Roll Center Discussion

[mjalaly]

The point was that to satisfy one thing you have to compromise on others and that is up to the designer to determine.

All the designer can do is ballpark it; the final compromise comes when the end user sets the car up for the track. Hopefully the designer allowed enough adjustments to dial things in.

sure but you can only really adjust a hand full of things. Moving inboard mounting points (i know they make mounts to change anti dive and sqaut though) and changing control arm length are kind of hard to do (you can to some extent but its very minimal). I can only adjust camber, toe, ride height and i can change the distance from the upper and lower ball joints in the positive position on my setup.

[Bobber]

To add a couple of things, per the attached force method, I also can’t see how lateral movement of the roll center affects the lateral forces on the chassis. Vertical forces are assumed to be in equilibrium in this method. Vertical forces are “provided” for by considering the differences in the lateral loads on the outer and inner wheels. Also attached is the unit load proportional method for easily finding and distributing the individual wheel loads into the force diagram to find the theoretically, mathematically true vertical roll center location. My suspension machine pretty much followed this theory. Why does that thing remind me of Barbarella every time I use it?

Is it garage time yet?

[Bobber]

I think I forgot to add this. Is anybody still awake?

[mjalaly]

Im awake!

I spoke to Dennis (drakan / dp1 designer, fsae judge) and he had this to say on the topic...


suspension design is all about compromise   the successful approach starts with determining priorities for your application, and then making compromises based on those priorities.   roll center migration in itself is kind of meaningless since you can't directly tie it to any specific dynamic characteristic.   as a rule of thumb, in a double-wishbone setup, the only way to limit lateral RC migration is to have arms that are equal or nearly equal in length, and make them as long as possible (this is what i do but for different reasons).  the main reason i do it is that it gives a predictable camber control curve and favorable camber in droop as well as in bump.  this makes the unloaded tire do its fair share of the work in corners.

short/long arm suspensions maintain camber in bump but completely trow it away in droop.  this is actually good if you're running a spool diff and huge tires on the back (like can-am and f1 cars in the 60's and 70's), because it allows the car to turn.  in anything other than that scenario, short/long arms are a disadvantage.

hope this helps.

dennis

[Omaha Vette Graveyard]

We're right at the limit of my understanding here, but I think sufficient lateral movement of the geometric roll center would tend push the car up or down (vertical force) lateral load.

When the roll center moves away from center the moment arm gets longer, provided the center of gravity is in the center of the car, which would mean less geometric roll resistance. That center of gravity then tries to roll around a point that is off to the side, which would mean lifting or dropping (vertical) force.

This is interesting, as it would mean that a roll center that moves laterally toward the inside of the turn in roll would actually promote greater roll and a more pronounced squat on the outside tire.

-Graveyard

[horizenjob]

Thanks Bobber for posting your worksheets. I'm still working thru it. Also thanks for the FSAE pointer MO, I need to go read that too.

I'm still looking for the most recent Vsusp work I did for Car9. My suspension has a considerable amount of lateral roll center movement. I did not consider that a priority at the time. Since I did this I asked Rob to add the force lines from the contact patches to the instant centers ( green dashed lines ). I think they are instructive and maybe more helpful than the roll centers. Things like weight jacking are directly evident looking at the green dashed lines, but since the roll center is an average of the two suspension sides, it's not clear to me how you understand this.

I still do not see a direct correlation between lateral roll center movement and changes in angles of the force lines. My front suspension manages to hold the force lines pretty consistently but the rear shows a small increase with roll. That would say the jacking forces are rising while the car is rolling and also the geometric roll resistance. The two suspensions are behaving a bit differently because the roll center is representing an average of the two sides, yet there is not enough weight on the inside tires to matter nearly as much.

What I was trying to do was to make the car tunable by making the rear roll center movement adjustable in it's response to the cars pitch. It could remain nearly neutral or it could move either up or down depending on squat and dive. If it understeered too much on entry or too much on exit - that can be tuned for to at least some degree. The graph is below and it shows how the roll center moves for 3 different length upper rear suspension arms. The upper arm is just a radius rod so a different length is easy. The height of the RC is plotted against bump and droop ( squat or dive ) vs. roll.

Bobber your worksheet is great. To repeat what you are saying, so I understand, you roll the chassis and apply a standard load to the tires and calculate the loads at all the suspension mounting points. Then you can scale those loads according to the weight transfer of the lateral acceleration. And nothing like a lateral location of a roll center shows up in this work...

It's late, I'll look some more tomorrow. Thanks again.

[Sam_68]

roll center migration in itself is kind of meaningless since you can't directly tie it to any specific dynamic characteristic.  

I suggest you quote this bit back to Dennis Palatov with the words:

Weight transfer

As a statement, it is blatantly and utterly wrong.

[Bobber]

"To repeat what you are saying, so I understand, you roll the chassis and apply a standard load to the tires and calculate the loads at all the suspension mounting points. Then you can scale those loads according to the weight transfer of the lateral acceleration. And nothing like a lateral location of a roll center shows up in this work..."

Yes. At any particular roll, I use that lateral force and the resulting fulcrum point to figure the required spring/ARB torsional resistance required to hold that roll. Anything stiffer (softer) will lessen (increase) the roll and I have to look at the diagram for that roll. I have all the diagrams mapped out so I can leaf thru them and have a poor man's VSusp.

There are no resultant EXTERNAL vertical forces involved in the force method so the roll center lateral location doesn't figure into this math. Everyone is invited to challenge the math.

[Sam_68]

I think I forgot to add this. Is anybody still awake?

See, now the two things that should be glaring at everyone with the drawings and info you posted here on force based roll centres are:

1) they are treating each end of the car in isolation and;
2) they make absolutely no reference to spring rates.

As a very simple mind exercise, ask yourself this: will the car be rolling around the force based roll centre shown in my diagram at one end, if the roll centre at the other end is in a different position and/or the springs offered differing degrees of roll resistance.

Cars roll around neither the geometric nor the force based roll centre. But you need the geometric roll centre to accurately calculate the weight transfer, which in combination with the spring rates then allows you to calculate something close to the actual attitude of the chassis as it is subjected to cornering forces.

[a.moore]

Are you talking overall weight transfer or weight transfer distribution (IE what percentage of the total number goes to the front and rear suspension)?

[Omaha Vette Graveyard]

If nothing else, wouldn't a lateral movement of the roll center change the length of the moment arm?

[mjalaly]

I think one of the moderators should change the title of this thread to match the conversation. And why did it take a.moore a week to jump in? Haha

[horizenjob]

I think one of the moderators should change the title of this thread to match the conversation.

It's your thread, if you want to change the title just edit the title in the first post...

[a.moore]

And why did it take a.moore a week to jump in?

The Mini in my garage isn't going to restore itself.

[Bobber]

Some more thoughts on my design methodology
(critiques invited and necessary)

Lateral weight distribution depends only on the vehicle weight, the lateral (centripetal) force, the height of the center of gravity, and the vehicle track. Suspension type has no effect on lateral weight transfer. Suspension type and geometry only controls how the wheel moves relative to the chassis. (Four link) Suspension geometry controls suspension roll stiffness by moving the effective fulcrum about which the individual control arm forces are resolved. The vertical distance of this fulcrum from the location of the lateral load (C.G.) can be thought of as a lever about which lateral forces are applied to the vehicle, but this is only for roll stiffness leverage and does not affect weigh transfer. As such, suspension geometry controls roll stiffness but has no effect on the forces causing the roll and the resulting weight transfer to the wheels. Whether the chassis is stiffly sprung or softly, the lateral weight distribution is the same. Only the body roll and suspension linkage movements are affected by the stiffness.
Now the caveat. This is for a system that I calculated how much the chassis rolls subject to the tire forces I used. These are for an internally stable system. We can now tune it by adding or subtracting the roll driving force from one end to the other. Note that these forces will be external to our closed system so this will cause a lateral weight transfer. This is done by making the suspension at one end more torsionally soft or stiff, driving those forces to the other end. These forces will additionally twist our balanced system and transfer weight laterally, independent of the centripetal weight distribution.

Knowing the roll stiffness of the suspension linkage, in my example case, 1988 lb x 3.36 in / 5 degrees = 1336 in lb per degree, I can then distribute it as I want, front to rear to add or subtract the resulting vertical wheel forces.

Ps: We can also use the unit load method to more easily distribute this load as I did above.