Comparison of Simple Feedback Control Structures for Constrained Optimal Operation

Abstract

Optimal operation of systems subject to constraints can be challenging, especially when the set of active constraints changes during operation due to disturbances. The use of traditional model-based real-time optimization strategies has limitations related to model-plant mismatch and computational effort, and therefore the use of simple feedback strategies in lower layers is a good alternative to avoid these issues on fast timescales. This work aims to evaluate the viability of region-based control structures using selectors and the use of a primal-dual feedback optimizing control framework on this kind of problem, through the study of two process systems with changing active constraints. While region-based strategies focus on the effective control in specific active constraint regions, the primal-dual approach allows for control in all possible regions, with the introduction of intermediate variables that estimate the Lagrange multiplier values. In the first case study, the traditional region-based strategy could not handle all constraint regions, and an additional logical switching was necessary to account for the remaining region. The implementation of primal-dual feedback optimizing control was flexible enough to control the system in all regions without the need for additional logic. The second case study presents more constraints than the first and increased nonlinearities, which makes finding controlled variables for the unconstrained degrees of freedom challenging. The primal-dual control framework was able to drive the system to the optimum in all considered regions. Therefore, this framework is deemed as a promising control structure for optimal operation in the presence of changing active constraints.