Mixed-Integer Programming for the ROADEF/EURO 2020 challenge

TL;DR

This paper introduces a novel optimization approach combining constraint generation and cutting planes to effectively solve complex maintenance scheduling problems in the electricity grid, demonstrating significant improvements on real-world instances.

Contribution

It presents a new method integrating constraint generation with innovative cutting planes for nonconvex problems, validated on a real-world electricity grid maintenance challenge.

Findings

Significant improvement in solution quality on challenge instances

Effective handling of nonconvexity through new cutting planes

Applicable to various problems involving uncertainty modeling

Abstract

In this paper, we present a novel approach for solving a maintenance scheduling problem from the French electricity grid company RTE, as presented in the ROADEF 2020 challenge. Our approach combines constraint generation with a new family of cutting planes to address the nonconvexity of the problem. We provide mathematical proofs and separation algorithms for the cutting planes and study the practical impact of our approach on the challenge instances. This method is applicable to a wide range of problems involving the modeling of uncertainty, and is shown to improve results significantly on real-world instances.