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.
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 simplify 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 presentation 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 hurdle. 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 discussing 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 functionality 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 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 formulations 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 economic 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 topics. 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 nonlinear 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).
• 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 currently a lot of research on integrating machine learning with advanced control such as MPC. This author definitely accepts the relevance of research on learning 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...