RESEARCH IN PROCESS CONTROL

Pradeep Deshpande                                                 Tel: (502) 852-0885
Chemical Engineering Department                     Fax: (502) 852-6355
University of Louisville
Louisville, KY 40292, USA                                  E-mail: pradeep.deshpande@louisville.edu

Technology Vision 2020, a document jointly put together by AIChE and others, envisions that tomorrow's process plants will run effortlessly under the command of computer systems constantly striving to achieve maximum productivity, producing high quality products, while minimizing energy consumption, in a safe and environmentally friendly manner. It predicts that nonlinear technologies will control and optimize complex processes, which will also be used to predict stream composition avoiding the necessity of using expensive hardware devices for measurement. Watchdog software systems based on expert systems, artificial neural networks, and statistics, will ensure that emergency situations are avoided and not allowed to get out of hand. Against this backdrop, we summarize the state of the art in process control and in this context provide information on research in this area at the University of Louisville.

The field of process control is relatively new. The tuning rules of PID controllers were proposed by Ziegler & Nichols and Cohen & Coon sixty plus years ago. Pneumatic control systems slowly gave way to electronic analog control systems in the fifties and sixties. Computer systems were first deployed for control in the late fifties and early sixties. Today, large plants are controlled with distributed control systems (DCS), worldwide. The entry of low cost microprocessors in control applications along with multitasking operating systems and PC-based data acquisition and control software has meant that even small to medium processes can derive the full benefits of computer control and optimization. While the chief beneficiaries of the advances in control and optimization until recently have been continuous processes, rapid progress in the application of modern technologies to batch and discrete parts manufacturing processes is taking place.

The availability of digital computers for monitoring, control, and optimization means that hardware realizability is no longer an issue. The proliferation of methodologies for improved control and optimization in the last two to three decades is due to this fact. Some of the notable advances in process control include (1) dead-time compensation, (2) feedforward compensation, (3) cascade control, (4) interaction analysis, variable pairing, multi-loop controller design, and explicit & implicit interaction compensation, (5) constraints handling, (6) optimization, (7) multivariable process identification, (8) stochastic control and time series analysis, (9) constrained model predictive control, (10) adaptive control and self tuning, and (11) statistical process control.

Process control research at U of L has spanned two and a half decades. During this period, a number of control laws and technologies have been developed by Prof. Deshpande and his students. The systems considered have included single-loop linear systems, multivariable linear systems, single-loop nonlinear systems, multivariable nonlinear systems, and chaotic systems. A unifying framework for engineering and statistical process control has been developed. Recently, the increasingly popular six sigma methodology has been scrutinized in the context of engineering process control, statistical process control, and emerging technologies. This work has resulted in five textbooks and over one hundred refereed publications and presentations. Several of these technologies have been used to enhance the state of the art of specific processes, one notable example is advanced control and optimization to desalination processes. Current research is focusing on fault detection, artificial neural networks, multivariable advanced control and optimization, and six sigma for process industries, with special emphasis on the application of these modern methods in industry.

Evidence of success of a research program is generally gauged based on innovation, ability to obtain sponsorship, and on how well the graduates do in the real world. Funding for research has come major organizations such as Exxon, Du Pont, NSF, and others, including major funding from an overseas sponsor. In the last twenty-five years, seventeen Ph.D. students and over fifty Master's students have graduated. The graduates are employed at some of the best known companies in the country including Exxon, Du Pont, Aspen Technology, Bechtel, and others.