roll center

[carenthu]

what are the ways to control roll center migration?

[Greg Locock]

geometry of the hard points. Not, to be honest, that it really seems very important. (Consider the roll centre migration of a watts link live axle. Compare it to a typical double wishbone IRS. Is there any objective difference in their performance on the track?)

[Tim.Wright]

I'm with Greg. I'll add that I've never ever seen either:
A: Any clear definition of what roll centre migration actually is
B: Any objective reasoning as to why the above undefined migration should be minimised

Furthermore when you discover that:
C: The body doesn't roll about the geometric roll axes/centers
D: force/moment resolution about the geometric roll centre is only valid for small lateral accelerations

You will realise that most of what you read about it is just handwaving. There's loads of other more important things to worry about.

Roll centres are basically a measure of the jacking of the left and right wheel. This is worth looking at as vertical jacking of an axle is generally bad from an aero, CG height and a subjective/driveability point of view.

[Greg Locock]

I once had an idle chat with a SAAB engineer who claimed that migrating the roll centres towards the contact patch was a good idea in limit handling. I suppose if you only have drawing board techniques then that might be an approach, but once you've got full vehicle models it sounds like one possibility among many. He never articulated why this was a good idea, I got the impression it was a manager's pet theory.

[Jersey Tom]

The premise of Saab and limit handling is amusing in and of itself.

[Tim.Wright]

That wouldn't happen to have been the famous Magnus Roland of Saab would it?

[thisisaditya]

what are the ways to control roll center migration?

Roll center migration of a car is prominently responsible for controlling the cornering characteristics. Roll center tends to travel about the vertical axis of the vehicle which is nothing but "Roll Center migration". Now how does this roll center travel affect? The answer is when it moves with respect to the center of gravity, greater roll angle results which in turn results in a greater moment arm acting on the vehicle which tend to "topple' the vehicle. Thus the roll center and the CoG must be as close as possible.
Coming to the ways in which this phenomenon can be controlled ; Front and Rear Roll centers depend on respective suspension geometries (Swing Length, Knuckle height,etc) also an increase in track width helps in decreasing roll angle.
Use of stiffer Anti roll bars is another solution.
Controlling the roll center migration is VERY important as understeer and oversteer is also affected by roll center geometry.

[Tim.Wright]

Roll center migration of a car is prominently responsible for controlling the cornering characteristics. Roll center tends to travel about the vertical axis of the vehicle which is nothing but "Roll Center migration". Now how does this roll center travel affect? The answer is when it moves with respect to the center of gravity, greater roll angle results which in turn results in a greater moment arm acting on the vehicle which tend to "topple' the vehicle. Thus the roll center and the CoG must be as close as possible.
Coming to the ways in which this phenomenon can be controlled ; Front and Rear Roll centers depend on respective suspension geometries (Swing Length, Knuckle height,etc) also an increase in track width helps in decreasing roll angle.
Use of stiffer Anti roll bars is another solution.
Controlling the roll center migration is VERY important as understeer and oversteer is also affected by roll center geometry.

The CG to roll axis lever arm doesn't in any way try to topple the vehicle. It merely add more roll angle to the sprung mass. Load transfer at the contact patch is independent of this which is the important thing.

Furthermore, moving the roll axis up to the level of the CG has no logical sense and will cause other problems which will reduce cornering grip.

From what I have seen the roll centre doesn't do much more than what you can't otherwise do with springs and bars. The phase delay between geometric load transfer (i.e. jacking from the roll centre effect) and the elastic load transfer (due to roll angle and springs+bars) is negligible at low frequencies and low at high frequencies. I've seen negligible changes in a response to a steering frequency sweep for roll centers +/-150mm from the ground (via simulation).

Like I said previously - while everyone seem to have strong opinions that the roll centre and its 'migration' is a critical suspension characteristic which needs to be controlled at all costs - noone has ever demonstrated this. In fact I have found just the opposite.

[Greg Locock]

I must admit that when I was a vehicle dynamics newbie I was very impressed by the concept of force based roll centres (FRCs), and less enamoured by geometrical based roll centres (GRCs). There is a paper around somewhere that plots the migration of the two as a car goes round the corner, they move quite differently.

However, when we were actually stuffing around with GRCs we found that they seemed to be telling us useful things about the car, and the changes made went into production (admittedly that is a pretty cheap change, not hard to sell).

But in 16 years I have yet to see a problem with RC migration, which given that I have been directly involved in 3 new suspension layouts and modified tens of others, seems to suggest that either our suspensions already have it under control, deliberately, or that some of the other kinematic targets stop the RCs from migrating, or that the dynamic models flag up problems that might be due to RC migration, and in curing those we suppress RC migration while sorting them out. Or something else.

Here's a longer discussion, unfortunately nobody really got to a conclusion

http://www.eng-tips.com/viewthread.cfm?qid=34231

The FSAE guys also have several threads.

Here's a good thread, pay attention to Ben in particular (others may well be worth reading). http://dsrforum.yuku.com/topic/2802/Rol ... ent?page=1

[thisisaditya]

@Tim.Wright :
I agree the effect of RC travel is yet to be demonstrated. But then atleast in theory, where will all the sprung mass moment that is generated due to large CoG and RC lever arm be transferred if not to contact patches through load transfer? Will it not contribute to (kind of) unpredictable handling?

[Tim.Wright]

The load transfer that you speak of (and in fact any lateral load transfer) will occur regardless of and independent of the CG-RC lever arm. Total load transfer depends only on the track width and CG height.

The roll centre only changes what portion of it acts through the suspension links (no delay) and which portion acts through the springs/bars (small delay).

I've seen nothing yet to suggest that large elastic load transfers (due to a large CG-RC lever arm) cause un-predictable handling.

[thisisaditya]

Oh okay! Thanks for the info

[Per]

The roll centre only changes what portion of it acts through the suspension links (no delay) and which portion acts through the springs/bars (small delay).

What does this delay actually imply? Does it mean lateral acceleration starts more gradually as the springs are being (de)compressed?

Would it be correct to say that the RC-CG lever has no direct impact on mid-corner performance (constant velocity and constant corner radius) but it does play a part in transient conditions (entry and exit)?

[Greg Locock]

Excellent. Those are just the right questions. If you flick through towards the back of Milliken he's got some plots showing the force buildup in various components as the car enters a turn.

If you can imagine a car with a beam axle and watts link at each end, if the cg is the same height as the watts link then when you enter the turn the lateral force at each tire patch is reacted by the watts to the cg with no roll, so the car responds quickly. If we drop the watts links right down then the body has to roll before the force can accelerate the CG (badly put). So the time constant of the transfer function (yaw velocity/SWA) is much longer, and is controlled by the roll inertia and the spring/sta bar roll stiffness. The shocks come into it as well, of course.

In a more likely suspension the same things happen , it is just a bit harder to wrap your head around.

So "Would it be correct to say that the RC-CG lever has no direct impact on mid-corner performance (constant velocity and constant corner radius) but it does play a part in transient conditions (entry and exit)?" Yes.

[bill shoe]

Can I + 1 an entire thread? Learning much.