Physics-based Modeling of Wildland-Urban Intermix Fires

Ronald G. Rehm and David D. Evans
National Institute of Standards and Technology, Gaithersburg, Maryland 20899

This talk discusses the theoretical and experimental program at NIST to develop physics-based models for fires in the Wildland-Urban Intermix. These fires arise when wildland burning invades the built environment. Fire models for ignition and spread must consider individual fuel elements of both vegetation and structures in order to assess fire risk of developed properties. Successful prediction of wildland fire spread has been accomplished through "operational" mathematical models based on empirical correlations for wildland fuels. They fail, however, when the fire spreads to the built environment where the empirical correlations no longer apply. Property owners and communities need guidance in managing the urban forest and built environment to decrease the risk of losses to wildfire. The Oakland and Berkeley Hills fire of October 21, 1991, and the Los Alamos fires of May 2000 are examples of community-scale fires that attracted national attention in the US. The potential fuel loadings for various land uses demonstrate that structures generally provide much higher loadings than wildland fuels do. While this comparison is useful, it could also be misleading since generally, not all of the potential fuel in either the wildland or the built environment will burn. Furthermore, generally, the time scales for ignition and the heat release rates for the wildland fuel and the fuel in the structures will be widely disparate, and these differences will influence both the spread rate of the fire and its persistence. As the main tool for these studies, we are using the mathematical/computational model originally developed at NIST to study fires in buildings and known as the Fire Dynamic Simulator (FDS). The FDS is based on the laws of physics (conservation of mass, momentum, energy and species), is in the public domain (downloadable from the Website http://www.bfrl.nist.gov/ ) and is widely used by Fire Scientists and Fire Protection Engineers around the world. It utilizes the methods of Computational Fluid Dynamics (CFD), particularly Large Eddy Simulations (LES), and includes models of combustion and of radiative transport that are commensurate in complexity with the CFD model.

This FDS model is now being generalized to study fire spread on a neighborhood or community scale. The model requires much higher resolution data than any earlier models, including the local topography and micro-meteorology; location, structural characteristics and material properties of buildings; and placement and burning characteristics of vegetation. It also requires quantification of burning behavior of a single tree or building. No operational models of fire spread today require such detail. In return, the generalized FDS model can be expected to provide a time dependent prediction of fire spread in wildland-urban neighborhoods, a capability that does not exist at present! Data on the burning of single fuel elements will be obtained in large-scale laboratory fire tests and full-scale field burns. Insight from the FDS simulations of burns on single parcels of land containing structures and vegetation are now being used to construct, in collaboration with the US Forest Service Pacific Southwest Research Station in Riverside, CA, a user-friendly, web-based model for assessing community vulnerability to wildfire. Data collected by the Riverside Laboratory on individual parcels of land in the community of South Lake Tahoe, California, are currently being used to demonstrate a prototype of this model