James B. Rawlings
Chemical and Biological Engineering
Chemical and Biological Engineering Professor James Rawlings' work on model-based predictive control has been described as "unusually broad and deep, extending across the entire spectrum from the shedding of light on the principles of model-based predictive control to innovations in education and drastically improved industrial practice." This award, in particular, cites two papers: "Model predictive control with linear models," published in the AIChE Journal, and "Stability of constrained receding horizon control," printed in IEEE Transactions on Automatic Control. Through this work, Rawlings and his PhD student Ken Muske contributed a groundbreaking theory, which allowed enormously improved and effective computer control algorithms to be developed and applied to processes with serious operating constraints.
All industrial processes have constraints on both actuators (valves full-open or full-shut, etc.) and process variables (temperatures, pressures or chemical concentrations, etc.). Before Rawlings' work, it was difficult to precisely control both types of constraints and also difficult to know if the actuator limitations would even allow operation at a desired set of temperatures, pressures or material concentrations. The best one could do was carry out extensive experimental studies or many numerical studies using simulations. If the process was inoperable with the current constraints, it was not possible to easily determine how to make investments to improve the actuator capabilities without extensive trial and error simulations. Unfortunately, even the results of these simulations did not give the structure of the constraint issue.
By contrast, his group's theoretical results show clearly the effect of constraints on process stability and operability — and thus how to select the best actuator range improvements to make an inoperable process operable. The results also allow one to design controllers for constraints on process variables so that safe and effective process operation is possible.
Rawling's results made possible a systematic control-system design approach, which can increase process safety, improve product quality, and lower production costs.
Rawlings developed an industry short course for control engineers and both Exxon and Eastman Chemical adopted his results as the theoretical basis for their standard model-predictive control designs. Furthermore, Aspen Technology, the dominant international vendor of computer control systems for the process industries, created a new generation of control algorithms based on Rawlings' papers.