Large-Scale Refinery Crude\nOil Scheduling by Integrating\nGraph Representation and Genetic Algorithm

Abstract

Scheduling is widely studied in process systems engineering\nand\nis typically solved using mathematical programming. Although popular\nfor many other optimization problems, evolutionary algorithms have\nnot found wide applicability in such combinatorial optimization problems\nwith large numbers of variables and constraints. Here we demonstrate\nthat scheduling problems that involve a process network of units and\nstreams have a graph structure which can be exploited to offer a sparse\nproblem representation that enables efficient stochastic optimization.\nIn the proposed structure adapted genetic algorithm, SAGA, only the\nsubgraph of the process network that is active in any period is explicitly\nrepresented in the chromosome. This leads to a significant reduction\nin the representation, but additionally, most constraints can be enforced\nwithout the need for a penalty function. The resulting benefits in\nterms of improved search quality and computational performance are\nestablished by studying 24 different crude oil operations scheduling\nproblems from the literature.