Designing Roll Moment behavior with increasing body roll.

[ksw100]

How should the roll moment (distance between the roll center and the center of gravity) behave with increasing chassis roll? i.e. Do you want the roll moment to increase, remain somewhat constant, or decrease as chassis roll increases?
Say this is for a road sports car like a McLaren, Corvette, Ferrari etc.

 

[BrianPetersen]

Given that the center of gravity is staying put in the vehicle, the question then becomes whether you want the force-based roll center to move around, and that's going to depend on the design of the suspension.

The only way it isn't going to move around some, is if the design of the suspension mechanically forbids it. Pure trailing arms. Semi-trailing arms. Swing axles (you're going to have other problems). Beam axle with panhard rod (you're going to have other problems). None of those suspension designs are appropriate for the type of vehicle that you are talking about.

For the type of vehicle that you are talking about, upper-and-lower-wishbone and more-or-less equivalent multi-link layouts are standard fare.

All of these are going to have the roll centers move around some due to the angles of the links changing with suspension movement. But, unless you've done something inappropriate in the design (links too short relative to expected suspension movement, etc) it's going to be neither here nor there compared to what else is going on.

BMW and Porsche commonly use MacPherson front suspension, and those have instant-centers that move all over the place, especially in rebound (extension). But ... what it does in extension is largely irrelevant because that's the inside front wheel, which is carrying scarcely any load during cornering, so who cares?

You don't want instant-centers high off the ground, to avoid having the car trip over its own swing axles. Hark, the Herald axles swing! Ralph Nader, you called? If the instant-centers are appropriately low, but not necessarily at ground level, it follows that they're also not going to be moving around much under the circumstances that matter. (Who cares what the inside front does?)

Focus on other stuff. Get the camber gain in compression right but balance that against caster change with suspension movement. Get the caster right. Get the roll-steer and bump-steer right. Get it so that there are not excessive changes in camber, caster, etc with suspension movement. If you do all that then the instant-centers will also not be doing anything extreme in the situations where it matters.

 

[BrianPetersen]

By the way, if you study Formula 1 front suspension (visible control arms!) don't be surprised if you conclude that there is almost zero camber change with suspension movement and very little antidive.

There is also almost no suspension movement because the springing and damping are so stiff. You can get away with this when operating on smooth tarmac surfaces. What little movement there is, of course is managed very expensively, because they can.

The cars that you mention, are also intended for being driven on tarmac, and are not expected to deliver a luxury-car ride. Suspension designs get a whole lot more complicated if compliance over rough surfaces is required and ride quality matters. You can make any (bad) suspension design work, if you don't let it move.

 

[GregLocock]

Shall we agree to define what OP means by roll centre? There are two definitions, the ic based one (geometric), and the force based one. The former was easier to work on a drafting board, the latter is (in my opinion) more useful for dynamics. It also accounts for compliances more easily than the ic method.

Unfortunately the definition matters because during a corner they move in different ways. That is, if you freeze the attitude of the car during a corner and work out your roll centres, you can plot the migration of both geometric and force based roll centres.

Fortunately it doesn't matter much because the effect of RC migration is small if you have a sensible setup. Any targets I have seen for vertical migration during roll seem to be based more on rule of thumb than any calculation. I have even see a target for lateral migration of the roll centre, at least one SAAB engineer thought it was important. RCVD mentions off centre RC in passing but spends no time on it.

So, what use it? Well, it doesn't migrate just in roll, it also migrates in bounce, so we can tune the balance of the car in a load (cargo) dependent fashion. So if you've got a ton of wood in the back we can dial in some understeer.


Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[ksw100]

In terms of definitions, I'm using these depicted:

http://www.researchgate.net/figure/...-and-moment-arm-Smith-1978-p30_fig2_329196488

Where the roll moment is depicted in the above picture by the distance between the ROLL CENTER and the CENTER OF GRAVITY.
Thank you!

 

[GregLocock]

OK GRCH it is. So the higher your RCH the more lateral contact patch forces are transmitted via suspension arms, and the more weight transfer you'll get at that end, and the more responsive your vehicle will be to steering input transients. With less roll you'll get less roll steer. You need to tie that up with load sensitivity of your tires and your understeer vs latacc targets (the only time I've worried about that is trailer tow). It's a typical vehicle dynamics muddle, you change one thing and everything gets affected. Of course changing the UCA hp (the easy way to change RCH) affects your kinematic roll steer anyway, and probably your roll camber .

That is why I build DOEs with all the hp in and all the responses, trying to do this stuff by hand drives me potty. Frankly I suspect that is why RCH migration vs latacc is not high on anyone's priorities, except yours!

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[ksw100]

deleted

 

[cibachrome]

While it sounds like you are only considering the total amount of roll per g vs. the sum of the front plus rear roll stiffness contributors, more important is how the front and rear roll moments get generated and whether nonlinearities in one or both can make the car's behavior 'better' (whatever that means to you as well). This implies a desire for a dynamic TLLTD which migrates front to rear or otherwise, or not at all, and also depends on knowledge of each axel's tire properties, especially the load transfer sensitivity. And don'r forget the roll moment due to caster, another one of those pesky items which consumes only a few book mentions, but is rich in the use of handwaving, smoke, mirrors, and LEGOs.