The foam insulation that is now a suspect in the Columbia disaster has been linked to space shuttle damage in NASA reports going back more than a decade.

Over the years, the space agency has made modifications to the insulation that sheaths the shuttle's towering external fuel tank, but despite those changes, pieces of the material have continued to break off and hit the orbiter during launch.

Even as Columbia departed last month on its fatal mission, NASA had resolved to develop yet another way to control the problem, spokesmen at the space agency said.

NASA yesterday released documents indicating that three pieces of debris from the external fuel tank -- initially assumed to be foam -- were seen hitting Columbia's underside about 82 seconds into its flight. Previously, NASA had said one piece of debris was involved.

Analysts at Boeing Co., one of NASA's prime contractors, assessed the potential damage while the shuttle was in orbit and concluded that the crew could return safely, according to the documents. In making their assessment, the analysts revisited damage Columbia sustained in a 1992 flight, when foam of a similar size from the same part of the external fuel tank struck the orbiter. NASA has not determined that the debris in Columbia's last flight was actually foam or that it caused the accident.

"You begin flirting with disaster when you have pieces [of foam] that are much larger, that can hit several tiles, and when there is the possibility that some of these tiles may not be perfectly bonded to the surface of the orbiter," said Elisabeth Pate-Cornell, chairman of the management science and engineering department at Stanford University, who led a study of the debris threat for NASA in the early 1990s.

In an interview with Washington Post reporters and editors this week, NASA administrator Sean O'Keefe said the agency had not been complacent about the potential for damage.

NASA workers apply the foam to the outside of the fuel tanks to prevent the cold liquid fuel inside from warming. Without insulation, the metal tank could become encased in ice from frozen condensation.

The shuttle's skin suffered damage from debris -- including ice, foam and unknown sources -- so routinely that NASA prepared a report after each flight assessing the extent and causes of the damage. Foam figured prominently in many of those reports, as did the particular region of the external tank where the debris that struck Columbia is suspected of having originated.

"The question is, have they taken these early warnings and used them appropriately," said Paul S. Fischbeck, an engineering professor at Carnegie Mellon who studied the debris problem with Pate-Cornell.

For its risk analysis during Columbia's last flight, NASA assumed the debris captured in launch pictures measured 20 inches by 6 inches, by either 10 or 16 inches, according to the documents released yesterday. The agency assumed it had fallen off of a "bipod ramp," near the area where a two-pronged support known as the bipod attaches to the tank. The bipod connects the tank to the underside of the orbiter near the shuttle's nose.

Within the bipod area, other problem spots have been the "bipod jack pad closeouts," which have lost foam in a number of past flights. These are squares where a structure is attached to the metal tank at the Kennedy Space Center to support the tank while it is mated with the orbiter. The supports are then removed, and workers must patch the squares with foam.

"It's always been a troubled area," a former NASA analyst said. "We've had many missions" where bipod foam came off, the analyst said.

Because it is not a smooth surface, the bipod region experiences increased aerodynamic stress, the analyst said. It also bears the strain of being attached to the orbiter, Fischbeck said.

In June 1991, Columbia hadn't even gotten off the launchpad when foam insulation began separating from a bipod jack pad. A NASA photo shows it jutting beyond the surface of the adjacent insulation. NASA replaced the protruding foam before launch, an agency report said.

A year later, the orbiter's lower surface sustained a wound measuring 9 inches by 4.5 inches by half an inch, spanning three of the fragile tiles that protect the shuttle's lower surface from the searing heat of reentry. That damage was most likely caused by the loss of foam from a bipod ramp, an August 1992 NASA report said. One swath of damage to the foam measured 26 by 10 inches, the report said.

The August 1992 report noted that something similar had happened on STS-7, the Challenger flight in 1983 that carried Sally K. Ride, the first American woman in space.

NASA responded to the 1992 debris hit by drilling additional vent holes in the foam around the bipod ramps on tanks for later flights. The holes were meant to allow pockets of gas trapped in the foam to escape without popping off pieces of the insulation.

Less than a year later, in April 1993, foam came off a Columbia bipod jack pad, exposing the tank's green primer coating. In addition, at least 21 other divots were spotted in the foam. A NASA damage assessment warned that loss of foam near the Orbiter's nose "is a potential threat to lower surface tiles."

The loss of foam was more dramatic when Columbia flew again in October 1993. A chunk of foam 28 inches long by 3 inches wide came off a section of the tank above the bipods, exposing at least 18 inches of primer, a NASA report said. "The loss of this foam may have contributed to the centerline tile damage on the Orbiter lower surface," the report said.

By November 1997, after the foam was reformulated, the popping off of small pieces had become "an expected occurrence," a November 1997 report said. NASA eventually responded by sanding the foam to eliminate a dense layer of material and reduce the amount of potential debris.

Still, foam continued to fall off.

On a flight of the shuttle Atlantis last fall, foam from a bipod ramp hit one of the solid rocket boosters but "did not cause significant damage," said Melissa Motichek, a NASA spokeswoman. That prompted a fresh review of the foam.

The study was ongoing, but when Columbia launched on Jan. 16, NASA was confident the issue was not a safety concern that should stop future missions, said NASA spokesman James Hartsfield.

Staff researcher Richard S. Drezen contributed to this report.

NASA's Sean O'Keefe speaks with Washington Post reporters during an editorial luncheon.