Gyrene, you are a gamer. I wouldn't use the term with such elitist disdain.
The problem is you've been indoctrinated to think "your" game is right, and AH is wrong. Then you come to AH and "everything is all wrong!!" -- when you've just learned things the wrong way to begin with.
(fill in whatever "game" you want, I'm not pointing fingers or anything)
Re: failures... so what are the criteria of a part failing? Over-stressing past the tolerances it was designed for? Fractures/wear from being used long after its life? Wear-down of metals, rubber seals, gaskets?
Well, if you look at maintenance in WW2 and in fact even up to today, you'll find parts are made with a tolerance greater than that which they'll be used, and are inspected every sortie for cracks, wear, tear, and changed out with brand new parts on a very frequent basis.
So if the "failure" is from some random (yes, I said it) flaw, you should consider the odds of an engine properly maintained, and properly used. You're talking failures as if running an Allison engine at 3000 hp in the Reno races, rather than the 1150hp war-time use it was actually designed and built for.
There's "failures" because people did something risky (i.e. push something well past a breaking point, as in Reno races), then there's "failures" because the parts randomly are bad. The latter just doesn't happen much.
So if these predictable failures are all outstanding circumstances that NEVER happen and situations that NEVER come up in aces high, why should we have them modeled?
it's like saying "Okay, we should model atmosphere out to 100,000,000,000 miles, model the pull of the moon's gravity on the gas in your tanks, and make is so that landing on Mars is much easier due to the lighter gravity" -- all well and good, but the problem is you don't need to model any of that because in the scope of this game it's a non-issue.