Orchestrating constrained programming and local search to solve a large scale energy management problem

This paper presents a heuristic approach combining constraint satisfaction, local search and a constructive optimization algorithm for a large-scale energy management and maintenance scheduling problem. The methodology shows how to successfully combine and orchestrate different types of algorithms and produce competitive results. The local search for production assignment is a simple yet optimal solution for the relaxed initial problem. We also propose an efficient way to scale the method for huge instances. A large part of the presented work is done to compete in the ROADEF/EURO Challenge 2010, organized jointly by the ROADEF, EURO and the Électricité de France. The numerical results obtained for the official competition instances testify about the quality of the approach. The method achieves 3 out of 15 possible best results.