Massive population evacuation in an urban context

Pierrick Tranouez (1) , Eric Daudé (2) , Patrick Taillandier (3) , Guillaume Czura (2) , Géraldine del Mondo (4)

1 DocApp - LITIS - Equipe Apprentissage
2 IDEES - Identité et Différenciation de l’Espace, de l’Environnement et des Sociétés
3 MIAT INRA - Unité de Mathématiques et Informatique Appliquées de Toulouse
4 STI - LITIS - Equipe Systèmes de Transport Intelligent

Increasing urban sprawl all over the world leads to an increase of vulnerabilities, as greater numbers of goods and people are exposed to hazards, both natural (flood, quakes, fires, tsunami, epidemic) and industrial (factories, plants). Population evacuation features amongst the tools political managers can use to mitigate these risks. When evacuations are decided, managers must strike a fine balance between displacing all the people needed to avoid injuries and fatalities, and not evacuating people that would finally not be struck by the hazard. Evacuated populations convoys must furthermore not lead to a sub-crisis where they become new vulnerabilities, or through disorder and density prevent the management of other vulnerabilities. There is regretfully a scarcity of tools to help risks manager make this kind of decision. We present here our works on such tools. Most risk management policies focus on a planar, continuous conception of space. Seveso Directives for example use different radius around the hazard center, a plant for example, to find which housings and business will be affected at which degree. We described in our presentation for UrbanNet 2013 how this approach was found lacking for handling road networks, both as vulnerabilities of the direct hazard, and as the means for a successful crisis management. In order to overcome this obstacle we proposed a city-wide agent based simulation called MOSAIIC to model the car traffic, both in normal and extraordinary situations. In MOSAIIC each driver is capable of strategical, tactical and operational planning and decision making. They have a list of destinations they try to reach, and choose a path to get there using their knowledge of the network. They accelerate, brake, change ways depending on their surroundings and personality. They choose alternative solutions if trapped in traffic jams of when facing a road networks altered from their initial knowledge. In this follow-up article we would like to discuss the data and its analysis we used to calibrate and validate simulations built with MOSAIIC to study theoretical all-car evacuation of the city of Rouen. Furthermore, since MOSAIIC, we started a new project, ESCAPE (Exploring by Simulation Cities Awareness on Population Evacuation), which aims at simulating massive evacuation:

Domaines: Intelligence artificielle [cs.AI]; Système multi-agents [cs.MA]; Modélisation et simulation

Pierrick Tranouez, Eric Daudé, Patrick Taillandier, Guillaume Czura, Géraldine del Mondo. Massive population evacuation in an urban context. UrbanSys 2018, Sep 2018, Thessalonike, Greece. ⟨hal-02111135⟩