Original articles
Radiant heat flux modelling for wildfires

https://doi.org/10.1016/j.matcom.2019.07.008Get rights and content

Abstract

Accurate predictions for radiant heat flux are necessary for determining exposure levels to personnel and infrastructure in the event of wildfires. However, detailed physics-based calculations of radiant heat flux are complex and current modelling practice involves significant simplifications in order to make these calculations tractable. We detail current practice for the calculation of radiant heat flux from wildfires and investigate modelling improvements that could benefit practical usage. Furthermore, we demonstrate that current limitations can be circumvented by more advanced physical models using newer generations of computational hardware. Such physical models could allow highly detailed and accurate calculations of radiant heat flux leading to improved risk assessments and planning in regions affected by wildfire.

Introduction

The ability of homes and infrastructure in wildfire prone areas to withstand the effects of fires is crucial to the safety of residents and the survival of structures. Predicting the probability of house ignition from a nearby wildfire is complex as the fire can impact a structure through direct flame contact, radiant heat and embers [4], [6], [17]. Although embers can fall on or within the structure and cause secondary fires [3], [4], [8], radiant heat flux is generally used to model potential fire risk [21], [22], [25] and is used by existing regulatory instruments in Australia [26].

Analytical models of radiant heat flux have been used to determine firefighter safety zones [5] as well as the minimum separation distance between houses and nearby vegetation [7], [11]. Although the physical relations governing radiative heat transfer are conceptually straightforward, application to general real-world scenarios are challenging due to the difficult calculations required [28] and such studies have generally used simplified flame geometries such as parallel surfaces. For geometries representative of real-world conditions more complex modelling techniques must be applied.

This paper reviews the practice of modelling radiant heat flux, in particular from an Australian perspective. The work here provides a context for technical modelling details that have been previously presented [13], discusses current practice and demonstrates limitations of simplified models currently in use. Furthermore, we show that dynamically calculated radiant heat flux models can be applied to realistic scenarios with complex topography, and provide a comparison between simplified models of heat flux and these fully dynamic models. Finally, we discuss improvements that could benefit practical usage in wildfire risk modelling.

Section snippets

Radiant heat flux

The radiation exchange between two surfaces depends on the temperature of the surfaces and the proportion of radiation leaving the emitting surface that strikes the receiving surface, known as the view factor. The radiant heat flux, q, with units Wm2 on an infinitesimal receiver surface at ambient temperature Ta K from a surface at a temperature T K is given by the Stephan Boltzmann law: q=Fσϵ(T4Ta4)where F is the view factor from the surface to the infinitesimal receiver, σ is the

Bushfire attack level model

Simplified models of radiant heat flux are currently used in practice within Australia, the most common is outlined in the Australian AS:3959 (2009) Standard for bushfire attack level (BAL) assessment. This uses an estimation of potential radiant heat flux to determine the level of fire protection required for construction in bushfire prone areas. There are two methods detailed in the Standard for estimating the radiant heat flux. The first ‘method 1’ is a tabulated set of values, whereas the

Physical modelling of radiant heat flux

A physical approach for modelling radiant heat flux is to directly evaluate Eq. (1) for the heat flux on a receiver from all nearby emission sources [25], [28]. Such physical models have been used in highly detailed wildfire reconstructions to calculate the average radiant heat on damaged and un-damaged houses to assess the relation between exposure and structure loss [18], [21], [22].

The major difficulty in computing heat flux using a physical model is in determining the view factor. Although

Dynamic radiant heat flux calculations

Wildfires are dynamic processes and structures subject to radiation from a wildfire will have a complex time-varying exposure to radiant heat flux which a static method cannot fully capture. The calculation times for the GPU-based DDA method are rapid enough for this to be used in conjunction with dynamic fire simulations to determine such temporal profiles of radiant heat flux. In these cases, the emission surface is a region of flame dynamically generated from a wildfire simulation. In this

Discussion

The dynamic radiant heat flux model can be compared directly to the existing simplified methods for calculating radiant heat outlined in the first part of this study. It should be emphasised that direct comparison is not entirely applicable as the dynamic model provides a temporal profile of the heat flux, whereas the simplified models provide an estimate of the maximum possible heat flux under certain fixed conditions. A map of the radiant heat flux calculated using the spatial BAL ‘method 2’

Conclusion

The calculation of radiant heat flux from a wildfire is challenging due to many factors including accurate determination of geometry dependent view factors in highly complex environments, difficulty in prediction of fire behaviour and the ability to accurately represent and compute the underlying physics. The need of such measurements for risk assessment has resulted in models incorporating simplifications for each of these factors, the most widespread of which is given in the Australian

References (30)

  • HiltonJ.E. et al.

    Effects of spatial and temporal variation in environmental conditions on simulation of wildfire spread

    Environ. Model. Softw.

    (2015)
  • AlexanderM.E.

    Estimating the length-to-breadth ratio of elliptical forest fire patterns

  • AmanatidesJ. et al.

    A fast voxel traversal algorithm for ray tracing

  • BlanchiR. et al.

    Investigation of bushfire attack mechanisms resulting in house loss in the ACT bushfire 2003

    (2005)
  • BlanchiR. et al.

    Property safety: judging structural safety

  • ButlerB.W. et al.

    Firefighter safety zones: A theoretical model based on radiative heating

    Int. J. Wildland Fire

    (1998)
  • CohenJ.D.

    Preventing disaster

    J. Forestry

    (2000)
  • CohenJ.D. et al.

    Modeling potential ignitions from flame radiation exposure with implications for wildland/urban interface fire management

  • EvansD.D. et al.

    Physics-Based Modeling for WUI Fire Spread - Simplified Model Algorithm for Ignition of Structures By Burning Vegetation

    (2004)
  • FLAMESOL

    Bushfire Attack Level Calculator

    (2019)
  • Geoscience Australia

    The Bushfire Attack Level Computation Software

    (2017)
  • GettleG. et al.

    Criteria for determining the safe separation between structures and wildlands

  • GouldJ.S. et al.

    Project Vesta: Fire in Dry Eucalypt Forest

    (2007)
  • HiltonJ.E. et al.

    Dynamic modelling of radiant heat from wildfires

  • HottelH.C.

    Radiant heat transmission between surfaces separated by non-absorbing media

    Trans. ASME

    (1931)
  • Cited by (13)

    • Behavior of child droplets during micro-explosion and puffing of suspension fuel droplets: The impact of the component mixing sequence

      2022, International Journal of Heat and Mass Transfer
      Citation Excerpt :

      It is well known that flame fluctuations are affected on droplet behavior during the heating [53]. It is absolutely relevant to conditions of real wildfires [57]. Another schemes of droplet heating including muffle furnace, high temperature gas, heating substrate are not good choice to produce conditions of real wildfires.

    • Data-driven surrogate model with latent data assimilation: Application to wildfire forecasting

      2022, Journal of Computational Physics
      Citation Excerpt :

      A full investigation calls for additional work and a larger project, which is beyond the scope of this initial attempt. For example, we can build the surrogate model from some other fire behaviour models such as FARSITE, FLAMMAP [32], SPARK [42], and BehavePlus [5] to simulate at various complexities. Future work will benefit from learning from these simulations, augmented with historical fire observations, to add probability of surface-to-crown transition and spotting spread in our framework.

    View all citing articles on Scopus
    View full text