Evolutionary Algorithms for Controller Tuning of Tert-Amyl-Methyl-Ether Reactive Distillation

Abstract

Efficient tuning of the coefficients used by proportional-integral-derivative (PID) controllers enhances their performance. For highly non-linear systems, optimization algorithms are required to make the PID controllers more responsive to disturbances. The production of tert-amyl-methyl-ether (TAME), an essential additive for gasoline, in reactive distillation columns integrates highly non-linear reaction and separation processes. On the other hand, TAME distillation is an azeotrope distillation process, therefore non-linearity of this process is more complex than that of conventional distillation. PID-controller tuning methods applying a genetic algorithm (GA) and a particle swarm optimization (PSO) algorithm are compared using a dynamic simulation that integrates the optimization algorithms with the HYSYS process simulator. The PID controller response trends are analyzed following the introduction of a significant disturbance to the TAME reactive distillation column (i.e., a ten percent change in the methanol feed temperature). The PSO PID controller tuning method that minimizes the integral of the absolute error (IAE) as its objective function significantly outperforms the GA tuning method. The novel PID-tuning methodology developed has more extensive application potential.