AHPCRC Projects
Project 2–1: Dispersion of Biowarfare Agents in Attack Zones Principal Investigators: Mark Jacobson, Gianluca Iaccarino, and Eric Shaqfeh (Stanford University) |
 |
|
|
 |
|
| Clockwise from top left: Simulations of wind velocity, black carbon tracer concentrations, surface solar wattage, and visibility for Oklahoma City, early morning. |
Simulation of turbulent air flow in the wake of a building. | |
| Graphics this page courtesy Mark Jacobson, Gianluca Iaccarino (Stanford University). | ||
Planning an emergency response to a toxic substance release requires knowing the air circulation patterns around rivers, buildings, and ventilation systems and knowing whether the substance is more likely to stay near the ground or remain high in the air. This knowledge enables responders to know how large an area to evacuate, how long the substance is likely to linger, and the speed and direction of toxin dispersal. Real-world trial-and-error methods for emergency response planning are largely impractical, making computer simulation a key tool in disaster response planning. Computer simulations take many variables into account, integrating real-world information from many sources to construct predictions and “what-if” scenarios. Simulations must not only factor in numerous parameters and variables, but they must also model agent behavior over a sufficiently long time span to produce useful results. Massively parallel codes and HPC facilities provide the computing resources necessary to run realistic simulations of this nature. AHPCRC simulations using downtown Chicago as a test case have shown significant differences in the dispersion of biological warfare agents as a function of local topographical features, such as rivers and buildings. Recent simulations of downtown Oklahoma City have been compared with field study data obtained under various weather conditions and at different times during the day. The computational studies take a two-tiered approach, in which different mathematical methods are used for localized phenomena and large-area effects. Information is passed back and forth between the models at each scale in order to improve the overall accuracy of the results. Another part of this study addresses the problem of simulating the flow of air or water around an arbitrarily shaped object, such as a building. In contrast with the city-scale simulations, this part of the project models phenomena at the individual building scale. Methods under development conserve several key physical properties, including energy, mass, and vorticity. Previous schemes have not included this capability. Work in progress includes modeling air flow over uneven terrain. |
||
