Furthering Councours77's musing:
Evidence of pre-existing fatigue damage does not *necessarily* mean that the fatigue failure was primary. It's entirely plausible that some external factor caused loading to the blade in excess of what the remaining material could support, at which time it would fail at the fatigue site and expose the defect.
You'd want to do some analysis -- normally you can tell what part of a fracture face was progressive and what part was sudden, assuming it's not too bashed up by subsequent events. In the state the blade existed in just before the failure, was the remaining strength enough to withstand the normal operating forces (within some reasonable tolerance)? This is a fairly simple calculation.
If it's near the margin, then the most likely scenario is indeed that the fatigue fault grew until time ran out and the blade snapped off under normal (-ish) loads. If it should have still held under normal conditions, even in its degraded state, then you'd need to look for additional precipitating conditions.
What those conditions might be goes into pure uninformed speculation (on my part at least). A whopping surge prematurely disassembles part of the inlet, say, and re-ingested debris whacks the weakest blade?
I understand the general thinking is that a surge should not fracture a blade -- but what if it were approaching the point where it was going to come off soon? Here's a question: is a surge symmetrical (or maybe I should say circumferentially uniform)? During a surge event, does a given blade rotate through drastically different aerodynamic conditions as it spins around? How many revs are made during the typical duration of the upset flow? I think I've seen in the FBO test videos I've watched that sometimes the visible exit of combustion gases through the fan is asymmetrical, at least for part of the event. That would create a cyclic load that is (probably) much different than the design loads.
For that matter, under normal operating conditions, is the pressure/velocity uniform circumferentially around the fan? I can imagine it might not be, especially under large AOA (thinking about a simple prop clawing through the air in climb attitude).
I agree that the simplest and cleanest scenario is that the blade failure was primary, but there are things about that which nag at me. Not the least of which being that this event doesn't seem to conform to what is seen in static blade-off testing. Relying on a testing regime that *may* not accurately capture real-world failure scenarios is not a comfortable place to be.
For the record, I'm neither a pro pilot nor an aviation engineer, but I am a mechanical engineer who has too many years of this sort of failure analysis under my belt. And because of that I am still in awe of the detail and quality of the investigations the NTSB does!
Looking forward to what the professionals say when it's all wrapped up. Back to lurking and learning.