Abstract
In the semiconductor industry, the capital expenditure of capacity investment is so high that the manufacturers tend to keep their fabs highly utilized. In addition, customer satisfaction has been recognized as an important performance measure especially for foundry companies. Demand fulfillment has a great influence on both capacity utilization and customer satisfaction and therefore has become a critical job. The semiconductor manufacturing process is very complex that each product is required to go through the process of mask creation, pilot run, and qualification so that it could be volume produced. This process will take a long lead time and high mask costs. Furthermore, the masks are not interchangeable between fabs. The above characteristics limit the production capability of fabs and products could not be arbitrarily assigned to fabs for production. Therefore, the NTO allocation affects the future demand of fabs very much. This research aims to develop a demand fulfillment model considering NTO allocation problem. The decisions of product allocation and capacity backup are considered to construct the model. The problem is formulated as both deterministic model and stochastic model to incorporate demand uncertainty for decision making. Finally, a numerical study is conducted to evaluate the performance of the stochastic model and results show that the stochastic model can achieve a lower cost as the demand scenario changes. Also, discussions about the differences of decisions under different level of demand volume are provided.