A New Model and Algorithm Design of Pooling Problem

The pooling problem is a non-convex optimization problem with applications in petroleum industry. It is typically modeled as a bilinear programming problem with linear objective and bilinear constraints. Considering the factors of oil volume and mass, we propose a trilinear model for the pooling problem in this paper. Based on a new non-monotone technique, we put forward an improved successive linear programming algorithm and establish the global convergence of the algorithm. Moreover, the numerical experiments show that the proposed trilinear model outperforms the conventional bilinear model and that the non-monotone technique is effective for solving the large-scale pooling problems.