A branch and bound network algorithm for interactive process scheduling

Abstract

A multi-facility multi-product production scheduling problem is considered in terms of a general class of process unit operations scheduling problems which are common in the refining and chemicals processing industries. A generalized network formulation is used to model the conversion of unit processing capacity to finished products. A specialized branch and bound algorithm is used to enforce the restriction that only one operation can be run per unit at any given time. The algorithm minimizes total costs, which consist of unit operating costs, processing costs, inventory holding costs, setup and changeover costs. A procedure is developed by which the setup and changeover costs are used to estimate bounds for the network model in the branch and bound algorithm. All other costs are incorporated in the network formulation. It is shown that the algorithm is more efficient in those problems for which the setup and changeover costs are small, or in problems in which a lower bound for the setup and changeover costs can be accurately estimated. The implementation of the algorithm in an interactive process scheduling system is discussed in terms of the human engineering factors involved.