Abstract
IBM's capacity optimization planning system (CAPS) is a decision-support system based on linear programming for strategic planning of semiconductor manufacturing capacity. CAPS finds the volume mix of products to maximize profit, constrained by the existing tool capacity, or identifies the tool capacity required to manufacture a specified mix of products. The major challenge was formulating the linear program to reflect the available manufacturing capacity. The formulation has to deal with parallel, unrelated tool groups that may perform the same operation at different rates, to capture the preferential order in which to use such tool groups, and to identify which are the true bottlenecks. Capacity planners use CAPS to reconcile product-demand forecasts with the available or planned manufacturing capacity and to generate requests for capital investments. The manager of site operations credits CAPS with identifying tens of millions of dollars in revenue opportunities and with avoiding significant unnecessary capital expenditures.