Can you drive an F1 car on the ceiling?

[jubuttib]

Hmm, could them be run at ceiling too (in theory)? I'm wondering if F1 is only one or if there actually is others too?

Depending on the setup, probably. Gumpert claims that their Apollo could, and that's a street car.

 

[K Szczech]

Yeah, Saleen S7 is also supposed to produce enough downforce to match it's own weight at 257 km/h.

But that's not enough to drive on ceiling. Not only you need to counteract gravity, but also produce more downforce if you want to have any traction and therefore ability to maintain speed and control. And the faster you go, the more traction you need to counteract drag.

So I'm guessing S7 would need to go well over 400 km/h to produce 3-4 g's and actually drive on ceiling.

Open wheel cars are much lighter so the speed at which they produce their own weight in downforce is lesser. They're also capable of reaching speeds when they do produce 3-4 g's of downforce.

 

[Marek Lesniak]

OK chaps, some good news!

http://youtu.be/ZHUnTqzRcSk

Got a few tips from Terence and yes, it is possible to drive upside down. The trick is, you have to set all surfaces near every 90* of banking to HATTarget=false (exclude from the HAT). On the video, you can see I had some issues with traction in the tunnel - that's because I excluded all tunnel parts from HAT (the whole loop was albo excluded from HAT). By doing that you basically drive on raw surface without any smoothing and stuff. Also notice, I eventually start driving in the opposite direction - that gave me better results and allowed me to finally get to the ceiling area

 

[K Szczech]

Yup

I believe this is rather a limitation in track technology itself - where racing surface is stored in form of heightmap, rather than full 3D mesh. If you recall - rF1 had HAT data for that.

I incorrectly assumed tyres will not work properly on non-HAT surfaces I guess I didn't give it enough thought or I would suggest disablng HAT myself in the first place.

Just in case low speeds will draw attention - LesiU was obviously using exaggerated downforce for that test.

 

[martymoose]

Nice one, how much extra downforce was needed for that or was it similar to the f1

 

[Marek Lesniak]

Yes, like K Szczech said, in my case speeds could be low, because I put A LOT of downforce. About 65kN at 200kph, while the car (it's a modified GT1 class car) have mass of 1200kg (so about 11.7kN).
Front end starts to unstick at about 95kph - this is where front downforce is equal to front axle load in that case. Of course, you need slightly more speed to be able to steer the car.

I edited the video's description. It is not possible to drive on ceiling on older tyre model (tbc), with surface not assigned to HAT. That works only with the TGM.

To sum up - there should be enough downforce to overcome the gravity, to be able to drive on ceiling.

 

[Terence Groening]

I incorrectly assumed tyres will not work properly on non-HAT surfaces I guess I didn't give it enough thought or I would suggest disablng HAT myself in the first place.

Yes, rF2 tires work on both HAT and non-HAT. Sometimes this works basically as expected, like if you ever drive one of the historics on Spa into the trucks in the paddock and get a tire jammed between the road and the bottom of the truck. At other times, like touching a tire against a guardrail, there's so many opposing contact points that it doesn't really work realistically (and sometimes even flips out, but I currently think that's actually due to a poor interaction between tire collision and regular collision).

 

[K Szczech]

there's so many opposing contact points that it doesn't really work realistically

Yeah, getting multiple contact points physically stable is a real bugger

By the way, I know you're using center point and radius for each triangle in collision body, but do you also build a tree out of these spheres while loading a mesh?
I've noticed CPU usage raising insanely while drifing F1 car within 0,5m of metal guardrails at Eau Rogue in some early builds (49 for sure and perhaps next one aswell) but it seems gone now.

That would be an important optimization hint for track modders to know we don't need to split long guardrails into shorter pieces just to make collision detection work on smaller domains.

 

[jtbo]

Yes, rF2 tires work on both HAT and non-HAT. Sometimes this works basically as expected, like if you ever drive one of the historics on Spa into the trucks in the paddock and get a tire jammed between the road and the bottom of the truck. At other times, like touching a tire against a guardrail, there's so many opposing contact points that it doesn't really work realistically (and sometimes even flips out, but I currently think that's actually due to a poor interaction between tire collision and regular collision).

45 degree camber limit has some to do also, I have tried 90 degree camber limit and that seem to create tires that work better with those random situations as if I have understood correctly, it then uses tire contact model to almost any tire/surface contact?

That has lead me to think that it is not actually only 2 wheel driving and motorcycles that can benefit from that, but also open wheelers with tire to tire contacts and contacts with guardrails etc.

However my understanding of things is rather limited as of now, so it can also be that changing that camber limit just masks underlying issue, that might be better to solve differently.