Apologies for the late reply. I'm catching up now with the thread.
To Tony:
False. Plain false.
NIST claim that the girder moved to the west an unspecified distance due to thermal expansion, and that their FEA shows the girder walking off.
They don't claim that the 6.25 in. the girder moved off the seat matched the thermal expansion of the beams. That's a mistake you're making all the time.
And in that regard, your analysis has a number of unexplained assumptions that are fatal to your claims. It's time for you to address them.
1. You assume that the exterior end of the beams is fixed. You have provided no justification.
2. You assume that the beam that pushed the girder is the closest to the connection. Your justification is that it's the longest one; however, that justification is not valid, since leveraging is able to amplify the displacement of the girder further than the elongation of the beam, and the farther from the connection, the bigger the amplification factor will be.
3. You assume that the seat is in the same position all the time. You've provided no justification.
4. Your analysis lacks any consideration of creep. NIST's FEA includes it.
And while on it:
5. You assume that the girder's web being completely off the seat is a perfectly stable condition. You give no justification.
Put simply, you assume that maximum expansion of the beams = maximum displacement of the girder from the center of the seat, but you fail to justify that assumption. That failure is fatal to your claims.
If you actually read what NIST says, as to how they determined that the girder had failed, you wouldn't say it was based on the FEA showing the girder walked off.
On the contrary, I said it because I read what they said. Their criterion for walk-off was a girder displacement of 6.25 inches (after the errata fix), as you correctly quote below:
In the fourth paragraph on page 527 of NIST NCSTAR 1-9, you can see them say for yourself
Walk-off occurred when beams that framed into the girders from one side thermally expanded and the resulting compressive forces in the beams pushed laterally on the girder from one side, sheared the bolts at the seated connection, and then continued to push the girder laterally until it walked off the bearing seat. A girder was considered to have lost vertical support when its web was no longer supported by the bearing seat. The bearing seat at Column 79 was 11 in. wide. Thus, when the girder end at Column 79 had been pushed laterally at least 5.5 in., it was no longer supported by the bearing seat. Additional factors that contributed to this failure were the absence of shear studs on the girders that would have provided lateral restraint and the one-sided framing of the northeast corner floor beams that allowed the floor beams to push laterally on the girder due to thermal expansion.
The bolded line shows they made a determination, not that the FEA showed it happening.
My reading is that the FEA showed a walk-off distance of 6.25 inches from the seat, and that was enough for the forensic structural engineers at NIST to consider the girder as failed.
That's where my point 5 above comes into play:
5. You assume that the girder's web being completely off the seat is a perfectly stable condition. You give no justification.
With the stiffeners included there would never be enough stress to cause the girder flange to fail so saying it did because the web was no longer directly over the bearing seat would not be an accurate measure for failure.
Newton's Bit has made a valid point regarding the girder rolling off. You have not made any study that includes creep and shows how the girder is stable with its web off the seat. Your only argument about the state of the connections is that you don't know how they could have failed.
Based on what was in the original final report as shown above, along with the erratum of June 2012 where they correct the seat length to 12 inches from 11 inches and then say the lateral travel was 6.25 inches, they are still implying it is from the beams expanding as they provide no other mechanism for the movement of the girder web off the bearing seat. Of course, the beams can't expand 6.25 inches so they have a problem.
You have a problem. Their FEA shows the girder walking off a 6.25 inch distance. Your argument from ignorance
WP is a logical fallacy, the same you're using when you claim that the bolts in the fin connections could not fail because you can't figure out how they could fail.
And that fallacy is all you offer as proof for both things, which you spice with reversed burden of proof: "It can't fail because I don't see how it can fail," (argumentum ad ignorantiam) "so you prove it can" (reversed burden of proof).
You will also have to excuse me if I don't see where NIST mentions creep in relation to their alleged girder failure.
It is part of their ANSYS FEA (see NCSTAR 1-9 Appx. E). The FEA as a whole shows the girder walking off. The beam expansion is certainly a major factor, but also certainly not the only one into play. We're dealing with a system full of variables; you're reducing your analysis to a single variable (beam expansion) without taking all others into account, when the whole model showed the walk-off (see COMBIN37 in figure 11-15 of NCSTAR 1-9 as well as the last sentence of page 482).
The girder would be stable with its web past the seat with the stiffeners involved and the K3004 beam connection to the girder intact.
The connection was not intact. And a bare assertion won't cut it.
Just to show the impossibility of the claim that girder A2001 could roll-off when its web was past the bearing seat at column 79, I did a sketch showing the reactions, moments, and the factors of safety (FoS) of the various failure modes such as shear tearout of the beam web, shear of the 7/8" diameter bolts, and shear of the fin plate welds.
The minimum FoS of any failure mode for just one out of the five beams to the east resisting the roll-off was 3.6 at 600 degrees C, so it is clear that this claim has essentially no chance of occurrence.
See the attached.
You're assuming all the bolts in their pristine condition. NIST's FEA shows that not to be the case.