Body and Frame

[aae3]

What is a good recommended torsional rigidity for the frame? I have been researching other cars and it seems that some teams recommend a torsional rigidity above 2,000 lbs*ft/degree and other teams are in the 1,400 lbs*ft/degree. I have also see in posted where one team will say "Stiffness is the only way to go," and another team will say, "There is no need for that much stiffness in a 2 day event." Any comments on this subject would be greatly appreciated. Thanks

[Bill Kunst]

The torsional virginity of the vehicle needs to be related to the stiffy of the suspension.

If you have a very stiff suspension, and not so stiff frame, the suspension will not be able to do its work properly. Now the question becomes "how stiff should the frame be in relation to the roll stiffness of the suspension?"

[aae3]

Posted October 17, 2006 01:10 PM Hide Post
The torsional virginity of the vehicle needs to be related to the stiffy of the suspension.

If you have a very stiff suspension, and not so stiff frame, the suspension will not be able to do its work properly. Now the question becomes "how stiff should the frame be in relation to the roll stiffness of the suspension?"


Thanks for the advice! So do you have an answer to find the relationship of the role roll stiffness of the suspension to the frame stiffness?

[C.Zinke]

A good frame stiffness for a typical FSAE Car is between 1000 an 1500 Nm/deg.

You have to understand why you need the relationship, otherwise you won't get any advantage of this information. You will notice that there a much more important relationships than this one.

[mtg]

I've never seen a textbook answer for this question, but what I've come up with is that the system should be looked at from a natural frequency perspective.

You should know the natural frequencies of the four suspension modes to begin with (ride, roll, pitch, warp). Then it makes it much easier to apply what Bill Kunst mentioned and pick a torsional natural frequency of the chassis based on the warp mode, and back calculate to a stiffness. Now how much higher to make the torsional chassis frequency than the warp frequency, my thoughts are high enough not to induce the natural warp frequency while driving as a minimum.

That's my theory on figuring out a torsional stiffness, designed to work with if you have no baseline.

[js10coastr]

Originally posted by mtg:
I've never seen a textbook answer for this question, but what I've come up with is that the system should be looked at from a natural frequency perspective.

You should know the natural frequencies of the four suspension modes to begin with (ride, roll, pitch, warp). Then it makes it much easier to apply what Bill Kunst mentioned and pick a torsional natural frequency of the chassis based on the warp mode, and back calculate to a stiffness. Now how much higher to make the torsional chassis frequency than the warp frequency, my thoughts are high enough not to induce the natural warp frequency while driving as a minimum.

That's my theory on figuring out a torsional stiffness, designed to work with if you have no baseline.

...after working on that SRF, I think that if you're TLLTD is a lot different from your static weight distribution then you'll need a stiffer chassis. Just a theory I thought up the other night...

[mtg]

Or if you're working on a SRF, you should cry yourself to sleep instead http://fsae.com/groupee_common/emoticons/icon_smile.gif Sorry, couldn't resist.

[KU_Racing]

In general, our basic factor for the chassis stiffness is that it should be 10 times the roll stiffness of the suspension at a minimum. More stiffness is usually better, so if you are over dont worry about it. That can be a good starting point if you arent sure of how to use bill's method, which is more complicated.

[aae3]

Thank you everyone for your help. We have been using strain gauges from National Instruments on our previous year's car to record raw data, and we have been using FEA to calculate the torsional ridigity of the frame. Which methods do most teams use to measure their chassis stiffness?

[Bill Kunst]

KU-
I was saying the same thing, I just thought that it would be good if aae3 researched that information himself.

aae3
There is still a long road ahead. You have to design your suspension layout first. Design it to handle well, and then design a frame to fit into the constraints of your suspension. Or vice versa, which can be just as hard and less successful. It is a chicken or egg thing. What part of the car is optimized. Eventually, after four or five cars, you or your team will get good enough that you know the general dimensions that you have to start with for either, which becomes alot easier.
Bill

[Mike Cook]

Has anybody here ever actually done testing with respect to torsional stiffness. I don't mean actual test your car for a value, but test the way it affects the vehicles handling. Because I have and I think you all are off in the weeds. I built two cars, both have the same suspension points, engine, wheel rates, etc. One was about 700 ft lb deg, the other was way over 2000 ft lb deg. I never noticed a difference between the two. Both had tiny sway bars which contributed the same percentage of total roll stiffness (something like 10%). Both cars we're very responsive to small sway bar adjustments. Talking with some people in the industry, they have noticed some of the same things, and in fact if you do most of the math, you see that about 4-5 times the suspension rate is good enough.

I will be honest though, I never compared the cars back to back in a slalom which is the only place I think the stiffer one might have been faster, due to having a higher frequency.

Mike

[js10coastr]

Originally posted by mtg:
Or if you're working on a SRF, you should cry yourself to sleep instead http://fsae.com/groupee_common/emoticons/icon_smile.gif Sorry, couldn't resist.

...or start looking for a new job. Funny story actually... I'll have to tell you some other time when the details are worked out.

Mike-

I've hear that "10 times" thing a lot and figured it was one of those numbers that was jus a "rule of thumb".

This is where my thinking of the TLLTD/static weight distribution comes in. It also touches in with that MTG mentioned about warp. I think that if you need to transfer a lot of load diagonally you'll need a chassis with a higher torsional stiffness. Maybe someone can play devils advocate...? I haven't had a lot of hands on experience with the subject to come up with any definitive say.

[Mike Cook]

Well, I was thinking that the warp mode (essentially one wheel bumps) will torque the chassis. Torsional stiffness will certainly affect your warp mode. For instance, when doing a wheel rate test on a old car, our chassis twisted quite a bit which had an affect on the wheel rate. Now this can directly be related to warp mode. So less torsional stiffness = less warp mode. How will this affect the car? Less weight transfer to the other side I suppose. But as was mentioned before, the big concern here is evaluating what freq domain your warp mode is and making sure that it won't resonate the chassis. I'm not really sure this is a big deal, but I haven't run the numbers.

I also haven't run the numbers on a slalom, but I'd be curious to see what freq the car is at for a tight slalom. If the chassis is not stiff you may have to worry about the chassis torsional freq.

A couple other thoughts on the subject:

A stiffer car is going to have better transient response. All this mumbo jumbo about 10x the suspension stiffness, I believe pertains to steady state handling. Basically what it is getting at, is that for a desired LLTD, a torsionally stiffness challenged car will have to increase the Roll stiffness distribution more than a non torsionally stiffness challenged car. Ok so what? More adjustment is built into the sway bars, the day goes on.

Second thought:

Torsional stiffness alone is not enough to really understand how the car will behave. The torsional stiffness between the front axle and cg, and the torsional stiffness between the rear axle and cg must be known. Think of the roll stiffness at one end of the car. The total roll stiffness is a combination of the suspension roll stiffness plus the torsional stiffness between the CG and that axle. Now imagine two cars, both with the same suspension stiffness and torsional stiffness. Car one is infinitly rigid between the CG and the rear axle - that is that all torsion happens between the CG and the front axle, and car two is the exact opposite. Now think about understeer/oversteer.

This alone is not really that useful (although interesting to think about) but it is important when designing cars. For example, lets say we for a give wb, tr, cg location, we can relation LLTD to understeer oversteer. Now we're building next years car. We know what we want but we're trying to improve the chassis. We are sucessful, but the stiffness between the cg and front axle becomes twice as stiff as before, and from the cg to rear axle stays the same. You assume that you should be able to use similar RSD as before - BUT is this really the case anymore?

Anyways, this is getting rather long. I put this forth more as a discussion than a decleration of correctness.


Final two things:
Stiffer = faster transients
Shifter Karts are still faster than us. Never forget this one.

[Garbo]

For the last few years, we have been increasing frame stiffness considerably (interestingly, the frames are not getting heavier... just more work goes into design) and we have found that the cars have been responding increasingly well to roll distribution tuning, allowing finer adjustment. We also do a physical torsion test on the finished cars to verify the FEA, normally the results are within 5% of predicted.

When designing a frame, particularly if you are working to a stiffness target (in any mode), it is important to not over-constrain the frame. We saw a 40% or more drop in predicted/tested stiffness, just from releasing track width and wheelbase degrees of freedom during testing... test and analyse with constraints similar to what the car will see on track. This goes for testing upright, hub, wheel, etc. stiffness as well as installed stiffness of the suspension.

We really started revising our targets and testing methods in 2005 and we have seen really good results.

Also, beware of quoted stiffness numbers. How was the test conducted? Was the frame loaded through the hubs or through some other method? How were the loads and constraints applied? Through creative constraints/load application, you can probably double your number but does it mean anything? I spent more time in design in 2005 & 2006 talking about how we came up with our constraints and targets than spouting numbers (over 2800Nm/deg, BTW but after what I've just said, you should probably ignore that).