Semiconductor capacity expansion based on forecast evolution and mini-max regret strategy for smart production under demand uncertainty

Abstract

Semiconductor industry has continuously migrated for advanced technology nodes with capital intensive capacities to fulfil the demands and maintain competitive advantages. Owing to demand fluctuation and shortening product life cycles, capacity expansion strategy is crucial for maintaining capital effectiveness and profitability. Limitations of the existing capacity expansion approaches can be traced in part to the lack of a framework within which the interrelated decisions can be integrated in light of demand uncertainty along the long lead time for capacity planning, while the forecasted demand changes in the rolling horizon. The research objective of this study is to develop a rolling horizon capacity expansion model based on forecast evolution with minimax regret strategy (MMR) to reduce the risks of capacity shortage and surplus to improve capacity utilization and capital effectiveness. In particular, this study applied both additive and multiplicative martingale models of forecast evolution (MMFE) to generate sets of possible demand scenarios for minimax regret capacity expansion model. To estimate the validity of the proposed approach, different types of demand scenarios based on realistic data from a leading semiconductor company in Taiwan are employed. A simulation model that integrates mathematical programming model is constructed using realistic demand scenarios to compare the performance of the proposed approach with conventional approaches. The results of the proposed MMFE + MMR approach have shown practical viability to provide a more robust capacity strategy to reduce supply chain risks while integrating the collaborative planning decisions for smart production.