Dynamic optimization and nonlinear model predictive control of a semi-batch epoxidation process

Abstract

We investigate offline dynamic optimization (ODO) and multi-layer control schemes to optimally operate the epoxidation of oleic acid. The process involves a two-phase reaction mixture with highly exothermic reactions wherein the reactor temperature needs to be controlled to guarantee safe operation. Optimal operating conditions are computed using ODO and the control scheme is responsible to enforce such conditions. We compare a conventional control scheme, which uses PI controllers, against two multi-layer control schemes: (i) using nonlinear model predictive control (NMPC) and (ii) using dynamic real-time optimization (DRTO) together with NMPC. With ODO and NMPC we achieve a reduction in process time greater than 75%, compared to a standard process, while still maintaining safe operating conditions. With DRTO and NMPC, the same reduction in process time is obtained, however with a maximum violation of the temperature constraint by 12%. This may be due to model-plant mismatch and large computational time required for the optimization.