• 0 Vote(s) - 0 Average
• 1
• 2
• 3
• 4
• 5
Wrong calculation of rolling resistance
07-29-2011, 04:07 PM,
 NaN Posting Freak Posts: 2,024 Threads: 120 Joined: Jan 2010

So we are using '94 formula and have a bug in the stiffness factor calculation, right?
07-29-2011, 04:07 PM,
 LBodnar Junior Member Posts: 37 Threads: 10 Joined: Jul 2011

false alarm! :o

Fz came from "D" and cancelled BCD's one out.
07-29-2011, 04:10 PM,
 LBodnar Junior Member Posts: 37 Threads: 10 Joined: Jul 2011

NaN Wrote:So we are using '94 formula and have a bug in the stiffness factor calculation, right?

No, forget '94. The formulas in the vdrift code are '96 and at least "touring" is '96 data too (but for racing slicks.)

I have gone through the Fx Fy Mz code and checked it all.
07-29-2011, 04:11 PM,
 NaN Posting Freak Posts: 2,024 Threads: 120 Joined: Jan 2010

It is correct, the Fz has been optimized out(Fz/Fz).
07-29-2011, 04:15 PM,
 NaN Posting Freak Posts: 2,024 Threads: 120 Joined: Jan 2010

This means touring should be slick. I wonder what carparts/tire/slick is?
07-29-2011, 04:19 PM,
 NaN Posting Freak Posts: 2,024 Threads: 120 Joined: Jan 2010

I will add rolling friction factor to the horizontal shift calculation:
btScalar Sh = b[9] * Fz + b[10];
Sh = Sh * rolling_coeff //scale by track surface rolling friction factor
07-29-2011, 04:28 PM,
 LBodnar Junior Member Posts: 37 Threads: 10 Joined: Jul 2011

NaN Wrote:I will add rolling friction factor to the horizontal shift calculation:
btScalar Sh = b[9] * Fz + b[10];
Sh = Sh * rolling_coeff //scale by track surface rolling friction factor

This does not seem very intuitive because Sh is an offset to slip ratio.
What value and meaning rolling_coeff has?
07-29-2011, 04:33 PM,
 NaN Posting Freak Posts: 2,024 Threads: 120 Joined: Jan 2010

It scales rolling friction depending on the surface(surface deformation).
07-29-2011, 04:46 PM,
 LBodnar Junior Member Posts: 37 Threads: 10 Joined: Jul 2011

NaN Wrote:It scales rolling friction depending on the surface(surface deformation).
Does it scale it up? Tarmac is the hard reference surface I assume and other surfaces all have higher rolling resistance so it should scale it up?

Tarmac rolling resistance comes from tyre rubber energy dissipation.

Soft ground rolling resistance comes from non-elastic ground deformation so it is not very tyre dependent apart from its geometry and normal force.

Have a look at p.41 Fig 2.7a http://books.google.com/books?id=yAo-6yQ...&q&f=false

Also it acts along the vector of tyre movement so should have both Fx and Fy components.
07-29-2011, 05:07 PM,
 NaN Posting Freak Posts: 2,024 Threads: 120 Joined: Jan 2010

The surfaces have two components rolling friction factor (values 1-40) and rolling drag (0 - 200) used to simulate different surfaces.

My interpretation has been: There is deformation(surface+tire). Energy is dissipated. The wheel has to come up for it.

I didn't write the code, so I might be wrong. Not sure how much difference it will make to remove the rolling friction factor. The important thing is to maintain the effect.
07-29-2011, 05:41 PM,
 LBodnar Junior Member Posts: 37 Threads: 10 Joined: Jul 2011

NaN Wrote:The surfaces have two components rolling friction factor (values 1-40) and rolling drag (0 - 200) used to simulate different surfaces.

My interpretation has been: There is deformation(surface+tire). Energy is dissipated. The wheel has to come up for it.

I didn't write the code, so I might be wrong. Not sure how much difference it will make to remove the rolling friction factor. The important thing is to maintain the effect.

They are pretty independent:
high friction, low drag: tarmac
low friction, low drag: ice
low friction, high drag: sand

I agree, but personally I would not put compliant surface rolling resistance inside Pacejka formulas. It just doesn't belong there.
07-29-2011, 05:49 PM,
 NaN Posting Freak Posts: 2,024 Threads: 120 Joined: Jan 2010

This is what we've got:
Code:
```[surface-00] Type = asphalt BumpWaveLength = 2.0 BumpAmplitude = 0 FrictionNonTread = 1.0 FrictionTread = 0.9 RollResistanceCoefficient = 1.0 RollingDrag = 0.0```

RollingDrag -> drag

I need to test how RollResistanceCoefficient affects the cars atm. Maybe it can be removed with an adjusted rolling drag.
07-29-2011, 05:54 PM,
 LBodnar Junior Member Posts: 37 Threads: 10 Joined: Jul 2011

I have just tried Genta saloon tyres on XS with caster fix on G27 and it feels so great!

Also tried Genta sports tyres on Z06 and they are very "grippy" - literally glued to the road.
07-29-2011, 07:46 PM,

I *think* those coefficients are set to zero like that because I was seeing tire forces generated without any slip, so the car would slowly start rolling on its own. That was a long time ago, though.
07-30-2011, 04:03 AM,
 LBodnar Junior Member Posts: 37 Threads: 10 Joined: Jul 2011

joevenzon Wrote:I *think* those coefficients are set to zero like that because I was seeing tire forces generated without any slip, so the car would slowly start rolling on its own. That was a long time ago, though.

They should be treated like any other friction force:

1) Fx only appear when the tyre is rolling
2) if tyre is stationary it only starts rolling when applied driving torque exceeds Fx(0) * Re

Stationary and low speed tyre physics is not directly covered by Fx Fy Mz Pacejka formulas because the thread width cannot be considered zero anymore. There are local contact patch stresses that need some running length to dissipate and this translates into non-zero time at low speeds. E.g. heavy steering in a stationary vehicle creates a twist in the contact patch and generates counteracting Mz. When you start rolling slowly it will exponentially dissipate over a few tyre turns. I will come back to this later though.

Forum Jump:

Users browsing this thread: 1 Guest(s)