Sept 3, 2019 release of Hulsey's WTC7 draft report: Analysis

Status
Not open for further replies.
Kostak Studios has now also put a video out that demonstrates very clearly that there is no dynamic analyses being performed in the "simulations" underlying Hulsey's released animations. Specifically, this video shows (via an analysis of pixel movement) that there is no movement within Hulsey's model other than the seemingly scripted movements of the penthouse drop and the subsequent uniform movement of the top 30 floors or so:



As suspected, there is no actual deformation or failure analysis being performed on the individual elements of the building during the scripted collapse sequences.
 
Last edited:
From Hulsey's Sep 03 presentation, I am only quickly dumping two bit of transcript that I have prepared - just so the work doesn't go to complete waste. Both are from the Q&A at the end.
The first - I initially thought Hulsey had the definition of "live load" entirely wrong, but then I looked it up, and it isn't so bad. Still slightly weird that he thinks that the evacuation of people caused a significant decrease of loads. WTC7 weighed on the order of 100,000 tons[*]. A significant percentage would be at least 5% of that - 5,000 tons, the weight of 67,000 people (at 75 kg a piece) - NIST estimated 4,000 occupants on 9/11 (NCSTAR 1A, page 56). Does he think the movement of people at work in an office building is relevant?
1:07:06 Keep in mind that when you design a building like this,
1:07:09 you're designing it for dead load, which means that's the
1:07:12 self-weight of everything in there, and then you're going
1:07:16 to be designing it for live loads, that's service ability
1:07:22 condotions, that's the use of the building. And that's
1:07:25 what you would size it for, and then you would size it to
1:07:29 being that you've got a factor of safety of somewhere
1:07:31 about 1.5, maybe even one-point, maybe even 2.
1:07:34 Ok. Ok. In this case, when that building came down, there
1:07:39 was nobody in it! There was no live load! There was only
1:07:44 dead load!
Content from External Source
The second snippet is about the formation or non-formation of the kink in the east part of the north wall at the beginning of global collapse:
Questioner:
1:09:08 I couldn't help but notice a discrepancy ... [unintelligible]
Hulsey:
1:09:21 Got a kink in it?
Questioner:
1:09:22 ...kink in it. The model did not.
Hulsey:
1:09:25 Right.
Questioner:
1:09:26 Is that significant? Is that something to you?
Hulsey:
1:09:27 Ours did. It was just not as deep as what it looked like
1:09:31 on camera, in the video. But it did have a kink.
1:09:36 We spent three weeks trying to find that kink,
1:09:38 and seeing what would cause that kink to occur.
1:09:44 Cause you know you wouldn't have expected it to occur
1:09:48 uh, but it was there! The other thing I didn't bring up
1:09:52 to your attention, which I should, is if you look at the,
1:09:55 if you look at the building, it has, it had a
1:10:01 surfacing on the outside, and windows and all that kind
1:10:04 of stuff. And if there was any relative movement of one
1:10:08 column to another, windows would be breaking,
1:10:11 you would see cracks and distortions in the exterior
1:10:15 sheathing, for those were non-structural. You didn't see
1:10:18 any of that, hardly any of it. So even in our computer
1:10:22 model, we were not getting very much relative movement
1:10:25 to cause that kind of stuff to happen. So that in itself
1:10:29 was have a significance. You just step back and
1:10:32 take an overview, and look at a building, it did not so
1:10:36 this [moves hands up and down in opposite directions]
1:10:38 Otherwise you would have had breaking of glass, you would
1:10:41 have had breaking of the exterior surfacing. And that's not
1:10:46 structural, it's not hardly any loaded that's capable of
1:10:49 transferring that.
Content from External Source


[*] 9 years ago today, coincidentally, I did an estimate of total weight, and of Potential Energy released by WTC7, in my other blog:
https://oystein-issues.blogspot.com/2010/09/gpe-of-building-wtc7.html
Short version: I had found earlier that the specific gravity of the twin towers was about 1/6 that of water, and applied the same to WTC; the volume of WTC7 was height x depth x average of north face width and south face width = 186 m x 42.7 m x ((75+100)/2) m = 694,942.5 m^3, and mass then 1/6 tons / m^3 x 694,942.5 m^3 = 116,000 tons.
PE released works out, conservatively, to 80 GJ.


Source: https://www.youtube.com/watch?v=TAEHhDCTaBw
 
Last edited by a moderator:
To understand what we could expect from Hulsey, here are two examples of what appear to be dynamic models from one of the modeling programs he used, Abaqus FEA:



This is from Abaqus Explicit, which is in the Abaqus FEA suite.



So graphics with this level of realistic detail are possible with the software Hulsey used.
 
That also the first thing I really noticed from just looking at the videos. The difference between what they show and what actually happened are sufficiently dramatic that they cast serious doubt on their models.

In a nutshell, the sides of the penthouse rotate around the TOP of the sides in their model, but in reality, they rotated around the bottom - which is exactly what you would expect.
Figure-4.24a-vs-reality.gif
Notice here the corner closest to the camera - on the right, and highlighted in blue in the simulation. In reality, it pivots around the base, just falling into the building to the left. In the simulation, it does a bizarre, inexplicable pivot outwards.

This is even clearer in the front view
Figure-4.20-strange-pivot.gif

Notice nothing underneath the penthouse is moving. Not only is this motion radically different to observed reality, there's also no explanation for why their simulation would give this result.


I've looked at the two simulation views of the penthouse: Something doesn't add up.

The front view shows the penthouse breaking in the middle, but the side view shows (accurately) the penthouse breaking ca. two thirds in, from the camera.

We're not talking about perspective here: I've counted the grid lines, and there's an even number of grid lines on each side of the penthouse (front view), but a smaller number of grid lines to the far and a larger number of lines to the near (side view).

When we look at the front view footage of WTC 7, the penthouse clearly breaks in a third og two-thirds of the length.

Am I right or do I need more coffee? If I am right, then Hulsey is using different simulations for each view.

/Claus
 
We're not talking about perspective here
I think we are. The grid spacing is irregular on the right side, and there's a truncated corner on the left. Here I've blanked out all the front panels. Metabunk 2019-09-12 07-09-57.jpg

And here's a view of the top of the building (from a photo I took in 1998), showing the irregular shape of the penthouse (viewed from the other side)
Metabunk 2019-09-12 07-13-20.jpg
 
Last edited:
Also from Hulsey's Sep 03 presentation, the moment where Hulsey categorically refuses to offer an explanation for the magical disappearance of all columns over 8 stories:

Questioner:
1:08:35 You said that the fire did not cause the collapse.
1:08:39 Do you any hypotheses of what DID cause the collapse?
Hulsey:
1:08:42 I'm not going there.
(Background laughter)
1:08:51 I can tell you what had to happen to make that, and I
1:08:54 showed you that.
Content from External Source

Source: https://www.youtube.com/watch?v=TAEHhDCTaBw&t=1h8m35s
 
In the old, closed thread on the Hulsey study, I had written this in post #3 on Jan 14, 2015 - more than 2 months before Hulsey officially started:
A CD simulation could cut the 13th-floor 44-79 girder just where NIST thinks it walked off, then cut the floors below 13 all the way down to the 8th as NIST says they fell, then cut column 79, 80, 81, then ... some more in the core, along with floor connections, and then perimeter columns along the 8th floor or so - and wonder upon wonder, it would look just like the real thing! :D
I wonder if they also simulate the demo charges themselves - either explosives, or incendiary steel-melting stuff ;D They'd have to offer numbers for the size of the charges, which could be used to estimate sound levels etc. Or, if they truly want to melt columns in a rapid and controlled manner, simulate heat transfer within the columns...
This was a mock suggestion for a very lazy, trivial and non-explanatory simulation approach. It turns out, Hulsey's approach was even simpler, even less realistic: He did not bother to to anything with floor connections. He did not even "cut" columns. No, he made them disappear magically. Sections 4.4 to 4.6 (the three scenarious to trigger global collapse: East core only, East core then west core; East core then west core then entire perimeter) most of the time uses the phrase "simultaneous failure of ... columns over 8 stories" (and variants), but what failure mode does that imply? I think the solution is found in Section 4.4 on page 98 (my bolding):
With the removal of six core columns on the eastern side of the building...
Content from External Source
Sounds as if he does not break, buckle, cut columns - he simply removes them over 8 stories.

Compared to my simplistic suggestion in January 2015, Hulsey dumbed his down in the following ways:
  • Ignored floor system failures
  • Did not "cut" columns, he just made them disappear without explanation
  • Did not attempt to replicate the actual sequence of failures (e.g. west core east-to-west progression)
  • Did not attempt to analyse the effects of cutting charges, as he refused to hypothesize any demolition scenario
 
Could it be sloppy language? He mean they were rendered structurally invisible... non performing columns???
No also in the report they talk about 'removed' literally; page 94
The linear static analysis results for the simulations where we removed Columns 79, 80, and 81 at different floor intervals are shown in Figures 4.2 to 4.7 below.
Content from External Source
 
In the old, closed thread on the Hulsey study, I had written this in post #3 on Jan 14, 2015 - more than 2 months before Hulsey officially started:

This was a mock suggestion for a very lazy, trivial and non-explanatory simulation approach. It turns out, Hulsey's approach was even simpler, even less realistic: ....

.....Compared to my simplistic suggestion in January 2015, Hulsey dumbed his down in the following ways:
  • Ignored floor system failures
  • Did not "cut" columns, he just made them disappear without explanation
  • Did not attempt to replicate the actual sequence of failures (e.g. west core east-to-west progression)
  • Did not attempt to analyse the effects of cutting charges, as he refused to hypothesize any demolition scenario
Whilst yours Oysten was a mock suggestion and not a serious predictin - Hulsey has also gone down other paths that were predicted.

In the same closed thread I referred to the influence of T Szamboti.
This comment at post #6:
Agreed. Especially if they come up with something a grade or two better than recycling T Szamboti's tired hypothesis.
Then at Post #10:
I wouldn't know - just that their three (?) recent WTC7 based claims have had the same core technical material which is Szamboti's.
...and since those early remarks I have been more pointed in predicting T Szamboti being the puppet master pulling the strings on Hulsey.

So the study has a few major areas of failure before we even go to the engineering details. The first three I can think of being:
1) The impossible "false global negative" claim that he proves "fire could not (or "did not") cause the collapse";
2) The "fudging" of the "simulations" to force them to look visually similar to the real event;
3) The wholesale importation of the Szamboti EPH descent mechanism;
 
I think we are. The grid spacing is irregular on the right side, and there's a truncated corner on the left. Here I've blanked out all the front panels. Metabunk 2019-09-12 07-09-57.jpg

And here's a view of the top of the building (from a photo I took in 1998), showing the irregular shape of the penthouse (viewed from the other side)
Metabunk 2019-09-12 07-13-20.jpg
Why would the grid spacing be irregular on one side?

If that is true, that totally screws up any visual clues.

/Claus
 
I don't know that Tony is a puppet master as you suggest.... but it seems that the truthers have a huge lack of structural engineers who have done anything in depth whether right or wrong... and kept at it. Gorden Ross I believe dropped out after his claims that the corners were destroyed and the towers then could not stand. Gerry has no technical explanation for the collapse. He likes to nit pick details.... right or wrong. In marches Hulsey with a university engineering department and shows nothing new.... not even showing NIST's blunders and fraud.

Why hasn't the truth movement put out what they suggest happened to the structure of these buildings with some level of specificity?
 
Why would the grid spacing be irregular on one side?

If that is true, that totally screws up any visual clues.
In a SAP2000 model, the lines should represent structural members. So for accuracy in modeling the collapse they should mirror the actual penthouse. So to verify this it might be visible in photos and should be there in the blueprints.
 
This shows the "posts" (colunns) of the penthouses:

NCSTAR 1-9 Fig 2-12 Roof Layout.jpg

Posts are closer to each other in the western part of its north wall.

(I count 8 posts in this figure, but more in the SAP2000 animation)
 
I was looking at the drawings for the penthouse, and got sidetracked to floor 13. The framing (girders between columns, and beams between girders and girders or columns).

I noticed when I overlaid sheet S-8 (attached) with Hulsey's figure 2.63 there was a discrepancy around the stairwells.
Metabunk 2019-09-13 10-06-38.jpg
Notice the vertical white line in the middle of each indicated area. That's a beam that goes the full length of the section in Hulsey's model, but is not there in the actual drawings, as there's a stairwell there, which is framed around.

Look at the area between columns 58,61,59, and 62.
Metabunk 2019-09-13 10-09-25.jpg

White lines are Husley, red lines are S-8.

It then started noticing this elsewhere. Hulsey's figure 2.63, supposed of floor 13, has quite different framing in the core to drawing S-8 (described as "Typical Floor Framing Plan, 8th to 20th & 24th to 45th")z

Here I've marked missing members in green, and added members in blue.
Metabunk 2019-09-13 10-30-10.jpg

Some detail:
Metabunk 2019-09-13 10-31-02.jpg

Besides all the changes, notice one in particular that's a bit different. On the right side, there's a beam that has moved
Metabunk 2019-09-13 10-32-39.jpg
The Blue one is Hulsey's lone beam, the green one is the S-8 drawing, it's 11 inches over to the left (the West).

The next beam over is also wrong. It should be inline with the beam south of it, and both beams should be in line with column 24. Instead they are out of line.
Metabunk 2019-09-13 10-55-01.jpg



Related, Hulsey's figure 2.59 similar seems to miss most of the framing in that region.
Metabunk 2019-09-13 10-42-02.jpg


Hulsey appears to be working of E12-13, which shows the same detail as S-8. He includes E12-13 on page 15, saying:
"Figure 1.10 shows the framing plan view for Floors 12 and 13. The drawings are difficult to read because of the number of framing members shown on the drawing."
Metabunk 2019-09-13 10-48-13.jpg


Here's a HD image of E12-13 https://www.metabunk.org/f/E12-13.jpg

That's his old ABAQUS model, what about the SAP2000 model? It's a bit hard to tell, as he displays it here with two floors overlaid. However, it does look like it might have proper framing?
Metabunk 2019-09-13 11-03-34.jpg

The next two figures clearly show the simplified framing, it's not specified where they are from, but probably ABAQUS.
Metabunk 2019-09-13 11-06-37.jpg

(This appears to be a perspective overhead view of two floors, which is a bit odd, that's why things don't line up)
 

Attachments

  • S-8.tif
    3.4 MB · Views: 618
  • Metabunk 2019-09-13 10-30-10.jpg
    Metabunk 2019-09-13 10-30-10.jpg
    683.8 KB · Views: 487
"but for now it is just based on my 'feeling', no hard data.
'Feelings' indicate bias and perhaps is not a relevant observation to state.
As for the video, I think something interesting is that these broken windows happen at the exact moment the penthouse begins to collapse, suggesting the entire structure underneath the penthouse has already failed - the broken windows indicate some collision on those floors. The penthouse is still completely visible and has only just begun to move as the first windows break. How is this possible?

I was under the impression that the penthouse collapsing is what caused the progressive collapse of the interior, leading to the exterior collapsing inwards? Even in the video, it seems like the interior columns are all just simultaneously collapsing; you say what we see is a 'progressive' collapse, but what is making these collapses automatically happen where fire was only on 6-8 floors? How is healthy structure supposedly all giving way at the same time in a fire that wasn't a raging inferno and burnt uncontrollably with unevenly distributed fuel (office materials as there was no plane on WTC7)?

I'm skeptical that fire can cause global damage to 3 buildings in 1 day when there are possibly only 1 or 2 recorded cases ever of a building ever experiencing global structural failure due to random burning of entropic fires with unevenly distributed fuel in a building, and even those burned for far longer, encapsulated more of the building and didn't reach near free fall for any duration of their collapses.
 
My idea of what they are doing to "fail" the columns is this:
They simply make all columns literally disappear.
Then the building falls at free fall.
Of course, after falling one floor, floor slab runs into floor slab - what happens then is a guess: Either the falling floor slab disconnects from its columns, or the slab pass though each other without interaction.
Either way, there is no deceleration - jolt. Someone call T.Sz.!

There is no observed jolt in real-life either.. I think you're missing the point. Dr. Hulsey recreated the observed collapse (even if his findings would support NIST it would be exactly the same) so why should there be a jolt? The fact is that the 'lack of jolt' is indicative that Hulsey's model is more accurate, T.Sz would not disapprove. The 'lack of jolt' indicates these columns were already taken out, the floors would all by falling simultaneously, not really touching, therefore, no jolt as they move hegemonically - which is why we see an entire block of building move as one in reality, at nearly free-fall.
 
The 'lack of jolt' indicates these columns were already taken out, the floors would all by falling simultaneously, not really touching, therefore, no jolt as they move hegemonically - which is why we see an entire block of building move as one in reality, at nearly free-fall.

I suspect we actually see that because it's not a real dynamic analysis. Of course, if it was, then removing all the columns magically instantly would have the same initial result. But there's the issue of what happens when the falling part of the building meets the bottom part. This is very different to the chaotic leaning gradual disintegration collapse of the Towers. Under Hulsey's hypothesis for WTC7 the columns would meet at full force because they are all still there. Of course, he can't actually demonstrate what happens, because his analysis is not real.
 
It's virtually impossible to remove an entire 2 story length column let alone all of them in the plan. What happens theoretically is of no consequence to any real world scenario.
 
There is no observed jolt in real-life either..
How do you know this? Can you present evidence for that claim?

But what we see in real life is irrelevant anyway - we are discussing the behavior of Hulsey's animations, not the real life events.
As structure collides with structure in those animations, there should be SOME kind of response, but none is apparent.

I think you're missing the point. Dr. Hulsey recreated the observed collapse
No, this is FALSE, he didn't. See my post #67 where I showed that Hulsey mentions only six of many more features of the observed collapse, but then only recreates 3 of the 6 he himself found noteworthy - and each one of the 3 is really forced onto the simulation artificially: All 3 features are input to the model, not result of any hypothesis.,

so why should there be a jolt?
Because after about 1 floor height of actual free fall (falling structure runs into no other structure), the falling stucture runs into standing structure - there is no more free fall. This should result in deceleration / a decrease of acceleration - which is called a "jolt" (jolt = sudden change of acceleration)

The fact is that the 'lack of jolt' is indicative that Hulsey's model is more accurate, T.Sz would not disapprove.
Urrrr... accurate as measured against what? Are you sure you aren't conflating T.Sz.'s "No Jolt" findings for WTC1/2 with Hulsey's WTC7 model? Different buildings!

The 'lack of jolt' indicates these columns were already taken out,
But in Hulsey's various models they, nor the floors, are not. We are discussing Hulsey's models, and how they fail do model the behaviour of real floors and columns. This is before any consideration whether this also fits the reality of WTC7's collapse.

the floors would all by falling simultaneously, not really touching, therefore, no jolt as they move hegemonically - which is why we see an entire block of building move as one in reality, at nearly free-fall.
But in Hulsey's models, there are floors connected to columns - and these floors crash into other floors. There shouuld be SOME structural/dynamic response - in Hulsey's model. Remember, we are discussion Hulsey's model.
 
Given that it now seems clear that Hulsey did not perform the dynamic modeling necessary to support his conclusion re the global collapse, I think it is worth taking a few steps back and looking at the consequences of that failure in context.

That failure certainly confounds Hulsey's immediate aim (i.e., demonstrating that his conjectured column deletion scenario would result in a collapse similar to what was actually observed), but that ultimately doesn't even really matter in the grand scheme of things. If Hulsey and his graduate students had competently created a true nonlinear dynamic model of the building, then it seems to me that there should be no technical barrier to them creating a demolition-via-structural-element-deletion scenario that closely resembles the collapse that was actually witnessed. If they were very clever, they likely could even come up with fanciful scenarios that don't involve obvious errors (such as requiring the magical unspoken deletion method do work on floors that were involved in fire, as conjectured in the released report). It would just be a matter of trial and error to come up with such "just so" scenarios whereby a collapse with certain visual characteristics could be induced by magically deleting structural elements in a certain sequence; in fact, it should logically be the case that there are infinite such "just-so" scenarios out there in the ether, waiting for Hulsey and co. to stumble upon them.

Such "just-so" modeling this is what I actually predicted Hulsey would do in the other Metabunk thread here:

Only time will tell what Hulsey ultimately chooses to do, but my impression was that the preliminary report power point presentation that has already been debunked in this thread was the full extent of the fire analysis Hulsey intends to do and that the second phase of the project would focus exclusively on coming up with a non-fire-based theory for the observed global collapse. In any case, whether or not Hulsey actually revisits his flawed fire analysis, the apparent difficulty he is having in the second phase should certainly give him and those inclined to believe him pause about accepting at face value his previously stated conclusion re the inability of fire to initiate a total collapse of the building. (If Hulsey actually believed in the soundness of the first part of his analysis, by the way, there is absolutely nothing stopping him from submitting such analysis to the Journal of Structural Engineering to rebut and correct the same's earlier publication of the NIST report, regardless of the status of his second phase analysis.)

My guess (based on a series of angry emails I received from someone affiliated with the project) is that the UAF report is being written primarily by Hulsey and Tony Szamboti at this point, and so I expect it is just a matter of those two playing with modeling software to come up with a "just so" set of explosives that could result in a global collapse that is substantially similar to what was actually observed. In fact, if the final theory offered by Hulsey isn't some variation of Tony's long-proclaimed theory that some sort of near simultaneous demolition of the core columns across 8 floors or so lead to the global collapse, I will donate $100 to AE911Truth.
Content from External Source
And then here:

I think you guys are making it more complicated than Hulsey and AE911Truth are going to make it. I don't want to speculate too much, but I expect that demolition devices are largely going to be assumed to be effective at felling columns and that it's just going to be a matter of Hulsey and whoever is working with him placing them around the building in a just-so pattern. Does anyone really expect a fulsome model of a hypothetical nanothermite cutting device, after all? Since Hulsey and AE911Truth have no basis for modeling exactly what such a device would be, they are just going to assume its effectiveness and model the demolition devices as cuts to columns that occur on timed increments. We all know that such an approach is obviously flawed, and we will be here pointing that out if and when Hulsey ever publishes anything concerning it, but obviously flawed is par for the game AE911Truth is playing, so get ready to be underwhelmed. These will not be serious models that actually meaningfully tread new ground re the type of demolition devices allegedly used; I guarantee it.
Content from External Source
My only real mistake was assuming they would actually be able to competently build the necessary dynamic model to back up their "just-so" scenarios. It seems they could not, but the underlying issue with their methodology remains the same: it proves nothing to recreate the collapse via magical just-so element deletion scenarios. The whole second half of Hulsey's report, flawed as it may be on a technical level, is just a colossal waste of time on a logical level. What Hulsey needed to do was test an evidence-based hypothesis and then see if that hypothesis had better explanatory power than's NIST's. (And even that wouldn't rule out NIST's scenario or other fire-induced collapse scenarios, but it would at least present a plausible alternative to NIST's chosen scenario.) But Hulsey instead failed to correct the many errors we previously identified here on Metabunk with respect to his local model and apparently just barged forward (as I predicted) with a mode of global model analysis that, even if it were properly executed, is devoid of any logical significance.
 
Last edited:
Given that it now seems clear that Hulsey did not perform the dynamic modeling necessary to support his conclusion re the global collapse...

Reminder from the UAF project page:
The research team is currently organizing and uploading all of its data into a format that can be readily downloaded and used. We expect to post the data sometime between September 16 and September 30, 2019.
Content from External Source
Within the next 2 weeks, we should have their data available, which should remove any doubt as to what their models did and didn't do...

...IF we have the software and sufficient hardware, and the software skills to use that data! Does anyone here have such capabilities?
 
Within the next 2 weeks, we should have their data available, which should remove any doubt as to what their models did and didn't do...
Although based on previous time estimates, likely towards the end of, or after, that time period. A significant issue is that the public comment period is up to November 1. So the later they release, the less time people will have to check their data. They should make the public comment period be at least six weeks after that data release.

But yes, that will be interesting. Hopefull it will be structured in a way that it can be easily accessed. They do say:
For anyone intending to download the data, please be advised that there are several hundred gigabytes, so plan your data storage accordingly.
Content from External Source
Several hundred gigabyes will take a long time to download though. Several days for many people - which really should also be factored into the public comment period. The large size will dissuade some people with bandwidth caps.

...IF we have the software and sufficient hardware, and the software skills to use that data! Does anyone here have such capabilities?
I have the hardware capabilities (a reasonably powerful computer with 64GB of memory and terabytes of space). I don't have ABAQUS or SAP2000. I do have some coding skills, so should be able to munge their data into a variety of visualizations.
 
Although based on previous time estimates, likely towards the end of, or after, that time period. A significant issue is that the public comment period is up to November 1. So the later they release, the less time people will have to check their data. They should make the public comment period be at least six weeks after that data release.

But yes, that will be interesting. Hopefull it will be structured in a way that it can be easily accessed. They do say:
For anyone intending to download the data, please be advised that there are several hundred gigabytes, so plan your data storage accordingly.
Content from External Source
Several hundred gigabyes will take a long time to download though. Several days for many people - which really should also be factored into the public comment period. The large size will dissuade some people with bandwidth caps.
It's tempting to interpret this all as deliberate obstruction.

I have the hardware capabilities (a reasonably powerful computer with 64GB of memory and terabytes of space). I don't have ABAQUS or SAP2000. I do have some coding skills, so should be able to munge their data into a variety of visualizations.
I think it could well be important to check and understand how the original software - the versions of ABAQUS and SAP2000 Hulsey used - interprets the data. There may be switches, toggles, parameters embedded in the construction data, or additional to it, that control what the software does or doesn't do.
 
...IF we have the software and sufficient hardware, and the software skills to use that data! Does anyone here have such capabilities?
Not me. But my usual word of caution. We run the risk of being baffled by BS if we go sifting the data before we validate how it is being used. (Or more likely INvalidate. ;) )

The errors are almost certainly in the assumptions, context framing and the structure of arguments. Such has been consistent experience through out the history of WTC 9/11 debate. And it has been a trick that T Szamboti has relied on whether by deliberate intent or otherwise. I identified the error with a 2007 Szamboti paper in my very first internet post when I said:
The paper referenced as Engineering Reality by Tony Szamboti is typical of many which look impressive in detail to the non-engineer. The complex calculations may even be correct but the base premises are faulty and the resulting conclusions can readily be demonstrated to be totally wrong.
Content from External Source
Other examples over the years include the two fatal false premises of Szamboti's "Missing Jolt" and the similar problem with some of Bazant's papers. (Specifically "crush down/crush up). And IIRC every step of the evolution of Szamboti's changing claims about WTC7.

I'm not saying that there will not be errors which can be revealed by application of data. BUT the significant and fatal flaws of argument will almost certainly be in how the data is used.. not in the numeric details.
 
I think it could well be important to check and understand how the original software - the versions of ABAQUS and SAP2000 Hulsey used - interprets the data. There may be switches, toggles, parameters embedded in the construction data, or additional to it, that control what the software does or doesn't do.
Yes. That is sort of "mid level". But the need is first to determine if what he is trying to do is valid THEN - IF it is valid - go down one layer and check the switches and other broad parameters. The actual numbers are yet another layer down into detail and probably of little importance because he will have errors at those two higher levels.
 
Yes. That is sort of "mid level". But the need is first to determine if what he is trying to do is valid THEN - IF it is valid - go down one layer and check the switches and other broad parameters. The actual numbers are yet another layer down into detail and probably of little importance because he will have errors at those two higher levels.
Yes, exactly. Certainly the goal ought not be to compare every hook and ornament against shop drawings - we are not Truthers! :p

The kind of thing I would be looking for is confirmation that the models are SET to behave in ways that have no correspondence to the physics of real steel frame structures. Such as not checking whether beams collide and thus allow them to pass through each other. Or having connections that never break and rather can be stretched dozens of meters, because they are not modeled with a condition or threshold above which they break.
 
While waiting for any data to be released by Hulsey I was looking at his talk where he showed his theory of collapse; at just before 58 minutes in the video on the http://ine.uaf.edu/projects/wtc7/ site which is called:
A Structural Reevaluation of the Collapse of World Trade Center 7 presentation by Dr. Leroy Hulsey on September 3, 2019 at the University of Alaska Fairbanks.
Content from External Source
Somehow I can't link it here; however the same part can be found on youtube :


(He claims it is the global collapse of floor 13, where as he later has to admit the east penthouse wasn't collapsed because of this, but that has been pointed out many times before.)

What I did see in this simulation/view is something different as what can be seen in the same collapse of the separate videos like this one:


If you look at the same collapse during the presentation of Hulsey you see the penthouse is falling much further down as in the other videos, almost as far as half way down the building around floor 24:
WTC_EP.png
 
Yes, exactly. Certainly the goal ought not be to compare every hook and ornament against shop drawings - we are not Truthers! :p

The kind of thing I would be looking for is confirmation that the models are SET to behave in ways that have no correspondence to the physics of real steel frame structures. Such as not checking whether beams collide and thus allow them to pass through each other. Or having connections that never break and rather can be stretched dozens of meters, because they are not modeled with a condition or threshold above which they break.

Unlikely that any BEAM would become completely free to collapse and drop straight down... Columns would likewise need to be translated laterally. The load would cause the columns to cripple/buckle because the cross sectional area required to support the load was so reduced. It appears that most of the more or less intact steel (columns) broke apart AFTER bracing was stripped off with floor collapse.

Failure columns means the loads they carried--- to be transmitted axially to the column below had to be redistributed... and this can over load the structures that carried the new increased load... causing them to sag, buckle and fail and that is how the structural failure rapidly propagates through the structure. It's complex and likely very hard to model accurately.
 
Unlikely that any BEAM would become completely free to collapse and drop straight down... Columns would likewise need to be translated laterally. The load would cause the columns to cripple/buckle because the cross sectional area required to support the load was so reduced. It appears that most of the more or less intact steel (columns) broke apart AFTER bracing was stripped off with floor collapse.

Failure columns means the loads they carried--- to be transmitted axially to the column below had to be redistributed... and this can over load the structures that carried the new increased load... causing them to sag, buckle and fail and that is how the structural failure rapidly propagates through the structure. It's complex and likely very hard to model accurately.
This does not address the post you quoted at all.

I was talking about how the simulation software used by Hulsey and team represents the various types of structural elements as they intersect. Not what happens in reality. Specifically, does the software at all consider that materials will tear or break when overloaded? Does it only determine material response at the connections and perhaps treat the length of beam between connections merely as mathematical lines in space - such that beams would pass through beams as, for example, two floor assemblies run into one another?
 
Mick,

Excellent analysis on the blueprints.

Without the data(!) and a mirror setup of his model, we can't test if Hulsey has deliberately left out/added/misplaced the structural elements in question in order to get a desired result.

What we can do, is conclude that there is a discrepancy between the blueprints and the model Hulsey used:

- Some elements are missing.
- Some elements are added.
- Some elements are misplaced.

The last is interesting: Hulsey makes a big point out of Column 79 on page 80:

"According to NIST, the initiating failure occurred at the connection between Column 79 and girder A2001 at Floor 13 due to a relative displacement of girder A2001 of 6.25 inches. Our ABAQUS analysis corroborated the findings of our SAP2000 analysis, showing that girder A2001 would undergo less than one inch of westward displacement relative to Column 79 under the worst-case fire condition. Examining all possible conditions, our analysis found that the westward movement of girder A2001 relative to the base plate on Column 79 was 3.68” for noncomposite, 2.18” for partially composite and less than 1” for composite behavior (the condition at time of the fire)."

But if the green in the S-8 drawing (your analysis) is misplaced by 11", what effect would that have?

These questions should definitely be put to Hulsey:

Why is there a discrepancy between the model and the blueprints?

What effects do the missing/added elements have on the simulated scenarios?

Mick, could you check Colum 79 and see if that is placed correctly?

/Claus
 
I think it could well be important to check and understand how the original software - the versions of ABAQUS and SAP2000 Hulsey used - interprets the data. There may be switches, toggles, parameters embedded in the construction data, or additional to it, that control what the software does or doesn't do.


IT consultant with +30 years of experience here.

There will be a whole system behind the simulations, and that will be the litmus test for whether Hulsey is honest or not.

The input data alone is worthless. Hulsey needs to publish everything, ready for installing on SAP2000 and ABAQUS.

An analogy:

1) I write a VBA program that analyses Excel data. The program relies on a database of constants and variables as well. The analysis proves Earth is flat.
2) I publish the Excel data only.

What are you going to do with the Excel data alone? It tells you nothing of how I got to my conclusion.

/Claus
 
I was chatting with @gerrycan on Twitter, and a few things came up, specifically for looking at the UAF presentation given by Hulsey, and most about the global collapse simulations.

Firstly, Hulsey never seems to mention "dynamic analysis" when showing the collapses. In one slide at 01:01:23 he says:
Metabunk 2019-09-23 15-14-10.jpg
Hulsey: The progressive collapse was simulated with the help of static analysis, by progressively taking away those columns that failed. So when you take out some columns out, the load is still there, it's got to be taken by something else, right? And so we progressively took them out, and took them out, and took them out, to get a handle on what it looked like.
Content from External Source
He then shows the odd fake-looking rotation which is in the report as figure 4.16 (except it's Penthouse44 in the presentation, Penthouse46 in the report)


Source: https://giphy.com/gifs/VCn5AOzBAr2kgkkTcY/html5


Not this is backwards to the order things are presented in the report, where he discusses static analysis first, and then the above animation, which is described as "dynamic analysis results"

Hulsey then goes on to discuss the static analysis. He says there's two colors, red and green. Red indicates failed, green indicates not failed. But of course initially none of the columns exceeded their load capacity, should all be green. In fact, Hulsely seems to have misunderstood what is the green means. Green is used to indicate columns that have not failed, but were manually removed to initiate or progress the collapse. The grammar and spelling of this slide indicate it was written by one of his students (both were Chinese, not native English speakers).
Metabunk 2019-09-23 15-28-47.jpg

The three slides later this is made very clear:
Metabunk 2019-09-23 15-30-25.jpg


"would fail just to keep the progressive collapse proceeding" - So this seems to indicate that they did a linear static analysis step, and the building settled into a steady state, so they selectively removed more columns to make the collapse proceed. They continue doing this several times. The last such slide they show is at 1:04:05
Metabunk 2019-09-23 15-34-47.jpg

This matches 4.12 in the report. But in the report there's a final image 4.13
Metabunk 2019-09-23 15-36-35.jpg

This shows everything failing except for two columns which are in tension (obviously they would immediately fail after this step)

He then shows a different version of the static analysis visualization, on two slides.
Metabunk 2019-09-23 15-38-11.jpg

Metabunk 2019-09-23 15-38-48.jpg

Why is this purple, and not multi-colored, you ask? It's because of those weird appendages at the bottom. Those are significant errors in the analysis. Members that have been displaced so far away that the color code used for displacement has everything close to 0 displacement, even though the roof has moved over 100 feet. The color scale automatically adjusts to the range of things in the model.
Metabunk 2019-09-23 15-41-08.jpg

Here the top is about 25E+12, i.e. 25,000,000,000,000 inches. So something has gone wrong with model, causing bits of it to be displaced thousands of miles below ground, and stretched out across that distance. Who knows if that has any effect on the physics. But it indicates a very glitchy model - and that's on top of it being hand tweaked with columns selectively removed.
 

Attachments

  • Metabunk 2019-09-23 15-26-13.jpg
    Metabunk 2019-09-23 15-26-13.jpg
    192 KB · Views: 506
There's also this image he shows for a flash. (not in the report)
Metabunk 2019-09-23 15-50-42.jpg

Metabunk 2019-09-23 15-51-27.jpg

Presumably, floor 13 at some point in the static analysis. Notice the fractured floor around C79 and C80
Metabunk 2019-09-23 15-53-28.jpg

This appears to sort-of match an earlier slide, also not in the report.
Metabunk 2019-09-23 15-54-02.jpg
Notice nothing breaks though.
 
Revisiting the static model visualization in the report, they are quite high resolution.
Metabunk 2019-09-23 16-02-53.jpg
That's a portion of 4.15, showing which columns on the left side are connected (green). I had mistakenly though these were at the back, but it seems they are at the side.

This does not match anything analysis result. The closest is figure 4.12:
Metabunk 2019-09-23 16-06-23.jpg

However the corner column is still there, and only the next two columns down are in tension, whereas five columns are in tension in the 4.15 visualization. So this seems to be a visualization of something else.

I suspect the figures 4.8 to 4.13 came from the simulation that produced the "purple" visualization. Metabunk 2019-09-23 16-10-19.jpg
Since this is obviously glitchy and inaccurate, another version was done, patching the glitches, and that is what was used in the report.

It is quite different. With three less floors (the bottom three) and a different set of smaller glitches.
Metabunk 2019-09-23 16-17-58.jpg
 
Here's a close-up of the newer glitches.
Metabunk 2019-09-23 16-20-20.jpg
Of additional interest there's a local origin (the cyan with X,Y,Z arrows) - a point about which things can be rotated or scaled. Rather an odd place for one. However, it might simply indicate they had accidentally selected an element before taking the screenshot.
 
I think the glitches are a result of how they have modelled the connections; They simply can't break or snap so every column they don't remove will simply be stretched into eternity.
There would be a lot more if that was the case. We see similar things in game development.

Metabunk 2019-09-24 05-36-36.jpg

There's a variety of quite different reasons why this would happen: code bugs, data bugs, engine bugs. It's a bug though.

At that point in the simulation, neither the purple or the (slightly neater) rainbow model are accurate representations of the actual physics of the building. Not that they ever really were, and it's not just the large distortions - things like the side columns stretching has no correlation with reality.
 
I was chatting with @gerrycan on Twitter, ...
Did he contribute any insight?

...
Firstly, Hulsey never seems to mention "dynamic analysis" when showing the collapses. In one slide at 01:01:23 he says:
Metabunk 2019-09-23 15-14-10.jpg
Hulsey: The progressive collapse was simulated with the help of static analysis, by progressively taking away those columns that failed. So when you take out some columns out, the load is still there, it's got to be taken by something else, right? And so we progressively took them out, and took them out, and took them out, to get a handle on what it looked like.
Content from External Source
I assume this means they never considered that, during load redistribution, any additionally load on any member will at first present a dynamic load that's (at least) double the static load, right?

...
Hulsey then goes on to discuss the static analysis. He says there's two colors, red and green. Red indicates failed, green indicates not failed. But of course initially none of the columns exceeded their load capacity, should all be green. In fact, Hulsely seems to have misunderstood what is the green means. Green is used to indicate columns that have not failed, but were manually removed to initiate or progress the collapse. The grammar and spelling of this slide indicate it was written by one of his students (both were Chinese, not native English speakers).
...
"would fail just to keep the progressive collapse proceeding" - So this seems to indicate that they did a linear static analysis step, and the building settled into a steady state, so they selectively removed more columns to make the collapse proceed. They continue doing this several times.
Frankly, I had all along understood the colors of the circles this way: Red means "fails structurally - a result of analysis", green means "did not fail, we select this for manual take-out in the next iteration"

...
Why is this purple, and not multi-colored, you ask? It's because of those weird appendages at the bottom. Those are significant errors in the analysis. Members that have been displaced so far away that the color code used for displacement has everything close to 0 displacement, even though the roof has moved over 100 feet. The color scale automatically adjusts to the range of things in the model.
...
Here the top is about 25E+12, i.e. 25,000,000,000,000 inches. So something has gone wrong with model, causing bits of it to be displaced thousands of miles below ground, and stretched out across that distance. Who knows if that has any effect on the physics. But it indicates a very glitchy model - and that's on top of it being hand tweaked with columns selectively removed.
Holey shmoley! :D
It's been obvious immediately that these simulations are glitchy, but that that they are THAT glitchy...
And to think that these obvious glitches are never discussed!
I understand it is not unusual for FEA to produce wildly divergent results when pushed beyond numerical limits that are difficult to estimate ahead of time. Such that any FEA result should receive a thorough reality-check. Obviously, Hulsey did not do this at all.
 
Status
Not open for further replies.
Back
Top