Context and project

In the context of designing ultra-flexible factories for M.F.P. Michelin, we need to locate different machines on the factory floor at the lowest cost. Given a set of machines and flows of materials between them, the problem of finding an optimal layout - known as the Facility Layout Problem (FLP) - consists in positioning and orienting the machines in a production plant such that the total cost of 1) the machines’deployment and 2) the machine-to-machine supplies transportation, is minimum. FLP is one of the most challenging problems encountered in plant optimization and has several known variants including:
1. uncertainty in the inter-machine flows,
2. constraints on the underlying placement structure (e.g. single line, multiple lines, grids),
3. different types of machines, having their own capabilities and restrictions.

Objectives

After a review of the literature on FLP, several models corresponding to different variants will be considered and cast as nonlinear optimization models. A study of these models will be performed based on the most efficient and recent methods of combinatorial optimization and linearization techniques. This theoretical work will lead to the development of a real-life optimization framework combining exact approaches (e.g. branch-and-bound, cutting planes, symmetry management) and approximate techniques (e.g. heuristics, metaheuristics). The resulting methodology will then be implemented and tested on real-life instances provided by M.F.P.Michelin.

Candidate background

The candidate should hold a Master degree in Computer Science/Applied Mathematics/Industrial Engineering or equivalent with good skills in applied mathematics in relation to optimization and operations research. The candidate should also like programming and be willing to learn CPLEX

More informations below