Predictive Computational Fluid Dynamics Simulation of Fire Spread on Wood Cribs

Abstract

Presently, there is a need for a robust numerical simulation approach to investigate the influence of various parameters on fire spread in large open framed structures. CFD-based methods can already be used for analyzing the fire conditions but they are difficult to apply for large calculations where the geometrical details of the fuel are sub-grid scale. In this paper we present a CFD-based fire spread simulation method that makes use of the ignition temperature model for pyrolysis and introduce a correction for the mesh dependency of the fuel surface area. Wood sticks, with an ignition temperature of 300°C and a specified heat release rate per unit area of 260 kW/m 2, were used as fire load. The method was validated using laboratory scale tunnel (10 m×0.6 m×0.396 m) fire tests with a longitudinal velocity of 0.6 m/s, demonstrating a 3% bias for the peak heat release rates and less than 33% biases for the fire growth rate. The method was then applied to room-scale fire spread simulations with uniformly distributed wood cribs at 600MJ/m2. The results show that, with the help of the surface area correction, the fine-mesh predictions of the heat release rate and thermal environment can be reproduced with coarser meshes and one order of magnitude lower computational costs. Due to the inherent inability of the large-scale CFD to resolve the flame temperature, there is a minimum size of the initial, prescribed fire area which is required for consistent fire spread predictions. Through this study, the authors attempt to build a reliable CFD modelling approach for fire spread and traveling fires.