Jump to content Jump to content

DR2.0 Physics In Motion: Measuring Oversteer/Understeer


Recommended Posts

(Credit goes to Jorge Segers book, “Analysis Techniques For Racecar Data Acquisition” for most of the theory here)

“I hate RWDs because they oversteer so much.”

“I can’t seem to tune out the understeer in this car!”

“The car won’t turn in, and then it oversteers :(“

When it comes to car preferences in DR2.0, much is made of the vehicle’s balance, or its tendency to oversteer or understeer. The driver might tweak setup, or pick a different car, or buckle down and drive through the problems. So is there a way we can track these changes, or even better, measure the amount of oversteer/understeer according to our particular driving preference? And then use this measure to track setup changes? Absolutely!

But first we need to define oversteer and understeer. We can look at it from two vantage points - the driver’s and the car’s. From the driver’s perspective, oversteer and understeer is quite easy - oversteer means the car turns in more than the driver expects it to, and understeer is the car not turning in the amount the driver expects it to. Compare this to how the car sees it - as a function of a car’s slip angle and its relation to ground speed and the corner radius, etc., all measured in a steady-state environment which a rally car never is in. So, obviously, let’s go with the driver’s definition and merge it somewhat with the technical definition to try and get something useful.

In this case, we will be using the car’s yaw rate vs. the car’s angular velocity to produce what Jorge Segers refers to as the attitude velocity, which is the difference between the two. Angular velocity is the car’s actual trajectory as defined by this equation:

Vangluar = Lateral G force / Ground Speed

If you are familiar with circular motion physics, this is a rearrangement of the equation for centripetal acceleration and is measured in rad/sec, which will be converted to deg/sec.

Figuring out yaw rate (the speed the car spins on its axis measured in degrees/sec) is typically measured with a yaw rate sensor at a car’s center of gravity. Unfortunately we don’t have that logged, but luckily the car’s actual direction measured in degrees IS logged, as well as the world position of the car’s CoG (assumed to be in the middle) and the front of the car, so we can combine and derive those two bits of data to get a complete picture of the car’s yaw motion, seen here: https://i.imgur.com/emkc963.png

Attitude Velocity is thus a simple subtraction: Yaw Rate - Angular Velocity. What we end up with is a trace that looks like this: https://i.imgur.com/xAKO4hq.png

So now we can look at the car’s balance depending on how it is moving. When Yaw Rate = Angular Velocity, both the car’s trajectory and it’s yaw are moving in tandem and are thus considered to be in neutral steer. When Yaw Rate > Angular Velocity, the car is yawing much faster than the car is actually turning, which results in mechanical oversteer. Finally, Yaw Rate < Angular Velocity results in understeer, which is the car not yawing enough in relation to the car’s trajectory. Looking back at the graph, anytime the graph goes positive, that’s oversteer - when it goes negative, that’s understeer. Neat!

(Remember, what we are describing is a specific MECHANICAL event. This is different from what we defined earlier with regards to driver perception of over/understeer. When you look at the traces you will notice that attitude velocity often goes positive on corner entry, especially with RWDs and AWDs. The car is rotating fast but it’s not turning in, and a driver might perceive this as understeer, but mechanically this would be considered oversteer. If a driver perceives the car as having too much oversteer, i.e. the car turns in too quickly, the car might actually be mechanically neutral and has very good grip, but the driver prefers to have a car that mechanically understeers or has less sensitive steering input as the driver is very uncomfortable driving a car that reacts so quickly. Be very careful of this when addressing your car or driver’s particular issues - words matter, and it is good to be clear in one’s descriptions)

From here we can do simple spot checks on slow corners or tricky spots by comparing traces and evaluating how much over/understeer we end up performing around corners. We can also plot this on a heat map and see where on a stage we tend to over/understeer, like this one: https://i.imgur.com/IBRXG8S.png. I don’t find this particularly useful in motec, though, simply because motec for some ungodly reason doesn’t let you zoom in on a heat map so kind of useless for long rally stages. However, you can just as easily look at the data trace and examine what’s happening. 

The REAL power of this particular trace, though, is tracking a driver’s performance. By looking at a series of setup changes or stages and tracking average attitude velocity, we can actually see what how much oversteer a car actually has throughout a stage (avg attitude velocity is near the bottom): https://i.imgur.com/y48iUK7.png

A higher number indicates greater oversteer, while a lower number trends towards understeer. Once you start looking at different cars, you can see particular trends, such as FWDs usually have numbers around 2 or lower, Tarmac RWDs from 4-6, and R5/rallycross can have numbers 8 or more. I’ve been testing the R5 Fiesta MK2 and tires on the New Zealand stages and have been able to push averages up to 11, which is high even for rallycross. 

We could take this even further and plot these averages against stage times, to see if you, the driver, prefer a car with more or less oversteer: https://i.imgur.com/6j73vmd.png

I added a trend line so you can see how in my case a higher attitude velocity (driving with more oversteer) seems to help improve my times. A more rigorous examination would look at variance and finding more consistency. In this instance, we look at which average attitude velocity gets us the fastest times, and then when deciding a setup, we shoot for that number, and push that as far as it goes until we start to lose performance/consistency. Nice!

Our driving performance, though, is only as good as our racing lines, so we have to be mindful throughout all of this that we still need to make sure that we are picking the right lines through corners and that the driver feels comfortable with their car. But how do we know what the right line is? We’ll find out in a future post!

Postscript: Knowing that we can track attitude velocity, the big question now is, at what velocity do we maximize our available grip? This is a hard question to answer and I'm not completely confident in answering it yet, BUT, when you plot lateral Gs against attitude velocity, it at least appears that we don't really get much more grip beyond 50 deg/s in oversteer and about -30 deg/s in understeer, but that's just from a visual guess. What seems clear though, is you can maintain pretty good grip just staying as close to neutral as possible.

Link to comment
Share on other sites

3 hours ago, merseyxshore said:

This hurt my brain, but full respect for the effort here. Codemasters, hire this person.

Ha, CM already builds their own driving simulators I think they're already way ahead of what I'm doing here 😅

I hope that I can find something  actionable for your every day driver though. Data is fun and all, but being able to find just that one easy driving tip that anyone can immediately apply to their driving is really what I want to strive for. I think this stuff is really interesting and worth talking about, but might be more for the hardcore racers out there.

Link to comment
Share on other sites


This topic is now archived and is closed to further replies.

  • Create New...