For racing or performance purposes, the chassis's torsional stiffness should be 5X greater than the car's roll stiffness. A well designed DSR race car would have a chassis good for ~4500 ft-lb/degree (guess) and that is with the CG slammed to the ground. A car with a higher CG would require more roll stiffness, and thus a considerably stiffer suspension. 2000 ft-lb/deg would be extremely flexible in my opinion.

To figure out how stiff the chassis should be would require knowledge of the CG height and suspension geometry info, or roll stiffness value for the car (how much torque does it take to roll the chassis 1 degree). Does anyone have an accurate Mitchell suspension geometry file for the 911? The whole purpose of a stiff chassis is for tuning the handling characteristics of the car. With a stiff chassis the car responds to minute adjustments to the sway bars etc... A weak chassis car does not. I would be pretty surprised if the 911 chassis is weaker that 5000 ft-lb/deg.

If you want a meaningful increase in the chassis stiffness there isn't any substitute for a roll cage designed for this.

Bob

"I would be pretty surprised if the 911 chassis is weaker that 5000 ft-lb/deg."

Bob,
Your '88 chassis might reach that threshold, but factory specs are hard to come by. The earlier chassis are more flexible due to thinner sheet metal.

Adding a roll cage is cheating (in this thread). The objective is to increase torsional rigidity via other, less visible/less intrusive means (if possible).

Sherwood

The difficult thing about the 911 chassis (especially the Cabriolet and Targa) is that most of the structure between the ends of the car are close to the same plane (the floor). The coupe isn't much better because the winshield and rear window seals don't allow any stress to go through the windows (probably a good thing), leaving the roof suspended by four skinny beams. Without more 3D structure that gets away from the floor it will be tough to increase the stiffness much. It is like trying to stiffen a flat piece of sheetmetal. Also, the key is TORSIONAL stiffness, so beefing up the side rails isn't really going to do much.

I did quite a bit of computer analysis on the 914 chassis in an effort to solve the exact same problem of stiffening the chassis without a roll cage. My conclusion was that a roll cage couldn't be avoided. Without that it would have been a ton of welding and fitting for a ~15% increase when a 100% increase was needed.

My thought with the 914 was that the side rails acted like torsion bars connecting the front and rear sections of the car, and that stiffening them would be a very effective way to stiffen the chassis. The computer analysis showed that there was a lot more going on and stiffening the side rails was just causing more flexing in the wheel wells, where the side rails terminate. In short, to really solve the problem using that method would have been a ton of work involving modifications from end to end.

Bob

I have been catching up on this thread and a couple of things stand out. There are pictures of a GT3 tub that clearly show a tunnel in the middle of the car going from front to rear. Wleding a plate to the bottom of this tunnel would make it a tube which would increase the stiffness considerably.

Also, there are a few pictures of cross braces in the front trunk of the 911. I would assume that the designer wanted to tie the strut towers to the front A-arm mount point but I'm not sure why. There are only torsional forces in the front A-arm/ torsion bar mounts so tying those together with a piece of rectangular bar would be a smarter thing to do. There aren't any vertical forces at those points because those are all at the strut towers and the other end of the torsion bars.

The main key to increasing the stiffness is tying together the points (strut towers and pivoting end of the torsion bars) where vertical loads enter the car with structure. Also, going stiffer than 5X the car's roll stiffness is overkill. The Bugatti is extremely stiff (44,000 ft-lbs/ deg) but I'm sure it is all needed. A car with 10X is great, but a car with 5X is great and weighs less. That tells me that the roll stiffness of these cars are pretty high and there is a lot to gain by stiffening the Porsche's chassis.

Bob

A couple of years ago i drooled over a book at my favourite bookstore in Stockholm. It was a german book, which in german detail covered tests of the torsional stiffness of a 911. The car was a coupe from between '74 and '89, I can not remember exactly. In addition to the total torsional stiffness between front and rear shock mounts they also measured the amount of displacement at several points between.

To the best of my knowledge, the torsional stiffness was measured to 14.400 Nm/degree.

I have forgotten the title of the book. The book was in german and had no color images so I did not buy it... Yes, I have hit myself a couple of times.

14,4000 NM/deg would be 10,620 ft-lb/deg which sounds pretty good. I will look into the roll stiffness just out of curiosity. Does anyone know how high the CG is in a stock 911?

IT would be great if you could measure the torsional stiffness - I think you'll have to measure the CG also. There is a procedure in Race Car Vehicle Dynamics, and a procedure in an old UpFixing (Vol. III?).

Here's what I have on the former:

Body flex – Torsional Rigidity

Mark Donahue mentions in his book "The Unfair Advantage" that the 911's that they worked on flexed at a rate 2,000 ft-lbs/degree, which he considered plenty stiff. Keep in mind what this means to your alignment. For every 2000 ft-lbs of torque you apply to the chassis, it will twist 1 degree. Imagaine that you have 500 lb springs on the front and that they are about 2 feet from the car's centerline. If you corner such that you compress the outside spring by 1 inch, this is 500 lbs at 2 feet or 2000 ft-lbs or torque. Add on a stiff roll bar and you most likely have even more torque being applied to the frame. Now your rear suspension is rolled 1+ degrees further then your front. This additional roll at the rear basically dilutes the effects of your sway bars. As a rule of thumb, your chassis stiffness should be about 20x stiffer then your spring stiffness if you want to keep it's contribution to the cars total movement to less then 5%.

So we have one source giving a value that is 5x that of another source - a big difference.

I would really be amazed if it was as low as 2000 ft-lbs/deg. I have designed small, open-wheeled race cars weighing 500 lbs with a CG height of 12". With a suspension roll stiffness designed to keep body roll under 1.5 degrees under 1.2g of lateral acceleration (500 ft-lb/g roll moment), the chassis stiffness requirement was ~2500 ft-lb/ deg.

The Porsche weighs in at 2800 lbs and has a GC height around 16" (guess) which makes the roll moment 3700 ft-lbs/g. The difference in track is not tremendous so it is a fair assumption that the roll stiffness of a Porsche suspension would be ~3000-4000 ft-lbs/deg. making the chasiss demands ~15000-20000 ft-lbs/deg for it to be functioning properly. That starts to get in the neighborhood of the Bugatti, which probably has a much stiffer suspension.

20X is pretty stiff and it seems like overkill. I have always been told that 5X is a minimum and anything over 10X was overkill unless composites materials are available (stiffness to weight are still good with composites). I have seen the body contribution profiles in SAE studies and it gets into diminishing returns. At 5X a race car will respond to tuning very well. All of this is excluding and aerodynamic downforce of course. Once that enters the picture things might change a bit.

I think the CG this is a red herring in and of itself. A high or a low CG isn't meaningful, but rather the length of the roll couple at either end (the distance from the CG to the roll center at either end of the car), or after thinking about it some more, the difference in roll-resistance at either end. So if both ends are equally stiff in roll (either geometrically or via springs and sway bars), stiffness may not be as important as situations (like FWD cars and 911's) where one end is stiff and the other end relatively soft. Those are situations where you're really twisting the chassis.

A high CG is bad news because it causes more weight transfer during cornering. A car will handle best when its weight is most evenly distributed on its four tires. The tire that has more load on it will break loose first. With a higher CG, more load will be placed on the outer tires when the car corners, reducing its handling capability. You are right though, the roll moment arm is the distance between the CG and roll center, though roll centers are usually within a few inches or the ground.

The roll resistance of the front and rear end should be pretty close in value but to make the car tunable the chassis needs to be stiff. If the car is understeering for some reason and you wish to make the car oversteer, you will need to increase the rear's roll stiffness, usually via a sway bar. If the chassis is weak, instead of more load being placed on the rear tires, the chassis simply flexes more and you drive with the same understeer as before, until you make a huge adjustment which swings the characteristics to the opposite extreme.

A stiff chassis isolates the front and rear suspensions so minute adjustments are meaningful and they alter the car's handling characteristics. With a weak chassis, suspension adjustments simply result in more or less flexing of the chassis.

I was just re-reading this idea and I think it is brilliant. And I don't see why it could not be done in an aesthetically pleasing way. For instance, why not coat the entire inner tub from the rear package shelf to the pedal cluster with a strong structural epoxy - the same stuff they use to assemble airplanes. Then lay a dozen or so layers of carbon fiber on top - essentially creating a carbon fiber monocoque using the original unibody tub as a mold. Then you can cover it all with carpeting and the thing should be TOTALLY stiff...

Is this too simple? Am I missing something obvious?

This makes sense to me: "high CG is bad news because it causes more weight transfer during cornering"

- John? Do you think the length of the lever arm predominates over the effect above (?)

____________________

Janus - Jim is talking about an oriented fiber material (anisotropic) - a common example is wood. We are just now starting to make carbon and other materials like this in the last few decades. Epoxy per se, would add a lot of wt for each quantum of stiffnes you get out...

I also think it is a great idea - I suppose it could be done on the inside of some panels so it didn't show. You'd have a lot of labor to remove whatever is there now (undercoating) and would have to shape the stiffener panel to the steel somehow.

Just glanced through the August, 2006 issue of Sport Compact magazine. They had several pages and photos on reinforcing and lightening a Honda S2000. Some good ideas. Also shows the extent at which a chassis can be modified for strength. I believe they managed to trim about 800 lbs. as well. I liked the swiss-cheesed door panels.

FWIW,
Sherwood

Randy
Don't forget that the cars Donohue was working with were most likely the ones at hand: IROC RSRs. These cars were stitch welded with reinforcements at the suspension pick up points in the rear and in the engine compartment. As they were not thin sheetmetal cars (like the first 500 73 RS Carreras) , these cars would have been stiffer than the production counterparts. The only roll bar they had was a simple ST style hoop with diagonal and Penske installed harness bar near the base of the back Y. These were bolted in to the welded in plates on the main hoop and the rear seat hinge pins.No B pillar attachments
So not much stiffness added there either
I cannot remember the source , but some where way back when in discussion with a major Porsche Guru, a claim of 20% increased stiffness from the stitch welding was bandied about.

fwiw- i think adding A and B pillar half tube "hidden" roll cage would add considerable stiffness to a chassis when incorporated with rocker stiffening , like Thoms Fitzpatrick posted. Coupled with stitch welding and front trunk bracing a reasonably light addition of weight in return for a stiffer chassis would be achieved. The hidden "cage"would by no means be a track legal item, but the car would be better for it with increase in handling capabilities

Sherwood - any pics you can post for those of us in the boonies re car mags...?

Sounds interesting. Any engineers care to comment on this idea?

During my quick scan of the article, the builder welded replacement panels on top of the existing panel for add'l strength/rigidity. I assume they fit neatly, but observation reveals where most unit-body cars need added reinforcement. For example, the builder ended up with double-thickness floor pans.

In other areas, he created cardboard patterns for boxed sections, tabs and gussets, then fabricated sheet steel puncutated with weight-reduction holes - usual racecar stuff.

Sorry Randy,
I was browsing through the mag - didn't buy it; maybe will return and pry $5 from my wallet.

Sherwood

Good thread.

And many ideas which I have visited over the years.

Firstly the lever arm effect is not the same as the effect of cg height on weight transfer.

So both of you are correct..you are discussing different things.

The CG height to track relatioship is the SOLE determinant of the total weight transfer to the outer wheels...assuming the sideways G is fixed..

The propertion transfrred by front and rear wheels is varied by the roll stiffness at each end

Roll stiffness depends on roll centre heights, and spring and antiroll bar stiffness.

So the lever arm between roll centre and CG does affect the twist put into the chassis...

Kind regards
David

Sorry!

Secondly, Carbon fibre is wonderful stuff..but...

Its not always realised that 1mm of anisotropic CF solid laminate is approximately as stiff as 1mm of steel sheet.

The benefit is that its much lighter--period..

And also that it is easy to use a core material to produce a thicker stiffer section with minimal weight.

Structurally then, all you will achieve with CF applied to the tub is a thicker steel tub.

it is rather hard in practice to achieve good adhesion over such an area ..even using vaccumm bagging..which is pretty essential with epoxy carbon..

carefully used, steel sheet and tubes will do the job, with some weight penalty, but will be much eaiser to create more section modulus..

Kind regards
David