'What about building 7'?

[pgimeno]

Your debate even prompted me to read Chapter 11 of NCSTAR 1-9. I am currently in 11.4.1, the "Discussion of Results for Case B Temperatures".

That, and 11.3.2, are the main chapters that gerrycan has failed to read, or at least to comprehend, as evidenced by his arguments.

Funnily enough, I'm not arguing that NIST was right or that their analysis is perfect or their scenario is the only possible. I'm only arguing that gerrycan's criticism of NIST is based on a strawman version of what NIST said, not on what they actually said.

 

[JSanderO]

I realize this is the horse before the cart thing... but... assuming the girder walked off...how does that lead to 79 buckling? (not say it couldn't... just asking how that works) And if it does buckle do we assume all the column 79s above came down? And then... how does that "sink hole" lead to the global collapse?

 

[Oystein]

I realize this is the horse before the cart thing... but... assuming the girder walked off...how does that lead to 79 buckling? (not say it couldn't... just asking how that works) And if it does buckle do we assume all the column 79s above came down? And then... how does that "sink hole" lead to the global collapse?

Yes, that's several steps ahead of where we are now.
pgimeno and gerrycan are debating what the individual failures of the floor framing were accorcing to NIST's Case B temperature ANSYS modelling. That is the static part of the simulation; it predicts how beams buckle, conections fail, and horizontal and vertical supports are lost.

One prediction is the walk-off of the girder connecting interior column 79 and exterior column 44. This was only one of several such failures. In the ANSYS model, when a beam or girder had failed, it was removed from further modelling. In reality, a girder that hat lost vertical support would fall down on the floor below.

In NIST's engineering judgement, the 79-44 13th floor walk-off was the final trigger to global collapse: The dynamic LS-DYNA model shows how it's fall triggered floor slabs 12 to 6 (?) to fail and fall in rapid succession.

Already before that, col 79 was laterally unbraced in the E-W direction at floors 10-12 (Fig. 11-49). The failure of the 79-44 girders exposed it over 8 (floor) to the north. I am not up to speed if any more connections to 79 were affected by the floor collapses.

Col 79 was either laterally unsupported over several floors or, where girders were still in place, pushed out of plumb by heated girders.
Either condition significantly reduces capacity. -> Euler buckling.
NIST ran the dynamic situation with LS-DYNA and found this is indeed what would happen, given the input of the Case B of ANSYS. So no need for us to rely on "imagination".

End yes, of course all of Col 79 above floor 13 would come down once it buckles, or how do you imagnine the load would be redistributed and 79 with surrounding floor loads suspended in mid-air? Col 79 bore the largest vertical load of all columns (Fig. 11-45: 7,282 kip; compared to 5767, 5437 and 2926 kip for its three neighbors, 76, 80 and 44 respectively)

 

[Animal]

Standing on the sideline, I enjoy gerrycan's and pgimeno's reasonably civil exchange, as both try to make their points succinctly, and support then with external documentation.

I'd encourage both to continue, ignore cheers and boos for the time being, and work out where your differences really are, so that a layman on the sideline such as myself understands the point.

Your debate even prompted me to read Chapter 11 of NCSTAR 1-9. I am currently in 11.4.1, the "Discussion of Results for Case B Temperatures".

The problem being that GC does not have the technical knowledge to begin the discussion, let alone grasp the intricacies of a structural design nor to try to "make points". TZ at least had enough of a technical background to present his case (though wrongly) GC is simply regurgitating talking pints without understanding what he really is talking about. Evident by not knowing the difference between beam and a girder, not understanding that a non-failed connection does not necessarily prevent movement of a structure....and not having read the NIST report to understand that they not only deleted certain minor elements to simplify the computer processing, they ADDED elements as well for the vary same reason.

 

[Oystein]

The problem being that GC does not have the technical knowledge to begin the discussion, let alone grasp the intricacies of a structural design nor to try to "make points". ... GC is simply regurgitating talking pints without understanding what he really is talking about. Evident by not knowing the difference between beam and a girder, not understanding that a non-failed connection does not necessarily prevent movement of a structure....and not having read the NIST report to understand that they not only deleted certain minor elements to simplify the computer processing, they ADDED elements as well for the vary same reason.

Maybe this is so - then both gerrycan and I have a chance to learn a few things, and stress the "E" in "ISF"

For example, I was never clear myself what the difference is between "girder" and "beam" - my intelligent guess is that, in NIST's diction, the main horizontal elements that two connect columns directly are the girders, while elements connected to at least one other horizontal member instead of a column is refered to as a beam. Reading Wikipedia, girders and beams are defined by the main load (girder: compressive and tensile; beam: bending under gravity load), but as both the girders and beams in WTC7 bear both vertical and horizontal loads, I believe the former "definition" is the "right" one here.

 

[Animal]

Maybe this is so - then both gerrycan and I have a chance to learn a few things, and stress the "E" in "ISF"

For example, I was never clear myself what the difference is between "girder" and "beam" - my intelligent guess is that, in NIST's diction, the main horizontal elements that two connect columns directly are the girders, while elements connected to at least one other horizontal member instead of a column is refered to as a beam. Reading Wikipedia, girders and beams are defined by the main load (girder: compressive and tensile; beam: bending under gravity load), but as both the girders and beams in WTC7 bear both vertical and horizontal loads, I believe the former "definition" is the "right" one here.

Putting it in simple terms...a girder is a beam that supports other structural members....such as other beams or columns. You could have a beam connecting to two columns and it may not be a girder. It is typically used in the industry to denote a structural member that is transferring a load. (A girder roof truss supports other roof trusses)

 

[gerrycan]

Maybe this is so - then both gerrycan and I have a chance to learn a few things, and stress the "E" in "ISF"

For example, I was never clear myself what the difference is between "girder" and "beam" - my intelligent guess is that, in NIST's diction, the main horizontal elements that two connect columns directly are the girders, while elements connected to at least one other horizontal member instead of a column is refered to as a beam. Reading Wikipedia, girders and beams are defined by the main load (girder: compressive and tensile; beam: bending under gravity load), but as both the girders and beams in WTC7 bear both vertical and horizontal loads, I believe the former "definition" is the "right" one here.

Very broadly speaking. Girders are primary and beams are secondary elements. A good way to think about it in the terms of the matter at hand is that to remove a girder would necessitate beam failure whilst to remove beams would not necessitate the failure of a girder. That said, different people have different ideas and the whole beam Vs girder debate will continue to roll on far beyond this topic and forum.
As for where the column 79 debate here, the question is now can the column experience lateral movement yet still retain orthogonal support via th
e 2 undamaged connections at the west and south faces.
I would have thought that the answer to that should be obvious. Apparently it isn't to some.

 

[pgimeno]

As for where the column 79 debate here, the question is now can the column experience lateral movement yet still retain orthogonal support via th
e 2 undamaged connections at the west and south faces.

No, that is not "THE" question.

NIST ran three FEA simulations. The first one is a LSDYNA analysis to check what failure modes should be accounted for. That analysis covered A2001, the five beams framing into it, the associated columns, the shear studs, and the concrete slab and metal deck.

Then a 16-floor ANSYS simulation to determine global failure initiation, which covered a good part of the east half of WTC7, where the temperatures were input from their fire simulator. That's the one where A2001 walked off.

And finally, a full 47 story model in LSDYNA to simulate collapse progression after the initiating event, which is of no interest at this point.

This is how they modelled the connection of A2001 to C79 in ANSYS, taking into account the results of the preliminary LSDYNA analysis:

I've highlighted in yellow the control element they used to model walk-off, named COMBIN37. NIST claims their analysis showed that that element reported walk-off after 4 hours of fire, and then they proceed to interpret the results, attributing the major part of them to beam expansion, yet reporting column displacement to the east at the same time, and enough of it to break bolts in other floors.

[ETA] So as a matter of fact, NIST is claiming that the ANSYS simulation they performed showed that control element COMBIN37 reported a walk-off, by whatever means. Factors that I can think of that could have influenced it include pushing of A2001 to the west by the beams to its east, displacement of the column to the east by the girder to its west, and leveraging on A2001 that caused its end to move more than the element that pushed it. There may be others that I am missing, but that's not the point.

The point is that they say that in the simulation, that element reported walk-off, by whatever means. [/ETA]

Now "THE" question is, what exactly are you questioning? Here are some options that come to mind, but maybe it's none of them so you tell me:

- The results of that simulation. That would implicitly mean blaming the ANSYS authors for making a faulty program that behaves unrealistically.

- The setup of the simulation itself. But I don't think that's the case, because according to NIST's reporting of the status of the connections and to what you're arguing, the column was perfectly held in place preventing it from displacing east, therefore the inputs to ANSYS were good and showed the connections pristine and able to hold the column (let's disregard for now the mobility of girder A2002, which you don't account for).

- NIST's reporting of the results of the simulation. That would mean you're saying NIST lied about the output of ANSYS, and that element COMBIN37 (the one highlighted in the picture) did actually not report walk-off but NIST said it did.

So what is it you're questioning, gerrycan?

 

[Spanx]

Very broadly speaking. Girders are primary and beams are secondary elements. A good way to think about it in the terms of the matter at hand is that to remove a girder would necessitate beam failure whilst to remove beams would not necessitate the failure of a girder. That said, different people have different ideas and the whole beam Vs girder debate will continue to roll on far beyond this topic and forum.
As for where the column 79 debate here, the question is now can the column experience lateral movement yet still retain orthogonal support via th
e 2 undamaged connections at the west and south faces.
I would have thought that the answer to that should be obvious. Apparently it isn't to some.

You could always take this opportunity to explain in detail what did happen.

 

[Oystein]

Putting it in simple terms...a girder is a beam that supports other structural members [...] It is typically used in the industry to denote a structural member that is transferring a load. (A girder roof truss supports other roof trusses)

...A good way to think about it in the terms of the matter at hand is that to remove a girder would necessitate beam failure whilst to remove beams would not necessitate the failure of a girder. ...

Thanks to both of you!

 

[Animal]

Very broadly speaking. Girders are primary and beams are secondary elements.

Just more proof that you don't know what you are talking about.

A good way to think about it in the terms of the matter at hand is that to remove a girder would necessitate beam failure whilst to remove beams would not necessitate the failure of a girder. That said, different people have different ideas and the whole beam Vs girder debate will continue to roll on far beyond this topic and forum.

There is no debate as to girder vs. beam. Only uninformed troofers has issues with it.

As for where the column 79 debate here, the question is now can the column experience lateral movement yet still retain orthogonal support via th
e 2 undamaged connections at the west and south faces.
I would have thought that the answer to that should be obvious. Apparently it isn't to some.

It is apparent troofers has issue with many obvious things.

 

[gerrycan]

Sure. For starters I would question this.....

This is how they modelled the connection of A2001 to C79 in ANSYS, taking into account the results of the preliminary LSDYNA analysis:

[qimg]http://www.formauri.es/personal/pgimeno/xfiles/11-s/NCSTAR1-9fig11-14-crop.png[/qimg]

No it isn't. Wrong figure. There is no top clip angle in that.
Your figure shows seated connections at the North, South and East face columns, and NOT for A2001 framing into the column 79 connection.
Your figure shows how A2001 would have framed into the C44 connection at the other end of it.

This is how NIST modelled the seated connection at C79
http://www.internationalskeptics.com/forums/showthread.php?p=10493153#post10493153

Also, what drawing does it say it is based on? I am pretty sure the drawing is called out wrongly there too.

Basically, so far you got the girder right, you just got the wrong column and the wrong connection type.

 

[Oystein]

gerrycan is right - pgimeno had shown Fig 11-14 in NIST NCSTAR 1-9, which is captioned "Figure 11–14 Analytical model for seated connections at exterior columns". He should have shown "Figure 11–15 Analytical model for the seated connection at Column 79."

However, both contain this element that pgimeno highlighted: "Control element to model walk-off (COMBIN37)"

So how does pgimeno's argument change, or become invalid, by switching to the correct figure?

 

[Animal]

gerrycan is right - pgimeno had shown Fig 11-14 in NIST NCSTAR 1-9, which is captioned "Figure 11–14 Analytical model for seated connections at exterior columns". He should have shown "Figure 11–15 Analytical model for the seated connection at Column 79."

However, both contain this element that pgimeno highlighted: "Control element to model walk-off (COMBIN37)"

So how does pgimeno's argument change, or become invalid, by switching to the correct figure?

To troofers......trivial errors like that are their proof that the only answer is CD.

 

[pgimeno]

Sure. For starters I would question this.....

Sure, I made a mistake and posted figure 11-14, which is not the one corresponding to the seat at column 79, instead of 11-15, which is the right one, but the rest of the message doesn't change. I've deleted the wrong figure.

Can you address my post now? I'll repeat it for you now with the right figure.

__________________________________________________

As for where the column 79 debate here, the question is now can the column experience lateral movement yet still retain orthogonal support via th
e 2 undamaged connections at the west and south faces.

No, that is not "THE" question.

NIST ran three FEA simulations. The first one is a LSDYNA analysis to check what failure modes should be accounted for. That analysis covered A2001, the five beams framing into it, the associated columns, the shear studs, and the concrete slab and metal deck.

Then a 16-floor ANSYS simulation to determine global failure initiation, which covered a good part of the east half of WTC7, where the temperatures were input from their fire simulator. That's the one where A2001 walked off.

And finally, a full 47 story model in LSDYNA to simulate collapse progression after the initiating event, which is of no interest at this point.

This is how they modelled the connection of A2001 to C79 in ANSYS, taking into account the results of the preliminary LSDYNA analysis:

I've highlighted in yellow the control element they used to model walk-off, named COMBIN37. NIST claims their analysis showed that that element reported walk-off after 4 hours of fire, and then they proceed to interpret the results, attributing the major part of them to beam expansion, yet reporting column displacement to the east at the same time, and enough of it to break bolts in other floors.

[ETA] So as a matter of fact, NIST is claiming that the ANSYS simulation they performed showed that control element COMBIN37 reported a walk-off, by whatever means. Factors that I can think of that could have influenced it include pushing of A2001 to the west by the beams to its east, displacement of the column to the east by the girder to its west, and leveraging on A2001 that caused its end to move more than the element that pushed it. There may be others that I am missing, but that's not the point.

The point is that they say that in the simulation, that element reported walk-off, by whatever means. [/ETA]

Now "THE" question is, what exactly are you questioning? Here are some options that come to mind, but maybe it's none of them so you tell me:

- The results of that simulation. That would implicitly mean blaming the ANSYS authors for making a faulty program that behaves unrealistically.

- The setup of the simulation itself. But I don't think that's the case, because according to NIST's reporting of the status of the connections and to what you're arguing, the column was perfectly held in place preventing it from displacing east, therefore the inputs to ANSYS were good and showed the connections pristine and able to hold the column (let's disregard for now the mobility of girder A2002, which you don't account for).

- NIST's reporting of the results of the simulation. That would mean you're saying NIST lied about the output of ANSYS, and that element COMBIN37 (the one highlighted in the picture) did actually not report walk-off but NIST said it did.

So what is it you're questioning, gerrycan?

 

[DGM]

So what is it you're questioning, gerrycan?

I think he answered it here:

Also, what drawing does it say it is based on? I am pretty sure the drawing is called out wrongly there too.

This is the connection as modeled not necessarily the same as plan. It's based on plan but also takes into consideration previous simulations. I'm thinking this is where the confusion lies.

 

[pgimeno]

This is the connection as modeled not necessarily the same as plan. It's based on plan but also takes into consideration previous simulations. I'm thinking this is where the confusion lies.

In any case, he is claiming that it's impossible for the girder to have walked off and therefore that the report is fatally flawed. What I'm asking is what part does he question from NIST's reported results that cause it to be flawed.

I've pointed out that the report says that the column displaced to the east too, not only A2001 to the west, and therefore that his reasoning about the impossibility of the walk-off is flawed, and he questions denies that the column moved, so the best answer I can think of given that he denies it, is that NIST lied when they reported the output of ANSYS. But I'm giving him the opportunity to explain if that's the case.

 

[gerrycan]

Sure, I made a mistake and posted figure 11-14, which is not the one corresponding to the seat at column 79, instead of 11-15, which is the right one, but the rest of the message doesn't change. I've deleted the wrong figure.

Can you address my post now? I'll repeat it for you now with the right figure.

__________________________________________________



No, that is not "THE" question.

NIST ran three FEA simulations. The first one is a LSDYNA analysis to check what failure modes should be accounted for. That analysis covered A2001, the five beams framing into it, the associated columns, the shear studs, and the concrete slab and metal deck.

Then a 16-floor ANSYS simulation to determine global failure initiation, which covered a good part of the east half of WTC7, where the temperatures were input from their fire simulator. That's the one where A2001 walked off.

And finally, a full 47 story model in LSDYNA to simulate collapse progression after the initiating event, which is of no interest at this point.

This is how they modelled the connection of A2001 to C79 in ANSYS, taking into account the results of the preliminary LSDYNA analysis:

[qimg]http://www.formauri.es/personal/pgimeno/xfiles/11-s/NCSTAR1-9fig11-15-crop.png[/qimg]

I've highlighted in yellow the control element they used to model walk-off, named COMBIN37. NIST claims their analysis showed that that element reported walk-off after 4 hours of fire, and then they proceed to interpret the results, attributing the major part of them to beam expansion, yet reporting column displacement to the east at the same time, and enough of it to break bolts in other floors.

[ETA] So as a matter of fact, NIST is claiming that the ANSYS simulation they performed showed that control element COMBIN37 reported a walk-off, by whatever means. Factors that I can think of that could have influenced it include pushing of A2001 to the west by the beams to its east, displacement of the column to the east by the girder to its west, and leveraging on A2001 that caused its end to move more than the element that pushed it. There may be others that I am missing, but that's not the point.

The point is that they say that in the simulation, that element reported walk-off, by whatever means. [/ETA]

Now "THE" question is, what exactly are you questioning? Here are some options that come to mind, but maybe it's none of them so you tell me:

- The results of that simulation. That would implicitly mean blaming the ANSYS authors for making a faulty program that behaves unrealistically.

- The setup of the simulation itself. But I don't think that's the case, because according to NIST's reporting of the status of the connections and to what you're arguing, the column was perfectly held in place preventing it from displacing east, therefore the inputs to ANSYS were good and showed the connections pristine and able to hold the column (let's disregard for now the mobility of girder A2002, which you don't account for).

- NIST's reporting of the results of the simulation. That would mean you're saying NIST lied about the output of ANSYS, and that element COMBIN37 (the one highlighted in the picture) did actually not report walk-off but NIST said it did.

So what is it you're questioning, gerrycan?

(COMBIN37) Can be switched on or off and varied in ANSYS. For example, in fig 11-14 it would be set to 4.5" to signify failure. NIST modelled column 79 to be the same as column 81, which had a seat plate that was 11" wide, and no stiffener plates. Their rationale for deeming an element to have failed due to walk off was that the girder web should no longer be supported directly below ie the girder had walked so that half of its width was no longer resting on the seat plate. In the case of an 11" seat plate this would be 5.5". Once the element was deemed to have failed it could be removed from the analysis.
When it was pointed out to NIST that the plate was actually 12" wide, they dutifully found a typo which transformed their original walk from 5.5" to 6.25". The problem is that the maximum expansion that could be experienced by any of the beams is less than 5.5" never mind 6.25".
The beam pushing due to thermal expansion was what NIST claim to have happened in their model.

Contact interaction between the end of the beam and the girder was modeled using a node-to-node contact element with an initial gap. When the connection was subjected to a large horizontal tension, 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.

Going back to figure 11-16 you can see that NIST did not account for stiffener plates at either side of the girder web and these would have increased the required walk off distance closer to 9" in order to have deemed this element to have failed.
A collapse has to start somewhere, and NIST chose girder walk off at column 79 as their initiating event. I cannot envisage how this would happen in the building itself, but for NIST to claim the failure to have happened in this manner in their model is just plain nonsense.
They made no account for the plates, and they did not model the connection correctly.

you still didn't mention which drawing the figure you posted was supposedly based on..

 

[gerrycan]

I think he answered it here:



This is the connection as modeled not necessarily the same as plan. It's based on plan but also takes into consideration previous simulations. I'm thinking this is where the confusion lies.

Not at all. I am questioning the validity of the connection failures around C79 that were carried into the LSDYNA model from ANSYS.

 

[pgimeno]

(COMBIN37) Can be switched on or off and varied in ANSYS. For example, in fig 11-14 it would be set to 4.5" to signify failure. NIST modelled column 79 to be the same as column 81, which had a seat plate that was 11" wide, and no stiffener plates. Their rationale for deeming an element to have failed due to walk off was that the girder web should no longer be supported directly below ie the girder had walked so that half of its width was no longer resting on the seat plate. In the case of an 11" seat plate this would be 5.5". Once the element was deemed to have failed it could be removed from the analysis.
When it was pointed out to NIST that the plate was actually 12" wide, they dutifully found a typo which transformed their original walk from 5.5" to 6.25". The problem is that the maximum expansion that could be experienced by any of the beams is less than 5.5" never mind 6.25".

That's not really a problem. I've already told you that NIST reports that the column (and therefore the girder seat) moved east, and there are other mechanisms such as leveraging that could have had a role in reducing the expansion needed for walk-off.

Are you questioning that the COMBIN37 element at the connection of A2001-C79 showed walk-off in ANSYS?

Are you questioning the parameters that NIST fed to COMBIN37?

The beam pushing due to thermal expansion was what NIST claim to have happened in their model.

They claim that the analysis showed that the girder walked off; that was necessarily detected by COMBIN37. They attributed it to the expansion of the beams as the major factor. But (1) leveraging allows for the displacement at the seat to be greater than the expansion of the beams, and (2) the column displaced to the east, increasing the relative distance between the girder and the seat, possibly by several inches.

Going back to figure 11-16 you can see that NIST did not account for stiffener plates at either side of the girder web and these would have increased the required walk off distance closer to 9" in order to have deemed this element to have failed.

That's not correct. That has been discussed in this forum in past. When the web was off the seat, the girder could roll off the seat; the connections would not be able to hold it.

A collapse has to start somewhere, and NIST chose girder walk off at column 79 as their initiating event. I cannot envisage how this would happen in the building itself, but for NIST to claim the failure to have happened in this manner in their model is just plain nonsense.

You keep saying that, but while you don't address my arguments rebutting the claims in which you base it, it will just have the status of a bare assertion.

you still didn't mention which drawing the figure you posted was supposedly based on..

They don't say. The legend only says "Based on fabrication shop drawings (Frankel 1985)".