Matlab for tire data/suspension design

[ubrben]

For ss geometric load transfer, your fastest and most accurate simplification is instant centers.

But with instant centre you need an estimate of the lateral force split between the inside and outside tyres - which is why you need a tyre model, to solve geometric load transfer iteratively.

The roll centre is a useful rule of thumb to get baseline ride and roll rates and a lateral load transfer distribution. I don't think Tom or I are suggesting it's "correct" in some absolute sense, but is it useful provided you understand the limitations? Without a doubt yes.

As George Box said - "All models are wrong, some models are useful". As a tyre engineer myself along with Tom, we can probably name a number of reasons you're full 6x6 matrix is "wrong" in some sense because it won't consider tyre temperature for example. Any model is an abstraction and you have to sometimes work on the pareto principle. 10% of the effort yields 90% of the answer. It applies to using a bicycle model in a lapsim and it also applies to roll centres in vehicle setup.

Ben

[RideRate]

Not sure that 'roll center' location is pointless. If I do a lateral force compliance test on an SPMM I can get the dFz/dFy slope at each corner. I can summarize both those rates at each axle with a (Y,Z) co-ordinate at their intersection ("roll center") assuming I know the track width as well. By using the front and rear axle co-ordinates together, I can summarize the dFz/dFy of all 4 corners simultaneously with a line ("roll axis"). I can do this at any combination of ride height and chassis roll displacement. That seems like it would be valuable.

What to DO with those relationships in a simulation, and how to resolve it into something useful... is admittedly beyond me. I'll admit I'm a bit blank on "just use instant centers" as well.

Mostly, there are quite a few interpretations on the whole roll center thing, anti's, geometric vs elastic WT, etc. You point this out between yourself and Claude. The only thing in common between all of them is I have never seen any real data to support which is most applicable, or most correct! In any event, I believe quasi steady-state simulation gets ya a long ways in designing a car, and in that case I just roll everything together into a global "TLLTD" variable where I'm not too concerned in the individual contributions. I'd like to nail dynamic as well though.

What is your background, by the way? Just curious.

I've never played with a suspension measuring machine, unfortunately, but the slope of the dFz/dFy I can see being very important, obviously. The intersection I'm not so sure is good information. Do you have unlimited access to one of these machines? I'd be interested in some tests as it would go along with a few of the things I'm working on.

I'm not real interested in revealing much of my background as I don't think it's important. I mostly grew up a fabricator. I started by dabbling in the off-road world and it's where my heart will always lie. That bled over into pavement. All this instilled an interest in dynamics, mechanics, statics, vibrations, fluids, etc. I have one b.s. from an accredited university and am working on another b.s. as a full-time student. Between degrees I worked at vehicle research labs and a transportation testing facility as well as a fun stint in the fab shop of a noise and vibrations company as my first job out of college the first time. So there's a little. Does that help?

[RideRate]

But with instant centre you need an estimate of the lateral force split between the inside and outside tyres - which is why you need a tyre model, to solve geometric load transfer iteratively.

The roll centre is a useful rule of thumb to get baseline ride and roll rates and a lateral load transfer distribution. I don't think Tom or I are suggesting it's "correct" in some absolute sense, but is it useful provided you understand the limitations? Without a doubt yes.

Yes, you do need an estimation of lateral force split. But you don't need a tire model to APPROXIMATE that. If you use a rollcenter, what is its assumption on relative forces between right and left? Now, is the rollcenter's assumption or a rough approximate using llt going to be more realistic? It's not even a contest. The rollcenter automatically neglects all these things that could at least be approximated more to reality.

I don't need rollcenter to get baseline ride and roll rates. The analysis will still be much more accurate using instant centers for each corner rather than trying to combine them with the rollcenter. Just as simple and more accurate. What more could you ask for? IMO, it's just a poor tool period. There are limitations in every model, but some more so than others.

I understand now I am dealing with tire guys. So I know what you two are knowledgeable about and what you find to be most important. Believe me, I totally understand that. However, I think things can be very well figured without detailed tire data. Using tire data correctly becomes very important to fully maximize car grip and performance but that's not what these calculations are trying to do.

[Jersey Tom]

I do have 'unlimited' access to an SPMM. With regard to public use, it is not free. If you're interested though, feel free to PM. If it's commercial it's one thing. Academic may be another.

You are indeed dealing with tire guys.. or at least one tire guy and one tyre guy! But we both do our share of instrumented vehicle testing and simulation, with some level of hands-on background.

I think we just have to clarify what you mean by saying that RC's are malarkey. With the SPMM I can measure the dFz/dFy slope for any corner of the car at any chassis orientation. The intersection of those two slopes is important only in that with the one point, you define both slopes instantly (knowing the track width). I call that intersection the force-based roll center. The nomenclature may or may not be entirely correct, but the way I figure it should define the geometric load transfer and jacking properties of the vehicle. All you have the plug in are the appropriate tire forces (either by a rough guess, or a tire model)

No?

The accuracy of kinematic RC's is a separate question

[RideRate]

If you only ever use the individual slopes to that intersection, I say you're calculating things accurately. Reason being the slope to that intersection is also the slope to the instant center. That is all that is important. The intersection location, height, lateral location, etc. isn't important. Hence if you design for roll center lateral migration you're missing out. If you sum forces at the roll center location, you're missing out. Stick to the slopes and that's fine, use the further 'common' applications of the roll center and you're not sticking to statics and dynamics principles. This goes for force or kinematic based. Make sense?

[Jersey Tom]

So in that event, the roll center location (vertical and lateral) isn't completely worthless, so long as it is only used in defining slopes to each IC, after which appropriate forces can be applied.

Agree?

[RideRate]

I agree with that, sure. But that's not using the concepts associated with a rollcenter, you're now using instant centers alone. So, why waste time calculating the rollcenter location when you have to find the ic's first anyway? So it's worthless in that it's not telling you anything you don't already know. Again, it's the other concepts associated and applied to the rollcenter that are the issues.

But, I think you're seeing where I'm coming from.

[PlatinumZealot]

I have to interrupt, where can i get these tyre graphs? (or spec sheets or whatever)

Right now I am attempting to make a suspension model. My only memory of dynamics is 3 years old.. so it is very blurry.

What other areas does one need to study (remember in my case)to make an accurate enough suspension model?

[Jersey Tom]

Tire data... try asking your tire company. Or, you can get it tested yourself. However, it is not cheap.

Bear in mind, you really don't need tire data to design a pretty good suspension.

[Belatti]

Bear in mind, you really don't need tire data to design a pretty good suspension.

what if pretty good is not enough?
what if I want the opposition look like turtles?

[Jersey Tom]

Making them look like turtles is all relative. I still think you can design a very good suspension without tire data at all. The kinematics are only a small piece of the puzzle. Even if you had an identical kinematic package to a competitor you could gain huge performance just by reducing compliance.

A lot of the usefulness in tire data, IMO, is in figuring out what baseline setup you need at a track... springs, bars, ride height, static camber. With enough track time you could get to the same point. For some series where there is little or no test time before a race, it makes a big difference!

Just gotta be careful. The big question is... how good is the tire data you've been given? If it's not accurate, then you can be designing around garbage, or just chasing fake trends.

Plus, there's a lot more to "race speed" than designing to max theoretical from tire data. A lot comes down to experience and knowing what you can get away with. If you aren't careful you could set yourself up for a lightning fast qualifying setup, which then loses all grip or blows the tires out after a handful of laps.

[Belatti]

Just gotta be careful. The big question is... how good is the tire data you've been given? If it's not accurate, then you can be designing around garbage, or just chasing fake trends.

In that case, I should forget about using tire data arround here. Unless a client wants to send a bunch of tire sets to Europe or the States and pay a test for them.

[fastback33]

So, i've come up with a paper doll game of the rear suspension. How do i upload the pic? Also how do i get it to do a system in roll and bump?

[Jersey Tom]

You can use a free image host like photobucket.com for uploading images.

[unsprung]

im asked to design the suspension for FSAE car. right now we know the track, wheelbase, the estimated cg form our prototype(which was not designed,just fabricated and modified over and over!! ). i need to give the chassis guys the suspension mounting points. i calculated the roll axis, just wanted to know how do u decide upon the angle between the "A" of the wishbone arms??