We consider a real production planning problem occurring in a glass container manufacturing company. The goal is to minimise the average stock level during a planning horizon of one year, maintaining customers satisfaction and keeping the production at its maximum rate. Besides the production capacity, the main constraints considered are related to the number of setup changes in machines, which is restricted in each factory. The main goal is to find the best balance between the number of setups and the average stock level. We propose a decomposition of the problem into two interrelated problems: a lot-sizing problem and a scheduling problem. For each problem we propose a mathematical model that can be solved using a commercial solver package. To be an efficient managerial tool, the method should provide quickly good solutions. Therefore we solve the lot-sizing problem through a relax-and-fix heuristic and discuss the results.

CEMAT - Center for Computational and Stochastic Mathematics