The remote possibility clause requires "every" failure mode to be forecast, identified, and considered in order for it to be a valid analysis.
Just how likely is it for an engineer to accurately identify every single failure mode that would result in a run dry mode for every single part and gear of a transmission?
That to me is something that is extremely remote in itself.
The question about the oil filter problem is that it identified itself to all an sundry at Broome, Australia. As soon as the first filter housing parted and allowed the oil to leak from the system it breached the "FAA approved Extremely Remote" clause in FAR Part 29.927 under which the aircraft was certified. (IMHO anyway).
The real failure was not the Oil Filter Housing Titantium Studs but the failure of anyone to realize the fact that had happened upon the first set of studs to fail.
The FAA and Sikorsky will have to explain that at some time in the near future I hope.
When the requirement to replace the studs was made and predicated upon a one year or 1250 flight hour limit that would suppose someone had analyzed the situation and determined the actual risk was more like two years and 2500 hundred hours flight time and halved the estimated time between failures or whatever they wanted to call it.
The fact it could be a fatal flaw seemed to escape notice as evidenced by the amount of time granted to do the replacement.
On top of that decision, the Operators of the aircraft have to explain why they failed to see a more urgent need to replace the problem studs.
These will be interesting questions for the FAA, Sikorsky, and the Operators of the S-92's in service today.