Community Tire Development Project - Building Better rF2 Physics Together

Looks like a well thought out approach so long as the driver feel is not compromised.
The less "driftable" the tires are, the less margin of reserve feedback they provide to the driver and thats already rather weak in rFactor2.

Crunching tire algorithms consumes the majority of the physics engine and ISI discovered a very efficient algorithm but not without its problems.
Thermal tire physics are way off from the real world in that they have a very narrow and somewhat linear band of thermal resistance characteristics.

Best luck with success.
 
For years sirmacers almost universaly spoke about how rF2 has best FFB, and now I am reading @John R Denman telling that tires feedback to the driver is weak. I couldn't disagree more. Also I couldn't disagree more about the statement that if tires will be made to be more planted, sharper, less driftable, more pointy, or however you'd like to call it, that they would reduce reserves of feedback to the driver. It is most certainly the opposite what happens. The real drawback is that such tire becomes more challenging, takes more consentration, it is more sporty, and for someone who can't really keep up with it, it perhaps could possibly seem then that feedback rate is going down.

As always when talking about physics, if talking gets to much biased about feelings, then it gets nowhere.

And could you elaborate about norrow thermal resistance characteristics ? Perhaps you already did in the past, and I forgot. Personally I couldn't say that I have reached high levels of understanding how to manipulate TGM thermal properties related parameters. But I can achieve things that I need in most cases.

Plus when talking about temperatures tires or brakes the mixing up of core temperatures to temperatures of the very surface get confused all the time. Obviously, the surface will be getting fluctuating in temperatures a lot more.

There are plenty of temperature related parameters in TGM realtime section.

1. WLFParameters are very interesting, but I have not advanced to handle those, at best I can shift glass transition temperature a bit. Can't tell with confidence what exat effects are, other than what is expected.
2. ThermalDepthAtSurface and ThermalDepthBelowSurface works well in controlling how steady or how much fluctuating temperatures are.
3. GroundContactConductance for wet and dry surfaces does work well helping to control how fast or slow tire will be cooling off. Taking in account pressure of tire.
4. InternalGasHeatTransfer and InternalGasHeatTransfer how much heat gets transfered to gasses around the tire material, taking velocity into account.

Then you can very easily control grip versus temperature with StaticCurve parameter.

Then you can alter how tires abrasive wear is versus temperature.

I don't understand where the issue about tire thermal conductivity/resistance is in rF2, is it there ?

To be honest my last test wasn't successful when I tried to register any tire thermal degradation, but it was less advanced Howston G4 based tire model, so I still don't know if that feature does actually function in rF2. It is very important for endurance races.
 
mantasisg said:
For years sirmacers almost universaly spoke about how rF2 has best FFB, and now I am reading @John R Denman telling that tires feedback to the driver is weak. I couldn't disagree more. Also I couldn't disagree more about the statement that if tires will be made to be more planted, sharper, less driftable, more pointy, or however you'd like to call it, that they would reduce reserves of feedback to the driver. It is most certainly the opposite what happens. The real drawback is that such tire becomes more challenging, takes more consentration, it is more sporty, and for someone who can't really keep up with it, it perhaps could possibly seem then that feedback rate is going down.

As always when talking about physics, if talking gets to much biased about feelings, then it gets nowhere.

And could you elaborate about norrow thermal resistance characteristics ? Perhaps you already did in the past, and I forgot. Personally I couldn't say that I have reached high levels of understanding how to manipulate TGM thermal properties related parameters. But I can achieve things that I need in most cases.

Plus when talking about temperatures tires or brakes the mixing up of core temperatures to temperatures of the very surface get confused all the time. Obviously, the surface will be getting fluctuating in temperatures a lot more.

There are plenty of temperature related parameters in TGM realtime section.

1. WLFParameters are very interesting, but I have not advanced to handle those, at best I can shift glass transition temperature a bit. Can't tell with confidence what exat effects are, other than what is expected.
2. ThermalDepthAtSurface and ThermalDepthBelowSurface works well in controlling how steady or how much fluctuating temperatures are.
3. GroundContactConductance for wet and dry surfaces does work well helping to control how fast or slow tire will be cooling off. Taking in account pressure of tire.
4. InternalGasHeatTransfer and InternalGasHeatTransfer how much heat gets transfered to gasses around the tire material, taking velocity into account.

Then you can very easily control grip versus temperature with StaticCurve parameter.

Then you can alter how tires abrasive wear is versus temperature.

I don't understand where the issue about tire thermal conductivity/resistance is in rF2, is it there ?

To be honest my last test wasn't successful when I tried to register any tire thermal degradation, but it was less advanced Howston G4 based tire model, so I still don't know if that feature does actually function in rF2. It is very important for endurance races.
Click to expand...

Most my experience comes from real world despite having been an rFactor user since 2007. FFB was basically reversed in rF1 and in rF2 its improved but still lacks the fidelity with turn-in, inside wheel slippage and threshold braking.
Crashing a hand built Ralt SuperVee is a little more costly than crashing a sim car. I've done both.
That's why race tire manufacturers are so emphatic about building reserve into all their tires.
In GTP racing the heat is so brutal that Michelin's new Pilot Sport Endurance racing tire was designed to provide more reserve and durability with reserve being the primary goal. In the real world reserve is the predictability that enables a driver to drive at the limit.

So yes, I feel that rFactor is weak with feedback that was never a focal point for S397. Compared to the games I can't really say much other than LMU which doesn't tune like the real world where rF2 tends to be close.

Lets take a deeper dive into tire heating.
Thermal increases occur mostly from rolling hysteresis, tire slippage, and even some heating from the brakes (which impacts tire pressure more than contact temperature) with cooling mostly from convection with the air, with some road conduction, infrared emission and particle release.

Convection is the vital one. At 40 MPH there is boundary layer of air about 1/2" that separates cold air from hot air. At 80 MPH that boundary layer drops to about an 1/8" which doubles the thermal transfer rate. At about 95 MPH the boundary layer slows enough that it again doubles the thermal transfer. Same technology is used in Impingement Ovens I've designed for rapid cook or solder reflow.

I've used MoTec to measure in rF2 tire temperature deltas and compared them to real world. Thermals are way off.
I haven't played with physics files much in the last few years since I'm not tuning anymore. Not even Mining Trucks (small chuckle).

I don't expect any miracles from MG with rF2 I just hope they survive. As for other "Sims" I have little interest even with LMU.
 
@John R Denman Thank you for great detail in response.

To be honest, if I wasn't modding cars for myself in rF2, I would have probably used rF2 and perhaps simracing at all 90% less than I did for past six years. I am also having good success for my personal liking of what I want from car/tires, and it looks like the end of potential of improved results is still rather far. That is encouraging and motivating. I am also happy about tires talk emerging, hopefully it will go for real, and won't be just some sort of signaling act, about supposedly carrying and hard thinking about physics.

A lot in rF2 can get better than it is depending on what car did you drive, if it was well configured. And if it wasn't, then it would be always interesting to find out what exactly causing what you don't like.

I think that a lot of content physics, car tires in rF2 were built by careful approaching, avoiding bold decisions when data was lacking or in similar circumstances.

It makes sense what you talk about tires needing reserve to be safe and drivable first of all. Maybe lots of race tires does achieve it to a degree. I tend to think similarly about tires. However, it also seems that there are always tradeoffs with tires, and giving tires reserves, safety margins, so they don't act too unpredictably and too instantly conflicts with the aim of achieving highest possible performance out of them. In simulations, however, it is possible to make tire that excells at everything and has no drawbacks. To an individual, lets say the custom taylored tire for you, or me, or just principally perfect tire theoretically - splendid grip in all circumstances, no wear, good at all temps, super drivable, have to put effort to spin out. But developer often makes tires for multiple thousands of people, not always genuinelly following principles of simply doing it for realism, because perception of gamers is more important. I guess this is why lots of tires in simracing are made to permit more and harder sliding, sometimes they are straight forwardly made for sliding and hardly even grip up normally. The worst part about it is that we can never agree what is that right spot to be at, despite lots of us having real driving experience, having seen tons of cars videos which would mean that we should know what to expect from cars. So even when right tire, on correctly configured car drops from time to time by some developer, it did never receive "this is it" message spreading all around, there is always something wrong for someone. On the contrary, I have seen developers scolded more than once.

Lets say simracing tires are BOP'ed. And that means not that they are made to be liked equally, but rather disliked equally by all. And this means success in simracing.

Enough of the silly rant. I am sorry.

I want to thank for sharing interesting information about how velocity impacts tire cooling. I have never paid proper attention to it, but I think this parameter might do it:

Code: 
ExternalGasHeatTransfer = (base, mult, power) - heat transfer coefficients to external air =
base+(mult*(vel^power)), where vel is linear velocity of tire.
 
Last edited:
mantasisg said:
I want to thank for sharing interesting information about how velocity impacts tire cooling. I have never paid proper attention to it, but I think this parameter might do it:

Code: 
ExternalGasHeatTransfer = (base, mult, power) - heat transfer coefficients to external air =
base+(mult*(vel^power)), where vel is linear velocity of tire.
Click to expand...
This is something I've been using heavily for some time now. Throughout the years I've seen the common odd behaviour on the surface temps where the cooling flatline early on straights, and even start to heat up. Possibly a more deep underlying issue with the crr, but even with a low crr you tend to get this behaviour. Generally I've been setting low 1st and 2nd params and purely adjust the cooling based on 3rd param, which has been working well. You can probably also get some more dynamic behaviour of the bulk temps if you run some extreme stuff on the InternalHeatTransfer, but that's a more experimental thing, and likely not that correct thing.
 
@Robin Pansar Thank you for the tip. I think people could share some findings like this, "secrets", that might help others to get best of tgm tires, hopefully.


I have small tips for rain tire parameters, although general tire grip composition is very important to have tire act properly in the wet, it is even more important then for a dry conditions.

THERE IS REALY BAD LACK OF ATTENTION TO WET TRACK DYNAMICS

- First, setting temporary wet grip loss parameter higher for wide tires, and obviously for less grooved tires. Wet tires tends to be narrower and grooved.
- Second, reducing dampness effects less for static friction and less for macro roughness, more for micro roughness and sliding friction obviously.
- Third. Adjusting wet track temps conduction respectably to how wet softer compound tire would be holding temps higher - supposedly.
- Fourth. Dropping glass transition temp a bit comparing to dry compound.
- Shift static curve so it gives bit better grip at lower temps, and worse at high temps than dry compound.
- Have load sensitivity a bit higher since softer compound would press into road surface macro irregularities fully sooner, also because it is treaded.

If you ever noticed how racecars tend to behave on wet track, it is not constant sliding and undrivability. Cars are still drivable, especially on proper rain tires. But they seem to become twitchier and snappier, this is because transient shift of grip to slip becomes less progressive and more instant, but the sliding usually still remains controllable as long as driver is good and careful. Even better, when drivers in wet are capable to overdrive and be faster a little bit more confidently and safely, they are heating up tires, which then allows fast drivers to go even faster. This doesn't happen in dry races.

Sure we don't have puddles and slippery main racing lines in rF2, but it doesn't make existing rF2 rain driving obsolete. I just don't see well made rain tires very often. Maybe it is also because they aren't made for the dry conditions properly in a way that would then simply work in the wet.

I'll tell this. Realistically behaving tire in the wet, must be realistically behaving tire in the dry too. You can't have rain tire which is not realistic in dry, but realistic in wet. You can though have good realistic dry slick tire, that is still unrealistically great in absurdly wet track....
 
I see interesting things here. Meanwhile, for the Mini, I have been doing some test work that can give a basic example of just a few tiny tweaks in the TGMs.
Some finetuning is still needed, but the behavior is becoming more realistic.

Example of before and after tweaking based on the recalculation of logical values that are still WIP.. My idea is to do some math based on tire size to get a proper air volume, contact patch, thickness, slip curve, expected lifetime, compound hardness, and some good averages.
This is why I said, Hybrids, based on the data I have reverse-engineered from the Pirelli and Michelin tires.

I started with A dev corner tire that was resized to the specifications of the race tire we needed, then did a proper calculation over it to make a new lookup and to make sure all was adjusted. However, as a good start, I always go for a few Spa Francorchamps laps, followed by other tracks. More track results will not fake the outcome. The car has that typical feeling of FWD and has a lot of fun factor to it. It felt a little bit off, hard to point the car, and especially for this tiny car, it felt like a tank. I did the same inputs at Eau Rouge, but the car behaved as if the track was very damp. (What you would see in Spa for real at a cold morning), in rF2 the track temp was 25.9 degrees.

The behavior on 2nd lap after a proper outlap, 3rd, 4th lap very similar:

peakdev.png




Then I switched in-game to use the updated TGMs that also had new calculations applied, a tighter grip curve, and better influence on temp vs grip. It has a little bit too much grip now to my taste, but when I watched real-life actions, the car behaves very similarly now, as the car physics are done very well in both examples: The feeling on FFB was instantly better, after 3 laps the tires behaved stably, and I could start to push the car. Something that, with the other tires, resulted in strange oversnap or serious drifts.


And the data again:
fixeddev.png



This is still undergoing development and is at an early stage. I am thinking of putting a GitHub online with table-based averages per tire size and the expected values you should have. However, I did not do that yet as I am not sure that it is ok just to show ISI, tgm code online. At least I will add a few lines in the info that are made clear in big font that the code falls under MSG/S397 copyright.

If that is a good idea, all can be tweaked/updated by participants to perfection, the more we test/develop them..

It is a lot of work to pull this off alone, you know ;)
 
Last edited:
Frederick Alonso said:
If that is a good idea, all can be tweaked/updated by participants to perfection, the more we test/develop them..

It is a lot of work to pull this off alone, you know ;)
Click to expand...

Indeed it is a lot of work and sometimes turns into Trial & Error.
(It was said that Edison discovered 9,999 ways not to make an incandescent bulb. I can relate to that.)

Thanks for all your work so far!
 
I know I shared this link previously. But this may have some additional notes to add. I really need to thank Lucien on this as it was my main source of info when I was working on the Swift tires.


As listed on that link, there are some items in the .TBC that not in the .TGM (SpringBase for example). The thing with the .TBC is that if you make a change, you need to close out and reload the whole game in order for the changes to take effect.
 
Interesting to see such replay comparison before and after. Bold difference in my eyes. I think it could possibly act even more sharply, but could be just right.

It is really doable when you try and get to know TGM. One thing I always struggle though to get dynamics that I want for longitudinal slip, especially for accelerations, always seems too planted for me.

Snappy tires are dinamic tires, it means they reach a lot of static friction, and at times kinetic friction and static friction can differ significantly, but also (unfortunately) alter very quickly. This is something that simracers simply can not accept, therefore true hardcore realism is never about to happen. But we can do it in a sim, at our own risk to be compared to iRacing negatively (of course iRacing is wrong in different ways).

Also in these situations aerodynamics is part of the blame, but it is a lot due to tire.

Also if you don't get tires acting this way, you also aren't going to achieve realistic wet track handling.


Look how fast and pro-actively driver has to react to grip switch from sliding, to static grip. Split second too late and he would be sent to the barriers. This is driver that I know personally, he is very talented and has a lot of great experience in racing. A year or so later this car was driven by young, unexperienced driver and he lost the rear at the same place in this track, he wasn't quick enough to straighten steering, and got a very expensive tank slapper into barriers.

It doesn't mean that in simulation every single slip has to be brutal, but there is video evidence that suggests it being so at some occasions. Ignoring this evidence is detrimental to simracing. And is also the reason why some people are over confident about obtained skills in simulations, when they aren't in fact prepared for these moments, when simulation is just not doing it.

@Chris Lesperance I have used this source too after you shared. I have never worked before with this advanced tire model of rF2. I have found it superior to previous ones I used (such as based on Howston Dissenter, or Cobra). It is quite complex to get it properly when yo uhave so much control over adjusting bristle springs versus rotational speed, pressure, temperature and base value... and having it repeated four times. But it gives a possibility to get pretty damn good tire, acting right in very wide dynamic range.
 
mantasisg said:
Click to expand...
I was at the gstand just above this cam angle. Remember that it was a colder moment and still some wet spots here and there. Always very tricky.
There was also an Audi that crashed that morning at top Raidillion. Also similar but the other side going with a snap straight into the walls to show how the car looks without the outer shell. :D
 
Sad for the cars, and hopefully no serious injuries.

Speaking of tires, seems like tires usually act snappier when they are not up to the temp. I guess there must be very simple logic to that. When sliding develops, they heat up, and end up warmer and grippier than before the slide, at least at surface level. Then driver doesn't anticipate it, or just being unlucky, rear grips up firmly and quickly while he is still correcting and BOOM
 
mantasisg said:
Sad for the cars, and hopefully no serious injuries.

Speaking of tires, seems like tires usually act snappier when they are not up to the temp. I guess there must be very simple logic to that. When sliding develops, they heat up, and end up warmer and grippier than before the slide, at least at surface level. Then driver doesn't anticipate it, or just being unlucky, rear grips up firmly and quickly while he is still correcting and BOOM
Click to expand...
Exactly that. The range for error on cold or overcooked tires is very tiny. Cold, they can really catch you off track if you are not careful.
The rubber is stiffer on cold vs on temp, while overcooked, they are gummy bear candies.

As we also see at the Lambo, a bit wet here and there. Cooler tires, then he snaps and tries to countersteer while they get up to better temp.
The grip kicks in, but the input was more than expected for the driver, while he goes into a slide and cannot save that obviously.
Not saying he is a bad driver, but experience on such conditions and the car itself might have played a role here.
 
overattack22 said:
https://imgur.com/a/TjnEZ17

Is there a way to add new code commands to make RF2 tires bounce up and down like this?

I think this physics is incredibly important in sim racing, especially when it comes to shock absorption for vehicles that drive over curbs.
Click to expand...
That is why I brought this up , so far not being able.. But we should investigate that with some people.


In your example, it is a similar behavior starting to oscillate on the F1 suspension, perhaps. Add to this that left and right have this non-parallel...
 
Robin Pansar said:
Never seen tyre flex like that from longitudinal load, but it's very prominent from lateral load. Got an old clip of it.
Click to expand...
https://imgur.com/a/3fcIZHv

Let me show you another scene.

The part I'm highlighting is
LIVE FOR SPEED and
AMS2 (latest patch).

In the case of tires,

when you step on a curb,

the tires move up and down.

RF2 calculates tire physics in real time, but it doesn't do this.

From my perspective, it's not that difficult, and
it seems possible to do the calculations by injecting new code commands.

However, stuidio397
isn't making these improvements...

They're only focused on developing new games.
 
Frederick Alonso said:
I see interesting things here. Meanwhile, for the Mini, I have been doing some test work that can give a basic example of just a few tiny tweaks in the TGMs.
Some finetuning is still needed, but the behavior is becoming more realistic.

Example of before and after tweaking based on the recalculation of logical values that are still WIP.. My idea is to do some math based on tire size to get a proper air volume, contact patch, thickness, slip curve, expected lifetime, compound hardness, and some good averages.
This is why I said, Hybrids, based on the data I have reverse-engineered from the Pirelli and Michelin tires.

I started with A dev corner tire that was resized to the specifications of the race tire we needed, then did a proper calculation over it to make a new lookup and to make sure all was adjusted. However, as a good start, I always go for a few Spa Francorchamps laps, followed by other tracks. More track results will not fake the outcome. The car has that typical feeling of FWD and has a lot of fun factor to it. It felt a little bit off, hard to point the car, and especially for this tiny car, it felt like a tank. I did the same inputs at Eau Rouge, but the car behaved as if the track was very damp. (What you would see in Spa for real at a cold morning), in rF2 the track temp was 25.9 degrees.

The behavior on 2nd lap after a proper outlap, 3rd, 4th lap very similar:

View attachment 56853



Then I switched in-game to use the updated TGMs that also had new calculations applied, a tighter grip curve, and better influence on temp vs grip. It has a little bit too much grip now to my taste, but when I watched real-life actions, the car behaves very similarly now, as the car physics are done very well in both examples: The feeling on FFB was instantly better, after 3 laps the tires behaved stably, and I could start to push the car. Something that, with the other tires, resulted in strange oversnap or serious drifts.


And the data again:
View attachment 56854


This is still undergoing development and is at an early stage. I am thinking of putting a GitHub online with table-based averages per tire size and the expected values you should have. However, I did not do that yet as I am not sure that it is ok just to show ISI, tgm code online. At least I will add a few lines in the info that are made clear in big font that the code falls under MSG/S397 copyright.

If that is a good idea, all can be tweaked/updated by participants to perfection, the more we test/develop them..

It is a lot of work to pull this off alone, you know ;)
Click to expand...
Yeah, I think it’s not a good idea to just show parts proprietary source code without permission, but if it’s just just small parts and mostly some values / parameters , then you could probably just ask for it. Since you already have access to TGM source code , and worked with them on Zolder etc. , then I guess you already have good relations with MSG/S397. BTW: have you collaborated with, or at least was able to share your ideas and work on rf2 tires with Michael Borda( as currently he’s probably the best expert on this , except maybe the TGM autor himself).
 
Emery said:
We all have multi-core computers these days (I've got 10 cores on the older one and a bunch more on the new one). Is it possible to run multiple instances of ttool, even if it required multiple screens?
Click to expand...

I asked ISI that 13 years ago ;) IE: Said I would run TGM profile 24 hours one day a month.

I would just like days back where you could do a fair dinkum burnout in the Historic :p
 
You must log in or register to reply here.
Back
Top