Managing Broader Product Lines through Delayed Differentiation Using Vanilla Boxes

In an attempt to reduce cost while maintaining good customer service, some of the leading manufacturers in the computer industry are delaying product differentiation (by exploiting component commonality) while managing broader product lines. In an environment where demands are stochastic, it seems a good strategy to store inventory in the form of semi-finished products (vanilla boxes) that can serve more than one final product. However, finding the optimal configurations and inventory levels of the vanilla boxes could be a challenging task. In this paper, we model the above problem as a two-stage integer program with recourse. By utilizing structural decomposition of the problem and (sub)gradient derivative methods, we provide an effective solution procedure. A special case, a variant, and several extensions are also discussed. In our computational section, we utilize our model to study several new research issues. We provide insights on the effect of demand variance, correlation, and capacity limitations on the optimal configuration and inventory levels of vanilla boxes and the performance of a vanilla assembly process. In addition, we compare the performance of the vanilla assembly process to make-to-stock and assemble-to-order processes and provide managerial insights on the conditions under which one might be better than the others. Finally, we discuss the characteristics of an IBM product line (which motivated this work) and the effectiveness of a heuristic tailored for that application.Stochastic Multi-Product Assembly Systems, Product Variety, Component Commonality, Demand Correlation, Inventory Management, Stochastic Integer Programs