Sandia Examines 'Option of Last Resort'

Scientists and engineers at Sandia National Laboratories are fairly confident that April 13, 2036 will be an uneventful day in the history of Earth. Still, they want to be sure it stays that way.

That's why a group at Sandia is employing a supercomputer with 13,000 AMD Opteron processors and 50 teraflops (trillions of floating point operations per second) of number-crunching power to determine how to stop an asteroid that could be headed our way on that day.

The asteroid, called Golevka, is unlikely to strike Earth. Scientists set the possibility at about 1 in 8,000. Moreover, most observers believe they'll fully understand Golevka's impending trajectory several years in advance of a strike.

Still, a growing number of experts say a little insurance plan might be in order. "When you're talking about saving or losing billions of lives, it's important to have a backup plan," notes Mark Boslough, a scientist at Sandia. "We need to have an option of last resort."

Destruction on a national scale

That's why Sandia's group has begun using the supercomputer, built by Cray Inc, to determine how to push Golevka off course, if it is indeed ultimately headed for Earth. Known as Red Storm, the computer has run repeated simulations of a scenario in which a nuclear-explosive-laden rocket crashes into the asteroid, detonates, and knocks the speeding rock off its collision course.

By most accounts, the simulations are necessary. The oblong-shaped Golevka, measuring about 320 m across, could cause enormous damage. If it strikes Earth, Golevka could unleash energy comparable to approximately 1,000 megatons (Mtons) of TNT and create a crater about 5 km in diameter. Moreover, it might lead to destruction on a national scale, scientists say.

"Golevka could take out a country larger than France and smaller than the United States," notes Bill Camp, director of computing and mathematics at Sandia. "The threat is very real."

If it strikes land, scientists say, Golevka will immediately incinerate cities and forests across a huge swath of Earth. In the ocean, the results would be no better. The speeding rock would ultimately vaporize when it strikes, possibly deform the ocean floor, and eject miles of superheated water vapor, melted rock, and other debris into the atmosphere, causing climate problems.

Given the catastrophic possibilities, Sandia scientists decided last year to put Red Storm to work on a simulation. In the simulation, they assumed that Golevka is speeding toward Earth, and that they can drive a rocket into it, approximately 150 m below its surface. There, somewhere near its center of mass, they'd detonate a nuclear device, possibly in one huge explosion, but more likely in a series of smaller nuclear detonations. That's where Red Storm comes in. The supercomputer's job is to help scientists determine the best way to push the asteroid off course. It computes a long list of variables — including stress, temperature, pressure, kinetic energy, volume fraction and more — in more than a billion computing cells inside and on the surface of the asteroid, millions of times per second. The mathematical problem is so "compute intensive" that no workstation could ever come close to doing it. For most of the runs, which take about 30 hours, Red Storm has employed 7,200 processors working in parallel. In bigger runs, it has used 9,000. Ultimately, it could use as many as 13,000 microprocessors.

Sandia mathematicians and scientists have rerun the simulation numerous times, employing different nuclear yields at varying depths beneath the asteroid's surface. In essence, they've created a matrix » using nuclear yields ranging from 10 ktons to 10 Mtons, versus numerous depths of detonation inside the asteroid.

"We've learned that if you don't do it just right, you might leave some big pieces of the asteroid behind, and they might still be on a collision course," Boslough says.

Sandia scientists liken the process of detonating the asteroid to using a cue ball to break a rack of 15 pool balls. By hitting the rack just right, they say, a good pool player can control the break, much as they must control the breakage of the asteroid, lest they cause large pieces to continue on toward Earth.

"At a great depth with a large nuclear device, we can break the asteroid to smithereens," Boslough says. "But while breaking it to smithereens makes for an interesting simulation, it's probably not the best way to do it."

Ultimately, Boslough says, scientists want control, rather than chaos, when they break the asteroid. Most agree that Newton's Law, which describes an equal and opposite reaction for every action, offers the most intelligent solution. By using a series of smaller detonations, they say, they hope to create "a cloud of debris" that will in turn push the remaining pieces off course.

"It works like a rocket engine," Boslough says. "By burning gas and shooting it out one side, the other side is accelerated in the opposite direction."

Putting a plan in place

To be sure, many scientists believe there will be plenty of advance warning if Golevka appears to be headed for Earth. It's due to swing past our planet in a near-miss scenario in 2012, and will come back for another near-miss around 2029. By then, experts hope to have a better handle on its status in 2036. If they do, and if the rogue asteroid then appears to be headed directly for Earth, engineers would have as much as seven years to begin building solutions.

Still, there's always uncertainty when scientists measure velocities and extrapolate times and distances in deep space, and Sandia scientists believe that the potential consequences warrant simulation now.

"Unless you have a lot of warning — decades of advanced warning — the only way you can divert an asteroid of this size is with a nuclear explosive," Boslough says. "And we don't know how much advance warning we'll have. That's why it's important to have a plan in place."

Moreover, Sandia's scientists consider Golevka a real threat.

"These things don't happen with high frequency, but they do happen," Boslough says. "We understand it's very low probability, but it has extremely high consequences."