Scheduling of Tests on Vehicle Prototypes Using Constraint and Integer Programming

Abstract

We address a scheduling problem in the automobile industry where hundreds of tests must be performed on vehicle prototypes. The objective is to minimize the number of required prototypes, while all tests are executed with respect to several kinds of constraints. We apply Constraint Programming (CP) to solve the complete problem. Then to complement CP in finding a good lower bound, we separately solve an Integer Programming (IP) model which is simplified to cope with the large scale of the problems involved. This model is based on two main principles: set covering to select prototypes with suitable components for all tests; and energetic reasoning to determine the number of prototypes definitely required over the time intervals between the distinct values of the release and due dates. Computational results show that CP can achieve good feasible solutions as confirmed by the results obtained from solving the simplified IP model.