Evacuation time estimate for a total pedestrian evacuation using queuing network model and volunteered geographic information

摘要

Estimating city evacuation time is a non-trivial problem due to theinteraction between thousands of individual agents, giving rise to variouscollective phenomena, such as bottleneck formation, intermittent flow andstop-and-go waves. We present a mean field approach to draw relationshipsbetween road network spatial attributes, number of evacuees and resultantevacuation time estimate (ETE). We divide $50$ medium sized UK cities into atotal of $697$ catchment areas which we define as an area where all agentsshare the same nearest exit node. In these catchment areas, 90% of agents arewithin $5.4$ km of their designated exit node. We establish a characteristicflow rate from catchment area attributes (population, distance to exit node andexit node width) and a mean flow rate in free-flow regime by simulating totalevacuations using an agent based `queuing network' model. We use thesevariables to determine a relationship between catchment area attributes andresultant ETE. This relationship could enable emergency planners to make rapidappraisal of evacuation strategies and help support decisions in the run up toa crisis.