more dodgy maths?
If an engine, which has shared the same history as one which has just failed, is asked to make more power than it ever has before (ie the 30 second rating) is the risk of it's failure in that 30second period still only 1 in 100000 x (3600/30) - ie 1 in 12,000,000 (based on the oft used 1:100000 per hour failure rate) ? (I don't think so)
If an old engine is at 10% of its beans remaining, and 30seconds burns all the beans of an engine life, then there might be some beans missing!