I have some experience with epicyclic gear drive design. And when I looked at a cross section drawing of this gearbox, one thing that seemed questionable to me was the rim section thickness of the 2nd stage planet gears. The radial section thickness at the face width center appears to be less than 1x the gear tooth whole depth. I recall being given a stern lecture by a very experienced gearbox designer about making sure to provide plenty of rim section thickness with planet gear configurations like this. The alternating 2/rev tension/compression stress cycles at the bearing race surface resulting from "bending" of the rim will be more of a problem than you would think. There is a great illustration on page 8 of appendix H that shows how the planet gear rim deforms under load
It is also important to apply a suitable analysis factor to account for unequal load distribution between the planet gears. For this particular arrangement of eight planet gears with good accuracy mounted on fixed carrier pins, AGMA recommends using an analysis load factor of 1.3. Unfortunately, the analysis report produced by
Romax Technology in appendix H assumes equal load sharing between all eight planet gears, which would be much lower loading than industry recommendations.