Originally Posted by
OnePerRev
....The gear design does not account for a rim fracture as a design point, relies on this mode to simply never show up. An overtorque test is the standard used, but that would not consider degradation of raceway over time, just tooth bending primarily....
Good point. Gear rim fractures are most likely to initiate at the tooth root fillet surfaces where tensile stress is concentrated. With current design practice, the gear rim section under the root fillets is made thick enough (typically >1.25 times the tooth whole depth) so that any fracture initiated at the fillet propagates thru the adjacent tooth rather than through the rim. Loss of a single tooth is preferable to failure of the rim structure.
Here is a good read on the situation with a thin section planet gear supported by a spherical roller bearing, if anyone is interested.