Many-objective optimization for scheduling of crude oil operations based on NSGA-Ⅲ with consideration of energy efficiency

Abstract

Abstract Crude oil operations in refineries are characterized as a hybrid system since it contains both discrete-event and continuous processes, and it is extremely difficult to schedule such a system. For scheduling such a system, initially the discrete tasks to be performed during the scheduling horizon is unknown such that heuristics and meta-heuristics are not directly applicable, which further complicates its scheduling problem. Moreover, there are large number of objectives to be optimized, including the minimization of energy consumption due to that crude oil operations consume large amount of energy and therefore lead to large amount of emissions. Furthermore, the energy optimization problem is characterized as highly non-linearity. Hence, the scheduling problem of crude oil operations in refineries belongs to the many-objective optimization problems and it is extremely challenging. This paper addresses this challenging scheduling problem of crude oil operations. This scheduling problem is first converted to a discrete dynamic resource allocation problem such that meta-heuristics can be applicable. Then, with the results of large number of experiments, this work innovatively proposes an NSGA-Ⅲ-based optimization algorithm to efficiently solve the problem for Pareto-optimal solutions. By the proposed method, the genes in a chromosome are generated one by one and, when generating a gene, safeness check is done according to the derived safeness conditions such that each gene is feasible. In this way, the schedule feasibility can be ensured. An industrial case study is given to test its performance and comparison is made with the existing state-of-the-art algorithms for many-objective optimization problems. The results show its good performance in terms of convergence, solution diversity, time efficiency, and its applicability to real-life refinery scheduling problems.