Originally Posted by Rolfe
Hold it, hold it, hold it!
Let's get this straight.
The steel did not get soft enough to deform, or even become pliable, as most people think of it.It did get weaker. A fine distinction, I know, but let's be as accurate as possible
Take a generic steel. It has a yield stress of 36000psi (2.48x10^8 Pa) at 70 degrees F. Heat it to 500 degrees F, and that value is at 3/4 the original. At 800 degrees F, it is about 1/2.
Stress is Force/area in compression. You also have to add in M*C/I for bending, where C and I are geometry-based values--the shape is important, here.. Ideally, the collumns are in pure axial load--that's the way they are designed.
So now we have 1/2 the strength, plus!!!! steel gets longer as it gets hotter!!!! Amazing physical property, known to everyone except CT'er's. If the heating is symetrical, it all gets longer at the same rate. If not, as was likely the case, things get complex.
The floor trusses got longer (and weaker), trying to push the support collumns. This puts them (the collumns) in bending--they were designed for axial loading. Additionally, the collumns themselves are trying to get longer, while at the same time, getting weaker-both by heating.
It does not take long to fail a beam by restraining the ends (loading them), and heating the beam. You can find the values for Coefficient of thermal expansion (units are in/in/degree F), and calculate the strain (elongation per unit length (in/in)) at any temperature. Look that strain up on a force/displacement curve for your material, and figure the stress. Amazing how fast it goes to failure...