Actually, the girder cannot be 547 inches long as that is the center to center distance between columns 44 and 79. It seems Nordenson had a slight error there also. ...
...
I was doing the 20,000 pounds from memory while writing the post. I think I checked the mass in the FEA, which would have been very accurate. By hand I get
33 x 130 girder A2001 = 45 feet x 130 lbs./foot = 5,850 lbs.
24 x 55 beams K3004, C3004, B3004, A3004 = 53 feet x 55 lbs./foot = 2,915 lbs. each
21 x 44 beam G3005 = 52 x 44 lbs./foot = 2,288 lbs.
12 x 19 lateral support beams K3007, G3007, S3007 = 4 feet x 19 lbs./foot = 76 lbs. each
So I get 5,850 + (4 x 2,915) + 2,288 + (3 x 76 lbs.) = 20,026 lbs.
The reason I get 6,633 lbs./inch with the little higher frequency here is that I used the steel mass as it is what is relavent and what the natural frequency is dependent on. In the earlier calculation I had used the Nordenson load of 46,000 lbs. and that was not correct. If the slab were included it would make the natural frequency and stiffness even lower as without shear studs it adds nothing but mass. You can't use the slab mass with the 0.52 Hz frequency in the frequency equation as the frequency will go down in the FEA due to the greater mass and no addition of stiffness.
You can make the steel weight 21,000 if you like, as I do some rounding and averaging here. However, beam G3005 is a 21 x 44, not a 24 x 55, and your estimate loses 11 lbs. per foot for it, and the beam is actually a couple of feet shorter than the K3004 beam, so that would be about 600 lbs.. Adjustment of your estimate for that beam alone takes it down to about 20,200 lbs.. Just to show you that this is not significant, for 21,000 lbs. the stiffness is 6,960.5 lbs./inch., deflection is 31.6 inches, and generated load is 219,881 lbs.. This is only a few percent difference and does not change things at all and the weight was much closer to 20,000 lbs. anyway as you had the G3005 error.
...
I am editing this post to add a more accurate weight calculated in the attached spreadsheet. The drawings are also provided for anyone who would like to check the figures. I used drawing 4349 in lieu of A2001, but with a 537 inch length. The weight of the steel frame turns out to be 19,795 lbs., which is just a little under 20,000 lbs. and would have made my analysis conservative as the stiffness would have been slightly higher using 20,000 lbs..