Decision models for capacity planning in a regeneration environment

Abstract

This paper presents an approach to capacity planning and coordination for the regeneration of complex investment goods. In order to capture the specifics of the regeneration environment, Bayesian networks are utilised to improve the accuracy of the workload forecast and mathematical models are proposed for the long-, medium- and short-term capacity planning and coordination. The long- and medium-term models maximise the total profit through the optimum determination of the quantity of goods to be regenerated in-house or at the sites of external vendors, the number of regenerated goods to be stored and the extent to which penalties should be tolerated for delayed regeneration orders. The short-term model determines the optimal allocation of human and machine resources to different orders. The models are validated through the use of real-world data supplied by an industrial partner. Sensitivity analyses are conducted in order to gain insights into the model. The application of the models leads to significant improvements for the industry partner due to the reduction in regeneration costs as a result of implementing varying plant capacities over the course of a year.