I got the idea to do this little experiment from interviewing Allen Lim. I had 5×6 threshold intervals one day and decided to try out different pacing strategies. The plan was to do the first interval at exactly 300 watts the whole time, then with each progressive interval add 20 watts to the ascents and subtract 20 watts from the descents. I rode the same exact rolling stretch of road, the Central Park loop from the north end to 74th street on the west side. I hoped to find out how much harder I could go on the climbs before I blew myself up and started slowing down.

Here’s the profile:

Here’s the power numbers for each run, along with normalized power and times.

I left runs 2, 4, and 5 as is, and crudely adjusted runs 1 and 3 to 308 watts. I subtracted 40 watts for rolling resistance, then calculated velocity by assuming it was cubically proportional to power. This doesn’t account for gravitational resistance, but hey, that’s the best I can do.

You can see how normalized power starts to deviate from actual power as the climbing power goes up. Perceived exertion matched the normalized numbers, too. The first two intervals were downright civilized, almost easy to do. Runs 3-5 got progressively harder, and by run 5 my thoughts were drowned out by my heavy breathing. It got increasingly harder to hold my target wattage on the downhills.

From the data it seems my sweet spot is around the +/- 50 watt area.

Here’s the power and speed files from runs 1 (yellow), 4 (red), and 5 (blue).

The teal line in the power file is 300 watts. The 1st run (yellow) stays pretty close to 300 all the way through. The red line has peaks and valleys, the blue line even more so. I coasted a few times on the 5th run to simulate what would happen if you really blew up and couldn’t pedal the descents.

The speed graph is the opposite of the power graph in that the yellow line (steady power) has more extreme peaks and valleys, while the blue line (+/- 80) is the flattest.

So?

There are two common reasons for hitting the climbs harder than the descents. One, you spend more time climbing than descending, assuming the course begins and ends at the same altitude. Two, you get more bang for your wattage buck at lower speeds, since air resistance increases as a square of velocity. The speed graphs, however, reveal a few more things.

The 4th run (red) was the fastest. Compared to the steady 300 watt run (yellow), I’d crest the rollers at a higher speed. The yellow run would be faster halfway down the descent, but the max speed at the bottom would be the same. Hitting the gas harder at the bottom on the red run preserved momentum longer, resulting in an apparent shift to the right on the graph.

Of course, only an idiot would ride a rolling course with steady power. Comparing the ideal run with the over eager 5th run shows what happens when you blow up and start to lose it.

This closeup of two of the rollers shows that if I climb too hard (blue) and have to coast on the descent I’ll begin the next ascent 1-2 mph slower (a). And though I’d climb the next roller with more power and lose less momentum, I’d crest it going the same speed as the red run (b). In other words, I’d start the climb faster on the red run and crest the roller at the same speed as the blue run, while expending 20 less watts. Starting faster and ending just as fast = faster.

Part V, Epilogue

I hit the same stretch of road two days later armed with my new knowledge. I accelerated to crest each roller, then grabbed a little recovery by pedaling at 250 watts. I stepped on the gas just before each descent ended to pick up a couple mph’s, then settled into a steady 350 watts going up. It was a real revelation, knowing exactly how hard to ride every part of the course. After a winter of training for power it was a real change to ride for speed and momentum, and being stingy with power.