Can anyone provide any info on thumbrules for sizing motor space heaters.

IEEE 43-1974 contains some excellent info which suggests that winding temperature should be 5C above ambient, and gives a formula H=DL/35 for attaining this, where
H=Heating power in KW
D= Diameter in feet
L= Length from endbell center to enbell center.

This info was removed in later rev's to the standard, but I don't know why.

I do have an example vertical motor 1250hp dimensions H=4.6', L=4.6' The above formula predicts space heater sizing of 600w.  We actually have 400w.  Measurements of winding temperature by RTD indicate approx 10C rise above ambient (and rtd can be proven accurate....we've got computer trend of rtd and ambient temperature... when we intentionally removed space heater power we saw the delta-T go exactly to zero.

So far so good, I'm not questioning the fact that we're getting higher delta T (10C VS 5C) with less watts then required (400W vs 600W).

We had two of six motors fail within about 6 months apart (historical failure rate is about one failure every 2 years among these six outdoor motors).  When evaluating the failed motors the shop told us the space heaters were undersized.  Their thumbrule is 1-1.5watts per horsepower, which predicts we need up to 1800watts.

Call me skeptical. In my mind the observed temperature is the proof of the pudding. All we need is temperature above ambient at all points within motor to prevent condensation. I'll admit there is some uncertainty in attempting to extrapolate from winding temp to other points in the motor.

Any comments on the above?
In particular:
What thumbrules are available for estimating space heater watts from motor size, weight, horsepower etc?  Have you ever heard of the 1-1.5 watts per HP rule?