When Constraint Programming and Local Search Solve the Scheduling Problem of Electricite de France Nuclear Power Plant Outages

摘要

The French nuclear park comprises 58 nuclear reactors distributed through the national territory on 19 geographical sites. They must be repeatedly stopped, for refueling and maintenance. The scheduling of these outages has to comply with various constraints, regarding safety, maintenance logistic, and plant operation, whilst it must contribute to the producer profit maximization. This industrial problem appears to be a hard combinatorial problem that conventional methods used up to now by Electricite de France (mainly based on Mixed Integer Programming) fail to solve properly. We present in this paper a new approach for modeling and solving this problem, combining Constraint Programming (CP) and Local Search. CP is used to find solutions to the outage scheduling problem, while Local Search is used to improve solutions with respect to a heuristic cost criterion. It leads to find solutions as good as with the conventional approaches, but taking into account all the constraints and in very reduced computing time.