Lap times as lab testing

This is a continuation of a discussion that starts here.

I have pointed out lap time testing as an example of one of the things that’s analogous to some kinds of lab tests when doing camera testing. Lap time testing is a more complicated case than car tests like 0-60, 0-100-0, stopping distance, or even skid pad testing, for two reasons:

• It is a scalar measure of a vector quantity: in this case the axes of the vector are handling, slip angle, transition control, acceleration, braking, and a flock of other things. Caveat: I note that 0-100-0 testing is a scalar measure of a 2D vector.
• It is more dependent on the skill of the tester than simpler tests.

Nevertheless, it is a popular fixture in auto reviews: Top Gear’s The Stig laps, Abbie Eaton’s trips around the Eboladrome — I’m not making that up — on The Grand Tour, and Car and Driver’s well-instrumented, well-documented Lightning Lap.

All but the C&D tests are intended more as entertainment than serious data-gathering to inform prospective purchasers, but I’m going to treat them all here are if they were entirely about finding out how fast a car can navigate a closed course.

The purpose of this kind of track testing is to find out how fast the car is. It is not to test the driver. The prospective purchaser is interested in just the vehicle performance, and anything that can be done to take the driver out of the picture is a good thing. There is a strong parallel here with lab testing of a camera or lens; the test should yield results about the lens, and not about the tester. This is a weaker formulation of the standard for any scientific experiment, where reproducibility is necessary for acceptance.

I question the relevancy of how fast a car is when driven at 10/10 is to someone who is going to use it on the street (other than bragging rights, I guess), but let’s set that aside. The skill required of the test driver is to consistently get very close to the maximum performance of the vehicle. Since we’re talking about street cars here, that doesn’t mean that we need to have world-class racers as the test drivers. Race cars, especially the most high-strung ones like F1 cars, take more talent and ability to drive at their limit than street cars. There are also many uncontrolled variables in the kind of lap time testing we’re talking about here that mean that small variations in the capability of the test drivers isn’t going to be the thing that makes the most difference on the lap times achieved by different drivers in the same cars in different circumstances.

That is not to say that you could put any skillful street driver into the test driver role and get consistent results (when Clarkson and company first introduced The Stig, they said that they needed someone who could consistently get fast lap times out of the cars, implying that they themselves could not; which seems entirely reasonable given the way we often see them driving, although that could just be because truly fast driving does not make as good television footage as hooning). That wouldn’t work at all. You’re going to need people who are experienced racers with a lot of track time under their belts. Even better if they’ve raced many kinds of cars, including race-modified street cars. Track driving instruction experience probably helps. Once they were fully-acclimated to the less-responsive, lower-performance handling of the street car under test, I’m sure that top-flight racers could do a good job of this kind of testing, and maybe be a few tenths faster, but this kind of test is not about tenths, it’s about seconds.

Let’s assume an excellent driver and a given vehicle, and take a look at some of the other things that can affect lap times.

The first is tires. With a small number of exceptions, the tires fitted to street cars when they leave the showroom are not suitable for track use in dry weather. At a minimum, they need to have their treads shaved, or the heat buildup from all the acceleration, braking, and cornering forces will cause the tread to chunk, or fly off in pieces. This is disconcerting, expensive, dangerous, and slows the car down. On dry pavement, you don’t want a lot of tread, and the fastest tires have no tread at all, and are called slicks. On Top Gear, use of slicks disqualifies a test lap from participation in the cumulative ranking. With most cars, softer (stickier) tire compounds are faster than hard ones. None of this is well-controlled in the three test regimes mentioned above. It’s obvious why: it’s just too darned difficult. There are many different tire constructions, tread patterns, and compounds, and also many different tire sizes. Fixing on one design and trying to fit different sizes of those tires to all the tested vehicles would be difficult at best and is probably impossible. Also, for best results tire pressures need to be carefully selected and monitored as the tires change temperature.

The second is wind. Even cars whose aerodynamic properties approach that of a brick will have different lap times as the wind picks up and changes direction. Figuring out the effect can get quite complicated, but I’ll run through a simple example to show you some of how it works. Imagine a circular track of 1 mile circumference, and a car going around the track at a speed of 150 mph, which in the absence of any wind is the aerodynamically-limited top speed based on the power available. Let’s further simplify by saying that the speed is entirely limited by aerodynamics and horsepower. Now let’s postulate a 30 mph wind from the north. When the car is going north, it’s going to be slowed by the wind, and when it’s going due north, it’s going to be going 120 mph. When the car is going south, it’s going to be pushed along by the wind, and when it’s going due south, it’s going to be going 180 mph. Since it’s going slower when it’s going against the wind, it’s going to spend more time at the speeds below 150 mph than at the speeds above 150 mph, and the average speed around the circle will be slower than 120 mph. Remember the old algebra test question: a car goes up a 1 mile hill at 30 mph; how fast does it have to go on the way down to average 60 mph?

Next is fuel. For low-performance street cars, any octane increase over recommended won’t help much, but for more exotic vehicles, 100+ octane racing gas will allow the computers to extract more power from the engine.

Next up is temperature. It affects the density of the air, which changes aerodynamics. It changes tire temperatures, which affects grip (each compound has a temperature range in which it works best). It also affects engine power.

Then we get to track condition. If the track is wet, that’s going to slow the cars down. (Top Gear and Grand Tour flag wet runs). If there’s oil or dirt or sand or dust on the track, that’s a minus for speed. If it‘s rained recently and washed some rubber off the track, that will increase the times, too.

Controlling for all those factors is very difficult and is not done by anyone that I know of who’s doing this kind of testing.

There is also a higher-level issue. The lap time test produces a scalar result – a single number. Many things about the car affect that result. Here’s a partial list:

• Mechanical grip at various speeds
• Aerodynamic grip at high speeds
• Acceleration
• Frontal area
• Coefficient of drag
• Braking grip
• Brake fade
• Handling balance at various speeds
• Weight distribution

How those (and many other) characteristics affect the lap time is greatly dependent on the layout of the course. Long straights favor cars with low drag and high horsepower. Tight corners favor ones with high mechanical grip and low weight. You get the idea. The courses chosen by Car and Driver and The Grand Tour are radically different from each other, with the C&D course weighting high-speed and high-radius curve performance much more highly than the Grand Tour course.

To the potential purchaser of a car, the way that the course design prioritizes qualities of the vehicle is not obvious, and even if the prospective customer could suss that out, it might not be at all the weighting desired by that person for street use.

To sum up the points I’m making here: the lap time test is beset with uncontrolled variables, tests vehicles under conditions not encountered in street driving, and weights vehicle traits in unknowable and arbitrary ways. That doesn’t mean that the lap times are completely useless to a prospective purchaser, but it does mean they shouldn’t be weighted highly.

The parallels between lap time testing and camera laboratory testing are weak. Skill is required to carry out both kinds of testing well, but the skills required to do excellent camera and lens lab tests can be learned by anyone with reasonable intelligence, a mastery of high-school mathematics, a budget that is modest by auto racing standards, dedication, care, and year or so of experience. That doesn’t mean that skilled lab testers are common – in fact, there seem to be far more people doing camera and lens lab testing who have not mastered their craft than those who have – but the innate talent required is, I believe, less than that for a good track tester.

In photographic lab testing, there are few tests that are innately quantitative and which combine many disparate characteristics into a scalar, although there are many that reduce some vectors to scalars, such as MTF50, SQF, and the like.