@conference {bnh-6292, title = {Incorporating firebrands and spot fires into vorticity-driven wildfire behaviour models}, booktitle = {23rd International Congress on Modelling and Simulation}, year = {2019}, month = {12/2019}, abstract = {

Complex modes of fire behaviour resulting from local coupling between the fire and the atmosphere are a significant challenge for rapid operational wildfire spread simulations. While threedimensional fully coupled fire-atmosphere models are able to account for many types of fire behaviour, their computational demands are prohibitive in an operational context. Two-dimensional fire spread models have much lower computational overhead, but are generally not able to account for complex local coupling effects and cannot provide a three-dimensional flow structure suitable for modelling the transport of firebrands. In this paper we investigate extending two-dimensional fire spread simulations to model local coupling effects resulting from wind flow over a ridge that can result in a number of non-intuitive modes of fire behaviour. These include fire propagation opposite to the direction of the prevailing wind on the lee slope of ridges caused by re-circulation on the lee slope, called vorticity-driven lateral spread (VLS). Furthermore we develop extensions of these two-dimensional models to incorporate three-dimensional firebrand transport and show that enhanced downwind spot fire formation can result under certain VLS conditions.

The spread of fires under VLS conditions is driven by vortices in the ground plane. A model for the production and effects of these vortices was incorporated into computational simulations using a vector potential formu-lation in similar manner to a scalar {\textquoteleft}pyrogenic potential{\textquoteright} model, detailed in earlier studies. Firebrands were incorporated using a Lagrangian scheme to model transport through the atmosphere and a sub-scale model for spot fire creation and growth. The firebrand transport took factors such as drag, gravity and buoyancy into account. As effect of plume buoyancy on firebrands under real-world conditions for this scenario is currently unknown, the plume buoyancy was parameterised using a exponential decay model. The sensitivity of the decay parameter in this model was then examined in relation to the resulting spot fire distribution and area burnt. All simulations were carried out using Spark, a wildfire prediction framework.

The coupled VLS and firebrand transport simulations indicated that a higher value of decay parameter, rep-resenting a higher cooling rate of the plume, acted to enhance the lateral spread as firebrands were lofted for shorter times and were caught in the vortices at the edge of the lateral spread region. In contrast, a lower value of decay parameter, representing a lower cooling of the plume, resulted in widespread downwind spot fires and larger burnt areas. This appeared to be due to longer lofting times resulting in firebrands being transported further downwind and away from the vortices within the lateral spread region. The model appears, at least qualitatively, to match observed lateral spread and {\textquoteright}deep flaming{\textquoteright} fire behaviour although many of the parame-ters in the model require further research and experimental calibration. Further development of the model may allow these complex modes

}, keywords = {dynamic fire propagation, embers, Firebrands, spotting, wildfire simulation}, doi = {https://doi.org/10.36334/modsim.2019.H7.hilton2}, url = {https://mssanz.org.au/modsim2019/H7/hilton2.pdf}, author = {James Hilton and Garg, N and Jason J. Sharples} }