Roll damping distribution

[MalcolmG]

I've spent the last couple of days working out desired damper curves, and I've come across something that I thought might warrant some discussion.

I've been looking at how much damping I want in roll. I've used a pretty accurate and detailed CAD model to find an approximate moment of inertia about the roll axis (yeah I used kinematic roll centres, but let's not get in to that discussion), and with a roll stiffness and baseline damping factor I've come up with a figure for desired damping in roll. So the question now is, how should this be distributed between front and rear axles? I'm not necessarily looking for hard and fast answers, but wanted to hear other's opinions on this.

Typically we design roll stiffness distribution to be adjustable in a broad range, centred about 2-5% further forward than the weight distribution, and last year through pre-competition testing we settled on a number pretty close to this.

It seems like having more damping stiffness at the rear could be beneficial on turn in, especially in a slalom, as once the chassis starts rolling you will get weight transfer occuring faster at the rear, causing a more rapid yaw response to get the car pointed in to the corner.

Thoughts?

[Mark_W]

i think you will find you roll damping arguement to be back to front. check milliken chassis setup section:
- moving cg forward increases turn in response
- increase front roll stiffness increases turn in response

damping is not covered but if you can derive why those two things might aids turn in, then you can easily work out roll damping effects.

[MalcolmG]

I am aware of those theories, and I do agree with them from my limited experience and analysis I've done, however I'm not sure that it's so cut and dry with damping. Damping doesn't have an effect on the stiffness until some rolling motion begins to occur, by which time you've already generated (or have begun generating) slip angles and lateral forces at the tyres, the phase shift you get with damping could effectively mean you get the roll resistance moving backward as the vehicle begins to turn in, which seems to me like it could be beneficial.

[MalcolmG]

Hmm, having said that, I just went through some data from some slalom testing we did earlier in the year, and times were faster when we shifted the damping stiffness toward the front

[Pennyman]

When we were testing, we settled on a similar setup just through driver feedback.

We ended up having to up the bump damping in the back which helped the car rotate and limited squat during acceleration, but we kept very little rebound to prevent the inner rear from lifting (kind of a band-aid fix in my opinion).

Because of known unknowns involved such as driver preference, it's hard to say where a good starting point will be. Really this subject comes down to the stopwatch in my opinion.

Your suggestion of starting with higher roll damping in the rear may work assuming the car doesn't understeer on initial point (affected by your ackermann or static toe-out setting).

Here's my best bit of advise:

Get the car done and test it!!

[exFSAE]

Originally posted by Mark_W:
i think you will find you roll damping arguement to be back to front. check milliken chassis setup section:
- moving cg forward increases turn in response
- increase front roll stiffness increases turn in response

damping is not covered but if you can derive why those two things might aids turn in, then you can easily work out roll damping effects.

I'd think increasing rear roll (damping) stiffness would increase turn-in.

I'd say.. pick a number, pick a shock curve such that you have lots of adjustment up and down.

[MalcolmG]

Well I'm not really asking for help on picking the curves, because my approach is basically what's been suggested, decide on something, make sure it's adjustable and test it.

I was really just interested to hear people thoughts/experience on the subject, as it's something I don't think I've ever heard discussed

[flavorPacket]

Originally posted by MalcolmG:
I've used a pretty accurate and detailed CAD model to find an approximate moment of inertia about the roll axis (yeah I used kinematic roll centres, but let's not get in to that discussion), and with a roll stiffness and baseline damping factor I've come up with a figure for desired damping in roll.

You need to double check how you're getting a roll inertia number. Roll inertia is commonly defined about the principal x axis (which is a real thing you can measure and has meaning), not an imaginary roll axis.

After you get this number, I suggest setting a list of goals for what you want to achieve with damping. Do you want to minimize transients? Do you want to generate a yaw moment to turn your car in? Do you want to control tire load variation? etc Then prioritize these objectives and implement them.

And TEST!

[MalcolmG]

But if you use the principal axis for getting a roll inertia, then it's likely your true roll inertia will be much higher, as you wont be rolling about the principal axis; I would think a kinematic roll axis is still a better approximation than the principal X axis

[flavorPacket]

And what exactly constitutes 'rolling about' something? I can resolve the motion of anything about anything: that's the basis for rigid body dynamics in an inertial frame. You need to use the inertia about the principal axis to actually get the dynamics of your rigid body (the sprung mass). Of course you can resolve that inertia about any point as well, but I personally think using the CG is suitable because it is easiest.

[BillCobb]

Set it to the same ratio as the total lateral load transfer distribution to start your evaluations. Then move it rearward in 2% increments until your driver says stop. The tip-in issues are more of a left to right (jounce to rebound) fraction as you get the Dynamic TLLTD issue resplved. The closer you get to the 'optimum' fraction, the less forgiving the car is for wear, pressure and throttle changes. But it can deliver a fast right off the hauler car. Just a bit knife edged, though. You will need to figure out or measure the total roll peak to steady state ratio, too, in order to get enough damping in the car. Don't assume its critically damped.

[GSpeedR]

Originally posted by MalcolmG:
But if you use the principal axis for getting a roll inertia, then it's likely your true roll inertia will be much higher, as you wont be rolling about the principal axis; I would think a kinematic roll axis is still a better approximation than the principal X axis

You want to use the principle axes; it will be simplest to treat these as body-fixed axes. Once you have these measured/calculated you can transform to any axis you want (or calculate Euler angles).