A, lower control limit, and upper
control limits are placed as the limit of control. The following formulas are used for
finding these values in which A2 is the constant to be looked up in Table 7.1 depending
on the subset size:

UC L−X Bar = X + A2 R

μ X Bar = X

L C L−R = X − A2 R
214 7 Structuring Control for Sustainability

Table 7.2 Constants used for control chart calculations

Sample size = m A2 A3 d2 D3 D4 B3 B4
2 1.880 2.659 1.128 0 3.267 0 3.267
3 1.023 1.954 1.693 0 2.574 0 2.568
4 0.729 1.628 2.059 0 2.282 0 2.266
5 0.577 1.427 2.326 0 2.114 0 2.089
6 0.483 1.287 2.534 0 2.004 0.030 1.970
7 0.419 1.182 2.704 0.076 1.924 0.118 1.882
8 0.373 1.099 2.847 0.136 1.864 0.185 1.815
9 0.337 1.032 2.970 0.184 1.816 0.239 1.761
10 0.308 0.975 3.078 0.223 1.777 0.284 1.716
11 0.285 0.927 3.173 0.256 1.744 0.321 1.679
12 0.266 0.866 3.258 0.283 1.717 0.354 1.646
13 0.249 0.850 3.336 0.307 1.693 0.382 1.618
14 0.235 0.817 3.407 0.328 1.672 0.406 1.594
15 0.223 0.789 3.472 0.347 1.653 0.428 1.572
16 0.212 0.763 3.532 0.363 1.637 0.448 1.552
17 0.203 0.739 3.588 0.378 1.622 0.466 1.534
18 0.194 0.718 3.640 0.391 1.608 0.482 1.518
19 0.187 0.698 3.689 0.403 1.597 0.497 1.503
20 0.180 0.680 3.735 0.415 1.585 0.510 1.490
21 0.173 0.663 3.778 0.425 1.575 0.523 1.477
22 0.167 0.647 3.819 0.434 1.566 0.534 1.466
23 0.162 0.633 3.858 0.443 1.557 0.545 1.455
24 0.157 0.619 3.895 0.451 1.548 0.555 1.445
25 0.153 0.606 3.931 0.459 1.541 0.565 1.435

b. R graph. Using the subsets Ai of set A graphed for XBar graph, R graph displays
the plot of the range values Ri along with the values ofR, upper control limit, and
lower control limit that are placed as the control limits. The constant values D4 and
D3 displayed on this graph need to be taken from Table 7.1 depending on the subset
size n.
7.2 Structuring Charts for Control Observations 215

UC L−R = D4 ∗ R

μR = R

L C L−R = D2 ∗ R

Example 7.1 (X-R Chart) Suppose a healthcare personnel collects data by using the equip-
ment. The patients’ measurements can be used to determine out-of-control conditions for
each patient for a particular illness. The collected subset data has each six data points and
the associated calculations are determined to be the following:

X = 126

R = 12

UC L−X Bar = X + A2 R = 126 + (0.483)(12) = 131.8

L C L−X Bar = X − A2 R = 120.2

UC L−R = D 4 ∗ R = (2.004)12 = 24.048

μ R = R = 12

L C L−R = D 3 ∗ R = 0

In real-life situations these values need to be calculated by either using the data or the
corresponding distribution.

Example 7.2 (Solar Panels) Suppose the goal of a six sigma project is to determine the
wasted space on roofs based on the panel implantation. The following data displayed in
Table 7.3 is collected by a six sigma team that reflects the summary values for the project.

There is certain fundamental information and rules that need to be known and investi-
gated if the team is new to the type of project. One example is the need for a maximum
75% space solar panel placement rule on roofs of a real-estate located in New York
State as the maximum solar panel capacity placement. For instance, if there is 65% space
utilization on one roof then this simply indicates 10% wasted space on the roof.
216 7 Structuring Control for Sustainability

Table 7.3 Table containing statistical quantities to summarize a data set of two solar panel types
Panel type 1 – space Panel type 1 Panel type 2 – space Panel type 2
utilized (%) – wasted space (%) utilized (%) – wasted space (%)
Average 70.43% 4.57% 68.63% 6.37%
Medain 70.71% 4.29% 69.02% 5.98%
Mode 70.71% 4.29% 68.90% 6.10%
SD 0.009147663 0.009147663 0.011587785 0.011587785
2SD 0.0182955326 0.018295326 0.023175571 0.023175571
Q3 70.95% 4.88% 69.37% 6.61%
Q1 70.13% 4.05% 68.39% 5.63%
Max 71.16% 9.00% 69.61% 10.41%
Min 66.00% 3.84% 64.59% 5.39%
IQR 0.82% 0.82% 0.99% 0.99%

Fig. 7.2 Cause-and-effect diagram for the solar panel utilization information

The following fish-bone diagram in Fig. 7.2 is designed for causes and their impact on
the outcomes.
Looking at the original data, further analysis of the space wasted and utilized by using
the box-whisker diagram can be used as well; the associated two figures are shown below.
These plots display the space utilized and wasted space for both panel types with two
7.2 Structuring Charts for Control Observations 217

Fig. 7.3 Space utilization box-and-whisker diagram

extreme outliers. For Type 1 panels (300-W), an outlier of 66% can be seen also the
minimum of the data set as an extreme outlier meaning the specific roof size 300 square
feet is the most system that needs improvement. This can also be seen in Table 7.3 for the
wasted space outlier of 9% attained for the 300-W system with a reflection in Fig. 7.3.
This means that the 300 square feet roof size is not utilizing the space correctly and may
be inefficient.
Considering Type 2 panel (400-W system), the extreme outlier is 64.59% attained for
the 300 square feet roof. The corresponding wasted space plot with the extreme outlier
was found to be 10%, which is even more than the 300-W system. Hence, we can conclude
that both 300-W and 400-W systems contain the most wasted space for 300 square feet
size roofs (Fig. 7.4).
The control charts designed for pre- and post-improvement are shown in Figs. 7.5 and
7.6. As can be seen on both graphs the associated systems are under control during both
pre- and post-improvements. The only change is in the output percentages as the system
is now faster and capable of producing more. The improvement idea presented in this
case is the use of robotics and the automation of transportation from one place to another
by using robotics. We can see from these two figures that the average loading time in
minutes (the output axis) for a robot is approximately 50% lesser than human’s loading
time.

7.2.2.2 Graphs for X-S Chart

An S-chart can be significantly useful for understanding the standard deviation behavior
of subset information when the data size is larger than 11 or the data is represented by
a continuous variable. Just like the X-R chart, X-S chart has two graphs; X graph and S
graph; however, the formulas for X graph for these two charts is different as it will be
seen below on the associated formulas.
218 7 Structuring Control for Sustainability

Fig. 7.4 Wasted space for the two types of solar panels

X Bar Chart for Solar Panel Post-improvement

35.50
35.40
35.30
35.20
35.10
35.00
34.90
34.80
34.70
34.60
34.50
1 2 3 4 5 6 7 8 9 10

X Bar X_D_Bar UCL_Xbar LCL_Xbar

Fig. 7.5 Solar panel transportation improvement results

• Graph of X values. This graph contains the subset average values that incorporate
the standard deviation average s as a factor in the upper and lower control limits. The
constant value A3 is given in Table 7.1. The average is the center line of the graph.

UC L−X Bar = X + A3 s
7.2 Structuring Charts for Control Observations 219

X Bar Chart for Pre-Improvement

65.80

65.60

65.40

65.20

65.00

64.80

64.60

64.40
1 2 3 4 5 6 7 8 9 10

X Bar X_D_Bar UCL_Xbar LCL_Xbar

Fig. 7.6 Solar panel transportation pre-improvement analysis

μ=X

L C L−X Bar = X − A3 s

• Graph of S. This is the graph designed for displaying standard deviations of all subsets.
The constant values B3 and B4 are given in Table 7.1.

UC L−s = B4 ∗ s

μs = s

L C L−s = B3 ∗ s

The IMR chart is structured for subsets of size 1 and this chart is left as an exercise to
the reader.
220 7 Structuring Control for Sustainability

7.2.3 Designing Control Charts for Discrete Data

The following are the four different types of control charts that we will cover in this
section:

• c-chart.
• u-chart.
• p-chart.
• np-chart.

7.2.3.1 Designing c-charts

A discrete sample data set with a constant sample size can benefit from a c-chart for
identifying control limits for defective parts. Let.

• v: Sample data set’s number of defects per unit.

• g: Sample size.

Upper and lower control limits can be calculated by using the following formulas:
v
μ=c=
g
√
UC L−c = c + 3 c

√
L C L−c = c − 3 c

The upper and lower control limits represent the ∓3 standard deviation (i.e., ∓3σ )
from the center line value that represents the six sigma range from the average. In the
case of a c-chart, the UCL and LCL are both single straight lines. Depending on the
expectations of the organization, the quality control limits for the data within the range
of ∓3σ signifies the 6σ range. The points that tend to be close to the center line are the
values that are expected to be close to the average expected value. Below LCL and above
UCL data points are out-of-control points. Negative numbers are not possible to attain
from these graphs given that the formulas are nonnegative.
The assignable causes need to be investigated in the case when the data is out-of-
control.

7.2.3.2 Designing p-charts

Data sets that have variational sample sizes with a size of 6 or more can benefit from p-
charts. This chart can be used for identifying a fraction of non-confirming/defective units
resulting from the sampled production. Let.
7.2 Structuring Charts for Control Observations 221

• vj : Defective number of parts in subgroup j.

• l: Number of subgroups.
• gj : Subgroup j sample size.

Upper and lower control limits can be calculated by using the following formulas:
vj
pj =
gj
p 1 + p 2 + · · · + pl
Center Line = p =
l
/
p ∗ (1 − p)
UC L− p = p+3∗
gj
/
p ∗ (1 − p)
L C L− p = p−3∗
gj

The upper and lower control limits represent the ∓3 standard deviation (i.e., ∓3σ )
from the center line value that represents the six sigma range from the average. In the
case of a p-chart, the UCL and LCL are not necessarily single straight lines but rather
piecewise straight lines. Depending on the expectations of the organization, the quality
control limits for the data within the range of ∓3σ signifies the 6σ range. The points
that tend to be close to the center line are the values that are expected to be close to the
average expected value. Below LCL and above UCL data points are out-of-control points.
Negative numbers are not possible to attain from these graphs given that the formulas are
nonnegative.

7.2.3.3 Designing u-charts

A data set with varying sample size can benefit from a u-chart. This control chart can be
useful for displaying the number of non-conformities per unit. Let.

• rj : Number of defects in the production of set j.

• l: Number of subgroups.
• qj : Sample size of set j.

A c-chart can be used for the number of defects in a sample while the u-chart is useful
for investigating the average number of defects per sample unit. The following are the
formulas The control chart graphing components can be calculated by using the following
formulas:
rj
uj =
qj
222 7 Structuring Control for Sustainability

u1 + u2 + · · · + uk
μu = u =
k
/
u
UC L−u = u + 3
qj
/
u
L C L−u = u − 3
qj

The upper and lower control limits represent the ∓3 standard deviation (i.e., ∓3σ )
from the center line value that represents the six sigma range from the average. In the
case of a u-chart, the UCL and LCL are not necessarily single straight lines but rather
piecewise straight lines. Depending on the expectations of the organization, the quality
control limits for the data within the range of ∓3σ signifies the 6σ range. The points
that tend to be close to the center line are the values that are expected to be close to the
average expected value. Below LCL and above UCL data points are out-of-control points.
Negative numbers are not possible to attain from these graphs given that the formulas are
nonnegative.

7.2.3.4 Designing Np-Charts

A data set with constant sample size of 6 or more can benefit from np-charts. An np-chart
can be useful for determining control limit conditions for number of nonconforming units.
Let.

• fj : Defective part number in subgroup j.

• k: Number of subgroups.
• n: Subgroup sample size.

The control limit values are the following formulas:

dj
pj =
n
p 1 + p 2 + · · · + pk
μ=n∗p=n∗
k
√
σ = n ∗ p ∗ (1 − p)

UC L−np = n ∗ p + 3σ

L C L−np = n ∗ p − 3σ
7.3 Out-Of-Control Conditions’ Handling 223

The upper and lower control limits represent the ∓3 standard deviation (i.e., ∓3σ )
from the center line value that represents the six sigma range from the average. The
UCL and LCL are both straight lines. Depending on the expectations of the organization,
the quality control limits for the data within the range of ∓3σ signifies the 6σ range.
The points that tend to be close to the center line are the values that are expected to be
close to the average expected value. Below LCL and above UCL data points are out-of-
control points. Negative numbers are not possible to attain from these graphs given that
the formulas are nonnegative.
The assignable causes need to be investigated in the case when the data is out-of-
control.

7.3 Out-Of-Control Conditions’ Handling

Designing a system to detect and prevent out-of-control points can be essential for a
system to have a stable behavior. Cause-effect type of charts need to be typically used
for determining causes to these data points for preventing the mistake that happens and
designing a control mechanism or system that can be manual or automated to either
stop or out-rule the occurrence of mistakes. The reaction to such occasions requires a
systematic approach. Two ways to react a mistake in a process when it happens are the
following:

(1) Halting operations due to mistake occurrence. There are certain mistakes that
require stopping the local or global operations depending on the nature of the mistake.
Such mistakes would be reflected on the control charts as extreme points due to the
changes in the natural trend of the data. Such a mistake can be anything, and a smart
approach would be to place a mechanism to warn the system’s personnel before its
occurrence.
• Depending on the conditions, the halting of the operations due to safety reasons can
be the case. This may or may not be reflected on a control chart depending on the data
collection strategy. Regardless of its nature, such a condition needs to be identified
and its nature needs to be studied to prevent before and after effects.
• Under certain cases, the production may stop due to operational reasons, and this is
typically the case in most of the halted operations. Just like the safety driven rea-
sons, the causes, and effects of halting the operations should be determined with the
associated control charts if possible.
(2) A warning Requiring Mistakes: Majority of systems require a warning system in
certain locations that allow the recognition of the mistakes made. Regardless of the
automated nature of such a system, the reaction to such a mistake can be eminent.
Advanced control systems for a variety of industrial applications exist today and it
would be the six sigma project team’s responsibility to determine such technologies
224 7 Structuring Control for Sustainability

if used for the project. Manual warning systems are frequently used by employing
staff and setting warning mechanisms that are triggered by the staff.

As always, a fish-bone diagram can be useful for assignable cause detection when
mistakes occur.

7.4 Reporting Outcomes and Completing Project

with Documentation

Reporting the six sigma project outcomes in a comprehensive manner is essential since
otherwise the techniques used, the resulting outcomes, and the steps of DMAIC method-
ology will be forgotten and would not be communicated with the associated stakeholders.
Some of the details of document finalization should include the following:

• The improvements need to be displayed wherever necessary for the associated person-
nel to be able to look at and follow to be able to design the system components. Such a
guidance is essential for changing staff as the existing staff would know how to follow
such methods and would not need it after a while in the workplace.
• Quality related visuals need to be displayed in a dynamic fashion to the employees
that take place in the operations/production if needed. Such dynamic displays help the
employees to be able to track the quality elements of their own operations and benefit
from reducing the costs.
• Meaningful charts and graphs need to be the produced that reflect the project outcomes;
not doing so can result in displays that can result in confusing displays that won’t be
understood by anyone if they are explained in the report.
• A copy of improvement documentation may need to be kept in the locations where it
is needed for employees to be able to look and make changes.
• If statistical quality control charts are displayed to employees, then they need to be
explained carefully in a way that they can understand and apply in their operations.
• The employees need to be trained about the wastes with the associated categories and
how these wastes can be eliminated.

Final reporting should not only contain an explanation of the DMAIC applications and
outcomes but also include the displays that would be placed for employees to follow on
their own. Such a design needs to be very elementary as the employees do not necessarily
have the fundamental knowledge of the associated concepts.
Another important aspect of the project completion phase is the preparation of a pre-
sentation to the customer and stakeholders that contain simplified explanations. The ability
to professionally present project outcomes won’t be covered here as it requires additional
7.5 Questions 225

training and developing skillsets. The following are some of the highlights that can be
included in the six sigma project team’s presentation:

• Definition of the problem.

• Pre-improvement SIPOC explanations.
• Nature of the collected data and KPIs.
• Identified wastes and their locations.
• Improvements needed and how they can be applied.
• Locations where improvements can be made.
• Pre- and post-improvement statistical significances.
• Analysis of pros and cons for improvement.
• DMAIC results attained for each phase.
• Conclusions and finalization.

The presentation of the outcomes depends on who the customer is.

7.5 Questions

The following data is covered previously in Chapter 4, Example 4.15.

Patient ID 1120 1121 1122 1123 1124

Reference value 6 8 10 12 14
Observations 1 6.8 7.2 9.6 11.9 13.8
2 7 7.1 10.1 11.8 14.1
3 5.6 7.8 10.4 12.1 14.2
4 5.8 7.9 10 12 13.6
5 6 7.9 9.9 12.3 13.7
6 5.9 8.3 10.1 11.9 13.6
7 6.7 8.2 10.5 11.8 13.9
8 6.7 8.4 10.4 12.2 14.5
9 6.6 8 10.1 12.5 14.3
10 6.3 7.8 9.5 11.5 14.5

Question 7.1 Use this example to structure an X-R chart. Interpret the under-control points
on the graph and explain the outcomes attained.

Question 7.2 Use this example to structure an X-s chart. Interpret the under-control points
on the graph and explain the outcomes attained.
226 7 Structuring Control for Sustainability

Question 7.3 Use this example to structure a p-chart. Interpret the under-control points on
the graph and explain the outcomes attained.

Question 7.4 Use this example to structure an np-chart. Interpret the under-control points
on the graph and explain the outcomes attained.

Question 7.5 Use this example to structure a c-chart. Interpret the under-control points on
the graph and explain the outcomes attained.

Question 7.6 Use this example to structure a u-chart. Interpret the under-control points on
the graph and explain the outcomes attained.
Epilogue

A basic introduction to Lean Six Sigma concepts is covered in this book from industrial
engineers’ perspective. Given that the focus of industrial engineering is to engineer any
industry, this book can be particularly useful when more advanced concepts are included
in applications in many different settings. In the accomplishment of such tasks, a broad
knowledge of statistics can be very helpful, and individual training can also be beneficial
for each one of you in background development and applications. Based on my expe-
riences with both industry and student groups, some of the industrial engineers could
naturally see and adopt changes during the projects’ completion while some others strug-
gled. It is essentially coming down to the interest of the individual to reduce waste and
developed experiences regarding such improvements. Excel is commonly used in appli-
cations; however, the use of additional technology such as Minitab, SPSS, and MATLAB
software can be helpful if you know how they can be used.
The information shared in this book is a result of my theoretical knowledge and exten-
sive experiences in Lean Six Sigma applications with a variety of organizations. It is also
a reflection of the Green Belt and Black Belt Six Sigma training certificate programs that
I initiated for industrial engineers. Growing in knowledge is the key, therefore continuous
improvement comes with continuous learning. You can also view your life as a six sigma
project with the value-added, business value-added, and non-value-added activities laid
out to be able to improve it for both your professional and personal life. Given that this
book has a certain content coverage, it is essential to be able to select the right application
materials. Incorporating a real-life project to the content covered in this book would be
an ideal approach for a course’s coverage.
Prof. Emre Tokgoz.

© The Editor(s) (if applicable) and The Author(s), under exclusive license 227
to Springer Nature Switzerland AG 2024
E. Tokgoz, Six Sigma and Quality Concepts for Industrial Engineers, Synthesis Lectures
on Engineering, Science, and Technology, https://doi.org/10.1007/978-3-031-55740-8
References

1. Arcidiacono, G., Calabrese, C., & Yang, K. (2012). Leading processes to lead companies: Lean
six sigma: Kaizen leader & green belt handbook. Springer Science & Business Media.
2. Devane, T. (2004). Integrating lean six sigma and high-performance organizations: Leading the
charge toward dramatic, rapid, and sustainable improvement (Vol. 4). John Wiley & Sons.
3. Galli, B. J. (2018). Risks related to lean six sigma deployment and sustainment risks: How
project management can help. International Journal of Service Science, Management, Engi-
neering, and Technology (IJSSMET), 9(3), 82–105.
4. Goldsby, T. J., & Martichenko, R. (2005). Lean six sigma logistics: Strategic development to
operational success. J. Ross Publishing.
5. John, A., Meran, R., Roenpage, O., & Staudter, C. (2008). Six sigma+ lean toolset: Executing
improvement projects successfully. Springer Science & Business Media.
6. Kaswan, M. S., Rathi, R., & Khanduja, D. (2020). Integration of green lean six sigma: A novel
approach for sustainable development. International Journal of Six Sigma and Competitive
Advantage, 12(4), 389–405.
7. McCarty, T., Daniels, L., Bremer, M., & Gupta, P. (2005). Six sigma black belt handbook (Six
SIGMA Operational Methods). McGraw-Hill Education.
8. Quality Tools, American Society of Quality (ASQ). Retrieved 21 February, 2023, from https://
asq.org/quality-resources/quality-tools.
9. Tang, L. C., Goh, T. N., Yam, H. S., & Yoap, T. (2007). Six sigma: Advanced tools for black
belts and master black belts. John Wiley & Sons.
10. Theisens, H. C. (2020). Lean six sigma green belt: Mindset, skill set and tool set. Van Haren.
11. United States Department of Labor, Occupational Outlook Handbook. (2013). http://www.bls.
gov/ooh/Architecture-and-Engineering/home.htm.
12. Voehl, F., Harrington, H. J., Mignosa, C., & Charron, R. (2013). The lean six sigma black belt
handbook: Tools and methods for process acceleration. CRC Press.
13. Montgomery, D. C., & Runger, G. C. (2010). Applied statistics and probability for engineers.
John wiley & sons.
14. Meran, R., John, A., Roenpage, O., Staudter, C., & Lunau, S. (2013). Six Sigma+ lean toolset:
Mindset for successful implementation of improvement projects. Springer.
15. ABB robotics, https://global.abb/group/en
16. FlexSim simulation software, https://www.flexsim.com/
17. Simio simulation software, https://www.simio.com/
18. Anylogic simulation software, https://www.anylogic.com/
19. Anylogistix simulation software, https://www.anylogistix.com/
20. Arena simulation software, https://www.arenasolutions.com/
© The Editor(s) (if applicable) and The Author(s), under exclusive license 229
to Springer Nature Switzerland AG 2024
E. Tokgoz, Six Sigma and Quality Concepts for Industrial Engineers, Synthesis Lectures
on Engineering, Science, and Technology, https://doi.org/10.1007/978-3-031-55740-8
230 References

21. Fanuc robotics, https://www.fanucamerica.com/

22. Mitsubishi robotics, https://mitsubishisolutions.com/industries/industrial-robots/
23. Kuka robotics, https://www.kuka.com/en-us
24. Universal robotics, https://www.universal-robots.com/
25. Tokgoz, E. (2024). Quality and lean six sigma for engineering technicians, synthesis lectures on
engineering, science, and technology, Springer Cham, 978–3-031–44,033-5.
26. Gadekar, R., Sarkar, B., & Gadekar, A. (2022). Key performance indicator based dynamic
decision-making framework for sustainable Industry 4.0 implementation risks evaluation: ref-
erence to the Indian manufacturing industries. Ann Oper Res 318, 189–249. https://doi.org/10.
1007/s10479-022-04828-8.
27. Kumaravadivel, A., & Natarajan, U. (2013). Application of Six-sigma DMAIC methodology to
sand-casting process with response surface methodology. The International Journal of Advanced
Manufacturing Technology, 69(5–8), 1403–1420.
28. Wong, K.C., Woo, K.Z., Woo, K.H. (2016). Ishikawa diagram. In W. O’Donohue, & A. Mara-
gakis (Eds.), Quality improvement in behavioral health. Springer, Cham. https://doi.org/10.
1007/978-3-319-26209-3_9.
29. Verma, N., Sharma, V., & Badar, M. A. (2021). Entropy-based lean, energy and six sigma
approach to achieve sustainability in manufacturing system. Arabian Journal for Science and
Engineering, 46(8), 8105–8117.
30. Basu, R. (2009). Implementing six sigma and lean. Routledge.
31. Issar, G., Navon, L. R., Issar, G., & Navon, L. R. (2016). Human resources (HR). Operational
excellence: a concise guide to basic concepts and their application, 53–55.
32. Govindan, A., & Li, X. (2023). Design and implementation of a fuzzy expert system for
an ergonomic performance assessment in modular construction operations using the DMAIC
approach. In: Walbridge, S., et al. Proceedings of the Canadian Society of Civil Engineering
Annual Conference 2021. CSCE 2021. Lecture Notes in Civil Engineering, vol 251. Springer,
Singapore. https://doi.org/10.1007/978-981-19-1029-6_31.
33. Rathi, R., Khanduja, D., & Sharma, S. K. (2016). Efficacy of fuzzy MADM approach in Six
Sigma analysis phase in automotive sector. Journal of Industrial Engineering International, 12,
377–387.
Index

A D
Advanced material handling, 203 Data collection, 97–99
Advanced technologies, 184, 194–205 Design improvements, 183–185
ANOVA, 166, 167 Discrete distributions, 5–14
App development, 203, 204 Documentation, 187, 188, 224
Artificial intelligence (AI), 202, 203
Augmented reality, 204
E
Automated guided vehicle, 200, 202 Efficiency analysis, 168–171
Automated transportation, 200, 203 Equipment effectiveness, 127, 128
Error, 162–166
Exponential distribution, 21, 22
B
Bar chart, 100, 107–110
Basic statistics, 4 F
B.E. S.M.A.R.T., 47 Finalization, 224, 225
Bias, 119–126, 131 Fish-bone diagram, 136, 137
Binomial distribution, 11–13 5S, 46, 180–183
Box plot, 113, 114 5W + How, 134–137
Business process mapping, 72, 74 Flow chart, 70–73
Functional flow diagram, 72, 73, 75

C
Cause-and-effect diagram, 136 G
C-chart, 220 Gage R&R, 117–126
Gamma distribution, 22–24
Continuous distribution, 2–14
Geometric distribution, 13
Control charts, 207–223
Control mechanisms, 176–180
Correlation, 137–152 H
Critical To Quality (CTQ), 30, 33, 47, 61, 63, Healthcare improvement, 198, 200, 201
67, 68, 77, 80, 81, 87, 126, 127, 139 Histogram, 108–110
CTQ hierarchy tree, 80, 81 Hypergeometric distribution, 14
Cumulative distribution function, 15 Hypothesis testing, 156–166

© The Editor(s) (if applicable) and The Author(s), under exclusive license 231
to Springer Nature Switzerland AG 2024
E. Tokgoz, Six Sigma and Quality Concepts for Industrial Engineers, Synthesis Lectures
on Engineering, Science, and Technology, https://doi.org/10.1007/978-3-031-55740-8
232 Index

I Q
Improving design, 183–185 Quality, 35
Input-Process-Output (IPO), 66, 67 Quality management, 40, 41
Interaction table, 153, 154 Quality tree, 80, 81

K R
Kaizen event, 48–50 Reacting to out-of-control, 208
Kano model, 87, 88 Repeatability, 117–120, 123, 124
KPI analysis, 155, 156 Reproducibility, 117–120, 123
KPI correlation, 137–147 Robots, 194–196

L S
Lean methodology, 45, 46 Sampling, 101–107
Lean techniques, 171 Severity analysis matrix, 167, 168
Linearity, 118, 120, 121, 125 Sigma term, 41–43
Linear model fitting, 125, 137–152 Simulation, 184, 195, 197, 200
Lognormal distribution, 24, 25 Single Minute Exchange of Die (SMED), 188,
189
SIPOC, 66, 68, 69
M
Spaghetti diagram, 77–79
Machine learning, 202, 203
Stability, 122
Measurement matrix, 126, 127
Measuring performance, 127 Staff training, 187, 188
Mission, 50 Surveying, 104–106
Mistake prevention, 175 System improvement, 191, 192

N T
Normal distribution-continuous, 17–19 Takt time, 116, 117
Normal distribution-discrete, 10, 11 Time dependent improvement, 190, 191
Np-chart, 222, 223 Total quality management, 34, 40, 41
Number of operators, 116, 117 Transportation redesign, 184, 186
Type 1 error, 162, 163, 165
Type 2 error, 163, 164
O
Out-of-control points, 222, 223
Overall equipment effectiveness, 127–129 U
U-chart, 221, 222
Uniform distribution, 19, 20
P
Pareto chart, 115
P-chart, 220 V
Pie chart, 111, 112 Value analysis, 30
Poisson distribution, 12, 13 Value Stream Map (VSM), 81–86, 169–171,
Poka Yoke, 189 193, 194
Probability density function, 15 Virtual reality, 204
Process efficiency analysis, 168, 169 Vision, 50
Project charter, 75–77 VSM icons, 82, 84–86
Index 233

VSM improvement phase, 169–171, 193, 194 Weibull distribution, 23, 24

VSM initial phase, 81–86

X
W
Waste types, 36–40 XBar-R chart, 212–214
Waste walk, 31 XBar-s chart, 217