Fatigue
The thing with fatigue is that it starts with grain growth in the metal. When you see the crack, you are already far down the road to failure, because of the intense stress concentration at the knife-edged leading point of the growing crack. Visible cracks increase rapidly in length.
A paper here yesterday published a picture looking back at the tail fin. What I think to be a major factor leapt right out at me. The fin is a double airfoil. In the area of the failure, the passing air is speeding up (and thus dropping in pressure) to pass the fin. I think you'll realize that you can't have pressure at a single point that is different in two directions.
Thus locally there is a drop in pressure on the hull at right angles to this surface that failed. That is, here the pressure differential at altitude is greater that the pressure used in the fatigue life test. Fatigue is very sensitive to the pressure range, moreso that to the actual level of stress.
Fatigue life testing is (or was) done by filling a hull with water, and pressurizing the water. Rapid cycles, and no danger of flying fragments. However, the pressure applied is perfectly uniform.
Now as for the panel-like appearance of the pop-out, when you have a dome-like deformation under pressure, the stresses at the restrained perimeter edges of the dome will be something like twice the stress in the center of the dome. This is very counter-intuitive but it is so.
The rip-stoppers are the restraint, although they are somewhat yielding. That and their rectangular pattern make for a situation that is very difficult to analyze, either by math or by models such as plastics that can show visible stress patterns in special lighting. Well, I may be showing my age by mentioning such ancient methods.
NTSB will of course etch the edges of the break and make a microscopic inspection for grain growth (compared to an as rolled sample of the material). I think you will see this will show fatigue damage was present.
Back to runway length-- I apologize for just throwing out "minimum" length. I should have said "minimum design standard", which is a hole-in-the-cheese eliminator approach. An operational minimum is entirely different, of course. I'll explain why as soon as I have time. Washington National was the classic example, as FAA HQ was on this field for many years.
OE
On reflection, I realized I should not leave the impression that this (local pressure reduction) is an unrecognized problem. Crown inspections seem to me to indicate an awareness there is some problem which must be guarded against.
Last edited by Old Engineer; 17th Jul 2009 at 12:43.
Reason: Added comment re crown inspections