There are multiple effects on bearings from heat, and all play a part:
1. Higher temps reduce the strength of the materials. No real effect on silicon nitride, and small effects on the steels at our temps.
2. The change in bearing preload due to bulk temperature changes and gradients. Potentially a big effect on our engines.
3. The lubrication changes with temperature and operating state.
There are probably others, as well.
It is hard to know the separate temps of each of the components, so it's hard to know exactly how your fits are changing. The easy thing to do is to run the engine on a stand as hard as you expect it to run, then stop it and immediately check endplay again. Under no circumstances should you be able to "feel the balls." Don't know how to say it better than that. (Well, I do, but as I have been "counseled" by my peers, Stunt Hanger is not the appropriate forum for technical matters. Flat earth society, you know?) You of course could fill in some of the temps by external measurement (infrared thermometer) of the case. The case spacing will grow in length more per degree than the crank due to the difference in materials. You might get an idea of the crank bulk temp by just looking at the end of it. Most of the other effects should actually reduce bearing preload during operation, although the contact angle will also affected by the axial thrust to a small extent. All model engines I know of have DB preload geometry.
Bearings self-heat as they run due to friction. More so when run at higher surface velocities. Smaller balls equates to higher surface velocities. A primary function of lubrication is to cool the bearings. Easy to see in the main, less so in the front bearing. But the heat has to migrate further to get to the front, so it’s not as bad as you might expect. And with a front rotor engine, you are dumping coolant thru the crank on every revolution.
I will point out that some engines I have had were more sensitive to bearing internal clearances. In my case, on Nelson .15s, they wanted a bit more clearance than the usual C3 fit coming from Boca. I talked to Henry about this and he agreed, saying that he had success using parts that were a bit more controlled than the C3 spec required. He stated a preference for a specific European supplier that had just what he needed. Unfortunately, he did not have any (many?) left.
You can get an idea of how hot an engine maker thinks his engines will be run by how much of a shrink fit he uses. You don't want the fit between the outer race of the main bearing and the case to become loose at max operating temperature or you will pound the bearing and fret the case. Think black junk coming out the exhaust. I would think this is less of a concern on the front bearing, and yet, many times the front is harder to remove. Perhaps in part because the main often comes out with the crank and you need a tool to simulate the same “ease” when doing the front. Perhaps because the clearance from temp soaking the smaller bearing is less.
When guys go to a catalog to buy bearings, there are often logical questions. "Should I get seals, not shields?" "I can't see how the front bearing gets enough lubrication. If I get double seals and a grease packed bearing will that be better?" And so on. Interestingly, in many cases you can get a variety of things to work, and seem to work well enough for that engine and their duty cycle to get acceptable life expectancy, that they conclude it is preferred. In fact, perfect. Well, it may be good enough, and still not be perfect. For example, grease packed bearings self-heat more. You are plowing the grease. Work is being done. There is no heat transfer due to flow-thru of lubricant. And bearings are often notoriously over-packed, not grease-plated. My preference is for a shingle shield on the front bearing (outside) and no other shields or seals. If the engine has a close fit between the shaft and the case between the bearings, the total leakage is controlled and limited. Check to be sure there is acceptable end play. And go have fun. I’ll say what every other Joe Bellcrank says---“Well, it’s working for me!”
I suspect that some of the control line speed guys are very knowledgeable about bearing selection and setup for model engines. We could probably learn from them.
Again, just a few more thoughts, for what they are worth. Forgive me for stating the obvious as I rambled.
The Divot