Abstract

Any reasonable production planning contains three important decisions on lot size, lead time, and capacity. The common approach in the literature is to divide the planning problem into lot sizing, lot sequencing, and lot splitting sub-problems. Very few studies, to the best of our knowledge, have been conducted on the interdependencies and three- way interaction of lead-time, lot size, and actual capacity usage. A particular lot size calculated by the sub-problem method, however, will likely yield an infeasible solution or at least result in schedule instability (nervousness). This is just because in most capacitated lot sizing models, the capacity constraints in the model only take into consideration the available time on each work station, ignoring the sequencing of lots, sublot sizes, and their effect on makespan and lead times. In this thesis we bridge the gap between lot sizing and scheduling in flowshops, and examine the use of the lot splitting and sequencing techniques to reduce schedule instability. A mixed integer programming formulation is presented, which enables us to simultaneously find the optimal lot sizes as well as the corresponding sublot sizes and sequence of jobs. With this model, small size problems can be solved within a reasonable time. The computational results confirm that this model can be advantageous in dampening the scheduling nervousness. For larger size instances, a Genetic algorithm is proposed.