I think that, of all the components on a multirotor, the bearings should be at the top of the "consumables" list and subject to a "planned replacement" policy. Bought in bulk, even high quality bearings are about $2 each or less. Replacing them every
n flights or
n hours would be cheap insurance, especially considering the potential damage/danger of an in-flight failure. They are effectively the
only moving part in the entire machine - the
sole mechanical contact point for the conveyance of hundred of watts of work from battery to propeller. To replace 16 bearings (for an octo) every, say, thirty flights is a consumables cost of about $1 per flight - insignificant!
Furthermore, they are exposed to the elements in a way and to a degree that invites early failure. They are
shielded bearings, not
sealed bearings: contaminants
can and
will get past the shields. This makes a strong case for using "
contact sealed" bearings for this application - these have rubber seals that make contact with both inner and outer races, effectively preventing the ingress of any contaminants. The penalty, of course, is increased drag. But on a large engine the power loss (and additional temperature rise) will be completely insignificant. The cost is slightly higher but it would constitute an overall
saving if it extends the planned replacement schedule from the above example of thirty flights to, say, sixty flights.
In addition to a planned replacement schedule, one should have a strict policy for replacement after
any "mechanical incident", whether it's an actual crash or just the slightest touch of a prop against the ground, a tree branch or whatever. In such an event the contact pathway for each bearing drops from the full complement of eight ball (in this size of bearing) to just one of those balls. That's potentially an
immense load through a
tiny contact area and is highly likely to cause damage to even the hardest ball and/or raceway. "Best practice" would certainly be to replace any bearing that has been subjected to such an event as a matter of course, rather than "hope it didn't do any damage". The damage may be microscopic in scale but it
will develop into an eventual failure, so why wait? One interesting thought here is that if you are going to replace all bearings on a regular basis then there would seem to be little point in using stainless steel units. The justification for using this material is only resistance to corrosion - the penalty is that stainless steel is not as hard as the chrome steel (or other types) normally used in bearings. This suggests that chrome steel bearings would actually be the better choice (and they're cheaper!) - they'll be replaced long before they have a chance to go rusty!.
I've found myself that the circlips and shim washers factory fitted to my MT4008's are of relatively poor quality. You should have a plentiful supply of high quality steel (not stainless) replacements on hand. Both circlip and washer should be discarded whenever they are removed and new items used for re-assembly. The cost is so insignificant that to do otherwise would be careless.
One of the significant differences between different qualities of motors is the tolerances to which the parts are made. The most significant measurements are the diametral tolerances of the bearing seats and their circularity/concentricity, and the diametral tolerance, circularity and straightness of the shaft. Deviations in any of these will have significant impact on the performance of the motor and, particularly, the life of the bearings since they will experience the full consequences of any such errors. Also important will be the material of the shaft as it affects the torsional stiffness - subjected to constant acceleration/deceleration, the shaft will flex, especially within unsupported lengths. This will induce high frequency vibrations that will target the bearings (since they represent the only fixed support points), causing accelerated wear. Of course, prop balance will do the same but at lower frequency - all of which adds up to an "extreme environment" for the poor little bearings. After any prop strike it would be wise to replace the shaft as well as the bearings (and hope that the impact didn't distort the bearing seats as well).
Spending extra money on ceramic, hybrid and/or higher ABEC-rated bearings doesn't seem justified. Ceramic or hybrid ceramic bearings exhibit lower wear rates, but a regular replacement schedule negates that advantage. A higher ABEC rating could only be justified for use in a very high quality motor where the tolerances (as mentioned above) are so good that the bearing's improved performance would not be rendered irrelevant by the neighbouring components.
Such an approach might seem like overkill to some but it represents a trivial amount of work and expense compared, for example, to a single-rotor heli. The bearings are, as I said earlier, the
only moving part on an MR and should be treated accordingly.
P.S. I haven't mentioned lubrication here at all - see
this thread