GPU based Computational Simulation of Aircraft Evacuation: Temporal and Spatial Analysis

摘要

The effectiveness of Aircraft Emergency Evacuation plays a vital role in the safety of the passengers on board an Aircraft, in case of Emergency landing. In this paper, the implementation and development of a Cellular Automata (CA) based simulator which can be used to simulate the Aircraft Evacuation process is presented. Given the seat-map of the Aircraft, number of passengers, passenger feature distribution and number of functional exits, the simulator can calculate the approximate time of Evacuation. For computational implementation, a bi-dimensional as well as uni-dimensional grid which represents the 2D Aircraft seat-map and an agent which control the passenger movement during Evacuation of the Aircraft is used. Each agent represents a passenger and is characterized by the properties of a human being. An agent has properties like age, sex, walking speed, response time, position and status. The CUDA framework for the parallel implementation of our algorithm/code which can be executed on GPUs (graphics processing unit) has been used and thereby speeding up the simulation process. Several test cases are performed and the results on Aircraft Evacuation times have been compared with existing data collected by Aircraft manufacturers. Detailed investigations reveal useful information on the relationship between Evacuation time and important attributes such as passenger age and sex, number of gates open, passenger distribution etc. The simulator facilitates the investigation of spatial and temporal movement of the passengers, as well as the visualization of pattern formation and collective behaviour.