CSE Symposium Keynote

Dr. Jay Boris, Chief Scientist
Laboratory for Computational Physics & Fluid Dynamics,
U.S. Naval Research Laboratory

TITLE: Dust in the Wind: Challenges for Urban Aerodynamics

DATE: Thursday, April 21, 2005
TIME: 3:00 P.M.
PLACE: 2240 DCL
1304 W. Springfield Ave., Urbana, IL

ABSTRACT

The fluid dynamics of airflow through a city controls the transport and dispersion of airborne contaminants. This is primarily a problem of urban aerodynamics rather than of meteorology. The space scales are short, tens of meters to a few kilometers, so meteorological effects enter primarily through the aerodynamic boundary conditions. The average flow, large-scale fluctuations and turbulence are closely coupled to the building geometry. Buildings create large rooster-tail wakes; there are systematic fountain flows up the backs of tall buildings; and dust in the wind also moves perpendicular to or even against the locally prevailing wind. The fact that surfaces are rough and have sharp edges somewhat simplifies the aerodynamics but realistic atmospheric factors also play a role. Requirements for better prediction accuracy demand time-dependent, three-dimensional CFD computations that include solar heating and buoyancy, complete landscape and building geometry specification including foliage, realistic

Computing urban aerodynamics accurately is a time-dependent, High-Performance Computing (HPC) problem. On the other hand, using this technology in the emergency assessment of industrial spills, transportation accidents, or terrorist attacks has very tight time requirements that suggest simple approximations that unfortunately produce inaccurate models. The trade-off has been the need to choose either a fast model or to live with accurate results. Using new fluid-dynamic principles, an urban-oriented emergency assessment system called CT-Analyst has been invented to solve this dilemma. It produces accurate results for airborne contaminant scenarios nearly instantly. Designed to predict all airborne contaminants including Chemical, Biological, and Radiological (CBR) threats, CT-Analyst has unique new capabilities and gives HPC accuracy while running much faster than other current alternatives.

This presentation explains how applied aerodynamics enables CT-Analyst to do this. A relatively few detailed CFD computations for a given area can be generalized to describe a wide range of wind directions and speeds, and all likely source locations using a new data structure called Dispersion Nomografs. We have developed a portable, entirely graphical software tool that embodies this new, high-resolution technology and runs on small personal computers. Real-time users do not have to wait for results because accurate answers are available with near zero-latency (that is 10 - 20 scenarios per second). Sequences of individual cases, such as a continuously changed source location or wind direction scans, can be computed and displayed as continuous-action movies. Since the underlying nomograf database is precomputed, qualitatively new real-time functions such as sensor data fusion, backtracking to an unknown source location, and evacuation route planning are possible with almost no computing delay.

BIOGRAPHY

Dr. Boris is an internationally recognized expert in developing and applying advanced computer technology to the solution of large-scale scientific and engineering problems, having conceived and led computational and theoretical studies in many areas of fluid dynamics, many-body dynamics, reactive flow, and plasma physics and developed many fluid and chemical kinetic models now in current use for simulating chemically reactive flows, hydrodynamics, solar physics, and astrophysical and atmospheric fluid dynamics. He is currently responsible for developing, supervising, and leading the theoretical and numerical research of the Laboratory for Computational Physics & Fluid Dynamics, an interdisciplinary group of engineers, numericists, fluid dynamicists, and reactive flow physicists. His current research interests include applications of parallel computing to CFD and reactive flow, development of parallel near-neighbor algorithms for molecular dynamics, and computational hydrodynamics for naval applications.