LAUSR

We consider the problem of determining an optimal release schedule for a production facility as it transitions from producing an older product to a newer one. We use simple queueing ideas to model the impact of the new product introduction on the throughput rate and cycle time of both products. We then incorporate the learning gained from accumulated experience in producing the new product to formulate a deterministic optimization model for release planning during product transitions. Since the model is nonlinear and non-convex we use a genetic algorithm to obtain near-optimal solutions. The structure of the optimal solution provides insights into the relationships between the different costs involved in the model. Computational experiments show that the release planning model produces high-quality solutions in reasonable CPU times, and reflects the behavior of the system realistically. The principal insight is that careful management of releases during product transitions has the potential to mitigate many of the adverse effects frequently observed in practice such as longer cycle times for both old and new products.