Multi-objective stochastic scheduling of job ready times

A fundamental scheduling problem is to determine a production start (ready) time based on customer-specified due dates. Typically, the objective is to delay the ready time in an attempt to minimize work-in-process inventory and maximize production system utilization. In many practical situations, highly variable service times complicate this problem. In such a case, the ready time implies a level of on-time completion confidence for each job. As the ready time increases, the on-time confidence decreases. This paper investigates the ready time/job confidence level tradeoff. A multi-objective model balances the ready time and confidence level maximization goals. The model involves combinatorial and numerical optimization and has an exceptionally complex state space. In view of these complexities, we investigate a pairwise interchange heuristic and a genetic algorithm search solution. Experimental results support solution through a process involving both the heuristic and the genetic algorithm.