Rolling Resistance: Hubs
 

Much is made of the rolling resistance of tires. But what about hubs? My hunch is that there's significant difference in the rolling resistance of hub bearings. Also, in cassette bodies. Does anyone know of any data to compare various hubs, and how much resistance increases with wear? I think there is more to it than spinning your wheels when the bike is on the stand.

Bearings are pretty efficient even when poorly adjusted or worn. And most of us will know when they are too worn.

I don't think there is as much there as you believe.

Nope, very very very very little resistance for anything but the most worn, filled-with-sticky-dirty-old grease bearings. Very very very little difference between loose balls and good cartridge bearings. Same with ceramic bearings. Almost all are hybrid (ceramic balls, steel races). They are engineered for very high rpm (not a bicycle), clean/hot (not a bicycle) and a pure radial load (definitely not a bicycle). The reason you seem them spin like crazy in videos is because they have no-contact seals and no to maybe a little super thin grease.

As mentioned the seals are going to create the most drag which is barely measurable between just the different bearings themselves. New seals will also generally have more drag than used seals after they've been "broken in". Plenty of comparisons on the web between a hub that has quality, new seals and one that has expensive bearings of some sort but has crappy seals which result in less drag when new and claims that the bearings are better. Agreed that hybrid ceramic bearings even the full ceramic $$$$ types are a waste of money for cyclists.

Story told by a Bicycling! magazine writer who had been sent to cover the 1977 Tour de France: he was strolling around the mechanics' area before one stage and picked up a front wheel from the bike of Bernard Thevenet, who went on to win that edition of the race. To the writer's surprise, the hub of the wheel felt crunchy, indicating that the cones were too tight. He told the mechanic, who glanced at the wheel and said (in French, it is to be presumed), "Doesn't slow him down."

I now realize that the mechanic was annoyed by the writer and probably adjusted the hub after he left. Still, it's likely that, given the effective lever length represented by the distance from the tire to the bearing surfaces, a hub could be very tight and still have almost no effect on speed.

The rolling resistance in even half-decently adjusted hub bearings is insignificant and the resistance of a freehub body is extremely low when coasting and zero when pedaling. My front wheels (and rear when the chain is off the cogs) will spin for a long time when given a push and will "pendulum" to a stop just because of the slight wheel imbalance. The makers and sellers of ceramic hub bearings have made a lot of money convincing riders that they can save 50% of the friction losses in hubs while not admitting they save 50% of an extremely small number.

I'll add to the pile-

For basic bicycle design, mechanical function and human interaction I suggest Bicycling Science (The MIT Press): Wilson, David Gordon, Schmidt, Theodor, Papadopoulos, Jeremy J M.: 9780262538404: Amazon.com: Books Andy

+1 this. Unloaded spinning in a stand tells you nothing about loaded rolling resistance on the road. The rolling resistance of any decently maintained and adjusted hub will vanish in the noise of tire rolling resistance, road conditions, etc.

rr-hubs
 

thanks all for your observations. Even so, it would still be interesting to see this tested out on a dyno, with a load on the saddle. FWIW, this notion arose after replacing some cheap wheels with a set of Pacentis with glassy smooth White Industry hubs. What an improvement! Not sure, however, how much was due to lighter, wider rims.

dunno but when a shop changed my rear wheel & hub of course, I picked up a solid 1mph to my avrg speed

To Andrew R.
just ordered Bicycle Science--thanks for the tip!

Ok, it was definitely the wheel. :notamused:

If it is one of your listed bikes, I’d guess the Trek 540 would be the “cheap” wheel bike. Even still 1mph gain is a lot to pick up with just a wheelset change. I’d still imagine weight is a bigger influence over hubs.

Another story if the wheels were truly cheap Wheel Master level quality vs White Ind wheels.

John

With your bike up in a repair stand, try pressing your fingertip against the hub seal where it meets the axle while the wheel is spinning freely. You'd have to press very hard to make even the slightest perceptible difference in wheel speed. No hub without major damage is going to present more than a tiny fraction of that much resistance.

I would think the weight of the hub plays a part rather than bearing resistance. I had a pair of hed ardennes plus and also farsports 50mm rims on dt swiss 350. Both front wheels felt about the same (no scale just using my hand to hold). But the rear wheels had significant weight difference with the hed being a little heavier. Although the farsports are more aero I doubt the 10watts saving or so could give me a 1- 1.5km/h increase of average speed. Last I calculate 10watts gives less than a 1km/h advantage. I could be wrong

For our use sintered bronze bushings would work. At 30mph bike wheels turn at about 350 rpm.

In my experience; old school serviceable bearings on solid axles perform better than any of the reasonably priced quick release axles. You can test this by cleaning and lubricating your front wheel bearings in a quick release wheel, reassembling and spinning whilst off the bike. Use approximately the same force when you lock it into the fork and see how long it takes to get to a halt. I switched two of my rosd bikes to solid axle solid tyres because of this. It's obviously not practical for off road cycling.

Sounds like you did all that instead of setting the QR hub slightly loose so it achieves proper bearing preload when the QR is closed.

That isn't some theory - every bicycle service manual will tell you to do this.

True. I'm just suggesting that the variability in proper preloading presents a problem for QR systems.

You can screw up the preload on any bearing. But it is not hard to get the preload on QR hubs right by setting them so play just disappears when the QR is tight. The axles can only compress so much, so the system is not that sensitive to variations in QR tension.

Did you really replace your QR axles with nutted just because of this?

​​​​​​

Not just because of this. On two road bikes that I also run solid tyres on I have though. There's an added assurance of not having to remove the wheels to repair flats

have not tested this but I think the hub's friction is a small contribution to the overall drag and resistance encountered by a bike. Speculatioin is that radius of the wheel (hub to ground) moots all but the most extreme cases of high rolling resistance. I could imagine a test that measures the temperature of the bearing - comparing the best possible hub in perfect adjustment with the best grease - to a worst case bearing. See if there are any friction losses that can be discerned with temperature. Portlandjim might have done an experiment like this in the Specialized lab, they had a rolling road test rig I believe.

In order I think the sources of drag and resistance on a bike are

Aerodynamic drag
Tire rolling resistance
and driveline losses

also note that aerodynamic losses become a bigger part of the overall drag budget as the speed increases. Draq increases as the square of speed, and the power required to overcome it rises as the cube.

/markp

This may be because of axle compression (slight decrease in length between the locknuts) by the quick-release when it is closed; this can put more than optimal pre-load on the bearings, if they're not properly adjusted in the first place.
Axle compression between the locknuts does not occur in solid-axle hubs as a result of tightening the mounting nuts.
The axle type should make no difference if the bearings are properly adjusted (not an insignificant "if").

would be worth an experiment to detemine if the axle "compresses" or does it bow ? I think the latter.

I'm in agreement with the rest of your posting however. The old school mechanics that I learned from would adjust the hub with just the slightest play, so that the play was taken up as the quick release was tightened. I agree that's not a factor with a solid axle with lock nuts.

/markp

Hi, Mark,
This has been discussed a lot, but the material science is clear: actually the axle compresses, while the skewer stretches. Steel is an elastic material, and like all such materials responds to stress (within limits) by expanding or contracting.

Bicycling Science, Fourth Edition, by Wilson and Schmidt (The MIT Press, 2020) has this to say, and describes the experiment you suggest:
"A bicycle wheel's quick-release skewer (the through-axle tension rod used to secure most modern wheels) applies considerable compressive force to the wheel's axle, shortening it by 0.02-0.04 mm and thereby "tightening" the bearing adjustment. The effect of this shortening and the resultant tightening can be felt by placing some washers on the axle to take the place of the bicycle frame and squeezing them with the skewer as if the wheel were installed. (This experiment will not simulate any bearing load that might arise from axle bending due to operating loads or preexisting frame misalignment, however.)" Pages 272-273

Jobst Brandt (quoted on Sheldon Brown's web site HERE ) had this to say in 2003:
"Wheels with quick release (QR) axles present an additional problem in that closing the QR alters bearing clearance. Closing the lever requires increasing manual force with a slight over-center feel near the end of the stroke. This lever force arises from compressing the hollow axle and stretching the skewer. The ratio of elastic length change between axle and skewer is that of their cross sectional area and active lengths."Although small, axle compression on QR hubs is large enough to alter bearing clearance and should be considered when adjusting bearings."

FBinNY has mentioned the bending effect of misaligned dropouts and operational loads as well. As to bending, a cylindrical piece of an elastic material (such as a bicycle axle) experiencing bending will elongate on the outside of the curve, and shorten on the inside of the curve.
These changes are very small and hard to measure... but they can be measured.
I hope this helps settle the matter. :thumb: