A hybrid solution strategy for production planning

Production planning with lot-sizing and scheduling in industries where the cleaning times depend on the sequence of the products can be modelled as a mixed integer problem. However, this problem is very hard, and even small instances cannot be solved exactly in a reasonable time. We propose a division of the whole production planning problem into its two subproblems, and to solve each them independently: first the lot-sizing, then schedule the operations that were determined for each machine, for each period. In this strategy, the lot-sizing part does not take into consideration the changeover times. After scheduling the operations, if the cleaning times make a solution infeasible, a constraint cutting this solution must be added. The lot-sizing problem must be resolved, and its solution rescheduled, until obtaining feasibility. We propose an algorithm for integrating lot-sizing and scheduling, and analyse the results obtained.