Friction Circle - the reality
#1
Lifetime Rennlist Member
Thread Starter
Friction Circle - the reality
Again, keeping this out of CC's thread...
Every data engineer I have ever spoken to says that the FC diagram is about the most useless tool out there. I agree. In addition to lacking a connection to where you are on the track, a basic problem is that the FC is not THE FC. Vehicle dynamics can make a huge difference in how much grip you have on a given corner.
Case in point, the keyhole at Mid-Ohio. Long ~180 degree turn that enters uphill, levels off and drops as you get half way through. If you take this corner with the car balanced (maintain neutral throttle through out) then your front tires will have very little grip and the car will slide off the outside and into the wall at a certain speed.
BUT, if you do a big lift off the throttle at the right time, the front tires get some weight and grip. The same speed and attempted lat g that put you into the wall with neutral throttle is not fine and you negotiate the corner.
In both cases, you have driven the edge of the FC (or gone over if you crashed) but the FC graph would not look the same for those examples. Looking at the FC of either of those, really tells you nothing about what happened in the corner.
Every data engineer I have ever spoken to says that the FC diagram is about the most useless tool out there. I agree. In addition to lacking a connection to where you are on the track, a basic problem is that the FC is not THE FC. Vehicle dynamics can make a huge difference in how much grip you have on a given corner.
Case in point, the keyhole at Mid-Ohio. Long ~180 degree turn that enters uphill, levels off and drops as you get half way through. If you take this corner with the car balanced (maintain neutral throttle through out) then your front tires will have very little grip and the car will slide off the outside and into the wall at a certain speed.
BUT, if you do a big lift off the throttle at the right time, the front tires get some weight and grip. The same speed and attempted lat g that put you into the wall with neutral throttle is not fine and you negotiate the corner.
In both cases, you have driven the edge of the FC (or gone over if you crashed) but the FC graph would not look the same for those examples. Looking at the FC of either of those, really tells you nothing about what happened in the corner.
#2
Originally Posted by SundayDriver
Every data engineer I have ever spoken to says that the FC diagram is about the most useless tool out there. I agree. In addition to lacking a connection to where you are on the track, a basic problem is that the FC is not THE FC. Vehicle dynamics can make a huge difference in how much grip you have on a given corner.
#3
Rennlist Member
Although i agree with everything else that you said, and I like your example (but I've never driven Mid-O), I wouldn't totally discount the Friction Circle.
IMHO, I also agree that it should not be used as the foundation for determining a univerally correct formula for the fastest way around any track.
But I think it does have some good use to describe a car's general abilities, and it is good for analyzing situations where you might be un-knowingly "leaving something on the table".
IMHO, I also agree that it should not be used as the foundation for determining a univerally correct formula for the fastest way around any track.
But I think it does have some good use to describe a car's general abilities, and it is good for analyzing situations where you might be un-knowingly "leaving something on the table".
#4
Lifetime Rennlist Member
Thread Starter
Jack,
I agree that the FC could show the car capabilities and even some changes. But the big thing it is missing is the ability to connect the data you see with a particular place on the track. So unless you are looking at pretty big differences in the envelope, I don't see how it tells you much since you often gain in one thing at the expense of another. Strip chart type data can show you the g's at a particular place along with the speeds. Overlaying those will tell you far more than the FC chart.
There is also at least one more problem with FC and that is the g data itself. Where does this data come from? In many of the less expensive systems, the sampling rate is not all that high (sampling rate is not the same as logging rate). The accelerometer is measuring both the real g's (cornering force in the case of lat g's) and the chassis flex. To be accurate, you need a high sampling rate (perhaps a 1000 per second) and then good filtering. Some of the entry level systems have slow sampling, some filtering and then fast logging which fools you into thinking the rates are high. The Stack and Pi systems I have had suffered from g readings that were hard to accept in many cases. OTOH, they tended to be consistent so were OK for overlay data. Long g's can be even worse as many systems do not measure this, but calcualte it from speed data. Nothing wrong with that except the speed signal may not be very good (could be a filtered speedo with fairly large error) and you tend not to know what is happening inside with the filtering of that signal.
Bottom line is that I have never found anything that I could get from the FC that I could not get, and in a better manner, from some other data aq tool.
I agree that the FC could show the car capabilities and even some changes. But the big thing it is missing is the ability to connect the data you see with a particular place on the track. So unless you are looking at pretty big differences in the envelope, I don't see how it tells you much since you often gain in one thing at the expense of another. Strip chart type data can show you the g's at a particular place along with the speeds. Overlaying those will tell you far more than the FC chart.
There is also at least one more problem with FC and that is the g data itself. Where does this data come from? In many of the less expensive systems, the sampling rate is not all that high (sampling rate is not the same as logging rate). The accelerometer is measuring both the real g's (cornering force in the case of lat g's) and the chassis flex. To be accurate, you need a high sampling rate (perhaps a 1000 per second) and then good filtering. Some of the entry level systems have slow sampling, some filtering and then fast logging which fools you into thinking the rates are high. The Stack and Pi systems I have had suffered from g readings that were hard to accept in many cases. OTOH, they tended to be consistent so were OK for overlay data. Long g's can be even worse as many systems do not measure this, but calcualte it from speed data. Nothing wrong with that except the speed signal may not be very good (could be a filtered speedo with fairly large error) and you tend not to know what is happening inside with the filtering of that signal.
Bottom line is that I have never found anything that I could get from the FC that I could not get, and in a better manner, from some other data aq tool.
#5
Race Director
The friction circle is a nice concept and useful from a theoretical concept to train a driver. The actual circle is just theoretical. Trying to plot data against it is pretty much worthless. Ever seen a FC plot with lines outside the circle? Some folks who don't understand the FC think it's where you've stepped over the line of max grip. Wrong. If that were the case the plot would be well inside the FC. A plot outside just means the theoretical FC is not large enough.
If a thousand monkeys, drive a thousand cars on a thousand race tracks, will one of their FC plots draw the Mona Lisa?
If a thousand monkeys, drive a thousand cars on a thousand race tracks, will one of their FC plots draw the Mona Lisa?
#6
Race Director
Originally Posted by SundayDriver
Bottom line is that I have never found anything that I could get from the FC that I could not get, and in a better manner, from some other data aq tool.
#7
Lifetime Rennlist Member
Thread Starter
Originally Posted by Geo
The friction circle is a nice concept and useful from a theoretical concept to train a driver. The actual circle is just theoretical. Trying to plot data against it is pretty much worthless. Ever seen a FC plot with lines outside the circle? Some folks who don't understand the FC think it's where you've stepped over the line of max grip. Wrong. If that were the case the plot would be well inside the FC. A plot outside just means the theoretical FC is not large enough.
Originally Posted by Geo
If a thousand monkeys, drive a thousand cars on a thousand race tracks, will one of their FC plots draw the Mona Lisa?
But mine are never very pretty though I think I saw one that looked a lot like Homer Simpson.
Trending Topics
#8
The Friction Circle is an accurate representation based on the assumption of a flat surface with uniform grip and that the vehicle is completely balanced. Unfortunately reality is that the surface has a complex slope, grip is not uniform and the vehicle has a dynamic balance characteristic. The model will still work in this case, but you need to take these variables into account.
#9
Race Director
Originally Posted by fizbot
The Friction Circle is an accurate representation based on the assumption of a flat surface with uniform grip and that the vehicle is completely balanced. Unfortunately reality is that the surface has a complex slope, grip is not uniform and the vehicle has a dynamic balance characteristic. The model will still work in this case, but you need to take these variables into account.
#10
fizbot, just as said above, its a teaching tool. Its a concept on a chalkboard just like the 4 contact patch examples. Now they would make for some fun blurry eyed drawings.
#12
For those who are reading these posts, please do not beilieve the vast majority of what you are reading. Most of the comments are flat wrong and reflect an amazing level of ignorance/misinformation.
If you are serious about understanding the subject, please study any vehicle dynamics book, Miliken being the most widely accepted, and you should will learn the truth.
If you are serious about understanding the subject, please study any vehicle dynamics book, Miliken being the most widely accepted, and you should will learn the truth.
#13
Lifetime Rennlist Member
Thread Starter
Originally Posted by ColorChange
For those who are reading these posts, please do not beilieve the vast majority of what you are reading. Most of the comments are flat wrong and reflect an amazing level of ignorance/misinformation.
If you are serious about understanding the subject, please study any vehicle dynamics book, Miliken being the most widely accepted, and you should will learn the truth.
If you are serious about understanding the subject, please study any vehicle dynamics book, Miliken being the most widely accepted, and you should will learn the truth.
Yep. Color is right as always. Don't listen to me or anyone else with real track experience. Do what Color says.
But please do me one favor...
Post the video.
#14
Race Director
Originally Posted by ColorChange
For those who are reading these posts, please do not beilieve the vast majority of what you are reading. Most of the comments are flat wrong and reflect an amazing level of ignorance/misinformation.
Hey, please skool us, OK?
#15
Race Director
Actually, better yet, I'll get a hold of a friend of mine who wrote the DA software for Competition Data Systems (CDS), now works with (among other things) 7 post shaker rigs, and gets calls from Ross Brawn, (yes, THAT Ross Brawn) in his office. We'll see just how much time absolute TOP level professional teams spend looking at FC plots (I'm already guessing none). Heh, I called him one weekend and we were talking about all the races on TV that weekend including F1, Champ Cars, IRL, Speed Touring and GT, ALMS, Trans Am, NASCAR, and I think something else. He had clients in every one of them. I'm sure he can give us the real scoop here.