【Abstract】  This paper established a mathematical model with nonconvex bilinear terms. It formulated the complex material flow in the petroleum refinery scenario based on the concept of the “P model”. The mathematical model described the nonlinear constraints such as linear and nonlinear mass and volume intersection flow blending of crude and middle material physical properties. Additionally, it described the complex inflow and outflow in secondary devices as nonlinear constraints such as delta base structure and physical property transfer. It is highly difficult to determine the direction and quantity of each material in the network of refineries. An improved interior point algorithm with an initial point strategy was proposed to find a high-quality feasible solution in a short time. The real instances from the petroleum refinery were employed to compare and analyze the solutions from the improved algorithm and commercial solver. The experimental results show that the proposed algorithm framework can balance the solution quality and computational efficiency and perform well in different scenarios of refinery material flow networks.