We developed a novel control strategy for an important class of plant-wide control problems via integration of linear and nonlinear model predictive control. The control strategy is formulated by decomposing the plant into linear and nonlinear subsystems. A model predictive control system for the decomposed plant is derived by applying linear control to the linear subsystems and nonlinear control to the nonlinear subsystems. Our initial work focused on basic design issues and application of the proposed method to a simple flowsheet comprised of a highly nonlinear chemical reactor and an approximately linear distillation column. We also developed more systematic plant decomposition and sequential controller solution procedures via application to a complex styrene production flowsheet. The sequential solution method was shown to be nominally stabilizing for nonlinear plants with a triangular structure.

Funding: National Science Foundation (CTS-9501368) and DuPont

Student: Guang-Yan Zhu (Ph.D.)

Collaborator: Prof. Babatunde Ogunnaike (Delaware)

Publications:

1. Henson, M. A. "Nonlinear Model Predictive Control: Current Status and Future Directions," Computers and Chemical Engineering, 23, 187-202 (1998). [PDF]
2. Zhu, G.-Y., M. A. Henson, and B. A. Ogunnaike, "A Hybrid Model Predictive Control Strategy for Nonlinear Plant-Wide Control," Journal of Process Control, 10, 449-458 (2000). [PDF]
3. Zhu, G.-Y and M. A. Henson, "Model Predictive Control of Interconnected Linear and Nonlinear Processes," Industrial Engineering and Chemistry Research, 41, 801-816 (2002). [PDF]