Estimating wheel travel during roll

[Jrud]

I wouldn't think this would give me as many problems as it has. I have a working spreadsheet that takes inputs, i.e. weights, springs, geometry factors and calculates wheel rates, damping coefficients, roll rates, tlltd, etc.

Most of the calculations have been verified, but I'm having trouble getting the wheel travel estimates under cornering to work out. Right now I'm working with a suspension that has softer springs and a very stiff bar in the front, soft bar in the rear. I've worked out all the wheel rates, roll rates and they match the numbers that I have from another source on this car.

I have a section on the spreadsheet that calculates suspension position at static, meaning how much the spring/damper compression, tire compression, overall chassis deflection, force in the spring.

Following this section, I have a set of calculations that are supposed to calculate wheel travel of the outside wheel at 1g. I am starting with the static wheel loads and adding the loads calculated from the lateral load transfer section (which I'm pretty sure are correct). I subtract the unsprung weight from this number, this is assuming that unsprung weight transfer is small and it is about 10% of the total weight transfer, so I think it should be alright. The sprung corner weight at 1g is divided by the wheel rate including the one-wheel bump arb rate to estimate the wheel travel relative to the chassis.

I am comparing this number to the estimated wheel travel based on the roll gradient. And they are greatly different. With the first method of calculation being much greater than the second.

So what am I missing? If I have the wheel loads, shouldn't I be able to relate that to suspension travel if I have rates and installation ratios for the springs?

If anyone has any ideas or any suggestions, I would appreciate it. I can provide more numbers and calculations if they would help. Thanks guys.

Tim

 

[GregLocock]

I think you need to provide the calculations.

I'm certainly not convinced that the single wheel bump arb load is correct for a start.

Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[Jrud]

Greg, what do you mean, about single wheel bump arb load? Do you think I'm just neglecting part of the arb's contribution during cornering?

As for the calculations, I'll start with the arb stuff.

I was provided a bar rate of 621 lb/in. Which is just the amount of force to move one end of the bar up 1". So to get the one wheel bump wheel rate. I used the provided bar rate time the ARB IR^2 and added to the previously calculated wheel rate.

My lateral load transfer calculations should be right. I'm using roll rates that were I calculated that match the ones that I recieved with the other vehicle information. And the TLLTD is the same as I recieved.

Alright, I think this is going to take too long to explain and probably easier for everyone if I just attach the spreadsheet. I cleaned it up so it's a little easier to follow and got rid of all the stuff that doesn't matter for this discussion.

http://www.theoryinpracticeengineering.com/tech/engtips.xls

The area of interest is at the bottom of the first column. Everything else should be pretty easy to follow, so I hope this helps make sense of my problem.

Tim

 

[GregLocock]

Unfortunately that style of spreadhseet you have written is practically uncheckable except by the writer. I'm not saying mine are any better.



Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[JsTyLz]

It looks like you have figured ou the deg/g, just divide the track width by two and use the sin functin.

So you have X [deg/g].
you multiply by 1 [g] and get Y [deg]
then SIN(Y)*(track/2) = vertical deflection

I am not sure if this is what you are looking for???

 

[Jrud]

Greg, Well then you should've seen the one I've been working with. I cleaned that one so hopefully it would've been easier to follow.

Well what would be easier going over the calculations of just discussing the method of figuring this out.

The thing is, I think I remember verifying my calculations when it was a suspension without a sway bar. So with throwing the sway bar into the mix, I'm missing something somewhere.

Jstylz, I have that and I have worked out those numbers. But I am also trying to work spring/wheel/bar deflections from lateral load transfer. And the numbers from these calculations aren't matching the deflection based on the roll gradient.

 

[hutch325]

Yeah, I agree that spreadsheets are nearly impossible to check. I tried to make one that would model the 3D kinematics but I gave up and started over in Matlab.

A couple things you could do:
- make your calculations done in a macro instead of in formulas embedded in the cells
- name your values, that way your formulas would look like "=mass*acceleration" instead of "=$C$4/B12" (for example)

 

[Jrud]

I think Matlab is just as hard to check if you aren't the one who wrote it.

I don't know, I'm used to the spreadsheet, so it works fine for me. It started a lot smaller and a lot simpler and then sections got added on. So the previous sections were right and didn't need to be touched anymore. There's only about 10 or so cells that need to be filled in to generate all the other numbers and plots that I have in the spreadsheet I'm working out of.

And everything else was working right, it's just this part that's throwing me off. I'm going to try and work everything out by hand to see where my mistakes are and maybe I'll post the hand calcs and someone might be able to help and see my mistake.

Tim

 

[Fabrico]

So what am I missing? If I have the wheel loads, shouldn't I be able to relate that to suspension travel if I have rates and installation ratios for the springs?

That would seem a bit one dimensional. There are two kinds of roll that use up suspension travel. One is roll that pivots, more or less, about the CG. The other is roll that pivots about the opposite side wheels.

 

[Jrud]

But I'm talking about steady state load transfer, like a skidpad situation. And trying to estimate spring/bar/wheel travel. And for right now, I'm just interested in the outside wheels.

I'm not sure exactly what you're trying to say. I didn't split up elastic and geometric load transfer, but like I said earlier, doing that wouldn't account for the amount of difference that I'm seeing in the numbers.

So the roll that you say pivots about the CG, you're talking about elastic load transfer. And I'm guessing when you say more or less about the CG, you mean between the CG and RCH. But the "roll that pivots about the opposite side wheel" what do you mean by that? Geometric load transfer? Or something else?

Tim

 

[sreid]

Since you clearly know what you are doing, is it possible you have the sign wrong for whether the roll bar adds or subtracts force from a spring?

 

[Jrud]

I don't think the sign is wrong, I think the magnitude might be wrong though. I'm trying to essentially model the roll bar and spring together as another linear spring. So I think I'm underestimating the amount of spring rate the roll bar is adding.

At first, I was using the rate of the bar at the endlink and modifying that to get the arb wheel rate for a one wheel bump. I was at first using this number in the deflection calculations, but it's not correct. Because during roll, as the one wheel goes up 1", the other wheel goes down 1", so that would double the spring contribution of the arb. But I think I tried it doing that and the numbers were still coming out wrong.

I still haven't had much time to really go through this, fine tuning my model doesn't pay the bills as well as I would like, and this isn't a critical part of the spreadsheet, so I've been putting off running the numbers.

Tim

 

[Fabrico]

Because during roll, as the one wheel goes up 1", the other wheel goes down 1", so that would double the spring contribution of the arb. But I think I tried it doing that and the numbers were still coming out wrong.

That goes back to what I was mentioning. As the body rolls, it's not necessarily one wheel goes up and one goes down in equal amounts. Depending on CG etc, etc, it can be down only, or any ratio inbetween.

In other words, a given body angle does not equate to a specific travel number.

 

[Jrud]

But you never clarified for me what exactly you meant? Refer to my post directly after your first one, there were a couple questions I had in there.

And I think you took that a little out of context. 1" is arbitrary. I'm just trying to figure out the contribution of the ARB to the wheel rate during roll.

Is assuming that on each axle, when one wheel goes up x", the other goes down x" during steady state roll? I know that the front and rear won't go the same distance, that's the whole reason I'm trying to figure this out. The roll gradient only tells you so much, especially with greatly varying roll rates front to rear.

Could you elaborate a little more on what you're trying to say? I think you're either underestimating my problem and what I have already gotten done. But you may be talking about something I haven't thought about or something I'm simplifying or assuming to make things simpler on myself, but it's something I can't because of the inaccuracy of it. So which is it?

I'm trying to break the car down into 4 corners. So for sake of argument, let's say I'm only working with a 1/4 car model, but I have information on CG, geometry, RCH, etc. So I know all the data about the car and now I need to model SS 1g cornering on just the front 1/4 of the car. What assumptions can I make? Can I used lateral load transfer in lb/g? And add that to the weights I already have for the front corner? And then what is the spring rate? Say I know the amount of force it takes to move one end of the bar up 1" while the other end is fixed. Would I use double this number and then modify it by the installation ratio of the ARB to get the wheel rate of the ARB? Would I use it differently? Am I missing some conversion?

I mean isn't that how a sway bar works and contributes in this situation.

Tim

 

[murpia]

In a situation where a 'rule of thumb' was required to turn an anti-roll bar rate into a wheel rate, we used the form of fixing one end of the bar and calculating the increase in wheel rate by moving one wheel in bump only. That way we could quote a N/mm wheel rate for the bar comparable to the N/mm we quoted for side springs. The same applied to 3rd springs (active in heave only).

Of course this is only a first approximation, especially if rising rates are involved anywhere. But it made it possible to look at a setup and e.g. see we had about twice as much side spring contribution to wheel rate as bar or 3rd. I wouldn't have used those numbers for a 1/4 car dynamic analysis, though.

Regards, Ian

 

[hemipanter]

I might tell the obvious here, but when the car roll we could think of the bar as beeing welded solid to the chassis in the middle.
Regards
Goran Malmberg

 

[hemipanter]

Another thing, I did made an eperiment with swaybars a few yeras ago. Moost of the time a swaybar has its arms in an angle to the twisting section. This is changing the spring ratio. Making the bar to act a litle as a transverse leaf spring. So, I made a few bars with the same lenght twisting section but diferent angle arms out to the wheel to see what acually happen to the rate. If anyone is interested in the outcome I could see If I find the paper showing the numbers and drawings.
Regards
Goran Malmberg

 

[Fabrico]

I'm only pointing out that right and left suspensions can, and most likely will, move different amounts in a steady state turn.

Some cars rise on one side much more than they sink on the other. Some are the opposite. Some rise on both sides :-0. Some, perhaps, rotate perfectly in place. Speed can change suspension jacking effects.

It is unlikely that the simple calculated movement of one wheel will provide an accurate picture.

 

[hemipanter]

Fabrico, right the roll situation has been discussed a lot since the movement of the chassis is a product of both roll and jacking. And dependent on suspension geometry the roll and jacking will show more or less agressive changes dynamically. This may appear confusing as the two movements is hard to separate from one another.
Apart from many others only using computers I am also very much in to making physical modells, which I also have done in this case. On such model it is easely seen that we deal with a jacking and roll movement, and that the sum looks like one side has a different deflection that that of the other during lateral load. However, setting the model up with long parallell A-arms vill eliminate this effect and both sides is getting the same deflection. It is also possible to set the model up to rise on both sides.
I might have som images for you to look at if you want to.

Regards
Goran Malmberg