> The hull on the other hand is ... is impossible to model accurately.
Why is the hull difficult to accurately model? (assuming the exact materials and precise dimensions were known).
Why is the hull difficult to accurately model? (assuming the exact materials and precise dimensions were known).
The modeling they did assumes a perfect material, and then adds some arbitrary factor of safety that will hopefully account for imperfections. It probably would have been fine if only used for one dive, but not after a substantial number of fibers had snapped as the hull flexed across numerous dives.
[1] composite materials are not as homogeneous as metals are, so most of the time we actually don't know the exact materials and precise dimensions. Only manufacturers that make enough parts to afford to do destructive testing can characterize the process enough to make reliable guarantees about part strength.
And you can produce reliable composite parts (you don't want homogenous, but them to come out exactly as spec'ed) under strict control, but it doesn't look like they necessarily did that here. They seem to have eg foregone autoclave curing of the tube section probably due to cost.