Abstract
The focus of this study is to compare the performance of deterministic and stochastic approaches for production planning problem under various degrees of uncertainties in customer demand and production yield. In an uncertain environment, the degree of uncertainty affects how the manager to model uncertain future system values. In this study, there are two modeling methods to represent the uncertain system values of demands and yields. Normally, an uncertain future system value can be assumed as a random variable. For a random variable with small variance, to improve computation efficiency and simplify mathematical formulation, a constant value of the expected value for the random variable is used in the first modeling method, which is called deterministic approach. The second method, called stochastic approach, adopts scenario-based forecasting method that generates a number of scenarios with assigned probabilities. Since, in this study, there are two modeling methods for two types of uncertain system values, future demands and yields, totally four approaches are used to solve the uncertain production planning problem. They are deterministic-yield deterministic-demand approach (DDA), deterministic-yield stochastic-demand approach (DSA), stochastic-yield deterministic-demand approach (SDA), stochastic-yield stochastic-demand approach (SSA). This study compares their long-term effectiveness under rolling horizon practice by using discrete-event simulation experiments. According to the experiments, DSA outperforms the other approaches.