however the property of a mild steel structure that hotrod seems to like," graceful ductile failure" I don't think I could support if the "graceful ductile failure" happened to be the roll cage folding in over my head!
You are misinterpreting my comment. The idea of graceful failure is well understood in engineering design.
You can never design a structure or component so it is absolutely guaranteed to never exceed its design limits and fail.There will always be that 1:1000 or 1:1000000 event where all your design planning assumptions turn out to be too conservative or the loads are way beyond anything you planned for.
Example suppose you design the roll cage over your head to survive a 20 G impact after the car flies at 200+ mph and lands on its top. Ungraceful failure would be for the car when experiencing a 21 G impact for the roll cage structure to break apart and the car parts and driver to fly in loose formation until they all come to rest.
Graceful failure of the cage would be for when subjected to a 21 G or 30 G impact for the cage to remain intact but deform significantly, perhaps putting a scuff mark on the drivers helmet where the top of the cage comes down 2 inches but still manages to protect his head.
Alive and injured is always better than almost alive. Mild steel structures when loaded beyond their ultiment strength tend to bend and stretch to release some of the over load without breaking.This in safety cage structures can be the difference between a deformed but still useful safety bubble inside the cage and a cage loop that buckles and breaks at the welds coming off the frame leaving the driver with ZERO protection.
Personally I believe rule making bodies should include energy dissipation structures into their safety cage designs. When they were designing safety casks for radioactive materials they tried several different designs to come up with a structure that would survive really extrodinary extreme impacts. Things like being hit by a high speed freight train at a RR crossing or having a truck loaded with a D-9 bull dozer cross the center line and side swipe the safety cask with the bull dozer blade at highway speeds (140 mph closing speed), the final designs have fins on the cask (they look sort of like an air cooled engine cylinder). Those fins are designed to take such an extraordinary impact and deform using up energy so that the cask structure itself although mangled still survives and maintains it integrity.
Rule makers are finally starting to consider such energy absorption structures. Champ cars now require an energy absorber on the back of the transaxle unit so if the car backs into a concrete wall at 200+ mph there is some give and dissipation of energy in the structure before an immovable force (concrete wall) meets an extremely rigid metal structure (transaxle/engine assembly) and gets pushed through the drivers seat back.
NASCAR after the Earnhart accident also included energy absorbing foam structures in the doors and of course went to the energy absorbing barriers.
Roll cages should have an energy absorbing structure on the top/outside of the primary cage tubes to absorb catastrophic impacts and use up energy before the cage structure has to deal with the impact.This way the impact of something like an engine block striking a main roll bar tube, would instead of denting or cutting the outer wall of the tube, would mash the external crush tube, blunting the impact on the structural tube, and protecting it from that point impact.
The ideal low cost crush structure could be something as simple as a second external tube of lighter wall thickness, designed to crush on hard impact and protect the structural tube it is attached to.