Investigators have recovered debris from the shuttle Columbia that appears to support the increasingly held belief that the doomed ship's left landing gear door remained in place as a plume of super-heated air entering through a breach near the leading edge of the left wing wreaked havoc inside the wheel well. The heat-weakened door may have deformed enough to permit jets of hot air to spew out around the corners as the wing literally burned up from the inside, its skin buckling and shedding tiles along the way.

On another front, examination of contrast-enhanced video shot during Columbia's Jan. 16 launching indicates the possibility foam debris from the ship's external tank may have struck at least two different areas on the orbiter's left side, one well in front of the left wing, according to sources close to the accident investigation.

Video of Columbia's launch could indictate more than one impact by material from the external tank. Photo: NASA video

Running a video clip on a laptop computer, one of the sources, who spoke on condition of anonymity, showed how a suitcase-size piece of debris falling away from the external tank slammed into the lower leading edge of Columbia's left wing. That much was already known. But playing the clip over and over again in a loop, he pointed out a subtle brightening as the debris fell past the left side of Columbia's fuselage near the extreme forward part of the wing known as the "chine" area.

"It tumbles and then I want to point out a subtlety to you," he said. "Watch the contrast against this black line (in the chine area) and you'll see what appears to be an apparent contact way up here. We're not sure if it actually contacts it or not. It may just be the light foam against the dark background. ... But you can see it against the dark."

One of the earliest signs of anything amiss aboard Columbia during re-entry was an unusual temperature increase in a water dump nozzle located on the left side of the fuselage just behind the crew cabin hatch. No one has yet been able to explain that telemetry, but the nozzle is located near the chine area in question.

Regardless of whether the chine area suffered an actual debris impact - the brightening may simply be the result of lighting effects as the debris tumbled toward the wing - the sources said investigators are increasingly convinced a major breach occurred at or near the site where that same piece of debris crashed into the bottom side of the leading edge of the shuttle's left wing near where it joins the fuselage. And they are increasingly convinced a plume of super-heated air entering the breach worked its way into the left main landing gear wheel well and that the landing gear door remained in place as the disaster unfolded.

Heat-damaged debris indicates the plume of rarified but extremely hot air shooting into the well heated the aluminum structure of the left landing gear door enough to cause it to deform slightly, possibly forming gaps at the corners that allowed jets of hot air to spew out.

The sources stressed that no conclusions have been drawn and that scenarios like the one outlined above may be incorrect.

"Think of what kind of disaster we'd have if we jump to a conclusion that fits all the facts and come up with the answer and it's the wrong answer," one of the investigators said. "So we have to be prepared to really look at multiple answers and not jump to a conclusion."

But so far, a leading edge scenario, or one very much like it, provides the best fit with telemetry from the shuttle and damage patterns seen in recovered debris.

"This event started to evolve almost immediately after EI (entry interface), it's fairly clear that the mechanism of failure had to be in place before the re-entry," one of the sources said. "So we're looking for mechanisms which lead to a very rapid problem occurrence after Ei.

"There was a lot of concern about the (landing gear) door because as everybody knows, that's a very weak area. And the initial photographs appeared to indicate the foam may have hit the door area or adjacent to it. Where we're at now has been really moved. In my opinion, it is low probability that we're looking at a strike on the door or a door issue or acreage issue around it. That it's much more likely that we have a problem or a breach that occurred up in the RCC area or the carrier panels adjacent to that."

He was referring to the 22 reinforced carbon carbon composite panels making up the leading edge of the left wing and narrow, removable panels equipped with protective tiles on one side that form a flush interface between the RCC panels and heat-shield tiles permanently bonded to the underside of the wing. The carrier panels are bolted in place with screws that go through the tiles on each end making such tiles inherently weaker than those without such holes. The screws are covered by a ceramic plug.

Three pieces of debris appeared to separate from Columbia's external fuel tank 81 seconds after launch. In the contrast-enhanced video released Tuesday, a chunk measuring 25 by 15 inches can be seen slamming into the left leading edge around RCC panels 6, 7 and 8. This same piece may have glanced off the chine area as well.

"It may be more complicated than we have assumed it to be because it's possible - not probable, but possible - that there may be multiple breaches," one of the sources said. "We know there were three pieces that came down, if three pieces come down you have to think in terms of there potentially having been multiple breaches. As we look at this, we may be seeing sprays coming from different directions, which will be confusing to us as we go through this."

Whether the known foam impact occurred directly on one of the RCC panels or a carrier panel or both is not yet known. But the sources interviewed for this story said relatively minor problems with either component likely would grow in the fierce heat of re-entry, quickly leading to a re-entry disaster like the one that claimed Columbia and its crew.

For example, if one of the so-called T-seals used to hold an RCC panels in place was damaged or missing, enough of a gap probably would exist to let super-heated air into the void behind the leading edge. Likewise, a missing carrier panel tile would leave an exposed surface susceptible to burn through. While the gas in this case is not ionized enough to meet the textbook definition of a plasma, it is more than hot enough to melt the wing's aluminum sub-structure.

"You've got a gas there that is at 10,000 degrees," one of the investigators said. "Because it's so rarified, the transfer of the heat from that creates a temperature in (the wing) of about 2,500, 2,800 degrees. The transfer mechanism in here is what they're struggling with. It's not strictly convection, it's not strictly radiation. You're dealing with a hundred thousand times lower pressure than you have on the surface."

The resulting plume may well have moved about as Columbia rolled through normal entry maneuvers and as the breach itself worsened, subjecting different areas to extreme heat.

"If you think of a fire hose that you're not holding and it's moving all around, we may find that's more what we're dealing with," one of the investigators said. "As it hits a hole that has different surfaces, the surfaces are going to be directing it in different directions. I think when (Columbia is) in the right bank, it may very well be forcing some of this air ... laterally down the spar and then when you go into the left bank, it actually moves up. It goes from one side to the other and you might find this thing changing over periods of time depending on the hole size as it eats away, what's in its way and those kind of things. So it may be much more complicated than just trying to point a finger and say it came in this hole and it went right through the wing here."

Despite those extreme temperatures and the mercurial nature of such a plume, investigators say it would be very difficult to trigger a disaster without having a fairly sizeable area of damage. Whatever happened to Columbia, it may have started out small and rapidly worsened.

"One RCC doesn't do it and a couple of tiles won't do it," one of the sources said. "So we have to look for things that may increase in size over periods of time. One of the things we're looking at is if one panel comes off, it's likely others will come off because you've got hot air ... melting pieces and you would have a cascading kind of event. People have talked about unzippering of tiles. We're talking about unzippering of RCCs."

Alternatively, a small but growing breach in the leading edge area could cause surrounding RCC panels to partially melt and collapse inward. So far, recovered RCC panel fragments and support hardware don't provide definitive clues as to which scenario may be more accurate.

But it now seems likely a plume of hot gas entered the wing just forward of the landing gear wheel well, burning through the forward skin of the wing, spraying aluminum slag back on the inner surfaces of nearby RCC panels and ultimately working its way into the main landing gear wheel well.

Damage found in recovered debris increasingly indicates the left landing gear door remained in place until the wing itself began breaking up late in Columbia's re-entry and that a plume of super-heated air entering the wing from the breach near the leading edge worked its way into the wheel well and then out through seals around the landing gear door.

Burn damage on recovered debris just inboard of the door indicates a plume of hot gas spewed out of the well's forward inboard corner in the direction of the shuttle's centerline. Recently recovered debris indicates the possibility of a similar jet from the outboard forward corner of the door."

"There is a similar one, but much smaller in nature, on the other side," one source said. "We have a piece on the outboard side that has a very small indication of the same effect. You can't have that effect if the door is gone. You have to have something that's holding the pressure in and allowing the vent hole to occur. Again, we're coming back to the theory that if the door is lost, the door is lost very, very late in the event.

"We also have another debris item," he added, "a door hook roller uplock. It's not the hook itself, it's the pin, uplock pin, the door latches to. And it appears to be eaten away on one corner, which may be a key to us for the direction of flow.

"We believe you're getting heat coming into the well, heat and air coming into the well, building up temperature on the inside and pressure on the inside. What's happening is then that temperature and pressure is allowing the aluminum corners to get a little bit soft and the pressure pushes up the corners a little bit and that's where the events are coming from. That also indicates, again, that the door is still on."

Adding credence to that picture, search crews in Texas have recovered numerous tiles that came off the door. Had a major breach occurred in the door area, investigators believe, fewer tiles would have been recovered. And in the initial video showing the debris impact, there are no obvious signs of tile damage. Admittedly, the resolution is low, but one would expect to see a lighter color on the belly of the orbiter if the black surface layer of many tiles was removed in an impact event. No obvious brightening, however, can be detected.

"No item would you rest a case on individually," the source said. "But when you put all these together, it appears to give you a story that we're really dealing with something which is farther up from the wheel well and all these things about the wheel well and tile fractures is probably not the story."

In the next few weeks, investigators at the Southwest Research Institute plan to begin a series of tests that could shed light on just what happened when the foam impact occurred. Pieces of foam insulation roughly the same size as the chunks seen falling away from Columbia's external tank will be fired into various wing components with a powerful "chicken gun" normally used to shoot debris into jet engines.

One RCC panel and support hardware taken from the shuttle Discovery will serve as a target, as will two RCC panels removed from Enterprise, a non-space worthy shuttle prototype. In addition, foam will be fired at a landing gear door removed from Enterprise and "acreage" tile to determine the effects of impacts with broad areas of tile on the underside of the wing.

Unlike earlier tests carried out at Southwest Research Institute in 1999, the upcoming runs will use large pieces of foam and include impacts on tile edges and other areas. The tests will not be exhaustive, but "they will, I think, provide a piece of information that's going to be critical to determining what kind of damage you can do, what kind of damage you can't do."

"This isn't going to just be a piece of carbon carbon on plywood," he said. "They're actually taking the structure and instrumenting it with accelerometers and strain gauges to figure out how much of a lateral push it gets, how much compression it gets and then do the before and after."

Said the other source: "I don't see where foam is going to hit an RCC and break the (RCC). I can see more likely where it might break some of the support structure underneath it, a T-seal potentially, or shear a bolt or something. But inherently, this is a very weak area."