Discussion in 'General Discussion' started by davehenrie, Sep 4, 2017.
What track do you drive in a straight line that many km
Tire deform constantly at every revolution, this mechanical work degrade the structure at a microscopic / molecular level through the whole tire, not only in the outer surface.
I am not sure if I am understanding correctly your message. So if my reply seems unrelated somehow is due to it. Sorry about that.
As I understand it tread wear has nothing to do with shoulder damage produced by fatigue due to a high hysteresis. They are different failure modes and would need to be considered differently and not additive.
Straight line driving does have some tread wear mainly due to convergence induced slip angle. I haven't tested if pressure is affecting this wear as it apparently should. Are you saying straight line driving has no wear?
Thanks for clarifying.
From what I've read the deformation happens mainly on acceleration and braking on a straight,but I guess for high downforce cars that's different because as the speed is increasing so is the load.
I always questioned what I read because if no wear happens then why would a Bugatti veyron shred its tyres in 15 mins at full chat
I'm confused. Any time I've looked at tyre wear - to high precision - in rF2, nothing happens on the straights. Considering tyres are rarely/never pointing straight, and I'd expect some wear even if they were, this doesn't seem right.
Spaskis, you said low pressure should produce heat and wear. What wear did you mean? And why can't that 'simply' be tacked onto what's already happening to an adequate degree to be believable, without striving for perfect modelling?
In RL you always want to use the lowest tyre pressure possible, because it means bigger contact patch -> more grip.
But the downside of low pressure is, you get more tyre/carcass flexing (especially from sidewalls), so more base heat through the whole tyre. When you enter next corner, and tread surface begin to slide in addition to that already pretty high base heat (from flexing), your tread surface overheat more quickly. Car begins to feel "greasy" and unprecise and you loose grip of course.
More surface heat means also more wear.
In rF2 temp conduction between rubber/carcass and surface seems a bit "dull" to me. Also there's to much grip if the tyre is to hot or to cold. This means in rF2 you can get away with heavy high speed slides, while in RL the tyre simply give up, because of very high surface temp.
After some tests I can confirm that there is no tire wear in the straights in rF2.
I can also confirm that in rF2 tire heating in straights is unaffected by
- tire pressure
- tire convergence
Low pressure should increase sidewall wear due to fatigue but this is not modelled ingame. It should also change the wear in the tread. Low pressures should wear inner and outer sides and high pressure should wear the center.
The last question I don't understand. In any case it is quite evident that, from a thermal point of view, rF2 tire model is far from being accurate.
The last tire model I fooled around with was the GTR2 tire(for the SCC mod), which was based on the first pre-patch release of rF1. You could set the tire to heat up from rolling friction or sliding friction. I altered the normal tires to heat up more from rolling frinction for the Le Mans cars as the normal tire was not really getting up to temp on those long Le Mans straights. The trade-off for GTR2 was that a tire that heats on the straight from rolling friction would burn-up in just a few laps at any other track. I had to repeatedly overstate NOT to drive the Le Mans carset on normal tracks and also not to bring ALMS or LMS cars to Le Mans as the tire would be cool most of the lap.
I took one look at the rf2 new tire and about popped my brain out my ears. What a huge difference. I would like to see if a mushy sidewall could be implemented when tire pressures drop too low. That might be sufficient penalty for all these rock bottom tire pressures. If the sidewall flexes too much, due to low pressures, the car should, in theory, resist turning in.
Ok, let me give an example so I'm clear.
I asked above about the effects you're saying low pressure should have, and you've answered as I've quoted.
First things first, sidewall wear/degradation/fatigue(?) is something that rF2 doesn't do anything with anyway. Let's assume for the moment we're accepting the game doesn't deal with that sort of thing, so we probably won't get tyres tearing themselves apart due to misuse, and we don't expect that to be modelled.
So that leaves heat, and wear. Here we run into a bit of a problem. The ttool shows uneven pressure with non-optimal pressures from what I've seen, though I'm not sure that translates through to heat and wear the way we might expect. More importantly, there's no telemetry to tell us. Wear is a single figure for the whole tread. So there could be uneven wear and we can't see it, except what we can actually see on the screen. I haven't done any extreme testing to see how well this works, though I feel like I've seen videos/screenshots of uneven wear, but I could be wrong.
As far as something we can measure and therefore test, we're back to heat. An excessively low pressure (remembering here we're talking about low to the point of being dangerous in reality - to use F1 as an example again, Pirelli is keeping teams above certain pressures for structural integrity. The teams want to go that low, which suggests grip/handling aren't an issue, so pressures low enough to enter bad-handling territory are probably well below what we see on the real cars) will lead to excessive heating especially on the sides of the tread. If that were to happen with very low pressures, then using the tools we have we could suggest that is behaving correctly. A parameter could be used to effect that kind of heating, and already the WLF parameters will increase wear for parts of the tread that are excessively heated (but not on straights ).
This reminds me a little of laserscanned tracks. Some people who want laserscanned tracks talk about non-LS tracks and how it's not fun because they can tell it's not right when they drive it. My question is, if a track is done close enough to correct to not be noticeably wrong, why is that any worse than LS? And again with the tyre model - though in a particular aspect it might not be right, if a parameter could approximate correct behaviour well enough to be indistinguishable from a theoretical perfect model (within reason), what's the difference?
So I ask what the real-world effects of something would be, to ascertain if that real-world effect could be dialed-in using another method. And if so, why not just do that instead of wanting the model to be perfect?
Anyway, this is all rhetorical. Talking about this stuff won't make anything in the game change. But I guess it kills time.
No, it also leaves handling.
I have a good idea from real life racing of how a tire would perform depending on tire pressures and also construction.
For example there's specific handling degradation related to using low tire pressures.
Or sidewall hardness affects how quickly the tire heats up and how easily it overheats, how snappy it is on the limit etc, but ultimate tire grip is not affected.
All these things I expected to be simulated but are not, apparently just dropping the tire pressures to their lowest setting is the first thing one would do and it hasn't failed me so far.
I think you are complicating things. Keep it simple.
Should pressure make any difference wrt heating? It should but it doesn't.
Should convergence affect heating?
It should but it doesn't.
Heating should come primarily from tire sliding and hysteresis. When two key setup variables like pressure and convergence that should affect the mentioned heating mechanisms do nothing, the whole thermal model must be questioned.
Regarding tire wear. You say now it is a unique figure for the whole tread. However when in other threads we talk about flatspotting apparently it is being accounted in a node basis.
Or one or the other, but not both.
So in conclusion. None of the real aspects that limit using extremely low pressures is being considered in the tire simulation.
No sidewall lifetime
No thermal influence.
No higher wear in tread sides wrt the center.
As Christos said, lowering tire pressures seems to be the first thing to do when messing with the setup. The only reason for not using minimums should be uneven tread temperature coming from a questionable thermal model. I am assuming here that temperature effect in grip is being correctly considered. However I have no proof that the same type of simplifications as for pressure effect are not being used.
Thanks for the explanations. Can you describe how load sensivity should be altered by pressure changes and how this could effect overall grip?
It is not neccesary to involve heat effects. The issue is not caused by thermal model. We tweaked tires to remove heat activity (constant temperatures) and no improvement was observed (fastest lap times were also achieved with extremely low pressures)
I think most of us have reached the same conclusions in this thread. That is, there are not mechanisms in place in rF2 (or most other sims) to effectively put a reasonable 'floor' on tire pressures, that in real world would have those tires fail or give other performance detractors. So right now, the advice to go 'as low as you dare' on pressures seems to be the right one.
I'd note, iRacing have this issue also on low ambient temperatures; the tires do become unmanageable on the first lap(s). iRacing have also introduced a huge fudge in 17S4 to actually have tire warmers on the FR 2.0 because too many people were crashing in these conditions!
What we seem to have is a collection of models that work in a very narrow band. I think of it like a line fitting exercise. If you have a single data point in the middle of the range, you might hit on parameters that make your simulated value reach that point. But as you move away from it up or down, the slope of the line not being correct gives you greater and greater errors. So, the model does not work over a reasonable range of values, but does appear to work for one specific set of values.
Lazza, Dave Henrie gave you a perfect example of why you don't want to fudge one parameter that you have to try and get a good answer in one case. In his example, he took one of the parameters available to him and fudged it to 'make it work' in one specific case; but then it didn't work in all other cases. This has been the story of racing sims since the beginning of time. This is why you can't do what you propose. It only works if the range of other input values that result still work to correlate to actual outputs. They don't. Fixing one thing in one area causes it to be off to a greater extent in another area.
I worked for a long time on optimization problems. Many of them fall foul of a 'local minima', i.e. some combination of values that appears to give a low result; increase or decrease the inputs, you get larger outputs. So you think you have found the 'global minimum', but in fact you haven't, you just need to get out of that range and then the outputs will again reduce. This doesn't directly describe what is happening here, its simply an analogy. But what we have a sets of input parameters, that if carefully controlled, can give realistic outputs some of the time. But the models don't work across the range of values we should have, and there are also limitations/exclusions in the models which mean they are missing key inputs and hence will never result in good outputs.
If thermal or wear effects are not considered, lower pressure means a wider contact patch. Distributing the load on a bigger surface will reduce contact pressure. Despite Coulomb's friction law states that friction coefficient is independent from load this is untrue in many cases. As contact pressure increases friction coefficient is normally reduced. The reduction is slight but noticeable. For sure smaller than the pressure change so increasing the load over a constant surface will always yield an overall higher friction despite the friction coefficient reduction.
So basically at the moment the only thing preventing from using lowest pressures could be handling problems or as said uneven temperature distribution. Grip will increase when pressure is reduced as far as contact patch increases as well.
It would be interesting to modify a car increasing the available pressure range without lower restrictions to see if grip keeps improving with ridiculously low pressures below 1 bar.
Does anybody know what happens if lower pressures compared to the range used for defining lookup tables are used?
I read Michael Bordas development diary but cannot remember how many dof are included in the lookup table generation. I'll probably take a more detailed look at it and why not mess a bit with tire tool to see if I get to understand better what the simulation engined is doing with all that data.
One of the best thing about simulation is that testing is free. You just need time, patience and methodology.
There is thermal influence. You get higher rubber and carcass temperature with lower pressure. There is also thermal condution from carcass/rubber to tread surface, it seems pretty low but it's there.
You can try the following: Drive a few laps and at the end of a straight you come to a hold (without locking) , you should see now (on Motec for example) that tread surface temperature rises for a few seconds. The reason is, you took away (almost) air cooling so the pretty cold tread surface gets heaten up by higher carcass/rubber temps, until both are even out and surface temp start to drop again.
A good indicator for this, is also the Motec chart from Raintyre (post #52).
I think (can't prove), there is also temperature generation from sidewall flexing.
I'm sure that the guys at studio 397, know all this (and a lot more). We know they working on tyremodel, so let's wait and see what future brings.
@SPASKIS I said that we can only measure overall tread wear. Telemetry provides one figure. I didn't suggest the nodes don't have individual wear. The point is we can't tell how uneven, or not, wear becomes in different scenarios because we can't measure it.
I'll make a longer test to see if it makes any difference. Yesterday test showed similar results under extreme pressure and convergence settings.
But please answer. What heats tires in a straight line? (In reality I mean).
After that think about a setup variable that should make a significant contribution to that heating factor. Test changing that value and compare results.
Well on a straight line you have friction (between tyre and tarmac) rolling resistance and suspension geometry (toe-in/out). Even if you don't accelerate or decelerate your driven Wheels have to work against drag which generates additional friction. There's still inner friction, so carcass flexing (vertical and torsion forces).
To get as much heat generation on a straight line i would maximize the following: Toe-in and -out, camber, downforce and springs/dampers. Tyre pressure to lowest.
I don't think there's one setup variable that have significant more impact then the other, but i think tyre pressure should have the greatest influence here.
I just did a short test with the McLaren 650S at Silverstone int. about 8 laps. One stint with 200kPa, and one with 140kPa. Low pressure gave me about 5-9°C more temp at RR, 15-23°C more at RL, almost same at FR and 2-10°C more at FL. all carcass and rubber temperatures (surface temps are not very viable here).
One question: Is there any parameter in physic files, which determines if the tyres are covered by fenders. For example a specific slick mounted on a open wheeler should cool down some "X" degrees on a straight. The same tyre mounted on a GT-car should cool down a lot less now.
>>>Grip will increase when pressure is reduced as far as contact patch increases as well.
That's the point i don't agree,
Due to a handful of effects (some of them included or described in overall as 'load sensivity') real tires cannot provide more lateral force when deformation reaches a certain point. The transformed shape simply can't translate friction into lateral force properly.
At optimal pressures, loads applied must be really big to reach the point where no more benefit is obtained. However, some Group C prototypes experienced strong issues on early 90's due to massive downforce that some tires couldnt translate into more corner speed because of their excessive deformation (especially at high speed corners).
Underinflated tires obviously deform much more. They need much smaller loads to reach the point where the whole tread is transformed to a shape which fails to produce the desired lateral force.
We call it load sensivity or not, there is no doubt this effect doesn't benefit the use of low pressures against high pressures to produce effective grip, hence there is a limited pressure range for optimum grip (which actually is lateral force).
That's why i think that excessive ratio load/pressure should be penalised in the sim with disturbing effects and drastic lose of grip, just because that is what happens on real tires (not only overheating or worst handling). Pressures like 15 kPa , 1kpa!! shouldnt be driveable. Suposedly, tgm lookup tables usually include results of tests at 0 tire pressure, but that is not working so far.
I pretty much agree with the mechanisms except that you mentioned toe in (that's friction as well).
I also agree on how to setup things to maximize heating on a straight. IMO in straight line, spring and dampers shouldn't make a difference considering a flat tarmac: tire loads won't change in such a case regarding suspension and damping.
In any case, if you had read my previous posts you would have noticed that I already had made those straight line tests with no difference in tire temps. Camber changed distribution but pressure or toe in didn't.
Your stints are not valid for such testing. I have a flat surface track test for this purposes.
So, as I said, heating aspects are not correctly implemented. Whatever is heating tires under straight line is not being sensitive to parameters that should make tire temperatures change. That's why I asked what should heat tires in a a straight line.
They probably affect during turning but it seems to be an indirect effect. I will also make some constant radius cornering tests soon to see what is happening there. Hopefully I will have some time over the weekend.
Separate names with a comma.