|Columbia Accident Investigation Board Public Hearing
Wednesday, May 6, 2003
9:00 a.m. - 12:00 noon
Hilton Houston Clear Lake
3000 NASA Road One
BOARD MEMBERS PRESENT:
Admiral Hal Gehman
Brigadier Gen. Duane Deal
Major General John Barry
Major General Ken Hess
Dr. Sheila Widnall
Mr. Roger Tetrault
Mr. G. Scott Hubbard
Mr. Steven Wallace
Dr. Gregory Byrne
Mr. Doug White
Mr. Steven L. Rickman
Dr. Brian M. Kent
Dr. Dave Whittle
Mr. Paul S. Hill
ADM. GEHMAN: Good morning, everybody. This public hearing
of the Columbia Accident Investigation Board is in session.
We have three panels of two people each to hear this morning.
The purpose of today's hearing is to put into the record and
let the board hear an update of the very latest data that
we have on data from the orbiter, information from the debris,
and information concerning the testing of the Flight Day 2
object which was observed orbiting with the shuttle. This
will bring the board completely up to date with the latest
information we have from all of the analysis that's been going
The first of our panels today, we're delighted to have two
people who have been working on this project since Day 1 and
are very knowledgeable in exactly what went on onboard the
We are grateful, gentlemen.
Doug White is the director for operations requirements in
the orbiter element of USA; and Dr. Gregory Byrne is the assistant
manager, Human Exploration Science, at JSC.
What I would like to do, first of all, gentlemen, is read
you a statement that you will attest that you are telling
us the truth. Then I would ask you to introduce yourselves,
say a few words about you, and then if you have an opening
presentation, we will let you have the floor and we'll listen
to your presentation.
So before we begin, let me ask you both that you affirm that
the information you're going to provide the board today is
accurate and complete, to the best of your current knowledge
MR. WHITE: I do.
DR. BYRNE: Yes.
ADM. GEHMAN: All right. If you would introduce yourselves,
please, and then we will start the presentation.
GREG BYRNE and DOUG WHITE testified as follows:
MR. WHITE: I'm Doug White. I'm director of operations
requirements for United Space Alliance. My responsibilities
include turn-around requirements, problem-solving for during
the turn-around, and in-flight; and I'll be presenting a summary
of the MADS data today.
DR. BYRNE: I'm Greg Byrne. My normal job at JSC is
manager of the Earth Science and Image Analysis Laboratory.
For the 107 investigation, I'm the lead of a much larger image
analysis team which includes imagery experts from across the
country; and I'll be presenting today some ascent video and
ADM. GEHMAN: Thank you very much. You can proceed.
MR. WHITE: Greg, why don't you go first.
DR. BYRNE: Okay. I understand, Doug, that you have
a long briefing. So I'm going to be short and just answer
questions as they come.
Can I have the first slide, please.
First of all, by way of introduction to the team, the image
analysis team consists of both NASA organizations and non-NASA.
As I mentioned, imagery experts from around the country. The
NASA organizations include Johnson Space Center, Kennedy,
Marshall, and Langley; and then outside of NASA we have independent
assessments from folks at National Imagery and Mapping Agency,
NIMA, and Lockheed Martin at three locations across the country.
So let me start with an overview of the imagery we have to
work with. You've seen these views already. They have been
released to the public. We have two primary cameras that we're
able to work with to analyze the debris event on ascent, the
debris that struck the wing. Two cameras. E212 and ET208.
I do have some short movie clips of these.
By way of introduction and background for these two views,
E212, the imagery that we had to work with was original. We
took the original negatives from the camera and had it digitally
scanned at the highest resolution. So we had the best-quality
digital imagery to work with from that camera, and that camera
gave us the best view of the bipod ramp area which was the
source of the debris. It also gave us the best view of the
debris itself for size measurement. The drawback to that view
was that we had literally no view of the impact area from
that particular view.
The other camera view is a video camera. It's called ET208.
We also had it digitally scanned from the original tape. The
advantage of that particular tape is that we do see the impact
area directly; but it being video, it's inherently less resolution
than the film. But it does give us a full view of the debris
all the way to the impact area.
Next slide, please. Also, by way of background, here's a layout
of the KSC area. It shows the relationship to the launch pad,
which is that circle right there, with the two cameras which
are south of the launch pad. Then that red line, that is the
orbiter trajectory going uphill. Now, the event happened at
about 81 seconds. It would put it right around there by that
Bubble 5. So these are the lines of sight to those respective
E212 was the closer one. It was about 7 miles away. ET208,
further south, was about 26 miles away. So the cameras were
distant from the orbiter but are essentially telescopes with
cameras mounted to them and they track automatically and so
we get a good view.
Next slide. Let's go ahead and go to the movie. Eric, if you
would key up that movie for me, please.
What we're going to show here is that ET212 view. It has both
the visible frames and what we call a difference mode of frames.
We'll show those side by side in movie format and then track
the debris on down. So on the right is the normal view, and
on the right is a difference view.
Just looking at the normal view first, the debris exits from
the bipod area and strikes the under side again. Again, we
don't see the actual strike, but we do see the debris cloud.
A close strike. It passes entirely underneath the wing. We
don't see any evidence of debris or a debris cloud coming
over the top of the wing. So that's an indication to us that
the strike was entirely on the under side of the wing, below
what we call the stagnation point on the leading edge.
The difference view highlights changes from one frame to the
next; and so it's useful for highlighting the debris because,
of course, the debris wasn't in the frames previous to the
event itself. So it does highlight the debris, and again you
can see it tracking on down. Unfortunately, what it does is
also exaggerate the size of the debris. So you can't use it
for size measurements, but it does give you a better view
of the debris itself and then the post-impact cloud coming
The cloud appears to be pulverized foam or perhaps tile. We
can't tell if it's tile or not, but upon closer inspection
-- and I'll talk about this later if I have time -- we do
see actual chunks of debris. You can see them as they pass
through this region here, the SRB. There are actual chunks
of debris in that view, as well.
Next slide, please.
ADM. GEHMAN: Greg, let me interrupt here with a question.
I think this is a good point. Are there launch commit criteria
for the number of cameras that should be working? Are cameras
a launch commit criteria?
DR. BYRNE: I don't believe they are, but I'm not the
person to ask.
MR. WHITE: No, they're not.
ADM. GEHMAN: So whether you've got one working, two
working, or four working just depends on whether you're having
a good day or not a good day.
DR. BYRNE: Okay. This next view is another movie view.
It shows the actual trajectory. We map the trajectory to try
to understand the character of the debris as it comes on down;
and what we'll see in this movie is that it appears that the
major piece of debris acts as a parent, so to speak, that
it spawns smaller pieces along the trajectory. So it's possibly
shedding smaller pieces and we can see them pass under and
then the major parent piece is the one that strikes the wing.
So let's go to that movie, please.
Another conclusion was that we saw no evidence of more than
one strike other than the major parent piece. Okay. Here again,
we'll see the event begin around the bipod ramp area; and
maybe we can go slowly frame by frame, if that's possible.
Yellow is the major parent piece. It originates here. Frame
by frame. The piece is spawning off. Little pieces in blue
and then other smaller pieces in red keep on coming down.
You see the other red and the blue pieces pass underneath
and then the parent piece striking and then here are individual
post-strike debris chunks that we're able to track and measure
sizes. We're still working on that.
Okay. Let's go to the next slide, please. The other camera
view, the ET208 video, again, as I mentioned, we see it all
the way from the bipod ramp to the impact area right there
on the leading edge. Again frame by frame, we can map it on
down; and let's play this movie very quickly.
I was asked to bring the best quality copies of these, and
that's not possible on a setup like this to view it in best
quality. For that we would need our laboratory facility or
something similar to it. We might not have any luck with this
one. It worked back at the facility. Okay. Why don't we go
on? I apologize for that.
Back to the E212 view. Once again, we can map frame-by-frame
the trajectory of the debris coming on down, just as we can
map frame-by-frame in the other view, and we can take those
two camera views together. Go to the next slide, please.
With those two camera views, we can define line-of-sight vectors
for every point along the trajectory or every place where
we see the debris in those frames and we can then use a two-camera
solution to derive a three-dimensional trajectory of that
debris as from source to impact. That's very important for
us to be able to determine the point of impact and three-dimensional
Next slide, please. Concerning the debris source, we have
a couple of lines of evidence that tell us that, yes, indeed,
it was the bipod ramp or the immediate area next to the bipod
ramp that was the source of the debris. I mentioned the three-dimensional
trajectory mapping that we do.
Here this red line is one of those trajectories that we've
mapped onto the CAD model of the external tank. So we take
the imagery and then we employ CAD models and overlay the
imagery on the CAD model and that gives us a graphical representation
of orbiter that we can overlay the trajectory onto for visualization
and, as you can see, there's the bipod ramp on the left side
of the tank. This trajectory maps it to right adjacent to
and on top of. That's an indicator that, yes, it was the bipod
In the next view, take the imagery itself. Next slide, please.
And we do some enhancement. As I mentioned, the E212 view
gives us a view of the bipod ramp but not a very good one;
but if we do a technique of frame averaging in which you overlay
multiple frames and do some enhancements and bring out detail,
you can see in this before-and-after view -- before being
on the left where we've averaged 22 frames immediately before
the shedding event and then some 21 frames immediately after
the shedding event -- if you look at the differences before
and after, and there's the bipod ramp. It's a slightly different
shade of color, slightly lighter color than the tank so you
can see it. And after, it's very subtle but there is a definite
change to that area. It's whiter, as if to expose the white
Next slide, please. We have measured the debris size, again
from the E212. We took a frame-by-frame measurement of the
debris. Here's one frame on the left and another on the right,
just to give you an example of how the apparent size of the
debris changes frame-by-frame. Obviously it's tumbling. What
it is, it's tumbling and so it is changing its orientation
relative to the camera line of sight. So in every frame it
has a different appearance but if you take this frame-by-frame
measurement and lay them all out, you can deduce from the
multiple frames an estimate of the size and our estimate is
given there, 24 by 15, in the length and the width. Now, we
weren't able to determine that third dimension, which was
depth; but we were able to determine that that depth is a
much smaller dimension than the other two. It's plate-like,
a length and a width and a much smaller third dimension, plate-like,
and that we could not determine from the imagery alone.
Next slide, please. 3-D trajectory analysis. As I mentioned,
we're able to map to the wing to determine impact locations;
and we had several analyses. Again, my team consists of many
different organizations, in many cases working independently
and so getting different results; but when you take them all
collectively, we are able to determine that the impact location
was in the range of Panels 6 through 8. Now, when I say impact
location, we have to keep in mind this is a big piece of debris
and that it's likely to strike multiple panels; but the center
line of the trajectory, at least in this model -- and this
is just one example of the several that were generated. Here's
the center line of the trajectory, and the center line intersects
the wing at that location right there. So in this model, X
would mark the spot of the center of the impact; but, of course,
it's a big piece of debris and then there's uncertainty in
that trajectory on top of that. So that would then spread
out our area of impact location across these three panels
and then the other trajectories are also showing some dispersion,
as well. So we can't exclude the possibility that Panels 5
and 9 were at least partially impacted. So that's our range,
6 through 8, plus or minus one, and more likely outboard than
Next slide, please. We did measure the velocity, but we weren't
able to pinpoint it. The total velocity, we got actually three
components of velocity; and when you add them all up, the
total velocity was in this range measured from the imagery,
610 to 840. Now, that's a wide range and I'm disappointed
our team was not able to pinpoint it any better than that,
but we're fundamentally limited by simply a few data points
to work with. When you're working with so few data points,
especially in four dimensions, X, Y, Z, and time, you can
get a wide range of answers; and that's why we have this wide
range. But I am confident that the total velocity, the true
velocity is within that range. But it takes more than just
imagery alone to nail down the impact velocity and so we've
needed to apply some physics to the problem. So we're turning
our results, our trajectory data over to the folks who are
working the fluid dynamics and applying some air-flow dynamics
to the problem to get a better estimate of the velocity.
Of course, all of this is going to feed into the impact testing;
and everything we've been doing up to this point has been
driven by the need to feed the impact testing. So our schedule
has been pushed to meet that schedule.
Next slide, please. In regards to what can we see on the bottom
side of the wing, ET208 gives us a direct view of the under
side of the wing and, again, these frame averages before and
after. On the left is before the event, before the strike
to the left side of the wing or rather the left wing. Then
on the right is the "after" view. Same averages. In the "after"
view, when you do the differencing, we simply don't see any
difference before and after. So that's an indication that
tells us that we simply can't see any damage. Of course, the
orbiter perspective is not the best in this view and our resolution
is not very good and we estimate the resolution would be about
2 square feet. What that means is in order for us to see damage,
we would need at least a 2-square-foot area of difference
to see it.
ADM. GEHMAN: Which is on the order of three or four
tiles square, I guess.
DR. BYRNE: Something like that.
ADM. GEHMAN: 2 tiles by 2 tiles something.
DR. BYRNE: Of course, that's presuming that the damage
would be in the form of tile removal to have a high contrast
between the dark normal tile on the top versus the white substrate
underneath. So that would assume a high contrast in the damage.
MR. WALLACE: What might you expect to be able to see
as far as damage to the lower surface of the RCC and the T-seals?
DR. BYRNE: We wouldn't expect to see any damage to
the leading edge. Again, I mentioned --
MR. WALLACE: I mean, is there a degree of damage that
you're confident you could have seen?
DR. BYRNE: Yes. About a 2-square-foot.
MR. WALLACE: Even in the RCC? Or talking about just
DR. BYRNE: Just in the acreage. I wouldn't expect to
see any damage in the leading edge because contrast is all-important
and a hole in the leading edge would be presumably a dark
hole against a dark background. In a view like this with the
resolution that we have, we simply wouldn't see it even if
it were a gaping hole, I think.
ADM. GEHMAN: I don't have any argument with that conclusion;
but what about the sharp edge, leading edge of the RCC there?
I'm thinking about a notch or something missing, even though
I agree when you've got the dark RCC against a dark hole against
a dark background, you can't see anything. But what about
the leading edge there? Is that enough definition there to
indicate some -- I mean, you've got that nice leading edge
against that nice white background.
DR. BYRNE: If there were a large enough gap, I think
we might be able to see it. If there were an entire panel
missing or two panels adjacent to each other missing, it's
possible that we could see it because it would show up against
the white background of the fuselage. So that's conceivable;
but, of course, we didn't see anything like that.
Next slide, please. The last slide, I mentioned the debris
post impact. The wing is up in here, and the debris after
the impact is sweeping on by. This is an area of work that
we're still pursuing to characterize better the size of these
chunks post impact and primarily to see, well, two things.
Is there any hardware in there? Can we say it's tile or can
we say it's a T-seal or something of that nature? That's a
very difficult task, of course, but also to characterize it
to compare it with what we see in the impact testing. My team
is also involved with the impact testing, doing the photogrammetry
in those tests to see does it make sense.
That's all I have.
MR. HUBBARD: Thanks, Greg, for that description. I've
got maybe four or five questions here, a number of which are
intended to just illuminate things that have been in the realm
of rumor and give you a chance to talk about this and perhaps
put it to bed if it's not factual. The first one has to do
with a statement that I have heard several people make that
there was another camera, a third camera. Some people have
called it Camera 204. So can you talk a little bit about that?
DR. BYRNE: I can, yes. There was another camera that
saw the debris. If we can pull up that map. The second slide,
I think. Camera 204 was well south of the other cameras. I
don't have a mileage exactly, but well south.
MR. HUBBARD: So much further down.
DR. BYRNE: Much further south. It did see the left
side of the orbiter with basically the same perspective as
208, but much further away. So a worse view in that regard,
Now, early on in the analysis, of course, our analysis team,
even during the mission, screening all of the imagery from
all the cameras, we saw that debris in 204; but early on in
the analysis, it was discarded as unuseful for analysis simply
because it was so much poorer in resolution. The debris looked
like a fuzzy blob. At that time, as I have mentioned, it was
disregarded. Since then, especially in regards to the velocity
calculation where we were strapped with having so few data
points to work with and in that sense any data point is a
good data point perhaps, one of the team members -- it was
the folks from Marshall -- went back to the imagery to try
to get more data points and they did access that 204 camera
and determined that possibly two frames, two data points from
E204 were useful for their trajectory analysis and subsequent
velocity calculations. So they did fold that into their calculation,
and we discussed that with them last week. Their result is
brand-new as of last week.
The bottom line is we don't know if it adds value or not.
Marshall did their analysis with 204 and then redid it without
204 and got the same result. So although the error associated
was much larger and they did determine that the error was
much larger, it didn't seem to hurt the analysis but didn't
seem to help it either. So that's the story on 204.
MR. HUBBARD: Okay. Very good. Thank you. So what you
presented today, Camera 212 and Camera 208, represents still
the best available evidence for all the calculations you've
DR. BYRNE: Correct.
MR. HUBBARD: The second thing has to do with the number
of objects. A lot of speculation about the spawning, how many
pieces came off and so forth. Can you just expand a little
bit on how many objects you have clear evidence that exist
and resolve that dispute a little bit?
DR. BYRNE: Right. Early on, that was a big question,
how many particles are we talking about, how many impacts
were there. To this day, I don't think we've had total team
consensus on that, simply because at the top of the trajectory
-- first of all, on 208 we only see one piece of debris throughout,
in that video view from far away. It's in 212 where you can
see more than one piece, but how many there are is still indeterminate.
There's almost a shell-game juggling act going on at the top,
and trying to pick out which piece is which and when is very
difficult to do; but we had determined early on that we think
we saw three pieces, three distinct pieces.
Now, whether they originated as three pieces from the bipod
-- in other words, came off in three pieces originally --
or whether they were spawned, that we have never been able
to determine because literally now you see them, now you don't.
It's that sort of game going on at the top. Even frame by
frame, when you see a piece of debris, the next frame it's
gone. So either it's a very thin piece that when it turns
edge on, you simply don't have the resolution to see it, or
whether it goes behind another piece, we don't know. So it's
very difficult to determine, but at one point we thought we
saw at least three distinct piece.
MR. HUBBARD: Okay. And the best evidence that's available
shows only a single strike.
DR. BYRNE: Only a single strike and that being of the
major piece and all these others.
MR. HUBBARD: Now, you did mention tumbling; but you
didn't talk about the rate. I've seen numbers and viewed these
videos, of course, several times. The sense from one group
was it was tumbling at about an 18-hertz rate, 18 cycles per
second. Is that still the case?
DR. BYRNE: Well, that was the measurement that was
done. Our partners at NIMA did a very innovative calculation
to try to discern the tumbling rate. What they did was look
at the different color channels in the film -- the red, green,
blue, RGB -- and the foam, being a shade of orange, would
stand out better in the red-green channel. So they looked
at the different channels and plotted frame-by-frame the intensity
of those three color channels and looked at the variation
in the intensity; and just in that rough calculation, that
variation in intensity came out to be 18 hertz.
Now, we all recognize -- and NIMA did, too -- that that's
very crude because we have so few data points to work with
that to try to do a frequency determination from so few would
give you an enormous error bar. But that was the only handle
that we had, the only analytical handle that we had at all
to try to determine rotation rate of that piece of debris.
I do not have confidence that the rotation rate was 18 hertz,
but that's all we have.
MR. HUBBARD: So the conclusion there is -- would you
say it is clearly tumbling but the rate is -- we've only got
one data point?
DR. BYRNE: It is clearly tumbling and in our analyses
we worked with the still frames to get the exact measurements,
but you have to work with the motion as well to get a big-picture
view of what's going on. In that motion, when you put the
debris in motion, you can clearly seen with your mind's eye
-- your mind's eye can integrate between frames and you can
determine at that time it is tumbling; but to take it the
next step and say what the tumble rate is, in an analytical
process, that's difficult.
MR. HUBBARD: The before-and-after picture you showed
of the bipod ramp area where it's dark, light, dark, light
-- and I think if you were able to flicker those, it might
be even more obvious.
DR. BYRNE: Yes. In fact, I should have brought the
movie form of that where they're overlaid before and after
and you can form that and it shows it clearly.
MR. HUBBARD: Do you have an estimate for how large
that bipod ramp area is?
DR. BYRNE: That's something we've been working on.
That also is very difficult because when you apply a software
routine to do the differencing, the software is detecting
the change in the image before and after. Well, when there's
so much noise in the imagery, which there is here at that
scale, then literally the entire image after looks different
because of the noise. So what we've done to date is do a manual
estimate of that area of change, and our area was consistent
with the size of debris. I believe we were getting somewhere
in the order of 30 inches by 15 or 16 inches of the size of
change. Again, consistent with the ramp itself, consistent
with what we measured.
ADM. GEHMAN: Scott, how you doing down there?
MR. HUBBARD: Ready to yield the floor, sir. I'm probably
dangerous because I have a little knowledge about this area.
ADM. GEHMAN: I'm watching the clock.
MR. TETRAULT: Greg, last week I think we were using
a velocity of approximately 640 feet per second; and I noticed
today that 640 is in the lower element of the range that you
threw out there. Would you describe what's been going on that
appears to have revised your calculations a little bit?
DR. BYRNE: Yes. As I mentioned, that was one of our
disappointments, that we weren't able to nail it down better.
The first four or five analyses that were done by the various
team members came up with a range of total velocities between
610 and 700, and the average of all of those were 640. So
that's what we put forward originally. Last week our friends
at Marshall came in with a new, different analysis. They used
a fundamentally different technique than some of the others.
They came up with a much higher velocity that was in that
higher number, 840.
Well, we had a peer review, so to speak, of that and with
all team members last week -- and this is brand-new, last
week -- and the Marshall analysis passed the peer review,
so to speak. We couldn't say, "You're wrong." In fact, I can't
point to any one analysis and say it's the best. I can't point
to any one analysis and say it's wrong -- because, again,
so few data points that we're working with in four dimensions.
You can fit almost any curve to those data points and get
a reasonable answer.
MR. TETRAULT: Does a higher velocity suggest a smaller
DR. BYRNE: Now, that's straying a little bit away from
our area of imagery alone; but in the transport analysis,
the next step that we're feeding our trajectory data over
to, in order to meet the transport analysis model, that is
true. The smaller mass would require a higher velocity.
ADM. GEHMAN: Okay. General Hess, you have a question?
GEN. HESS: I just have a couple here. Real quick. In
your earlier comments, you kind of qualified the bipod ramp
as being the source, by saying we have a couple of lines of
evidence that indicate. Do you have any lines that indicate
that it's not the bipod ramp?
DR. BYRNE: No.
GEN. HESS: Looking at the video, I know that most of
your effort almost entirely was focused at the debris, the
debris strike. Have we analyzed the video beyond 81 seconds
to see if the debris is --
DR. BYRNE: Oh, yes. What I've shown here is a tiny
fraction of the whole analyses that we've been doing; and,
yes, we have looked thoroughly at from pre-launch all the
way through SRB sep and beyond. We have looked for any and
all indications of events before and after, debris coming
off after the 81-second event and so forth. The answer is,
no, we don't see any debris other than some normal stuff that
we see all the time, SRB slag near the sep.
GEN. HESS: Has your work with all this post-video analysis
given you any ideas about what the current state of the art
in terms of what the cameras are and what they should be that
would have helped you do this better?
DR. BYRNE: The return-to-flight effort is a big one
and a lot of that is focused on enhancements, upgrades of
the imaging capability of the orbiter. That's one area that's
being closely looked at, what can we do in terms of launch
cameras to better our capability to analyze. That's still
in work. High-definition TV might be one way that we need
to go. The film cameras are good. You really can't do better
than film, but we're strapped fundamentally with the problem
that here we are on the coast and the orbiter is moving away
from the coast very quickly. So we're going up and away from
our camera assets and so just losing sight of it very quickly.
ADM. GEHMAN: I'm going to have to interject myself
here so we can get on. We'll reserve the opportunity to ask
more questions later, but let me ask two quick ones. This
level of photo analysis takes a considerable amount of time.
It's taken a couple of months now. Would I be incorrect in
saying that this level of photo analysis, for example, these
20- and 30-time enhancements and things like that, would not
be available during the 14 or 16 days of the mission?
DR. BYRNE: No. They were, actually. That before-and-after
view of the under side of the wing, for example, was something
that we had done during the mission and, again, to see if
there were any damage. It's interesting that much of what
I am presenting here, we have concluded after three months
and thousands of manhours across the country, much of what
I'm presenting is similar, if not exact, to what we had reported
a week after launch, during the mission.
ADM. GEHMAN: That's important. Thank you. And the last
one is you did not discuss what you can determine about the
angle of impact with respect, for example, to the plane of
the wing or however else you want to measure it. Very briefly,
can you say something about the angle?
DR. BYRNE: Yes. The three-dimensional trajectories
that we measured were three-dimensional, X, Y, and Z. So from
those trajectory analyses we were able to measure a range
of impact angles. Almost all of it was in the X. However,
we did measure a slight Z component, upward and into the wing,
of approximately zero to 3 degrees; and in the Y component
there was a small outboard Y. The range was about 2 to 10
ADM. GEHMAN: All right. Good. Thank you very much.
MR. WHITE: If you could pull up the presentation. I'm
going to talk about the MADS data. That's the Modular Auxiliary
Data System. This is a separate data system from the operational
instrumentation system that we were able to see realtime.
This data is only recorded on board, and we were very lucky
to find the recorder intact and the tape in very good shape
and able to pull that data off.
Go ahead to the second slide.
ADM. GEHMAN: Doug, I think it's useful for the people
who have been following this that this is the recorder that
the board has been referring to as the OEX recorder.
MR. WHITE: That's correct.
ADM. GEHMAN: We're going to properly name it here.
MR. WHITE: Well, the MADS system is the name of the
entire system, which is the avionics, the electronics to condition
and report the signals and the sensors and the wires connected
to them. The recorder itself was an early model of the recorder,
which was called the OEX recorder, the Orbiter Experiments
Recorder. In the subsequent vehicle, we just called it the
MADS recorder; but the version that was on 102 was called
the OEX recorder.
On 102, it had the most sensors of any of the vehicles for
the MADS system because it was the first vehicle built. Through
the years, some of those sensors have broken and fallen off
line and during the recent major modification a lot of the
sensors were removed or the wires were cut and just left in
place, but there were 622 measurements on board, located throughout
the vehicle. Most of those are pressure, temperature, and
strain measurements; and I've broken down into three large
categories there. You can see the left wing, about 259 --
we had more of our measurements there than anywhere else --
right wing, about 220; and then other places altogether, 143.
The avionics to condition all of these signals, all of these
wires run to the mid-body, about Bay 8 of the mid-body, and
then they're recorded actually on the OEX recorder, which
is in the crew module. As I said before, none of this data
is available to us realtime during the flight.
Next slide, please. First thing I'm going to talk about here
is failures of this data. What we see mostly in this data
is all of these sensors beginning to fail and going off line,
with a wildly variable signature where they oscillate between
off-scale high and off-scale low. To us that indicates that
the wire bundles that contained these measurements in the
left wing were being burnt through and being destroyed. Most
of that happens between about 480 seconds to 600 seconds from
entry interface; and for those of you working in GMT, that
would be 13:52:09 to 13:54:09 in GMT time.
ADM. GEHMAN: Entry interface being?
MR. WHITE: Entry interface is when you first start
to encounter a little bit of the atmosphere. That would be
13:44:09. So I broke that down between temperature, pressure,
and strain gauges in the left wing, the right wing, and then
other measurements we were interested in. You can see the
What this chart tells us is that we saw, surprisingly, some
failure signatures over in the right wing. There were a number
of right wing pressure sensors that went off line, about 30
of them, and that is because they have commonality with left
wing measurements, they share a common piece of avionics in
the avionics boxes that condition the signals, and as things
were being shorted or destroyed in the left wing, that affected
measurements in the right wing. So we've been able to tie
those events together.
The other thing you notice from this chart is that there were
two measurements only that did not eventually fail in the
left wing, and those hung in all the way through the loss
of vehicle. Those two measurements are strain gauges which
are on the wing surface or on the spar actually that runs
in front of the wheel well. That's the 1040 spar. If you look
at the wire routing for those particular measurements, those
two measurements peel off from the main bundle in front of
the wheel well and stay there as opposed to running farther
back into the wing. That tells us that the damage that was
going on was farther back in the wing and that the wire bundles
were being burned farther back in the wing rather than up
near the front of the wheel well, because those two measurements
did hang in there.
There were 241 measurements that are what we call snapshot
measurements. By design, they only take data for a few seconds
at a time and then they go off line and the recorder goes
and looks at something else. So you only see these little
snapshots, bits of data, and it's very hard to determine whether
those are failing or not. We suspect that they failed the
same way that the other measurements in the left wing did,
but we just don't have the data that will show us that.
MR. WALLACE: Can you discuss the time sequence -- maybe
you'll get to this later -- with respect to the first off-nominal
indications in the telemetered data?
MR. WHITE: Yes. I'm going to talk about that and, depending
on how much time we have, I have another version of this which,
last time I was here, I talked about the operational instrumentation
data in sort of a graphical sequence, marching through the
time line. I have one of those available if we have time to
get that done today, but I thought I'd start off with showing
you the data and showing you where it looked off nominal and
we'll talk about the sequencing, too.
Next chart, please. Just real quickly all I wanted to talk
about in this chart here was we said we saw these measurements
oscillate wildly between off-scale high and off-scale low
and can we explain that from an instrumentation system point
of view that these were, indeed, failure signatures of these
measurements and not really data that it was trying to tell
us. We have done that. We've had our instrumentation system
experts go and look at how the system could fail and if you
shorted this wire to that wire, could you get the signature
that you observed in the data. The answer is, yes, you can
pick from what we saw in the data just any combination of
shorting or variable resistance between wires to get the observed
The other thing we see is that sometimes after this oscillation,
off-scale high, off-scale low, that it looks like a measurement
returns to a normal state or something that reads real data.
This has to do with bias, the way the measurement was set
up and its residual voltage in the system; and it should not
be interpreted as real data. So after you see the data do
one of these wild swings, you shouldn't believe anything that
you see afterwards.
Next chart, please. Let's go one more. We'll concentrate on
the leading edge of the left wing which is, as Greg told you,
where we narrowed down the strike to the Panel 5 through 9
region. We did have some measurements in the left wing, near
Panel 9 and 10. We had two temperature measurements, one in
the clevis area where the RCC attaches between Panel 9 and
10. That's on the outside of the spar but inside of the RCC.
We had another temperature measurement on the back side of
the spar, so inside the wing. There's a third temperature
measurement in that area, which is on the skin just behind
Panel 10; and there is also a strain gauge measurement in
that area which tells us the relative strain in that spar.
Those are all the ones that you can see highlighted right
in this area here.
I've also highlighted the wire run that feeds measurements
along the wing leading edge. There's a group here and a group
out there and some here and some back in here. Each of those
measurement numbers and each of those times is the time when
those went off line. So you can see the ones in the leading
edge went off line almost all together. The only one that
stayed around for a while was this one temperature measurement
here on the back side of the spar. That hung around for 522
seconds after entry interface, but the rest of them failed
early and we'll talk about those sensors right there at Panel
9 and what they showed us. Again, that tells us that something
was coming through the left wing and destroying that set of
leading edge bundles first before it got to some of the other
sensors in the wing.
Next chart. This is just a wiring diagram of the back of the
wing. If you start over here -- these are from photos from
the last major mod of Columbia. This is looking on the side
of the wheel well. Here are some major bundles here that run
down the side of the wheel well, but the bundles for the leading
edge of the wing go off this way and you can see there's several
different bundles here run across the wing. This is the back
side of Panel 9 and 10 region, which is down here; and I've
got some more pictures of this here later, showing some of
the measurements. This particular one is a pressure measurement
and a temperature measurement. They go through the wing here,
and then they run on down the back side of the wing.
Next chart, please. This is just a close-up of the bundles
along the side of the wheel well inside the left wing, and
we've just numbered them arbitrarily. We started at the front
side, but they change their routing and switch over each other.
So the order that you see here happens to be 1, 4, 3, and
then this is the wing spar and you can see the wires going
down the leading edge of the wing there.
Next chart, please. This particular chart is in the Panel
8-9 region, and I highlighted the split there. This is the
back side of the wing, looking forward. These are wire bundles
running down the wing spar. We, again, arbitrarily labeled
these A, B, C, D, E, and you can see measurements there and
which bundle they were in, Bundle A, C, or D, and when they
failed. Just lining these up in time order, it appears to
us that the damage was maybe higher or at least the wing spar
began to fail higher up before it worked its way through.
There's one measurement here at the bottom, the one that lasted
the longest. We're not quite sure because it's very difficult
to tell from the photos whether it's routed in Bundle D or
Bundle E. That's this temperature measurement here which is
under this red piece of tape. This is the temperature measurement
I mentioned that's on the back side of the spar.
Next chart, please. This is just a graphical way to look at
all of those wire bundles failing. We pulled out the ones
from the leading edge which we showed in purple; and you can
see how quickly those failed, starting here about 480 seconds
after entry interface. You can see how quickly those failed
relative to the other bundles that I showed you, the larger
bundles that ran down the side of the wheel well, Bundles
1, 4, and 3. Also you notice that Bundle 3 had the two measurements
that never did fail, had 117 measurements in that and only
115 failed. That's because two of those peeled out of that
bundle very early in front of the wheel well.
I also tried to indicate, just for timing, some of the other
major events in the time line that we're familiar with that
we were able to get from the realtime flight data. So you
can compare when these events were happening relative to those
other events. For example, the first orbiter debris event
is way down here.
Next chart, please. We'll talk about some ascent data that
we got from those Panel 9 temperatures. This again is just
a graphic to show you where things are located. This is a
skin temperature measurement which is on the skin behind Panel
10. We had two temperature measurements, one in front of the
wing and one behind the wing, and then we had one strain gauge
measurement right here. Then in a side view you can see the
one that's in the clevis there of the RCC and then the one
that's on the panel behind.
Next chart. Again, just to get you oriented physically, looking
at the back side of the wing, this is the strain gauge here
about the center of Panel 9. There's the temperature gauge
on the spar. This is the feed-through for the temperature
gauge that goes inside the RCC but outside of the spar, and
then there's that lower skin temperature measurement that
I was talking to that passes through the skin right there.
Next chart, please. So this data compares the temperature
rise for the Measurement 9895 -- that's the one on the back
side of the spar -- to data from other flights. The RCC cavity
is vented. So as you go uphill, the air comes out of the cavity.
So you normally see a cooling kind of a trend, which is why
all these measurements drop down a couple of bits. Then as
you go through ascent, you get ascent heating and the measurement
tends to warm up a little bit.
What we see here on STS 107, which is the black line, is it
drops down a few more bits than the other ones do and it rises
back up a few more bits than the other ones seemed to do.
Now, this in itself is not conclusive that we actually had
a hole in the wing at this point and that we did have abnormal
heating on this spar, but it's just something a little bit
different than what we have seen. We've looked at some more
data than what I presented on this chart. We have found some
flights where we were able to see the dip maybe as big as
this one was, but we still haven't found any that rose back
up quite as much as what we saw here.
GEN. BARRY: Can you argue that this is definitive evidence
that there is a breach?
MR. WHITE: No, I cannot argue that it's definitive
evidence; but if I were to put this in a big scenario that
says there was a breach at this time, then this certainly
would be supporting evidence for that. But I would not hang
my hat on this evidence alone. This is not strong enough to
say that there definitely had to have been a breach, but it's
not inconsistent with the fact that there might have been
a breach at this time.
Next chart, please. This is just comparing in numbers what
I just said, the other flights, how many bits down it went
and how many bits back up. For 107 here, we did indicate that
it's a little bit different than other flights.
Next chart, please. Let's go talk about the entry data. Again,
we'll talk about the leading edge area here on Panel 9. This
is an under side view. There's also pressure measurements
MR. WALLACE: Doug, can you sort of equate bits to degrees?
MR. WHITE: I believe, on that measurement, one bit
is about 5 degrees, I believe. On the order of 5 or 6 degrees.
So there were some pressure measurements we'll look at back
here and other measurements along the side wall and the lower
skin, as well. Again, that's the inside of the RCC, showing
the two temperature measurements we had there.
Next chart. This is that lower skin measurement that's just
behind Panel 10, and we compared it to other measurements
on this flight. You can see that one gets a little hotter
and then the next chart will show you that this area right
in here is anomalous heating. This is a little hotter than
that measurement ever got on other flights during the entry,
and this little bump right in this area here also appears
to be a little outside of our experience base.
Next chart, please. Here's that same measurement in the black,
plotted against that same measurement for other flights. You
can see this area here that I talked about is a deviation
from the heating we've had before. This measurement normally
comes up and flattens off. So we saw a little bit higher.
Then all of this stuff here you see, that's the failure signature.
That's where the measurement goes unreliable, where we believe
the measurement itself or the wires to the measurement were
being burned through; and then any of the data out here you
can't believe, even this little bit out here at the very end.
You also see this little bump here which is a little bit different
than we've seen before.
Next chart, please. This is just some graphics showing you
some of the temperature measurements along the side wall.
Next chart, please. Some more toward the aft.
Next chart. We'll talk about this data. Here's some of that
data, plotted for side wall temperatures; and you see some
off-nominal heating in these two particular measurements.
These are on the side wall fuselage. You can see this measurement
rising here, and this one rising here is off-nominal heating.
This is not something that you would have seen from other
Next chart, please. Again, these are measurements on the OMS
pod. We saw a curious effect on the OMS pod. We saw lower
heating for a portion of the flight and then we saw higher
heating. So that tells us the vortex that comes along and
normally would heat the OMS pod was moving around. It was
off of the OMS pod early, when it normally would have been
there, and then it was more intense on the OMS pod later.
So this black line here, these measurements are actually below
where they would have been for this period of time in other
flights; and then where all these arrows are about here, all
of these measurements start going high again and getting higher
heating than they would have been in other flights.
Next chart, please. Getting back to the wing leading edge
at Panel 9, the approximate area where we believe the impact
Next chart. Again, just the back side view to help you remember.
This is the strain gauge, temperature gauge inside, temperature
gauge outside, and then the lower skin temperature.
Next chart. So I put all of those on the same graph, and this
is the graph that says the first events we saw happening were
in this area. These are earlier than the wheel well measurements
that I talked about last time. The first thing we see is this
strain gauge measurement go up and off, and this is the off-scale
failure again. But about 290 seconds is when we see the start
of the off-nominal rise.
Here you see the two temperature measurements in the blue
and the purple. They began rising earlier than we've ever
seen before; and again, they all failed about the same time
right here in this region. This one other strain gauge measurement
that I showed you was one of the snapshot measurements. So
you only have a little bit of data in here and here. You can
argue that this might have been off nominal, but we just don't
really have enough data to say. Definitely this part here
and then down before it failed was off nominal, and this is
an indication that because of temperature and heating in this
area that the strain and the load was shifting and that there
was something happening to the leading edge of the wing in
this region, the Panel 9 region. Again, as I said, this is
the earliest indication, about 290 seconds after entry interface
-- this is the first indication of something going wrong that
we saw in the vehicle data. This measurement, again, I already
showed you a couple of times. This is the skin temperature
measurement, again showing deviation. There's this little
hump here and then higher heating before it goes off scale,
ADM. GEHMAN: In front of me, I have the advantage of
having the Rev 15 of the time line; and what you classify
as start of peak heating occurs at Time 50:53, is what arbitrarily
is called start of peak heating, which works out to entry
interface plus 400 seconds. So you are seeing temperature
rises and some strain prior to peak heating?
MR. WHITE: That's correct.
ADM. GEHMAN: So what's happening is that as the vehicle
heats up, so are these leading edge.
MR. WHITE: Right, these leading edge. Inside the RCC,
where we wouldn't be expect it to be heating up, before peak
heating -- I mean, peak heating, like you said, is kind of
ADM. GEHMAN: It's still hot.
MR. WHITE: It's still hot. We have heating all the
way from the beginning of entry interface. So what we're seeing
is that heating manifesting itself inside the RCC cavity where
we would not expect it to manifest itself. So again, this
is a good indication that at this point we did have some sort
of breach in the RCC.
Any more questions here? We'll move on and talk about the
pressure data a little bit. Next chart.
I'm not going to go through each one of these sensors, but
you can see they're all arrayed in more or less the same Y
location away from the fuselage. This is the lower surface.
We also have a lot of pressure measurements on the upper surface
that I won't talk about. This band right here, the forward
8, we see some interesting measurements here; and I'll go
Next chart. These are on ascent. So we're back to ascent now
and looking at the pressure on ascent to see if we can determine
anything going on on ascent from these pressure measurements.
What we see is all the measurements decaying, as you would
expect. As you go uphill, the pressure gets less and less;
but there's one measurement here which is behind the Panel
9-10 region. We see this bump at about 84 seconds or so, then
coming back down, and then another spike farther out. Now,
to us that's an indication -- we don't worry so much about
the particular value that it went up to but the fact that
it took two jumps is an indication to us that something hit
that sensor, either clogged the port or moved it or did something
to the sensor to cause it to have those two spikes.
Also there's another sensor. There are two types of pressure
sensors. One's called a statham sensor, which is mounted on
the surface of the skin and has essentially a very short tube
that goes through the tile to sense the pressure. Excuse me.
I said those backwards. That's the Kulite. Then the statham
sensor is mounted inside the vehicle, away from the point
where the tube goes through, and has a rather long tube running
inside the vehicle and then poking through the skin. So the
statham sensor, which happens to be right next to this, we
don't see this kind of a spike on, because the actual sensor
and wiring and everything was inside and protected; but if
you had something hit in the tile where this Kulite sensor
was mounted right on the skin, you could have done damage
to it. So this data tells us that we did have some kind of
a hit in this region, but it doesn't tell us anything more
exact than that.
GEN. BARRY: Two quick questions. We know the impact
occurred at 81. So this is about 85, .
MR. WHITE: Right. So this number is a little bit downstream
from the leading edge of the wing. So there could have been
something tumbling or coming back a few seconds later that
affected this sensor.
GEN. BARRY: When you say tumbling back, you mean like
something could have gotten loose and then just rolled back?
MR. WHITE: Right. It could have been debris. It could
have been that the tile where the sensor is was damaged and
then suffered some further damage, some bits of it came off
or part of the sensor became de-bonded somehow or was affected.
So there could have been a delayed reaction from the hit.
GEN. BARRY: We know that sensor's not 100 percent reliable.
Have we got any indications of any previous flights where
we have these kinds --
MR. WHITE: No, we have never seen these kind of spikes
before on pressure sensors.
MR. HUBBARD: Just to be clear, again, you're not measuring
here -- what you're saying is not a pressure change. You're
saying it is something, it's an electrical signal as a result
MR. WHITE: Well, it's possible that that was -- especially
the first one. The second one is a lot harder to explain as
a real pressure change. It's possible there was some sort
of real pressure change in this region here. Again, that would
be a result of the instrument being affected and maybe the
flow around that instrument being changed. So there was temporarily
a higher local pressure around that measurement; but it also
could be just an effect of the instrument being damaged, as
ADM. GEHMAN: And you're confident that the time line
differences between the camera time hacks and the MADS data
recorder time, that you don't have a second and a half of
MR. WHITE: No, these are pretty good times. So whatever
it was here was a little bit delayed from the impact that
Greg told you about.
Next chart, please. This is another measurement which was
again in this same region farther back from the leading edge
where we believe the strike happened and you can see the pressure
here -- this is compared to other flights of Columbia. You
can see the pressure there just kind of decayed off a little
bit faster. Again, that could have been from debris plugging
the tube or something like that to cause it to have apparently
lower pressure earlier than the rest of the flights, the earlier
flights would have shown.
Next chart. Finally, there are three measurements, again in
this same band, that show a very odd behavior around 102 seconds
here. Two of them go down, come back up; and one of them makes
a jump up. This one we haven't been able to explain yet as
any kind of hit or anything, thus appears to be some sort
of glitch in the instrumentation system. Again, it's something
we've never seen before and it's odd that all three measurements,
which are not -- two of them are located together. This one
and this one are close together. This other one's a little
farther up. It's odd that they would all have the same behavior
at the same time and then return to what appeared to be sort
of a normal reading. Just kind of connect the line here. It
looks like it came back to where it would have been. So we're
not sure what to make of this yet. It's something else we're
still looking at. Again, this is ascent data; and the scale
along the bottom is seconds from liftoff.
That's all I had, as far as showing you pictures of the data.
If you wanted to go in and look at how these things relate
in time, we can go into the time line charts.
ADM. GEHMAN: Let's see if there are any questions before
MR. TETRAULT: Is it possible to go back to your Viewgraph
MR. WHITE: Sure.
MR. TETRAULT: I have two questions. On the upper right
and the lower right, there are two pressure sensors, if we
get back there.
MR. WHITE: Okay.
MR. TETRAULT: See the pressure sensors in the upper
right and the lower right? Those have wires which run back
into the bundles, but those are also cut at Times 495 and
497, which to me would suggest that the breach had to be close
enough to --
MR. WHITE: Talking about it might have been over here
MR. TETRAULT: Right. You had mentioned that you thought
the breach was in No. 9.
MR. WHITE: Well, from Greg's data, it's anywhere from
5 through 9. To get a little off of this, our forensic evidence
says that it was more likely in this region of Panel 8. So
it's very possible that it was over here and got these wires.
MR. TETRAULT: That's what I'm trying to get at is to
catch that wire right here and this wire right down here,
you would probably have to have some breach that would be
in this area or further over to the right.
Now, the other question that I have is this one here, this
Temperature Sensor 9895. You indicated that there's a certain
degree of ambiguity as to whether it comes down and goes out
this run or goes back up.
MR. WHITE: Right. It's hard to tell whether -- I don't
know if you can see this or not. The wire runs down here.
It's hard to tell whether it doubles back in this bundle here
and it runs up this way or whether it just stays in this bundle
and goes that way.
MR. TETRAULT: It is, however, I've been told, that
you have a specification requirement that does not allow you
to make a pigtail like that on a wire run, so that it would
be more likely that, in fact, this wire run goes down this
MR. WHITE: That's correct. Yes, sir.
MR. TETRAULT: I see that as important because this
wire run comes back up and joins these wire runs at Panel
No. 7; and because of the lateness of this sensor going off,
it would tend to preclude the breach from being over here
in 7 since it joins the other wire bundles.
MR. WHITE: That's correct.
MR. TETRAULT: Would that be a good assumption?
MR. WHITE: That's a good assumption, yes, sir.
MR. TETRAULT: Okay. Thank you.
MR. WHITE: Did you want to get into the time line?
ADM. GEHMAN: Yes. Please. I'm thinking we have about
20 more minutes.
The two leading edge temperature sensors in the vicinity of
RCC Panel No. 9, which are labeled 9910 and 9895, I think.
I was looking through. You did not actually plot that temperature
MR. WHITE: Yeah, let's see. If we go back -- I'm sorry,
go back to Chart 26. Sorry to back up. Let's see. Can you
get Chart 26 of the previous presentation back?
Those are plotted here. It's just difficult to see because
of all this noise from the strain gauge. They're the two.
The purple and the blue. Sensor 9910 is the blue, and 9895
is the purple. So you see the one from the blue begin to rise
here. That's the one outside the spar, in the RCC cavity,
and then followed behind by a rise maybe somewhere in here
for the one inside the cavity, and then both of them get very
hot very quickly and then begin to go off scale. As I said,
in this particular graph, because I plotted everything together,
it's masked in here by the failures of the strain gauge. Here's
the first temperature rise and then the one outside the spar;
and then here's the temperature rise, maybe somewhere in this
range, of the one inside the spar.
ADM. GEHMAN: I want to make sure I'm reading this right.
In the case of the blue one, which is 9910, which is outside
the leading spar, both the temperature rise and also the time
scale is significant in that this almost certainly could not
be a cut wire or burning insulation or a slow ground or --
MR. WHITE: No, sir, we believe the data is real data
up until right here, somewhere in this area here; and then
it becomes very difficult to tell when it starts to go vertical.
ADM. GEHMAN: Now, in the other one, 9895, which is
the lower one, that argument's a little bit harder to make
because both the temperature rise is --
MR. WHITE: It's more subtle.
ADM. GEHMAN: It's more subtle and it varied over a
small period of time, but your conclusion is that that also
is a legitimate temperature rise.
MR. WHITE: Yes. Both of these we believe are real,
to somewhere in this point here. We believe those are real
indications that we had heat inside the wing at that point.
Now, whether or not the breach was farther down and we just
had convective heating coming down to that part or whether
the breach was nearby -- and you heard some of the other arguments
why it should be farther upstream, maybe in the Panel 8 region
-- but we do believe that was real evidence of real heat inside
ADM. GEHMAN: Now, for the temperature sensor outside
the spar, the area between the spar and the cavity in there
between the spar and the RCC, it's hot in there.
MR. WHITE: Yes.
ADM. GEHMAN: Because the RCC is not really an insulator.
MR. WHITE: Right. The RCC, it re-radiates. We have
a lot of insulation inside the RCC, in the front of the spar,
to protect the spar and protect it from the re-radiation of
the RCC; and that temperature sensor is buried down underneath
ADM. GEHMAN: My next point. 9910 is actually buried
inside the insulation.
MR. WHITE: Yes, sir. It's down in the clevis where
the panel would attach, and then there's lots of insulation
over top of that.
ADM. GEHMAN: Right. Okay. Thank you very much. Go ahead
with your time line.
MR. WHITE: Let's see if we can get the other presentation
up. All right. This is similar to the time line I showed you
the last time I was here for the operational instrumentation
data and we've mixed in some of those time line points here.
There's an awful lot of ones here. I'll maybe skip some, and
there's some that I just left out of here even putting this
together, just to try to make it more brief. This is not every
single event we have on the time line and I'm not going to
walk you through every single failure of every sensor here,
but I'll try to look at this in a big picture.
Next chart. Now, these are some of the sensors that I decided
to plot. I did not plot all 622 of the MADS measurements,
just some of the ones that are more interesting. We also plotted
some of the OI measurements that you're familiar with here
in the wheel well and some of the ones in the wing. Again,
these are the sensors that we were just talking about here,
and you'll see this area start to have things happen first.
We also tried to keep a color-coding, trying to show what
was on what bundles. The blue ones here on this blue bundle
which is No. 3 which runs down the side of the wheel well
and also splits off and runs along the front of the wing.
Bundle No. 4 is this pinkish one. Bundle No. 1 is the yellow
one, and you can match those up with the pictures I showed
As we walk through this, I'm going to keep score over here
on how many sensors in a bundle had failed, but you won't
necessarily see a dot for each one. So sometimes you'll see
these numbers jump a lot and you won't necessarily see that
many dots change color.
Next chart. So this is now our new first event that we have
at 13:48:39 or 270 seconds -- I believe I said 290 in the
other one. Because the rise is so small, you can put a tolerance
around the front of that. But that's the strain gauge measurement
on the front spar there near the Panel 9-10 interface and
we see that begin to rise off nominal. That's real data we
believe that says something is happening to the strain in
the wing leading edge spar at this time.
Next chart, please. Again, we see that first rise we just
talked about, 9910. That's the clevis. It begins its very
Next chart, please.
ADM. GEHMAN: And that's only 20 seconds now.
MR. WHITE: Right. We've only gone now to 13:48:59.
So not very far in the time. As we get closer in, you'll see
lots of events starting happening within seconds of each other.
The next thing we notice again from the MADS data which we
did not have before is now we have an OMS pod temperature
sensor which is now showing cooler. As I talked about when
I showed you the data, some of those temperatures went down.
That says the vortex has now been disturbed and is not hitting
the OMS pod the way it normally does. So this temperature
here showed a little blue, to indicate it's cooler than it
normally would have been.
ADM. GEHMAN: Even though you're not going to show every
sensor of all 600 and whatever, you have more than one sensor
that does that.
MR. WHITE: Yes. We have several in the OMS pod, and
I think I have some of them highlighted in here.
ADM. GEHMAN: So it can be corroborated.
MR. WHITE: Yes. It's not just one lone sensor doing
this. We see cooling trends on a number of OMS pod sensors,
we see them on the side wall temperature measurements here,
and then we see off-nominal heating trends as well in this
Let's see. Go on to the next one. All right. This is a comm
dropout. We're still way out off the coast of California.
Next chart. Another comm dropout.
Next chart. This is another corroborating measurement. This
is payload bay surface temperature again going cooler than
it normally would have been at this point in the flight. Shows
a little blue dot there.
Next chart. Another comm dropout.
Next chart. All right. Now we see the lower surface temperature.
This is the one behind Panel 10 on the surface, and it's starting
to rise. It says we've got some kind of heating that's now
getting to the surface from probably through conduction through
the skin of the vehicle. It's starting to heat that up right
there. Again, all of these events are now earlier than anything
we had seen in the operational instrumentation data before.
Next chart. Comm dropout.
Next chart. Another comm dropout.
All right. Now, we're back to the spar temperature itself.
This is the one in the inside. Now it's beginning its rise;
and we're at 425 seconds past entry interface, or 13:51:14.
ADM. GEHMAN: Once again, peak heating is arbitrarily
defined as some number 40 seconds ago, if it turns out that
400 or 404 or something like that.
MR. WHITE: Yes, sir.
ADM. GEHMAN: So we are now at peak heating.
MR. WHITE: Yes, sir, we are now at peak heating.
All right. Now we see OMS pod temperatures where we're seeing
cooler measurements here and here. We're seeing hotter measurements
than we would expect, a little further back on the OMS pod.
So right about here.
All right. Next chart. Somewhere in between maybe a slide
or so ago that I showed you and maybe a slide or so from now,
we believe that the wing leading edge spar got breached. It's
hard to tell from the data exactly where that might have been.
In a few seconds, I'm going to start showing you a lot of
sensors dropping off line. So we know that it had to have
breached before the sensors drop off line. It's difficult
to tell exactly when that wing leading edge spar was breached,
though. This is at 52:05; and this is now where we're starting
to notice something different in the aero. This is data that
we had seen before, and it could correlate with a time that
we started to make the hole bigger or had burned through the
wing leading edge.
Next chart. Another comm dropout.
Next chart. Now, this is something different; and we can't
really explain this yet. We've tried to get our thermal folks
to explain it; they can't. We've tried to get our instrumentation
folks to explain this instrumentation failure, and they can't.
We did not see this data until we got the MADS data, but there
is a temperature measurement up where the chin panel and the
nose cap attach and one of those measurements began an off-nominal
rise. If you look at the plot of the data, you'll see it going
on a normal kind of slope and then it takes a jump, a higher
heating rate, and then for some reason it cools back down
and joins where it would have been at that time if it had
just kept going and continues on its way.
So we don't know what to make of that either physically --
it's hard to explain something heating up and then cooling
down and getting back to exactly where it would have been
if it had kept on its same rise rate -- but instrumentation-wise
it's also difficult to explain it. It's different than the
vent nozzle temperatures that we talked before from the OI
data. There when you see a higher heating rate and they cool
back down again, they're offset from their slope where they
would have been. So that extra heat stayed there and they're
a higher temperature but at the same rate. Here it actually
comes back to the same temperature it would have been and
then resumes. So it's kind of odd, and we don't know how to
Next chart. All right. These are the first measurements that
we start to see go off line. So at this point here, 5216,
we know the wing spar has been breached and that we are burning
wire bundles. So there's one back in the back of the wing
here. This is a left wing upper surface pressure that goes
off and a corresponding right wing upper-surface pressure
that shares a common power supply in the MADS system. Both
of those were affected.
ADM. GEHMAN: Doug, can I ask you to go back one or
two. I want to go back to the first aero event, I think, which
is 5205, I think. First clear indication of off nominal. I
happen to have your detailed line here. The QBAR and the pressures
here are still extremely low.
MR. WHITE: Extremely low. Yes, very low.
ADM. GEHMAN: We're talking, according to this, 22 pounds
per square foot or something like one tenth of a pound per
MR. WHITE: Yes, sir.
ADM. GEHMAN: So even though we've got some aero events,
the aero pressure --
MR. WHITE: It's less than 1 percent of atmospheric
ADM. GEHMAN: It's practically nothing.
MR. WHITE: Yes. That's correct. But yet we can see
an effect in the way the vehicle's flying.
ADM. GEHMAN: Also, in about another 11 seconds, we're
going to project that the heat penetrated the spar. So even
though we've got extraordinarily low pressures here -- in
other words, we don't have anything like a jet, a high-velocity
MR. WHITE: But the amount of air that's there is very,
very hot. There is a lot of heat there.
ADM. GEHMAN: A lot of heat.
MR. WHITE: And the wing spar actually may have been
penetrated at this point. In another few seconds, as you said,
we'll start seeing sensors drop off line. So we know that
the wing spar was breached somewhere before that. The timing
of how soon it was breached versus how soon wires start to
drop off line, we haven't nailed down yet. So it could have
been breached right here at this time.
ADM. GEHMAN: But this is almost exclusively a thermal
event at this point.
MR. WHITE: Yes, sir.
ADM. GEHMAN: I mean, it becomes an aero event later.
MR. WHITE: Yes.
MR. TETRAULT: You have done some testing, heat-testing
of Kapton wiring and how long it takes.
MR. WHITE: Yes, we have.
MR. TETRAULT: It's my understanding -- and I haven't
seen any data -- it seemed, at 2,000 degrees, to take quite
a lot a long time.
MR. WHITE: Depending on where the bundle is or where
the wire is and how big the bundle it's in, because you know
it provides some heat sink and stuff, there's a lot of variables
in there. They're still trying to devise some more testing
to get a better feel for the kind of heat rates you can put
into bundles, but it's not inconceivable that you could breach
the spar and less than 30 seconds later you could start burning
ADM. GEHMAN: As we did.
MR. WHITE: Yes, sir.
GEN. BARRY: One quick question on the nose sensor.
We've had failures before in MADS data sensors.
MR. WHITE: Oh, yes. We have failures, yeah, maybe a
couple per flight, where the sensor fails for one reason or
GEN. BARRY: You can tell a difference between a failure
and one that --
MR. WHITE: Yes, sir. The folks that are used to looking
at the data at every flight can tell when it's failed and
we put them on a list and depending on how much time we have
in the turn-around -- because these measurements are all Crit
3, that means that we don't need them for anything in flight.
It's good data to have and engineers like to see this data,
but we don't rely on it for anything in flight. So if they
have time to fix them during the turn-around, they'll fix
them. Otherwise we'll just fly with a piece of paper that
says this one's broken and we'll fix it when we can.
GEN. BARRY: A point to be made. The ones you're showing
in this briefing are ones that you determined --
MR. WHITE: Yes, sir. These were all working measurements.
Right. I'm not showing you any that were determined to be
bad here. Yes, sir.
Let's see. Keep going a little farther. Okay. We talked about
the clevis. We talked about the first sensors going off line.
DR. WIDNALL: Could I ask a question. Where is the wire
that they share in common? You said they both went off line
at the same time. You said they share a common something or
MR. WHITE: Well, the power supply and the avionics
for the MADS would be about here in the mid-body; but the
wiring that they would share would be wiring that comes from
here into the avionics box and this wiring here, this blue
wiring that runs along the spar and then connects in through
here to the mid-body and then over to the MADS avionics boxes.
We believe what happened is, because of a short or a burn-through
in this blue bundle here along the leading edge, that it pulled
down the voltage to the power supply, which also dropped this
DR. WIDNALL: Because otherwise it's sort of mysterious.
MR. WHITE: Yes. We believe we can correlate the right
wing ones with the left wing ones where they have failures.
This particular point here, 5217, is the previous earliest
measurement that we had seen. This is from the OI data. This
is where we thought things were beginning to happen. Again,
if the wing is breached somewhere in this area and we have
hot gas entering the wing, there may be enough that gets around
into the wheel well just a little bit to cause that temperature.
You remember that was just a bit flip and it was very small;
but it is possible, with heat coming in through the wing,
that we are now seeing that sensor begin to respond.
ADM. GEHMAN: Now, that is significant, what you just
said. The temperature rises that we saw on those two spar
temperature lines were measured in big numbers, hundreds perhaps.
MR. WHITE: Yes, sir. And I indicated those by making
these dots red which says that these were quite significantly
out of what they should be at this time, greater than -- well,
let's see, I guess in the color-coding here it would be greater
than 30 degrees by this time. It gets significantly hotter.
Here this is a very small temperature range.
All right. Next. This is a strain in the spar, the 1040 spar
that runs in front of the wheel well. Again, we believe we're
seeing off-nominal measurements here because of the shifting
loads within the wing as the heat begins to damage things;
and this is one of the two measurements that never did drop
You notice here in my count I'm starting to show how many
have failed in Bundle No. 3, which is the blue bundle here
and down the side.
Next chart. A couple more sensors drop off line. Again, these
are all connected to this leading edge bundle here again,
which is the one that you would expect to fail first, the
ones I showed you in the back of the spar, and probably haven't
gotten over to start burning any of these yet.
Next chart, please. Okay. The measurements for the temperature
here on the leading edge. The surface temperature behind Panel
No. 10 on the lower surface and the one in the clevis are
starting to look off nominal. It looks like they're being
damaged at this point and that we can no longer trust the
Next chart, please. This is the spar measurement itself and,
again, the lower surface pressure measurement here showing,
again, unreliable data, showing damage trend to the wire.
Next chart. Another comm dropout.
Next chart. You notice we're still at 52 minutes and only
27 seconds now. We haven't gone very far forward.
ADM. GEHMAN: We're going to go second by second here.
MR. WHITE: Pretty much. So if you want to jump a little
faster. But you can also notice that my count is increasing
here. I've got two failed in Bundle No. 1. I've got 20 failed
in Bundle No. 3.
ADM. GEHMAN: Well, just go ahead and just clip through
them. You don't need to describe each wire that breaks because
the next significant events --
MR. WHITE: Next chart. This is OMS pod temperatures.
These are the supply water and waste water vacuum vent nozzle
temperatures that we talked about before. Showing a little
off-nominal heat rise. Again, we still haven't been able to
explain how that correlates with anything that was happening
back here in the wing.
GEN. BARRY: A point to be made. Is this about the time
we had our first telemetry reading on the previous operational
MR. WHITE: Yes. That was actually a few seconds before,
when we saw this one in the wheel well rise.
GEN. BARRY: 52:17. So all this that you've shown is
MR. WHITE: But is very close. Yes. This is only 52:32
Next chart. Okay. There's another measurement off line.
Next chart. There's some brake temperatures. Again, we had
seen these before. That's starting to rise. More heat in the
wing. More heat in the wheel well.
Next chart, please. Okay. Supply water dump nozzle.
Next chart. Another comm dropout.
Next chart. The attach clevis now went back to nominal.
Next chart. This is the one on the temperature on the spar.
Now it's starting to go off line; and we're still at 52 minutes,
now 51 seconds.
Next chart. More sensors off line.
Next chart. Vacuum vent nozzle begins to rise.
Next chart. Now that front spar temperature finally does go
off line. So the size of the hole here must have increased
enough to take out that sensor.
Next chart. Some more skin temperatures going off line.
Next chart. This is where we start to see roll moment happen.
So now the damage into the wing has begun to be serious enough
to affect the roll of the vehicle.
Next chart, please. Some more sensors off line. Now we're
only at 53 minutes. We've barely gone a minute, and you can
see the wire failure counts are pretty high -- 9 of 11, 99
of 138, and 6 of 35.
Next chart. This is an OI measurement that went off line.
Next chart. Some more. These were ones from the OI that went
ADM. GEHMAN: Now, these are the four elevon actuator
temperatures that went off essentially at the same time.
MR. WHITE: Yes, sir.
ADM. GEHMAN: And this was then noted in mission control
MR. WHITE: Yes. These are the ones that alerted something.
The MCC began to notice something that was wrong, that these
four should not have failed all nearly at the same time.
ADM. GEHMAN: So you might say this was the first indication
people on the ground had any idea that anything was happening
that was unusual.
MR. WHITE: Yes, sir. That's correct. The temperature
rises that we had in the wheel well were pretty subtle and
were hard to pick up if you didn't know -- you know, it's
only going back and looking at it that we know and pick this
up. But these measurements failing here were picked up immediately
and, as you said, were the first indication to the folks on
the ground that they had a problem.
ADM. GEHMAN: And depending on what displays were being
displayed at MCC. So even though those wheel well temperatures
are telemetered to the ground, they may not be actively looked
at at every instant.
MR. WHITE: Yeah. I can't answer that. I can't be sure
what the MCC looks at routinely.
ADM. GEHMAN: We do know, based on the video and audio
recording, that the loss of these four elevon actuator line
temperatures was noted and reported and this is when the conversation
MR. WHITE: Yes, sir. And then this, position-wise,
we're still not quite at the California coast yet.
Next chart. OMS pod temperatures now start to rise. This is
one that was cooler earlier. It's now starting to rise. You
can see other parts of the OMS pod. This one is still cooler,
and this one is very hot. So we've shifted the vortices around
Next chart. More pressure measurements going off line. Strain
Next chart. Some side wall fuselage temperatures rising now.
Some of these had also been cooler and now are getting hotter.
Next chart. Again, another side surface temperature behaving
Next chart. Comm dropout. Now some more strain measurements
and elevon return line temperatures going off line.
Next chart. Now my supply water dump nozzle, my vacuum vent
nozzle returned to nominal.
Next chart. Another hydraulic system elevator -- excuse me,
elevon actuator return line temperature going off line.
Next chart. Now, the strain. This is the other measurement
that hung in there but, again, is showing an off-nominal reading
in front of the wheel well on this spar. Again, it tells us
that the load is being redistributed within the left wing.
I can't tell you exactly what damage would have caused these
measurements to behave the way they did, but there was damage
and it was causing the load to redistribute.
Next chart. This is now the first debris sighting. We're over
California, and so this was the first debris event. Again,
it could have been tile falling off the lower wing. We know
we had a lot of heat in here that damaged all these sensors
in here. It could be upper-wing skin. It could be upper-wing
tile. It could be lower-wing tile. We see a number of tile
that indicate that they fell off because they were melted
off from the inside, not that they were damaged or melted
off from the outside.
ADM. GEHMAN: Of course, this is the first observed
MR. WHITE: First observed debris. There could have
been debris earlier. Of course, we haven't found any tile
out in California or any debris of any sort out in California
that would tell us exactly what it was. We don't have any
confirmed debris until we get all the way into Texas.
Next chart. Another debris event.
Next chart. Third debris event.
Next chart. Fourth debris event.
Next chart. Fifth.
Next chart. Lower-wing surface temperature going off line.
You can see now pretty much failed all of my instrumentation
DR. WIDNALL: Actually this is kind of directed to Greg
but related to what you were talking about.
I looked at your image analysis work on some of the re-entry
where you're looking at these debris, and I'm very excited
about what I saw in your briefing. I assume you are trying
to infer ballistic coefficients of these various debris pieces
from some kind of relative deceleration of those debris relative
to the shuttle.
DR. BYRNE: My team takes the first step in that process.
We analyzed the motion of the debris as it shed for all of
these events where we've made some good progress in analyzing
the motion relative to the orbiter.
DR. WIDNALL: When you say motion, you mean deceleration
DR. BYRNE: Yes. We then turned our motion measurements
over to Paul Hill's team. I think Paul's going to speak later.
Then his team then calculates from those a ballistic coefficient.
DR. WIDNALL: When do you think those are going to be
available? Is he going to talk about that today?
DR. BYRNE: I think he will. I haven't seen his charts,
but I believe he is. In addition to the motion analysis that
we're doing on these debris events, we've also done the time
line; but we're also looking at the luminosity, looking at
the intensity of the light given off by the debris and trying
to use that to determine what other characteristics we can
from that mass and area in particular. We're making some progress
DR. WIDNALL: Great. Well, I look forward to that. Really
MR. WHITE: Let's see. We'll just continue to flip through
these. This is more temperatures in the wheel well now starting
to rise. Again, we believe the heat's been in the wing for
some time now, maybe for as much as two minutes, and it's
conceivable that we're starting to get higher heating in here
because of conduction or flow in through the opening in the
front of the wheel well.
Next chart. Another comm dropout.
Next chart. More sensors going off line.
Next chart. This is a point in the aero where we start to
see the aero change. This is the reversal in the roll moment
that you see from other charts. The roll moment was going
negative and for some reason it turns around and it starts
to grow and go positive. So, again, some possibly significant
structural damage within the wing itself or possibly a large
piece of skin being shed to affect the aerodynamics of the
vehicle at this point.
ADM. GEHMAN: Or jetting.
MR. WHITE: Possibly, yes, sir.
ADM. GEHMAN: Or just some kind of a change in the geometry.
MR. WHITE: Somehow or another the shape -- either because
of internal damage, the external mold line changed, or pieces
came off. There's a number of ways that we could have affected
Next chart. More temperatures on the fuselage going up. Again,
this one was an OI one that we knew about from before.
ADM. GEHMAN: Okay. I'm going to ask you to just flip
forward. I think what we want to get to is 59:32.
MR. WHITE: Actually I only carried this through about
where the wheel well, in our estimation, was breached.
ADM. GEHMAN: Then I do have a question about that,
about the MADS data, because the MADS data does two things
that the previous data, which was telemetry down to the ground,
do not do. One is that it fills in the 25-second gap. Remember
when we have loss of signal, then we have these 32 seconds
which was retrieved, of which there was 5 seconds of data,
25 seconds of gap, and then 2 seconds of data. So this recorder
was running during those 25 seconds.
MR. WHITE: Yes, it was.
ADM. GEHMAN: Anything significant from those 25 seconds?
MR. WHITE: From the left wing -- and you can even see
from where we are here -- almost everything in the left wing
had gone off line by this time; and what we see over in the
right wing, except for those that failed sympathetically with
left wing measurements, those measurements all hung in there
and appear to be good. So there's no new, startling data in
that gap that says there was anything significantly wrong
with the vehicle.
ADM. GEHMAN: And the sensors in the mid-body fuselage
were all working.
MR. WHITE: Appeared to be working and except for the
ones we know of, temperature measurements that were higher
than they should be, there were no indications of anything
internal to the vehicle going off line.
ADM. GEHMAN: Right. That's one area of information
that the MADS data provided that fills in a nice gap for us.
That indicates that the vehicle was intact and the electrical
system was working and the right wing, at least, was on.
Then another thing that the MADS data does is it continues
about -- I forget what the number is -- 9, 10, or 11 seconds
longer than the telemetered OI data. I don't know the exact
numbers, but it goes for about another 9 or 10 seconds.
MR. WHITE: That's correct. Another 9 or 10 seconds.
ADM. GEHMAN: Is there anything there?
MR. WHITE: Once again, the MADS data, once we pretty
much failed everything in the left wing and the higher temperatures
that we've been seeing all throughout entry, again, there's
no startling data in that extra 9 seconds either.
ADM. GEHMAN: Okay. Board members?
GEN. DEAL: I've got one. It goes back to your very
first slide. You started talking about how some of the instrumentation
has been taken out and some of it was broken. Can you give
me a little bit more insight into what was broken? Did we
look into why it was broken? For example, were any of them
strain gauges or anything like that?
MR. WHITE: Yeah, I don't have the list. There are probably
a handful, maybe a dozen or so, that were off line for this
flight; and I could get you a list. I just don't know off
the top of my head which ones. I assume it's a little bit
of each -- pressure, strain, and temperature.
GEN. DEAL: Just curious if any analysis had been done
about why they broke.
MR. WHITE: I don't know the answer to that. They work
these things on a routine kind of basis.
MR. TETRAULT: Somewhere in the 300-second area, you
showed one of the first sensors on the OMS pod going low.
In fact, there were, as I recall, four sensors on the OMS
pods that went low just somewhere in that time frame. For
those to go low, you talked about the flow of the air was
obviously changing at that particular point. Wouldn't that
suggest that there was something on the top of the wing that
had to be missing at that particular point? We've talked about
issues of foam striking the bottom of the wing; but at that
point, for that to go low, wouldn't there have to be something
that was missing on the top of the wing?
MR. WHITE: Well, we've done some wind-tunnel testing
where we just arbitrarily took sections out of the leading
edge of the wing; and actually I believe about the Panel 5
region, if you took Panel 5 out, you can actually get cooler
temperatures along the side of the OMS pod.
MR. TETRAULT: But that's a full panel, which wouldn't
include the top of the wing.
MR. WHITE: That's a full panel, right. What I'm trying
to say, I guess, is that we haven't done any wind-tunnel testing
with some sort of a protrusion or a missing hole or anything
on the top of the wing to see what that would do to OMS pod
temperatures. One of the things we have to do to finish our
scenarios is to make sure we can understand the aerothermal
in such a way that we can get increased and decreased heating
as the time line progresses. But I don't have any data right
now that says, yes, something on the top of the wing would
cause me cooler temperatures. I do have some data that says
some configurations of leading edge damage could get me cooler
ADM. GEHMAN: Correct me if I'm wrong here. Is this
not a rather unique aero environment because at a 40-degree
angle of attack and a 70-degree roll angle -- talking about
the top of the wing and the bottom of the wing leads you to
a funny conclusion.
MR. WHITE: It's not like a regular air flight. Right.
ADM. GEHMAN: It's more like a blunt surface, and so
it really presents a real aero challenge.
MR. WHITE: Yes. It's quite difficult to go figure out
exactly how the vortices shift around.
ADM. GEHMAN: Right. But we're going to do that.
MR. WHITE: We're pursuing it. Yes, sir.
MR. HUBBARD: Any thoughts on the source of the comm
dropout, communications dropout?
MR. WHITE: There have been some theories -- and again,
these are just theories -- that perhaps as we were shedding
material, if it had metallics in it, that that would interfere
with the comm, if you were melting away parts of the insulation
on the leading edge spar that perhaps you would get enough
metal in the stream behind the vehicle to interfere with the
comm. But there isn't any way we can prove that. That's just
MR. HUBBARD: As far as you know, the transmitter was
working and receiver in TDRSS was working. So something interfered.
MR. WHITE: Yes, sir. Right. The only reason we described
it as anomalous is that you look at other flights of 102 for
these inclinations and these look angles to the satellite
and we didn't see this number of comm dropouts. So we just
flagged them as anomalous.
MR. HUBBARD: Thank you.
ADM. GEHMAN: Well, thank you very much, Mr. White and
Mr. Byrne. I know that what you've shown us here today represents
the tip of the iceberg for the amount of work that's been
done by not only yourselves but a great team of people that
reach way, way down into both your organizations. We appreciate
very much not only this presentation and your willingness
to dialogue with us in a very frank manner but also the hours
and days and days and days of work that you and your team
have put in and will continue to put in because we have several
mysteries here that we can't explain.
The board is very grateful for your cooperation and also for
the energy and the zeal by which you and all your people have
pursued this. We both have the same goal to find out what
happened here; and we're going to have to find out what happened
by good, hard, roll-up-your-sleeves kind of detective work.
You and your folks are doing that. So we're very grateful.
You are excused.
The board will take about a ten-minute break while we set
up for the next panel, and we'll be right back.
ADM. GEHMAN: All right. We're ready to recommence.
For the next panel, we're going to discuss the object that
was observed on Flight Day 2, 3, and part of Flight Day 4;
and we're very pleased to have two experts join us here today,
Mr. Steve Rickman and Dr. Brian Kent.
Gentlemen, before we start, I'll ask you to affirm that you're
going to tell us the truth; and then I'll ask you to introduce
yourselves and say a little bit about your background and
where you work. Then the board would be pleased to listen
if you have a presentation or an opening statement.
Before we begin, let me first ask you to affirm that the information
you will provide the board today will be accurate and complete,
to the best of your current knowledge and belief.
THE WITNESSES: We do.
ADM. GEHMAN: Introduce yourselves, tell us where you
work and a little bit about your background, and then we'll
have an opening statement.
STEVE RICKMAN and BRIAN KENT testified as follows:
MR. RICKMAN: My name is Steve Rickman I'm chief of
the Thermal Design Branch here at the Johnson Space Center.
I got involved in this particular endeavor because if you
look at the outside of the vehicle, there's a lot of things
on there that are either thermal protection or thermal control
related. So I got involved in this effort; and it's been a
very, very interesting challenge. I have a Bachelor of Science
degree from the University of Cincinnati in aerospace engineering.
I have a Master of Science degree in physical science from
the University of Houston at Clear Lake.
DR. KENT: My name is Dr. Brian Kent. I work for the
Air Force research laboratory in Dayton, Ohio. I'm a specialist
in radar signature measurements. I've been working in this
particular area for 26 years, the majority of my adult career.
I have a Bachelor's and Master's in electrical engineering
and a Ph.D. The Bachelor's from Michigan State, the Master's
and Ph.D. from Ohio State. I direct most of the activities
not only within our own facility for signature measurements
but I also chair a multi-service panel that works signature
standards for the Army, Air Force, Navy. That is involved
in the National Institute of Standards and Technology. So
I've been actively involved in quality control efforts in
signature measurements for a number of years.
ADM. GEHMAN: And normally we can find you at the Air
Force research lab at Wright Patterson Air Force Base. Is
DR. KENT: That's correct, sir.
ADM. GEHMAN: Please go ahead.
MR. RICKMAN: Okay. If I may have the cover slide for
our presentation, please.
First of all, I would like to thank the board for the opportunity
to appear this morning. This has been quite an effort. It's
involved a number of agencies, NASA, and various organizations
within the United States Air Force; and it's truly been a
team effort. What our effort has focused on was trying to
get an understanding from a ballistics and a radar cross-section
standpoint of the object that we refer to as the Flight Day
2 object that was observed coming off of the Columbia from
Next slide, please.
ADM. GEHMAN: In accordance with the board's long-standing
tradition of never letting any presenter getting past the
first viewgraph, may I make the observation that the object
was not observed coming off the Columbia.
MR. RICKMAN: Yes. Perhaps I didn't state that correctly.
It was a post-flight --
ADM. GEHMAN: What I mean is there's no -- unless you're
going to tell me something I don't know here -- we don't have
any observation of anything coming off the Columbia.
MR. RICKMAN: That is correct.
ADM. GEHMAN: It was observed on orbit accompanying
the Columbia. One hour it wasn't there, and the next hour
it was there.
MR. RICKMAN: Yes.
ADM. GEHMAN: And we don't know how it came off or what
-- we don't have any observation of anything coming off the
MR. RICKMAN: That is correct, sir. We have some charts,
I think, that will clarify that.
ADM. GEHMAN: Thank you very much.
MR. RICKMAN: Here's our plan for today. We first want
to give acknowledgement to the organizations that have been
involved in this rather large effort, give you a little bit
of background on what we know about the object, talk about
our approach to better understanding it through the radar
cross-section testing and the ballistics analysis. I'm going
to give a brief description of all the shuttle hardware tested.
Some of the items I have here today. Then I'm going to turn
it over to Dr. Kent, who will give a summary of all of the
UHF radar cross-section testings and ballistics analysis,
and then we'll wrap it up and along the way we'll be happy
to answer any questions you may have.
Next chart, please. I mentioned before that has truly been
a collaborative effort. It involves the Department of Defense,
the United States Air Force, and NASA. You see all the organizations
that are listed up there. We could not have done it without
the support of all of these organizations, and it truly has
been a joy to work with these groups. Everybody's been very
helpful and professional, and anything that we had in our
way has magically disappeared and we've been able to do our
job. So we're very appreciative of that.
Next chart, please. A little bit of background information.
While up on orbit, there were 3180 separate automated radar
or optical observations of Columbia collected. There were
collection sites at Eglin Air Force Base, Beale, Naval Space
Surveillance, Cape Cod, Maui, and Kirtland Air Force Base.
It's important to note here that each observation was individually
examined after the accident. The debris piece was detected.
It was a very laborious effort of post-flight examination.
It was the most laborious post-flight examination that the
Air Force Space Command has ever conducted for a shuttle mission.
It required just over 285 manhours just in the first week
alone after the accident.
The Air Force catalogs these things, and you can see the catalog
numbers there. It's been referred to as Object 90626, but
I think we'll just refer to it as the Flight Day 2 object
from this point on.
Next chart, please. This is an example of some of the data
that we've been looking at. Just to give you some orientation
here, along the bottom is Greenwich Mean Time. This object
separated on Flight Day 2. The best time that they have for
a window of separation is somewhere between 15:15 and 16:00
on Flight Day 2. That would have been January 17th. You can
see how it tracks away from the shuttle's orbit, which is
shown in red there, and it's expressed in terms of delta time
(seconds). So this is seconds of separation. The various symbols
that you have on the curve there show the various sites that
gathered the data.
Next page, please. What we do know about the object is it
has certain ballistic characteristics or a B term. What we're
looking for are objects that match this ballistic term or
B term and what we have up there is the B term there, drag
coefficient C sub D, area-to-mass ratio. CD times A over M.
And we're looking for objects that fit the .10 meters square
per kilogram and that's believed to be known within about
plus or minus 15 percent.
The estimated physical size of the object was between approximately
.4 meters by .3 meters. So it's roughly square, and the object
was initially in a semi-stable or slow rotation on January
17th. And Dr. Kent actually has some of the data to share
with you to show how over time the object began to spin up.
The first day it was rotating about once a minute. The next
day, in a Cape Cod pass, it was rotating about once every
7 seconds. The day after that, it was rotating about once
every 3 seconds; and it actually fell out of orbit approximately
60 hours after it separated from the orbiter.
Next chart, please. Okay. Well, what else do we know about
the Flight Day 2 object? We also have radar cross-section
data. That was taken in the UHF frequencies at 433 megahertz
and it varies between minus 20 decibels per square meter to
minus 1 decibel per square meter and Dr. Kent will give you
a better understanding of exactly what that measurement entails.
With high importance, we've also bounded what the confidence
level is within plus or minus 1.33 decibels.
The next chart. The way we approached this -- and I'm going
to show you a couple of picture here in a minute -- is we
had to take a look at what we would see on the outside of
the vehicle, what had the potential to get away from the vehicle.
In my organization we tend to break those things into two
classes, what we call thermal protection materials, or TPS
-- those help protect the vehicle against the high entry heat
loading. In that category I also put the leading edge subsystem
or reinforced carbon-carbon components that there's been a
lot of discussion of. And then we also have thermal control
system, or TCS components, which would be representative of
what you would find in the cargo bay. Those components are
there more to protect the vehicle from the extreme temperature
swings that you would get while going around in orbit, hundreds
of degrees above zero to hundreds of degrees below zero in
a very, very short time.
So we basically applied two gates that any object or any candidate
object had to get through. It had to match not only the RCS,
the radar cross-section information. It also had to measure
the ballistic coefficient, but also we're very mindful of
the fact that there's been a lot of debris collected, a lot
of forensic evidence down at the Cape. So obviously if something
shows up on the floor down at KSC, it's something that we
can exclude; or if it was something that we carried with some
interest previously, once it is found, then we can exclude
that, as well. So candidates failing to match even one of
those criteria are excluded as possibilities for the Flight
Day 2 object.
Next chart, please. I mentioned before this is an overview
of the thermal protection system constituent materials. We
try to be very methodical in our approach to performing this
investigation. We have various materials on the outside of
the vehicle. The light blue -- and it doesn't really show
up very well here -- represents the LI 900 or the 9-pound-per-cubic-foot
density tiles. We also have 12-pound-per-cubic-foot density
tiles and 22-pound-per-cubic-foot density tiles. Those comprise
the lion's share of the acreage of the bottom of the vehicle.
On the side of the vehicle, we have a blanket insulation that
we refer to as AFRSI, Advanced Flexible Reusable Surface Insulation.
We also call it fibrous insulation blanket. That's good to
a lower temperature than the tiles. This is in a more benign
area of the vehicle. We also have FRSI, which stands for Flexible
Reusable Surface Insulation or felt reusable surface insulation.
It's a needled Nomex felt. We also have AETB-8 tiles. I believe
those are vacuum-based heat shield.
The tile materials are all going to look very similar to one
another. As a matter of fact, I have a sample tile right here.
This is the 22-pound density tile. They vary in size and shape
as you go around the vehicle, but by and large on the bottom
acreage they're approximately 6 inches by 6 inches. So this
would be representative of the shape; and, of course, the
thickness varies as a function of location. As you can see
here, by just testing a handful of materials, you can cover
the lion's share of the outside of the vehicle.
Can I have the next slide, please? I already showedyou a picture
of the tiles. We tested them in a number of different varieties.
For example, the LI 900 tiles, we weren't sure what would
happen to the radar cross-section if we also included the
RTV adhesive on the back and the strain isolator pad, which
is Nomex felt. We also didn't know what densification of the
tile would do. Densification is a process that we do that
increases the density about .15 inches at the bottom of the
tile and helps it adhere to the vehicle. So we tested in a
densified and undensified state. LI 2200 tile looks the same.
Here's AFRSI and FRSI.
May I have the next slide, please? There was also interest
early on on testing carrier panels or segments thereof. I
have with me here the actual mockup of a carrier panel that
we tested up at Wright Patterson Air Force Base here. It consists
of 22-pound density tiles, a metal support plate on the back,
and also an insulation called horse collar, which is Nextel
with a sheet of Inconel in it. So this was tested early on.
At the time we found great interest in that sample. We ultimately
asked for and received some flight assets, in particular some
actually flown four-tile and three-tile variant carrier panels
that have more hardware on them; and we got those up to Wright
Pat for testing, as well. We also tested the horse collar
all by itself.
Next chart, please. Given the intense interest in the carbon
system, we had some flight assets sent up to Wright Pat. We
had a flight RCC panel tested. We have some Incoflex ear muff
spanner beam insulation. As a matter of fact, I have that
right here. This is Inconel over a serochrome batting, and
this would be located behind the wing leading edge panel.
So it's normally inside of the wing.
And then our latest area of focus has been on the actual T-seals.
This is a T-seal that's undergone testing up at Wright Pat,
Next chart, please. Once we had some preliminary measurements
on the reinforced carbon-carbon pieces, we needed to do a
little bit of refinement; and one of the best ways to do that
was to retrieve some pieces from the debris from Columbia
down at KSC. What we were looking for are different classes
of objects, different classes of carbon objects, like what
I refer to as carbon acreage. It's essentially a piece out
of an RCC panel. So we tested a few samples with that, with
and without lips. We also tested segments of RCC T-seals to
get a better idea of what fragment of a T-seal might give
you the appropriate radar cross-section.
Next chart, please. Okay. That's the outside of the vehicle.
Now, if you look inside in the shuttle cargo bay, there are
a number of thermal control system materials there. When you
look out over the cargo bay and you see a lot of white, what
you're really looking at is a material called beta cloth.
Beta cloth is a glass fiber material. A lot of times it has
a Teflon sizing over it; but if you look at something that
creates the cylindrical surface of the cargo bay, what you're
actually looking at is multilayer insulation.
Multilayer insulation is a very good thermal control insulator.
You can have temperature gradients of a couple of hundred
degrees across a sample of about this thickness. If you were
to cut into this, what you would see are alternating layers
of an aluminized plastic like Kapton or Mylar and Dacron spacer
mesh. So there is metalized layers in here. You'll also note
that this has metal quilting in here in the form of a stainless
steel wire to help it from electrical grounding.
If I can go to the next slide, please. We tested a variety
of multilayer insulation blankets -- some from payloads, some
from the cargo bay itself. We even tested logos off of payloads.
I should mention that it's my understanding that they did
a survey post flight from the video coming down to see if
they noticed any difference in the cargo bay. I believe about
60 percent of the cargo bay is observable from the cameras,
and no differences were found. So if there was an object that
was conspicuously missing from the cargo bay, it would have
likely been detected from that survey.
Next chart, please. In addition to the multilayer insulations,
there's various types of bulk insulations that we have in
the cargo bay. If you were to look inside here, you would
see a glass batting that's inside here. This is beta cloth
with the familiar quilting material on it, this is Kapton
on the back, and this protects the vehicle, in regions it
needs to, from higher heat loads.
There's actually three different varieties of this bulk type
insulation. The one I found pretty interesting to look at
was this one. This is actually the type of insulation that's
beneath the cargo bay radiators. I should point out and did
not point out but at mission the last time of about 3 hours
and 8 minutes, the port side radiators were deployed. So if
there was an object under there that could have possibly escaped,
that might have given it an opportunity to do. Those radiators
stayed deployed through mission elapsed time about 3 days,
7 hours, and 50 minutes; and then they were redeployed again,
I think, on the 11th day of the mission. So this is the type
of blanket that you would see beneath the radiators.
We also had a question from a board member a week or so back,
asking us is it possible that any tool might have been left
beneath the radiator. We did a little bit of checking into
that. The only thing we were able to find as a possibility
would be a crimping tool that would be used for blanket snaps.
We had some ballistic analysis done on that, and we'll be
talking about that today.
May I have the next chart, please? I'm going to turn it over
to Dr. Kent now. Now, one final thing I did want to mention,
though, just so people are aware of it, is there was an attitude
maneuver that corresponds with the time just prior to when
we think the object was released. What was happening at the
time is the shuttle was flying in a cargo-bay-to-earth tail-on
velocity vector attitude. That happened at mission elapsed
time -- well, the GMT on it would be January 17th. I believe
it was 14:42 GMT. The vehicle yawed 48 degrees, biassing the
right wing into the velocity vector, and then I think it was
at 15:17 GMT they went back to the tail-on velocity vector
attitude. The nearest maneuver to that, prior to that, was
about mission elapsed time 8 hours. After that, the next maneuver
wasn't until about mission elapsed time 48 hours.
MR. WALLACE: Mr. Rickman, could you characterize that
maneuver you just described? Now, I understand it to be an
extremely benign maneuver. Would that be accurate?
MR. WHITE: Yes. I'm glad that you brought that up.
This particular mission had approximately 500 attitude maneuvers
in it, and we've flown missions before where we've had many
maneuvers. So this is very run-of-the-mill. This is very,
very benign, yes; and I believe this particular maneuver was
done for an IMU alignment to support a given payload, an initial
MR. WALLACE: In terms of it imposing any stresses?
MR. RICKMAN: Actually this particular maneuver was
done with the vernier jets. Those are about 25-pound thrusters
as opposed to the primary RSC, which I believe is somewhere
in the neighborhood of 800-pound thrust. So, yes, it was done
with very gentle jetting.
ADM. GEHMAN: Thank you very much.
DR. KENT: Okay. What I'd like to do now is to proceed
directly into the summary of radar signature and ballistics
analysis. I'd like to acknowledge my coworker, Dan Turner,
who worked many hours with this, as well as my collaborator
out at Space Command, Mr. Robert Morris.
The key point I want to make here on this chart is we've invested
about a thousand hours in this activity since the 3rd of March
but I also want to point out, too, that we did testing not
only at UHF band, which is the subject of what we're talking
about today, but we also did a significant amount of RCS testing
at FAA radar bands -- that's the L and the S band -- as well
as the ascent-tracking radar that's used when the shuttle
goes up. It's C band. That information has been turned over
separately to the flight directors; and I believe Mr. Hill
will be commenting later on how that particular data is going
to be used as part of the debris characterization recovery
efforts. This particular discussion, we'll solely discuss
the UHF testing in relation to the Flight Day 2 object.
Next slide, please. What I want to start off with is to very
quickly review the actual data that we have in hand. As we've
talked about, it was observed by multiple sensors. I'm going
to concentrate on the two sensors that were used that are
characterized in radar signature terms. Those were what we
call the Pave Paws radar, located at Cape Cod and at Beale
Air Force Base. I then will give you a brief description of
our test facility and how we use it to actually simulate the
same radar signature conditions that were observed for the
on-orbit measurements and how we're comparing the two. Then
I'm going to basically walk through these candidates that
we've examined and show you how very quickly you can, either
from a ballistic standpoint or an RCS standpoint, move a large
number of the classes of objects off the table and focus our
activities only on a few of interest. Then I'll give you a
quick summary at the very end.
Next slide, please. This basically I'm showing are the four
most reliable on-orbit observation measurements of radar signature.
The one in the white, which I did differently, is the one
observed at Beale on the 17th of January. What I've indicated
there is something that we did throughout the effort but I've
added to this particular piece of information. We've added
on top of the data, which is in black, a red and a green line
that indicates our level of fidelity or understanding or,
let's say, level of accuracy of the data that we believe has
been taken. This is very important because if you have a certain
data range that's like this and you're trying to match another
object to it, it's very important that the fidelity range
of your actual measurement falls within the actual on-orbit
observed, or else it becomes excluded. So we thought it was
very important very early on to get the information necessary
to assess the accuracy of this data so that we really knew
what we were starting with.
So you notice the first yellow chart in the upper-right corner
here is the first-day data. You notice this very slow, over
60-second period here of the revolution of a tumble of approximately
a period of about once per minute. By the time of the second
day, you can see that the tumble period has increased; and
by the third day it's gone up quite a bit, shortly before
Next slide, please. What we glean from this particular information
was on the Flight day 2, 3 and 4 tracks is that the observed
RCS varied from, for instance, Flight Day 2, approximately
minus 18 to minus 4 decibels per square meter. The Beale data
tracked around minus 17 to zero; and that's not too unusual
because, remember, they're observing this particular target
at different spots in the United States. So that particular
object, if it were floating around, would present a different
angle to those two radars. Day 3 and Day 4 tracks varied between
minus 15 and minus 2, minus 13 and minus 1.75; and you can
see the fidelity.
I should also point out that these particular radars, since
they're designed to penetrate through radar, operate in what
we call circular polarization. That means that the actual
electric field that's radiated from these radars rotates,
and this allows superior coverage through bad weather. It's
used by Doppler radars, for instance. In this particular case
the data was transmitted left circular and received right
circular; and as you'll see, the way that we actually take
measurements are in linear polarization and then we mathematically
combine them to simulate the same numbers.
Next slide, please. This is the advanced compact range at
Wright Patterson Air Force Base. That's where my day job is.
Basically, it's a major facility. It's an occult chamber.
It's designed to take radar signature measurements from very
low frequencies, around the television band, all the way to
very high military frequencies. The actual signatures that
we're talking about in this particular comparison at UHF are
433 megahertz, is kind of on the low to mid range of what
our capabilities are. The facility is capable of testing actually
a very large object, so that objects on the size of what's
on the table here are well within our capabilities; and because
the levels that we're talking about are fairly high in signature,
it didn't present any significant technical challenges in
terms of doing the measurements.
Next slide, please. This is, for instance, a setup showing,
for instance, that one blanket that Steve just showed you
here. That's mounted on a very low cross-section foam. In
other words, this foam piece here that actually holds the
target has a very low radar scatter, does not contribute to
the experiment, and we can also subtract out its residual.
Now, this big reflector that you see in the background, essentially
what this is like, you can think of it like the equivalent
of a telescope. By putting a radar very close to a reflector
at its focus, basically what that does is allows us to simulate
a very large separation between the radar and the target,
like what was really observed on orbit, in a very small or
compact space. That's where the name "compact range" comes
Next slide, please. I wanted to start off just to kind of
ground you in terms of the data. This is one of the test cases
that we run before we do any kind of experiment. It's just
one of many. This is strictly a 12-inch-by-12-inch metallic
conducting aluminum plate. The reason we wanted to present
this to you is you'll notice for a square plate this oscillatory
behavior here. What we're looking at is we're talking about
aspect angle or orientation angle. So in other words, if this
is my plate, when we talk about aspect angle, that's the orientation
of the plate relative to the radar. So if my radar is out
here and I talk about zero degrees, that means I'm looking
normal or perpendicular to this plate. As I move it out to,
say, 180 or zero or whatever, I'm going off normal here. So
the peak scattering for a flat plate tends to be when you're
normal to the plate and the lowest level tends to be off normal
and that depends on the frequency of the radar that's actually
illuminating the object.
I should also point out that radar cross-section, the physical
property that we're measuring, is not a function of weather.
It's not a function of atmosphere or any of those kinds of
things. It's a physical property that relates to how much
radar energy is scattered from an object, based on what's
The second thing I want to point out to you is what we normally
do is that we normally take these two linear polarizations.
The vertical, which is the VV, and the horizontal, HH, are
always referenced to the ground; and then we construct what
we call the circular polarized data, which is the on-orbit
data, which rotates continuously. So I wanted you to see that
because you'll see the patterns of these kinds of shapes are
going to be very similar to this standard that we use so that
we know everything is working.
Next slide, please. So I'm going to give you a kind of a close-up
of one of these and not going to show you them in large groups
because very quickly we're able to eliminate a large number
of these classes.
This would be typical, for instance. This is the AFRSI fibrous.
It's approximately a 2-inch-by-12-inch piece, and what you
have down here is this particular scale is a radar cross-section
in decibels per square meter. Now, this looks like a linear
scale, but actually think of it in a logarithmic sense, in
the sense that something that's minus 40 is four orders of
magnitude lower in radar cross-section than something that's
zero. So what I've drawn on this right here is this box.
This is the maximum and minimum range of the on-orbit observed
values. Now, the minimum range is not nearly so much as important.
In other words, the observed eye can actually, in terms of
a measurement that we make, can be less than that because
we have a lot more signal that we can do than they do on orbit.
But what's important is this maximum value. You need to be
at or in excess of that maximum value somewhere in the aspect
presentation of this target for it to be a viable candidate.
So looking at this particular device here, this AFRSI, one
of the first things you notice is that it's nowhere close
to the box. As a matter of fact, it's orders of magnitudes
off. The RCS for this thing isn't anywhere close to where
it would need to be to be a Flight Day 2 object; and, therefore,
by default, it's immediately eliminated.
Next slide, please. So let's look at large groups of them
because I broke them off into several classes. The first,
items that we rejected because the RCS is clearly too low.
These included the FRSI, the tiles of all varieties -- and
that's no surprise. Because what's tile mostly? It's mostly
air. They're very lightweight, and it's basically a block
of air with a little bit of structure on it. As a result,
it inherently has very low cross-section.
We tested both the 9- and the 22-pound variety of these things;
the signatures are way too low. So we were able to eliminate
the tiles very quickly. The beta cloth that we were talking
about on the back of the insulation were also tested. For
the most part, those are also much too low. These Freestar,
the logos that are typically put on, are nonconducting. There's
no metal in them, and it's metal that contributes a lot to
the radar signature. So again, those were also way too low.
Next slide. Continuing that, we started off in measuring what
Steve referred to as the carrier panel mockup. We did some
initial measurements, but we also found out that there were
some differences between the mockup that was provided to us
and the real carrier panels. So we ended up measuring both,
just to be thorough.
Now, what we find, again here this box is the range of the
on-orbit values. The blue is this equivalent circular polarization,
and what you notice for the most part that it doesn't get
anywhere close to the peak value observed in any of the configurations
that we looked. I should also point out that for the more
complicated parts, because of their shapes, we generally oriented
them in two or three different axes, usually trying to highlight
the presentation that we would know would produce the highest
radar cross-section so that we would get an idea, since we
really don't know the angle between this object that might
be tumbling in space and the radar, what its exact RCS is,
what we do know is that it took swings in a maximum to minimum.
If we couldn't even come close to producing a maximum swing,
then likely that object was also eliminated.
Next slide, please. Finally the fibrous thermal blankets,
the carrier panel by itself, the collar seal by itself, and
the 22-pound tiles, again, were just not anywhere close to
where they needed to be from a radar cross-section standpoint.
So those particular items are immediately taken off the table.
Next slide, please. The next RCS results I'm going to do --
and I'm going to be intermixing a few ballistic results as
well with these things -- are on this class of what I call
lightweight thermal blankets as, again, Steve is going to
be showing you here in a minute. In this particular case what
you'll see when you look at these things is you say, "Oh,
look, the RCS is very close to the box. It must be a good
match." Well, two things I want to let you know. That shouldn't
be too much of a surprise because most of these thermal blankets
have metalized layers in them. They should look very much
like the metal plates that I showed you earlier that we used
as a test case.
The other thing that I'd like to point out, Steve, if I could
borrow this, is one might say, "Well, but that's a real flat
surface and these are kind of crinkly." You need to keep in
mind that the radar wave length that we're looking at, this
thing is on the order of 2 feet and, because of that, local,
small, minor variations in the actual shape are not going
to seriously hurt its radar signature. That will also become
important later as we start talking about RCC fragments. So
as I look at almost all the classes of thermal blankets, which
are all variations on a theme, some type of metalized layer,
some type of metalized Kapton, they all look like they could
fit very well within the RCS rate; but as I'll show you in
a minute, the area-to-mass or ballistics coefficient is not
right. I'll show you that data in just a second.
DR. WIDNALL: Wait. I have a question. I would certainly
agree with you that the area and the mass are probably not
right, but there is also the issue of drag coefficient and
I wanted to know what kind of drag coefficients would you
assume. I'm not trying to make these candidates, but you need
a drag coefficient. What do you use?
DR. KENT: Right. I think Robert Morris and the space
community are using a drag coefficient that I believe -- again,
you're asking a little bit outside my area, ma'am, but I believe
the number was .2 --
MR. RICKMAN: It was 2.2.
DR. WIDNALL: 2.2?
MR. RICKMAN: 2.2 for a drag coefficient, which is a
rectangle on the broad side and then for the tumble, they
time-average the area that's presented.
DR. WIDNALL: Okay.
DR. KENT: I'll have that figure for you in just a minute.
DR. WIDNALL: More fineness on that.
DR. KENT: These insulation space blankets are also
thermal materials. There's two others that I've included in
this particular category where the area and mass is wrong
but in this case the item was much too heavy and too large
and that's, of course, a full, intact RCC edge which we tested
initially just to kind of baseline what kind of signature
level we would get at UHF frequencies if an entire edge was
intact for whatever reason. Clearly, it's at or much above
the observed values. And I should point out this particular
RCC reinforced carbon-carbon edge has the T-seal installed
in the end and that will be important because you'll see a
lot of the pattern characteristics from a side aspect are
the same because it's the T-seal that's doing a lot of the
Next slide, please.
MR. TETRAULT: Excuse me. One of the RCC panels that
you tested, did it have a spanner beam attached to it when
you tested it, the original one?
DR. KENT: The answer to that question is, yes, it did.
I believe the picture showed that.
MR. TETRAULT: Will you make sure that we understand
which ones have metal attached to them and which ones don't
on your testing, please?
DR. KENT: Okay. With the exception of the RCC panel,
none of the other items that we had had any kind of metal
attachments, no bolts or anything else; but as long as you're
on that topic, sir, I will point out that if we're talking
about bolts that are like 2 or 3 inches long, at these radar
wave lengths -- again, the radar wave length's about like
this, and a bolt's like this -- it's going to have quite a
low scattering value and it's going to be very non-directive
in one of the two radar polarizations. So it's going to be
quite lower than the observed values that we're talking about
Again, I borrowed this chart from my compatriot at Space Command,
Mr. Morris, showing you the series of these lightweight blankets.
What I'm showing you is the B term or the ballistic coefficient.
I've labeled the various items down here. The important thing
is that the Flight Day 2 value here is the solid red line
and the dotted lines are its approximate level of uncertainty.
So it's not a matter like, well, these things are a little
off. They're a lot off. They're quite a bit removed from the
possibility. So it was fairly easy, again from a ballistics
standpoint, to eliminate these particular items, mostly because
they're too light. Now, again if somebody says, "Well, what
about a piece twice that size?" Well, keep in mind its area
to mass. So making the same material a larger piece is not
going to change this value. So again, that was one of the
reasons why these were not very strong candidates.
Next slide, please. Now, I'm going to show you a series of
charts where the RCS and ballistics begin to converge. The
first item, I'm showing you an example -- I believe this was
actually released in the press conference last week -- was
the wing spar insulation piece that Steve is holding up here.
It was a good match both in signature and insulation. Most
of these, I'm only showing you one view. There were actually
many views in terms of radar looks at these particular targets.
A whole T-seal was tested and shown to be well within the
bounds, both from a side aspect and a top aspect. Most recently
one of the things that had dawned on us when we actually tested
a T-seal -- and I'm going to use this. This, by the way, is
the attachment flange for a T-seal. One of the things that
dawned on us, because these are fairly strong scatters, that
the thing that fits inside of here -- which, of course, is
the RCC edge -- would also be a strong scatter. So we made
a recommendation a week and a half ago for us to look at what
we call acreage candidates or basically pieces of RCC that
would be on the order to find out how big a piece that we
would have to have to have it to be on the order of the RCS
for the Flight Day 2 object.
Now, you just don't go breaking away a piece of a perfectly
good, expensive RCC. So the methodology we decided to use
was to go down to the actual floor, look on the symmetrical
right-side area and look for fragments of RCC that were on
the order of the size that we felt as though would be appropriate
for signature. So keep in mind that even though we are measuring
debris components, obviously they're not the Flight Day 2
object because these were recovered parts from the right side.
They were used to bound the RCS or radar signature of the
RCC panel acreage.
So these last two items down here, which is what we call Fragment
2018 and 37736, which are just designators that they use for
the recovered pieces, both measured very close to the on-orbit
range; and these things, even though they don't see much in
this particular picture, are quite irregular. The parts can
be roughly squarish, but they can have some curvature or they
can have a lip on them. The point of the matter is that carbon-carbon
is fairly conducting and so it behaves quite a bit, again,
Next slide, please. Now, I do want to talk in particular about
this item. There seems to be a great deal of interest in the
T-seal; and we, of course, tested a whole T-seal as part of
our initial test package. What we really wanted to do was
to test a half T-seal; but again, you don't take a piece of
flight hardware and destructively cut it apart.
So what we tried to do is we looked again on the right side
of the vehicle and recovered the largest intact fragment that
had been recovered from the right side in the vicinity of
the area of interest on the right side which was -- again,
I think this was a top piece in Panel 10. It's a piece of
T-seal that's approximately 33, 34 inches long. But I will
point out to you that it did not have its attachment flange,
which is this part right here on this particular scrap that
we had, nor did it have very much of the apex, as you see,
a kind of C-shaped devices. So what I tried to show on this
chart here is this actual green area is the approximate acreage
of that part that was recovered and we believe through analysis
that you're going to have to recover a T-seal that's going
to have to have part of the apex or part of this flange area
in order to bring the RCS closer to a match.
If you just take a look at this particular T-seal, what you'll
find is that the circular polarization value looks a little
bit low. In another orientation, it turns out that one of
the polarizations is well within the limits and one is under.
This is again the classic issue of the fact that when you're
creating circular polarization from two linear datas, both
polarizations have to be high; and in order for this part
to be more reactive to the circular polarization, it has to
have some curvature. So we feel very confident that this particular
item, even though this graph is a little bit low, that we
cannot eliminate as a class a T-seal half that includes the
attachment flange or part of the apex in terms of radar signature.
Next slide, please.
GEN. BARRY: And that can mean the top part and the
DR. KENT: Yes. It could be the top section, or it could
be the bottom section.
ADM. GEHMAN: Could you go back one, Doctor? The chart
there on the left-hand side, the on-orbit radar cross-section.
That looks to a layman like that's a pretty good match.
DR. KENT: Well, you see, remember, the on-orbit minimum
to maximum falls in here; and the point is that we know we
observed values that are close to the top of the box. So what
we're looking for are what I'll call these blue lines that
are very close to the top of the box at some point in its
aspect orientation. As a matter of fact, if you look at the
carrier panel, for instance, you'll find that it is consistently
about minus 5 at its most advantageous orientation; and the
problem with that is we know that the carrier panel's it.
We've measured the whole thing. There's no more to add, so
it can't get any larger. In the case of this fragmented T-seal,
we know that there are pieces of it that we would have liked
to have had but we didn't have.
ADM. GEHMAN: So the fact that your results for any
azimuth fall completely inside the box is interesting but
you need more reflectivity.
DR. KENT: Yes. It's most important that it crests the
top of the box, touches or exceeds the top of the box; and
you don't want it to exceed the top of the box but just a
tiny aspect angle because then you get into the whole question
of whether you'll ever present that favorable orientation.
It turns out that T-seals have a particularly nice property
because in this plane where it has the T, it has a very, very
broad radar pattern in this plane, which means orientation
is -- it's very insensitive to orientation if that part of
the T is intact.
Next slide. Now, here are the ballistic coefficients for what
I'll call more interesting components. This is the RCC and
carrier panel components. Now, this is a different scale than
the one I had before. The other one went up to 1.2; and the
maximum on this one only goes up to .3. So we're really blowing
this up. Here again is the observed Flight Day 2 value. You
notice the uncertainty bars look larger, but that's only because
of the change in scale.
I'm showing you a couple of things. First of all, what I wanted
to show here is initially when we were looking at carrier
panels, before those were no longer an RCS candidate, an intact
carrier panel didn't make it anyway and you had to actually
explain away one of the tiles or add in the collar in order
for it to behave appropriately. The ear muff seal, I think
it's called the spanner insulation piece that Steve showed
earlier, fits well within the ballistics. The interesting
thing is we had an analysis run for this particular briefing
on one of these pieces, which is about 100 square inches,
and it fits right where it needs to be. Now, since I produced
this chart, I got an E-mail from Mr. Morris yesterday. He
ran the ballistics on all four of the scraps that we did;
and all four of the scraps met the beta term criteria, well
within the experimental limits.
ADM. GEHMAN: All four of the scraps of what?
DR. KENT: Of RCC. If we could go back a slide, please.
ADM. GEHMAN: RCC pieces.
MR. TETRAULT: Did all of those RCC pieces include a
DR. KENT: They didn't include a web but they were --
MR. TETRAULT: A web. An angle. So that it had a rib.
DR. KENT: No, actually this one did not.
MR. TETRAULT: You had one with just plain acreage,
and it passed the test.
DR. KENT: Right. It's not quite flat. We had one that
was attached as an edge. I believe that's the one here, No.
37736. It's got an edge. There were two others, as well, we
reported to the board. Basically it turns out -- again, remember,
the radar wave length is this big and these lips are only
a small fraction of this wave length. It helps to have it,
but it's not a crisis to have it. The important matter is
the acreage or the size of the piece.
Go forward two charts, please. So basically in this particular
chart what I did, of course, is these had failed the RCS and
so far that the T-seal, which I would like to point out initially
one looks at this thing in either its tumble or its spin axis
and it's not hitting the mark but, of course, you could have
any state between those two and because they bound the observed
value, both the T-seal or half T-seal still fall within the
Next chart, please. So keeping in mind that the flight day
object must meet the observed physical properties of these
components, I can't stress enough that these are primarily
exclusionary tests. We started with 31 materials. If the items
do not meet one of these two criteria, they cannot be the
Flight Day 2 object. At the end of the day, as you'll see,
the items that meet both the RCS and the ballistic criteria
is this spanner beam insulation, sometimes called the ear
muff -- of course, it's excluded if it's not exposed -- and
I think that's been discussed in the past -- a whole T-seal;
a T-seal fragment that includes an attachment flange that's
this part, this end of it or the apex, kind of the middle
of the C; or an RCC panel acreage. 90 square inches is the
minimum if you're worried about it just having enough radar
signature; but if you want to have a little bit of leeway
to account for the fact that you don't have all the control
of the orientation, probably on the order of 130 to 140 square
inches of RCC would also agree with this object. It needs
to be roughly square, within about 20 percent. Otherwise one
of the dimensions has to get a little bit bigger. Again, that
does not hurt the area-to-mass or the ballistics. And the
curvature, again, is okay because, remember, the wave length
is large compared to local curvature of these pieces.
I would point out that we have been asked by the CAIB to screen
an upper carrier panel; and because that's coming out of flight
spares and it's taking some time to arrange, that item has
not been done yet.
Next slide. Steve.
MR. RICKMAN: Okay. Let me just do a quick wrap-up here.
What we tried to do is roll up everything into a one-page
summary chart that you can take a look at. What I would offer
up is looking at the right-most column, and what we did is
we came to our conclusions on these. The green represents
items that we feel are excluded -- again, noting that the
ear muff is excluded if it's not exposed; otherwise it does
meet the criteria.
From all of the testing and analysis that we've done, we feel
that RCC T-seals as a class cannot be excluded and RCC, what
we call acreage or pieces of the panel, cannot be excluded;
but there's another point to be made there that the panel
acreage itself would have to be on the order of .33 inches
thick for it to have the correct ballistics. Just so you know
the area for a constant thickness piece item, the area-to-mass
ratio will scale up. So if it meets the area criteria that
Dr. Kent discussed and it meets the thickness criteria, then,
again, as a class, you cannot exclude it. It turns out that
on the lower panel acreage in the Panel 8 to 9 region, you
do have RCC panel acreage that is of this thickness; and it
varies elsewhere. That's pretty much all we need to say on
that particular chart.
ADM. GEHMAN: Thank you very much.
Board members, any questions for these real smart gentlemen?
GEN. BARRY: You're going to have the final panel testing
MR. RICKMAN: Are you referring to the upper panel,
sir? We need to get the paperwork going to get that out of
the flight inventory, and we'll be starting to work that ASAP.
MR. TETRAULT: Let me go back to the 8 or 9 area and
whether or not it has that .33 requirement. Is the .33 in
that area only on the spar rib, or is it on the acreage itself?
MR. RICKMAN: Sir, it's on the acreage. I did verify
ADM. GEHMAN: You said that in the case of a candidate
that was just flat acreage, RCC acreage, you need something
that's between 90 square inches and 120 square inches, which
is roughly the size of a piece of paper or a little bit larger.
DR. KENT: Right. It could be larger than that, of course,
for the orientation; but if it gets much smaller than that,
then that peak signature doesn't come anywhere close to the
top of that box that I drew around all those charts.
ADM. GEHMAN: Very good. Board members, anything else?
Gentlemen, you've kind of briefed us there on how much work
was involved in this; and we really appreciate it. This object
orbiting with the Columbia is a great mystery and we don't
know if it's related or not, but we had to move heaven and
earth to describe what either it is or it is not because it
fits into this pattern of circumstantial evidence. It's very
difficult to prove the negative, but your help has been instrumental
in us characterizing what we have here. We think we made great
strides in clarifying what we've got up there even though,
as you have said at least five times, we can't prove anything.
So on behalf of the board, for both yourselves and also the
teams that you represent, please accept our thanks. You are
excused and --
DR. WIDNALL: I did have a question. Sorry.
ADM. GEHMAN: Hold it.
DR. WIDNALL: My favorite question. Why do things tumble?
ADM. GEHMAN: In space.
DR. WIDNALL: Why does the frequency of tumble increase
for this object? Is that correlated with coming down into
slightly denser regions in the atmosphere? What's going on?
MR. RICKMAN: I think it could be a number of reasons.
I think if you have an irregularly shaped object and you have
the center of aerodynamic pressure at a location different
than the center of mass, then as you get lower and lower,
you're going to have increasing aerodynamic forces on there
that would tend to get the object to spin up.
DR. KENT: And if you take a look at, for instance,
even the samples, the pieces of acreage that we've tested,
they're highly irregular pieces. You, know, one side will
have a lip; one side won't. So we have no idea if it were
something like that. The chance of a nice, symmetric, clean,
square shape coming out are quite low; and it's probably going
to have some kind of differential pressure on it.
ADM. GEHMAN: Thank you very much. We're going to stay
here. You all are excused.
We'd like Mr. Whittle and Mr. Hill to please come out and
take their seats; and we'll get moving on this right away.
(Next witnesses seated)
ADM. GEHMAN: All right. Thank you, gentlemen.
Our third panel is ready. They consist of Mr. Dave Whittle.
Mr. Whittle was and has been the director of the mishap investigation
team since Day 1. He's been in charge of picking up the debris
and the recovery efforts, all recovery efforts and all coordination
efforts with all the agencies that were helping with this
investigation. Mr. Paul Hill is a flight director and has
been responsible for the sighting studies and videography.
So we're now going learn what we can learn about debris, where
it's found, and what we can determine from debris analysis.
So, gentlemen, before we get started, I would like for you
to affirm that you're going to tell us the truth. I'll read
a statement to you and ask you to affirm that you agree to
this. Let me ask you to affirm that the information you will
provide the board today will be accurate and complete, to
the best of your current knowledge and belief.
THE WITNESSES: Yes, sir. I will.
ADM. GEHMAN: Would you please introduce yourself and
say a little bit about your background and what your day job
is, and then we'll listen to your presentations.
DAVID WHITTLE and PAUL HILL testified as follows:
MR. WHITTLE: I'm David Whittle. I work for NASA in
the shuttle program office. I have an electrical engineering
degree from the University of Texas at Arlington and an MBA
degree from the University of Houston at Clear Lake. I have
accident investigation training from the NTSB school, from
the NASA school, and a Certificate of Air Safety from the
University of Southern California School of Aviation Safety.
MR. HILL: My name is Paul Hill. I'm ordinarily a space
shuttle and a space station flight director. For the last
few months, I've been leading a team looking at primarily
early sightings and videos.
ADM. GEHMAN: Good. All right. We're running considerably
late, but we would like to ask you if you would like to make
a presentation or an opening statement. If it's all right
with you, we'll kind of ask our questions as they go along.
Whichever one of you is ready, go ahead.
MR. WHITTLE: I'm ready. On February the 1st, I stepped
off the airplane at Barksdale Air Force Base to start the
first part of this search, what has turned out to be the largest
search of this nature in the United States, in the history
of the U.S., perhaps the world. In the process of this, we've
involved over 30,000 people from virtually every state in
the United States. We've involved over 130 Federal, state,
and local agencies in various roles, from major to not so
major. It started off with thousands of volunteers from the
people of East Texas. My E-mail every day for the first few
weeks was full of people writing me, wanting to help, wanting
to assist. We got a lot of phone calls. So we had a lot of
people from all over, wanting to help.
Early on, what we were trying to understand is the distribution
and the magnitude of where the debris was. As you well know,
when you visited me at Barksdale, we were literally putting
pins in maps to help us understand how the debris was distributed
and where we should be applying our efforts. As time went
on, we got a lot more scientific than that.
We had reports from a great majority of the states in the
union. We also had one report from Jamaica and one from Bermuda,
of people reporting what they thought were shuttle debris.
In many of those cases, they were not debris; but people were
seeing all of the publicity and wanting to do their part.
As the magnitude and the position of debris became more and
more evident, we developed a methodology and a technique that
we felt would allow us to return the great majority of debris.
The major players in the retrieval and in doing that was NASA,
both the U.S. and the Texas Forest Service, FEMA, and EPA.
They did the lion's share of the debris retrieval.
We closed our Texas search on April the 0th. At the end of
that time, we have physically on the ground covered, with
people walking, over 700,000 acres. We have searched over
1.6 million acres with our air assets, which primarily was
helicopters. We've mapped 23 miles of the bottom of Lake Toledo
Bend and Lake Nacogdoches. The U.S. Navy Supervisor of Salvage
was a major player in our underwater operations, and they
dove on over 3100 targets in Toledo Bend and over 326 targets
in Lake Nacogdoches. The days that I was out there, the water
temperature was 47 degrees. The visibility under water was
about inches. As of April the 30th, we have about percent,
a little over 84,000 pounds, over 82,000 pieces, and that's
continuing to change today in that we're still getting calls
As much as I would like to find something west of the state
of Texas, right now our westernmost piece, as you know, has
been the single tile that was reported by a farmer in Littlefield,
Texas. That does not mean that we don't think there is something
out west. In fact, we have been working and still continue
to work in that area.
Analysis from radar, from video, from trajectory resulted
in nine what I tend to call NTSB boxes, but nine boxes that
were identified where there was a high potential of having
something in that area. Sometimes these boxes were large,
and sometimes they were small. Four of those boxes were in
Texas; and with the end of the search on April the 30th, we
have completed those boxes. As a matter of fact, the last
box and the box that I really personally felt the most confidence
in was in Granbury, Texas.
Before they left, the Forest Service sent 00 people out there
to search that box. We sent 800 people out there for about
two days, searching what I thought was a very high-probability
box. And it wasn't just me. A lot of people did. We did find
one tile, but we really felt like there was perhaps some metal
in there. There may still be, but we searched it very good.
So that completed our Texas searches. The other boxes have
been searched in other ways at an earlier time.
That did leave five boxes that were to the west, and those
boxes are in New Mexico, Utah, and Nevada. We have finished
searching the New Mexico boxes a few days ago; and, in fact,
they found about four or five items. It's to be determined
whether or not they're shuttle. They've been sent to Kennedy
for analysis. There is an Air Force base around there, and
there's a very high possibility that aircraft type material
could be in that area. So we need to sort out is it shuttle
or is it not.
We are still working in the boxes that's in Utah and in Nevada,
and I expect before the end of the month that that will be
complete. We're ground-searching those things. Weather has
been a major factor in that we've been kept out of those because
of snow and other conditions.
We didn't really give up on the West Coast even. We did that
one time even. We had an effort to walk along the coast of
California, knowing that there's a possibility that things
might wash up on the beach. In fact, that showed no results;
but we feel like that there are groups who walk the beaches
routinely that were briefed about what might wash up and something
may show up in the future.
In doing all this, I've used a U-2, a DC-3, forest penetration
radar, hired parachutes, 37-plus helicopters, 10-plus fixed-wing
aircraft, imagery from two different satellites, more than
one type of hyperspectral scanner, forward-looking infrared
radar, the Civil Air Patrol. And, yes, the rumor's true, I
even tried to use a blimp.
The one tragedy that came out of this is that we did lose
a helicopter that two people died in. One of them was a U.S.
Forest Service person. The other was a helicopter pilot from
the Grand Canyon area. Other than that, the safety record
in injuries and to the 5,000-plus people that we had in the
field every day was remarkable.
As of April the 28th, we opened the Columbia Recovery Office,
and that's located across the street here in the emergency
operations center in the control center. We ran parallel for
two days with the operation in Lufkin to make sure there was
no hiccoughs, no disconnects. In fact, that place is up and
operating and we are receiving calls, anywhere from 10 and
16 a day. Our intention is to respond to all of those.
We have a contract with the same people who are picking up
and cataloging and logging the debris for the normal search.
When necessary, we'll send those people out, even if it takes
decontamination. We have the skills. We have a storage area
that we have at the NASA Bloom Base in Palestine. So if things
are large enough that they can't be FedExed, we will take
them up there and store them and then get them down to Kennedy
at the appropriate time.
General feeling is that we're going to see a great, big peak
around November, when hunting season happens. We've done an
awful lot to educate the hunters and we've provided packages
for when they get their licenses, where they give some numbers
to notify us if they run across things. Unfortunately, there
are a number of potentially hazardous items still out in the
woods someplace. Those are primarily pyrotechnics and there's
a couple of fuel tanks that probably have been open and probably
are safe, but you don't know.
All of the local emergency response agencies, all of the county
judges, all of the people that would be affected by that have
been notified. We passed out circulars. We passed out fliers,
pictures, information. So, hopefully, no one will get injured;
and if they find it, they know who to call and how to get
it back in.
At some point in time, the Columbia Recovery Office here will
close. The phone will not go away. We have a toll-free number
that can be called, and the phone will not go away. It will
be answered by Kennedy Space Center. That will continue for
a long period of time. As people find things, they can call
it in. In fact, I think you can still call a number for Challenger.
So that will continue on, and we will close the CRO here.
The number of people. Like I said, there's over 130 Federal
agencies. The number of people to thank is endless, and I've
named a few of those agencies already. Interestingly enough,
there's been a great deal of interest in our operation from
other areas, in that, with the heightened awareness of terrorist
threats and things like that and Department of Homeland Security,
the size and magnitude of this operation has been one that
has piqued interest and that they have deemed might be a model
for following in the event of a similar type response. So
we've had a lot of people come down and talk to us and see,
try to understand how we did this, how we put it together,
and how it worked so well.
ADM. GEHMAN: Thank you very much. Any questions?
I'll ask one, Mr. Whittle. I am interested in this last point
you brought up, in the sense that, from our visits to you
and also from what I understand from reading reports, that
the level of local, state, and Federal cooperation was remarkable,
maybe unprecedented in a large operation where you have lots
and lots of people. And you didn't mention how much this cost
either. So there was a considerable amount of money involved
in this. My understanding is -- and I think most people agree
-- that the level of local, state, and Federal cooperation
has not been exceeded in other major instances in this country.
Do you have any idea at all as to what to attribute that?
MR. WHITTLE: I get asked that a lot. I think that there
was a single-mindedness. Everybody felt ownership, and there
was a single purpose. You know, it almost became a family.
From the people out in the fields, the U.S. Forest Service
folks that were 12 hours a day out there, marching through
the fields, sleeping in tents at night, they were all really
dedicated to this and proud to be there. That was kind of
the attitude for everybody.
ADM. GEHMAN: And the cost? I could ask FEMA, I guess.
Really FEMA paid.
MR. WHITTLE: FEMA paid a great deal of that, and the
costs are going to be in the 300 million-dollar ballpark.
They said I was really good at spending money.
ADM. GEHMAN: You did a great job, and I'll just make
a comment here for the board that we have authorized the expenditure
of a few dollars to create an official memento that we intend
to give to all those people, a piece of paper, a parchment
with a nice certificate in which we recognize all those organizations
and then some kind of small coin or medallion that we can
give to those people that we would like to recognize all the
people that took part in that. I happen to know that you have
an accurate list of who it was that you want to recognize.
MR. WHITTLE: Yes, I do.
ADM. GEHMAN: Thank you very much. The boards wants
to recognize that work, and we will do that. Thank you very
MR. HILL: I had a few charts that I brought. Mostly
pictures to give you an idea of where we ended up with the
various facets of analysis. On the next page I summarize more
or less everything that we did on the team. I don't intend
to go into a lot of detail. I can say at a high level we took
the public reports, we took the video, we analyzed the video
to try to come up with trajectories for the debris we see
coming off, build footprints. We use those footprints to then
go search radar data bases with the NTSB to find signs of
that debris falling down through the radar. We arranged the
AFRL radar testing, some of which you heard about just a little
while ago, for both the Flight Day 2 object and to give us
some sense of truth on whether or not we could, in fact, track
the most likely debris in the air traffic control radar or
the C bands that we use for ascent.
We also have been talking some about luminosity and spectral
analysis, and I'll talk about a little bit of that here in
a few minutes. And we went through various other sensor data
both with the DOD and with NOAA and the USGS. I can summarize
all of that to say that outside of telemetry we have from
the vehicle, the OEX data, and the public video, we really
have no external data that adds any engineering value yet
to the investigation.
We have some ongoing work. If you go to the next page, on
that last piece let me just mention on the bottom bullet we
have not yet run the tests at Ames to try to use luminosity
to estimate mass and drag of the objects that we see in video.
We have a good test plan; and we're in the final throes now
of deciding if we, in fact, are going to manufacture those
test samples and conduct those tests. We pretty much have
dropped the spectral piece of the analysis just because confidence
is so low that we would get meaningful data.
Everything else you see on here is the open work. It really
is just final cleanup work. We have a handful of videos still
to process through to calculate relative motion and trajectory
for the individual pieces of debris. We've gone through all
the radar data base that coincides with our generic debris
footprint from California all the way to Texas. We have a
few backup passes we want to make through that radar data
base, and we have some final analysis to do with the radar
test data that we already have in house. I'll describe what
some of that is here in a few minutes.
The next three pages are debris time lines. You've seen iterations
of these, and I think you have this copy. This is the latest
and greatest copy from April, and I admit it's difficult to
read here in the resolution that I brought.
The big-picture story is, as you've already seen, we know
we were dropping debris from California to Texas. Chances
are we were dropping debris in areas that do not show up as
white dots on this trajectory. These are the ones that we
had best angles, best lighting, and we were fortunate to catch
in video. Our expectation is if we had more videos from different
angles, we would probably have more white dots on here.
ADM. GEHMAN: The white dots represent the position
of the orbiter when the debris came off; they don't represent
MR. HILL: That's correct. That's the point in time
when we clearly see a distinct piece of debris coming off
the vehicle or a couple of indications of flares which you
see out here over eastern New Mexico. There's also a flash
there over early Nevada and there's a debris shower. So we
have 20 distinct pieces of debris we capture in video plus
this thing we call a shower, which looks like some large piece
that then splinters into many pieces and then the two flares.
The next two pages just show you the same information with
where the people were standing that took the video and their
field of view. Most notable, we added one, way to the south
in San Diego, which in spite of the range they were at and
the 5-degree elevation on the horizon that the video captured
the orbiter, they, in fact, capture the flash and Debris 6
in their video.
On the next page, it shows you the rest of the trajectory
to Texas. You can see the about minute-and-a-half-or-so video
gap we have from eastern New Mexico to across Texas. I guess
the other thing I would point out is -- and I think you have
heard this before -- while we appear to have relatively continuous
coverage from that point over east New Mexico all the way
back to California, there are places in the video where the
tracking was not good or that the angle was not good and we
actually can't see the orbiter at all times. But it's pretty
On the next page. This is an early generic footprint that
we generated from the East Coast all the way to Texas. This
is based on some top-level assumptions on where tile would
fly if we were to be shedding tile all the way from California
to Texas. That area in the middle would be the non-lifting
box, which would be our highest-probability area where we
would expect to be finding debris as we drop across the CONUS.
On the next page, this is the latest and greatest set of footprints
we have for relative motion that we have, in fact, measured
off of all the debris. There are a handful still of individual
pieces of debris that don't show up here with specific footprints.
We have those videos in work, but this already gives you an
indication that we have near-continuous footprints, even based
on really good trajectory analysis. So from California almost
all the way to Texas, we have almost continuous overlap, which
clearly makes your chore of going out and searching out west
a large one. If each one of these large rectangles represents,
say, a single tile, looking for a tile in an area like that
is a huge task.
Again, that thin, dark area in the middle, that would be that
non-lifting area. That is our highest confidence area where
we would expect to find the debris.
ADM. GEHMAN: You're talking about these little lines
MR. HILL: Yes, sir. If for whatever reason the debris
was to take on some amount of asymmetric lift -- if, for example,
it was to drop as a flat plate and not be tumbling -- it could
venture off into the wider part of the rectangle.
On the next page, this is an old overlap map. We have an updated
one that we're doing some work on to refine, but just to give
you an idea how we tried to sharpen the pencil a little bit
to come up with better areas to search out west rather than
that large swath, we took the areas where all the highest-probability
boxes overlap and you see those as the darker regions on this
map. So those would be the places that, based on ballistics
and trajectory analysis, would give you the highest probability
to find something if we were to put people on the ground to
search. You know, for comparisons, that first one you see
there over the Nevada-Utah border, that's about a 300-square-mile
box. It's still very large if you're looking for, say, a single
piece of tile.
I guess I'll also point out that I keep mentioning a single
tile. We don't necessarily know these are tiles. Our expectation
is what we see coming off is something small.
Last thing I'll say on this picture. If you look over Texas,
you see a very faint overlap area, just kind of a light gray;
and towards the end of that light gray box is where Littlefield,
Texas, is. That's where that Littlefield tile was found. Our
analysis shows that if that tile came off in that size, then
it would have been shed somewhere in the Flare 1, Flare 2
area over eastern New Mexico.
Next page. Now, going back to Dr. Kent's radar tests, what
this shows you is for the radar data that we have finished
the analysis for. All of these circles show what the detection
ranges are for each one of those radars. The large black circle
would be the range of the radar in and of itself. The smaller
dotted lines would be tuned to specific materials. The thing
to note is the green circle out to the red circle, the relatively
larger circles, those are all the leading edge components.
The little light blue circle in the middle, that would be
individual tiles or tile material. So the thing you would
conclude from this, of course, is very low probability, at
best, of us being able to detect tiles falling through any
of these footprints.
You can see the ballistic footprint above these radars. Now,
there are other radars that you see up here in red X's that
we have not mapped. The analysis is still in work. I expect
to have that in the next week or two. My expectation when
we finish is there are only going to be a few cases where
we have a possibility of detecting tile anywhere over the
ballistic footprint, which was not happy news for us because
it does give us less confidence that the radar threads that
we're finding in many cases really could be tiles. They could
still be some other leading edge type of component; but as
you can see, it would have to be something relatively large.
On the next page I have a couple of different footprints.
The thing I would like to point out is in the lower right
you see the large black cross. I sent some folks back within
the last few weeks to look through the thousands of reports
that we have from witnesses that just saw something in the
sky. These are reports that have gotten a lot less attention
from us once we saw the video and we found we could calculate
engineering data from the video.
We went back through all the reports and we tried to pull
out the reports from people that saw things that could have
been anywhere in any of our actual footprints. Of those, this
one report was the one that stands out as the only one that's
significant. This fellow was in a camping site 70 miles north
of Las Vegas, saw the orbiter fly overhead. Ten minutes later,
looking due east, he saw something bright falling out of the
sky, between him and a peak that was in front of him. This
is where he was standing, overlaid on top of the Debris 1
footprint, a relatively old Debris 1 footprint. On the next
page, similarly on a Debris 6 footprint. You see our high-probability
box just to the east of where he was standing.
If you go to the next page, this is a close-up of one of those
overlap footprints. That small green rectangle you see just
east of where he was standing in Delamar Lake is Radar Search
Box 8. We've already had NASA folks on the ground out there
that put where he saw this object within a mile of our last
radar return in Search Box 8. I haven't heard the results,
but it was my understanding that by mid-week last week we
had people on the ground, actively searching that area for
MR. WHITTLE: We did, yes.
ADM. GEHMAN: Dave, you want to comment on that?
MR. WHITTLE: Yes, we do have people out there; and
that box may be finished today. As of yet, we haven't found
ADM. GEHMAN: Thank you.
MR. HILL: On the next page. I'm not going to read all
these. What I'll tell you, though, is the radar search boxes
or the NTSB search boxes that Dave mentioned, those are listed
in this table and the next page. All of those overlap areas
you saw on the overlap map, they all show up here. The Delamar
Lake sighting shows up on here. What we have done is these
two pages summarize the 21 search areas that we have out west,
and that's a combination of our radar search boxes, witness
sightings, or trajectory footprints. They're in priority order,
based on how good the data is, say, from radar, how close
the radar thread or the witness sighting is to our high-probability
areas et cetera. The only other thing I would point out is
you can see you don't have to go very low on this list and
the areas you are talking about searching are enormous. The
one that I have highest confidence in from a ballistics perspective
would be that Priority No. 7, which I already mentioned is
300 square miles. The next one after that is 1200 square miles.
I have absolute certainty that our trajectory analysis is
good and that the objects we see coming off in video are,
in fact, in these areas; but as Dave and I were talking about
a little while ago, sending people out to a 300-square-mile
area or a 1,200-square-mile area to look for something that
could be a tile is a tough job.
ADM. GEHMAN: All right.
MR. HILL: Skipping on to page 16. I'm not going to
go into a lot of detail. I'd just like to explain this is
the evolution of our generic footprints over Texas. So this
would be our post-breakup debris footprint. Within an hour
or two of the accident, the February 1st release was published;
and that really was just a dark line that essentially was
under the ground track. That was a really simplified analysis
just to give us a place to start. Within three days that was
expanded with Monte Carlo sims to that gray rectangle, giving
you a larger footprint. By February 7th we had a better time
on the estimated breakup. That moved that gray box up to the
right, which gives you then that purple rectangle. That's
a function of we continued the left roll, so we continued
to get a little bit more lift. That moved then your debris
After two months of detailed analysis and adding in real weather
and much more sophisticated Monte Carlo simulations, we ended
up with that yellow feather-shaped footprint that you see
there or the orange feather-shaped footprint. The yellow one
is based on a breakup time or an end of lifting of 13:59:37,
and then 25 seconds later we ran another case for lifting
that continued and that gives you that second orange footprint.
You go on to the next page. This just shows you where those
areas are over Texas and Louisiana.
On the next page, interestingly, the NASA 220 center line,
this is the line that Dave Whittle and company used to search
in East Texas and Louisiana. That center line was based predominantly
on their observations of where debris was being found, and
it matches up very closely to the center line for the orange
footprint. You can see in the upper right it's only about
a mile off at the end from the center line of our 1400 footprint,
and also the difference in the center lines between the yellow
and the orange footprint is about 4 miles on the east end
and about to 2 miles on the west.
On the next page, this just gives you an idea of where the
significant items were. This isn't everything found; this
is just from the significant items list. You can see how they're
distributed relative to the footprints. You can also see up
in the upper right where the SSME power heads were found,
right on the center line of that orange footprint.
Then my last two charts. This is a combination of all the
radar hits in the NTSB data base from 13:59 to 14:10. You
can notice the high concentration of those radar returns right
in the middle of the footprint. A lot of the rest of what
you're seeing is just standard noise.
If you go to the next page, this is a combination of the data
from 14:30 to 14:40. You can read this essentially as background
noise or clutter that you would typically see in this view.
If you go back one page again. Again you can see the high
concentration, which gives us good confidence that we've definitely
broken the code on how to generate these types of footprints.
I guess the last thing I would say is, were we to have to
go through this exercise again, we have done enough work now
that we could generate these footprints at this same level
of accuracy within about two hours of the accident.
That's everything I have.
ADM. GEHMAN: Board members?
Mr. Hill, what do you think is remaining for your working
group to do?
MR. HILL: Primarily processing the last handful of
videos to calculate relative motion and good footprints on
the remaining western debris and then summarize everything
that we've done.
DR. WIDNALL: I'll ask my favorite question. What drag
coefficient did you use?
MR. HILL: Drag coefficient. You know, I'm not positive.
We used an L over D of zero to .15.
DR. WIDNALL: I saw that.
MR. HILL: And we actually measured the ballistic number
from relative motion. So we didn't have to pick a drag coefficient.
DR. WIDNALL: Then in order to generate the footprint,
you would have to -- I mean, if you were trying to estimate
where the thing landed.
MR. HILL: Even with the footprint, we based that on
ballistic numbers, independent of individual CDs of objects.
GEN. BARRY: Paul, have you given up on the Caliente,
MR. HILL: I'll speak for myself. Personally, where
the Caliente, Nevada, radar search boxes appear in our ballistic
footprint gives me lower confidence that it's something that
belongs to us, just because it's so far off our non-lifting
box. So my confidence is not high that that is something that
belongs to the orbiter. I think it's good radar data; I just
don't think it belongs to us necessarily.
DR. WIDNALL: I was intrigued basically by Greg Byrne's
image analysis. Are you planning to use image analysis to
try to estimate? I mean, if you actually had a ballistic coefficient
of a piece of debris, based on, you know, you might be able
to say that's a tile or that's a part of an RCC, because they're
MR. HILL: Well, what we have done is we've used the
ballistic coefficients that we've measured to sort of bound
which objects fall in the category of the ballistic numbers
we're seeing in video. So typically the ballistic numbers
we're measuring in relative motion range from about .5 to
on the order of about 5 pounds per square feet, which, in
fact, exactly brackets the full range of intact tiles. There
are pieces of other internal components, leading edge components
that, if you were to break them down small enough, would also
fit in that category. I guess another conclusion you could
reach is because those are the ballistic coefficients we're
measuring, we don't think we're seeing anything large coming
off in video. I don't know if that answers your question.
DR. WIDNALL: Well, I guess my own view is that probably
many of those debris are tiles. I mean, I literally cannot
imagine 14 or 20 pieces coming off the shuttle without the
thing just melting. So I guess I have to believe a lot of
them were tiles and I would assume that you could identify
that from the trajectory, that these would decelerate much
faster than structural elements.
MR. HILL: We can definitely show that the ballistic
behavior we see of those objects is consistent with an intact
tile or a tile fragment. It doesn't tell us for sure that
it is, but it is consistent.
ADM. GEHMAN: All right. Well, thank you very much.
Mr. Hill and Mr. Whittle, both of you represent the top of
an iceberg of a lot of people -- particularly Mr. Whittle,
who's got 30,000 people working for him on one day or another.
Also, Mr. Hill, your group has done a lot of work to help
us understand what happened; and we're very grateful. We're
grateful to not only you two but also all the people that
you represent. We'd like you to pass that on to everybody.
You've done a great job, and we thank you for your candor
and your willingness to discuss these things with us here
at this hearing.
This hearing is closed, and we'll be having a press conference
right here in this room in 34 minutes.
Thank you very much.
(Hearing concluded at 12:24 p.m.)
Back to Event