Hardfire: Physics of 9/11

[psikeyhackr]

Arrogance and ignorance is an ugly combination and make for so many content-free posts.

.
I am complaining about IGNORANCE!!!

I am complaining about not knowing the numbers and weights of the perimeter wall panels.

You people are DEFENDING the MAINTENANCE of IGNORANCE.

ROFLMAO

psik

 

[femr2]

Despite the fact that the LS-DYNA, when faced with the Non-linear geometry and Non-linear material properties can't converge in any reasonable delta-t time step

LS-DYNA seemed to cope with convergence for the WTC 7 model...

 

[BigAl]

.
I am complaining about IGNORANCE!!!

I am complaining about not knowing the numbers and weights of the perimeter wall panels.

psik

What would you do with the numbers if someone gave them to you?

Do you think that the aluminum panels were part of the load-bearing structure?

 

[rwguinn]

LS-DYNA seemed to cope with convergence for the WTC 7 model...

Different collapse Mechanism
And I'd like a link to that, please.

 

[R.Mackey]

Different collapse Mechanism
And I'd like a link to that, please.

There seems to be a fundamental confusion about modeling tools used.

In the WTC Towers report (NCSTAR1-6D, to be specific), the models used to estimate progression to failure were run using ANSYS, not LS-DYNA. LS-DYNA was used, instead, in the aircraft impact simulations -- a much more similar scenario to the WTC 7 collapse model.

LS-DYNA is a totally different type of model, essentially a transient model whereas ANSYS is a static model. It therefore can handle free bodies and collisions without failure to converge, but it does so at a spiraling cost in complexity. I imagine the WTC 7 model was ground-breaking in its size, and would have been impractical to run in 2004. LS-DYNA analyses are also prone to large uncertainties, since minor changes in the early stages of a dynamic situation will propagate throughout the problem.

Today, LS-DYNA could be used to model the WTC Towers collapses, much like WTC 7 was, but that was never a requirement on the team. Nobody (except a few irrelevant lunatics) needs that kind of detail to understand the collapses of the Towers. A far, far simpler model suffices to prove that they were inevitable, as femr2 would discover if he'd actually use his model properly, rather than just brag about it.

To complain that the ANSYS models "should be made to converge" in an inherently unstable situation, as a poster in this thread has done, betrays a total lack of understanding. It just can't do that. You can get closer to the point of instability by increasing your modeling fidelity, but you do so at a staggering additional cost, for virtually no improvement in results.

 

[aggle-rithm]

.
I am complaining about IGNORANCE!!!

And I am complaining about RELEVANCE!!!

ROFLMAO

psik

Well, the sky may be falling, but at least you're keeping your sense of humor about it.

 

[FineWine]

.
I am complaining about IGNORANCE!!!

I am complaining about not knowing the numbers and weights of the perimeter wall panels.

You people are DEFENDING the MAINTENANCE of IGNORANCE.

ROFLMAO

psik

And what happened when you contacted NIST? Don't tell us that you didn't bother to make a phone call after all your blather.

You keep complaining about not knowing the weights of the perimeter panels, but the engineers have told you repeatedly that the exact weight is insignificant. Either tell us why they are wrong or drop this tiresome rant.

 

[rwguinn]

There seems to be a fundamental confusion about modeling tools used.

In the WTC Towers report (NCSTAR1-6D, to be specific), the models used to estimate progression to failure were run using ANSYS, not LS-DYNA. LS-DYNA was used, instead, in the aircraft impact simulations -- a much more similar scenario to the WTC 7 collapse model.

LS-DYNA is a totally different type of model, essentially a transient model whereas ANSYS is a static model. It therefore can handle free bodies and collisions without failure to converge, but it does so at a spiraling cost in complexity. I imagine the WTC 7 model was ground-breaking in its size, and would have been impractical to run in 2004. LS-DYNA analyses are also prone to large uncertainties, since minor changes in the early stages of a dynamic situation will propagate throughout the problem.

Today, LS-DYNA could be used to model the WTC Towers collapses, much like WTC 7 was, but that was never a requirement on the team. Nobody (except a few irrelevant lunatics) needs that kind of detail to understand the collapses of the Towers. A far, far simpler model suffices to prove that they were inevitable, as femr2 would discover if he'd actually use his model properly, rather than just brag about it.

To complain that the ANSYS models "should be made to converge" in an inherently unstable situation, as a poster in this thread has done, betrays a total lack of understanding. It just can't do that. You can get closer to the point of instability by increasing your modeling fidelity, but you do so at a staggering additional cost, for virtually no improvement in results.

Mea Culpa. That's why I asked for a citation.
I am not familiar with LS-DYNA. I have used Ideas^R to run non-linear geometry and Non-linerar materials, simultaneously.
On a simple steel structure, (bus roll-over single column sim), for a 6 inch overall deflection, analysis time was <1 hour up to yield, and >12 hours for the remaining 3 inch deflection, with several restarts required, and iteration step definition of <.001 inch...
Back to what NIST model was SUPPOSED to find out...

 

[femr2]

There seems to be a fundamental confusion about modeling tools used.

In the WTC Towers report (NCSTAR1-6D, to be specific), the models used to estimate progression to failure were run using ANSYS, not LS-DYNA. LS-DYNA was used, instead, in the aircraft impact simulations -- a much more similar scenario to the WTC 7 collapse model.

LS-DYNA is a totally different type of model, essentially a transient model whereas ANSYS is a static model. It therefore can handle free bodies and collisions without failure to converge, but it does so at a spiraling cost in complexity. I imagine the WTC 7 model was ground-breaking in its size, and would have been impractical to run in 2004. LS-DYNA analyses are also prone to large uncertainties, since minor changes in the early stages of a dynamic situation will propagate throughout the problem.

Today, LS-DYNA could be used to model the WTC Towers collapses, much like WTC 7 was, but that was never a requirement on the team. Nobody (except a few irrelevant lunatics) needs that kind of detail to understand the collapses of the Towers. A far, far simpler model suffices to prove that they were inevitable, as femr2 would discover if he'd actually use his model properly, rather than just brag about it.

To complain that the ANSYS models "should be made to converge" in an inherently unstable situation, as a poster in this thread has done, betrays a total lack of understanding. It just can't do that. You can get closer to the point of instability by increasing your modeling fidelity, but you do so at a staggering additional cost, for virtually no improvement in results.

Well, as someone sitting in front of the ANSYS product suite, I beg to differ, Ryan.

ANSYS LS-DYNA combines the LS-DYNA explicit finite element program with the powerful pre- and postprocessing capabilities of the ANSYS program. The explicit method of solution used by LS-DYNA provides fast solutions for short-time, large deformation dynamics, quasi-static problems with large deformations and multiple nonlinearites, and complex contact/impact problems. Using this integrated product, you can model your structure in ANSYS, obtain the explicit dynamic solution via LS-DYNA, and review results using the standard ANSYS postprocessing tools.

You can also transfer geometry and results information between ANSYS and ANSYS LS-DYNA to perform sequential implicit-explicit / explicit-implicit analyses, such as those required for droptest, springback and other applications.

Oh, and for rwguinn:

NIST WTC 7 LS-DYNA Model Animations

 

[R.Mackey]

Well, as someone sitting in front of the ANSYS product suite, I beg to differ, Ryan.

Ha! I learned something. I knew ANSYS also offered fluid modeling tools and so on, but I didn't know about the connection to LS-DYNA. Thanks.

However, the point remains -- regardless of branding or fusion of tools, the WTC Towers modeling was done with a relaxation model, not a transient model. The WTC 7 collapses were modeled with a transient model. You can copy grids from one model to the other, but there's no guarantee it will work or is even appropriate to do so.

Again, you could model the WTC Towers collapses with LS-DYNA, but it wasn't done. The convergence failure they discuss in NCSTAR1-6D is due to their different algorithm. They didn't even attempt a dynamic collapse simulation, and there's no reason why they should.

 

[cmcaulif]

where did those pressure impulse diagrams come from?

 

[R.Mackey]

where did those pressure impulse diagrams come from?

If you mean the ones in my presentation, they came from NIST, NCSTAR1-2B, chapter 10 (as marked on the slides themselves). Chapter 10.5 describes their creation, using LS-DYNA.

 

[Heiwa]

There seems to be a fundamental confusion about modeling tools used.

In the WTC Towers report (NCSTAR1-6D, to be specific), the models used to estimate progression to failure were run using ANSYS, not LS-DYNA. LS-DYNA was used, instead, in the aircraft impact simulations -- a much more similar scenario to the WTC 7 collapse model.

LS-DYNA is a totally different type of model, essentially a transient model whereas ANSYS is a static model. It therefore can handle free bodies and collisions without failure to converge, but it does so at a spiraling cost in complexity. I imagine the WTC 7 model was ground-breaking in its size, and would have been impractical to run in 2004. LS-DYNA analyses are also prone to large uncertainties, since minor changes in the early stages of a dynamic situation will propagate throughout the problem.

Today, LS-DYNA could be used to model the WTC Towers collapses, much like WTC 7 was, but that was never a requirement on the team. Nobody (except a few irrelevant lunatics) needs that kind of detail to understand the collapses of the Towers. A far, far simpler model suffices to prove that they were inevitable, as femr2 would discover if he'd actually use his model properly, rather than just brag about it.

To complain that the ANSYS models "should be made to converge" in an inherently unstable situation, as a poster in this thread has done, betrays a total lack of understanding. It just can't do that. You can get closer to the point of instability by increasing your modeling fidelity, but you do so at a staggering additional cost, for virtually no improvement in results.

OK, so if the little planes colliding with the WTC towers are destroyed 100% at impact at 500 mph just making a hole + local failures, why doesn't the little top part of WTC 1 - your part M - just get stuck on top on WTC 1, when colliding with the bigger lower part at 10 mph? Or at least get damaged?

 

[cmcaulif]

Well, as someone sitting in front of the ANSYS product suite, I beg to differ, Ryan.

If they wanted they probably could have used the arc-length method, a nonlinear static analysis that doesn't 'blow up' at the onset of instability like the Newton Rhapson method (which is common for nonlinear analysis). This method could give postbuckling behavior, but is not suitable for the impacts occurring during collapse.

 

[psikeyhackr]

And what happened when you contacted NIST? Don't tell us that you didn't bother to make a phone call after all your blather.

You keep complaining about not knowing the weights of the perimeter panels, but the engineers have told you repeatedly that the exact weight is insignificant. Either tell us why they are wrong or drop this tiresome rant.

.
Insignificant was it?

We know the weight of the heaviest perimeter wall panel because it was mentioned in a 1970 engineering magazine, 22 tons.

There were 59 columns on each side of the buildings but two columns were part of the corner pieces. So 57 columns were composed of perimeter wall panels. With 3 column sections on each wall panel that is 19 wall panels per side. The panels were 36 feet high and the levels were only 12 feet so only 1/3rd of any given panel were on a level. So that calculates out to:

19 * 4 * 1/3 * 22 tons = 557.33 tons of steel per level in the perimeter.

That would be the amount of steel for perimeter columns and spandrels starting at the 10th floors of the buildings. Now since we don't know the numbers and weights of each type we can't determine where the transitions were and what the weights were on higher levels. But counting the corner piece it had to start off at about 560 tons per level. That doesn't count the floor slabs and trusses which were about 1100 tons and the steel in the core.

So how did these engineers explain to you that 560 tons was not significant? Did they even give you any numbers?

psik

 

[FineWine]

.
Insignificant was it?

We know the weight of the heaviest perimeter wall panel because it was mentioned in a 1970 engineering magazine, 22 tons.

There were 59 columns on each side of the buildings but two columns were part of the corner pieces. So 57 columns were composed of perimeter wall panels. With 3 column scetions on each wall panel that is 19 wall panels per side. The panels were 36 feet high and the levels were only 12 feet so only 1/3rd of any given panel were on a level. So that calculates out to:

19 * 4 * 1/3 * 22 tons = 557.33 tons of steel per level in the perimeter.

That would be the amount of steel for perimeter columns and spandrels starting at the 10th floors of the buildings. Now since we don't know the numbers and weights of each type we can't determine where the transitions were and what the weights were on higher levels. But counting the corner piece it had to start off at about 560 tons per level. That doesn't count the floor slabs and trusses which were about 1100 tons and the steel in the core.

So how did these engineers explain to you that 560 tons was not significant? Did they even give you any numbers?

psik

So it turns out after much ranting and raving that you do have a pretty good idea of the weights of the perimeter panels. In other words, the information has always been available to you and you are perfectly capable of estimating the numbers you claim to want. Your performance has been just a noisy show. The engineers here keep asking what the difference is if your estimate is a little high or a little low. When are you planning to explain your obsessive need for the exact weights?

 

[psikeyhackr]

So it turns out after much ranting and raving that you do have a pretty good idea of the weights of the perimeter panels. In other words, the information has always been available to you and you are perfectly capable of estimating the numbers you claim to want. Your performance has been just a noisy show. The engineers here keep asking what the difference is if your estimate is a little high or a little low. When are you planning to explain your obsessive need for the exact weights?

.
I have said plenty of times that we knew the weight of ONLY ONE perimeter panel.

But I notice you are not trying to explain how those ENGINEERS could say that 560 tons is not significant. I am not interested in estimates. It makes sbsolutely no sense that we do not have the numbers and weights of ALL of the perimeter wall panels.

So explain how 560 tons on one level of the building is NOT SIGNIFICANT.

psik

 

[tsig]

.
I have said plenty of times that we knew the weight of ONLY ONE perimeter panel.

But I notice you are not trying to explain how those ENGINEERS could say that 560 tons is not significant. I am not interested in estimates. It makes sbsolutely no sense that we do not have the numbers and weights of ALL of the perimeter wall panels.

So explain how 560 tons on one level of the building is NOT SIGNIFICANT.

psik

Why do you hate engineers so? You could be one if you were smart enough.

 

[GlennB]

.
19 * 4 * 1/3 * 22 tons = 557.33 tons of steel per level in the perimeter.

So how did these engineers explain to you that 560 tons was not significant? Did they even give you any numbers?

psik

You're coming across as close to insane.

All along, nobody has suggested that (say) 560 tons is insignificant. People have said that the difference between 550/560/570 is insignificant. If you extrapolate the known dimensions/masses up the towers - as suggested to you many times - you'll get close to the correct figure. Good enough for worthwhile calculations.

But where has anybody said that the entire mass of a floor of perimeter columns is "insignificant" ?

 

[Newtons Bit]

You're coming across as close to insane.

All along, nobody has suggested that (say) 560 tons is insignificant. People have said that the difference between 550/560/570 is insignificant. If you extrapolate the known dimensions/masses up the towers - as suggested to you many times - you'll get close to the correct figure. Good enough for worthwhile calculations.

But where has anybody said that the entire mass of a floor of perimeter columns is "insignificant" ?

Well, he needs SOMETHING to attack the NIST report with. He can't claim that he's smarter than engineers, or that engineers don't understand FALL of PHYSICS, if there's nothing wrong with the report.