On Solving Stochastic Production Planning Problems via Scenario Modeling

The problem of deciding how much capacity to acquire and how to plan its utilization is a strategic problem that deserves careful analysis. Scenarios are used to model the multistage decision process and the recourse actions available to the decision maker are introduced. In general, the Simplex method performs better on the compact representation and the IP methods perform better on the splitting variable representation. The most appealing aspect of stochastic programming models is the idea of recourse, i.e, modeling what future adjustments or options the decision maker may have and optimizing with respect to them. Based on our experimentation, the computational effort required by the IP method with the predictor-corrector modification for the splitting variable representation is comparable with the computational effort required by the Simplex method along with our crash procedure, applied to the compact representation. The size of these models grows exponentially with the number of decision stages.