Monday, January 19, 2015

Some kilometres are longer than others

With the spate of attempts at the UCI world hour record over late-2014 and into 2015 due to the revised UCI rules making the record within reach of more riders, it has naturally sparked interest in discussing what matters for best performance in the event.

Jens Voigt started the latest round of hour record attempts at the UCI's Aigle track
I recently saw some chat on a triathlon forum speculating about who could do what distance and so on. All in good fun, but none of them actually go to a track to find out. If they did, they'd realise it's not quite as simple (or as hard) as they might make out.

It pretty much comes down to optimising four main elements:
  • maximising sustainable power output for an hour
  • minimising the physical resistance factors of riding on the track
  • technical execution / skill
  • logistics & resources
Some might add psychological factors to that list, but ultimately I consider these to be expressed within the outcomes of each of the above.

Regarding logistics, there are of course UCI requirements to be permitted an official attempt an hour record, e.g.: minimum time in anti-doping bio-passport program is mandatory at elite level or dope testing at age group level, application submitted in advance for approval to relevant levels of cycling administrations, all the technical requirements including international level commissaires to supervise, a UCI approved track, use of timing equipment, start gates, specified date and time of attempt, etc. You can't just rock up and ride whenever you like. Well you could but it would never be a sanctioned attempt.

Then of course you need to factor in enough solo rider time on track for preparation, and that costs money and time as well. Quality indoor tracks are not always local, and even if they are, getting solo time on the track is not always so easy, let alone cheap. For an elite professional rider whose job is to race on the road, it may be difficult to devote sufficient time to the task of preparing properly for a track event.

Of course assuming the paperwork is all in order and you can do your training, then sustainable aerobic power and aerodynamics are king and the rider's ratio of power to aerodynamic drag area is the single most important factor for how far they will go in the hour. But W/m^2 is not the only factor.

There are other physical resistance force factors, like the influence of air density which is a function of altitude, temperature and barometric pressure (and to a much lesser extent, humidity) and the rolling resistance of the track and tyres chosen. I discuss some of these in the following items:
Altitude and the Hour record Part I
Altitude and the Hour record Part II

Which leaves us with technical execution and skill factors, of which there are a couple of key items, namely:

Riding good lines

Riding a good line involves a couple of components, one is pretty obvious and involves not riding further than you need to around the bends. Ride wide and you ride further. Pretty simple given the track is all but two semi-circles joined together with two straight sections. OK, the actual shape of tracks are more subtlety curved but that's close enough to describe why riding wide adds distance to your travels around a lap.

Design of the Glasgow Velodrome

If you ride 10cm wider in the turns, you add 10cm x 2 x PI = 62.8cm per lap.

If the extra width is measured on the track's surface, well the actual addition to the distance the wheel travels is reduced by the cosine of the banking angle. e.g. say the track's turns are, on average, banked at 40 degrees, and you ride 10cm above the black line. Then the actual additional track radius ridden is cosine (40 degrees) x 10cm = 7.7cm, and the additional distance per lap = 7.7cm x 2 x PI = 48.1cm. Nearly half a metre.

Do that over 200 laps or so for an elite hour record and you'll ride ~100 metres more than you need to. And that's for riding only a hand's width above the black line.

London Velodrome used for the 2012 Olympics
Another more subtle ride line factor involves the shape and design of the banking and in particular the transitions from the straights to the turns and back again, and whether it's advantageous to ride a slightly wider line in the straights to aid the transitions. On the straights you don't suffer the same severe distance penalty of riding a wider "radius" as you do when riding wide in the turns, so you can explore marginal gains in this manner.

However there is no simple or single answer to this, it depends on the rider and the track geometry - all of which have subtle differences. This is a somewhat more complex optimisation problem and I'm not going to delve into it here.

So putting aside these subtleties, the shortest distance around the turns is to ride the track's black measurement line* - ride any further out from the black line and you ride more distance each lap than is necessary. For the hour record you only get credit for the official lap distance each lap, which is typically 250 metres per lap on most modern standard indoor velodromes although some tracks are shorter and some are longer.

* it is possible to ride inside the black line, however in such timed track events like the hour there are foam blocks placed around the inside line of the track to ensure the riders don't. Very skilled riders can however ride fractionally under the black line on some tracks but it is risky as hitting the foam blocks can disrupt your effort and wash off some speed. The shape of the track in that small space between the black line and the wide blue section varies from track to track and it can be good or not so good to ride in.

Foam blocks discourage riders from riding inside the black measurement line.

Now why are some kilometres longer than others?

Office distance for the hour record =
(Official lap distance)  x  (Number of full laps completed within the hour)
+ a pro-rata distance calculated for the final incomplete lap

I won't go into the formula used by the UCI to calculate the pro-rata distance of the final lap (that's actually deserving of a blog post on its own as the regulations are remarkably confusing).

It matters not how far you actually ride, you'll only be credited with the official minimum lap distance per lap. This is why track riders and coaches are focussed on lap times and not with bike speed, since lap times are the integral of all performance elements. Power meter and other data loggers are of course valuable in parsing out the individual elements of performance that go into attaining lap times, and helping to prioritise development opportunities.

How good are riders at riding the minimum distance necessary?

It varies. Quite a lot. Skilled track riders are typically much better, which is what you'd expect. But what sort of penalty would an unskilled rider face if they started out on a track effort?

Of course we can do lots of maths to figure out how much extra distance on average a rider might cover if they ride wide by so much, but in reality riders move up and down the track, sometimes riding a good line, other times not so good. Some riders are just better at it than others and some adapt to the track more quickly than others.

It'd be so much better if we could simply measure what people actually do rather than speculate.

Which had me thinking. I have some data like that already...

Not so long ago I was doing some performance testing involving half a dozen pro-continental road racers at an indoor 250m velodrome. One of the features of the data logging system used for the tests is an ability to calculate the distance ridden per lap using the wheel's speed sensor data combined with track timing tapes to know precisely when they pass a specific points on the track. With some clever maths this is enough to nail the actual distance ridden each lap to high precision.

In amongst the test data were some solo efforts of at least 10% of the distance of an elite hour record attempt (i.e. 20+ laps of consistent effort) and several such runs by each of the six riders. I figured the runs needed to be long enough to reasonably approximate what a rider might be expected to do over a longer distance/duration.

Of course absolute accuracy of the distance the wheels travel depends on having an accurate wheel circumference value and that value not changing a lot while riding. So I'm not going to assume that the absolute accuracy was perfect, even though the absolute error might typically be somewhat less than 1%. More than that would require an error in tyre circumference assumption of 20mm, which is a lot for those used to measuring such things. However in our favour is that even if such an error existed, it would be a consistent bias error.

So rather than concern myself with absolute accuracy, I thought I'd compare the measured average lap distance for each run with the shortest recorded legitimate lap. In this way if there is any bias error, it's impact on this analysis is minimised (i.e. both measurements would be out by the same proportional amount). By legitimate lap, I mean a full lap not ridden below the black line.

Here's a table summarising data collected from the six riders (in no particular order). Each rider has multiple runs although I haven't identified the riders in the table. What the first column shows is the average lap distance per run less the minimum legitimate lap distance for that same run. The distances are of course distance travelled by the wheel.

Now the riders possibly could ride a tighter line than they actually did for their shortest legitimate lap, meaning that these distances likely underestimate the extra distance ridden when compared with riding very tight to the black line.

For the moment though let's assume the shortest lap they rode during each run was the best they are capable of doing. Since they actually did it, I think that's a reasonable assumption.

The average extra distance ridden per lap varies from one rider to another. One rider consistently rode only 0.3-0.4 metres more per lap than their shortest distance lap, while another was consistently riding more than 2 metres extra per lap on average compared with their shortest legitimate lap. The rider with smallest extra distance per lap had a track racing background.

The second column shows what that average extra lap distance would mean if extrapolated to riding 200 laps of a 250m track (an official distance of 50.000km). For one rider they would be riding nearly half a kilometre further than their track skilled team mate. Yet if both completed exactly 200 laps in the hour, each would be credited with riding precisely 50km, even though one rider's wheels had travelled nearly 500m further than the other's.

In this case, 50.5km = 50.0km. Some kilometres are longer than others.

So what's that extra 500m cost in power terms?

Well for a rider with a CdA of ~0.23m^2, that extra 500 metres travelled requires they output ~11-12 watts more than if they were able to ride a a better line.

Or they'd need to find a 3% reduction in CdA to make up for their skill deficiency.

Remember these were well skilled, well trained and experienced pro-continential road racers and finding an extra 10W or losing another 3% of aero drag coefficient isn't such an easy thing to do.

So no matter your current skill level and experience, if you're expecting to ride such an event yet you have never trained to become proficient riding on the track, well you might want to chop half a kilometre or so from your estimated distance covered based on your power and aero data alone.

Better still, just get to a track a find out what you can actually do.

Likewise, when estimating power, or W/m^2 from the official hour record distances, you might need to add some watts for the technical proficiency of the rider. The less proficient, the more power is required to attain the same official distance.