Guys, come to talk about tires

I have been enjoying new ARB updates in 0.14, there is certainly change to vehicles once again to better!

I think that many that are blaming tires to be problem might, in addition to controller issues, have setup that is not quite optimal for their driving.

For example here is one race car:
Caster 9.3
Camber F -3.2
Camber R -2.5

Toe F -0.1
Toe R 0.16

That is set up to over 2.6Hz springs small downforce 300kph top speed.

Setting over 9 degrees of caster in BeamNG is not quite simple. After playing etki enough, I have noticed it is bit limited in setup range for a race car.

One 70's RWD econobox I had there was 7.8 degrees of caster if my memory does not fail me horribly.

W123 street car form has caster of 8 degrees.

C230K is bit like K-series I guess? Runs 10 degrees of caster http://www.peachparts.com/shopforum/tech-help/95898-c230k-coupe-alignment-question.html
(k-series does not show anything in caster value part of tuning window from some reason)

Caster is good for at least two things, steering feel and dynamic camber, which means that tire gets more negative camber when you turn your wheel. Negative camber and lack of toe can remove steering feel so more caster is needed to compensate.

So for me it appears that BeamNG is running bit low on caster, but I don't know if that does do much to what people are experiencing, maybe higher caster would help with steering feel?

Anyway setting up car as close to race car does improve how BeamNG race variety vehicles feel and reduces the impression of understeer when entering corner too fast.

To get best out of tires one really needs a good setup and as team gathers more data and experience on car setups, I think we see improvements from that area too as we have continuously seen before.

Then again does setup tuning page show caster correctly, etki shows 5 degrees which is almost half of what one would except?

 

Well... I can only speak for what I have laying around...

5° may not sound like much, but I would say it's good enough... really depends on the camber being run, the amount of toe, and the width and sidewall of the tires.

For instance... My Corvette has really fat, low profile front tires, and only runs -1.0° of camber in the front tires and the toe is set at -0.110... thus they get away with merely 3.0° of caster angle.

Fast forward a few years, and the exact same car now running the same camber of -1.0° and 0 toe now uses 5.5° caster angle... interesting...

Generally, the more camber you run, the higher caster you will want. That will keep the inside tire "flatter" with the ground on the inside of corners while cornering and really bank in the outside tire (giving a much higher effective camber to the outside tire)

to make a long story short, here are some setups they used in the 80's and 90's on Corvettes presented very nicely (just scroll to the bottom of the page)

http://cartechstuff.blogspot.com/2013/02/c4-corvette-alignment.html

As for how that feels... well throw some sports tires under it and it feels like a million bucks... another thing to remember though that plays a big part in this is roll bar stiffness. I personally like rather stiff front roll bars as that usually keeps the car planted and under control. Then I like just stiff enough rear sway bars to get the back to be a little curious and then cut it off there... nothing too crazy.

That will make a MASSIVE difference in how the car drives alone. Sway bars are crazy important.

 

sorry, it was just some weird sarcasm.

i've been silently following this thread because it's really informative and full of people trying to extract knowledge from themselves and others. this thread should be pinned for an example of what all forum threads should look like

 

That is interesting as you have one of the masterpiece of American car engineering that you have experimented yourself, but as you said camber affects it a lot.

German cars tend to have lot more caster for given camber, which is curious, why they run twice the amount you have good experience from?

In racing world however -3 degrees of camber at front is what they usually aim to, but obviously there are bit more variables.

My BeamNG crashed again, I was adjusting camber, does not like Thrustmaster wheel plugged in at all. Anyway I made good progress with only few crashes to get caster to 9 degrees and camber, toe etc. dialed in.

Test drive was quite good, but damper settings clearly were off, I must say vagueness at corner entry does dissappears with caster increase on etki you can get lot more pointy feeling of the car, I would say that two laps around nordy did felt much closer in turn in of corners than what I experience in AC for example, but more tuning is needed.

Also etki is heavy and with correct 4 pot engine weight at front it tends to have more weight on rear axle than on front axle, so hesitation at turn in kinda comes with that, however there is definitely something in low caster vs high caster in game world.

Sway bars indeed are important to feel, but least you can get away with, better, usually. Especially at rear it limits rear wheel grip at corner exit, so often running without rear arb if possible is helping, however usually that is not possible as front would need so stiff one or springs would need to be too stiff.

This new ARB code/update did make cars much better during cornering and especially when shifting from straight to corner or corner to another, hard to describe, but I would say transition is now smoother and feel more like real thing, before it was bit like hitting a hammer when coming to corner.

Well, that alone does affect a lot to feel and again is hard to tell apart from tires.

All that suspension setup is so very important to get car behave and it depends much from torque curve, driving style, weight distribution etc.

I guess I need to send my creation of Frankenstein to you when I get it working way I want

 

I'll have to screw around with some of the cars and set one up... the 200BX ironically is probably the closest to my car, so I will give that a shot. If I can get it to within 80% of what my real car feels like, it would feel epicly good in game.

 

There is this kind of thing, adjusting caster moves axle instead of strut top which is what IRL is moved when adjusting, so going from 5 to 9 my front axle is quite forward:


Obviously there are reasons why 5-7 degrees is used with most cars, it can be that what I see as 9 is not 9 degrees of caster really or that these cars are really not designed to work with such caster.

Also there probably are good reasons why lower node of hub is moved instead of upper.

Learning and understanding more of this is helping to craft better setups of course.

It might be that to get numbers where I want, some suspension bits need to be bit moved around, it is really good already, but I feel that it is bit a stretch from current node setup, there are times when something feels like giving away a bit.


However tires work really great, threshold is very clearly there and thing is pretty fast, but I'm not sure if BeamNG racing tires like so much camber actually, needs lot more skidpad and track testing to find out how to get best out of the rubber.

 

Welp... I spent a good long while on this... probably could get it better but I think it's pretty good as is right now... take it for a spin, see what you think. This is made off of the 200BX, and all it is is a configuration file.

Once installed into the game, just go over to the save, load tab and load up "test car"... should work just fine.

This car handles good to me and my driving tastes... it may not feel good at all to others... take that as fair warning.

 

Not read through everything here but i say this, tire physics are probably one of the best, if not the best in the sim genre. Might not be popular with the simracers because it hasnt attracted that crowd yet and hasnt got the cars and tracks to match.

My 2 cents are that its down to the cars chassis/settings. Some cars suck due to not being jbeamed 100% realistic or its simply jbeamed 100% realistic and is a pig on the road, for example the first C63 AMG, understeer galore.

I watched a vid of Jommy Broadbent where he tried to complete a lap on The Ring, crashed all the time, bs'd somewhat about BeamNG. Made me decide to try it out myself, with a car i knew was well jbeamed and set up. Check this out (on version 0.13):

 

It was kinda fine during the acceleration, but then at corner entry I felt something was lacking a bit, had quite bit of understeer to my liking, but I guess everyone to their tastes

Spoiler

There was probably a mod used which I don't have installed.

--- Post updated ---

That config has AWD, which makes it quite prone to understeer, damping is also differently set that I have used to, I think that track day setup could be optimized for that track even further.

It is not so easy to get setup to work though, etki with M102 DTM tuned, while I like front end setup, rear is too loose with 10" wide tires, with 12" it is better, but then front too would need 12" slicks and I think that is what they used in DTM at late 80's early 90's?:

Two corners I messed up quite badly, entry at horribly wrong line etc. and then it also liked to plow straight a bit, but

Would probably need to set even lower FFB strength as Caster of 9 is making everything quite strong, but also more detailed/faster.

But reason why simracers don't use much BeamNG is not tires, setups does affect first impressions etc. of course and as they need work, that might make some to create those impossible to drive videos, but I think real reasons are:
-UI is too slow to bare when building setups, also layout of items on tuning page is not very handy at all, you have to move a lot and it is very slow, so changing 4 clicks of something and two of something else can take up to 30 seconds or more.
-Can't get telemetry to MoTec (unless this has changed?)
-Timing functions/hotlapping are not so great. Huge glowing checkpoints are bit arcade feeling, in typical sims timing is built on track with sector times and no glowing things (hey this is WIP Alpha still)
-Online gameplay/leaderboards are not present (for obvious reasons)
-No race weekend, typically that is how even hotlapping is set up, you are on track, building setup, with menus for garage and timing etc. It adds to experience of being there, immersion.
-TV cameras build to trackside specific locations, rewindable replays and such, which are very difficult to make in BeamNG.
-Tire wear, temps, flatspots etc.

Most of those points are WIP and going to improve, but as there was earlier some question about why sim racers don't use BeamNG, I think those are reasons, not so much of realism of tires as on that BeamNG is doing quite good job.

For comparision Assetto Corsa lap and bit more, car is some 400kg lighter than etki race version, but etki pulls more G's at corners and has higher cornering speeds (it is worth noting that etki has added downforce and that there is huge difference in grip when accelerating, not saying anything which has that wrong or right):
video is ready in 30 minutes or so.

 

There is one other simulator using a physically based model and first principles to simulate their tires. It's iRacing. If you haven't read either of Dave Kaemmer's tire blogs, I would highly suggest doing so, as they are trying to do mostly the same thing you are, and have had great success compared to the other emprical models in a lot of areas.

 

iRacing is so horribly expensive that that for me at least it is something that really does not exist.

Setup or tires, well, I decided to test default setups today a little, these are just my observations and fails at driving

For reference At late 60's TransAm racing series cars did pull bit over 1G's, 80's sport cars (ZR1) did pull 1G with street tires, ZR1 Corvette of today, 2019 model gets 1.24G's on 200ft skidpad, that is on street tires. (note that temperature, skidpad size, driver etc. affect to results and they are not necessarily directly comparable, but gives you roughly there idea). (I did test on largest circle on gridmap, that is roughly d183m/ c580m, so obviously not comparable to 200ft results directly. At least aero effect can sent things off)

Every bit of understeer means same car with same tires, but different setup can get higher skidpad result, same is true with oversteer, but with oversteer things are just different.

So my verdict is that some cars are not having optimal setup, not all cars need to have optimal setup, but some cars might need more tuning and maybe something from there is what then makes people to think it is the tires.

///////////////
Nice to drive fast class, which is neutral to mild oversteer for me.

Bus I found to be very nicely balanced, small bit of oversteering with torque, but pretty much perfect tune for racing. It just needed few mods to get up to speed.


Gavril Barstow Kingsnake, very nicely balanced, I felt it being pretty much neutral.


Bruckell Moonhawk Track, pretty much like Kingsnake, close to neutral, easy to put to oversteer


////////////
Understeer class


Ibishu Covet, FWD, so understeer comes with it, but this is easy to point in turn, can make line changes even while pulling 1G, using brake trick can do 1.1G without too much trouble.


Ibishu Pessima Hillclimb config, 1.2G's with a stretch, hillclimb cars seem not to do much more, need to test Sunburst of that variety, downforce is not helping yet at this speed?:


ETK I-Series Track config, 1.0 to 1.1G pretty close to Kingsnake and Moonhawk tests, but understeering clearly, so there is probably more to be gained by setup changes.


ETK K-series Tracday config, this has AWD, but manages to pull less G's than older I-series track config and generally is bit uninspiring, Kingsnake is equally fast in corners while being more interesting to drive.

--- Post updated ---
First, do note that g-forces on circle test are not same as those on race track, because of banked turns, you can usually pull more G's on racetrack. Also laid down rubber helps on dedicated race tracks.

For reference, this is probably quite close comparable car to track variants in BeamNG, this one shows G-forces which to my eye go to 1.5G, easily:

That video has everything, slips, slides, even RAIN and you can see how strongly rain affects grip of that vehicle.
Notice how he still pulls 1G at corner before long straight while sliding at wet surface.

This is best of the best, note how hard and early he is on throttle at corner exits and how rough the ride is, but car seems perfectly planted, despite insane amount of power it has, at speed aerodynamics at work, we don't have anything even near to this level in BeamNG:


This would be something like SBR4 not really sure if there is any config that would match this, but it is very worth to check video and try to match even hillclimb variant to that:

 

I used a splat map last week to develop a deformable terrain shader, in my effort to create a deformable snow shader in Unity. Perhaps it might be worth looking into splat mapping techniques when it comes to deforming geometry. Image processing is pretty fast, and you could get the GPU working on that quite nicely. It wouldn't be a huge undertaking to create a sort of two way interaction between the two things. It would be rudimentary at first, with room to refine it later.

 

Send that video to jimmer and show him how to drive on beamng.
But, I think he doesn't realize that beamng simulates each node of the object (car) while other similar games mostly simulate it like one solid object.

Thats just my opinion though..

 

Quick bump but still on topic.. I put race suspension and Autobello Type 2 tyres on an Ibishu Covet and it handled really well and extraordinarily realistic.

 

Well... those are the newest tires in the game, so I am sure the dev's probably threw everything they have learned at them... thus resulting in the best of the best compared to everything else.

 

I know this is a double post but I want anyone that's interested to get a notification as they will probably want to look at this.



So I don't know why... but this guy decided to launch his truck on drag radials at different tire pressures.... the video isn't scientific by any means, but you can actually gather a decent amount of data from something silly like this.

Mostly the really low pressures... you can see just how much the tire bounces back and forth when launching... so much so it almost hits the leaf springs. I wonder if they would also do this in Beam when running next to no pressure... I would try right now, but I am currently not home, so that will have to wait till later.

Edit: Just saw this on his channel as well... this guy is nuts, but again this is very interesting to me how well it does. More useful information I guess.

 

i love that guy, his instagram is a laugh as well. my favorite clip from that is actually when we get to see the tire grip and launch the truck forwards, it's incredible to watch

 

Rubber compound, softness etc. is very much important, sure some tractor thread helps, but as long as things are somewhat dry, different rubber has huge differences.

D15 with 0PSI is still quite well held up by rubber sidewalls, not sure if there is something or not, but especially at rear default tires hold quite well without any air in them. There is something like 1/4th compression left of tire profile height when pulling axle down while having 0psi in tires.

Definitely difference to IRL, but then again real tires have much more detail in construction and if tires work well with normal pressures, then how much weight there would be on those extreme tests?

It could be that air pressure effect could have more spring and sidewall less spring, but on lateral forces would it then be off there?

Better brains might understand if that needs more examining or not.

 

I know its been a while since we all were talking about tires and traction and what not... but I have come across something that I need some help with and this seems like the best place to go about it... so yeah...

Ok... can you explain to me exactly how friction is generated in fluids? ie. Mud and sand as I think both are kinda treated as fluids ish... Also... how is buoyancy calculated now? It seems to have something to do with the collision triangles, but I am not sure...

The other day I saw somewhere where you (or someone) stated that friction in a fluid is calculated by the change in angle of a surface in fluid over time? Is that correct? How exactly does it work? Also, this used to be generated on a per node basis by defining specific volume and surface area on a per node basis. Is this not done anymore?

The reason I ask is I am working on a... well... contraption. I don't want to completely ruin the surprise, so I am just gonna say... I am making a tank. Anyways, this particular tank has rubber tracks and whatnot like a tire, just it doesn't use the same geometry. Instead it uses the geometry of the track (naturally) with a sharp turn at each end and a nice, big, flat area at the bottom.

Now... the issue I am having is that of several with how fluid friction seems to be calculated now...

1. The track has tremendous traction on pavement, can't really say that's a surprise, but when I enter mud and sand, it seems to have almost no traction... while, being a tracked vehicle, this is where it should shine above and beyond most average vehicles. In mud, it seems to generate almost no traction whatsoever. Only just enough to propel it slowly forward in flat mud in the testing area... while sand seems to just sink in a bit and then have almost no traction whatsoever. It just slides around like its on ice.

2. Buoyancy... uh... how is it calculated? My tracks, despite having collision triangles, seem to sink right through the mud and sand as if there is nothing there... like... it just passes right through it like it doesn't exist.

Are these calculations done exclusively on the collision triangles now, or do the nodes still get some say in this? From what I can tell with my testing, it seems like the node properties (at least for volume and surface area) seem to be depreciated. Also... how friction and buoyancy in a fluid calculated? If I know I might be able to optimize my design to work better with the current system. I thought about adding paddles, but that would add WAY to many nodes and beams, and I am pushing the limits with my current iteration already (more optimizations will be done naturally... right now it is definitely a POC)(Proof of Concept or Piece Of Crap... take your pic)

At the moment I am just very confused with this project...

I should have great floatation on mud and sand despite not having any air to hold it up while also having fairly decent traction since the shear force is divided up over a much greater area than a normal tire. Side to side traction I expected to be sub par as there is nothing to push against unlike the sidewall of a tire... that seems to be pretty realistically calculated... i am just rather confused...

 

Buoyancy should be calculated with tires, I don't know if it uses anything else. Water is treated as very viscous air in all other cases.