"Harm's Way," Feature Article, August 2005

harm's way

Engineering software and microtechnology prepare a defense against terrorism.

by Jean Thilmany, Associate Editor

what would happen if a group of terrorists were to casually slink into the New York subway system and unleash an airborne contaminant like Sarin or mustard gas?

That's a scenario no one wants to think about. Yet it must be planned for. That's why researchers at government-associated labs across the nation are quietly working on the best ways to swiftly ready emergency responders in the event of a bioterrorism attack and to deal with the repercussions hours and days after the dreaded event.

Researchers at the Army High Performance Computing Research Center in Minneapolis are currently working on highly detailed models that depict up close how certain airborne particles would snake their way through a segment of a particular U.S. city. These models take into account very specific weather and wind flow.

The researchers run their huge models on Cray X1E supercomputers and rely on a powerful in-house computational fluid dynamics program they've been perfecting for 16 years. Because the research center develops technology for the U.S. Army and the Department of Defense, it'll pass those models to the Defense Department, along with related technology that allows for spur-of-the-moment modeling. That related software lets officials swiftly simulate a bioterrorism hit to best determine how to respond in the hours following an attack.

Over at Sandia National Laboratories, researchers have created a small, wall-mounted unit powered by a microchip that continuously monitors the surrounding air to check for harmful biological agents. Emergency responders could take a similar, handheld version of the unit into the field to pinpoint just who's been exposed and exactly what to do for those people.

In both cases, the scientists are calling upon hardware and software also used by mechanical engineers.

The Army research center makes use of supercomputers running CFD software, which mechanical engineers commonly use to model airflow and fluid flow, often in the aerospace and automotive industries.

The heart of the Sandia scientists' unit is powered by a tiny chip, commonly called a lab-on-a-chip. This chip, based on the same technology as the silicon chip that powers your computer, essentially shrinks all the beakers, pipettes, and processers housed on the lab bench to a single chip about the size of a pea. It contains all the on-board tools needed to analyze air.

Wind Between Buildings

Sharouz Aliabadi, Northrop Grumman professor of engineering at Jackson State University in Mississippi, works in conjunction with the Army High Performance Computing Research Center to put the powerful CFD software to work. The 16-year-old research center develops technologies it passes on to the U.S. Army and the Department of Defense.

For the last 16 years, Aliabadi's team has been perfecting their CFD technology to carry out a series of large-scale simulations—accurate to the foot—that show in three dimensions how a biological contaminant would disperse on the wind through large U.S. cities, including New York, Washington, and Chicago.

These simulations, already created, can be called upon in time of disaster. They take into account the wind speed and direction on a particular day, airflow between particular buildings, and cloud cover. They do this by incorporating the MM5 weather model, an updated version of the model originally developed at Pennsylvania State University in the early 1970s. That model uses geological survey and land use data to derive weather information.

Those weather models, however, simulate atmospheric conditions over large portions of the country. They divide the country into segments between three and 15 miles wide, far too large a scale to accommodate the complex geometry of a few city blocks, Aliabadi said.

To bring those calculations down to the neighborhood level, he combines the CFD and the weather model, thereby projecting wind speed and temperature onto the much-smaller-segmented 3-D CFD model.

"That way, we can provide more accurate simulations of chemical dispersion and take into account the predicted future weather conditions," Aliabadi said.

The computational program he and his team use for fluid analysis—in this case, wind and airflow—contains a mesh comprising hundreds of millions of cells. Such a tight mesh, or grid, when overlaid with the weather pattern, lets simulators calculate where even tiny portions of contaminant can drift on the wind. The problem is, it can take an hour to run the model, and in an emergency, every second counts.

The Army High Performance Computing Research Center in Minneapolis is working on detailed models that depict how certain airborne particles would snake their way through a segment of a particular U.S. city.

For that reason, the Department of Defense asked for a large number of possible dispersement scenarios for major cities to have on hand. That way, department officials could quickly draw upon the closest-matching simulation in the pandemonium following a release of dangerous airborne substances. They'd refer to the simulations to give emergency responders directions on how best to respond to the release.

"They have these simulations for many different scenarios," Aliabadi said. "If you want to know how the air flows behind a building in the city, you store this, and if the release happens, you then have the flow pattern in the city at hand."

The array of on-hand scenarios also takes into account the point of release and the agent sent into the air. Aliabadi and his team have run the scenarios ahead of time to save precious response time.

"Those kind of scenarios require a lot of computational time, so you have to precompute this and have it stored," he said. "It's a very large data set and takes a lot of time to compute. But if something happens, you already have a good idea of how it will disperse. You wouldn't be 100 percent sure, of course, but sure enough that you'd have a good enough idea of how to respond."

Aliabadi, however, can't say exactly how Defense Department officials plan to store or to use the already-performed simulations.

"We're actually a Department of Defense collaborative laboratory, and we transfer this technology to them," he said. "How they're going to implement it is beyond my expertise."

To create a model that's accurate to the particular attack, Aliabadi and his crew have a second method for officials to call upon in the event of an attack.

His group's simulation capability could, shortly after an attack, model what just happened to predict how the contaminant will travel on the air in the next hours and days. Those quickly produced, but huge, models would be populated with actual strike statistics and would therefore be even more accurate than the simulations prepared in advance, Aliabadi said.

"We'd need to model a situation within seconds after it happens," he added. "We have a short amount of time to do a simulation on such a very large scale."

He envisions firefighters, police officers, and ambulance drivers responding according to scenarios already on hand at the Defense Department. Then, after officials run a simulation with actual strike specifics, they can tailor their emergency response.

"That way, they could evacuate the most exposed area. Then, using the new model, they can focus on the area that may be exposed in the near future," Aliabadi said. "You could see with the new model the time this chemical would take to disperse, and how long it would take to travel from point A to point B."

A supercomputer could return initial simulations in a matter of about five minutes.

Aliabadi said they would be low-fidelity simulations that relied upon a coarse mesh—rather than the mesh made up of hundreds of millions of points—but they could still return quick, useful answers.

"They might not be very accurate, but those rough answers can be given to agencies so they can start an evacuation," Aliabadi said.

A later simulation run with the very fine mesh would likely be used days following the immediate response. Because of the computing power involved, this second simulation might take a day or even two to return results, he added.

"These could be used during phase two, to help put people's lives back together," Aliabadi said. "It'd be done on a finer scale with higher detail and a greater degree of accuracy," he said. "This would show how you want to decontaminate these areas."

Inconspicuous on the Wall

Sandia's handheld or wall-mounted monitors are about the size of a telephone. They don't analyze events, but they can test continually for the presence of harmful airborne agents with the help of analysis chips and the software that powers them.

Lab-on-a-chip technology performs analyses in a fraction of a minute that would take hours with traditional laboratory methods, said Art Pontau, bio detection program manager at Sandia National Laboratories in Livermore, Calif.

Sandia researchers have inserted the chips within small machines to function as sniffer dogs, detecting chemical agents released into the air. The lab is looking to partner with industry to license the system, called MicroChemLab, for commercial use, Pontau added.

Although those units were originally developed for national security, defense applications, and first responders, a variety of applications exist for their use in the chemical diagnostic market, he said. Possible commercial markets include air and water quality detection, medical diagnosis, biotechnology, and industrial process control.

In all those activities, detecting unwanted, dangerous, or even benign chemicals can be important.

Sandia Laboratories' handheld or wall-mounted analysis units are approximately the size of a telephone. They rely on analysis chips to test for harmful airborne agents.

The lab's MicroChemLab system verifies chemical, biotoxin, and pathogen signatures in the environment. In tests, the small unit detected seven different forms of the biotoxin ricin, a highly toxic compound that comes from the castor beans used to make castor oil. About one million tons of the beans are processed every year and the residue, when boiled down, is ricin. So it's relatively easy for terrorists to get their hands on the stuff.

Ricin detection was tested because it's a particular threat to national security; it can be easily and cheaply produced. Less than a pinpoint of the substance can kill a human if ingested, injected, or inhaled.

Sandia is also working with two companies to develop chemical detection units that will continuously monitor water systems.

That system would analyze biospecies in the water every half-hour and could operate for weeks between maintenance cycles, Pontau said. Sandia is now testing a field unit and expects to have actual water-testing applications up and running in about a year.

"There's a little probe inserted directly into water lines; a small amount of water is brought into the system and processed by concentrating species of concern and throwing away everything else," he said. "We're basically analyzing microliquids."

The species of concern are extracted and concentrated through solid phase extraction and other concentration techniques. They need to be condensed because they appear only in concentrations of a few per liter and the unit analyzes a microliter of fluid.

"We want to make sure the microliter contains the agent we're looking for," Pontau said.

Another handheld unit made by Sandia looks rather like a handheld telephone: It runs on batteries, contains the chip, and can carry out hundreds of analyses over the course of the day. Pontau envisions first responders taking these units with them on emergency calls.

"They could analyze a white powder on the spot to see if there's anthrax in it," he said.

Sandia researchers, with funding from the Department of Defense, seek to perfect a similar wall-mounted unit so it could one day be hung in a subway station or in another public area. The unit would while away the day collecting air samples and analyzing them.

Those systems would concentrate an air sample to the microliter and analyze for particles sized from 1 to 10 micrometers, the sizes that would be most harmful if breathed.

No one likes to think about a bioterrorism attack, but it's a good thing that researchers like Pontau and Aliabadi are willing to have our backs.