Cars Caterham Roadster 500 1.3

[fasfdasf]

Well even with no angle dependence it will almost flip if you go flat out. Speed and how much throttle you keep during the jump plays a major role than any aero since rear stays on ground if you floor it through then you essentially power the rear under you and it's the rear grip and power that is making you flip. I tried it now a few times, if I lift off during jump the car doesn't go anywhere near crazy in the air. I will have to eventually calculate the speed he's doing in the video if he's going full throttle no idea I don't do that with many cars in AC over this jump as it will ruin your steering, line and braking ability.

Even making rear lift more with angle than front would have to be crazy imbalanced for it to have an effect.

The aero is still work in progress and will get updated on all the Cats.

The data we're using are from this Cat standing in a wind tunnel:

As usual this doesn't seem to be a quick and easy fix. Even the comments in the source for the data literally say: interesting result and quite disturbing. And that's exactly how it drives and also seems to be in line with other less precise source from CSR variant. The front to rear lift balance though cannot be verified from a 2nd source and may be "faulty"/not correct.

The Kamil's Academy Cat, I think it was Kamil, had even higher front lift balance but it's total lift was low so one didn't notice it as much, with these more realistic lift values this high very high front lift balance is what makes it want to flip over and also lose so much front grip overall compared to the value being better spread out front to rear.

Some flipping may also be from suspension but I find it quite unlikely, it's more likely the car is still a bit murderous due to some polish needed on the rear suspension, but for a quickly done thing it's usable.

The front-rear lift balance will have to change most likely and quite a bit of testing be done with any reference possible. Does DPR have any aero data, no idea but it would certainly be of big help.

The article you're referencing states about 35kg of lift at the front and 2kg at the rear at 100mph. But if I check the tyre loads in AC, I get about 66kg of lift at the front and 26kg of downforce at the rear at 100mph instead. Is that as intented?

The Wings app does show 39kg of lift with a 94% front bias. But that value only considers the vertical forces of the wing elements. The drag on them will create a backwards torque on the car, pushing the rear wheels down and lifting the front wheels up, which would explain the differing values above.

 

[JackCY]

The article you're referencing states about 35kg of lift at the front and 2kg at the rear at 100mph. But if I check the tyre loads in AC, I get about 66kg of lift at the front and 26kg of downforce at the rear at 100mph instead. Is that as intented?

The Wings app does show 39kg of lift with a 94% front bias. But that value only considers the vertical forces of the wing elements. The drag on them will create a backwards torque on the car, pushing the rear wheels down and lifting the front wheels up, which would explain the differing values above.

There is a dev app for aero and it should report same values as I have calculated. I can read them out of my data at 160km/h without launching AC and AC from my testing when verifying was spot on using the very same formulas and had equal lift, drag and balance values.

On 1700 and consequently this roadster the lifts and balances are as follows at 160km/h. Hold on maybe I converted miles/h to km/h wrong still it should be 160 or there about, yeah yeah it's supposed to be 160.934km/h for matching, will correct that, otherwise it's calculated at 20C and 150m elevation both if which you have to match in AC for testing no idea if matching elevation is even doable in AC I use skid pad oval track for testing. Anyway from my testing the calculation I have seemed to be spot in equal to what AC calculates at 160km/h +- AC rounding errors. Plus you have to open another dev app to see physics real speed not what you somewhat guess from the analog graphical speed meter. The wheel loads and weights, there are at least 2 I think and it's a whole another topic and I would not bother trying to guess aero from changes on tyres loads from it as it's probably not very accurate and hard to do without logging data, overall a bad idea and tons of corrections needed to even be able to judge the aero accurately from wheel loads. Use the AC aero app.

Downforce: F -33.6112 kg = -74.100 lb, R -1.8603kg = -4.101 lb, balance 94.76% toward front (50% = 50/50 balance), total -35.4715 kg = -78.201 lb, body has none

The target to match was -33.6112 kg and -1.8597kg.

Of course this all aero thing is not ideal it should be calculated by AC realtime from the 3D model, Kerbal Space Program does that to some degree especially with a free mod to it and is quite damn accurate. In racing sims though I have not seen a good aero simulation yet. Why do they keep using static or LUT Cl and Cd coeaficients? Well because they can get that from wind tunnel machines and it's cheaper to do than to program a better realtime in game aero physics simulation.

I've tried finding a 3D aero simulation program but could not find anything, certainly not free, that would allow to load the 3D model, say let it simulate over night and then get the Cl and Cd values out of it or at least something for extra reference. If someone found such program let me know but I bet they are all proprietary and very hard to get.

There is another issue of car angle at 160km/h from aero but that's easily and already fixed, minimal impact. The matching of 160 vs 160.9344 km/h I will fix that but again the impact will be minimal.

The lift balance of 94-95% is what's making the car flip so much, it loses front grip but not rear the rear wheels don't really spin much instead they power the car and flip it around. If you guys can find videos of Caterham jumps send them to me with a time code in the video at best ones that show somewhat precise speed at moment of jump.

Absolute lift values are probably quite correct but the balance I don't know what they did to their car or in the windtunnel but it seems off.

 

[fasfdasf]

Those little inaccuracies like the exact speed or air pressure don't bother me. All I know or heard of before is that the aero balance in the dev app is incorrect as it doesn't take drag forces into account. I don't mod myself and I don't know how modders get their aero balance correct either, so someone who knows more would need to chime in.

 

[JackCY]

I have not heard anything about the aero dev app, only car engineer user app and even that was rumors mostly false with a minor labeling error in there. In the data physics wise it seems to be OK.

As far as aero simulation complexity goes, it is rather spartan so to speak and you can do that same on a paper by hand that AC does, except computer does it many times a second and you would do it once a minute. The formulas used seem to be the same.

I have not played with drag balance to test how AC reacts to it especially since there are 4 tyres holding the car steady it would need some extreme measures data wise to go and verify AC behavior is as it should be. Same as I make aircraft the drag is behind CG a little and same way it seemed official cars have it as well. This makes the car stable instead of trying to twitch and spin around from aero.

Verified on roadster 500 right now, -34kg lift and 94% balance at around 160km/h reported by AC.

 

[garyjpaterson]

All I know or heard of before is that the aero balance in the dev app is incorrect as it doesn't take drag forces into account.

Yes I believe what you are saying is correct - I often forget it but the effect is clear on this particular car (being so light, it doesn't take huge forces to start making big differences).

Whilst your inputs are correct, the output of the drag induced downforce is effecting the car but not being displayed in the app.

I'd ask @mclarenf1papa if you want someone who knows what they are talking about to explain it.

 

[fasfdasf]

I have not heard anything about the aero dev app, only car engineer user app and even that was rumors mostly false with a minor labeling error in there. In the data physics wise it seems to be OK.

As far as aero simulation complexity goes, it is rather spartan so to speak and you can do that same on a paper by hand that AC does, except computer does it many times a second and you would do it once a minute. The formulas used seem to be the same.

I have not played with drag balance to test how AC reacts to it especially since there are 4 tyres holding the car steady it would need some extreme measures data wise to go and verify AC behavior is as it should be. Same as I make aircraft the drag is behind CG a little and same way it seemed official cars have it as well. This makes the car stable instead of trying to twitch and spin around from aero.

This post is where I heard about it: http://www.assettocorsa.net/forum/index.php?threads/discuss-rfactor2.36890/page-48#post-910629
What I think you could test is move all the wing elements to the same height as the CoG, that would get rid of vertical forces caused by drag I guess. See how that changes the handling and the behaviour on jumps.

 

[JackCY]

The heights don't matter that much, tested it right now all zeroed to CG.
Getting on the original angle dependency -50kg total lift at -50deg during a jump. It's not that big a lift in kg but the balance is what's off and has to change most likely. It's the only thing that will make a difference.

Will keep testing it, see what the heights mess up if anything, it's possible.

The drag induced downforce/lift makes sense but then... should AC even be doing that...
Let us define wings then and do the simulation instead of AC wanting this and that number, then making down/lift values out of drag? When those are already defined elsewhere? Then not even showing that to the user? Is it really that bad in AC again?

Gotta try the Nordschleife jump see how it goes, suspension test track is too bumpy.

 

[Ghoults]

On the R500 there is an air exhaust vent on top the nose that they did not have on that wind tunnel test that should reduce front lift.

 

[JackCY]

Yes there are tweaks to lift and drag that the more modern variants have I have some values premade for that.

The wing height does seem to have an effect in AC on car rotation "=" lift/downforce. It helps a bit but not a to point where you could go over the jump in question safely at 212km/h. It still goes bonkers, only less.

---

"Success", can go over at 215km/h no high flip no nothing crazy. Here is what I have set now, no angle dependence as most AC cars don't have either, drag slightly below CG. Thanks, the height is probably having a bigger impact than anticipated, will dial it in eventually.

 

[JackCY]

Ben has a quick fix patch and some other previous patches, not sure he had the time to test and upload. I've tested the aero changes myself, checked wheel loads and there are much more complex changes needed to the aero than simply playing with drag position. I will have to recalculate it so it's easier to tune.

Also AC seems to have moved to CPU bottlenecking in recent patches, no idea why, could be the Intel fiasco and issues they knew about since 1992 and ignored. If I run the built in replay benchmark it runs "fine" 155fps but with maxed out 91% CPU load where as before the CPU load was around 70% (60-80%). Online often FPS down by some 25-40% and CPU maxed out. Even offline on a simple track with one car the FPS looked abysmal.
Anyone knows what's up with AC lately? Is it some changes in 1.15-1.16 versions compared to older?

1440p used to load my poor GPU quite well, now the GPU is 45% loaded online in a race... There is definitely some CPU bottlenecking going on.

 

[JackCY]

Could not figure out what's up with AC performance, either Windows/Driver or AC patches, I've tried other things related to CPU and RAM performance. I have a history of benchmark runs and I can clearly see how the CPU load goes up in AC over time. Can get 120+fps offline but online it's a 60-100fps unable to load the GPU due to CPU bottleneck. IMHO too many drawcalls and outdated graphics API for what it needs.

As far as Caterham FPS performance goes, Ben knows what the issue is, a number of objects, sort of has not been "flattened" to retain easier editing. Not sure which of the Cat variants are flattened right now if any.

Aero, I'm still stuck with something else outside AC, at least now it's clear what the issue with aero is, quick fix was a quick fix, I'm aware of the complexities and I have to first work out where drag position is supposed to be more precisely before tuning lift to get desired total combined lift values on axles. Slapping values into AC is "easy" but researching and getting data takes tons of time. The good news is as far as I'm aware is that lower drag versions will have lower lift on front from lower drag, plus some small lift reduction due to front "wings" and vents.

The front-rear lift difference with quick fix seemed OK but lift needs to be adjusted to correct absolute values on each axle. It did drive better but it did not have a rear lift off on a Nordschleife test jump and as a result with enough power such as this 280hp roadster it can push the rear under itself quite a bit.

 

[Kristaps Dzelzgalvis]

Also the front tires are always stone cold, because of the front lift.

 

[JackCY]

Also the front tires are always stone cold, because of the front lift.

Yeah that's quite normal. Though the ST are fine for me temperature wise, you have to push all Cats to get temperature to tyres, very light = light stresses on tyres. I can't speak for other tyres on this roadster, Ben put those there.

 

[Nico]

Hey, I apologize if I offended you with my rating, it wasn't my intention. I just wanted to point out a minor detail.

 

[JackCY]

I found at least some solution to my AC FPS issues so I can use AC again without a slideshow feeling.
x86 variant of AC is dead, very bad in recent versions, x64 runs fine and similar to what x86 used to run before. It's still poor GPU load in multiplayer but at least as bad as it used to be and not worse. Gotta use x64 with recent AC versions.

Newest ReShade 3+ works with x64 now and it has gone a long long way over all the years since SweetFX and others started the shader tweaks. Now has an ingame overlay for easier configuration.
No idea why game devs very often when they allow anti-aliasing in some form do not add a luma sharpen filter as well as an option as it is pretty much a necessity to combat blur caused by enabled AA. Still no depth even with latest ReShade, seems they killed depth buffer access in one of the old AC patches for good and even newest ReShade won't dig the depth out, a shame it was very useful for videos to add depth dependent effects and made AC look way more modern.

These are often the ReShade/SweetFX settings I use in AC for those interested, shadow-midtones to taste, tweaked it just now to be less aggressive and more natural looking:

[LiftGammaGain.fx]
RGB_Lift=0.960000,0.965000,1.000000
RGB_Gamma=1.030000,1.025000,1.000000
RGB_Gain=1.005000,1.000000,0.995000

[LumaSharpen.fx]
sharp_strength=0.400000
pattern=1
sharp_clamp=0.035000
offset_bias=1.000000
show_sharpen=0

Obviously this also depends on what PP filter you use in AC.

 

[Ben O'Bro]

Hey, I apologize if I offended you with my rating, it wasn't my intention. I just wanted to point out a minor detail.

Hey no worries. Model is far from perfect. Then the ''5* rating only'' becomes a useless quest. Plus i had a few too many pints. So that happened

 

[Nico]

I admit I may be a little bit too nitpicky when it comes to small details, and I never really tried "deep" modding, just some easy texture work. I also admit it's only noticeable on the yellow skin with the marked tyres, and can be somewhat disguised by using the fully black tyres from the other skins. That doesn't take away the fact that it's fun to drive, and somewhat challenging. It took me like four attemps before I could do a clean, fast drive on the Glava Zete hillclimb.

 

[Kristaps Dzelzgalvis]

That doesn't take away the fact that it's fun to drive,

Wait til you get the updated version
Current one is understeer galore

 

[Ben O'Bro]

Same data i sent to @Kristaps Dzelzgalvis a while ago

i won't include in an update until we have a more finished/polished version
so in the meantime if you want to have a try to see the difference,
see link below:

http://benobro.org/sub/r500_jack_data.7z

• rename data.acd and keep the data folder from the zip file
• the ini file at the root of the 7z archive is a setup to load that jack drives, for me street tyres are too harsh on my abilities!
  (put setup in C:\Users\$user$\Documents\Assetto Corsa\setups\bo_caterham_roadster500 etc.)

 

[João Tuning]

Hello. In case you did't work on the flames yet, I think these flame values are better
POSITION=-0.8,0.16,-0.525