A two-stage stochastic program for scheduling and allocating cross-trained workers

Abstract

A two-stage stochastic program is developed for scheduling and allocating cross-trained workers in a multi-department service environment with random demands. The first stage corresponds to scheduling days-off over a time horizon such as a week or month. The second stage is the recourse action that deals with allocating available workers at the beginning of a day to accommodate realized demands. After the general two-stage model is formulated, a special case is introduced for computational testing. The testing helps quantify the value of cross-training as a function of problem characteristics. Results show that cross-training can be more valuable than perfect information, especially when demand uncertainty is high.