NIST's Rationale For Not Releasing Simulation Data

TOTAL COP OUT. It is easier to get it RIGHT than it is to get it WRONG. You are struggling here Mick, ridiculous.

I've worked on 3D modelling projects similar in scale. The idea of being able to model hundreds of thousands of elements with 100% accuracy is a bit optimistic. There are always errors where humans are involved.
 
I've worked on 3D modelling projects similar in scale. The idea of being able to model hundreds of thousands of elements with 100% accuracy is a bit optimistic. There are always errors where humans are involved.
So let's suppose we are going to design a building. The building has hundreds of thousands of elements. Are you seriously saying that it would be optimistic to expect a team of highly qualified engineers to get this right on a CAD based program?????
 
So let's suppose we are going to design a building. The building has hundreds of thousands of elements. Are you seriously saying that it would be optimistic to expect a team of highly qualified engineers to get this right on a CAD based program?????

They were not designing a building. It's a different process. They are essentially taking blueprints, and converting it to digital form manually. I would expect a pretty high accuracy rate, over 99.9%, but not 100%.
 
I cannot see any justification for withholding the data. WTC 7 was unique, (as debunkers often remind us), as are most high rises. Ergo, the data is only pertinent to 7 in terms of security and as 7 no longer exists in its previous form, the data can be of no value to any potential terrorist. Also it appears there is no need for terrorists to have these details as they are evidentially capable of accomplishing the destruction without them, (or they were able to get the details beforehand).

On the other hand, it would be of great benefit for designers and engineers to have the data so that they can better design buildings to withstand such situations.
 
I cannot see any justification for withholding the data. WTC 7 was unique, (as debunkers often remind us), as are most high rises. Ergo, the data is only pertinent to 7 in terms of security and as 7 no longer exists in its previous form, the data can be of no value to any potential terrorist. Also it appears there is no need for terrorists to have these details as they are evidentially capable of accomplishing the destruction without them, (or they were able to get the details beforehand).

On the other hand, it would be of great benefit for designers and engineers to have the data so that they can better design buildings to withstand such situations.

You basically just restated the OP. Goto #1 and read the discussion from there.
 
They were not designing a building. It's a different process. They are essentially taking blueprints, and converting it to digital form manually. I would expect a pretty high accuracy rate, over 99.9%, but not 100%.
Yes, they weren't designing a building from scratch. They had the structural drawings to work from. And you think that makes it more difficult?
 
Yes, they weren't designing a building from scratch. They had the structural drawings to work from. And you think that makes it more difficult?

It's a different type of thing, creating vs. transcribing. An analogy might be writing a novel vs. copying a novel. When you are writing a novel there's a lot of room for variance, things work in all kinds of different ways. You just need to avoid spelling errors etc. But in copying a novel, you've got to be 100% on every single word.

It's just data entry. People make mistakes in data entry.
 
It's a different type of thing, creating vs. transcribing. An analogy might be writing a novel vs. copying a novel. When you are writing a novel there's a lot of room for variance, things work in all kinds of different ways. You just need to avoid spelling errors etc. But in copying a novel, you've got to be 100% on every single word.

It's just data entry. People make mistakes in data entry.
Yes, people make mistakes in data entry. It just so happens that every mistake these people made re WTC7 favoured their theory. Get real.
 
Yes, people make mistakes in data entry. It just so happens that every mistake these people made re WTC7 favoured their theory. Get real.

But the addition of a stiffener under the C79, which you pointed out as an error, would not favor their theory, as it would make it harder for that seat to fail.
 
But the addition of a stiffener under the C79, which you pointed out as an error, would not favor their theory, as it would make it harder for that seat to fail.
You still haven't looked at the drawings Mick. The stiffener plates that were omitted were on the girder, not below it. The were attached to the web and bottom flange. The connection that NIST told us was there is entirely different to what actually was. Why do you insist on arguing with what is on the structural drawings for the building?
 
You still haven't looked at the drawings Mick. The stiffener plates that were omitted were on the girder, not below it. The were attached to the web and bottom flange. The connection that NIST told us was there is entirely different to what actually was. Why do you insist on arguing with what is on the structural drawings for the building?

I was talking about the extra stiffener in the seat you mentioned in the other thread.
 
I was talking about the extra stiffener in the seat you mentioned in the other thread.
The stiffener shown under the girder seat at column 79 in figure 12-25 of the NIST WTC 7 report is not an extra stiffener. It is the type of seat support used on column 44, and others like column 81 shown in figure 12-26. It is an error because it is not the actual support plate under the girder seat at the column 79 side of girder A2001, which was a 2" thick x 18.88" wide x 14" high plate that was welded to the side plates of column 79.

Of course, the reality that there were girder stiffeners at the column 79 side of girder A2001 is a different and even more significant point, as they would require the girder to be displaced about 10 inches to the west before a flange failure could occur and cause the girder to fall off its seat. This is clearly impossible as the maximum expansion possible for the beams framing into the girder from the east is 4.75".

I have attached the two pertinent drawings which show the actual girder seat support plate at the column 79 side of girder A2001 and the flange to web stiffeners at that end of the girder. If you haven't yet you should look at them Mick, as you seem confused.
 

Attachments

  • 9114.pdf
    1.1 MB · Views: 495
  • 1091.pdf
    2.1 MB · Views: 514
Last edited:
I gave it as an example of an error that that did not make thing "better" for NIST's case. Here the change in type of seat support would not have made the seat more likely to fail. It's a minor point.

You will also note that the column itself is modelled as a simple box, which removes a vast range of possibilities. The entire global model is a simplification, as it's only intended to simulate collapse, not the collapse initiation. In this model the disputed connections had already failed, so the details are not incredibly relevant.
 
In this model the disputed connections had already failed, so the details are not incredibly relevant.

And there lies your problem.

As you say, the cartoon was designed and programmed after assuming that column 79 had buckled. And that it had buckled due to lack of lateral support over many floors. And that had been caused by a cascade of floors which itself had been initiated by a girder on floor 13 at column 79 dropping from its seat.

Saying that the details 'are not incredibly relevent' has to be the biggest understatement ever. They are so irelevent that they are meaningless. They relate to a fiction. NIST has spent millions of tax dollars designing and building a simulation that was wrong from the very start.

Now - consider what NIST would have done if they had modelled the elements at that girder connection correctly and discovered that it couldn't fail ? Would they have made different assumptions based on a less likely collapse sequence ? And then programmed it using a different initiation point ? Or would they have concluded that if their most likely collapse sequence was proved wrong then their less likely one would be forced to prove wrong also.

They would then have been forced to admit to themselves (and their paymasters), that others running their program data would be able to discover that the simulation had been deliberately designed to engineer a global collapse after a false assumption had been made from the start point.

That could not be allowed to happen.

So the subterfuge of 'jeapardise public safety' was used to withhold the data from peer scrutiny.
 
I gave it as an example of an error that that did not make thing "better" for NIST's case. Here the change in type of seat support would not have made the seat more likely to fail. It's a minor point.

You will also note that the column itself is modelled as a simple box, which removes a vast range of possibilities. The entire global model is a simplification, as it's only intended to simulate collapse, not the collapse initiation. In this model the disputed connections had already failed, so the details are not incredibly relevant.

There was a single floor partial model of the northeast corner that should have been much more detailed. It would have been discerned there that the actual connection configuration (with girder stiffeners) would not allow the girder to be pushed off the seat.

So gross simplification, due to the need to reduce the number of elements for computational efficiency in a global model, is not an excuse for the erroneous claims of the report.

It seems the flange to web stiffeners at the column 79 end of the A2001 girder were omitted as they would have prevented the alleged girder failure.
 
Last edited:
Saying that the details 'are not incredibly relevent' has to be the biggest understatement ever. They are so irelevent that they are meaningless. They relate to a fiction. NIST has spent millions of tax dollars designing and building a simulation that was wrong from the very start.

I think you misunderstand which model we are referring to here - the LS-DYNA 47 story model. Here the damage was applied from the ANSYS model, so the connection had already failed, so, as I said, the details on that particular connection were not important in that model. Of course they were important in preceding models.
 
The details of the connection that NIST says was the beginning of the global collapse should have been the most researched, and accurate part of the entire investigation. Anyone who understands that should then see clearly that they must have deliberately misrepresented that connection. Being sidetracked by the smoke n mirrors of a simulation does not take the focus from that simple fact.

Its clear that they had to discover a means of explaining why the support to the penthouse gave way because that was seen to be the first sign of movement. They then assumed that it was simply fire. Having limited their options in that way they had to discover a way that 79 could buckle. That led to the need for all lateral support over many floors to take place, which itself led to a desire for many floors to cascade. Such a cascade had to start where fire had been observed. And then a start point had to be identified.

Having taken that route they then had to programme all the ANSYS and LS-DYNA to show the desired result.

I have just described drylabbing.

And that is the rationale for NIST not releasing simulation data. It would prove devastating to their credibility.
 
There was a single floor partial model of the northeast corner that should have been much more detailed. It would have been discerned there that the actual connection configuration (with girder stiffeners) would not allow the girder to be pushed off the seat.

So gross simplification, due to the need to reduce the number of elements for computational efficiency in a global model, is not an excuse for the erroneous claims of the report.

Nor is it an excuse they use. It is the reason that the global collapse model had simplified geometry.

The model you mention is the initial partial LS-DYNA model that was used to establish failure modes and criteria for the ANSYS model. As you know this model WAS a lot more detailed.

upload_2013-10-20_6-29-40.png

It omitted the web/flange stiffener plate, however in this simulation failure of the flange was not actually observed as the result of the simulation was the girder being "rocked" off its seat. So in this instance the presence of that stiffener would not have seemed to have made a difference.

And based on this rendering, they correctly modelled the seat at 12" in this model.
 
Nor is it an excuse they use. It is the reason that the global collapse model had simplified geometry.

The model you mention is the initial partial LS-DYNA model that was used to establish failure modes and criteria for the ANSYS model. As you know this model WAS a lot more detailed.

upload_2013-10-20_6-29-40.png

It omitted the web/flange stiffener plate, however in this simulation failure of the flange was not actually observed as the result of the simulation was the girder being "rocked" off its seat. So in this instance the presence of that stiffener would not have seemed to have made a difference.

And based on this rendering, they correctly modelled the seat at 12" in this model.


Unfortunately, the LS-DYNA model you discuss, that says the girder was rocked off, has omissions of its own that would have prevented the rock-off. Take a look at the attached drawing E12,13 and Figure 8-27 in NCSTAR 1-9 and see if you can tell what was left out that would have prevented the north most beam (G3005) from buckling and the rock-off from ever occurring.
 

Attachments

  • E12,13.pdf
    1.5 MB · Views: 541
And would you agree that in this model the width of the seat is 12", assuming the girder flange is 11.5"?
 
This isn't school Tony. Just tell me what your theory is.
It is the S3007, G3007, and K3007 lateral supporting beams from the exterior that drastically reduce the slenderness of G3005. They cut the unsupported length down to 25% of a beam without them, and that would increase the buckling force needed by 16 times, since it is a function of unsupported length squared in the denominator.

The G3005 beam does not buckle when those lateral supporting beams are included. Why are they omitted in the analysis?
 
Last edited:
And would you agree that in this model the width of the seat is 12", assuming the girder flange is 11.5"?

This is a moot point as it doesn't matter if the seat is 11 or 12" wide since the girder stiffeners would have required at least 8 inches of lateral travel to the west and the beam expansion from the east would have been a maximum of 4.75".
 
It is the S3007, G3007, and K3007 lateral supporting beams from the exterior that drastically reduce the slenderness of G3005. They cut the unsupported length down to 25% of a beam without them, and that would increase the buckling force needed by 16 times, since it is a function of unsupported length squared in the denominator.

The G3005 beam does not buckle when those lateral supporting beams are included. Why are they omitted in the analysis?

Doesn't seem like a huge impediment to me, you've still got A/B/C/K3004 affecting about 90% of the length of A2001. And you are not account for any failure in the lateral supporting beams or their connections. But I agree the model would have been more accurate with them.
 
This is a moot point as it doesn't matter if the seat is 11 or 12" wide since the girder stiffeners would have required at least 8 inches of lateral travel to the west and the beam expansion from the east would have been a maximum of 4.75".

A moot point? Then why does gerry keep going on about it?

And I'd quibble that 4.75. I think you get more like 5.5 if you actually integrate the instantaneous CTE. But perhaps a topic for another thread.
 
A moot point? Then why does gerry keep going on about it?

And I'd quibble that 4.75. I think you get more like 5.5 if you actually integrate the instantaneous CTE. But perhaps a topic for another thread.
Quibble away. Show your working though. The reason that I "keep going on about it" is that NIST got the dimension wrong initially. Mick, you did the calculation yourself. The girder does not travel near 5.5", but it would be interesting to see you dispute this in a meaningful way. ie Show your figures.
 
Quibble away. Show your working though. The reason that I "keep going on about it" is that NIST got the dimension wrong initially. Mick, you did the calculation yourself. The girder does not travel near 5.5", but it would be interesting to see you dispute this in a meaningful way. ie Show your figures.
A moot point? Then why does gerry keep going on about it?

And I'd quibble that 4.75. I think you get more like 5.5 if you actually integrate the instantaneous CTE. But perhaps a topic for another thread.
I did integrate it and used the actual CTE at every degree and allowed the beam to grow at a differential amount at every degree and result is 4.75". Do you want the spread sheet?
 
Doesn't seem like a huge impediment to me, you've still got A/B/C/K3004 affecting about 90% of the length of A2001. And you are not account for any failure in the lateral supporting beams or their connections. But I agree the model would have been more accurate with them.
What you might not realize is that when G3005 is stiffened with the lateral support beams from the exterior its axial stiffness is much greater than the lateral stiffness of the girder between column 44 and A3004 and it deflects the girder a significant amount to the west as a result of its expansion. This puts A3004 in tension. Beams can't buckle when they are in tension. There is nothing putting B3004, C3004, and K3004 in compression either.

So it looks like another omission was made which worked in favor of the story told by politicians. Do you see a pattern?
 
Back
Top