Measuring bike efficiency

I’m just watching a GCN show - Si’s Hyper Bike KOM Challenge which is mostly about drivechain efficiency and I wondered how I’d test the efficiency of my bike - from pedal to road?

Has anyone done this?

I’m thinking you need to measure power at the pedal and at the road to do this…I have pedal power meter, but I can’t think of a meter that measures power output on the road…presumably some kind of treadmill based tool for bikes would be needed?

There’s two reasons for this, curiosity about the power loss through the bike, and a certain paranoia that something in the way I’ve built my bike is costing me watts :open_mouth:

You need two data points. One on the input side, before the drivetrain (Crank > Chain > Cassette > Hub) and one at the output side. You need to measure and record at both points, then repeat for different test conditions to see what difference, if any, there is in the test states.

The obvious option would be pedal power meter and hub power meter. The issue is more about the accuracy of the measurement tools to get any real resolution that you can trust for accuracy.

A typical power meter is +/-1.5% accuracy, and that is notable, when we are likely talking about efficiency difference of 1-2% total for some differences (clean vs dirty chain).

The reality is that unless you have lab quality gear, you probably can’t get any meaningful data. You are better to trust the lab testing from places like Friction Facts.

However, the notable issue now is that FF was purchased by Ceramic Speed, and I see a potential conflict of interest from that arrangement now.

What luck! Here’s what you were looking for. :rofl:

Some quick google:

Again, you probably have little if any chance of learning more than what is lab tested.

1 Like

Thanks, but I’m thinking about the whole system not just the drivechain.

Labs tell you about the average you might expect, not what your bike is actually achieving.

Ha ha, yes that could do the trick…not sure I have $7999 lying about though. Maybe if I can find a company that owns one I can use…

Sure, I get what you want to know. Adding the wheel/tire and such is even more tricky.

We are talking about mere watts here from one change to the next. It may well add up to something more when you combine multiple variables (chain, tire, tube, etc.), but you still have to measure it in some meaningful way with enough resolution and control to actually KNOW that the variable changes are the cause of the differences.

As a whole, rolling system assembly, the only way is via the input (pedals) and output (speed) in an extremely controlled setting (indoor lab). You have to apply a known and stable power, then see what the outcome of that is from the system. Then you change one variable, retest with the same input, and compare for any differences. It’s time consuming and tedious work that must be done well to get good data. Not a small undertaking.

You might be able to use something like rollers and a very precise speed sensor to measure that with a known and stable power input, but I still think it will be questionable from an accuracy standpoint.

1 Like

Yeah sure, I agree with you there.

I was just wondering whether rollers might do it too - are there any out there that display watts or perhaps have a resistance setting that is known to be say x watts at a given rpm?

It’s just a thought experiment for now.

1 Like
  1. Reliable pedal power meter installed on bike.

  2. High accuracy speed sensor on the bike

    • This can be a regular magnet speed sensor, but increased accuracy with more magnets installed. If you do 2 magnets, the speed sensor is set to 1/2 the actual wheel circumference. You can (and should) use more magnets if needed and adjust the wheel circ as appropriate. The more times you trip the sensor, the more likely you will catch a speed change.
  3. Pedal at a determined power input, and a specific gearing (to be repeated for all tests).

  4. Record the wheel speed.

  5. Change one variable, and repeat the test with the same input (within reason) and record the speed. You may see an increase or decrease in speed in accordance with the variable change.

Do any of the “wheel on” smart trainers measure power at the roller?

1 Like

Yes, but they all have no better than +/-3% accuracy (some are +/-5% or worse), so it is not a great tool for the purpose.

For perspective, that means a 200w effort could be 194w to 206w as measured. That’s a 12w swing with just the measuring device. If you are hoping to see if one tire is more or less efficient, you will only be able to see ones that are largely different, at best.

The accuracy of the tool matters as much or more in this setting.

Sure I’ve posted about the accuracy of power meters myself a few times - the inaccuracy in the roller meter is mulitplied by the inaccuracy of meter at the pedal too.

However there are other stats to consider. The likelihood the both my pedal meter is at maximum positive inaccuracy, and the roller at maximum negative inaccuracy (thereby giving the maximum discrepancy in readings) is quite small. It’s far more likely that neither are massively inaccurate - I’m comfortable with that level of confidence in the readings.

Which wheel on smart trainers do you know nearer the 3% accuracy? I’m only finding ones that do virtual power for some reason.

The Snap claims +/-3%.

All of these current wheel-on trainers don’t use a strain gauge based PM. They use the electrical system and flywheel speed to estimate power to a relatively high degree of accuracy.

As to float within a tolerance range, sure, there is a tendency to be on one end or the other. But the reality is unknown and should probably not be assumed when we are talking about differences that fall within those ranges. We will get a number every time, but the accuracy of that is in question each and every time.

It is especially so with something like a wheel-on trainer that relies on the system of bike, tire, wheel, related bearings in the bike and trainer, and the pressure of the roller and tube, to get to a value. Change any one and you may or may not be getting good data.

Testing like this can be fun and enlightening, but only if you have real confidence in the tools and the methods of testing.

1 Like