The changeover scheduling problem with time and cost considerations: Analytical results and a forward algorithm

In a multiproduct, discrete-item manufacturing environment, it is often more economical to have one ‘flexible’ machine to produce several products rather than to have a dedicated machine for the production of each product. We consider the problem of scheduling multiple products on a single-finite-capacity resource when faced with product-dependent, but sequence-independent, changeover costs and times. The problem horizon is assumed finite and the demand time-varying and known over this horizon. The objective is to find a feasible schedule that meets all demand on time and minimizes the total changeover cost. With the ‘delivery-on-time-or-else’ mentality and the trend toward shorter order-to-delivery lead times, scheduling and managing changeovers has become just that much more important. We first present analytical results to reduce the feasible solution space and then develop a branch-and-bound algorithm to solve this NP-complete problem. Computational results are presented that show that the algorithm can solve realistic size problems in a reasonable amount of time.