.....With your scheme, for each shaft you need at least a pair of sensors in quadrature, probably more for multi-bearing shafts. Whereas for external vibe sensors, one can share a number of shaft/gears/bearings and/or more than 1 sensor can look at the same element to give corroboration. With the amazing "power" of signal averaging, current hums technology is pretty good at detecting the sort of shaft breathing you are referring to. Sorry but I remain unconvinced that proximity/position sensing is a better way to do it..........
HC- I agree that a single triax accelerometer mounted externally can monitor multiple sources of vibration (passing frequencies of bearing rolling elements, gear meshing frequencies, etc.) for abnormalities associated with spalling type failures. With sophisticated signal processing and accurate mapping of the gearbox's normal vibration signature, it is even possible to determine which component in the gearbox is having a problem. However, this vibration monitoring technique would not detect a transverse shaft fracture until the fracture propagation was significant enough to alter the structural stiffness of the shaft and produce a dynamic imbalance force.
It was not my intent to propose using the technique shown in that image as a fix for the EC225 MGB. I only meant to demonstrate that there are methods to monitor for such failures. As you correctly noted, it would be difficult to implement such a system for each rotating shaft in the MGB.
In practice, there is typically no monitoring for these types of failure modes. While it may seem illogical, especially given what occurred with the EC225 MGB bevel shaft, the reason for this is that such failure modes are not considered to be "credible" in a classic failure mode/effects analysis. In theory, these failures are prevented from occurring by a combination of careful design and analysis, use of special materials, tightly controlled manufacturing processes, special inspections, etc.
Regards,
riff_raff