Employing economic model predictive control for improving efficiency of the batch reactor carrying out decomposition of the Di-Cumyl-Peroxide

Abstract

Background:

Batch reactors are used to produce agro-herbicides, often operating under metastable conditions. Exothermic reactions within these reactors can rapidly release large amounts of chemical energy, increasing the reaction rate and risking a runaway excursion. To prevent excessive heat dissipation, the reactor must operate within a lower mesophilic temperature range. The decomposition of dicumyl-peroxide is a notable example of a runaway reaction, characterized by multi-order thermo-kinetics with an Arrhenius temperature dependency. This decomposition rate is highly sensitive to factors such as Damkohler’s number, exothermic enthalpy, oxygen presence, substrate concentration, and initial temperature variations. Many kinetic parameters are not directly measurable in real-time applications.

Method:

(i) An economic stage cost is derived and integrated into a model-predictive-control (MPC) law to regulate coolant flow and manage exothermic heat dissipation, mitigating unmeasured disturbances and faults. (ii) For estimating unmeasured reactor states and uncertainties, a cubature Kalman filter with a singular-value-decomposition approach has been utilized. (iii) Event-triggered scheduling combined with cubature Kalman filter, reduced communication resource usage.

Significant findings

: The economics MPC controller tested on the dicumyl-peroxide system is compared with a standard nonlinear MPC rule, considering factors like cooling period, decomposition rate, and observer accuracy which ensures the effectiveness of the proposed algorithm.