Thursday, March 13, 2008

Funky things with a Power Meter #77

OK, maybe I haven't demonstrated 76 other funky things with a power meter but then I have discussed or demonstrated the following:

OK, that's only 17 dots, so this will really be Funky thing # 18. I should add that most of what I've written is based on tools and analysis developed by guys far cleverer than I but is simply there to demonstrate a number of power training ideas and principles. If anything up there sparks your curiosity, then just click on link or look it up via the index at the right to find posts grouped into various categories.

So what's Funky Thing #18 all about?

The Chung Method

Well it's linked to aerodynamic field testing but using a different methodology, known as "The Chung Method", developed by a data analysis guru and regular power training forum contributor Robert Chung. It also acts as a proxy for developing an elevation profile of a loop course without the aid of an altimeter (now that's the really funky bit). It works best when you ride a course that passes the same point more than once (the more times the better). What am I on about?

From analysing the power & speed data from a power meter file for a typical ride, estimates for both the coefficient of drag-area (CdA) and coefficient of rolling resistance (Crr) can be made. These key measures indicate the degree to which air resistance and the road surface serve to retard our forward progress or how hard we have to push on the pedals to overcome these forces at any given speed.

The method works by using the equations of motion for a cyclist (well a slightly cut down version), with a few assumptions thrown in (such as a low wind day). If we know a little more data about the course and the conditions, the estimates of CdA and Crr derived and elevation profile obtained can be pretty good. Good enough that changes in rider position, equipment (or conditions) can be readily detected and that the elevation profile be correct to within a few metres.

And it does not require the usual protocol for field testing, that of doing multiple runs under highly controlled conditions. Just use ordinary power meter data from a loop course. It helps if you have a near windless day (a little wind is OK).

Now Robert's paper which discusses
this method in detail can be found here:

So I thought I'd have a go and with the aid of a spreadsheet posted on one of the training forums I frequent, I applied it to a sample of my own data.

I picked a training file from Boxing Day 2006. Here is a pic of the training loop I rode that day, a popular local training ground - Centennial Park in Sydney. Grand Drive is a 3.8km roughly circular loop, flattish. There is also an option to climb up a hill to the Ocean Street gates, then across to the Paddington gates before descending back down to Grand Drive, which adds about 2.5km to a loop. Sydney-siders would be pretty familiar with the Park.

Here is the graph of my power and speed file for the day chosen. It was a tempo effort of 90-minutes duration where I did laps of Grand Dr with a climb up the hill to Ocean St every second lap. You can see that by and large I kept my power output within a range and let my speed vary (not that that's necessary for this method - it's just what I happened to do that day).

Using this data and the spreadsheet with the funky formulas which use the equations of motion, here is the chart produced showing the ride elevation profile of my ride in Centennial Park that day. On the chart I show the CdA and Crr estimates needed to provide a consistent elevation for the same points in the ride. Since I already knew the elevation difference from the lowest point to the highest point in the park, that helped me adjust the CdA and Crr numbers such that the profile provided an accurate representation of the course (to within a few metres).

That's quite remarkable if you ask me. Now it was just a training run, not a time trial, so I was on my training bike, probably riding with my hands on the hoods, maybe occasionally on the bar tops going up the hill. A cool morning too, so probably a bit of extra clothing on for warmth. Hence the relatively high CdA of 0.384. A Crr of 0.005 was settled on and seems to be a reasonable estimate for the mostly decent quality hotmix/asphalt surface in the Park.

You will note some variations, particularly the opening kilometres and the final lap, where the profile varies from the consistent elevations shown from km 6 to km 40. I suspect that during this middle section of the ride I was using a consistent position on my bike.

For the final lap, since the derived elevation data doesn't match the other laps, it appears that either I rode in a different position, changed clothing, conditions changed (perhaps the wind) or I was mixed in with other riders. I'm not sure, I can't recall. But the change is very distinct with this method and is one way of assessing the impact of changes to equipment and/or rider position.

The forum thread where this sparked my interest in having a look at it myself is here:
Slowtwitch Chung Method

So what's Funky Thing #19 gunna be?

As Robert Chung would say, "Hmmm...."


Lawrence said...

Thanks for listing the links Alex. I am now a PT convert, and after 2 mths of use, still getting used to its quirks and idiosyncracies. The one I like most is my PT's need to throw in one cadence reading of 244 (always 244) per ride. Cute. Annoying but cute. It's like a little signature it feels it must include in each new file.

Alex Simmons said...

I was asked which method of aero testing is most accurate?

How about I leave that for the experts to debate. See this thread on the wattage forum:

You may need to be a member to view it.