Control Engineering

Series Editor
William S. Levine
Department of Electrical and Computer Engineering
University of Maryland
College Park, MD 20742-3285
USA

Editorial Advisory Board

Okko Bosgra William Powers
Delft University Ford Motor Company (retired)
The Netherlands USA

Graham Goodwin Mark Spong

University of Newcastle University of Illinois
Australia Urbana-Champaign
USA
Petar Kokotovic
University of California lori Hashimoto
Santa Barbara Kyoto University
USA Kyoto
Japan
Manfred Morari
ETH
Zurich
Switzerland
Guillermo J. Silva
Aniruddha Datta
S.R Bhattacharyya

PID Controllers
for Time-Delay Systems

Birkhauser
Boston • Basel • Berlin
 Guillermo J. Silva Aniruddha Datta
IBM Department of Electrical Engineering
11400 Burnet Road Texas A&M University
Austin, TX 78758 College Station, TX 77843
USA USA

S.P. Bhattachaiyya
Department of Electrical Engineering
Texas A&M University
College Station, TX 77843
USA

AMS Subject Classifications: 30-02, 37F10, 65-02, 93D99

Library of Congress Cataloging-in-Publication Data

Silva, G. J., 1973-
PID controllers for time-delay systems / GJ. Silva, A. Datta, S.P. Bhattacharyya.
p. cm. - (Control engineering)
ISBN 0-8176-4266-8 (alk. paper)
1. PID controllers-Design and construction. 2. Time delay systems. I. Datta,
Aniruddha, 1963- II. Bhattacharyya, S. P (Shankar P), 1946- III. Title. IV. Control
engineering (Birkhauser)

TJ223.P55S55 2004
629.8'3-dc22 2004062387

ISBN 0-8176-4266-8 Printed on acid-free paper.

©2005 Birkhauser Boston
All rights reserved. This work may not be translated or copied in whole or in part without the writ-
ten permission of the publisher (Birkhauser Boston, Inc., c/o Springer Science+Business Media Inc.,
Rights and Permissions, 233 Spring Street, New York, NY, 10013 USA), except for brief excerpts in
connection with reviews or scholarly analysis. Use in connection with any form of information storage
and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now
known or hereafter developed is forbidden.
The use in this publication of trade names, trademarks, service marks and similar terms, even if they
are not identified as such, is not to be taken as an expression of opinion as to whether or not they are
subject to property rights.

Printed in the United States of America. (SB)

987654321 SPIN 10855839

www. birkhauser. com

 THIS BOOK IS DEDICATED TO

My wife Sezi§ for her loving support and endless patience, and my parents
Guillermo and Elvia.
G. J. Silva

My wife Anindita and my daughters Apama and Anisha.

A. Datta

The memory of my friend and mentor, the late Yakov Z. Tsypkin, Russian
control theorist and academician whose many contributions include the
first results, in 194-6, analyzing the stability of time-delay systems.
S. P. Bhattacharyya
Contents

Preface xi

1 Introduction 1
1.1 Introduction to Control 1
1.2 The Magic of Integral Control 3
1.3 PID Controllers 6
1.4 Some Current Techniques for PID Controller Design . . . . 7
1.4.1 The Ziegler-Nichols Step Response Method 7
1.4.2 The Ziegler-Nichols Frequency Response Method . . 9
1.4.3 PID Settings using the Internal Model Controller
Design Technique 11
1.4.4 Dominant Pole Design: The Cohen-Coon Method . . 13
1.4.5 New Tuning Approaches 14
1.5 Integrator Windup 16
1.5.1 Setpoint Limitation 16
1.5.2 Back-Calculation and Tracking 17
1.5.3 Conditional Integration 17
1.6 Contribution of this Book 18
1.7 Notes and References 18

2 The Hermite-Biehler Theorem and its Generalization 21

2.1 Introduction 21
2.2 The Hermite-Biehler Theorem for Hurwitz Polynomials . . 22
2.3 Generalizations of the Hermite-Biehler Theorem 27
ii Contents

2.3.1 No Imaginary Axis Roots 29

2.3.2 Roots Allowed on the Imaginary Axis Except at the
Origin 31
2.3.3 No Restriction on Root Locations . 35
2.4 Notes and References 37

PI Stabilization of Delay-Free Linear Time-Invariant

Systems 39
3.1 Introduction 39
3.2 A Characterization of All Stabilizing Feedback Gains . . . . 40
3.3 Computation of All Stabilizing PI Controllers 51
3.4 Notes and References 56

P I D Stabilization of Delay-Free Linear Time-Invariant

Systems 57
4.1 Introduction 57
4.2 A Characterization of All Stabilizing PID Controllers . . . . 58
4.3 PID Stabilization of Discrete-Time Plants 67
4.4 Notes and References 75

Preliminary Results for Analyzing Systems with Time

Delay 77
5.1 Introduction 77
5.2 Characteristic Equations for Delay Systems 78
5.3 Limitations of the Pade Approximation 82
5.3.1 Using a First-Order Pade Approximation 83
5.3.2 Using Higher-Order Pade Approximations 85
5.4 The Hermite-Biehler Theorem for Quasi-Polynomials . . . . 89
5.5 Applications to Control Theory 92
5.6 Stability of Time-Delay Systems with a Single Delay . . . . 99
5.7 Notes and References 106

Stabilization of Time-Delay Systems using a Constant Gain

Feedback Controller 109
6.1 Introduction 109
6.2 First-Order Systems with Time Delay 110
6.2.1 Open-Loop Stable Plant 112
6.2.2 Open-Loop Unstable Plant 116
6.3 Second-Order Systems with Time Delay 122
6.3.1 Open-Loop Stable Plant 125
6.3.2 Open-Loop Unstable Plant 129
6.4 Notes and References 134

PI Stabilization of First-Order Systems with Time Delay 135

7.1 Introduction 135
 Contents ix

7.2 The PI Stabilization Problem 136

7.3 Open-Loop Stable Plant 137
7.4 Open-Loop Unstable Plant 150
7.5 Notes and References 159

8 PID Stabilization of First-Order Systems with Time Delay 161

8.1 Introduction 161
8.2 The PID Stabilization Problem 162
8.3 Open-Loop Stable Plant 164
8.4 Open-Loop Unstable Plant 179
8.5 Notes and References 189

9 Control System Design Using the P I D Controller 191

9.1 Introduction 191
9.2 Robust Controller Design: Delay-Free Case 192
9.2.1 Robust Stabilization Using a Constant Gain 194
9.2.2 Robust Stabilization Using a PI Controller 196
9.2.3 Robust Stabilization Using a PID Controller . . . . 199
9.3 Robust Controller Design: Time-Delay Case 203
9.3.1 Robust Stabilization Using a Constant Gain 204
9.3.2 Robust Stabilization Using a PI Controller 205
9.3.3 Robust Stabilization Using a PID Controller . . . . 208
9.4 Resilient Controller Design 213
9.4.1 Determining fc, T, and L from Experimental Data . 213
9.4.2 Algorithm for Computing the Largest Ball Inscribed
Inside the PID Stabilizing Region 214
9.5 Time Domain Performance Specifications 217
9.6 Notes and References 222

10 Analysis of Some P I D Tuning Techniques 223

10.1 Introduction . 223
10.2 The Ziegler-Nichols Step Response Method 224
10.3 The CHR Method 229
10.4 The Cohen-Coon Method 233
10.5 The IMC Design Technique 237
10.6 Summary 241
10.7 Notes and References 241

11 P I D Stabilization of Arbitrary Linear Time-Invariant

Systems with Time Delay 243
11.1 Introduction 243
11.2 A Study of the Generalized Nyquist Criterion 244
11.3 Problem Formulation and Solution Approach 248
11.4 Stabilization Using a Constant Gain Controller 250
11.5 Stabilization Using a PI Controller 253
X Contents

11.6 Stabilization Using a PID Controller 256

11.7 Notes and References 263

12 Algorithms for Real and Complex P I D Stabilization 265

12.1 Introduction 265
12.2 Algorithm for Linear Time-Invariant Continuous-Time
Systems 266
12.3 Discrete-Time Systems 276
12.4 Algorithm for Continuous-Time First-Order Systems with
Time Delay 277
12.4.1 Open-Loop Stable Plant 279
12.4.2 Open-Loop Unstable Plant 280
12.5 Algorithms for PID Controller Design 284
12.5.1 Complex PID Stabilization Algorithm 285
12.5.2 Synthesis of Hoc PID Controllers 287
12.5.3 PID Controller Design for Robust Performance . . . 291
12.5.4 PID Controller Design with Guaranteed Gain and
Phase Margins 293
12.6 Notes and References 295

A Proof of Lemmas 8.3, 8.4, and 8.5 297

A.l Preliminary Results 297
A.2 Proof of Lemma 8.3 301
A.3 Proof of Lemma 8.4 302
A.4 Proof of Lemma 8.5 303

B Proof of Lemmas 8.7 and 8.9 307

B.l Proof of Lemma 8.7 307
B.2 Proof of Lemma 8.9 308

C Detailed Analysis of Example 11.4 313

References 323

Index 329
 Preface

This monograph presents our recent results on the proportional-integral-

derivative (PID) controller and its design, analysis, and synthesis. The fo-
cus is on linear time-invariant plants that may contain a time delay in
the feedback loop. This setting captures many real-world practical and in-
dustrial situations. The results given here include and complement those
published in Structure and Synthesis of PID Controllers by Datta, Ho, and
Bhattacharyya [10]. In [10] we mainly dealt with the delay-free case.
The main contribution described here is the efficient computation of the
entire set of PID controllers achieving stability and various performance
specifications. The performance specifications that can be handled within
our machinery are classical ones such as gain and phase margin as well as
modern ones such as Hoo norms of closed-loop transfer functions. Finding
the entire set is the key enabling step to realistic design with several design
criteria. The computation is efficient because it reduces most often to lin-
ear programming with a sweeping parameter, which is typically the propor-
tional gain. This is achieved by developing some preliminary results on root
counting, which generalize the classical Hermite-Biehler Theorem, and also
by exploiting some fundamental results of Pontryagin on quasi-polynomials
to extract useful information for controller synthesis. The efficiency is im-
portant for developing software design packages, which we are sure will
be forthcoming in the near future, as well as the development of further
capabilities such as adaptive PID design and online implementation. It is
also important for creating a realistic interactive design environment where
multiple performance specifications that are appropriately prioritized can
be overlaid and intersected to telescope down to a small and satisfactory
xii Preface

controller set. Within this set further design choices must be made that
reflect concerns such as cost, size, packaging, and other intangibles beyond
the scope of the theory given here.
The PID controller is very important in control engineering appHcations
and is widely used in many industries. Thus any improvement in design
methodology has the potential to have a significant engineering and eco-
nomic impact. An excellent account of many practical aspects of PID con-
trol is given in PID Controllers: Theory, Design and Tuning by Astrom
and Hagglund [2], to which we refer the interested reader; we have chosen
to not repeat these considerations here. At the other end of the spectrum
there is a vast mathematical literature on the analysis of stability of time-
delay systems which we have also not included. We refer the reader to the
excellent and comprehensive recent work Stability of Time-Delay Systems
by Gu, Kharitonov, and Chen [15] for these results. In other respects our
work is self-contained in the sense that we present proofs and justfications
of all results and algorithms developed by us.
We believe that these results are timely and in phase with the resurgence
of interest in the PID controller and the general rekindling of interest in
fixed and low-order controller design. As we know there are hardly any
results in modern and postmodern control theory in this regard while such
controllers are the ones of choice in applications. Classical control theory
approaches, on the other hand, generally produce a single controller based
on ad hoc loop-shaping techniques and are also inadequate for the kind
of computer-aided multiple performance specifications design applications
advocated here. Thus we hope that our monograph acts as a catalyst to
bridge the theory-practice gap in the control field as well as the classical-
modern gap.
The results reported here were derived in the Ph.D. theses of Ming-Tzu
Ho, Guillermo Silva, and Hao Xu at Texas A&M University and we thank
the Electrical Engineering Department for its logistical support. We also
acknowledge the financial support of the National Science Foundation's
Engineering Systems Program under the directorship of R. K. Baheti and
the support of National Instruments, Austin, Texas.

Austin, Texas G. J. Silva

College Station, Texas A. Datta
College Station, Texas S. P. Bhattacharyya
October 2004
 PID Controllers
for Time-Delay Systems