3 présentations

• 
  13h30 - 13h55

  Scheduling in some reconfigurable manufacturing systems

  • Fayez F. Boctor, prés., Université Laval

  A new type of manufacturing systems, called reconfigurable manufacturing systems (RMS), is proposed to allow combining the flexibility of flexible manufacturing systems (FMS) and high capacity of dedicated manufacturing lines (DML). Reconfigurable manufacturing systems are designed to admit several configurations that allow the production of a family of parts that can be divided into subfamilies. Each subfamily requires one of the possible configurations and, often, moving to manufacturing a new subfamily requires to reconfigure the system. Reconfiguring the system requires a major sequence-dependent setup while manufacturing parts within a same subfamily requires no or minor sequence-independent setups. This talk addresses scheduling issues in reconfigurable flow shops, in reconfigurable flow lines and in reconfigurable parallel flow lines. Three versions of the reconfigurable flow shop scheduling problem are considered. Mathematical formulations as well as heuristic solution methods will be presented for the reconfigurable flow shops, the reconfigurable flow lines, and the reconfigurable parallel flow lines. The presentation reports on a numerical experiment undertaken to evaluate the proposed heuristic solution methods. Some industrial settings will be presented to help illustrating the specificities of the discussed problems.

• 
  13h55 - 14h20

  Solving the Job Shop Scheduling Problem with the TitanQ Quantum-Inspired Solver

  • Ali M'hammedi Alaoui, prés., InfinityQ Technology Inc.

  Solving the Job Shop Scheduling Problem (JSSP) is one of the most challenging NP-Hard Combinatorial Optimization problems. It serves as a key component in production systems, supporting process optimization and assisting in cost minimization. While there are many approaches in literature proposed to solve the JSSP, most of these approaches are mainly metaheuristics or Mixed-Integer Programming solvers like Gurobi known to be limited in performance and scaling. Quantum computers and quantum-inspired solvers serve as an alternative that could tackle previously unsolvable optimization tasks if effectively formulated in the Quadratic Unconstrained Binary Optimization (QUBO) method. Venturelli et al. (2016) proposed a hybrid approach to solving the JSSP using QUBO and classical improvement methods. We introduce an approach relying on formulating the problem as QUBO encoding both the constraints and the objective function in one single expression. This approach has the benefit of adding more constraints and objective terms to the QUBO rather than altering the logic of the algorithm. Experimentations using this approach are done on a set of JSSP instances available online using the TitanQ quantum-inspired solver system developed by InfinityQ Technology.

• 
  14h20 - 14h45

  Vessel scheduling in restricted waterways

  • Marie-Sabine Saget, prés., FSA-Laval University/CIRRELT
  • Jacques Renaud, Université Laval, CIRRELT
  • Maryam Darvish, Université Laval

  This paper introduces a time-discrete framework for managing ship traffic in restricted waterways. We partition these waterways into multiple segments to accommodate varying width restrictions and time-indexed water level fluctuations over the planning horizon. Using integer linear programming, we solve the model considering various instances to assess the impact of channel restrictions with exact methods. We then focus on evaluating the computational performance of our approach using data from the Traverse du Nord. This narrow channel, located near Quebec City, is part of the St. Lawrence River system, linking the upper Midwestern United States and the Canadian provinces of Ontario to the rest of the world. It is subject to tide fluctuations, which constrains the timing and size of ships able to navigate the channel.

  Our results indicate that the average computation time remains under 4 minutes, even for scenarios with up to 60 segments and 30 vessels.