Experimental study and large eddy simulation of effect of terrain slope on marginal burning in shrub fuel beds

https://doi.org/10.1016/j.proci.2006.07.222Get rights and content

Abstract

This paper presents a combined study of laboratory scale fire spread experiments and a three-dimensional large eddy simulation (LES) to analyze the effect of terrain slope on marginal burning behavior in live chaparral shrub fuel beds. Line fire was initiated in single species fuel beds of four common chaparral plants under various fuel bed configurations and ambient conditions. An LES approach was developed to model fire spreading through a fuel bed with a subgrid scale turbulent combustion model based on a flame surface density concept. By examining two fuel bed slope configurations, it was found that upslope fire spread depends not only on the increased radiant heat transfer but also on the aerodynamic effect created by the interaction of the flame with the inclined surface. Under certain conditions, the convective heat transfer induced by this interaction becomes the dominant mechanism in determining fire spread success. Seventy-three (or 42%) of 173 experimental fires successfully propagated for slopes ranging from −70% to 70%. It was found there exists a critical slope above which fire spread in these live fuel beds was successful, and below which fire spread was unsuccessful. This critical slope for marginal burning varied widely with fuel moisture content and fuel loading. A stepwise logistic regression model was developed from experimental data to predict the probability of successful fire spread. It is expected that this model may be helpful in providing guidelines for prescribed fire application.

Introduction

Wildland fire commonly occurs in Mediterranean climates that are characterized by cool, moist winters and hot, dry summers [1]. Fuel complexes of various shrub species prevalent in these areas exhibit significant fire behavior. The shrub complexes are known by various names such as fynbos (South Africa), mattoral (Chile), garrigue (France), and chaparral (California, US). Approximately 85% of the vegetation area burnt during the 2003 fires in southern California was in chaparral, indicating the importance of managing chaparral vegetation. Prescribed fire is one of the alternative strategies used by land management agencies to reduce wildland fuels and prevent wildfires [2]. Because of the fire risk in chaparral and other living fuels, prescribed fires are often conducted under marginal burning conditions. These conditions occur when environmental variables such as wind speed and direction, air temperature, relative humidity, and topography, and fuel conditions such as type, moisture content, and continuity result in low-intensity fire that may or may not spread successfully. However, even when all variables are seemingly within the prescribed conditions, often ignition results in little, or no fire spread, due to undesired extinction. Consequently, the cost of prescribed burning increases as significant equipment and personnel allocation costs are incurred prior to ignition. It is thus desirable from both an economic and scientific perspective to gain a better understanding of marginal burning conditions.

Chaparral grows extensively on hilly terrain, indicating the importance of understanding marginal burning in live fuels occurring on flat and sloped (up and down) terrains. The literature review reveals that only a limited number of experimental studies have investigated slope effects in dead fuels and virtually none have studied wildfire spread in live fuels under marginal burning conditions [3], [4], [5], [6], [7], [8], [9]; most of our previous laboratory experiments were limited to zero slope conditions [10], [11]. Towards this end, a dual approach including experimental and numerical modeling was adopted to study the effect of terrain slope on marginal burning.

Over the last forty years, a significant amount of interest in numerical modeling of wildland fire spread has been generated within the scientific community. Operational models [12], [13], [14] refer to computer-based, semi-empirical models that are currently utilized in the field for decision support systems. They do not include a two-way coupling between the fire and the fire-induced fluid flow in the atmosphere. Furthermore, they were designed primarily for dead, not live, fuels. Coupled atmosphere-fire research models [15], [16], [17], [18] on the other hand focus on improved methodologies, often limited in scope and designed to better understand specific physical processes. Research models for fire spread at intermediate length scales (typically laboratory scale fire spread experiments through fuel samples) have included simple gas-phase turbulent combustion models and an appropriate form of the Reynolds-averaged Navier Stokes (RANS) equations, including transport equations for gaseous chemical species arising through pyrolysis of solid fuel [19], [20], [21], [22]. These research models are appropriate to overall length scales of the order of a few meters.

In this paper, a three-dimensional large eddy simulation (LES) approach to model a fire spreading through a fuel bed is developed. In LES, the super-grid scales are accurately resolved, while the subgrid scales (SGS) are modeled through appropriate closure models. A recently proposed SGS turbulent combustion model, based on the well-known concept of flame surface density, was applied to model finite rate chemistry effects [23]. This LES approach is motivated by our desire to develop a physically more accurate method to simulate fire spread through a porous shrub fuel bed, compared to previous approaches [20], [22]. When fire spreads upslope, it is observed that propagation depends not only on the increased radiant heat transfer rate [24], [25], [26], [27], but it is also influenced by the complex aerodynamic flow created by the interaction of the flame with an inclined surface [28]. By obtaining instantaneous and detailed data on turbulent flow, heat transfer, and fuel combustion, LES is a promising tool to model this complex, dynamic two-way coupling between the fire and the fire-induced fluid flow. Given the obvious difficulty with large scale fire spread, the current LES is restricted to model fire spread through shrub fuels at laboratory scales.

Section snippets

Experimental setup

Fuel beds (2.0 m long and 1.0 m wide) were prepared with one of four live chaparral species: manzanita (Arctostaphylos spp.), chamise (Adenostoma fasciculatum), hoaryleaf ceanothus (Ceanothus crassifolius), and scrub oak (Quercus berberidifolia) that are typical of the chaparral found near Riverside, CA. Fuel was cut in the morning in the North Mountain Experimental area at an elevation of 1160 m, bagged, and transported to the laboratory. The test bed was prepared and ignited early in the

LES methodology applied to fire spread

In a LES, the instantaneous, time-dependent, three-dimensional governing transport equations for various field quantities such as density, velocity components, species mass fractions, etc. are spatially filtered using a characteristic filter-width Δ [23]. This results in a set of transport equations for the filtered or resolved field. They include new terms that characterize sub grid scale (SGS) phenomena, occurring at scales below Δ. These SGS terms are modeled in terms of resolved quantities,

Aerodynamic effect induced by fire and slope

Each test bed was placed on a large platform that is designed to tilt up to an angle of 45° (slope = 100%) to the horizontal to model a terrain slope. This configuration is designated setup A as illustrated in Fig. 1a and models the actual arrangement of wildland fuels in mountainous terrain. Because of safety concerns involving use of the large tilting platform, another slope setup B was designed to lift one side of the fuel bed to different angles (see Fig. 1b) while keeping the heavy platform

Conclusions

In examining two fuel bed slope setups, the large eddy simulation approach is useful in analyzing the detailed fire structure and the coupled physical processes. Upslope fire spread depends not only on the increased radiant heat transfer rate but also on the aerodynamic effect created by the interaction of the flame with the inclined platform. Under our experimental conditions, the convective heat transfer induced by this interaction becomes the dominant mechanism in determining fire spread

Acknowledgments

The funding source for this research is the USDA/USDI National Fire Plan administered through a Research Joint Venture Agreement No. 01-JV-11272166-135 with the Forest Fire Laboratory, Pacific Southwest Research Station, Riverside, CA. We appreciate the efforts of Joey Chong, David Kisor, Lulu Sun, and Watcharapong Tachajapong in assisting with the experimental burns.

References (33)

  • T. Hirano et al.

    Combust. Flame

    (1974)
  • D.D. Drysdale et al.

    Fire Safety J.

    (1992)
  • X. Zhou et al.

    Proc. Combust. Inst.

    (2005)
  • D. Morvan et al.

    Combust. Flame

    (2001)
  • X. Zhou et al.

    Combust. Flame

    (2005)
  • P.J. Pagni et al.

    Proc. Combust. Inst.

    (1973)
  • F.A. Albini

    Proc. Combust. Inst.

    (1967)
  • P.A. Santoni et al.

    Fire Safety J.

    (1998)
  • Y. Wu et al.

    Fire Safety J.

    (2000)
  • B.P. Leonard

    Comp. Meth. Appl. Mech. Eng.

    (1979)
  • L.R. Green, USDA Forest Service, Gen. Tech. Rep. PSW-51, 1981, pp....
  • J.L. Dupuy

    Int. J. Wildland Fire

    (1995)
  • C.E. Van Wagner

    Can. J. For. Res.

    (1988)
  • J.M.C. Mendes-Lopes et al.

    Int. J. Wildland Fire

    (2003)
  • D.X. Viegas

    Int. J. Wildland Fire

    (2004)
  • Cited by (53)

    • Experimental and numerical simulation of multi-component combustion of typical no-charring material

      2023, Energy
      Citation Excerpt :

      For example, the light hydrocarbon mixture (replaced by CO, CH4 and H2) was applied for wood combustion by Galgano and Zhou et al. [7]. CO, H2, light hydrocarbons (replaced by CH4) and heavy hydrocarbons (replaced by C6H6) were used for biomass combustion by Porteiro et al. [8,9]. However, there is no doubt that the rough replacement of light hydrocarbons and heavy hydrocarbons would result in a large deviation from real gas components, and this problem can be further tried to be solved by the experiment measurement of TG-FTIR technology [10,11].

    • Predicting the fire spread rate of a sloped pine needle board utilizing pyrolysis modelling with detailed gas-phase combustion

      2018, International Journal of Heat and Mass Transfer
      Citation Excerpt :

      Meanwhile, the reduction of peak-flame temperature resulting from the emission of radiation, can then significantly modify the chemical kinetics and affect other quantities that are temperature sensitive, such as formation of NOx, unburned hydrocarbons and soot particulates [17]. Most physical fire spread models incorporate simple chemistry combustion models that often predict combustion products as yield fractions [18,19]. The integration of detailed chemistry that resolved all necessary kinetics to facilitate better prediction of the evolution of combustion generated species, and the coupling of radiation model that considers the radiative heat transfer from soot particulates as well as major gas-phase species, is critical for this work and therefore employed.

    View all citing articles on Scopus
    View full text