S2U Community Physics Experiment

[offom]

Weights of suspension are being added to the chassis, as I said. Therefore curb weight is around 740. By looking to the suspensions files, each wheel and spindle has a mass and inertia.
20 front spindles, 30 front wheels
24 rear spindles , 36 rear wheels
+
5 driver head.
+
1 kg fuel tank definition in suspension file
+
25 kg fuel in chassis file
=141 kg + 600 kg for chassis=741 kg.
(776 was due to misremembering, sorry)

Those suspension files are about 110 kg and push the weight balance to a bit rear.

Also, there is something wrong with load transfer. When standing still, turning the wheel tilts the car (due to caster geometry, thats ok), lets say to the left side (fronts are directed to the right), front "right" and rear "left" tire (why opposite ?) loads up, the counter wheels lighten (by a huge amount like x0.8 times wheel load).

Like:
(kg)
Default,
Fronts: 250 250
Rears: 275 275
Full right lock,
Fronts: 100 400
Rears: 425 125

[ermo]

Originally Posted by offom

Weights of suspension are being added to the chassis, as I said. Therefore curb weight is around 740.

My question was related to where you found the evidence that this is the case? Which game engine output led you to this conclusion?

I'm just curious -- I'm not saying your are wrong or anything.

[offom]

Editing suspension parts' weights changes wheel load, therefore their weights are being used by the engine. That simple.

[ermo]

Originally Posted by offom

Editing suspension parts' weights changes wheel load, therefore their weights are being used by the engine. That simple.

Heh. Did you check it by setting a very high load sensitivity (so that more mass added to the suspension would make the tyres grip a lot less) or what?

Again, I'm merely curious. And kudos for some pretty clever thinking there!

[tulib202]

Originally Posted by offom

Editing suspension parts' weights changes wheel load, therefore their weights are being used by the engine. That simple.

Well i wonder what JDougNY would have to say about that

[brrupsz]

Originally Posted by offom

Weights of suspension are being added to the chassis, as I said. Therefore curb weight is around 740.

This part of Shift is taken from pMotor, so I will paste this comment from isiforums about inertia:

Originally Posted by LesiU

rFactor takes inertia from GENERAL section and substracts (after calculations) wheels/spindles inertias (from unsprung masses) to get inertia for sprung mass.

Link to original discussion: http://isiforums.net/f/showthread.php/7108-Inertia

But in short no - it doesn't weight 740kg.

Really guys, if you want to know more about Shift engine then read tutorials related to pMotor games - GTR2, GTRE, Race and especially rFactor. Everything "physics" is the same in Shift except tyre model. There is no point in reinventing the wheel.

[offom]

Quoted from "Knut Omdal Tveito":
//
This is the only one not used in the PM:
[BODY]
name=body mass=(0.0) inertia=(0.0,0.0,0.0)
pos=(0.0,0.0,0.0) ori=(0.0,0.0,0.0)

The rest of the pm file is very important. In the hdv file you specify the total mass and inertia of all parts combined, and the pm file gives mass and inertia of the unsprung parts.

To find the mass and inertia of the sprung chassis you take the total minus the sum of the unsprung (PM) parts:
m_chassis=m_hdv-SUM[m_unsprung]
//

So resultant weight distribution, inertia and mass is being defined in chassis file. Is it ? Or something wrong with graphic illustration ?

I changed mass of the wheels 15-30-50-70 kg, the result is on the screenshots.

[ermo]

The way I read those posts, the total mass (sprung + unsprung) is the mass as defined in the .cdf, the mass of the suspension/wheels alone is defined in the .sdf (so unsprung mass) and the sprung mass is calculated as .cdf mass - .sdf mass or total mass - unsprung mass?

I think it is fair to say that it could be interesting to create an experiment designed to show the impact of wheel mass on the tire load, given a fairly load sensitive control tyre?

Looking at offom's post with pictures, I'm noticing that the radius of the traction circles seem to grow when looking at the pictures from left to right, and the 'current grip in use' percentage seems to shrink from left to right?

It would probably be best to conduct this test on a fairly straight and level stretch of road, too.

[offom]

Ermo, I will check that again at a level, nice road. But I think something is different in this game from the earlier usage of this physics engine.

Edit: "Load transfer in steer-lock" behaviour is same with LFS.

[brrupsz]

Originally Posted by offom

Also, there is something wrong with load transfer. When standing still, turning the wheel tilts the car (due to caster geometry, thats ok), lets say to the left side (fronts are directed to the right), front "right" and rear "left" tire (why opposite ?) loads up, the counter wheels lighten (by a huge amount like x0.8 times wheel load).

Like:
(kg)
Default,
Fronts: 250 250
Rears: 275 275
Full right lock,
Fronts: 100 400
Rears: 425 125

You say it's tilted due to caster geometry and I think the cross-load is a result of that tilt.
Full right lock:
front left has a negative camber; front right has a positive camber

My guess is this load difference is a result of ackerman, front right tyre is turned more than front left. So front right has more camber angle than front left, in result "pushing" the front right of the car "up". Rear left is loaded as a result.

I'm not good with suspensions however. I remember that rF has a problem with lack of grip at slow speeds due to "divide by zero" (like every sim based on math, duh!). My guess is that checking loads while standing still might not be precise but I'm not sure.

[offom]

Mainly caster causes self-centering effect because it makes trail length longer (added with pneumatic trail). Also, it pushes the inner front wheel under and inner side up. Yes, you are right, inner wheel turns more due to the ackermann geometry and that results in slight increase of center of gravity (another reason for self-aligning) -(btw, caster is for high speed stability, kingpin is for overall and cornering stability).

I think games "need to" increase this inner wheel's load portion "to be able to" tilt the car opposite direction. After that, rear tires just follow the situation.
I find the telemetry view a bit messed up anyway because as an example, cars are turning at 1.5 - 2 G's while sliding all over the road.

Thanks for your precious answer&comment Brrupsz.

If you have some questions about suspension, feel free to ask please.

[brrupsz]

There are ways to get rid of the sliding at high G's. It's not a problem with telemetry but with tyres values. I will try to write something about it tomorrow.

[offom]

Originally Posted by brrupsz

There are ways to get rid of the sliding at high G's. It's not a problem with telemetry but with tyres values. I will try to write something about it tomorrow.

Tires should not be able to carry that forces. Expected values are around 1.0 G I guess, and after that point forcing the wheels should result in loss in G force...

Edit: G force telemetry in S2U reads DryLatLong grip coefficients. As they are greater than 1.0 in all the cases, G force result is boosted up.

[brrupsz]

Basically the best thing to do with Shift tyres (.. and hardest at the same time) is to find the perfect balance between grip, stiffness and mass of the car (lighter cars do need less stiff tyres - otherwise they become very tricky to drive).

I think you were right that drop off part of slip curves is not used. I tested this few times and from what I remember slip curve with drastic dropoff after the peak, which in rfactor caused sudden loss of grip in shift caused nothing.

Load sensitivity curve can be used to smooth the load transition of stiff tyres (so loss of grip won't be as sudden, because it will start sooner). However as you've said - it can be turned off (flat curve) and tyres still can be smooth. It's all about balance. Flat line value for testing: LoadSens=(-5e-7, 1, 40000.0)

Lat/LongPeak values are not important at all. At least I didn't felt any difference from changing it (Sakurablack discovered the same thing a while ago). My guess is - if slip curves are not used and stiffness of the tyre is described by stiffness values then LLPeaks are just a place holders for AI. They do crash a lot when these values are zeroed - to be specific the first two - (0.2, 0.2.

[offom]

Also I'm suspecting if curve's shape is not being used at all. Once upon a time I had replaced it with a cubic spline to experiment and it had blown the tires. But, maybe straight line (as in default) is the only way calculations are made. Maybe.

Good to watch:
http://www.youtube.com/watch?v=ifNVGhG2jOU