+2.2% efficiency, measured while pedalingTests carried out at MSPORT Olgiate O.
The treated Shimano chain, tested in two different conditions with three measurement systems. The result is always the same.
53×16 · −2.2%
53×16 · −2.3%
53×19 · −2.2%
Of everything fitted to a road bike, the chain is the component that gets argued about the most and measured the least. Frames have wind tunnels, wheels have rolling-resistance test rigs, tires have published Crr curves. For the drivetrain, more often than not, all that is left are impressions: it runs better, it feels smoother, you can feel it.
The problem is not a lack of curiosity. It is that measuring a chain is hard, for one specific reason.
Why a chain is so hard to measure
A bike's power meter sits upstream of the drivetrain. Whether it is in the pedals, the crankset or the hub, it reads the watts the rider puts into the system, not the watts that reach the ground. Between the two values sits the efficiency of the chain, and that efficiency appears nowhere on the bike computer.
Worse, a chain does not have a fixed efficiency. In one of the few serious scientific studies on the subject, published in a mechanical engineering journal, a group of engineers at Johns Hopkins University showed that the efficiency of a chain drive depends mainly on how taut the chain is and on the diameter of chainrings and sprockets, and that the largest share of the losses is not friction turning into heat.
Two chains can run equally warm and still have different efficiencies. That is exactly what we found.
To isolate the difference between two chains, then, you need a rig where the external work is imposed from outside and not chosen by the rider. If the resistance is fixed, the power read at the pedal becomes an indirect measure of drivetrain efficiency: the more watts it takes to do the same thing, the worse the chain. We built two independent tests on this principle.
If the number changes, is it the chain or the instrument?
The first test answers the most obvious objection: what if it is just the meter telling stories?
On a Specialized bike we fitted two measurement systems at the same time, Garmin Vector pedals and an SRM crankset, recording the same pedal stroke in parallel. The bike was placed on resistance rollers with a constant load. Between sets, only the chain was replaced: same frame, same wheel, same gear, same rider, same session.
- Rig
- Resistance rollers, constant load
- Gear
- 53×16 · cadence 80 rpm
- Repeats
- 3 per condition
- Measured with
- Garmin Vector + SRM, in parallel
Mean power by condition, three measurement systems
The line joins the two conditions; the thick translucent bar is the standard deviation of the repeats.
| Meter | Stock | Shimano | Δ |
|---|---|---|---|
| Garmin Vector | 181.3 ± 2.5 | 177.3 ± 3.5 | −4.0 W |
| SRM | 186.3 ± 1.5 | 182.0 ± 3.6 | −4.3 W |
What is interesting is not so much the number as the fact that the two meters disagree with each other in absolute terms but agree on the difference. Vector and SRM read the same pedal stroke with about 5 watts of systematic offset. But when the chain changes, both move in the same direction by the same amount.
A calibration offset behaves like this: it stays constant. It does not change because you changed the chain. The fact that two independent instruments record the same 4-watt change is the strongest check this test could offer.
Cadence and heart rate stayed comparable — 80 rpm in both conditions, 135.7 vs 134.7 bpm for the Vector and 136.0 vs 134.7 for the SRM — which rules out the difference coming from a different way of pedaling.
To confirm the result in a completely different test condition, we moved on to the second test.
Removing the roller calibration from the picture
The second test removes the roller and its calibration: we moved to a cycling treadmill, with speed and gradient imposed by the belt.
This is the key point of the protocol. On a treadmill at fixed speed and gradient, the work against gravity is determined by physics, not by the rider. The rider cannot push harder: push harder and you accelerate, and the speed is no longer 20 km/h. The system settles on a single value of power at the wheel. What varies is how much power it takes at the pedal to produce it.
- Rig
- Treadmill, 20 km/h · 5% gradient
- Mass
- ≈ 68 kg total
- Gears
- 53×19 at 57 rpm · 39×19 at 78 rpm
- Readings
- 6 per condition
The second gear was there to verify that the effect did not depend on the chain tension of a single combination.
Individual readings on the treadmill
Each dot is one reading. Values repeat, so the dots stack.
| Stock chain | Shimano | |
|---|---|---|
| Mean power | 193.75 ± 0.89 W | 189.50 ± 0.93 W |
| Coeff. of variation | 0.46 % | 0.49 % |
| Actual speed | 20.0 – 20.2 km/h | 20.0 – 20.2 km/h |
| Chain temperature | 24.5 °C | 25.5 °C |
| −4.25 W (−2.19%) · 95% CI 3.3–5.2 W · ES 4.69 · p < 0.001 | ||
Note the repeatability: coefficients of variation below 0.5% in both conditions. On a rig like this, measurement noise is minimal, and the four watts stand out clearly.
And then there is the temperature. One degree of difference, with the more efficient chain slightly warmer. If the difference were just friction dissipated as heat, we would expect the opposite. It is the same conclusion as the Johns Hopkins study: heat does not explain the efficiency of a chain. The gain comes from something else, most likely from the way the links articulate under load and from the elastic deformation at every meshing.
What four watts are really worth
Four watts out of 194 do not sound like much. In energy terms it is even less: over four hours of racing that is about 61 kJ, fifteen kilocalories, a quarter of a banana. But energy is the wrong metric. What matters is time.
| Situation | Speed | Gain |
|---|---|---|
| 5% climb, 10 km | 16.6 → 16.9 km/h | ≈ 38 s |
| 8% climb, 6 km | 11.5 → 11.7 km/h | ≈ 38 s |
| Flat, 40 km | 33.8 → 34.1 km/h | ≈ 34 s |
These are estimates from a standard power model, and the order of magnitude holds whatever assumptions are made: the speed gain stays around 2%.
Half a minute on a thirty-five-minute climb. In a hill-climb time trial, that is the difference between the podium and fifth place.
In summary
Two test conditions, three measurement systems, three different gear ratios: the same result every time, about four watts less to produce the same work. These are watts that the stock chain dissipates among its links and that the treated Shimano chain gives back to the pedal stroke.
The Procoating treatment
Procoating is a proprietary nanotechnology treatment, developed and applied by Zerofactory. Its composition and process parameters are protected as trade secrets, so its value is documented through measurement, like the tests in this article.
We make the Procoating treatment available for the drivetrain: chain, cassette and chainring. We are available to carry out a pilot treatment on the components of your choice, so you can verify the result with your own instruments. More information: www.zerofactory.it.