Coupled microscale-mesoscale modeling of contaminant transport in urban environments

The threat of a chemical and(or) biological weapon release in an urban environment is increasingly becoming a concern to the military as well as the civilian sector. Accurate and detailed contaminant transport modeling that accounts for complex physical processes and varying meteorological conditions in geometrically-complex environments is required to effectively support consequence management doctrine, training, and operations. Complex-geometry Computational Fluid Dynamics (CFD) models coupled with atmospheric models provide the framework for an effective chemical/biological modeling system. This research involves the one-way coupling of FAST3D-CT, a high-resolution contaminant transport model, with COAMPS, a meso-scale meteorological model. Methodologies were developed for the spatial and temporal interpolation of the COAMPS-generated wind field to be applied as transient boundary conditions to FAST3D-CT.

The relative difficulty in coupling a meso-scale atmospheric code to a micro-scale CFD code can be understood when comparing the relative length and time scales. At the finest resolution in production use, COAMPS utilizes a roughly 3-km square horizontal grid with stretched vertical spacing (10 m first zone) and a corresponding time step of 10 seconds. Length and time scales used by FAST3D-CT are on the order of meters and tens of milliseconds, respectively. With such disparate scales, it is intuitive that a sample of COAMPS data at its finest time step would be required to accurately represent boundary conditions to the CFD code. However, it would be computationally expensive to write out data from COAMPS at that resolution. Furthermore, even a simple linear interpolation scheme used to determine boundary conditions within FAST3D-CT incurs a computational cost if performed at each time step. In order to efficiently utilize the COAMPS-generated wind fields as boundary conditions to FAST3D, the sensitivities to sampling rate, interpolation frequency, and representation of atmospheric turbulence on the resolved micro-scale grid need to be identified. To this end, an investigation as to these sensitivity effects on solution accuracy was conducted and the results are presented herein.

Initial investigations using actual hindcast data for the March 20, 1996 passage of a cold front through the Washington, DC, area were made with promising results. This cold front presents a 140 degree shift in wind direction over a 3-hour period. Procedures for the investigation of coupling and sampling rate effects on numerical stability and accuracy were developed. Sample calculations are made on a test grid comprising simple block buildings with heights comparable to the DC Mall area.