Transcript: Columbia Accident Investigation Board Roundtable June 12, 2003 (part 1)
National Transportation Safety Board Conference Center
429 L’ Enfant Plaza, SW
Washington, D.C.
MS. LAURA BROWN: Okay. I think we still have a few stragglers outside, but I’m gonna get started here.
My apologies to the people on the phone bridge. We’re gonna try to get questions in from you before we loose the bridge at 3 o’clock, but I’m gonna get Scott started here. He is gonna give you an update on where we are with the foam testing.
MR. SCOTT HUBBARD: Okay, thank you very much.
As we’ve done a few times before, I’m happy to give you a briefing this afternoon on the latest tests against foam hitting the reinforced carbon panel six. We’re gonna go to the first set of slides, please?
Right, let’s go to the first – there we are. I think everybody knows that we conducted the tests at Southwest Research Institute in San Antonio. Several of you were there as we conducted the tests and you know that we have a large outdoor facility – it’s a nitrogen powered gas gun that fire a piece of foam through a 30 foot barrel and that that impact occurs against a mock-up of the wing leading edge and it’s built as close to flight specification as possible, including in key areas actual flight hardware.
Let’s go to the next slide. The projectile that we used is the same foam that was used for the bipod ramp, the so-called BX-250, and I’ve got the dimensions and the velocities and the masses, both the planned and the actual. We have, thanks to a lot of people and a lot of analysis, settled in on a 1.67 pound foam piece traveling at 775 feet per second as the projectile that is most representative of what happened during the accident. So, the planned mass was 1.67, the weighed actual projectile in this test, 1.68. Dimensions, as you see there, about five-and-a-half by eleven-and-a-half by twenty-one inches in this case. The actual dimension length will change slightly, that length and width are determined by the – or the width and the height are determined by the barrel and the length is adjusted in order to get the mass.
The mass is the constraint that we work to, the 1.67 pounds. We planned for velocity of 775 feet per second. We got 768, so within a percent or so, the angle of the impact relative to the wing leading edge, 20.6 degrees. And the angle of the gun barrel, that is to say whether it was vertical or slanted was straight up and down and we call that a clocking angle of zero degrees. So, that was the set-up and the projectile.
Next picture. The test occurred at Southwest Research June the 6th and we hit within a quarter inch of the planned target so that accuracy of this gun seems to be quite reproducible and quite good. The foam, as you’ll see in the high speed video in a minute and for those of you who were there, you know it broke into small and large pieces, resembling to a large extent, what we’ve seen in the actual video of the accident.
I will point out that in the video that most all of you have seen of the strike in the accident, it appears that there were more smaller pieces than in what was observed at San Antonio and the tests that were done at Glenn Research Center prior to this investigate – or prior to this experiment showed that in a vacuum, it looks like the foam breaks up into smaller pieces, so that accounts for the difference. However, when we measured the actual loads, we found that there was no difference. So, that’s why even though the appearance is slightly different, we are confident that we’re reproducing the actual event itself.
You see the post test. Many of you saw this for yourselves. The projectile had been inked so as to leave streaks and outlying the area of the impact and it started at the impact area in T-seal 6 and then went on to cross the T-seal – started in panel six, went across the T-seal and then on to the fiberglass panel seven. And you can see little dips of foam that were caught in the area between panel six and the T-seal.
Next slide. Okay, here is the first of two high speed videos that show the impact, show the piece breaking up and if you look closely, you can see the deflection of the panel. I’m going to show in a little bit, another view from the inside of the wing leading edge that shows the violence of this impact much more directly. The deflection of the panel is less in RCC than in fiberglass and that is directly attributable to the material. The material itself reinforced carbon is a much stiffer material, although it does not have the strength of fiberglass. Fiberglass is more flexible.
So, as you can see the break-up, the impact, the streaks and what we’ll do now is go to some of the analyses after the tests of what happened. Initially, we measured about a three inch crack. We’ve gone and made a much more careful measurement and the total length measured along the entire length of the crack is about five-and-a-half inches. It goes all the way through the thickness of the reinforced carbon panel. It is located on what is called the lock side. I have the model here and just to remind everyone or perhaps explain it if you haven’t seen this before that a reinforced carbon panel and the T-seal have two sides. One is called the flip side and that’s the side where it – this element, the T-seal could actually slide back and forth a little bit and take up minor differences.
And then there’s the lock side. There’s a groove – there’s a groove here where the T-seal locks any two. So, this is the slip side, this is the lock side, this T-seal has on it a corresponding groove and then a smooth edge. And so, this is the lock side here of the T-seal, this is the lift – slip side of the T-seal, so the damage that was observed was on the lock side, between panel six and panel seven.
It – the damage wrapped around the entire rib and showed up eventually as a three-quarter inch crack observable on the surface. Most of the length of the observable damage is seen from the inside. Next slide.
This is – some additional pictures showing the crack from different vantage points. I think the one thing to note here is although the shuttle program largely deals in English units, the people who have made these measurements have put up centimeter scales. So, just be careful. Having had Tom Young on the panel earlier, we know that it’s important to distinguish between English units and metric units. So, these rulers here in centimeters, so if you download these from the Web site, be sure that you recognize that the dimensions in the briefing are all in English units.
MS. LAURA BROWN: Is there a Martian in there?
MR. HUBBARD: Not that I know of.
So, that was the first damage that was visible. We reported that right after the test. A little bit later that same day, a more careful inspection, closer inspection, showed that the T-seal was cracked. This is an actual piece of T-seal here and what’s cracked is not the surface here, but rather what’s called the “web” – and that is this piece down here. It’s the chunk that goes in-between the leading edge and the next panel, it’s this unit right here, not the outer surface, but this piece, that’s where the crack is, about two-and-a-half inches long.
And an x-ray that was done at Southwest Research the next day show that this crack extends all the way into what they call the fillets (sp), that is the material at the bottom here. So, there is damage that is not apparent to the naked eye and that’s one of the issues that we’ll continue to look at with this particular panel is how much damage is there that is, in fact, hidden from simple visual inspection.
Next slide. There is an additional crack that was found on the flange of the panel six. Out here, you have a way to attach these panels to the rest of the structure. You have to have a mechanical attachment with bolts and what are called spanner beams and so forth. There is a panel – I mean a crack there right around this fitting a so-called sheer fitting. It’s part of what attaches this to the rest of the structure, about three-eights of an inch long. This is a point where stresses get transferred from this unit into the rest of the structure that holds it to the orbiter. It doesn’t appear to go all the way through, but it is evidence of more damage and damage that gets down to the structure of the orbiter.
Next chart. This is something that we noticed right on the site at the time which was about a one inch square divot, a piece of material knocked out of the carrier panel. It looks like this was a repaired area. And this is, of course, makes you think about all the repairs that have occurred to the orbiter over the years and I think this will be an element of what needs to be examined in the future is whether or not the repaired areas, in fact, are as strong as they need to be. This particular piece of damage was relatively small.
Next chart. Now, here are some displacements. When you look at what is called the step in the gap. The step refers to how smooth the panel is. When you put it all together, it’s supposed to be a very, very small difference across the surface. Perhaps, I think, the specification is 40 thousandths or less difference, and then you look at the gap. That’s how closely do these pieces fit together. Is there a significant seam here or not? That is also supposed to be in the tens of thousands per specification. Well, we found that the step and gap showed a change of about twice the flight spec post-test, compared to pre-test. This may be due to the fact that the foam got lodged in here and spread the T-seal relative to the panel. There was also a change in – on the other side of the T-seal and the slip side is down here of about 30 thousandths. So, they were significant in the sense of being a factor to more than what you would set as a flight specification, although small in the absolute number.
Next thing I want to show is a very interesting video from inside the wing leading edge. And I’m just gonna let it play first and then I’m gonna come back and describe in a little more detail what you’re seeing. You can see the violence of the shot, the shaking, and then if you’ll watch briefly at the end, you will see, well, I see it’s looped here. Okay.
Let’s go to the next slide and come back to this. These are three stills taken from this high speed video. We were lucky in having six cameras in there. One of them pointed at the exact area where the crack appeared. So you can, in fact, if you look a the video carefully on your monitor, see before the impact, you can see the crack beginning to appear right here, you can see the crack growing, and then you can see – see this difference in space in here? You can see this actually pulling apart to a maximum deformation then snapping back together again.
So, why don’t we go back and repeat this and watch, oops. Watch right up in here.
Yes, so you see some foam puffing through there at the end. You know, this looks like a long time period, but remember you’re looking at millis – you know, tenths of milliseconds here. And you can see if you look over here the deflection and the shaking of the RCC panel. I the longer version of this video, you see very briefly at the end the flash of light, which is actually daylight coming through. The force of the impact was sufficient to actually spread the panels apart briefly, let the daylight in and then it closed back up again. It says we opened it up to the atmosphere, the external part of the wing-leading edge.
So, I think the importance of this is to show that this is not a static event. This is a dynamic event and that this crack, depending on what was happening, what kind of force it was under, could be much larger and then close up again.
All right. Let’s go on then to the comparison of predictions. Overall, the impact loads were pretty close to what was predicted. We measured 2,600 pounds. It was predicted was 3,500. The stresses in the various areas, the rib, and again, the thing that was cracked was a so-called rib area. That and out here in the surface in this were about as predicted. However, the panel response that was further away from the impact was lower than anticipated and that’s probably because this didn’t deflect as much as the analysts thought it would.
Next? Now, an interesting fact that appeared when we looked at the sensor data is that this failure occurred very early – almost as soon as the foam hit the panel, and we have the scale model piece of foam here, almost as soon as it impacted, this crack occurred. And the foam was still diving into and compressing against the wing leading edge well after that happened. Now, it could be that we simply have very simplified models and we don’t understand yet all of the complex interactions, and that we’re actually looking at a system response, not just a simple piece of reinforced carbon.
The other thing that was surprising and I’ll show you some numbers in a minute, was the fact that even though we greatly exceeded the predicted design limits, the shell here itself did not break. The break again was over here at the edge. And we don’t know quite what that means yet. Partially, it’s that we have a very limited information on the dynamic properties of this material and the whole leading edge structure.
So, let me show you a little bit – next slide. This is where the various sensors were. This material is rated at around 4,500 PSI, pounds per square inch. It failed at 4,940. So, just over the load limit. That’s a little unusual because ordinarily these load limits are not a sharp cliff that you just fall off. Usually there’s some margin in there.
On the other hand, we have a gauge out here at the face sheet that recorded three times the allowable limits and it didn’t fail. Now, that may be because edges are more sensitive than the properties of the big face sheet here. Maybe there was some little spot, a little crack or something that was invisible over here that served as a propagation point. We don’t know that yet and we may never know it. The T-seal crack, though, occurred over here at – very near this impact point at a measured value at about three times the breaking point.
The next slide shows the predicted stress levels. The group at Sandia did some pretest prediction for us and where they showed the stresses building up out in here in the center of the sheet, in the center of the shell and over here at the edge are very close to what was actually measured. So, I feel like we’ve got a good handle on that part of it, even though some things occurred much earlier, like the break on the rib, much earlier than expected, and some other areas didn’t fail at all when, you know, a factor of three greater says you might expect that.
So, after the test was done, after we made these observations checked the comparisons, we went and did some non-destructive evaluation. Let me show you in this next little piece of video something called thermography. Watch right here. What you’re seeing is a flash of heat, a flash of infrared radiation, heat radiation, against this material. It heats up and then it cools off and you watch this with an infrared camera basically and then what you see in doing this kind of test is that the crack appears right here.
So, if you look at the next slide, this is a snapshot of that and it shows how the thermography can identify the crack through the panel. In fact, it goes beyond what you can see with your naked eye and this was part of the recommendation that we’ve made, which is that the whole fleet have non-destructive evaluation techniques, perhaps like this one used to detect cracks and flaws and other things that might be a problem down the road.
So, the next slide shows the conclusions thus far. I think we’ve established the failure mechanism. We’ve established that you can, in fact, crack reinforced carbon by using the foam under the conditions observed in the accident. Saw two types of damage to the panel. We saw damage to the T-seal, carrier panel and we saw a shift in the positioning of the entire structure. But, I want to caution that this is just a single data point. We don’t know yet what the potential would be for this crack growing and for hot gas penetrating into the rest of the leading edge.
This panel sticks out, it’s gonna be sent off to Kennedy Space Center for a very detailed examination, basically a CAT scan, a x-ray scan – computer-aided tomography is what that stands for – and that it’s a detailed x-ray of the entire 3D structure of the material that may tell us more.
My conclusion is it’s too early to extrapolate to the complete foam impact story for this single data point and we need to do more, which is the point of the next slide.
All right. What do we do next? Our objective all along has been to look at this place on the fault tree to see whether or not the foam can create a breach that would be sufficient to simulate the observed damage in the accident. See if the tests match the sensor evidence, the debris evidence, and so forth.
So, before we begin the tests on the reinforced carbon in the panel eight through ten area, which other evidence is saying was the most likely breach, we need to know four things, maybe more. First of all, we need to improve the models a little bit. Secondly, we need to know that by picking that particular spot right there, which is based on analysis, that we had, in fact, loaded the panel in a representative way. We need to know whether or not the angle of the gun and, therefore, whether or not the angle in which this strikes a so-called clocking angle makes a difference. And we need to also understand whether or not having this RCC panel next to an RCC T-seal, next to a piece of fiberglass makes a difference. It may be that we end up saying that we want to use multiple RCC pieces. That would be taking resources away from the inventory, but it may be required in order to establish in a much more direct way, the types of damage that can be created by the foam.
So, what we’re going to do is the following: we’re gonna go back and do some fiberglass tests on panels five through seven. We use the same mass and velocity and so forth. But, we’re going to instrument panel eight. Panel eight is a very unusual panel. It’s the one where some of the debris evidence is saying the breach occurred. Most of the rest of these panels are basically – basically rectangles. I mean if you look at the footprint of this, it is essentially rectangular.
The footprint of panel eight is more like what’s called, if you remember your geometry, a trapezoid. It is spread apart. The two sides aren’t parallel to each other. They go off in an angle. Panel eight is the largest panel. It has some of the most unusual structure in the whole leading edge. And we want to know how this impact propagates down the system and, therefore, whether or not we ought to have reinforced carbon at least at two panels before we do the next set of RCC tests.
So again, we’re gonna do a couple of things. One is, we’re going to aim about three inches below the first shot to see how we load up the panel. We’re going to rotate to the thirty degree clocking angle and see if the analysts are right. If that ends up damaging the fiberglass, it doesn’t matter because we’ll be done and we’ll be moving on to the panel eight, nine, ten area. And then, we’ll apply all of that to the tests in the eight, nine and ten region. And I think I’ve got one more slide which is the schedule for this.
We’ll resume the fiberglass test I just described on Monday, June the 16th. Second fiberglass test will be on Wednesday the 18th, and then the dates for the reinforced – first fiberglass and then reinforced carbon for the panels eight through 10 will be at the end of June.
So, I’m – that concludes my briefing. I’d be happy to take questions. I think we’ve got about a minute and a half for the phone bridge.
MS. LAURA BROWN: Let me just check and see if anybody’s left on the phone bridge first. Cindy (sp), are you on the bridge? Okay.
MS. IRENE BROWN: Hello, this is Irene Brown, if I could ask a quick question?
MS. LAURA BROWN: Okay, quick and you may get cut off and – .
MS. IRENE BROWN: Okay.
MS. LAURA BROWN: – I apologize if you do, but go ahead.
MS. IRENE BROWN: Scott, my understanding in your last briefing, these – there are not replacements for these panels and I’m just wondering what is available to be installed back on the shuttle for flight?
MR. HUBBARD: It varies. There are several RCC panel nine – panels nine. There is one spare panel eight and so forth. It varies by panel what’s available and what is and is a spare. We’re very cognizant of the fact that there’s a single spare panel eight and if we decide to make that the impact point, it will be with all due consideration. But, given the system effects I described, the fact that it may be necessary to have the RCC next to each other to really have the representative test, it may be required that we do that. It takes about eight months to build a new one. I understand that NASA Shuttle Program has already put in an order for a complete replacement set.
MS. IRENE BROWN: And it’s not your intention that these objects are to be flown on the shuttle, these test objects?
MR. HUBBARD: Oh, no, no.
MS. IRENE BROWN: Okay, thank you.
MR. HUBBARD: These are – .
MR. PHIL CHEN: This is Phil Chen.
MS. LAURA BROWN: Phil, one question.
MR. CHEN: Okay, Scott, when you are talking about a zero degree clocking angle, you’re talking about you can use it on slate on at the panel? No side angle whatsoever?
MR. HUBBARD: No, zero degree clocking angle has only to do with the – whether the gun barrel is straight up and down or turned at an angle so that the straight up and down gun barrel is zero degrees as it was for the tests. On June the 6th, we intend to rotate the gun barrel by one notch and that will cause the foam to hit with the full edge, the full leading edge, striking the RCC panel. That’s what we mean by clocking angle.
MS. LAURA BROWN: Okay. Anybody else from the phone bridge?
Okay, I think we’re gonna leave you guys, so thanks for being with us and we’ll go on to the folks here. Matt, go ahead.
MR. MATTHEW WALD: Scott, Matt Wald, New York Times.
What is the potential relevance of the damage you saw to the ability to lose a mystery Day Two object? Yesterday, O’Keefe said the damage that you inflicted couldn’t possibly be seen by an astronaut, unless you got two inches from it, implied you couldn’t see it with a national security asset – but didn’t go into the idea that with this kind of damage trigger with thermal conditioning on orbit, losing a bigger piece later?
MR. HUBBARD: That – you make a good point and I’ve thought a lot about this, along with the rest of the team. What we saw was a snapshot in time. We did a single test and then very carefully preserved the panel and the T-seals and so forth, so as to not let anything grow or change.
On orbit, we were at 81, 82 seconds, so we’re only at a fraction of the time of the – of the entire mission. I just looked up a few facts on this. We are 66,000 feet. The SRB separation is 126 seconds. An engine cutoff is 502 seconds. So, we’re only at about 18 percent of the altitude that you get at MECO.
Talking to our astronaut colleagues, these are events that you feel – the SRB separation, MECO and so forth. So there’s – even though we’re past the so-called Max Q, past the maximum pressure on the orbiter as it ascends, the dynamic with aerodynamic loads, there’s still a lot of other events that go on. In addition, of course, you go through thermal expansion and contraction as you go from seeing the sun and being several hundred degrees positive Fahrenheit to several hundred degrees negative Fahrenheit, so there’s a big temperature swing.
How this could affect the panel is unknown at this time. That’s a part of going and taking the x-rays. There are various analytical techniques you can use to test this or to simulate this. But, that’s gonna take a while. So, all I can say at the moment is that what we had there was a snapshot in time of one test. It’s conceivable those cracks could have grown and propagated and changed throughout the life of the mission.
MS. LAURA BROWN: Okay, Bill?
MR. BILL HARWOOD: Bill Harwood, CBS.
Just to look ahead at the test schedule, just the course of the geometry are great versus seven and nine, which I’ve thought about a lot. I mean, I realize there’s no spare for eight, but I guess what I’m trying to figure out is just the geometry alone is going to give you a different system response. It’s a strike on eight versus nine. I’m just wondering the relevance of these tests and maybe extrapolate to, you know, what you get if you actually hit a real eight?
MR. HUBBARD: You mean the relevance of what we just did?
MR. HARWOOD: Well, yeah. In the context, maybe the exact same hit on eight because of geometry would give you a different response or a more violent or less or whatever, how do you extrapolate from one of these flat panels to this one that has got the weird geometry to convince yourself that the test is valid and representative.
MR. HUBBARD: Right, that’s exactly that – the point I was hoping to make which is that the system response and the different shapes of these panels tends to drive you toward firing a shot at panel eight, even though it’s the only spare in the fleet. We don’t – we wanted to do that deliberately and thoughtfully though. And we’re using our best analytic tools to figure out if we can extrapolate or not.
I think that what you’re seeing here is physics in action. I mean we’re doing the type of research and development work necessary and parallel with an accident investigation to understand the impact and part of that is weighing doing on one hand a simulation versus on the other hand an experiment. And given what’s happened thus far, I’m leaning in the direction of doing an experiment, but we’re going to do some more work next week and then make the final decision.
