Linear programming-based approach to scheduling of crude oil operations in refinery with oil type mixing requirement

With both discrete-event and continuous variables, the process of crude oil operations in a refinery is characterized as a hybrid system. The scheduling problem of crude oil operations is naturally combinatorial and challenging. Moreover, with oil type mixing requirement, it may involve nonlinear constraints. No software tools have yet been designed for industrial use due to its difficulty. By studying the problem in a control-theoretic perspective, our prior work proposed a two-level architecture and derived schedulability conditions. For cases where crude oil mixing is required, this work studies the scheduling problem at the upper level to maximize the productivity and process effectiveness in handling different types of oil by distillers. Based on the schedulability conditions, a linear programming-based approach is presented. By this approach, the discrete event and continuous variables are decoupled and the main computation involved is to solve linear programming problems. Thus, it is very efficient and applicable to industrial-size problems.