Critical Errors and Omissions in WTC7 Report Uncovered

Status
Not open for further replies.
So the bit that they chose not to model in finer detail is the bit that they say is the most important to their explanation. That makes less sense than a chocolate firegaurd Mick.

They chose not to model everything in finer detail. It was a global model optimization. Because it was the global collapse model. You know that at the point the collapse started the individual connections in question (lots) had already been marked as failed, so modelling in finer detail would have done nothing at all.
 
They chose not to model everything in finer detail. It was a global model optimization. Because it was the global collapse model. You know that at the point the collapse started the individual connections in question (lots) had already been marked as failed, so modelling in finer detail would have done nothing at all.
Nonsense. I think you will find that they are also talking about the ANSYS model which was partial. ANd the connections marked as failed - who decided they had failed, the computer simulation that did not model them accurately? You need to address the whole issue, not take unrelated and inaccurate detail and run with it.
 
Nonsense. I think you will find that they are also talking about the ANSYS model which was partial.

The ANSYS model was an entirely different type of thing, and used a combination of standard and custom break elements for the connections.

Analytical Model for Seated Connection at Columns 79 and 81

The analytical model for the seated connection at Column 79 is shown in Figure 11–15. The same model
was used for Column 81, but the girder was connected to the column at a different angle. Also note the
difference in the location of the fillet weldit was at the top of the top clip angle for Column
79relative to the fillet weld along the sides of the angle in the STC connections at the exterior columns.

The top clip angle used in STC connection at Column 79 and Column 81 was weak in tension and needed
to be explicitly represented in the connection model. Therefore, the failure of the connection was
governed by tension failure of the top clip angle or bolt shear followed by beam walking-off the seat and
loss of vertical support. The analytical model accounted for: 1) shear failure of individual bolts, 2) beam
walking off of the seat, and 3) horizontal weld failure of the top flange clip angle. The failure of the
connection was governed by bolt shear followed by walk-off before loss in vertical support.

At each bolt location one break element, one elastic beam, and one non-linear spring were defined. The
bolt shear strength was modeled with break element type USER103 with K1 = 1, i.e., failure occurred
when the SRSS of the horizontal force components in the x- and y-directions exceeded the temperature dependent
shear capacity of the bolt. The elastic beam represented the stiffness of the bolt connection.

The non-linear spring was used to capture slip in the connection before bolt bearing against the hole edge.
Slotted bolt holes were used at the top flange clip angle while standard bolt holes were used at the bottom
seat plate.

A node-to-node contact element with an initial gap representing the clearance between the end of the
beam and the column was used to model the contact between the beam and the column when the bolts
sheared off and the beam moved enough to touch the surface of the column. The bolts could fail in shear
under a tensile, compressive or lateral force, but the beam would only walk off the seat under a tensile or
lateral force. The failure of the bolts and weld was a prerequisite for the beam end walking off the seat.

The travel distance for walk off was 6.25 in. along the axis of the beam and 5.5 in. lateral to the beam. A
walk off failure occurred when the bolts sheared at the seat connection, the bolts sheared or the weld
failed at the top clip angle, and the beam walked off the seat. A control element was used to model beam
Walk-off in the axial direction. Beam walk off in the lateral direction was monitored during the analysis.
Content from External Source


I presume you here would argue that the COMBIN37 control element had incorrect parameters in one dimension?
 
Last edited:
I presume you here would argue that the COMBIN37 control element had incorrect parameters in one dimension?
Why would I argue that? Neither you or I have any input data for this element, so it is meaningless. What we do have data for is the elements around this connection and the structural drawings do not match NISTs illustrations and analysis. Note the second sentance in the following from NCSTAR 1-9, maybe you missed it first time round, you say
The ANSYS model was an entirely different type of thing
You need to elaborate on this. They are similar to the extant that they both fail to model the connection correctly.
target elements.jpg
 
The global model had much simpler 3D geometry and connections, because there were a lot more of them, and it was a different type of simulation (involving collisions from falling debris), so more computationally complex.

What exactly do you think the the "finer detail" would have done, in both models?
 
The global model had much simpler 3D geometry and connections, because there were a lot more of them, and it was a different type of simulation (involving collisions from falling debris), so more computationally complex.

What exactly do you think the the "finer detail" would have done, in both models?
Included the stiffener plates and prevented the walk off that NIST allege happened. You do not even know what the detail used was, so your question is meaningless really, has no frame of reference and is a logical fallacy.
 
The crux is that if the correct elements had been included and then modelled correctly in all the analysis methods utilised then they would have concluded that the girder from 79 to 44 was not the initiating event.

That is really all that this thread is about.

And as gerry uncovered critical errors and omissions that are indeed critical if they would have such a fundamental effect then his input remains debunked.
 
Included the stiffener plates and prevented the walk off that NIST allege happened. You do not even know what the detail used was, so your question is meaningless really, has no frame of reference and is a logical fallacy.

But the stiffener plates would affect the break elements in the ANSYS (the COMBIN37), that's independent of the level of detail in the geometry.

The geometry LOD only affects the walk off determination in the very first LSDYNA model.

(By "geometry" here I mean the 3D polygonal stuff, the shell elements, apologies for any terminology confusion, comes from my game development days).
 
But the stiffener plates would affect the break elements in the ANSYS (the COMBIN37), that's independent of the level of detail in the geometry.

The geometry LOD only affects the walk off determination in the very first LSDYNA model.

(By "geometry" here I mean the 3D polygonal stuff, the shell elements, apologies for any terminology confusion, comes from my game development days).
Mick, please answer this very simple question. How much of the input data for the NIST simulations have you seen?

edit - how long does it take to type "none" ?
 
Mick, please answer this very simple question. How much of the input data for the NIST simulations have you seen?

None (except a brief scan over the LSDYNA model without connections).

But what I'm discussing here is your claim that the LOD is significant. I think you think the connection failures in the larger models actually came from buckling of the geometry (the flange), when they really came from the break elements. Hence the LOD issue you bring up is irrelevant, and you should retract it.
 
None (except a brief scan over the LSDYNA model without connections).

But what I'm discussing here is your claim that the LOD is significant. I think you think the connection failures in the larger models actually came from buckling of the geometry (the flange), when they really came from the break elements. Hence the LOD issue you bring up is irrelevant, and you should retract it.

What you are trying to do is refute clear and transparent data that has been presented to you with models for which you have no data. Also, I do not believe that you have studied the simulation outputs enough, or you would not make such statements. Also, you are still avoiding the howler you came out with earlier, about 'one direction'. I think it would be very enlightening if you explained this statement now. I suspect you know this, and will not respond specifically. As for retractions, you should retract your blind support of an obvious deception on NISTs part and just admit that they got it hopelessly wrong. The fact that you do not have one shred of input data to back up the models that you are trying to use as evidence is deceptive. These models will become relevant if you get a hold of input data for them, which you won't, so they're not. Debunked and in denial.
 
Also, you are still avoiding the howler you came out with earlier, about 'one direction'. I think it would be very enlightening if you explained this statement now. I suspect you know this, and will not respond specifically.

"One direction" referred to "west". To be more accurate it should have been "east/west", or "one dimension/axis" (as both flanges had a stiffener plate). But seeing as we were not really talking about walk off to the east, then I just said one direction.

Now, about this level of detail? How would it have made any difference? Isn't it just the break elements that would change?
 
"One direction" referred to "west". To be more accurate it should have been "east/west", or "one dimension/axis" (as both flanges had a stiffener plate). But seeing as we were not really talking about walk off to the east, then I just said one direction.

Now, about this level of detail? How would it have made any difference? Isn't it just the break elements that would change?
Hang on, didn't you actually state that i had maybe debunked the walk off in one direction? Here's what you actually said
You have possibly debunked the failure criteria for one connection in one direction for one mode of failure. That's really it
What other modes of failure do NIST propose for this connection, and what directions is that failure supposed to happen in. I do not think that what you just said in #652 relates to the above quote.
As for levels of detail, you have no data whatsoever for this model, so you need to drop that one until you do. This is not a high school debate after all. Your assertions should be backed up with data, unlike NISTs models.
 
Yes I did say you had maybe debunked the walk off in one direction.

But you keep saying that the level of detail is a problem, and you have been entirely unable to explain why. Can we address that?
 
Yes I did say you had maybe debunked the walk off in one direction.

But you keep saying that the level of detail is a problem, and you have been entirely unable to explain why. Can we address that?
What we do know is that NIST made serious errors in what they wrote. They got the seat plate length wrong by confusing 11 with 12". The missed out the stiffener plates at the connection. They haven't released the input data for their model, even though we have the structural drawings and their estimates of temperature, so can logically work out how the connection would have reacted. They supposed that C81 and 79 were similar, which is totally wrong. They continually contradicted themselves in illustrations showing the connection.
But then on the other hand, there you are, touting their simulation outputs as if they were meaningful, while at the same time admitting that you have no data for them. The level of detail that we do have re this connection is minimal, and what we do know is that it is hopelessly wrong. Why would that instill confidence in their model simulations in anyone? Exactly what motive do you have for defending NISTs story Mick? I just don't understand how someone who is apparently intelligent cannot see their lies laid bare. Give me ALL the detail that you have re this connection (which is basically what you have got from this thread and the related discussion) and I will take you through the problems with the analysis again, but you need to accept first that evidence requires data to back it up, and blind assertion is no substitute for that.
 
What we do know is that NIST made serious errors in what they wrote. They got the seat plate length wrong by confusing 11 with 12". The missed out the stiffener plates at the connection. They haven't released the input data for their model, even though we have the structural drawings and their estimates of temperature, so can logically work out how the connection would have reacted. They supposed that C81 and 79 were similar, which is totally wrong. They continually contradicted themselves in illustrations showing the connection.
But then on the other hand, there you are, touting their simulation outputs as if they were meaningful, while at the same time admitting that you have no data for them. The level of detail that we do have re this connection is minimal, and what we do know is that it is hopelessly wrong. Why would that instill confidence in their model simulations in anyone? Exactly what motive do you have for defending NISTs story Mick? I just don't understand how someone who is apparently intelligent cannot see their lies laid bare. Give me ALL the detail that you have re this connection (which is basically what you have got from this thread and the related discussion) and I will take you through the problems with the analysis again, but you need to accept first that evidence requires data to back it up, and blind assertion is no substitute for that.

I'm not defending NIST's story. I really can't see why they chose to focus on that one walk off.

Can you just explain what difference changing the level of detail would make, given that the failure criteria was in the break elements.
 
And there's absolutely no need to go further.

This guy has avoided answering more than forty questions of mine, on the way to delivering "work" less exacting...

[...]

It's important for everyone to keep in mind that from the mainstream or Official Inc. perspective there isn't only no need to go any further into the technical details of the report, there was no need to have a simulation of an investigation in the first place. Everyone that takes the official story/theory at face value already knew what happened on the day of the attacks or shortly thereafter due to reports in the corporate media, so there was no need to begin any real process of verification/falsification based on specifications that could be "critical."

It's impossible for there to be a critical error in a theory about what happened when people already think they know what happened.
 
Last edited:
I'm not defending NIST's story. I really can't see why they chose to focus on that one walk off.

Can you just explain what difference changing the level of detail would make, given that the failure criteria was in the break elements.
How can I possibly comment on what difference a change in parameter would make when neither of us has a datum point for that parameter. Your question makes no sense. Give me the parameter and the datum point then ask the question.
 
How can I possibly comment on what difference a change in parameter would make when neither of us has a datum point for that parameter. Your question makes no sense. Give me the parameter and the datum point then ask the question.

You were the one criticising the level of detail. So why were you criticising it?

The "parameter and datum point" would simply be (for example) the number of polygons used to model the flange. Here in the first ANSYS model:
upload_2013-10-18_9-19-9.png

The ANSYS seems like it was modeled with just contact and break elements:
upload_2013-10-18_9-22-7.png

And the full LSDYNA was a combination of contact and shell elements.

upload_2013-10-18_9-24-14.png
upload_2013-10-18_9-23-16.png

So what's exactly is the problem you are talking about with the level of detail?
 
So what's exactly is the problem you are talking about with the level of detail?
..............
.........................he says having just posted half a dozen or so illustrations, none of which represent the connection details accurately. I will respond in full in a later post.
 
..............
.........................he says having just posted half a dozen or so illustrations, none of which represent the connection details accurately. I will respond in full in a later post.

We were talking about level of detail. Specifically
SRemember also that NIST stated that they could not accyrately model the connections at the girder because of the target element size of 0.15 - 0.3M. Another fact that you have not acknowledged yet.

Hence I attempted to acknowledge it, and discuss it in depth. So explain to me what the target element size is affecting in the above models.
 
.
Fig 12-22 is of an exterior column, I included it to show the spring elements.

But the connection said to have failed, causing the girder to drop and cause the cascade of floors leading to 79 buckling was shown in Fig 12-25.

And the Fig 12 - 25 Seat connection global model column 79 is totally wrong.
 
.


But the connection said to have failed, causing the girder to drop and cause the cascade of floors leading to 79 buckling was shown in Fig 12-25.

And the Fig 12 - 25 Seat connection global model column 79 is totally wrong.

But does it have anything to do with the target element size? Or is this another topic?
 
But does it have anything to do with the target element size?

It has everything to do with accuracy. The seat in fig 12-25 is not in any way accurate. And if the basis of an analysis is inaccurate then a representation using any target size is bound to compound that error.

That connection in fig 12-25 is totally different to the real drawings.
 
So if the target element size is irrelevant, then why does Gerry keep bringing it up?

Where did I say it was irrelevant ?

What I actually said was this :- " And if the basis of an analysis is inaccurate then a representation using any target size is bound to compound that error."

Target size is crucial to provide accuracy ( As you well know ) But as I said, if the basis is wrong then even using the very best target size available, the end result will be wrong. GIGO.
 
Where did I say it was irrelevant ?

What I actually said was this :- " And if the basis of an analysis is inaccurate then a representation using any target size is bound to compound that error."

Target size is crucial to provide accuracy ( As you well know ) But as I said, if the basis is wrong then even using the very best target size available, the end result will be wrong. GIGO.

So you are saying the error itself has nothing to do with the element target size? Just the output.

Wouldn't a smaller target element size compound the error more, as it is more accurate?

Really though I just want Gerry to explain what he feels is significant about:
SRemember also that NIST stated that they could not accyrately model the connections at the girder because of the target element size of 0.15 - 0.3M. Another fact that you have not acknowledged yet.

I think there is some misunderstanding about what the three different models were used for.
 
So you are saying the error itself has nothing to do with the element target size? Just the output.

You love to re-phrase peoples input to put words in their mouth don't you.

I will try again. This time please read what I put and refrain from going into your paraphrase mode.

" if the basis is wrong then even using the very best target size available, the end result will be wrong. GIGO."

As Fig 12-25 bears no resemblance to the real connection, and therefore the base is wrong - then the end result will be wrong.

And much reliance was then placed on that wrong result.
 
You love to re-phrase peoples input to put words in their mouth don't you.

I will try again. This time please read what I put and refrain from going into your paraphrase mode.

" if the basis is wrong then even using the very best target size available, the end result will be wrong. GIGO."

As Fig 12-25 bears no resemblance to the real connection, and therefore the base is wrong - then the end result will be wrong.

And much reliance was then placed on that wrong result.

Sorry, I'm trying to figure out what you mean. Paraphrasing something as a question is a way of checking if your understanding is right. Hence the question mark at the end.

So how does:
" if the basis is wrong then even using the very best target size available, the end result will be wrong. GIGO."
really different from my paraphrasing:
"So you are saying the error itself has nothing to do with the element target size? Just the output."

Gerry seems to be saying that NIST used the target element size as an incorrect excuse for getting things wrong. But the target element size in the global model is a necessary optimization due to lack of computing power. When they actually did focussed studies on the individual parts to establish failure modes, like in fig 8-23, they used smaller element sizes. And with the ANSYS model, I'm not really sure if element size comes into it at all.
 
For reference, here's the basic scale of the LSDYNA global simulation model, the computing hardware used, and the time taken per run:

To model the damage and sequence of failures in WTC 7 following the initial failure event, failure
models were used within the framework of the LS-DYNA finite element analyses. For the global
collapse analyses of WTC 7, shell element sizes were 6 in. to 12 in. to maintain a model size of
approximately 3 million elements (see Table 12–1). With this resolution, the gradients around a fracture
could not be accurately resolved, and the damage criteria required adjustment to obtain the appropriate
strength and ductility of the structures (see NIST NCSTAR 1-2B). Simulating the global response of
WTC 7 required the same level of refinement throughout the model.

All simulations were run on a high speed Linux Beowulf compute cluster. The head node had two 64 bit
AMD Opteron 250 2.4 GHz processors, 4 GB of RAM, and 1.5 terabytes of RAID 5 disk storage. Each
compute node had two 64 bit AMD Opteron Dual-core 285 2.6 GHz processors. Six of the compute nodes
had 8 GB of RAM and the remaining two nodes had 16 GB RAM.
[total 9 quad core computers]

The global simulation was first initialized under gravity loading over 4.5 s of simulation time. Then
damage and temperature initialization states were applied over 4 additional seconds. Collapse propagation
in the global model required approximately 16 additional seconds after initialization, for a total of about
25 s of simulated time, which generally took up to 8 weeks using 12 cores across three nodes. The
analysis was performed using the double precision version of LS-DYNA for better computational
accuracy.
Content from External Source
 
" if the basis is wrong then even using the very best target size available, the end result will be wrong. GIGO."
really different from my paraphrasing:
"So you are saying the error itself has nothing to do with the element target size? Just the output."

Your own interpretation is completely different to my own input. I am saying that the error itself has EVERYTHING to do with the output. Its the element target sizing thats irelevent. If the input is wrong then it doesnt matter how you process the data - the output will be wrong too.

The girder seat total configuration in 8-23 is just plain wrong. No underseat plate. No stiffeners. The girder is shown central to the column incorrectly. And is set at right angles to the column - incorrectly. Displaying incorrect elements using a finer programme cannot make the elements themselves correct. How can they ? And if input is wrong then output is wrong.
 
Your own interpretation is completely different to my own input. I am saying that the error itself has EVERYTHING to do with the output. Its the element target sizing thats irelevent. If the input is wrong then it doesnt matter how you process the data - the output will be wrong too.

Eh? But that's what I said.

You: "the element target sizing thats irelevent"
Me: "the error itself has nothing to do with the element target size?"

What am I missing?
 
Last edited:
To model the damage and sequence of failures in WTC 7 following the initial failure event, failure models were used within the framework of the LS-DYNA finite element analyses.

I have deleted the bulk of that quote because it it is meaningless if the initiation is wrong. If you programme all that data based on an incorrect initiation event then you have just wasted much taxpayers money.
 
I have deleted the bulk of that quote because it it is meaningless if the initiation is wrong. If you programme all that data based on an incorrect initiation event then you have just wasted much taxpayers money.

But very clearly it was not modeled based on that one walk-off event. The initialization of the LSDYNA model was the application of damage from the ANSYS model, which had multiple independent failures, not just one.

And in fact that girder does not even see to play a significant part in the simulated collapse. Which would actually seem to be more accurate (based on gerry's objections), not less.
 
But very clearly it was not modeled based on that one walk-off event. The initialization of the LSDYNA model was the application of damage from the ANSYS model, which had multiple independent failures, not just one.

And in fact that girder does not even see to play a significant part in the simulated collapse. Which would actually seem to be more accurate (based on gerry's objections), not less.

lol You are really infatuated by that cartoon arn't you. NIST put forward a sim based on incorrect input. It then bears no resemblance to the words they use to describe the initiation. And then you use that as evidence that their words are wrong. Priceless.
 
lol You are really infatuated by that cartoon arn't you. NIST put forward a sim based on incorrect input. It then bears no resemblance to the words they use to describe the initiation. And then you use that as evidence that their words are wrong. Priceless.

All I'm doing here is seeking clarification on what Gerry's complaint about element size was. We've already discussed the difference between the narrative and the simulation.
 
Status
Not open for further replies.
Back
Top