Using Agent-Based Simulation of Human Behavior to Reduce Evacuation Time

Abstract

Human factors play a significant part in the time taken to evacuate due to an emergency. An agent-based simulation, using the Prometheus methodology (SEEP 1.5), has been developed to study the complex behavior of human (the ‘agents’) in high-rise building evacuations. In the case of hostel evacuations, simulation results show that pre-evacuation phase takes 60.4% of Total Evacuation Time (TET). The movement phase (including queuing time) only takes 39.6% of TET. From sensitivity analysis, it can be shown that a reduction in TET by 41.2% can be achieved by improving the recognition phase. Emergency exit signs have been used as smart agents. Modified Ant Colony Optimization (ACO) was used to determine the feasibility of the evacuation routes. Both wayfinding methods, the ‘familiarity of environment’, which is the most natural method, and the ACO method have been simulated and comparisons were made. In scenario 1, where there were no obstacles, both methods achieved the same TET. However, in scenario 2, where an obstacle was present, the TET for the ACO wayfinding method was 21.6% shorter than the one for the ‘familiarity’ wayfinding method.KeywordsEvacuation planningPrometheus methodologymulti-agent simulationAnt Colony Optimizationcognitive behavior