I bought a Gibli last fall while preparing for my World Hour Record last fall, determined to chase down every last watt. Over the last five years I have done tons of tunnel and track testing, I felt my setup was already close to peak aero. But I thought having a CdA sensor gave me the opportunities to test more gear and more combinations in real conditions to see what combinations work best.
I pulled out old equipment, bought new pieces, and ran countless tests: helmets, skinsuits, shoe covers—you name it. I expected to maybe uncover an extra 1–3 watts. Instead, I found 16 watts over my already optimized setup. And this was without touching my position. Side note: don’t put too much stock in vendor-provided tunnel numbers.
Which brings me to the bigger question: why aren’t we talking more about CdA sensors? At this point, I see them as every bit as essential as a power meter. As a Masters athlete, I don’t have much headroom left to produce more power. If I want to get faster, I have to get more aero—and I know that’s true for a lot of people here as well.
Not all of us have time for testing. I know it does sound weird since we’re all training a bunch and so the easiest gains are in aero testing rather than training.
But in my case, all of my riding is either on the trainer, on my commuter to/from work, or rides with specific goals (e.g. a long ride, or social ride). For me, with a family, finding the time to carve out an hour here or there where I can tweak this or that while in full race garb is the hard part. Its easy to ride at 6am indoors when its dark and snowing - harder to carve out an hour on Saturday afternoon away from the kids. That, and I live in Canada, so the number of months available to bike outside is already reduced.
Though I’d love to hear if anyone else in my position has been able to figure out a process.
I have a GiBLI and have made a few runs. Plan to keep going with this and would like to know what you did to save the 16 watts? How many runs have you made?
I have Aerosensor and have found some interesting things for sure.
For me helmets seem to matter little. I have tested all the latest stuff and use Met Drone 2. Rudy is so heavy but I will continue it as an option.
For me r28 neo socks and baselayer matter a lot! I am not sponsored a single bit by r28. I have also tested their new neo kit vs my kit with baselayer and it seems to be another marginal gain in my testing so far.
Still have a Notio. I hear better things about the current sensors. Sure. But until it can do laps of data with zero post ride laptop work and you don’t need a near perfect road and wind day. ……it can be a frustrating venture.
I mean literally like if I have a rolling couple mile lap with plenty of wind I can still do it and it gives reliable lap results with zero tweaks.
Well, I for one am glad this topic was raised. I will be looking for a sensor sometime this spring. Tired of guessing the effectiveness every change I make to my setup actually has on the road.
I have one and my experience has been mostly more frustrating and time-wasting than useful, for the investment in money and more importantly in time I’ve put in.
Don’t get me wrong, I had a few testing sessions that have been useful but they were for changes that made a larger difference. Like a 15 deg aerobar tilt was a lot faster than 21 deg tilt for me by like 10-2 watts IIRC. That was clear cut.
For things that changes 1-3-ish watts, I feel like I could not trust the results 100% in the sense that the uncertainty/repeatability level was of the same range. That’s after taking some decisions on the data analysis task that can swing things one way or another.
And even if I felt I could make a call, it would not necessarily repeat another day. And maybe part of it is the conditions of the day (i.e. yaw).
I do not know if it comes down to the sensor itself (aerosensor), my protocol, experience or expectations, choice of testing venues… or simply coming down to typical weather conditions here.
It is always very windy and gusty here, I do not have access to an oval track. So I’ve been using the out-and-back method. It does not take much wind/gusts for the calibration factor to vary enough to cause issues in data analysis for the Aerosensor.
Low wind days are very rare here (Southern Ontario Canada) and even if I watch the weather carefully for opportunities, it rarely fits in my schedule when it happens, or low wind time periods happen in early morning or evening when the temperature changes faster.
Also difficult to find a good long stretch of road with almost non-existent traffic and could not find a good place with trees for wind protection within reasonable driving distance.
It’s pretty much all wide open fields here for quiet roads.
So I’ve had quite a few times where I spent 3-4h to test (including prep, driving to the place, etc) only to find out that the results were inconclusive or that the difference between the setups were small and within the uncertainty range.
I only have experience with one sensor, but my take is that you need to be careful about a lot of things, be meticulous and invest a lot of time to get useful results when you get down to smaller gains. This is where experience can make a huge difference I am sure.
So I think this is definitely not for everybody. I am a mechanical engineer by trade (so somewhat familiar with testing protocols) and it hasn’t been a smooth road for me.
The 16w were saved just doing testing. For example; we have seen a lot of chatting about tire and wheel combinations and what is the fastest. You can’t just look at this in isolation, the fork and downtube and the rest of the bike play a role. You need to test the whole system to determine what is the fastest.
I did a lot of my testing on interval days, so it fit right into my plan. I have done only done 10 or 12 days of testing.
I can completely understand your frustration. I have been testing for 11 years via different modes, and found many times the juice is not worth the squeeze. This even includes tunnel testing when I found my baseline at the end didn’t match my starting baseline.
With the sensor I’m using, its been very stable even with strong cross winds. I tested helmets on a day with 25mph crosswinds winds to compare to a windless day. The results of which was faster were the same but raw numbers were different, and what I would have expected from each helmet at high yaw angle. For me traffic has been the biggest issue, a car zooming by at 50mph really screws the CdA reading for about 20-30 seconds. You can’t have too many of these before the run is trash. I have had other frustrations, the unit I use allows you to set a geo start and end point, so went you enter the start point the recording start and when you past an end point the recording stops. I have had a couple of time where this didn’t work and had to re-due runs. This is likely not the sensors fault but the phones gps. Other issues were finding a good location to test, and the mounting setup of the system.
With the minor frustrations aside, I have found my unit to be completely consistent with results even down to a watt or two which has been extremely valuable at improving my CdA. Having an app that allows you to see test results while in the field and even watching your live CdA as you ride are reasons I purchased it. I feel like, I have only started to capture ways to improve my CdA, so far, I have only done equipment testing. I’m the same size as Dan Bigham and he has a sub .16 CdA, Over the next year, I want to get closer to that.
What CDA are you seeing? My CDA with this setup is .285 ish:
XL Speed Concept w stock cockpit & (2) 7 degree angle spacers, Zipp 808 Firecrest and tubeless 28mm Continental 5000s TLR, Castelli PR2 Speed, Rudy Wing Dream, with no aero sox. Also front bottle between the arms and two bottles behind the seat on a 45 degree angle.
I have recently bought a Gibli and will be doing first test runs as soon as weather allows. It’s a shame there’s no publicly available advice (‘rules of thumb’) on how to go about testing (beyond what is in threads here and social media groups; all of which is useful but falls short in aggregate of being a guide or framework to follow). Perhaps it’s out there and I missed it?
I found good advice / info on suitable courses and Chung method. But little to nothing on how best to establish an initial position (low stack, short reach vs. high stack, long reach; etc) and how best to approach testing changes to different combinations of pad stack, reach, pad width, hand width and tilt. (And likely saddle position changes required as a consequence of other changes).
Maybe there’s too much intellectual property value at stake for those who know (and not a large enough market to make it worth putting in a book)?
Anyway, I’ve written up an approach for myself that I will try but seldom I have I felt so like I was shooting in the dark! Would be delighted to compare notes with anyone interested in being open about sharing details of their testing strategy.
Sorry if it wasn’t clear - I mentioned it later on, but the name of the sensor is not helping haha! It’s the one simply named “aerosensor”.
There’s a lot of people that seems to be getting good data with it. Like I said I think a big part of my limitation is finding the time and the fact I do not have easy access to good testing grounds in good conditions for out and backs. Having access to an oval track would be good to log laps.
I do not have experience with other sensors to be able to judge how it compares.
They are not cheap, but less expensive than a top preforming wheel, new frontend, a skinsuit and helmet and God forbid a new bike. Knowing what I know now after using my sensor, I would purchase it before purchasing any of the above.
