roll center movement question

[Paul V.]

Alright Suspension gurus,

I'm going to the well until it runs dry, so if you ladies and gentlemen don't mind here is another question.

Do you want the roll center to stay stationary in relation to the chassis or to the ground on jounce?

My reasoning says chassis to keep moments about the car predictable for the driver, but a past suspension designer has said the ground.

Any advice is appreciated.

Paul Vaughan
University of Alabama in Huntsville FSAE

[Denny Trimble]

Well, if the roll centers stay fixed vs. the ground, then as the driver transition from full braking (pitch) to full cornering (level), the roll moments will change. This would be a bad thing for the balance of the car through the turn. RC's should stay static vs. the chassis as much as possible, given the other compromises of the system.

[D J Yates]

I second that.

If the roll center follows the movement of the center of mass then the distance from NRA to the center of mass stays constent and hence, the roll moment stays constent. A ratio of the vertical displacements equal to 1 is ideal.

Any opinions on what is an allowable amount of camber gain in bump? Obviously, less is best. But how much is acceptable?

[Brian Smith]

its a compromise. If you improve camber controll in bump then you worsen camber controll in roll. Look at the track. Are their more performance gains in good cornering or in accel and braking. Its your choice.

[kwancho]

Sorry to pull this thread up from the dead, but I had this exact debate with my vehicle dynamics prof last week. Everything I've heard says to move the RC with the chassis, but he says to keep it fixed vs. the ground. His reasoning is that the change in CG height from pitch is very small, because the overall pitch angle is small. I couldn't come up with a good response. Thoughts?

[js10coastr]

If your chassis stays stationary relative to the ground, your roll center could stay stationary in relation to the ground AND the chassis.

http://fsae.com/groupee_common/emoticons/icon_smile.gif

[Richard Pare]

The Geometric Roll center is best kept quiet reletive to the chassis. Even though the pitch motions are "small", they are very significant. If the RC changes a lot reletive to the chassis, the driver will be chasing an ever-changing dynamic.

[Storbeck]

Did your prof give a reason for keeping it fixed relative to the ground? Sounds like HE couldn't come up with a good arguement. Basically he said the reason it should be done his way rather than yours is because it doesn't matter.

I say keep it stable relative to cg of car.

Also, it's not so much the need to keep the roll moment constant but the need to keep the relative roll stiffnesses the same front to rear. If the roll center stays steady with the ground then under braking when the nose goes down and the tail goes up you will have a relatively higher roll stiffness up front and a relatively lower roll stiffness in the rear. Vice versa for accel. Seems like it could create understeer on corner entry and oversteer on exit, pretty much the opposite of what you want. Based on that you could even argue that you want the roll centers to actually get closer to the cg as the suspension unloads and further than the cg as the suspension loads.

Something I'd like to have been able to try during my time in fsae land.

It's a shaky theory though. Constant geometric roll stiffness is probably the most predictable.

[kwancho]

Originally posted by Storbeck:
I say keep it stable relative to cg of car.

Right. Everyone agrees we want to keep the roll moment constant.
However, his point is that with pitch, the CG doesn't change. The pitch angle for us is what, 1 or 2" over 65?



Also, it's not so much the need to keep the roll moment constant but the need to keep the relative roll stiffnesses the same front to rear. If the roll center stays steady with the ground then under braking when the nose goes down and the tail goes up you will have a relatively higher roll stiffness up front and a relatively lower roll stiffness in the rear. Vice versa for accel. Seems like it could create understeer on corner entry and oversteer on exit, pretty much the opposite of what you want. Based on that you could even argue that you want the roll centers to actually get closer to the cg as the suspension unloads and further than the cg as the suspension loads.

Good point, but that's assuming the CG of the front moves down as much as the suspension moves. Like, under braking if the suspension moves 1", the chassis moves 1", and thus, the CG goes down 1". But, assuming 0 antis, the pitch center is in the center of the car, yes? And the axes are relatively close to the ends of the car, so the CG will move some lesser portion of that.

[Storbeck]

I may have just had a minor revelation thanks to your prof and your post. I'm starting to think he's right.

What all of the books say is that the roll center needs to stay fixed relative to the cg of the vehicle, so it must follow the sprung mass. But the cg of the vehicle is roughly in the center of the wheelbase, so when the car squats or dives the cg stays relatively unmoved vertically, so that the relationship between the front and rear roll centers and the cg would drastically change if they followed the sprung mass at thier ends, rather than the cg in the middle.

The world doesn't make sense to me anymore. I have to think about this a little more.

I seems so obvious right now, I can't figure out why this never occured to me.

[kwancho]

Yeah... That's kinda where I'm at.
Olley has a formula that can keep the RC location constant with bump. It's based solely on UBJ and LBJ height, and A-arm lengths. It works too, but it requires some compromising. I'll post it if I remember to write it down from the book at work tomorrow.

[Kevin Hayward]

Just to add a couple of thoughts. If you keep the roll moment arm is just one of the factors of the weight transfer equation. There is the close to instantaneous weight transfer that is due to the RC height relative to the ground. The idea being that if your RC is at the CG height there should be no roll, but there is still the same weight transfer.

The way in which the tire is loaded is quite important. Inelastic vs. elastic weight transfer. If you try and keep one constant for any movement then the other one will alter.

For the purpose of handling especially during turn in I would be attempting to have better control over inelastic weight transfer. This would make me want to keep the RC moving less compared to the ground.

Just my 2c

Kev

[kwancho]

Kevin -
Interesting. So, if you turn in on the brakes, and release them gradually on entry, the roll center height changes, and the geometric weight transfer changes. In this case, the roll center is moving up, increasing the geometric weight transfer (also, the geometric transfer is increasing purely because lateral acceleration is increasing). Any thoughts on how big of an effect the changing RC height has on turn in? Using Claude's logic, if weight transfer = grip, does that mean having the roll center move up on turn in = more grip?

edit: Olley's formula (if my math is correct):
height of RC change =
t/(2h) * (H2/R2-H1/R1)
where:
t = track
h = distance between UBJ and LBJ
H1 = height UBJ
H2 = height LBJ
R1 = length upper A-arm
R2 = length lower A-arm

Thus, to minimize the change in RC height, H2/ H1 = R2/R1. The lengths are usually dictated by camber curves and the lower one is ~1.5x the length of the upper one. Then, assuming the wheel center's at 10" (20" tire), H1 = 8", H2 = 12", which is way too close to the center of the wheel.

[Buckingham]

Alex,

Be careful not to think of cornering in one term (like grip). There is a magnitude of grip and there is an over/under damped response time as to how the vehicle reaches the final magnitude. Some parameters will affect the magnitude, some will affect the response time and some will affect the damping.

[kwancho]

Right. If I wasn't completely crazy, Claude used grip in this sense to mean lateral acceleration. So... my real question is does lifting off the brakes on entry cause the front of the car to turn in harder due to the roll center height change (in addition to all the longitudinal effects). Or is it meaningless/moot/impossible to isolate this effect because so many other things are happening (roll = elastic weight transfer, longitudinal effects, roll damping)?

[Alan]

...Or is it meaningless/moot/impossible to isolate this effect because so many other things are happening (roll = elastic weight transfer, longitudinal effects, roll damping)?

Add compliance to that list.

[kwancho]

So... after reading the Olley book some more, he says it is "usual to ignore the slight side shift of the roll center O which occurs when the car rolls". The Milliken guys say, uh, actually "the side shift in the roll center may be large and it may make sense to use a more sophisticated type of analysis".
Is there some type of simplified equation for lateral roll center migration with roll like Olley's for bump? I'd like to know exactly which parameters affect it, and which don't.

[Kevin Hayward]

Alex,

When the RC moves up and down during braking/accel there will undoubetdly be a shift in the ratio of geometric and elastic weight transfer will alter. Given that it is control of the elastic weight transfer that will affect the roll moment distribution this will cause a change in balance. However I would say that the longitudinal weight transfer will have a bigger influence.

The effects of a changing RC height during Turn-in are probably not dominant. Will have to give it more thought though.

Claude's grip / weight transfer argument is only partially useful. Yes in a 4 wheeled vehicle anytime there is grip there is weight transfer. An equal sign definitely does not work though. More weight transfer doesn't necessarily mean more grip. I think it is that whole cause / effect relationship, an Claude puts it in a way that makes you think about the issue a bit more.

On your equations, your SAL will have a greater effect on the camber change than the ratio of the upper and lower arms for any case where the arms are reasonable long. For example an open wheel single seater. Check this on your kinematics programs (or excel) for the ranges of movement seen in a FSAE car. I would definitely not lock yourself into a ratio of 1.5x. In fact I would think it more common to use SAL for camber and then alter the upper arm for RC control.

Kev

[Kevin Hayward]

Alex,

Sorry, forgot to mention in my last reply that the reason I would have more focus on geometric weight transfer is that it happens a lot faster. Tires take time to generate grip. For sims I think the standard is half a revolution of the wheel before full grip is reached. At lower speeds this time is more significant. You can get a situation where you ask too much of the tyres too quickly.

This is very evident in the way you drive the cars. For example if you go quickly to your steering angle during turn-in it is quite easy to induce understeer, however for the same car if you decrease the speed to the same steering ange you can have oversteer. A similar effect can happen with loading. Sometimes delaying the load to the tires will allow them to generate more grip over when you want/need it.

Using this stuff for control is very dependant on a lot of things. Temperature of tires is very important. In some situations you want to load tires very quickly to build temp, in others you want to be gentle to avoid going over the curve.

A basic answer is that for the first couple of cars I thought it was best to design for a RC that was consitent with reference to the chassis, the next couple of cars I though it was better to have it versus the ground. Now I know that I don't know and all of it is trade-offs.

Cheers,

Kev

[Marshall Grice]

my main concern would be car setup. If you design the RC to stay fixed relative to the ground regardless of ride height and you set the car up with a different ride height(parallel wheel travel) what does that do to your roll stiffnesses? Example, you raise the car 1" front and rear. This should translate to 1" higher CG. So now your CG is 1" higher and your roll centers are at the same spot. This may not have an effect on overall balance if the roll centers do stay in the same spot relative to each other. It would increase roll however.

I agree that pitch movement doesn't effect the overall CG height very much. The pitch would be due to significant load transfer. This load transfer would significantly change the handling balance of the car(like Kevin said). Taking a braking example. more load on the front means more grip available from the front(oversteer) and nose down geometry means(with RC relative to chassis) the RC has moved down relative to CG(oversteer). If the RC were relative to the ground, pitch change would have little affect on balance.

Seems that RC relative to the ground may not be such a bad idea. that just seems weird to me though.

[GSpeedR]

Originally posted by Alex Kwan:
So... after reading the Olley book some more, he says it is "usual to ignore the slight side shift of the roll center O which occurs when the car rolls". The Milliken guys say, uh, actually "the side shift in the roll center may be large and it may make sense to use a more sophisticated type of analysis".
Is there some type of simplified equation for lateral roll center migration with roll like Olley's for bump? I'd like to know exactly which parameters affect it, and which don't.

I like the procedure given by Mark Ortiz in one of his RCE articles. He uses a method analogous to modeling anti-effects (pitch, dive, etc) in the roll front-view plane. The lateral location of the RC is not dependent upon the orientation of the instant centers but on the distribution of lateral force between the inside and outside tires (specifically he calls it "undefined"). He develops a "resolution line" 75% of the track width away from the outside tire to account for the outside tire generating 75% of the total force between the pair. The roll center is never outside of the track width, which I completely agree with...if the RC is outside the track then it's more a "heave node" or something (remember the pitch and bounce nodes in the x-z plane). In any case, ignoring the effects of the tires is a poor approximation. The RC height is the average height of the force-line intercepts on this resolution line...similar to the SAL method everyone has always used.

75% is an estimation since we don't know how much force a tire is producing between a pair. If you have the tire data you can determine a more realistic lateral force percentage based on the measured inputs to the tire (vertical load, alignment, estimated temp, etc.). I never trusted it when ADAMS tells me my RC is 3.72 miles from the center of the vehicle in roll.

[GSpeedR]

Originally posted by Marshall Grice:
my main concern would be car setup. If you design the RC to stay fixed relative to the ground regardless of ride height and you set the car up with a different ride height(parallel wheel travel) what does that do to your roll stiffnesses? Example, you raise the car 1" front and rear. This should translate to 1" higher CG. So now your CG is 1" higher and your roll centers are at the same spot. This may not have an effect on overall balance if the roll centers do stay in the same spot relative to each other. It would increase roll however.

In many/most cases I think the RC will also raise when you raise ride heights, as the force-line angles will increase (ie raise the IC heights). You'd need to recalculate the roll moment to say whether it's bigger or smaller.

[nopanon]

Discussing the influence of roll center movement on roll stiffness, it shouldn't be forgotten that there is more to roll stiffness than RC height. I think that the progressiveness in your suspension, which is most likely there, will have a much greater influence on how your roll stiffness distribution alters with picth, than what the roll center movement does in the normal case.

[Marshall Grice]

edit: Olley's formula (if my math is correct):
height of RC change =
t/(2h) * (H2/R2-H1/R1)
where:
t = track
h = distance between UBJ and LBJ
H1 = height UBJ
H2 = height LBJ
R1 = length upper A-arm
R2 = length lower A-arm

so exactly what does equation tell you? it's obviously unitless. My numbers give me "1.75" and i'm just curious if that is like an in/in relationship or what

[kwancho]

Yeah, inch of RC movement for inch of bump, I think... Methinks I wrote down that equation wrong. I'll check tomorrow.

[awhittle]

Design the roll center height to be wherever you want it verticaly and such that in any position where the suspension can get to the roll center does not move horizontaly more than an inch or so. If the camber curves are what you are wanting to prove to the judges, you are done and move on to the next subject. The vertical distance from the CG to the RCH will change some with vertical suspension travel. You just have to accept that.

[GSpeedR]

Unless you're thinking with a kinematic RC I don't believe you can constrain it laterally.

[awhittle]

DohertyWins

Try these numbers in your suspension program. You can get real close to static lateral.

AW
http://www.ncs-stl.com/images/sampleSus1.jpg
http://www.ncs-stl.com/images/sampleSus2.jpg

[GSpeedR]

Originally posted by awhittle:
DohertyWins

Try these numbers in your suspension program. You can get real close to static lateral.

AW

With a kinematic RC, yeah I agree you can keep it at the center of the car. There are different ways of calculating the RC location (force center like ADAMS, or tire-based like Ortiz) and a method based on control arm angles seems to be the least accurate way.

[Marshall Grice]

AW,
I'm curious why you say to keep the horizontal RC location as stationary as possible. I've never seen/heard that the horizontal position is relavant to handling balance. Care to explain?

[js10coastr]

Originally posted by Marshall Grice:
AW,
I'm curious why you say to keep the horizontal RC location as stationary as possible. I've never seen/heard that the horizontal position is relavant to handling balance. Care to explain?

asymmetrical lateral movement of the RC will combine pitch and roll under lateral and long acceleration... screws with drivers.

[kwancho]

Lateral migration doesn't affect the roll moment length, but it does affect what direction and how much the suspension travels. If the RC moves to the inside wheel, the inside wheel sees less compression and the outside more extension. If it moves past the half track, both go into extension (or compression if it;s to the outside).

[Mike Cook]

Alex, I like to think that if the roll center moves to the outside wheel contact patch, that tire can provide no force to resist the rolling moment, therefore all the elastic lateral load will happen at the other end of the car. Do you agree?

[awhittle]

If the front RC is moving left as the rear is moving right this can't be good. If it stays in the center at all possible combinations the above will never happen. We have enough to worry about from the drivers seat as it is. Who would ever want even more confusing input to the driver.

AW

[kwancho]

Originally posted by Mike Cook:
Alex, I like to think that if the roll center moves to the outside wheel contact patch, that tire can provide no force to resist the rolling moment, therefore all the elastic lateral load will happen at the other end of the car. Do you agree?
Yes. The outside suspension doesn't deflect, and thus doesn't provide any stiffness.

[awhittle]

Some day we need to start two threads.

1) Impressing judges and intertaining people that like to run calculations

2) Making a car fast and reliable and how it differs from thread 1

AW

[GSpeedR]

Originally posted by awhittle:
Some day we need to start two threads.

1) Impressing judges and intertaining people that like to run calculations

2) Making a car fast and reliable and how it differs from thread 1

AW

Some day I am going to make two sandwiches.

1) Ham and swiss with mustard and bacon

2) Peanut butter and jelly and how it differs from sandwich 1

DW!

http://fsae.com/groupee_common/emoticons/icon_biggrin.gif

[Marshall Grice]

asymmetrical lateral movement of the RC will combine pitch and roll under lateral and long acceleration... screws with drivers.


but any non ground level RC already does that. How does moving it sideways introduce anything different?

About the only thing I can say about horizontal RC movement is that your instant centers are no longer at the same height(left vs right). With out knowing the forces involved i don't think that tells you anything useful.

[GSpeedR]

Originally posted by Marshall Grice:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">asymmetrical lateral movement of the RC will combine pitch and roll under lateral and long acceleration... screws with drivers.


but any non ground level RC already does that. How does moving it sideways introduce anything different? </div></BLOCKQUOTE>

Inclined roll axis will combine roll with yaw. If the roll axis is rotated about the vertical z axis, then you have coupled roll and pitch. Example thread (http://fsae.com/eve/forums/a/tpc/f/125607348/m/57510718721/r/17010499721#17010499721)

[ben]

Originally posted by DohertyWins!:
I like the procedure given by Mark Ortiz in one of his RCE articles. He uses a method analogous to modeling anti-effects (pitch, dive, etc) in the roll front-view plane. The lateral location of the RC is not dependent upon the orientation of the instant centers but on the distribution of lateral force between the inside and outside tires.

The resolution is determined by the instant centre so I think Ortiz's method does include the effect of IC position.

Any approach that looks at % anti-roll or behaviour of force resolution lines automatically leads you to consider roll centre movement w.r.t the ground rather than the CG.

The obvious question to ask is "Does the car roll about the kinematic roll centre?" The answer I think is no. This being the case should I care where the kinematic RC is at all?

Ben

[murpia]

As mentioned above, for non-aero cars at least, the CoG will not change very much in height during typical manoeuvres, except in the case of very large jacking forces (either lateral or longitudinal). Also, nothing much happens in a racecar unless you react a force through the contact patches, which are (usually!) on the ground. So, initially it makes sense to analyse, and constrain any 'roll-centre' movement (if that's your decision), with respect to the ground.

IMO the kinematic roll-centre concept is a useful one as far as it goes. It's the final stage of the instant centre & virtual swing axle 2D drawing and you'll need those when analysing the amount of scrub, camber gain etc.

However to conduct a complete vehicle performance analysis you will need to consider the distribution of contact patch forces and the application of those forces via the suspension links to the sprung mass. Whether you pass through the intermediate stage of instant centres, virtual swing axes etc. or just go straight to the multi-body simulation of your choice, is up to you. Once complete you should look for (and try to avoid) big car balance shifts as a result of weight transfer effects. Maybe these will result from a migration of the intersection of the instant centre geometry lines (in which case constrain them with different geometry), but then again, maybe not.

And you will always find someone out there who will challenge your assumptions, or your tyre data!

Regards, Ian

[GSpeedR]

Originally posted by ben:
The resolution is determined by the instant centre so I think Ortiz's method does include the effect of IC position.

The lateral location of the "resolution line" does not depend on the IC location (unless I've misunderstood the article), but on the distribution of tire force. The RC height is dependent on the force-lines and thus the ICs, but that is height only. I should note that he declares the lateral position as undefined.


The obvious question to ask is "Does the car roll about the kinematic roll centre?" The answer I think is no. This being the case should I care where the kinematic RC is at all?

Ben

I still think it is an OK way to make general predictions about handling behavior, particularly when you are at the track and not in front of your computer.

[murpia]

Originally posted by DohertyWins!:
I still think it is an OK way to make general predictions about handling behavior, particularly when you are at the track and not in front of your computer.

For sure, if you decide to raise or lower a 'roll-centre' and all else being equal, you can add weight transfer to an axle by raising the 'roll-centre' and vice versa.

But I can't resist asking why you wouldn't be in front of your computer at the track, particularly when changing your suspension geometry? Surely you'll need to run your quasi-static multi-body simulation to understand the effect of the camber-gain, ride-height, motion-ratio, anti-forces and jacking-forces changes you've just made at the same time...

Regards, Ian

[Drum]

Interesting discussion. I've given up hope that I'm going to ever have all the answers with regards to vehicle dynamics, and I've come to the realisation that you just have to use what you've got. Kinematic roll centres are not ideal, but are a good representation of what happens between the non-suspended and suspended mass. Add some basic information about the tyres and you will have a basic idea of what is going on between the system and the ground, however the output can only be as accurate as the input.
One of the projects I'm working on at the moment is actually a "quasi-static multi-body simulation" (sorry for stealing that description, but I like it!). I should mention that I don't expect a driver to sit in the car in between testing runs while running a simulation like this, but if there is a possibility of making a geometry change at the track then tools like this can give you a good indication of what to expect before you leave the office.
I found the debate interesting about roll centres either following the suspended mass or the chassis. I don't expect there is a black and white answer, there are advantages on both sides. To confuse the issue more, I decided to run the two scenarios through our quasi-static multi-body simulation. The solid line represents the car with roll centres stationary with respect to the ground. The dashed line is the car on which the roll centre height changes with the suspended mass height when the car is in bump or droop.
http://www.optimumg.com/images/Hosted%20Pics/RC%20Ground%20or%20Chassis.gif
I'm sure that the effects could be shown with a greater contrast if I had used step inputs of lateral and longitudinal acceleration, but chose to use some data typical to what you may see on track.
The inputs of mass, cg, shock absorbers, inertias, etc are only general and were not matched with the input acceleration. To look at the basic concept this was not neccessary.
I got lazy in this example and simply constrained the height and lateral position of the instant centres for the "RC Ground" run, and then only adjusted the height of he instant centres to achieve a roll centre that will follow the chassis in bump and droop. This had the additional effect of causing the roll centre to migrate toward the inside wheel during cornering.
This lateral movement of the roll centre reached only 110mm, but the effect on the "Vertical Wheel Position" plot can be seen clearly (a negative value indicates that wheel is in bump). The effect of moving the roll centre toward the inside wheel had an anti-jacking effect as expected.
I included a plot of the front lateral weight transfer, split into unsprung, geometric and elastic weight transfer. It isn't the best example to demonstrate the differences here, but if you get a magnifying glass you can see that the geometric transfer of the car with the 'RC Chassis' is a little less and the elastic weight transfer of the suspended mass is a little more. Ideally a combination of significant lateral and longitudinal acceleration would magnify this effect. I'm sure that it would also show the total weight transfer as being less in 'RC Chassis' due to the anti-jacking effect of lateral roll centre movement.
So..... this perhaps is not a clear cut answer for the roll centre vertical movement question, and my feeling is that whatever answer you come up with will be a compromise. You just have to use what you have got to come up with the best solution you can.

Pat Drum
OptimumG
RMIT Racing Fan Club

[kwancho]

So, some inboard lateral migration in a turn does mean anti-jacking forces? I was trying to work it out in my head, and I couldn't figure it out. If, for example, the RC migrates to the inside halftrack, the inside suspension doesn't move at all, while the outside one rebounds a lot. What does that mean for weight transfer? Assuming the height doesn't change, the geometric weight transfers should be the same. How does the elastic weight transfer happen if the springs don't compress?

[Drum]

Alex

My understanding of the effect of lateral RC migration I think is the opposite to what you have described. The way I see it is if the roll centre migrates toward the inside wheel, there will be less movement on that wheel (less rebound) and more movement on the outside wheel (more bump). The way that I picture this is to exaggerate the situation where the RC moves towards the inside of the corner, lets say the RC is not between the wheels but another 1/2 track away from the car.
If the car tries to 'roll' about this point, the radius of rotation will be 1 track width. From the front view, if you roll the suspended mass about that point on the inside of the corner, the whole suspended mass will be either compressing all springs or extending all springs. Given that the centre of gravity of the suspended mass is above ground, the direction of roll will result in all springs being compressed (what I described as an anti-jacking effect before).
I'm not very good at explaining things without drawing crazy diagrams, usually to double check that I'm not telling fibs.
You are right in correlating the elastic weight transfer to the spring compression. If there is more spring compression, this means there is more elastic weight transfer. For a given lateral acceleration, centre of gravity and track width you know the total (final) weight transfer of a vehicle of known mass. If the roll centre location is changed yet the centre of gravity is constrained, the total weight transfer will remain the same. Therefore you can assume that the gain in elastic weight transfer was 'borrowed' from the geometric weight transfer. However, if the centre of gravity was free to move, the jacking/anti-jacking effects could move the cg and the total weight transfer would not necessarily be the same.
This weight transfer only speaks of the vertical load on the tyre. If the roll centre is not between the two tires and a lateral force is translated to the roll centre, when this force is resolved geometrically into vertical and lateral loads at the contact patch the two tyres will have conflicting lateral loads.
If the roll centre is out past the outer wheel, it has been suggested that the outside tyre can experience lateral grip initially in the wrong direction. This is something that I have not got my head around, perhaps it is at this point where the differences between the kinematic roll centre and the force based roll centre become more obvious.
I'm keen to hear people's thoughts on this..... or maybe someone who could explain it better. Big Bird??

Pat Drum
OptimumG
RMIT Racing Fan Club

[kwancho]

Shit, totally got that reversed.
Alright, so, we agree that lateral movement of the RC does change the components of the weight transfer. Basically, what you're saying is that the anti-jacking forces from an RC that migrates inward would squat the chassis, thus lowering the CG and reducing the amount of weight transfer, and which then changes both geometric and elastic weight transfer, which then causes the RC to do some more migrating, etc. etc. right? Isn't that bad both for consistency of loads on the tires and for driver control?

[Drum]

The total reduction in weight transfer from that anti-jacking effect is not that significant, I just got a bit carried away explaining why my theory was far from perfect.
The squatting of the car will not neccessarily cause more roll centre migration, it did in my example that I used earlier on but that is purely a function of your chosen geometry.
The consistency of loads on the tyres is an interesting subject, I'm guessing you mean how you want the tyres to be loaded throughout the corner from a transient perspective. If you are lacking response you may need more geometric weight transfer. If you lack overall grip you may need less geometric weight transfer.
The feedback of the vehicle to the driver is important, and roll centre control laterally throughout the corner will contribute to how the driver perceives the car behaviour. If non-linear motion ratios are being used then roll centres migrating laterally could then be used to shift the roll stiffness distribution to the front or rear mid-corner. And don't forget that the shock absorbers may need to work in different ways with a significant migration of the roll centre, and perhaps deserve a thought when it comes to the transient roll stiffness distribution.
Even after all this discussion, I wonder how many teams would benefit from designing enough lateral roll centre migration into their suspension to have to worry about the adverse effects. I think that a good starting point is just to keep it somewhere in the middle and be able to explain why it does or it doesn't move vertically on your car.

Pat Drum
OptimumG
RMIT Racing Fan Club

[Jeff The Pyro]

From an old fashioned theoretical standpoint, which assumes that the geometric roll center is in fact the point that the car is rotating around, then the decreased jacking force would be explained by drawing a force vector between the outside contact patch and the roll center. as the roll center moves farther from the contact patch, the angle of this vector becomes more horizontal resulting in a decreased jacking force.

After putting a lot of thought into this, i'm starting to come to the conclusion that in the end the location of the geometric roll center is not particularly useful. I was talking to some of the design judges about this after formula West this year about the older school of thought that it was a very bad thing for the roll center to pass through the ground plane at any time, and bad in general for the roll center to move. I dont remember the names of the judges but they were saying that a lot of this theory had to do with a correlation between the change in heights of instant instant centers and roll center migration, in other words its a good way to approximate the more complicated effects of whats going on at each individual corner and lump them into one easy to analyze value.

The conclusion i'm turning over in my head is that keeping the geometric roll center stable generally results in a stable car, as has been historically shown, but is not *necessary* for this to occur. Where the roll center theory falls apart is when its positioned right around ground level. At this point a tiny change in instant center height at one side of the car can make the roll center jump literally miles away in the horizontal direction. The other part was that if the roll center crossed the ground plane, the jacking forces would reverse direction, theoretically resulting in instability. If you think about it logically, however, neither of these are particularly large concerns. If you approximate a car as a single mass sitting on a spring, its a balance of force between gravity and any other external forces, and the "zero displacement" reference condition is set at equilibrium so that gravity can be neglected. Once you add gravitational force to the jacking force and take the sum, there really isnt any "reversal" of force going on, simply a reduction, no different than if the roll center changed height while far from the ground. (a reversal of force would be when the jacking force completely overcomes the car's weight and it pole-vaults off the ground and flies into orbit...) Besides this when the roll center is near enough to ground level jacking force is so small that its reversal in direction would be pretty meaningless anyways. The horizontal migration issue also isnt much of an issue if you look closer at it, especially since its pretty much accepted that the car isnt actually going to rotate around this point.

Anyways what the judges were explaining to me was that what you really need to pay attention to are the individual instant center heights above the ground, along with the lateral force contributed from each tire, because these are what dictate the net jacking force. The idea given to me was to draw a line from the initial instant center location to the contact patch, and to try to keep the instant center on that original line at all times, or else to have it deviate in a smooth fashion if you do want a jacking force change for some reason.

This seemed to be on the right track, but the more i think about it i'm still not sure if its quite there, because keeping the force vector in the same direction wont really stabilize the jacking force because its vertical component is still dependent on the magnitude of the vector (ie a function of lateral acceleration). Imagine a car taking an s-type curve with an above ground roll center/nonzero jacking force. There is no jacking force while the car is going strait however as soon as the car takes some lateral acceleration, a jacking force will develop. Now the car turns the other way through the second half of the S curve. it momentarily passes through a point of 0 lateral acceleration, at which point the jacking force drops to zero, then as it turns the other way the jacking force develops again. This type of "bouncing" behavior doesnt seem favorable at all from the aspect of stability and predictability to the driver, especially if it is great enough to cause a significant disturbance in ride height. It seems like the best way to avoid this is to keep all instant centers exactly at ground level at all times, then all jacking effects are simply nonexistant.

The real question is, will this unsteady jacking force actually cause significant unfavorable handling characteristics? Most of the major instability occurs during transient states where the driver would never really expect the car to be "steady" anyways... obviously if the driver was in a steady state turn and suddenly the jacking force changed this would unsettle the car a lot, but the most unstable part only occurs inbetween strait line and steady state turning, where all sorts of other effects are going on (rapid longitudinal and lateral weight transfer, body roll, etc.) and a small change in net jacking force may be completely overshadowed by these other effects.

[ben]

Originally posted by DohertyWins!:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by ben:
The resolution is determined by the instant centre so I think Ortiz's method does include the effect of IC position.

The lateral location of the "resolution line" does not depend on the IC location (unless I've misunderstood the article), but on the distribution of tire force. The RC height is dependent on the force-lines and thus the ICs, but that is height only. I should note that he declares the lateral position as undefined.


The obvious question to ask is "Does the car roll about the kinematic roll centre?" The answer I think is no. This being the case should I care where the kinematic RC is at all?

Ben

I still think it is an OK way to make general predictions about handling behavior, particularly when you are at the track and not in front of your computer. </div></BLOCKQUOTE>

The resolution line is developed by drawing a line from the contact patch to the IC - so yes it is included. The amount of force each resolution line has to resolve is then determined by the tyre force distribution.

Drum: Cool looking sim. Is it full multi-body a la ADAMS or is it lumped parameter with a sprung mass and four unsprung masses? Also what tyre model are you using?

Ben

[murpia]

Some weighty posts overnight!

Anyway, I think the consensus now is that the (old fashioned?) concept of drawing an intersection between the lines connecting the instant centres with the contact patches and calling that the 'roll-centre', and furthermore expecting the car to actually roll around the axis formed by joining the front and rear 'roll-centres', is useful but fundamentally flawed. Certainly as a minimum the difference in contact patch forces need to be considered!

It's the 'lateral migration' concept that seems to cause the most confusion. Many designers are proud to announce they have kept their 'roll-centre lateral migration' low or even zero. I think we've agreed that in fact it's the height of the individual instant centres, and the effect on the jacking forces, that's most important. Dramatic changes will likely make for less predictable handling as geometric load transfer is traded for sprung load transfer (in the first approximation assuming equal contact patch forces and no CoG height change).

Always remember too that the jacking forces are a result of contact patch forces, and don't just 'turn on' instantly but build up with slip angle just like any other tyre force. Hence the concept that 'high roll-centres are good for initial response' is a limited one that needs further justification (in my view).

Drum, your simulation is actually going one step beyond what I was thinking, so well done. I was considering a model that for given lateral and longitudinal accelerations and steer angle, solves for the car attitude and contact patch forces (based on full kinematics and at least a basic tyre model, and I expect an iterative solver is required). Obviously some points will prove unsolvable!

Regards, Ian

[awhittle]

Watch these two movies of the same car on the same course and then put into prospective stady state cornering loads in the big picture. Remember that the mission of FSAE is to design an "autocross car for manufacture and sale".

http://www.ncs-stl.com/mike/Movie.wmv

http://www.ncs-stl.com/mike/Movie_0007.wmv

AW

[GSpeedR]

Originally posted by ben:
The resolution line is developed by drawing a line from the contact patch to the IC - so yes it is included. The amount of force each resolution line has to resolve is then determined by the tyre force distribution.

Nope. Here's a clip from the Ortiz article I was just talking about Roll Center (http://www.auto-ware.com/ubbthreads/showflat.php?Cat=0&Number=489&an=0&page=1#Post489) :


I have also said that the roll center, properly assigned, should be considered a point in side view (of the car), and its lateral position should be considered undefined. It lies in the transverse plane containing the wheel center in all cases, or, in side view, it lies straight down or straight up from the wheel center. So we really need only one number to define its position, namely its height. This height is not the same as the height of the force line intersection. Rather, it is the mean height of the two force line intercepts on a line I call the resolution line.

The resolution line is a vertical line in the front view, positioned according to distribution of lateral force generated by the two tires. For example, if the right front tire is generating 75% of the front lateral force, the front suspension resolution line is 75% of the track width away from that tire.

Unfortunately, we do not know this distribution of lateral force exactly, in most cases. We have to estimate it. That means our modeling of the suspension's behavior is only as good as this estimate. This is unfortunate, but ignoring the fact doesn't make it go away. The behavior of the suspension really does depend on the distribution of lateral force. To predict the jacking forces each of the individual wheels generates, and thereby calculate an anti-roll or pro-roll moment, we must not only know the suspensions' geometry, but also the forces at the contact patches. Any analysis method that takes this into account, even using an estimate for the lateral force distribution, is better than a method that ignores this factor altogether.

The resolution line is different from the "force-line" connecting the contact patch to the IC at least in the article I've been referencing.

[Marshall Grice]

Originally posted by awhittle:
Watch these two movies of the same car on the same course and then put into prospective stady state cornering loads in the big picture. Remember that the mission of FSAE is to design an "autocross car for manufacture and sale".


AW

what perspective is to be taken from these videos?

personally i'd say this car is in serious need of steady state cornering improvement.

[awhittle]

Originally posted by Marshall Grice:
perspective is to be taken from these videos?

personally i'd say this car is in serious need of steady state cornering improvement.

The point I was trying to make is nothing is ever static in autocross. I have full time wireless datalogs coming off the Megasquirt on that car. Between watching the video and watching the AE.... The trick to making a quick autocross car is not making it fast... The trick is making the car drivable at these fast speeds. I have seen my car go from .75 g acceleration to 2 g braking to 2 g cornering in 3/4 sec on Topeka concrete. I have seen karts fly threw the air and land in ways I can't believe at Nationals. Right now I autocross a 2002 CRG shifter cart and learned that all the things that I thought about chassis design was sacred was ALL in question. The new A-Mod car will be designed with me warring different colored glasses with what I have learned in the CRG. I enjoy getting people to think out of the box regarding everything they design. The things that will bite every new design will be the things that 6 mo before you would have never even considered. Things that you debated for weeks may never be a problem for either side of the debate. I have seen all sorts of chassis/suspensions work over the years and I am convinced that alignment and choice of shocks has a far greater influence on handling than RC movement.

spent your time on driveability and less on design. The greatest number of points is all in the autocross and endurance. The rest is just icing on the cake.

AW

[murpia]

Originally posted by awhittle:
Watch these two movies of the same car on the same course and then put into prospective stady state cornering loads in the big picture. AW
These videos are of an car where aero is significant (I hope, looking at the size of that wing!). So the CoG height is likely to be different low speed to high speed, and much of the above discussion is invalid.
Regards, Ian

[murpia]

Originally posted by DohertyWins!:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by ben:
The resolution line is developed by drawing a line from the contact patch to the IC - so yes it is included. The amount of force each resolution line has to resolve is then determined by the tyre force distribution.

Nope. Here's a clip from the Ortiz article I was just talking about Roll Center (http://www.auto-ware.com/ubbthreads/showflat.php?Cat=0&Number=489&an=0&page=1#Post489) :


I have also said that the roll center, properly assigned, should be considered a point in side view (of the car), and its lateral position should be considered undefined. It lies in the transverse plane containing the wheel center in all cases, or, in side view, it lies straight down or straight up from the wheel center. So we really need only one number to define its position, namely its height. This height is not the same as the height of the force line intersection. Rather, it is the mean height of the two force line intercepts on a line I call the resolution line.

The resolution line is a vertical line in the front view, positioned according to distribution of lateral force generated by the two tires. For example, if the right front tire is generating 75% of the front lateral force, the front suspension resolution line is 75% of the track width away from that tire.

Unfortunately, we do not know this distribution of lateral force exactly, in most cases. We have to estimate it. That means our modeling of the suspension's behavior is only as good as this estimate. This is unfortunate, but ignoring the fact doesn't make it go away. The behavior of the suspension really does depend on the distribution of lateral force. To predict the jacking forces each of the individual wheels generates, and thereby calculate an anti-roll or pro-roll moment, we must not only know the suspensions' geometry, but also the forces at the contact patches. Any analysis method that takes this into account, even using an estimate for the lateral force distribution, is better than a method that ignores this factor altogether.

The resolution line is different from the "force-line" connecting the contact patch to the IC at least in the article I've been referencing. </div></BLOCKQUOTE>

I think maybe there's a bit of confusion here about the term 'resolution line'?

Mark Ortiz is saying draw a vertical line biased to one side in the ratio of contact patch lateral forces, and use this line rather than the car centreline to generate a 'roll-centre' height. He calls this his resolution line.

Some of us here are suggesting you use the true resolution of the contact patch forces on the sprung mass to calculate your car behaviour. While these forces act along the lines between the contact patch and instant centres (you could call _these_ the 'resolution lines') the length of these lines is arbitrary depending on your vector scale of forces and the effect on the sprung mass needs to be calculated using a free body diagram and vector algebra (or the quasi-static multi-body simulation of your choice).

Maybe I'll post a drawing...
Regards, Ian

[GSpeedR]

Originally posted by murpia:

Some of us here are suggesting you use the true resolution of the contact patch forces on the sprung mass to calculate your car behaviour. While these forces act along the lines between the contact patch and instant centres (you could call _these_ the 'resolution lines') the length of these lines is arbitrary depending on your vector scale of forces and the effect on the sprung mass needs to be calculated using a free body diagram and vector algebra (or the quasi-static multi-body simulation of your choice).

Maybe I'll post a drawing...
Regards, Ian

I'm pretty sure I understand that method that you (and Ben) are describing (forced-based RC's, right?), though I've always heard/used "force-line" instead of resolution line. The 2 methods are slightly different though, not sure which one is better.

[awhittle]

Could someone post a napkin sketch of what Ortiz article is trying to describe. I am a mechanical engineer and like pictures and arrows. At 2 gees, is there any weight left on the inside tire.

AW

[Jeff The Pyro]

the resolution line method doesnt really make sense to me. the inside tire is creating a force vector that is pointing inside the turn (IE the force line between the IC and the tire patch is in tension, whereas the outside tire force line is in compression), which means the vector itself should never actually intersect this "resolution line". You could just use the line of action, but it wouldnt make sense because the sign conventions would be all screwy. Think of it this way... assuming the inside and outside IC's are at teh same height, they are actually creating jacking forces in opposite directions of each other. The outside tire IC just has a larger amount of lateral force because of load transfer and therefore overpowers the jacking force associated with the inside tire. The only true way to calculate the net jacking force is to figure out the load conditions at each tire and sum the vertical components.

It just seems like every way that anyone has suggested as a method to calculate the roll center are just "guesses" that coincidentally come up with believable results. The only true way to do it would be to do a full dynamic analysis and plot the velocity vectors at two points on the sprung mass... then draw the normal planes to these two vectors and where they intersect would be the roll axis.

[ben]

I pretty much agree with that.

I don't see the point of Ortiz's method in arriving at a single roll centre. If you're going to make an assumption about tyre forces you could just as easily resolve the vertical equilibrium of each tyre using the wheel rate and IC position.

Ben

[murpia]

Originally posted by DohertyWins!:
I'm pretty sure I understand that method that you (and Ben) are describing (forced-based RC's, right?), though I've always heard/used "force-line" instead of resolution line. The 2 methods are slightly different though, not sure which one is better.

Not quite, what we're saying is that the more you look at the problem the more you realise that the concept of a single, convenient dimension called the 'roll-centre' (either force or geometry based) is flawed. Real-life racecars just aren't that convenient I'm afraid.

What is important are your jacking forces. When designing your car you should keep in mind an objective for those, be it zero net jacking force or otherwise. Also, you need to decide if you require some geometric weight transfer, or want to react it all through the springs / bars.

To go back to the original thread title 'roll-centre movement' then, it seems the answer could be 'it doesn't exist so don't worry'. But, as Jeff says, sometimes these sort of 'guesses' do help, and certainly constraining the movement of the mythical geometric 'roll-centre' is very likely to co-incidentally result in well controlled contact patch geometry and minimal variation in net jacking force, both of which are decent design objectives to have set yourself.

Regards, Ian

[ben]

In my experience the static unrolled kinematic roll centre is worthwhile as a quick and dirty assumption for specifying intitial ride and roll rates as per RCVD.

Beyond that I always focused more on camber control and scrub radius variation.

Unless you know how accurately or not the kinematic roll centres represent reality, spending too much time tracking roll centre migrations and assuming they are informative about the actual loads on the tyre seems a little bit pointless.

Ben

[awhittle]

Ben

I think you are coming around to the way I am seeing the issue.

Load distribution thru the Heims and into the chassis, upright design, a clean rocker/shock/motion ratio, camber curves, caster, ackermen, unsprung weight... All these things are all far more important when you consider the video I posted earlier and the beating these cars take when you finaly make then quick.

AW

[GSpeedR]

OK, I'll buy the "the whole concept is inherently flawed" argument but if we're using these methods to make good guesses then I think it's worth to gain as much insight as possible. If the focus point of your analysis is to understand your jacking forces, then it would be best to use a method that at least takes tire forces into account. It's still quick and dirty; it doesn't even require more pencil lines on your napkin, but it does a better job of answering "why" and that's the point IMO.


But, as Jeff says, sometimes these sort of 'guesses' do help, and certainly constraining the movement of the mythical geometric 'roll-centre' is very likely to co-incidentally result in well controlled contact patch geometry and minimal variation in net jacking force, both of which are decent design objectives to have set yourself.

Along the lines of what I said in this post above, there could be some danger in the above. You've designed your suspension to contrain this geometric roll center in order to acheive certain goals (minimal jacking variation, etc). However, if your method sucks and your design does not minimize jacking variation then you might be in trouble. If you method made a more educated guess then maybe you're results would be better (or maybe the same). It's at least worth the effort if you are using RCs (whether they exist or not) to make design decisions, which a lot of teams do.

Use the most accurate method within your resources, whatever that may be.

[murpia]

To dig up an old thread, this (http://www.ee.ic.ac.uk/CAP/Reports/2001/ASCInteractions.pdf) is worth a read.

Regards, Ian

[Goran Malmberg]

I am just comming up with an Idea of mine here to see what you think.
Let's say we have a Tw of 1600mm and decide to set the "Rc"� at 21mm in height. Therefore we design the A-arm so that the forcelines become 1,5 dgr. The intersection of the Fl will then be in the middle of the car. We apply a side load at Cg that result in 75 % weight transfer. If the geometry is such that the Rc stays the same spot during the roll, we got 1,5tan*75=1,96 for + jacking, and 1,5tan*25=0,65 for – J. The car will have net 1,31 in + J.
If we design the A-arm geometry so that the outer wheel Fl become 1 dgr and the inner wheel Fl 3 dgr, the intersection of the Fl will be at 75 % distance from the outer wheel to the inner wheel, still at the same height of 21mm. Now 1tan*75=1,31 and 3tan*25=1,31 for zero J.
The actual motion of roll will still be about the same spot in the middle of the car. What appear to make a difference in the "rollaxlespot"� is that the car is not being "jacked"� in the last example. A difference is of course that the camber curve will alter between the two examples during roll.
Goran Malmberg

[Goran Malmberg]

I hope I am not to boring by showing a few images. The images show a half scale model of a car. The tires each side, is free floating sideways and can be set to take different part of a side load that is applied at Cg. The springs has a strange possition, but this is to make the "car body" more free floating, just like lever balance scale.
image 1
http://www.hemipanter.se/Sae%201.jpg
The outer tire has 75% of the load and the intersection of the Fl is not exactley at the same sideway location and therfore we got a tiny bit of -J. There is an "eye" where we can see a dot on the middle that show how the body roll is behaving. The eye is located in the middle of the car at the static Fl intersection height.
image 2
http://www.hemipanter.se/Sae%202.jpg
Here I located 100% load on the outer wheel and the Fl intersection is also following out to the inner wheel but now giving a little more lift on the inner wheel as seen in the eye and the metric scales.
image 3
http://www.hemipanter.se/Sae%203.jpg
In this case we got extream angle A-arms, just to see what happens. Both wheels is set to take 50% of the side load. This means that the resolution line of Ortiz will be in the middle of the car. Fl itersection does also stay in the middle, meaning that left and right Fl taking the same load but in the opposit direction, resulting in zero J. The roll arm above the "Rc" is giving the same deflection each side then.
image 4
http://www.hemipanter.se/Sae%204.jpg
75% loaded outer wheel, the Fl intersection still pretty much in the middle of the car resulting in big J forces, the dot is outside the eye by 4,5 mm.
image 5
http://www.hemipanter.se/Sae%205.jpg
100% on the outer wheel and no load on the inner wheel. Naturally resulting in almost twice the J lift, 8 mm.
As the side load is tha same in all last 3 examle we got tha same total deflection in al cases, 16 mm. In the first case 8 mm each side and the last is 16mm on inside wheel.
Compared to image 3, we now got the same roll action from the roll arm, but also all the geometric forces on the outside force line, giving a 8mm lift of the chassis.

Goran Malmberg

[JHarshbarger]

I got flashing error signs for the images. Could just be the system I'm on.

[Goran Malmberg]

Originally posted by JHarshbarger:
I got flashing error signs for the images. Could just be the system I'm on.
Hmmm, I just open the side myself and it worked Ok. Lets see what other people end up with.
If there is a problem i will see what I can do.
Regards Goran

[kwancho]

Errorz.

[Goran Malmberg]

Originally posted by Alex Kwan:
Errorz.
Is is Ok now?
Goran

[murpia]

Goran,
This looks to be a physical rig, not a computer model, am I right?
Looks like an interesting experiment, can you describe the rig a bit better, for example: how are all the forces applied and balanced to achieve equilibrium?
Regards, Ian

[Goran Malmberg]

Murpia,
This is a one axle physical model set up on a big board mounted on the wall. Just for you to know, I also have a 2 axle model working in the same way to study the "roll axle" phenomenon.
The right wheel is fixed at centre of tire to road contact patch by a bearing. On the left side the same spot also has a bearing but is hanging in a long bar mounted in another bearing high up the board, making the tire free floating sideways.

A-arms are mounted on movable mounting points on the "chassis" by loosening screws for infinite adjustments of angles and another length.

The springs were at first mounted as usual from the lower A-arm to the top of the chassis, which is the reason for the shape of the top section of the chassis. However, this "coilover" design coursed bindings giving false readings at lighter loads. So, the springs were mounted for pulling loads.

Load is applied to the model by a cord from the Cg and at the left wheel contact patch. The cord is running horizontally out to-over an idler wheel down to vertical and a can for weights.

There are also the already mentioned "eye" that could be positioned anywhere in front of the chassis. A dot is positioned where the physical Rc is believed to be and the eye right over it.

There are scales at all important spots and adjustable arrows to measure track width, deflection, camber etc.

I like physical models as I they show what is actually going on, and the Rc discussion has running hot in magazines, forums etc.

We do have a rollcentre, and we do have jacking forces. And I find all of it useful, depending on what we want to calculate. Many times Rc and jacking forces seem mixed up with each other in an unlucky manner. We also fight against an older use and definition of Rc.

Goran Malmberg

[JHarshbarger]

Is this for the Pantera or the Corvette?

The amount of time you must have put into this physical model is...confusing. Why not use a suspension program such as SusProg3D? Tons of calculations all performed simultaneously...it makes small modifications a lot less tedious. For the initial setup and design, I can see why you'd make the physical model though.

[Goran Malmberg]

1
Is this for the Pantera or the Corvette?
2
The amount of time you must have put into this physical model is...confusing. Why not use a suspension program such as SusProg3D?
3
For the initial setup and design, I can see why you'd make the physical model though.

Joel.
1
My "modelwork" is not for any prticular car. It is that I like to see things i physical materia.
2
I have had a number of discussions here, but when people look at one of my model, we could say that have seen things with their own eys.
dont missunderstand me, program like "SusProg3D" is great.
3
Basically I have used what I call the "zero car" idea when constructing the Corvette. Which is an idea very much based on what I have seen by looking at my models.
Cheers
Goran

[murpia]

Originally posted by Goran Malmberg:
We do have a rollcentre,
Thanks for the explanation Goran.

By stating you do have a roll centre, I assume you mean that using your rig you can identify a single point which the sprung mass (body) appears to roll around?

In the spirit of the thread above, my question to you is therefore this:
If you changed your rig so that the left wheel is the one mounted by a bearing and the right wheel is free to translate, then for equivalent loading* would you identify the same point as the roll centre?

Regards, Ian

* presumably the distribution of contact patch forces is handled by varying the loading between the body weight can and the right wheel weight can. So for e.g. a 75%/25% split you could have 100N at the body, 75N at the right wheel and assuming negligible friction, the small angle approximation and equilibrium you know you have 25N at the left wheel. So for my question above you need to load the left wheel can with (body can - right wheel can) to achieve the same rig situation.

[Goran Malmberg]

Originally posted by murpia:
Thanks for the explanation Goran.
1
By stating you do have a roll centre, I assume you mean that using your rig you can identify a single point which the sprung mass (body) appears to roll around?
2
In the spirit of the thread above, my question to you is therefore this:
If you changed your rig so that the left wheel is the one mounted by a bearing and the right wheel is free to translate, then for equivalent loading* would you identify the same point as the roll centre?

Regards, Ian


1
Right. As one side of the car goes lower and the other higher, there must be a centre of rotation somewhere in between. My example "zero car" is a square one with one spring each corner, and if we move weight from one side to the other it will rotate about its middle. This is so even if we include A-arms that produce an instant centre and forceline intersection spot. So, to me it appears that the lateral "Rc" location is governed by load transfer and wheelrate, rather than forceline intersection. Presumed that we by Rc mean the actual spot the car roll about.

Then we got the rollarm length which is the CGH-geometric load transfer (just to say anything in short).

2
This is no problem to do as one only changes the direction of Cg force to the other side.
What happen is that the INSIDE wheel get ALL the load. Then, (depending on its forceline angle) we will probabley get -J. But the rolling motion will still be about the middle of the car.
------
We must separate the (two) chassis motion to see what is Roll and Jacking movements. Or else we will be confused about the "spot in the eye" movement. If we have NO Cg sideway force, and just apply force to the model "floating wheel"
we will se a vertical movement of the chassis without any roll whatsoever.
Regards
Goran

[Goran Malmberg]

Originally posted by murpia:
1
What is important are your jacking forces. When designing your car you should keep in mind an objective for those, be it zero net jacking force or otherwise.
Ian
1
Agreed, we must establish how much geometric load transfer we want the car to have. And how we want it to behave during suspension movement.
This got little to do with the rollcentre issue, as Rc is living a life of its own, even if the two are related.
2
The Rc is affected by the geometry, but so are the trackwidth and camber, as we to a big extent can see the "extreem angle A-arms" image 5.
3
If the Rc stays in the middle of the car we will have the same deflection each side during roll. Now, this may be a little confusing as if a large amount of J is present
as in image 5. The actual roll from the rollarm
is the same as in image 3, which create 8 mm of D each side. But image 5 show 16 mm D left and 0 mm D right side.

This set up has 50% Rch to Cgh, therfore 50% of the side force goes to J and 50% to roll. In image 3 the 50 % J is cancelled out, so there is only roll movement present. In image 5 all the geometric forces is creating J. As the geometric forces is the same as the 50 % creating roll, it will couse the same total deflection against the wheelrate = 8 mm each side. The difference is that it comes in the form of 8 mm LIFT each side, leaving for 8+8=16mm left and -8+8=0 on the right side.

The jacking and roll together makes the chassis to describe sort of an arc like movement with a centre that may appear to be at the right of the chassis. But the roll of the chassis is still at the same angle as in image 3, so a single roll spot for the chassis movement is not present as roll and jacking is doing their own "work".

The dot in the eye seem to move upp and to the RIGHT. One might draw the conclusion that the chassis is moving with a forceline centre in the middle of the right wheel contact patch. This is not so, the "to the right movement" comes from the fixed possition of the right wheel and that the bad geometry is shortening the track width, moving the chassis to the right.

Regards
Goran Malmberg

[murpia]

The July 2007 (Volume 17 Issue 07) issue of Racecar Engineering has a good article by William C Mitchell entitled 'Roll Centres - Myths and Reality'.

Definitely worth a read.

Regards, Ian

[Kurt Bilinski]

Bringing this back from the dead.

Shouldn't ICs should be related to lateral tire load? That is, as lateral force increase, and the outside tire is effectively doing most of the work, the importance of the inside tire's IC seems questionable. I understand the Mitchell comments about how the force line's *slope* is more important than an arbitrary intersection of ICs from both tires. I'm beginning to think that as corner reaches a maximum, only the loaded tire's force line/slope should be of concern. What the other side's doing seems unimportant since it's not contributing (much) to the overall cornering force. So if I design my suspension to keep the fully loaded tire's FAP the same height (as at rest), do I need to care about anything else?

I realize this is only one part of a larger picture but I really want to understand it. My life was much simpler before I learned all about this new-age FAP stuff, now my head hurts...

[exFSAE]

What makes you say the inside tire isn't really important? Are you lifting wheels? If so that would probably be an issue in itself. Up until that point, inside tire still contributes to cornering and IMO should be made the most of.. no?

Or am I reading you wrong?

[Kurt Bilinski]

Somewhat.

For example, if the outside tire is contributing 75% of the cornering force, it seems like that tire's FAP is three times as important (75/25) to be kept in the right place, while the inside tire's FAP is three times less important?

A small amount of knowledge is a dangerous thing.

[js10coastr]

Originally posted by Kurt Bilinski:
Somewhat.

For example, if the outside tire is contributing 75% of the cornering force, it seems like that tire's FAP is three times as important (75/25) to be kept in the right place, while the inside tire's FAP is three times less important?

A small amount of knowledge is a dangerous thing.

yes, but the coeff of friction for the inside tire is a lot higher than the outside so it is more than three times less important.

[exFSAE]

Originally posted by js10coastr:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Kurt Bilinski:
Somewhat.

For example, if the outside tire is contributing 75% of the cornering force, it seems like that tire's FAP is three times as important (75/25) to be kept in the right place, while the inside tire's FAP is three times less important?

A small amount of knowledge is a dangerous thing.

yes, but the coeff of friction for the inside tire is a lot higher than the outside so it is more than three times less important. </div></BLOCKQUOTE>

I think you mean its less than three times less important.

[Kurt Bilinski]

I know what you meant, but because of how i phrased the example, that the outside tire has 75% of the cornering force, takes that into account.

[Kurt Bilinski]

Here's my real-world (non-FSAE) example.

I'm using Miata uprights at the front of my next mid-engine car project. The Miata's factory front track width is about 57", but I want to go to 60". However, widening the track and lengthening the A-arms messes with things, making it impossible to keep the KRC static laterally, as well as being unable to keep the FAPs static. But, with some work, I can make the outboard FAP remain constant, at the expense of both the KRC and inboard FAP.

Is this a good thing or a big mistake? My theory is, keeping the FAP constant on the heavily loaded outside tire is a good thing. (though I realize there's are other factors as important...)

[exFSAE]

Is it good or bad? You tell me. It's easy to go down the road of intellectual hoo-hah, but the objective is still to maximize what grip you get out of the tires.. while providing crisp predictable handling.

Widening the track, with all your FAP and KRC stuff, does it make your peak grip at that condition go up or down? (I would imagine it will go up) Of course this assumes you have pretty good tire data to begin with. If you don't, then I'd default to the assumption that in general wider track = less WT = mo' grip. Then again if it really provides a negligable amount more grip then I'd axe the idea in the favor of simplicity.

From the handling side.. you tell me. With your specific vehicle parameters (inertias, geometric and elastic roll resistance, roll/yaw coupling, blah blah) how is having FAPs that juggle around going to affect turn in? Yaw gain and damping? Phase lag?

If you've got some amount of analytical software, what does that change specifically do to the forces and moments applied to the vehicle, and can we say if thats a good or bad thing? If your theory is your change is good.. why specifically.

[StefanDavisBRC]

Hi people, I am a rally driver competing in the British Rally Championship and desperate to absorb as much technical info I can to relate to feelings within the car. I have done alot of research lately and a massive thing I have noticed is that to drive fast the biggest feeling I use is the feeling given back from using the tyres slip to the best effect. This however is cool and means you can be fast in any car with the feeling sensitivity but this does not mean that the car will be set to the optimum in order to provide more benefit from the tyre slip useage/grip available. Where can I read up on more fundamental stuff because I can only figure about 70% of this out without grounding knowledge. Great thread some wizz kids on her for sure.

[J. Vinella]

http://fsae.com/groupee_common/emoticons/icon_biggrin.gif

Oldie but a goodie.

(bump)

[ashutk]

Originally posted by kwancho:
Kevin -
Interesting. So, if you turn in on the brakes, and release them gradually on entry, the roll center height changes, and the geometric weight transfer changes. In this case, the roll center is moving up, increasing the geometric weight transfer (also, the geometric transfer is increasing purely because lateral acceleration is increasing). Any thoughts on how big of an effect the changing RC height has on turn in? Using Claude's logic, if weight transfer = grip, does that mean having the roll center move up on turn in = more grip?

edit: Olley's formula (if my math is correct):
height of RC change =
t/(2h) * (H2/R2-H1/R1)
where:
t = track
h = distance between UBJ and LBJ
H1 = height UBJ
H2 = height LBJ
R1 = length upper A-arm
R2 = length lower A-arm

Thus, to minimize the change in RC height, H2/ H1 = R2/R1. The lengths are usually dictated by camber curves and the lower one is ~1.5x the length of the upper one. Then, assuming the wheel center's at 10" (20" tire), H1 = 8", H2 = 12", which is way too close to the center of the wheel.
I was going through RCVD when i stumbled onto something and i quote "On race cars the height of the lower ball joint should be as low as possible on the upright for structural reasons". would anyone tell me what those stuctural reasons are??
If i keep the lower ball joint that low, olleys formula would give me absurd ratios
a high LBJ to UBJ ratio and thus higher ratios of the arms, much less than practicable ??
please try and explain

[Claude Rouelle]

Torque = Force x Distance, correct?

Think for example abut the braking torque. Is it easier to resist it with a long or a short arm?

[ashutk]

Thanks claude.
But going on with the above discussion, when i use olleys formula the UBJ and LBJ get very close to the center as in the above case 8" and 12".
8" would give large forces during braking.
what should i start compromising on ??

[Claude Rouelle]

Ashutk,

1. What do you call "absurd' ratios?

2. Do you "use" this formula or do you "practice" it?

Have you make numerical examples of this formula and compare the results to a simple 2D sketch,or Excel spreadsheet or simple lines intersection math or a drawing software?

3. Be practical: In a 13' or a 10" or a 8 " rim you will not have a lot of space anyway.

[MCoach]

Although I haven't seen it in a formula SAE application, there is always the concept of the 'tall knuckle' design. Honda famously uses this. If you are worried about forces 'going absurd' then this may be something for you to look into. Bill Cobb, who comments on here every now and then has mentioned that Honda has great K&C results, which may or may not be partially due to this design.

Honda Accord Corner (http://cdn.2carpros.com/automotive_pictures/55316_accordrotor_1.jpg)

Being practical, there isn't a whole lot of space inside these tiny car wheels which sometimes see cornering forces beyond 2Gs. You can imagine this dumps a lot of force into wheel assembly and control arms. Placing the upper ball joint above the tire gives favorable values for compliance and reacting forces. But, hey, what am I other than a bench engineer on a forum.

[Claude Rouelle]

Like this one (a racing version)?

http://downloads.optimumg.com/..._to_HBJ_distance.jpg (http://downloads.optimumg.com/images/High_LBJ_to_HBJ_distance.jpg)

[MCoach]

Yes, Claude, just like that!

I haven't looked too hard, but would have to guess this came from something competing in a near stock sedan class such as the Continental Tire Challenge or Le Mans GT2. I notice the radial mount location for the calipers as well as the lightening pockets around the bearings, which tips me off to think that this upright was cast for the race car and is not shared with the road going version.
I wouldn't think this is an FSAE upright due to the distance from one caliper hole to the other in relation to the total part height. It doesn't look the right proportions.

Either way, that's a better representation of how it is done in racing.

[Claude Rouelle]

Nope it is not an American car or a car racing in the US.

Two additional comments:
- I too like radial caliper mounting. More assembly stiffness.
- Nice brake and bearing cooling included in the upright design and manufacturing (look at inlet about 1/4 height from the top)

[MCoach]

My next guess then would have to be the Japanese Touring Car Championship/ Super GT class.

If not, I'm curious what other cars run such a knuckle design...

[Claude Rouelle]

Not Japan either.

[M. Nader]

I was wondering what that top part was to do, very interesting to know they use it for cooling. would want to know how air is routed inside though.

is it perhaps used in an endurance racing series? a lightened part for a Ferrari for a Mclaren 12c?

P.S i was also soon faced with a problem to solve by shifting the upper wishbone as seen, and although practical i didn't like the looks of it (who wants to see things lurking outside the wheel!) and tried to put everything inside the wheel but it just wasn't a good option at the time. fortunately though things did solve themselves for me at the end.

[ed_pratt]

This has got me really curious, and seeing as it's a friday....

Has the steering ball joint been significantly moved from it's original location?

I'm guessing it's touring car, but it doesn't look much like a focus upright...

[Markus]

VAG four-link front upright for a touring class series?

[Owen Thomas]

Big sturdy brake mounts and channels for rotor cooling, plus the need for hugely extended moment arm on the upper ball joint makes me think it's for something (relatively) heavy taking corners much harder than the thing was originally made for. I'm throwing my lot in with a touring series vehicle. Maybe some European GT series?

[Claude Rouelle]

Where this upright comes from is not important. It was just an illustration of upright pick up points outside the rim space ... and a reminder (call it teaser if you want) that all interesting or innovating technologies do not necessarily always comes from US, Europe or Japan http://fsae.com/groupee_common/emoticons/icon_smile.gif Let's focus on FSAE.

[M. Nader]

You really should give us more than that given the thought everyone put in! will help me sleep better tonight. http://fsae.com/groupee_common/emoticons/icon_smile.gif

My next guess other than endurance racing would be an offroad car, Baja style.

[nowhere fast]

Tokyo Denki has placed the rear upper ball joints outside of the rim in the past, so it has been done before in FSAE.

Side note: I really liked the few Tokyo Denki cars I've seen. They are possibly the most realistic 'car for the weekend autocrosser' that I've seen in person; simple, light, compact.

http://farm9.staticflickr.com/8241/8615262397_38dd9cbb60_c.jpg