I agree that if you can get CdA for a time window, that's better than for a lap -- but even if all you get is CdA for a lap, yaw might still be useful. If the lap is shaped like a "D" or maybe a triangle or square, but with an elevation drop during part of the lap, then varying speed across and within laps could help you identify changes in drag with yaw even if they're not time-averaged.

As usual we agree. In your protocol (with a specific course layout and execution protocol) you are able to associate a change in drag with specific position/yaw condition.

When trying pul out the yaw/drag effect you are even more dependant on wind being constant which I absolutely do not get 80+% of the time. Maybe I need to go an test in Paris :-)

I can see two useful charts for these devices. One is the traditional Yaw vs CDA or drag and the other the time spent at various yaw angles. Any other yaw charts you have seen to be useful ?

Come to Paris anyway.

However, if things go well, maybe I'll have better ideas about yaw charts in a few weeks.

Question:

Are some of the really really low CdA's we see from top TT, pursuit folks attainable by folks at lower speeds? Or is it such that there is a theoretical low per your body/bike based on your speed?

Like.....a pro going 32+mph could have a minimum CdA of like 0.180 but a joe going 27mph only a minimum of 0.200 or something?

I am not sure I'm answering your question but here are a few points

a) I think some of the CDAs attributed to pros are grossly underestimated, ie we think he's a .17 and he is a .195. That or I only got to see the slow guys.
b) Yes, CDA can drop with speed. I would have to dig up the data, but it's not from a .180 at 27mph to .200 at 32mph. It's much smaller than that
c) One way to find out is aero test on a good descent at 35mph. If you are a .23 on the flats, you won't be a .20 on the downhill. You would see it well in a virtual elevation plot.
d) I tested a few riders that had tunnel references, tested at mere mortal speeds and the numbers made sense with their tunnel numbers. Often you'd test, give him a number and get a "ya, that's what I got in the tunnel, on another bike, on another team".
e) One thing we sometimes forget is the size of these guys. They are stick men. There is a strong stickness to CDA correlation.
f) I have seen several riders that got a CDA of X in the tunnel and Y in the velodrome or road.

Underestimate a person's CDA is the new overestimating someone's FTP. You'd be surprised what reality is. I certainly was on both fronts.

How much of this is "social"? If power is generally fixed, but speeds keep going up and times keep dropping, then of course people will start self reporting .16 and .17 CdAs. We will be at .10 before long, say 2030.



E

---

Eric Reid AeroFit | Instagram Portfolio
Aerodynamic Retul Bike Fitting

“You are experiencing the criminal coverup of a foreign backed fascist hostile takeover of a mafia shakedown of an authoritarian religious slow motion coup. Persuade people to vote for Democracy.”

How real do you believe some of these CDAs are ?

My hunch is that people are getting lower CDAs but mostly through more availability to testing and they are not going from 0.25 to 0.18.

I was part of the tests they did with Lionel when filming chasing the Lion. The first run was with his 2015 Kona position including the Camelback. He was a .260. Very soon he was 0.24. By the end of the session he was a 0.23. He since moved on to a new bike, new fitter, lots of testing and in his hour record he achieved what ? 0.22 ? (which includes much better tires)

I don't think frames, helmets or wheels have changed much in the last 5 years. Lots of data to show that. Put a 2016P5 with a LGP09, HED disc/90 against any combo today. A watt ? Two ?

I do think fabrics and assembly of suits is improving slightly.

I think rubber has improved quite a bit. I don't think there was much under .0035 (that could last more than 4km) a few years ago.

But I think the biggest gains come from the availability of testing from multiple resources. There are more tools and more knowledgeable people. In 2016 very few riders had been tunnel tested, which was the only way to test. Today, every rider has been tunnel or velodrome tested with some Chunging and creative use of sensors. I'd love to hear a presentation from Dan Bingham on the gymnastics of data/sensors he uses to do his magic.

There are riders (like Sebi) that optimized early and really profited from it. But it has now become the norm. And I think this has driven a lot of the improvements. But soon the "elites" will hit the peak of aero.But the masses have a lot of work to do.

There is still a lot of room for improvement on the test tool front. We are seeing the tip of the iceberg but I think steam is running out for a few of the vendors and I think there is potential for a large paradigm shift in the next 18 months.

Thanks. Makes sense.

To me, I'm always most interested in the day's work. Run with current, run with new things and see which is better. The "actual" or "true" value to me is very "meh, whatever". I just care that the new things to try are faster. What the actual # is to me doesn't feel important because ultimately you have to push the pedals and hold the position and make your run.

This has just been a curiosity of mine. I wonder if some of the silly low CdA's may come from velodromes with poor assumptions for CRR. Where they assume a higher CRR than reality, and that contribution to power winds up there instead of where it belongs.......in CdA. Like, you assume a CRR of 0.004 but it is actually .003 and all of that .001 power lies there versus it taking more power to push air......which is the reality.

I have a fun experiment I'm trying soon with arms/hands that probably has been tried, but I can't find anyone trying it before. If it works, I'll probably never tell anyone. If it doesn't work, I'll post up about it. LOL. It's a bit weird, but would be "legal" in the fit rules.

This is one of my theories. Most of the people doing field testing don't do the types of protocols that would allow them to estimate Crr, so they just fix it at the same figure across tests.

A Crr of 0.0035 instead of 0.0025 gives about a 0.01 m² lower CdA. I think, good tires on a wooden track are below 0.003 (see attached diagram with Crr tests on a wooden drum test rig).

Using the speed from a wheel speed sensor for getting CdA may underestimate CdA by 0.005 m², depending on the turn radius, the speed and other variables.

How low does CdA have to be in order to be 'silly low'?
For a pure TT or Track rider the lowest achievable CdA is likely lower than a triathlete as more time for training in position and no running to worry about affecting.

In my day job, we often have to evaluate whether an estimate or an answer is plausible, or reasonable, or right, not by looking at the number but at the process of how the number was achieved. If the process was solid then we accept the number even if it was outside the range we originally thought it'd be in.

Don't know. These other folks in here are the experts.

All I know is when a "joe" posts up a figure in the sub-0.200 range, like in the lower 0.190's, I raise my eyebrow a bit. Some folks are talented, some lucky, but in my view it takes a lot more work and talent to get that low than the collosal aligning of planets required to just get lucky to have a sub-0.200. I say "joe" as somebody that plugged some numbers from a power meter ride into aeroweenie's calculator and posted up a CdA onto the chat forum.

It's a tricky thing calling stuff out like that. Like the "440w hour record Wiggins" thing. On the one hand, we stand there with no proof other than some numbers that look funny. On the other hand I am a person in engineering, life, and politics "extraordinary claims require extraordinary evidence". The Sagan standard. "Bigfoot exists". Alright, catch him and let's take some DNA samples. "I put out 350w for an hour on my TT bike but only go 25mph, what's wrong?". I'm from Missouri..........show me.

I think some of the lowest recorded values are in the sub-0.17 range, but a) these aren’t really big/tall people b) it has usually taken years and some serious work to get there.

I think it is fairly safe to say that 10 years ago sub-0.2 was rare, but the 0.19 range is now possible for most people. I would say that the biggest improvements since that time are skinsuits.

Currently at the World Championships in Belgium, and the TT earlier in the week had some impressive numbers. We have seen in wind tunnel tests as well as road tests with the AeroLab sensor (following Crr calibration) some riders do indeed achieve sub 0.17 CdA, but they are absolutely at the top level world-wide. Moreover, the sub 0.17 was indeed only achieved after years of training, and includes some very unique aspects such as custom design skin suit boundary layer trips leading to a significant decrease in CdA beyond about 39-48 kph (the effect is local, so there is no one single magical speed you need to hit).
The riders at worlds were hitting 49 to 54 kph average speeds for the 48+ minute solo effort, nearly 34 mph. Keep in mind several technical corners required the riders to slow down, even brake, and this usually does not factor into the math of what is truly needed to be going this fast on average. The riders have no choice but to be sub 0.19 CdA in order to be competitive. Our Worlds TT course-specific math was showing a 0.2 CdA would require approx. 475 [W] for sub 48 minutes in order to be on the podium. When you drop CdA to 0.17, you get closer to a humanly possible effort in the neighborhood of 400 [W].

---

Chris Morton, PhD
Associate Professor, Mechanical Engineering
co-Founder and inventor of AeroLab Tech
For updates see Instagram

Thanks for this diagram.

---

Blog | Twitter| Bike CdaCrr app

Thank you for this. We are doing some simulations for a WT team with the Tokyo and Worlds course and your number make sense. We will be off your numbers a few watts here or there, but this is what I am thinking

Only 2 guys went under 48 : Ganna and Van Aert. I don't have their power files but I am pretty sure they were north of 400w. Those guys could be much closer to .190-.195 and hit the times they did. Ganna ,I am pretty sure would be closer to 440watts. What a monster.

You go down the list and they are all big power guys. While they are not Ganna, they are much closer to 400w, and their time delta with Ganna is not in line with massive CDA improvements. That .19ish 400w range aligns with their time behind the leader.

Then you have a guy like Dan B. 16th place, 2Min 11 seconds down. I have never seen a Dan B power file, but let's pretend his CDA is .165 like some of the rumors floating around. He would need about 340-350 watts to do it in 2m11 down. I suspect he is more powerful than that. At a .17 somewhere between 350-360w. I can see this.

My point here is I believe VERY few people are in that .17 range. Very few are able to reproduce on the road and in typical race distances. The winners are still producing BIG power numbers. I do have the files of the guys that are closer to 0.20, do close to 400 watts and their times are pretty aligned with this. But they are 2.5+ minutes down. They are (IMO) the pretty dialed, pretty strong riders in the WT. Not the Dan B aero bullets or the Ganna super engines.

You are right these are guys extremely dialed, the very best equipment. No tri pros or amateurs are getting close to them. But I think the guys that can do .17s on a road TT are pretty, pretty, rare.

If you have files from the Olympics we can compare to those as well but so far I am seeing the same trends.

All this IMHO

Dan Bigham? Very very slippery.

---

Chris Morton, PhD
Associate Professor, Mechanical Engineering
co-Founder and inventor of AeroLab Tech
For updates see Instagram

Yes, Dan Bigham. Very well known for his slipperyness.

Also very well known to find ways of testing that go far beyond what the current state of technology would allow.

The kind of guy that knows what works in a given sensor and what doesn't. He throws out all the bad data, combines it will good data from elsewhere and gets results.

It's cooled off enough here that now an after work or weekend day session are possible in the nice 70deg weather. 90+ I would have fried like an egg in that Tempor helmet.

Either way, next line of questioning for the sensor folks........

When you get a "statistical sample" of data, typically your "period" is very long. Like "the temperature at noon in Raleigh in July". Well, you only get a shot at one data point every 24 hours. However, with aero testing like this you are sampling at the sample rate of your worst rate sensor. Let's say 1 second for sake of argument or easy math. I understand that accelerations eff with the CdA calc. So you want to limit that, OK. So you need to reach a state of homeostasis of the terrain or wind or your power not causing major accel. Now, with this..............what's the deal with sooooo many runs when the sample rate is 1 second. Let's say you're up to speed, no more accel, minimal wind, road is flat.................you go for 30 seconds. That's 30 sample points.

Can someone explain the testing/statistical value in covering identical road 4, 5, 6x for multiple runs? I mean, if the road is so shit you can't hit the same miniscule bump each time.....sure. But if the road is fresher paved and you can hold the same line..........

I see the 30 seconds once as almost like sticking a thermometer in water to get a temp. Your bike accel and homeostasis would be waiting for the water/thermometer to even out in their temp/energy to take a reading. Then you would sample the temp every 15 seconds for like 2 minutes.

I just keep feeling that the multiple many many repeated runs just may beg for more disturbances than they eliminate.

You needed all those runs with virtual elevation so you could sufficiently get the map of your elevation to skew over enough distance/time versus known elevation that you could use that delta to get at your CdA.

With the sensors though.......that shouldn't be as necessary. Right? You're sampling wind speed, power, wheelspeed, etc...... like 1/sec or more.

I just can't understand the idea of sooo many out/backs or sooo many laps. It's been like 15 years since I was in college and had a stats class.

 
What are you testing with that requires 4,5,6x identical road ?

I guess that's what I'm asking.

I mean this has lots of runs: https://notio.ai/...you-measure-your-cda

In there MTM is doing 1.5mi laps to the tune of 6 or 7 laps per setup. Sure, it's MTM. Not some wannabe like me at the regional novice TT level. But.......that's 10mi per setup.

That's where I got this from.

I mean, I was just going to do 3 trips down the street at 30sec of steady data per trip with each setup.....with a final trip of 30sec back on the old setup to verify. It's longer than 30sec per lap, 30sec is the steady part.

I am starting with front end height/angle. I'm tilted a little now, trying a bit more.

All wind tunnels are different, but in my experience the measurement time for a single yaw angle is in the 2 minute range, so even the most controlled environment requires some time to average drag of a constantly moving body.

If you are doing Chung testing it helps to have multiple laps so you can more easily and accurately level the VE profile.

Notio sensor. With a custom home DIY out front rod mount to get it away from cockpit and hands. Like one of you experts posted once.

No biggie, just means driving to my loop instead. There is one town over a looong traffic circle in a park I can hold 28mph on due to gentle curve. I was hoping to use the street at house to save drive time.

note: PM me if you're interested in a never used Notio sensor

---

wovebike.com | Wove on instagram