As a SIM novice, but knowing enough statistics to be dangerous, my assumption is that there is a difference in calculating fatigue for RP and SP and that this is done so as not to disadvantage RP.
Now, I didn't figure out the exact function, but here is what seems to happen. Your RP pitches a limited number of innings, and uses them in smaller bunches, say 1 IP per appearance. Once this guy has pitched as an RP and you switch him to SP, the SIM recalculates his fatigue factors based on new projections of how long it takes to reach the IP limit (in the case of the guy converted from RP to SP its reduced. Probably if you went the other way around, the fatigue factor would increase). So, you RP goes from 90/90 to 78/80 because, (1) his pitch count and IP go up as an SP and if its set at, for example, 50-60 pitches or around 5 innings, his fatigue prediction increases because after one game, if he makes the 5 IP, he is pitching more innings (say 5 versus 1.66) and he will reach his IP limit sooner; (2) because he is scheduled to throw more IP in his start, he needs to rest more to recover.
This is just my guess. I have no idea if this is what really goes on, but logically it makes sense.