Wildland-Urban Interface and Wildland Fires

This webpage contains an overview of ongoing work at the National Institute of Standards and Technology in the areas of wildland-urban interface and wildland fires.
last updated February, 2008
Disclaimer: Any links to non-Federal Government Web sites do not imply endorsement of any particular product, service, organization, company, information provider, or content.

Postdoctoral opportunities

Contents

Contacts
1 page overview

pdf 218 kB
ppt 6MB




Introduction
Through both experiments and model development the Building and Fire Research Laboratory at the National Institute of Standands and Technology is seeking to better understand, from a fire safety point of view, fire behavior in the Wildland-Urban Interface (WUI). Fires in the WUI spread through both vegetative and structural fuels. Vegetative fuels can include those characteristic of wildlands (trees, grasses, understory growth, and ground litter) as well as those purchased at nurseries for home or community landscaping purposes (trees, mulch, grasses, and ornamental plants). Similary, a wide range of structural materials and designs are used in the construction of exterior walls, windows, soffits, eaves, decks, roofing assemblies, etc.
Fire Spread  Mechanisms in the WUI
vegetation-to-vegetation fire spread vegetation-to-structure


Fire spreading within vegetative fuels approaches a community.
 Fire spreading through vegetation in the upper part of the photo ignited a single structure.
structure-to-structure

(John Gibbons)
Fire spread occured without significant particpation from burning vegetation or significant flame contact from adjacent structures.

Fire spread between structures due to direct flame contact.


Currently our effort is divided into three catagories:
  1. Fire behavior in structural fuels
  2. Fire behavior in vegetative fuels
  3. Fire behavior in the intermix of vegetative and structural fuels
The sections below contain examples of ongoing work in each of these areas.

Large Fire Laboratory (LFL) Experimental Facility
Experiments are conducted in the Large Fire Laboratory at NIST which consists of  a 9 m x 37 m test area containing a 6 m x 6 m and a 9 m x 12 m hood for oxygen consumption calorimentry measurements (heat release rates). A description of some of the measurement capabilities in the LFL in "The NIST 3 Megawatt Quantitative Heat Release Rate Facility."

Simulations
Fire simulations are performed by a modified version of the structural fire code the Fire Dynamics Simulator (FDS) and visualized by Smokeview (a one page handout on Smokeview is here). The version of FDS used here in being modified to simulate firespread through vegetative fuels. This version of FDS is call WFDS (Wildland-Urban-Interface Fire Dynamics Simulator). WFDS is a physics based numerical modeling approach which includes all modes of heat transfer (convection, conduction, and radiation). Smokeview uses OpenGL to interactively render WFDS outputs (temperature, heat release rates, smoke, heat and mass fluxes, etc.) on animated color contour slices, isosurfaces, and solid surfaces. Both WFDS and Smokeview run on a variety of computer systems including Windows, Linux, and SGI. FDS can be run on on single or multiple processor computers.

Fire Behavior in Structural Fuels

Structural fuels: Experiment on structure-to-structure fire spread
The separation distance between structures and the materials used in their construction are both important factors in structure-to-structure fire spread.  For exampe, as seen in the above photograph from a 2003 WUI fire in San Diego, CA (USA Today),   Experiments have been conducted to investigate the likelihood that fire can spread from one structure to another. The materials and design used in the structures were either typical of current building practices or included a fire barrier. In these experiments a fire was started and allowed to burn in a typically furnished room. Eventually the fire exited a window and an adjacent structure either ignited (in the case of typical construction, shown below) or did not  ignite (when a fire barrier was included in the wall construction). In these experiments the structures were separated by 6 feet (1.8 m) which is currently allowed in some local building codes.





Downloads:
  • movie (150 MB) of experiments with and without a fire barrier in the wall construction
  • factsheet (pdf) of experiment
In the above photographs a structural fire exits an enclosure through a window and ignites an adjacent wall.

Structural fuels: Simulations of structure-to-structure fire spread


Downloads:
The snapshots above are from preliminary simulations of the structure seperation experiments discussed above. These simulations were conducted on 4 processors using measured heat release information from the room fire in the experiments. Further work is required in the modeling of the target wall which has been assumed to be spruce in the simulation.

Fire Behavior in Vegetative Fuels

Vegetative fuels: Single, isolated, tree burns (experiments and simulations)
Experiments of burning trees have been conducted in order to validate heat fluxes and heat release predictions of simulated trees using WFDS and other modeling approaches.


The sequence of snapshots are of a 2.4 m tall Douglas fir; top row are the experimental burn, bottom row are Smokeview rendered WFDS predictions.
Movies of experimental burns:
- mpeg movie (12 MB) of a 1.5 m, 3 m, and 3.8 m tall Douglas fir trees burning
- quicktime movie (5 MB) of 5 m Douglas fire
Comparison of experimental and computer simulated tree burn:
-  mpeg movie (17 MB) or avi movie (60MB) of a 2.9 m (10 ft) tall Douglas tree burning, mass loss rate, net radiation flux 
Additional information is available in the NIST technical report:  "Physics-Based Modeling for WUI Fire Spread - Simplified Model Algorithm for Ignition of Structures by Burning Vegetation"



Vegetative fuels: Grassland fire simulation results



The homogeneous fuel and lack of terrain variation in the grassland fires of Australia and Brazil make these fires good candidates for use in model validation.  The figures below are some examples of simulation results of Australian grassland fires from a current validation study.

Australian grassland experiment
200 m x 200 m plot; 5 m/s wind left to right
Ignition: over 56 s two field workers walked in opposite directions starting from the center of the left-hand-side fire break.





WFDS computer simulation of the experiment
Ignition procedure was simulated. Grass fuel is modeled as a subgrid fuel along the base of the gas phase. Convective and radiative heat transfer is accounted for. See here for preprint of paper in the Intnl. J. Wildland Fire.

Animation of WFDS simulation: mpeg (17MB) or avi (38MB)

Note: simulation domain extends ~ 700 m in all directions. Only the 200 m x 200 m grassland plot in the WFDS simulation is shown here.








The figure at left shows the steady state spread rate from the grassland simulations (symbols), BEHAVE (solid line), and from Eq. 4 in Cheney et al. (Prediction of Fire Spread in Grasslands, Int. J. Wildland Fire, 8: 1-13, 1998). 


Vegetative fuels: Intermix of vegetative fuels leading transition to crown fire.
Many of today's forests have historically dense accumulations of dead fallen material which pose a fire threat to the overall ecology of the forest and nearby communities. An important question in forest management is, therefore, how much of this material should be removed to reduce the fire threat to acceptable levels. In the example below WFDS is used to simulate a surface fire spreading through a forested region part of which has underbrush.
Plan View

Side View of Vegetative Fuel
In the above figures grass and pine needle fuel are colored green and black, respectively; the underbrush, tree trunks and tree crowns are colored blue. In the movies (see links below) this blue color changes according to the temperature of the fuel (red is hottest).

Fuel/Wind Specifications:
- grass fuel loading is NFFL 3 (tall grass, 0.674 kg/m^2)
- pine needles: 5 cm depth, 20 kg/m^3 bulk density
- underbrush: 0.5 m - 2.0 m height, 1 kg/m^3 bulk density
- tree crown: 7 m - 14 m height, 0.24 kg/m^3 bulk density (160 trees in 80 m x 80 m area)
- wind is 6 m/s from left to right

 Computational Specification:
- domain is 320 m x 320 m x 80 m (160x160x20 grid cells)
- horizontal grid resolution is 2 m, vertical is 2 m near ground and stretches to 8 m at top.

Animations from Smokeview:
- plan view quicktime movie (1.3 MB) showing fire spread
- tour view avi movie (15 MB) showing fire spread and smoke)

The animations clearly show the loss of tree crowns in areas where the underbrush was present and provided a ladder fuel for fire spread from the pine needles to the crowns. Additional information is in a talk given at the March 2004 Core Fire Science Caucas meeting in Reno, NV.


WFDS Simulation of a Stand Burn Similar to the International Crown Fire Experiment
Two wall assemblies are placed 10 m and 20 m down spread of the stand. This mimics similar experiments performed during the International Crown Fire Experiments conducted in the Northwest Territories of Canada. The simulated stand is approximately 1/4 the size of the experimental stand. A movie of the simulation is here. Time histories of the radiant fluxes on the walls are shown in the movie. The magnitudes and duration of the rise to maximum value of the fluxes are similar to the experiment. 


Vegetative fuels: Simulation of surface fire spreading under tree

wfds_one_tree_movie.avi (8.5 MB)
Example of a fire spreading through an excelsior surface fuel under a 6 m tall conifer.



Vegetative fuels: Simulation of enclosure effects: excelsior fuel bed burning in a wind tunnel

Two different simulations were conducted. The two lower images in the figure above show a side view (on the left) of the wind tunnel and an end view looking downwind (on the right). The wind tunnel on the bottom row of the figure has the same cross-section (3 m x 3 m) dimensions as the experimental facility used by the USDA Foreset Service in their Missoula, Montana laboratory. The wind tunnel in the upper row has crossection dimensions which have been doubled (6 m x 6 m). Both wind tunnels are 16 m long. A fuel bed of excelsior 1 m wide, 8 m long and 20 cm tall is placed in the center of the wind tunnel floor. A 1.8 m/s wind blow from left to right. The fuel properties of the excelsior are from an experiment conducted by Catchepole et al. (Rate of Spread of Free-Burning Fires of Woody Fuels in a Wind Tunnel, Comb. Sci. Tech., 131, 1-37, 1998). The

As can be seen in the figure above and the movies (links are below) the walls and ceiling of the wind tunnel sufficiently confine the buoyant plume that the plume itself it acts as a barrier to the incoming flow.

Links to movies:
- fuel_bed_wind_tunnel.mpg (4.6 MB)
- fuel_bed_wind_tunnel.avi (12 MB)


Fire Behavior in the Intermix of Vegetative and Structural Fuels


Simulation of a grass fire spreading into structures with different fuel treatments

Fuel/atmosphere specifications:
- grass fue loading is 0.3 kg/m^2 (between NFFL1 [short grass, 0.166 kg/m^2] and NFFL 3 [long grass, 0.674 kg/m^2])
- tree crowns have a bulk density of 4.72 kg/m^3
- underbrusg bulk density 4.72 kg/m^3
- wind is 1 m/s from right to left

Computational specifications:
- domain is 80 m x 80 m x 100 m (80x80x36 grid cells)
- horizontal grid resolution of 1 m; vertical grid is 1 m at ground and stretches to 3 m at top

Animations from Smokeview:
- quicktime movie (9 MB)
- avi movie (12 MB)

A grassland line fire spreads into structures which have varying amounts of surrounding vegetative fuel loads. The structure on the upper right-hand side has the highest vegetative fuel loading. As seen in the animations the ignition of the underbrush surrounding this structure leads to a crown fire.

Firebrand Experiments

Firebrands are a recognized, but under researched, component of fire spread and ignition. In order for firebrands to play a role in the overall fire spread process they must be generated, transported, deposited on a flammable target fuels, and finally ignite the target fuel. Relatively little work has been done on the generation, deposition, and ignition stages of a firebrand evolution.

Work at NIST is focused on the ignition of a range of target fuel types (pine needles, cedar shingle crevices, paper, hardwood mulch, cut grasses). Each target fuel is tested at different moisture levels and wind conditions. A number of brand types and brand conditions (flaming versus glowing, single versus multiple brands) were used. Brands that were generated from burning Douglas trees (2.4 m and 5 m tall) are being collected and measured.
Brand shapes used in ignition studies
The brands below were ignited and dropped on target fuels of different moistures and wind enviornments.

Disk shaped. This shape has good lofting capabilites and is an approximation of firebrands generated from burning bark and roof shingles.
Disk dimensions: diameter = 25 mm; thickness = 8 mm
Disk dimensions: diameter = 50 mm; thickness = 6 mm
A paper covering disk shaped brands is here..

Cylindrical shaped: This shape approximates branches and twigs based on brands collected from burning Douglas firs tree in NIST Large Fire Laboratory (see image at right).

Size of Douglas fire dowels at left are:
length = 76 mm; diameter = 10 mm
length = 51 mm; diameter = 5mm


Examples of target fuel beds



shredded paper
 pine needles
 shredded hardwood mulch


Illustrative results



no igntion from glowing firebrand
 flaming ignition from single flaming firebrand
 flaming ignition from multiple glowing firebrands


Brand Generator: Creating a mass flux of brands


In order to study the ignition properties of target fuels under more realistic setting a brand generator (pictured above), capable of producing a mass flux of firebrands, is being constructed. This project is still in a preliminary stage of characterizing the size, mass, and burning state of the firebrands generated from different "feeder" materials. For example, among the materials tested to date are mulched Douglas Fir trees and 50 mm long Douglas dowels. See below.

Mulched Douglas Fire tree.
movie of brand generator
(12.4 MB, avi - might not work with quicktime)

50 mm long, 5 mm diameter Douglas Fir dowels
movie of brand generator
(14 MB, avi - might not work with quicktime)


Realistic Inputs for WUI Fire Simulations

Currently NIST is funding the Coeur d'Alene Tribe GIS Program to create datasets for WUI input into WFDS. These input included the location of vegetative fuels (trees [location and size], grass, shrubs), location of non-fuel surfaces (roads, gravel patches, rock outcrops, etc.), structural fuels. A number of datasets within the Coeur d'Alene Tribe Reservation Boundaries are being used: Light Detection and Ranging (LiDAR) elevation data was the main data set (2 m resolution). In addition, 1 m resolution color imagery from the National Agricultural Imagery Program (NAIP) was used. These data sets from remote sensors were supplimented with ground surveys which can provide "truthing" of the remote sensing data. Also, following the protocols of the NFPA 1144 (Standard for Protection of Life and Property from Wildfire, 2002) building materials could be rated. Examples of this effort are show below.

Southern Study Area. Derived from a combination of LiDAR and color imagery.  Structures are colored according to NFPA rating of the roofs, siding, decks, and eaves.
WFDS represetation of the above 500 m by 500 m area using input data created by the Coeur d'Alene Tribe GIS program. Trees are shown in blue. The roofs and sides of structures are colored according to their ratings from field observation using the NFPA 1144 standard (a simplified scheme of red, blue, green for most to least combustible was used at this point).



Presentations (in reverse chronological order)

#
 Title
 Venue
17 The NIST Wildland-Urban Interface Program pdf Presentation to CALFIRE and City of San Diego Fire Department; February 2008
15 Measurement of firebrand production and heat release rate (HRR) from burning Korean pine trees Asia-Oceania Symposium (IAFSS), Sept. 20-23, 2007, Hong Kong
14 On the use of a firebrand generator to investigate the ignition of structures in wildland-urban interface fires 11 International Conference on Fire Science and Engineering (INTERFLAM), September 3-5, London, UK (2007)
13
 Firebrand production from burning vegetation 5th Intnl. Conference Forest Fire Research, Combria, Portugal, November, 2006
12
 Numercial modeling of fires spread through trees and shrubs " "
11
 Quantifying and ranking the flammability of ornamental shrubs in the southern United States 3rd Intn'l Fire Ecology and Management Congress, San Diego, Nov. 13-17, 2006
10
 Fire spread and structural ignitions from horticultural plantings in the wildland-urban interface " "
9
 Flammability rankings of commonly used horticultrual plant in the southern U.S. " "
3
 Numerical simulations of grassland fire behavior from the LANL-FIRETEC and NIST-WFDS models (paper) (presentation) (avi)

 EastFIRE Proceedings, May 11-13, 2005 George Mason University, Fairfax, VA
presentation
2
 Physics-Based Modeling of Community Fires International Interflam Conference, 10th Proceedings. Volume 2. July 5-7, 2004Edinburgh, Scotland, Interscience Communications Ltd., London, England, 11 p., 2004
1
 Neighborhood-Scale Fire Spread Fire and Forest Meteorology, 5th Symposium. Joint With 2nd International Wildland Fire Ecology and Fire Management Congress. Proceedings. November 16-20, 2003, Orlando, FL, 8 p., 2003


Archival Publications (in reverse chronological order)

# Title
 Journal
8 Mass and size distribution of firebrands generated from burning Korean Pine (Pinus Koraiensis) Trees Fire and Materials, to appear
7 On the development and characterization of a firebrand generator Fire Safety Journal, to appear
6 Experimental investigation of firebrands: Generation and ignition of fuel beds Fire Safety Journal, 43, 226-233, 2008
5 Firebrand generation from burning vegetation Int'l J. Wildland Fire, 16, 458-462, 2007
4
 A physics-based approach to modeling grassland fires (preprint)
accessory publication with model details is here or here 		Int'l J. Wildland Fire, 16, 1-22, 2007
(note the accesory pub. is here too)
3
 Ignition of mulch and grasses by firebrands in WUI fires
 Int'l J. Wildland Fire, 15: 427-431, 2006
2 An evaluation of the fire plume properties simulated with the FDS and Clark coupled wildfire model Canadian J. Forest Research, 36, 1-15, 2006
1
 On the ignition of fuel beds by firebrands
 Fire & Materials, 30:77-87, 2006


Other publications (in reverse chronological order)

#
 Title
 Venue