Saturday, March 1, 2014

comparing rides: power histogram and trend in L-R balance

I've seen blogs where people just post their power numbers week after week. It gets quite dull. I'm risking making the same mistake here. But I started this blog as sort of a depository for whatever I find interesting, and this was sort of interesting.

I took a break from running this week to give my legs a chance to recover from the Lake Chabot 30 km trail run. But I've been seriously craving cycling, especially riding to work in the morning (75 km), and this week gave me an opportunity to get a few SF2G rides in.

I have gotten comments from 2 independent sources that 75 km commutes aren't "recovery". But riding at my commute pace, while fatiguing, is much less traumatic than running. And getting some endurance work in should keep my metabolism pointed in the right direction. It's a small gamble, but then I do miss riding, and need to get my little fix before starting another running ramp to the shining goal of the Woodside Ramble 50 km trail run in six weeks.

Anyway, I rode in twice this week, on Tuesday and Thursday. These were similar rides: identical routes, both ridden solo. The Tue ride was dry, the Thu in intermittent rain, but I was never cold so the rain wasn't much of an issue. In both cases I measured power with Garmin Vectors, which I've not yet validated.

What motivated this post was the similarity of the power histograms. Here's the distribution of power, in 10-watt bins, where the spectra are fit with weighed 4-parameter curves, asymmetric variants on hyperbolic secants.

histograms

The spectra are close, except that the second ride is shifted around 20 watts to the right. Indeed, my legs felt tired on Thursday, but notably less crappy than they had on the Tuesday ride. So perhaps this is working out to be a recovery week, after all.

Then I move on to the L-R balance:

histograms

The pattern continues of a right bias at low powers moving up to an equal balance, or perhaps slight left bias, at higher powers. But with each ride, in order, the right bias at low power has become less profound. It was around 54% on 16 Feb, then 53% on 22 Feb, then 52% on 25 Feb, and now close to 51%. By my next ride it may even be gone completely. So why? Is this a pedal artifact? Or is it that my left leg had an issue which is resolving itself?

I decided to do power histograms separately for each leg. I plot first the left, then the right. You can see in the second ride the left-leg curve is shifted right. The left leg produced more power. The right-leg data are also shifted right, but only barely. The right leg power was more constant. The histograms for the two ride, left and right leg, are surprisingly similar, almost eerily so. But then it's the same route, ridden roughly the same way.

histograms

histograms

Of course, I don't know whether these data are the result of the pedals changing or my legs changing. It's possible the pedals are "settling in" after initial installation. But it also seems quite plausible I'd tweaked something in my left leg running and it's improving over time. And if so, that's really interesting. Actionable intelligence? I'm not sure, but if it's a hint that something's wrong, that should be good to know.

3 comments:

Ruud_G said...

Hi djconnel. Nice blogpost. It s good more and more people start to look at their left right data and post about it, although -apart from arguments in the recovery sphere- is not that actionable at the moment. That is I have not seen an article yet which really adresses the merits of this data and to use it in training to enhance performance. I think I am gonna look at my data as well in a while although its not the same left right data as the garmin vector data (power2max type s). And yes. I also have to make sure I dont fall in the trap of posting too much on powernumbers from training ;) http://laymanpowermeter.blogspot.nl/

djconnel said...

Thanks for the comments! I view LR at this point more as a check that nothing's amiss. Best, I think, would be to see if it has any actionable correlations, like to fit issues or injury. I wonder how many coaches, for example, who have access to data from multiple riders, have gotten anything useful out of it yet.

But I am suspicious of total power at the moment. I didn't use any washers, trying to minimize Q, and I wonder if that's an issue. The White Industries crank I am using has a flat outer face, so it seemed if any crank didn't need washers it was this one. As soon as I get new batteries for the Powertap I'll do a comparison.

Ruud_G said...

Yes would be interesting to see!