Thanks Astro - those articles talked about the geometry and pre-fabrication, but how did they actually bend the trusses with the exact required curvature required? Some of those sections are quite long and the diameter of the steel is quite large. Was it done when the trusses were originally produced, or were tubes of the appropriate diameter produced and bent when cold? I’m assuming a bloody big bender (not the Futurama variety) was used but then how was the profile and curvature controlled?
Somewhat off topic, but bending large steel plates for ship hulls has been done with a torch for many years. I interviewed an engineering candidate who did his thesis on a computer automation of this method.
The way it works is that the plate is at rest and without any force on it other than holding up its own weight. If you aim a torch at a spot for a while, two things happen. The steel there gets softer, and it tries to expand. Since the rest of the plate is still cold, and stiffer, the hot spot deforms a little bit. As it cools, it also gets stiffer, and so the new deformed shape mostly remains. Since it was getting to be neutral with respect to the rest of the plate while it was hot, as it cools it winds up in tension, and tries to contract and pull the rest of the plate around it as it goes. This introduces curvature.
So, they go around the surface of the plate doing heat treatments in spots or along lines, and adjust how hot they get a spot to how strong of an affect the want there, and so forth.
We all know what a warped mess welding can make of straight parts because of this effect. What this method does is harness that amazing power.
In my callow youth I worked at a fabricator with some of the largest rolling equipment on the east coast U.S. which included plate, pipe and structural rolls. All long before CNC or even punch tape controllers. The pressman would have templates made up to whatever radii required and would test frequently after judicious pinch roll adjustments. The hardest part(s) were the ends which tend to tangency. Pipe was easier than plate though avoiding helix could be tricky. Pull through mandrels were used unless concentricity specs were loose. All the rolling for large work had to correlate with overhead crane & operator who could inadvertantly have effect on the work.