1. Mick West

    Mick West Administrator Staff Member

    Source: https://www.youtube.com/watch?v=u1CZmtR8gno&feature=youtu.be&t=21m21s

    (Skip to 21:21 if not auto-started there)

    This first post is the transcript, with minor asides omitted and some slides added for context. Discussion to follow.

    Mick: The NIST study, the ANSYS simulation came up with a set of damage that the fires did to the building, and you can see here these little black dots are what they refer to as being "Full Connection Damage" and these beams here, the red beams, are beams that have suffered significant damage themselves. So this is Column 79 over here. ... There's damage that is shown throughout the building.

    (1) The JB Podcast Episode 22-Mick West & Tony Szamboti Debate - YouTube 2018-01-13 15-52-58.
    And when they took this pattern of damage from this simulation here, the ANSYS simulation, and then they applied it to this full height model, it caused a collapse.
    (1) The JB Podcast Episode 22-Mick West & Tony Szamboti Debate - YouTube 2018-01-13 15-53-52.

    Mick: But it didn't actually use that particular connection that you just talked about [A2001 to C79] in this simulation.

    Tony: Oh sure, sure it did, sure it did. It had to.

    Mick: No, it didn't

    Tony: It had to Mick.

    Mick: But it didn't though.

    Tony: Well what did they use? Tell me that they used then Mick, because they say in Chapter 12 they use what they got from the ANSYS model.

    Mick: They did, they used that damage pattern. But that damage pattern didn't have that particular ... column connection failure, A2001 to column 79, That wasn't the first thing that actually failed. There was actually a number of simultaneous collapses.

    (1) The JB Podcast Episode 22-Mick West & Tony Szamboti Debate - YouTube 2018-01-13 15-54-59.

    Mick: Which you can see here, there's collapses over here and over here [indicates the two circled areas above]. This say the "ANSYS-based damage application is resulting in floor structure failures around Column 79 to 81", which as you know is all the interior East side of the building, so they didn't actually..."

    Tony: Well where is that, ... they haven't told us that in the report. [Mick note: this IS from the report]

    Mick: It's quite apparent though, when the girder falls,

    Tony: you can't just show ... we want an analysis ... this is not an analysis

    Mick: If you look at this particular image here, which people are familiar with, the girder in question, A2001, I think is somewhere over here, it didn't actually fall down through the other floors.
    (1) The JB Podcast Episode 22-Mick West & Tony Szamboti Debate - YouTube 2018-01-13 16-03-01.

    Tony: You need to justify, when you give a presentation like this, and you're making a claim, you have to justify it, you can't just say this is what we found, ...

    Mick: You were saying you found that some particular thing didn't happen earlier, I think if I did the same ... But this isn't really something that's actually up for debate really. We know that the ANSYS damage contained other connection failures.

    Tony: No, we don't know! They haven't released their data.

    Mick: well, look at this slide here:
    (1) The JB Podcast Episode 22-Mick West & Tony Szamboti Debate - YouTube 2018-01-13 15-54-59.

    Mick: There's two areas in which collapse is happening simultaneously. Now this collapse over here toward the center of the building is completely separate from this collapse over here.

    Tony: Mick, if they left off the stiffeners which prevented the failure I showed earlier...

    Mick: No, I agree with that Tony...

    Tony: .. who's to say they might not have left off something here?

    Mick: Yes, they may well have done, but my point here is that you are focussing just on this one connection. And your analysis of that one connection is pretty good. Because it's quite comprehensive, and you did identify a number of things, like the stiffener plates that were missing, and the differences in the width of the plate, and other things like the amount of thermal expansion, things like that, it's all valid. And that's something that in that limited case does actually make a difference. But what we've got to look at is the actual global case. And when NIST did it they didn't use just that one collapse of that one girder. They used a large number of other collapses of other girders.

    Tony: I hear what you are saying. I saw you say this on the internet, okay, recently. But you have to be specific. You can't tell people we found some other areas where it collapses - well how did that happen? You have to show. They are not showing that.

    Mick: That's what came out of their simulation.

    Tony: Well, I want to see the results, the simulation is going to show the stresses ...

    [J.B. asks about missing data]

    Tony: I'm questioning why Mick supports it, if he doesn't have a basis of justification for supporting it.

    Mick: Right, well what I'm trying to explain to people is your criticism of the NIST report hinges around this on connection, the connection of girder A2001 to column 79, and the fact that you think it couldn't get off the seat, and if it did then it wouldn't damage the floors below. However, that's not what NIST are claiming in the global model. They claim that, they describe that as a possible initiating event, or a probably initiating event, because they did some simulations of just that one thing because it seemed like a likely thing.

    Tony: It's the main hypothesis Mick.

    Mick: Now, you've identified some problem with the simulation, and that's perfectly valid, that we should look at wether that's actually possible, if that girder did actually go off its seat or not. And it's possible that it did not. That is something that should be looked into. However that doesn't invalidate the entire study because the actual simulation that they used did not that as a necessary component.
  2. Mick West

    Mick West Administrator Staff Member

    So I was a bit surprised at this bit of the conversation with @Tony Szamboti
    I think there's a general misconception that in NIST's modeling this:

    NCSTAR 1-9 WTC7_unlocked.pdf (page 395 of 797) 2018-01-13 17-20-16.

    Fed directly into this:

    and then into:

    NCSTAR 1-9 WTC7_unlocked.pdf (page 654 of 797) 2018-01-13 17-12-52.

    But it didn't, in fact in the 16 floor ANSYS model that Tony refers to, that connection did not actually initially fail.

    This has been discussed at some length years ago, but I think it would be useful to get a more definitive description of what's actually going on. Here's the old discussion, prompted mid-thread by @gerrycan posting a video tracing A2001 as it fell in the global model.

    Source: https://www.youtube.com/watch?v=VQkylMIuH-g

    In subsequent posts I start out thinking NIST never said A2001/C79 failure was an initiating event, but then discovered some places where they did (obviously). However in large sections of the various NCSTAR reports the initiating event is discussed just in general terms as failure of the floors around C79-C81. Like in this post:

    This all feels like partially covered old ground, and I'm likely missing something. But it's Saturday evening, and I shall have to continue later.
  3. Christopher 7

    Christopher 7 Member

    I gave you the NIST quote than confirms the walk-off of the girder at column 79 was the event that triggered the cascade of floor failures.
    Here it is again.
    NCSTAR 1-9 Vol.2 pg 611 [PDF p. 677]
    There were other failures but they did not trigger a progressive collapse. Without the walk-off of that girder between columns 79 and 44, the cascade of floor failures does not begin.

    mod add link http://ws680.nist.gov/publication/get_pdf.cfm?pub_id=861611
    Last edited by a moderator: Jan 14, 2018
  4. deirdre

    deirdre Moderator Staff Member

    you've been a member for 3 months. Mick is talking about realizing this in 2013. I think you are missing the point of his question.

    the global model shows the floor could have collapsed without the 79-44 walk off. (although when I picture the entire structure and read EVERYTHING nist says about what was happening prior, I don't think Tony's claim that it couldn't have walked off is correct anyway, but that's me.)
  5. Christopher 7

    Christopher 7 Member

    Mick made a statement, he did not ask a question.
    That's the same claim he made in September and it is not true. The NIST quote I posted clearly says that the walk-off of the girder at column 79 is what triggered the cascade of floor failures that led to the buckling of column 79. They offer no other explanation as only one is necessary. The collapse never would have started without that girder falling and starting the cascade of floor failures around column 79.

    Tony pointed out several reasons why the NIST scenario is impossible. The most damning is the blatant fraud of leaving out the web stiffeners that would have prevented the bottom flange from folding.
  6. deirdre

    deirdre Moderator Staff Member

    the thread title is the question.
  7. Mick West

    Mick West Administrator Staff Member

    Trying to find exactly where in the NCSTAR 1-9 report this connection acquired such significance. Here it is discussed in the context of the damage from the ANSYS model in section 11.4.1 Discussion of Results for Case B Temperatures. There's quite a large list of failures.

    Then in Section 13.1 we have:
    And yet this is not reflected in the LS-DYNA model which is supposedly initiated with the Case-B 4.0h ANSYS damage. Here's a more detailed tracking of A2001 (green) in the LS-DYNA full height model.

    Source: https://youtu.be/cqh1Ye3Mt9s
  8. Mick West

    Mick West Administrator Staff Member

    The progression of damage in the LS-DYNA global model is described in NCSTAR 1-9, Chapter 12, section 12.4.4 (page 572, pdf 638)

    NCSTAR 1-9 Structural.
    This matches what we see in the animation.
  9. Mick West

    Mick West Administrator Staff Member

    To be clear here, in the close up animation the exterior walls and columns are not show. A2001 goes from 79 to the exterior wall column 44, which is (from this view) to the right. You can only first see it here, which I unfortunately missed when coloring it in.
    Combined tracking of A2001 2018-01-15 11-10-38.

    Then you see more in the next frames
    Combined tracking of A2001 2018-01-15 11-11-19.

    Combined tracking of A2001 2018-01-15 11-11-47.

    At this point it's still attached to C79 and C44, just sagging
    Combined tracking of A2001 2018-01-15 11-12-13.

    There's some valid criticism of NIST here. Saying that A2001 initiated the collapse, but not actually showing it in the their global model. Instead they seem to be dumping a bunch of accumulated damage in at once, and the collapse happens without A2001 being the cause.

    Note when they refer to thing on floor 13, that means on the floor of that floor. The girder on floor 14 from C79 to C 44 does not fall until the columns buckle:
    Combined tracking of A2001 2018-01-15 11-37-53.
    • Useful Useful x 1
  10. Oystein

    Oystein Active Member

    Hehe I opened a thread on this very topic at ISF (formerly JREF) a bit over two years ago - but it didn't get anywhere:

    I raised the problem that NIST incurs all of the ANSYS connection failures at the exact same instance in the LS-DYNA model, when in fact they would have occurred at different points in time. This changes the dynamics of the collapse - significantly, I'd guess.
  11. Mick West

    Mick West Administrator Staff Member

    I think it's all a bit greek to most people.

    Quite significantly. As it is you get a simultaneous buckling failure of SIX girders, none of which is A2001 (79-44).
    Tracking WTC7 Girder C44-79 at Floor 13 more.mp4 2018-01-15 13-36-12.

    There's also a failure of 81-80 on floor 12, but that does not seem to buckle.

    So the LS-DYNA model does not really work well as a continuation of the ANSYS model. The sudden dumping of damage is not realistic. But those are the types of shortcuts when the model takes 8 weeks to run a full 25 second simulation.

    It demonstrate that progressive collapse of those floors leads to global collapse. But does not gel with the narrative of A2001 being a "key girder".

    I'm surprised that Tony was not familiar with this, and that it's not been raised more often. When I've raised it with @Gerry can in the past he's just said things along the lines of "that's how crap the model is".

    It's odd, as it seems like a big deal, and clearly people have noticed it before. Maybe it's just too difficult to explain to people.

    There's no detailed NIST analysis specific to A2001 walk off, is there? It's just a product of the 16-floor ANSYS model?
  12. Mick West

    Mick West Administrator Staff Member

    Not only does the LS-DYNA model seem to show A2001 not failing, the graphical representation of the output of the ANSYS model does not seem to show it either:

    Section 11.3
    NIST_NCSTAR19_909257 DRAFT.pdf (SECURED) 2018-01-15 15-40-42.

    (This is the diagram from the draw report which used nicer vector graphics, but the data is the same)

    Notice in the horizontal damage it's marked as "Full Connection Damage"

    But in the vertical, it is marked as "No connection damage".
    NIST_NCSTAR19_909257 DRAFT.pdf (SECURED) 2018-01-15 15-45-33.

    So it looks like according to the 16 floor ANSYS simulation, A2001 did not walk off its seat.

    So why did NIST say that it did?

    In addition, all the girders along 79-81 are marked as Buckled/Failed.
  13. Mick West

    Mick West Administrator Staff Member

    I said the draft graphics above were the same as the final graphics. but it turns out not. Here's the final report's version:
    NCSTAR 1-9 WTC7_unlocked.pdf (SECURED) 2018-01-15 15-55-48.

    This is a rasterized version of the vector graphics, done so the pdf displays faster, but it makes the dots look jagged here because they did not use anti-aliasing.

    However that allows us to see that they manually changed the vertical connection status at C81 and C79. As they were added with a circular brush in a program like Photoshop they have anti-aliased edges, meaning soft grey pixels.
    8*) * 2018-01-15 16-04-57.

    However, the Summary description is unchanged in both versions, describing two girders as having walked off the bearing seat.

    This piece of consistency is then spoilt by the LS-DYNA model, where those two connections remain firmly attached.

    All seems rather messy.
  14. Mick West

    Mick West Administrator Staff Member

    A manual change was also made to the C81 connection (girder from C26). This appears to be the only other one.
    NCSTAR 1-9 WTC7_unlocked.pdf (SECURED) 2018-01-15 16-31-48.
    The Case C damage state diagram seem unchanged.
  15. Mick West

    Mick West Administrator Staff Member

    I've added "(and NIST's)" to the title of this thread, as it's not really clear if A2001@C79 failed in the ANSYS OR the LS-DYNA simulations.
    • Like Like x 1
  16. Mick West

    Mick West Administrator Staff Member

    One must also not forget the other simulation NIST did involving A2001, section 8.8, simulation that results in the girder NOT walking off, but failing via buckling.
    NIST_NCSTAR19_909256 DRAFT.pdf (SECURED) 2018-01-15 17-24-06.

    NIST_NCSTAR19_909256 DRAFT.pdf (SECURED) 2018-01-15 17-25-25.

    This one was described as:
    This analysis is generally rejected by AE911 as the W24x55 beam next to col44 is missing some connecting beams that would have inhibited buckling. That's one of the points made by Hulsey and repeated by Tony in the "debate" with me.

    However this is not something that got into the ANSYS model or the LS-DYNA.

    This raises a somewhat discomforting possibility, that someone at NIST mistakenly thought that this study actually proved the push-off theory, and that mistakenly got shoehorned into the report.

    This seems unlikely though, as there were three changed to the ANSYS -> LS-DYNA data, not just the one that might arise from such a mistake.

    A more likely explanation is that different runs of the ANSYS model indicated the three walk-offs.

    It's certainly not clear though.
    Last edited: Jan 15, 2018
  17. Nada Truther

    Nada Truther Active Member

    For me it is a bit like Greek, translated into Spanish, written in Japanese, and spoken back in Klingon sign language.

    Carry on, I'll just watch from here.
  18. Jeffrey Orling

    Jeffrey Orling Active Member

    All of these models are

    WHAT IF...

    there is no "hard" data... some perhaps derived from videos and "historical" understanding of fire... and steel structures, statics and so on on.

    Ergo it is totally absurd to take these models are representing anything other than assumptions made to create them.

    Add to this that the fire event was incredibly complex and the system under stress were also extremely complex.

    This amounts to debating how many angels can dance on the heat of a pin.
  19. Mick West

    Mick West Administrator Staff Member

    That's not the point. The point is the high level interpretations on the model (by both AE911, and NIST themselves) don't match the models. How well they match history is another aspect entirely.

    AE911 go on and on about how A2001 on floor 13 could not have walked off the C79 seat. It's a big part of their $300K study. But in the draft diagrams of the ANSYS model results, and in the actual detailed results of the Global model it did not walk off (fail vertically). Nor did it walk of in the local model of just that corner, although it did fail vertically off the other side.

    NIST, in a few bits of the report describe A2001 as a key girder, and say it did walk off. The diagrams were hand edited to indicate it (and two other girders) walked. But the bulk of the reports just refer to general floor collapse.

    In NIST's presentation back in Nov 2008 we have:
    WTC7RevisedTechnicalBriefing_111908.pdf 2018-01-16 09-14-00.

    So there's a slide (#33) showing the girder was pushed off its seat, right next to slides (#31, #34, #35 & #36,)based on simulations in which that does not happen.
  20. deirdre

    deirdre Moderator Staff Member

    so... Hulsey and his students wasted 2 years trying to show something that the NIST models already show? That's messed up.

    and the NIST global model clearly shows that a2001 doesn't have to have fallen first for progressive collapse to have happened anyway. My poor little brain is like "huh?". o_O
  21. John85

    John85 Member

    The NIST reports do then describe several different and slightly incompatible sequences to account for the loss of support around column 79, but the buckling of this column is nonetheless meant to be what turned a series of somewhat undetermined local failures into a global, progressive collapse. It's definitely right to study column 79, and I think we can all agree that if NIST proposed a mechanism, even if only one of several, that only works if a series of errors are inputted, some significant doubt is cast on NIST's work. Are we able to dismiss this doubt by looking at other possible initiation mechanisms? Well no, because they have ignored requests to release their model, and refuse to correct their work for errors already identified.
    • Disagree Disagree x 1
  22. Mick West

    Mick West Administrator Staff Member

    Their global model collapses, but it it does not use the "series of errors".
  23. John85

    John85 Member

    By what mechanism? Is there any girder walk-off supported by data?
    • Funny Funny x 1
  24. Mick West

    Mick West Administrator Staff Member

    Perhaps you should re-read the thread?

    NIST say that the ANSYS model showed walk-off of three girders, two on floor 13 and one on 14. This was not indicated on the damage diagram in the early draft, and was manually added to the diagram in the final report.

    The global models do not have any girder walk off. All the initial failures come from buckling, which is the mechanism of the initial collapse in the global model.
  25. Christopher 7

    Christopher 7 Member

    Mick, You missed something. The girder does fall about 1/10th of a second after your 3rd screen capture from the LS-DYNA simulation. (lower image)

    Key girder failure in LS-DYNA sim.
  26. Mick West

    Mick West Administrator Staff Member

    I know it falls, I don't think I missed anything. It appears to be a buckling failure.
  27. Christopher 7

    Christopher 7 Member

    A lot of people, including myself, were confused by the “rock off to the west” analysis. As it turns out, the “rock-off” analysis was an interim analysis. NIST carried the shear stud failures forward to the 16 story ANSYS and 47 story LS-DYNA analyses but not the “rock-off” part.

    NCSTAR 1-9 p. 349 [PDF p. 393]

    p. 353 [PDF p. 397]
  28. Christopher 7

    Christopher 7 Member

    I'm not one to make excuses for NIST but it appears that NIST noticed or was informed of the error and corrected it.
  29. Mick West

    Mick West Administrator Staff Member

    What error? They used the vertically undamaged connections in the global model. What are you suggesting happened? That they had them undamaged in the diagrams by mistake, then corrected the diagrams, but not the data transferred to the global model.
  30. Christopher 7

    Christopher 7 Member

    As I noted in post #25, they did use the vertical damage on the global model. As you noted,
  31. Mick West

    Mick West Administrator Staff Member

    No they don't. If they did then it would have immediately fallen and not sagged.
  32. Christopher 7

    Christopher 7 Member

    ??? You said that it was just sagging but still attached.
    It was sagging as it was being pushed off the seat but it did not fall due to the sagging or it would have fallen before it came into view in the simulation.
  33. Christopher 7

    Christopher 7 Member

    Furthermore, it is unthinkable that they would not include the trigger that started the collapse. That was a critical part of their presentation as you noted above.

    Here is the video of the August 21, 2008 press conference where NIST presented its findings on the collapse of WTC 7.

    Source: https://vimeo.com/11955064

    Starting at 9:43, Shyam Sunder explains the NIST collapse hypothesis.

    The critical part starts at 12:55 when Sunder explains the NIST hypothesis of the failure of the girder between columns 79 and 44 on floor 13 being the failure that led to the total collapse of the building.
    Last edited: Jan 17, 2018
  34. Mick West

    Mick West Administrator Staff Member

    It's sagging in the middle. It's attached at the ends. It seems to fail because the beams pull it down. It has not failed in the vertical direction at the start of that simulation.
  35. benthamitemetric

    benthamitemetric Active Member

    It is interesting that the ultimate failure mode for A2001 that seems to be shown in the LS-DYNA simulation is similar to the failure mode identified by ARUP (and later confirmed by WAI).

    This thread definitely raises some good questions about the internal consistency of NIST's approach, in any event, and I think Mick has done a good job parsing through the various descriptive narratives. Given the volume of the NIST reports and how it was published in an unsearchable format, such parsing is certainly a chore.
  36. Mick West

    Mick West Administrator Staff Member

    Why is it unthinkable? Regardless, it very clearly happened. SIX other girders collapsed in the global model. A2001 only fell later.
    In the frame above A2001 on Floor 13 has not yet visible moved.

    I've attached a 5fps full size version of the girder tracking with is easier to frame advance.

    Attached Files:

  37. Jeffrey Orling

    Jeffrey Orling Active Member

    If the sequence was anything like what is shown in this animation... it's absurd even to consider a single node as THE straw that broke the camel's back. It appears that there were many failures all interacting leading to the collapse. And this looks reasonable. But it could be different and ALSO look reasonable.

    So where is the CD "version" of this showing how the frame came undone and the motion we saw? Was this collapse engineered? If it was then why can't truthers engineer a CD now with years to work on it???
  38. Oystein

    Oystein Active Member

    Haha great finds, Mick! The photoshopped connection failure dots - excellent! Makes one wonder why none of the 2964 architects and engineers who have a problem with the NIST reports ever found these problems!

    My hunch is this: Backwards-engineering, they needed a failure at col 79 first, to match the observed collapse sequence. Perhaps ANSYS didn't give them enough of this, so they looked manually into the state of the key connections within ANSYS, and found some failure even though ANSYS' criteria were not triggered. That would actually be a smart way to use FEA results, though it's sensitive to bias.

    At the same time, there seems to be some difficulties with editing the entire report. I'd guess that different teams worked on different chores and chapters, some of the work having been done asynchronous with the chapter sequence.

    Chapter 8 has evidence for the plausibility of several failure modes.
    Chapter 11 shows that multiple connection failures around columns 79-81 are to be expected from the wandering fires on many floors.
    Chapter 12 shows that multiple beam failures around columns 79-81 can progress to global collapse.

    The safety recommendations adress systemic failure modes and progression proneness. Those would be valid even if NIST doesn't nail the actual details of collaps initiation.

    I think it is a good idea at this point to take NIST's detailed hypothesis exactly as the marketed it: Their most probable scenario at the time (almost 10 years ago), though not very probable in absolute terms.

    We know there was fire
    We know there was no CD
    We know there was a collapse.
    We thus can be sure that fires caused the global collapse - and NIST shed valuable light on how such a thing can occur, even if the ultimate cause, the straw that broke WTC7's neck, has not been demonstrated beyond doubt.
    • Agree Agree x 1
  39. Mick West

    Mick West Administrator Staff Member

    But in the global analysis C79 still buckled first, even though the vertical connection failures were not there.
  40. Oystein

    Oystein Active Member

    That's the global, asynchronous editing then - perhaps LS-DYNA work was already too advanced (and had eaten too much of the budget) by the time they decided to foucs on A2001?