Assessing the cost of robust capacity allocation for serving dynamic customer demand

Abstract

Long-term production capacity planning has to anticipate future shifts of customer demand. The resulting gap between currently utilized production capacity and available capacity determines the robustness of a production network. In this paper a new approach for assessing the costs of a robust capacity allocation is presented. To this end production networks are modeled by multiclass queueing networks that can be approximated by fluid models. The stability radius reflects the smallest shift of customer demand that destabilizes the network and is used as a measure for robustness. Based on findings for this measure an optimization problem for a cost-oriented capacity allocation is set up.