The online racing simulator
Tire physics
(88 posts, started )

Poll : Which curve is more realistic?

LFS current curve (in red)
48
My attempted curve (in black)
26
Quote from JTbo :Drift cars do use lot of caster, intention is keeping more negative camber when wheels are steered, I think?

if im not mistake they use low caster to keep the camber on the outside loaded wheel (which in nromal caster operation is the inside wheel) from going too far into the positive values

Quote :I think that ISI's engine is not so badly flawed, only how it is used is, I could try to create full LFS tires and suspension of course to see how different it will be, just bit lot of work to do and does it serve as platform to test how different tire curves would affect / benefit handling in LFS as it is still ISI engine, I don't know

axus made lfsish curves for gtr2 a while ago ... they helped a lot to make the cars feel like an actual thing on tyres instead of something that randomly spins

Quote from DaveWS :Right, I am wondering if this curve is more realistic? NOTE that the scale goes right up to 90 degrees.

if that curve was true countersteering would be impossible

and why would they drop of anyway ?

the 0 at 90° is debateable but everything else is wrong
#27 - JTbo
Quote from Shotglass :if im not mistake they use low caster to keep the camber on the outside loaded wheel (which in nromal caster operation is the inside wheel) from going too far into the positive values



axus made lfsish curves for gtr2 a while ago ... they helped a lot to make the cars feel like an actual thing on tyres instead of something that randomly spins



if that curve was true countersteering would be impossible

and why would they drop of anyway ?

the 0 at 90° is debateable but everything else is wrong

Yes, indeed low caster it have to be as more caster more positive camber it gives in normal setup if my brain worked it out right this time, it should be sleeping actually

In GTR2 there is inertia, dampers and tires that use really weird values, imo, also there was something else that I have forgotten already, maybe better so I think.
Quote from JTbo :I think this is because how tire curves are made to them, they are just like black curves in first post pics.

Drift cars do use lot of caster, intention is keeping more negative camber when wheels are steered, I think?

I think that when I get my work done with this mod, it will be huge shock to ISI sim lovers, I already hear comments how all tire curves are wrong etc.

I think that ISI's engine is not so badly flawed, only how it is used is, I could try to create full LFS tires and suspension of course to see how different it will be, just bit lot of work to do and does it serve as platform to test how different tire curves would affect / benefit handling in LFS as it is still ISI engine, I don't know

Try these FFB settings. My aim was to make it feel similar to LFS (ish).

http://forum.rscnet.org/showpo ... 3066965&postcount=147
#29 - JTbo
Slip angles are only part of the issue. I thing the lateral grip curves are close. Longitudinal curves are not right. Pressure and its effects on heating are still a problem as are pressure and its effect on sidewall flex. Those issues are more out of ballance then the grip curves. Did not vote.
Quote from tristancliffe :Why can't that tyre generate friction (and therefore force) at 90°

Surely at 90° slip angle the tyre cannot produce any lateral force because it's being pushed sideways! Any forces applied laterally would just cause the tyre to rotate around its axis, wouldn't it?
#32 - JTbo
Quote from StewartFisher :Surely at 90° slip angle the tyre cannot produce any lateral force because it's being pushed sideways! Any forces applied laterally would just cause the tyre to rotate around its axis, wouldn't it?

But if tyre is being pulled sideways and it has not rigidly attached it would rotate and eventually fall down to it's side, maybe I'm wrong but there have to be some forces here or then I'm missing something again?
Quote from StewartFisher :Surely at 90° slip angle the tyre cannot produce any lateral force because it's being pushed sideways! Any forces applied laterally would just cause the tyre to rotate around its axis, wouldn't it?

yes and no
basically youre right which goes with what i said that a 0 at 90° is somewhat correct as the situation is much more of a longitudinal braking with 100% slip ... but of course if you interpret those graphs the right way a force at 90° makes sense
your wrong though about the rotating ... why would it the force is acting in the dirction of movement with the force centroid being somewhere on the direction of movement ... no torque whatsoever

btw its curious that the endpoints of the decelleration curves and the lateral curves are different in rfactor ... from my understanding tyres should generate the same force at 90° slip and 100% braking slip (not considering sidewall flex camber and anything else that would lead to a non flat contact patch)
Quote from DaveWS :Todd Wasson, I understand now what you have been saying about the sliding of the tyres in your post which axus pointed out.

However. Say that a car is driving around a roundabout at a constant speed. The front wheels are turned to the slip angle which will provides the most lateral grip. Then the front wheels are turned in to say 45 degrees slip angle. Surely then the lateral (in relation to the car) force is going to drop a lot? I have drawn I diagram to explain what I mean.

Tire lateral force curves like the ones the OP posted are lateral force in the tire's plane, not the car's. In the car's reference frame of course increasing slip angle beyond a point will reduce lateral force on the car right down to 0 at 90 degrees slip angle. However, the tire is indeed producing full lateral force in its own reference frame (the tire plane), only now it's acting to slow the car down in a real hurry.

Here's a fancy version of what you posted from some real tire data:

http:/performancesimulations.com/files/tire2.JPG

When I talk about lateral force, I'm talking about it in the tire plane.

Quote :
Edit: Ok, I now believe that the lateral curve I did is wrong. There should be be no "sudden drop" after the peak, but it should drop of linearly (spelling?) after the peak?

It should look very much like the LFS curves in general. The only tires I've ever seen with any significant drop off after the peak in lateral force are big truck tires (even in the dry) and other tires on the wet. Even a street or racing tire in the wet doesn't drop off as much as the curves that were posted originally as an update. Big truck tires from semi trucks and so on do though.

Try googling "tire lateral force curves" or something similar. There are plenty of examples out there. Just make sure you know for sure whether you're looking at a real set of tire data or an artist's rendition of what they think tire data looks like. Unfortunately most text books are chock full of bogus curves which is probably what led to all this misunderstanding in the first place. Also, make sure you know whether you're looking at dry or wet test data when you find it.

Quote :
Also, any more opinions on the longitudinal curve?

Longitudinal curves generally do indeed drop off somewhat at high slip ratios. It's highly speed sensitive though so it's tough to say, and this sort of data is even more rare than lateral force data is. However, it's fairly predictable how it will vary, but I'm not going into that aspect of tire modelling here. :shhh: Sorry

The LFS curves look quite good though in longitudinal as well as lateral. I wouldn't change much, and if I did, you'd barely notice it in sim anyway. It's that close
Quote from StewartFisher :Surely at 90° slip angle the tyre cannot produce any lateral force because it's being pushed sideways! Any forces applied laterally would just cause the tyre to rotate around its axis, wouldn't it?

Put a tire on the ground facing forwards so it will roll away from you if you push it. Now, drag it sideways to the right. The tire is operating at a 90 degree slip angle. There will most certainly be a force to the left

The confusion here in the thread seems to be because some are talking about slip angle in the car's reference frame, while others are speaking of it in the tire's.
Ah, real research and test data found Carry on

Edit: Interesting quote from your link:

"The authors in [5] develop a nonlinear estimator which consistently returns true parameter estimates in simulation to within 3%. The accuracy afforded by the new estimator structure motivated the experimental characterization of the relationship between tire inflation pressure and longitudinal stiffness estimates. This current work goes on to characterize the influence of tire inflation pressure, normal load, tread depth, frictional heating and surface lubrication on longitudinal stiffness and effective radius for two different types of tires. "

And this was freely available on the web, go figure Bet it took all of a couple minutes to find this little tidbit, eh?

Anyway, this is science. This is the difference between real data being used and folks adjusting and fudging and manipulating things to make them "feel right," which has no bearing whatsoever in determining what's realistic from an engineering point of view (and is really the mortal enemy of the real racing industry, where very often all anyone is trying to do is figure out why the car *doesn't* feel right or do what they thought it should be doing all along.)

Science doesn't work that way, and neither does your car. As perhaps illustrated above, there is some rather serious thought given to this sort of topic by folks that aren't quite the epitome of stupidity

Great find. Read it and then read some more and keep looking. The stuff you want to know is available out there online

Looking forward to reading the debates this generates and seeing more folks dig up food for thought!
Quote from jtw62074 :The confusion here in the thread seems to be because some are talking about slip angle in the car's reference frame, while others are speaking of it in the tire's.

Ah, that makes sense. I was assuming the longitudinal/lateral directions were relative to the car, not the tyre. Thanks for putting me right!
Quote from DaveWS :http://www-cdr.stanford.edu/dynamic/WheelSlip/

P.S. I'm not so sure if this will bring up more debates, it's a bit "deep" don't you think? I could be wrong of course.

you can tell they know what theyre talking about from their use of tex

one of the thinks ive always wonderes about it the noisiness of such measurments ... does the tyre really generate that much noise itself or is it the measurment equipment
or from a different perspective is the tyre behavious in sims too idealized
Quote from jtw62074 :The LFS curves look quite good though in longitudinal as well as lateral. I wouldn't change much, and if I did, you'd barely notice it in sim anyway. It's that close

Todd - You are far and away one of the most knowledgeable people on this forum about such things (not to mention physics engines in general) so I think most of us really value your opinion. I wonder if you could expand a bit on this and attempt to explain how LFS's longitudinal grip could be close to reality (enough that we wouldn't notice your suggested changes) yet cars still have more grip than one would expect when the tires are spinning - at the start of a race, for example.
Quote from Shotglass :you can tell they know what theyre talking about from their use of tex

one of the thinks ive always wonderes about it the noisiness of such measurments ... does the tyre really generate that much noise itself or is it the measurment equipment
or from a different perspective is the tyre behavious in sims too idealized

Tires do generate quite a bit of noise during these tests. The graphs that you see published are processed to average/cancel out the noise. I can only recall one pure data sample that's online somewhere (no link handy atm, sorry), and the fluctuations were on the order of maybe 2-4% or so. However, they're very rapid so in the end for all intents and purposes you are essentially running the processed curves anyway.

A sim doesn't lose anything from leaving out this noise. I'm not aware of any vehicle dynamics simulations that make any attempt at duplicating it all. There are other unknowns with the tires, the chassis flex/compliances, etc., that are much more significant than tire force noise.
Quote from Cue-Ball :Todd - You are far and away one of the most knowledgeable people on this forum about such things (not to mention physics engines in general) so I think most of us really value your opinion. I wonder if you could expand a bit on this and attempt to explain how LFS's longitudinal grip could be close to reality (enough that we wouldn't notice your suggested changes) yet cars still have more grip than one would expect when the tires are spinning - at the start of a race, for example.

The deal with starting a race is that the slip ratio curves are not really fixed. They're quite speed sensitive and vary with time and temperature. When you're nearly at rest and spin the wheels you may very well have a slip ratio of 10 or 20. While it's tempting to try to extrapolate out a longitudinal force graph measured at 60mph from a slip ratio of 1 out to 10 or 20 and say "this is about what the force must be at 5 mph with a slip ratio of 10 or 20," it doesn't really hold true. If the measurements were really made at that very low speed at extreme slip ratios, that would be another story.

I don't recall ever seeing any data out to slip ratios of 10 or 20 or anywhere even close to that. Usually that's not an interesting thing for whomever is paying big bucks for the tire testing so it just doesn't ever probably get investigated. Additionally it's rather difficult to really test things in slip ratio much past the peak on acceleration.

I have plans for a way to model all this and got a thumbs up on the ideas from a couple of guys that do a lot of tire testing, but am not prepared to describe how that will work publically any time soon. Sorry to be a poo about it, but this one is where folks will have to do their own research.

Some curves you see in books or online under tractive, positive slip ratio (acceleration rather than braking) show a curve that climbs up to a peak, then suddenly and very quickly swoops down to 60-70% of the peak value and flattens out. You'll see something similar in some sims and folks will emphatically insist that that's how the tires really work. They don't. Real curves look a lot more like the LFS ones than some of these other ones floating around.

Case in point: Have you seen the super slow motion tire video that's been floating around showing drag racing tires? At very low forward speed right at the start there is actually quite a significant amount of slip. Those big drag tires may not be the same animal as others, but it's probably not unreasonable to say that there is a period of time there in at least one of those clips where the slip ratio was as high as 3 or more. If they were losing grip they wouldn't be operating the tires that way. However, at much higher speeds they do indeed lose a lot of grip if the slip ratio goes up that high.
Quote from jtw62074 :Some curves you see in books or online under tractive, positive slip ratio (acceleration rather than braking) show a curve that climbs up to a peak, then suddenly and very quickly swoops down to 60-70% of the peak value and flattens out. You'll see something similar in some sims and folks will emphatically insist that that's how the tires really work. They don't. Real curves look a lot more like the LFS ones than some of these other ones floating around.

I know you've said this before - but it doesn't tie in at all to what is seen in reality with race cars. Excessive wheelspin off the line in a race car results in them dropping back in the field signifigantly.
There is a drop in grip at high slip ratio. Said that a few times now. My point there was that the shape of the curve is not like what you see shown in many places where at 10% slip you have grip of 1, then at 20% or 30% you have something like 1/2 or 2/3 of that at low (or any) speeds. What about at 1000% slip? Sure, there's likely to be a big drop there.

"Dropping back significantly." Yes, of course. But how much grip have they lost there? Are they at 50% the normal grip? 70%? 90%?

I agree that in LFS the wheelspin off the start should probably give less grip than what you currently get. Depends on the tire and especially the rubber compound though. Some don't change a whole lot at high slips unless they're getting really, really hot in the process.
Ok, fair enough. You did seem to suggest that changes you would make to the curves in lfs (if given the opportunity) may not even be noticeable, hence my query I don't know how much grip is reduced when watching them on tv, but it's clear that it's far more than LFS currently exhibits.
Interessting discussion again about that topic

I really hope that this longitudinal grip is fixed. It would make starts even with the current signal system a lot more enjoyable, as everyone would try to give a good amount of throttle.
If you do it right, you should be quick. If you just do a full throttle start, like most people now, you are a bit slower and some starter behind you will overtake you.

And if you try the right amount and fail, by giving not enough throttle, you should be very slow at start.

Much more fun at starts.

I really hope that this is fixed before the free start system is implemented. This would hide this last problem of LFS tire physics because of different reaction times to the signal.
Quote from jtw62074 :Tires do generate quite a bit of noise during these tests. The graphs that you see published are processed to average/cancel out the noise. I can only recall one pure data sample that's online somewhere (no link handy atm, sorry), and the fluctuations were on the order of maybe 2-4% or so. However, they're very rapid so in the end for all intents and purposes you are essentially running the processed curves anyway.

A sim doesn't lose anything from leaving out this noise. I'm not aware of any vehicle dynamics simulations that make any attempt at duplicating it all. There are other unknowns with the tires, the chassis flex/compliances, etc., that are much more significant than tire force noise.

I've always thought that the use of smooth curves rather than noisy ones is why "sim tyres" don't regain grip as well as real tyres when recovering from a slide Since the last physics update in LFS though they are very good at this aspect, but still feel it could be better The slicks are probably very close to what they should be, but the road tyres while already good still need abit more attention imo and I would guess road tyres generate more noise...

My interpretation of the noise is it would soften the transitions abit and give more "feel" for being able to recover from a slide (not my field of expertise though so probably talking out of my ass ) A little knowledge makes you more dangerous than having alot they say

It occurs to me that the areas that have the least amount of data available and therefore are the hardest to get right are locked tyres when under heavy braking, grip in very high slip angles (such as recovering from a slide) and wheel spin at low speeds.

But I think your probably are correct when you say that averaged curves when compared to the raw noisy curves wouldn't yield great differences in how a sim feels, I'm guessing it has more to do with how tyres heat up, cool down, how they deform under diferent pressures and different circumstances and how this is simulated thats would have the greater impacts, as well as chasis flex etc makes the mind boggle

Quote from Blowtus :I don't know how much grip is reduced when watching them on tv, but it's clear that it's far more than LFS currently exhibits.

I see alot of variation already in the starts, so I don't think its far off... maybe just a few % is all. For example a flat out start at BLGP in the XFG gives an average 0-100 time of 7 sec where a controlled start averages 6.8 sec and a totally botched start you loose more than half a sec. So I think it's not too far off just needs alittle more tweaking And obviously the more powerful the car the more the variation should be.

Quote from RIP2004 :I really hope that this is fixed before the free start system is implemented. This would hide this last problem of LFS tyre physics by different reaction times to the signal.

It's also interesting that the curves for longitudinal grip is different for acceleration and braking although probably not as different as some texts suggest, I wonder if this is already a factor in LFS... and the curves for the driven wheels would be different to the no driven...
I see where you're coming from of course. Getting it proper might really take more than one curve. Also, I don't know if that longitudinal slip curve there is for braking or traction. If it's traction, what does it look like at slip ratios of 2-3 or higher? For braking, that curve looks very close to other data I've posted here and at rsc in the past, much more so than in any other sim I've seen data plotted from.

The curves that end up being spit out in my model will look different at different speeds. This will be done at a pretty low level so I can't be sure just yet what they'll look like, but they'll come about as a result of other parameters specific to the type of rubber itself and a couple other things. I haven't checked to see what a slip ratio of 3 at 5mph versus one at at 100mph will look like yet, though. And data like that is quite rare. I don't recall seeing any myself yet either

Tire physics
(88 posts, started )
FGED GREDG RDFGDR GSFDG