Large Eddy Simulation of an Urban 2000 Experiment with Various Time-dependent Forcing

Under the sponsorship of DOE and DHS, we have developed a CFD model for simulating flow and dispersion of chemical/biological agents released in the urban environment. Our model, FEM3MP, is based on solving the three-dimensional, time-dependent, incompressible Navier-Stokes equations on massively parallel computer platforms. The numerical algorithm uses the finite element method for accurate representation of complex building shapes and variable terrain, together with a semi-implicit projection method and modern iterative solvers for efficient time integration. Physical processes treated in our code include turbulence modeling via the RANS (Reynolds Averaged Navier-Stokes) and LES (Large Eddy Simulation) approaches, atmospheric stability, aerosols, UV radiation decay, surface energy budget, and vegetative canopies, etc.

Predictions from our model are continuously being verified and validated against data from wind tunnel and field experiments. One such example is a recent model evaluation study performed by Chan, et al. (2004), using data from Release 1, IOP7 of the Urban 2000 field experiment conducted in Salt Lake City (Allwine, et al., 2002). The experiment selected for the study was conducted under light and highly variable winds, with a ground level, line source of SF6 released from the South side of the Heber Wells building for 60 minutes. Results from their simulations demonstrate that time-independent forcing is inappropriate and, in order to simulate successfully the dispersion experiment, it is necessary to use appropriate time-dependent forcing on the inflow boundary.

In this study, the effects of turbulent fluctuations of the inflow velocity on the dispersion results are investigated. To this end, LES simulations were performed with various time-dependent boundary conditions, which were constructed from the 1-second sonic data collected on the rooftop of the City Center building without time-averaging, with a 2-minute running time-averaging, and with a 5-minute running time-averaging, respectively. The figures below show a comparison of the time-averaged concentration patterns (for time=50-55 minute of the release) on the z=1 m plane predicted by simulations with boundary conditions based on the 5-minute time-averaged sonic data (left panel) and those based on the actual 1-second sonic data (right panel), respectively. Data from the gas samplers near the source are superimposed as small squares with the same color scheme. Although there is a reasonable agreement between the predicted concentration patterns and the observed data from the simulation using the 5-minute time-averaged sonic data as boundary conditions, the predicted plume has missed a number of gas samplers in the West and Northwest of the Heber Wells building. On the other hand, results from the simulation with time-dependent forcing based on the actual 1-second sonic data are in much better agreement with the measured data, attributing apparently to a more realistic representation of turbulence coming into the computational domain from the inflow boundary.

In addition to using field measurements, we are also investigating the feasibility and performance of our model, with time-dependent forcing provided by the COAMPS model (Leach, et al., 2002). In general, nested COAMPS runs are made first and results from the finest nest are used to generate the required time-dependent forcing for FEM3MP simulations. More specifically, time-dependent boundary conditions are constructed by combining appropriately the random perturbations based on the values of TKE and the mean velocity components of COAMPS predictions and passed on to FEM3MP model simulations.

More detailed results and findings from the above simulations will be reported at the conference.

References

Allwine, K., J. Shinn, G. Streit, K. Clawson, and M. Brown, 2002: Overview of Urban 2000, Bulletin of the American Meteorological Society 83 (4), 521-536. Chan, S., T. Humphreys, and R. Lee, 2004: A Simplified CFD Approach for Modeling Urban Dispersion, 2004 AMS Annual Meeting, Seattle, WA, Jan. 11-15, 2004. Leach, M., S. Chin, J. Leone, Jr., G. Sugiyama, and H. Walker, 2002: Urban Effects in Numerical Models and Evaluation with Field Experiments Data, Part III: Comparisons to Tracer Data, 4th Symposium on the Urban Environment, Norfolk, VA, May 20-24, 2002.

Acknowledgements

This work was performed under the auspices of the U.S. Department of Energy by the University of California, Lawrence Livermore National Laboratory under contract No. W-7405-Eng-48.