Journées de l'optimisation 2018

HEC Montréal, Québec, Canada, 7 — 9 mai 2018

Horaire Auteurs Mon horaire
Cal add eabad1550a3cf3ed9646c36511a21a854fcb401e3247c61aefa77286b00fe402

WA2 Green routing

9 mai 2018 10h30 – 12h10

Salle: Banque Scotia (69)

Présidée par Samuel Pelletier

3 présentations

  • Cal add eabad1550a3cf3ed9646c36511a21a854fcb401e3247c61aefa77286b00fe402
    10h30 - 10h55

    Leverage the long recharging times of electric vehicles in city logistics

    • David Cortes, prés., Université Technologie Troyes
    • Hasan Murat Afsar, Université de Technologie de Troyes
    • Caroline Prodhon, Université de Technologie de Troyes

    Electric Vehicle usage represents environmental and economic benefits. Here a variant of e-VRP is studied. We allow visiting a customer by walking while the vehicle is charged. A hybrid-ILS is tested on literature instances: the total recharging time is reduced up to 2.5% and the total distance up to 2.8%.

  • Cal add eabad1550a3cf3ed9646c36511a21a854fcb401e3247c61aefa77286b00fe402
    10h55 - 11h20

    Measuring emissions in vehicle routing: new emission estimation models using supervised learning

    • Hamza Heni, prés., Université Laval
    • S. Arona Diop, Université Laval
    • Leandro C. Coelho, Université Laval
    • Jacques Renaud, Université Laval, CIRRELT

    Based on real-world data of instantaneous fuel consumption, time-varying speeds observations, and traffic data related to a large set of shipping operations, we propose effective methods to estimate greenhouse gas (GHG) emissions. By carrying out nonlinear regression analysis using supervised learning methods, namely Neural Networks, Support Vector Machines, Conditional Inference Trees, and Gradient Boosting Machines, we develop new emission models that provide more prediction accuracy than classical models. Extensive computational experiments under real datasets show the effectiveness of the proposed machine learning emissions models, clearly outperforming the Comprehensive Modal Emissions Model (CMEM) and the Methodology for Estimating air pollutant Emissions from Transport (MEET) in the prediction of hot running traffic emissions according to root mean square error metrics. Based on performance indicators we show that MEET underestimates real-world GHG emissions by 24.94% and CMEM leads to an overestimation of emissions by 13.18% according to observed fuel consumption, while our best machine learning model (Gradient Boosting Machines) exhibited superior estimation accuracy and is off by only 1.70% considering real-world driving conditions.

  • Cal add eabad1550a3cf3ed9646c36511a21a854fcb401e3247c61aefa77286b00fe402
    11h20 - 11h45

    The electric vehicle routing problem with energy consumption uncertainty

    • Samuel Pelletier, prés., HEC Montréal
    • Fan E, McGill University

    In urban environments, freight electric vehicles (EVs) are often exclusively charged at a central depot. We therefore introduce the EV routing problem with energy consumption uncertainty, in which EVs must be routed so that no vehicle ends up stranded even in the worst-case energy consumption scenario. We solve small instances using robust optimization techniques, and we propose a metaheuristic to solve large instances.

Retour