Tell ya what, Tony.
I'll take the first two, and give you a 2nd chance to try some of the others.
BTW, the answers to 3 of the other 5 are contained within this reply.
Here goes...
___
Thank you Oystein. Your comment rang true enough that it caused me to go back and check it out. I don't know how I missed the significance of this but yours and Marks comments got me to see it a different way than I had been interpreting it:
Oystein pointed it out to you, and you pointed it out to me.
That is the value of these discussions.
1-9 Vol.1 pg 353
This analysis demonstrated possible failure mechanisms that were used to develop the leading collapse hypothesis further. The failure modes in this model were incorporated into the 16 story ANSYS and 47 story LS-DYNA analyses
Yes, this is what NIST says. But you are reading it incorrectly.
There were THREE models constructed: a preliminary FEA, the ANSYS model & the LS-DYNA model.
1. the preliminary FEA model: a static thermal / mechanical model of a single floor of the Northeast corner only.
The
ONLY PURPOSE of this model was to help build the second model accurately. The first simple model helped the engineers determine which failure modes were
likely (not definite!), and which components needed to be modeled in detail and which could be simplified.
This very simplified model ran for only 4 seconds (real time), and all the data needed for it came out in the first 3.
None of the failure modes or temperatures or effects seen in this model made any appearance in the final failure theory or summary.
2. the ANSYS model: a dynamic thermal & mechanical analysis of 16 lower floors.
The purpose of this model was to help determine the initiating event.
This model started with fires beginning in the SW corner of the building. It took about 3 hours (real time) for the fires to work their way around to the SE corner, and the rest of the results that are reported in NCSTAR1-9 start from the 3 hour mark.
The temperatures of all the components were determined by the FDS simulation. Nothing burned for any artificially long time, or at any artificially high temperatures.
3. the LS-DYNA model: a dynamic mechanical model (the thermal component was only to get the material properties correct at FDS calculated temps) of the whole building's response that started at the moment that c79 was unsupported. Note that all column buckling was calculated in this model.
The purpose of this model was to track the failure progression thru the building.
This model used the mechanical collapse output of the 2nd model (& temp profile from the FDS) as input, and then calculated the mechanical response of the building to the initiating mechanical failures thru total collapse.
All of the above is exactly appropriate for each failure analyses.
___
Reading about the 1st two models should show you that the
failure modes found in the preliminary analysis were not "incorporated" (as in "shoe-horned") into the ANSYS & LS-DYNA analyses.
Rather, the information gained in the preliminary analysis guided "what to look for", or "what options in the final analyses to enable".
It told them, for example, that in the ANSYS model, they had to model the studs independently (they couldn't assume them fixed to the concrete), they had to model the bolts & welds & connectors, and they had to allow the beams & girders to rotate & buckle.
These are all choices that are made in any FEA. And the more of these choices that you allow, the larger your degrees of freedom & the more complex & longer your analysis time.
The proof that this is true is that the ANSYS & LS-DYNA failure modes (girder pushed off seats, for example) were
completely different from the preliminary analysis failure modes (torsional buckling of c44-c79 girder). Also the temperatures used in the ANSYS & LS-DYNA models were much lower than those used in the preliminary model.
However, I see a problem with GIGO in the subsequent analyses. There are numerous fraudulent aspects in the "rock to the east" analysis.
There are precisely zero "fraudulent aspect" of the preliminary model. There is merely your ignorance of how engineers perform failure analyses.
The preliminary analysis (i.e., "Rock to the east" model) is irrelevant to the final conclusions. It simply informed the engineers which issues were significant issues and which issues were insignificant.
Note that the preliminary analysis was INTENDED to highlight, to exaggerate the effects that were likely to occur in this specific structure. So they intentionally select conditions that were extreme, to see what broke & how it broke.
The final failure mode analysis, the ANSYS model, used the results of the first model to determine which components had to be modeled in detail & which components could be simplified.
This is standard engineering failure analysis.
It is "fraudulent" only to the clueless.
The preliminary model was NOT the basis for their conclusions.
The final (ANSYS) model was the bases for their conclusions.
As they say explicitly:
NIST said:
The boundary conditions and temperatures were selected to create maximum shear forces on the stud connectors and beam and girder connections. Note that, in the detailed finite element analyses of the 16-story ANSYS model (see Chapter 11), no boundary conditions were applied to the floor slabs, and the temperatures of both the steel and concrete were derived from a thermal analysis based on fire dynamics calculations.
NCSTAR1-9 vol 1, pg 349
1) they applied 4 hours of heat at 1100oF but the fires only burned for 20 to 30 minutes in any location
This statement is simply false.
Nowhere in any NIST analysis was any temp set at 1000°F & held for 4 hours. I find it incredibly difficult to believe at this late date that someone hasn't already called you on this error / fabrication.
You are, accidentally or intentionally, conflating two different analyses.
1.
In the preliminary analysis (which contributed nothing to the final failure causes):
a) they ramped the temperatures up from room temp to max temp over the course of about 2.5 seconds & watched the effects for around 4 seconds. They were looking at static response of the components, not dynamic response, so this technique is completely valid. (NCSTAR1-9, Fig 8-25)
b) the max beam temps ramped to 600°C (1100°F) & girder temps ramped to 500°C (930°F). This was a short cut and they intentionally set the temperatures too high for this
preliminary analysis.
b) they only needed to perform about 3 seconds of analysis time before they got the information that they were after. (See Table 8-2). None of the failure modes discovered here were included in the final "what caused the collapse" summary.
2.
In the ANSYS model (which were the basis for the final failure summary):
a) the temps in all components were calculated, the result of their Fire Dynamics Simulator, based on many local conditions such as office contents, fuel load, oxygen flows, etc. These temps were not set to exactly 500° (beams) or 600°C (girders), but rather ranged between 150° to 700°C, depending on time, location & fire conditions.
b) the 4 hours was from the start of the fire (in the SW corner) to the buckling of col 79. The heating in the Eastern side of the building did not start until 3 hours into the simulation, when the fires in the SE corner were just starting.
2) they heated the entire area all at once but that is not what happened in the actual fire.
Wrong in both cases.
In the preliminary case, they ramped up the temp over 2.5 seconds. This was a
static analysis so the ramp time does not matter.
In the ANSYS case, they ramped the temp UP & DOWN, locally & dynamically, as calculated by the FDS software. To the best of this software's ability, this is exactly how fast, how long & how hot the temperatures actually rose in each modeled component in the building. With a couple of simplifying assumptions that were verified thru detailed thermal analysis.
___
There ya go, Tony.
Not terribly hard, is it?
tom
)
