rF2 tire size (ride height) vs pressure issue

Several threads have been published regarding rF2 tire model and the strange effect pressure seems to have.

However I have not seen in any of them anything related to tire size. I have been performing several tests and I have found a weird behaviour regarding this aspect.

Tires get bigger with lowest pressure than with highest pressure. For both C6R and FISI the result is the same. About a 2 mm ride height increase with cold tires.
In C6R, initial pressure ranges from 117 to 212 kPa. One would expect that a tire with higher pressure would both increase in size and stiffness. Hence deflection under load would be smaller compared to a lower pressure one. However, exactly the opposite is happening.

It should be noted that this pressure values get bigger in difference when tires get hot causing a bigger variation of ride height. But in the wrong sense.

It should be noted as well that asymetric ride height provokes a different tire load combination in the 4 wheels.

For example having FL and RR at lowest pressure and FR and RL at highest, creates an over 10% difference tire load between both wheels in the opposite direction to what would be expected.

In reality, uninflated tires would be less loaded than inflated ones but motec says the opposite and it is clearly related to the unreal variation of ride height. When you uninflate two tires in diagonal the whole cars is rising and the uninflated tires get higher load.

If this basic tire shape is not being simulated properly, we cannot expect that tire temperature distribution to be right.

Any basic motec user can check this. I hope it gets addressed by ISI.

Enviado desde mi GT-I9505 mediante Tapatalk

 

I hope the motec plugin is reading the correct sides

 

If you could check if nothing is wrong I would appreciate it. I am relying on telemetry as if it was god.

Ride heights seem to have no bug whatsoever. If I lower ride height setting for both FL and RR, most of the load gets transferred to the opposite wheels FR and RL (as expected). With highest suspension settings and with extreme ride height values in both diagonals you can make 90% of the car weight rest in only two wheels.

I have also found that changing suspension spring settings doesn't alter ride height either. However this could be due to different length springs being used in order for the car to maintain ride height. For an average 3350 N tire load, the difference in deformation using 123 vs 315 N/mm springs would be 27,2 mm vs 10,6 mm (16,6 mm).

This difference is not seen in the telemetry as said. Both setups show the same ride height values immediately after leaving the garage and stopping the car (all tests have been performed this way).

This results would be OK if 123 N/mm springs where around 16-17 mm longer than 315 N/mm ones. I cannot say that this is wrong. I don't actually know how this is TYPICALLY adjusted in a racing car.

However the difference between a fuel loaded vs empty car does show more consistent differences with respect to variable stiffness. A fuel loaded car shows lower ride height values in the telemetry. Especially in the rear end which is closer to the CoG of the fuel tank. This difference gets significantly bigger with minimum stiffness vs maximum stiffness (which is what would be expected)

Note: ARB settings where set to maximum in all tests to minimize the weight transfer between wheels in the same axle.


Enviado desde mi GT-I9505 mediante Tapatalk

 

Interesting finds mate. Especially the fuel tank CoG.

 

It seems the "method" of the ISI setup pages is about the end result, not how you would make changes in real life (having to account for things like ride height change with spring rate, etc). I'm thinking its a much simpler way for the end user, as you'd have endless pages of setup options to do it as per real life.

I remember Richard Burns Rally had a "real life" type of setup page. For the alignment settings, you would have to change things like shock tower height, toe-link length, etc...then further down the page it would give you the end-result of the changes.

 

I would love to have that in rF2! That's why I liked RBR so much other than sheer fun of driving in that thing.

 

The spring setting vs deflection issue may have the explanation I gave regarding different spring lengths. However, the tire pressure issue is an error unless telemetry has an issue.

I have also detected that tire stiffness is also higher at lower pressure which is also wrong.

Summarizing: the tire size/stiffness vs pressure is not showing consistent results. I will make more tests to try to see if it is a size problem or a stiffness problem.

I will better document everything commented so that anybody interested can check it out.

Enviado desde mi GT-I9505 mediante Tapatalk

 

Last time I saw the tire deflection data from Lazza posted, which seemed exactly same as the suspension deflection data (diff unit of course). I suspect the tire deflection data could be wrong, as I pointed out.

the stiffer of the tire, the more difference between the data sets shall be...

 

well, i realised that ISI is really quiet about tire pressures and tire model in the first place.. CPM didn't bring us what we expected.. After fast laps, the garage is showing weird temperatures spread and that just something i learned to live with

Until some fix/update...........

 

This begs the question, why use complicated tyre models (compared to rF1) if they can't handle the basics properly (temps and pressures)?

 

Maybe because if no one did then we would still consider flying as a bird only thing?

rF2 tyre model works. Simple as that. That's the main reason it should be and is in use. The only thing is that like everything in software department, has loose ends. With time these loose ends will shrink and results will get closer and closer to real ones. Not having few, for some essential, features of a tyre is frustrating but maybe same day next year many things that we were complaining about are gone? I have lots of stuff to keep me busy until then and not having them now doesn't stop me having fun from time to time with rF2.

 

First, I don't think I managed to reverse (left side / right side) any of the channel values in my plugins, which still doesn't rule out the game providing reversed values - though that seems even less likely, all things considered.

This is just a difference in approach. As has been pointed out above, the game doesn't just let you change the spring settings and have nothing else change. It still respects your ride height settings, and so adjusts things to suit. I believe it bases this on an empty fuel tank, hence fuel does make a difference afterwards.

No, we (you and I) were plotting tyre deflection vs tyre load. Suspension deflection does graph a little differently to tyre deflection as you'd expect.

The real question is what you call "properly". A lot of sims, rF1 included, let you control the temperature spread across the tyre very well using camber and pressure, and nearly all setup guides (based on sims) start with adjusting camber and pressure to equalise temperatures (or aim for an 'optimum' delta). Is this realistic? Is rF2 realistic? Is realistic somewhere in the middle? If we don't know for sure the answer, then it doesn't make a lot of sense to bemoan rF2 being 'wrong'.

 

I have checked that tire stiffness correctly depends on tire pressure. The smaller the pressure, the smaller the stiffness. In the C6R plotting tire deflection vs tire load you get 463 N/mm for 212 kPa and 291 N/mm for 117 kPa.

We will consider a full stiff vehicle (springs and ARBs) with max pressures as reference. Reducing FL and RR tire pressures should imply that part of their load would go to FR and RL tires. The effect would be similar as if you slightly reduce the ride height of those suspensions.

However, the effect in rF2 is exactly the opposite. These uninflated tires get a higher load and a very big deflection due to their confirmed low stiffness. This would imply that these wheels' lowest point has also been lowered significantly under no load.

This can only mean that the wheel has grown in size (which would be an error) or that it has reduced in size as would be expected but has also been further separated from the chassis to compensate the loss of load.
In any case the behaviour is unrealistic and very confusing.
I understand that the approach for the springs is different since in a real car a part is replaced by another one which maintains ride height. This appears to be very well done.

As an example of this, if instead of reducing tire pressures you reduce FL and RR springs, those wheels get a slightly lower load compared to the reference case. This is the expected behaviour.
If you reduce those wheels suspensions heights the result is similar and can almost completely unload lowered wheels. Again, it is the expected result.

However for tire pressure this compensating approach (if confirmed) is unnatural. Reducing tire pressure does not imply replacing any part of the car. You just open the valve and let the air escape. Hence any ride height compensation for tire pressure would be confusing. Especially if the final results ends up being the opposite one to reality.

With actual behaviour in order to get the expected result after lowering a tire pressure, it would require as well to slightly reduce ride height in that wheel suspension.

Enviado desde mi GT-I9505 mediante Tapatalk

 

I have been further researching and several other weird things turn out:

1. The mentioned pressure vs tire stiffness relationship is static!!! It does not dynamically change as tire pressure increases due to temperature. It only depends on INITIAL tire pressure. This dependency should be updated as pressure changes.

2. The mentioned tire stiffness or tire load vs deflection relationship is completely linear throughout the work range. I would expect a very non linear pattern which would depend on tire pressure.

Considering the above giant simplifications, I wonder what kind of contact patch are we supposed to get....If tire load calculation is messed up due to basic load vs deformation error, no matter how accurate is the contact patch calculation for that load, the result will be wrong.

In my opinion tire load should be among the values to be seek in the lookup tables. Apparently and according to telemetry, tire deflection vs tire load remains constant thoroughout a race, only depending on the selected initial pressure.

Another thing that is very suspicious is the radical change in tire loads, ride heights, suspension positions and tire deflections when switching between player and AI by pressing "I" key. I know that AI is using a simplified tire model, but that alone cannot explain the massive differences shown.

To finish I would like to ask:

- if someone understands the relationship between tire deflections, susp position and ride height. I would expect ride height variations to be caused by tire deflection plus suspension position variations but there is a difference I cannot explain.

- If someone knows why in a stopped car where dampers are not providing any force, how suspension force is not proportional to tire load. Only tire deflection is. I would expect all the tire load going through the suspension spring !?

Enviado desde mi GT-I9505 mediante Tapatalk

 

Maybe SpringMultiplier must be considered?

 

I agree it would be very useful but I am reluctant to think that it will happen. I think it is much more likely that we will continue with lazza's or other people's plugin.

I am using telemetry data but I really don't know if those logged values are accurate. I am kind of reluctant to accept that tire load vs deflection relationship is linear and only initial pressure dependent. However that is what available data says.

Maybe lazza can put some light here since he perfectly knows what is being logged and its origin.

Enviado desde mi GT-I9505 mediante Tapatalk

 

Regarding complicated tyre models producing weird results, it is not just rF2's problem. AC and iRacing suffer it from too. One would think that more simple Pacejka-model with some added features (tyre damage) was more useful for all concerned. Keep the complicated stuff in virtual laboratory until it has reached some real credibility (said in hindsight, of course).

None of this stops enjoying the sims though.

 

Interesting. I can't offer any experience in this area, but it would seem to agree with some of your findings. The static nature of this (if true) would be a bit worrying.

There are other simplifications for AI, not just the tyre model, I believe.


Regarding the data source for my plugins, I'm 99% sure I didn't manage to do some typos in my original telemetry plugin (the code that references the various fields for each tyre is grouped for each tyre; so the first group has all [0], second has [1], etc... a typo would be easy to spot, and I'm pretty sure I would have considering I first made a motec plugin in ~2009 for rF1 and the telemetry plugin is just an evolution of that same code), and my current DAMPlugin has a more automated process for tyre/suspension ('corner') channels. It's more likely I could accidentally reference the wrong data, but it would be the same for all 4 corners, and they would stay in the correct sequence.

As for ISI's data given to plugins, it also seems unlikely to mix up wheels because it would use a similar structure; the definition of the fields is static, then you just reference a particular wheel. So mixing up wheels would apply to all channels, not just a channel or two.

 

Weird thing is, is that I see our sims we drive as the labs. rFPro came about after rF1 was pumping the jams. So in a way, I'd see the work done on the tech here going out into the real world once it works well over a wide range of inputs.

Devs need to push the tech on over time, they have to do new and better things as they can. I also think we forget that there are issues with all the solutions out there. I found nothing with the old rF1 system that would ever punish me properly for poor driving technique. I do remember reading stuff from people working with those old models that also found flaws in them, just things that don't work right.

 

As I said that is correct. However you failed to read what followed that text and my following post.

- what is not correct is that the mentioned tire stiffness maintains constant while pressure increases throughout the stint.

- what is not correct is that tire deflection has a linear relationship with tire load ranging from 0 N to 6000 N.

- what is not correct is that after reducing pressure in opposite wheels (in diagonal), tires magically get closer to the ground (considering an unloaded car) and despite their lower overall stiffness, they end up having a higher tire load than the ones with high pressure.

Considering all these inaccuracies, it would seem to me that the lookup tables generated by the fem model that are stored in the tgm files are only used immediately after leaving the garage and hard coded into the physics. However, that would not explain why a linear model for tire deflection vs tire load is used. It makes absolutely no sense to me.

The only explanation that I have is that the real tire deflection used ingame is not being actually logged in the telemetry. However, motec says it is a logged channel so I don't know what to think... I really have a hard time to believe that tire model is as simplistic as it appears according telemetry.

Enviado desde mi GT-I9505 mediante Tapatalk