Numerical modeling of the effects of fire-induced convection and fire-atmosphere interactions on wildfire spread and fire plume dynamics

It is possible due to present day computing power to produce a fluid dynamical physically-based numerical solution to wildfire behavior, at least in the research mode. This type of wildfire modeling affords a flexibility and produces details that are not available in either current operational wildfire behavior models or field experiments. However before using these models to study wildfire, validation is necessary, and model results need to be systematically and objectively analyzed and compared to real fires. Plume theory and data from the Meteotron experiment, which was specially designed to provide results from measurements for the theoretical study of a convective plume produced by a high heat source at the ground, are used here to evaluate the fire plume properties simulated by two numerical wildfire models, the Fire Dynamics Simulator or FDS, and the Clark coupled atmosphere-fire model. The study indicates that the FDS produces good agreement with the plume theory and the Meteotron results. The study also suggests that the coupled atmosphere-fire model, a less explicit and ideally less computationally demanding model than the FDS; can produce good agreement, but that the agreement is sensitive to the method of putting the energy released from the fire into the atmosphere.; The WFDS (Wildfire and wildland-urban interface FDS), an extension of the FDS to the vegetative fuel, and the Australian grass fire experiments are used to evaluate and improve the UULES-wildfire coupled model. Despite the simple fire parameterization in the UULES-wildfire coupled model, the fireline is fairly well predicted in terms of both shape and location in the simulation of Australian grass fire experiment F19. Finally, the UULES-wildfire coupled model is used to examine how the turbulent flow in the atmospheric boundary layer (ABL) affects the growth of the grass fires. The model fires showed significant randomness in fire growth: Fire spread is not deterministic in the ABL, and a probabilistic prediction method is warranted.; Of the two contributors to the variability in fire growth in the grass fire simulations in the ABL, fire-induced convection, as opposed to the turbulent ABL wind, appears to be the more important one. One mechanism associated with enhanced fire-induced flow is the downdraft behind the frontal fireline. The downdraft is the direct result of the random interaction between the fire plume and the large eddies in the ABL. This study indicates a connection between fire variability in rate of spread and area burnt and so-called convective velocity scale, and it may be possible to use this boundary-layer scale parameter to account for the effects of ABL turbulence on fire spread and fire behavior in today's operational fire prediction systems.