NIST Wildland-Urban Fire Models (Downloads)

(these models are in the early stages of validation)
This page contains links to NIST's wildland-urban interface fire computer models and example input files. Validation of these models is ongoing but far from complete. Please keep this in mind when using wfds for problems outside the current scope of validation. See download sections below for information on the current state of validation.
last updated February, 2009

In general please use the FDS discussion group and issue tracker when you have questions (see http://fire.nist.gov/fds/). Refer to WFDS. My (William [Ruddy] Mell) e-mail address is ruddy@nist.gov

For new users please send me an e-mail so that I can inform you when new versions, capabilities, and a user guide are available.


Contents
WFDS model
  1. Overview of WFDS
  2. WFDS user information
  3. Status of WFDS development
  4. WFDS for raised fuels (e.g., trees) and surface fuels; the fuel element model
    1. Validation studies
    2. Program download
    3. Input files
      1. spread through pine needle bed under one tree
      2. spread through grass under tree stand
      3. stationary line fire upwind of tree stand
  5. Download for WFDS for surface fuels (e.g., grasslands); the boundary fuel model currently not supported
    1. Validation studies
    2. Program download
    3. Input files
      1. two-dimensional grassfire simulation
      2. 1 processor Australian grassfire
      3. 2 processor Australian grassfire
      4. 1 processor laboratory pine needle bed


WFDS

Overview of WFDS

WFDS (Wildland-urban interface Fire Dynamics Simulator) is an extension of NIST's structural fire dynamics simulator  (FDS) to fuels that include vegetation. WFDS uses computational fluid dynamics methods to solve the governing equations for buoyant flow, heat transfer, combustion, and the thermal degradation of vegetative fuels. The solution method makes use of large eddy simulation (LES) techniques to solve the gas-phase equations on computational grids that are too coarse to directly resolve the detailed physical phenomena. The development and validation of WFDS is part of NIST's wildland-urban interface project

WFDS user how-to information

Quick way (for experienced users of wfds or FDS):

Running WFDS is identical to running FDS. The visualization package from NIST, called Smokeview, also works for WFDS. Just download the appropriate WFDS program and input file from below and type (from within a command window for the MS Windows users)
     wfds.....exe  input....
where wfds....exe is the wfds program and input... is the input file.
In the example above it's assumed the wfds program resides somewhere in the path. If the program is in the same directory as the input file type
    ./wfds....exe  input....

For new users (who have not run either WFDS or FDS in the past):
  1. Installing WFDS: New users of WFDS will find it easiest to first install and perform simple runs with FDS and the companion visualization tool Smokeview (also developed at NIST). See http://www.fire.nist.gov/fds/downloads.html to obtain the self-extracting installation for FDS, Smokeview, sample FDS input files, and documentation.
  2. Running WFDS and viewing its output is the same as for FDS. Users are encouraged to read the FDS and Smokeview user guide. A user guide for WFDS is under development. Once FDS is installed then the WFDS program, downloaded from the links below, can be put in the same directory as FDS (the default installation directory is C:\Program Files\...). The user can then type the WFDS program name in a command window to run WFDS (exactly as they would for FDS). The user can also put the WFDS program in the folder where they plan to run it. (See the FDS manual.)
  3. Input files. Sample input files are given below in the download sections. The quantities that define the vegetation are straightforward to understand. Some discussion of these are in the download sections. In general, users should define the thermophysical parameters of the vegetation with values that are similar to what's in the input files. If you have questions contact me ruddy@nist.gov.

Status of WFDS model

WFDS There are currently two ways of representing vegetative fuel in WFDS:

(1) Fuel element model. This model is fully integrated into FDS (links to the executable are also given below). In this approach the vegetation occupies a specified volume  (e.g., trees crowns).  In general, this model is to be used when the grid resolution can span the vegetation with a number of grid cells. Realistic fire spread through grass, or other surface fuels, can be simulated if the computational grid is sufficiently fine.

(2) Boundary fuel model. This model is currently being incorporated into FDS. It is
limited to surface fuels (e.g., ground cover, grass). Links to the out-dated version, based on FDS4, are below.

As mentioned above, the procedure for running WFDS is the same as for FDS (only some of the inputs are different as can be seen in the sample input files below). Consult the FDS user manual ( http://fire.nist.gov/fds/documentation.html ) and the discussion group (http://groups.google.com/group/fds-smv) if you are having trouble and I'm not available.

Download links and input files

WFDS for fuels in a specified volume such as tree crowns (fuel element model)
Validation of fuel element model to date:
  1. Simulation of tree burning experiments at NIST.
    Report of ongoing validation studies is "Numerical modeling of fire spread through trees and shrubs," to be presented at 5th Intn'l Conf. Forest Fire Research, Portugal.
    Archival paper to Combustion and Flame is under review.
WFDS program downloads with fuel element model.

Note: This implementation of WFDS is fully incorporated in the current version of FDS, so if you have the current version of FDS you already have WFDS with the fuel element model representation of vegetation. When posting to the FDS Issue Tracker or the Discussion Group regarding the application of FDS to wildland or wildland-urban interface fires please refer to WFDS in your text to facilitate searches and so we can streamline our response.
  1. WINDOWS 
  2. LINUX 
Input files using fuel element model
Vegetation can be described by a rectangular, conical, or cylindrical volume. In the input files below you'll find surface fuels described by a rectangular volume (e.g., pine needles), crown fuels as conical volumes, and tree stems as cylindrical volumes. The stems are present to serve as wind breaks and do not burn. If you are including stems please following the format of the input files.

CPU timings given below were obtained on a duo quad core Linux computer with 3.5 GHz processors.
  1. Fire spread through surface fuel with one tree

Single processor input file:
 input_surf_onetree_1proc.txt

Run by typing (in Windows command window)
wfsd32.exe surf_onetree_1proc.fds

Two processor input file:
input_surf_onetree_2proc.txt

Run by typing (in Windows command window with MPICH2)
mpiexec -n 2 wfds32_mpi.exe surf_onetree_2proc.fds

Burning pine needle bed (5 cm deep), temperature plume, tree foliage, and tree stem. Domain is 16 m long (160 grid cells, dx=10 cm), 3 m wide (30 grid cells, dy=10 cm), and 6 m tall (60 grid cells, dz = 5 cm at ground to 20 cm at top). Mirror or symmetry boundary conditions are used along y = 0 plane. Axes units are meters. Simulation time is 60 s which requires about 2.2 cpu hours. About 200 MB of memory is required.  The 2 processor simulations required 1.2 cpu hours on each processor (total elasped wall clock time was also 1.2 hours, this includes start up and writing out data files)
  1. Fire spread through surface fuel with many trees on flat terrain. No significant crowning
 
Single processor input file
input_grass_trees_flat_1proc.txt


Two processor input file:
input_grass_trees_flat_2proc.txt


About 90 identical 6 m tall, 3 m wide, cone shaped trees are randomly distributed across a 30 m wide and 25 m long area. Axes units are meters. A 0.5 m tall grass is underneath the trees. The grass is ignited on the upwind side. The wind speed is uo = 2 m/s at the x=0 plane.
Grid resolution is 0.5 m in all directions. Note that, unlike the more resolved (~7.5 cm grids) isolated tree simulations, validation of the fuel element model at this resolution has not been completed. The simulation time is 60 s and requires about 25 cpu minutes for a single processor run and about  About 200 MB of memory is required. The trees in the above image are colored according to the temperature of the crown vegetation.
  1. Stationary line fire with transport of smoke downwind

no input file presently

Fire, trees, and smoke plume. Same domain size and tree definition as above. Axes units are meters. Fire is held stationary with bed dimensions of 2 m deep by 30
m long with a heat release rate of 500 k/m^2 (based on Australian grassland fires).  Simulation time is 30 s which requires about 9 cpu minutes on a 3.8 GHz processor. About 200 MB of memory is required.


Mock-up of deep fuel bed experiments.
input_deep_fuel_bed_2p5mBY2m.txt

Semi-2D mock up of deep fuel bed experiments. This is preliminary since fuel properties and bulk geometry are unknown. Fire spread is clearly  too fast - which is also due to assuming a 2D geometry. Single processor run required about 3.5 minutes for 60 s simulated time.
3D Run of deep fuel bed experiments
input_deep_fuel_bed_3m2m2m_4mesh.txt


Input file that has examples of the use massless tracers
input_dfb_tracers.txt


Input file that has

Deep fuel bed simulation in 3D domain 3 m x 2 m x 2 m. Four meshes are used. When each mesh is assinged a processor (i.e., four process run) the 60 sec simulation takes about 20 minute (but the fire extinguishes after 10 s).





WFDS for surface fuels such as grassland fuels (boundary fuel model)
Under development - All links below are for a currently unsupported version of WFDS
Validation of boundary fuel model to data:
  1. Simulation of Australian grassland fires.
    "A physics based approach to modeling grassland fires." to appear Intn'l J. Wildland Fire.
WFDS program downloads with boundary fuel model
  1. WFDS based on version 4.05 (February 7, 2005) of FDS:
    1. <>WINDOWS single processor executable wfdsp1_win_bndyfuel.exe  (2.2 MB)
    2. <>LINUX single processor wfdsp1_linux_bndryfuel.exe (2.7 MB)
      <>
    3. <>LINUX multiple processor mpi executable wfdsp1_linuxmpi_bndryfuel.exe (3.3 MB)
      run by typing mpirun ##  wfds_linux_5-19-05_fds4p04_mpi.exe >& out & where ## defines the number of processors. NOTE you must have a file named  "fds.data" in the same directory as the executable, this file contains only the name of the input file that WFDS is supposed to read.
    4. <><>
Input files with boundary fuel model
Vegetation is present on the bottom of the computational domain as a surface fuel. It can be thought of as being "painted" on. Note that input variable names differ from those in the fuel element model above. Both models will have idential input variable name when then are incorporated into fds5.
  1. Single processor, two-dimensional grass fire
    input_2d_fireF19.txt

Two-dimensional temperature slice. This case also outputs tracer particles carried by the plume and by the inflow (not shown here). Australian grassland fuel. 200 m (100 grid cells) by 120 m (72 grid cells) domain (2 m horizontal grid resolution, 1.67 vertical resolution). Axes units are meters. Grass is ignited at upwind edge. The wind speed is uo = 3 m/s at a height of zo =2 m and depends on height according to u = (uo)(z/zo)^(1/7).  Simulation time is 60 s (this takes about 1.2 cpu miniutes on a 3.8 GHz processor).  Requires about 68 MB of memory.
  1. Single processor, three-dimensional grass fire using AU grassland fuel
    input_fireC064_1grid.txt

Fire perimenter and smoke plume. Australian grassland fuel parameters. 300 m by 300 m horizontal domain (0.6 m grid resolution); 80 m tall domain (0.6 m cell near ground, stretches to 2.2 m at top). Computational grid is 180(x)  by 180(y) by 72(z); over 2 million cells. Inert fuel break (dark color) surrounds grass. Axes units are meters. Grass is ignited at upwind edge in a time dependent manner (from center out to edges) to recreate field ignition procedures. The wind speed is uo = 7 m/s at a height of zo =2 m and depends on height according to u = (uo)(z/zo)^(1/7). Simulation time is 125 s (this takes about 9.6 cpu hours on a 3.8 GHz processor).  Requires about 1.2 GB of memory.
  1. Two processor, three-dimensional grass fire using AU grassland fuel input_fireC064_2proc.txt

Fire perimeter and smoke plume. The domain is split between the two processors as bordered by the colored lines. Same fuel parameters, domain size, grid resolution as case above (input_fireC064_1grid.txt) but runs on two processors. Axes units are meters. Required 5.6 cpu hours on dual 3.8 GHz processor computer. For details on running with multiple processors see the FDS user guide.

  1. laboratory experiment
    three-dimensional laboratory simulation of fire spread along a pine needle bed.
    input_lab_pineneedles.txt

Fire line and smoke plume is shown. Domain in 4 m (x) by 1.25 m (y) by 2.2 m (z) with 80 (x) by 25 (y) by 45 (z) grid cells. Horizontal grid resolution is 5 cm; vertical starts at 2.5 cm at bottom and stretches to 20 cm at top. The fire spread from left to right in zero ambient wind. Green rectangle shows were the pine needle bed is. The color should change from green to something darker to denote what has been burned. This capability will be added to Smokeview. Simulation time of 300 s required about 3 hours of cpu time on a 3.8 GHz processor;66 MB of memory were required.