Non-dominated sorting-based strategy for optimizing the mixture of initiators in polyethylene reactor

Abstract

Multi-objective optimization (MOO) of low-density polyethylene (LDPE) production in a tubular reactor is performed for three problems with three different objectives. For the first problem, the objective is maximization of productivity and minimization of cost of initiators. For the second problem, the objective is maximization of conversion and minimization of cost of initiators. While for the third problem, the objective is maximization of productivity, minimization of cost of initiators, and maximization of conversion. An inequality constraint on reactor temperature is also enforced to prevent the tubular reactor from a runaway condition. The non-dominated sorting–based strategies are utilized to tackle the optimization problem with Aspen simulator as model-based optimization for LDPE production in a tubular reactor. The strategies are non-dominated sorting genetic algorithm II (NSGA-II), non-dominated sorting grey wolf optimizer (NSGWO), and non-dominated sorting whale optimization algorithm (NSWOA). The inputs for MOO decision variables are mass flowrates of tert-butyl peroxypivalate (TBPPI), tert-butyl peroxyacetate (TBPA), tert-butyl 3,5,5 trimethyl-peroxyhexaonate (TBPIN), and tert-amyl peroxyacetate (TAPA). Performance matrices like hypervolume, spacing, and pure variability are examined to choose the most effective MOO approach. Findings showed that the NSGWO is the most effective MOO approach due to the discovered solution set providing the most precise, diverse, and appropriate in the homogeneity allocation points along the Pareto front (PF). The highest productivity, lowest cost of initiators, and highest conversion obtained by NSGWO are 549.369 Mil. RM/year, 7.5589 Mil. RM/year, and 31.685%, respectively.