Objective tyre model / physics discussion thread..

[Niels Heusinkveld]

Edit: URL fixed

Yo Folks,

Yes, lets try to objectively keep the LFS physics against the light. So instead of 'I feel this' and 'I feel this is realistic', lets see how we can attatch some proper numbers and facts to what we think and feel.

This Lx type video was posted in the 'racing' subforum but I think its worth an attempt at an analytical physics debate, not going by 'feel' but going by videos / numbers.

Here is the vid:
http://www.westfield-sportscars.co.uk/westvid/misanomegabusa2.mpeg

Interesting because there is both 'sane keep it gripping' driving and 'insane try to slide the car around' stuff in one video.

Here are my observations, perhaps something can be concluded out of them..

1) It seems that he does the sliding in 2nd and perhaps once even in 1st gear

2) when not deliberatly sliding and/or in higher gears the car NEVER requires opposite lock to keep it on the track despite what seems to be full throttle.

3) the first slide (1m04) isn't really initiated by throwing the car into the turn but just by using a lot of throttle on turn in, in 2nd gear (I think). Perhaps the sudden transition from lift off to throttle unsettles the car.

4)lateral (sideways) grip barely seems to drop with wheelspin. Once he unsettles the car with power on corner entry it seems more forward force from the outside rear tyre that cause yaw torque than loss of lateral grip that causes yaw slide.

5) at 1m20 he gets mid turn oversteer again with power, not really assisting it with the steeringwheel. (I tend to flick the car into turns)

6) The oversteer seems brief as it seems to be a full throttle corner exit but the sliding stops before exitting the turn...

7) The corner exit at 1m31ish is interesting as its one of those oversteer guaranteed things in LFS, yet his car is planted..

8) 1m45 he really seems to want a slide but has trouble getting one going :) After his first correction a big slide is no longer possible.

9) 2m25 he gets nicely sideways under max power.

Now I think I can make something out of this. Based on the tyre model discussions, most notably the 'new' model by Todd Wason, it seems that traditional combining of longitudinal and lateral forces isn't quite realistic. Now this is just another debate with people being against and people being pro but I think I can now sketch a bit what happens in real life and what happens in sims, notably LFS.

I'm the non-expert on this but I'll give common sense a shot. During opposite lock in LFS, you turn against the corner. If you wouldn't, the lack or rear sideways grip will make you spin. You seem to balance the forces around the car centre of gravity by steering the 'wrong way'. This makes the front tyres roll more or less in the direction the car is traveling at, which basically stops them from wanting to enter the turn. Now what seems to be the case in LFS, when you get wheelspin there is a considerable drop in lateral grip. This seems to be what causes oversteer.

Now I *think* I'm beginning to understand what happens in reality. It isn't so much the drop in lateral force that causes oversteer, it is the torque from mostly the outside driven tyre around the car centre of gravity that tries to push it around.

Tyre models are tricky to understand as there is always a combination of longitudinal and lateral forces at play. With risk of being totally wrong I've attatched a sketch which hopefully nods in the right direction. I'm only looking at pure lateral force in LFS (assuming longitudinal force is relatively small) and looking at pure longitudinal force in the 'real(??:))' sketch, assuming lateral grip stays relatively constant, more or less based on drifters pulling about 90% lateral G compared with non drifters, which I heard somewhere, and have no idea whether that is actually true..

Would it seem plausible that such a thing is happening in LFS? I think it might explain its handling to some extent. Imagine driving LFS with the sketch on the right being the case and lateral grip being relatively constant.

Ever so slight slalom turning in LFS tends to cause oversteer by the bucketload. With more constant lateral grip you would certainly not get the oversteer simply because the amount of rear lateral grip available is not much less than what the fronts have available. You would only get oversteer when applying a lot of power in a turn. You would force the rear around the front instead of passively having it overtake the fronts because they have no grip. I.E. you would actually have GRIP on the outside rear tyre that causes oversteer (as the grip causes forward force) instead of LOOSING sideways grip.

Any constructive thoughts on this one? :)

[Hankstar]

Will any of this be in the exam?


:D


sry, not really constuctive was it?

[J.B.]

Will any of this be in the exam?

No but this will: long-ass post by Todd Wasson (http://forum.rscnet.org/showpost.php?p=2500155&postcount=152)

Niels, the video doesn't seem to be working. Maybe you could upload it somewhere?

[Hankstar]

long-ass post by Todd Wasson (http://forum.rscnet.org/showpost.php?p=2500155&postcount=152)


Phwoah. I'm dropping out :Looking_a

[Breizh]

http://www.westfield-sportscars.co.uk/westvid/misanomegabusa2.mpeg

[himself]

Devs should implement the swinging of the key-hanging like in the video too :D

[Bob Smith]

As far as I can tell, both affects are present in both LFS and real life. However the first one may well be too dominant in LFS. We know Scawen will look into it (if he hasn't already), only he can truly compare how LFS is working to an ideal scenario. It would be very interesting if he would share his conclusions.

[Niels Heusinkveld]

Bob, indeed, there will probably be some loss of lateral grip, and the wheelspin will result in some forwards force. The interesting thing is that both ends of the scale produce the same basic effect (car wants to oversteer) yet the difference can be considerable.

How does beta testing LFS work? I've rarely seen Scawen talk about physics / tyre models. I can understand that as everyone here probably has a different opinion about what is wrong and what isn't, but surely there must be something more constructive some players can mean for LFS. I doubt beta testers have insight in the inner workings of the tyre model?

Its hard not to have any control over what happens with LFS physics, as S1 and S2 alpha are so amazing in dynamic feel and 'instant response' and the feeling of being connected, yet are pretty far off when it comes to tyres, providing it is not some strange weight transfer bug..

[Cue-Ball]

Does Scawen even have beta testers for LFS? I was under the impression that the three devs test changes, then release them as a patch when they're satisfied. I don't think I've ever seen mention of beta testers.

[Renku]

The next time you exit LFS, there should be a screen and under section "beta testing" you should see some names.

[Bob Smith]

Yep, and none of them will ever talk about what they do. It's hard enough to just get them to admit it even when they're name is in the credits... dunno what the devs made them do to promise. :x

[Breizh]

Don't you mean 'what they made the devs do to promise [ND]'? :razz:

[sinbad]

Anyone else look at that video and think that guy's not exactly Kimi Raikonnen? :)

These comparisons are very difficult to make. I'll agree, footage of Caterhams of all kinds makes them appear much more "tight", grippier and more stable, but it's not a case of watch video + see where he slides + decide if an LX would slide there or not = diagnosis of tyre problems.

If it ever comes to be that the only way to make a car oversteer in LFS is by mashing the throttle pedal and waiting for the tail to try and overtake the nose, then something has gone seriously wrong and we're in ridge-racer territory :)

[himself]

How come we still have Alpha version and have Beta testers? :smileypul

[Niels Heusinkveld]

Lets try to stay on topic guys :tilt:

Some more observations on LFS in general..

0)Effect of small changes. I recall messing with tyres in RACER that relatively small changes could have a noticable effect. Perhaps sims can feel 'quite wrong' when values are 'not astronomically wrong'.

1) Drifting. Much hated but its possible in real life and some videos show that drifting in LFS doesn't seem too far off. This is slightly odd as one would expect the tyre model, if wrong in the area that I think it is (first post), drifting to be harder somehow. Why is it easier than other sims to keep a slide (suggesting more constant lateral grip with wheelspin) when the tyres loose lateral grip so easy and start sliding? Sounds contradictory..

2) lateral vs longitudinal grip. Based on limited information it seems that cars pull slightly more longitudinal G (braking/accelerating) than lateral G. (cornering). It seems that this is not the case in LFS. When small changes have noticable effect perhaps road car performance along the lines of 1G lateral and 1.2G longitudinal would improve things considerably

3) Buildup and peaks. Another thing I notice when driving LFS is that you enter 'slip / slide' status without having the idea that the forces and grip build up. You simply seem to go from a left to a right slide (I don't mean huge oversteer slide, but just not in a 'stick' zone as you leave skid marks and warm the tyres.. where real tyres would expected to be in their 'stick' zone. Its hard to say but perhaps the buildup of available long. and lat. force with small slip angles or ratios is too sudden? (see pics attatched). LFS also doesn't have the longitudinal peak. Now its debatable if threre really is a noticbale peak in real life, you would assume there would be a small difference. Now a locked tyre in LFS slows you down best, until it overheats..

--
I really don't know it all for sure..

[Fetzo]

If it ever comes to be that the only way to make a car oversteer in LFS is by mashing the throttle pedal and waiting for the tail to try and overtake the nose, then something has gone seriously wrong and we're in ridge-racer territory :)

hey in ridge racer you have to lift the throttle first, steer and floor it again! it's not that simple! you won't need any brakes in this game though :P
:hidesbehi

[Cue-Ball]

Although I'll fully admit I don't know as much about tire models as some of the guys who frequently discuss it, I do know how a car reacts in real life and I'm able to compare that to how a car reacts in LFS. In my opinion the main physics flaw has to do with lateral grip, but it's not as simple as just saying it's too low. In my street car if I get the tail out then let off the gas the rear end snaps back into line. Even on wet pavement and high treadwear street tires the rear end will 'snap' back rather quickly. I should think that this would be even more pronounced in a race car on dry pavement with sticky tires. However; LFS does not exhibit this behavior. The curious thing is that if you drive on the skidpad the tires behave as they should. So lateral grip seems to be fine in some instances (on the skidpad), but for some reason lateral grip seems very, very low in others (when "drifting"). I think this helps to explain why drifters seem to like LFS so much as well. It's easy to get the tail out and keep it there, assuming you can keep the car from coming all the way around. The slick cars exhibit the same behavior with the difference being that it takes a lot more pushing to get the tires loose in the first place. But once you get them loose, even the slicks won't snap back into line as they should.

I don't know what would cause this or how to address it within the sim, but this is the #1 problem with the physics in my opinion. As much as I love LFS, this really needs to be addressed. Several of my friends refuse to play LFS and instead play GTR/rF/etc because there is so much more grip available. Going from GTR to LFS is like driving on wet, oily pavement. This is something we've heard time and again from people playing the demo and I think they're correct when they say there is a problem.

[Hyperactive]

One thing that really strikes out is that in LFS when you get a slide and then try to get it back in line, it is much harder than this video shows. But to compare LFS and some racing car you need better video. That guy mostly just drifts the car. Actually I drove LX4 just little ago and seemed to be somewhat close to that, though the difference in power is just...big.

LX6 can't be driven by that driver. With same throttle control that he has in that vid he would just kill himself in LX6. :)

[bobbfwed]

Although I'll fully admit I don't know as much about tire models as some of the guys who frequently discuss it, I do know how a car reacts in real life and I'm able to compare that to how a car reacts in LFS. In my opinion the main physics flaw has to do with lateral grip, but it's not as simple as just saying it's too low. In my street car if I get the tail out then let off the gas the rear end snaps back into line. Even on wet pavement and high treadwear street tires the rear end will 'snap' back rather quickly. I should think that this would be even more pronounced in a race car on dry pavement with sticky tires. However; LFS does not exhibit this behavior. The curious thing is that if you drive on the skidpad the tires behave as they should. So lateral grip seems to be fine in some instances (on the skidpad), but for some reason lateral grip seems very, very low in others (when "drifting"). The slick cars exhibit the same behavior with the difference being that it takes a lot more pushing to get the tires loose in the first place. But once you get them loose, even the slicks won't snap back into line as they should.
I think this is very close. Driving a small "sport truck" has shown me that getting the back end out isn't too dangerous as long as you know when to lift off the throttle; as soon as you do so, the back end uncomfortable comes back into alignment. Drifters at the tracks I've seen always get "drifting tires" which are very hard compound tires. This does two things, makes it much easier to loose grip, and second it make the tires wear much slower. If drifters at the track has as little traction as the slicks in LFS they wouldn't need drifting tires.

[J.B.]

What makes it difficult to discuss LFS tyre physics is the simple fact that we don't know a lot about how LFS tyres work. We don't even know if it is more physically based or data based (empirical). All we can do is interpret features such as deformable structure, advanced temperature, pressure, flatspot and dirt modelling as indicators that the model is based more on physical ideas than a Pacejka type data approximation model that IIRC ISI, Papyrus and NetKar use.

Another thing we can do is compare the LFS driving style to videos like the one Niels posted. I watched the video and then drove a few laps in the LX4 around Fern Bay Club to do some comparisons. I used Bob's road going setups, trusting that he had made the effort to set the values as real to life as possible (IIRC his Dad actually owns a Lotus 7 type car).

What I tried to see is if the stability seen in the beginning of the video could be achieved by simply staying a fair bit below the limit in terms of speed. The answer was no. The car was always controlable but I was using more opposite lock as soon as I got on the throttle than the driver in the video. Comparing to later parts in the video, where he is holding long, controllable slides, I found LFS to behave in a very similar manner.

So what I feel is that LFS cars like to get into a slide a bit more easily than they should but are as nice and controllable as they should be once they are sliding. This is also apparent when you look how realistic the LFS drift videos look. For me the controlability of an instable car is one of the features of LFS that makes me regard it as having the best physics in sim racing today. In ISI based games I am all the time scared of getting the tail out because I will have a very hard time to stay in control of the car, especially avoiding a snap back reaction in the opposite direction.

I think there does seem to be a problem with LFS tyres losing too much lateral grip when power is applied and but we cannot say what the cause for this is. Are longitudal and lateral forces calculated seperately and then joined using some kind of traction circle hypothesis or is the force vector a result of a more low level, physically based model? We don't know.

We know that Scawen is very aware of these discussions and we can be sure that improvements and tweaks will come in the future. Here is a relevant post by Scawen:


Quote:
Originally Posted by SladiVadi
May I ask what's so bad about the tyres?


Originally Posted by Scawen:
Well they are close to behaving correctly at all times, but not quite as good as real world tyres in some conditions.

It's mainly about some things they do when they are near their limit of grip, and just over the limit, and how it's affected by applying and removing the power.

BTW, if anyone asks me for more detailed answers, i won't give any. Nothing's clear until i've coded up several testing mechanisims, tyre test rigs and so on when i get back on the case with the physics.

And here are links to threads on this forum that discuss and analyze the tyres of LFS:

http://www.lfsforum.net/showthread.php?t=4375
http://www.lfsforum.net/showthread.php?t=4214
http://www.lfsforum.net/showthread.php?t=3667

Also I'd like to add that it is nice to see that there are no "STFU, LFS is teh PERFECT!!111!" kind of posts here that are often seen in certain (http://forum.rscnet.org/showthread.php?t=240639) other places. ;)

[Niels Heusinkveld]

Thanks for that quote from Scawen, at least that explains his lack of response to the matter.. At least he's 'on it' (as expected but nice to confirm. :)

[deggis]

I'll pop up this thread...

What do you think, are we waiting too much for the next physics patch? I came to LFS after S2 was released but as far as I know the same grip issue has been in LFS since the very beginning (I presume the issue was even worse in S1 days). Have Scawen ever before "admitted" this grip issue than just recently? Or have there been same kind of discussions about this issue in S1 days? The question is why the tyre grip would change now because it has been there for 3-4 years?

[Blowtus]

I think because everything else feels a lot closer and better these days, it stands out more. It certainly was more of a problem in s1 though.

[tristancliffe]

You make it sound like it's easy to simulate a tyre properly. No one has managed it yet in industry, and there isn't one sim that has a really good tyre model. LFS might have a problem at low speeds (but nothing really noticable at racing speeds if you ask me), but it makes up for it in so many other areas. Why don't people see that? Why do they only see bug after bug after bug?

Yes, the tyres were worse in S1. I'd imagine the GTR cars would be undrivable on S1 tyres. But at the same time, LX6 aside, the cars in S1 weren't powerful enough to make the tyre behaviour that noticable, and everyone was aware that LFS was catering for a niche market in a very unique way, and could understand that it wasn't perfect.

[Niels Heusinkveld]

It will be interesting to see a physics update when Scawen is ready with it.

Meanwhile I've uninstalled all ISI games. I would bet a pint on its tyre model being more flawed than that of GPL. Sure it does flashy suspension settings, brake duct cooling and suspension modelling, but the tyres ARRGGG the tyres. I couldn't handle it anymore! :) Meanwhile I've put my faith on Scawen and Stefano.. :) (and Kaemmer)

2006 might be very good if two out of those three come out with something new :)

[B2B@300]

I agree with Tristan in that I think the tyre model is very good as it is, aside from the obvious low speed grip issue...

I have read race drivers comments that driving at the limit does feel very much like the average person would expect driving on a wet road to feel, when you are near the limits of traction it doesn't take much to over do it :shrug: our main problem and the problem of all sims is the lack of seat of pants feel which often makes us slow to react in a sim.

Having said that I do think there is some small issues at the limit which affects handling, and if I was to take a punt at what it would be I'd say [I "think" that the changes in the slip angles between front and rear tyres and even left and right tyres has a small issue when the tyres are in the different states (grip, tread walking, no grip - I don't know proper terms), which affects the handling of the cars at the limit also how slip angles are affected differently between the free rotaing wheels and driven wheels]. But I have every confidence Scawen will be able to identify any areas that need improvement and then come up with away to do it. It's hard for us to identify as we simply don't have the information or tools necessary to determine anything useful :shrug: (without scientifc test instruments any observations we make are still very subjective and therefore most likely flawed) Scawen does have or is able to get the necessary tools and info though :tilt: as he is able to code tools (tyre test rig) and can obtain data to compare it too.

I don't actually see the current tyre model as having bugs :) but as improve it a little more and it will go from being awsome to absolutely awsome :D and that's why I look forward to the next physics patch :tilt:

Now as for S3 tyre models :D I look forward to seeing brake heat simulated (which heats the tyres) and the current tyre heat model revisited, also more refinement in the differences between the treaded tyres and slicks (e.g. how dirt sticks to different tyre types and how the different tyres react in the transition phases of grip to loss of traction, also I guess the relationship between lateral and longitudinal grip [traction circle] would vary for different tyre types also) and then there is the track :D

[bobbfwed]

Another problem I have seen.
In real life:
If you get into a rear-wheel-drive car and hold the break and give the car gas (clutch work is involved) the front tires will grip as long as you want, while the back tires will slip. This is a very common practice for drag racers who want to warm up their rear tires before a race. And even with drag car with large rear tires and smaller front tires, because the rear were quickly put into traction loss, the smaller fronts can keep traction and not slide -- this of course is only to a point, eventually the rear tires (if large enough or with soft enough compound) will over power the front tires' grip.

In LFS:
I use pedals with the break and gas pedals being on separate axis. I have tried to do burnouts on RWD (I'm pretty sure I used one of the GTR RWD cars -- I can't remember, and I am not home to check) cars by holding the break, holding the clutch (not necessary), giving gas, then releasing the clutch. The result is the rear tires squealing and the front tires (if the break is pressed enough) sliding while locked up. So I messed around with it a bit trying to give my front tires better traction -- slowly letting on the break, get them right to traction loss without going over; but to no avail. I went into the setup and even put R4s on the rear, and R2s on the front. But still the front tires just slide.
I find this all very odd, because the rear tires with complete traction loss (only slipping friction) have more traction than the front tires with no traction loss -- and these cars, in general, should be front heavy, so the front tires should have more traction than the rear innately.

[mrbogeyman]

Well, you are nearly there.

IMO, this is easily fixed with the setup, you maybe just tweaked the wrong settings. Try putting brake bias forward as well. Another factor is that in much of the setups for LFS they are setup for racing, and generally have quite a tall 1st gear. So try shortening 1st gear to give the rear wheels more torque.

I usually don't have any bother pulling a burnout if I want to, and if I can't it is usually the setup.

[bobbfwed]

Yeah, but my point is that this shouldn't be a problem at all.
The traction of the "stable" front tires should overwelm that of the rear tires which are in traction loss.
The diference between traction and traction loss is too little. This is why people who "slide" around corners in LFS are able to keep up with those who keep the car in alignement. In real life, if you are getting your backend out at all around corners, you'll be all but left in the dust.

[mrbogeyman]

i think it is just another side effect of the incomplete tyre physics. they seem to lack lateral grip, so some small and controlled slides can be perfectly quick.

there is more to it than that i think, but its a whole other story....

[Tweaker]

I don't know if anyone ever saw this post I wrote, but this is about the only problem I have with the tire physics/model in LFS (and generally how it is right now with focus on road cars)

http://www.lfsforum.net/showthread.php?p=66005#post66005

Seeing that we also have low-speed grip issues as well and all....

But this is all primarly on the issue of having pre-heated tires. As if it was used to hide even wierder low-speed physics at cool temperatures.

And Bob's quote here sums it up kind of: :)

Yeah it's weird the tyre pressures have swapped since S1. While it used to be "set them as high as they would go", now it's "set them as low as you can without melting them".

[Woz]

Another problem I have seen.
In real life:
If you get into a rear-wheel-drive car and hold the break and give the car gas (clutch work is involved) the front tires will grip as long as you want, while the back tires will slip. This is a very common practice for drag racers who want to warm up their rear tires before a race. And even with drag car with large rear tires and smaller front tires, because the rear were quickly put into traction loss, the smaller fronts can keep traction and not slide -- this of course is only to a point, eventually the rear tires (if large enough or with soft enough compound) will over power the front tires' grip.

In LFS:
I use pedals with the break and gas pedals being on separate axis. I have tried to do burnouts on RWD (I'm pretty sure I used one of the GTR RWD cars -- I can't remember, and I am not home to check) cars by holding the break, holding the clutch (not necessary), giving gas, then releasing the clutch. The result is the rear tires squealing and the front tires (if the break is pressed enough) sliding while locked up. So I messed around with it a bit trying to give my front tires better traction -- slowly letting on the break, get them right to traction loss without going over; but to no avail. I went into the setup and even put R4s on the rear, and R2s on the front. But still the front tires just slide.
I find this all very odd, because the rear tires with complete traction loss (only slipping friction) have more traction than the front tires with no traction loss -- and these cars, in general, should be front heavy, so the front tires should have more traction than the rear innately.

In LFS even with clutch help off and the clutch on an axis there is still Auto clutch issues to deal with. In LFS it is not possible to stall the car so if the car is put in a position that will stall the engine the clutch will activate itself. Watch the clutch bar and see if it moves when you try to do a burn out

[JeffR]

In real life, if you are getting your backend out at all around corners, you'll be all but left in the dustOk, not hanging out the rear end, but every car needs some amount of working slip angle to corner. The smallest working slip angles are 2 degrees with IRL cars on ovals. Most high end downforce race cars have a working slip angle of 3 or 4 degrees. Bias ply slicks as used on most non-downforce cars are higher still.

The other issue is that optimal cornering grip is achieved with some tire slippage. Do a web search for "traction control race", and you'll find that most modern traction control systems allow some amount of slippage, instead of just stopping the slippage. Typically the slippage is 8% to 12%. This might be something a driver can do for a short period, like a qualification lap, but not for an entire race, which is why modern traction control systems produce better lap times.

Of course there are a few high end race series that don't use traction control, like USA Champ Cars (formerly CART), and Nascar, but then again, with the ECU's, it's difficult to check for. Nascar tries to monitor engine exhaust sounds to try and detect traction control. Most of the high end race classes do allow traction control, and even "club level" racing, such as Mazda prototype (like Indy lights), or Dutch Supercar Challenge, where the only rules are power to weight ratio limits for each class, otherwise there are no restrictions at all (active suspension, XTRAC sequential shifters, traction control systems that use individual wheel sensor / braking and engine ecu, ... are all allowed).

One issue is that in real life, it's difficult to detect if there is traction control. http://insiderracingnews.com/jf111802.html

Anyway, this is a bit off topic. I think the real issue is that it's not possible on a PC to truly simulate the dynamic physics of a tire / contact patch, so some sort of simplification has to be made.

With real life racing, optimal lap times occur with some slippage, which is one issue racing games have a problem with. In addition, in order to drive at the limits, a driver is going to spend a significant amount of time beyond the limits, and the car has to be forgiving enough to allow this.

In a racing game, the cars need to be more forgiving than in real life, or else a player just ends up memorizing control inputs, which is not a good thing, since the goal of any racing simulator should be to simulate a real world racing experience, where you can go reasonably go beyond the limits without incident, even if the physics has to be modified a bit in order to accomplish this.