Technical FAQ: How much drag can we eliminate by dropping two bearings?

In Technical FAQ, Lennard Zinn elicits a detailed response from Friction Facts on the drag saved by reducing a rear hub to two bearings

This week we’ll tackle a reader question on wheel bearings and drag, specifically targeted for the rear wheel. The in-depth answer from our friend Jason Smith at Friction Facts should give you plenty to chew on until next Tuesday.

How much drag would we save by dropping two rear-wheel bearings?

Dear Lennard,
I have a question about bearing drag and the watts required to spin the bearings. I know you do a lot of work on bearing seals and such, and figured you may know the answer to a question I have.

We are looking into a new hub design and when pedaling the hub would only have two bearings spinning as opposed to the normal four. When coasting, all four bearings would be spinning. Would you happen to have any idea of how many watts would be saved by having two less bearings spinning on a rear hub? I’m assuming standard steel cartridge bearings (not ceramic).
— Boyd

Dear Boyd,
Jason Smith at Friction Facts thinks about questions like this all of the time, and he has answered you very thoroughly below.
― Lennard

I am having difficulty visualizing your reader’s design specifics for his hub, so I’ll answer as best as I can using a few assumptions.

Based on the data from previous bearing tests, a basic analysis can be performed to estimate hub drag due to the bearings.

By referencing the Friction Facts bottom bracket efficiency test results, we know the drag found in sealed cartridge deep groove ball bearings is primarily due to three factors: seals, lubricant viscosity, and the ball-race interaction. Interestingly, the drag contribution from the seals and lubricant was slightly higher than the contribution due to the ball-race interaction.

For the most part, the drag due to the seals and lubricant is not dependent on load, yet is dependent on RPM. That being said, the drag due to the ball-race interaction is dependent on both load and RPM.

To analyze hub drag in general, it might be best to start with an analysis of a front hub — a simple symmetrical (non-disc brake) configuration with two sealed cartridge bearings under load. The assumptions are as follows: 1) The load on the hub is 100 pounds (50 lbs on each bearing);
2) The wheel is spinning at 314 RPM (about 25 mph); and
3) Drag from any external hub seals/dust covers (aside from the bearing seals themselves) is a factor but not taken into account in this analysis.

Based on the bearing efficiency data from the bottom bracket efficiency test, using similar loading conditions yet adjusting for higher RPM of the wheel versus crank speed, we can speculate that the two hub bearings in this case would consume anywhere from just under one watt to over three watts, with the variance based on the quality and efficiency of the two bearings installed in the hub. For this conversation, let’s assume very high efficiency bearings are installed in the hub, which would create one watt of total drag (0.5 watts for each bearing). Of the 0.5 watts of drag created in each bearing, possibly 50 percent of the drag is created by seals/lubricant and 50 percent is created by the ball-race interaction, based on previous test results.

Drag analysis of the rear hub gets more complicated. More than two bearings are present. Typically the main support bearings are of a different design than the cassette bearings. The loading conditions vary based on the rider’s instantaneous power output (the reactive force of the chain tension increases the total load seen by the main bearings). The RPM of the cassette bearings varies based on coasting or pedaling conditions. The asymmetrical lateral placement of the load bearings within the hub can produce cantilever effects and increase the effective load on the main bearings.

But we can get to a rough estimate of rear hub drag using some assumptions. Assuming a 250-watt rider creates a steady state 50lb chain tension, a cantilever loading factor of 1.2, 100lb load, 314 RPM, all cartridge bearings, then the drag produced by the two main bearings (high efficiency bearings) would be approximately 1.5 watts or thereabouts.

If the rider were pedaling, the cassette bearings would not be spinning; therefore, these bearings would not contribute to drag.

If coasting, the main bearings would see the chain tension portion of the load drop, but the drag due to the cassette bearings spinning would enter the picture (and the drag due to the ratcheting pawls as well, but this is not covered in this analysis). I speculate the drag due to the cassette bearings during coasting would be about 0.3-0.5 watts, mostly due to the lubricant viscosity, and not the ball-race interaction.

It is most likely a wash — the drag contribution due to all of the bearings in the rear hub combined is similar whether pedaling or coasting.

Based on this analysis, for a bike with high-efficiency bearings installed in the front and rear hubs, the total drag from both hubs is likely around 2.5 watts. Hubs with lower efficiency bearings would most likely create about five-to-six watts, given similar conditions.

Back to your reader’s original question — for the loads and RPM seen in a bicycle application, a decrease in the number of bearings supporting the same load would decrease the overall drag seen in the hub.

The total drag due specifically to the ball-race interaction would be similar, regardless of the number of bearings used. This is because adding bearings given the same overall load would decrease the drag in each individual bearing, yet because the number of bearings increases proportionally, any savings would be cancelled out. However, as mentioned earlier, the total drag due specifically to the seals and lubricant is fixed and does not vary with load. Therefore, adding bearings will increase this type of drag, and conversely, decreasing the number of bearings will decrease the total drag due to seals and lubricant.
— Jason

Feedback on last week’s column regarding SRAM lever replacement

Dear Lennard,
Lennard, you can buy SRAM singlespeed brake levers to get a replacement body for a Doubletap lever. They are easy to disassemble and reassemble, as the Doubletap mechanism is so simple.
— Gordon Ong, City Bicycle Works
Sacramento, Calif.

Dear Gordon,
Thanks for that suggestion, and thanks to the other readers who sent this same suggestion to use the composite lever body from this lever. Another reader sent me an eBay link.

On page 134 of Zinn and the Art of Road Bike Maintenance, 4th Edition, I have an exploded view of the guts of a SRAM DoubleTap lever.
― Lennard