As circumstantial evidence builds suggesting a loose piece of insulating foam brought down the space shuttle Columbia, the board investigating the disaster is starting to focus on how a shuttle program devoted to safety could have missed the danger.

While its members are rigorously avoiding jumping to a final conclusion, new data examined by the Columbia Accident Investigation Board (CAIB) point to a breach in the leading edge of the left wing near the spot hit by a slab of foam during liftoff as the accident's cause.

Early playback from a data tape recorder that survived the fall to Earth shows temperature sensors in the left wing overheating much sooner after Columbia reentered the atmosphere than was detected in telemetry, in a pattern consistent with superhot gases eating into the wing from the vicinity of the foam impact. Meanwhile, examination of foam on a new tank at the Michoud Assembly Facility in New Orleans found more voids where water could collect, boosting chances the foam would pop off again during launch as it also had before Columbia was launched.

Ultimately, the CAIB hopes enough of the physical data will line up sufficiently straight to indicate the direct cause of the accident, or at least allow it to be inferred. But Adm. (ret.) Harold W. Gehman, Jr., the CAIB chairman, said the panel's report will also try to present "contributing and root causes" that NASA and Congress can use to shape future human spaceflight doctrine and U.S. space policy.

"This accident, in our view, is not necessarily a random data point on a continuum graph," he said. "It probably fits into some kind of an overall context. That context could be budget patterns, changing priorities, perhaps the psychology of continued success. It could be the context of workforce patterns, the way the workforce has changed or the way they look at the shuttle program for being operational or research and development."

John Logsdon, director of the Space Policy Institute at George Washington University, heads a CAIB working group that's starting to tackle the "context" questions in some detail. Gehman said Logsdon's "Group No. 4" will take a "top-down" look at shuttle history, examining budget trends in detail and reviewing past studies of shuttle operations to pinpoint factors and trends underlying the accident.

At the same time the panel's other three working groups will feed the results of their interviews, some of which are conducted on a not-for-attribution basis to enhance candor, to the Logsdon panel in an effort to underpin its theoretical work.

"Either their work will suggest to Group No. 4 an area that they ought to look at, like the normalization of deviation, or as Group 4 reports to the board on what they've found, we will find specific evidence that confirms or denies their theories," Gehman said.

"Normalization of deviance" is an organizational theory that was applied to NASA by Boston College sociologist Diane Vaughan in her comprehensive 1996 study "The Challenger Launch Decision." Vaughan, who is scheduled to testify at a CAIB hearing Apr. 23, described it as a situation in which organizations "proceeded as if nothing was wrong in the face of evidence that something was wrong."

When Columbia broke up on reentry, NASA had already started working on the problem of insulating foam dropping away from the shuttle's big external tank during launch, based on its last launch and the launch of STS-112 in October 2002 (AW&ST Feb. 10, p. 41). But at the Flight Readiness Review (FRR) that cleared Columbia for launch, the issue of falling foam was not raised because managers considered it had been laid to rest in the FRR for STS-113, which did not shed foam during its Nov. 23, 2002, liftoff.

Air Force Maj. Gen. John L. Barry, a member of the CAIB, said board members were still investigating cryopumping-the flash evaporation of water in voids in the foam-as the cause of foam separation. Barry said technicians working for the CAIB have found 14 voids in the right side and at least 18 on the left side in practice cuts on External Tank 120 as they prepare to dissect the bipod area of External Tank 94, a twin of the tank that flew with Columbia.

Despite progress identifying debris that drifted away from Columbia on flight day two, the CAIB was still not certain it had identified the source of the object. Experts at Wright-Patterson AFB, Ohio, who compared some 3, 100 recorded radar returns from the object with the radar reflectivity of 29 different leading-edge samples measured on the ground, concluded that the object was most likely one of the heat-resistant carrier panels that joins reinforced carbon-carbon (RCC) wing leading-edge sections to the thermal tiles aft of the leading edge.

But the same searchers who found the recorder from the Orbiter Experiment Support System (OEX) in Texas have found the carrier panel believed to be closest to the foam impact site, meaning it was still attached to Columbia when it broke apart over Dallas. That carrier panel connected RCC Panel 6, where enhanced photo analysis places the main foam impact, to the wing. Although carrier Panel 5 is still missing, a CAIB official said visual examination of carrier Panel 6 found no damage that might be associated with an adjacent missing piece. That could be significant if it holds up to closer examination, because carrier Panel 5 was located forward of Panel 6. Temperature data from the OEX recorder suggest superheated gases were blasting down the cavity behind the U-shaped RCC panels from a breach forward of RCC Panel 9, where the first of the OEX sensors was located.

As described by CAIB member Roger E. Tetrault, the sensors near Panel 9 all registered dramatic changes well before flight controllers in Houston first noticed anomalies in the telemetry. The first was a strain gauge right behind the aluminum wing spar behind RCC Panel 9 that started to increase and went off nominal 270 sec. after entry interface, when Columbia started back into the atmosphere.

That was 206 sec. before an unexplained yaw movement of the shuttle that was the first anomaly recorded at Mission Control Center in Houston. Next came a temperature sensor positioned in front of the spar behind RCC Panel 9. It started to rise 290 sec. after entry interface, and failed 2 sec. later with a reading of only about 50F. Tetrault attributed the low temperature reading to insulation packed around the sensor, and suggested it failed because its wire was cut.

A temperature sensor on the leading side of the left Orbital Maneuvering System pod aft of the wing followed, registering unusually low temperatures beginning 344 sec. after entry interface and continuing low for 196 sec. before swinging up as high as 1,200F-about twice the normal level.

"There's something else that happens that brings the temperature rapidly up, which may be burning aluminum or a number of other effects that we have to go look at," Tetrault said.

Finally, a temperature sensor behind the wing spar at RCC Panel 9 started to go off nominal at 425 sec. after entry interface, rising to 450F before failing 520 sec. after entry interface and 63 sec. before the first anomaly was picked up on the ground.

The CAIB will keep sifting through what Gehman called the "treasure trove" on the OEX recorder from about 670 functioning sensors across Columbia that continued to produce data for 15 sec. after all telemetry was lost. But Tetrault said the evidence to date-from OEX, telemetry, debris, imagery and other sources-has "narrowed this down pretty well to the leading edge of the wing. It remains to be seen exactly what failed," he said, listing RCC panels, the "T-seals" between RCC panels, the stainless steel structure that supports the RCC panels, the carrier panels and the bolts that hold it all together.

"We have to sort our way through all of those and make a determination of which one is the real failure mode," Tetrault said. "If we fail to do that properly and we get the wrong one, then we could have a future accident."

Gehman said the OEX data would help engineers at Southwest Research Institute refine upcoming ballistic tests that will essentially shoot pieces of foam at RCC panels and other leading-edge materials to see if the impact could have broken something or jarred it loose. Those tests have been delayed slightly to accommodate the new data and refined calculations of the impact site. Barry's working group also is examining the effects of aging on leading-edge components, including the tiny holes etched into the RCC material by paint primer from the shuttle launch pads, and the action of salt air on critical components over time.

Making the final call on the cause of the accident will be just a little bit harder for the CAIB and NASA because the agency did not request access to classified military satellites and ground telescopes to examine Columbia while it was still in orbit. Ron Dittemore, NASA's shuttle program manager, said right after the accident that he and his top staff decided not to request use of the "national assets" because past experience had shown them to be of little value and because an analysis of the foam impacts had concluded they were not a safety-of-flight issue.

Since the accident, the National Imagery and Mapping Agency (NIMA) has offered, and NASA has accepted, use of national assets to examine future shuttle missions "on a routine basis, without specific tasking from NASA," according to a Mar. 25 letter from Administrator Sean O'Keefe to Air Force Lt. Gen. (ret.) James R. Clapper, the NIMA director. The letter was copied to Gehman and Director of Central Intelligence George Tenet.

"This will be very helpful as we continually assess the condition of the shuttle during on-orbit operations," wrote O'Keefe. "Significantly, your willingness to employ NIMA assets during targets of opportunity without specific tasking will be another useful source of information to help us assess the potential of on-orbit anomalies."

O'Keefe said the agreement would eliminate the "administrivia" that might have made managers reluctant to request classified information that might not have been needed during Columbia's last mission. Details of the data transfer remained to be worked out, O'Keefe said, including how NASA would communicate specific engineering needs to NIMA during future missions.

Special Report: Columbia Disintegrates