Roll Steer Plot 2

[SusTestEng]

Can someone post, or point me in the direction of a "proper" roll steer plot. What I am looking for is an example of ideal conditions for a REAR multilink (or other similar) suspension design. The project I am currently working on has some issues with roll oversteer and I what some fuel to back up what I am feeling in the car. Basically I am looking for a toe patern plot in relationship with body/suspension roll.

 

[NormPeterson]

Hi Greg,

Anyone who believes that the average owner will modify their pressures to suit hasn't met many average owners.

Agreed. It's doubtful that many even realize that some cars already have more than one set of mfr-recommended pressures posted, depending on the loading (at least two of my cars' owner's manuals and/or pressure stickers contain such info). And I'd expect that most of the folks that do are involved in some form of motorsport competition or high performance on-track lapping.

I can also picture that various loadings of a transient or quasi steady-state nature such as that due to lateral and longitudinal accelerations could affect the loading used as a design basis, if only as a simplified concession to fatigue.

Norm

 

[Z8]

Is it possible for you to tune the lateral force steer to counter the roll oversteer?

Thanks.

 

[GregLocock]

Yes, but it is going to lead to the possibility in real events where the two will be fighting each other. This is less important on the race track, where your roll and lat acc should be in phase.


I just had a long conversation with someone who thinks about this stuff long and hard and he said:

1) Roll oversteer of 0.1 /is/ on the high side (sorry Norm, you were right), and "proved" it by pulling the plot of the best handling car (tm) which was more like 0.03. Fair enough, but I still have plots showing 0.1 for good handling cars.

2) People do strange things to their roll and bump steer plots, depending on which they think is most important. This may explain why luxury sports have funny roll steer curves - they were trying to optimise the bump steer response.

3) We managed to persuade ourselves that compliance oversteer (at the back) should preferably be as linear as possible, and MUST account for variations in pneumatic trail. This doesn't apply to cheaper suspensions that may need a bit of help.



Cheers

Greg Locock

 

[Andrew1966]

This is a very interesting thread. I realise that in this instance it is probably too difficult to justify a change in the rear roll centre height, however I wanted to propose REDUCING the rear roll centre height rather than increasing it.

Several things happen when you adjust the roll centre height, including effecting the load transfer across the axle. Reducing the roll centre height should reduce the load transfer across the axle which will increase the effective cornering stiffness of the axle and inturn reduce the slip angle and add a little bit more understeer. It will also mean that a higher percentage of the load transfer goes through the springs, shocks and ARB hence the vehicle should hopefully become a little more responsive to tuning of these components. The total roll moment will have been increased so you will need to re-check your roll stiffness to ensure the roll angle is still within your target range.

Out of interest what is the roll axis of the vehicle? I have done some tests in a driving simulator where I changed the roll axis by changing the rear roll centre height. The base condition had a 5degree inclination. I dropped the rear roll centre to get a parallel roll axis and the vehicle felt very 'crisp', however when I increased the roll axis to 10degree and again drove through the lane change, the vehicle went into a series of large 'fish tails' as the rear tried to break away.

I hope this helps, yes I realise that there are many coupled effects as a result of changing the roll centres etc...

Andrew

 

[SusTestEng]

This is a fairly old topic, but I thought I would wrap it up a bit and give some insite to others what I found.

This is still an issue with the vehicle, but I think I masked it pretty well. What has been done since this is started:

Redistributed the roll stiffness: I reduced the front spring rate by more than I reduced the rear, and I increased the rear ARB. The vehicle is now about 57.1% Front roll stiffness dist, with a load transfer rate ratio of about .90. This is a sporty FF vehicle. I wish the load transfer rate ratio was less, but basically the gyometry does not allow it.

Increased the front compliant understeer: after reviewing the front compliance steer, I noticed it was very stiff. I reduced the front crossmember mounting bushing stiffness. I wanted to just change the rear mounting stiffness(like the VW golf), but two different bushing rates was not accepted by the chassis dept. And this way we can commonize with another vehicle.

So basically I have reduced the front suspension response and let the rear do more of the work. This also has reduced the yaw rate acceleration to become more linear, which in turn masks the sudden change in roll steer. Having more rear roll stiffness has also reduced the actual amount of roll, so it limits the amount of roll steer.

Biggest problem NOW: A 4-5Hz bounce in the rear. It makes a very uncomfortable jiggle in your stomach. I am assuming the damping force around 0.1-0.2 m/s is too much and not allowing the rear suspension to stroke. The problem is we can't use a long jounce bumper, so I need damping force to creat a reaction for the steering feel. I need to reduce that area in the rear and in the front to balance it back out to regain the steering response.

 

[SusTestEng]

I also increased the bushing stiffness of the inner/rearward upper arm bushing of the rear suspension. This helped a bit to reduce the knucle rotation of the multilink system.

Anyone who reads this: NEVER offset the spring/damper knuckle fitting point from the centerline of the rear knuckle if the vehicle is not AWD. And if possible, NEVER offset the rear ARB reaction fitting point from the centerline of the rear knuckle. These two issues have created so MANY headaches this past six months. On a rear multi-link suspension, fix the knuckle to the centerline of the knuckle and mount the ARB to the shock or similar centered position.

 

[NormPeterson]

I think I follow most of what you're saying.

I wonder if maybe you also want the lines of actions of the springs/bars/shocks to be consistent with the intended path of the loaded ball joint, at least as seen in side view. That's one of the immediately noticeable things about the C5's rear suspension when you get to see one up on a lift, BTW.

Regarding the "tinkering" with compliances - what about local chassis structural compliances? It's easy to assume that the chassis is rigid, but it's really just another spring in series, and will deflect under load. Maybe you could get some help from the structural guys?

Norm

 

[GregLocock]

"Regarding the "tinkering" with compliances - what about local chassis structural compliances? It's easy to assume that the chassis is rigid, but it's really just another spring in series, and will deflect under load. Maybe you could get some help from the structural guys?"

That is a very good point. My models account for body (and arm) complance in a variety of ways, but as of the next program we'll be switching to a fully compliant model of the body, which is derived from the finite element analysis of the bodyshell. It is not unusual to see mounting points that are only slightly stiffer than the bush that bolts into them. This means that very large changes in the bush's stiffness are needed to get a measurable difference in the car, and that the isolation at that bush is very poor.



Cheers

Greg Locock

 

[2SlowJoe]

Many people assume that for an independent suspension, the plot of toe vs wheel center travel that results from a ride event is the same as the one that results from a roll event, or a single wheel event. This assumption assumes that the left side of the vehicle is independent of the right. This is not true.

Consider the poster’s rear multilink suspension. In a ride event the wheel may toe in with jounce travel, yet in a roll he observed the event the effect as the opposite. As the vehicle rolls, the ARB applies forces to the knuckle that will tend to rotate it and cause a steer effect.

The original poster posed a solution of linking the spring and ARB as close to wheel center as possible to minimize the moments that it produces on the knuckle and therefore minimize the steer effect.

I have a different view on things. With careful thought to the geometry, you can have the stab bar moments on the knuckle help you. If you link the ARB away from wheel center, but in a fashion that causes toe in with jounce (for the rear) you can achieve a toe curve that has little to no toe change in a parallel wheel travel event (good for tire wear) and has the level of toe in you require during a roll event.

In practice you can not really achieve zero steer in ride and your target roll steer in roll for all vehicles because you will have different spring rates and different ARB rates on different vehicle trim levels. For a multilink rear, what you can do is link the stab bar in a manner that toes the wheel in the opposite direction that the offset spring force toes the wheel. This should bring your compliant roll steer number closer to the one predicted kinematically.

 

[2SlowJoe]

Another thing...

If this was a front suspension things get even more complicated since the left side is linked to the right through the steering system compliance. With rack compliance, a net rack force will cause a steer effect.

Anything that could put a moment about the kingpin can cause a net rack force. In a roll event if there is a net rack force, it will cause a steer effect.