A new relaxation technique based on fractional representation to solve bilinear models: Application to the long horizon crude oil scheduling problem

This paper proposes a novel relaxation technique based on the fractional representation of bilinear terms. This technique is embedded into an iterative two-step MILP-NLP algorithm based on piecewise relaxation and domain reduction strategies. To evaluate the performance of the algorithm, it is compared to the recently addressed iterative MILP-NLP algorithm based on piecewise McCormick relaxation techniques over a variety of instances. Our method is also applied to the crude oil scheduling problem as an application. The results confirm the efficiency of the proposed algorithm from both solution quality and running time