Crude Oil Distillation Optimization Using Surrogate-aided Constrained Evolutionary Optimization

Abstract

Evolutionary optimization of a crude oil distillation operation is a time-consuming task. Therefore, this paper proposes a surrogate-aided constrained evolutionary optimization algorithm (SACEO), in which surrogate models' establishment and management are combined to search for an optimal result. By establishing global and local surrogate models with the goal of maximizing profit, an adaptive constrained optimizer is developed for the global and local surrogate models' infilling operations and optimization searching. Thus, time-consuming strict mechanism models can gradually be approximated by continuously updating the surrogate models until the optimized results are obtained. The optimization results obtained for benchmark systems and the crude oil distillation system indicate that SACEO is similar to other constrained optimization algorithms in terms of its optimization accuracy and stability, while the number of evaluations of time-consuming models can be considerably reduced. Thus, the economic efficiency of crude oil distillation processes can be improved while satisfying the production conditions.