CORS / Optimization Days 

HEC Montréal, May 29-31, 2023


HEC Montreal, 29 — 31 May 2023

Schedule Authors My Schedule

IDII Industrial Day II

May 31, 2023 01:30 PM – 03:10 PM

Location: Walter Capital (blue)

Chaired by Yan Laporte

4 Presentations

  • 01:30 PM - 01:55 PM

    La planification intégrée des horaires et des véhicules en transport public

    • Loïc Bodart, presenter, GIRO Inc.

    GIRO propose des logiciels qui optimisent la planification et la gestion des opérations des sociétés de transport public (bus, transport ferroviaire de passagers) et d’opérations postales dans plus de 25 pays à travers le monde. Nos algorithmes sont, non seulement réputés pour leur puissance d’optimisation, mais aussi pour leur flexibilité permettant ainsi de s’adapter à la réalité spécifique de chaque client.
    Nous présenterons un algorithme d’optimisation assez unique et innovateur dans le domaine du transport public. Cet algorithme permet de créer ou d’améliorer un horaire-maître de façon à bonifier la qualité du service tout en respectant un ensemble de normes de service, et en tenant compte des coûts d’opération des véhicules. L’algorithme utilise une formulation de programmation linéaire en nombres entiers et il est résolu de façon itérative à l’aide d’une approche de décomposition heuristique, qui permet d’obtenir de très bonnes solutions rapidement.

  • 01:55 PM - 02:20 PM

    Airline crew planning optimization

    • Daniel Villeneuve, presenter, IBS

    IBS is an international company whose vision is to redefine the future of travel through technology innovation. The Montreal office is the optimization center of excellence within IBS. Being part of the Aviation Operations Solutions division, we are responsible for airline crew planning optimization. This talk will cover some of the challenges we face when building optimization-based systems to meet real industrial demands. We will highlight the need for multiple skills related to operations research and computer science.

  • 02:20 PM - 02:45 PM

    Understanding your Optimizer

    • Troy Taillefer, presenter, AlayaCare
    • François Lacoursiere, AlayaCare
    • Nadia Lahrichi, Polytechnique Montréal

    Optimizer solver programs are very complex. Understanding and verifying the quality of their solutions is a daunting task. AlayaCare has developed a multi-criteria optimizer for a Vehicle Routing Problem to schedule and route home care workers. How optimal are the solutions? Can a decision in the solutions be explained in terms of constraints and objectives? For a multi-criteria optimizer, a framework for understanding optimality is needed, the concepts of pareto front was selected and visualizations were developed. Pareto front concepts had limited applicability to hierarchical prioritization of objectives of the optimizer. Pareto front would apply better when using the weighted sum of objectives. These limitations lead to an effort to compare the optimizer to another solver, OR-Tools was selected. To explain the decisions of optimizer solutions, first procedures to manually inspect solution files and constraint violations were developed. These procedures did not scale at all to larger problem instances and were very time consuming even for small problem instances. To address these problems a program to analyze scheduling and routing constraints was developed to automatically explain the optimizer decisions in terms of availability, travel time, service time constraints. This work is based on time window compatibility ideas found in Joubert and Claasen, 2006.

  • 02:45 PM - 03:10 PM

    Modeling Passenger Behavior and Demand to Optimize Rail Revenue Management

    • Yan Laporte, presenter, ExPretio Technologies

    ExPretio is the result of over two decades of academic research in pricing and revenue optimization by revenue management practitioners and experts in operations research and management science.
    We address an important need in the field of revenue management and optimization, where the existing solutions were adaptations from other industries, but which were not well-suited to the rail market.
    The aim of revenue management is to allocate resources to different products at different price points for every origin destination served by a train to maximize revenues. But rail companies operate in a world where revenues are not the sole desired outcome.
    One challenge was to infer the effect of price and availability changes on each passenger segments and how they are interrelated.