Report of the PRESIDENTIAL COMMISSION on the Space Shuttle Challenger Accident


Volume 5 Index


Hearings of the Presidential Commission on the Space Shuttle Challenger Accident: February 26, 1986 to May 2, 1986.


Centered number = Hearing page
[bold number] = Text page.

[1294] 2228


CHAIRMAN ROGERS: The Commission will come to wider.

Dr. Littles.

Dr. Littles, if you would introduce your panel, please.

(Witnesses sworn.)







DR. LITTLES: Yes, sir. My name is Wayne Littles, and I am associate director for engineering at Marshall, and on my left I have Bob Ryan, who will be discussing the loads and joint dynamics, and Harold Scofield will be discussing the flight reconstruction and trajectory.

I would like to begin with a brief overview of the fault tree and scenarios, and update where we have changed since we last presented them to you two weeks ago, and point out the areas we will be providing information on today to allow closure of some of these areas. If you would go to Chart W-1, please.

(Viewgraph W-1.) [Ref. 3/21-3]

DR. LITTLES: Okay, looking down in the right-hand corner you will note that we have changed the external tank. When we made our last presentation, we had that as a yellow, which was a possible contributor. We are going to be recommending today with the




information that we present to you that we change that to improbable.

CHAIRMAN ROGERS: Doctor, could I interrupt just a moment to say that we are going to try if it is possible to conclude this session by 1:00 or 1:30. If it turns out that it runs longer than that, then we will take a recess and come back after the recess, but if it is possible, we would like to end at 1:00 or 1:30, and I thought for the convenience of everyone here I would like to announce that.

Okay, proceed, Doctor.

DR. LITTLES: Okay, go to Chart W-2, please.

(Viewgraph W-2.) [Ref. 3/21-32]

DR. LITTLES: This is the fault tree which is the expansion of the external tank in SRM. When we presented two weeks ago we had a number areas associated with the external tank in yellow. We were still looking at the possibility of damage and debris possibly causing impact to the tank. We will present information to you that concludes that that is highly unlikely. We are also looking at a structural flaw as a possible contributor, and we will present data on that.

Could we go then to Chart W-3, please?

(Viewgraph W-3.) [Ref. 3/21-33]




DR. LITTLES: This is the chart that provides the tailed scenarios associated with the hot gas leak, and when we provide the information to close the tank, we will also be providing the [1295] information and recommending that we close Item 2 on the scenario tree, which is the hydrogen leak impinging on the SRM and causing that failure.

We will also be providing information and recommending that we close the Item 4C, which is the leak check port, and we will be providing details of the load item, Item 5. Bob Ryan will be discussing that in detail. We had that in green when we talked to you before. We are going to provide more detailed data today. Also, since we spoke to you last, we have added two items to the scenario tree.

Since we have closed out loads and determined that the loads are within the design limits, as we discussed before, they could still be a possible contributor to a potentially damaged joint. We have reflected that by adding the item on the bottom right, liftoff flight loads effect on the degraded seal, so that is an added item. We have also added Item 7 to make this chart more complete. We have considered the potential of a case rupture.

We have not provided that information yet to




the task force. We will be doing that next week. But based on what we know, we consider that to be improbable, and we are carrying that as green contingent upon presenting that information to the task force and getting them to approve that next week.

So, when we close those items out, what is remaining then are the items associated with the joint itself Those items deal with potential damage to the joint during mating which is Item 3, that is, damage to the O-rings and/or the tang/clevis. The items associated with secondary O-ring defect in the joint or a delayed actuation of the O-ring, and Item 6, which is putty holding the pressure off the joint combined with low temperature.

So, we are left with items that - those items which either individually or potentially in combination, defects, will and/or possibly a low temperature, and possibly a combination of those things being the thing that initiated the failure in those joints, and we will 90 through at the end, the testing that we have under way to try to define which one of those items or which combinations might have caused the failure. Could we go to Chart 5, please?

(Viewgraph W-5.) [Ref. 3/21-34]

DR. LITTLES: I will now be getting to the




information which we have accumulated in summary: fashion, and recommend to you based on these data that we close the tank. The investigation approach - we reviewed all of the flight data for 51L for the tank, and the only anomaly that we saw in the tank itself, and this was mentioned by Dan Germany when we went over the time line, was that there was a change in the ullage pressure rise rate at 66 seconds, and that got progressively worse with time. Other than that, all of the flight data was perfectly nominal.

We have reviewed all of the Challenger external, tank history, all of the build paper. We have reviewed all of the x-rays. That has been done by four different people on all the x-rays, and we have found no irregularity in all of that paper review which would indicate that anything on the tank was a contributor to the anomaly.

We have evaluated all of the potential anomalies, the hypothetical anomalies that are on the anomaly tree, and I will now address each one of those items in summary fashion and provide the rationale to you as to why we think they should all be closed out. Could we go now to Chart W-2, please?

(Viewgraph W2.) [Ref. 3/21-32]

DR. LITTLES: Okay, this is just recalling


[1296] 2234


that tree again to point out the things that we want to cover again. It is going to be the liftoff damage and debris. The potential of a structural flaw, premature year shape larger detonation or structural overload. Could we go now to Chart 6?

(Viewgraph W-6.) [Ref. 3/21-35]

DR. LITTLES: Okay, this item on the fault tree deals with the potential failure of the external tank due to a structural flaw. We have reviewed, as I have said, all of the build paper x-rays, and we found no inappropriate dispositions.

Now, we did find in that review one x-ray which had an indicated defect. It was about four-tenths of an inch long. It was a weld that was located on the minus Z axis on the 2058 ring frame. Now, that is not in the area where we had the anomaly. It is on the minus Z axis of the vehicle. We have analyzed that flaw, and the analysis indicates that it should not have contributed to the failure. As a matter of fact, the critical flaw in that area would be something like 5.2 inches, and you would have to have a flaw of something like .65 inches before it would break through and have a leak.

The growth rate of a flaw in that area with the stresses that are imposed on that weld is very, very




low. We would expect it to grow something less than 1/10,000ths of an inch per cycle. We have also uncovered the piece of hardware that contained that flaw, and there is no evidence looking at that piece of material that the flaw was involved in the failure, so we conclude that is improbable that a structural flaw initiated the failure of the tank.

CHAIRMAN ROGERS: Dr. Littles, in questioning some of the Martin Marietta people who had the responsibility for this, apparently they found out that an employee failed to discover this imperfection that you have described, and they apparently have interviewed him, and so forth. Is this a problem that you find recurring that there are mistakes on the part of the contractors that maybe are getting more regular? In other words, there is more sloppy work being done by your contractors in some of these things?

DR. LITTLES: Well, I don't know that I have all the information to answer that question completely. I am not aware of any progressive degradation in quality. It does happen from time to time. There are human mistakes that do come about, but I am not aware of any degradation in quality.

CHAIRMAN ROGERS: Well, I think that is one of the things the Commission will want to focus on in its




report, whether we need to beef up the supervision of the contractors in some of these things. I think it is fortunate in this case that you found the portion that we are talking about, and obviously it didn't contribute to the accident.

DR. LITTLES: Yes, that was fortunate, but as I indicated, it was not in the right orientation at all. It was on the minus Z axis, which puts it around pretty far from where we see the plume. It was fortunate that we found it.

CHAIRMAN ROGERS: In other words, the imperfection did show up on the x-ray. It just wasn't reported by the employee?

DR. LITTLES: Well, it did show up on the x-ray. Now, my understanding is that something like a half a dozen people had re-reviewed that x-ray and I was told that only about half of them actually saw it. So it wasn't something that jumped out at you. It wasn't too obvious. So it [1297] wasn't something that he failed to report, but something he just failed to detect, just a human error.

Now, if I could have Chart 7, please.

(Viewgraph W-7.) [Ref. 3121-36]

DR. LITTLES: Okay, this item deals with the structural overload. As I indicated, Bob Ryan will




discuss the loads in more detail, but what we have found is that the liftoff and flight loads on the tank were well within the design limits. There was nothing greater than 80 percent of the maximum, and of course there was the safety factor, on top of this. The ullage pressure was maintained well within limits. There was no cursion on pressure. The trajectory was nominal up to the point of the incident. There was no excessive heating that could have caused an overload of the structure, and we have also done analysis that indicates that if we had lost some TPS in the area of interest during this time frame, that the beating would have been low enough that we still wouldn't have degraded the structure to the point of having the failure. So, we concluded that it is improbable that a structural overload from flight loads or excessive ullage pressure or excessive heating initiated the failure.

Chart 8, please.

(Viewgraph W-8.) [Ref. 3/21-37]

DR. LITTLES: Okay. This is an item I discussed when we last were together, but just for completeness I have included it again, and it deals with the premature detonation of the linear shape charge. The bottom line, of course, is, we did find these linear shape charges with the recovered hardware, and they were




intact, so they were not involved in the incident.

Chart W-14, please.

(Slide W-14.) [Ref. 3/21-38]

DR. LITTLES: Okay, this item deals with damage to the tank from debris which could have caused a leak. The ice team when they went out, and they went out more than once that night, but when they went out at T minus 20 minutes, they saw no evidence of a leak.

Could we see Chart W-14 please, Okay. This is a photograph that was taken during the night that contains the area of interest, and what I wanted to point out to you was that they have a very good view of that area. The right hand SRB is on the left of the picture there, and the area of interest is, of course, around toward the tank, so they had a very good view of that. If you could give me Chart 14A, please.

(Slide W-14A.) [Ref. 3/21-39]

DR. LITTLES: Fourteen-A is just a little closer view of that area, so you see, they had a very good view in that area, and if there had been a leak there, I think they would have seen it. There was no evidence from the cameras on the pad, of any debris. If you could give me now Chart 10, please.

(Viewgraph W-10.) [Ref. 3/21-40]

DR. LITTLES: You have seen these camera locations described by Dan Germany earlier. What this


[1298] 2239


sketch is is a composite of a number of the cameras. He mentioned E-63. You see that one down at the bottom right. And E-60, to the left, and several cameras that are shooting across from 90 to 270. You see those cameras give you a pretty good view of this area.

After liftoff at about one and a half seconds there is also a camera which is E-31, which is in the top right, which gives you a view of another portion of that area, and so you see there is really only a small portion that you can't see between liftoff and the second and a half, and in reviewing those films we see no evidence of any debris in that area.

One of the possible contributors to debris, one thing that we were concerned about earlier was that there had been a modification made to the covers for the hold-down posts and there are springs in there, and those springs, several of them were missing. There has been a good bit of work done on that, and it has been concluded that those could not have been a contributor to debris. An analysis done by KSC indicates that the springs could not physically get out of that cover prior to greater than eight-tenths of a second. Hence, since we saw the puff of smoke at .68 seconds, that is not consistent, and so we conclude that that was not a debris source.

You will also note in looking at the sketch on




the screen, Chart W-10, that the area of concern, the right-hand SRB, is well away from the fixed support structure on the stand, which is where the ice was. That was of some concern that night. And so if a piece of ice were going to cause debris, it would have to come off of that stand, and it would be a very rough trajectory to get it around to the point of interest, and it is improbable that that could have happened, but there has been some analysis and tests done associated with ice, and what those analyses and tests indicate is that for a reasonable size piece of ice, that you would not get penetration of the tank

There has been some recent testing done with ice particles as large as a quarter of an inch in diameter and four inches long fired out of a gun at a velocity of 300 feet per second, and they did not damage the tank. Of course, it goes through the thermal protection that is on the tank, but it actually shattered when it hit the metal underneath, and there are a number of other tests like that, but a particle of ice falling off the fixed support structure couldn't attain anything like 300-foot-per-second velocity, and so we conclude that ice is not a probable source of debris relative to penetrating the tank and causing a leak.




There have been some other tests done which are of some interest relative to both the ice team seeing the leak and a jet impinging on the solid rocket motor. There have been tests performed at Denver, and I am going to show a videotape in a moment showing these tests. These tests indicate that very small quantities of hydrogen escaping from the tank would be visible. As a matter of fact, tests have been done down to something like 3/1,000ths of a pound per second, and you can still see the hydrogen coming into the atmosphere, so if the team, if it had been leaking when the team had gone out, they probably would have seen it.

There have also been tests done at Marshall where we have impinged a burning jet of hydrogen, again, fairly small quantities of flow onto the materials which are present on the solid rocket motor in the area of interest. There is a significant quantity if you go back to Chart 14-A, please.

(Slide 14-A.) [Ref. 3/21-39]

DR. LITTLES: There is a significant quantity of instafoam which is on the ET/SRB attach rings. It is about four inches deep. There is also some cork on the ring that closes out the pins on [1299] the joint, and of course there is paint in the area, and if you impinge small quantities of burning hydrogen on those materials,




you see significant quantities of smoke and flame, and what I am doing is trying to relate this to the fairly small quantity of smoke that we saw with the initial puff at .68 seconds. Could we see that videotape now, please?

(A videotape was shown.) [Not published.]

DR. LITTLES: There is sound with this. Are we going to get it? Okay, it is narrated, but we are not going to get the sound. Okay, the test conditions are on the screen. The first test, what you see is on the left. You see there is a small hole. This is something like .003 pounds per second. There are three series of tests on here at varying flow rates, but .003 for the equivalent conditions we had at KSC was the lowest case that was, run. That wasn't the limit of visibility. It was just the lowest one they ran. Again, here is another view. You can still see it coming out.

CHAIRMAN ROGERS: So your conclusion is, the ice team would have seen that leak if there had been one?

DR. LITTLES: Yes, sir, I am confident they would. They had a very good view, as we showed in the earlier photographs. They could see the area very well, and you can see that you see that leak very well.




VICE CHAIRMAN ARMSTRONG: Is that water condensation you see?

DR. LITTLES: Yes, it is cold hydrogen with the moisture content of the air making it visible.


DR. LITTLES: Okay, the next series of tests are those that I mentioned, and this is .01 pounds per second of hydrogen being ignited, and you see the test fixture there. It will be coming from your left to the right. What's on the right are simulated portions of the material on the solid rocket motor. And the burning jet will be impinging on the material on the right, and you can see the smoke and flame characteristics.

There are four of these tests. The narration tells what they were, and I didn't bring that with me, because I thought we would be hearing the sound, but the first one here was, as I recall, just the jet impinging on paint, and you don't see a great deal that comes from that. You can see some smoke, but not a large quantity.

The next one that we will see will be one of the two that is of primary interest because it will be the burning jet, I believe, impinging on the instafoam, and the instafoam again is on the ET attach ring, in




that area, and you can see large quantities of smoke and flame, and this had a simulated quantity of instafoam, and it burned for quite a long period of time, and again, it is a very low flow of hydrogen.

This again was just some smaller quantities of material, paint primarily. I believe this one also had the cork and the hypolon paint. And the last one you will see will be as composite panel which has the instafoam, the cork, and other materials that are in that same area.


DR. LITTLES: You can see that there are fairly copious quantities of smoke and flame with that simulated material, and this, of course, is qualitative to some extent. But it would seem to [1300] indicate that you would see potentially more smoke than we saw from that initial puff. Okay, could we go to Chart W-10, please?

(Viewgraph W-10.) [Ref. 3/21-40]

DR. WALKER: Could you just repeat what the size of the orifices were in the first set of tests?

DR. LITTLES: That orifice was .06 square inches, I believe, and the flow rate was .01 pounds per second. Are you referring to the burning test?

DR. WALKER: No, I was referring to the first set of tests.




DR. LITTLES: The first set of tests. I don't have those written down in front of me, Dr. Walker, but I will get those for you. I remember the flow rate was .003, but I don't remember what the diameter was. I will get that for you.

Chart W-10, please - I am sorry, W-11. I am out of sequence.

(Viewgraph W-1 1.) [Ref. 3/21-41]

DR. LITTLES: So even though we don't feel that it was likely that we had a leak from the tank, we have also done some analysis to try to determine what the effect of the leak would be on the joint. What we have done is, we have used the maximum amount of leakage which would not be detected by the instrumentation. That turns out to be something like four pounds per second of hydrogen, and we have mixed that stochiametrically, and really it would be almost impossible to get a stochiametric mixture. So that is a conservative assumption. We have mixed that stochiametrically and burned it and impinged it directly on the membrane and the joint.

Now, what happens when you do that, of course, is that it doesn't stay in one place for the whole period from zero up to the time of the incident, because the velocity of the vehicle, the aerodynamics around the




vehicle are going to be causing that plume to sweep back away from the point. When we did that analysis, we came up with a maximum temperature on the membrane of about 650 degrees, and on the O-ring of 360 degrees Fahrenheit.

Now, we have done tests that indicated that after you get the O-ring sealed and have pressure behind it, that you can maintain that. It will maintain a seal up to a temperature of 1,000 degrees.

GENERAL KUTYNA: Dr. Littles, when you did those tests, what kind of air flow did you assume on the vehicle? Did you just assume a Mach 2 air flow, or did you actually look at all of the eddies and currents that were going around those tanks at the time?

DR. LITTLES: We used the vehicle aerodynamics. This question has been raised. As a matter of fact, I was going to mention later that this is one area where Thiokol has requested some additional information, and, as a matter of fact, we have Thiokol and the people from Rockwell who were responsible for the flow fields, and our own people, meeting today to review this analysis, and particularly to review the flow fields that we are using.

Now, there were some assumptions made in the flow field that we used. We used the free stream flow




field. In the area where we have this leak, of course, between the tank and the SRB, at some point in flight after you start getting a lot of shocks in that area, the assumption we made would not be completely valid. So that there is an assumption there, but there are also assumptions associated, as I mentioned, with the fact that we have assumed that it is stochiametric, which is [1301] a worst case. We haven't accounted with the large leakage that we have used here for any cooling effect. There would be a cooling core in the center of this jet. So there are some questions about the assumptions we made there, and those are being worked today, and I think we will be able to resolve that, as I said, because there are a lot of conservative things in here even with that simplified assumption that we have made.

There is another point, too, that I think I should make. Well, two points. One is that, you see, we have a reasonably large margin between the temperature we are predicting and what we think or what the test data says the O-rings could withstand.

And the other thing, I guess, is that we don't feel, based upon the other data, and I think there is quite an accumulation of data now, that the tank was involved. What we are doing here is assuming the worst




case leak to try to see what that would do to the motor, but we have no basis for believing that such a leak existed. So, we are trying to resolve that point.

DR. WHEELON: Dr. Littles, I find your conclusion both reassuring and just a little surprising. Let me read the conclusion. Tests indicate that the O-ring will maintain the seal at 1,000 degrees Fahrenheit. The reason I find that a little surprising is that in prior sessions we have heard a good deal of description of confusion and differences about what the low temperature end validity of that seal would be, whether it would function at 30 degrees or only at 54, and yet you seem quite sure that at the upper end of 1,000 degrees you know that the seal will perform satisfactorily at the high end.

Can you describe the tests that give you that confidence and explain to us why we don't have similar confidence at low end?

DR. LITTLES: Yes, these tests are tests that have been conducted since the incident. Of course, the spec on the O-ring is 500 degrees.

DR. WHEELON: We have heard a lot about the specs. Let's talk about the test.

DR. LITTLES: It was specifically oriented to this type of analysis and the possibility of having




something impinging on that joint, so there were special tests that were conducted, and what they did was, they had a test rig which had a fairly small O-ring in position, seated it with pressure at the right pressure, 900 psi, and then they just heated that up and monitored it with a thermocouple, the temperature in the O-ring area, and heated it up to the point where they got the O-ring to break loose structurally. And there were, I believe, three or four of those tests, and the temperature at which you had the structural failure was something above 1,000 degrees, and so that is the basis of that statement.

CHAIRMAN ROGERS: The last time you were here, the Commission raised the question of whether there should be outside independent people involved in these tests, and I will now ask General Kutyna to make a comment about that. I understand that we have, the Commission has worked out an arrangement to provide for that.

GENERAL KUTYNA: Mr. Chairman, those people are in place now. We took them out to Thiokol on Wednesday. They are there through today, and they will be down at Marshall next week, and so they will be involved in this particular test and this analysis to make sure that it checks out.




CHAIRMAN ROGERS: How many people are there?

[1302] GENERAL KUTYNA: There are six people, Sir: representatives from the Rocket Propulsion Lab, from the National Transportation Safety Board, from the Air Force's Space Division, and from industry support to the MX Missile Office.

DR. WHEELON: Mr. Chairman, may I continue my question with Dr. Littles just a little further? If it is so easy to determine the high temperature behavior of the seal, why isn't it equally simple to determine the low temperature end? Can you explain that for me and perhaps other people on the Commission?

DR. LITTLES: I don't believe it is complicated to determine the low temperature performance of the seal, and we are doing that in more detail as a part of this investigation. The thing that is a little more complicated relative to the failure scenario we are pursuing is not the performance of the seal itself at low temperature, but the performance of the seal in combination with the dynamics associated with the joint. That makes the testing of that total effect a little more complicated, or at least the building of a test rig to do that a little more complicated.

As I will report to you in the test section, we do have our dynamic test rig functioning properly




now. We are getting good data out of it. I believe within the next few days, week to ten days, we will have the information required to resolve that particular failure scenario.

DR. WHEELON: Okay, given that this test is easy to perform either at high temperature or low temperature - I accept your judgment there - is it true that we had done these simple tests prior to the launch of 51L, either at high temperature or low temperature, or are these new tests that are being done for the first time?

DR. LITTLES: These particular tests I am referring to are tests and test fixtures which have been designed and fabricated and are being used specifically to evaluate the scenarios or hypotheses for failure. They are all associated with the failure investigation.

DR. WHEELON: So they were not done prior to the launch?

DR. LITTLES: These particular test fixtures were not used prior to launch, no, Sir.

DR. WHEELON: Were any tests of this nature done prior to launch?

DR. LITTLES: To my knowledge there were no subscale tests which incorporated dynamics. There were some subscale tests looking at the O-ring capability of




the seal under the ignition pressure transient. Those tests had been conducted down to something 30 degrees or a little below, I believe - I know it was 30 degrees; it might have been a little lower - as a part of an evaluation of the joint that had gone on, but there had been, to my knowledge, no subscale tests that included joint dynamics.

DR. WHEELON: Had there been any full-scale tests?

DR. LITTLES: Yes, of course, because the motor firings themselves have joint dynamics, yes, and those have been conducted down to a temperature of 40 degrees on one of the - it was either the development motor or qualification motor, I can't remember which, but that motor was at 40 degrees on the joint, calculated 40 degrees. The environment was 36. And there was an analysis done which indicated that the joint was 40.

MR. RUMMEL: However, those full-scale tests were not really representative of the flight article as I understand it, because additional putty was inserted at the field joints prior to those tests. Is that correct'?

DR. LITTLES: There were some modifications made to the putty. The concern was, as I understand it,


[1303] 2253


that they wanted to make sure that they didn't have gaps in the putty, and there were in some localized areas. It has been reported to us by Thiokol that there were some areas in there where they had to go in and do some adjustment to the putty. It wasn't over all joints. And it wasn't over the total area, we are told, on any joint, but there were some areas where they made adjustments to it. That is correct.

MR. RUMMEL: Those same kind of additions of putty, just to be clear, have not as a matter of practice or really at no time was that same thing done to flight articles. Is that true?

DR. LITTLES: No, sir, not to my knowledge.

MR. RUMMEL: Thank you.

DR. LITTLES: Could I have Chart W-17, please?

(Viewgraph W-17.) [Ref. 3/21-42]

DR. LITTLES: So, we conclude, based on upon what we have just discussed, that it is improbable that damage to the tank initiated a leak or, based on the analysis, that a leak within the detectable limits of the tank instrumentation would have served to initiate the SRM failure.

Chart W-3, please.

(Viewgraph W-3.) [Ref. 3/21-33]

DR. LITTLES: Having concluded, then, that




there was no leak in the tank, we did get a failure of the tank, of course. We have indicated that at 67 seconds we saw a decrease in the ullage rise rate, and what I would like to address now is what caused that, and also, while we are on this chart, with what we have just presented here, we recommend to close Block 2 on this chart.

Could we go now to Chart W-18, please?

(Viewgraph W-18.) [Ref. 3/21-43]

DR. LITTLES: As you have seen from the film and time line earlier this morning, there was a hot gas leak from the motor which was evident from that film at about the 58 to 59 second time frame. We have used the data, the chamber pressure data that we have from flight, and we have established a leakage from the size hole that the change in chamber pressure would indicate.

We have looked at the heating that would result from that type of plume coming out, and impinged that type heating rate on the external tank, and we have concluded that within the time frame between 58 to 59 seconds, when we first see the hot gas leak coming out of the motor, and the time at 66 seconds, when we see the change in ullage pressure in the tank, that there is more than sufficient heat during that period of time to




bum a hole through the tank. As a matter of fact, with the heating rates that we could get, which could range up to 880 BTU per hour foot per square second, you could burn a hole in the tank in a couple of seconds, and so we conclude then that it is probable that the leak from the SRM initiated the tank failure rather than vice versa.

This concludes the section on the tank, and we recommend, based upon what we have here, that the tank be colored green on the color code indicating that it is improbable that the tank initiated the failure. As I mentioned, this information has been presented to the task force. Martin Marietta agrees with this, and Marshall also agrees with the recommendation that we close it. There is some, as we mentioned a minute ago, some residual work that we are doing with Thiokol. They are reviewing some of our analysis. We are going to provide some additional [1304] data to them, so there is some residual work going on there. But on balance we concluded that the tank is not involved in the failure at this point in time.

Could we see W-20, please?

(Viewgraph W-20.) [Ref. 3/21-44]

DR. LITTLES: Relative to the leak check port, Chart W-20 is a subtier failure tree that was used to evaluate the leak jet port, as was discussed this




morning with the time line in the film. Dan Germany showed you where the cameras were relative to the liftoff. I am not going to go through all of the details that we went through in this investigation because the bottom line is that in reviewing all of the film, looking at all of the cameras we have, there is nothing visible from that location.

Could we go to W-21, please?

(Viewgraph W-21.) [Ref. 3/21-451

DR. LITTLES: W-21 is the picture I had up earlier, which shows the locations of the camera, and what you can see from that is that looking at cameras E-63 and E-60, you have a very clear view of that leak check port, and in reviewing all of the film, we have never been able to see anything that comes directly from that port.

DR. COVERT: Dr. Littles, it might be helpful if you would indicate if that leak check port is on the top or the bottom of the right-hand booster. The picture you have does not indicate where the leak check port is.

DR. LITTLES: If you go to Chart W-23, please, I believe that shows it.

(Viewgraph W-23.) [Ref. 3/21-47]

DR. LITTLES: I am sorry, it was W-22.

(Viewgraph W-22.). [Ref. 3/21-46]




DR. LITTLES: On the left part of that figure there you see the arrow pointing to the location of the leak check port on the right-hand booster, and what this is is an artist's concept of the view that you would have from camera E-63, and where you might see the puff of smoke emanating if it came out of the leak check port, and in looking at all of the film, we don't ever see an area, a clear area on the motor itself between the leak check point and the tank. There is always smoke over there, up against the tank. So we conclude that it doesn't come from the leak check port, but it comes from somewhere around out of view, coming around the corner there.

Could we see Chart W-23, please?

(Viewgraph W-23.) [Ref. 3/21-471

DR. LITTLES: This is just the conclusion relative to that that the smoke is plainly visible at .678 seconds. The source itself is not visible, and, as Dan indicated this morning, all of the film indicated that it probably originates between 270 and 310 degrees, and we conclude from this that it is improbable that the smoke origin was from the leak check port, and we recommend closing out that failure scenario.

Okay, that concludes the leak check port. We would like now to move into the section on trajectory




and flight dynamics reconstruction, and Harold Scofield will cover that for us.

DR. SCOFIELD: Mr. Chairman, Dr. Ride, gentlemen, I joined ABMA in 1956, and have been with Marshall Space Flight Center since its inception, and since that time my speciality has become the analysis of Right dynamics. I am currently the chief of the Control Systems Division, [1305] which applies the analysis of flight dynamics to the synthesis of auto pilots and pointing control systems.

If I could have Chart S-2, please.

(Viewgraph S-2.) [Ref. 3/21-481

DR. SCOFIELD: This pitch is about reconstructions. Now, reconstruction is a computer simulation, and in that computer simulation we attempt to include all of the environmental effects that we can in order to recreate what happened, in this case on 51L, of course.

So, our task will be to construct math models of the environment, and we hope to get out of these reconstructions three things, truth checks against failure scenarios. That is, if we can detect the effect of some attribute of a failure scenario, and if we can simulate it and can understand it, then that leads to increased confidence that that failure scenario happened




or that it didn't.

Secondly, the data that is produced from these reconstructions is needed for loads analysis to determine the dynamic loads on the vehicle, again to be reconstructed for the 51L flight, and Bob Ryan to your right will cover that part of the pitch later.

And lastly, of course, the reconstructions enable us to tell whether the vehicle systems operated properly, particularly the guidance and control system in this case. Now, the chart that is on the screen graphically shows you how these analyses fit together.

On the left, the propulsion reconstruction feeds into the trajectory. First, I will talk about propulsion reconstruction. Secondly, I will talk about the winds aloft and the atmospheric reconstruction. That is in the upper left corner of the chart. That also feeds into trajectory. And then on the right part of the chart is a block called plumes and aero. And what that is is a modeling of the failure plumes, the hot gas leak from the SRM using photo data, and the calculation of revised aerodynamics due to that plume. Those aerodynamics have an effect particularly on the control system. They do have some effect on the trajectory also. You will find that those effects are





Fourth, we generate a coordinated time line, which we need for the trajectory reconstruction, and with all that data we are able to compare our mathematical simulation on the computer with observed data and feed that into a control system simulation which generates the dynamics or a more detailed explanation and understanding of the dynamics from which we can estimate the loads.

Chart S-4, please.

(Viewgraph S-4.) (Ref. 3/21-49]

DR. SCOFIELD: The main propulsion system performance was reconstructed. That is the Space Shuttle main engine system that I referred to. We determined the propellant loads at main engine start command and the thrust and flow rates during flight. We reconstructed the SRM performance, and that included the internal pressures, the flow rates, and the thrust during flight now. On the screen you have a plot of SRM right chamber pressure, internal pressure, versus time. The heavy line is the measured data from 51L for that right SRM. The dot-dash line that it almost covers is the mean of our flight experience in the first 24 flights, adjusted to 51L conditions. We do that by running tests on each batch


[1306] 2261


of propellant. Each batch of propellant is a little different, you see, and we make small motors as well as the big ones that we use from each batch propellant, and fire them to ascertain the differences in performance.

The temperature also has an appreciable effect on this parameter, and that is not the parameter of the temperature of the atmosphere, but, rather, the temperature of the propellant inside the rocket. It varies nowhere near as much as the outside temperature, but nevertheless it is important. Those two main parameters were used to shift what those first 24 flights would have done had they been flown on the day that 51L was launched, and had they been constructed from the same batch of propellant.

DR. WALKER: Are those test firings done at Marshall or at Thiokol?

DR. SCOFIELD: At Thiokol, sir. Now, we have two attributes of this curve that need to be discussed. The first on the left is marked right SRM performance higher than predicted. That is not a large deviation. That is about 1.4 standard deviations above the mean, and you would expect variations on the order of one standard deviation in examining this data. Now, furthermore, there are some mechanisms which are not in this reconstruction which tend to make us believe that




that pressure would be a little high, for that motor, anyway.

Nevertheless, it is not important in our opinion. The attribute on the lower right, marked SRM leak begins, of course, is very anomalous. You see that it crosses, "it" being the heavy line, crosses the dotted line, which are marked "expected limits", and we would estimate the probability of that behavior at being less than three parts in 1,000 if we only had to worry about random things, and so we conclude that that is not random. We would have concluded that anyway. Even if we had not had a failure on this mission, we would have been very much concerned about this phenomenon. Of course, we know from photo coverage that the SRM leak does begin at about that time that we see the divergence of the internal pressure.

Now, we use - our propulsion people use this divergence from the expected to construct the size of the leak that must have been evident in this right SRM. Now, of course, that leak had a zero area at the start, but it was estimated that it grows to about 45 square inches at 72 seconds. With that data, we can construct a thrust profile from that leak, and we can use it in subsequent reconstructions.

Chart S-5, please.




(Viewgraph S-5.) [Ref. 3/21-50]

DR. SCOFIELD: All right. At this point, with no other reconstructions accomplished, we were able to go to work on liftoff dynamics. We simulated liftoff dynamics with a three-dimensional translation and rotational program that included the flexibility of the shuttle and the initial conditions for this liftoff-that is, the conditions with which the vehicle left the pad - were chosen to match the flight records telemetered from the flight and the first one or two seconds, and we obtained a match for that telemetered data using the values that are listed there which were within the nominal range of expected variations.

That is, the space shuttle main engine thrust rise was within the expected variations in that parameter, and likewise the SRM ignition timing differences and the ground winds which were measured that day, of course. This gives us confidence that we understand this part of flight [1307] and it enables us to conclude that the control system, for example, worked properly during the liftoff phase of flight.

We also compared the SRM gimballing activity for 51L with STS-6. STS-6 was also the Challenger, and it carried the TDRS satellite also, so it had, essentially the same payload, and those plots are very




similar up until the point when the vehicle clears the launch supports, the haunches, as we call them, the supports the SRBs sit upon.

After that, they diverge some, and we attribute that to ground winds. That is a minor thing. We also evaluate the liftoff films to make sure no collision occurred with the ground structure, and we were able to determine, of course, that there was no collision with the ground structure. Now Chart S-7, please. [Ref. 3/21-51]

(Viewgraph S-7.) [Ref. 3/21-52]

DR. SCOFIELD: Next, we needed to reconstruct the environmental model for 51L.

DR. WHEELON: Let me make a point, before we leave that matter for the last time. During this initial part of the liftoff sequence the vehicle has vibrational modes. It twangs like a reed. At what frequency does it twang, in its resonant mode, as it lifts off.

DR. SCOFIELD: The first main mode is essentially three hertz.

DR. WHEELON: Three cycles per second.

DR. SCOFIELD: That's right, and it was contained in the simulation, and indeed that has to match in order to obtain the match that I claimed we do. We were able to see that twang from the photo coverage. You can actually see the vehicle bend or rather you can see the




vehicle move, and you must include the bending of the vehicle in your simulation to account for how it moves, and so that was included.

DR. WHEELON: So the vehicle in its principal mode is twanging three times a second. Do you make any correlation between that and the fact that the smoke was apparently puffing three times per second?

DR. SCOFIELD: Well, that has been discussed earlier today. I can only say that I have seen that data, and it is only a very rough correlation.

DR. WHEELON: It is better that no correlation.

DR. SCOFIELD: No, sir, it is not no correlation. That is too many negatives. The puffs of smoke do appear to be very, very approximately at a three-puff-per-second rate. However, we cannot account for that. We are still working on it. We feel that it could be a coincidence, but of course we are not going to ignore that as evidence, and we are going to pursue that until we are sure.

DR. WHEELON: It could indeed be a coincidence, but it is worth noting in passing.

DR. SCOFIELD: It certainly is.

VICE CHAIRMAN ARMSTRONG: It is at least not inconsistent.




DR. SCOFIELD: It is not inconsistent with the mechanism that opens that gap according to the vibration of the vehicle, and I think Bob Ryan will talk more about that later, as a matter of fact.

DR. WHEELON: Thank you.

DR. SCOFIELD: Surely. Now, next we reconstructed the atmospheric environment, and that is mainly winds aloft and pressure and temperatures aloft for this type of analysis. We need that [1308] data in order to predict where the vehicle goes and what sorts of air pressure are exerted on it. The ground winds were also reconstructed, and the ground temperatures, and the main fault histories also, and that was mainly to supply it to other teams. Our team was in charge of producing these reconstructions for our own use and for others' use.

Now, the plume of the SRM leak was modeled from photo coverage, and you see the chart which contains some line drawings on the right as seen by Camera E-207 at 60.6 seconds. Now, you can actually see the plume from that camera at that time of flight. The aerodynamicists estimated the size of that plume, and they estimated how much choke flow they get in the channel between the SRB and the tank, and they estimated the differences that would be evident in the plume




downstream of that leak.

And they produced what they called aerodynamic increments or really new aerodynamic data that would account for the forces and moments on the vehicle due to this phenomenon. Now, on the left part of that same chart, if you could go to the left side, please - thank you - we have an aft view. Unfortunately, we don't have a camera that shows this. It would be really great if we did. We have modeled the plume from the SRM leak along a 45-degree line to the vertical. Some people think that is known fairly well, say, within ten degrees. Others don't.

They think that we cannot reduce that photo data that accurately, and so we are doing a sensitivity analysis at this time, assuming a 30-degree error in the location of that plume, and we are going to see if we can tell any difference in the flight dynamic reconstruction that would aid in pinning down where the SRB leaked.

Now, that is the kind of thing that we are after. And of course the leak was a fan-shaped phenomenon. It wasn't a round hole. And so we expect that this could have a small effect. If it does, we won't be able to help pin down where that leak came from. But, on the other hand, if it does make enough




difference, we shall.

The maximum aero force due to the aerodynamics, the revised aerodynamics of the plume, are about 130,000 pounds, and I think by anybody's rule book that is a lot, and you will see in a few moments it does move the vehicle. We provided all of this data from the atmospheric reconstructions and from the propulsion reconstructions to JSC, and to all others, contractors included, that needed the data.

Chart S-9, please.

(Viewgraph S-9.) [Ref. 3/21-53]

DR. SCOFIELD: Now, next we determined a coordinated time line, and a trajectory using the winds aloft, the propulsion, and the SRM leak with aerodynamic increments. We have already obtained excellent time line agreement with the JSC time line activities, and we have obtained excellent agreement with both Marshall Center photo activity and the JSC/KSC photo activities. And we are talking about ten milliseconds here. We know these times very well. We may have some difference of opinion on the interpretation of the films, but we know when the phenomena occurred, and we know what is on the film.

The reconstructed trajectory was successfully compared with a JSC product called the best estimated


[1309] 2269


trajectory. Now, the best estimated trajectory is constructed from radar data, and it is a very accurate thing. Those radars are quite precise, and the other data is used there, too, to determine true record of where the vehicle went, altitude versus time, for example.

Now, our simulation is not constructed from that data. Our simulation is constructed strictly from the environmental models. And so we compare these two. Now, before you here we have a chart of what we call the dynamic pressure versus time. The dynamic pressure is air pressure.

It is the same force you feel on your face when you ride a bike, and it is a very important parameter in predicting air loads, and so we tried to very carefully reconstruct this parameter. The solid line is the simulation and the dot-dash - no, I am backwards. The solid line is the best estimated trajectory from the radars, and the dot-dash is reconstructed, and you can see that they are virtually identical.

This transient at the top is due to winds aloft, and that is not abnormal. You see excursions in this parameter when the vehicle encounters winds. Chart S-10, please.




(Viewgraph S-10) [Ref. 3/21-54]

DR. SCOFIELD: Next, we attempted a reconstruction of the flight dynamics between 50 and 72 seconds. We were interested particularly in these times of flight for obvious reasons. That is where the SRM leaked, and we have quite interesting dynamics in that region of flight. The match was a successful match, and the data acts as a check on the SRM leak model. The data was used or is being used in loads analysis.

Now, I might remark that Rockwell also is doing the same sort of activity, and we provide a check against each other. Rockwell is a systems contractor for JSC as well as building the orbiter, and they are doing these same studies, and so they have made a reconstruction similar to this one, and we have compared notes, oh, every two or three days here.

CHAIRMAN ROGERS: What do you hope to conclude by these reconstructions?

DR. SCOFIELD: Well, we hope to provide data that shows that the loads were within limit loads after you do the loads analysis. We hope to conclude that the SRM leak was in the position that I showed, and we hope to conclude that the guidance control system of 51L operated properly. We have not documented all of those things, Mr. Chairman. Some of them we are quite sure




of, but we haven't studied the test.

CHAIRMAN ROGERS: When do you think you will complete that?

DR. SCOFIELD: The first week of next month.

CHAIRMAN ROGERS: Okay. Thank you.




DR. SCOFIELD: All of the parameters of vehicle motion were within predicted envelopes, every one of them. Now, the predicted envelopes were generated early in the program. I don't mean a predicted envelope for 51-L here. I am speaking of the dispersions analysis that one goes through to provide data to thermal analysts and to provide data to loads analysts and that sort of thing, and those are large variations. You take very conservative assumptions about what you know regarding the environment in the vehicle, and those variations are, oh, 30 percent above what we saw in this flight.

Now, all of the parameters that were telemetered and reconstructed from 51-L were within flight experience up to 65 seconds. At 65 seconds we did begin to pick up some effects of the [1310] failure phenomena that drove us outside of our tube of accumulated flight experience. The peak gimbal angles, SRM gimbal angles, were within flight experience, but the total travel of those SRM gimbals was outside our flight experience.

Now, what I mean by that is if you take each increment of angle, each telemetered number, take the absolute value and add them all up, it gives you a measure of how far that rocket had to gimbal that angle




during the flight, and that parameter for this flight was 132 degrees. The largest previous was 125.

Let's have Chart S-11, and we will begin to show why that was.

(Viewgraph S-11) [Ref. 3/21-55]

DR. SCOFIELD: This is a plot of pitch attitude error versus time. Pitch is the up and down motion in the plane of flight, and the attitude error is the angle between where the vehicle is and where the guidance system wants it to point, the command angle. And once again you have 51-L measured data, dot dash, and the simulation in the solid lines.

Now, note the region between 40 seconds and 60 seconds. We have quite a transient response there. All of those variations are due to winds aloft, and we can reconstruct as we did from 50 seconds-on a simulation of this. We think that is very close to that. That is a very good reconstruction. We have not only this angle within 1 degree or so, but the slopes of the angle matches up very well with the telemetered data.

At 66 seconds or so we have a response to a planned roll maneuver that comes out of the guidance system. That is normal, but the long feature in that transient response leading to the lower right corner is




the response to the SRM leak. If the SRM had not been leaking, as hypothesized in all of these reconstructions, that curve would have turned back up and gone towards zero before 70 seconds.

So we know very well - it's a dramatic difference. We know very well that the SRM was leaking at that time. We know fairly well where it was. We have a fairly good measure of what the plume was. We have a pretty good measure of what the flow was. It all adds up. That is the bottom line.

Okay, Chart S-12, please.

(Viewgraph S-12) [Ref. 3/21-56]

DR. SCOFIELD: This plot is actuator extension on the left SRB versus time. Actuator extension is in inches, but actuator extension in inches is approximately the same as gimbal angle in degrees. The point I want to make from this chart is the response to wind gusts between, oh, 35 seconds and 60 seconds. There is a lot of activity there, and even though this parameter was within our flight experience, once again, the dynamic activity was not within our flight experience. This has been referred to by one of the guys on our team as a busy wind. It is a lot of motion.

Now, if we are incorrect about how much that joint opens up due to the loads, then this data would




become very important. We don't think that this type behavior would open a gap in a healthy SRM seal. It is certainly not designed that way.

MR. ACHESON: How close to the limits of your flight experience were the winds that were experienced in this frame of time?

[1311] DR. SCOFIELD: They were very close, sir, up around 60-something seconds. That peak up at 62 seconds is barely within our flight tube. The one at 41 seconds is well within our flight tube. We usually get more activity down in mid-flight, 40 to 60 seconds, than we do at 60 to 70 seconds. That is a healthy maneuver for that time of flight.

Now, I might remark that if we had a degraded structure at that time, these data could even be more important. We don't know the answer to that question. We don't really have a good way of determining the effect of this type, of behavior on a degraded SRM structure.

GENERAL KUTYNA: But, Mr. Scofield, what you are saying is we saw the flame appear at about 58 seconds, and if I can use an analogy, this chart shows the Shuttle going on a fairly bumpy road prior to that time with loads on that SRB. If you had a weakened joint that had sealed itself on lift-off, this could be




the factor that opened that joint up again, is that right?

DR. SCOFIELD: We are very much interested in that theory. We are working on it.

CHAIRMAN ROGERS: Can you give us a little more certain answer on that?

DR. SCOFIELD: Not at this time, no, sir. What we are waiting on is the next presentation, and the next presentation will show you some pretty sporty analysis that needs to be done to try to evaluate the degraded structures, or even a normal structure, for that matter.

If I could have Chart S-13, please.

(Viewgraph S-13) [Ref. 312 1-571

DR. SCOFIELD: This is a summary of findings. The propulsion performance was within the predicted limits until 69 seconds. At that time the right SRM broke outside our limit band. The SRM hole opened at approximately 60 seconds and grew to 45 square inches by 72 seconds. The measured winds aloft, propulsion reconstruction and aero increments based upon observed SRM plume give us a trajectory reconstruction that matched the best estimated trajectory, and a dynamic simulation that matched the flight data very well. We conclude that the control system worked properly until 72 seconds. It did




not after 72 seconds. When we got into structural breakup at 72 seconds, it was a bit much to ask of that system. It wasn't designed for that, of course. The parameters were within flight limits from the preflight simulation and were within the flight experience through 65 seconds. However, the SRM gimbal angle duty cycle, as we call it, was large due to winds aloft.

Now, that concludes my prepared presentation.

VICE CHAIRMAN ARMSTRONG: In determining the flight derivatives for the region of interest here with the plume, the CM-alpha/CM-beta, did you do those, determine those new values analytically, or did you use tunnel models?

DR. SCOFIELD: We did those analytically. Rockwell also did, and we compared notes, and we have a wind tunnel program planned. However, that probably will be several weeks off yet. Those things, as you undoubtedly know, take a little time to do. We have to make a model which includes-we have to make a wind tunnel model which includes some sort of representation of that plume, of course.

VICE CHAIRMAN ARMSTRONG: But with the analytical ones you can do a bit more sensitivity analysis with the geometry, is that correct?

DR. SCOFIELD: Of course, and that is going on today.


[1312] 2278


DR. LITTLES: Mr. Chairman, the next section will be presented by Byb Ryan, and there are two subjects. One is the data that we have used to conclude that the loads were well within the design limit, and the second section is the work that we are doing relative to the joint dynamics and applying the specific 51-L loads to that joint.

MR. RYAN: Mr. Chairman, Commission members, I have been in aerospace engineering for 30 years. I started out with ABMA in January 1956, and went with NASA at the organization of the Marshall Space Flight Center, and been Chief of the Structural Dynamics Division since 1974.

What I would like to do today is give you a status report because I don't have a full report of all of the structural analysis, it is not complete, of the 51-L mission, including the aft field joint in particular, since it is of primary interest. And so I will split my presentation into those two parts. The first part I want to discuss is the overall load.

So if I could have Chart R-1.

(Viewgraph R-1) [Ref. 3/21-58]

MR. RYAN: The objectives of this analysis that we are doing is to determine the loads experienced by 51-L for all flight events, compare those loads to




the expected design loads and determine if excessive or unusual loads occurred. And then, thirdly, we want to extract from those loads forcing functions for special dynamic analysis. In particular, we want to run response analyses of the SRM, solid rocket motor field joints to see how the gap would open for the different events.

Could I have Chart R-2, please?

(Viewgraph R-2) [Ref. 3/21-59]

MR. RYAN: Those events that we will be reconstructing the flight loads for are on the pad, the liftoff, the in-flight - the in-flight covers all of the dynamic regions that have been discussed by Dr. Scofield as the major events of the gust, the roll maneuver and so forth. To do this, a dynamic model of the 51-L vehicle, its payload, the MLP pad for the lift-off phase, etc., has been generated, and the vehicle then is flown in a simulation through all of these different flight events and different environments that have been created that Dr. Scofield talked about. Also, we made use of telemetry data and certain measured data that were available on the pad for our work.

This work is a joint effort that is carried out by Marshall Space Flight Center, Johnson Space Flight Center and Rockwell International at Downey.




The next chart, which is R-3 -

(Viewgraph R-3) [Ref. 3/21-60]

MR. RYAN: - I have chosen to illustrate the dynamic situation that really occurs during the liftoff phase of the maneuver. This is the most dynamic situation that we have. It is where you saw the puff of smoke. It is where you see all of the structure wringing out. A lot of things are going on at that time, and I would like to take just a minute to step through those basic events for you so that you might get a feel for what takes place.

I have chosen only two parameters to plot out of many parameters that we could plot versus time. The time base I am using here is when we give the commands to start the flow in the Shuttle main engines, all the way through to approximately four seconds after lift-off. Now, the two parameters I have chosen to plot for you are at the aft field joint of the right solid booster, [1313] since that one is of concern to everybody. As I say, we could plot all kinds of parameters at any other location on it.

Now, if you look at the left hand side during the first two seconds, there is really nothing going on in terms of loads and so forth. The vehicle is sitting on the pad. The weight of the vehicle you see there




indicated by the axial load, that trace, and that is a compressive load in the aft field joint which is like the weight of it sitting on it.

The other parameter I have chosen to plot is the bending moment in the Y direction, that is in the pitch direction, which would be the bending moment that would bend the vehicle this direction if you looked at the model there.

Now, at approximately 2 seconds after we start all the process, the SSMEs start coming up in thrust, and as they come up in thrust, you can see a slight decrease in the axial load. In fact, the thrust of the SSMEs are lifting the vehicle up, particularly the orbiter, reducing some of the weight that is in the SRM joint, and the vehicle is bending over, and it is bending over away from the orbiter. Of course, the orbiter and everything is bending with it, and you can see that in the films. In fact, that deflection is quite dramatic, and if you look down at about 5-1/4 seconds, you see the peak of that bending moment. At that time the SRB, the solid rocket booster motor, has bent approximately 24, 25 inches, and the tank is bent like 32 inches.

Now, we give the signal to light the solid rocket booster motor and to release the hold-down bolts




that hold the vehicle on the pad such that we have a minimum amount of energy stored in the vehicle, and that occurs there, as you can see, a little past six seconds on the chart is that minimum moment part. You see literally the vehicle is being bent over, is being pushed, which rolls the SRBs, and we are storing a tremendous amount of energy in that structure. You are winding it up like a rubber band, so to speak. And then when the SRB thrust comes up and we get a thrust-to-weight of greater than one, the vehicle lifts off the pad, and it releases from the pad, and all of that energy is released into a dynamic motion.

Now, you can see that occurring from about 6.25, 6.35 seconds on in the MY, and you see that frequency that everybody has been talking about wringing out. That is actually bending moment wring-out in the structure at the aft field joint.

Now, later I will talk about how that corresponds to an effective gap opening, but the structure is wringing out in some very complex structurally dynamic modes at this time, this being mainly a 2.7 to 3 hertz mode, but there are two or three modes in this region that the vehicle is wringing out. So those are the basic events that are going on.

DR. COVERT: Mr. Ryan, could I ask a





If we had plotted the axial load curve earlier in time, you would have it for, not for the whole assembly but for each solid rocket booster itself, would it be possible to see the dynamic effect on the joint and the pins as the load transfers from being down to being up?

MR. RYAN: Yes, sir, that occurs, as you see that, around 6

DR. COVERT: No, I am talking not about the whole assembly coming off of the booster, but, rather, as I pressurize the rocket case.

MR. RYAN: Yes, sir, it would do the same thing. It goes in tension just due to pressure.

[1314] DR. COVERT: Are you going to talk about that today?

MR. RYAN: Yes, sir, I will talk about that later in the second part.


MR. RYAN: So those are the dynamic events that go on that affect the joint and affect the loads.

Now, to evaluate that we did three things. We had the film coverage that you have seen some of many times, I am sure, and we observed those films, and compared it with several flights. In particular, we




compared 51-L with STS-6, which was a comparable flight, and the deflections on the SRB on 51L was 24 inches, and on STS-6 it was 28 inches. These are within, the envelopes of what we have been seeing and indicates that everything was okay, that it was a nominal flight at liftoff. The frequency of that oscillation that occurs is about a quarter of a hertz for both vehicles, in fact, for all of our vehicles, which indicates that the pad effects and so forth were the same. That's the important conclusion there. [Ref. 3/21-61]

We have two other modes that go on there. When the Shuttle main engines come up, due to the fact that the nozzle is not full and the thrust wallowing around in the nozzle, those loads appear to be nominal. One is a 25 hertz load and the other is a 30 hertz load. Dr. Scofield talked about the clearance, and that was nominal.

Now, R-5, please.

(Viewgraph R-5) [Ref. 3/21-62]

MR. RYAN: Now, we also evaluated the loads during the twang motion we had. On each hold-down post of the SRBs, we had strain gauges, and we have those on the flights to tell us how much loads go into those SRB hold-down bolts or the posts. Those are designed by that twang motion that I showed you earlier when that MY




moment peaks. We evaluated those, and you can see on the chart that I have here, if you look down at the base of that chart where the arrow points to the base of the solid rocket booster motor, you can see that the design bending moment was 347 million inch pounds. We actually experienced by measured data on this flight 291 million inch pounds, which is about typical of what we see. If I transfer that same measured data up to the aft field joint of the right solid rocket booster motor, the design limit load there would have been 248 x 106 million inch pounds. We experienced 208 by transferring that measured data to that point. And so, in addition, we could observe some of the posts. We couldn't observe them all. But we could also observe from the film, we could observe from the strain gauge standpoint each one of those four hold-down posts on each SRB released nearly simultaneously, which tells us that there was not undue energy being transferred and undue motion being transferred into the vehicle at lift-off.

VICE CHAIRMAN ARMSTRONG: How did you determine that near-simultaneous? [Ref. 3/21-63]

MR. RYAN: By the timing, sir when the strain gauges show that the load has been released on that hold-down bolt and they were within a couple of milliseconds of each other, which says that there was very little abnormal energy or




unsymmetrical energy or unexpected energy in that part of the liftoff.

Could I go to Chart R-7, please?

(Viewgraph R-7) I Ref. 3/21-64]

[1315] MR. RYAN: The third way that we have evaluated the loads by the different parties that I mentioned earlier is a reconstruction of the liftoff loads used in the dynamic model. I want to use this chart, R-7, to show you where I am going to talk about some loads later. If you notice the Ps that are at the back end of the graph, you have the struts. We called the loads for each strut a P-1, P-2, P-3 and so forth, and if you will notice there that P-1 through P-7 are the ET to orbiter struts and are indicative of the loads that the orbiter relative to the ET are experiencing during the liftoff. And then if you notice the P-8 through P-13 are the loads, the struts that attach the two solid rocket booster motors to the external tank and are indicative of the loads experienced there.

Now, on Chart R-8

GENERAL KUTYNA: Before you go off that chart, is P-11 about where we think we saw the flame?

MR. RYAN: It is very close, yes, sir.

(Viewgraph R-8) [Ref. 3/21-65]




MR. RYAN: All right. On Chart R-8, on the first seven flights of the Shuttle, development flights, we had measured data on some of the struts at the strain gauges, and we measured that data, and I am comparing the mean of that data and the standard deviation of that data with the predicted reconstructed loads for 51-L, and then on the far right column I have the design loads, the limit loads.

Now, these do not have the safety factors in them. All loads are multiplied by a factor to put margin in them above this. So that is not in this limit load. And so you have got the safety factor above that. So you can see that the 51-L reconstructed loads of the struts are well within the design loads, and are comparable to the loads that we have experienced in the first seven flights.

GENERAL KUTYNA: Mr. Ryan, you know what I am going to ask you, but if you look at P11, that strut I just asked about, if you look at STS-1 through 7, we have got a number of about 70, and on 51-L it is 141.

Would you explain why that is not a concern?

MR. RYAN: Yes, sir, I would be glad to.

On STS-1 through STS-5 and on STS-7, we flew a heavy external tank. That tank had a marginal condition in the bulkhead region, and because of that we put the




margin back into it by preloading the struts. In other words, we put the struts in compression so that when you loaded the tank with its cryogenic propellant and it shrunk, then it would make up that difference and take that load out.

GENERAL KUTYNA: As you look at that load, however, this is in the area of that joint where possibly it failed.

Would there have been any structural deformations as a result of that load that might have been a factor?

MR. RYAN: It should not have, sir, with a 306 versus a 141 SR design and then the 1.4 times that. So that would be a big margin, really.


DR. RIDE: If you compared the 141 with others that we have seen on P-11 using the lighter weight tank on Flights STS-8 and subsequent, is that in the general ball park of experience?

MR. RYAN: Yes, it is. We have not had measured data, though, Dr. Ride, on these other flights, and so we have to do it just strictly by analysis.

[1316] Okay, let me go to Chart R-9, please.

(Viewgraph R-9) [Ref. 3/21-66]

MR. RYAN: Taking the same loads and the same




reconstruction of the liftoff motion to the SRB or the solid rocket motor field joints, I have put on that chart the design loads in terms of an equivalent load, just to keep it fairly simple. All of the joints are designed to the same load. The forward joint determines that design load, and that is the - 17.2 x 106 pounds.

Now, as you come down the stack, because of bending and because of pressure drops and hoop strains and so forth, that load drops for the aft field joint in 51-L - these are the reconstructed values-the forward field joint would have been - 15 x 106. If you drop down to the aft field joint, the one with the leak, you see a - 4.1 x 106. So you could conclude, and some people have concluded, without damage to that joint, then, another joint should have leaked first because it should have had a higher load and a slightly higher opening. There is not a large difference, but there is a difference between the loads in those joints, and the forward joint does have the highest load.

All joints are designed alike. They are equal and interchangeable.

Now, Chart R-10, please.

(Viewgraph R-10) [Ref. 3/21-67]

MR. RYAN: Dr. Scofield talked about the reconstruction of the trajectory and the loads during




the in-flight regions, the roll maneuver through the maximum dynamic pressure region. That reconstruction has not been completed, and we do not have all of the loads for that reconstruction. However, to get a feel for what those loads were and what we were dealing with there relative to what was seen, we took a simulation that was made at a couple of times, the L minus three and a half hour wind and the L minus zero wind, plus estimations of the loads that occurred due to the gimbal angle excursions that he showed you, and predicted what - and we think they are very conservative loads - predicted what we will get out this coming week out of the total loads reconstruction. And you can see compared to the design loads that in the struts the loads are a good bit lower than the design limit loads.

I have put one load indicator there for the tank. You can see the loads going into the tank there where the struts go into the tank are substantially lower than the design loads. And you can see that even in the SRM right field joint, that that load is substantially lower than the expected design load.

So on Chart R-11 then we conclude that the 51-L systems loads, the overall loads of the vehicle were lower than the design limit loads and were essentially within our flight experience. However, the [Ref. 3/21-68]




effects of those loads on a degraded solid rocket booster aft field joint is in the process of being assessed.

And the next presentation that I am going into, if there are no questions on this, I will talk about where we are at in that analysis.

DR. COVERT: I have a question or two, Mr. Ryan. As I understand what you have been talking about, you are dealing with static strength of the structure and static loads of the structure, is that correct?

[1317] MR. RYAN: I am giving you the equivalent static design loads, yes, sir, derived from dynamic and static loads.

DR. COVERT: Is metal fatigue a critical factor in any of the structural design of this solid rocket booster case?

MR. RYAN: Well, I don't have the total answer on that, Dr. Covert. We will be glad to give you a review of that when you come to the task team. It is designed for 20 reuses. It is basically a pressure vessel, and that is the essential design consideration.

DR. COVERT: If a crack of suitable length existed, is it possible that a case failure could have existed while the case was, to use General Kutyna's words, driving over that bumpy road?

DR. LITTLES: We have added, Dr. Covert, one




item that I mentioned earlier on the fault tree associated with case rupture, and we intend to present the data on that next week to the task force. But relative to your specific question, I believe the program is set up such that if we had a flaw which would have been driven to a leak-through type situation with the loads we had on this flight, since they are within design limits, then we should have screened that out in the proof test that we run on the case. But we are accumulating all of those data, and we will specifically address your question relative to the fatigue in closing out that action item.

DR. COVERT: I just want to mention in passing that, if I recall, in October of 1983, a crack of sufficient size did slip through the inspection, and it was fortunately caught by the proof testing, so that I guess my point is that it is a little premature, I think, to paint this thing too green.

DR. LITTLES: Well, we are certainly accumulating that data, and we will pursue that.

DR. COVERT: Thank you.

MR. RYAN: Are there any other questions on the first part?

DR. WHEELON: Perhaps just to clear my own mind, if I may, to sum up what I think I heard you say,




the forces and the bending moments, or the torsion on the SRB was about normal. There was nothing unusual in that save the temperature, which is outside your expertise, but there is more work to be done to translate these forces and moments into the tang-clevis area, is that right?

MR. RYAN: Yes, sir, that is what I am talking about next.

DR. WHEELON: Thank you.

CHAIRMAN ROGERS: You may proceed.

MR. RYAN: Could I go to Chart R-13, please?

(Viewgraph R-13) [Ref. 3/21-69]

MR. RYAN: The objective of this analysis that we are conducting both at Marshall and at Thiokol is to reconstruct the 51-1 SRM field joint response, in other words, the gap openings that some of the questions that are being raised here, for all of the events using the mated data, the natural environments, the induced environments, and the reconstructed 51-L loads.

If you will look at Chart R-14 -

(Viewgraph R-14) [Ref. 3/21-70]

MR. RYAN: - we feel that we have to start with the mating of the SRM segments. Where you actually start with the first segment, mate the second one to it, take all the dimensions and so forth there, and the


[1318] 2294


loads, etc., that they would experience and move through the stacking, the transportation, the on-pad, and then the other conditions that I have talked in the previous section, which was the lift-off dynamics, the roll maneuver, and the maximum dynamic pressure or the in-flight regime which really covers all of the areas that Dr. Scofield talked about from approximately 35 seconds on.

Chart 15 then

(Viewgraph R-15)[Ref. 3/21-71]

MR. RYAN: - is the approach that we are taking to do that analysis. We are developing finite element static and dynamic models of the joint, the field joint, and the structure around it, carrying enough of the structure out through the segments to adequately describe what happens there. We are conducting tests at Thiokol now on joint rotation, and we are tuning these models such that they match or duplicate the tests that are being run at Thiokol on joint rotation.

We are adding to these models, then, the SRM segments, the propellant effects, and the structure, like the ET attach ring and so forth. Then we are characterizing the initial joint condition. In other words, every joint that is mated is slightly different




because of the ovality and so forth between the tang and clevis, and we are in the process of characterizing those conditions. And then they will be used in conjunction with this model and with this dynamic analysis to say then what was the joint opening on each of the joints, in particular, the aft field joint of the right solid rocket booster.

Then we will determine then the static and dynamic response of the gap at the primary and secondary seal.

Now, Chart R-16 is a chart that kind of summarizes for you some of the different types of models that we have.

(Viewgraph R-16) [Ref. 3/21-72]

MR. RYAN: If you look at the top right hand corner you see the finite element model of the clevis and tang. You see just under it the deformed model. In other words, if I load that with internal pressure or if I load it with a line load or a punch load, you see that that clevis deflects in particular, and it deflects away from the tang, creating a gap between the seals and the joint.

Now, down below that is the aft attach ring. That is our model of the aft attach ring that the struts are attached to, and it is attached to the overall SRM




segment. Now, the picture on the left hand side of the chart shows all of that put together with a one-diameter length of our model of the segment fore and aft of that aft field joint.

Okay, now, taking that model and taking all of the forces that were recreated or reconstructed in the liftoff loads analysis that I talked in the first session and Dr. Scofield talked in the liftoff sequence, took all of those and got out of that, extracted out of that the forcing functions, including the internal pressure rise of the SRM, and all of the loads that would go into that section of the solid, the inertial loads, the point loads going from the struts, etc., we reconstructed that and drove that model. And on Chart 17 -

(Viewgraph R-17) [Ref. 3/21-73]

MR. RYAN: - I show you a maximum and minimum response during the short period Of time when the SRM pressure is coming up, and all that dynamic load that I showed you earlier is going on.

[1319] Now, the main effect that you see here is the effect of the pressure bulging and stretching the SRM, and you see that gap opening that is given in mils there. You see that gap opening, and you see it occurring very fast due to that pressure. As that




pressure builds up in the solid, the thrust comes up, it opens, and it opens very fast, and you can see there that it is over only about 600 milliseconds. Then you see in addition, and it is not as clean as it looked on that MY a while ago, but you see in addition the three-hertz oscillation in that gap opening, although it is a very small opening, plus or minus a mil and a half or so, you do see that opening taking place, opening and closing of that gap. It is not really opening and closing. It is opening. It is just a delta opening and closing relative to the open position.

GENERAL KUTYNA: Now, Mr. Ryan, this is a pretty important chart. What it says is that gap opens as much as .025 to .030 as soon as you light off that SRB.

MR. RYAN: As soon as I reach maximum pressure, yes.

GENERAL KUTYNA: And what you are telling me is if we had a metal-to-metal contact in the area of that seal, we could open that gap as much as .025 to .030, and the seal would have to follow that opening very quickly, within a half a second, to remain in contact with the other piece of metal.

MR. RYAN: Yes, sir.

GENERAL KUTYNA: And the second thing you are




telling me is that 3-hertz; opening or vibration of that gap, possibly, or at least does not dispute the data that says the smoke was puffing at about three cycles per second.

MR. RYAN: That's right, it doesn't dispute it. It doesn't fit it necessarily, but it doesn't dispute it. That's a very small opening and closing, but it does happen.

DR. COVERT: Mr. Ryan, one other minor point here. My memory is not very good, so please refresh it. This 25-mil change corresponds to what change in the percent of squeezing on the 0-ring between the static condition and in the as-deflected?

DR. LITTLES: I believe about a .040 is a squeeze of somewhere around 12 percent, I believe. I think that is correct. So we are reducing it by whatever fraction that is.

DR. COVERT: And the 25 would take it down around 7 percent?


DR. COVERT: Thank you.

DR. WHEELON: Could I just editorialize on this data for a moment and perhaps say the obvious a second time?

We have this rapid expansion of .025 probably




in this clevis-tang area, and we are counting on the seal to follow it, which is to say, expand and take up that difference, and it is true that cold seals expand more slowly or have more difficulty in doing that adapting than warm seals do.

MR. RYAN: There are two things that cause a seal to seal, though. One of them is the fact that the aerodynamic pressure around that gap is more on the inside than it is on the outside, which drives the seal out also. So you are not depending just on resiliency to get the seal out into the gap to seal. So you have to put the two effects together, sir.

The next chart, let me go to R-19.

(Viewgraph R-19) [Ref. 3/21-74]

[1320] MR. RYAN: R-19 is just a static analysis where we can put a little more detail in than we can in a dynamic analysis, but we can't show the dynamics, of the same pressure rise buildup rate, and it shows essentially the same thing that I showed previously, but without the dynamics.

Now, we are continuing this analysis. We have not run it for the Max Q region, the roll maneuver and these kind of things that you saw, and that is in process. When we get the reconstruction of the loads, we will be doing the same thing. We are also




continually trying to refine this model and to try to predict more accurately just how these gaps and seals behave.

DR. COVERT: Mr. Ryan, I think this is part of the answer to the question that I asked you earlier about the rise time in the longitude. This looks - I realize the motions are small, but it looks like a fairly rapid change in position.

Is there an impact load as this thing seats at the other side of that motion?

Now, remember, these pins are - you can sort of put them in with your thumbs, so there is a certain amount of slack in the system, and I am sure you have been on a Pullman in the middle of the night when they have stopped and started suddenly.

MR. RYAN: Yes, sir, this is a nonlinear problem, obviously, and you are talking about a very difficult analysis. The results I presented to you today was a linear analysis that didn't take into account those contacts. That is very complex and very time consuming for us to step through iteratively to do that. We are looking at that. In the particular data I showed you there they did not contact. But there are conditions where they will contact, and that does give you a bouncing and a twanging motion, and as I say, we




are looking, and we know we can do some of that type of analysis. But it is very complex.

DR. COVERT: Is there any evidence when you examine the holes in either the clevis or the tang where these big pins go through that there is permanent set of the order of maybe .003 to .005, that sort of thing? Do the holes stop being round and become sort of elongated?

MR. RYAN: I can't answer that, sir. Maybe Dr. Littles can. I haven't been involved in that side of it.

DR. LITTLES: I can't answer that specifically. I am not aware of that, but I would have to check on that.

DR. COVERT: Would you get that for me?

DR. LITTLES: Yes, I would.

DR. COVERT: Thank you.

MR. RYAN: That concludes my discussion, Mr. Chairman.

DR. LITTLES: Mr. Chairman, we need to change speakers at this point.

CHAIRMAN ROGERS: I assume that all the tests you are conducting are for the purpose of assisting the Commission and NASA, too, in deciding the cause of the accident. Are you also going to use that material, or are you thinking about it in connection with redesigning




the seal?

MR. RYAN: Yes, sir.

DR. LITTLES: We will certainly use anything that come out of those tests or the analysis which indicates any problem with that joint, in the redesign, certainly.

CHAIRMAN ROGERS: But it is primarily to determine the cause of the accident?

[1321] DR. LITTLES: At this point in time, yes, sir, these tests and analyses are to determine the cause of the accident.

CHAIRMAN ROGERS: Aren't tests being run to think about redesigning the joint?

DR. LITTLES: We are certainly looking at things that would be necessary to change in that joint to resolve the kind of things that may have gone wrong, and we are looking at that, yes, sir.

MR. RYAN: There are analyses being conducted now, parametrically, to determine some of that at Marshall and Thiokol, too.

CHAIRMAN ROGERS: Do you have separate teams doing that?

DR. LITTLES: Yes, sir. It is different people. It is not the people who are involved in the investigation, although they are aware of what we are




doing and we are keeping them up to speed, but it is a different group of people.


DR. WHEELON: Mr. Chairman, I was going to hold this question until later, but I sense we are about ready to lose the right people from the witness stand, so perhaps I could flash back.

Could I invite your attention, Dr. Littles, to W-10? This was the top view of the Shuttle and the orbiter, and it shows the camera obscuration I think in a very skillful way it localizes the rotation around the SRB, the position of the probable leak that caused the black smoke at takeoff, and incidentally precludes the possibility that it was the check port doing the leaking.

(Viewgraph W-10) [Ref. 3/21-40]

DR. WHEELON: Now, this addresses two of the three dimensions, namely, where around the azimuth is the leak likely to take place by camera elimination. Have you thought about the vertical, the other dimension, where along the length of the SRB was the smoke coming from? How consistent is that with the up and down location of the field joint? And incidentally, if there were a leak in the clevis-tang combination, it seems to me the smoke would be shooting




up relative to the stack. And that is in fact what it does. So you are trying to put together the other dimension of the three dimensions of this leak, potential leak location puzzle?

DR. LITTLES: Yes, sir, we have tried to do that, and as you know, the smoke, when you first see it in the film, is not at the location of the leak check port. It is some distance up. And it is apparently coming from the hidden location, and what we have tried to do, we have done an analysis looking at a jet that would emanate from an opening of the dimensions we have in the tang clevis, and using as parameters the pressures that you have in there, and all of those variables, to try to establish at what point in time we might have had to have that leak coming from the joint area to see the smoke at the vertical locations at a given time. And the bottom line of all of that analysis is there are so many variables in it that, depending upon what assumptions you make within the bounds of those that are reasonable, that you can get that time to be anywhere within, where it might first come out, within 50 milliseconds of when you first see it or as far back as 400 or 450 milliseconds. And so that shows you something about what kind of-actually, you can predict the vertical location as well, because those would be




substantially tied together.

There are just so many variables involved in that problem that you just can't tie it down very well. But we have tried to do that.

[1322] DR. WHEELON: But simplifying all of those variables in that very complicated problem, simplifying it just a bit, isn't it likely that if the leak were occurring in the shaded area, that is to say, the obscured area, and if the flow was going vertically, because that after all is the way the opening in the tang-clevis combination is, that you would first see on any one of these cameras the puff of smoke above the opening?. It would have shot out of sight vertically and then come into view only as it expanded and expanded out?

Would you agree that that is still consistent with the idea the smoke was coming from the O-ring?

DR. LITTLES: Very much so, yes, sir. We feel it does come from the joint, yes, sir.

DR. WHEELON: What was the significance of your remarks about the timing, the delay?

DR. LITTLES: What we were trying to establish was when in the timeframe of the ignition sequence that the leak occurred. It would be of interest to know whether it is close to the maximum pressure time or




whether it is back down during the early phases of the ignition transient, and that relates to one of the scenarios we are working, which is Scenario 6, which deals with putty holding the pressure off the O-rings, the joint rotating, and then a lack of resiliency and the gas blowing through the putty and by the O-rings. So that is the area of interest there in Scenario 6.

DR. WHEELON: But before, again, you get away, isn't it clear that these optical observations of the black smoke near liftoff and the flashes later on come from about the same angular azimuthal area on the tank and the same vertical or linear dimension, and doesn't the web of evidence seem to be closing in on the location of whatever has happened, or do you want to reserve on that?

DR. LITTLES: I think the location of what is happening, we are narrowing it down very well.

DR. WHEELON: That's my sense, you've really got it.

CHAIRMAN ROGERS: I think if there is no objection, we will take a recess for lunch and resume after lunch.

Why don't we say 1:45, and Dr. Littles will not get away.

(Whereupon, at 1:00 o'clock p.m., the




Commission recessed, to reconvene at 1:45 o'clock p.m., this same day.)



[Please note that some of the titles to the references listed below do not appear in the original text. Titles are included to identify and clarify the linked references- Chris Gamble, html editor]
1323] [Ref. 3/21-31 1 of 2] Fault Tree and Failure Scenario Update. [Ref. 3/21-31 2 of 2] 51-L Fault Tree.

[1324] [Ref. 3/21-32] 51-L Fault Tree. [Ref. 3/21-33] SRM Hot Gas Leak Failure Scenarios.

[1325] [Ref. 3/21-34 1 of 2] External Tank. [Ref. 3/21-34 2 of 2] External Tank: Investigation Approach.

[1326] [Ref. 3/21-35] Fault Tree Evaluation.

[1327] [Ref. 3/21-36] Fault Tree Evaluation.

[1328] [Ref. 3/21-37 1 of 2] Fault Tree Evaluation. [Ref. 3/21-37 2 of 2] Fault Tree Evaluation.

[1329] [Ref. 3/21-38] Photo of Right-Hand SRB attach point to the ET.

[1330] [Ref. 3/21-39] Closer view of Ref. 3/21-38.

[1331] [Ref. 3/21-40] Camera Location.

[1332] [Ref. 3/21-41] Damage to External Tank at Liftoff (continued).

[1333] [Ref. 3/21-42] Damage to External Tank at Liftoff (continued). [Ref. 3/21-43] Overheating External Tank From SRM Hot Gas Leak.

[1334] [Ref. 3/21-44 1 of 2] Leak Check Port (Smoke at liftoff). [Ref. 3/21-44 2 of 2] Failure Tree Chart.

[1335] [Ref. 3/21-45] Camera Location. [Ref. 3/21-46] Camera E-63.

[1336] [Ref. 3/21-47] Probable Location of Smoke.

[1337] [Ref. 3/21-48 1 of 3] Trajectory and Flight Dynamics Reconstruction.

[1338] [Ref. 3/21-49 1 of 2] Propulsion System Performance. [Ref. 3/21-49 2 of 2] Right SRM Internal Pressure.

[1339] [Ref. 3/21-48 2 of 3] Trajectory and Flight Dynamics Reconstruction. [Ref. 3/21-48 3 of 3] Flight Reconstruction In-Flight.

[1340] [Ref. 3/21-50] Liftoff Dynamics. [Ref. 3/21-51] Atmospheric Environment and Aerodynamics.

[1341] [Ref. 3/21-52] SRM Leak Model.

[1342] [Ref. 3/21-53 1 of 2] 51-L Comparison of Reconstructed and Best Estimated Trajectory Dynamic Pressure Profile.

[1343] [Ref. 3/21-53 2 of 2] Trajectory and Flight Profile Reconstruction. [Ref. 3/21-54] Flight Dynamics Reconstruction.

[1344] [Ref. 3/21-55] Flight Dynamics Reconstruction.

[1345] [Ref. 3/21-56] 51-L Left SRB Actuator Extension Measured Data.

[1346] [Ref. 3/21-57] Summary of Findings.

[1347] [Ref. 3/21-58 1 of 2] Load Analyses.

[1348] [Ref. 3/21-58 2 of 2] Objectives. [Ref. 3/21-59] Flight Events.

[1349] [Ref. 3/21-60] Liftoff Sequencing: Axial Load and Pitch Plane Bending Moment.

[1350] [Ref. 3/21-61] Liftoff Film Evaluation.

[1351] [Ref. 3/21-62] Space Shuttle Liftoff Transient. [Ref. 3/21-63] 51-L Measured data.

[1352] [Ref. 3/21-64] Shuttle Strut Identification.

[1353] [Ref. 3/21-65] Liftoff Measured Strut Loads Versus 51-L Calculated.

[1354] [Ref. 3/21-66] Space Shuttle Pre-liftoff Transient.

[1355] [Ref. 3/21-67] Max "Q" Results.

[1356] [Ref. 3/21-68] Findings.

[1357] [Ref. 3/21-69 1 of 2] 51-L SRM Field Joint Analysis.

[1358] [Ref. 3/21-69 2 of 2] Objective. [Ref. 3/21-70] Scope.

[1359] [Ref. 3/21-71] Approach.

[1360] [Ref. 3/21-72] SRB 3D Solid Element/1 Dia. Length Model.

[1361] [Ref. 3/21-73] Max-Min Primary Seal Gap Opening Liftoff.

[1362] [Ref. 3/21-74 1 of 2] Static Gap Results.

[1363] [Ref. 3/21-74 2 of 2] 2D Axisym Elem. Static (Joint Rotation Test).

[1364] 2308




CHAIRMAN ROGERS: Will the Commission come to order, please?

DR. LITTLES: Mr. Chairman, we will now go into the final three topics. The first one will be a short review of an O-ring analysis model. This model has been in existence for some long period of time. We are using it in this investigation, and Garry Lyles will tell the Commission what the pedigree of that model is and how it has been verified, and then Rick Bachtel will go into the work that has been going on relative to the joint thermal flow analysis. Primary emphasis of this analysis is trying to establish whether we could have had a continuous link from the time we first saw the puff of smoke until 58 or 59 seconds.

(Witnesses sworn.)






MR. LYLES: My name is Garry Lyles. I have been an employee of Marshall Space Flight Center for ten years in the area of propulsion systems analysis, propulsion analysis, dealing in the area of internal flow dynamics, and as Dr. Littles stated, I will present the analytical modeling that has been done to predict O-ring erosion and the model's application as a tool to evaluate possible 51L O-ring failure mechanisms.

Could I have Chart L-1, please?

(Viewgraph L-1.) [Ref. 3/21-75]

MR. LYLES: As an introduction to the O-ring erosion scenario and the two types of O-ring erosion that we have modeled and can cause erosion of the O-rings, what I would like to do is talk from the picture, and I would like to zoom in, if I could, on the area of the putty and the 0-rings just a little bit. The hot gas enters the cavity between the putty and the primary O-ring through a path in the putty, this path that we have been calling a blow hole.




The blow holes range in size. They are nominally one-inch width and flow between the insulation to the primary O-ring, and this causes impingement of a hot gas jet on the O-ring, and this impingement erosion continues until the cavity in front of the primary O-ring equals motor pressure, and this occurs during the pressurization transient of the SRB or the solid rocket motor. [Ref. 3/21-76]

[1365] That is the impingement erosion, and that scenario assumes that the primary O-ring seals. The other type of erosion that we have modeled is the erosion which could occur if the primary O-ring does not seal, and in fact we get leakage by the primary O-ring. In this case, we predict what we call blow-by erosion of the primary O-ring. The hot gas then continues on into the cavity between the primary and secondary O-rings, pressurizing that cavity until the two cavities equalize, and we also predict impinging erosion on the secondary O-ring.

Could I have Chart L-4, please?

(Viewgraph L-4.) [Ref. 3/21-77]

MR. LYLES: Chart L-4 is a description of the analysis and the modeling that we have done in this analysis. The first part of the analysis is a cavity pressurization model which we have modeled as a simple




lumped-parameter cavity pressurization model in which we solved the conservation of mass and energy equations by numerical integration.

The model includes the solid rocket motor pressurization transient which is a function of time, of course. It models the change in volume of the seal cavities as the seal rotates or the tang and interleg of the clevis move apart. It also models the heat transfer from the gas to the metal parts and predicts a cavity pressure and temperature which predicts how long a gas jet can flow into the cavity, and how long it can impinge on the primary O-ring.

We also have an impinging heat transfer model or an erosion model of the O-ring based on impinging heat transfer coefficients in which, again, the hot cast jet pressurizes the O-ring cavity, impinges on the primary O-ring, and we erode the O-ring by stagnation point heat transfer.

CHAIRMAN ROGERS: If I may ask you a question, is this sort of a general study of the 0-rings, or is this trying to simulate what happened in the accident, or what?

MR. LYLES: This analysis originated about a year ago, and it originated because we were getting some erosion of the O-rings, and we wanted to study that and




determine if we could the cause and try to -

CHAIRMAN ROGERS: In other words, this has been a study that has been going on for a year?

MR. LYLES: Yes, sir, and it started about a year ago, and it was presented in, I think, July of last summer.

CHAIRMAN ROGERS: And why was it started?

MR. LYLES: It was started because flight data on disassembly indicated that we were getting some erosion on the primary O-ring.

CHAIRMAN ROGERS: Which flight caused you the most concern or gave rise to the study?

MR. LYLES: Well, it was - I think it was just the general fact that we were getting erosion. I am not sure it was a specific flight. We did

CHAIRMAN ROGERS: Well, what prompted it? There must have been something that prompted it.

MR. LYLES: Well, I think at the time the model was first created, we were getting erosion, we started getting erosion in the field joint, and we had seen some erosion in the nozzle joint, and those data points indicated to us that an analysis needed to be performed to determine the cause of this.

CHAIRMAN ROGERS: How large a group worked on this?


[1366] 2313


MR. LYLES: This was originated by Dr. Salita at Morton Thiokol, and he developed the model. We then, after the model was presented to us, we brought the model in-house and audited the model, and have exercised the model, and have made some modifications ourselves to it.

CHAIRMAN ROGERS: Who worked on that project?

MR. LYLES: Well, Dr. Salita, at Morton Thiokol worked on it. Out of my group I have worked on it. And a couple of engineers in my group have run the model.

CHAIRMAN ROGERS: Did it have a name? Did you have a name for the group that worked on it?

MR. LYLES: Yes, sir. The engineers in my group were Sam Lowry and we've got another young engineer, John Hutt, who worked on the model, and we are the ones.

CHAIRMAN ROGERS: Were any of you involved in the telecon the night before the 51L launch?

MR. LYLES: No, sir.

CHAIRMAN ROGERS: Were any of you asked to give information about the work you had been doing for that group on the telecon?

MR. LYLES: No, sir.

CHAIRMAN ROGERS: Do you have a report of the work that you have been doing on that joint, anything in writing that your group has produced?




MR. LYLES: Dr. Salita has published two reports on the work that has been done.

CHAIRMAN ROGERS: When were those reports made?

MR. LYLES: I think Part 1 of the report was issued in July of last summer. The second report, I believe it was the following month.

CHAIRMAN ROGERS: Did they point out in those reports- are you familiar with the reports?

MR. LYLES: Yes, sir.

CHAIRMAN ROGERS: Did they point out that there was concern about the O-rings and what had happened previously?

MR. LYLES: The reports were mostly a technical writeup on the modeling work itself. Part 2 showed a parametric analysis, and compared that analysis to the flight data. It also gave some analysis looking at what the limit cases could be on O-ring erosion, and it indicated that you could predict the type of erosion that we saw in flight.

CHAIRMAN ROGERS: Were those reports given wide circulation, or were they closely held?

MR. LYLES: I am sorry, I can't answer that. I got a copy. And there are several copies around my office. I assume it got wide circulation.




CHAIRMAN ROGERS: Was it fairly well known as far as you know at Marshall that these studies were being conducted?

MR. LYLES: Yes, sir, I think that analysis had been presented a couple of times.

CHAIRMAN ROGERS: And I assume the same thing was true, obviously, in Thiokol too?

MR. LYLES: Yes, sir.

CHAIRMAN ROGERS: Okay, go ahead. Thank you.

MR. LYLES: Could I have Chart L-5, please?

(Viewgraph L-5.) [Ref. 3/21-78]

[1367] MR. LYLES: The other type of erosion that we have modeled we call blow-by erosion, and in this model we treat the jet as a hot jet created by an assumed initial blow-by area beneath the primary O-ring, which allows gas to impinge on the secondary O-ring. We erode the primary O-ring by simple pipe flow heat transfer relations. We then spread the jet and erode the secondary O-ring by impingement heat transfer as we have shown in the primary O-ring.

The model has several parameters that affect the magnitude of erosion that you would predict. Those parameters are the width of the jet or the area of the blow-hole that you get, which influences the rate at which you pressurize the cavity and which, of course,




drives the time at which you are eroding the O-ring.

DR. WALKER: I have a question. How did you determine the size blow-holes to use?

MR. LYLES: We based our estimates of blow-hole size on the disassembly of the spent motor cases from flight, as when they take the joints apart, if there is evidence of a blow-hole through the putty, they measure the blow-hole and over that range of measured data we assumed that we get nominal blow-holes of that size.

Now, we did in our parametric study look at a wide range of blow-hole sizes, and we did show that when you calculate, when we do a worst case analysis, that we do bound the flight data, that is, we predict erosion greater than the available data we have from the flight motors.

DR. WALKER: I would just like to pursue the question of the analysis of the blow-holes observed for a moment. Who carries out that analysis, and is that analysis made for every single case which is disassembled?

MR. LYLES: Well, it is more of an inspection than an analysis, and yes, sir, I believe the putty is inspected every time the joints are disassembled. We have data on all of the flight joints except, I think, one that was lost at sea.




DR. WALKER: Could the Commission get copies of that, of those records?

MR. LYLES: Yes, sir.

CHAIRMAN ROGERS: Mr. Lyles, were you consulted at all prior to 51L by anybody about O-rings and the study you were doing.

MR. LYLES: Yes, sir, we made a presentation on the modeling work and our calculations on erosion before 51L.

CHAIRMAN ROGERS: When was that?

MR. LYLES: It was last summer, in the August time frame.

CHAIRMAN ROGERS: But not after that. I mean, the summer of 1985?

MR. LYLES: No, sir.

CHAIRMAN: ROGERS: Was there any discussion with you after the accident about the work you were doing?.

MR. LYLES: Well, when it became apparent that it was a possible joint failure, we immediately turned and started exercising the erosion model again, and looking at the parametrics that we have run, we haven't come up with any different conclusions than we had at the time, that is that the parametric analysis that we have run bounded the data from an impinging erosion





There is one thing that I should say, and that is, these are two completely - not completely different problems, but they are different problems, the impingement erosion and blow-by erosion. The model does not predict when you would leak past the primary O-ring. We have to [1368] assume in the model that you could leak past the primary O-ring, and then we can calculate an erosion rate to the O-ring, but the model is not sophisticated enough to tell us when the O-ring would leak. That is a major assumption in the model, and we have to assume the size of the hole, and so on and so forth.

CHAIRMAN ROGERS: I am not quite clear. Did you finish your work on the model before the accident?

MR. LYLES: No, air. The work on the model is ongoing, and we have been trying to improve it all along.

CHAIRMAN ROGERS: Did you reach any conclusion before the accident from your model work?

MR. LYLES: The conclusion that was reached by running the analysis was in. the range of the blow-holes that we had seen, and even for blow-holes smaller than that and for worst case analysis, that we would not show enough erosion to bum completely through the O-ring,




and it would leak by impinging erosion.

We did show that if in fact the primary O-ring leaks, that erosion would, of course, continue until you filled the secondary cavity, and you could get large amounts of erosion of the primary seal, and then all you have got left is the secondary seal. On one flight, we did have erosion of the secondary seal, as you probably know, on the nozzle joint, and we matched that data very well, and again, by worst case analysis we showed that you would not get enough erosion on the secondary nozzle joint to bum through, and the worst case analysis was in fact a worst on worst analysis.

It is assumed that all of the gas that passed by the primary seal did impinge on the secondary seal, and in fact it doesn't. It has to go around a 90-degree bend, and it spreads in three dimensions, and we had to take that into account to match the flight data that we had.

CHAIRMAN ROGERS: Were you familiar with the criticality 1 decision that was made in, I guess, December of 1982, which said that the secondary, O-ring could not be counted on if there was a failure of the primary seal?

MR. LYLES: No, air, I was not.

CHAIRMAN ROGERS: You weren't familiar with






CHAIRMAN ROGERS: And your work came to the opposite conclusion, I assume you felt that maybe the secondary O-ring would hold?

MR. LYLES: For the nozzle joint, and we did come to the conclusion that the secondary O-ring would seal on the nozzle joint because it does not go through the same rotation and problems that the secondary seal has on the field joint.

CHAIRMAN ROGERS: You weren't talking about the field joint then?

MR. LYLES: No, air.


MR. LYLES: Okay, the model was then validated. The erosion model was validated by subscale hot fire data. If I could have Chart L-8, please.

(Viewgraph L-8) [Ref.3/21-79]

MR. LYLES: Chart L-8 shows the calibration of the model or the predicted model results versus the results from the subscale test data, and in this test the blow-hole was simulated by a [1369] rectangular orifice upstream of a Viton O-ring. The field joint was simulated in these tests, and it, the data, showed that analytically we could match the erosion, and we did vary




the width of the upstream orifice to simulate the varying widths of the putty blow-hole, and we varied the cavity volume to calibrate the pressurization model that we have in the analysis, and we are showing a plus or minus 12 percent variation on the measured data, and the picture is there. It just shows - it just represents the type of erosion that we are seeing on the subscale test versus the one case of the flight erosion, and it shows the same type of erosion that we saw in flight.

MR. COVERT: Mr. Lyles, talking about validation of the model if I understand the model, there is an undefined heat transfer coefficient that allows for the cooling of this gas stream, and there is also an undefined heat transfer coefficient at the stagnation point. Did you do other tests to validate the model before you did the comparison with the subscale data? Or does this data represent the best fit of the results from your model based on the selection of those two coefficients?

MR. LYLES: This does represent the best fit of that subscale data based on varying the heat transfer coefficients to the metal, and I think they had to vary the discharge orifice, discharge coefficient some.

DR. COVERT: Thank you.

DR. WALKER: Mr. Lyles, were you asked or did




someone else as a result of your analysis try to determine whether this analysis suggested there was a serious safety problem?

MR. LYLES: I think this analysis showed that we did have a problem in the joint with erosion. I don't think this model showed that we had - that we were getting real close to a flight failure. As I said, the worst on worst case that -by analysis that we put on this model did bound the flight data, and we predicted much larger erosion than we had seen. And when we used that worst on worst case, we still did not show that we would bum through an O-ring unless the 0-ring did not seat and we got blow-by erosion.

And in the case where the O-rings don't seat, if both O-rings don't seat the model becomes moot, really.

DR. WALKER: Did you consider the possibility that the secondary O-ring would become unseated as a result of the rotation?

MR. LYLES: I was really - at the time we were doing this analysis, really not up to speed on the dynamics of the joint rotation on the field joint.

DR. LITTLES: I think we are getting a little beyond Garry Lyles' involvement in this. He was just an analyst who was doing this analytical work but he was




not involved in the safety issue or the criticality issue or those kinds of things. So it is a little bit beyond what he does, I think. He is not really involved in those things.

VICE CHAIRMAN ARMSTRONG: I would like to ask with respect to this chart, because I think it may have confused some. We talk about prediction versus measured, and I understand that measured has to do with the after the fact measurement of erosion on recovered seals from previous flights. Is that correct?

MR. LYLES: That is correct.

[1370] VICE CHAIRMAN ARMSTRONG: But the prediction has to do with what your model would say the erosion would be given certain conditions appropriate to those flights. Is that correct or not?

MR. LYLES: Well, we can't specifically analyze a flight. We don't have any idea. We can't predict what the blow-hole would look like on a specific flight. All we can really do is take the data that is available to us and try to calibrate the model with that data we have, and it is really not good enough to make a specific prediction on a flight, because we don't know yet how to handle the putty blow through.

VICE CHAIRMAN ARMSTRONG: Thank you. Let me ask it a different way then. It is a comparison of the




actual erosion as measured post-flight with the prediction that your model would hypothesize, given the conditions as close as you might be able to guess them appropriate to that particular seal. Is that correct?

MR. LYLES: That is right.

VICE CHAIRMAN ARMSTRONG: And it does nothing - the prediction has nothing to do with predicting into the future?

MR. LYLES: No, sir.

VICE CHAIRMAN ARMSTRONG: And one should not assume from this prediction that you could predict what the erosion on 51L might have been or what the prediction on the next flight would be.

MR. LYLES: No. We did - all we did was a parametric analysis, and with the seeming randomness of the blow-hole and the putty, there is no way that we could predict a future flight erosion.


MR. LYLES: If I could have chart L-10, please.

(Viewgraph L-10.) (Ref.: 321-80]

MR. LYLES: This chart just says that we do have a reasonable analytical tool to predict erosion within the bounds of the prediction. We are still improving the model, and we are using it to evaluate the




O-ring failure mechanisms due to erosion for 51-L failure. Included in that we are improving the analysis relative to O-ring heating as we blow by the O-ring. We will try to correlate that with the mass and mass flow that has been analyzed for the black puff of smoke.

If there are no more questions, that concludes my presentation.

CHAIRMAN ROGERS: Thank you very much.

MR. BACHTEL: My name is Rick Bachtel, and I have been with the Marshall Space Flight Center for the past 21 years, and I have been working in the area of heat transfer thermodynamics and thermal analysis, and since the 51L incident I have been assigned to the SRM failure analysis team in the area of SRM thermal analysis.

What I am going to talk to you today about is some of the work that we have been doing about concerning the flow through the clevis that would be initiated at the leak. We are dealing with a subset of the scenario that says that the leak occurs at liftoff as a result - or the puff of smoke indicates that there is a leak at liftoff, that the smoke is either obscured or disappears, however the leak continues until 58 seconds, when it becomes obvious in the form of the plume that we see.


[1371] 2326


The purpose of the analysis we did was to go off and see if that kind of scenario is feasible and to understand the events that happened between the puff of smoke and 58 seconds, and of course the alternate to the scenario that I am going to discuss is that the leak stops and then starts again in 58 seconds.

The objective of the analysis was twofold. One of them was to determine the thermal response of the joint during that continuous leak, and the other one then was to go back and look at what kind of parameters either flow or dimensional or whatever would be required to sustain a 58-second or 60-second leak consistent with the observations.

(Viewgraph B-1.) [Ref. 3/21-81]

MR. BACHTEL: When we are assessing the thermal analysis there are three events we were looking at. One of them is when the exit temperature or the exit gas temperature, the gas coming, emanating from the joint, when that temperature exceeds 3,000 degrees, it was felt that at 3,000 degrees, that it would be clearly luminous, that it would appear as the plume which we saw.

Another one is, when the outer surface comes to a temperature of 2,150 Fahrenheit, again, that was a visibility factor. It was felt that at that temperature




it should be white hot and may represent what we saw. It is also indicative of a possible failure of the case, of course, and then the third event which we are looking for is any kind of failure of the clevis or the joint that would allow a blowing hole, which is pretty much obvious what we saw at that time.

Some of the parameters which we were assessing to estimate whether or not we could have 60-second leak was the leak flow rate. We tried to determine what leak flow rate or how small leak flow rate we would have to have before we could get out to 60 seconds without a catastrophe, what kind of putty blow-hole size, the blow-holes that Mr. Lyles was talking about, how small they would have to be to be consistent with a 60-second leak, and then what kind of clearance between the tang and the clevis that would limit the flow that would be consistent with a 60-second leak.

Now, if you would put up Chart B-3, please.

(Viewgraph B-3.) [Ref. 3/21-82]

MR. BACHTEL: We had two different models we were using. One of them was a two-dimensional-model, which is pretty much like what you see on the table here as far as this cutaway. It is one circumferential slice of the clevis and tang. It is between the pins. There is no pin in that particular model, and that model is




basically a parametric tool. It has fairly good detail thermally of what is going on in here. It is not as complicated as a three-dimension model such that we could turn around the model as rapidly and run some parametrics. It is pretty much appropriate for doing analysis up until when it is we get a failure of the joint.

Now, once we get a failure of the joint, we have to go to a three-dimensional model if we want to assess how the failure progresses in the other dimension. Both of the models, of course, include the heat transfer and thermodynamics that go on inside the joint, the heat transfer outside the joint due to the aerodynamics of flight, which tends to keep the joint cool, the melting of the steel, the recession of the steel, the opening of the gaps, and thus the increase of the flow rate, the ablation and the increase in size of the putty blow-hole with time, the recession of the [1372] O-rings which then allows more flow. And when we do this analysis we do it in two different ways.

One of them is, we run a fairly simplified flow analysis which allows us to do parametrics, and then another method is, we depend upon a separate flow analysis which is a rigorous approach to the flow, a three-dimensional approach which takes into account the




spreading as you come through the clevis, the interaction of the pins, et cetera.

We take the output from that analysis and feed it into the thermal analysis we are doing, and then finally we are trying to couple those two together, and that is in work currently, and we should be able to report on that in another couple of weeks. If you would go to Chart B-4.

(Viewgraph B-4.) [Ref'.3/21-83]

MR. BACHTEL: This is a detail of the two-dimensional model. The grid work you see on the chart represents the elements that the clevis and the tang are broken into. Within those elements, there are subelements or nodes which allow those elements to become smaller as the metal shrinks, rather, I am sorry, as the metal melts. As the metal melts, then the gap between the clevis and the tang becomes greater and the flow rate becomes greater, and you fairly rapidly cascade the flow rate and the melting process until you go catastrophic, and of course the purpose of the analysis was to find out how small these various gaps either in the NBR, either in the insulation and in the putty coming into the O-ring area here, how small the O-ring clevis clearance had to be or the other clearances to get a 60-second leak.




Now, if we go to Chart B-5.

(Viewgraph B-5.) (Ref. 3/21-84]

MR. BACHTEL: These are the results of some of our analysis. The box at the top is where we ran some constant flow parametrics to find out what kind of flow rate it would take to get the 60-second leak, and I have got the chart up now. As you can see, it is a fairly small flow rate. It is on the order of .004 to .005 pounds per second. Now, this would be continuous flow rate during that 60 seconds to keep from having either a case failure or an exit gas greater than 3,000 or what have you at 60 seconds.

Now, that .004 pounds per second relates to about 700 cubic inches per second if it was a volumetric flow expanded to the atmospheric pressure. The next thing we did then was look to see how small the blow-holes would have to be to keep the flow rate that low and thus last 60 seconds. The 0-2, which is the smallest one I have on the chart, of course, shows a failure at about 49 seconds. The 0-1 gets us out to 70 seconds. As Mr. Lyles reported, the blow-hole sizes were normally on the order of half an inch, so the minimum or the maximum blow-hole size that I can tolerate to be able to have the joint last for 60 seconds with a continuous leak is almost an order of




magnitude below what we normally see.

The third box shows the kind of analysis we did for the tang to clevis clearance. Of course, if this closes up enough, it limits the flow, which then allows us to get up to 60 seconds again. The kind of clearance we had to squeeze this down to in order to keep the joint alive for 60 seconds was on the order of .002 inches.

Mr. Ryan showed you earlier this morning where we normally open up to .025 inches, which is almost ten times greater than that. So in almost all cases our analysis shows that if the puff [1373] of smoke indicates a leak that was to continue, that we should have destroyed the joint within about 10 to 20 seconds.

So, if we go to the next chart, B-6.

(Viewgraph B-6.) [Ref. 3/21-85]

MR. BACHTEL: I will carry this just a little bit further. We have also done some three-dimensional analysis. One of the things that the three-dimensional analysis does is, it picks up conduction circumferentially around which tends to spread the heat out and makes the joint last a little bit longer, and there is a demonstration of that here. When we ran the two-dimensional model on a specific case, we took the joint out at about 24 seconds. The same case on a




three-dimensional model got us out to 35 seconds, so that gives us a little bit of time.

The box at the bottom shows our best estimate of the flow case. Now, this is a rigorous flow case which is the beat estimate of what the blow-hole size would have been, what the O-ring erosion would have been based on some of Garry's work, how the flow would have spread once it comes through the clevis. As the flow goes through the hole in the O-ring, it tends to spread out, which then dissipates, which could possibly let the joint last longer, and again, we are showing on the order of 16 to 20 seconds in this case when the joint should have failed.

So, if you go to Chart B-7.

(Viewgraph B-7.) [Ref. 3/21-86]

MR. BACHTEL: In summary, then, our analysis which we have done to date tends to indicate that the scenario that the leak continued from the puff of smoke all the way out to 50 seconds is probably not the proper scenario. There is still some work that has to be done in this area. For example, one of the things that may limit the flow and which also could give us the other scenario, and that is a stop flow, would be the deposition of aluminum oxide in these gaps, so we are going to start looking at that or we have started




looking at that. We are not prepared to report on it yet. That kind of analysis, as I said, would support both this analysis or this scenario of a continuous leak, an intermittent leak. It may start and stop on the way up, or even a stop at, say, six, seconds, and then reissue at 58 seconds. The analysis does show, however, that if we were to reestablish the leak at about 50 to 60 seconds within ten seconds we would expect to go catastrophic, and that concludes what I had to say

CHAIRMAN ROGERS: Thank you. And that test on the aluminum oxide will be completed in a couple of weeks?

MR. BACHTEL: We are starting to do some analysis. It is a very difficult analysis to do because it involves empirical data. It involves a lot of statistics. It is not an exact analysis. It will only show a probability, and it is going to be based primarily on some test data which we are now acquiring, and we expect to have within a couple of weeks some type of a report on that analysis. Yes, sir.

CHAIRMAN ROGERS: Thank you very much.


[Please note that some of the titles to the references listed below do not appear in the original text. Titles are included to identify and clarify the linked references- Chris Gamble, html editor]
1374] [Ref. 3/21-75] Aft Segment/Aft Center Segment Field Joint Configuration. [Ref. 3/21-76 1 of 2] O-Ring Erosion Analysis.

[1375] [Ref. 3/21-76 2 of 2] Erosion Scenario.

[1376] [Ref. 3/21-77] Analytical Modeling of Pressurization and Erosion of O-Rings.

[1377] [Ref. 3/21-78] Analytical Modeling of Pressurization and Erosion of O-Rings (continued).

[1378] [Ref. 3/21-79] Simulation of Hot Subscale Data.

[1379] [Ref. 3/21-80] Application to STS 51-L.

[1380] [Ref. 3/21-81 1 of 2] Thermal Analysis of SRM Field Joint Leak.

[1381] [Ref. 3/21-81 2 of 2] Thermal Analysis of SRM Field Joint Leak.

[1382] [Ref. 3/21-82 1 of 2] Two- and Three-Dimensional Model.

[1383] [Ref. 3/21-82 2 of 2] Method.

[1384] [Ref. 3/21-83] Two-Dimensional Thermal Model.

[1385] [Ref. 3/21-84] Summary of 2-Dimensional Thermal Analysis.

[1386] [Ref. 3/21-85] Summary of 3-Dimensional Thermal Analysis.

[1387] [Ref. 3/21-86] Summary.



[Ref. 3/21-87 1 of 2] Testing to Evaluate Scenarios.

DR. LITTLES: I will now conclude, Mr. Chairman, with a summary of the testing which we have under way to try to narrow down some of the remaining




scenarios. If I could go to Chart W-25, please.

(Viewgraph W-25.) [Ref. 3/21-87]


[1389] DR. LITTLES: The testing that we have underway relates to these remaining items on this scenario, Items 3, 4A, B, and D, and Item 6. In addition to that, some of the subscale motor testing that we are doing relates to the analysis that Mr. Bachtel just discussed relative to the continuous flow or stop-leak flow analysis. I will summarize some of those test results for you. There is other work that is still going on there. And then I will get into a schedule chart and discuss the specific tests related to these scenarios and when we will finish those or when we hope to.

Could I have Chart W-27, please?

(Viewgraph W-27.) [Ref. 3/21-88]

DR. LITTLES: This chart depicts one of the two subscale motors that we are using, and this particular one has a full-scale cross-section of the tang and clevis. It is roughly 11 inches in diameter, but you can see if you look at the joint area that it is identical to the design of the case standard size joint design.

It has capability of-or we have the capability of inducing defects with this motor of the O-ring or defects on the ceiling surfaces, scratches,




and we also have the capability to induce defects in the putty, and we can run the motor over a range of temperatures to evaluate the temperature effect on any kind of defect or parameter that we put in there, and one of the key things we are going to be doing with this motor and the other one is looking at putty performance at various, temperatures. The motor simulates the pressure in the actual SRM. We can get up to 1,000 psi.

It operates from 10 to 75 seconds, depending upon how we configure the propellant in there. If we could go to Chart W-28, please.

(Viewgraph W-28.) [Ref. 3/21-89]

DR. LITTLES: I just selected a few of the tests here, an initial test to show you what kind of variation we are getting, and these tests are where we have had simulated defects in the motor to see what kind of results we get relative to the continuous flow or stop-leak type flow evaluation.

The first one there is one where we had 6-inch cutouts in the O-ring, very large defects there. We also put a half-inch wide fault in the putty, and we had a .040 inch extrusion gap and burned for 20 seconds. We had a complete burnthrough on that one of the tang near the primary O-ring at 6 seconds, and so it failed very quickly. The second test was




identical to the first, except here we had only eighth-inch flaws in the O-rings, had an eighth-inch section cut out, and in parentheses there you see that we think we had about one-sixteenth of an inch after we had it actually put together and squeezed, but in this case it burned for 20 seconds without visible leakage, and when they disassembled it, they found soot between the 0-rings, and the gap in the primary O-ring which filled with a combustion residue, so the difference between those tests was the magnitude of the defect, but there was a considerable difference in the result.

VICE CHAIRMAN ARMSTRONG: How do you account for that? Would you guess the O-ring grew back together?

DR. LITTLES: Well, if somebody had told me before we had the test that we could have gotten this result, I guess I frankly wouldn't have believed it. What they saw was that in the gap of the O-ring, they said it was filled with a combustion residue. To me it seems unlikely that you could get that with a hot gas flow through there. But indeed it did occur. It seems an unlikely result, but we found some other similar results. I will mention another one in a second that is [1390] not on this chart. On Test 3, again, it was the same as 1 except both O-rings were ground down from their normal diameter




of .280 down to .200 for about 3/1 e of an inch, and in this one, the burnthrough, it burned through the tang in about 12 seconds.

Now, since then, as a matter of fact, a couple of days ago, and I don't have the detailed results yet, but we ran a test similar to three, and the only difference was that we put some instrumentation in this one to get more thermal data. Other than that, as I understand it, the conditions were identical, but this one ran for the full 20 seconds.

It had aluminum oxide deposited in the area of the initial-of the cutback O-ring, of the defect, and it stopped the flow there and moved it around, so there is a considerable amount of variation that you can get with these motors, and I think they are not very predictable relative to what a defect will do relative to progressing the damage, which makes the analysis that Mr. Bachtel talked to you about extremely difficult, if not impossible.

VICE CHAIRMAN ARMSTRONG: Let me ask you to look at this again to make sure I understand it properly. I am comparing test 2 and test 3. And these were essentially identical tests. In test 2 you just completely removed a section of the O-ring that was 1/8 inch across.




DR. LITTLES: That is correct.

VICE CHAIRMAN ARMSTRONG: And in No. 3, you had a complete O-ring but you just made it a little narrower, for 3/16ths of an inch?

DR. LITTLES: That is right.

VICE CHAIRMAN ARMSTRONG: So one would think on the surface that test 2 would be much more severe than test 3.

DR. LITTLES: I would think so.

VICE CHAIRMAN ARMSTRONG: And yet in test 2, the more severe test, you didn't get any damage. In test 3 you had a complete collapse.

DR. LITTLES: That is correct.

VICE CHAIRMAN ARMSTRONG: So one would draw the conclusion that it is difficult to be able to predict the characteristics of these burnthroughs past an O-ring.

DR. LITTLES: That is correct. That is the conclusion I draw.

CHAIRMAN ROGERS: I guess one of the things that occurs to me in these tests, and maybe you can explain it, weren't tests like these run before the accident? I mean, knowing that the O-rings were such a-or seemed to be such a serious problem, weren't tests of this kind run before the accident?




DR. LITTLES: Well, these are failure tests that are being conducted here to try to duplicate or to try to understand the condition that we might have had if we had a puff of smoke and then a continuous leak.

CHAIRMAN ROGERS: Well, I guess my question really is broader than that. Were somewhat similar tests run to determine what would happen to O-rings under certain conditions?

DR. LITTLES: No, sir, no tests were run that I am aware of like this. We always knew, I think everyone would agree, that one cannot tolerate the kind of conditions we are setting up here with an O-ring. I don't think there is anyone who would have predicted that you could sustain that kind of defect and have a successful burn of the.

[1391] CHAIRMAN ROGERS: That is really not quite my question. My question is, in the beginning, say, of 1985, when concern was expressed about the O-rings, weren't tests run then either by Thiokol or by Marshall to decide some of these concerns, to answer some of the questions that were raised at the time?

DR. LITTLES: The analysis and the supporting tests that Mr. Lyles discussed were ongoing at that time. As a matter of fact, they had been started a number of months earlier. Those analyses and the




supporting tests that confirmed that analysis were used to evaluate the degree of erosion that one might expect in a worst-worst case on the primary, and so those kinds of tests and analysis were conducted relative to the phenomenon that we had observed, which was impingement erosion of O-rings.

CHAIRMAN ROGERS: Do you or does Thiokol to your knowledge have any history that has been prepared about O-rings and their failure? The reason I ask that is because in 1982 it was known that the O-rings were a problem, and the criticality was changed from criticality 1R to criticality 1.

I would have thought there would have been a lot of tests to determine reliability of the O-rings during that three-year period, and I would have thought that there would be a history of the O-rings and how much you could rely on them, and so forth. To your knowledge, are there any such reports or any history of the O-rings that has been prepared?

DR. LITTLES: Well, there was a history of the O-rings relative to the erosion that had been seen in flight. Those results were available on every occasion when we had erosion in flight, whether it was on a field joint or on a nozzle joint. It was looked at, that specific case was looked at. Those results were




discussed, and the flight readiness reviews for the subsequent flights, and were reported all the way up through channels, and that history was maintained on a continuing basis.

So, yes, sir, that history was available.

CHAIRMAN ROGERS: Well, the Commission would want to get all of that information before we make our report. As I remember some of the original testimony, though, by some of the people who made the decision to launch was that they were not familiar with the problem with the O-rings at all.

DR. LITTLES: Well, I don't recall that testimony, but I do know based on personal experience that those anomalies that have been experienced in flight have been reviewed in flight readiness reviews.

CHAIRMAN ROGERS: So it is your opinion that that concern was generally known both at Marshall and Thiokol among people at the top of the agency.?

DR. LITTLES: Yes, sir. I can't imagine that it was not known because it has been, I know, covered in flight readiness reviews. We always discuss any anomaly that we have. We evaluate it relative to its potential consequences on the subsequent flight, and we discussed it in Right readiness reviews. And so I am not familiar with the testimony you refer to, but I personally




couldn't imagine anyone not being aware of a flight anomaly.

CHAIRMAN ROGERS: Well, of course, the astronauts say they were not aware of it. You have seen that testimony, I guess.

[1392] DR. LITTLES: Yes, sir. I have seen that. I am not personally familiar with their participation in flight readiness reviews. If someone had asked me, I would have said I thought that they were probably represented, but I don't know that, but I-do know for a fact that they have been discussed in the level 2 flight readiness reviews where I would have expected at least a crew representative to have been present. So I know they have been discussed.

CHAIRMAN ROGERS: So I assume then when you heard statements by the astronauts that they did not know about the problem in the flights where they were involved, that surprised you?

DR. LITTLES: Well, like I say, I am not familiar with their direct participation in flight readiness reviews, and so I don't know what their normal communication channels are with those who do participate.

CHAIRMAN ROGERS: Well, let me ask the question more directly. Were you surprised or are you




surprised now when you learn that a commander of one of the space shuttles was not told after the fact what had happened in connection with the O-rings on his particular flight? Did that surprise you, or does it surprise you today?

DR. LITTLES: It would surprise me. As a matter of fact, I couldn't believe that that incident had not been covered in the subsequent flight readiness review. Now, again, and I would be surprised if anyone who had been involved in that flight didn't somehow find out about it, but again, I don't know where the communication channels are. I know it was covered in the flight readiness review, but what that commander's opportunity for getting that information might be, I am not familiar with.

CHAIRMAN ROGERS: Okay. We will come to that later on in the investigation. Thank you.

MR. LYLES: Mr. Chairman, can I make a statement for the record? I would like to clarify something that I said before.


MR. LYLES: I told you that I was not aware of the criticality 1 issue. That is a true statement. However, I think all of us knew that if the secondary O-ring leaked, that it was a bad situation.





MR. LYLES: If the secondary O-ring leaked, that was a bad situation.

CHAIRMAN ROGERS: The criticality 1 that I am talking about said that if the primary 0-ring fails, you can't rely on the secondary O-ring.

MR. LYLES: I was not aware of that criticality. I was aware that the effort that was going on, that Thiokol was going through a redesign effort on the field joint, and we were looking at different O-rings and things like that, but as far as the specifics of all that, we were working the erosion problem.

CHAIRMAN ROGERS: I see. Thank you.

DR. LITTLES: Could we have Chart W-30, please?

(Viewgraph W-30.) [Ref.3/21-90]

DR. LITTLES: This chart depicts the second subscale motor that we are using. This is a five-inch motor. It is a little smaller than the other one. It can burn between three and 24 seconds. We can get up to 800 psi pressure in this one, and run for 24 seconds. Again, we can vary the temperature over a wide range, and again, we can simulate degraded O-rings and sealing surfaces and putty. Would you focus in on the bottom


[1393] 2345


right, please? You will see that this one does not simulate the actual joint, so it is not quite as good in that respect as the other motor, but you can get a lot of good, qualitative data with this motor. Could we see Chart W-31, please?

(Viewgraph-31.) [Ref.3121-91]

DR. LITTLES: I will summarize again just a few of the tests that have been run with this motor, and you will see the same kind of trends as with the other motor, a lot of variability. Test 3 had an eighth-inch section of O-ring missing, no putty in this one. It had a 0.03 gap and burned for 3 seconds. And it had smoke at ignition with flame at 1.2 seconds, and the O-ring was mostly consumed, and heavy metal damage on that one.

The next test, test 4, was the same as 3, except there was a 0.004 gap rather than a 0.03, and it burned for 24 seconds. This one we had continuous smoke but no fire, no O-ring damage at the cut, O-ring erosion and heat effect away from the cut, but no metal damage. So the primary variable there was just the gap, and there was a considerable difference in performance.

Could I see W-32, please?

(Viewgraph W-32.) [Ref.3/21-92]




DR. LITTLES: Test 6, which is the first one on this sheet, we had a half-inch vent in the putty at this time with a scratch on the sealing surface. The O-ring wasn't damaged, but just a scratch on the sealing surface. It burned for 24 seconds with a 0,019 gap at 70 degrees. It had a small plume immediately on this one, at the defect, and it shut off and then reappeared 90 degrees away from that original defect, and there was considerable damage on this one. It had molten metal and very heavy damage.

Test 7, again, was the same as 6 except here we had the putty intact. There was no vent or defect in the putty, and we cooled this one down to 80 degrees, and it had two scratches on the surface. Now, this one, nothing happened at all for ton seconds, and then we got some flow for five seconds, and then it stopped. The significant item out of this test, at least to me, is that the putty appeared to hold that pressure off of the 0-ring for ten seconds, and that is very significant, of course, because you need to get the pressure on the O-ring during the ignition transient, particularly with the joint rotation situation that we talked about.

DR. COVERT: Dr. Littles, I think that these are very interesting results, and I wonder if you in your test matrix intend to repeat these several times,




in other words, take the configuration exactly the same, for example, as Test 7, and then just refire it, and then build a new fixture, and refire it.

DR. LITTLES: Yes, sir, we are going to do that. You see so much variability with the same general type test. I think you have to run several of the same type to see what really can occur, and we are going to do that. That will probably be a little later, though, because one of the key things we are going to be using these motors for in the next few days is to support failure evaluations for the remaining scenarios. We are going to concentrate on that, and then we will do some more tests relative to the leak stop leak type thing.

DR. COVERT: I would like to pursue this just a little bit further. I wasn't particularly interested in, although I think it is important, the leak stop leak, but in all of these tests the level of variability is such that it suggests to me that if you are going to base any serious conclusions on it, you might want to have some repeat points.

[1394] DR. LITTLES: I agree. I certainly do. And even more specifically, I guess, even more importantly for those there we are verifying scenarios, we will have to run a number of tests on those in those areas where we find results that are important to us and might




support conclusions one way or the other. I agree with you. We have to run a number of tests. That will be very important.

DR. WALKER: Can we just return to a point that I think you made? At 30 degrees evidently the putty becomes quite stiff and can withstand considerable pressure. Is that the point you were making?

DR. LITTLES: Well, that is a point I was making. I was going to continue with that, though. We are running some additional tests at higher temperatures in these motors. I had hoped to have a higher temperature test or two by this time, but I haven't gotten it yet. But one of the things that we have found from the putty tests we are doing, we are doing tests with putty over a range of humidities and temperatures and defects and all these kinds of things.

And we haven't completed that by a substantial amount, but one of the things we are learning out of that is that the putty is temperature-sensitive. It will hold pressure at lower temperatures for longer periods of time, but even at ambient conditions, 70 degrees, the putty still will hold pressure off for long periods of time relative to the half-second of the ignition transient. So, yes, it is temperature sensitive, but even at ambient temperature it will hold




pressure for long periods of time relative to the time of interest that we are looking at.

Now, those tests, there is still some work that has to be done in that, because those initial tests were conducted in a fixture which did not incorporate the dynamics of the joint. In other words, the joint was held fixed, and actually you get some motion of the insulation around the putty, and we have a test fixture which has just started into test, and we will be getting a lot of results out of that, we hope, in the next week to 10 days which incorporate that feature.

DR. WALKER: Now, it was my understanding from discussions at Morton Thiokol that the way in which the joint was to operate was for the putty to transmit the pressure pulse to the O-ring in order to seat it, and these results would suggest then that that theory of the operation of the joint may have been faulty.

DR. LITTLES: That is absolutely correct. That is correct. And that is the hypothesis of scenario 6, that that concept of the joint operation which people had believed was correct is not in fact correct.

GENERAL KUTYNA: Dr. Littles, this one interests me. You are interested in the putty, but I am interested in the leak-stop-leak. You had a leak and




then it sealed itself, and this is one of the few that it has done that on. How did it seal itself.? What was the mechanism? What was the structure of the material in there?

DR. LITTLES: I am not exactly sure how that sealed. I asked the other day somebody to go check on that. I am not sure whether it sealed because of some internal mechanism or whether the putty itself sealed. And we have run some tests, some lab tests in configurations that don't simulate this joint, and we are going to have to do some more, where we have seen the putty have a blow-hole breakthrough, and then seal itself. So, it could be that the putty sealed itself here, or it could be something internally.

GENERAL KUTYNA: This is, of course, very similar to the actual shuttle launch, in that we had a leak and possibly it sealed itself. I had heard at Marshall that in some of these cases [1395] the material that sealed the leak was kind of a glassy substance rather than something very tough, and if it were glassy and brittle, it might be something that could break off later in flight-break open later in flight after you had loads imposed on it.

DR. LITTLES: That is correct.

GENERAL KUTYNA: Do you have any of that experience to report?




DR. LITTLES: No, we don't have anything that simulates that.

GENERAL KUTYNA: Are you aware of that glassy substance, though?

DR. LITTLES: Yes, I would not be at all surprise that that could happen either with the loads that we were putting on it at that point in flight or, you know, we were, of course, beginning to increase in motor pressure. At about that time the pressure was going back up, which imposes a little additional opening on that joint, and so there were things going on which, with a damaged seal, and possibly having some deposition like you referred to could have caused it to open up.


DR. LITTLES: Chart W-33, please.

(Viewgraph W-33.) [Ref.3/21-93]

DR. LITTLES: This is just a summary of observations, and I think we have touched on most of them as I have gone through. There is more work, as I said, to be done in this area, but my conclusion, I guess, at this point in time, it is obvious, is that it is a highly variable phenomenon which I don't believe is amenable, frankly, to a very detailed analysis. There are just too many variables to try to do it, and I think it is possible based upon what we have seen that you




could have had a leak that stopped and started some time later. I think at this point in time it is highly unlikely based upon the analysis that Rick Bachtel reported that you had a continuous leak at one location for that long. I think that is highly unlikely.

Chart W-34, please.

(Viewgraph W-34.) [Ref-.3/21-94]

DR. LITTLES: Okay, we have a number of tests that are going on relative to the other elements of the joint scenario. The first item on the chart there is testing that we are doing relative to the assembly, potential assembly damage that we have discussed before. We have a test fixture at Marshall which is a partial joint tang/clevis. It is a segment about two and a half feet wide, and we have it in a fixture, and we are using that to simulate various off-tolerance situations in the mating to look at potential damage of the tang and clevis or the O-ring, and depending upon the results of that, and we may go further and do some full-scale short attack tests on that.

Item 2 deals with scenario 4-A.

MR. ACHESON: May I ask a question about 1? 1 see that 61G is to be destacked in April. DR. LITTLES: Oh, yes. I failed to point that out.




MR. ACHESON: Is there any reason to believe that 61-G experienced particular assembly problems, out of roundness or difficulty in stacking?

DR. LITTLES: 61G, as you might recall from our last discussion, 51L had a negative dimension which puts the tang back over the O-ring section of the clevis of .393, I believe it was. 61G is not a direct analog of that. It had a negative dimension, I believe, of .274, but there are some [1396] reports that there was some difficulty in making that mate, and we think it would be very informative to take it apart and see what the condition is, but it is not a direct analog of 51L.

MR. ACHESON: Thank you.

DR. LITTLES: So then, back to Item 2, the seal test relative to a defective seal, and this relates primarily to the closeout photo that we have discussed. We are still doing some work on that to try to determine whether what we see there is indeed a defect. We have a test fixture that we put together to do that with, and to take some photographs and see if we can simulate that, depending upon the results of that, we might go to our dynamic test fixture which I will discuss in a minute, or maybe even to some hotfired tests to close that out.

Item 3 is testing related to ice in the




joint. We have a visual fixture that has been built to look at what might happen when you get ice up near the second O-ring or ice and grease above it in a column to potentially unseat that O-ring. We are doing tests on that. Depending upon what comes out of that, we will probably do some tests in some of the other fixtures possibly in the joint motor, the motor that has the joint simulated very well, test 104, and probably even in the dynamic test fixture.

Could we see Chart W-35, please?

(Viewgraph W-35.) [Ref. 3/21-95]

DR. LITTLES: The first item on this chart is the relative to scenario 4D. This is the scenario which deals with the situation that was discussed on the Monday night before the flight, where you have cold situation, you have O-rings which have degraded resiliency, and you have the joint rotation. We have this test fixture in tests now. We have gotten the first series of tests back. We ran the first series of tests with the minimum squeeze that we had on the 51L conditions, and we ran it with a gap opening of 0.020. In other words, we had it set at the minimum squeeze, and then what we do with this test fixture is that we open that gap at a rate which simulates the gap opening on the actual motor tang/clevis during flight,




and of course it is simultaneously-you have the pressure that you see during the ignition transient imposed on it. Those first series of tests, there were seven run at a range of temperatures. There were two run at 75 degrees, one at 40, two at 25 degrees, one at ten degrees, and one at minus ten, and in all cases the seal performed properly except at minus ten degrees, and in that case it did not, but down through ten degrees it did perform properly.

Now, we are now going into another series of tests with that fixture where we are going to impose the maximum squeeze condition. In other words, in looking at the way this joint is assembled and the tolerances, it looks as though you can have a situation where you have, rather than-you can have a situation where you have almost metal to metal contact, and to us that seems to be a worse situation relative to the response of the joint than the minimum squeeze, and so we are going into that test series now. We will set that up with the maximum squeeze and then run through the full joint rotation, and we expect to have those tests probably some time next week, so that is a very interesting set of test to us.

DR. KEEL: Dr. Littles, could I just ask a question there for clarification? You say you are




simulating the gap opening due to rotation, but you aren't simulating, as I understand it, the gap opening due to this going back from out of round to round, if you will, based upon this load analysis you have done.

[1397] DR. LITTLES: That is correct. It doesn't do that, because it only moves in one direction, and doesn't have that in it, but again, I think the worst condition is going to be where you have the maximum squeeze, and in that case it is going to be going in one direction anyway. The effect of the rounding situation where you have the maximum squeeze is much less than it is where you have the minimum squeeze. If you have the minimum squeeze, what you actually get during the rounding process is that when that takes place in the neighborhood of 50 to 150 psi, you actually get more squeeze put on that O-ring, and it continues to open up, but you are right. That feature is not there.

The next item is Item 5, which are the putty tests. I have mentioned these already. We have that second fixture ready now. It is in tests. We are conducting a range of parameters there, temperature and relative humidity. And again, we expect to complete that by the end of the month and have the data evaluated by the end of the first week in April.

Item 6 is also




DR. WALKER: One question on Item 5. Are there putty tests both at Marshall and at Thiokol? Or just one place or the other?

DR. LITTLES: Well, the primary tests are being conducted at Thiokol. That is where we have this more high fidelity feature. We had conducted some tests at Marshall on very simplified lab rigs. We have done some work there, but the predominant work is being done at Thiokol. Item 6 is another very important series of tests for us. We have a test fixture for this which we can use to simulate the dynamic situation that I discussed under Item 4D.

But it can also be used to simulate the situation that you get where the putty holds the pressure off of the O-ring. The gap or the joint opens, the gap opens, and then you get a pressure through the putty, and you hit the O-rings while you already have a gap underneath them, and that is the situation that we will be simulating with this test fixture, and that test fixture is now available, and we should be testing it next week, so in the next week to ten days we hope to get a significant amount of data relative to these prime scenarios, and that concludes my presentation, Mr. Chairman.





think these reports of test results are a substantial contribution to the Commission's investigation. And I would like to ask Dr. Littles a couple of questions by way of summary. I think the time line we have seen is a substantial improvement in accuracy and understandability, and my understanding of the information you have presented today is that there have in the past been a lot of reports, both in the popular press and the trade press and so on of potential external tank leaks, and being contributory to the event, and I understand that your analysis to date indicates that in fact although you have still some people that are reviewing the conclusions that you have drawn, that at this point in time you have no evidence to indicate that leakage to the external tank was contributory first or even existent. Would that be a fair conclusion?

DR. LITTLES: That is a good summary of that, yes.

VICE CHAIRMAN ARMSTRONG: Secondly, there have been a variety of speculations regarding potential loads, both unusual loads either at liftoff or at the combination of Max Q and wind sheer events, and a detailed look at the load profiles up to this time indicates that those are in fact in the normal range.


[1398] 2359


However, if there were earlier alternate failures, it is possible that loads could have contributed to the subsequent breakup, that normal loads would have contributed to the subsequent breakup?

DR. LITTLES: Yes, I think that is a possibility.

VICE CHAIRMAN ARMSTRONG: And thirdly, you find no evidence to indicate that the leak check port had any applicability to the failure?

DR. LITTLES: No, we don't think it did. I think the evidence indicates that it did not emanate at that point.

VICE CHAIRMAN ARMSTRONG: And now I am going out on a limb a little bit with the last presentations. We had earlier, some had earlier thought that the prediction of O-ring erosion was perhaps impossible or an unpredictable kind of event, and I think Mr. Lyles' presentation indicates that in fact there is a certain predictability to O-ring erosion, given the proper initial conditions and characteristics of the flow. But at the same time the tests that you have done in your latter presentations indicate that in fact if there are abnormalities present of one sort or another, then the way in which a joint might fail are quite variable and even unpredictable.




DR. LITTLES: I think that is true. I think that the work that was done, the analysis that was done to develop the O-ring erosion model and the test data did a good job of bounding that problem, but when you get the situation, the situations that we are simulating in these tests where you have defects that allow blow-by and don't give the joint or the O-ring an opportunity to seal, I think it is a highly variable situation in that case, and very difficult to predict.

VICE CHAIRMAN ARMSTRONG: Thank you. That is all I have, Mr. Chairman.

CHAIRMAN ROGERS: Just one other. I assume that the conclusion you reached last time is still valid, and that is the joint still seems to be the area of most suspicion, the one you are most concerned about. Is that correct?

DR. LITTLES: That is correct. I believe we have eliminated all of the other possibilities except these items we have just discussed here relative to the tests, and those all deal with the joint.

CHAIRMAN ROGERS: Thank you very much. I think there has been real progress since the last session. Thank you. There may be some other questions. If not, thank you very much for a very good presentation.




(Whereupon, at 3:25 p.m., the hearing was adjourned.)


[Please note that some of the titles to the references listed below do not appear in the original text. Titles are included to identify and clarify the linked references- Chris Gamble, html editor]
1399] [Ref. 3/21-87 2 of 2] SRM Hot Gas Leak Failure Scenarios.

[1400] [Ref. 3/21-88] Full Scale Joint Cross Section Motor. [Ref. 3/21-89] Status.

[1401] [Ref. 3/21-90] 5 inch Subscale Motor. [Ref. 3/21-91] Summary of Selected Tests.

[1402] [Ref. 3/21-92] Summary of Selected Tests. (continued).

[1403] [Ref. 3/21-93] Observations.

[1404] [Ref. 3/21-94] 51-L Failure Analysis - Test Activity Summary. [Ref. 3/21-95] 51-L Failure Analysis - Test Activity Summary (continued).

March 21, 1986 Session (part 1) | Volume 5 index | April 3, 1986 Session