A Robust, Computationally Efficient Methodology to Set Service Levels for Components in Assemble-to-Order Environments

May 31,2002 We consider a situation where end-items are assembled from a set of components in an assemble-to-order fashion. These components are either made-to-stock or procured from outside vendors. We assume that the component inventories are maintained using independent one-for-one replenishment policies. Components can either be unique to a specific end-item or common across several end-items and can also be differentiated on the basis of their costs and their replenishment lead-times, which are assumed to be deterministic. We assume that the arrival processes for the end-items are independent renewal processes. We also assume that there is a constraint on the system-wide safety stock. Under such a setting there is no prevailing methodology to set the base stock levels for the component inventories. In this paper we address this issue by formulating an appropriate optimization problem. The solution to this optimization problem provides us with the base stock numbers for the components. We compare the quality of our solutions to the quality of the solutions provided by an equal allocation policy through a simulation study for a set of sample problems. For the sample problems studied our method outperforms the equal allocation policy under a wide variety of circumstances lending to the viability and robustness of our approach.