This paper addresses a production planning problem that arises in small-scale furniture companies, where the demands and setup times of bottleneck operations are random variables that can be approximated by a discrete and finite number of scenarios that are weighted by their corresponding probabilities of occurrence. The problem is modeled under multiple scenarios via two-stage stochastic programming with recourse. To control the variability of the second-stage costs, we propose a restricted recourse model that generates a set of solutions that are less sensitive to the scenario changes because the variability is limited to a given tolerance. Numerical experiences indicate that, in some situations, risk-averse solutions with good service levels are not excessively expensive to obtain.
Production planning; Furniture industry; Stochastic programming; Risk-aversion; Restricted recourse
