The Seventh Asia-Pacific Conference on Wind Engineering, November 8-12, 2009, Taipei, Taiwan A NUMERICAL STUDY ON TRAVEL DISTANCES OF FIREBRANDS BY WIND Young-duk Kim 1, Yoshihiko Hayashi 2 Yohanes Joni Tri C 3 and Seung Jeong Baek 4 1 Professor, Environmental Engineering, Kwandong University, Korea, kimyd@kd.ac.kr 2 Researcher, Building Research Institute, Japan, hayashi_m@mopera.net 3 Researcher, Environmental Engineering, Kwandong University, Korea, joni@kd.ac.kr 4 PhD candidate, Environmental Engineering, Kwandong University, Korea, baeksj04@naver.com ABSTRACT In this study, CFD simulation is performed in order to assess about traveling distances of firebrands by wind. Burning embers, commonly called firebrands, are lofted by a fire s buoyant plume and transported downwind to In the present work, the influences of the inflow wind velocity, firebrand diameters, generating site, and initial generating velocity were investigated. A three-dimensional physics-based model is used to precompute the steady-state gas flow and thermal fields induced by a crown fire into which firebrands will be injected. Another preliminary study of the thermal degradation and combustion of woody fuel particles is conducted to determine the burning characteristics of firebrands. Then the trajectories and burning rates of cylinder and disk shapes of firebrands lofted by the crown fire plume and transported downwind are calculated for various values of the turbulence intensity, Uwind, and of the fire intensity, I. Results show that for firebrands that remain longer in the thermal plume, the distance covered upon landing is independent of the initial particle diameter and pyrolysis temperature. Finally, the landing distances of the firebrand spatial distribution produced from the firebrand generator are presented by comparing in each set-up conditions. It shown that, the comparison between experiment and calculation are good agreement on traveling distances of firebrands. KEYWORDS: FIREBRAND, FIRE, BURNING EMBERS, WIND EFFECT, 1 Introduction The spotting process, whereby flaming or glowing embers, commonly called firebrands, are lofted by a fire plume and transported downwind to ignite new fires (spot fires) ahead of the main fire, is an important mechanism for wildland fire spread. Understanding how these hot firebrands can ignite surrounding fuel beds is an important consideration in mitigating fire spread in communities. Ignition due to spotting is one of the most difficult aspects to understand in these fires. Due to the sheer complexity involved, it is useful to delineate the firebrand problem into three main areas: the generation from vegetation and structures, subsequent transport through the atmosphere, and the ultimate ignition of fuels after firebrand impingement. Most firebrand studies, experimental and numerical, have focused on firebrand transport. Prediction of spot fires and firebrand distribution requires understanding of the firebrand behavior, including their transport in convective plumes associated with different fire intensities and wind speeds, and their potential to ignite fuel beds on landing. Recently, Hayashi et al. (2007) have determined the size and mass distribution of firebrands produced from burning trees and they have also conducted experiments investigating the ignition of fuel beds to firebrand attack. The general lack of
knowledge of the type of firebrands that are produced as well as the type of materials that may be ignited has greatly hampered further understanding of this problem. The present work focuses on effect of firebrand shapes and wind velocity. The experimental test from Hayashi et al. on the development and characterization of a firebrand generator are used for reference model and comparing the results on traveling distances of firebrands. 2 Methodology 2.1 The Gas-phase model The gas flow is described by the Favre-averaged balance equations for mass, momentum, enthalpy, and species mass fraction. Turbulence is modeled using the k-epsilon with additional buoyancy-driven production/ destruction. In Table 1 the boundary conditions are summarized for the different variables and the constants of the turbulence model are listed. The relative humidity is assumed to be constant and equal to 40%. At the upstream boundary, the incoming airflow is taken as that corresponding to the atmospheric surface layer with a uniform flat terrain. The soot concentration is determined from the species equation and treated as gas-phase species. Radiation model of the discrete transfer method is used to provide the radiation source term for the energy equation of gas and the radiation flux to the solid surface. The absorption and emission of gas and soot are considered. Scattering is negligible due to the small diameter of the soot. Convection terms are discretized using the upwind difference scheme. In order to accelerate convergence, all the gas variables are under relaxed using inertial relaxation. The gas phase and each class of the solid phase (i.e., vegetation elements such as dead and living needles, leaves, and branches) in thermal nonequilibrium is assumed to be interdispersed and coupled by appropriate interaction terms. Each of them is treated as a continuum and governed individually by a set of time-dependent equations. Table 1 Boundary condition for Gas-phase model Initial conditions T 0 300K P 101325Pa 0 2 g ( 0,0, g) m / s Turb. model k Radiation: Discrete transfer method Boundary conditions Inflow U= uniform flow (0 m/s and 9 m/s) Outflow Relative pressure: 0 Pa Firebrands generator Heat transfer: 993K Ground No slip condition and Roughness wall Wall Free-slip conditions 2.2 Firebrands model Firebrands are thought to be cylinder and disk-shaped particles, and in this study the disk-shaped particles are used. The behavior of firebrands in airflow can be represented by air solid two-phase flow. Therefore, the scattering of firebrands can be presented by Lagrangian transport equation. A schematic representation of cylinder-shaped and disk-shaped dry woody fuel particles studied here is shown in fig. 1. As the particle undergoes thermal degradation and combustion processes, it loses mass and volume. Pyrolysis is a subsurface volumetric
chemical process whereas heterogeneous combustion (char oxidation) occurs at the outer surface of the particle. Consequently, the particle loses mass via in-depth pyrolysis and heterogeneous combustion; and it loses volume only from the heterogeneous combustion. Fuel properties and thermo kinetic constants are summarized in Table 2 for the species of trees considered. Fig. 1 Schematic of a cylinder shape and a disk-shaped dry wood particle undergoing pyrolysis & char oxidation Table 2 Firebrands properties Parameter Unit Value Wood Density: w kg/m³ 50 Wood heat conductivity: w W/m/K 0.24 Specific heat of wood: C pw J/kg/K 1466 Char density: c kg/m³ 50 Char heat conductivity: c W/m/K 0.1 Specific heat of char: C pc J/kg/K 1100 Pyrolysis temperature: T pw0 K 993 Reaction heat of pyrolysis: Q pw0 J/kg 418 Pre-exponential factor: A 0 1/s 725 3. Outline and results of Numerical Simulation 3.1 Computational details The computational domain is 2000 cm long, 1000 cm high, and 1000 cm wide. The simulation model was adopted from fire tunnel experiment, which was performed by Manzello et al. The burning fuel occupies a volume of as an idealized representation of the cylinder 225 cm in height, the location of the firebrand generator is shown Fig. 2. To track the evolution of the size and mass distribution of firebrands produced, the 1500cm of test section is used. Fig. 2 Simulation domain
Grid-independent solution of tetrahedral mesh type is generated for examining grid dependence. The numbers of mesh scheme is about 1.2 x 10 6. The meshes grid-stretching ratio in the wake region is restricted to less than 1.2 for reducing the difference of cut-off wave number between neighboring grids. The total number of nodes is about 88,000 nodes and 250,000 numbers of elements. Table 3 lists the cases simulated in this study. The inflow wind velocity of 0 m/s and 9 m/s is chosen and six cases of simulation conditions are used for comparison in this study. The density firebrand of 50 (kg/m 3 ) is chosen to apply in three cases in this simulation. For the case 1, 2, 4 and 5 the cylinder-shape firebrand is used with surface temperature of firebrands is 993K, while in the case 3 and 6 the disk-shape firebrand is used with surface temperature of firebrands is 993K Case Inflow Table 2 Firebrands properties Firebrands properties velocity (m/s) Density (kg/m 3 ) Diameter (mm) Length (mm) Pyrolysis temperature (K) 1. Cylinder shape 0 50 8 50 993 2. Cylinder shape 0 50 12.5 50 993 3. Disk shape 0 50 25 6 993 4. Cylinder shape 9 50 8 50 993 5. Cylinder shape 9 50 12.5 50 993 6. Disk shape 9 50 25 6 993 3.2 Firebrand trajectories The vertical trajectories of firebrands (cases 1 to 6) are shown in Figs. 3 and 4. Firebrand trajectories for wind speed of 0 and 9 m/s at vertical center section and gas temperature contours are shows to provide a visualization of the fire plume. Fig. 3 shows the trajectory of firebrands with 0 m/s flow and it clearly shows in cases 1, 2 and 3 the firebrand trajectories are similar. Also the temperature distributions in each case are also similar. Under (no wind) conditions, the firebrand distribution occurred close from the exit of the generator (dragon). (a) Cylinder shape 8mm, 50mm(L) (case 1) (b) Cylinder shape 12.5mm, 50mm(L) (case 2) (c) Disk shape 25mm, 6mm(L) (case 3) (d) Experiment of firebrands generator Fig. 3 Vertical trajectories of firebrands with 0 m/s flow
Fig. 4 shows the vertical trajectories of firebrands of cases 4, 5 and 6. Firebrand trajectories with inflow wind speed of 9 m/s at vertical center section and gas temperature contours are shows to provide a visualization of the fire plume. Fig. 4 shows the trajectory of firebrands with 9 m/s flow and can be seen from the figures, the tendency that the shape of firebrands could give effect to the firebrands traveling distances. At 9m/s, the firebrand distribution occurred farther from the exit of the generator (dragon). (a) Cylinder shape 8mm, 50mm(L) (case 4) (b) Cylinder shape 12.5mm, 50mm(L) (case 5) (c) Disk shape 25mm, 6mm(L) (case 6) (d) Experiment of firebrands generator Fig. 4 Vertical trajectories of firebrands with 9 m/s flow 3.2 Percentages of firebrands landing distances Fig. 5 and 6 shows the comparison percentages of firebrands travel distances between experiment and calculation in the x direction. Fig. 5 displays the spatially resolved locations where the firebrands landed under no flow conditions (0 m/s). As can be seen from the figures, the largest percentage of the number of firebrands lofted in the range of 1.5 (dimensionless) from the firebrands landed. The results show, the distribution of firebrands landed between case 1 and 2 are similar in the landing distances, while in case 3 the percentage of firebrands landing distance are more shorter than cases 1 and 2. This mean the cylinder shape of firebrand may the farther on landing distance than disk shape of firebrands. (a) Cylinder shape 8mm, 50mm(L) (case 1) (b) Cylinder shape 12.5mm, 50mm(L) (case 2)
(c) Disk shape 25mm, 6mm(L) (case 3) (d) Cylinder & disk shapes (Case 1,2 & 3) Fig. 5 Percentages of firebrands traveling distances with 0m/s flow Fig. 6 displays the spatial distribution of the firebrands landed used at 9m/s. At 9m/s, each of firebrand shapes resulted in the largest percentage of the number of firebrands lofted in the range of 2.5 3 (horizontal distance) from the firebrands landed. These detailed findings regarding firebrand lofting may be used to validate a model of firebrand transport In Fig. 6 the distribution of firebrands landed between case 4 and 5 are similar in the landing distances, while in case 6 the percentage of firebrands landing distance are more shorter than cases 1 and 2. This mean the cylinder shape of firebrand may the farther on landing distance than disk shape of firebrands.. (a) Cylinder shape 8mm, 50mm(L) (case 4) (b) Cylinder shape 12.5mm, 50mm(L) (case 5) (c) Disk shape 25mm, 6mm(L) (case 6) (d) Cylinder & disk shapes (Case 4,5 & 6) Fig. 6 Percentages of firebrands traveling distances with 9m/s flow
Conclusion A numerical model has been developed to gain better understanding of the wind effect to scattering firebrands. A parametric study is performed to determine the effect of the fire intensity, wind conditions, and firebrand properties on firebrand behavior. It is found that the distance reached by the firebrand varies almost linearly with wind speed, while it depends very weakly on fire intensity (pyrolysis temperature) and the diameters of firebrands. The results are summary as follow: 1. The Lagrangian trajectory model that takes into account the forces of drag, pressure, and gravity on the firebrands was applied to predict the trajectory of firebrands. 2. The firebrand trajectories distribution trends are good agreement with the experiment. 3. Under baseline (no wind) conditions, the largest percentage of the number of firebrands lofted in the range of 1.5 (dimensionless) from the firebrands landed. 4. The lofting distance increased as the wind speed was increased. At 9m/s, each of firebrand shapes resulted in the largest percentage of the number of firebrands lofted in the range of 2.5 3 (horizontal distance) from the firebrands landed Acknowledgment The work described in this paper was partially supported by SOC Project (05-GIBANGUCHUK-D03-01) through the Design Criteria Research Center for Abnormal Weather-Disaster Prevention (DCRC- AWDP) in KICTTEP of MOCT, KOREA and RIC(N) of Kwandong University References S.L. Manzello, J.R. Shields, A. Maranghides, W.E. Mell, Y. Hayashi, D. Nii, On the size and mass distribution of firebrands produced from burning Korean Pine trees, Fire Mater., 2007, in review. Nicolas Sardoy, Jean-Louis Consalvi, Bernard Porterie, A. Carlos Fernandez-Pello, Modeling transport and combustion of firebrands from burning trees. Combustion and Flame (2007) V. Babrauskas, Ignition Handbook, Society of Fire Protection Engineers, Fire Science Publishers, Issaquah, WA, 2003 (Chapters 11, 14). C.S. Tarifa, P.P. Del Notario, F.G. Moreno, A.R. Villa, Transport and Combustion of Firebrands, Reports of GRANTS FG-SP-114 and FG-SP-146, U.S. Dept. of Agriculture Forest Service, 1967. A. Muraszew, J.B. Fedele, W.C. Kuby, Combust. Flame 30 (1977) 321 324. Y.Hayashi, Y. Ohmiya et al., Study on flames and thermal plumes induced by fires under windy condition. The 17th national symposium on wind engineering (in Japanese), 173-184, 2002 Di Blasi et al., Modeling and simulation of combustion process of charring and non-charring solid fuels, Prog. Energy Combust Sci, 19, 71-104, 1993 S.L. Lee, J.M. Hellman, Combust. Flame 13 (1969) 645 655. H. Otake, H. Huang et al., Simulation of flames and thermal plume in urban fire under windy condition, J. of Industrial Science Research, University of Tokyo, Vol.56, 11-16, 2004 S.L. Manzello, A. Maranghides, W.E. Mell, Firebrand generation from burning vegetation, Int. J. Wildland Fire 16 (2007) 458 462. S.L. Manzello, J.R. Shields, A. Maranghides, W.E. Mell, Y. Hayashi, D. Nii, On the size and mass distribution of firebrands produced from burning Korean Pine trees, Fire Mater., 2007, in review. F. Albini, Transport of firebrands by line thermals, Combust. Sci. Technol. 32 (1983) 277 288.