Use of available storage to improve scheduling in an automobile assembly plant

Abstract

The Final Assembly Plant at General Motors-Holden's Automotive Ltd converts painted body shells into drive-away vehicles. It encounters difficulties when processing certain sequences of vehicles with high work contents, so GMHAL wishes to schedule its input to reduce or eliminate such undesirable sequences. GMHAL has a set of empirical rules for delineating undesirability.

The Painted Body Storage (PBS), which precedes Final Assembly, has 4 lanes that can be used to partially reschedule a vehicle sequence. Information on vehicle work content is available prior to arrival at the PBS, and GMHAL wants advice on using this data and the PBS to achieve a more satisfactory input to Final Assembly.

The Study Group devised three approaches.
1. Use the rules to show which short sequences are desirable and devise input and output strategies for the PBS to achieve these consistently. Choice between the strategies requires further investigation.
2. Model Final Assembly to produce an optimality criterion for vehicle sequences and use combinatorial optimization methods to optimise it over possible PBS outputs. A characterization of these outputs was derived.
3. Suggest that the initial production be suitably scheduled, which may substantially reduce the difficulties at the PBS stage.