Abstract
Phenomena like overcapacity and declining sales prices have damaged the earning potential of semiconductor manufacturers. The increasing pace of product innovation also places a premium on the ability to effectively plan the production of a fab. Under such an uncertain environment, a fuzzy production planning model is constructed in this study to serve as a robust production planner for a wafer fabrication plant. The incorporation of the fuzzy set theory is to facilitate the treatment of uncertain factors. Fuzzy concepts or techniques applied in this study include the specification of parameters in triangular or trapezoidal fuzzy numbers, the adoption of fuzzy comparison in dealing with the fuzzy objective function or expanding inequalities, projecting outputs by the fuzzy dynamic production function, and defining the satisfaction level of uncertain demand. Besides, a simple way to calculate fuzzy interpolation is proposed to catch the trend in price. The improvement of yield is better described by an expert-opinion-referenced fuzzy learning model. A systematic procedure is also proposed to provide better estimates for various time measures. All these methods are based on fuzzy concepts and helpful in the preparation of fuzzy-valued inputs to this model. The FLP model is at last converted into an equivalent LP model solved by any existing LP software.Finally, a practical example of a DRAM fab is used to demonstrate the proposed methodology. Based on it, both the fuzzy production planning model and Leachman's model are constructed for comparison. According to the results, the application of this methodology has at least the following advantages:(1) It needs less time to obtain the optimal solution.(2) The best or worst performance of the fab during the planning horizon, under present operating conditions, can be simultaneously estimated.(3) A more flexible release schedule and a more precise finish schedule can be provided.(4) This model is a more precise description of the fab than traditional models.