since there will be lots of new materials and bonding methods in these structures, so lots of new uncharted territory, then if nothing else it would be good to conduct destructive testing just to validate the computer predictions. So if the computer says it will break after 15,000 cycles but it goes 25,000 cycles that shows an error with the computer modelling. Otherwise this is one computer checking another computer, and the first real world test is likely to be with passengers onboard. And we've seen where that suck-it-and-see approach got us...
G