Parallel Performance of a New Model for Wildland Fire Spread Predictions

摘要

A new approach to simulation of wildland fire spread is outlined, and preliminary results of computations and parallelization of the algorithms are presented and discussed. The fire spread model is based on a porous medium representation of the forest with heat source terms replacing treatment of detailed chemistry. The formulation is implemented in conjunction with a synthetic-velocity form of large-eddy simulation (LES) which is capable of producing far more detailed results than can be obtained with usual forms of LES, but yet is very efficient and easily parallelized.