Techniques of Model-Based Control

About

Features

Description

The state-of-the-art publication in model-based process control—by leading experts in the field.

In Techniques of Model-Based Control, two leading experts bring together powerful advances in model-based control for chemical-process engineering. Coleman Brosilow and Babu Joseph focus on practical approaches designed to solve real-world problems, and they offer extensive examples and exercises.

Coverage includes:

• The nature of the process-control problem and how model-based solutions help to solve it
• Continuous time modeling: time domain, Laplace domain, and FOPDT models
• Feedforward, cascade, override, and single-variable inferential control approaches
• One and two degree of freedom Internal Model Control
• Model State Feedback and PI/PID Implementations of IMC
• Tuning and synthesis of 1DF and 2DF IMC for process uncertainty
• Estimation and inferential control using multiple secondary measurements
• Basic and advanced techniques of model identification and model-predictive control

The appendices review the basics of Laplace transforms, feedback control, frequency response analysis, probability, random variables, and linear least-square regression.

From start to finish, Techniques of Model-Based Control offers the real-world insight that professionals need to identify and implement the best control strategies for virtually any process.

Downloads

Downloads

Below you will find links to download software, the authors' course syllabi, and supplemental chapter files.

Software

You can use the software below to reproduce the results of the examples in the text and to solve the problems at the end of the chapters. (Please note that the software downloads are in .zip format.)

1. IMCTUNE: A zipped collection of m-files
   Runs under both Matlab 5.3 and 6.0. Requires Control System and Optimization Toolboxes. (See also the Updates tab for this download.)
2. MODELBUILDER
   Requires Matlab 6.0, Identification Toolbox. Optimization toolbox recommended.
3. PIDTUNER
   Requires Matlab 6.0.
4. SIMULINK CASE STUDIES. Require Matlab 6.0 and Simulink 4.
   1. Naphtha Cracker
   2. Shell Fractionator
   3. Temperature and Level Control in a Mixing Tank
   4. Pressure and Level Control
   5. Temperature and Level Control
   6. Heat Exchanger
   7. Tennessee Eastman Process

Syllabi

Below are .zip archives containing the syllabi used for the authors' graduate and undergraduate courses. (Please note that the syllabi are Microsoft^® Word documents.)

• Coleman Brosilow's Courses
• Babu Joseph's Courses

Sample Content

Online Sample Chapter

One-Degree of Freedom Internal Model Control

Table of Contents

Preface.
Acknowledgements.
1. Introduction.
2. Continuous-Time Models.
3. One-Degree of Freedom Internal Model Control.
4. Two-Degree of Freedom Internal Model Control.
5. Model State Feedback Implementations of IMC.
6. PI and PID Parameters From IMC Designs.
7. Tuning and Synthesis of 1DF IMC for Uncertain Processes.
8 Tuning and Synthesis of 2DF IMC for Uncertain Processes.
9. Feedforward Control.
10. Cascade Control.
11. Output Constraint Control (Override Control).
12. Single Variable Inferential Control (IC).
13. Inferential Estimation Using Multiple Measurements.
14. Discrete-Time Models.
15. Identification: Basic Concepts.
16. Identification: Advanced Concepts.
17. Basic Model-Predictive Control.
18. Advanced Model-Predictive Control.
19. Inferential Model-Predictive Control.
Appendix A. Review of Basic Concepts.
Appendix B. Review of Frequency Response Analysis.
Appendix C: Review of Linear Least-Squares Regression.
Appendix D: Review of Random Variables and Random Processes.
Appendix E: MATLAB and Control Toolbox Tutorial.
Appendix F: SIMULINK Tutorial.
Appendix G: Tutorial on IMCTUNE Software.
Appendix H: Identification Software.
Appendix I: SIMULINK Models for Projects.
Index.

Preface

Preface

The design and tuning of any control systemis always based on a model of the process to be controlled. For example, when an engineertunes a PID control system online the controller gain and the integral and derivative timeconstants obtained from the tuning depend on the local behavior of the process, and this lo-calbehavior could, if desired, be well approximated by a mathematical model. It turns out,however, that even in the prosaic task of tuning a PID controller, much better control systembehavior can be obtained if the local mathematical model of the process is actually obtained,and the PID tuning is based on that model (see Chapter 6 on PID tuning). The foregoing notwithstanding, in this text the term model-based controller is used primarily to mean controlsystems that explicitly embed a process model in the control algorithm. In particular, weconsider control algorithms such as internal model control (IMC), inferential control (IC)and model-predictive control (such as dynamic matrix control or DMC), which have foundapplications in the process industry over the last few decades.

The book focuses on techniques. By this we mean how the algorithms are designedand applied. There is less emphasis on the underlying theory. We have also used simple ex-amplesto illustrate the concepts. More complex and realistic examples are provided in thetext as case study projects.

We have written the text with two types of audience in mind. One is the typical industrialpractitioner engaged in the practice of process control and interested in learning thebasics behind various controller tuning methods as well as advanced control strategies beyondtraditional PID feedback control. Our aim is to provide sufficient understanding of themethodologies of model-based control to enable the engineer to determine where and when such control strategies can offer substantial improvement in control as well as how to implementand maintain such strategies. The second audience that we have in mind is studentsin senior or graduate level advanced process control courses. For such students, we havetried to provide homework exercises and suggested projects to enhance the learning process.It is assumed throughout that the student has convenient access to modern computing systemsalong with the necessary software. Most of the problems and examples cannot be carriedout without the use of such tools.

Chapter 1 gives an overview of the hierarchical approach to process control. Chapter 2reviews the various types of models used in process control with emphasis on continuoustime models. Chapter 3 gives the development of the basic model-based control structure(IMC). The latter two chapters form the basis of the further developments in both continuousand discrete time implementations.

From this point on the reader can take two possible paths: the first path focuses on thedevelopment of theory and structures for continuous-time implementation using the IMCstructure. The second path focuses on discrete-time (computer-based) implementation ofmodel-based control.

Chapters 4 through 13 cover the first path. Chapter 4 focuses on simultaneous setpointand disturbance rejection using a two-degree of freedom control structure. Chapter 5 showshow to handle control effort constraints using model state feedback. Chapter 6 shows therelationship between classical PID controllers and the internal model control structure.Chapter 7 shows how to design one-degree of freedom controllers in the presence of modeluncertainty. Chapter 8 shows how to tune two-degree of freedom controllers. Chapters 9through 11 focus on multiloop control structures such as feedforward, cascade, and constraintcontrol. Chapter 12 discusses control using secondary measurements (called inferentialcontrol). Chapter 13 extends the concepts of Chapter 12 to inferential control using multiplesecondary measurements using disturbance estimation methods.

Chapters 14 through 19 cover the second path dealing with discrete-time computer implementationof model-based controllers. Chapter 14 introduces the models used in discrete-timerepresentation and Chapters 15 and 16 discuss algorithms used to identify such modelsfrom plant test data. Chapters 17 and 18 discuss the computer implementation of model-basedcontrol using the model-predictive control framework. Chapter 19 extends this to inferentialcontrol using secondary measurements.

As an aid to the various possible readers, we have provided an extensive set of appendicesthat contain the background material necessary for the material in the main body ofthe text. We have indicated at the beginning of each chapter the prerequisite material, and wesuggest that the appropriate appendices be reviewed prior to reading the chapter. The followingmaterial is reviewed:

• Laplace transforms and block diagrams.
• Frequency response methods.
• Linear least square regression.
• Probability theory and random variables.
• MATLAB and SIMULINK software.

We have used the MATLAB/SIMULINK software system as the platform uponwhich to develop software that provides added functionality and a convenient interface forsolving otherwise complex problems. Among the reasons for this choice is that theMATLAB platform provides the tools required to implement and test various control conceptswith relatively little effort required to learn how to use the software. However, there is other software that provides similar functionality, and the reader is encouraged to use which-evertools are most comfortable.

The website (http://www.phptr.com/brosilow/) associated with this text contains the following material:

* IMCTUNE:

A user-friendly interface to a set of MATLAB m-files that enables the reader to design and tune both IMC and PID controllers for single loop, feedforward and cascade control systems. In addition to MATLAB and SIMULINK, IMCTUNE requires the Control System and Optimization toolboxes in addition toA user-friendly interface to a set of MATLAB m-files that enables the MATLAB 5.3.1 or later versions.

* MODELBUILDER:

A set of MATLAB m-files to generate both discrete time and continuous time process models from input/output data. Requires the Identification Toolbox.

* PIDTUNER:

A set of MATLAB m-files that implement and test classical PID tuning. Requires the Control System Toolbox.

* SIMULINK case study models:

MATLAB/SIMULINK models of a number ofprocess systems including the Shell fractionation column, a Naphtha cracker simulator and the Tennessee Eastman problem. These models can be used in course projects.

* Microsoft PowerPoint slides

containing the text material.

* MATLAB m-files, mat-files and data files, SIMULINK mdl-files

and MicrosoftWord files related to the examples and exercises in the text. These files are identified by chapter and example number. The * .mat files for the examples in Chapters 3 through 11 are associated with IMCTUNE.

* Syllabi

used by the authors in their respective undergraduate and graduate courses.

The reader is strongly encouraged to download the software listed above and to useit to reproduce results of examples in the text, and to solve the problems at the end of thechapters. Our experience with teaching using this text material indicates that hands-on exercisesusing simulated processes is important to get a good understanding of the concepts.

Hence the reader is strongly urged to experiment with the software. We have provided anumber of exercises that require computer implementation and testing of the concepts.While the normal reader will have had an introductory course in process control, itis possible to use at least parts of this text in an introductory course. For example, ColemanBrosilow has used Chapters 3, 5, 6, 7, and parts of 9 and 10 in a first undergraduate course in process control for more than ten years. Babu Joseph has been using the Appendix mate-rialin the laboratory sessions associated with the undergraduate course. The entire materialin this book can be included in a graduate course on process control. However, the bookshould be supplemented with some additional material on multivariable control.

The SIMULINK case studies provide comprehensive test beds for implementingand testing the various concepts and algorithms presented in the text. The visual feedbackprovided by the simulation case studies is valuable in understanding the performance andlimitations of the control algorithms. Experience with the simulated examples can smooththe transition to real-world applications.

Updates

Updates & Corrections

Download a .zip archive containing these two files (noise_amp_e.m and noise_amp_e2.m) to resolve a problem encountered when using IMCTUNE.

Submit Errata