Sway bar design

[jonesy_HB]

Hello - this is my first year on a formual student team and I have been given the task of designing adjustable sway (anti-roll) bars for both the front and rear suspension. I have read RCVD chapter 22 and have worked through the calculations in the SAE manual for designing torsion bars. I have come up with a basic design for the rear suspension, but I was wondering if anyone could recommend a material for the bar/tube. Any other information or recommended reading would be greatly appreciated.

Thanks

[TomF]

Well, the best way to go is usually Aluminium or steal, depending on what stiffnes you want. Just play around with your diameters and Torsional stiffnes. Just take metal since than you have elastic behavior, which can be modeled.

Good luck!

[GTmule]

Generaly some 4130 tube works out pretty well. I can't get my mind around using aluminum, but that's mostly because steel dubing has appropriate GJ's in pretty standard sizing. I wouldn't wanna use aluminum rod or tube, as to get enough stiffness, you might end up with inch and a half OD tube, or 3/4 bar, which is just kinda cheesy.

so yeah, 4130 is what I'd use.

[Denny Trimble]

To paraphrase some of Carroll Smith's writings on the subject:
-If you don't heat treat 4130 (quench + temper), you might as well make the parts out of mild steel.
-If your swaybars yield, bad things happen, so make sure they don't.

Heat-treated 4130 has worked well for us.

[Matt Gignac]

I would stay away from aluminum because of fatigue considerations. Mind you, it could probably still be done, but I think steel would still be better. Then there's the issue of packaging the torsion bar sine it will have to be larger.

[GTmule]

<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Originally posted by Denny Trimble:
To paraphrase some of Carroll Smith's writings on the subject:
-If you don't heat treat 4130 (quench + temper), you might as well make the parts out of mild steel.
-If your swaybars yield, bad things happen, so make sure they don't.

Heat-treated 4130 has worked well for us. <HR></BLOCKQUOTE>

Agreed

The yielding would be fun to figure out.....a bit of wedge, anyone?

[jack]

from what i remember UWA didnt have a rear ARB, and they won design, twice. was it taken off the car when i saw it? did they just have their suspension that sorted out? anybody else notice this?

[fade]

you could get your motion ratios to work so that the added roll stiffness usually supplied from the roll bar was instead added from a rising rate rocker. you would have to watch your ride rate though.

[Denny Trimble]

Our 2003 car went to Detroit without a rear ARB because it worked better with the rest of the suspension setup. It's common on some race cars to eliminate the rear ARB, as a tuning tool (adjusting the stiffness to zero).

The car only cares about ride and roll stiffnesses at each end (expecially front roll resistance vs. rear roll resistance, as a ratio). You can get there with several spring and bar rates and nonlinearities.

[fsae racer]

<BLOCKQUOTE class="ip-ubbcode-quote"><font size="-1">quote:</font><HR>Originally posted by Denny Trimble:
Our 2003 car went to Detroit without a rear ARB because it worked better with the rest of the suspension setup. It's common on some race cars to eliminate the rear ARB, as a tuning tool (adjusting the stiffness to zero).

The car only cares about ride and roll stiffnesses at each end (expecially front roll resistance vs. rear roll resistance, as a ratio). You can get there with several spring and bar rates and nonlinearities. <HR></BLOCKQUOTE>

I hear what your saying denny, as F/R track will hugely affect the ratio that you are referring to. Further, that ratio definitely has a huge effect on the overall balance of the vehicle. However, I still could not imagine having a well balanced FSAE car w/o a rear arb. Would you care to share why you think it was fairly well balanced w/o the rear arb?

[Mark Bacchetti]

Along the lines of what Denny said...

Look at sway bars as an extra spring for fine tuning your steady state lateral load transfer distribution. A suspension designer can choose to design a suspension without a rear sway bar as long as you can package a sway bar (with enough adjustment) that will give your driver a chance to tune to a desired steady state balance.

It ultimately comes down to how you design the suspension. Milliken has some great simplified equations for the designer to achieve this during initial design stages.

[Denny Trimble]

Nick,
I can see why you're puzzled, so this info might clear it up for you. The car had a 45:55 weight distribution, 51" front track / 48.5" rear track, and equal and linear front and rear installation ratios. But, our only spring rate choices for our Fox shocks were 225 and 300, so we ran 225's in the front and 300's in the rear. So, there was already a big difference in front to rear roll resistance ratio due to the springs. We tested with a few different rates on front and rear ARB's, and ended up happiest with none in the rear. For 2004 we had better shocks with springs in 10lb increments, so we did end up running a rear ARB.

[IsheeM]

Since we are one the topic of ARB. I have a question regarding what roll rates to achieve. More specifally, in Milliken & Milliken p.585 he begins his iterations by assuming front and rear roll rates of 1222 lb-ft/deg and 873 lb-ft/deg respectively. He doesnt explain why he assumed these particular roll rates and I cant figure out what roll rates to assume to begin the iteration. Obviously the initial assumptions dont change the calculated roll rates very much at the end of the iteration, but in order to calculate ARB rates I need to know what roll rates to assume. In an effort to figure out what roll rates to assume, I followed his second simple example and calculated a roll rate for the total car as being about 130 lb-ft/deg using a roll gradient of 1.5, but this number is low compared to the figures I get for front and rear roll rates at the end of my iterations. So, in case you forgot the question, what front and rear roll rates should i try to achieve or how can i calculate them.

Another problem I encountered was that in my iterations I am getting a roll gradent of about .75 which I think is too low. But that may be another issue.

Thanks,

[mario_rbr]

What about using :
Roll Rate = (Track Width^2 * Ride Rate) /2

where
Ride Rate = (Wheel Rate * Tire Rate)/(Wheel Rate + Tire Rate)

where
Wheel Rate = Spring Rate * IR^2

Do you guys mind to post here the values you´re getting of roll rates, or Front/rear ratio?

[Gareth]

I believe UWA's anti-roll is built in to their hydraulic system. That was (may have changed) one of the downsides of their design because it wasn't simple to change the roll distribution of their car. I'm thinking they had 'classical' roll bars in parallel to get around this for tuning. Once they found their 'magic numbers' then the hydraulics could be rebuilt with the better distribution. Is this right? And just because UWA won design with something, doesn't mean it's the best way to do it, not to discredit UWA. I'm sure they'd agree that you must have your own reasons for your design choices and that's what matters.

I found it useful to look at how your LLTD changes with pitch/heave. The UW car had shuttle bars in 2004 and 2005, which had fairly poor kinematics. I found that reasonable MR's were possible, but they were very ride height sensitive. Essesntially, as the car pitched the LLTD changed in ways that were undesirable. This was discovered on the '05 design and I showed those plots in design...which got me an 'F-smart' from Mr Rouelle...one of my favourite compliments. ;-) That's what drove our change to U/T-bars in '06. More recently I have realized that the RCVD chpt 22 calc's are incomplete as they lump the tire with the springs. Assuming you have some sense wrt the kinematics, the only tricky part is combining the springs in the right order. The springs and anti-roll bars are in parallel, which sit on top of the tires in series. It's pretty obvious once stated, but can introduce some error into your calculations. I guess all I'm saying is that you can have infinite kinematic roll stiffness (say RC's at the CG) and still have chassis roll through the tires. With a load distribution that's not 50/50, this will create some diagonal load transfer. Finally, make sure you have good 'roll center parity' (thanks Aaron). That just means that your RC's move vertically with your CG during heave so that your roll couple stays constant.