Cleaner Air at Less Cost

Across the United States, cities are working to comply with the National Ambient Air Quality Standard for ozone, a primary constituent of smog. Ozone levels relate directly to the amount of volatile organic compounds (VOCs) dumped into the air by all manner of human activity, from drycleaning to transportation to assaulting underarms with aerosols. Compliance is costly -- estimates run as high as $25 billion annually by the year 2000, and the outcome for many control strategies is difficult to anticipate.

The current federal effort to manage VOC emissions takes a shotgun approach. Some VOCs contribute more to ozone levels than others, but compliance guidelines target VOCs collectively. This undermines the goal of cleaner, safer air, says air-pollution control engineer Ted Russell. "Since one VOC might be cheaper to manage than another, a city strapped for cash could attain compliance using the most expedient, economical approach while harmful VOCs enter the atmosphere."

Russell and fellow researchers Michelle Bergin and Erik Riedel are using high-performance computing to design scales that can be used on a nationwide basis to help cities target the worst VOCs. Using Pittsburgh Supercomputing Center's DEC Alpha SuperCluster, they have shown such scales are feasible. The new approach could save tens of millions of dollars while attaining VOC reductions equal to or greater than what is achieved currently.

VOC Reactivity: A New Approach

Thirty miles above the Earth, the ozone layer protects living things by filtering out the Sun's damaging ultraviolet rays. Earth-bound ozone offers no such benefit. It reduces crop yields by $3 billion annually and contributes to lung disease. The problem occurs when hydrocarbons, a class of VOCs, react in the atmosphere with nitrogen oxides, a combustion byproduct. Mixed by winds and cooked by the sun, they combine to create smog. Automobile exhaust is a primary culprit, simultaneously spewing hydrocarbons and nitrogen oxides.

By offering a means to compare VOC reactivity levels in like products, such as gasolines, a reactivity scale can weigh the ozone-causing potential of one VOC source against another. Each VOC is assigned a value that reflects the amount of ozone it will produce. But design depends on the urban setting in question and the level of nitrogen-oxide emissions expected, so a scale for Los Angeles, for instance, might not be useful for New York.

Despite atmospheric differences among cities, says Russell, data suggest differences in VOC reactivities across the nation tend to balance one another out. Using existing scales as a springboard. the Russell team designed a scale that uses a "relative" approach to measuring compounds.

In the case of gasolines, for instance, the most important issue is how an alternative fuel measures up against the status quo fuel. When used for Los Angeles, for example, the relative approach produced results very close to those from the scale specifically tailored to that environment.

"You can use these scales more globally," says Russell, "and you can do that because of the relative relationship between reactivities shown by our calculations." Moreover, results can be used to target VOCs, improving management strategies across the United States. "A relative scale will reveal, for example, that emissions from source x are twice as reactive as source y, so it's twice as beneficial to reduce a pound from source x."

A Dynamic Picture

While existing scales were created using atmospheric modeling without spatial variation (zero-dimensional) and based on ozone episodes of short duration, Russell employed 3-D modeling to simulate a three-day episode. The 3-D approach handles movement and mixing of compounds, offering a dynamic rather than static atmospheric representation. "If you envision the study area as a box," says Russell, "the atmosphere in the zero-D model is exactly the same everywhere -- at the ground, all the way up, and from one end of the city to the other."

To complete these calculations in time to meet a California Air Resources Board deadline, Russell and his colleagues teamed PSC's SuperCluster with 20 DEC Alpha processors at Carnegie Mellon University. Running continuously on this network of 30 machines, their distributed code finished the modeling in a week. The research produced three new scales: one for predictions of peak ozone scenarios, another to gauge vegetation exposure, and a third to judge exposure to humans. "One of the powers of 3D modeling," says Russell, "is we get a better spatial representation of population exposure, and what you're really interested in is the level of ozone people are exposed to."

Russell looks forward to adapting his models for use on the CRAY T3D. "The T3D will let us develop models faster and link them to other models faster." In particular, his plans include coupling his air pollution models to geographic information systems (GIS), significantly broadening the research possibilities. GIS can hold a huge range of information that make it possible to better assess exposure to specific populations. "You could look at population by income and gauge their exposure to pollution. Or with the click of a button you could determine the effects of heavy usage of a particular highway system."

Three-dimensional air-quality modeling provides detail missing from past simulation efforts, accounting for dynamic variables such as movement and mixing of compounds as they are influenced by source location, elevation, exposure to sunlight, time of day, wind conditions, temperature and other factors. This image, from modeling by Russell and colleagues with data from 1987, uses satellite imagery to represent the landscape. The red cloud encloses the region where nitrogen oxide concentration exceeds the EPA standard. The yellow cloud encloses the region where concentration is above normal but not exceeding EPA standard.

Researchers: Armistead Russell, Carnegie Mellon University.
Hardware: DEC Alpha SuperCluster
Software: User developed code.
Keywords: National Ambient Air Quality Standard, ozone, volatile organic compounds, VOC, hydrocarbons, nitrogen oxides, air pollution, EPA, smog, air quality.

Related Material on the Web:
Ted Russell's Page.
Projects in Scientific Computing, PSC's annual research report.

References, Acknowledgements & Credits