Using microsimulation software to model large-scale evacuation scenarios. The case of Sangüesa and the Yesa dam collapse

The use of computer models to analyse evacuation scenarios and human behaviours in emergencies has greatly increased due to their capabilities to simulate these processes. These models have been commonly applied to different types of infrastructures in isolation, and little is known about their use for large-scale evacuation. This paper explores the capabilities of micro-simulation tools for modelling large-scale evacuation, and also presents the development and application of a computer based large-scale modelling tool. A novel methodology is proposed here in which different sub-models are connected to represent the different levels of the evacuation scenario. The methodology has been applied to a real case study: a small town in Spain, Sangüesa, which is in risk of flooding as a consequence of a potential collapse of the Yesa dam. The inputs used were taken from survey data, census data, plans of the town, data obtained from the emergency plan, data from other research studies (flooding models and acoustic models), and a literature review process on human behaviour in emergencies. The study was focused on the evacuation scenario at night (2:00 AM assuming the majority of residents are sleeping) as one worst-case scenarios. Four scenarios were developed in order to analyse the impact of (1) two pre-evacuation time distributions identified for residential evacuation at night and (2) group behaviours (with and without) during the evacuation process. Results showed that the time needed to evacuate the whole town (between 33 and 44 min) is not sufficient, as in case of a dam collapse the water could reach Sangüesa in about 23 min.