The post mortum of most, if not all, detectable drive system faults that led to an accident were "missed" not by operator error, but by:
1. System diagnostic programming errors which showed false/erroneous readings which could not detect any faults.
2. Incorrect limits which were set too high and missed the event.
3. (Related to 2) Trying to use sensors located in non-optimum locations to monitor the target component. In the case of the 92, how close is the accel used to the faulty bearing?
4. Routine false alarms which destroys trust in the system, so if something real happens it is lost in the noise (G-REDL). I attended a CAA meeting where operators reported a 100 to 1 false alarm rate with North Sea HUMS.
HUMS is pretty good at detecting the second occurence of what causes an accident after fixing the system short comings highlighted in the accident review. In case of the 92 it appears it requires 3 or more examples.
The Sultan
Last edited by The Sultan; 13th Jan 2017 at 20:57.
Reason: Dropped word