• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

How NASA Failed: Dissecting a Disaster

Posted on: Tuesday, 26 August 2003, 06:00 CDT

WASHINGTON -- The shuttle Columbia was doomed in part because NASA relied on flawed computer simulations and mathematical formulas that failed to accurately predict damage to the shuttle from flying pieces of foam, documents released by the space agency and the group investigating the accident show.

Nuclear and aviation industry safety experts who are familiar with the sorts of analyses NASA used say space agency officials were wrong to rely on the simulations to gauge risks.

Instead, the experts say, they should have performed tests on shuttle components to determine the damage that foam from the fuel tank might do to its wing.

These simulations and formulas -- some done before the launch, others done during the mission -- often were based on erroneous assumptions about the ability of the shuttle to withstand damage. NASA managers assumed it was safe for Columbia to return to Earth even though foam had struck the wing.

The foam hit the left wing 82 seconds after liftoff and broke through one of the carbon panels designed to protect the wings from the intense heat of re-entry. Columbia broke apart as it re-entered Earth's atmosphere Feb. 1. All seven astronauts were killed.

Today, the Columbia Accident Investigation Board will issue its findings on what caused the accident. In addition to detailing how the blow from the foam led to the breakup, the board will document how decisions by NASA managers underestimated the peril Columbia faced and how they ignored warnings, according to previous public statements from board members.

The independent board's report also is expected to find that the NASA safety office was ineffective and underfunded.

Among the areas to be discussed is NASA's reliance on computer simulations and mathematical models. Before the mission -- and again after fuzzy photos showed the foam had hit Columbia's wing -- simulations and other analyses led to poor decisions by NASA managers, documents show.

The simulations appear to have blinded them to danger signals about foam on earlier flights and led them to dramatically underestimate the threat to Columbia.

In one case, during the Columbia mission, at least 75 shuttle experts with NASA and its contractors were so concerned about the poor quality of data in an analysis they were preparing that they recommended taking photos of the shuttle in space, sources say. Those photos could have shown whether foam had damaged the left wing.

Instead, NASA managers trusted the analysis that the experts said was so flawed. The analysis wrongly estimated where the foam hit and concluded that the shuttle had not sustained significant damage. Had NASA known the extent of the damage, the agency might have launched a rescue mission to try to save Columbia's crew.

Engineers connected to the analysis say they are angry and puzzled that NASA would not seek photos from satellites or telescopes. NASA managers say they were not told of the limitations on computer simulations and analyses.

By necessity, NASA relies heavily on computer models and other analysis tools to provide clues about what happens to the shuttle from launch until re-entry. It cannot easily recreate the extreme temperatures of space in a lab. Used properly, such analyses can improve safety. Indeed, analysis techniques routinely help design safer jet wings and nuclear plants.

But safety experts in high-risk industries say they've learned the hard way mathematical formulas and computer simulations cannot fully mimic nature. Several major aviation crashes and the Three Mile Island nuclear accident were caused by poor analyses, experts say.

''You need to take into account the uncertainties,'' says Elizabeth Pate-Cornell, chairman of the management sciences and engineering department at Stanford University. ''It's the old story: garbage in, garbage out.''

NASA relied on other flawed analyses:

* After a large piece of foam peeled off the fuel tank and hit the solid rocket booster attached to the side of the tank during the launch Oct. 7 of Atlantis, NASA officials considered the incident so severe it could have grounded the shuttle. But a shuttle manager told investigators NASA contractor Boeing had run a computer model that determined there was little risk the foam could hit the shuttle. Col. James Halsell Jr., a former launch manager who heads efforts to get the shuttle fleet flying again, said the analysis gave officials confidence to launch again. However, a spokesman for Boeing said last week that it used the model to determine where the foam had broken off during the launch, not where the foam might hit.

* As Columbia orbited, engineers erroneously concluded the foam could not damage the wing's carbon panels. The analysis was based on earlier tests of small ice pellets hitting the panels. But those tests could not predict whether a large chunk of foam would cause damage.

* A computer program known as Crater, which engineers used to predict whether the foam damaged the ceramic tiles beneath Columbia's wing, had not been designed to account for impacts from large pieces of foam.

-----

On the Net:

NASA

More science, space, and technology from RedNova

Copyright © 2003 USA Today. All rights reserved.

More News in this Category