Fire spread and percolation modelling

Abstract

Simple percolation theory, which has recently been used as a mathematical model of fire propagation, predicts values for critical exponents that differ from those obtained by a laboratory simulation of percolation in which matchsticks (with ignitable heads) are randomly placed in a square lattice. The most noticeable discrepancy is that the theory predicts that at critical percolation a fire-front decelerates, whereas the experiments indicate acceleration. Although site-bond percolation theory yields qualitative insights into expected fire-behaviour, its present formulation appears to be quantitatively incorrect. This implies that actual fires are in a different universality class to site-bond percolation. © 1990.