Advanced Chemical Process Control 1St Edition Morten Hovd Full Chapter PDF

Advanced Chemical Process Control
1st Edition Morten Hovd

Visit to download the full and correct content document:
https://ebookmass.com/product/advanced-chemical-process-control-1st-edition-morte
n-hovd/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Chemical Engineering Process Simulation 1st Edition

Nishanth G. Chemmangattuvalappil
https://ebookmass.com/product/chemical-engineering-process-
simulation-1st-edition-nishanth-g-chemmangattuvalappil/
More incidents that define process safety First Edition

American Institute Of Chemical Engineers. Center For
Chemical Process Safety
https://ebookmass.com/product/more-incidents-that-define-process-
safety-first-edition-american-institute-of-chemical-engineers-
center-for-chemical-process-safety/
Process Control 4th Edition Sohrab Rohani (Ed.)
https://ebookmass.com/product/process-control-4th-edition-sohrab-
rohani-ed/
Advanced Control of Power Converters Hasan Komurcugil
https://ebookmass.com/product/advanced-control-of-power-
converters-hasan-komurcugil/
Advanced Data Analysis & Modelling in Chemical
Engineering 1st Edition Denis Constales
https://ebookmass.com/product/advanced-data-analysis-modelling-
in-chemical-engineering-1st-edition-denis-constales/
Human Factors Handbook for Process Plant Operations:

Improving Process Safety and System Performance 1st
Edition Ccps (Center For Chemical Process Safety)
https://ebookmass.com/product/human-factors-handbook-for-process-
plant-operations-improving-process-safety-and-system-
performance-1st-edition-ccps-center-for-chemical-process-safety/
Chemical Engineering Process Simulation 2nd Edition

Dominic C.Y. Foo
https://ebookmass.com/product/chemical-engineering-process-
simulation-2nd-edition-dominic-c-y-foo/
Process Safety for Engineers : An Introduction 2nd

Edition Ccps (Center For Chemical Process Safety)
https://ebookmass.com/product/process-safety-for-engineers-an-
introduction-2nd-edition-ccps-center-for-chemical-process-safety/
Process Control Instrumentation Technology 8th Edition

Curtis D. Johnson
https://ebookmass.com/product/process-control-instrumentation-
technology-8th-edition-curtis-d-johnson/
WILEY END USER LICENSE AGREEMENT
Go to www.wiley.com/go/eula to access Wiley’s ebook EULA.
Putting Theory into Practice
Morten Hovd
Author All books published by WILEY-VCH are carefully
produced. Nevertheless, authors, editors, and
Prof. Morten Hovd publisher do not warrant the information
Norwegian University of Science and contained in these books, including this book,
Technology to be free of errors. Readers are advised to keep
7491 Trondheim in mind that statements, data, illustrations,
Norway procedural details or other items may
inadvertently be inaccurate.
Cover Image: © Voranee/Shutterstock
Library of Congress Card No.: applied for
British Library Cataloguing-in-Publication Data

A catalogue record for this book is available
from the British Library.
Bibliographic information published by

the Deutsche Nationalbibliothek
The Deutsche Nationalbibliothek lists
this publication in the Deutsche
Nationalbibliografie; detailed bibliographic
data are available on the Internet at
<http://dnb.d-nb.de>.
© 2023 WILEY-VCH GmbH, Boschstr. 12,

69469 Weinheim, Germany
All rights reserved (including those of

translation into other languages). No part of
this book may be reproduced in any form – by
photoprinting, microfilm, or any other means –
nor transmitted or translated into a machine
language without written permission from the
publishers. Registered names, trademarks, etc.
used in this book, even when not specifically
marked as such, are not to be considered
unprotected by law.
Print ISBN: 978-3-527-35223-4

ePDF ISBN: 978-3-527-84247-6
ePub ISBN: 978-3-527-84248-3
oBook ISBN: 978-3-527-84249-0
Typesetting Straive, Chennai, India

To Ellen
vii
Contents
Preface xvii
Acknowledgments xxi
Acronyms xxiii
Introduction xxv
1 Mathematical and Control Theory Background 1

1.1 Introduction 1
1.2 Models for Dynamical Systems 1
1.2.1 Dynamical Systems in Continuous Time 1
1.2.2 Dynamical Systems in Discrete Time 2
1.2.3 Linear Models and Linearization 3
1.2.3.1 Linearization at a Given Point 3
1.2.3.2 Linearizing Around a Trajectory 6
1.2.4 Converting Between Continuous- and Discrete-Time Models 6
1.2.4.1 Time Delay in the Manipulated Variables 7
1.2.4.2 Time Delay in the Measurements 9
1.2.5 Laplace Transform 9
1.2.6 The z Transform 10
1.2.7 Similarity Transformations 11
1.2.8 Minimal Representation 11
1.2.9 Scaling 14
1.3 Analyzing Linear Dynamical Systems 15
1.3.1 Transfer Functions of Composite Systems 15
1.3.1.1 Series Interconnection 15
1.3.1.2 Parallel Systems 16
1.3.1.3 Feedback Connection 16
1.3.1.4 Commonly Used Closed-Loop Transfer Functions 17
1.3.1.5 The Push-Through Rule 17
1.4 Poles and Zeros of Transfer Functions 18
1.4.1 Poles of Multivariable Systems 19
1.4.2 Pole Directions 19
1.4.3 Zeros of Multivariable Systems 20
1.4.4 Zero Directions 22
viii Contents
1.5 Stability 23
1.5.1 Poles and Zeros of Discrete-Time Transfer Functions 23
1.5.2 Frequency Analysis 24
1.5.2.1 Steady-State Phase Adjustment 26
1.5.3 Bode Diagrams 27
1.5.3.1 Bode Diagram Asymptotes 27
1.5.3.2 Minimum Phase Systems 29
1.5.3.3 Frequency Analysis for Discrete-Time Systems 30
1.5.4 Assessing Closed-Loop Stability Using the Open-Loop Frequency
Response 31
1.5.4.1 The Principle of the Argument and the Nyquist D-Contour 31
1.5.4.2 The Multivariable Nyquist Theorem 32
1.5.4.3 The Monovariable Nyquist Theorem 32
1.5.4.4 The Bode Stability Criterion 32
1.5.4.5 Some Remarks on Stability Analysis Using the Frequency Response 35
1.5.4.6 The Small Gain Theorem 36
1.5.5 Controllability 37
1.5.6 Observability 38
1.5.7 Some Comments on Controllability and Observability 39
1.5.8 Stabilizability 40
1.5.9 Detectability 40
1.5.10 Hidden Modes 41
1.5.11 Internal Stability 41
1.5.12 Coprime Factorizations 43
1.5.12.1 Inner–Outer Factorization 44
1.5.12.2 Normalized Coprime Factorization 44
1.5.13 Parametrization of All Stabilizing Controllers 44
1.5.13.1 Stable Plants 45
1.5.13.2 Unstable Plants 45
1.5.14 Hankel Norm and Hankel Singular Values 46
Problems 47
References 49
2 Control Conﬁguration and Controller Tuning 51

2.1 Common Control Loop Structures for the Regulatory Control Layer 51
2.1.1 Simple Feedback Loop 51
2.1.2 Feedforward Control 51
2.1.3 Ratio Control 54
2.1.4 Cascade Control 54
2.1.5 Auctioneering Control 55
2.1.6 Split Range Control 56
2.1.7 Input Resetting Control 57
2.1.8 Selective Control 59
2.1.9 Combining Basic Single-Loop Control Structures 60
2.1.10 Decoupling 61
Contents ix
2.2 Input and Output Selection 62

2.2.1 Using Physical Insights 63
2.2.2 Gramian-Based Input and Output Selection 64
2.2.3 Input/Output Selection for Stabilization 65
2.3 Control Configuration 66
2.3.1 The Relative Gain Array 66
2.3.2 The RGA as a General Analysis Tool 68
2.3.2.1 The RGA and Zeros in the Right Half-Plane 68
2.3.2.2 The RGA and the Optimally Scaled Condition Number 68
2.3.2.3 The RGA and Individual Element Uncertainty 69
2.3.2.4 RGA and Diagonal Input Uncertainty 69
2.3.2.5 The RGA as an Interaction Measure 70
2.3.3 The RGA and Stability 70
2.3.3.1 The RGA and Pairing of Controlled and Manipulated Variables 71
2.3.4 Summary of RGA-Based Input–Output Pairing 72
2.3.5 Partial Relative Gains 72
2.3.6 The Niederlinski Index 73
2.3.7 The Rijnsdorp Interaction Measure 73
2.3.8 Gramian-Based Input–Output Pairing 74
2.3.8.1 The Participation Matrix 75
2.3.8.2 The Hankel Interaction Index Array 75
2.3.8.3 Accounting for the Closed-Loop Bandwidth 76
2.4 Tuning of Decentralized Controllers 76
2.4.1 Introduction 76
2.4.2 Loop Shaping Basics 77
2.4.3 Tuning of Single-Loop Controllers 79
2.4.3.1 PID Controller Realizations and Common Modifications 79
2.4.3.2 Controller Tuning Using Frequency Analysis 81
2.4.3.3 Ziegler–Nichols Closed-Loop Tuning Method 86
2.4.3.4 Simple Fitting of a Step Response Model 86
2.4.3.5 Ziegler–Nichols Open-Loop Tuning 88
2.4.3.6 IMC-PID Tuning 88
2.4.3.7 Simple IMC Tuning 89
2.4.3.8 The Setpoint Overshoot Method 91
2.4.3.9 Autotuning 95
2.4.3.10 When Should Derivative Action Be Used? 95
2.4.3.11 Effects of Internal Controller Scaling 96
2.4.3.12 Reverse Acting Controllers 97
2.4.4 Gain Scheduling 97
2.4.5 Surge Attenuating Controllers 98
2.4.6 Multiloop Controller Tuning 99
2.4.6.1 Independent Design 100
2.4.6.2 Sequential Design 102
2.4.6.3 Simultaneous Design 103
2.4.7 Tools for Multivariable Loop-Shaping 103
x Contents
2.4.7.1 The Performance Relative Gain Array 103

2.4.7.2 The Closed-Loop Disturbance Gain 104
2.4.7.3 Illustrating the Use of CLDG’s for Controller Tuning 104
2.4.7.4 Unachievable Loop Gain Requirements 107
Problems 108
References 112
3 Control Structure Selection and Plantwide Control 115

3.1 General Approach and Problem Decomposition 115
3.1.1 Top-Down Analysis 115
3.1.1.1 Defining and Exploring Optimal Operation 115
3.1.1.2 Determining Where to Set the Throughput 116
3.1.2 Bottom-Up Design 116
3.2 Regulatory Control 117
3.2.1 Example: Regulatory Control of Liquid Level in a Deaeration Tower 118
3.3 Determining Degrees of Freedom 121
3.4 Selection of Controlled Variables 122
3.4.1 Problem Formulation 123
3.4.2 Selecting Controlled Variables by Direct Evaluation of Loss 124
3.4.3 Controlled Variable Selection Based on Local Analysis 125
3.4.3.1 The Minimum Singular Value Rule 127
3.4.3.2 Desirable Characteristics of the Controlled Variables 128
3.4.4 An Exact Local Method for Controlled Variable Selection 128
3.4.5 Measurement Combinations as Controlled Variables 130
3.4.5.1 The Nullspace Method for Selecting Controlled Variables 130
3.4.5.2 Extending the Nullspace Method to Account for Implementation
Error 130
3.4.6 The Validity of the Local Analysis for Controlled Variable Selection 131
3.5 Selection of Manipulated Variables 132
3.5.1 Verifying that the Proposed Manipulated Variables Make Acceptable
Control Possible 133
3.5.2 Reviewing the Characteristics of the Proposed Manipulated
Variables 134
3.6 Selection of Measurements 135
3.7 Mass Balance Control and Throughput Manipulation 136
3.7.1 Consistency of Inventory Control 138
Problems 140
References 141
4 Limitations on Achievable Performance 143

4.1 Performance Measures 143
4.1.1 Time-Domain Performance Measures 143
4.1.2 Frequency-Domain Performance Measures 145
4.1.2.1 Bandwidth Frequency 145
4.1.2.2 Peaks of Closed-Loop Transfer Functions 146
Contents xi
4.1.2.3 Bounds on Weighted System Norms 146

4.1.2.4 Gain and Phase Margin 147
4.2 Algebraic Limitations 148
4.2.1 S + T = I 148
4.2.2 Interpolation Constraints 148
4.2.2.1 Monovariable Systems 148
4.2.2.2 Multivariable Systems 149
4.3 Control Performance in Different Frequency Ranges 149
4.3.1 Sensitivity Integrals and Right Half-Plane Zeros 149
4.3.2 Sensitivity Integrals and Right Half-Plane Poles 150
4.3.3 Combined Effects of RHP Poles and Zeros 150
4.3.4 Implications of the Sensitivity Integral Results 150
4.4 Bounds on Closed-Loop Transfer Functions 151
4.4.1 The Maximum Modulus Principle 152
4.4.1.1 The Maximum Modulus Principle 152
4.4.2 Minimum Phase and Stable Versions of the Plant 152
4.4.3 Bounds on S and T 153
4.4.3.1 Monovariable Systems 153
4.4.4 Bounds on KS and KSGd 154
4.5 ISE Optimal Control 156
4.6 Bandwidth and Crossover Frequency Limitations 156
4.6.1 Bounds from ISE Optimal Control 156
4.6.2 Bandwidth Bounds from Weighted Sensitivity Minimization 157
4.6.3 Bound from Negative Phase 158
4.7 Bounds on the Step Response 158
4.8 Bounds for Disturbance Rejection 160
4.8.1 Inputs for Perfect Control 161
4.8.2 Inputs for Acceptable Control 161
4.8.3 Disturbances and RHP Zeros 161
4.8.4 Disturbances and Stabilization 162
4.9 Limitations from Plant Uncertainty 164
4.9.1 Describing Uncertainty 165
4.9.2 Feedforward Control and the Effects of Uncertainty 166
4.9.3 Feedback and the Effects of Uncertainty 167
4.9.4 Bandwidth Limitations from Uncertainty 168
Problems 168
References 170
5 Model-Based Predictive Control 173

5.1 Introduction 173
5.2 Formulation of a QP Problem for MPC 175
5.2.1 Future States as Optimization Variables 179
5.2.2 Using the Model Equation to Substitute for the Plant States 180
xii Contents
5.2.3 Optimizing Deviations from Linear State Feedback 181

5.2.4 Constraints Beyond the End of the Prediction Horizon 182
5.2.5 Finding the Terminal Constraint Set 183
5.2.6 Feasible Region and Prediction Horizon 184
5.3 Step-Response Models 185
5.4 Updating the Process Model 186
5.4.1 Bias Update 186
5.4.2 Kalman Filter and Extended Kalman Filters 187
5.4.2.1 Augmenting a Disturbance Description 188
5.4.2.2 The Extended Kalman Filter 189
5.4.2.3 The Iterated Extended Kalman Filter 189
5.4.3 Unscented Kalman Filter 190
5.4.4 Receding Horizon Estimation 193
5.4.4.1 The Arrival Cost 195
5.4.4.2 The Filtering Formulation of RHE 196
5.4.4.3 The Smoothing Formulation of RHE 196
5.4.5 Concluding Comments on State Estimation 198
5.5 Disturbance Handling and Offset-Free Control 199
5.5.1 Feedforward from Measured Disturbances 199
5.5.2 Requirements for Offset-Free Control 199
5.5.3 Disturbance Estimation and Offset-Free Control 200
5.5.4 Augmenting the Model with Integrators at the Plant Input 203
5.5.5 Augmenting the Model with Integrators at the Plant Output 205
5.5.6 MPC and Integrator Resetting 208
5.6 Feasibility and Constraint Handling 210
5.7 Closed-Loop Stability with MPC Controllers 212
5.8 Target Calculation 213
5.9 Speeding up MPC Calculations 217
5.9.1 Warm-Starting the Optimization 218
5.9.2 Input Blocking 219
5.9.3 Enlarging the Terminal Region 220
5.10 Robustness of MPC Controllers 222
5.11 Using Rigorous Process Models in MPC 225
5.12 Misconceptions, Clarifications, and Challenges 226
5.12.1 Misconceptions 226
5.12.1.1 MPC Is Not Good for Performance 226
5.12.1.2 MPC Requires Very Accurate Models 227
5.12.1.3 MPC Cannot Prioritize Input Usage or Constraint Violations 227
5.12.2 Challenges 227
5.12.2.1 Obtaining a Plant Model 228
5.12.2.2 Maintenance 228
5.12.2.3 Capturing the Desired Behavior in the MPC Design 228
Problems 228
References 231
Contents xiii
6 Some Practical Issues in Controller Implementation 233

6.1 Discrete-Time Implementation 233
6.1.1 Aliasing 233
6.1.2 Sampling Interval 233
6.1.3 Execution Order 235
6.2 Pure Integrators in Parallel 235
6.3 Anti-Windup 236
6.3.1 Simple PI Control Anti-Windup 237
6.3.2 Velocity Form of PI Controllers 237
6.3.3 Anti-Windup in Cascaded Control Systems 238
6.3.4 A General Anti-Windup Formulation 239
6.3.5 Hanus’ Self-Conditioned Form 240
6.3.6 Anti-Windup in Observer-Based Controllers 241
6.3.7 Decoupling and Input Constraints 243
6.3.8 Anti-Windup for “Normally Closed” Controllers 244
6.4 Bumpless Transfer 245
6.4.1 Switching Between Manual and Automatic Operation 245
6.4.2 Changing Controller Parameters 246
Problems 246
References 247
7 Controller Performance Monitoring and Diagnosis 249

7.1 Introduction 249
7.2 Detection of Oscillating Control Loops 251
7.2.1 The Autocorrelation Function 251
7.2.2 The Power Spectrum 252
7.2.3 The Method of Miao and Seborg 252
7.2.4 The Method of Hägglund 253
7.2.5 The Regularity Index 254
7.2.6 The Method of Forsman and Stattin 255
7.2.7 Prefiltering Data 255
7.3 Oscillation Diagnosis 256
7.3.1 Manual Oscillation Diagnosis 256
7.3.2 Detecting and Diagnosing Valve Stiction 257
7.3.2.1 Using the Cross-Correlation Function to Detect Valve Stiction 257
7.3.2.2 Histograms for Detecting Valve Stiction 258
7.3.2.3 Stiction Detection Using an OP–PV Plot 260
7.3.3 Stiction Compensation 262
7.3.4 Detection of Backlash 263
7.3.5 Backlash Compensation 264
7.3.6 Simultaneous Stiction and Backlash Detection 265
7.3.7 Discriminating Between External and Internally Generated
Oscillations 266
7.3.8 Detecting and Diagnosing Other Nonlinearities 266
xiv Contents
7.4 Plantwide Oscillations 269

7.4.1 Grouping Oscillating Variables 269
7.4.1.1 Spectral Principal Component Analysis 269
7.4.1.2 Visual Inspection Using High-Density Plots 269
7.4.1.3 Power Spectral Correlation Maps 270
7.4.1.4 The Spectral Envelope Method 271
7.4.1.5 Methods Based on Adaptive Data Analysis 272
7.4.2 Locating the Cause for Distributed Oscillations 273
7.4.2.1 Using Nonlinearity for Root Cause Location 273
7.4.2.2 The Oscillation Contribution Index 273
7.4.2.3 Estimating the Propagation Path for Disturbances 274
7.5 Control Loop Performance Monitoring 278
7.5.1 The Harris Index 278
7.5.2 Obtaining the Impulse Response Model 279
7.5.3 Calculating the Harris Index 280
7.5.4 Estimating the Deadtime 281
7.5.5 Modifications to the Harris Index 282
7.5.6 Assessing Feedforward Control 283
7.5.7 Comments on the Use of the Harris Index 285
7.5.8 Performance Monitoring for PI Controllers 286
7.6 Multivariable Control Performance Monitoring 287
7.6.1 Assessing Feedforward Control in Multivariable Control 287
7.6.2 Performance Monitoring for MPC Controllers 288
7.7 Some Issues in the Implementation of Control Performance
Monitoring 290
7.8 Discussion 290
Problems 291
References 291
8 Economic Control Beneﬁt Assessment 297

8.1 Optimal Operation and Operational Constraints 297
8.2 Economic Performance Functions 298
8.3 Expected Economic Benefit 299
8.4 Estimating Achievable Variance Reduction 300
8.5 Worst-Case Backoff Calculation 300
References 301
A Fourier–Motzkin Elimination 303
B Removal of Redundant Constraints 307

Reference 308
C The Singular Value Decomposition 309

Contents xv
D Factorization of Transfer Functions into Minimum Phase Stable and

All-Pass Parts 311
D.1 Input Factorization of RHP Zeros 312
D.2 Output Factorization of RHP Zeros 312
D.3 Output Factorization of RHP Poles 313
D.4 Input Factorization of RHP Poles 313
D.5 SISO Systems 314
D.6 Factoring Out Both RHP Poles and RHP Zeros 314
Reference 314
E Models Used in Examples 315

E.1 Binary Distillation Column Model 315
E.2 Fluid Catalytic Cracker Model 318
References 320
Index 321
xvii
Preface
Half a century ago, Alan Foss [1] wrote his influential paper about the gap between
chemical process control theory and industrial application. Foss clearly put the
responsibility on the chemical process control theorists to close this gap. Since then,
several advances in control theory, some originating within academia, while others
originated in industry and was later adopted and further developed by academia,
have contributed to addressing the shortcomings of chemical process control theory,
as addressed by Foss:
● The extension of the relative gain array (RGA) to nonzero frequencies and
Graminan-based control structure selection tools have extended the toolkit for
designing control structures.1 Self-optimal control [4] provides a systematic
approach to identifying controlled variables for a chemical plant.
● Model predictive control (MPC) has great industrial success [3].
● Robustness to model error received significant research focus from the 1980s
onward.
● Control Performance Monitoring has, since the seminal paper by Harris [2],
resulted in tools both for monitoring and diagnosing the performance of

individual loops as well as for identifying the cause of plantwide disturbances.
The aforementioned list notwithstanding, many would agree to the claim that
there is still a wide gap between control theory and common industrial practice in
the chemical process industries. This book is the author’s attempt to contributing to
the reduction of that gap. This book has two ambitious objectives:
1. To make more advanced control accessible to chemical engineers, many of whom
will only have background from a single course in control. While this book is
unlikely to effortlessly turn a plant engineer into a control expert, it does contain
tools that many plant engineers should find useful. It is also hoped that it will give
the plant engineer more weight when discussing with control consultants – either
from within the company or external consultants.
1 This author is aware that many colleagues in academia are of the opinion that the
frequency-dependent RGA is not “rigorous.” This book will instead take the stance that the
dynamic RGA has proved useful and therefore should be disseminated. It is also noted that the
“counterexamples” where the steady-state RGA is claimed to propose a wrong control structure are
themselves flawed.
xviii Preface
2. To increase the understanding among control engineers (students or graduates

moving into the chemical process control area) of how to apply their theoretical
knowledge to practical problems in chemical processes.
The third approach to reducing the gap, i.e. to develop and present tools to sim-
plify the application of control, is not a focus of this book – although some colleagues
would surely claim that this is what proportional integral derivative (PI(D)) tuning
rules are doing.
The reader should note that this book does not start “from scratch,” and some
prior knowledge of control is expected. The introduction to the Laplace transform
is rudimentary at best, and much more detail could be included in the presenta-
tion of frequency analysis. Some knowledge of finite-dimensional linear algebra is
expected. Readers who have never seen a linear state-space model will face a hur-
dle. Still, the book should be accessible to readers with background from a course in
process control.
Readers with a more extensive knowledge of control theory may find the book
lacks rigor. Frequently, results are presented and discussed, without presenting
formal proofs. Readers interested in mathematical proofs will have to consult the
references. This is in line with this author’s intent to keep the focus on issues of
importance for industrial applications (without claiming to “know it all”).
The structure of the Book

This book has grown out of more than three decades of learning, teaching, and dis-
cussing the control of chemical processes. What has become clear is that process
control engineers are faced with a wide variety of tasks and problems. The chapters
of this book therefore address a range of different topics – most of which have been
the subject of entire books. The selection of material to include is therefore not trivial
nor obvious.
● Chapter 1 presents some mathematical and control theory background. Readers
with some knowledge of control may choose to skip this chapter and only return
to it to look up unfamiliar (or forgotten) concepts that appear in the rest of the
book. This chapter definitely has a more theoretical and less practical flavor that
much of the rest of the book.
● Chapter 2 addresses controller tuning for PI(D) controllers, as well as control
configuration. The term control configuration here covers both the control func-
tionality often encountered in the regulatory control layer (feedback, feedforward,
ratio control,...) and determining which input should be used to control which
output in a multi-loop control system.
● Chapter 3 focuses on determining what variables to use for control. Typically,
there are more variables that can be measured than can be manipulated, so the
most focus is given to the choice of variables to control.
● Chapter 4 presents limitations to achievable control performance. Clearly, if it
is not possible to achieve acceptable performance, it makes little sense trying to
design a controller. Understanding the limitations of achievable performance is
Preface xix
also very useful when designing controllers using loop shaping, as presented in
Chapter 2.
● Chapter 5 is about MPC, which is the most popular advanced control type in
the chemical process industries.2 In addition to presenting MPC problem formu-
lations per se, important issues such as model updating, offset-free control, and
target calculation are also discussed.
● Chapter 6 presents some practical issues in controller implementation. This list
is far from complete, some of the issues included are well known and may be
considered trivial – but all are important.
● Chapter 7 addresses control performance monitoring (CPM). The number
of controllers in even a modestly sized chemical process is too large for plant
personnel to frequently monitor each of them, and automated tools are therefore
needed for the maintenance of the control. The chapter also includes some tools
for finding the root cause of distributed oscillations – oscillations originating at
one location in the plant will tend to propagate and disturb other parts of the
plant, and hence it is often nontrivial to find where the oscillation originates and
what remedial action to take.
● Chapter 8 addresses control benefit analysis, i.e. tools to a priori assess the eco-
nomic benefit that can be obtained from improved control. This author admits
that the chapter is disappointingly short. Developing generic tools to estimate such
economic benefit is indeed difficult. On the other hand, the inability to estimate
economic benefit with some certainty is also one of the major obstacles to more
pervasive application of advanced control in the chemical process industries.
What’s Not in the Book

Selecting what to cover in a textbook invariably requires leaving out interesting top-
ics. Some topics that are relevant in a more general setting, but which this book does
not make any attempt to cover, include
● Nonlinear control. Real-life systems are nonlinear. Nevertheless, this book almost
exclusively addresses linear control – with the main exception being the handling
of constraints in MPC.3 Linearization and linear control design suffices for most
control problems in the chemical process industries, and in other cases static non-
linear transforms of inputs or outputs can make more strongly nonlinear systems
closer to linear. The book also describes briefly approaches to nonlinear MPC
(with nonlinearity in the model, not only from constraints). Such complications
are needed mainly for control of batch processes (where continuous changes in the
operating point are unavoidable), or for processes frequently switching between
different operating regimes (such as wastewater treatment plants with anaerobic
and aerobic stages). Although linear control often suffices, it is clearly prudent to
verify a control design with nonlinear simulation and/or investigate the control at
different linearization points (if a nonlinear model is available).
2 And, indeed, the chemical process industries is where MPC was first developed.
3 Constraints represent a strong nonlinearity.
xx Preface
● Modeling and system identification. The availability of good system models are
of great importance to control design and verification. This book only briefly
describes how to fit a particularly simple monovariable model – more complete
coverage of these topics is beyond the scope of this book.
● Adaptive and learning control. While classical adaptive control seems to have been
out of favor in the chemical process industries for several decades, there is cur-
rently a lot of research on integrating machine learning with advanced control
such as MPC. This author definitely accepts the relevance of research on learn-
ing control but is of the opinion that at this stage a research monograph would be
more appropriate than a textbook for covering these developments.
In leaving out many of the more theoretically complex areas of control theory,
readers from a control engineering background may find the book title somewhat
puzzling – especially the word Advanced. Although some of the topics covered
by this book are relatively standard also within the domain of chemical process
control, this author would claim that much of the book covers topics that are
indeed advanced compared to common application practice in the chemical process
industries. The Introduction will hopefully enlighten readers from outside chemical
process control about the unique challenges faced by this application domain and
contribute to explaining the prevalence of linear control techniques.
August, 2022 M. Hovd

Trondheim, Norway
References
1 Foss, A.S. (1973). Critique of chemical process control theory. AIChE Journal
19: 209–214.
2 Harris, T.J. (1989). Assessment of control loop performance. The Canadian Journal
of Chemical Engineering 67: 856–861.
3 Qin, S.J. and Badgwell, T.A. (2003). A survey of industrial model predictive con-
trol technology. Control Engineering Practice 11 (7): 733–764.
4 Skogestad, S. (2000). Plantwide control: the search for the self-optimizing control
structure. Journal of Process Control 10: 487–507.
xxi
Acknowledgments
It is noted in the preface that this book is a result of more than three decades of
learning, teaching, discussing, and practicing chemical process control. Therefore,
a large number of people have directly or indirectly influenced the contents of this
book, and it is impossible to mention them all. The person with the strongest such
influence is Professor Sigurd Skogestad at NTNU, whom I had the fortune to have
as my PhD supervisor. At that time, I had an extended research stay in the group
of Professor Manfred Morari (then at Caltech). Discussions these outstanding pro-
fessors as well as with my contemporaries as a PhD student, both in Skogestad’s
group and in Morari’s group, greatly enhanced my understanding of process con-
trol. I would particularly like to mention Elling W. Jacobsen, Petter Lundström, John
Morud, Richard D. Braatz, Jay H. Lee, and Frank Doyle.
In later years, I have benefited from discussions with a number of people, includ-
ing Kjetil Havre, Vinay Kariwala, Ivar Halvorsen, Krister Forsman, Alf Isaksson,
Tor A. Johansen, Lars Imsland, and Bjarne A. Foss. My own PhD candidates and
Postdocs have also enriched my understanding of the area, in particular, Kristin
Hestetun, Francesco Scibilia, Giancarlo Marafioti, Florin Stoican, Selvanathan
Sivalingam, Muhammad Faisal Aftab, and Jonatan Klemets.
Special thanks are due to my hosts during sabbaticals, when I have learned a
lot: Bob Bitmead (UCSD), Jan Maciejowski (University of Cambridge), Sorin Olaru
(CentraleSupelec), and Andreas Kugi (TU Vienna).
M. H.
xxiii
Acronyms
AKF augmented Kalman filter

AR auto-regressive (model)
BLT biggest log modulus (tuning rule)
CCM convergent cross mapping
CLDG closed-loop disturbance gain
CPM control performance monitoring
DNL degree of nonlinearity
EKF extended Kalman filter
EnKF ensemble Kalman filter
FCC fluid catalytic cracking
FOPDT first-order plus deadtime (model)
GM gain margin
IAE integral (of) absolute error
IEKF iterated extended Kalman filter
IMC internal model control
ISE integral (of) squared error
KISS keep it simple, stupid
LHP left half-plane (of the complex plane)
LP linear programming
LQ linear quadratic
LQG linear quadratic Gaussian
MHE moving horizon estimation
MIMO multiple input multiple output
MPC model predictive control
MV manipulated variable
MVC minimum variance controller
MWB moving window blocking
NGI non-Gaussianity index
NLI nonlinearity index
OCI oscillation contribution index
OP controller output
PCA principal component analysis
PID proportional integral derivative
xxiv Acronyms
PSCI power spectral correlation index

P&ID piping and instrument diagram
PM phase margin
PRGA performance relative gain array
PV process variable
QP quadratic programming
RGA relative gain array
RHE receding horizon estimation
RHP right half-plane (of the complex plane)
RHPZ right half-plane zero
RSS residual sum of squares
RTO real time optimization
SIMC simple (or Skogestad) internal model control (tuning rules)
SISO single input single output
SP setpoint (reference value)
SOPDT second-order plus deadtime (model)
SVD singular value decomposition
UKF unscented Kalman filter
xxv
Introduction
I.1 Why Is Chemical Process Control Needed?
Many texts on control implicitly assume that it is obvious when and why control is
needed. It seems obvious that even a moderately complex process plant will be very
difficult to operate without the aid of process control. Nevertheless, it can be worth-
while to spend a few minutes thought on why control is needed. In the following, a
short and probably incomplete list of reasons for the need of control is provided, but
the list should illustrate the importance of control in a chemical process plant.
1. Stabilizing the process. Many processes have integrating or unstable modes. These
have to be stabilized by feedback control, otherwise the plant will (sooner or later)
drift into unacceptable operating conditions. In the vast majority of cases, this
stabilization is provided by automatic feedback control.1 Note that in practice,
“feedback stabilization” of some process variable may be necessary even though
the variable in question is asymptotically stable according to the control engineer-
ing definition of stability. This happens whenever disturbances have sufficiently
large effect on a process variable to cause unacceptably large variations in the pro-
cess variable value. Plant operators therefore often use the term “stability” in a
much less exact way than how the term is defined in control engineering. A con-
trol engineer may very well be told that, e.g. “this temperature is not sufficiently
stable,” even though the temperature in question is asymptotically stable.
2. Regularity. Even if a process is stable, control is needed to avoid shutdowns due to
unacceptable operating conditions. Such shutdowns may be initiated automati-
cally by a shutdown system but may also be caused by outright equipment failure.
3. Minimizing effects on the environment. In addition to maintaining safe and stable
production, the control system should also ensure that any harmful effects on
the environment are minimized. This is done by optimizing the conversion of
raw materials,2 and by maintaining conditions which minimize the production
of any harmful byproducts.
1 However, some industries still use very large buffer tanks between different sections in the
process. For such tanks, it may be sufficient with infrequent operator intervention to stop the
buffer tank from overfilling or emptying.
2 Optimizing the conversion of raw materials usually means maximizing the conversion, unless
this causes unacceptably high production of undesired byproducts or requires large energy inputs.
xxvi Introduction
4. Obtaining the right product quality. Control is often needed both for achieving the
right product quality and for reducing quality variations.
5. Achieving the right production rate. Control is used for achieving the right produc-
tion rate in a plant. Ideally, it should be possible to adjust the production rate at
one point in the process, and the control system should automatically adjust the
throughput of up- or downstream units accordingly.
6. Optimize process operation. When a process achieves safe and stable operation,
with little downtime and produces the right quality of product at the desired
production rate, the next task is to optimize the production. The objective of
the optimization is normally to achieve the most cost-effective production. This
involves identifying, tracking, and maintaining the optimal operating conditions
in the face of disturbances in production rate, raw material composition, and
ambient conditions(e.g. atmospheric temperature). Process optimization often
involves close coordination of several process units and operation close to process
constraints.
The list aforementioned should illustrate that process control is vital for the oper-
ation of chemical process plants. Even plants of quite moderate complexity would
be virtually impossible to operate without control. Even where totally manual oper-
ation is physically feasible, it is unlikely to be economically feasible due to product
quality variations and high personnel costs, since a high number of operators will be
required to perform the many (often tedious) tasks that the process control system
normally handles.
Usually many more variables are controlled than what is directly implied by the
list above, and there are often control loops for variables which have no specifica-
tion associated with them. There are often good reasons for such control loops – two
possible reasons are:
1. To stop disturbances from propagating downstream. Even when there are no direct
specification on a process variable, variations in the process variable may cause
variations in more important variables downstream. In such cases, it makes sense
to remove the disturbance at its source.
2. Local removal of uncertainty. By measuring and controlling a process variable,
it may be possible to reduce the effect of uncertainty with respect to equipment
behavior or disturbances. Examples of such control loops are valve positioners
used to minimize the uncertainty in the valve opening, or local flow control
loops which may be used to counteract the effects of pressure disturbances up- or
downstream of a valve, changes in fluid properties, or inaccuracies in the valve
characteristics.
I.2 What Knowledge Does a chemical Process Control

Engineer need?
The aforementioned list of reasons for why process control is needed also indicates
what kind of knowledge is required for a process control engineer. The process
Introduction xxvii
control engineer needs to have a thorough understanding of the process. Most

stabilizing control loops involve only one process unit (e.g. a tank or a reactor),
and most equipment limitations are also determined by the individual units.
Process understanding on the scale of the individual units is therefore required.
Understanding what phenomena affect product quality also require an under-
standing of the individual process units. On the other hand, ensuring that the
specified production rate propagates throughout the plant, understanding how the
effects of disturbances propagate, and optimizing the process operation require an
understanding of how the different process units interact, i.e. an understanding of
the process on a larger scale.
Most basic control functions are performed by single loops, i.e. control loops with
one controlled variable and one manipulated variable. Thus, when it is understood
why a particular process variable needs to be controlled, and what manipulated vari-
able should be used to control it, 3 the controller design itself can be performed using
traditional single-loop control theory (if any theoretical considerations are made at
all). Often a standard type of controller, such as a proportional integral derivative
(PID) controller, is tuned online, and there is little need for a process model. Other
control tasks are multivariable in nature, either because it is necessary to resolve
interactions between different control loops or because the control task requires
coordination between different process units. Process models are often very useful
for these types of control problem. The models may either be linear models obtained
from experiments on the plant or possibly nonlinear models derived from physical
and chemical principles. Some understanding of mathematical modeling and system
identification techniques are then required. Nonlinear system identification from
plant experiments are not in standard use in the process industries.
Optimizing process operation requires some understanding of plant economics,
involving the costs of raw materials and utilities, the effect of product quality on
product price, the cost of reprocessing off-spec product, etc. Although it is rare
that economics is optimized by feedback controllers, 4 an understanding of plant
economics will help understanding where efforts to improve control should be
focused and will help when discussing the need for improved control with plant
management.
A process control engineer must thus have knowledge of both process and con-
trol engineering. However, it is not reasonable to expect the same level of expertise
in either of these disciplines from the process control engineer as for “specialist”
process or control engineers. There appears to be a “cultural gap” between process
and control engineers, and the process control engineer should attempt to bridge
3 Determining what variables are to be controlled, what manipulated variables should be used for
control, and the structure of interconnections between manipulated and controlled variables are
quite critical tasks in the design of a process control system. This part of the controller design is
often not described in textbooks on “pure” control engineering but will be covered in some detail
in later sections.
4 It is more common that economic criteria are used in the problem formulation for so-called real
time optimization (RTO) problems, or for plant production planning and scheduling, as shown in
Figure I.1.
xxviii Introduction
this gap. This means that the process control engineer should be able to commu-
nicate meaningfully with both process and control engineers and thereby also be
able to obtain any missing knowledge by discussing with the “specialists.” However,
at a production plant, there will seldom be specialists in control theory, but there
will always be process engineers. At best, large companies may have control the-
ory specialists at some central research or engineering division. This indicates that a
process control engineer should have a fairly comprehensive background in control
engineering, while the process engineering background should at least be sufficient
to communicate effectively with the process engineers.
In the same way as for other branches of engineering, success at work will not
come from technological competence alone. A successful engineer will need the
ability to work effectively in multidisciplinary project teams, as well as skills in com-
municating with management and operators. Such nontechnical issues will not be
discussed further here.
I.3 What Makes Chemical Process Control Unique?
Control engineering provides an extensive toolbox that can be applied to a very wide
range of application domains. Still, process control is characterized by a few features,
the combination of which make process control uniquely challenging:
● Physical scale. Although small-scale industrial plants do exist, more common are
large-scale industrial production plants, with hundreds of meters or even kilo-
meters from one end to the other, and the need for some degree of coordination
between different sections of the plant.
● One-of-a-kind plants. Most plants are unique. Although many of the individual
components may be standard issue, their assembly into the overall plant is typi-
cally unique to the plant in question. Even in the rather rare cases when plants are
built to be nominally identical, differences in external disturbances, raw materials,
and operating and maintenance practices will mean that differences accumulate
over time. The sheer scale of industrial plants also mean that the number of ways
in which plant behavior may differ is very large. This differs from, say, automo-
biles or airplanes, where the aim is to produce a large number of identical units.
From a control perspective, the main consequence for chemical process control is
that the cost of modeling and model maintenance must be borne by each individual
plant. Model-based control design or plant operation will therefore often have to
be based on rather simple and inaccurate models.
● Plant experimentation. Another aspect of the point above is that in order to learn
plant behavior or verify control designs, experiments often have to be performed
on the plant itself. Experiments disrupt plant operation and can also involve the
risk of damage or accident. Again this differs from control applications in mass
produced items, where damage to a few items during testing is often expected.
● Lots of data, little information. In many chemical process plants, fast and reliable
measurement of key variables are not available. In some processes, measurements
Introduction xxix
of temperatures, pressures, and flows are readily available, while composition

measurements require either costly online analyzers with long time delays or
require manual sampling and laboratory analysis. In other cases, harsh process
conditions means that not even pressures and temperatures can be measured,
and control must be based on measurements obtained externally to the process
itself (such as currents and voltages in electrochemical plants). Again, this is
in contrast to other application areas of control, such as motion control where
positions and velocities often can be measured fairly accurately.
I.4 The Structure of Control Systems in the Chemical

Process Industries
When studying control systems in the chemical process industries, one may observe
that they often share a common structure. This structure is illustrated in Figure I.1.
The lower level in the control system is the regulatory control layer. The structure
of the individual controllers in the regulatory control layer is normally very simple.
Standard single-loop controllers, typically of proportional integral (PI)/PID type, are
the most common, but other simple control functions like feedforward control, ratio
control, or cascaded control loops may also be found. Truly multivariable controllers
are rare at this level. The regulatory control system typically controls basic process
variables such as temperatures, pressures, flowrates, speeds, or concentrations, but
in some cases the controlled variable may be calculated based on several measure-
ments, e.g. a component flowrate based on measurements of both concentration and
overall flowrate or a ratio of two flowrates. Usually, a controller in the regulatory
control layer manipulates a process variable directly (e.g. a valve opening), but in
some cases the manipulated variable may be a setpoint of a cascaded control loop.
Most control functions that are essential to the stability and integrity of the process
are executed in this layer, such as stabilizing the process and maintaining acceptable
equipment operating conditions.
The supervisory control layer coordinates the control of a process unit or a few
closely connected process units. It coordinates the action of several control loops
and tries to maintain the process conditions close to the optimal while ensuring that
operating constraints are not violated. The variables that are controlled by super-
visory controllers may be process measurements, variables calculated or estimated
from process measurements, or the output from a regulatory controller. The manip-
ulated variables are often setpoints to regulatory controllers, but process variables
may also be manipulated directly. Whereas regulatory controllers are often designed
and implemented without ever formulating any process model explicitly, supervi-
sory controllers usually contain an explicitly formulated process model. The model
is dynamic and often linear and obtained from experiments on the plant. Typically,
supervisory controllers use some variant of model predictive control (MPC).
The optimal conditions that the supervisory controllers try to maintain may be
calculated by a RTO control layer. The RTO layer identifies the optimal conditions
by solving an optimization problem involving models of the production cost, value of
xxx Introduction
Production planning/
scheduling
Real time optimization
Supervisory control
Regulatory control
To manipulated variables From measurements
Process
Figure I.1 Typical structure of the control system for a large plant in the process industries.
product (possibly dependent on quality), and the process itself. The process model is
often nonlinear and derived from fundamental physical and chemical relationships,
but they are usually static.
The higher control layer shown in Figure I.1 is the production planning and
scheduling layer. This layer determines what products should be produced and
when they should be produced. This layer requires information from the sales
department about the quantities of the different products that should be produced,
Introduction xxxi
the deadlines for delivery, and possibly product prices. From the purchasing depart-
ment, information about the availability and price of raw materials are obtained.
Information from the plant describes what products can be made in the different
operating modes and what production rates can be achieved.
In addition to the layers in Figure I.1, there should also be a separate safety system
that will shut the process down in a safe and controlled manner when potentially
dangerous conditions occur. There are also higher levels of decision-making which
are not shown, such as sales and purchasing, and construction of new plants. These
levels are considered to be of little relevance to process control and will not be dis-
cussed further.
Note that there is a difference in timescale of execution for the different layers.
The regulatory control system typically have sampling intervals on the scale of 1
second (or faster for some types of equipment), supervisory controllers usually
operate on the timescale of minutes, the RTO layer on a scale of hours, and the
planning/scheduling layer on a scale of days (or weeks). The control bandwidths
achieved by the different layers differ in the same way as sampling intervals differ.
This difference in control bandwidths can simplify the required modeling in the
higher levels; if a variable is controlled by the regulatory control layer, and the
bandwidth for the control loop is well beyond what is achieved in the supervisory
control layer, a static model for this variable (usually the model would simply be
variable value = setpoint) will often suffice for the supervisory control.
It is not meaningful to say that one layer is more important than another, since
they are interdependent. The objective of the lower layers are not well defined with-
out information from the higher layers (e.g. the regulatory control layer needs to
know the setpoints that are determined by the supervisory control layer), whereas
the higher layers need the lower layers to implement the control actions. However,
in many plants, human operators perform the tasks of some the layers as shown
in Figure I.1, and it is only the regulatory control layer that is present (and highly
automated) in virtually all industrial plants.
Why has this multilayered structure for industrial control systems evolved? It is clear
that this structure imposes limitations in achievable control performance compared
to a hypothetical optimal centralized controller which perfectly coordinates all avail-
able manipulated variables in order to achieve the control objectives. In the past,
the lack of computing power would have made such a centralized controller vir-
tually impossible to implement, but the continued increase in available computing
power could make such a controller feasible in the not too distant future. Is this the
direction industrial control systems are heading? This appears not to be the case.
In the last two decades, development has instead moved in the opposite direction,
as increased availability of computing power has made the supervisory control and
RTO layers much more common. Some reasons for using such a multilayered struc-
ture are:
● Economics. Optimal control performance – defined in normal control engineering

terms (using, e.g. the H2 - or H∞ norm) – does not necessarily imply optimal
economic performance. To be more specific, an optimal controller synthesis
xxxii Introduction
problem does not take into account the cost of developing and maintaining the
required process (or possibly plant economic) models. An optimal centralized
controller would require a dynamic model of most aspects of the process behavior.
The required model would therefore be quite complex and difficult to develop
and maintain. In contrast, the higher layers in a structured control system can
take advantage of the model simplifications made possible by the presence of
the lower layers. The regulatory control level needs little model information to
operate, since it derives most process information from feedback from process
measurements.5
● Redesign and retuning. The behavior of a process plant changes with time, for a
number of reasons such as equipment wear, changes in raw materials, changes
in operating conditions in order to change product qualities or what products are
produced, and plant modifications. Due to the sheer complexity of a centralized
controller, it would be difficult and time-consuming to update the controller to
account for all such changes. With a structured control system, it is easier to see
what modifications need to be made, and the modifications themselves will nor-
mally be less involved.
● Start-up and shutdown. Common operating practice during start-up is that many
of the controls are put in manual. Parts of the regulatory control layer may be in
automatic, but rarely will any higher layer controls be in operation. The loops of
the regulatory control layer that are initially in manual are put in automatic when
the equipment that they control are approaching normal operating conditions.
When the regulatory control layer for a process section is in service, the supervi-
sory control system may be put in operation, and so on. Shutdown is performed
in the reverse sequence. Thus, there may be scope for significant improvement
of the start-up and shutdown procedures of a plant, as quicker start-up and shut-
down can reduce plant downtime. However, a model, which in addition to normal
operating conditions, is able to describe start-up and shutdown and is necessarily
much more complex than a model which covers only the range of conditions that
are encountered in normal operation. Building such a model would be difficult
and costly. Start-up and shutdown of a plant with an optimal centralized control
system which does not cover start-up and shutdown may well be more difficult
than with a traditional control system, because it may not be difficult to put an
optimal control system gradually into or out of service.
● Operator acceptance and understanding. Control systems that are not accepted by
the operators are likely to be taken out of service. An optimal centralized control
system will often be complex and difficult to understand. Operator understanding
obviously makes acceptance easier, and a traditional control system, being easier
to understand, often has an advantage in this respect. Plant shutdowns may be
caused by operators with insufficient understanding of the control system. Such
shutdowns should actually be blamed on the control system (or the people who
designed and installed the control system), since operators are an integral part of
5 A good process model may be of good use when designing control structuresfor regulatory
control. However, after the regulatory controllers are implemented, they normally do not make any
explicit use of a process model.
Introduction xxxiii
the plant operation, and their understanding of the control system must therefore
be ensured.
● Failure of computer hardware and software. In traditional control systems, the
operators retain the help of the regulatory control system in keeping the process
in operation if a hardware or software failure occurs in higher levels of the con-
trol system. A hardware backup for the regulatory control system is much cheaper
than for the higher levels in the control system, as the regulatory control system
can be decomposed into simple control tasks (mainly single loops). In contrast,
an optimal centralized controller would require a powerful computer, and it is
therefore more costly to provide a backup system. However, with the continued
decrease in computer cost, this argument may weaken.
● Robustness .The complexity of an optimal centralized control system will make it
difficult to analyze whether the system is robust with respect to model uncertainty
and numerical inaccuracies. Analyzing robustness need not be trivial even for tra-
ditional control systems. The ultimate test of robustness will be in the operation of
the plant. A traditional control system may be applied gradually, first the regula-
tory control system, then section by section of the supervisory control system, etc.
When problem arise, it will therefore be easier to analyze the cause of the problem
with a traditional control system than with a centralized control system.
● Local removal of uncertainty. It has been noted earlier that one effect of the lower
layer control functions is to remove model uncertainty as seen from the higher lay-
ers. Thus, the existence of the lower layers allow for simpler models in the higher
layers and make the models more accurate. The more complex computations in
the higher layers are therefore performed by simpler, yet more accurate models. A
centralized control system will not have this advantage.
● Existing traditional control systems . Where existing control systems perform rea-
sonably well, it makes sense to put effort into improving the existing system rather
than to take the risky decision to design a new control system. This argument
applies also to many new plants, as many chemical processes are not well under-
stood. For such processes, it will therefore be necessary to carry out model identi-
fication and validation on the actual process. During this period, some minimum
amount of control will be needed. The regulatory control layer of a traditional
control system requires little information about the process and can therefore be
in operation in this period.
It should be clear from the above that this author believes that control systems in
the future will continue to have a number of distinct layers. Two prerequisites appear
to be necessary for a traditional control system to be replaced with a centralized one:
1. The traditional control system must give unacceptable performance.

2. The process must be sufficiently well understood to be able to develop a process
model which describes all relevant process behavior.
Since it is quite rare that a traditional control system is unable to control a chemi-
cal process for which detailed process understanding is available (provided sufficient
effort and expertise have been put into the design of the control system), it should
xxxiv Introduction
follow that majority of control systems will continue to be of the traditional struc-
tured type.
In short, the layered control system is consistent with the common approach
of breaking down big problems into smaller, more manageable parts, and as such
agrees with the keep it simple, stupid (KISS) principle.
I.5 Notation
x Vector of system states.

ẋ The time derivative of the state vector (for continuous-time systems).
u The vector of manipulated variables (the variables manipulated by the con-
trol system to control the plant), sometimes also referred to as inputs. In some
literature, the vector u is also called the control variables.
y The controlled variables (the variables that the control system attempts to con-
trol). Often, the vector y is also identical to the vector of measured variables.
d The vector of disturbance variables.
vk The vector v at timestep k (for discrete-time systems).
vi , vj Elements i and j of the vector v.
M Capital letters are used for matrices.
Mij Element (i, j) of the matrix M.
For the linear(ized) system in continuous time:
ẋ = Ax + Bu + Ed
y = Cx + Du + Fd
A, B, C, D, E, F are matrices of appropriate dimension, and
G(s) = C(sI − A)−1 B + D
Gd (s) = C(sI − A)−1 E + F
are the corresponding plant and disturbance transfer function matrices, respectively.
An alternative notation which is often used for complex state-space expressions, is
[ ]
A B
G(s) =
C D
That is, matrices in square brackets with a vertical and a horizontal line contain
expressions for the state-space representation of some transfer function matrix.
Matrices A, B, C, D, E, F are used also to define dynamical linear(ized) models in
discrete time:
xk+1 = Axk + Buk + Edk
yk = Cxk + Duk + Fdk
Introduction xxxv
where the subscript (k or k + 1) defines the discrete sampling instant in question.

For simplicity of notation, the same notation is used often for continuous- and
discrete-time models, and it should be clear from context whether continuous or
discrete time is used. Note, however, that the model matrices will be different for
discrete and continuous time, i.e. converting from continuous to discrete time (or
vice versa) will change the matrices A, B, E.6
6 Whereas the matrices C, D, F describe instantaneous effects (not affected by the passing of time)
and will be the same for continuous- and discrete-time models.
1
Mathematical and Control Theory Background
1.1 Introduction
This chapter will review some mathematical and control theory background, some
of which is actually assumed covered by previous control courses. Both the coverage
of topics and their presentation will therefore lack some detail, as the presentation
is aiming
● to provide sufficient background knowledge for readers with little exposure to con-
trol theory,
● to correct what is this author’s impression of what are the most common miscon-
ceptions
● to establish some basic concepts and introduce some notation.
1.2 Models for Dynamical Systems

Many different model representations are used for dynamical systems, and a few of
the more common ones will be introduced here.
1.2.1 Dynamical Systems in Continuous Time

A rather general way of representing a dynamical system in continuous time is via a
set of ordinary differential equations:
ẋ = f (x, u, d) (1.1)
where the variables x are termed as the system states and ẋ = dxdt
is the time derivative
of the state. The variables u and d are both external variables that affect the system. In
the context of control, it is common to distinguish between the manipulated variables
or (control) inputs u that can be manipulated by a controller, and the disturbances
d that are external variables that affect the system but which cannot be set by the
controller.
The system states x are generally only a set of variables that are used to describe
the system’s behavior over time. Whether the individual components of the state
Advanced Chemical Process Control: Putting Theory into Practice, First Edition. Morten Hovd.
© 2023 WILEY-VCH GmbH. Published 2023 by WILEY-VCH GmbH.
2 1 Mathematical and Control Theory Background
vector can be assigned any particular physical interpretation will depend on how
the model is derived. For models derived from fundamental physical and chemical
relationships (often termed as “rigorous models”), the states will often be quanti-
ties like temperatures, concentrations, and velocities. If, in contrast, the model is an
empirical model identified from observed data, it will often not be possible to assign
any particular interpretation to the states.
Along with the state equation (1.1), one typically also needs a measurement
equation such as:
y = g(x, u, d) (1.2)
where the vector y is a vector of system outputs, which often correspond to available
physical measurements from the systems. Control design is usually at its most simple
when all states can be measured, i.e. when y = x.
Disturbances need not be included in all control problems. If no disturbances
are included in the problem formulation, Eqs. (1.1) and (1.2) trivially simplify to
ẋ = f (x, u) and y = g(x, u), respectively.
Since we are dealing with dynamical systems, it is hopefully obvious that the vari-
ables x, y, u, d may all vary with time t. In this section, time is considered as a con-
tinuous variable, in accordance with our usual notion of time.
Together, Eqs. (1.1) and (1.2) define a system model in continuous time. This
type of model is rather general and can deal with any system where it suffices to
consider system properties at specific points in space, or where it is acceptable to
average/lump system properties over space. Such models where properties are aver-
aged over space are often called lumped models.
For some applications, it may be necessary to consider also spatial distribution
of properties. Rigorous modeling of such systems typically result with a set of
partial differential equations (instead of the ordinary differential equations of (1.1)).
In addition to derivatives with respect to time, such models also contain derivatives
with respect to one or more spatial dimensions. Models described by partial differ-
ential equations will not be considered any further here. Although control design
based on partial differential equations is an active research area, the more common
industrial practice is to convert the set of partial differential equations to a (larger)
set of ordinary differential equations through some sort of spatial discretization.
1.2.2 Dynamical Systems in Discrete Time

Although time in the “real world” as we know it is a continuous variable, control sys-
tems are typically implemented in computer systems, which cyclically execute a set
of instructions. Measurements and control actions are therefore executed at discrete
points in time, and to describe system progression from one time instant to subse-
quent instants we will need a discrete-time model. Such models may be represented
as:
xk+1 = f (xk , uk , dk ) (1.3)
yk = g(xk , uk , dk ) (1.4)
where xk , yk , uk , and dk are the discrete-time counterparts to the system states,

outputs, inputs, and disturbances introduced above for continuous-time systems,
and the subscript (k) identify the timestep (or sampling instant). Thus, xk is the state
x at timestep k, while xk+1 is the state at the subsequent timestep. Note that although
the same letter f is used to represent the system dynamics for both continuous-
and discrete-time systems, these functions will be different for the two different
model types. The measurement equation, however, will often be identical for the
two model types.
1.2.3 Linear Models and Linearization

Many control design methods are based on linear models. It is therefore necessary
to be able to convert from a nonlinear model to a linear model which is (hopefully) a
close approximation to the nonlinear model. This is called linearization of the non-
linear model.
A systems is linear if the functions f and g (in (1.1) and (1.2) for the case of
continuous-time models, or in (1.3) and (1.4) for the case of discrete-time models)
are linear in all the variables x, u, and d. Thus, a linear continuous-time model may
be expressed as:
ẋ = Ax + Bu + Ed (1.5)
y = Cx + Du + Fd (1.6)
where A, B, C, D, E, F are matrices of appropriate dimensions, and the matrix ele-
ments are independent of the values of x, u, d. Linear models for discrete-time sys-
tems follow similarly.
Linearization is based on the Taylor series expansion of a function. Consider a
function h(a). We want to approximate the value of h(a) in the vicinity of a = a∗ .
The Taylor series expansion then provides the approximation:
𝜕h || 1 𝜕2 h |
h(a) = h(a∗ + 𝛿a) ≈ h(a∗ ) + | 𝛿a + 𝛿aT 2 || 𝛿a + · · · (1.7)
𝜕a |a∗ 2 𝜕a |a=a∗
where the notation |a=a∗ indicates that the value a = a∗ is used when evaluating the
derivatives.
1.2.3.1 Linearization at a Given Point

When linearizing a dynamical system model, we terminate the Taylor series expan-
sion after the first-order term. The underlying nonlinear system is therefore natu-
rally assumed to be continuous and have continuous first-order derivatives. Assume
that the linearization is performed at the point:
⎡ x ⎤ ⎡ x∗ ⎤
a = ⎢ u ⎥ = ⎢ u∗ ⎥ = a ∗ (1.8)
⎢ ⎥ ⎢ ∗⎥
⎣d⎦ ⎣d ⎦
The terminated Taylor series expansion of (1.1) then becomes
dx d𝛿x 𝜕f || 𝜕f || 𝜕f ||
= ≈ f (a∗ ) + | 𝛿x + | 𝛿u + 𝛿d (1.9)
dt dt 𝜕x |a=a ∗ 𝜕u |a=a ∗ 𝜕d ||a=a∗
Similarly, we get for (1.2)

𝜕g || 𝜕g || 𝜕g |
y = y∗ + 𝛿y ≈ g(a∗ ) + | 𝛿x + | 𝛿u || 𝛿d (1.10)
𝜕x |a=a∗ 𝜕u |a=a∗ 𝜕d |a=a∗
where it is understood that y∗ = g(a∗ ).
𝜕f | 𝜕f | 𝜕f | 𝜕g | 𝜕g |
Next, define A = 𝜕x | ∗ , B = 𝜕u | ∗ , E = 𝜕d | ∗ , C = 𝜕x | ∗ , D = | ,
|a=a |a=a |a=a |a=a 𝜕u |a=a∗
|
F = 𝜕d |
𝜕x |a=a∗
Linearizing at an Equilibrium Point The point a∗ used in the linearization is usually

an equilibrium point. This means that
f (a∗ ) = 0 (1.11)
g(a∗ ) = y∗ (1.12)
Thus, we get
dx
= A𝛿x + B𝛿u + E𝛿d (1.13)
dt
𝛿y = C𝛿x + D𝛿u + F𝛿d (1.14)
Linearizing a discrete-time model is done in the same way as for continuous-time
models. The only slight difference to keep in mind is that for a discrete-time model
at steady state xk+1 = xk , and therefore f (a∗ ) = xk when linearizing at a steady
state.
Deviation Variables It is common to express the system variables (x, u, d, and y) in

terms of their deviation from the linearization point a∗ . When doing so the 𝛿’s are
typically suppressed for ease of notation – as will be done in the remainder of this book.
It is, however, important to beware that when converting from deviation variables
to “real” variables, the linearization point has to be accounted for.
To illustrate: A model for a chemical reactor is linearized at steady-state conditions
corresponding to a reactor temperature of 435 K. If the linearized model, expressed
in deviation variables, indicates a temperature of −1, the corresponding “real” tem-
perature would be 434 K.
Linear Controllers Are Not Linear! It appears that many students, even after intro-
ductory control courses, do not appreciate that our so-called “linear” controllers are
only linear when expressed in deviation variables. In “natural” variables, the typi-
cal “linear” controller is in fact affine, i.e. they have a constant term in addition to
the linear term. This can lead to many frustrations, until the misunderstanding has
been clarified – which might actually take some time, because the importance of this
issue will depend on both controller structure and controller type. Consider a simple
feedback loop, with a (linear) controller K controlling a system G, as illustrated in
Figure 1.1.
u*
r e u + y
K G
–
Figure 1.1 A simple feedback loop with a one-degree-of-freedom controller and possible
“output bias.”
This type of controller is called a “one-degree-of-freedom controller,” since

it has only one input, the control offset e = r − y. We can make the following
observations:
● Clearly, it does not matter whether the reference r and measurement y are
expressed in “physical” variables or deviation variables, as long as the same scale
is used for both. This is because the controller input is the difference between
these two variables.
● Consider the case when the controller K is a pure proportional controller,
i.e. u = K(r − y) with K constant. It is then necessary to add u∗ as an “output
bias”1 to the controller output, as indicated by the dashed arrow in the figure.
● Consider next the case when the controller K contains integral action. In this
case, the “output bias” is not strictly necessary, since the value of the integrating
state will adjust for this when the system reaches steady state. However, an out-
put bias may improve transient response significantly when putting the controller
into operation.2
Consider next a loop where the controller has separate entry port for the reference
and the measurement, as shown in Figure 1.2. This type of controller is used
r u*
y* – u y
+
K G
–
Figure 1.2 A simple feedback loop with a two-degree-of-freedom controller and possible
“bias” on both controller inputs and controller output.
1 Some system vendors may use different terminology.

2 See also Section 6.4 on Bumpless Transfer.
when one wants to treat the measurement and reference signals differently in the
controller. We note that
● In this case, we need to subtract the value of the measurement at the linearization
point, y∗ , from both the reference and the measurement.
● Whether to add u∗ to the controller output is determined by the same considera-
tions as for the one-degree-of-freedom controller.
1.2.3.2 Linearizing Around a Trajectory
It was noted above that it is most common to linearize around a steady state.
However, in some cases, one may want to linearize around a trajectory, i.e. around
a series of consistent future values of x, u, and d. This most commonly occurs in
nonlinear model predictive control (MPC). Each time an MPC controller executes,
it solves an optimization problem that optimizes system behavior over a “predic-
tion horizon.” However, for some strongly nonlinear problems, using the same
linearized model for the entire prediction horizon may not give sufficient accuracy.
In such cases, one may choose to linearize around a trajectory instead.
Given the present state, a prediction of the future manipulated variables (typically
obtained from the previous execution of the MPC), and predicted values for future
disturbances, the nonlinear model can be used to simulate the system in the future.
This gives predicted future states that are consistent with the present state and the
predicted future manipulated variables and disturbances.
For each timestep in the future, the linearization is performed around the pre-
dicted state, manipulated variable, and disturbance values. This will give different
matrices A, B, CD, E, F for each timestep. In this way, a nonlinear system is approxi-
mated by a linear, time-varying model.
Linearizing around a trajectory clearly complicates the model. In addition to the
added complexity of having to ensure that the right model matrices are used at the
right timestep in the future, one also has to remember that the linearization point
varies from timestep to timestep (resulting from f (a∗ ) ≠ xk in the discrete-time equiv-
alent of (1.9)). This adds additional complexity when converting between physical
variables and deviation variables.
1.2.4 Converting Between Continuous- and Discrete-Time Models

It will often be necessary to convert from continuous- to discrete-time models (and
less frequently necessary to convert the other way). Process models based on first
principles modeling will typically result in continuous-time models. Often, control
design is performed with a continuous-time model. The continuous-time controller
is thereafter converted to a discrete-time controller for implementation in a com-
puter. There are also controller types that are more conveniently designed using
discrete-time models. The most notable example of such controllers are the so-called
MPC controllers which will be described in some detail later in the book.
To convert from continuous to discrete time, we need to
● choose a numerical integration method for the system dynamics, and
● determine (assume) how the external variables (u and d) change between the time
instants for the discrete-time model.
Another random document with
no related content on Scribd:
MP25125.
Bobby Riggs. 30 sec., sd., color, 16 mm. © Pfizer, Inc.; 12Sep73;
MP25125.
MP25126.
Interviewing interactions. 11 min., sd., color, 16 mm. (Effective
interviewing techniques) © Edutronics Systems International, Inc.;
1Jun73; MP25126.
MP25127.
Introduction to C I C S. Edutronics Systems International, Inc.
Produced in cooperation with On-Line Software, Inc. 11 min., sd.,
color, 16 mm. (CICS applications programming) © Edutronics
Systems International, Inc.; 29Jan74 (in notice: 1973); MP25127.
MP25128.
The Applications program interface. Edutronics Systems
International, Inc. Produced in cooperation with On-Line Software,
Inc. 15 min., sd., color, 16 mm. (CICS applications programming)
Add. ti.: The Application program interface. © Edutronics Systems
International, Inc.; 29Jan74 (in notice: 1973); MP25128.
MP25129.
Management of major resources. Edutronics Systems
Inc. 21 min., sd., color, 16 min. (CICS applications programming) ©
Edutronics Systems International, Inc.; 29Jan74 (in notice: 1973);
MP25129.
MP25130.
Management of supplemental resources. Edutronics Systems
Inc. 23 min., sd., color, 16 mm. (CICS applications programming) ©
Edutronics Systems International, Inc.; 29Jan74 (in notice: 1973);
MP25130.
MP25131.
The System flow. Edutronics Systems International, Inc. Produced
in cooperation with On-Line Software, Inc. 16 min., sd., color, 16
mm. (CICS applications programming) © Edutronics Systems
MP25132.
O S / V S J C L differences. Edutronics Systems International, Inc.
14 min., sd., color, 16 mm. (Virtual storage concepts) © Edutronics
MP25133.
The Principles of virtual storage. Edutronics Systems
International, Inc. 17 min., sd., color, 16 mm. (Virtual storage
concepts) © Edutronics Systems International, Inc.; 29Jan74 (in
notice: 1973); MP25133.
MP25134.
A Matter of indifference. 50 min., sd., b&w, 16 mm. Appl. au.: Jack
M. Hanick & Leonardo Dacchille. © Jack M. Hanick & Leonardo
Dacchille; 2Feb74; MP25134.
MP25135.
Managing in a crisis. BNA Communications, Inc. 27 min., sd.,
color, 16 mm. (The Gellerman effective supervision series) © BNA
Communications, Inc.; 25Feb74 (in notice: 1973); MP25135.
MP25136.
Managing the disturbed employee. BNA Communications, Inc. 29
min., sd., color, 16 mm. (The Gellerman effective supervision series)
© BNA Communications, Inc.; 25Feb74 (in notice: 1973); MP25136.
MP25137.
The Voter decides. Ginn and Company. Made by Centron
Corporation. 16 min., sd., color, 16 mm. (Ginn instructional film
series on political roles) © Ginn and Company; 4Apr73; MP25137.
MP25138.
The Political animal. Ginn and Company. Made by Centron
Corporation. 15 min., sd., color, 16 mm. (Ginn instructional film
series on political roles) © Ginn and Company; 1Aug73; MP25138.
MP25139.
The Role of the interest group leader. Ginn and Company. Made by
Centron Corporation. 15 min., sd., color, 16 mm. (Ginn instructional
film series on political roles) © Ginn and Company; 4Apr73;
MP25139.
MP25140.
1974 full size car body highlights. Chrysler Corporation. 12 min.,
sd., color, Super 8 mm. Appl. au.: Ross Roy, Inc. © Chrysler
Corporation; 22Jan74 (in notice: 1973); MP25140.
MP25141.
Dodge Colt comparison. 12 min., sd., color, Super 8 mm. Appl. au.:
Ross Roy, Inc. © Chrysler Corporation; 23Jan74 (in notice: 1973);
MP25141.
MP25142.
Basic door and body adjustments. Chrysler Corporation. 12 min.,
sd., color, Super 8 mm. Appl. au.: Ross Roy, Inc. © Chrysler
Corporation; 18Mar74 (in notice: 1973); MP25142.
MP25143.
Sell your product, sell yourself. 12 min., sd., color, Super 8 mm.
Appl. au.: Ross Roy, Inc. © Chrysler Corporation; 17Jan74 (in notice:
1973); MP25143.
MP25144.
Handling walk ins. 12 min., sd., color, Super 8 mm. Appl. au.: Ross
Roy, Inc. © Chrysler Corporation; 5Mar74; MP25144.
MP25145.
Someone with authority. 12 min., sd., color, Super 8 mm. Appl.
au.: Ross Roy, Inc. © Chrysler Corporation; 17Apr74; MP25145.
MP25146.
Power steering problems and diagnosis. 12 min., sd., color, Super
8 mm. Appl. au.: Ross Roy, Inc. © Chrysler Corporation; 15Apr74;
MP25146.
MP25147.
Garner Ted Armstrong. Program 546. Ambassador College. 29
min., sd., color, videotape (3/4 inch) in cassette. © Ambassador
College; 24Jan74; MP25147.
MP25148.
College; 5Feb74; MP25148.
MP25149.
MP25150.
MP25151.
MP25152.
MP25153.
MP25154.
Rembrandt. 30 sec., sd., color, 16 mm. © Pfizer, Inc.; 31Oct73;
MP25154.
MP25155.
Limestone: Iowa’s buried treasure. 26 min., sd., color, 16 mm. ©
Iowa State University a. a. d. o. Iowa State University of Science and
Technology; 22Feb74; MP25155.
MP25156.
Point of proof 800 series cornhead. 7 min., sd., color, Super 8 mm.
© International Harvester Company; 12Feb74; MP25156.
MP25157.
Point of proof low profile combine. 8 min., sd., color, Super 8 mm.
© International Harvester Company; 22Jan74; MP25157.
MP25158.
Point of proof 345 spring tooth harrow. 6 min., sd., color, Super 8
mm. © International Harvester Company; 22Jan74; MP25158.
MP25159.
Point of proof rotary cultivators. 7 min., sd., color, Super 8 mm. ©
International Harvester Company; 22Jan74; MP25159.
MP25160.
Mitedos para operar el tractor hydrostatic. 10 min., sd., color,
Super 8 mm. © International Harvester Company; 22Jan74 (in
notice: 1973); MP25160.
MP25161.
Know your combine-corn. 14 min., sd., color, Super 8 mm. Add.
ti.: Know your combine-grain. © International Harvester Company;
10Dec73; MP25161.
MP25162.
Overview. Edutronics Systems International, Inc. 21 min., sd.,
color, 16 mm. (VS/DOS) © Edutronics Systems International, Inc.;
11Feb74; MP25162.
MP25163.
Working set and locality. Edutronics Systems International, Inc. 17
min., sd., color, 16 mm. (Virtual storage concepts) © Edutronics
Systems International, Inc.; 28Feb74 (in notice: 1973); MP25163.
MP25164.
An Approach to growth: awareness training. Griff Ruggles. 26
min., sd., color, 16 mm. © Griff Ruggles; 1Nov73; MP25164.
MP25165.
Outdoor play: a motivating force for learning. A Steve Campus
production. Produced in cooperation with Queens College of the City
University of New York, Board of Higher Education. 17 min., sd.,
color, 16 mm. © Steve Campus Productions, Inc.; 29Nov73;
MP25165.
MP25166.
Hockey: the power play. A Dick Borden production. 11 min., sd.,
b&w, 16 mm. © The Coca-Cola Company; 1Jan74 (in notice: 1973);
MP25166.
MP25167.
1974 cars: low speed crash costs (foreign models) 20 min., sd.,
color, 16 mm. © Insurance Institute for Highway Safety; 28Mar74;
MP25167.
MP25168.
Parent counseling child care system. A John Sutherland
production. 83 min., sd., color, 70 mm. (Rocom Parentaid film
system) © Hoffmann LaRoche, Inc.; 29Oct70; MP25168.
MP25169.
Rolling tissue. 30 sec., sd., color, 16 mm. © Pfizer, Inc.; 17Apr73;
MP25169.
MP25170.
Teaching sign language to the Chimpanzee Washoe. 48 min., sd.,
b&w, 16 mm. (Project Washoe) Appl. au.: R. A. Gardner & B. T.
Gardner. © R. A. Gardner & B. T. Gardner; 4Feb74 (in notice: 1973);
MP25170.
MP25171.
Concrete finishing. A Marshall Maintenance production. Produced
in cooperation with J. W. Ferrell Company, Inc. 12 min., sd., color, 16
mm. © Marshall Maintenance; 6Mar74; MP25171.
MP25172.
Energy conversion. 3 min., si., color, Super 8 mm. in cartridge.
Appl. au.: Doubleday Multimedia, division of Doubleday and
Company, Inc. © Doubleday and Company, Inc.; 7Nov73; MP25172.
MP25173.
Inertia. 3 min., si., color, Super 8 mm. in cartridge. Appl. au.:
Doubleday Multimedia, division of Doubleday and Company, Inc. ©
Doubleday and Company, Inc.; 7Nov73; MP25173.
MP23174.
Matter: spaces between molecules. 3 min., si., color, Super 8 mm.
in cartridge. Appl. au.: Doubleday Multimedia, division of Doubleday
and Company, Inc. © Doubleday and Company, Inc.; 7Nov73;
MP25174.
MP25175.
How to make an animal puppet. 4 min., si., color, Super 8 mm. in
cartridge. Prev. reg. NM: compilation & editing. © Walt Disney
Productions; 20Sep73; MP25175.
MP25176.
Radioactivity. 2 min., si., color, Super 8 mm. in cartridge. Prev.
reg. NM: compilation & editing. © Walt Disney Productions;
20Sep73; MP25176.
MP25177.
Radioactive fission. 2 min., si., color, Super 8 mm. in cartridge.
Prev. reg. NM: compilation & editing. © Walt Disney Productions;
20Sep73; MP25177.
MP25178.
Where today’s cats came from. 4 min., si., color, Super 8 mm. in
MP25179.
Atomic structure of matter. 3 min., si., color, Super 8 mm. in
MP25180.
Lawn groom. 1 min., sd., color, 16 mm. Appl. au.: Martin Latter. ©
Pamar, Inc.; 22Apr74; MP25180.
MP25181.
The Frozen war, America intervenes in Russia, 1918–1920. 30
min., sd., b&w, 16 mm. Appl. au.: William F. Malloch, R. C. Raack &
Patrick Griffin. Prev. reg. 20Aug73, MU8721. NM: revision &
additions. © Patrick Griffin, R. C. Raack & William F. Malloch
operating as a partnership under the fictitious business names Cadre
& Cadre Films; 1Dec73; MP25181.
MP25182.
M F T / M V T lecture 1. 17 min., sd., b&w, 16 mm. (S/360
operating system core dumps) Appl. au.: Consultants Associated,
Inc. © Edutronics Systems International, Inc.; 1Aug72 (in notice:
1971); MP25182.
MP25183.
M V T lecture 2. 17 min., sd., b&w, 16 mm. (S/360 operating
system core dumps) Appl. au.: Consultants Associated, Inc. ©
Edutronics Systems International, Inc.; 1Aug72 (in notice: 1971);
MP25183.
MP25184.
MP25184.
MP25185.
MP25185.
MP25186.
system core dump) Appl. au.: Consultants Associated, Inc. ©
MP25186.
MP25187.
MP25187.
MP25188.
M F T / M V T lecture 7. 20 min., sd., b&w, 16 mm. (S/360
operating system core dumps) Appl. au.: Consultants Associated,
Inc. © Edutronics Systems International, Inc.; 1Aug72 (in notice:
1971); MP25188.
MP25189.
MP25189.
MP25190.
MP25190.
MP25191.
Lecture 1. 14 min., sd., color, 16 mm. (S/360 operating system,
operations operator training) Appl. au.: Consultants Associated, Inc.
© Edutronics Systems International, Inc.; 1Jun72 (in notice: 1971);
MP25191.
MP25192.
MP25192.
MP25193.
MP25193.
MP25194.
MP25194.
MP25195.
MP25195.
MP25196.
MP25196.
MP25197.
MP25197.
MP25198.
MP25198.
MP25199.
MP25199.
MP25200.
MP25200.
MP25201.
MP25201.
MP25202.
The Falcon, prince of birds. Broyles, Allebaugh and Davis, Inc. 26
min., sd., color, 16 mm. © Broyles, Allebaugh and Davis, Inc.;
10Dec73; MP25202.
MP25203.
Boy of Quebec. A Coronet film. 13 min., sd., color, 16 mm. ©
Coronet Instructional Media, a division of Esquire, Inc.; 28Jun73;
MP25203.
MP25204.
Japan: the land and the people. A Coronet film. 2nd ed. 16 min.,
sd., color, 16 mm. © Coronet Instructional Media, a division of
Esquire, Inc.; 3Oct73; MP25204.
MP25205.
Climates and seasons. A Coronet film. 10 min., sd., color, 16 mm.
© Coronet Instructional Media, a division of Esquire, Inc.; 4Oct73;
MP25205.
MP25206.
College; 6Feb74 (in notice: 1973); MP25206.
MP25207.
College; 5Nov73; MP25207.
MP25208.
MP25209.
MP25210.
MP25211.
MP25212.
MP25213.
MP25214.
The World of darkness. 24 min., sd., color, 16 mm. Prev. pub.
11Jan73; MP23932. NM: abridgment. © National Geographic
Society; 8Jan74 (in notice: 1973); MP25214.
MP25215.
Spain: a journey with Washington Irving. 24 min., sd., color, 16
mm. Prev. pub. 15Jun66, MP17683. NM: abridgment. © National
Geographic Society; 20Jul73; MP25215.
MP25216.
Reptiles. 24 min., sd., color, 16 mm. Prev. pub. 25Nov68,
MP18961. NM: abridgment. © National Geographic Society;
20Aug73; MP25216.
MP25217.
Australia: down under and outback. 24 min., sd., color, 16 mm.
Prev. pub. 7Feb69, MP19013. NM: abridgment. © National
Geographic Society; 12Sep73; MP25217.
MP25218.
The National parks: America’s legacy. 24 min., sd., color. 16 mm.
Prev. pub. 11Oct68, MP18649. NM: abridgment. © National
Geographic Society; 12Sep73; MP25218.
MP25219.
Polynesia. 24 min., sd., color, 16 mm. Prev. pub. 25Apr69;
MP19210. NM: abridgment. © National Geographic Society;
5Nov73; MP25219.
MP25220.
Earth: the restless planet. 24 min., sd., color, 16 mm. Prev. pub.
8Feb73; MP23931. NM: abridgment. © National Geographic Society;
16Nov73; MP25220.
MP25221.
Gravity. 4 min., si., color, Super 8 mm. Appl. au.: Doubleday
Multimedia, division of Doubleday and Company, Inc. © Doubleday
and Company, Inc.; 19Feb74 (in notice: 1973); MP25221.
MP25222.
Chemical change. 3 min., si., color, Super 8 mm. Appl. au.:
Doubleday and Company, Inc.; 19Feb74 (in notice: 1973); MP25222.
MP25223.
Matter: weight — mass. 3 min., si., color, Super 8 mm. Appl. au.:
MP25224.
Gyroscope. 4 min., si., color, Super 8 mm. Appl. au.: Doubleday
Multimedia, division of Doubleday and Company, Inc. © Doubleday
and Company, Inc.; 19Feb74 (in notice: 1973); MP25224.
MP25225.
Rotary motion. 3 min., si., color, Super 8 mm. Appl. au.:
MP25226.
From the source. 15 min., sd., color, 16 mm. © Southern California
Rock Products and Ready Mixed Concrete Associations; 1Jun73;
MP25226.
MP25227.
For the rest of your life. 16 min., sd., color, 16 mm. Appl. au.:
Motion Picture Unit, University of Iowa. © National Society for the
Prevention of Blindness, Inc.; 1Mar74 (in notice: 1973); MP25227.
MP25228.
Group 7A elements: Physical properties. Pt. 1. A Kalmia film. 4
min., si., color, Super 8 mm. (Physical and chemical properties of the
elements) Appl. au.: John T. Fitch & Robert M. Fitch. © Kalmia
Company; 27Oct72; MP25228.
MP25229.
Group 7A elements: Chemical properties. A Kalmia film. 4 min.,
si., color, Super 8 mm. (Physical and chemical properties of the
MP25230.
Group 7B elements: Manganese. A Kalmia film. 4 min., si., color,
Super 8 mm. (Physical and chemical properties of the elements)
Appl. au.: John T. Fitch & Robert M. Fitch. © Kalmia Company;
27Oct72; MP25230.
MP25231.
Group 7A elements: Physical properties. Pt. 2. A Kalmia film. 3
min., si., color, Super 8 mm. (Physical and chemical properties of the
MP25232.
Chemical hazards. A Kalmia film. 4 min., si., color, Super 8 mm.
(Safety in the laboratory) Appl. au.: John T. Fitch & Robert M. Fitch.
© Kalmia Company; 27Oct72; MP25232.
MP25233.
Personal safety. A Kalmia film. 3 min., si., color, Super 8 mm.
(Safety in the laboratory) Appl. au.: John T. Fitch & Robert M. Fitch.
© Kalmia Company; 27Oct72; MP25233.
MP25234.
Fire in the laboratory. A Kalmia film. 4 min., si., color, Super 8
mm. (Safety in the laboratory) Appl. au.: John T. Fitch & Robert M.
Fitch. © Kalmia Company; 27Oct72; MP25234.
MP25235.
Source data controls. Edutronics Systems International, Inc. 10
min., sd., color, 16 mm. (EDP systems controls) © Edutronics
MP25236.
Validity checking controls. Edutronics Systems International, Inc.
11 min., sd., color, 16 mm. (EDP systems controls) © Edutronics
MP25257.
Master file controls. Edutronics Systems International, Inc. 6 min.,
sd., color, 16 mm. (EDP systems controls) © Edutronics Systems
MP25258.
Processing controls. Edutronics Systems International, Inc. 18
min., sd., color, 16 mm. (EDP systems controls) © Edutronics
MP25239.
Organization controls in E D P systems. Edutronics Systems
International, Inc. 12 min., sd., color, 16 mm. (EDP systems controls)
© Edutronics Systems International, Inc.; 29Jan74 (in notice: 1973);
MP25239.
MP25240.
System 370 I/O device overview. Edutronics Systems
International, Inc. 9 min., sd., color, 16 mm. (370 I/O device
operations) © Edutronics Systems International, Inc.: 14Nov73;
MP25240.
MP25241.
The 3270 information display system. Edutronics Systems
International, Inc. 11 min., sd., color, 16 mm. (370 I/O device
operations) © Edutronics Systems International, Inc.; 14Nov73;
MP25241.
MP25242.
The 3420 magnetic tape unit. Edutronics Systems International,
Inc. 12 min., sd., color, 16 mm. (370 I/O device operations) ©
Edutronics Systems International, Inc.; 14Nov73; MP25242.
MP25243.
The 3410 magnetic tape unit. Edutronics Systems International,
Edutronics Systems International, Inc.: 14Nov73: MP25243.
MP25244.
The 3505 card reader. Edutronics Systems International, Inc. 11
min. sd., color. 16 mm. (370 I/O device operations) © Edutronics
Systems International, Inc.; 14Nov73; MP25244.
MP25245.
The 3211 printer. Edutronics Systems International, Inc. 16 min.,
sd., color, 16 mm. (370 I/O device operations) © Edutronics Systems
International, Inc.; 14Nov73; MP25245.
MP25246.
The 3330 disk storage drive. Edutronics Systems International,
Edutronics Systems International, Inc.; 14Nov73; MP25246.
MP25247.
The 3525 card punch. Edutronics Systems International, Inc. 10
min., sd., color, 16 mm. (370 I/O device operations) © Edutronics
Systems International, Inc.; 14Nov73; MP25247.
MP25248.
min., sd., color, videotape (3/4 inch) © Ambassador College:
28Mar74 (in notice: 1973); MP25248.

Advanced Chemical Process Control 1St Edition Morten Hovd Full Chapter PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Advanced Chemical Process Control 1St Edition Morten Hovd Full Chapter PDF

Uploaded by

Copyright:

Available Formats

Advanced Chemical Process Control

1st Edition Morten Hovd

Chemical Engineering Process Simulation 1st Edition

More incidents that define process safety First Edition

Process Control 4th Edition Sohrab Rohani (Ed.)

Advanced Control of Power Converters Hasan Komurcugil

Human Factors Handbook for Process Plant Operations:

Chemical Engineering Process Simulation 2nd Edition

Process Safety for Engineers : An Introduction 2nd

Process Control Instrumentation Technology 8th Edition

Putting Theory into Practice

British Library Cataloguing-in-Publication Data

Bibliographic information published by

© 2023 WILEY-VCH GmbH, Boschstr. 12,

All rights reserved (including those of

Print ISBN: 978-3-527-35223-4

Typesetting Straive, Chennai, India

1 Mathematical and Control Theory Background 1

2 Control Conﬁguration and Controller Tuning 51

2.2 Input and Output Selection 62

2.4.7.1 The Performance Relative Gain Array 103

3 Control Structure Selection and Plantwide Control 115

4 Limitations on Achievable Performance 143

4.1.2.3 Bounds on Weighted System Norms 146

5 Model-Based Predictive Control 173

5.2.3 Optimizing Deviations from Linear State Feedback 181

6 Some Practical Issues in Controller Implementation 233

7 Controller Performance Monitoring and Diagnosis 249

7.4 Plantwide Oscillations 269

8 Economic Control Beneﬁt Assessment 297

A Fourier–Motzkin Elimination 303

B Removal of Redundant Constraints 307

C The Singular Value Decomposition 309

D Factorization of Transfer Functions into Minimum Phase Stable and

E Models Used in Examples 315

resulted in tools both for monitoring and diagnosing the performance of

2. To increase the understanding among control engineers (students or graduates

The structure of the Book

What’s Not in the Book

August, 2022 M. Hovd

AKF augmented Kalman filter

PSCI power spectral correlation index

I.1 Why Is Chemical Process Control Needed?

I.2 What Knowledge Does a chemical Process Control

control engineer needs to have a thorough understanding of the process. Most

I.3 What Makes Chemical Process Control Unique?

of temperatures, pressures, and flows are readily available, while composition

I.4 The Structure of Control Systems in the Chemical

Real time optimization

To manipulated variables From measurements

● Economics. Optimal control performance – defined in normal control engineering

1. The traditional control system must give unacceptable performance.

x Vector of system states.

where the subscript (k or k + 1) defines the discrete sampling instant in question.

Mathematical and Control Theory Background

1.2 Models for Dynamical Systems

1.2.1 Dynamical Systems in Continuous Time

1.2.2 Dynamical Systems in Discrete Time

where xk , yk , uk , and dk are the discrete-time counterparts to the system states,

1.2.3 Linear Models and Linearization

1.2.3.1 Linearization at a Given Point

Similarly, we get for (1.2)

Linearizing at an Equilibrium Point The point a∗ used in the linearization is usually