I was not challenging Greg L's power loss estimate due to gear box distortion. I was just thinking out loud about what 20 HP of bearing or coupling heating would be like.
My take on gearbox losses as a percentage of transmitted power ("gears are 98% efficient - woo-hoo") in conventionally oil lubed gearboxes changed when we were testing a fancy water cooled geared spindle. It was spec'ed to transmit about 35 Kw, but used about 15% of that just to squirt oil around while transmitting nearly zero real shaft work. I posted some temp/time test results of that planetary gearbox we embedded in a machining center spindle. All tests >>should<< be real low HP since no machining is actually being done. (3% of zero is zero)
Note The gearbox oil and spindle housing got pretty toasty real quick at the higher speed ranges. The monitored Coolant flow and temp increase indicated the drive motor was working MUCH harder at 8000
rpm than just a few thousand rpm slower. And the power window on the motor's vector drive agreed. Kind of made sense with all 14 gears sitting in the bath tub squirting oil out thru the gaps between their teeth.
The gearbox manufacturer kept reciting the 3% efficiency mantra, claiming that many BTUs could not possibly be generated, but the gearbox would simply have none of it. Only when they re-ran their tests at higher speeds, and kept track of how many gallons of French river water they were actually pumping thru their test rig, did they begin to
believe our data.
When the BRM H-16 started making a lot more power improved oil management was listed as an important reason why.
Everybody thinks their oil pan baffle/scraper is worth 50 HP at 8000 rpm. I bet some might be.
I believe if grossly misaligning a pump or fan caused it to consume several extra HP the passively cooled bearing and coupling temps would skyrocket. Thus I also believe claims of energy savings from any typical alignment or balance improvement are, well, hopeful.