Finite Capacity Production Planning with Random Demand and Limited Information

Abstract Production planning has a fundamental role in any manufacturing operation. The problem is to decide what type of, and how much, product should be produced in future time periods. The decisions should be based on many factors, including period machine capacity, pro¯t margins, holding costs, etc. Of primary importance is the estimate of demand,for manufacturer's products in upcoming periods. Our focus is to address the production planning problem by including in our models the randomness that exists in our estimates for future demands. We solve the problem with two variants of Monte Carlo sampling based optimization tech- niques, to which we refer as \simulation based optimization" methods. The ¯rst variant assumes that we know the actual demand,distribution (assumed to be con- tinuous) with which we approximate the true optimal solution by averaging sample estimates of the corresponding expected value function. The second approach is use- ful when we have limited information about the demand,distribution. We illustrate the robustness of the approach by comparing a three mass-point approximation of the continuous distribution to the results obtained using the continuous distribution. This second approach is particularly appealing as it results in a solution that is close to optimal while being much faster than the continuous distribution approach.