as the bending stress on the girder flange would be extremely high, so high it would almost be as though nothing was there. Since the seat was 12 inches wide the girder would have to move at least 6 inches sideways to fall. You could actually say a little more than that since the web had width beyond 6 inches from the edge of the seat.

NIST does say the girder expanded longitudinally and that this is what caused the girder to seat bolts to break. Column 79 was a built up column with extra plates welded to its flanges, but those plates did not extend much beyond the flanges of the column to provide much of a channel. The girder was also angled relative to column 79, so even with a couple of inches of girder expansion any sideways movement of the girder would not be restrained by column 79. Look at Figures 8-20 and 8-21 on pages 348 and 349 in NIST NCSTAR 1-9 Vol. 1 to see what I am saying here.

However, there is virtually no chance the beams could expand 6 inches. The expansion of 52 foot long steel beams at 600 degrees C (1,112 degrees F), after being heated from room temperature of 21 degrees C, is 5.07 inches. This is also assuming that they remain completely straight and don't sag at all. So in the worst possible case the center of the web would have still been an inch away from the edge of the seat. In reality, NIST needs the beams to get to 700 degrees C (1,292 degrees F) to expand 6 inches, but here it is really hard to make a case that they wouldn't be sagging and it is pretty hard to heat the beam to that temperature.

This also assumes all of the beam expansion went in the direction of the interior girder, which would require an infinitely stiff exterior and is also unlikely.

What gets even more silly is that in the WTC 7 report they also try to say that the beams could have pulled the girder off its seat by the beams going into lateral torsional buckling and shortening. I think this is what they are trying to say can happen even at 400 degrees C. However, this won't happen either for a couple of reasons; if the girder could be pushed or even rotated in the horizontal plane there would be nothing to resist the beams and cause them to go into compression, and even if they did buckle the shortening can't be more than the expansion in a horizontal member as it then relieves the load.

To your point about the girder being restrained by the channel of the column, it is this trapping between the flanges of column 44 at the exterior that NIST tries to use as a restraint to claim buckling of the beams would have occurred, but that can't happen for the reasons given above.

Of course, the above is allowing for no shear studs on the girder, which is extremely unlikely and Frankel Steel project engineer John Salvarinas' 1986 paper about the construction of WTC 7 shows they were called for there.

If NIST actually had a real natural explanation here they would have proved it with a physical test. They didn't.