| In response to consumer demands, the range of products offered by manufacturers is becoming increasingly complex. One of the ways that companies are dealing with this challenge is to offer products that can be assembled from a collection of independent components. One of the central difficulties facing the makers of such 'configured-to-order' products is the following: The precise quantity and collection of resources required to satisfy an order is not known until a customer has actually placed an order. The supply chain tools currently available to practitioners are not prepared to deal with such uncertainty and existing academic models do not adequately address the problem in a practical manner. This thesis responds to this challenge by addressing a series of optimization problems and framing them within the context of a popular business process called sales and operations planning.; Within this context, three related optimization models are formulated and solved using stochastic programming techniques. These models are referred to as the explosion, implosion and component rationing problems, respectively. The explosion and implosion problems both seek to maximize profit, but where the explosion model determines component requirements given product demand, the implosion model determines the appropriate product sales targets given certain restrictions on component supply. The component rationing problem seeks to maximize revenue for a given component supply and product demand by appropriately setting the component threshold levels that will reserve certain components for certain product orders. In all three problems, uncertainty is present because the quantity of components required to satisfy an order is unknown at the time that the decisions are made.; Computational studies performed using problem sets derived from data provided by IBM show that, with respect to the explosion and implosion problems, there is significant benefit to accounting for the uncertainty associated with how products are configured. For the explosion problem, accounting for this uncertainty results in improved profit, revenue and serviceability. For the implosion problem, accounting for this uncertainty results in improved serviceability. As for the component rationing problem, it is shown that rationing is often able to increase expected revenue over a first-come-first-served allocation policy. |
