30 replies to this topic

[shaun979]

I'm looking for torsional stiffness figures for racing chassis, but particularly for formula car carbon tubs. Champ car, LMP, GP2, F1, F3000 figures welcome.

Recently tested 5 tubs and was surprised that torsional stiffness was around only what non-exotic sporty production street car would be, yet these car experience so much more load both inertia and aero.

Some old SAE papers by Clemson university put old Winston Cup chassis at around 15500 ft-lb/deg only. Again this is in the street car range which I find surprising. There are even open top street cars that are as stiff. What makes some street cars like F50s and Veyron end up with figures 1.5 - 3 times this stiffness?

Theoretically an infinitely stiff chassis is best for suspension response, but at what point do you start to gain weight whilst not really gaining any more advantage?

[Greg Locock]

To be honest you start from a false premise. Road cars see wheel velocities far beyond what a circuit car does. (6 m/s)


"F50s and Veyron end up with figures 1.5 - 3 times "

really? measured in /exactly/ the same way? (Just checking, let's just say that Ferrari never used to care about it)

There is an old argument that says that so long as the impedance of the point you attach your suspension to is say 10x that of the spring/damper combination then anything more is polishing a turd.

That presupposes that you know how to work those impedances out, and that the factor of 10 is meaningful.

If you do you'll find that the required torsional stiffness is surprisingly low - indicating that the 10x number is a bit sus.

[cheapracer]

Originally posted by Greg Locock


.......the impedance of the point you attach your suspension to is say .......

Congrats Greg, your the first person I have seen to note that important part of the total sum.

[phantom II]

On NASCAR Tech last week they said that the chassis that is selected for tracks that have similar properties is determined by its flex. They gave no values. It may appear to be a low tech sport on the surface.

Originally posted by shaun979
Some old SAE papers by Clemson university put old Winston Cup chassis at around 15500 ft-lb/deg only. Again this is in the street car range which I find surprising.

[RDV]

Greg Locock-There is an old argument that says that so long as the impedance of the point you attach your suspension to is say 10x that of the spring/damper combination then anything more is polishing a turd.

...yes, you would not believe the shine on the turd I'm racing... forced to wear sunglasses even in the pits....;)

[McGuire]

Originally posted by phantom II
On NASCAR Tech last week they said that the chassis that is selected for tracks that have similar properties is determined by its flex. They gave no values. It may appear to be a low tech sport on the surface.

Far as I am concerned it can easily be too heavy but it can never be too stiff. That would be like a motor that makes too much torque.

Problem with the Nastycabs is they are constructed from a few hundred feet of DOM 1010 steel tubing (round and rectangular, all dimensions) MIG and TIG welded together. Very large and complex structure, like a gorilla cage, and it is harder than you might imagine to build two in a row perfectly the same... though they are getting there. This is why drivers and crew chiefs tend to have one or two favorite chassis, while the fab shop keeps building newer and "better" ones that get run once or twice and then parked in the corner. They used to assume it was dimensional drift but now they are down to compliance and frequency. (A good part of that might be pre-weld fitment and post-fab stress auto-normalization. There are only about a zillion welds in there.)

When Cheatin' Chad is on the television I recommend you should believe approximately every other word he says, but no more. I bet when he was a little boy he made his Mommy offset truck arm bushings for her birthday. LOL what a lying pirate, and he looks straight into the camera doe-eyed. One car owner suggested to a crewchief I know that he should watch Chad's TV show, said "it's really good, you could learn something." Wish I could have witnessed that explosion, probably took the roof off the hauler.

[imaginesix]

Originally posted by Greg Locock
There is an old argument that says that so long as the impedance of the point you attach your suspension to is say 10x that of the spring/damper combination then anything more is polishing a turd.

Not sure what 'impedance' means in this context, but would it be fair to say that a 6-wheel F1 car would have better impedance than an equivalent 4-wheel vehicle (it would have lower point loads anyways), so would be more likely to benefit from stiffening of the chassis?

[NRoshier]

I think Greg was making the point that the chassis may be easily stiff enough as a structure, but stiffness is easily lost is the suspension pickup/mounts. When I interviewed Ron Tauranac he clearly stated that this was the case and that lots of attention had to be paid to the suspension/damper mounts.
The Ferrari figures...do we know if they are testing empirically from suspension mount to suspension mount or did they just twist the monocoque?

[shaun979]

To be honest you start from a false premise. Road cars see wheel velocities far beyond what a circuit car does. (6 m/s)

Can you explain how wheel velocity is a factor in putting twist forces into a chassis? A heavily aero loaded formula car having one corner run over a good size curb at speed versus a lightly sprung and damped long travel road or rally car hitting a bump and getting its wheel velocity up... to me it would seem the former chassis experience more torsion force.

"F50s and Veyron end up with figures 1.5 - 3 times "

really? measured in /exactly/ the same way? (Just checking, let's just say that Ferrari never used to care about it)

I'm pretty sure across manufacturers they measure it slightly differently and I'm not sure how they measure it but basically the range was 30,000 - 60,000 Nm/deg for these cars...published figures by manufacturer. I understand what you mean by suspension pick up points, so in that case why do they even bother making the chassis that stiff? Why not save weight and reduce stiffness?

[Greg Locock]

You make a good point about the higher shock rates compensating for the lower wheel velocities. When I get my Ferrari book back I'll do a comparison.

(I use stiffness, impedance and compliance interchangeably in this post, and throw in resonant frequencies as well... generally stiff structures have high impedances, high frequencies and low compliances)

The weasel word in my first post was the 10x bit. That's a very crude rule of thumb, basically saying that if you get 90% of the effect you wanted than that is as good as 100%, practicallly. 100% would require a stiffness ratio of infinity, which is obviously not practical.

A mechanical impedance is what we measure with a hammer and an accelerometer, it is the ratio of acceleration (or velocity or displacement) at a given frequency, to the force. mathematically it is the solution to x/F in the equation

F=-m*w^2*x+j*w*c*x+k*x

For a given system and direction, at a particular frequency w rad/s. Note that c and k are often amplitude dependent, and that in real systems c is almost always frequency dependent as well. The frequency range of interest in this context is 0-10 Hz... and you could learn a lot by just measuring the static stiffness with a dial gauge and a hydraulic jack.

Note that a shocker has an impedance, as does a spring, as does a mass. Usually we compare the magnitudes of the complex impedance that the equation gives, that can be misleading.

A good rule of thumb for suspension mounting points is that you'll have to bust a gut to get above 5000 N/mm, and I often see stuff down at around 2000. With good design 20000 is achievable, and I have seen 50000.... but not on a realistic car sized structure. On a spindle or knuckle, measured relative to the wheel beraing, much more is possible, mostly beacuse the leverage is less.

For road cars there is no doubt that torsional stiffness way above 10x is worth having - BMW have made a fetish of this, and I think it pays off (shake, squeak and rattle, structural feel,- all track body stiffness) . However the big gains are all over by the time your body torsional mode hits 25 Hz for a sedan. In stiffness terms that is around 15-20000 Nm/degree (NASCAR cars have been measured at around 60% of this)

The 2000-5000 number I mentioned before is strongly affected by this torsional number - if the body rails are bending then it doesn't matter how much you reinforce the clevis locally, it'll still try and get out of the way of the load. Much more common of course is to see poor detailing at the clevis throwing stiffness away.

For a race car it may not be worth pushing it as hard as the tradeoff with weight becomes more important.

I'd add that if 2 labs measure torsional stiffness it may take some dialogue before they can agree on a value... This is crazy, with care and attention you can do it in your garage. I'll try and dig out the results I got compared with the FEA of the same chassis. If you are doing it yourself, I recommend ignoring torsionals, and just measure the deflection at the 4th corner with the chassis supported at the other three. The number you get is a mixture of bending and torsion, but it is equivalent to single wheel bump. Then do the other end of the car. Otherwise the method shown in that Winston Cup paper is fine.

Bored now?

[McGuire]

Nicely explained, excellent post.

[imaginesix]

It is, except that now I really should use that information to try and figure out the answer to my question. Dammit.

[cheapracer]

Originally posted by Greg Locock
If you are doing it yourself, I recommend ignoring torsionals, and just measure the deflection at the 4th corner with the chassis supported at the other three. The number you get is a mixture of bending and torsion, but it is equivalent to single wheel bump. Then do the other end of the car.

I believe this was Frank Matich's favorite method for getting his chassis to where he wanted.

[Lukin]

Some guys did a SAE paper on it (one of the guys went on to work at Renault)
http://www.sae.org/t...rs/2000-01-3554
The paper is interesting, looks at dynamic and static comparisons and comes back to the '% of overall roll stiffness' rule of thumb.

There is also an old thread about stiffness.

[Greg Locock]

OK, here's the deflection along each side of the rail, for Mr Hebel's ladder chassis, compared with the results from the FEA. As I said, this is a single corner deflection test. To get the correlation I had to model the test rig as well, not just the chassis. If your FEA guy gets better correlation than this... good. It took us a couple of weeks to get a test procedure that worked, Mr Hebel is a very patient and thorough man.

These plots are deflection at various points along the rails (mostly, a few are in the centre of x members), in response to a load applied at in line with each wheel, with the other equivalent points located on pin joints. The x axis is distance along the chassis measured in m, the vertical deflection is m and I'm not telling you what the load was.

front

http://www.geocities...hebel_front.png

rear

http://www.geocities.../hebel_rear.png

However he pulled a dirty rotten trick on me and mod'd the chassis by welding in a new cross member, to see if I could predict the difference.

http://www.geocities...y/hebel_mod.png

Phew (and you can bet that that took a while)

If the links don't work

http://www.geocities...ery/gallery.htm

Lukin - that's an interesting looking SAE paper. I'll read it tomorrow.

[shaun979]

Originally posted by Greg Locock
Bored now?

No not at all. All that information is really interesting, though I don't understand the equation because I don't know what most of the letters stand for and their units. If you could explain or list a book and chapter they came from I will follow up on it.

Actually we didn't use the Clemson paper method. We used an old Riley method which is also a pure torsion test.

===

Lukin thanks, I don't have that specific paper, but I came across a separate paper that included part of it in one of its sections, including the % overall roll stiffness graph. It seems to agree with Greg's numbers. I'll probably have to buy the full paper now or just go ahead and get PT-90. That old thread on torsional stiffness was good too. I like the questions you ended it with.. maybe if we bring them up on this thread again, they might be discussed further?

===

It's looking like the real high torsional rigidity chassis are that way for reasons other than suspension response because, 60,000 Nm/deg for example, is far beyond the point of diminishing return in that specific area.

Thanks guys.

[Greg Locock]

w is the frequency in radians per second (Hz *2 *pi)
x is the displacement

m is the mass
c is the damping
k is the stiffness

Best book on this equation and related stuff is Rao "Mechanical Vibrations" but WE Thomson is cheaper easier to find and more interesting.


But you don't need it, in my opinion. If you are just doing static torsion that is fine.

By the way the Ferrari F1 car in 2000 had a torsional stiffness of just less than 7000 Nm/degree, (~5000 ft lb/deg roughly) axle to axle.

That is not much is it?

[McGuire]

Originally posted by shaun979



It's looking like the real high torsional rigidity chassis are that way for reasons other than suspension response because, 60,000 Nm/deg for example, is far beyond the point of diminishing return in that specific area.

Right. While Bugatti's claim is 60k Nm/deg, that is around double the stated value for recent road vehicles of that general type. I'm not saying the claim is false, just not sure how it was derived or what it means. Ford GT etc. are around 30K and that is pretty stout in itself.

[lateralforce]

For road cars the stiffer the better, since stiffer chassis means that it is much more durable in the long run. First torsional mode should be around 30Hz just so that it is far enough from any of the engine modes and the air cavity mode for NVH reasons.

[shaun979]

Originally posted by Greg Locock

By the way the Ferrari F1 car in 2000 had a torsional stiffness of just less than 7000 Nm/degree, (~5000 ft lb/deg roughly) axle to axle.

That is not much is it?

Again, that's really surprising to me. They must be going all out for light weight. I was also thinking about how high downforce might play into torsional stiffness requirements but haven't had time to properly structure and phrase it. I will come back sometime when I have.

Claude Rouelle in one of his seminars showed chassis torsional frequency being measured. It was supposed to be factored into the turn entry F:R roll phasing somehow. I was so busy trying to write down and digest other things that this was one of the points that I partially missed. Question time was very limited because everyone had multiple questions too. I assume you try to keep the roll phases from matching the chassis torsional frequency so as to not excite the chassis? I may have this completely wrong since I'm going off memory. If so what's the real purpose of measuring the frequency and how does it factor into roll phasing?
Thanks again.

[Greg Locock]

Torsional frequency for a road car is something above 20 (and preferably 28) Hz, it is not going to have much interaction with a road car's roll mode which is 4 or 5 Hz from memory.

I don't have any feel whatsoever for what the modal alignment chart for an F1 car looks like, but I bet it isn't pretty.

[Melbourne Park]

Makes me wonder how a Morgan Plus-4, with its timber underneath, ever got round a corner. Fun to driver too. Well, if you like that sort of thing.

[Greg Locock]

It'll get around the corner, but it'll shake (vibrate) as it does so.

[Melbourne Park]

Originally posted by Greg Locock
It'll get around the corner, but it'll shake (vibrate) as it does so.

With them they had very hard suspension, the theory was the car flexed instead. Their ride was quite harsh ...

[Catalina Park]

Originally posted by Melbourne Park
Makes me wonder how a Morgan Plus-4, with its timber underneath, ever got round a corner. Fun to driver too. Well, if you like that sort of thing.

Timber underneath? Morgans have a steel chassis and a wooden body frame. Stupid journos have been mixed up with the "wooden frame" rubbish for a long time.

Chassis.


Body frame.

[Melbourne Park]

Originally posted by Catalina Park
Timber underneath? Morgans have a steel chassis and a wooden body frame. Stupid journos have been mixed up with the "wooden frame" rubbish for a long time.

Chassis.


Body frame.

But they are today's Morgans. Go back a few years, they only recently switched I think. Not all wood though.

Still I'll check - there's Plus 4 that's often parked outside a few blocks from where I live, I'll crawl under and have a look at it maybe.

[imaginesix]

So the wood contributes nothing to the strength of the structure? I don't think so.

[Catalina Park]

Originally posted by Melbourne Park
But they are today's Morgans. Go back a few years, they only recently switched I think. Not all wood though.

Still I'll check - there's Plus 4 that's often parked outside a few blocks from where I live, I'll crawl under and have a look at it maybe.

They are not todays Morgans. The bottom photo is a 1955 Morgan being restored.
They have always had a steel chassis. You have been conned by stupid journalists.

[Melbourne Park]

Originally posted by Catalina Park
They are not todays Morgans. The bottom photo is a 1955 Morgan being restored.
They have always had a steel chassis. You have been conned by stupid journalists.

I have a nasty suspicion you are right. Damn it - I always wondered how such a flexy thing could work.

But thanks for the contribution, and the photos are nice. A very clean place too.

Still timber boats can be stiff. In the Dragon Class of yachts, old 30 foot one design racing boats that were in the Olympics for years, the composite glass boats cost one third of the cold moulded timber boats, which really are a composite timber boat, with thin layers of timber at different angles I think, and I know they are extremely expensive to make.

What a shame.

[Melbourne Park]

Originally posted by Melbourne Park
I have a nasty suspicion you are right. Damn it - I always wondered how such a flexy thing could work.

But thanks for the contribution, and the photos are nice. A very clean place too.

Still timber boats can be stiff. In the Dragon Class of yachts, old 30 foot one design racing boats that were in the Olympics for years, the composite glass boats cost one third of the cold moulded timber boats, which really are a composite timber boat, with thin layers of timber at different angles I think, and I know they are extremely expensive to make.

What a shame.

And from an Aussie Morgan club site:

For those who may be a little unfamiliar with the subject, Morgan's are a hand made British Sports Car which are still made today in a small factory privately owned by the Morgan family in Malvern Link, Worcestershire. They are meticulously fabricated by hand in the same traditional, old-fashioned manner as when the factory first opened in about 1910. They have always, and still have, a simple steel chassis, wooden body, sheet metal panels and delightful lines. The four wheeled models have had the same basic chassis design and suspension for the last 60 years.

After that, they talk of a newer one.

[NRoshier]

The latest morgan uses an aluminium chassis and the 'wood' is a heat formable laminate, far removed from the 'traditional' methods of the 1955 car.
Re wood and stiff...indeed you can make a wood structure stiff, just look at all of the wooden planes. However a more relevant example would be the Marcos cars and the Costin Amigo, which were according to report, light and stiff...but challenging to manufacture.