Simulation-based evacuation planning using state-of-the-art sensitivity analysis techniques

With the continuous growth of cities and the issues of insufficient space, indoor parking garages have become a vital component of transportation systems. Since emergencies will occur sooner or later, planning and organizing a successful evacuation plan for parking garages, with the aim of reducing the evacuation time and improving the driver's behavior during evacuations, is needed. This research presents a systematic approach for the development and evaluation of evacuation strategies in such environments, through state-of-the-art sensitivity analysis that provides faster convergence and lower computational burden. To demonstrate the proposed methodology, a simulation model is developed using AIMSUN for a three-level indoor parking garage located in Athens, Greece. Four scenarios with different evacuation strategies and plans are considered. A multi-step sensitivity analysis framework is implemented to overcome the uncertainties that emerge from the model assumptions, model inaccuracy and shortage of data. Sensitivity analysis is used to evaluate the consequences of the proposed scenarios and estimate the total time to evacuate the parking garage. The full experimental design resulted in the automated execution of more than 80 thousand simulations, corresponding to approximately 65 processor-days. The results of this research show that the developed methodology provides significant reduction in evacuation time (from 35% to over 40%). This process provides useful and credible ranges of expected evacuation times under all plausible outcomes.