Abstract
Capacity planning is the calculation of the required numbers of machines for production. Due to the demand uncertainty, capacity planning is becoming more and more difficult. This study focuses on multiple periods, multiple product types, multiple machine types, and demand uncertainty capacity planning problem, and uses two approaches to solve the problem. The object function of the approaches is to minimize the backorder cost and machines purchasing cost. The first approach formulates the problem as stochastic programs. The model assumes there are several demand scenarios, and the probability of each scenario is known. The second approach formulates the same problem with known demand by mixed integer programs, and then uses decision tree to deal with the demand uncertainty. Our experiment compared object values from stochastic programming and expected costs from decision tree analysis with the expected value with perfect information（EVwPI）under different numbers of scenario. According to the experiment results, the stochastic programming approach is superior to decision tree analysis.