Estimating thermal radiation fields from 3D flame reconstruction

Abstract

Designing fire safety into a building requires a designer to think through issues that include fire ignition, growth and spread. The dominant mechanism of spread is radiative heat transfer from flames. Therefore, in order to understand how the fire might develop it is necessary to determine the thermal radiation field surrounding a fire.

This requires being able to calculate the heat flux for any target location. The accuracy of heat flux calculations depends on the accuracy of its components, which are the flame's emissive power and the shape factor between the flame's surface and the target. The essential requirements for determining shape factor are to define the geometry of the flame surface, and to have a method for obtaining the shape factor for that geometry. Previous methods for predicting heat flux have modelled the flame as simple regular geometries because methods for calculating shape factors for other geometries were either too complex and error prone, or did not exist. Hankinson [7] overcame this limitation with a method for calculating shape factors that can be applied to multiple, irregular flame geometries. However, these geometries must be defined in order to apply the method. Therefore, until now Hankinson's method has been restricted in its use by an inability to define multiple, irregular geometries. In this study we focus on accurately determining the shape factor by presenting a method to define flame geometry using image-processing techniques. The images of the fire are recorded using multiple cameras and the flame is reconstructed in 3D. Once the surface of the flame has been defined it is possible to calculate the shape factor at a given target. Then, using the shape factor and a suitable estimate for the emissive power of the flame, the heat flux at the target can be determined. Repeating this process for multiple targets builds a thermal radiation field. This can then be used to find out which adjacent objects are threatened and which are safe.

An introduction to some of the relevant aspects of radiative heat transfer is given, followed by a detailed description of the method. The method is first applied to a controlled flame, represented by a propane diffusion burner with an electronically regulated fuel supply. Next, the method is applied to an uncontrolled flame, represented by a burning item of upholstered furniture. For this experiment, the recorded heat flux data was graphed against calculated shape factors and the line of best fit was obtained. The slope of the line represents the emissive power of the flame, and values for Pearson's linear correlation coefficient were found to range between 0.955 and 0.998.