"arm 36g"
Altho it has been a couple or three pages since this term was mentioned, I just wanted to inject a note of caution about assuming that this means the tail of the aircraft impacted the water with a resultant deceleration of 36 g's. The cited strength of 120,000 newtons is about 27,000 pounds.
For example, it has got to be highly unlikely that this part, which resists the vertical component of forces on the rudder, would be designed for 36 times the force resulting from the static weight of the rudder.
We are just not furnished any info (English version) as to how this 36 g's is derived (I intend to see what the French version says). It may be that the 120 kilonewtons is the strength at a midsection of the arm which is oversized to allow for the making of adequate connections at the ends.
One visible end has a single hex bolt as the connector. It is difficult to easily see how large this bolt is, or how in the assembly the +/- 30 degrees of rudder deflection is accomodated by the lower joint. (Is this deflection correct?)
The thing to know in this is that a single bolt hole will induce a surprisingly high amount of stress concentration, and that the smaller the bolthole, the worse the problem can be-- depending in part how fatigue from accumulated service enters the situation.
This in turn depends on whether and which parts of the joint and arm are made of plastic. We are not told, nor is it clear which part broke-- there is just a lot of fuzz in the picture. Where plastic is involved, the impression I am getting is that there is a lot of ad hoc design even now, 50 years after the early research. (I was such a researcher, doing original research, with glass-reinforced parallel-filament plastics, that long ago.)
I would assume the design loading here for vertical component of rudder loading is to prevent rudder disablement following a tailstrike on takeoff.
For information, an AB representative testified that the design basis of the VS attachment was the gust loading (testimony at the AA Brooklyn 2001 hearing). Presumably this was the crosswind gust allowance on takeoff and landing, altho that was not said, IIRC.
There also appears, somewhere in [possibly other] testimony or furnished data, that in the history of these AB fins, a change was made in the percentage to be used for ultimate load. It was an increased percentage over what may have been the working load (as I would call it for clarity in my view), and was a substantial increase.
However, any discussion of the terms limit load and ultimate load in regard to stresses gets a bid complicated. These terms occur in the CFR (USA) and are complete with percentages dating to 1965 without change, IIRC. This situation was sufficiently murky to many (my characterization, true) that NTSB inquired of the AB rep exactly what "limit load" meant, in the case of the fin.
The reply (hearing cited above) was that it was the load not exceeded, without elaboration IIRC. Further clarification was not sought by NTSB, again IIRC. My conclusion was that it was the load at which, if exceeded, the fin would in theory be replaced.
One example of a replaced fin is of course the Canadian AB that overflew Miami and turned back to its origin in the Caribbean. There you would have the say it was the ultimate strength that had been exceeded at some of the attachment points. The reinforcing fibers leading to the attachment points in these fins (not the rudders) are carbon, and that is true of all at least through the 340 as far as I know.
I also believe that the statements that AB did not approve reversing the rudder from lock to lock in flight might have been backed up by reasoning that this would cause the limit load on the VS attachments to be exceeded, altho I have never heard a specific reason given by AB. The margin between limit load and ultimate load in plastic fins does appear to narrow with age of A/C, based on the sparse data from AB.
This is hardly a full discussion of the structural aspects, and is already far more verbose than I had intended. Just hitting the high points for now.
OE