How much torsional rigidity is enough?

[Jeff The Pyro]

Assuming an average sized competitive car (450lbs or less)... at what point does the extra weight added required to stiffen the chassis further become cause a larger disadvantage than having a less stiff chassis?

We designed our chassis in CAD to be about 3000ft*lbs/deg (never actually got around to testing it)... and then arrived at competition to find that we were right in the middle of a rather wide range, even among successful cars that placed very well in the design event... off the top of my head, a guy from Penn State told me they tested theirs at 1300 ft-lbs/deg, with a car that weighed around 375. at the complete opposite end of the spectrum, one of the guys from western australia told us 7800N-m, which is about 5600 ft-lbs/deg. both cars were quite fast, both placed in the design finals.

my initial thoughts were that something around the middle lower range would be the best compromise... somewhere around 2000ft*lbs/deg. most of this is from looking at the roll rates of our car... with a slightly rearward weight distribution the ideal roll couple came out slightly front biased... at any rate, in steady state through a perfectly smooth 1G turn, the chassis is subjected to ~130ft*lbs of torque... if the chassis rolls 1 degree, this torque will result in a total of ~.07 degrees difference in roll front to rear.... with a 46 inch track this correllates to one set of wheels being skewed ~.25" (IE one side up .25", the other side down .25")... though this sounds rather drastic, this "twist" would be split between front and rear tracks, resulting in a skew of only half that. Looking at your options at this point, suppose you increased the torsional rigidity to about 3000... is it really worth the extra weight in order to reduce the wheel deflection from chassis flex by only a couple hundredths of an inch? on the other side... how much weight could you save from the chassis by dropping it... and would it be worth it?

obviously there's other things to consider... like single wheel bumps, which would probably put considerably more stress on the chassis than a steady state turn.

just thought i'd throw this out there to see what you guys think... since obviously nobody has arrived at an answer that everyone agrees on.

[Kevin Hayward]

Jeff,

Those torsional stiffness values should be hub-to-hub. You will find that a lot of the deflection, at least for a decently designed chassis, will occur with the suspesnion components. So a large torsional stiffness comes at the cost of weight in the suspension as well (at least for equivalent materials). However a significant amount of slop can be found in the joints as well as significant deflection in the brackets.

We take a fairy holistic approach to stiffness. It does create problems in going super light-weight. But I would put significant value on both suspension and chassis stiffness.

It is worth trying to find the parts and areas that contribute the most deflection. That way you can put your effort into working on those areas. A badly designed upright package, for instance, can significantly reduce hub-to-hub stiffness without saving a lot of mass.

It is also worth testing at relatively low loads. As you increase the torsional load on the chassis during the test you see a type of artificial stiffening as all the slop in joints etc is taken up.

There are some papers out in SAE land that detail how much chassis stiffness is "enough". As of yet I have not read one which offers a completely plausible answer. My main issue with the papers I know is that they do not account for damping forces. If someone knows of one that does I would love to read it.

I would say that it is definitely worth having a stiff chassis. All the numerical and kinematics design is almost useless without it. From an empirical standpoint our 2002 (Oz) car had about half the stiffness of our Mono cars ... actually about what you mention for your vehicle. It is nowhere near as responsive to suspension changes during testing. I have already mentioned in previous posts that most of our vehicle tuning revolves around the dampers ... I leave you to put two and two together.

Anyway that is my basic stance on chassis stiffness. It is one of those things taht has an effect on the performance of the vehicle that is much harder to quantify than say overall vehicle mass. Just because some performance gains are hard to quantify it does not make them insignificant. As a community I would say our ignorance in vehicle analysis is only exceeded by our arrogance in stating our conclusions http://fsae.com/groupee_common/emoticons/icon_smile.gif

Cheers,

Kev

[Wham Bam]

You gotta go at least an order of magnitude above your wheel frequencies.

"Hit the thing and see if it sings like your mother"

[Charlie]

You see a large variance because:

1. Methods for testing vary widely
2. People are NOT doing physical testing, which makes thier numbers marginal at BEST I would never beleive anything that wasn't physically tested, because I've been there, and typical students just learning FEA are apt to make all kinds of mistakes.
3. Cars are different, wheel rates are different, and that is how you should determine what your needs are.

[karsten]

we got the worst torsional load case form
our suspension-guy. then we set an target
ratio between tire motion cause by
spring deflection and motion caused by
framedeflection. we set the ratio to under 10%.

in our case problems/differences with fea compared
to phy testing were caused by insufficant knowledge about the impact of the engine
on torsional stiffness. in fea we did
some calculations with rigid engine and
without engine. the truths was something in between. so you need phy testing to tune/refine
your fea-modell.
but phy testing is difficult aswell. the
stiffness value is one point, the other is
the chart deflection vs distance from rear
axle. the chart is interessting to find weak
zones, but you cannot compare it the different
frame and testing methods.



karsten
lionsracing.de

[gr3pz]

Originally posted by Charlie:
You see a large variance because:

1. Methods for testing vary widely
2. People are NOT doing physical testing, which makes thier numbers marginal at BEST I would never beleive anything that wasn't physically tested, because I've been there, and typical students just learning FEA are apt to make all kinds of mistakes.


Charlie,

I want to do physical test on torsional stiffness http://fsae.com/groupee_common/emoticons/icon_rolleyes.gif , how many methods you know? Can you explain how the methods work? Thank you.

-gr3pz-

[Denny Trimble]

http://students.washington.edu/dennyt/fsae/fsae_search_howto.jpg

http://students.washington.edu/dennyt/fsae/fsae_search_results.jpg

Don't click here (http://fsae.com/eve/forums?a=search&reqWords=torsional+stiffness+test&forum_scope=825607348)

[CMURacing - Prometheus]

denny, in the time you spent making all those screenshots, couldn't you have just posted that model of your rig again?

[Mark Bacchetti]

SAE Papers on this topic.

"Design, Analysis and Testing of a Formula SAE Car Chassis" 2002-01-3300

"The Effect of Chassis Stiffness on Race Car Handling Balance" 2000-01-3554