I've no direct experience of designing or building them, but I've also looked into doing it a bit.

The IVA is a bit of a joke, it goes into corner radii for switchgear, but the chassis (for steel) is a cursory inspection of the finished and painted chassis for any obviously crap welding etc. I'd warrant your FE calcs would be enough to satisfy that.

Coming on to the FE itself (which I do know a little about) while it is easy enough to model most things, the issue is always boundary constraints. You'd want to test chassis stiffness but just what forces do you want to test it against in terms of bump etc? Would you want to model impacts? How do you deal with local stiffness? (think wings which bend when you press on them etc)

On top of that, when building the actual model, how would you model the connections between one material and the other? You'd need to look into how that's done in the aero and marine industries to determine the appropriate modelling methods.

It may be worth getting hold of the Australian IVA equivalent, they demand allsorts of proof the chassis if of acceptable design and construction even if steel and may offer some guidance on composite construction.

I've just had a look in the blue book (2011) and it assumed a steel construction, referring only to reinforcement plates welded in place. I guess again you'd want to consider an impact situation in your FE calcs and ensure that the roll cage deforms and intrudes dangerously on the passenger space before the feet punch through the floor (use iterative increased in loading to determine the appropriate area for the reinforcement plates.)

Finally I guess it's obvious to say, but as you're registering it for road use, I doubt it's a flyweight hillclimber, so build in generous safety factors!