Fire spread and percolation in polydisperse compartment structures

In this paper, we employ a cellular automata and percolation analysis to model fire spread in polydisperse amorphous massively multi-compartmented structures (e.g. naval vessels, high-rise buildings, warehouses, or nuclear plants). Various shapes and sizes of compartments are considered. Each compartment is composed of n(c) equal-size cells. It is found that increasing n(c) increases fingering and lacunarities of fire patterns, and subsequently front roughness. However, this also increases the probability of fire propagation throughout the system as the percolation threshold presents a power-law decrease with n(c) -1 for small values of n(c). For large polydisperse compartments, the propagation/non propagation transition seems to be size-independent. A special emphasis is put on the dynamics of fire propagation. Further study is needed to evaluate network properties that should help in developing better strategies to reduce fire consequences.