thanks for the heads up on the images. his position and equipment in 16 suggests to me that a 14% improvement is indeed extraordinary.
fwiw, a std road helmet to a full-on aero helmet is about a 3% delta at beta=0.
14% is a lot, but it can and does happen, although it's usually not solely through positional change alone (although I have seen some awful starting positions and crappy clothing), e.g. we might see double digits if we also consider impact of wheels etc (you might be surprised at what people think is aero, or that wheels don't matter - seriously!). And the unecessary paraphernalia they have on their bike!
For those whom have managed to self set up a reasonably decent TT position I'm working pretty hard to find a 5% CdA improvement through positional changes in one 2 hour testing session, but it can be found and that's not an uncommon outcome, obviously it's an individual thing.
Did that recently with a recent world elite track champion and another former pro turned masters racer (just loves his racing) and managed a CdA reduction due to position adjustment of ~4%. Obviously we also look for other improvements, with helmets being the obvious one.
I've seen examples where lowering the rider's front end down from a similar position to your posted photo has made their CdA higher
, I've seen it result in very little change, and I have also seen it result in substantial improvements. What you notice happen for one individual just does not universally translate. Indeed we found this testing the elite world champion rider the other week - we had to bring his front end up
in order to lower his CdA. It seemed counterintuitive, but we had 4 hours of solo test time to re-run multiple times and be absolutely sure.
Some people are just aerodynamically gifted and some respond (in a CdA sense) to positional changes far more than others. Aero is weird like that.
And the best helmet choice for that same WC rider was one of the most unusual as well. I almost said to not bother with that helmet as it never wins when performing aero helmet comparison testings. Well it did for this guy. Aero is weird like that.
On some people I have tested road to aero-helmet difference that is negligible (including myself), and yet I've seen substantial differences with same helmets on others. It is so individual. Aero is weird like that.
As for controlling for conditions, yeah definitely something to stay on top of and venues need good inspection and repeated use to gain insight into what can/does happen (but a new tunnel presumably has same issues). Which is why it's great to see professionals like these guys working hard to nail such things. Personally I monitor and adjust for air density changes continually (every run is checked before/after), and venues I use don't have other stuff like volleyball going on as I book the venue for exclusive use, so things like loading dock doors are a non-event or readily spotted. Also I try to have baseline runs repeated during and at the end of a session to see how consistent they are.
Heck, on the right days I've got some pretty good testing data at outdoor tracks. Precision suffers a little, so it then becomes a matter of working out what level of change can be reliably detected. My team mate runs our local LS wind tunnel, just a shame it doesn't (yet) have a bike testing rig as we could on the same day compare results.