Quote from Woz :

The reason that people say its too easy compared to real life to keep the car drifting is because IRL you have FEAR to deal with. Never underestimate the effect fear has on your judgement. IRL once the rears step your body feels the massive shifts in G and you need to work out how and where and how to best balance the car before you are off and slam into something.

Yeah, I'd have to agree with that. The only time a car is really hard to control when it's sliding is when it's set up for final oversteer, which is not something you typically want in the first place. Pieter Bervoets told me in his Donkervoort that full opposite lock drifting, even on racing slicks, was actually very easy to do once you practiced in a parking lot for awhile. And this is in a 650Kg car with 280HP... He demonstrated this to me going through a corner once. Normally he'd use a lot of throttle at the apex and simultaneously straighten the wheel and just steer out of the turn with the throttle. Then, he went ahead and used a lot more throttle and we went quite sideways. Full opposite lock for a second or two, then bang, straighten the wheel and let off that excess throttle and the car straightened up immediately.

Fear was plentiful in my mind throughout that, even though there was nothing to hit but a couple of small ditches on either side of the road


Edit: Oh, one more thing. Typically even on public road ways when there's an accident or near accident involving a spin, the car generally does not just spin around. Rather, it goes very sideways and the driver overcorrects. I.e., the slide left, turn right, then don't begin to straighten the wheel until the car is pointing forwards again. By then it's too late because the car is yawing even faster now in the opposite direction. Slide left, then spin around to the right. I was a passenger in a '71 Firebird as a teenager that did several such oscillations before finally spinning around three complete times and stopping dead in the middle of the freeway. 120mph off road excursions need to be dealt with carefully

Quote from Woz :

That said if you do want to feel real fear in LFS you could always attach electrodes to you nuts and use outsim to trigger a 2000V shock when you crash. I bet you drifting on a track like South City will not be as simple then lol

Excellent lol

Just wanted to echo Blowtus -- I find your technical input very interesting, Todd. Thanks for taking the trouble.

Good point about typical slide / spin accidents. I once saw an accident a bit like the one you describe, ahead of me on a UK motorway as we were approaching some roadworks. Thankfully nobody was hurt, but afterwards I couldn't help thinking "Wow, that looked exactly like a NR2003 slide." If we all went out and deliberately spun cars at high speeds every day, we'd probably be a lot more qualified to comment on tyre models. (Not a suggestion, just a hypothesis.)

Very nice posts, Todd! Thanks

In the meanwhile, I made some accelleration tests, comparing longitudinal grip to slip ratio. The following graph was done with the FZR's R2 tyres at 120kPa pressure with a live camber at around 0°.

From the graph we can see, that the "missing" falloff curve from the slip angle tests is indeed there when accellerating/braking, and this also shows us that a spinning wheel accelleration is NOT as fast as a just-on-the-limit one. In the attached example the optimal slip ratio was at about 6-8% (if I interpret that value shown in the graph correctly).
A part of the most extreme forces closing in to 2g were caused by the short normal load peak you get on accelleration, but the most values were in the 4900-5000N range and the curve doesn't look all that different (actually almost no difference) when limiting the data to this range.

I also made the test with a higher tyre pressure. As expected, this shifted the longitudinal forces as a whole down a bit and the optimum slip ratio went closer to 4-6%.

just did some playing around with the xrt on the skidpad, thought I'd put up a few points on how *I find* it differs from reality. I'm a long way from an expert, just a goon with a longtime interest in cars / vehicle dynamics, and a bit of time spent on racetracks / skidpads. I drive an rx7 turbo, similar weight, size, power curve / output, open dif, standard suspension, moderate camber angles, good quality but high-ish profile tyres, good mechanical condition, etc. I've tried to emulate all this as closely as possible with the setup, though have had to pay more attention to how the car drives as opposed to any sort of strict accordance with numbers - none of the cars I've driven start this slow slide sideways when you get them at the limit under neutral throttle situations

I realise the xrt is *not* my car, I'm not trying to suggest they should behave exactly the same, but I believe with the wide range of setup options available I should be able to set the xrt to exhibit similar traits - this is not the case.

low speed physics - my car simply will not donut (spin on the spot) it just drives forward under power, whether a rear wheel is spinning fast or not. too much steering lock overloads the front wheels and it ploughs straight ahead. With an open dif and similar suspension settings the xrt spins around and around on the spot. No, this one in itself is not critical to the game, but is an illustration of a current failing, imho. Power understeer (from a rwd) *has* been dramatically improved (since the patch) in all situations though, I find.

engine response - boost still rises really slowly. could be argued as just a car design thing of course, just can't imagine any sort of car company would release such a thing 5k clutch dumps in anything resembling a 'normal' car of this ilk would result in lots of smoke and screaming tyres.

off the line traction - tried the same as above in some of the non turbo cars. Feels like the tyres have the same forwards traction in a static state as they do once already loaded / deformed in that direction. Real tyres 'feel' like they suffer some form of shock loading type situation and are unable to cope with such instantaneous changes in acceleration without a noticeable reduction in traction. I really know bugger all about the mathematical physics of tyres in a situation like this, would be interested in any input

Gah, I'm up too late already. bring on the disagreements, I'll try out some of the other points I had in mind another time. I realise the current ones are skirting around the issue under the spotlight so far.... got a bit sidetracked on the skidpad

I think LFS is a *fantastic* race simulator

Glad you guys liked the posts.

Here's an example of a couple of things I was talking about:

http://performancesimulations.com/files/tire1.JPG

Check out that top right picture. Those are two different tires that have the exact same dimensions. I.e., width, aspect ratio, etc., are identical. Even though the peak grip is very close to the same with both tires (at least out to the slip angle max they tested to there), the shapes of the curves are quite different. This is strictly due to internal differences in the cords, so really, a tire engineer has a great deal of control over the shape of the curves. Nowadays it's all done with FEA on computers of course, and those simulations give quite accurate results.

In the bottom left picture, and again this is real tire data, you can see another pair of tires that are remarkably different. The radial rises very linearily and then rolls off into the peak very quickly, very much like the LFS tires do (that's the most extreme example I've seen though). The bias one is much more progressive. This data is probably 30 years old so the peaks come in at greater slip angles than modern tires do, but it does illustrate the large flexibility a tire designer has in controlling the shapes of the curves.

If you were driving that radial tire you'd say it was far less forgiving than the bias one. It's not that the grip drops off after the peak, but rather the force suddenly stops rising. In the seat of your pants that feels like you lost grip, but really it's the shape of the curve that's responsible for that. The force suddenly stops rising over a very small slip angle range, so it feels like it breaks away suddenly, so it's not really a "losing grip" phenomenon at all, but it sure can feel like it

would be interesting to see the variance between the current 'super sports' style tyres... tyres with large amounts of grip have always 'felt' like they 'break away' more sharply to me.

Quote from AndroidXP :

Very nice posts, Todd! Thanks

In the meanwhile, I made some accelleration tests, comparing longitudinal grip to slip ratio. The following graph was done with the FZR's R2 tyres at 120kPa pressure with a live camber at around 0°.

From the graph we can see, that the "missing" falloff curve from the slip angle tests is indeed there when accellerating/braking, and this also shows us that a spinning wheel accelleration is NOT as fast as a just-on-the-limit one. In the attached example the optimal slip ratio was at about 6-8% (if I interpret that value shown in the graph correctly).
A part of the most extreme forces closing in to 2g were caused by the short normal load peak you get on accelleration, but the most values were in the 4900-5000N range and the curve doesn't look all that different (actually almost no difference) when limiting the data to this range.

I also made the test with a higher tyre pressure. As expected, this shifted the longitudinal forces as a whole down a bit and the optimum slip ratio went closer to 4-6%.

But why isn't it possible to outaccelerate an FZR with same tyres and same setup significantly if he just goes full throttle? And why is traction control also slower with FZ50 for example? If it isn't tyre physics what is it?

I guess because it's extremely hard to ride on that limit. All the time spent feathering the throttle costs you probably more than simply letting the tyres spin. Testing with the BF1, a TC start at 8% slip was considerably faster than a burnout start. But yes, maybe the grip should be even less when the tyres spin like mad, instead of staying exactly the same at anything > 12% slip. Who knows

BAL00 had a great point about this in another thread, taking into consideration drivetrain and chassis flex which isn't modelled in LFS. Point being that with those factors affecting real life launches, it's considerably easier to ride that fine line IRL than it is in LFS. If someone doesn't post link I'll find it when I get time.

this whole 'flex' business appears to have become a popular point to latch onto as the cause of all failings. I don't really see how myself, has anyone suggested why it would have such a great effect, or is it just something complicated sounding that hasn't been modeled at all?

Quote from Blowtus :

this whole 'flex' business appears to have become a popular point to latch onto as the cause of all failings. I don't really see how myself, has anyone suggested why it would have such a great effect, or is it just something complicated sounding that hasn't been modeled at all?

IIRC bal00s post was about the much less popular 'drivetrain flex' as opposed to chasis flex (which seems to be a favourite cure-all ).

Play in the drivetrain causing a continuous variation in torque - which has the effect of blunting the 'knife edge' between not enough and too much.

seems like this could have an effect on the handling of the more powerful cars depending on diff settings?

I guess it just seems like a very hopeful thought to me - the assumption that since Scavier are gods, everything is modeled perfectly *this time* and it's only what they haven't yet done that is letting it down...

Quote from Blowtus :

this whole 'flex' business appears to have become a popular point to latch onto as the cause of all failings. I don't really see how myself, has anyone suggested why it would have such a great effect, or is it just something complicated sounding that hasn't been modeled at all?

Since when?

If you read the post I mentioned it makes good sense. Where else has it been latched on to as a cause of failings? You asked for a reason that certain things could happen since Todd has pretty much taken the tired "Real Tires Lose Grip - Adjust The Curves!" nonsense and shown it for what it is, and enlightend us with the truth regarding the behaviour of tires. Bal00 had some reasons for this in a different thread, and noone else that I've read had the presence of mind to realize the effects of what he says. Since I seriously doubt it's modelled in LFS (prove me wrong) then this is what you asked for - a reason that some things can happen since the tire model doesn't appear to be at fault for the happening you were wondering about.

Quote from col :

IIRC bal00s post was about the much less popular 'drivetrain flex' as opposed to chasis flex (which seems to be a favourite cure-all ).

Play in the drivetrain causing a continuous variation in torque - which has the effect of blunting the 'knife edge' between not enough and too much.

seems like this could have an effect on the handling of the more powerful cars depending on diff settings?

Indeed, it seems very plausible. Well articulated in few words Col.

I would also surmise based on his idea that to a lesser degree, chassis flex plays a similar role. Anyone that's seen a high powered drag car come off the line with enough traction (with seemingly enough torque to twist the body enough that is looks like it will snap like a twig) won't need an explanation as to why.

Note the vertical osciallations at the front the cars in such circumstances that continue long after launch as well. Power absorbing, energy storing, transient reducing elasticity is visually apparant, IRL but not in LFS. It's not logical to think that it doesn't have a pronounced effect on buffering power delivery to the wheels.

Quote from Ball Bearing Turbo :

Since when?

If you read the post I mentioned it makes good sense. Where else has it been latched on to as a cause of failings?

Sorry, your initial post sounded like you wanted to lump chassis and engine flex together and blame it for everything I feel like I've seen quite a few posts that carry on as though chassis flex is the only thing left to be done for the sim to be perfect... maybe it's just a feeling though

I agree that it's worth bringing up - but it is just one hypothesis without a lot of backing behind it. I don't have enough knowledge of the physics behind it to be able to suggest how much effect it really has, and I couldn't see anything in the other thread to suggest anyone there did either. Tweaking of all the other bits has brought us great strides, suggesting they can't be improved further just seems kind of odd to me.

Quote from Blowtus :

Tweaking of all the other bits has brought us great strides, suggesting they can't be improved further just seems kind of odd to me.

Hey, it'll never be perfect that's a given. I think though that there's a level of refinement present now that we can start looking for other reasons for some things, since the generic tire behaviour - even as measured by Android's graphs, and Todd's input of seemingly endless know-how with tire modelling - seems very good since the patch. I'm sure the devs know what the limitations far better than we do though

Quote from AndroidXP :

I guess because it's extremely hard to ride on that limit. All the time spent feathering the throttle costs you probably more than simply letting the tyres spin. Testing with the BF1, a TC start at 8% slip was considerably faster than a burnout start. But yes, maybe the grip should be even less when the tyres spin like mad, instead of staying exactly the same at anything > 12% slip. Who knows

I think this kind of start : http://rapidshare.de/files/20129298/FTwithoutTC.wmv.html

should be significantly slower than this kind of start :

http://rapidshare.de/files/20129603/manualstart.wmv.html

Even if it isn't spot on perhabs. But it should be much closer ...

I didn't manage to start BF1 without TC. It spins after a few moments full throttle ... perhabs a setup thing? Can you show me two starts of BF1? I'd like to see TC being faster than full throttle ... completly same setup except TC, automatic shifting and no steering axis assigned ...

But I think it is the thing you mentioned last. Spinning wheels should get much less grip if spinning even faster than that.

Slip ratio 0,2 ; 0,3 ; 0,4 ... its all the same and near enough to perfection. It should get lower and lower. In LFS slip ratio 1,0 would still be fast I think

well, even in the F08 the NONE full throttle start is significantly faster than the full throttle spinning wheels start.

So the physics is on a good way already. it might not be perfect, but in some of the fast cars it is already better to start with a decent pedal work.

I made a custom layout where I just tested the accelleration in first gear.
With TC at 8% my fastest time was something like 2.85, without I never got below 2.93, IIRC.

I think the difference is far too small to warrant not going pedal-to-the-metal in cars where the stability isn't an issue to begin with.

Quote from Vykos69 :

well, even in the F08 the NONE full throttle start is significantly faster than the full throttle spinning wheels start.

So the physics is on a good way already. it might not be perfect, but in some of the fast cars it is already better to start with a decent pedal work.

But that's just because full throttle start spins the F08 after a few seconds. At least at my tests. So even if it just begins to spin that costs time. With TC it doesn't spin or begin to ...

Waiting for next tyre update.

Quote from AndroidXP :

I think the difference is far too small to warrant not going pedal-to-the-metal in cars where the stability isn't an issue to begin with.

In LFS but not in real life.

Quote from RIP2004 :

But that's just because full throttle start spins the F08 after a few seconds. At least at my tests. So even if it just begins to spin that costs time. With TC it doesn't spin or begin to ...

Read what you're typing. FO8 doesn't have TC. You're thinking of BF1. And the FO8 can be launced with full wheelspin starts and is perfectly controllable. Unlike BF1.

You can easily launch the BF1 with full wheelspin through the complete speed-range of the first gear. It's just extremely slow .

Vain

@Bob Smith :

I meant F08. It isn't stable with full throttle start. You have to make corrections with your steering which cost time. At least if you are looking at very small differences in time ...

http://rapidshare.de/files/22468812/F08.wmv.html

No steering axis assigned ... so no things that could manipulate the result.

And TC was a bad term. I meant controlling throttle manually. If you push the throttle less, it isn't starting to spin and you don't have to correct => costs less time and is faster on a straight.

@Vain : If you change the gear ratios perhabs :

http://rapidshare.de/files/22469186/BF1.wmv.html

___________________________________

You'll really notice the problem if taking the GT cars. IRL there are huge differences between different drivers at the start. And no single one is just pulling away with full throttle.
In LFS it is completly different. Nearly everyone is just pushing the pedal to the metal. And it is kind of a perfect start. Even if less throttle would make a difference of 0,0001 seconds. IRL the difference is HUGE. Huge enough to overtake without a different setup or slipstream ... just by getting a better start.

Having tested myself, yes, the FO8 will spin with a full throttle launch and no steering inputs... but it doesn't take a lot to keep it straight. I mean really, it's to be expected the car will begin to turn one way, so I'd still call that stable.

Indeed. Having no steering axis assigned seems counterproductive to this whole examination.