SCRO / Journées de l'optimisation

HEC Montréal, 29-31 mai 2023


HEC Montréal, 29 — 31 mai 2023

Horaire Auteurs Mon horaire

CHLE Collaborative humanitarian logistics and evacuation network design

31 mai 2023 13h30 – 15h10

Salle: Hélène-Desmarais (bleu)

Présidée par Valérie Bélanger

4 présentations

  • 13h30 - 13h55

    Multi-mode Evacuation Planning under Traffic Congestion

    • Alfredo Moreno, prés., HEC
    • Aura Jalal, HEC Montréal
    • Valérie Bélanger, CIRRELT, HEC Montréal
    • Marilène Cherkesly, ESG UQÀM
    • Marie-Eve Rancourt, HEC Montréal

    Disasters can threaten entire communities, requiring evacuation. To perform a safe and efficient evacuation, the proper use of transportation modes is desirable and necessary. In this study, we consider multi-mode evacuations, including self- (or car-based) evacuation and supported- (or bus-based) evacuation. The first evacuation type concerns people who use their own vehicles to drive to assigned shelters. Supported evacuation is arranged by public authorities and relies on buses stored and dispatched from pickup location to the shelters. Travel times on the road links are flow-dependent due to traffic congestion, which results from the simultaneous and massive movement of vehicles toward the links during evacuation operations. We propose a mathematical model to decide the location of shelters, the routes used during evacuation, the pickup points to collect evacuees, and the flow of evacuees on the network over the time horizon. Traffic congestion is incorporated using BPR function. The objective is to minimize the total evacuation time, i.e., the total travel time spent by evacuees from source to destination nodes, including possible delays and waiting times due to traffic congestion.

  • 13h55 - 14h20

    Wildfire evacuation planning under traffic congestion and network disruption

    • Aura Jalal, prés., HEC Montréal
    • Alfredo Moreno, HEC
    • Valérie Bélanger, CIRRELT, HEC Montréal
    • Marilène Cherkesly, ESG UQÀM
    • Marie-Eve Rancourt, HEC Montréal

    Wildfire evacuation planning is subject to significant uncertainty due to the unpredictable nature of time, location, and severity of the disaster. Depending on the magnitude of the disaster, the network capacity can be affected by partial or total disruptions of the road links. Also, the speed of vehicles can be reduced by traffic congestion and smoke conditions, increasing the evacuation time. Our study addresses evacuation planning under traffic congestion and network disruption uncertainty. We propose a stochastic mathematical model to minimize the expected evacuation risk. We use the natural hazards definition of risk by minimizing the exposure in time and location of evacuees to the potential hazard during the evacuation. The model decides the shelter location, and the flow and route definition for evacuees in a multi-period planning horizon. Network disruptions are incorporated by a stochastic parameter defined as the number of disrupted lanes by road link, which is based on the wildfire behaviour data. The traffic congestion is included by means of BPR-based formulation. We generate instances based on real data and perform computational experiments to analyze the performance of the proposed framework.

  • 14h20 - 14h45

    Operations research applications for coordination, cooperation, and collaboration in humanitarian relief chains: A framework and literature review

    • Birce Adsanver, prés., HEC Montréal
    • Burcu Balcik, Ozyegin University
    • Valérie Bélanger, CIRRELT, HEC Montréal
    • Marie-Eve Rancourt, HEC Montréal

    Humanitarian space engages a large number of actors, which necessitates coordination for efficient and successful disaster response. Cooperation and collaboration are also essential to cope with the sheer size of challenges and limited resources. The need for improved coordination, cooperation, and collaboration (3Cs) in humanitarian relief chains has been increasingly highlighted in literature and practice. Over decades, the humanitarian sector has made growing efforts to support 3Cs, and a substantial amount of research has explored 3Cs by using conceptual, empirical, and analytical methods. Nevertheless, there exists no study that provides an overview and analysis of the Operations research (OR) approaches that address the design and management of 3C mechanisms in humanitarian relief chains. In this study, we review the existing literature to present a holistic view of discussions and derive a conceptual framework for the 3C mechanisms in humanitarian operations. Based on our framework, we then analyze studies that develop OR methods to support decision-making for improved 3Cs in the humanitarian relief chains. Finally, we identify the current gaps and avenues by considering the problem and methodological aspects.

  • 14h45 - 15h10

    Humanitarian shelter network design and evacuation planning problem: An application to flood preparedness in Haiti

    • Maedeh Sharbaf, prés.,
    • Valérie Bélanger, CIRRELT, HEC Montréal
    • Marie-Eve Rancourt, HEC Montréal

    The risk of flooding is growing worldwide, and evacuation is a commonly used strategy to protect people from the impacts of floods. During an evacuation, locating proper shelter sites is important in supporting the security of affected people. Pedestrian-based evacuation in vulnerable regions is challenging due to the difficulty of walking in flood water. In collaboration with the World Bank, we present a decision-support tool which addresses the problem of shelter network design and evacuation planning through the use of a Stackelberg game. In the proposed model, the leader is the authority who decides about shelter locations and the followers are the evacuees who react to upper-level decisions. As a first attempt toward incorporation of human behavior in an optimization model, we contribute to the literature by modelling the response of the population to an evacuation order (e.g., evacuation participation rates, mobilization times, route choices of the evacuees). The problem is formulated as a risk-based bi-level optimization model with time-varying characteristics (e.g., evacuee behavior and disaster propagation) and tested using socio-demographic and GIS data of Haiti. This framework is used as a paradigm to capture three main sources of uncertainty in evacuation planning: demand (population), supply (shelter), and network (evacuation path).