Comparison of CFD and operational dispersion models in an urban-like environment

Abstract

Chemical plants, refineries, transportation of hazardous materials are some of the most attractive facilities for external attacks aimed at the release of toxic substances. Dispersion of these substances into the atmosphere forms a concentration distribution of airborne pollutants with severe consequences for exposed individuals. For emergency preparedness and management, the availability of assessed/vali- dated dispersion models, which can be able to predict concentration distribution and thus dangerous zones for exposed individuals, is of primary importance. Air quality models, integral models and analytical models predict the transport and the turbulent dispersion of gases or aerosols after their release without taking into account in detail the presence of obstacles. Obstacles can modify the velocity field and in turn the concentration field. The Computational Fluid Dynamics (CFD) models on the other hand are able to describe such phenomena, but they need to be correctly set up, tested and validated in order to obtain reliable results. Within the project Europa-ERG1 TA 113.034 "NBC Modelling and Simulation" several different approaches in CFD modelling of turbulent dispersion in closed, semi-confined and urban-like environ- ment were adopted and compared with experimental data and with operational models. In this paper the results of a comparison between models describing the dispersion of a neutral gas in an idealized urban-like environment are presented and discussed. Experimental data available in the literature have been used as a benchmark for assessing statistical performance for each model. Selected experimental trials include some water channel tests, that were performed by Coanda at 1:205 scale, and one full-scale case that was tested in the fall of 2001 at the Dugway Proving Grounds in Utah, using an array of shipping containers. The paper also suggests the adoption of improved statistical parameters in order to better address differences between models, and to have a more straightforward method for comparing models suitable for emergency preparedness aims.