Sequence-dependent scheduling at Baxter International

We solved Baxter International's IV (intravenous) systems production scheduling as a sequence-dependent scheduling problem. Unlike problems previously studied, Baxter's production line has a large number of components and parts (IV systems) with a nonrepetitive schedule. We compared cluster analysis and mathematical programming as alternative methodologies to solve the scheduling problem. Mathematical programming produces solutions that do better in optimizing setup times. Cluster analysis produces results that managers prefer over mathematical programming because cluster analysis reduces resource requirements relative to mathematical programming. Baxter's implementation of the scheduling model based on cluster analysis on four production lines reduces setup times and saves $165,000 annually.