Comparing two rides avg. speed vs. power - 13.5% more power = 2.5% faster speed?

So I just did two of the exact same 10 mile ride and want to understand if the power to speed numbers make sense. The only change between the two in terms of equipment was different tires. They were on different days so weather conditions varied, but were similar enough. Neither was particularly windy, similar temp, but the second ride was significantly more humid and a bit warmer.

Ride 1: 20.0 mph avg. with 215 watts avg. power.
Ride 2: 20.5 mph avg. with 244 watts avg. power. on this ride I was in the drops/more aero position quite a bit more, but can’t quantify this obviously.

So my question/confusion is this…given the 13.5% power increase from ride 1 vs. ride 2 I would have thought that the speed increase would be more than 2.5% faster. I know the speed vs. power relationship isn’t linear and there are other variables, but as I stated, many of them are marginal differences in my view. I’d argue that conditions on ride 2 should have set up faster all else equal given the warmer/more humid weather and increased time in the drops (this was a solo race…no drafting).

Any thoughts on how to think about this?

Power and speed are not linear….as you start to go faster, the power required to go an extra 1 mph increases exponentially.


It seems reasonable to me - and there are so many other variables that need to be looked at.

I read an aero article that talked about the length of time it takes for ariflow to settle down around you after you change positions for example - reaching and drinking from a bottle meant a very surprisingly long impact on airflow even after you got back into position. So, its very hard to compare unless everything is tightly controlled, but 0.5mph is a significant bump when you’re already at 20mph.

Just to make sure, you did calibrate your powermeter before each ride, and it is the same one?

To help this thread go into the weeds…riding two different tires could easily account for the difference. A change as little as .001 Crr is about all it would take.

Further, what you think are “marginal differences” and don’t matter actually do matter. Having tried to tease out power/speed issues before unless the conditions are flat, no wind, exact same temp/humidity 0.5mph for that average power is just noise. One little rise in the road if you don’t apply power the same or a little breeze even for a moment will destroy speed.


Yes, calibrated and same one.

1 Like

I think @Landis has nailed it above.

I hear you. These were my Garmin to Garmin results for cleanest comparison. The race had me at 20.1mph. Many of the fastest riders were in the 22-23mph range. Feels like a 2-3mph jump to me would be insanely higher watts. Not suggesting I am the fastest rider or should be in that range, just I am struggling to reconcile the results. Maybe it just is what it is. Seems I’d need to put out 350-400 watts avg. to hit those type of speeds. I am about 160-165lbs for reference.

Just trying to understand…this was comparing a race to an individual effort? Or two races?

Changing frontal area or drafting results in huge variation in power/speed relationship…

Were these 10 mile TTs on the same course? Out and back?

What was your time difference, apart from the speed? Being 20W down cost me a minute.

Also…its likely you don’t need to ride at 300-400W, you need to be more aero.

1 Like

Bike section of a Triathlon. No drafting allowed.

1 Like

Perhaps, but I was fairly aero…and more so on ride 2 for sure. This point won’t help compare my two rides, but may play a difference between my speed and others. I am assuming the fastest times all had aero bars. I don’t have a tri bike, so no aero bars…I just mostly rode in the drops. No idea how big of a difference that could make…

Play with my and you’ll find that being more aero will be a bigger factor than power. Since I’ve changed position and concentrated in staying in it, I am less powerful but I am relatively substantially faster.

1 Like

I won’t pretend to know, since I also have plenty of instances where the differences in watts don’t explain all the differences in speed, including some where they move in opposite directions. Take heart: it implies there’s a lot of stuff to help you go fast beyond increasing your watts.

One possible way to test the reasonableness of your underlying expectation is to use Best Bike Split and see what percent change in speed it predicts from a 10% increase in output. It won’t explain reality, but at least it’ll tell you what theory would predict.

For comparison - how much do you weigh?

It’s interesting because I often ride at close to the same speed - 19.5 to 20.5 mph and rarely exceed 190-210 avg. watts over call it 30 mile rides with similar elevation profiles.

All this to say that I was super happy with my average power (Garmin said I had my highest 20 min average ever, which I believe), but surprised by the relatively low speed.

I did a solo century last year in which I averaged 227w/230NP and averaged a little over 20mph, and in a recent race where I was basically setting the pace on the front on the first lap and I averaged 22.2mph on a loop at around 260w avg/262 NP both relatively flattish loops but subject to winds (the century loop being coastal)

Gonna 2nd the suggestion to try mywindsock, I find it more accurate than best bike split for some reason and it seems to align with me pretty nicely with the default cda settings

As a science teacher the equation you are looking for is Ek= 1/2mv2 or kinetic energy is half the mass x the velocity squared - mass stays the same but in m/s 20mph means squaring 8.9 and 21mph 9.4…increase the speed a bit and the energy needed increases quite a lot!..of course wind, air pressure etc also vary but velocity is key! :grinning:…and if your really bored a watt is a J/s so you can covert to power as well… :grimacing:

So I just paid for my windsock. Seems interesting, but not entirely sure how to use this. Is it just a matter of trying different things on the bike and comparing results/data afterwards?

The cda stuff says it needs accurate rolling resistance and drivetrain resistance data. No idea how to figure that out…

Any thoughts?

Hi. I have many years of experience doing Robert Chung field analysis to calculate CdA.

Various factors can impact your final speed. In my experience testing over many years in different parts of the USA, it is uncommon for 2 days to act exactly the same. It can “feel” the same but when you measure it, it’s not at all.

First is Air Density (rho). A very fast (low rho) day versus a slow (high rho) day can create a ~1mph difference, based on speed.

You mention humidity. Wet days can greatly impact rho. Interesting that you needed more power on your high humidity day, because typically this can create a faster day (counter intuitively) so there are other factors at play.

Anyway, without measuring it, we are all just making guesses. For quick and dirty, you can look up Wunderground history and look up Rho for both days you tested. Throw those number into’s rho calculator and have a look for yourself.

Second, Rolling Resistance (Crr) from tires make yet another ±1mph. Being fair, it’s closer to a 0.7mph difference going from puncture resistant extremely slow tires to race tires. But these are nominal extremes. And once again, this depends on speed tested, pressure you pumped up, and if you actually took the same lines or not.

There are MANY other factors you may not have controlled for. PM Calibration, your position, your kit, when you changed your tires did you go from better to worse air tripping treads, system weight, where you applied your power if the course is even the slightest bit rolly… invisible slipstreams and headwinds created by cars & trucks (yes, it shows up in CdA data and you must throw these sample sets out), was the wind REALLY as similar as you remember?, and this last one doesn’t apply to you but does to others who I have seen test: self-test bias.

My recommendation is to consider the factors I mentioned above, figure out which variables require tighter control, be very diligent with note taking of all things that CAN be measured, then retest dozens of times (throwing out bad runs) until you can extract clean data from your dataset.

It’s a ton of work, btw.

But without a clean dataset, a lot of A vs B comparisons are too noisy to draw a meaningful conclusion. Good luck!