Look what the cat hacked in - s02eo2

So true! I've received some data from the tyre manufactor for the tyres used for formula 3 I'm working on. Do you know what is the speed they've been testing their tyres? 20 km/h! Quite clearly they do those tests with some specific purpose that might not be aligned with what we need to correctly model our tyres.

The F3 engineering team is in fact working on some mathematical model trying to get some more reliable data using their own telemetry deducting as many information as possible from the few telemetry channels they have available.

Anyway, I really appreciate your comments, @Lord Kunos . It's very good to have some more insight from the dev team about how they work with the tyre parameters and to know more about their design decisions.
The biggest problem for modders is sometimes hard to understand how to use those parameters in the best possible way.

It would be great if we could have a Q&A here on the forum about the most complex part of the physics.
Maybe we can take one real example and trying to show how to model it correctly with few hints/confirmation from the dev team.
Could this work?
Obviously no rush. We all know how busy you are

 

Not everyone tests at those speeds though ... i have seen tests with up to 170 km/h

 

And still they are not 100% useable out of the box because of the missing purpose of the experiment and the real environment on a race track. It gives more an idea how it would perform, but in the end only the oldschool way of testing it on a car and a track will show.
This is always the problem with lab conditions.

Interesting.

 

And that is very nice until you keep that linearship approach and see how after ~7500N the lateral force starts to drop, take a very high downforce car (the tyre that misterbeam posted is from a high downforce car) and take Eau Rouge flat with the huge loads + weight transfer to the external tyres and tell me how fun was to lose the rear brutally without any warning

Just for the record, if you take the last 3 data points (so ignoring the first point at low loads) and try to make a tyre with AC values, you end up with this :

DY0=2.0958333
DY1=-0.282

And your lateral force vs load looks like this if we keep going up to 10000N (and we can reach more than those 10 kN in certain cars when cornering at high G's) :



I added the rear tyre of the Lotus Exos for comparison purposes.

 

Oooohhh physics battle / showdown !

/popcorn

 

@David Dominguez Wait, but isn't the red line a problem with data rather than model? I mean we definitely need non linear scaling to achieve something resembling all 4 points in AC, but only looking at the last 3 points, the relation >is< pretty linear (even curving the other way a little) and regardless of how AC models it's tyres, the red line in above chart represents how the actual data behaves - hinting that at some point lateral grip drops with increasing downforce during high speed cornering?.
I mean the grip should plateau a bit at very high loads, but the above conclusion is even more extreame than what AC models atm.

edit (not related directly to the above):
___________________
To better illustrate the problem, here is a graph of Exos tyre (red) and Misterbeam data (blue) (constructed from the above DY0/DY1 data), even around the 1500-6000N of avilable datapoints, the AC tyre is much much less sensitive to loads.

So it's not even that the tyres are too grippy at high speed, but the sensitivity itself, compared to Misterbeam's data is much smaller. (and it goes even more extreame when we include the 4th point).

Frankly, looking at the above, I'm in the same boat as Stefano on this - there is more to data than just points.

 

Not quite, if you really follow the line based on those data points, you would get the blue line, whereas with the linear method you end up with the red one. And as you can see, the bigger the load the worse it gets.

 

Just try these in game, and see what happens ...

DY0=2.0958333
DY1=-0.282

 

That just shows that some AC tyres have little load sensitivity. Hence why you can corner at >5.5G with a Lotus 98T.

 

Battle of the Giants.. better to go out of the way and take some pop or chips
But I like it, if there wouldnt be any discussion (in this data based way, or try to lean on facts) we wouldnt learn n' move.

 

98T is a corner case, but still where do you actually see 5.5? In a slightly cambered 310km/h, 100%boost, qualifying tyres, full downforce cornering. That car barely goes 3.5g on default setup. And coefficient of friction at top speed cornering are actually within reasonable limits - something around 1.3-1.4. Now i don't know what would happen to a hot racing compound at high loads, but even solid rubber at that point would have friction coefficient of something like 1.1-1.2 so not that far off. And it's not like the surface pressures are all that big as the tyre has quite big contact path at that point. ***
If anything it's the too perfect aerodynamics that is causing unreasonable effects at high speeds. (and possibly too magical/simplified qualifying tyre).
You can't blame results of a complicated system where near every element has been stressed to the limit on one variable.

Hmm it was interesting, but without rest of the car tuned properly around it it it's super hard to take any conclusions. Although at Eau Rouge there was slight drop in grip and while it looked a bit more realistic, the feeling was not quite right~.

Yeah, but extrapolating any data range to above twice it's original range, especially on absolute minimum amount of points to create ^2 trendline is just going too far imho. As awesome it is to have real data, as Stefano mentioned, without knowing full well how and what for it was created, it's almost impossible to arrive at reasonable conclusions without any way to crosscheck it. (even more so on only 4 points).

*** edit: Actually i wonder what are surface pressures of F1 tyres under extreme high loads... would not be surprised if they were smaller than a family saloon with thin eco tyres or something like a wheelchair haha.

 

While listening all the stuff that Stefano finds interesting enough to try make happen, or stuff that he wants to do for Assetto Corsa, just makes me want to kidnap @Lord Kunos for a week and clone him dozen times over, so all those things he keeps mentioning, but time is the factor, can be done NOW.

I heard that a thing called 'patiance' is a good thing... I don't have that thing.

ps. WTB patiance!

 

I don't know where to start concerning the 98T, the car is much much faster than the real life one! the cornering speed is just not representative of what you can see on onboards from that era! and you can see that on RSR world records, especially on high downforce tracks like mugello (1.21.1 just 3 secs slower than the F10 from FVA) and the ring (5.20)! people are lapping in 1.52s - 1.50 on Spa, which is like 6 seconds faster than 86 pole lap (1.54.3 if we take into consideration the old layout which was like 2 secs faster)!

The car drives as fast as a modern F1 ... check that section that starts around 2:17, that's simply not right!

That's what i wanted you to see, the combined slip conditions producing weird effects, and yes it doesn't feel right ... because you just keep understeering unlimitedly

 

I'm not suggesting AC is perfect but is it really valid to compare times between AC and FVA?

The 98T has 1000bhp and trick qualy tyres, would it really be that slow?

 

Current f1 cars should be a lot slower than the 98T then! right ?

 

That's not what I said is it?

 

I mean you can't define a car with an engine and tyres ! those turbo V6 were quite difficult to manage especially with the turbo lag, the car even with ground effect (which was really sensitive to ride height changes on early designs), didn't have as much downforce as modern ones! and besides to use ground effect they needed to use very stiff springs (they didn't have heaves at that time) which was compromising the handling!

You need to read the discussion above to see how we jumped to the 98T ... but i am pretty confident it was not cornering like it does on AC!

 

Prove it, or disprove it. Evidence people!? Assumption is nothing but the mother of all f***ups.

http://www.superstreetonline.com/features/news/0708-sccp-lateral-g-skidpad-testing/
Kinda like that.

http://www.lotustalk.com/forums/f3/cornering-grip-g-force-comparison-83465/#post1408780
Post about the article, kinda synapsis.

 

Guys, you've worked on mods of modern high performance cars. You perfectly know how difficult is to find good, reliable data and how sensitive some information are, if ever available. The most of the time even big teams don't have such detailed information about what's going on on their own car, and for sure if they do have it will not share it with the world.

Sorry to say, but now the discussion here seems to be related to one tyre parameter that is supposedly not linear, and you're trying to confirm this comparing lap times of 2 different F1 cars, or with perceived cornering speed from videos (even on different tracks without knowing any other aspect)?

Those cars are modeled most probably with time restriction (I'm sure Aris can't afford to spend 1 year on a car) and for sure with many assumptions. Do you really think that all data for a F1 are shared even with a licence? And even in case of Senna's car? A car of the '86?

Let's be realistic! Discussions like these will never end and will never prove anything at all, will just create confusion and give more material to people like our nice friend Hexx...

This is the model we have right now in AC. As every mathematical model isn't perfect and will fail in some situation.
Modders should try to find the best fit for the car they want to model.

I'm sure Kunos and Aris has plenty of experience to best judge what's the best model to use for their own simulator.
Simplicity sometimes is a bless since it will allow to have something more controllable, especially when it's very hard to find good reliable data.
Not always something that is more complex gives better results.

Please take those words for what they are. I don't want to offend anyone.

I understand what you want to do. We all strive to get our mod as close as possible to the real one and trying to use all the info we have available.
As an example I've much more info about the aero of my car than I can use with the current aero model (waiting for the interaction between wings, that will hopefully come sooner or later). I simply tried to fit the data I had in AC model in the best possible way. Is it perfect? no! but selecting the normal range of use it will give values that are very close to the real values with aero interaction.

Peace

 

nvm