(328c) From Process Control to the Enterprise: Contributions of Nonlinear Model Predictive Control

Increased competition has forced the chemical industry to consider on-line optimization and nonlinear model predictive control (NMPC) to enhance profitability while meeting various product/process constraints. Improvements in computing hardware and mathematical programming have made the use of optimization based on a detailed, rigorous first-principles models feasible and realizable. Along with collaborations with Prof. Jay Lee and his group, these optimization applications include:

- model-based optimization has been used to improve the operating recipes of semi-batch polymerization reactors.

- NMPC to track power profiles to track operations of reversible solid oxide cells that generate or consume hydrogen.

- economic NMPC (eNMPC) that performs dynamic optimization of nationwide gas pipelines with robust stability guarantees.

- dynamic real-time optimization for utility networks with uncertain demands and operating conditions.

This talk highlights enabling key advances in problem formulations and solution strategies for NMPC and dynamic real-time optimization that related to input-to-state stability, multi-stage NMPC and eNMPC, and recent developments in infinite horizon control. A detailed presentation of these topics will also be complemented with demonstrations with the applications listed above.