WRT 1: length of run and more replicates helps average out errors and variations. This is a pure guess based upon all the 1-2 mile segments on my 10 mile TT. I get pretty much the same CdA for those short flat segments on windless days. But day to day variation can be large (temp, wind, humidity, setup, pressure, etc). By same.....like 0.158, 0.155, 0.160.......that kind of sameness. But I do not repeat (replicates) and average. You can easily do that on 1 mile runs. Do them many times and average for condition 1. Change your whatever for condition 2, repeat. Whether you do ABBA depends how important it is to you. I have to do this homework for hand and arms myself.
WRT 2: I try to hold the same power generally speaking because it is training primarily. As an aside, you can tease out Crr from CdA by doing runs at say 20 mph and then at 25 mph. The correct Crr and CdA will only uniquely solve both runs. On my river road with the fast tires that I use, Crr is 0.0045. I tend to go 23 ish mph on my up and back runs since it is usually at least an 80 mile day. On my halfpipe, I hit 38+ mph at the bottom and then am crawling at the very top.....I haven't given a lot of thought to it because this is all I have but I tend to think the very high top speed and very, very slow low speed help WRT accuracy. I dunno. Head hurts today, could be wrong.
