Thermal expansion rates in steel are not state secrets. The data comes from the structural drawing dimensions, and accepted equations. Where do you think the spreadsheet is wrong. We have had it checked independently, what do you see as a problem with it and why? Do you want to talk to the team who did the analysis with us?
NIST don't make a simple claim along the lines of "beam A expanded by X". They ran a couple of different highly complex simulations, the results of which were that the girder fell off the seat. In the more detailed local simulation, the girder was twisted and rocked off its seat. Renderings of this failure are given. In the larger global simulation the end result was that the girder was unseated, and this was described as the girder being walked off by the expansion of the beams. However what actually happened to the girder and the individual is not described. No renderings are given.
Gerry makes the case that there were inaccuracies in the NIST report. A stiffener plate is missing from a diagram (and NIST have said nothing about it). A width is specified as 11" rather than 12" (since corrected). On a simplistic level of just considering one beam pushing a girder a certain distance in one dimension then these inaccuracies would seem to make a difference. But we don't know if these inaccuracies are in the actual model used, of if they are simply omissions and errors in the report itself. [Edit: actually the correct distance was used in the analyses, as RP notes below] We also don't know what difference [the stiffener plate] would make to the actual simulations.
What we DO know is that the simulations identified plausible failure modes for this type of building.