Hybrid methods using genetic algorithm approach for crude distillation unit scheduling

This paper presents a novel approach for solving special kind of large scale mixed integer linear programming problem with large integrality gap. Virtues of both stochastic and deterministic methods are utilised in order to achieve near global solutions within practical solution time. Two level optimisation method is proposed where upper level optimisation determines key integer decision variables using genetic algorithm and lower level optimisation determines rest of the variables using deterministic method. The capabilities of algorithm were tested with the crude distillation unit scheduling problem. Hybrid approach generated satisfactory solutions with considerably less computational time compared to deterministic methods.