47540-AC9
Next Generation Regulatory Controller for Chemical Processes
Introduction. Without effective process control, safe, efficient and economic operation is virtually impossible in refineries or other chemical processes. Advances in control theory have led to the development of such powerful techniques as Model predictive Control (MPC) capable of delivering overall high control systems performance. However, in the control system hierarchy, the effectiveness of any advanced control system responsible for high performance operations remains limited by the base regulatory PID controllers operating below it. As processes and process operations have become more complicated, meeting the stringent demands on these manufacturing processes effectively in the 21st century requires not just more advanced control techniques, but also a fundamentally re-imagined alternative next generation regulatory controller that takes proper advantage of modern digital electronic technology to overcome the weaknesses of current PID controller technology.
Results. In the first phase of this work, by using a combination of Bayesian Estimation (for current disturbance estimation), the statistical theory of stochastic forecasting (for future disturbance prediction), and least squares optimization (for control action calculation), we have successfully completed the development of a controller (i) whose tuning is more transparent, and (ii) which is also more effective than currently available PID technology. Specifically, we have developed a 4-mode control scheme that takes full advantage of modern electronic hardware components, and whose tuning parameters are directly related to controller performance attributes (robustness, set-point tracking, and disturbance rejection); its achievable performance is also better than that of the PID controller, and it can be designed and implemented much more directly and transparently.
We have used concepts from robust stability analysis to develop easy-to-apply tuning rules which have been tested in simulation on a model of an industrial polymerization reactor. Some preliminary applications tests have also been carried out recently on a laboratory scale simple water tank level control experiment.
Significance. Our results thus far demonstrate that the new controller, which combines the simplicity of the classical PID controller with the efficiency of model predictive control, while avoiding the tuning problem associated with both, is a viable alternative regulatory controller. The simulation results confirm the transparency of the tuning and the preliminary application on a simple process provides an encouraging proof-of-concept ahead of the full-scale application to a more complicated pilot-scale physical vapor deposition process that is prototypical of 21st century manufacturing.
Publication resulting from this work
Kapil Mukati, Michael Rasch, Babatunde A. Ogunnaike, “An Alternative Structure for Next Generation Regulatory Controllers. Part II: Stability Analysis, Tuning Rules and Experimental Validation, Journal of Process Control, Manuscript submitted.
