A methodology for defining building evacuation routes

This paper concerns an innovative methodology aimed towards specifying the best evacuation plan in a given scenario - where a scenario is characterised by the enclosure geometry, the population\u2019s capabilities and the population distribution within the building. The best evacuation plan is assumed to be the one which minimises the movement time of the last evacuee. The problem faced is an optimisation problem where the cost function to be minimised is the last evacuee\u2019s movement time and the research space is defined by all the possible evacuation plans in the given scenario. As this cost function could be a multi-peak function and since the research space is discrete and extremely large, a random search algorithm has been adopted. Given an evacuation plan, the proposed methodology models the evacuation over time with the aim of finding the movement time of the last evacuee. The modelling of the evacuation over time has to obey to two criteria. First, as it is embedded in the optimisation process and it has to be repeated a very high number of times (potentially for each possible evacuation plan), the computational cost is a critical issue: therefore the model has to be \u201csynthetic\u201d and microscopic models result not suitable. Secondly, the model cannot neglect flight behaviour and crowd crushes: therefore the need for a Dynamic Network Loading model which allows to keep track of the location of moving queues in the network, to predict spillbacks and dissipation and to model capacity drop due to clogging effects as well. The overall model has been implemented in an object-oriented simulator that allows analysis of large multi-floor buildings. The simulator has been applied to an illustrative problem of evaluating the best evacuation plan for a school: the application results are reported in the paper. The methodology application field is the straight forward evacuation, before the onset of hazardous conditions. In this context, the best evacuation plan is chosen off-line and the building occupants could be trained on the escape routes they have to follow from their origin rooms to the exit