A mixed-integer formulation for back-off under constrained predictive control

The steady-state economic optimum in chemical process plants generally lies at the intersection of constraints. However, in order to maintain feasible operation in the face of disturbances, the steady-state operating point needs to be moved some distance from the constraints into the feasible region. The optimal back-off is a function of the plant dynamics, the type and magnitude of the expected disturbances, and the plant control system. This paper considers calculation of the optimal back-off with regulation under constrained predictive control. The resulting optimization problem is multilevel in nature, and is formulated and solved as a mixed-integer quadratic or linear programming problem for which global optimality is guaranteed. Case studies comprising CSTRs in series and a fluid catalytic cracking unit illustrate the application of the strategy.