how does weight transfer actually work?

[Tim.Wright]

I've been thinking about it some more and I think i understand it....so technically it would be wrong to say weight transfer only comes from cg height, track widths, gforce, and overall weight. It is a function of all the other things, which track widths and cg height are part of. I can picture it as if the chassis is just a beam with a lever sticking straight up to the CG height, and there are front and rear solid axles attached to it by coiled springs (so if the chassis rolls from a force on the lever, the spring will twist the middle of the axle to push one end down and the other end up). a stiffer spring (aka roll stiffness?) would add resistance, so the end of the axle will push down harder, meaning more weight transfer. if the front has a softer spring than the rear, the rear would then have to take more of the weight transfer to make the moment sum around the central beam be 0. Does this sound like a good model? I wish i could draw it to be more clear...It is similar to something i read earlier, so i think i'm on the right track.

No, I believe you're wrong. There is no "resistance" added by the springs. At all.

All that happens is that when the springs give way (and they do, no matter how stiff they are, they are not infinitely hard), then the body moves and the CG moves. Then the balance changes, thus the reactions on each wheel. End of story.

I thought his explanation was not bad and in the correct direction. The resistence that he speaks of is analog to a mechanical "impedance".

The model he described is also fundamentally correct. A longitudinal beam the length of the wheelbase (linking front and rear "roll centres") with a vertical beam connecting to the CG and two rigidly mounted lateral beams at the front and rear axle can constitute a "chassis". Connecting a spring from the ends of the lateral beams to the ground can represent the suspension. This little model will predict your roll rate with about 80% accuracy (in my experience).

The best explanation of the traditional 1d.o.f roll model that I have heard of was from a Rouelle seminar 10 years ago with the aid of a little experiment. You need:
3 people
1 set of corner scales (or 4 bathroom scales)
1 broomstick

• Stand 2 of the people facing each other and spaced apart by the length of the broomstick. (wheelbase)
• Give each of them one end of the broom stick (roll axis)
• Put the corner scales under their feet (contact patches)
• Now these 2 people and the broomstick represent the roll model. One person is the front suspension, the other is the rear. The broomstick is the chassis, and the point/height at which it is held at each end is the roll centre of the suspension. The corner scales are measuring the load transfer.
• A 3rd person stands near the middle of the broomstick and applies a lateral force (g-force) AND and twisting force (roll torque) on the broomstick.
• The 2 people on the corner scales need to restrain the broomstick from moving laterally or twisting by opposing the forces applied by the 3rd person. This mimics the reaction of the springs and bars.
• Watch the measurement on the corner scales changing as this is done

Then try some variations such as have the front person hold the broomstick lightly and the rear person hold it strongly (representing a forward biased roll distribution). Change the height at which the broomstick is held to simulate different roll centre heights. Change the spacing of the suspension people's "feet" to simulate different track change.

Even if you don't do the experiment, its a pretty easy one to visualise and will give you a basic understand of how a suspension works in roll.

Like I said, its not perfectly correct, but it will get you within 20%. Maybe even 15% if you include tyre stiffness in your overall suspension stiffness.

[olefud]

[
No, I believe you're wrong. There is no "resistance" added by the springs. At all.

Technically yes. Roll resistance is ultimately provided by the tire contact patches. But the springs are part of the structure between the roll center and contact patches and allocate the “roll resistance” between the axles. They provide variable intermediate axle roll resistance along with the ARB. But the total roll resistance and weight transfer don’t change but are just reallocated between the axles.

[Greg Locock]

Here's the formula for simplified weight transfer

http://www.mediafire.com/view/t2m2qi33c ... ansfer.PNG

deltaFz is the weight transfer, rho is the radus v is the velocity, 1 means front axle 2 means rear axle

hprime is the distance between the roll axis and the cg vertically

subscript 2 means rear axle 1 means front axle. b means body, Gx means mx*g

C is the roll stiffness of that axle, springs + bar s is track, p is rch

[thepowerofnone]

Unmentioned so far, but the next addition to complexity here would be torsional stiffness of the chassis.

I don't mention this to nit-pick, more to state an obvious truth which hasn't yet been explicitly mentioned: every single load bearing member (which is everything if the vehicle is in a cornering state since everything has some inertia) could be included in this calculation, if the level of complexity were increased to an absurd degree; the realistic answer to your question is that you got the big three (lateral force, height of CoG and track) and increasing to Greg's level of complexity is probably plenty. Going past the addition of chassis torsional rigidity and the influence of engine torque, there aren't many other significant variables, except for asymmetric aerodynamic loads caused by a) ride height asymmetry and b) effectively a cross wind caused by the cornering if those are significant on your vehicle.

[RideRate]

This is a great question. And one thing that drives me mad. Every mechanical engineer should learn and understand the fundamentals of this problem early in their education, but alas many graduate and still don't get it. I see it all the time.

One of the issues comes with definitions and assumptions of what one means by just saying 'weight transfer'. I prefer 'load transfer' as the term, but we all seem to mean the same thing. I hear engineers all the time see a car with a lot of roll and say, "look at all that weight transfer." Which to me sounds ridiculous because by my definitions the magnitudes of body roll and load transfer are not proportional and in fact have nearly nothing to do with one another.

Of course all the time I'll hear, "soften/stiffen the front/rear springs to increase/decrease weight transfer." Again, by my definitions that makes no sense and one has next to nothing to do with the other.

In a steady state corner, load transfer (meaning the total load that is removed from both inside tires and now carried by both outside tires) is only a function of trackwidth, cg height, and cornering force. Period end of story. Cornering force can be found from lateral accel and vehicle mass, so saying I need cornering force is analogous to saying I need to know the mass.

deltaWeight = (corneringForce * CGheight) / trackwidth

deltaWeight is the total change in load of the sum of the inside tires or the sum of the outside tires.

So for a left turn: (load removed from LF + load removed from LR) = (corneringForce * CGheight)/trackwidth = (load added to RF + load added to RR)

So load transfer is that which is moved in total laterally across the car and only depends on cg height and trackwidth.

The amount removed from each individual corner is dependent on stiffnesses and becomes a deforms problem. This is what's known as your balance, TLLTD, roll stiffness distribution, etc. Also the individual loads may depend on counteraction of driveline torque and other stuff that cross loads the car.

But key point is, even with the variability in individual loads, the sum of the total change on the left side and the total sum of the change on the right side will be known and will be a function of cg height and trackwidth.

Once you understand this you can start to imagine how a car with equal front and rear roll stiffnesses does not torsionally load the chassis. And thus realize that the chassis torsional stiffness performance in roll mainly needs to be designed around just the difference in front and rear roll rates.

So the only way roll displacement would ever really matter is if in the process of cornering the body roll significantly moves the cg up or down. If this happens you just adjust load transfer by the new cg height. And then fun things begin to emerge in how this manifests.