Bicycles Stack Exchange is a question and answer site for people who build and repair bicycles, people who train cycling, or commute on bicycles. It only takes a minute to sign up.
Sign up to join this community
How accurate are the power numbers (watts) from a Tacx Flow?
Multiple aspects : Absolute number : can they be compared with numbers from other power meters ? Relative : if ride two times at 200 W for one hour then did I do the same effort ? During a ride: if I start at 200 W and after an hour am still riding at 200W then do I still do the same effort ? Are the numbers still the same after using the same Tacx Flow a couple of years ?
The Flow appears to be quite consistent though, depending on the mode in which it is used, it can be quite inaccurate.
Below is a plot of reported power for speed on the Flow, with each line representing a different "scale factor." All of these data were collected at a coast down calibration of 0, with the same tire, at the same ambient room temperature; however, the red dots represent data that were collected 4 months later than the black dots. As can be seen, when conditions are matched, the consistency of the Flow appears to be quite good.
The plot above compares the Flow to itself. Similar to what zenbiker has noted above for the Tacx Fortius, when compared against an on-bike power meter the Flow tends to underreport -- though as one might expect from the plot above it underreports consistently. In my own case, with my own tires, a better agreement between the Flow and a Power Tap (which had been statically-checked for accuracy) could be achieved with a coast down calibration of 4 and a scale factor of 120, though I suspect that could differ from tire to tire.
A greater problem for the Flow is its ability to adjust load while in "ergo" mode. In that case, one sets the power level and then, in theory at least, one can pedal at different cadences and speeds while the Flow adjusts the load to keep power constant. Sadly, here the Flow was overmatched. The plot below shows the percent error for three different scale factors (80, 100, and 120) when plotted against the ratio of "ergo power" to speed. That is, suppose you set the Flow to hold its power in ergo mode at 250 watts and pedaled at 20 km/h, then 30 km/h, then 40 km/h, and so on. The ratio of ergo power to speed (in km/h) would be 12.5, 8.3, 6.25, and so on. If the Flow were able to adjust its load at different wheel speeds, each group of points would be horizontal and parallel, and, in an ideal world one could adjust the scale factor to move some group of points up or down until it hovered over the zero error line. Why look at the ratio of power to speed? Power is the product of wheel speed and wheel torque (and a constant) so dividing the ergo power setting by wheel speed gives the torque at the Flow's roller.
What you will see is that the groups of points are far from horizontal so the error between the power desired and actual power (as measured with a Power Tap) is not at all constant. At high wheel speeds the Flow is unable to generate enough load to maintain constant power. Whether that inability is related to the load generator itself (which is also used in the I-Magic) or the Flow's control head, I could not determine.
The Tacx Power measurement is accurate in terms of consistency, meaning 200w on Sunday is 200w on Monday, as long as your trainer is set up consistently.
It is generally showing a higher number than most other Power meters. My Powertap and my Fortius, when run concurrently differ by about 10%. As for long term accuracy, I've had my Fortius three years. I do have to recalibrate occassionally, but it not difficult, and seems consistent otherwise.
However, they both (when in functional, operating condition, something that the Powertap seems to struggle to maintain) seem to always differ by 10%. So the training method is accurate with each separately. I don't mix numbers.
I am looking forward to trying the Garmin Power Pedals. Tacx is supposed to allow an ANT+ reading from them directly, rather than using calculated methods to provide power readings, and I think that will be both more consistent and more accurate, if the quality is good.
