A computationally efficient particle-puff model for concentration variance from steady releases

This paper presents a computationally efficient particle-puff model that can be used to calculate both average concentration and concentration variance in turbulent flows. The model is self-contained in the sense that it does not need externally supplied parameterizations of puff spread, but rather calculates the spread with an internally-built module for relative dispersion of particles— all existing particle-puff models rely on parameterizations of puff spread derived from theoretical considerations and observational data for dispersion of a point source in isotropic turbulence. Preliminary evaluations show that the model performs as well as a computationally demanding two-particle trajectory model in predicting the mean concentration and concentration variance in two different flows: isotropic turbulence and a wind tunnel boundary layer. Because of its numerical efficiency, the present particle-puff model can be built into conventional, regulatory air quality models as a module to calculate the mean concentration and concentration variance in the near field. The reason for the numerical efficiency of present particle-puff model is that it needs to track only thousands of particle-pairs in order to calculate the puff spread, while (the original) two-particle Lagrangian stochastic (LS) models need millions of particlepairs to achieve statistically stable predictions.