An Optimization Framework for Solving a Large Scale Scheduling Problem

The base of the model is a simplification of a general N-dimensional allocation formulation in [3]. The Just-In-Time (JIT) principle is considered and is achieved by a piecewise linear objective function penalty, for each order, penalizing late and early orders. The rolling horizon implementation used a time window with 20 groups of orders, which corresponds to a 3 week schedule. The framework has been applied for production scheduling at a Finnish supplier of packaging materials based on ethylene and polypropylene. The studied plastic producing company had over 3000 orders in 1 year to be scheduled on 4 of their printing machines. However, only a group of two machines is considered at one time, due to the similarity of these machines. In paper the scheduling model in GAMS implementation is presented and, furthermore, the rolling horizon implementation is described.