Effect of changing torsion bars compared to sway bars?
 

Is there an easy way to decide how much roll stiffness you will get from a bigger torsion bar as opposed to a bigger sway bar? Intuition suggests that the sway bar would produce more anti-roll for a given change in strength than the same change in a torsion bar. Can anyone offer an explanation?

That's the question: which offers more roll stiffness per % of size change.

Fatter torsion bars create a higher "wheel-rate" ( think "spring rate"...but measured as forces acting on the wheel), so the thicker bars will help cornering AND braking/acceleration motions ( pitch and dive).

Common convention for other cars says adding fatter anti-roll bars doesn't affect ride, whereas adding fatter suspension springs does. Common convention for our 911's tells us the opposite.

Because the "spring" effect of either is based on diameter ( "D" to the fourth power, quite sensitive to diameter changes).... AND....also the length of the lever arm the operates on either of these types of springs ( anti-roll bar or torsion bar).....it is not a simple matter to say, "Im going from "X" to "y" and expect such-and-such change.....not that cut-n-dry.

Sorry can't give you %%'s but I can give you my seat of pants feel:

I have uped my sways to the '86-89 thicker front and rears. It did improve the sway but I gained some looseness when I hit corners to quickly. I was also worried with the OEM '85 T-bars it may stress the sway bar mounts too much and snap the mount welds. Like many have experienced here.

With the combination of Strut Bar, Thicker Sways and Larger T-bars 22/29, she rides like she is on rails. I do not worry about snapping the Rear Sway Bar Mounts. I also learned how to control her thru the quick corners.

Caveat, albeit stiffer but I carry alot of stuff in my daily so it isn't that bad. At the track when I unload her, she is pretty stiff.
Jim

Wil,
Looking thru old notes I found a nice piece someone sent me a long time ago. I'm sorry to say I don't know who did it. Perhaps it was you? Or perhaps if he sees this he will chime in? MYSTERY SOLVED. BURGERMEISTER DID IT.
In any event, if you look at the items highlighted in orange by me, you will find my summary of this author's view that the front sway bar contributes 37% of the total front roll stiffness and the rear sway bar 23% of the total rear roll stiffness... given all of his assumptions including size of the various bars etc. Does that sound like it's in the ball park? I hope you can read these photos of the author's spreadsheet. I can send you or anyone else who is interested the Excel file if you want.


http://forums.pelicanparts.com/uploa...1300651216.jpghttp://forums.pelicanparts.com/uploa...1300651413.jpg

Where is Fleigler?

I suspect that was Burgermeister who sent that through to you. Based on much of the work he did I can estimate the following:

Increasing T-bars 3mm diameter front and rear = approximately .6 degrees less lean at 1 G cornering.
Stabilizer bar diameters would need to be increased by roughly 2.5 mm front and rear to have the same -.6 degrees per G effect.
Swapping out the bushings with something much stiffer could also have a similar effect.
Tire compliance is much smaller- going from 16s to 18s would probably only get you in the range of .2 degrees.

A stock car starts out rolling about 4 degrees per G. Many race car setup books tell you to aim for roughly 2 degrees total roll on cars such as ours.

This is before studying the spreadsheet posted above:

Well, by my calcs, the motion ratio for the front anti-sway bar* (*=aftermarket) means it twists about 108.4% of the torsion bar for a given wheel displacement, so the motion ratios are pretty close.

If the sway bar is measured in # per degree, that usually means that half of that degree comes from one wheel, and half from the other. The torsion bar being measured in # per degree means that all of that degree comes from one wheel. This means that for a given wheel displacement, the sway bar contributes twice as much force as the torsion bar, or that the front sway bar contributes 2/3 of the front roll stiffness.

Torsional stiffness # per radian is JG/L
J=(pi/2)r^4
G is about 11600000psi
effective L for the sway bar is about 36 inches
effective length for the torsion bar is about 22 inches, IIRC.

For actual stock front carrera suspension diameters I have the stab bar contributing half of the front roll stiffness, 40% of the rear roll stiffness, but those are not from my own calculations/ measurements. Bushing rates and other variables have a big effect on the actual contribution beyond the bars themselves.

It is truly amazing how much and in such detail some of you know about these cars. There cannot be many issues that have not been covered on this board.

Yeah, but it boils down to some seat-of-the-pants decisions. Sure, if you have data from a track and want to do the math, you can tune the car that way.

Or, you can decide on how the car is going to be driven and ask someone who has set up dozens (like Chuck Moreland or Dave B from TRE) for some practical guidance.

You will need to know the car, the weight, the ride height, the track width, the wheel sizes and what tires. Then you need to know the type of roads or track surfaces where the car will be used.

Personally, I'd rather make the tires stay on the surface rather than over focusing on body roll, whether treating it with t-bars or sway bars. Remember what Mark Donohue supposedly said, use heavier spring rates or heavier sway bars; you don't need both.

Of course that would be relative to what you start with.

Apples and oranges. Both ARB (sway bars) and T-bars (spring rate) influence body roll but the underlying reasons for the effects are far from similar.

Milt, I was thinking the same thing, give the experts my parameters that I want to run and let them sort it out. That then is the starting point to fine tune. Most of the heavy lifting has been done. It is a wonderful science.

Agreed. Unless you are very sophisticated you are far more likely to get yourself in trouble figuring it all out entirely on your own. Understanding the issues, however, can be very useful when working through the various options.

Max,
Had a chance to study the chart above yet? The still anonymous author of that work, possibly Burgermeister, suggested the front roll bars contribute 37% of the antiroll. If I understand you correctly, each side individually contributes twice that number or 74%...again, given his assumptions about sway and torsion bar sizes etc. So you and he are in the same ball park?
Jim

i like your last comment.

skid pads and lap times are a good way of tuning.

there are a lot of good books out there. its an old book, but start with "how to make your car handle"

You realize that the chart you're referring to has after market 24mm front and 33mm rear torsion bars, but stock swaybar sizes? That's why you're getting so little contribution from the swaybars- was that the intent?

Change the Tbars to the stock 19 and 25 and see what results...

fwiw, the early RUF 911s went to larger swaybars only. ...for Nurbergring.

What bugs me about these discussions is that rarely does anyone talk about --what should be an obvious bit of the discussion- That is; what are the road/track conditions, and the speed you intend to move over it?

That is, the whole idea of suspension is to balance the needs of the car/driver against the height variations of any given piece of road. ... a slight dip at 30mph, can become an abrupt hit at 130.... unless the suspension is tuned to those conditions.

..airport cone-killing autocross does not have the same requirements as rally racing.

edit, looks like Milt was on top of this.

Good points all. I agree that the only real way to find out what you want is to drive the car on the track and check results and "feel." When I started this thread though I didn't give a fig about real world conditions, wanting only to understand the relative effects of a change in the sway bar vs a change in the torsion bar. So far, the answer seems to be that the sway bars typically do 2/3 to 3/4 of the job, torsion bars the rest.

No. Sway bars are 50% to 40% of the job, torsion the rest (on a stock car).

I see where you got 2/3 from Flieger (though I think he's overestimating), no idea where you got 3/4 from.

My car is unique in having 22mm front torsion bars and a 22mm aftermarket through-body anti-sway bar. That must be how the 2/3 makes sense to me.

You would take the one wheel displacement, divide by 11.5 inches to get radians of twist on the torsion bar. Multiply the JG/L for one torsion bar by that radian value. Divide by 11.5 inches to get the change in force on the wheel.

Take that same one wheel displacement and multiply by 13/69. Multiply the JG/L for the sway bar by this other radian value to get the torsion. Multiply by 6.5/69 to get the change in force on the wheel.

Add the two forces to get the weight transfer on one wheel. The other wheel has that amount of force taken off of it. The sway bar is just connecting the two torsion bars. If the sway bar had nothing to react it on the other side, it would not add any force.

If you just use 1 inch of wheel displacement, you should be able to see how much the sway bar contributes relative to the torsion bar.

For extra accuracy, take better measurements of the lever arms and use sine of the wheel travel/lever arm. For small angles, sin=tan=angle itself

Ok. I've spent so much time staring at my copies of the spreadsheet that I forget that it is a spreadsheet. Calculating with 18mm sway bars front and rear, and 22f/26r torsion bars, we get sway bar contributions of 26% front and 25% of the total; that is the sway bars do 1/4 of the total roll resisting and the torsion bars do 3/4 of the job in this particular case? And, per Max, these numbers should be doubled at both ends to give an accurate picture?

If the spreadsheet is working correctly you should not need to double those numbers. 22/26 are stiffer than stock Tbars (suspiciously soft in the back relative to the front, are you sure they are not 28mm?). 18mm sways front and rear are very soft swaybars. Using those numbers I get results that agree with what you posted, but as I say those are weird settings. You probably want closer to 50% swaybar, 50% springs.

Torsion bars are 22/26, were 21/26 until last weekend. I am a fan of giant understeer because I spend a lot of time at Willow Springs trying to go fast around big sweepers. My best times (in an earlier car) there were with 23/26's and I may get there yet. After a little experimentation with this new (to me) setup though, and given the relatively modest contribution of the sway bars to date, I'll probably kick up the front bar!
As the spreadsheet isn't mine, I'm not sure if Max' doubling point in it is addressed or not. If these numbers agree pretty much with your calculations, and given the author's obvious capabilities, you may well be right.

I would encourage you to get adjustable swaybars to tune the handling balance you want, not use the torsion bars as much. The main springs do the important job of setting suspension frequency and maintaining ride height, while adjustable swaybars let you tune understeer/ oversteer and body roll separately.

The stock bars result in essentially the same spring rate front and rear. At 22/26 you have the front 30% stiffer than the rear. This can lead to the rear collapsing under acceleration, lots of inside wheel lift, etc. I'd suggest you try 22/27 or 22/28 (front still 6% stiffer than rear).

You can then use adjustable swaybars front and rear, the equivilant of roughly 25/ 23, to tune in all the understeer you want without collapsing the rear, confusing the shock settings or getting funny front to rear resonant frequencies.

I haven't seen any of those symptoms you describe, even at 23/26; the most dramatic effect of such settings so far for me is hard turning at lower speeds. Which I can live with if I can add speed in the faster big sweepers. But my next move will nevertheless likely be to a bigger rear bar together with an adjustable front. Thanks for your very helpful commentary, Jim

All the technical mumble jumble aside, I can promise you that you will get a much better handling car (faster lap times) on the track if you change only TBs vs only changing SBs.

Thank you. The best suspension setups rely very little on sway bars other than fine tuning and adjustments for changing conditions and different drivers. All the calculations might be fun but the bottom line is controlling roll with sway bars rather than spring rate is not the best method for a high performance car. Over loading with sways might feel real good at low G but the car will become a handful at high
g. What it will do can also be changing as you go through a corner. Under steer on entry and over steer mid corner or on turn out or vice-versa. T bars, sway bars and dampers all have to work together as a system.

The doubling is for looking at the roll stiffness. The spreadsheet has values I did not measure- like the stiffness of the A-arm tube, and the G50 (which I do not have) torsion tube. His bars are also different sizes, and the front bar is an under-A-arm type (he mentions an angled portion) so the calcs are different.

Bottom line is that the change in vertical force on one wheel multiplied by the whole track width gives the contribution to roll stiffness for that end of the car.

The change in force comes from the torsion bar and sway bar, but for a given vertical wheel displacement, the sway bar twists 2*108% more than the torsion bar. The 108% is my estimate of the motion ratio between sway bar and torsion bar for a given wheel displacement. The 2 is because it has one side going up and one side down.

This means multiply the sway bar stiffness by 2 times the angle and the torsion bar stiffness by just 1 times the angle. That gives the change in force on the wheel, and multiplying by the track gives the total roll moment reacting the car's inertia.

So, changing the sway bar size/stiffness will have more bang for the buck than changing the torsion bar stiffness in terms of reducing body roll. You get twice the force if you add a given amount of stiffness to sway bars than to torsion bars. (Note stiffness means JG/L, J=(pi/2)r^4 not jut adding diameter.)

I thought that was the meaning of the thread from the title.

thanks Max.
Burgermeister has confessed he is in fact the author of the spreadsheet I show at the top of this document so the mystery is solved.

Apparently when the Audi team showed up at pikes peak and handed the car over to Bobby Unser to win with, he immediately removed the sway bars and upped the springs.

I don't know much about 911s but from what I understand, torsion bars are the springs.

The argument is that in uneven road conditions where fancy independent suspension is important, sway bars make the suspension "dependent" (not quite as bad a s a solid axle but...) why transfer the forces from one wheel that hit s a bump, over to the other wheel that is otherwise is happy. (One wheel hits, the other reacts. The seat of the pants feel is a sudden rear end stutter).

On smooth surfaces where that isn't as much of an issue, then the advantage a sway bar provides by limiting body roll makes it worth it (unless it is wet and you want more weight on the outside tires to squish the water away :-). But, in unfair advantage, Donahue keeps upping the springs until it is too stiff, then backs it down one notch and then starts adding sway bars.

Thats it. I'm out of borrowed wisdom. I'm all ears at this point.

There is an interesting point here...

Consider a car that accelerates corners at 1G, accelerates at 1G, and has both a 70" track and wheelbase. If you put no swaybars on such a car it would roll in corners the same amount as it pitches in acceleration, meaning the wheel movement would be roughly the same if it's accelerating or cornering.

The average car looks nothing like the above- an Audi R8, for example, has a 64" track and a 104" wheelbase, and it's only going to accelerate about 50% as hard as it corners. So with a wheelbase 63% longer than the track and 50% acceleration vs cornering the result is a car that with springs only will have ~3x the roll as it does in pitch. Hence swaybars are required, roughly doubling the spring rate in roll so the total roll is only ~50% more than pitch.

Now consider another Audi, the S1 pikes peak mentioned above. 59" track, short 87" wheelbase, and with between 500 and 700 hp and 4wd it likely accelerated as hard as it cornered. Suddenly roll is already only 47% more than pitch vs roll out of the box, which is where the R8 got to after the addition of the swaybars. So why add the swaybars?

Many racing 911s are closer to the S1 than the R8. Due to the short wheelbase and wide track a 930 has 51% more wheelbase than track (vs 63% for the R8 and 47% for the S1). A racing 911 is also going to be able to generate pretty extreme accelerations (I was seeing over .75G in acceleration in the wet on street tires at last weekend's autocross, so ~65% of the cornering force). Do the math and you still need a swaybar to control roll some, but not as much as a more traditional car.

Seems to explain the experience people have noted above pretty nicely.

imagine how the poor sway bar toting Audi engineers felt when Unser started dismantling the result of their months of hard work :D

I'd like to resurrect this thread and ask a related question. My calculations tell me that about 53% of the roll resistance at the front, and 66% at the rear was carried by the torsion bars on my stock 74 before any mods. Reading between the lines here there are many who believe that it is desirable in a race car to have most of the roll resistance handled by the torsion bars as opposed to the sway bars. So how much is desirable? 70%? 80%? Bobby Unser obviously thought the number was high!

Money in the spreadsheet the front 998 and rear 1044 Total Roll Stiffness seems almost equal. Is this what you are trying to do? My question is how much is too much? If you can have a moderate amount of stiffness and then adjust up for different tracks with the sway bars and shocks then it would be great. I guess testing is the only way to determine where to stop.

Sorry, the spreadsheet data isn't current so doesn't mean anything. When I plug in my numbers I get the figures I quoted above. But the question I'm trying to answer is different: how much of the front and rear roll stiffness should be carried by the torsion and sway bars at each end?

The torsion bars will define your pitch center and so effect the ride over two wheel bumps and how the car responds to a straight vertical (four wheel) bump. You want the pitch center to coincide with the center of gravity's longitudinal position. So, make the torsion bar balance accordingly. Then adjust the roll stiffness distribution with anti-sway bars. Lever length has some leeway. Basically the droplink should be 90 degrees to the arm at static height for the best kinematics. So a somewhat long arm, but not too long to the point that the arm starts flexing.

I would say get as much roll resistance from the torsion bars as you can while still adhering to the balance I mentioned above, and while still leaving enough to the anti-sway bars so that you can tune the balance to the extremes you feel is necessary (rain,big willow,streets of willow, road use, etc.)

If you did not need to tune the roll balance at all you could just run one anti-sway bar if at all.

So there is theory behind the widely held view that bigger springs/torsion bars with only minimal help from anti-roll bars is the fastest setup, assuming of course that you aren't doing something stupid to get yourself in this state?