A Hybrid Numerical Model for Simulating Atmospheric Dispersion

Abstract

A hybrid numerical model for simulating atmospheric contaminant dispersion is developed. The hybrid numerical scheme employs an hp-adaptive finite element method coupled with a Lagrangian particle transport technique to solve the governing equations for atmospheric flow and species transport. A random walk/stochastic approach is used to generate Lagrangian particles that define the contaminant dispersion traces. A coarse mesh using low order shape functions is initially generated. Both the mesh and shape function order are subsequently refined and enriched in those regions where high computational error exist. Compared with fine mesh and high order numerical solutions, the hybrid scheme produces highly accurate solutions with reduced computational cost. A general probability distribution is used in the particle transport module for the random component of motion due to turbulent diffusion. Results depicting contaminant transport and dispersion in the atmosphere are presented. The computational efficiency of the hybrid numerical model is also discussed.