|Columbia Accident Investigation Board Public Hearing
Tuesday, March 18, 2003
Hilton Houston - Clear Lake
3000 NASA Road One
Board Members Present:
Admiral Hal Gehman
Major General John Barry
Brigadier General Duane Deal
Mr. Roger E. Tetrault
Dr. Sheila Widnall
Dr. James N. Hallock
Mr. Steven Wallace
Mr. Steven Labbe
Mr. Jose Caram
Dr. John Bertin
ADM. GEHMAN: Good morning. We'll go
ahead and get started. This morning we're going to
talk about aerodynamic and thermodynamic events that
took place when the Columbia reentered the atmosphere.
We have two panels this morning. The first panel
consists of the NASA engineers and scientists who are
trying to find out what happened to the Columbia; and
then the second panel is an outside expert, as we
This morning we have Mr. Stephen Labbe,
the chief of the Applied Aeroscience and Computational
Fluid Dynamics Branch of NASA; Christopher Madden, the
deputy chief of the Thermal Design Branch of NASA; and
Joe Caram, an aerospace engineer in the Aeroscience and
Flight Mechanics Division of NASA.
STEVE LABBE, JOE CARAM, and CHRIS MADDEN
testified as follows:
ADM. GEHMAN: Gentlemen, thank you very
much for helping us through this. Before we begin, we
don't swear people in; but I will read you an oath of
affirmation and ask you to state that you will give
information that's complete and correct, to the best of
your knowledge. So before we begin, let me ask you to
affirm that the information you provide the board today will be accurate and complete to the best of your
current knowledge and belief.
THE WITNESSES: Yes.
ADM. GEHMAN: All right. Would you, please in order, please introduce yourselves, tell us a little bit about your background and your current job and not only your full-time job but your role in the MRT.
MR. LABBE: My name is Steve Labbe. I'm the branch chief for the Applied Aeroscience and Computational Fluid Dynamics Branch here at Johnson Space Center. I've been with NASA since about 1981. Prior to February 1st, our branch was not really heavily involved in the shuttle program because it was primarily it's an operational system. We were working on the future. Since February 1st, we have been heavily involved in the investigation and supporting the efforts with a team that crosses the agency and the country.
ADM. GEHMAN: Thank you.
MR. CARAM: My name is Joe Caram. I work
in the Aeroscience and Flight Mechanics Division. For
the last six years, I've been the chief engineer for
the X-38 project for my division. So prior to
February 1st, that's what I was doing. Prior to that, I was in Steve's branch, working in the area of
aerothermodynamics, where I focused on the shock-shock
interaction region of the wing and boundary layer
MR. MADDEN: I'm Chris Madden. I'm
deputy branch chief of the Thermal Design Branch in the
Johnson Space Center. My background includes
thermoanalysis of TPS systems for reentry spacecraft.
Some of that's included analysis of shuttle flight
anomalies and other computation roles on the shuttle.
For the mission, the Columbia mission, our branch was
providing consultation for the work done by USA and
reviewed for that.
ADM. GEHMAN: Thank you very much.
Gentlemen, you may start. Who's first? Steve?
MR. LABBE: I'm going to start this
Good morning to everyone. I just wanted
to thank you for the opportunity to come and present
our efforts that have been in support of this. We have
a whole bunch of material. So I suggest we just get
Go to the second chart, please.
What we're going to cover today, I'm
going to give you kind of an introduction and then describe our analysis process, the current approach,
what we're doing. In our approach right now, we're
starting with an assumed initial damage and then trying
to propagate that to reproduce the aero and thermo
response. We're assuming the damage existed. We're
not trying to find necessarily the root cause. Once
our results are then completed, we hope that they will
point towards the root cause; but we start with the
damage. We're also looking at about the first 600
seconds of entry. We're trying to get from what
happened from entry interface to the point where we
believe there's a breach in the wheel well and the
temperatures start rising. So if we can get that
solved, we feel we'll have made a significant
contribution to the investigation.
The reason the three of us are up here
together is it's an integrated approach. We don't
believe that just aerodynamics or aerothermodynamics or
thermo by itself would be a good answer. We need to
all be consistent, and our results have to all work
together. So there's the three of us here, and we're
part of that integrated team.
Next chart, please. This is just a brief
snapshot of the organization, and it's really trying to
give you a picture of the breadth of the scope that we're working. We have support from numerous NASA
centers, the Boeing Company and its different
divisions, Lockheed Martin, Sandia National Labs, and
the Air Force research lab at Wright Pat. So we have
quite a range of expertise, and they are supporting us
in a large variety of areas that we represent.
Next chart. Okay. The approach.
Basically we're trying to, as I said, start with damage
and then take specific actions to investigate how the
scenario that comes up can be used and explain the key
data events. The first poster board there attempts to
illustrate that on the left here, the tall one. I
guess the easiest way is to just talk about it from
What we have plotted along the top and I'll go into much more detail there is the change in aerodynamics that we saw during the mission. It's versus time, and you can see that it's not zero. It's drifting negative and then eventually drifts positive. Down below, it starts here and then drifts so.
What we then wanted to do was find some key events in the instrumentation that corresponded with those changes. So the first thing that we noticed was this first off-scale low temperature I'm sorry, the bit flip in the wheel well was the first thing that we noticed, the change in the temperature. This is the brake line temperature in the main landing gear wheel well, and at this point it started an upward trend that continued. So this was the first point; and we correlated that, tried to correlate that to the aerodynamic events.
The second point is when we see our first
off-scale low temperature. So the first, Point A,
suggests a breach, a first initial breach into the
wing. There must have been an ingestion of hot gas in
order to create that change in the wheel well, and
we're going to get you into the details of why we
believe that. The second one is a burn-through of the
wire bundle that holds all of those instruments, so
that whatever was being ingested had to be able to burn
through that wire bundle.
When we get to the wheel well breach
here, we see a significant rate of change. Instead of
just drifting up, now we see a large increase in the
rate of change. Also that corresponds to a change in
the aerodynamic trend where it was drifting negative
and now is starting to go back positive.
So that's the idea. Line up these key
events and analyze each one of those and more or less
provide what we're calling a piecewise integration of the event as opposed to some time-dependent,
multi-physics solution that would explain it from time
zero through. We would never get through that
ADM. GEHMAN: Pardon me for interrupting.
On the top chart, I presume that's time after EI along
MR. LABBE: That's correct.
ADM. GEHMAN: In seconds?
MR. LABBE: Yes.
ADM. GEHMAN: Hundreds of seconds.
What's the vertical axis? On the top.
MR. LABBE: This is a residual or change in aerodynamic it's a coefficient form, but it's rolling moment. We express that in coefficient terms. I'm going to show you a lot more detail on this, but this is the change. We would expect it to be drifting, bouncing back and forth around zero. Instead, it's biased off to one direction.
ADM. GEHMAN: What is the big fluctuation
right at the beginning up there?
MR. LABBE: Very early in flight, the dynamic pressure is so low that the technique we use here, we're not going to be able to resolve down to this coefficient form. You're essentially accuracy of the data available. The initial spike there that you see is a roll. This is the first bank maneuver.
DR. WIDNALL: You and I have talked about
this before, but have you applied this analysis to
earlier flights and satisfied yourself that what can be
identified as off-nominal is, in fact, accurately
off-nominal? I know we've talked about that before.
MR. LABBE: Yes. We applied the same
tool to STS 109, which was the previous flight of
Columbia; and where we see similarities in these types
of traces, we assume that that is just the accuracy
level of our capability. Where they drift apart,
that's when we start believing that we're seeing
something off nominal.
DR. WIDNALL: Okay. I'm sure you'll
share that data with us.
MR. LABBE: Absolutely.
MR. TETRAULT: You talked about a sensor
bit flip. Would you define a bit flip for us?
MR. MADDEN: The bit flip is just the resolution of the instrumentation. So if the temperature change is 1 1/2 degrees 1 degree, you may not necessarily see that change. So it has to change over about a degree and a half. Then it will register a change in the data system. So that's why you see the step-wise plots. It's not a smooth plot because the resolution of the data isn't that tight. So when we say bit flip, we are just saying a change in temperature of about a degree and a half Fahrenheit.
MR. WALLACE: These rolling moments when
we see later on what I thought to be yaw corrections,
is it a pure rolling moment, or is there a yaw element
MR. LABBE: There's both. There's actually all three axes roll, pitch, and yaw. There are some deltas that we extracted. This is just the roll axes, but I'll be showing you both the yaw and the roll axes.
ADM. GEHMAN: Referring to the top chart
again, the big spike is a roll reversal or something
MR. LABBE: The spike here is a roll
reversal and the technique that we use is not as
accurate during a roll reversal but you get a lot of
rates in the vehicle.
ADM. GEHMAN: You say you're going to go
into that in a little more detail?
MR. LABBE: Yes, sir.
ADM. GEHMAN: Okay. Fine.
MR. LABBE: I think we've pretty much covered what's on this chart, and the next chart is
really just another version of the poster. So I'd like
to move on to Chart 6.
This is just a definition of what we're
defining as these key events, A, B, C, and D. I kind
of alluded to this, but there's a hole damage size,
there's a breach in the wing at what we call
488 seconds. That's when we see that bit flip. So
what can we do? What kind of hole or damage can be
created from entry interface to 488 seconds that could
produce that initial change in the instrumentation?
Then we go on to the next step. Step B
is we burn through that wire in another 42 seconds. So
if we pick a location and we have a burn-through, can
it then also burn through the wire 42 seconds later?
Then we have the breach into the wheel well at 600 seconds where we see the rate of change. Of course, that has to be consistent with the initial breach and the burning through the wire. So you can see how we're trying to piece all of these together. Then finally we see this change in the fuselage wall temperatures; and whatever is producing that, is the damage consistent with that and how we've propagated it to generate that. Aero, thermal, debris everything has to be correlated or we did not prove a specific scenario.
Okay. Chart 7. Just another way of
looking at the same thing. I really just spoke to
this. We're looking at all the data, the flight data,
whether it's debris evidence, flight profiles. We're
more or less handed a failure scenario from the failure
scenario team that's developing those, and then we go
and do our analysis and tests in the aero and thermal
analysis and tests. We produce our results; we get
them back together. Are they consistent? That's the
flow of the whole integrated analysis, and what I'm
going to do now is take you through the aerodynamics
side of that analysis.
If we go to Chart 8, which is also
represented here in the poster board. So this is going
back to the very beginning, February 1st, you know.
What we were trying to do is what happened. We needed
to reconstruct the flight, essentially. We had data.
We had flight data that was telemetried to the ground.
We knew the mass properties. We took that data. We
had some tools that had been developed from various
programs, the X-40 out at Boeing and the X-38 here at
JSC, to get delta aerodynamics from that flight data.
And I'm going to explain to you how we do that.
Out of that tool, we get our change in aero. We put that into the flight control simulation
and then we compare what the flight control simulation
predicts the response of the control surfaces, the
ailerons, the elevons, the body flaps, the jets firing,
to what was actually indicated in flight. And we
iterate around that loop several times to make sure
that we have a good comparison. That was our early
focus, and that probably took two to three weeks for us
to get worked out. We're working on actually working
on a final iteration right now.
We do have a good match, and so we're now
transitioning into what we call the damage assessment
phase. This is, again, more or less saying the same
thing. We have this assumed damage. I'm going to go
build a model, whether it's in the wind tunnel or
computationally, with that damage. I'm going to take
measurements or make predictions or make calculations.
I'm going to look back at what it's producing. Is it
consistent with my change in aerodynamics that I
reconstructed? I'm also going to be looking back to
the integrated team to make sure we're consistent with
each other and the other inputs. And we went down to
the Cape on Friday to look at the recovered debris and
to try to understand that so that when we're looking at
different scenarios, we're also considering what's been found there. Ultimately, if we're successful, we have
this piecewise integration of the change in
Next chart. Okay. How do we reconstruct
the aerodynamics? We have a data base, a very
well-defined data base. The shuttle's been flying for
0 years; and this data base has been established
through wind tunnel testing, flight testing. It's well
defined. We take the flight data, the flight
conditions, the Mach number, the angle of attack, the
mass properties, the control surface settings, where
they are. We feed those into our data book, and it
will predict the nominal aerodynamic coefficients that
we should get out of that configuration and that flight
We also take the flight data in Step 2 and we put it in the equations and motions for aircraft. Out will come from that what was happening in flight. Now, in flight we were what we call trimmed. The vehicle was not yawing or rolling. It was in a steady, controlled flight, even though it was experiencing these moments. So the second part of that equation essentially becomes zero and the delta and when we go into the data book and we were putting in several degrees of aileron or is there some side slip on the vehicle, these would produce a moment. So when we delta those down at the bottom in the third step, we're going to get some change in the aerodynamics that the vehicle was experiencing in order for the flight control system to have commanded these settings on aileron and the other control effecters. So that is the process we use to define our delta aerodynamics.
The next two charts go into the details
of those results. These are some busy charts, but
these really tell the aerodynamic analysts the story of
what was happening. This is a change in yawing moment
coefficient. Just the change in yaw. Yaw is nose left
and right, and it's versus time. We have GMT time on
the bottom and time from entry interface across the
top. What we would expect on a nominal flight is, like
I say, some scatter like so, which would stay near zero
the entire time. What we saw on 107 after we did our
analysis was this change in the yawing moment that
started off drifting very slowly, plateaued, and then
sometime around 80 seconds or so before loss of signal,
it started to increase rapidly. Then just prior to
loss of signal, it increased rather dramatically before
we ran out of essentially any available data, which is
about 5 seconds after the LOS.
DR. WIDNALL: I just want to know whether you feel that that dramatic increase is a valid either
measurement or computation or both.
MR. LABBE: I think so now. When we
first looked at it, we were not sure, but we've gone
back and the team that is recovering the data to
support our analysis has confirmed those measurements
by trying to look at two sources for it. So, yes, I
believe that is really valid.
DR. WIDNALL: Also, with the earlier times. I mean, you mentioned, back one chart, with the earlier times you mentioned, you know, scatter in the data. So would you say from I can't read your T from zero from here. 13:50 something or other. Way back at what would be time equals zero on that graph.
MR. LABBE: It's time actually about 300
seconds from EI. 13:50.
DR. WIDNALL: Is that little drop towards
negative and then that slight negative plateau, is that
a valid indication of off nominal, or would you
consider that part of the noise in your data?
MR. LABBE: I would consider that part of
the noise for this. When I went back and looked at
STS 109, it showed the same signature time frame.
DR. WIDNALL: So in some sense the valid
begins at 13:52.
MR. LABBE: That's correct. 13:52:17, which also happens to be we did not look at the data first, but that happens to be when the brake temperature bit flip is also occurring.
A little bit more. There's several lines
here that represent the Boeing simulation or analysis
technique and the JSC analysis technique, and then the
black line represents the model we gave to the flight
control community for them to use in the simulation. I
won't go through each one of these, but the idea was we
correlated things with time on the time line. Yellow
is off nominal. Green is a nominal event such as
starting of alpha modulation or a roll reversal. Then
this red box, this is a design limit for asymmetries.
We do expect to see some asymmetries in flight and
occasionally we see those, but you can see its level is
here and near the end of flight we are on the order of
five times that level. So something very dramatic
happened at that time which led to the loss of control.
GEN. BARRY: Could you put some of this
in context with dynamic pressure? If I remember right,
at loss of signal it's about 80 pounds per square foot.
Now, that would equate to about 180 miles per hour,
MR. LABBE: At sea level. I think it's a little about 150 miles per hour, somewhere in that neighborhood, though. Yes.
GEN. BARRY: Of course, the air molecules
are so far between. We really do have low dynamic
pressure. Can you give us a context of, you know, if
there's any kind of movement of the orbiter, how much
of a transient force is going to have to be in this
case a roll or a yaw moment to be able to counteract
this? We know the RCS jets are still functioning here.
MR. LABBE: Here we're in about, say,
10 to 20 thousand what we call foot pounds. So you're
pushing with 20,000 pounds a foot away, and that's the
kind of moment. That's just a couple of degrees of
aileron. One jet firing can manage that. Near the end
when we go off in this total value here, that's about
60,000 foot pounds. That requires all four jets,
three or four degrees of aileron, the side slip.
Everything the vehicle had to try to counteract that
moment, it was using. That's what the flight data
shows, and that's what our simulation shows. So that's
a very large moment.
GEN. BARRY: If you were to put this in
context, if you were trying to put your hand outside in
an airplane at 180 miles an hour, you would get some
kind of feel for not only do we have the flight control elements on the orbiter trying to control but you also
have the RCS jets doing the best job they can to try to
hold this in control.
MR. LABBE: That's right. If you hold your arm outside the car, you can feel that trying to pull your arm back. That's the moment is what you're feeling about your shoulder and you're talking maybe, you know, 10 pounds and a couple 20 pounds of moment. 20 foot pounds of moment. Not very much at all. And we're talking about several over a hundred thousand foot pounds of moment.
ADM. GEHMAN: Steve, you've got it marked
right here is the roll reversal. This spike right here
is a normal spike associated with the roll reversal and
the stop of the roll reversal.
MR. LABBE: That's correct.
ADM. GEHMAN: I don't know. I mean, the
magnitude of it may be a little greater than normal,
but a spike normally occurs.
MR. LABBE: Yes. The techniques work
best when you're in trim. When you're actually doing a
maneuver, you're not exactly trimmed; you're producing
rates and roll and yaw. So the technique shows a
residual there. It's really the accuracy of our data
base during a dynamic move versus static trim flight.
ADM. GEHMAN: But this one over here is
not explained by the roll reversal, though.
MR. LABBE: No, it's not, although we believe that is a normal response that has been seen on previous flights post roll reversal where there's either a change in the density in the atmosphere or the vehicle is adjusting. And we have gone back and seen the flight control team specifically has seen that type of signature in other flights.
ADM. GEHMAN: Yesterday we heard that
there's kind of a magic altitude of around 42 miles or
miles which, of course, works out to about
20,000 feet, something like that, at which reentry
vehicles seem to hit a wall. Could you tell me about
what the altitude of the orbiter was at that time?
MR. LABBE: I believe it's around two
hundred ten to twenty thousand. Very close. I could
get you an exact.
ADM. GEHMAN: But we're close. I mean,
we could go look it up.
MR. MADDEN: Right. 210.
MR. LABBE: About 210,000 feet, roughly.
ADM. GEHMAN: You've got debris shedding
down here. This is kind of Debris 1 through 6, as I
read that. Correct?
MR. LABBE: That's correct.
ADM. GEHMAN: Debris 6, as we learned
yesterday, was the first large thing that came off.
Then Debris 14, have you got that marked here?
MR. LABBE: I do not have that on this
particular chart, no. It's later in the time line. Do
you have a time for that?
MR. MADDEN: Debris 14.
MR. LABBE: 14 is roughly 13:55, 56 time
frame. About a minute and a half later there.
ADM. GEHMAN: 13:56.
MR. LABBE: Right in there. Yes.
ADM. GEHMAN: Okay. So the two big
pieces of debris come off and it doesn't appear to
trigger an aerodynamic reaction.
MR. LABBE: Can we go to the next chart?
ADM. GEHMAN: Okay.
MR. LABBE: Okay. This is the same plot,
but now I'm looking at a change in rolling moment.
That was change in yawing moment; this is change in
rolling. Here you do see a definite correlation
between that large debris. Somewhere between Debris 5
and 6 is when we see this event where the rolling
moment was drifting negative, the change in rolling
moment, and it changes direction. It starts its positive trend. We think this is a very key point for
us in trying to understand what happened. Something
changed about the configuration, some damage. Since we
know we were shedding debris, something significant
happened there to change the trend in rolling moment.
Debris 14, a minute and a half later.
Again, we don't necessarily see that.
ADM. GEHMAN: It's right about here.
Now, what kind of a change in the aerodynamic, the
external aerodynamic posture of the vehicle would cause
a change in the slope from going one way to going the
other way? I mean, damage on the opposite side?
MR. LABBE: I don't think so. You know,
you've asked the 64,000-dollar question there, I
believe. That's what our work is going to be. You
know, what it suggests early on is that I was losing
lift on the left wing and then something changed to
start creating lift on the left wing or pushing up on
the left wing. Whether or not that's opening up a
large cavity on the lower surface, I'll show you some
results from the wind tunnel that would suggest an
opening of a fairly large cavity on the lower surface
actually results in what I can think of is the damage
is so significant it's creating locally a very high
pressure that is on the lower surface of the wing and starting to push up on the wing as opposed to just
disturbing the flow.
ADM. GEHMAN: As a non-aviator, let me
ask kind of a basic question. Is it possible that the
aileron trim, elevon trim, which is, of course, a
measurement that you use which is not standard with
airplanes, but it is possible that the orbiter, trying
to correct one difficulty, created lift under the wing
by the way the elevons are set?
MR. LABBE: I'm not sure I follow the
ADM. GEHMAN: Well, in other words, in an effort of the guidance and control system to correct the yaw, for example, that the orbiter trimmed itself in such a way as to actually you know, like putting your flaps down?
MR. LABBE: Right. The way the orbiter
flies hypersonically is not your conventional aircraft.
There's no rudder available. It's mass, because you're
up at a high angle of attack, and you're using aileron
and side slip and then the jets, of course, as your
third effecter to try to trim both in two axes, yaw and
roll. Everything that we've seen about the flight says
that the vehicle was doing, the flight control system
was doing the proper response to these changes in these moments to trim out both yaw and roll. So we were not
trimming one and sacrificing the other; they were both
ADM. GEHMAN: You answered my question.
MR. LABBE: Okay.
ADM. GEHMAN: So we don't have an
explanation for this?
MR. LABBE: No, not yet. That's our
damage assessment work.
ADM. GEHMAN: It's not consistent with
GEN. BARRY: You understand, of course,
the roll reversal occurs at 56:30. We've got it on the
green box there. We have most of our shedding
occurring before that because Debris 14 goes off at
56:55. And we have a roll reversal and, of course,
what you see after 56:30 there after the roll reversal,
how much off nominal is that, compared to other shuttle
approaches? The roll reversal is normal; but at
56:30 she goes right off, you know, starting to
MR. LABBE: Right. How much off nominal
GEN. BARRY: Yes. Exactly.
MR. LABBE: I guess, you know, one thing to say, with all that damage, the vehicle executed a
perfectly nominal roll reversal in the middle of the
flight. So despite all the damage, the flight control
system still was commanding the vehicle to do exactly
what guidance was telling it to do. In this level here
these are small and during that time period are not
anything significant. It's almost like the damage has
returned the vehicle back to its original flight
characteristics; but then, of course, starting here we
see a rapid increase and then essentially going off the
cliff there at the end.
Okay. We move on to page 12. It's really just a summary of what we found. I think we've discussed just about everything here. The one thing I would like to point out is that the results we see initially a negative roll and a negative yaw, and there's been a lot of discussion about asymmetric boundary layer transition. When you experience that on the orbiter, these two increments will have opposite sines. So if you have positive yaw, you'll have negative roll or vice versa. We saw the same sine on this. This indicates to me that whatever was happening early on is not asymmetric boundary layer transition; it's some damage. And just basically the bottom line is at the end, just before loss of signal, we were at or approaching rapidly the trim capability of the vehicle.
Okay. The next topic I want to discuss
is now our damage assessment, what is causing this. We
have our events, our A, B, C, D and loss-of-signal
events where we're trying to look at the aero
characteristics I just showed you and now go and try to
produce some damage and do some tests and analysis that
will generate those signatures. We have wind tunnel
testing being done at Langley in their facilities, and
we're employing computational fluid dynamics from very
simple tools to our highest fidelity tools. Like I
said before, we are assuming damage and then creating a
model and then measuring or calculating that and then
mapping it back to the events.
On page 14. This is just a chart from
Langley. They've been doing an outstanding job in
supporting us, and we also have a poster of this.
Basically this summarizes the three hypersonic tunnels
they have there that we are employing in our
investigation. There's a shuttle trajectory here
versus Mach number and altitude and then we have the
Mach 6 tunnel here and they have a Mach 10 tunnel and
then you've heard about maybe the CF4 tunnel.
We do our initial screening in a Mach 6 tunnel, and there's a lot of questions about how that was applied since Columbia was at Mach 20 and above when we were seeing these events. When you're at Mach 6, you have all of the physics of hypersonic flow and they are listed there but you don't have chemistry. Because of the speed and temperatures, there's a lot of chemistry that goes on.
One way to simulate that chemistry is to
go into this CF4 tunnel, and it changes what we refer
to as the ratio of specific heats. But what that does
to the vehicle is brings the shock much closer, the bow
shock much closer to the vehicle. Expansions are much
deeper. Compressions are much stronger. So by going
into the CF4, we can take a step much closer to flight.
We still don't get up to this point here. Loss of
signal is actually at Mach 18 or so, but that's where
we can employ computational fluid dynamics to get to
that next step.
DR. WIDNALL: I want to understand just a
little bit more about the CF4 tunnel. When you say it
changes the specific heat, how is that actually
accomplished? Is that because the gas is actually at
the real temperature or because there's a different
form of simulation?
MR. LABBE: Maybe Joe can help me out here. All we've done is change the gas from air to
CF4. It's freon.
DR. WIDNALL: So you've basically changed
the gas to freon; but, for example, the same
temperature on the vehicle would be a low temperature.
MR. LABBE: Relative.
DR. WIDNALL: Relatively low. So it's
not like an arc jet simulator or something like that.
MR. LABBE: Okay. I just wanted to give
you a snapshot of the tunnels and how we're applying
The next chart shows some damage. Here's
a picture of the Mach 6 tunnel. There's the model
inside the tunnel, and we have a model here for you to
also look at. They're about 10 inches long. So
they're about three quarters of 1 percent in scale.
We've been taking IR images so we can get thermal
imaging of the model at the same time we get
aerodynamics. And we've gone in and just done some
damage where we notched out the wing leading edge or
drill some holes behind the wing leading edge to
represent carrier panel damage or even this is like a
side shot of the wheel well cavity where we've created
a cavity in the lower surface of the wing. What I'd
like to do is show you some results of that testing.
Next chart, please. It's again another
complicated chart, but what we have across the top is
our thermal imaging. You're looking at the lower
surface of the wing, and you have missing RCC Panel 6.
You have a gouge or essentially what's representative
of tile damage right in the middle of the main landing
gear door and then you have the holes drilled through
the wing, which would represent damage to the carrier
panel. What you see here is that the state of the
boundary layers essentially indicated by the thermal
imaging where you see the increase in heating, we know
that we've tripped the boundary layer and it's gone
turbulent for this particular run. These are very
preliminary results. We like to use those tunnels. We
want to use Mach 6 and CF4. This was just the Mach 6
aero results. So it's premature to draw too many
conclusions just from this set of results.
We have just completed similar testing in
the CF4 this week, and we'll be looking at those real
soon. What this shows is basically we're not getting
much in the wind tunnel, not much change to the
aerodynamics, even for taking out a notch that would
represent an entire missing panel. Yaw or roll.
MR. WALLACE: Just to clarify, when you
simulate the missing RCC panel, your model doesn't simulate any flow through the wing and exiting?
MR. LABBE: That's right. It's just an
external type of notch, and it's a limitation of the
MR. TETRAULT: Let me pursue that
question a little bit. We have debris that is both of
the left wheel well forward corners, and the debris
indicates that there was a flow coming out from the
wheel well outward at those corners. The inboard
corner was flowing inboard, and the outboard corner was
flowing outboard. What would that do to the flow
field? Would that create lift, or what's your sense of
how that would affect the flow field?
MR. LABBE: Like I said, we were there
Friday and we saw the debris and we were puzzled by the
flow patterns. I think if you have a jet, if it's
coming out with a strong enough rate that you create a
jet or create enough flow out of there, it will set up
a shock in front of that which will create a high
pressure which would be on the lower surface which
would push up on the wing and would probably create
more lift. Obviously by the time we've gotten to that
point, though, there must be other damage. So exactly
how those all work together is our challenge.
MR. TETRAULT: But it could create a lift, as long as that jet was still there?
MR. LABBE: Yes. I would think so. Now,
we're looking at all that debris. We are, in our own
minds, wondering what happened prior to breakup and
what happened post breakup.
ADM. GEHMAN: Let me ask another layman's
question here. The patterns that we see up there don't
change whether you're in the right-wing-down or
MR. LABBE: That's correct. These are all, you know angle attack, the plots show angle attack of three angles of attack 39, 40, and 43, I believe, is what was tested. The aerodynamics of the vehicle are a function of angle of attack and angle of side slip and Mach number, not bank angle. We bank about the velocity vector. So whether you're left wing down or right wing down, what the vehicle sees is the same from the aerodynamic standpoint.
ADM. GEHMAN: That's intuitively not
MR. LABBE: I understand.
DR. WIDNALL: Just to pursue that a little bit, in your reconstruction you have really verified that beta, the side slip angle was zero in other words, there's no question about that, that the side slip angle was zero?
MR. LABBE: In our reconstruction, beta
starts out at zero early in flight, but sometime around
the time when we see the first change in aero, it
starts drifting negative. By loss of signal, it's
hanging out at about 1 degree negative.
DR. WIDNALL: Right. I understand that.
But do you have beta through the roll reversal, all the
maneuvering, so you have a graph of beta as a function
MR. LABBE: Yes.
DR. WIDNALL: Less than 1 degree?
MR. LABBE: Right. During roll maneuvers
it might go up to several degrees.
DR. WIDNALL: I'd like to have a copy of
MR. LABBE: We'll get you that.
Okay. So not a whole lot of damage. There are some CFD results here
ADM. GEHMAN: Excuse me again. Since we
can't really read the scale on that chart, can you give
us some kind of indication of whether that's a little
heat, a lot of heat, severe heat, life-threatening
MR. LABBE: I'll let Joe answer that.
MR. CARAM: As you look at the images, you can see that the areas we see of red are indications of fully developed turbulent boundary layers. So you have two types of boundary layer characteristics laminar or turbulent. The turbulent provides higher heating, on the order of two to three times what you see for the laminar heating.
DR. WIDNALL: You used a key word, and I
want to make sure that I understand this chart. The
dashed line on the graphs is your calculated
differential aerodynamics that you would hope the wind
tunnel tests would go to?
MR. LABBE: That's correct.
DR. WIDNALL: So your wind tunnel tests are the solid lines with the dots on them, the dashed line is what you had hoped to get out of that particular wind tunnel test to explain, and then that triangle you said was a CFD is that what you said?
MR. LABBE: That's correct.
DR. WIDNALL: Okay. That's very
interesting. So you're saying that the CFD actually
predicts what you hoped the wind tunnel tests would
show. Is that what you're saying?
MR. LABBE: That's what I'm saying. And if we go to the next
DR. WIDNALL: Wait a minute. I mean, I
know this is a nasty question because I understand the
limitations of CFD, but to what do you attribute the
difference between CFD and the wind tunnel tests?
MR. LABBE: Okay. The CFD, this is an Euler calculation
DR. WIDNALL: It's a challenging
MR. LABBE: But this is a calculation
that doesn't have a boundary layer. And I believe
what's happening is when we are tripping the boundary
layer here, we're getting offsetting changes. So when
I do this computation, I don't have a boundary layer
and I'm not getting the offsetting changes.
DR. WIDNALL: Okay.
DR. HALLOCK: My experience is primarily
below Mach 1, but one of the issues you have when
you're dealing with wind tunnels is matching Reynolds
number. Here I see it is 10 to the 6. What is it
really for the shuttles itself, and is that a problem
MR. LABBE: This was run at roughly
.4 million, which is based on the length of the
orbiter. When we are in the flight regime that we're
studying, where we're interested is about half a million up to about 2 million.
MR. CARAM: 2 million.
MR. LABBE: So this particular test was
at a little bit higher Reynolds number.
GEN. BARRY: If you could just put to bed
one final question that we keep getting. Is there
anything that could have been done, whether the orbiter
rolled left or right, to minimize the heat as it was
reentering, based on any of the testing you're getting
on wind tunnel or otherwise?
MR. LABBE: I don't believe bank angle
changes your heating profile at all. So the answer
would be, no, I don't believe so.
MR. CARAM: No.
MR. LABBE: Okay. The next chart does
show just a snapshot of that CFD analysis. This is
again done by Langley, using a code call FELISA, and we
took out the same RCC Panel 6. You can see the flow
patterns, essentially showing the pressure
distribution. There's a shock forming. These three
thermocouples on the side of the fuselage that showed
temperature increases, the shock is in the vicinity of
that. We're doing this at Mach 23.8. So it's very
close to flight conditions. These figures here show
the blue is a clean configuration and then the red would be with the notch and we're showing that the
stream lines are tending towards the fuselage. So
there's a lot of indications here that wing leading
edge damage is consistent with some of the patterns
we're seeing here.
DR. WIDNALL: Could I have a question? I
mean, I think that's a very exciting result. So what
you're saying is that the temperature increase on the
side of the vehicle could be explained by a shock
coming off of this notch in the leading edge? That's
the first time I've seen this.
MR. LABBE: Okay. And Joe is going to
show you a lot more of that. But, yes.
MR. TETRAULT: Does that explain the
temperature that's far forward, the temperature
increases in the dump values?
MR. CARAM: No, that does not.
MR. TETRAULT: It does not get to that, it only gets to the side body
MR. CARAM: That's correct. The flow's
not going to be moving forward on the vehicle. It's
only going to be moving aft.
MR. LABBE: Okay. The next chart just
goes into a little bit larger damage. Basically we
talked about the wheel well. They took a metal model at Langley and machined out a representative cavity
that would represent the main landing gear wheel well.
And there's two depths to that, basically, a very deep
and then a more shallow. That's what the H over L is
representing. It's kind of hard to see; but if you
look closely, this shock that's forming in the wheel
well in this cavity is much stronger for the shallower.
ADM. GEHMAN: You'll have to describe
what we're looking at here.
MR. LABBE: Okay. I'm sorry. This is a
Schlieren photograph. What we use that to do is to see
shock structure in the flow field. So what you're
seeing is a bow shock on the orbiter vehicle and then
embedded inside of that is a secondary shock where this
cavity is and you can see there's this faint line that
goes up here is indicative of the shock forming in the
wheel well. Those are forming when you have abrupt
changes in the flow field. You end up forming shocks,
and that would be an area where you could expect high
So the results, this is a later time in
flight. Now we're 860 seconds and again the same
format on the plot that Sheila pointed out where we
have the flight data and we think we should be
approximating with this type of damage and then the wind tunnel results. And we're getting in the
neighborhood. In the rolling moment, the yawing
moment, we're only producing about half of what is
expected. But that's essentially the technique. We'll
look at this. We'll map it back. We're going to get
these results out of the CF4 tunnel which will be
closer to flight. We talked about the changes. This
bow shock will be much closer to the body in the CF4,
which would be much more like flight, which should
change some of these characteristics of what we're
ADM. GEHMAN: But that particular
measurement was if there was no landing gear door,
landing gear is gone and you've just got a hole there
because the landing gear door has been ripped off.
MR. LABBE: That's right. In this particular, we've done calculations with landing gear and main landing gear deployed or testing. I just don't have those charts.
DR. WIDNALL: I was confused by this
chart. Are these two pictures of two different landing
gear configurations, one deep and one shallow?
MR. LABBE: That's correct.
DR. WIDNALL: And on the two graphs, is
that rolling moment and yawing moment?
MR. LABBE: Rolling moment, yawing
moment, and we actually tested three depths.
DR. WIDNALL: Okay. Fine.
MR. LABBE: So you're seeing the shallow,
the deepest, and then there's an intermediate.
DR. WIDNALL: Okay. So everything is on
this single page for these two different kinds of tests
or actually three, I guess. Three tests.
MR. LABBE: Three tests. And the
shallowest actually produces the largest change. I
think Joe might be able to explain that in the future
Okay. That was just a snapshot of the
work we're doing, and we're just getting started on
this damages assessment. So my last chart is just kind
of a summary. We've looked at these things. One thing
that surprised us is when we put this initial damage in
the Mach 6 tunnel, we got very small increments and not
big enough to explain flight. The CFD suggests maybe
there's still something to that. We're going to
evaluate those and resolve those differences, apply our
higher fidelity tools.
DR. WIDNALL: Well, would that single
notch explain perhaps some of earlier part of the
off-nominal aerodynamics before you get into the catastrophic failure?
MR. LABBE: Yes, it could. What's
puzzling is that if it's also explaining the side wall
temperatures, those don't happen until 600 seconds or
DR. WIDNALL: Good point.
MR. LABBE: So that's one where we're not
integrated with the thermal and so maybe it's not wing
leading edge early on or it's a different panel. So
we're going to be looking at multiple panels missing
and other panels missing, and that's really where our
future work is focused is to first do a survey of the
wing leading edge and then start looking at other
damage scenarios that try to produce that and then
eventually get our higher fidelity CFD analysis tools
to get to the actual flight conditions and high
fidelity models of this damage.
GEN. BARRY: As you do the piecewise
integration, so just your aerodynamic element, just
some quick answers. One RCC does not account for what
you see. Yes or no?
MR. LABBE: No.
GEN. BARRY: Okay. How about four?
MR. LABBE: To be determined.
GEN. BARRY: Okay. How about a landing gear with an RCC, landing gear down?
MR. LABBE: Landing gear down, we didn't
do both; but I guess if you could put them together,
landing gear down increments look very similar to just
prior to loss of signal.
GEN. BARRY: Okay. Final question is:
As I think you told us, if the main landing gear door
is gone, the gear is still up, that will not give you
enough to qualify, from what you've seen
MR. LABBE: It would be sometime earlier
in the flight, where the increments have not grown to
the large level we've seen just prior to LOS.
MR. WALLACE: Some of your initiating
scenarios seem to be distinct. I mean, are you looking
at sort of things in combination? I'm also curious as
to whether does it remain an issue of Columbia's
historical wing roughness as a factor.
MR. LABBE: As far as the scenarios, most
of the scenarios that have been developed start with a
single damage that was relatively small that grew. I
believe the scenario team is now, as we bring in some
results, starting to rethink some of those, could it
have been something more substantial early on; but
that's kind of the iterative nature of this evaluation.
As far as the roughness on the orbiter wing, I think, Chris, maybe you could from a TPS standpoint, my understanding is that was recognized and there was a lot of effort to make the Columbia wing as smooth as possible by eliminating the sources of that roughness. So it was a very smooth wing.
MR. MADDEN: As far as, you know, the
signatures we saw were not anything related at all to
any sort of early transition.
MR. TETRAULT: Do any of your future test
plans include multiple breaches in the wing?
MR. LABBE: Not right now. But I am open. Our test plans are very fluid. So right now we are trying to I think the next thing we're going to do after we get the wing leading edge is drill large holes in the wing so you actually have a flow from the lower surface through the upper surface and see what results we get out of that.
MR. CARAM: As well as multiple panels.
GEN. BARRY: A follow-up on Steve
Wallace's question. The last STS flight that had a
really early transition from laminar to turbulent flow
was STS 73, I think. That was like 893 seconds. Every
one after that was pretty nominal. Now, that can be
qualified by working the issue and trying to smooth out Steve Labbe, Joe Caram, Chris Madden - March 18, 2003
the wing and in between flows and at the maintenance,
is that correct, when we do the OMM?
MR. LABBE: It's either that or Joe, I mean
MR. CARAM: I would agree; but when you
go back to STS 73, the cause of that that we
established is a protruding gap filler. The material
that resides in between the tiles sometimes displaces
and can reside there in the flow, and it was on the
order of about a half inch to an inch in size and
sitting about 20 percent along the center line down the
vehicle length. And that, we've shown in ground tests
that we can achieve boundary layer transition because
of that kind of disturbance.
MR. WALLACE: How do you identify whether
there's boundary layer transition? What's the
MR. MADDEN: Well, you would see it on
the surface temperature. You would see an immediate
rise in temperature on the surface. I was referring to
the off-nominal events we saw. Clearly things are
happening well before even the earliest transition
we've ever seen; and in terms of the roughness, I think
what we should do is get you a little report or a white
paper on what's been done on the orbiters to make them smoother.
ADM. GEHMAN: I definitely want to let Steve get off stage here; but I, too, have one more question. One of the first things you said was that you know pretty much about nominal shuttle reentry aerodynamics normal. But in my experience, I have experience in aircraft development and procurement and I won't mention anything specific but I remember being in a position of authority in the US Navy when an aircraft we were buying had several hundred test flights, several thousand hours of test flying, and we discovered a new, completely new and unexpected aero control problem, which was all in the front page of the papers and everything like that. It caused us a considerable amount of heartache to fix it and convince Congress that we had it fixed. So I must admit that I require a little convincing that after 113 flights and a few thousand seconds in transition that you say you know a lot about shuttle reentry and the aerodynamics of all that.
MR. LABBE: I'll offer you one thing and see if this what makes the orbiter different from, say, a military aircraft is that while we have a very broad flight envelope in speed, we fly the exact same profile over and over and over again. So each flight is essentially flying the same profile. We're not trying to expand this envelope that has very large differences at flight conditions. And we've learned a lot. Believe me, we've had a lot of instrumentation. We by no means had it figured out on the early flights. We did flight maneuvers and so, because of the repetitive nature of the entry profile, along that profile we have it very well figured out. If we diverge from that profile, then what you say is exactly true.
ADM. GEHMAN: I think we better get on to
Joe here or whoever is next. Thank you very much,
MR. LABBE: You're welcome.
MR. CARAM: Page 20, please. Again, this
is just revisiting our flow charts. So now we'll be
taking about the aerothermodynamics environments.
Next page. This is just a simple chart
to try to explain to you the process that we go through
when we provide aerothermodynamic environments to the
thermo community. Our answers, in and of themselves,
aren't the final product. We have to provide those to
the thermo analysts for the analysis of the structure.
As inputs to us, we need the trajectory conditions, how the vehicle's going to be flying through the atmosphere, its speed and density profile. We also need the configuration both the nominal and, in this case, what kind of damage scenarios are we assessing.
So as I think Dr. Bertin has already gone through with you, heating is a result of the exchange of kinetic energy of the vehicle to thermal energy in the gas. So you have now high-temperature gas flowing around the vehicle. As it flows around the vehicle, it imparts that energy to the vehicle. So when you consider what you have to do and look at when you're providing aero heating environments, you have to consider the physics and the chemistry of the flow. The physics phenomena, bow shocks, as Steve was talking to you earlier, the shock interaction on the wing, the boundary layer, the state of the boundary layer, whether it's laminar or turbulent and the transition in between the two, any kind of separation zones and reattachment for instance, the body flap, if that were to deflect down into the flow the flow upstream of it would separate and then you would have a reattachment point on the body flap where you would see higher heating. So anywhere there's a geometric change, we need to consider that in providing those heating environments.
Chemistry aspect. After I have the
physics modeled, we want to take a look at the
chemistry. As the air passes through the bow shock, it
is heated up to approximately 8,000 to 9,000 degrees
Kelvin. At that point the air molecules, the N2, the
nitrogen molecule, and the oxygen molecule can split.
So they dissociate, and that requires energy to occur.
So it's an endothermic reaction.
So now you have these atoms flying around
the vehicle. So that changes the chemistry of the
flow, and that can have certain effects. You look at
the shock angles. The shock angles can come closer to
the vehicle. The pressure distribution can change
slightly. And you're looking at the difference in
When you talk about TPS environments,
thermal protection systems that have partially
catalytic coatings on them, you can gain an advantage
by not absorbing the heat in the flow field because it
doesn't allow those atoms to recombine on the surface.
That's called partially catalytic heating, and the
shuttle's TPS is coated with those coatings. So during
those times when you have this dissociation, you can
gain some advantage in the heating environment. So you
have to account for these various physical and chemical phenomena before we provide heating environments for
the thermal analysts.
DR. WIDNALL: I'm very interested in this question of surface catalysity. I've got some data actually that NASA did and it came out of Professor Bertin's book, but you guys had the courage to run on one of your flights I think it was STS 2 where you painted, oh, seven tiles on the shuttle with a surface catalytic coating, a coating that allowed this recombination of O2 and N2 to occur on the surface.
MR. CARAM: That's correct.
DR. WIDNALL: Roughly speaking, the
temperature on the surface of those tiles went up by
about a factor of 2 to 3. That's the result.
MR. CARAM: Okay. I would believe that
to be true because what you're doing is recovering the
energy in the boundary layer.
DR. WIDNALL: Right. And just to get
these temperatures sort of on the record, when the gas
dissociates from behind the bow shock, roughly
speaking, you're looking at a temperature of, what,
3200 degrees Rankine? I mean, that's the temperatures
that I got from Steve.
MR. CARAM: At the edge of the boundary
DR. WIDNALL: At the edge of the boundary, there's what I would call stagnation temperature. About 3200. And the reason it's as low as it is is because
MR. CARAM: The partially catalytic
nature of the material.
DR. WIDNALL: Well, no, you ripped the
gas apart, so the temperature's gone down; but you
still have this energy potential, should you have a
fully catalytic surface, to drive that temperature back
MR. CARAM: That's correct.
ADM. GEHMAN: Let me follow on to that.
The TPS system, a particularly high reusable system,
it's painted to prevent that catalytic action.
MR. CARAM: It's coated.
MR. MADDEN: Reaction-cured glass
ADM. GEHMAN: That's right. But how deep
is that coating and is it possible that that coating
could be torn or damaged?
MR. MADDEN: Well, okay. The short
answer is that every mission there is multiple small
damages on the tile. So practically every mission has
tiles with coating damaged, which would imply chipped and missing. From that standpoint, the tiles are very
robust to survive having the coating missing.
ADM. GEHMAN: What I'm getting at is that
not only does damage to the smooth surface of the TPS
create little aerodynamic spots, it also provides an
opportunity for catalytic recombination.
MR. MADDEN: Yes. And also without the
coating, the tiles suffer from reduced infrared
re-radiation cooling effects. So it's a bit of a
double whammy; but the bare tile, even though it's not
coated, I don't think is very catalytic either.
ADM. GEHMAN: Okay. That's what I was
MR. TETRAULT: To go back to Sheila's
train of thought and inquiry, if you had an exposed
wing spar, wouldn't you have a catalytic surface?
MR. CARAM: Before the surface itself
oxidizes, yes. But as it heats up and the oxygen
penetrates that surface, it will perform an oxidation
layer. And Chris has some material on that for y'all
today. And that oxidized layer is partially catalytic.
GEN. BARRY: Let me just ask a question
on the RCC. At the boundary on the surface of the RCC,
temperatures can get as high, between Panels 7 and 12,
MR. CARAM: 2950 degrees Fahrenheit.
GEN. BARRY: And how far in front is the
boundary layer and what is the temperature, let's say,
inches forward of that?
MR. CARAM: Well, as you get to the wing,
you're starting to expand over that wing and the
boundary layer is getting thinner.
GEN. BARRY: At the edge of the boundary
layer, what's the difference in temperature?
MR. CARAM: Probably around 3,000 degrees
Fahrenheit. So not significant. Not a significant
difference in the edge of the boundary layer.
GEN. BARRY: Maybe I'm asking the
question wrong. When you get in front of the boundary
layer, what is the temperature? We were told at one
time it may be as high as 10,000 degrees.
MR. CARAM: The gas temperature can be
as high as between 9 and 8 thousand degrees Kelvin.
GEN. BARRY: So you go from the edge of
the RCC to just 6 inches forward and the difference is
almost 7,000 degrees.
MR. CARAM: That's correct.
GEN. BARRY: Okay. Now, if you get a
nick or a little bit of damage to the RCC and you have
this recombination that you just discussed, does that bring that 10,000 degrees closer in and reduce that
MR. CARAM: No, it does not. It's what
the available energy is in the boundary layer itself.
It's not bringing that shock layer closer. It's just
how you exchange the energy in the boundary layer
around the vehicle. So at the boundary layer edge,
you're seeing around maybe 4,000 degrees Fahrenheit;
but that is also changing as you go down through the
DR. WIDNALL: Wait a minute. I've got a question. The material that John is talking about, if the leading edge is damaged, is carbon. Carbon reacts chemically with the available oxygen and that will, in fact, release
MR. CARAM: I didn't understand that he was mentioning
DR. WIDNALL: Yeah, he was talking about
a damaged leading edge.
(To Gen. Barry) I think you were.
Weren't you talking about a damaged leading edge?
MR. CARAM: I misinterpreted his
question. This is more in Chris' area, but you could
start oxidizing the carbon and that can result in the
carbon receding or ablating. Steve Labbe, Joe Caram, Chris Madden - March 18, 2003
MR. MADDEN: An uncoated carbon panel I think that would have been briefed on this an uncoated carbon panel will oxidize because the carbon's going to react with the oxygen. And it's quite rapid. But as far as surviving a mission, I think, even though you get some damage, in most cases you don't eat through the entire thickness of the carbon. There's catalysis and oxidation on top of each other.
MR. WALLACE: Did you see that in your
observation of the debris in Florida?
MR. MADDEN: No. The debris in Florida is we don't know what happened there.
DR. WIDNALL: I have another question.
MR. MADDEN: But there was a lot of bare
carbon that looked fresh. Shiny. It didn't look like
it had been oxidized very much at all.
DR. WIDNALL: You seem to be using the
word "oxidation" and "oxide" as if it forms a
protective coating. Another word for oxidation is
"burning." I mean, the experiments that I've seen that
NASA has done indicate that damage to the leading edge
of a carbon-carbon burns a hole completely through the
MR. MADDEN: An existing hole would grow,
and then a damaged panel would oxidize the bare carbon and eventually would grow a hole.
DR. WIDNALL: Yeah. You would eventually
get a hole in the carbon.
MR. MADDEN: It depends on which panel
you're talking about and how rapid.
DR. WIDNALL: The only question we're
talking about is: What does eventual mean? How many
seconds is eventual? That's what we're talking about.
MR. MADDEN: We performed analysis for
the investigation on panels with existing holes and how
fast they grow and how fast they eat away at the spar.
DR. WIDNALL: I realize that you're going
to present later; but as we're talking about this
thermal environment, I would also raise the same
question with respect to aluminum. I mean, it
certainly is true that in our common experience of
aluminum, oxide is a protection for aluminum.
Otherwise we wouldn't have airplanes and we wouldn't
have chairs and all the other things that are made out
of aluminum. But aluminum oxide at a temperature of
3,000 degrees Fahrenheit is not a protection. The
melting point of aluminum is 700 degrees.
MR. MADDEN: Right. It's going to melt
and go away before you see that effect.
DR. WIDNALL: Yeah. Very quickly. It is not a protective coating for aluminum at the kinds of
conditions we're talking about, and I think that is a
subject we want to pursue in more depth.
MR. MADDEN: Well, we've got a chart or
two on that, as well.
ADM. GEHMAN: Okay. Board, let's let
MR. CARAM: Next page, please. Page 22.
Just to go over some of the models and techniques we're
applying in order to provide these environments. The
orbiter has an existing external heating data base that
we're using to provide the local heating around the
various damage sites that we're considering. We're
also using a plume model that was developed for
micrometeoroid penetration, so I means small
penetrations on the orbiter. However, for total
environments, both the convective and plume, the models
don't exist for the size and scale of damage that we're
considering. So we're having to develop those
techniques as we go.
We're also using engineering analysis or
correlations that we have available to us, and I'll
show you an example of that on the following page when
we're dealing with cavity flow heating. We're also
using what we have for existing computational solutions on the orbiter. We have the orbital experiment data
from STS 2 that's been calibrated with the
computational data. We also have pre-use test data.
We're also using, as Steve described
earlier, the current activities at Langley and the wind
tunnel testing that we're doing to look at the local
and heating environments as a result of damage to the
early metal. What we're trying to do with the more
high fidelity tools such as computational fluid
dynamics is to verify those environments because we are
going through different environments as we're coming
through the atmosphere. Early on, it's more applicable
to use a breakthrough as the Monte Carlo technique; and
since we are assessing damage that existed, we're
assuming, at entry interface, you want to verify that
the heating environments that we're providing are
accurate in those regimes.
So the following page gives you an example of the cavity heating models that we're using. The cavity heating for instance, many of you have one tile lost or three tiles lost. The heating down in that cavity will vary, a function of the length over depth ratio. And that ratio changes the heating. If you have a ratio of 14, over 14 you have a closed cavity and under 14 you have what's called an open cavity flow. It does not say it's penetration. It's a description of a flow inside that cavity. So with open cavity flows you tend to have less heating on the floor than you do with closed cavity flows because with closed cavity flows the flow has the opportunity to reattach to that floor and then start heating up the floor there before it separates again and reattaches on the outside of the cavity.
You also have to consider whether the
boundary layer is laminar or turbulent upstream. That
can change how much energy is being provided inside
that cavity. So it could change the types of
coefficients you're using. Typically you apply
coefficients down the cavity and you assume upstream is
the nominal heating. So you have the nominal heating
factors times the cavity factors, and that's how you
derive your heating.
Most of this data was established from
-D environments, 2-D testing. There's some data with
three-dimensional effects, but that data is just along
the center line of the three-dimensional object. Why I
mention that is because if we're assessing cavities on
the carrier panel tile areas, that flow is sweeping
outboard on the wing leading edge and it's highly
three-dimensional. There's a lot of cross-flow. So again, I want to be sure that the environments we're
providing are accurate.
So the next page is an example of how
we're doing that. Again, this is the schematic of the
open cavity flow typical for a single lost tile. On
the right you see a close-up view of the pressure
distribution from a CFD solution from an STS 2 CFD
solution using the LAURA code at Langley. Forward, the
nose is this direction. Outward is the wing. You can
see the outline of the main landing gear door. The
symbols in red are higher pressure. The blues are
lower pressure. And the high pressure in this region
is a result of the shock interaction zone. So you have
a higher pressure leading up from the leading wing edge
and then flowing inboard and aft from that region.
So we take information from the external
flow field and provide that as input conditions to a
cavity flow solution. And this solution here is a
direct simulation Monte Carlo solution of 2-D cavity
flow at high altitude. Why I wanted to present this is
because what the direct simulation Monte Carlo is doing
is giving you an indication of what the high altitude
effects are doing in your cavity flows. So you can see
it's almost a merge between what you have for open
cavity, between that and a closed cavity flow. So we want to know that information in order to make sure our
heating environments that we provide the thermal guys
Next page. This is an example of the
wind tunnel testing we've been conducting at Langley.
These particulars runs are from a Mach 6 air facility.
I will be showing you runs from the CF4 facility.
Again, as Steve mentioned, we've been looking at
notched wing leading edges. On the left you see a
nominal configuration orbiter, a side fuselage heat
transfer. This was done with the infrared system at
Langley. In order to acquire heating rates, we
measured the temperature, assume a short delta time in
the tunnel where the image was taken, and then 1-D
thermal analysis to back out the heat transfer
The two reds dots indicate the side wall
fuselage temperature measurements that showed
off-nominal behavior. The red zone is the shock
interaction zone on the wing leading edge, and this
area here is the attachment of the flow coming around
the chine of the vehicle, scrubbing along the side of
the vehicle. So this is what it pretty much looks like
in a nominal configuration.
When you take out Panel 6, as Steve showed you previously, you then have this shock
impinging on the side fuselage. In this case since
we're in the air facility, so we're at Mach 6 at air,
you see that it doesn't show that it interacts with the
sensors at this location.
So we also took a look next page at Panel 9. Again, here is Panel 6 in comparison going further out on the wing, removing Panel 9. Again, Panel 9 is in the region of the double shock interaction zone. So not only do we have the effect of Panel 9 but you also have the effect of the higher energy because of that double shock interaction zone. So can you see between the two that Panel 9 moves the disturbance further aft on the vehicle.
DR. WIDNALL: You said these were Mach 6?
MR. CARAM: These were Mach 6 at air.
DR. WIDNALL: Okay. I mean, at Mach 20
those shocks are going to lean over.
MR. CARAM: Next page.
DR. WIDNALL: You got it.
MR. CARAM: In order to do that, we're first using the CF4 facilities; and we're also using our computational techniques, as well. As we talked about earlier, this is a comparison between the air facility Panel 6 and Panel 9 to the CF4 facility, which simulates the high-temperature gas effects. Again, what we're trying to do with that by changing the gas is to model the high-temperature gas effects; and what you're getting there is that the shocks are moving closer to the boundary, to the body. The pressure distributions are changing slightly, and this is the result. So you see that seen for Panel 6 you see the heating or in this case this is just a temperature map. This is qualitative data only at this point in the analysis, but the high-temperature area moves slightly aft from Panel 6. With Panel 9, it moves further aft and the distribution changes. So you're getting the effect of the simulated high-temperature gas in this facility and at this point you can say that Panel 9 shows the influence over those gauges.
ADM. GEHMAN: Joe, speak about heating
MR. CARAM: Okay. We really aren't
seeing any changes forward of these damaged locations.
other than this flow right here. Forward, where the
vent nozzles are, you're not seeing any changes where
those are occurring. Now, you have to realize when
you're doing this experimental technique you're taking
snapshots of the image right after the model's inserted
into the tunnel. These imaged times can vary. The model baseline temperatures can vary. So you might see
small differences in the reduced heating that you get
out of the test, but in this case we're not seeing
hardly any changes as expected within the uncertainty
of the test techniques aft forward. Most all the
effect is on the side wall and aft.
DR. WIDNALL: Did you go above Mach 6? That's my question. My question is a geometric question, not a real gas question. If you were able to and I understand the limitations of tunnels if you were table to run such an experiment at Mach 20, your shock would be way leaned over from Mach 6 geometrically.
MR. CARAM: No, because the
DR. WIDNALL: Are you saying it gets into a Mach number independence regime
MR. CARAM: At a point. But then you
have the chemistry effects that take over. So those
chemistry effects will change your Mach angles, your
bow shock angles. So it's not going to change
significantly. When we obtain heating data in both
these facilities, it matches with flight within
5 percent. So you're not seeing a large change in the
way the flow is flowing around the vehicle. It
accurately models the hypersonic flight environment. Steve Labbe, Joe Caram, Chris Madden - March 18, 2003
MR. TETRAULT: Would you bear with me a
minute because I don't know much about wind tunnel
testing. I know nothing. So let's start from there.
What you're doing is looking at the external or
exterior environment here. Can you use the wind tunnel
test to test the internal environment? Like you just
put a notch in the wing. Can you go up and down the
wing and see what the thermal conditions, say, inside
an RCC panel is, using this mechanism?
MR. CARAM: This is the scale and type
model we are testing.
MR. TETRAULT: Well, you could drill
holes in it, right?
MR. CARAM: We could drill holes through
the wing, but it would be very difficult to obtain the
heating and the proper scaling inside that area. On a
larger scale? Possibly.
Next picture. Next page, please. All
right. As a follow-on, again, we're trying to verify
these environments; and we're using the higher fidelity
techniquese. This gives you an idea of where we're at
currently in this process. We've established a common
service grid. Since we have these multiple
organizations working on this problem, one of the
issues with computational fluid dynamics is that we can have differences just because of the grid topology. So
we've established a common one between all the
organizations, and so all the organizations will be
using a similar topology.
We can use that same grid system to
implement or embed damage in various locations on the
wing leading edge, along the fuselage of the vehicle.
And we'll be using those to provide and verify the
environments for the damage scenarios. So we can do
both the nominal geometry and damage. We're also
continuing to do the wind tunnel testing both in air,
as an initial screening, because that facility is able
to turn around the tests faster than the CF4 facility,
so we'll do initial screening in air and then go to the
CF4 facility to observe the simulated high-temperature
So out of this, we get not only updated
heating environments going to the thermal analysis
group but we also provide inputs to internal heating
environments. We have the outside boundary layer
conditions at the local areas where the damage or
breach is occurring that we're trying to model. And
since we are accurately trying to provide the heating
distributions, as a by-product you have the pressure
distributions and from there you can provide the aerodynamics. So we can provide that information to
the aerodynamics communities for the various damage
configurations that we're looking at.
DR. HALLOCK: Depending a lot on the CFD and also the other types of models here and you're sort of referring to them as being the truth of what's going on how do we know these models are actually predicting or calculating what's actually going to happen?
MR. CARAM: We're using the wind tunnel
data, as well. So what we're trying to do is
calibrate, for instance, at the Mach 6 conditions; we
want to run those conditions, as well. If you can
establish that you can correlate well with that data,
then by changing your free stream Mach number and
adding the chemistry in, we feel confident that we can
get the accuracy that we need. We'll also have to do
grid resolution studies, so to make sure that there is
no grid sensitivities in the solutions that we obtain
out of the CFD.
DR. HALLOCK: Do these models include the chemistry effects also
MR. CARAM: Yes.
DR. HALLOCK: or are you actually adding that upon the solutions?
MR. CARAM: No, they're embedded into the
Now, we've talked about the external
environment. I want to move on to the external
environments. This is a more difficult, I believe, and
less established approach. Now, I know this is a busy
chart; but it's actually quite simple to go through.
It just gives you a road map of how we're trying to
handle the internal environments.
Again, one of the customers for the
internal environments group is the external
environments. So they feed right into the internal
environments group. What the internal environments
group does is provide heating environments not only for
plumes but looking at, beyond the plume flow field,
what is the internal convection inside the wing, where
is the energy being distributed inside the wing and the
wheel well. To do that, we're requiring several phases
of the analysis.
We've already provided this 1-D heating methodology. This is a plume model that gives you the heating along the axis of the plume only. It's fully equilibrium heating. So it's going to be the worst-case heating and also captures the turbulent reattachment. So it is the worst-case heating as far as plume heating is concerned; but in order to look at the various scenarios, we need to have models that provide off-axis heating. So you have to assess whether, if your plume's not impinging directly on the object that you're worried about for instance, the wire bundles we have to provide heating environments off axis. So that's what this is attempting to do, and we'll be updating our models for that.
Then there's other kind of configurations
of plumes. You have wall-bounded jets. So there's a
jet orifice that is immediately adjacent to a wall. So
the heating along that wall is going to be different
than what you would see with an asymmetric plume.
DR. WIDNALL: Can you tell me how you
would do the calculation of a flow impinging on a flat,
bare aluminum plate that is, in fact, a leading edge
MR. CARAM: If we can go to the next
chart, I think I can try to do that. Basically what
you're looking at is a description of a plume entering,
for instance, the interior area or the spar of the
vehicle. On the outside, you have the boundary layer.
Then you have this external pressure. It's that
external pressure in combination ratio to the internal
pressure which will obtain what is your geometry of your plume. And this plume can exist, this core
environment can exist up to 20 diameters or greater,
0 whole diameters or greater downstream. And that's
where you're getting your high heating area.
DR. WIDNALL: Roughly speaking, what is
the stagnation temperature of that jet and what is the
MR. CARAM: Again, well, it depends on what your external conditions are and how big the hole is. So a large enough hole, you can probably swallow the entire boundary layer. So you can have gas temperatures up to 9,000 degrees Kelvin entering
DR. WIDNALL: Then you're assuming the
gas is not dissociated.
MR. CARAM: No, it can be dissociated at
that temperature. It is dissociated at that
temperature. It requires that temperature for
DR. WIDNALL: Right. But the outside
gas, the stagnation temperature is basically 3200,
based on the fact it's already dissociated.
MR. CARAM: But if you're swallowing the
entire boundary layer and beyond that, you can get
basically the post-shock gas temperatures.
DR. WIDNALL: Anyway, order of magnitude. Fine. Okay. So you're saying that you could have a
dissociated gas flow at a temperature of 9,000 degrees
Kelvin hitting some structure.
MR. CARAM: Yes.
DR. WIDNALL: Then what boundary
condition would you assume for that structure?
MR. CARAM: As far as the chemistry is
DR. WIDNALL: Yeah, as far as the
chemistry is concerned.
MR. CARAM: We're applying equilibrium
heating. So it's fully catalytic.
DR. WIDNALL: And reactive.
MR. MADDEN: Not right now.
DR. WIDNALL: Not right now. Okay.
MR. CARAM: At this point when you have fully catalytic, you're obtaining all the heating from the chemistry that you're going to
DR. WIDNALL: So assuming no chemical
MR. CARAM: No chemical reactions with
the material. That's correct.
MR. TETRAULT: Is one RCC sufficient to
swallow the boundary layer, the entire boundary layer
so that you're getting the 9,000 K in?
MR. CARAM: I would say so.
MR. MADDEN: Just because you swallow the entire boundary layer you still have to transfer heat from that gas. So just because the gas is 0,000 degrees doesn't mean this surface it's impacting is 10,000. That heat has to be transferred via another boundary layer.
DR. WIDNALL: You also have stagnation,
which is going to raise the heat.
MR. MADDEN: It still has to transfer the
DR. WIDNALL: Yes, but it will raise the
temperature. The stagnation will raise the
temperature; and then you're, I would say, halfway
MR. MADDEN: I don't understand. What
you do you mean, halfway?
DR. WIDNALL: Well, if you stagnate a
high-speed jet, you're going to get an increase in
MR. MADDEN: Correct.
DR. WIDNALL: Then the viscous process that transfers through the boundary layer
MR. CARAM: It's true. It's almost like
having another bow shock.
DR. WIDNALL: Yes, exactly. It's like
having another bow shock.
MR. CARAM: Agreed.
DR. WIDNALL: So it's an internal reentry
MR. CARAM: Which again, on the scales that we're talking about for this type of damage, we're having to create these models because if you have a large enough damage for instance, in this picture you have, eventually you will get turbulent mixing with the available or ambient flow in the cavity; but if your hole is large enough or you're close enough to the structure, you can have underdeveloped plume heating and that can be on the order of two to three times higher heating than you would see with a fully developed core flow. So again we're building these models. We're updating them for these phenomena for off-axis heating and for wall boundary jets. So these tools are in work, and we provide those tools to the thermal community.
Next chart, please. Part of this
analysis also involves, outside of the plume
environment, where is the energy going inside the wing.
Currently we're using the orbiter baseline venting
model to provide that information. You have the various vent locations in the fuselage, in the mid wing
going aft to the aft wing and then out the spar. You
also have the vent going into the wheel well.
What this doesn't provide is information
on what the high-temperature gas effects are because
now that you're ingesting high-temperature gas, it can
change the way the mass flow is being distributed
inside the wing and the fuselage. So what we do is, in
conjunction with thermal analysis that Chris has been
doing, we can get an idea of where the energy is being
distributed inside those volumes. We're also looking
at the possibility of what we call unmodeled vent areas
such as drain holes or gaps between closeouts. To the
venting guys, these are just bonuses; but to us it's
critical because that will determine where the mass
flow is going inside the vehicle and where the energy
is going. Our colleagues at Marshall are developing
complete orbiter venting models that account for these
high-temperature gas effects using a quasi approach.
It's not modeling the chemistry precisely but if you're
changing just some of properties of the gas as it goes
through the volume. The idea with this is that we can
then capture the phenomena and then couple it with a
thermal model so we can get an idea of how that energy
is not only being distributed inside that volume but Steve Labbe, Joe Caram, Chris Madden - March 18, 2003
also being deposited onto the various surfaces.
Next page, please. This is an example of
that. This is a thermal model of the internal wing.
You have the truss structure and the spar areas. Each
of those are being modeled thermally, and coupling that
with a venting model will give us an idea of where the
energy is being distributed. We need this in order to
reduce the number of scenarios that we have. Yes, we
can burn through a wire bundle; but where is the rest
of the energy going? We have sensors inside the wing,
the fuselage, that don't respond. So we're using that
not only to test against the data that went off nominal
but to test against the data that remained nominal
until LOS. So it gives us a way to differentiate the
different surfaces. So we're coupling this model of
the mid wing and aft to a wheel well model in the
forward glove, and this is being done at the Marshall
Space Flight Center.
Next page. Again, this is just a summary
of the forward plan. I pretty much discussed all the
items here and where we're headed. We've already
provided a simple plume model to assess heating at the
core. We are expanding that for off-axis heating,
taking a look at different types of plumes. We're
using as calibration these benchmark cases you were mentioning earlier, Dr. Hubbard, to verify that the
modeling that we're doing is accurate before we're
applying the flight conditions and then using that
information to upgrade our engineering model.
So we're not applying the CFD directly,
we're using it to build the engineering model so they
can apply it in the thermal analysis. We have the wing
venting model coupled to a thermal model in work.
We're also looking at CFD of the wheel well so we can
get an idea of what the internal flow structures would
be when you have a penetration of the main landing gear
ADM. GEHMAN: It seems to me that this is
a real challenge because in the case of the external
thermodynamic heating models that you do, you have the
aerodynamic forces to bounce them against. In other
words, you've got kind of a check and a balance here.
MR. CARAM: Exactly.
ADM. GEHMAN: But internally you've got
no check. You've got nothing other than the
temperature sensors. It's a one-dimensional theme
here. And you could hypothesize any internal
rearrangement of those spars and sturts and thin
aluminum walls in there and you've got nothing to check
it against. Other than the heating scenario, you don't have a second scenario. And as we have hit on pretty
hard here, once you get the very, very hot gases in
there, the aluminum doesn't stand up very long.
MR. CARAM: No.
ADM. GEHMAN: So you could make yourself
a new thermodynamic path in seconds and you've got no
second part of analysis to check that.
MR. CARAM: That's correct. That's why
we think that these temperature plots and our
interpretation of them is important in how we define
our scenarios. We have the first bit rise as
indication to us that there was a breach, but later on
you have a rapid rise in those temperature
measurements. At that point there's a breach inside
the wheel well so that the hot gas has penetrated at
that point. So that's just the various parts of the
piecewise analysis that we're doing.
ADM. GEHMAN: Are you finished?
MR. CARAM: Yes.
DR. WIDNALL: Can I have a question? I
just wondered at what point in your CFD analysis would
you allow the aluminum to interact and react with the
MR. CARAM: I don't think we have
currently models to account for that in the computational area.
DR. WIDNALL: Do you have the resources
to find out?
MR. CARAM: I'm working with some of the
folks at Boeing Huntington Beach who are looking at the
combination of the heating and the thermal response.
MR. MADDEN: We're going to get a group of guys together to go and address that. Now, I don't think it's coupled with CFD per se, but we're going to look at hole growth and the effects of oxidation, any possible
DR. WIDNALL: This is obviously an
extremely difficult area. I mean, nobody would ever
build a reentry vehicle out of aluminum. So clearly
you're trying to do the kinds of calculations that we
have just never thought about doing. There are some
resources. In fact, a lot of this early work was
really done by NASA Ames. A lot of the expertise that
exists in this area belongs to NASA.
ADM. GEHMAN: Okay. Before Chris gets
started, I'm going to declare a ten-minute break here
so we can pay attention. For the members of the press
in the room, please, this is not a press conference.
So leave them alone. You all are excused for ten
ADM. GEHMAN: Gentlemen, thank you very
much. We are not concerned about time up here. We've
got to get this right, and you're a great source of
information. So the only time constraint I have is
that we don't want to overstay our biological warning
signs that we're not paying attention anymore. So
thank you very much for bearing with us.
Okay. Chris, you have the floor.
MR. MADDEN: My name is Chris Madden.
I'm in the thermal design branch. I just wanted to
start off with a summary of what we've been doing. Our
branch has been part of this investigation, performing
thermoanalysis and support of test planning and
What I'm going to show you is a series of
preliminary results. The first several slides, you'll
start to see that, with enough damage, you can breach
the vehicle in several different ways. And this is the
way we attack the problem in the first few weeks of
vehicles is: Hey, can this damage blow a hole in the
wing? Can this do it? Can this do it? And the answer
always kept turning out that, well, if the damage is
big enough, sure, big enough damage is always going to
breach the wing. You'll see some of that in the slides.
So I just want to caution everybody that
if you see a slide that says a hole burned through in
0 seconds, it doesn't say that's it; it says that
could be it. And what we've done is evolve from that
and after getting frustrated with everything shows that
it could be the culprit, we started going to this plan
where we're saying, look, okay, while the configuration
is semi-stable before we have the debris shedding a
little before 600 seconds, what can we learn or what do
So there's several knowns that we've had
to make engineering leaps in saying that, okay, at
488 seconds when we saw our first bit flip, that was
the breach. So that's a time hack we're going to have
some level of faith in for the time being so that we
can perform some analysis based on that. Based on that
488 seconds, 42 seconds later the first measurement was
lost. So I'm going to show you a plan on how we're
going to take that 42 seconds to determine where the
damage site was and how big it was. We've also got
another time hack at the wheel well temperature rise.
We're going to say, okay, our engineering leap is that
was breach of the wheel well. So now you've got
12 seconds between the first breach in the wing to the breach in the wheel well, and we'll try to figure out
how that happened.
ADM. GEHMAN: I think that the board
understands the assumptions you're making for the
purpose of building a mathematical and an engineering
model of what happened, but I can assure you we don't
necessarily agree with those assumptions. What I mean
is the breach could have occurred two weeks before
MR. MADDEN: Sure. And it certainly
didn't happen after.
ADM. GEHMAN: We understand the mechanism
of why you've got to pin something down so you can do
the analysis. So we're with you.
MR. MADDEN: Okay. I appreciate that.
Okay. So the next slide, this is part of
the energy balance stuff we did at the beginning. I'm
going to show you a series of slides of what we've
done. This is explained, the early bit flip or a small
temperature rise on the brake line. The analysis
assumed here that you boil a hole, and here we did it
at 480 seconds. This is the amount of energy in BTUs
per second that enters into the wheel well.
Okay. The next slide shows the predicted
based on that energy coming into the wheel well via the healthy vent would, indeed, see a temperature rise on
the same order of magnitude that we saw in the flight
data. So the shorter answer is that, yes, a sudden
ingestion of hot gases into the wing, flowing into the
wheel well, would be indicative of the bit flip that we
saw on that very first measurement. So this is kind of
lending credibility to something happened at 488. Now,
agreed, it could have happened earlier and you're just
now seeing the heat coming in because the gas, although
as I think we discussed before, has a high heat
transfer rate to the surface, the amount of mass
involved in the gas is low and therefore the amount of
BTUs the gas molecules can contain is low. So you may
not see the temperatures until this time, anyway.
MR. TETRAULT: You're using just a 5-inch
diameter vent hole to calculate this? You have not
added any of these additional transfer patterns?
MR. MADDEN: Right, this is a healthy
wheel well assumed. The other thing you see from this
analysis is that, at least for this measurement, later
on you're going to need additional heat to explain the
temperature rise. There is another measurement here on
this poster that it start going up at about
0 seconds. There's some other ones that begin rising
at 600. For some reason this brake line was delayed a little bit. This was behind a fiberglass cover, so
that could explain that.
DR. WIDNALL: Could I just raise a
question? Sort of philosophy. I mean, I think this is
the point where one then needs to begin to challenge
the model because you have a conclusion on this slide;
and your conclusion is additional heat is required to
explain the flight data. So I think that's a point at
which we need to challenge the model because then I
would ask the question: Does your model include a
directional jet or is it what I would call a heating
and vent kind of analysis that you would use if you
were trying to build an air conditioning system for
your house? It's kind of a different kind of analysis.
MR. MADDEN: And this model is certainly
challengeable because this is an engineering method
where we just broadcast. All we know at this time is
that this amount of BTUs per second came into the wheel
well. How it's distributed, we have to wait on CFD.
So at this point all I was trying to say was: "Can be
ADM. GEHMAN: Maybe I misunderstand, and
I'd like to understand it. What I read from this,
though, Sheila, is that this graph supports your
position. What I mean is that just by the model he has here, which he has a healthy wheel well with nothing
broken except heat's getting into it, works for a few
seconds but then after that it doesn't work anymore.
DR. WIDNALL: Right. I think that's
right. It's just that when you see something like
this, you really have to make sure that you understand
the model and that it's pointed out that the model
itself is the simplest level of calculation that one
MR. MADDEN: Sure. Excellent point.
This is a very simple energy balance type analysis.
ADM. GEHMAN: One of the things that I'm
really interested in, of course, is that I'm interested
in the very first off-nominal reading.
MR. MADDEN: And this is it.
ADM. GEHMAN: I understand that, but you
have little red dots here that show flight, actual
telemetry data. Of course, you didn't put all of them
on there; but you've been monitoring that temperature
MR. MADDEN: Right.
ADM. GEHMAN: So the point is that you started here at this is EI. Is that correct?
MR. MADDEN: Correct.
ADM. GEHMAN: So you started here because Steve Labbe, Joe Caram, Chris Madden - March 18, 2003
that's kind of where the interesting part is.
MR. MADDEN: Right. It had been decaying
down slightly; and you see that in this plot, too.
ADM. GEHMAN: My question, though, is
that because these temperatures were essentially
nominal and even though you are in extraordinarily thin
atmosphere with very few air molecules, the orbiter is
heating up out here.
MR. MADDEN: It's heating up on the outer
ADM. GEHMAN: Yes. Where does peak heat
start? Do you know where peak heat starts?
MR. MADDEN: At about 300 seconds. Okay.
What you're seeing is inside the well wheel.
ADM. GEHMAN: I understand that; but I'm
saying even back here at 200 seconds or 250 seconds or
300 seconds, even though you're not at peak heating, as
the orbit decays, as the orbiter comes down, the
heating increases. External heating.
MR. MADDEN: Okay. Correct.
ADM. GEHMAN: And so what I'm trying to get at is whether or not we should feel that whatever access allowed the external heat to get in, whether it was a preexisting condition or whether it started whether that access opened right about here.
MR. MADDEN: That's challengeable. Whether or not that this was the first bit flip just because the hole was there the whole time in the wing and you just see the bit flip just because that's the period of time it took for this low-density gas to raise a high-density brake line to 1 degree.
ADM. GEHMAN: Or if just 1 or 2 seconds
or 10 seconds before here is when the fault manifested
MR. MADDEN: Right.
ADM. GEHMAN: We don't know.
MR. MADDEN: We don't know; and that's
why we're making these assumptions, to see if the whole
story fits. If it doesn't, we'll have to revisit
ADM. GEHMAN: Are we on the same sheet of music here? In other words, in my mind I don't know. And, of course, it bears on a lot of things because if the fault just manifests itself right here, even though the aerodynamic pressures are practically nothing but might be enough to remove something or cause something that was weakened, then all this stuff about on-orbit photography and stuff becomes irrelevant because if there was no fault that you could see I mean, it was a weakness clearly and something failed. So, I mean, it's important to know whether or not the orbiter had a preexisting condition that started, you know, way back over there, which then didn't manifest itself heat-wise until you got enough heat.
MR. MADDEN: Right. There is another
piece of analysis that we don't have in our charts that
we did make that assumption that, okay, let's say the
hole was there the whole time. Those transients, the
analytical transients didn't really jump up. There's
no reason for them to jump up at that time. In fact,
it wasn't enough heat for them to really respond until
out here at 7 or 8 hundred seconds. That's another
little piece of data that kind of suggests that
something happened there. I'm not saying it's a fact.
ADM. GEHMAN: Right. If I could ask Steve a question here. Back to this first graph over here. You say that these numbers 'cause they're ratios and they're ratios of irrelevant numbers at that particular time but because of this bias that the orbiter had in its control surfaces, where, compared to 400, 300, 500 seconds after EI do you start believing your own data?
MR. LABBE: For what's on that particular
plot, I would say it's more like 500 plus seconds.
ADM. GEHMAN: Yeah. So it's right in here.
MR. LABBE: It's close to that, but maybe
a little bit further, maybe another 20 or 30 seconds
ADM. GEHMAN: So even though we have an
indication of temperatures, we have another indication
of what the aero surfaces were doing that are to the
left of whatever, this 480 seconds after.
MR. LABBE: So it's close. I would say
if you look at that plot where you see the downward
trend where you see the slope really go away from zero,
right there, that's where I'm saying I have a clear
indication. What's happening before that...
ADM. GEHMAN: That horizontal but left
bias, you're less confident.
MR. LABBE: I'm less confident because
when I did STS 109, I got very similar results.
ADM. GEHMAN: Very similar things.
MR. TETRAULT: With regard to bit flips I'm going to go back to this that's the indicator that shows that you're going off nominal. Can you tell me, off nominal to what? Is that the average for that STS for that orbiter in terms of prior history? Is it average of the entire fleet? What is it off nominal to?
MR. LABBE: It's off nominal to what
would be to our data base, which is for the entire
fleet. So it's off nominal from previous flights.
Now, we haven't gone back and applied this analysis to
every single flight; but what you would expect to see
again is even if you had a slight bias down like that,
was that that would stay there, maybe drift back
towards zero. It's not going to get significantly away
MR. TETRAULT: This is important because a slight change in when you make a call of what's off nominal can change the entire time line of where the heat is coming from. So I would like to continue to explore this just a little bit. In terms of when you make that call and I've looked at some of these plots that we have and they appear absolutely straight to me and all of a sudden there's a call that it's off nominal how accurate do you feel that call that it's off nominal is?
MR. LABBE: Okay. I think what we've done and what's not shown here is you look at the rolling moment, you look at the aileron response, you look at the side slip
MR. TETRAULT: I'm talking about
off-nominal calls on just temperature sensors.
MR. MADDEN: Well, the previous missions have they've always kept decaying down. Although the surface of the wheel well on the door is being heated to very high temperatures, that heat soaked back into the structure of the door and then, via radiation, into the brake lines. It doesn't occur until much later. So this we're pretty confident was a beginning of off-nominal event. There have been bit flips before, but they've always kind of came back down. So the typical response is a downward trend and you might see a flip up but it would come back down and stay down.
ADM. GEHMAN: Of course, you have the
same measurement in the right wheel well.
MR. MADDEN: Correct.
ADM. GEHMAN: Which doesn't show anything
MR. MADDEN: Right. And it does the
typical decaying down until much later.
Next chart, please. The next few charts
are the quick assessment of how extensive the tile
damage would need to be to burn through the skin of the
wing. In this case we can predict, and what you're
seeing is temperature versus time for the outer face
sheet and inner face sheet of the sandwich. Our simulations can predict the burnthrough in this case is
Next slide, please. This shows it on the
landing gear door; and this, based on the configuration
of the structure itself and the heating rates and
heating factors and the size of the damage, it's
earlier. That's more around the time where the breach
was observed. I'm not saying it's the door. I'm not
saying it's the wing. It's just showing that it's
highly dependent on the damage you have to assume.
Like I said, at some point there's going to be enough
damage to burn through the wing.
GEN. BARRY: Chris, let me ask you a
question on temperature inside the wheel well. What's
your best guess, if you have any, of the temperature
getting about 700 degrees? The reason I'm asking that
question is the pyrotechnic inside the wheel well is
supposed to cook off at about 700 degrees.
MR. MADDEN: There is massive pieces of
structure from the flight data on the strut actuators
that don't rise over, I think, 120 degrees or so. You
would think that the pyro would be the same order of
magnitude. We will have a chart. I'm very unsure of
the math model. We have a math model of an entire wheel well, and we will confirm that the pyro didn't go
GEN. BARRY: Did not go early.
MR. MADDEN: Right now I think it's very
DR. WIDNALL: I have a question. You did
a calculation of burnthrough of the skin, and obviously
what's the skin made out of?
MR. MADDEN: Aluminum.
DR. WIDNALL: You know what I'm going to
ask. What sort of boundary condition did you use for
the surface catalysity and/or reactive behavior of the
MR. MADDEN: The reactive behavior was
not simulated. There's no oxidation for those
DR. WIDNALL: So you basically got a
melting hypothesis as opposed to burning.
MR. MADDEN: Right. And thermomechanical
effects were not simulated. So we're just trying to
see can you get to the melt temperature; and, of
course, you can.
Okay. This is analysis of the thermal
barrier and pressure seal around the door, if the tile
adjacent to the thermal barrier is severely damaged and you basically expose that cavity in the pressure seal
to the external environment. You see two different
assumptions here, but basically they both do the same
thing. The pressure seal will fully demise a little
before 500 seconds. So again, bad enough damage, you
can breach the wing. And this one is via the wheel
well. We're not concentrating on this one so much
anymore because of the timing between the wire burn and
the pressure or temperature rises seen in the wheel
Next slide. Okay. This is analysis to
explain the side wall temperature rise. What we did
here is at 600 seconds we applied ten times the normal
convective heating environment to the exterior of the
TPS in this region; and that, we actually
back-calculated it ten times. That shows that the
analysis can predict the flight data with ten times the
heating rate to the surface. That ties into what Joe's
studies have done. His team has shown that you get a
bump factor two to ten times. This is at the upper end
of that, but it's the correct order of magnitude and in
the same ballpark. So the conclusion here is that this
could be explained by external heating due to shock
hitting the side wall.
GEN. BARRY: But it could be explained by convective heating.
MR. MADDEN: Internal convective heating.
There is enough heat, if it's distributed to this zone.
We don't at this point know how the air flows within
the mid fuselage and whether or not it would make it
back to this region and heat the back side of the
sensor, but certainly it is possible.
DR. WIDNALL: Can I ask a question? I
know I could do these calculations myself, but are you
also considering thermal conductivity through the
MR. MADDEN: Right.
DR. WIDNALL: So that's part of it.
MR. MADDEN: Right. We looked at that,
the conduction effect. We looked at a very hot wing,
can it conduct up to this sensor quick enough to see
this response; and it really couldn't conduct fast
enough to see this. So conduction was ruled out.
Next slide, please. Okay. This is just
to show you that we've also been looking at leading
edge damage. The Huntington Beach guys have developed
math models of damaged RCC in support of the
micrometeoroid studies that were performed. They've
used those techniques to assume a hole size and then
they simulate the thermal response of the insulation Steve Labbe, Joe Caram, Chris Madden - March 18, 2003
and the fittings and the spar and they're able to show
that Panel 9 with the four to six initial holes
starting at the beginning here could burn through the
spar within about 500 seconds in other areas where it
wasn't predicted to burn through and oxidation was not
DR. WIDNALL: Do we know what the
catalytic properties of that pillow insulation are?
MR. MADDEN: That's Inconel covered, and
it's likely catalytic.
DR. WIDNALL: Okay. Was that considered
in their analysis?
MR. MADDEN: The plume, yes.
MR. CARAM: Anything that was applied was
DR. WIDNALL: Fully catalytic.
MR. MADDEN: I'm glad to make it to the
next chart where you've got a couple of bullets on
chemistry. As you pointed out, we didn't design for
aluminum to be in this atmosphere. So areas where we
have addressed it is for reentry of space debris. We
have done some co-development and studies of that, and
we have included chemical convective heating in those
simulations. It's an engineering method where it's
basically ratioed to the heat rate and the heated formation of aluminum oxide.
I ran that code when I understood you
were curious about this. This simulation is a
ballistic trajectory. This isn't the shuttle flight.
It's just an aluminum sphere on a ballistic or reentry
flight; and it's showing that the heating due to the
oxidation of the aluminum, assuming it's bare, was
0 percent of the total heating. And it's pretty
constant the whole way up. I also included aluminum
nitride formation. That's exothermic as well, and that
was another 7 percent. The assumptions that went into
this analysis assumed that all the available oxygen and
nitrogen contributed, all of the heat from the
exothermic reaction itself is liberated to the surface
and not carried on into the flow. So a worst case, if
you will. So in engineering terms, it's a fairly small
percentage of the total convective heat, at least for
this case, a sphere.
DR. WIDNALL: I obviously want to look at
that more closely.
MR. MADDEN: And I do and I will point
that out. So if you're looking for a reason why, you
know, that's one of the reasons why.
The Koropon could also hinder it while
the debris still had Koropon on it. That likely goes away at 400 and so degrees, though. Then I here try to
point out that, well, the aluminum oxide could
self-arrest basically and perform a protective coating
on the surface of the aluminum and knock that chemical
heating back down. I don't know how much of that
happens. I'm certainly not an expert in that area.
DR. WIDNALL: Well, it was kind of
interesting because yesterday we got a very different
picture from the reentry of, what was it, a steel tank
from the Delta 2, I guess.
MR. MADDEN: And that was Dr. Ailor
pointing that out. And I think it was a titanium tank.
DR. WIDNALL: Right. Some tank made out of
MR. MADDEN: Right. Titanium, I think
the reactions there are an order of magnitude higher in
terms of heat.
DR. WIDNALL: No, but I think what was
pointed out was that aluminum layer deposited on a
titanium tank would act as a fuel and destroy part of
the tank that otherwise would not have been destroyed.
MR. MADDEN: That's certainly
interesting. The titanium use on orbiter is very
limited. I think it's limited to pressure lines,
hydraulic lines and things like that. So I'm not so worried about any titanium reactions with hot aluminum.
I do want to check into this more, along with some
other pieces of physics, and see if we really
understand how holes grow in aluminum. Right now it's
been real simple engineering.
Next slide. We also understood you were
curious about the catalytic heating. As Joe
summarized, atomic recombination effects are going to
be probably more significant than chemical heating. A
lot of times, it's a 30 to 40 percent bump factor. An
aluminum surface will act as a catalyst and encourage
this recombination and liberate additional heat to the
surface. A lot of times this could be 30 to
40 percent, if you're using finite-rate chemistry
calculations. In our cases, for the plumes we're using
equilibrium heating; and that's very close to fully
catalytic, anyway. So I think in terms of catalysis
and the plume heating analyses and analysis we're doing
on the plate burning, we've already accounted for
Okay. And then these points, I just
wanted to point them out. The extent, to my knowledge,
is pretty limited here; but things like auto-ignition,
the studies that you see in the literature, I think, a
lot of times it's at very high pressure and what's
called oxygen-rich environments. Here I'd have to say
we're oxygen poor; and you certainly, as you descend in
an atmosphere during post-breakup, you're going to see
these effects probably a little more enhanced than you
would in the early part of the flight which we're in.
Like I say, I do want to address the
oxidation, just to make sure we understand what's going
on there. The melting, of course, any ignition effects
and any sort of vaporization or sublimation of the
aluminum. So we're going to get a team, a group
together to address that.
Okay. The next slide. I just want to
summarize. The work we have going on now is what we
call our engineering methods phase. Again, we're kind
of concentrating in this area of time here. If you
notice, we're talking about bit flips. Okay. So we're
trying to explain. The ability to explain these bits
flips before the configuration really goes chaotic
after 600 seconds, 700 seconds, you know, it's a tough
So what we had to do was make these big
assumptions like we've gone through. The bit flip at
is a breach in the wing. The wheel well rise at
0 seconds is a breach of the wheel well, and the
off-scale low is a burning of the first cable. That's very likely. But these two, admittedly, they're
engineering leaps we feel we have to make to create
knowns so that we have the same number of equations and
unknowns through our solution space so we can get
solutions and argue about them and refute them and
discuss them and see if they make sense.
ADM. GEHMAN: But the second assumption
there, the wheel well temperature rise around
600 seconds is a breach of the wheel well, that doesn't
necessarily mean it's breached through the door,
MR. MADDEN: Correct. And for these
solutions, we're going to try to breach the wall, the
internal wall. Okay. And what that means in terms of
brass tacks? Burn a cable in 42 seconds. We're going
to figure out how to do that and the wheel well wall in
12 seconds. Again, we'll cross-check the aerodynamics
and the forensic data.
These charts are a sample. There's a
whole series of analysis we're doing on varying the
distance away of the hole size. There's a lot of
parameters. So I just wanted to show a sample of a
plume being applied to a flat plate; and we get a
temperature response on the next slide, 46.
For various hole sizes, you'll see the temperature transients versus time; and the ones that
exceed the aluminum melt temperature in around
0 seconds are going to go into the next series of
plots on the next page. You see that show up right
there. That would be hole size you need. In this case
the spar and the distance away it would need to be to
burn that wall in 112 seconds. From this distance
away, we can go and look at each panel. Okay. This is
Panel 5 region. The hole size needs to be 3 inches.
Okay. Then now we are going to cross-check that to the
wire-burning analysis and also the aerodynamics.
Next slide, I think, is the wire burning.
Here it's kind of explaining how the cables of what we
call the bundle, which is the whole series of wires
that you see in the pictures, those consist of smaller
harnesses and then cables. So we're developing this
math model and correlating it to some burn tests that
were performed to make sure that we at least
macroscopically and engineering-wise can predict when
these cables fail.
The next slide shows some initial results
from that type of analyses. You see the time to
failure and the distance away from the plume. These
types of data will be compiled into very similar plots
that you saw for the flat plate, and they'll be cross-checked to see. Because we have to burn a wire
in 42 seconds that's right next to a wall that we burn
in 112 seconds, assuming we just have one plume. So
we'll make sure that those make sense with respect to
GEN. BARRY: Chris, the wiring you're
burning is Kapton wire, right?
MR. MADDEN: Correct. Kapton coated.
DR. WIDNALL: Another question. You are going to run some experiments on Kapton. Are you planning to run any experiments, say, with an arc jet with dissociated oxygen and the right kind of
MR. MADDEN: Yeah, we're starting to
think about arc jet tests.
DR. WIDNALL: of aluminum plates or honeycomb or structures and compare that with your analysis?
MR. MADDEN: Yeah. I guess two things. We started thinking about arc jet tests for burning the wires. It consists of a test where you have a hole, you blow the arc jet gases on it and see how fast it burns the wires. Coupled with that test, we could look at we were initially thinking of having that hole in the plate that the hole goes through water-cooled, but we could do tests where we
DR. WIDNALL: Basically burn it.
MR. MADDEN: cool it and see how fast it grows. So we certainly should think about that.
GEN. BARRY: Let me ask you about the
assumptions on the wire bundles. We understand that in
Columbia it was different. In the well wheel area,
there were like four large bundles as opposed to the
other orbiters have like seven; and there's a lot of
wires in there that were disconnected that didn't go
anywhere because they had been disconnected from
sensors over the years.
MR. MADDEN: Right.
GEN. BARRY: Did they have the right
diameter and the right combination?
MR. MADDEN: Well, we think so in terms
of diameter. The cables that are in those bundles, I
think there were only seven that were being recorded;
and all seven of those eventually failed. Where they
were within the bundle is unknown. So that's another
thing we have to deal with. What we're going to do is
assume that some of them are very embedded into the
bundle and assume those are the ones that go later and
slower; and, in fact, the tests that the guys at JSC
are performing on the burning include the effect of
being inside the bundles.
GEN. BARRY: When you do the testing, is
it going to include not just going to the center of the
bundle but going through the different sides of the
circumference, I would assume?
MR. MADDEN: Well, it's got to hit a
GEN. BARRY: But it could be at an angle
and not go right to the center, is what I'm saying.
MR. MADDEN: Yes. Of course. What we
have to assume here is that the plume is hitting, is
smart enough to hit cable. And that's likely not the
case but it's certainly bounding. This is going to
give us the farthest distance away that the hole in the
skin needs to be to burn that cable in X amount of
time. If it's off axis, it would have to be closer in.
GEN. BARRY: Or hotter.
MR. MADDEN: So we'll be able to determine a region that could exist
ADM. GEHMAN: Or bigger or hotter.
MR. MADDEN: Yes, sir.
MR. TETRAULT: Let me go to the RCC
panels. As I understand it, you've run two thermal
analyses, one on a 4-inch hole and one on a 6-inch
hole. Why aren't we looking at things like T panels
and an entire RCC section?
MR. MADDEN: Let's see. The cases we're
running for thermal analysis were holes in the panel.
Why aren't we looking at missing panels?
MR. TETRAULT: Yeah, or T sections. Does
anybody know what the equivalent size of a missing
T section would wind up being, if you look at the line
that then becomes available for air to pass through?
MR. MADDEN: With the missing T seal? Of
course, that's a function that's to protect that gap
between the panels.
MR. TETRAULT: Right. So what would the
gap be in an equivalent hole size?
MR. MADDEN: You still haven't breached
the wing in those cases. And there's a whole other set
of analyses that kind of the earlier part of my slides
that were trying to explain how do you get from the
entry interface to the letter A.
MR. TETRAULT: It depends on how all the
RCC panels line up. In fact, if the T seals are
missing, it may give you a gap.
MR. MADDEN: Well, it will give you a
MR. TETRAULT: A gap in the leading edge.
MR. MADDEN: But not the spar.
MR. TETRAULT: Not at first.
MR. MADDEN: Correct.
MR. TETRAULT: I'm trying to compare a missing T seal to the analysis that you've run based on a 4-inch hole or a 6-inch hole. I mean, what kind of
ADM. GEHMAN: Order of magnitude.
MR. TETRAULT: Is it less than a 4-inch
MR. MADDEN: I would say it's less than.
And, Joe, would the heating effects be
reduced because it's not concentrated?
MR. CARAM: It would be distributed around the leading edge panel. The T cell, as I recall, is about a quarter-inch thickness. So you have to fit it in between two panels. So you're talking three tenths, four tenths of an inch in thickness for a gap. So then you have to account for the area around the circumference of the leading edge. But the characteristic dimension would be your smallest dimension. That would be the size that would dictate the size of the jet that you're getting in between there, would be the slot width and not the circumference area.
MR. TETRAULT: Just one other comment.
You talked about your calculation on the wheel well seal and it's probably not as significant at this
particular point as it might have been. But if you
look at it from the fact that the heat and the pressure
did enter the wheel well and then escaped out the
corners, as the debris seems to indicate, then the seal
had to have failed at some point in that. So it may,
in fact, be an important number at some later point, so
put that in your time line.
MR. MADDEN: Maybe so. But the debris
that you see, the evidence you see in the debris is an
outward flow. And that would obviously come from
higher pressure on the inside and erosion from the
MR. TETRAULT: Right. That's exactly
what I've said.
MR. CARAM: Which meant you already have
the penetration into the wheel well and the damage is
done at that point.
MR. MADDEN: And we're talking about
areas out here now in terms of time and we're really
trying to figure out what the condition was right here.
DR. HALLOCK: Have you been looking at
the fact that when you have the roll reversal occur, it
looks to me that we are seeing the plume actually
moving grossly and just because of the fact that it has gone back now and is on the left wing, which may sort
of start some of these calculations all over again in
different locations? Have you seen that effect, or are
you looking at that issue?
MR. CARAM: Again, when the orbiter is
flying, it's typically the heating distribution around
the vehicle is dictated not by roll angle but by angle
of attack and angle of slicing.
DR. HALLOCK: Now assume that you're
actually in the wing itself at this point, so you do
have this plume moving around, trying to figuring out
where to go. If you look at before and after when you
do get the roll reversal, it's a different regime where
some of the problems are happening. In one case you're
seeing the shorts of the wire and in the other regime
you're starting to see all the temperatures starting to
change. It's as though the plume was at one point and
then when it completed its roll, it's suddenly pointing
somewhere else and finding a new path.
MR. CARAM: Well, the wire bundles that
he's talking about run right alongside the wheel well
wall, on the outboard side of the forward bulkhead. So
the plume doesn't have to move around much to get to
MR. TETRAULT: Let me be sure you understood the comment I made last time. On the
corners, if there's a vent that's there, the wheel well
door had to be there, otherwise the vent wouldn't have
occurred. So breaking the seal and the time line for
breaking the seal may play into your overall scenario
to tell you how long the door was there. So I just
wanted to be sure that you understood the comment that
I was making.
MR. CARAM: Valid point.
GEN. BARRY: Let me put you on the spot a
little bit. Now that we've gone through the analysis
that you've gone through on a basic attempt to put all
this together synergistically, what can we eliminate as
an entry point for the heat? If we follow the heat,
what can we eliminate right now as an entry point?
MR. MADDEN: We diamonded the door.
Okay. From the list of scenarios that the team at JSC
has come up with, there are several of them that we
called diamond; and we basically tabled them and
concentrated on three or four scenarios that we felt
were more likely. One of the ones we diamonded off was
any sort of breach through the door. The main reason
for that was the wires. If you see in a time line, the
first wire was burnt before you see hardly any
temperature rise in the wheel well. So for a jet to find its way through the wheel well, out a vent, and
find a wire and raise that to 900 degrees before seeing
any indication in the well itself, we felt, was quite
unlikely; and so we are tabling those sorts of analysis
at this time.
GEN. BARRY: But you haven't tabled
either in front of the main landing gear under the
wing, either in front or behind it, but you have
eliminated on the main landing gear door.
MR. MADDEN: I wouldn't use the word
"eliminate." Probably we might get ourselves into
trouble reporting this, but I'd let the shuttle program
maybe answer those types of questions.
GEN. BARRY: Okay. We're getting close.
ADM. GEHMAN: Okay. Anybody else?
Well, thank you very much, gentlemen. I
appreciate your patience with us today and your energy
and the zeal and the professionalism by which you are
approaching this. We admire it very much.
I've made several notes here. Several of
the board members have mentioned what about this and
what about that and what about this and the other
thing. It occurred to me that we are now at the point
where some of these future tests should be mutually
agreed upon because if we have some favorite scenarios that we want explored, we should let you know about
that so you can take them into account when you're
designing tests and things like that. So I think
that's very important.
The second area that I noted is the area of the initial assumption concerning a breach. It's not clear to me and I don't want to settle it right now, just in the interests of time but it's not clear to me how you have a scenario, the real scenario, the data from the Columbia, which suggests to me a changing geometry, and yet what we're trying to do is take a single event and backtrack it. In other words, you take a 4-inch hole. Well, it might have been a 4-inch hole at one point, but it might have been a half-inch hole and an 8-inch hole later on. So I'll have to reconcile in my head how you propagate a casualty over time versus one of those graphs. I don't want to get into it right now, but I think it's very interesting.
I would like for you to also pass on to your colleagues I know that you represent the tip of an iceberg of a lot of people who are working very, very hard and diligently to try and solve the riddle of this tragedy. We realize that, and I would like to have you pass on to all of your colleagues our admiration and our thanks for all the work that they are doing. They don't get to go to press conferences and things like that like we do and they don't get a lot of notoriety, but I know how hard they're working and I know how hard they want to solve this, too.
Thank you very much. You are excused.
And we will call John Bertin, if he's here, and we'll
go right to work.
testified as follows:
ADM. GEHMAN: Dr. Bertin, welcome. Would
you please introduce yourself and tell us where you
hang your hat and what you do for a living.
DR. BERTIN: On Continental Airlines,
coming back and forth to Houston.
When I graduated from Rice with a
Master's, I went to work for the manned spacecraft
center across the street and then got my Ph.D.
part-time at Rice and went to UT Austin and taught for
0-something years. I did some research on the shuttle
before it flew. Did some things with reentry heating,
tile misalignment, shock-shock interactions. And then
after it had flown, we did some analysis on asymmetric
transition and anomalous findings from some of the flights, with some of the people who have been giving
the presentations up here today. After my kids grew up
and they were all out of the house, I left Austin and
went to Sandia for a few years; and I teach now at the
Air Force Academy.
ADM. GEHMAN: Thank you, sir.
DR. BERTIN: Can you get 18 up here? For
I thought since we talked about
temperatures and we talked about catalysity and we
talked about some in degrees Kelvin and some in degrees
Fahrenheit and some in degrees Rankine, I thought what
we might do is talk about the flow field in general,
with one set of nomenclature and what have you.
So if you look at the orbiter coming in
in this orientation, it's at approximately 40 degrees
angle of attack. So the velocity vector is coming in
like this and the flight path angle, and it's not
rolled or yawed or anything like that. It's at an
angle of attack of about 40 degrees. So you see it in
this picture here.
Okay. This is a wind tunnel test and
they talked about Mach 6 in the wind tunnel and it
didn't do this and it didn't do that. So let's look at
and talk about Mach number and hypersonics and some general features. So if you're going to be flying in a
vehicle in the atmosphere, the Mach number is going to
be velocity over the speed of sound, whether you're in
the wind tunnel or in the atmosphere. So in the
atmosphere, no matter what altitude you're at, the
speed of sound is about a thousand feet per second. So
the Mach number is about the velocity at which you're
flying in thousands of feet per second divided by a
thousand. So if you're at Mach 6 in flight, you're
flying 6,000 feet per second.
Now, there's a lot of kinetic energy in that flow and as the flow so if you're doing a wind tunnel test, to have that much energy, you damage the wind tunnel. So what they do is they run the speed of sound down to where it doesn't simulate the same gas chemistry. So the gas chemistry in a wind tunnel is very, very different than the gas chemistry in flight, even though both flows are hypersonic. Okay.
So now the vehicle is flying along at,
say, 6,000 feet per second. It's at an angle of attack
of 40 degrees. Why isn't it flying at a low angle of
attack like airplanes, which fly about like that,
right? Because the heating goes as density to the
one-half velocity cubed divided by the bluntness.
Since the velocity cubed is large, the heating is large. So what you want to do is counter that by
giving as blunt a vehicle as you can. Okay.
So as the vehicle is flying through the
air, the air is coming rushing along at 6,000 feet per
second and it has to turn to go parallel to the flow's
surface. To do that, it goes through a shock wave.
I'm sure you've heard the witnesses talk about hearing
the sonic boom. The sonic boom is caused by that shock
wave. See this thing going up here? That's the bow
shock wave. Now, it decelerates and turns the flow.
So as the flow decelerates from a high kinetic energy
flow to one of low kinetic energy, the temperature's
going to go way, way up.
So the temperature is going to be the
atmospheric temperature up here a few hundred degrees
and it's going to be much, much higher back here,
depending upon what part of the vehicle you're in and
where you are. But in this region near the nose,
you're going to see the highest temperature and we'll
use equilibrium and we will use degrees Rankine.
You're going to see temperatures of 10 to 12
thousand degrees Rankine.
Now, obviously there's going to be some chemistry going on with these kinds of temperatures and you're going to see a strong shock wave. So the density's going to be changing very dramatically. The pressure's going way up, the temperature's going way up. Density is changing very dramatically. And you know how when you look in water and a fish is here but it looks like it's over here because the light rays are bent? Well, that's what's happening here. The light rays are bent. They pass light rays through the tunnel, and the density changes allow you to see the light being bent. So you can see the density changes downstream of the shock wave; and they're caused by, like I say, the pressure changes and the temperature changes and what have you. So up here the temperatures are on the order of, say, 10,000 degrees, maybe ,000 and again there's some chemistry, there's some non-equilibrium, there's some things going on.
Now, the flow expands around from the
nose. Just like when you put your hand out the car
window and stuff like that, you feel the force on your
hand. Well, the flow accelerates as it goes around
your arm, right, and you feel the velocity. You drive
down the street and in the windshield you see the
stream line patterns taking place in running rain
across your windshield. So there are stream line
patterns coming here and the flow accelerates to where
the temperature of the air in this region is more like to 8 thousand degrees Rankine because the pressure
has dropped. The flow is accelerated. The pressure
has dropped and the temperature has dropped. So 10, 12
thousand up here. Very high heating rates because it's
got a small nose radius and temperature dropping 6 to 8
thousand outside of the boundary layer.
Okay. What is a boundary layer? If
you've ever gone to the beach on a cold wintery day, if
you're late and it's sunny, if you lay down, you feel
relatively warm. If you stand up, you feel much
colder. Right? Well, what you're feeling is the
change in velocity as it goes from the surface to a
much higher velocity a few inches away from the
surface. So that's a boundary layer. So you're going
to get shear going on in that. Just like when you rub
your hands together, you're going to get heating.
So the boundary layer causes the air, by
rubbing against each other and fluid particles, to give
you more heating than you'd expect from just a few
thousand degrees. So you want to protect the vehicle
from this heating. Okay. So you put high-temperature
materials along the parts of the body that have a small
radius, like the nose and the wing leading edge, and
you can use less robust materials over areas where the
I said the heating varies as rho to the
one-half D cubed divided by the bluntness. Well, only
a fraction of that energy actually gets transmitted
into the vehicle. Most of it goes flowing past the
vehicle in the air stream et cetera.
Now, we talked about the tiles. The
tiles fill most of the area here, and they're black.
They have a thin coating, and the thin coating does
several things. We talked about the catalysity and
non-catalysity. So the thin coating is non-catalytic,
but it's also like Scotchgard. If you look at the
thing, if the vehicle's sitting on the pad and it
rains, if the tiles didn't have the coating, they'd
soak up a lot of the water. So the coating prevents
some of the water from getting in.
If you go back to your freshman physics
course and you did the little heat transfer thing, the
energy coming in can be radiated back out, right? And
if the energy is radiated back out, what's the best
color for radiating outward? Black. So the coating is
a thin, black coating that gives you several type
features; and it goes to a much lower temperature.
ADM. GEHMAN: Let me ask a question.
Back to wind tunnel, if is this a good time to talk
about wind tunnels. As I understood you, the way they achieve the very, very high speeds, the very, very high
Mach numbers without tearing the wind tunnel apart is
by changing the gas in the wind tunnel to where the
speed of sound is a lower speed of sound.
DR. BERTIN: They run the wind tunnels
where the speed of sound is an order of magnitude, or
close to it, lower than would be normally in the normal
ADM. GEHMAN: So they're using some other
DR. BERTIN: Or they're taking the air
and causing the pressure and temperature to drop way
low. The temperature is just above liquefaction of
oxygen. If you ran the tunnel any differently, you'd
get liquid oxygen going down your tunnel.
ADM. GEHMAN: So any other properties,
then, of the results that we should be suspicious of?
DR. BERTIN: That's going to give you
some changes in the density ratio. And the density
ratio, I think Dr. Widnall talked about how the shock
wave is going to change its inclination as you go up in
Mach number. It's going to change its inclination as
you go up in density ratio. And the density ratio in
the flight case is near 20. Maybe 12, maybe 15,
maybe 20. The density ratio in the wind tunnel is going to be 6. So it's going to have a much different
shock structure. We're going to talk about that in
terms of the shock-shock interaction in the Kirtland
ADM. GEHMAN: That was going to be my
next question. I'll wait.
DR. BERTIN: Okay. So we have these
things going on. So the wind tunnel is just a
simulation of parts of the flow, and what you want to
look for is some general overall things that you can
then compute and then correlate them in some fashion.
Okay. So we have basically now these tiles over much of the surface, and they're giving us many features. We've got carbon-carbon along the wing leading edges, and they go to higher temperatures. So the boundary layer is relatively thin, maybe a few inches by the time you get to the end of it, and so the flow going over that adjusts into from the zero velocity, the Mach 2 or 3 locally, so the Mach number in this region is supersonic. So if you had a disturbance way down here, it would not feed forward.
You asked the question about would any of
this explain what happened to the water being dumped.
I don't think so. That may be a problem, but it would
be a different function because the disturbances won't propagate upstream unless you have some strong shocks
that make the flow subsonic.
ADM. GEHMAN: Well, the reason I asked the question was because one of the gentlemen said that in experiments with the body flap that they had the first time they entered, they had the wrong pitch set in the body flap and when they started moving the body flap, there were some changes in the shock pattern, the properties of flow.
DR. BERTIN: There will be some changes
in the shock pattern, but they'll be limited to the
region within a few distances of the body flap. So
they can propagate upstream because you're having shock
waves, but they won't propagate unless you've got a
spectacular flow. They won't propagate very far
upstream. So you have that.
Then if you look at the model from this
standpoint and you rotate it about its velocity vector
so it's still a 40-degree angle of attack, if you look
at this picture, it's going to have a shock wave over
the bow, the fuselage, the nose region, right, and the
shock wave is going to wrap at fairly close angle, like
this. If you imagine that you just rotate the model
from like this to like this, you'll have shock waves
that occur that kind of envelop, form an envelope over the fuselage.
What's going to happen when those shock
waves reach the wing leading edge? Because the bow
shock wave will be at about this point on the body,
right? So what's going to happen? There's going to be
a shock wave set up for the wing leading edge. And
when the shock wave from the bow shock wave intersects
the shock wave from the wing leading edge, you're going
to get an interaction that could cause the heating to
go up, depending upon what the sweep of the wing is
relative to the oncoming flow. So it works out where
this kind of delta-wingish type thing has relatively
low severity in the shock-shock interaction. If the
wing were onswept, you would have great severity in the
shock-shock interaction because you're taking a flow
going this way and causing it to intersect a flow that
has a much stronger shock that's going this way.
So if you are missing maybe not one panel
but maybe two panels and maybe it's downstream from the
initial column that you had and stuff like that, then
you've got like two teeth missing from the leading edge
and you've got a little notch in there. Now the flow
can go in that notch and create a shock pattern that,
in my mind, kind of looks like what the Kirtland
photograph might be telling you, in that something is not missing, something is added. And it could be the
density gradients of the shock waves in a shock that's
been changed shape because you've had some damage that
has grown in time. So that would explain some of the
Then the other thing is, if you look at
airplanes flying in high-humidity air, the pressure is
higher on this side, right, in general, and lower on
this side because you're generating lift. So when you
get to the wing tip, you form a vortex, wing tip
vortices if you're a pilot for the trailing weight and
counter hazard. If you're an engineer, you've got
these beautiful pictures and wind tunnels and stuff
like that. If you're in CFD, you've got beautiful
pictures and colorized computer outputs. But for a
variety of reasons, you have a vortex. And the vortex
is basically a horizontal tornado and the velocity can
be very, very high speed and circulating, just kind of
like the flow going down your sink or a tornado that's
being spawned by a front coming through and stuff like
that. So you look at that.
Now, if that tornado came from someplace
in here through your gap in the shock wave, not only do
you change the shock wave out here but you get the
possibility of some kind of vortex coming and striking part of the vertical fuselage. And it could be only
I remember back when I looked at the data from the Gemini project, the GT2 was an unmanned test vehicle and the Gemini had umbilicals that brought the electronic wiring from the booster into the command module or the spacecraft and the umbilical the Gemini came in at a slight angle of attack and a vortex pattern that had been set up by the flow over the umbilical caused minute holes to occur in the surface of the Rene 41 of the Gemini and they had little holes.
So the vortex can be very localized and
it can be very hot, depending upon where it touches
down and how much it touches down and what the shape of
the vehicle is. But you can see a progressive
situation where if you lose a panel or two, you'll get
a vortex that could scrub the vertical surface and you
get a shock that forms with the shock-shock interaction
that creates the image of something that is different
than just the main planform of the vehicle. I say you
could, 'cause I need more looking at that.
Okay. Is that kind of good as far as overall as far as where temperatures are high and how they change and what they do?
ADM. GEHMAN: You covered this but I want to be sure I understand that when we're talking about these boundary layers, in accordance with this picture back here, for example, we're looking at boundary layers which are kind of spreading apart and are measured in tens of inches or something like that toward the tail but at the nose we're talking about
DR. BERTIN: It's going to grow. And this is a 100-foot long vehicle. So it will grow over the length of the vehicle so that it's, say, fractions of an inch, so negligible at the nose, grows to a few inches and greater toward the trailing surface. That's why when you have surface roughness like misaligned tiles, a misaligned tile toward the end of the vehicle is not nearly going to have as dramatic effect as a misaligned tile or a chip in the front of the vehicle because the boundary layer is so much thicker that the disturbance doesn't
ADM. GEHMAN: But on the front edge of a leading surface like the RCC or the nose of the vehicle, these boundary layers are compressed down to fractions of an inch. So the distance between the temperatures that the vehicle sees, 2750, 2900, and these 10,000-degree temperatures which are measured in little bits of
DR. BERTIN: The differences between the temperature at the edge of the boundary layer, being
or 8 thousand degrees and the temperature wall being
or 3 thousand degrees are going to take place over
fractions of an inch, which is why the heat transfer
rates become so large because those temperatures are
Then another thing that's going to
happen, like I say, is if you imagine rubbing your
hands together, you're going to get some friction and
the temperature within the boundary layer may even be
greater than the temperature at the edge or at the wall
because you have this frictional dissipation going on.
Because you're going so fast. Your air particles are
moving so fast that the rubbing together creates the
heat transfer that's unique to hypersonic flight.
DR. WIDNALL: John, you're talking about
basically the temperature distribution around the
vehicle for a gas that is fully dissociated.
DR. BERTIN: In equilibrium.
DR. WIDNALL: In equilibrium.
DR. BERTIN: The numbers I gave you were equilibrium. If you had non-equilibrium or you're fully dissociated, your temperatures would be a little higher. If you had recombination, your temperatures would be a little bit different again. So your temperatures mine were based on kind of an equilibrium model.
Now, if you do some computations and you
keep it to simple global areas, because the heating is
such a small fraction of the total energy available,
within about a 20-some percent model, whether it's
fully catalytic or non-catalytic over the length of the
thing will not change too dramatically. Now, if you
compromise a leading edge and you expose some metals
and stuff like that, then, yes, your catalysity
probably is a major factor like you've been suggesting.
DR. WIDNALL: You know, this is a subject
I absolutely hated in graduate school, I have to tell
you; but my reading and study of this indicates that
the effective surface catalysity has a larger effect on
temperature than it does on heat transfer rate.
DR. BERTIN: Well, now, you remember the
temperatures were backed out of the heat transfer rate.
I mean, the heat transfer rate's going to be backed out
of the temperature.
DR. WIDNALL: Right. But the temperature
that's being affected is sort of stagnation temperature
DR. BERTIN: Okay. You're referring, I believe, to some tests that were initiated at Ames
DR. WIDNALL: Well, not only that. Just
thinking about a stagnation point.
DR. BERTIN: Now, at a stagnation point you're going to have the velocity of the gas is going to be different than it is going to be moving around the vehicle. So your residence time is going to be a little different and so your effects are so you would have to take that into account. You'd have to take the shape of the vehicle into account and you'd have to take whether you were looking at the stagnation point, whether the stagnation point was catalytic and the surface was not and things like that.
ADM. GEHMAN: Speaking just aerodynamically, forgetting all about heat even though I've already learned you can't do that. Shuttles have returned safely from voyages in which as many as a dozen tiles were missing and weren't even there. Based on the presentations that you've heard and based on your knowledge of this leading-edge shock wave kind of thing, on the shuttle what kind of a deformation I'm not asking you to predict what was missing but what kind of deformation in order of magnitude should we be looking for? Are we talking about inches or feet in order to significantly change the shock wave and, therefore, the shock wave also determines the exterior heating wake?
DR. BERTIN: If I were trying to relate the aerodynamics and most of the stuff I've done has been aero-thermo and it's been with the heating and transition environment and not with the small increments to the aerodynamic coefficient but if you had one of the T fillers missing and stuff like that, I think the mechanism for heating would be different than if you had two or three of the RCC panels missing. Because I think with just a filler bar missing, I think you'd start the process and you'd have some situation where you would have to do some analysis of flow in a narrow gap. Because if you just did from a two-dimensional analysis of like the flow in a cavity like he was showing some of those things with the flow coming, the flow would pretty much skip over a T cavity.
So you'd have some flow getting in and it would have to start a process that led to more damage, in my mind, to get significant changes. I think people who have looked at the data that they're obtaining at Langley have said that having the one little RCC missing, the No. 6 one, did not give them the aerodynamic changes that they saw later on. And I would believe that. I would believe the T would give only slight changes, that what it grew into when it lost maybe like I say, if you were going to suppose or opine when it grew into something that had multiple RCC pieces missing so that you had kind of the bow shock changing significantly, that would change your aerodynamic forces significantly and that would be consistent with some of the later things going on.
DR. HALLOCK: Can I ask you a question?
We've heard the term "shock-shock interaction" used
many times. I think it would be useful if we could
define what that means; but also, as part of that, go
back to the fact that, as you mentioned, you can even
see this in a photograph at Kirtland. The question is:
Why can you see this in a photograph?
DR. BERTIN: If I have a vehicle like this, it's going to have a shock wave that looks
ADM. GEHMAN: You're welcome to sit down,
even though I know all professors do better waving
their arms around.
DR. BERTIN: I'm Italian.
If you look here, this is the shock wave standing off from the surface; and it causes the flow to change direction and the pressure to increase. And with the pressure increase, the temperature increases. So it would be about some small distance off the surface. If you rotate it and look at the picture in this plane you're not rotating the model, you're just looking at the picture in a different plane you'd see also the shock wave having about the same standoff distance, right? So it would come in and intersect this surface. Right? But what's going to happen to the surface out here? Because that shock wave is only changing things inside within its dimensions. So it's only changing things between the shock wave and here. So when it hit the wing, it wouldn't affect this at all out here. Right?
So another shock wave has to form to
cover this part of the body, and it would depend on
what the angles were and the radii and how fast you
were going. So you'd have a situation where you had a
shock wave up here and a shock wave here. Now, when
they intersect, you have changes in pressure that are
different in here than they are out here.
So there has to be something happening in the fluid mechanics to change so that the pressures become continuous and you don't have just sudden gaps in your flow and stuff like that. So the interaction you get depends on whether the wing is like this or like this or like this. So the bow shock is going to be and then all that is changed by the fact that if something happens so that you get a stronger shock, you'll move the flow.
So if I put a gap in here, a significant gap in here, when the flow comes down here, it strikes not the wing first but it strikes the teeth that are missing and kind of flows into that cavity and splashes up against the rear wall of this. So it creates a shock wave going out. And that shock waves gives you density gradients just like these and they'll cause light waves to be bent and give you a different pattern in the flow picture. So you could see and they unfortunately didn't have them. But if you roll the model to where you were in basically at a 90-degree bank, you would see the shock-shock interaction structure.
On the X-15 back when they did the last
flights of the X-15, they hung a hypersonic research
engine underneath it and they hadn't taken into account
the fact that there was a bow shock wave coming off the
main fuselage of the X-15 and there was a second shock
wave, completely different, coming off the hypersonic
research engine which was kind of underslung off the
ventral fin. And when the shocks came together, it caused a strong change in heating. The perturbations
in heating can be factors of 10, 30, and more when you
get the shock-shock interactions, depending on what the
sweep angle is.
For the shuttle without damage, the sweep
angle is such that the interaction effects are
relatively benign. So that while there's a shock-shock
interaction, the highly swept leading edge prevents you
from having strong interactions. If had an unswept
leading edge, you would have strong interactions and
very large heating going on. So that would be
something to look at.
MR. TETRAULT: Doctor, I'm told that the
shock-shock interaction occurs normally at RCC Panel
No. 9 on the shuttle. Is there anything that would
cause that to move, say, to a different location, say,
closer to the fuselage or further out on the wing?
DR. BERTIN: I'm assuming that RCCs
possibly were lost in time so that, in a very early
one, maybe one would be missing, maybe more, but then
because the understructure is exposed, that some
additional damage occurred and other ones would have
come off in some fashion. Just from my standpoint,
with just one missing, you could get the damage that
maybe was observed eventually; but for seeing the Kirtland one, I think you'd have a pretty good piece
MR. TETRAULT: I wasn't talking
specifically about any damage. I'm just talking about
in normal flight, I'm told that the RCC Panel No. 9 is
the location of the intersection of the shock wave.
DR. BERTIN: Oh, yes, 9.
MR. TETRAULT: My question is: Is there
anything that could happen in flight that would change
where that shock wave location would be? I mean, if
you are experiencing yaw, for instance, would it tend
to move closer to the body?
DR. BERTIN: If you're going to change the orientation of the vehicle, the angle of attack, the yaw angle, these things the shock-shock interaction pattern would be a function of geometry. It would be a function of angles.
MR. TETRAULT: But simply going from
right wing down to left wing down would not change that
intersection. Is that correct?
DR. BERTIN: If all you were doing is changing the bank angle from this to this where you had the 40-degree angle of attack, you shouldn't change. Now, if you change the yaw and roll angle at the same time
MR. TETRAULT: Then it would move.
DR. BERTIN: then it would change some things. If you change the angle of attack, it would change. If you significantly changed your Mach number so that the gas chemistry changed, that would change. In other words, the shock-shock interaction pattern at, say, Mach 15 in the wind tunnel might be substantially different than the shock-shock interaction pattern in flight because in flight you would have significant real gas effects, you'd have significant dissociation. In the wind tunnel, you'd probably have a perfect gas and a density ratio of 6.
ADM. GEHMAN: Doctor, you heard the previous presentation in which Steve Labbe mentioned the Columbia data showed a relatively early roll and yaw bias to the left or showed control surfaces trying to control that relatively early, earlier than previous different from other flights. Can you draw any conclusions or insights from that? Particularly what I'm interested in is the statistics, the chart that he showed where it showed this left bias very, very early, before the first temperature rose, before the first debris came off.
DR. BERTIN: Based on my limited experiences with the aero increments, the only thing that I was looking for when I talked to him about these very items was he talked about I was thinking that one of the possibilities would be premature boundary layer transition due to damage on one side as opposed to the other. 'Cause I was worried about that being one of the multiple players in a breakup scenario. So I believe he in fact, several people on the panel, in my conversations with them I believe the fact that they got the same sine for the increments of the yaw and roll and for when they got asymmetric transition, they always got opposing sines, that that was one fact that says, okay, it's probably not premature boundary layer transition on this particular flight.
Another thing. If you go back and look
at all the things, the sensors that went out, there
were several near the trailing edge near the elevons
and stuff like that; and it worried me that maybe
that's a sign that those were going out early because
of premature boundary layer transition. But if you
look, almost all the ones that went out early went out
because they came from bundles that were near the left
main gear area. So you could trace the ones that went
out near the trailing edge back to bundles that went
near the damage area, and the other ones that stayed on came from other parts of the vehicle.
Then the third piece of collaborative
information. The vehicle broke up at altitudes that I
think are just above where we had ever seen the
earliest transition, or close to it. I don't think we
had ever seen transition that early, even in the
anomalous flights. So for those three reasons, in my
mind, I kind of said, okay, damage notwithstanding,
there was not a premature transition event that led to
some additional failures.
MR. WALLACE: In the anomalous flights and I understand there were cases with Columbia where the boundary layer transition took place maybe at numbers as high as Mach 18 versus typically Mach 6
DR. BERTIN: I think it's Mach 8 and
50,000 feet. Mach 8, give or take one, and
50,000 feet, give or take about 10,000 feet.
MR. WALLACE: In those anomalous events,
if you know, did the boundary layer transition happen
sooner on one side or the other?
DR. BERTIN: There was one that was significantly asymmetric and I think most of them could be traced it's been a long time since I looked at those data, but I think asymmetry was significant as far as its resulting affecting of force on one flight. In two others, it was just early.
MR. WALLACE: Is it fair to say we have the piece well, you talked about the shock-shock and we have the shock-shock on either side. So I guess my question is, having stood under the orbiter down at KSC, it looks like one big wing to me
DR. BERTIN: Yeah.
MR. WALLACE: But does the boundary layer transition can happen really distinctly separately on either side?
DR. BERTIN: Boundary layer transition is the growth occurs because of disturbances grow to where the flow breaks down to where it kind of swirls and twirls. So you could have a piece of damage, a tile bar filler I believe that was one of the sources of one of the flights where they had the gap filler sticking up about half an inch or more, and it would trip the area. It would affect the flow downstream of it because, again, we're locally supersonic so disturbances won't propagate upstream. So if you put a gap filler bar up in front, you would have the transition promotion in kind of a wedge downstream of that.
MR. WALLACE: So you could just have kind of a localized area where
DR. BERTIN: Localized but broad
coverage. But it would start at the bar and go down in
some kind of wedge.
DR. WIDNALL: John, how would you
calculate the temperature at the stagnation point of an
aluminum sphere that was reentering the atmosphere at
DR. BERTIN: I assume you mean the entire temperature profile and not just the temperature at one
DR. WIDNALL: Well, I'm interested in the
temperature stagnation point. I mean, I assume that's
easiest to do relative to everything else you might
want to calculate.
DR. BERTIN: Okay. But just like with the vehicle in general, the stagnation point has a temperature at the surface and it has a temperature of the air outside
DR. WIDNALL: I'm interested in the
DR. BERTIN: At the surface temperature?
I would assume you would use, depending upon your
altitude, a Navier-Stokes code or the classical fluid
mechanics code with chemistry. And I'd be willing to
bet people at Ames have a code like this to calculate the chemical reactions which would be dependent upon
the density and the velocity of the vehicle at which
you were flying. And then you would have to do a kind
of thermal surface response. But you would have a
non-equilibrium flow with a surface catalysity of the
material in there and you could get a pretty good idea
of what the temperature would be of the material. And
like I say, I think it would be very sensitive, if you
had an aluminum sphere, as to what your thermal mass
was. Because the aluminum sphere would not only be
catalytic but it would be a good conductor. So some of
that energy would be immediately conducted into the
DR. WIDNALL: Well, let's make it a
thin-shell aluminum piece.
DR. BERTIN: Okay. A thin shell. With
an adiabatic back piece?
DR. WIDNALL: Whatever.
DR. BERTIN: Okay. Then you could do
some similar things like in your heat transfer model
just have the thing go up in response to the
environment you put it in. You could do a
non-equilibrium computation with a reacting surface.
And, in fact, I would think codes exist for simple
shapes like the sphere that you're talking about.
ADM. GEHMAN: In the debris associated
with this tragedy, there are some 25 spheres which have
been recovered. All the fuel thanks, 25 out of 30.
DR. WIDNALL: But I don't think any of
them were made out of aluminum, were they?
ADM. GEHMAN: No, but the question I'm asking is when you take everything into account that you know that this orbiter was subjected to, starting at about Mach 17 and then finally breaking up at Mach 15, something like that, and you take into account the discussion we've had about chemical reactions and catalytic reactions and ionization, what would you suggest that we should be looking for in the debris? I mean looking for in the chemical sense that is, in the sense of deposits and discoloration and oxidation. What kind of testing and metallurgic kinds of evidence should we be particularly sensitive to that might give us a clue as to how this thing started, particularly if we can juxtapose the left wing and the right one?
DR. BERTIN: Jim Arnold and Don Rigali and Howard Goldstein and I were here about a week ago, and we talked a little bit about forensic-type looking at the deal. And I think several people have talked about how the flow there's indications along the front part of the wheel well, the left landing gear box, the door covering, that indicate the flow was actually from the inside out, that there were stream line patterns in the surface there and there were dark and you could see little stream line patterns. So this would indicate that, in my mind, that at least at this time frame there was flow going through, whether it was, as you talked about, a 4-inch hole, an 8-inch hole that had grown to whatever, flow had gone in there, was impinging on the tires and kind of coming back out, not necessarily filling all the cavity but impinging on the tires and coming back out. So I would think you would want to do some things with doing some analysis of the surface in that area, find out if it was aluminum from some places, if it had tire type things in it. There was, like I say, a recommendation made by Jim Arnold that I thought had some good things in it.
Backing up in time to find out when
things first started, I would think you would want to
do something like some free molecular flow calculations
in some scenarios where you either had a pock of damage
or a crack or a split or some kind of realistic T bar
filler missing, some kind of realistic thing to see how
that would affect the back surfaces and the aluminum
facings and stuff like that to start your damage pattern. And then something in the middle where you
had a jet of hot air going in through the damaged
substructure and creating more havoc. But then you're
in kind of a continuum environment at about
70,000 feet, even before you get to peak heating.
That's kind of what I put together. And then you have
the tests at Langley with the RCC panels missing to see
if you could kind of reproduce the Kirtland photograph.
ADM. GEHMAN: My last question is, based on your fairly extensive knowledge of both aero- and thermodynamics, do you feel that after this total of 13 flights in this regime that is winged, manned, recovered flights, that our knowledge of this region is are we at the beginning, the middle, or are we fairly mature in our knowledge of this region of science?
DR. BERTIN: I think it was Mr. Caram that said that or one of them said that almost every trajectory flies right down the same path. So it's not like we've had, you know, each one as a new environment. They kind of go along the same path. If you overlaid the velocity altitude time, there would be very nominal type performance. So in that sense we have a lot of experience with what happens nominally if nothing has broken, if nothing has come off or if what has if a tile is missing, it's a tile in some place that's fairly benign or there's a structure underneath that caused the tile, the heating to be conducted internally and not out. So I'd say from that standpoint we have a lot of information.
From the standpoint of what could happen if something came off and hit something and damaged it in ways that had not been done before, it's a very unique and very harsh environment. I doubt that we'd know even something as simple as the initial flow field that caused, say, the initial say, a T gap had been missing or a small hole in the RCC. I think that would be a challenge to look at and say, okay, I think this happened in detail.
Like I say, if it's nominal, everything's
sealed and things going on, we've got a lot of
information. If we're substantially away from nominal,
it's a very, very harsh environment and very, very
sensitive to the individual details, I would think, of
what actually happened.
MR. TETRAULT: Let me ask you one question. As we heard in the last presentation, very late in the event, the rolling motion seemed to change and that change required that there be lift under the left wing. They talked about running analysis based on the wheel well door being open which might have created it. Can you think of anything else that would create lift on
DR. BERTIN: The shock-shock interaction if you look at the orbiter like this, the normal shock-shock interaction, like I say, is going to trace a bow shock that comes along here, intersects the wing in about here, and then another shock that's going to be like this. If we had the two or three pieces missing by that time here and we had a shock that looked like that, which is kind of what the Kirtland photograph and again, don't overinterpret this that needs more work. If the Kirtland photograph is saying we had a shock-shock interaction like that, that's going to be a much stronger shock. The modified one, the one with the pieces missing is going to be a much stronger shock with much higher pressures than the original shock would have been at, say, while the vehicle was still intact.
MR. TETRAULT: And that could have gone
under the left wing?
DR. BERTIN: That could have caused the
pressure to be higher and giving you an asymmetric
DR. WIDNALL: I was rather intrigued by the suggestion that came up earlier that perhaps a jet
coming out of the wheel well door could create a local
shock in the area around the wheel well and increase
pressure. Of course, obviously it depends on the
DR. BERTIN: It would do that also
because obviously if the flow is coming from inside
out, that's not a normal passage. So as it comes
oozing out or flowing out or however fast it was coming
out, it would thicken the boundary layer; and
thickening the boundary layer would change basically
the flow over the surface some small amount.
DR. WIDNALL: But it could lead to a
shock, a local shock.
DR. BERTIN: And it could lead to a shock
interaction that would cause locally higher pressures
in the area of the landing gear well.
MR. WALLACE: The prior panel described
their various scenarios that they were then going to
try to fit into the aero picture, the thermo picture
and on the sensory. Any thoughts on those scenarios in
terms of other scenarios that you might suggest?
DR. BERTIN: I think they're working with
the tools they have in a logical sequence of steps. If
I were kind of setting up things, I might try to spend a little additional time kind of coming up with
cartoons of what the flow might look like if it were
coming out of the wheel well and saying how much does
that modify the flow, to try to get me some additional
things that I could compare with some of the
observations that you made. Like if the thing was
generating more lift at some point, could I get a
shock-shock interaction to explain that, could I get
flow coming out of the wheel well creating a shock.
So in addition to the things they are
doing, which are certainly good steps along this line,
I think I would try to get a cartoon strip saying like
this is what's happening here, this is what's happening
here, this is what's happening here and try to get some
It's a very, very difficult problem to do
either experimentally or computationally. So you kind
of want to, like I say, have some pictures of what do
you think is happening and then run some tests or do
some computations to see if that's what you get out of
your models. I think somebody pointed out the fact
that the model fell apart later on. Well, why did it
fall apart? If one third is going like that or one
third is going like that. So do you need to upgrade
your model? Do you need to improve the rigor of what you're looking at?
MR. TETRAULT: I know you haven't had the opportunity to see some of this debris, but let me describe at least one of the vents that's coming out and maybe give you a sense of how large it may appear to be. Then maybe you can tell me whether you would think this would be fairly significant in terms of disruption to the boundary layer. It appears that the vent goes out and actually covers three adjacent tiles, which would mean that it would be probably in the range of 18 inches. It actually melts the tops and surfaces of those tiles. So it would have to be extremely hot. And it is perpendicular to the normal flow that you would expect the boundary layer to be going over the aircraft at. And you see it
DR. BERTIN: You're talking about the
main landing gear cover?
MR. TETRAULT: Right. This is the
forward inboard corner.
DR. BERTIN: Yeah. I looked at some
pictures. It actually even erodes away the metal.
MR. TETRAULT: It erodes away the metal
structure on the inside. That's the aluminum
structure, which you expect because it's obviously very
hot. It's hot enough to actually erode ablative tiles.
DR. BERTIN: No, but you can see on what's left of the tile patterns, you can see black surface which is stream lining out. Yes, I would think that would cause a significant increase in the boundary layer thickness, some strength of a shock 'cause, I mean, if gas is coming out, it's changing the surface of the pattern.
In other words, if gas is coming out, it
would be like if I took a little jet in here and blew
air into the surface. That's going to cause the flow
to turn, have to turn around the jet. So you would
have a shock wave. You would have a shock boundary
layer interaction. How strong that was would depend on
how much flow you were coming out with, but that would
certainly be a parameter that would be in addition to
the notch on the wing leading edge. And it goes in the
ADM. GEHMAN: Thank you very much, sir.
I'm going to ask that, by virtue of being the chairman
here, I'm going to ask one last question. Then we'll
close up shop here.
Based on your knowledge, would you make a
recommendation to us as to how much latitude there is
in the reentry profile, you know, to reduce heating or
to reduce stress, even if you wanted to increase heating but reduce stress or something like that? How
much latitude is there in the reentry profile?
DR. BERTIN: That's not one of my areas of expertise. But in talking to others, your entry angle is somewhat limited because if you and you have the weight of the vehicle. So unless you can throw things overboard to significantly change the weight of the vehicle, your entry angle has a certain range that you can come into. And you're going out there. I mean, you're orbiting at 20,000 feet per second. You've got a lot of energy. You've got to dump a lot of that energy, and there's only so much drag you can do, it's my understanding, with the flight path. So limited.
ADM. GEHMAN: That's what we've been told
by several people, and I just wanted to get your
Dr. Bertin, thank you very, very much for
helping us solve this mistery. Your knowledge and your
professionalism and your ability to explain complex
things to us is very, very greatly appreciated. We
appreciate you taking time to help us with this; and if
we have any further questions, we probably will get
back to you. Thank you very much.
The press conference will start promptly at 1:00 o'clock, for any of you that are interested.
For those of you that aren't, have a nice day. For
those of you who don't have any choice, be here anyway.
(Hearing concluded at 12:32 p.m.)
Back to Event