A GPU Implemented 3F Cellular Automata-Based Model for a 2D Evacuation Simulation Pattern

This study presents the principles of a Cellular Automata (CA) based model that incorporates an enhanced version of the floor-field model targeting the fire spreading representation, thus called fire-floor-field (3F). The aim of the model is to simulate evacuation processes fast and reliably, in order to act as the core module of a near real-Time effective anticipation system. To this direction, the model takes advantage of massive parallelism, an inherent feature of CA, by employing the efficient response of the floor field model and the accurate computational reproduction of fire-front evolution. Furthermore, a Graphic Processing Unit (GPU) based implementation of the proposed model is presented. Such a realisation aims at further speeding up the response of the model and it reinforces the fundamental goal of rapid activation. The model is validated quantitatively and qualitatively by being applied in the case of the two-dimensional (2D) simulated evacuation of the Building B, of the Department of Electrical and Computer Engineering, Democritus University of Thrace, under fire spreading conditions.