Concept2 BikeErg Raw Data
Through the ErgData application, it's possible to record and review stroke / revolution data from a Concept2 ergometer. Below we capture and analyse pieces performed on the BikeErg at different damper settings.
Through the ErgData application, it's possible to record and review stroke / revolution data from a Concept2 ergometer. Below we capture and analyse pieces performed on the BikeErg at different damper settings.
We want to see the effect that damper setting has on pace at a given RPM on the BikeErg. An in-depth discussion of damper setting and drag factor is beyond the scope of this experiment; Concept2 has an excellent article describing what damper is, what drag factor is, what they are not, and what damper to use.
Starting at damper 10, 2:00 of work would be completed on the BikeErg and recorded by ErgData. The athlete would try to get as high an RPM as possible over the 2:00, but not hold any particular RPM. This allows us to accumulate data mapping RPM to pace over a range at a given damper.
This would be done 10 times; the first at damper 10, the second at damper 9 and so on until damper 1.
Each work piece was pulled from the Concept2 logbook, along with per-stroke (per-revolution) data. Raw data is available here. Erroneously recorded values (value too high, recorded pace invalid with associated RPM etc) are commented out of the dataset.
Each damper setting graph series in each graph below is labeled D<num>, where <num> is the number the damper was set on.
Tap each series label in each graph's legend to disable / enable the series on the graph. Mouse-over or tap the graph to see values of pace for each damper at a given RPM.
The raw data was cleaned, with repeated pace/RPM values aggregated to a single data point to remove noise. At this point, it is possible to infer the trajectories of RPM-pace pairs resembling curves.
By plotting each series as a line, we can see a non-linear relationship between pace and RPM. We can also see that a higher damper setting yields a faster pace at the same RPM for all recorded RPM.
For example: at damper 1, an RPM of 80 gives a pace of 2:23.9. At damper 10, an RPM of 80 gives a pace of 1:28.0.
Passing a cumulative average function across each dataset helps reduce bumps, at the cost of reducing accuracy. However, this reinforces our previous observation relating pace, RPM and damper.
While our observation that a higher damper setting at a particular RPM yields a faster pace is not revolutionary (no pun intended), it is interesting to see the effect that damper has across a range of damper settings.
Note that a dusty flywheel or air pressure changes due to altitude will generate different values to the ones documented above.
The methodology of having the athlete cover a range of RPM for 2:00 was flawed. Naturally, the athletes increased from idle to a “comfortable” RPM over a short period of time, stayed at that RPM, and then increased RPM nearing the end of the 2:00.
This gave a larger variance of pace at low and high RPM, which may skew the results.
A better approach would be to select a set of RPM values (perhaps 30, 40, 50, 60, 70, 80) and have the athlete hold that RPM for a set amount of time before moving to the next. This way we collect a greater number of points at each known RPM, reducing jitter.
Drag factor should follow damper setting to some degree; a lower damper setting is a lower drag factor, higher damper setting is a higher drag factor. However, the test for damper 10 showed a drag factor lower than damper 9, 8, and 7:
Unsure what happened here.
It would be nice to re-run this and generate some more accurate data. This would increase accuracy of our damper curves to allow better comparisons between RPM and pace.
Special thanks to @be.like.scott and @ktna89 for the loan of their BikeErg and help in running the tests.