rec.autos.simulators

How does camber thrust work? If I have a real life car with negative camber
on all 4 wheels ....... /----x----\ ....... and it's parked on a flat
surface, then it appears to me that the tyres shouldn't generate any lateral
force at all.

I've seen some posts where people talk about the "camber thrust on the left
tyres generating inward-pointing lateral forces that exactly counter the
lateral forces generated by the right hand tyres" but that doesn't make
sense.... if that *was* correct then a car with super grippy tyres on the
right and with super-slippy tyres on the left (eg. if the left tyres were
resting on a patch of ice on the road) would crab sideways when parked.....

So assuming there's no camber thrust when parked, is there also no camber
thrust when driving steadily in a straight line?

And as for cornering..... if I think about a car cornering steadily, then
the car is "leaning" on the outside tyres, which are generating a lateral
force to turn the car (yes, the inside wheels are too, but I'm just
considering the outsides at the moment). In this case I figure that camber
thrust will be generating centripetal lateral force, which "helps the car
turn". It's as if the camber thrust only appears when needed...... if you
lean on the outside tyres then they will push back at you, but if you don't
lean on them, they won't push (kinda like brakes..... max brake pedal when
parked on the flat gives zero braking torque, but max brake pedal at 100 mph
generates lots of brake torque...... so the brakes only generate as much
torque as is "needed").

Racer has a quirk whereby stationary cars with high negative camber can
scoot sideways across the tarmac..... it's as if the camber thrust is being
generated all the time, and not only when needed. (this would be like your
brakes *always* generating a braking torque so your car would reverse when
you brake when parked).

Can someone help? I'm not sure how much I'm getting right and how much is
complete rubbish :0)

Cheers
Jim

Hmm... I'd guess it's possibly load sensitive (a lightly-loaded tyre
will have a smaller contact patch, so less room for conicity and
slip angle to develop?) which again would seem to fit the real-world
evidence.

Must admit it's not something I've modelled, though, and my camber
adjustment on my model racers is limited to keeping the tyres
wearing flat :-)

Jonny

Sounds good to me. If that's the case then, we'd not get any camber thrust
when the car is parked (because the cone isn't rolling), so I can only
assume the magnitude of the thrust will increase as speed rises. Pacejka's
formulae don't accept speed as an input, so I guess this would have to be
something that Ruud would need to add in alongside the Pacejka....?

I also assume that the magnitude of the camber thrust force will increase
with tyre load, and hence when cornering the outside tyres will generate
more thrust than the inside tyres and so there'll be a net increase in grip
due to camber thrust? I wonder how camber thrust varies with camber angle.

Now take this and put it on a stationary car with assymetrical tires as
you described.  The tire carcasses will flex as necessary until they
reach an equilibrium, and then nothing further happens.

So why do racers use negative camber?  Imagine a tire with zero camber
going through a corner.  You put a lateral force on it, and the carcass
bends inward.  The outside sidewall gets pulled further under the wheel,
and the inside sidewall gets pulled inward so that it is no longer under
the wheel.  As a result the outside of the contact patch ends up with a
lot of pressure on it, and the inside ends up with very little.  This
results in less effective use of the contact patch and a drop in lateral
force capability.  Adding negative camber to the tire pulls the contact
patch back under the wheel and evens out the pressure, potentially
regaining the full use of the contact patch.

It just so happens that the force difference experienced by doing this
is close enough, within the large range of experimental error inherent
in tire testing, that it appears similar to the camber thrust
measurements made on a test stand.  Therefore people assume that they
are getting camber thrust and don't bother to look any deeper.  But
that's not really what it is.  And that's probably why the camber thrust
models used in most sims aren't very convincing.  I think you would
pretty much need an FEA model of the tire to get it right.

I thought motorcycles corner because when you try to "tip" the
cycle onto its side while going straight, you exert a torque /
angular momentum force (damn, what's the english term for that?
"Drehmoment" in German) on the rotating wheels.

As the resulting force is actually the cross product of the two
force vectors the resulting force acting on the cycle makes it
turn and not just fall over.

All of this might be massively wrong, but that's the process
generating the "original" force.

Cheers,

rehoov.

--
mail replies to Uwe at schuerkamp dot de ( yahoo address is spambox)
Uwe Schuerkamp //////////////////////////// http://www.schuerkamp.de/
Herford, Germany \\\\\\\\\\\\\\\\\\\\\\\\\\ (52.0N/8.5E)
GPG Fingerprint:  2E 13 20 22 9A 3F 63 7F  67 6F E9 B1 A8 36 A4 61

If a "perfect" tire is forced to roll straight along a road with no steer
angle (and thus no slip angle), then there is no side force.  If the wheel
is then leaned to one side (inclination in tire axis system) a lateral
force is produced.  Or, if the path is not constrained to be straight, then
the tire will roll off at an angle.  If the tire is then halted, the
lateral force will remain unless the tire if free to move sideways
a small amount (relax the stress in sidewalls).

See tire chapter in RCVD for lots more detail.  Also, I think there is
more in Tony Foale's new book on motorcycle dynamics (recommended!)

-- Doug Milliken (on holiday with poor email connections)
   www.millikenresearch.com

I regard camber thrust as follows: say you have a car with cambered
wheels only on the right side. It's driving forward.
Now it tends to turn right (or left, depending on the direction of
camber). Imagine a kart that just doesn't want to turn in. This often
is just a result of one side being so much off-camber. If it doesn't
want to turn left, it probably wants to turn right twice as good.

So you're driving forward, the cambered wheels pull the car to a side
(just like indeed a conically shaped object would).
That force is the camber thrust.

In Pacejka, it's added into the calculations next to pure slip angle.
Problem with Pacejka that it indeed doesn't have speed as an input,
which is just plain wrong at low speed. Therefore, at low speed you
can use relaxation lengths (which are physically very correct, as they
give you a spring effect).
In Pacejka, with camber at speed=0, you STILL get a force, much like
there is a Sv and Sh variable which directly shifts (hence the 'S')
the graph. So at 0 slip angle you'd still get a force (normally used
for conical effects and the other one, which I forgot, lol).
At low speed, what you should probably do is cut out camber effects
and any Sv/Sh movements. Well, what *I* should do. ;-)

Racer's Pacejka graphs seemingly currently all go through the origin
nicely, otherwise (if Sh!=0 or Sv!=0) you'd have seen this a long time
ago (Ashley mentioned this a year or so ago while playing with Pacejka
curves).

Now this doesn't mean the *force* should end up at 0. As a car
standing sideways on a hill must still have lateral force buildup,
otherwise it would slide slowly downwards, since the gravity force
pulls the car down the slope, and the lateral tire forces must fight
that.
But that should be doable with just the relaxation slip angle, which
should reach an equilibrium.

You could see Pacejka's lateral force as:
Fy=f(slipAngle+slipAngleEffective)

where slipAngleEffective=camber*magicFactor to see the effect of
camber as some sort of slip angle (though it isn't).
But slipAngle is relaxed (in beta6) while camber isn't. To avoid
having to do both (and get a mess) it's easier to throw out camber at
low speed, where you'll hardly notice it anyway). And keep the car
sideways on the hill using a tricky slipAngle equilibrium.

You'll quickly be able to test if that works in beta6. :)

Hope this clarifies instead of mystifies. ;-)

Ruud van Gaal
Free car sim: http://www.racer.nl/
Pencil art  : http://www.marketgraph.nl/gallery/

Thanks for any suggestions,
jonas

Regards,
Hal Raymond
DaimlerChrysler Corporation

<still waiting for somebody to try an FEA tire model in a sim>

So stiffen the front ARB and you get the same overall lateral weight
transfer as before but you get more weight on the outside front and less on
the outside rear (and less on the inside front). [1]

Tyre load sensitivity does the rest.... that's the thing whereby doubling
the load on a tyre doesn't double the max force that the tyre can generate -
instead, the max force increases by less than a factor of 2. So to optimise
grip you want equal weights per wheel. So stiffer ARBs mean more weight
transfer at the front and so the load sensitivity reduces relative front
grip. And you get understeer.

Don't take this as gospel because I'm not 100% sure about this.

Jim

[1] According to GPL Foolishness the weight on the inside rear is
unchanged - can anyone tell me why??

Ooh yeah the gyroscopic effect..... precession is it?