Car Failures and DNFs

Car failures that knock a car out of a race have gone down over time. But not, perhaps, the way you might think.

The Data

We’re looking at data from 1980-2021 race 34. I went through all the races and grouped together the reasons for DNFs into some broad categories. Today, a driver who dnfs because of an engine failure is just listed as ‘engine’. In the 1980s, there were specific as to what particular part of the engine went.

But we’re looking at the broad picture, so I needed a smaller number of categories.

Percentage of DNFs

There are three very broad categories NASCAR can choose from when attributing a DNF to a car: the car retired because it was in a crash, because the car quit, or because something happened to the driver. (Some were parked, some quit, some cited fatigue.)

Below, I show you the status of all cars from 1980-2021 race #34.

A stacked vertical bar chart showing the status of all cars as percentages for 1980-2021 race #34
The driver percentages are so small that matplotlib doesn’t draw them.

The percentage of DNFs has gone mostly down, as you would expect.

  • In 1980, only about 60% of the drivers were running at finish.
  • In 1990, we were up to nearly 70%
  • Today, it’s more like 80%-90%.

But you’ll notice that there are oscillations in the rates after 2000. In other words: I would have expected the RAF rate to either go up slowly or to plateau. (The plateau is because race car drivers are always going to crash.)

And, in fact, today’s race car drivers crash a little more on average than they did back in the 1980s.

A vertical bar chart showing the number of DNFs per 36-race season attributed to crashes

But crashes aren’t producing the oscillations. It’s the car failures that have changed in way I hadn’t expected.

Car Failures

Let’s just plot car failures that cause DNFs. I normalized each year to a 36-race season to compensate for running different numbers of races in different years.

A vertical bar chart showing dnfs due to car failures

In the early 80’s, the average number of DNFs in a single race was about 12.

Car failures go down steadily from 1980 to about the mid 1990’s — which was the trend I expected to see over the entire time range.

But there’s a bit of a peak around 2004 and another peak between 2009 and 2013.

Types of Car Failure

As you might guess, engines are far and away the biggest cause of car failures from 1980 to the present day.

A vertical bar chart that shows the number of different car failure mechanisms that cause dnfs

Engines

Over the years, we’ve almost has as many engine failures as we’ve had crash dnfs. But, of course, they’re not spread out equally. Below, I show you the car failures that were classified as engine failures from 1980-2021

A vertical bar chart showing DNFs due to engine failures

A couple interesting things here.

  • In 1982, there were an equivalent of 289 engine failures.
    • Over 36 races, that’s about 8 engine failures PER RACE.
  • There was a steep decrease in engine failures from 1980 to the mid 1990’s. Since then, the numbers haven’t gone down as quickly
  • We’ve had 19 engine failures this year!
    • I went back to double check because I didn’t realize it was that many. But it was.
    • We’ve already beaten the number for 2019 and 2020.

I wondered if perhaps the larger number of engine failures was correlated to new rules about teams needing to use the same engine for multiple race, but I don’t think that’s it. My reasoning is because I plotted engine failures by owner. Here’s 2021.

A vertical bar chart that shows the number of engine failures by owner for 2021 through race 34

Top-tier teams and the teams that get their engines from top-tier engine builders make up 5 of the 19 engine failures — about a quarter of the total. And if you look back over the last half-dozen races, it remains true that the struggling teams suffer the most engine failures.

That makes sense, right? Engine building has consolidated and these expert builders have a very low failure rate. While the NextGen car hits the track in 2022, new engines won’t be part of the debut. While everyone will have the same car essentially, they won’t have the same engines.

Brakes

Let’s look at another type of car failure: brakes. This is where things got interesting.

A vertical bar chart showing a huge increase in brake failures in the early 2010s.

Start and Parkers

Start and parkers were teams that ran a bare minimum of laps and then retired. They rarely used more than a set of tires and some didn’t even bring pit crews. But they collected enough prize money to do it again next week. I made an attempt to gauge how much of an effect start-and-parkers had by flagging any driver who dnf-ed and

  • Finished less than 15% of the laps the leader finished
  • Didn’t crash
  • Wasn’t sidelined by the damaged vehicle policy

Start-and-Park Stats

  • The year with the most start-and-park retirements was 2011 with 142.
    • That’s out of a total of 1548 cars, so we’re talking about 10% of the field.
  • Other years with a lot of early retirements:
    • 2012 (117)
    • 2010 (69)
    • 2009 (66)
    • 2004 (55)

The most popular reasons for S&P DNFs are (in order of frequency):

  • brakes
  • electrical
  • overheating
  • power train
  • vibration
  • oil system issues

So How Many Brake Failures Did We Really Have?

I made the graph again, this time separating out drivers who finished fewer than 15% of the leader’s lap total.

A stacked vertical bar chart that show drivers who ran less than 15% of the laps before DNF-ing.
In 1985, Darlington, Dover North Wilkesboro and Martinsville seemed to eat brakes really quickly.

You’ll notice that my algorithm caught some drivers in the 80’s and 90’s who were definitely not start-and-parkers. Cars were much less reliable in those days. But the ones and twos are nothing compared to the huge numbers in the mid 2010s. There were more start-and-part brake issues than actual brake issues in 2005, 2011 and 2012.

BUT: There was still a peak in brake issues in the early 2010s, and a smaller peak around 2004. You can see the same phenomenon in other categories, like overheating, powertrain, electrical and vibrations.

So Car Failures Really Did Increase in the Early 2010s?

Even when we take start-and-park cars into account, there was an increase in car failures in the mid 2010s, plus a slightly smaller increase in the mid 2000s. Why?

There are a lot of factors in play, but here’s my theory

  • The number of teams grew in the late 1990s and early 2000s as NASCAR popularity boomed.
    • Some of those new owners treated NASCAR as an investment. They expected to make a quick buck.
    • This was before the days of alliances. It was every team for itself.
    • The workforce was stretched thin as NASCAR expanded.
  • The 2008 recession forced teams to merge — and some to shutter entirely.
    • There were still investors who took advantage of teams’ struggles. It worked for some (Roush) and not for others (Evernham).
    • The recession made it hard to get sponsorship; some mid and lower-tier teams were barely squeaking by.

Compare those situations to today.

  • Numerous companies with one- or two-car teams have given way to a smaller number of top-tier owners with the maximum number of teams.
  • New teams can buy complete cars and lease top-notch engines from established teams (as long as they have the money)
  • The playoff scheme puts a premium on reliability. If you’re on the bubble coming into Martinsville this weekend, about the last thing you need is for your engine to let go, or your battery die. The penalty for a dnf is much larger than it was pre-chase/pre-playoffs.
  • The trend I showed above with engines — that failures happen mostly in the lower-tier teams — extends to other car parts.

But everyone has the same chance of crashing!

Bonus Graph: Crashes vs. Car Failures

If we plot the crashes vs. total car failures on one graph, you can see where the numbers cross over.

a vertical bar chart comparing crashes vs. car failures on the same graph
  • In 1980, you were more than four times more likely to DNF by car failure than by a crash.
  • In 2000, you were equally likely to DNF by car failure as by a crash
  • In 2021 (so far), you are about twice as likely to DNF by a crash than a car failure

Be the first to comment

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.