Multimodal Mesoscopic Approach in Modeling Pedestrian Evacuation

A recently funded research project in Italy concerned reduction in risk exposure in urban areas through the definition and implementation of evacuation procedures. One of the research aims was to specify and calibrate a system of models able to simulate the transportation system when a population has to evacuate due to a forthcoming disaster. This paper summarizes the activities conducted to calibrate a set of cost functions to be implemented within a dynamic network loading (DNL) procedure to simulate pedestrian outflow related to the evacuation of a building. We show how flow models are calibrated on the basis of data collected during experimentation at a test site and describe the results of their implementation within a multimodal dynamic loading model used to simulate evacuation procedures. A comparison between experimental data and simulation results shows that the use of appropriate simulation models can realistically reproduce user behavior and then how such models can be used as a support for drawing up effective evacuation plans. The experimentation involved the evacuation of a primary school in an Italian town, but the method applied can be easily adapted to any public building with homogeneous characteristics.