SCRO / Journées de l'optimisation

HEC Montréal, 29-31 mai 2023


HEC Montréal, 29 — 31 mai 2023

Horaire Auteurs Mon horaire

HLPI Hub Location Problem I

31 mai 2023 13h30 – 15h10

Salle: Saine Marketing (vert)

Présidée par Mario José Basallo Triana

4 présentations

  • 13h30 - 13h55

    Efficient Solution Approaches for the Bi-Criteria p-hub Median and Dispersion Problem

    • Prasanna Ramamoorthy, prés., Indian Institute of Technology Delhi
    • Navneet Vidyarthi, JMSB, Concordia University
    • Manish Verma, McMaster University

    In this paper, we study the bi-criteria p-hub median and dispersion problem, which is modeled as a bi-objective Mixed Integer Program. The first objective is to minimize the total cost of routing the flows through p hubs and the second objective is to maximize the minimum distance (or dispersion) among the selected p hub locations themselves. We present two exact solution approaches that guarantees to obtain the entire non-dominated Pareto frontier. The first is a cutting plane method in which a p-hub median problem with a particular dispersion distance is solved at each iteration. Three formulations of the problem, based on the different type of cuts and preprocessing, are presented. We study the dominance relationship among the three formulations and validate the theoretical findings using computational results. The second approach is based on benders decomposition with several improvements. Using this approach we are able to solve large scale problem instances of the bi-objective problem.

  • 13h55 - 14h20

    Data-Driven Fleet Management in Hub Network Design with Demand Uncertainty

    • M. Saleh Farham, prés., Postdoctoral Fellow, Alberta School of Business
    • Borzou Rostami,

    Efficient vehicle management is crucial for achieving economies of scale in freight distribution and parcel delivery systems. However, fleet managers face significant challenges due to the uncertain nature of customer demand. This study addresses the fleet management problem in hub network design under demand uncertainty.
    To ensure all shipments are delivered with a certain level of guarantee, we consider the uncertainty in demand during the planning phase and determine the optimal number of vehicles required. We propose a mathematical programming formulation for the problem and develop an exact solution methodology. The effectiveness and efficiency of the proposed model and solution methodology are evaluated using real-life and benchmark datasets.

  • 14h20 - 14h45

    Sustainable hub location under uncertainty (CANCELED)

    • Gita Taherkhani, prés., Loyola University Chicago
    • Seyed Mojtaba Hosseini, University of California Irvine
    • Sibel Alumur Alev, University of Waterloo

    In this paper, we study the sustainable design of hub networks under uncertainty for truckload and less-than-truckload transportation. In particular, we study a profit-maximizing hub network design problem in which the demand of some origin-destination pairs can remain unserved, where satisfying the demand depends on the profit to be obtained from serving it. We additionally focus on sustainability by modeling both a carbon tax and a carbon cap in our problem setting. We develop a model in which, in addition to transportation and hub installation costs, the carbon tax is also explicitly included in the objective function. Moreover, to ensure that the total amount of greenhouse gas emissions emitted by trucks does not exceed the carbon cap, we incorporate an emission limit on the entire transportation network. We model emission on each arc as a convex function of transport load on the arc and consider separable and non-separable cases to cover a wide range of practical aspects, including different carbon cap policies as well as the robustness of solutions. We provide piecewise linear approximations to these emission functions to ensure the feasibility and tractability of our models. To provide a more reliable solution framework for this problem, we take the demands as stochastic parameters and then develop a Benders-decomposition-based algorithm coupled with a sample average approximation scheme for solving our stochastic problem. We implement several enhancement techniques beyond standard implementations of these algorithms to guarantee scalability to large-scale instances.

  • 14h45 - 15h10

    Intermodal hub network design with probabilistic service level constraints

    • MARIO JOSÉ Basallo Triana, prés., HEC Montreal - CIRRELT
    • Jean-François Cordeau, HEC Montréal, GERAD, CIRRELT
    • Navneet Vidyarthi, Concordia University

    We analyze hub network design models with probabilistic service level constraints for the service time. The total service time on a given origin-destination path is considered to be a random variable that accounts for the transport time on the path along with the waiting and processing times of the flow units at the different hubs. Hubs are modeled as M/M/1 queueing systems from which we obtain the total sojourn time distribution. We consider an arbitrary density function for the transport time, which is independent of the transport flow. Probabilistic constraints are formulated for the total service time, which is the convolution between the sojourn time distribution and the transport time density function. The resulting constraints are non-linear and we use piece-wise linear functions to approximate them. We propose a cutting plane algorithm to solve the formulations. Several computational experiments show the impact of different service levels and transport time requirements on the hub network structure and the modal shift between the different transport modes. In general, the formulations produce incomplete inter-hub networks in which the increased service level leads to a decrease in the number of inter-hub connections.