Preventive crude oil scheduling under demand uncertainty
Debashish Panda (Prof. Manojkumar Ramteke)
18th May 2017, Thursday, 04:30‐05:00 pm (25+5 mins), ChE Seminar hall
Crude oil scheduling is an important aspect in refinery operation as it enables the efficient processing of the crude which increases the profitability. The problem often becomes challenging due to the presence of combinatorial constraints, discrete variables and uncertainties. A study has been carried out to addresses the crude oil scheduling under commonly present demand uncertainty. For this, a proactive two-stage approach has been developed to solve such optimization problems. A discrete-time model is used with structure adapted genetic algorithm (SAGA) for solving single- and multi-objective scheduling optimizations. In the first stage, an initial schedule is generated with the nominal parameters provided a priory. The same schedule is then checked for robustness in order to incorporate demand uncertainty. This method is applied to three different industrial size problems with periods of 8, 21 and 42. The objective function in the single objective formulation is maximization of profit. However, in multi-objective optimization, an additional objective of minimization of fluctuation in crude oil supply to crude distillation units is used as it provides a better control and operability of the plant. The proposed method is capable of generating robust schedule which works for demand uncertainty in all three different types of crude oil scheduling problems.