Okay. I have access to:
A) Set of 404 tubies (non dimpled), conti sprinter 21 rear, conti podium 19 front
B) HED3 clincher and my regular Mavic (that’s “Maveek”) Ksyrium ES front wheel, Vittoria Corsa EVO CX 23.
I am pretty sure that the 404 set is more aerodynamic, definitely lighter and better for tris. The second seems more like an indoor track setup, but what ever, it is fun to test stuff out; and this should be a refreshing little workout as part of my final peak period.
I do not have access to a power meter, a wind tunnel, legs that don’t fatigue (had to return those to Matel due to lead levels in the paint) and control over the weather. HR monitor, speed, cadence, inclinometer are in my watch.
Proposed tests:
10 minute trainer warmup
- “Normal” ride with HR at top of HR zone 2, 15 km circuit, slightly flat - wheelset A
- 5 minute recovery ride, change wheels
- “Normal”, same as above - wheelset B
- 5 minute recovery ride, change wheels
- “Hard” ride with HR middle to top of zone 3, 15 km circuit, slightly flat - wheelset A
- 5 minute recovery ride, change wheels
- “Hard”, same as #3 - wheelset B
Analytical Review:
- Average speed, cadence comparison
- Maxmium speed, cadence comparison
- Fastest time for wheelset, average of time over 2 runs
Any advice on adjustments to the testing methodology or the analytics? I know this test is noisey.
I know this test is noisey.
You said it first. 
HR is not enough of a precise or reliable indicator of external work rate to capture the differences between those setups. Your error rate would exceed the measurement ranges. In order to pick up aerodynamic effects, you need dead-calm conditions and a dead-flat course. No conclusion will be possible without that type of ride, and without a power meter.
However, I’ll save you the work. The clincher setup will certainly offer lower rolling resistance. If you put Vitto tubulars on those 404s and glued them HARD with track glue, then it might make it a tie. Keep in mind that if you glue them correctly, your odds of getting them changed quickly on the side of the road goes down dramatically. You’d have to cut the tires off with a razor, and then your road-mounted spare will be terribly slow since it will be glued loosely.
As for aerodynamics, the Mavic front wheel is costing you speed. Probably not enough to offset the slow tires on the 404s. Keep the Hed rear and find a cheap used clincher 404 for the front. Sell the 404 tubies. Then you’ll be flying.
I’ve had reasonable results with this type of testing in the past. However, you need to make one important change.
Instead of single loops of 15K, I recommend 3 or 4 repeats of 3-4k. With the repeated measurement you can determine the CI (confidence interval), and determine if any differences in mean are statistically significant. (That is, are any differences real or just due to noise and lack of repeatability.)
So test 1 would be the same test, but with 3 or 4 measurements within the one test. I don’t see a need for a recovery ride between test 1 and 2; in fact I would just change wheels quick and immediately move to test 2 with only the minimal ride required to get your HR back to the approximate average expected during the test. You will be able to see fatigue in the results if it is significant. If fatigue is a factor, each successive test will take longer than the prior. In which case, you need to re-do the test with shorter distances and/or easier loads.
Also remember to try to fix HR OR speed during the test. One of these two variables need to be fixed. You then need a seperate test where you map HR to speed so that when doing the analysis you can normalize all measurements to the targeted speed or HR. (Lets say you target a HR of 125. Your measurements will likely fall in a range of 123-127. So if you know the relationship of HR to speed you can adjust the speeds for each measurement for the error in average HR during the test.)
FYI - when I do this type of testing I use a loop that is only about 1.5 miles. The resulting CI is such that I can believe results that are more than about 1/8th MPH.
Paul