Can anyone help i am looking for values for the torsional rigidity values of other teams chassis to write a report critacising ours. Any info or links would be greatly appreciated. cheers

We were told by a design judge from Ricardo that we should strive for 1200 lbs per degree. I was reading on the University of Missouri Rolla site last night that they strived for 2200 lbs per degree. That can be seen here http://web.umr.edu/~formula/library/sae_paper/paper.html

Thanks.
David Money

That was a great help, thank you.

Our frame is around 1700 lbs/degree. This is probably too stiff and our next frame we will aim for a 1400 lbs/ degree and shave off at least 20 lbs off the front frame.

1700 ft*lbs/deg could be too stiff, or not stiff enough. It depends on what your roll stiffness distribution is, among other factors. And saving 20 lbs for a reduction in stiffness of 300 ft*lbs/deg sound a little extreme. 20 lbs is 55 ft of .035 tubing or 30 ft of .065 tubing!!! If you can save that much weight on just the front frame, imagine how much more weight you guys can shave off.

Hello Everybody... we have already heard about different way to find the torsional stiffness but I still don't know how to implement all that in a FEA software... we would like to know the best (and a kind of standard) way to figure this out...

Is there any way to get this torsional stiffness value directly from the FEA software...

Any help please

David Lemire
Team Leader
McGill Racing Team

Our 2002 frame tested at 2450 ft-lbs/deg. We too are probably going to sacrifice some of that stiffness for weight gain for 2003. But we only weighed in at 475 lbs. (wet) at competition.

Travis Rouse - Test Pilot - The University of Texas (Austin)

The stiffness numbers you ar giving mean a pile of beans. The real key is how you test for your stiffness. Is it loaded through the suspension points. 90% of the game when it comes to chassis stiffness is having good suspension load paths.

1. In my opinion most cars are not as stiff as they say they are. Because of poor testing procedures.

2. Most cars at the competition have too soft of frames and probally don't relize it.

Jeff

I agree that having a frame which is not "rigid" will cause difficulty in the tuning of the chassis. However, at which point do you say that it is stiff enough? I am sure we could all build cars with an insane amount of torsional stiffness, but what for? When the car is driving, where do the torsional loads come from? For a car that is designed to run with an equal F/R roll distribution, having a chassis that weighs 100lbs but has a torsional stiffness of 3100 ft*lbs/deg doesn't make much sense. When you could have a much lower weight and torsional stiffness which will suffice for your car.
Anyway, just my .02.

F/R roll distribution ASSUMES all we are dealing with is stedy state cornering. The real world is that the chassis is the mass that the dynamic loads from the shock and spring push and pull. The load that they inflect on the chassis is huge. The loads are almost impossible to calculate with any degree of confidence. I would guess the loads could be measured but only on a sorted chassis with sorted shocks/springs. We seldom have that much time. It is true that there must be some point of diminishing returns. I have not found that point. My car is in the 3000 ft-lbs/deg at 1000 lbs with driver and about 800 lbs of downforce at speed. The adjustibility of the shocks, springs and sway-bars is a God-send.

see http://ncs-stl.com/racecar/MVC-786F.JPG

http://ncs-stl.com/racecar/MVC-787F.JPG

A simple test that has some validity (but subjective) is to replace the shocks with solid links. Support the two rear tires and one front tire (or even better alignment plates in place of the wheels/tires). One front tire will be hanging in the air. This will apply as much stedy state twist that cornering can possibily impart. Now measure the twist of the chassis at the shock bulkheads. I have found that about 1/16" twist at the shock mounts (about 14" apart) is enough.

see http://www.ncs-stl.com/Images/Sample2.jpg

I have seen cars that this test, with shocks and springs attached, only twist the chassis and don't compress the springs at all. I threw the car away and started over.

Hope this helps

Andy Whittle

[This message was edited by awhittle on November 19, 2002 at 09:06 PM.]

True, there are obviously other loads being imparted to the chassis. However, like you said, it is difficult to quantify these dynamic loads. The team I used to be on, decided that keeping the torsional stiffness an order of magnitude greater than the difference between the F/R roll stiffness was adequate for our conditions. (driver skill or lack there of, suspension design, CG location ect.) This seemed to provide a stiff enough frame that changes to the suspension, changed the driving characteristics of the car. BTW what is the stiffness/weight of your chassis?

Good points by all.

An analysis can be done by measuring the torsional stiffness of the chassis with loads applied through the suspension and the actual springs and rates into an infinitely stiff frame. This will give you the maximum theoretical chassis stiffness. Then, if you vary the frame stiffness from zero to a reasonable number, the chassis stiffness vs. frame stiffness will be found to be logarithmic. If a "specific" stiffness (stiffness/weight) function is known for the approximate type of frame geometry you are using, a plot of % theoretical max chassis stiffness vs. weight can be made. This may help determine the chassis stiffness you may want to aim for.

To optimize your model, (and to validate it once your car is built) the use of rigid links for twist testing (both in the model and in real life) as Andy stated earlier is very useful. You can get objective measurements and calculate the accuracy of your model!

If you don't have time to check your model with real world tests, make sure your control arms, rocker geometry, and load paths from the suspension are modeled as accurately as possible. This is where much of the frame stiffness may be lost.

Hope you find this useful!
Erich Leonard

[This message was edited by Erich Leonard on November 20, 2002 at 12:43 PM.]

[This message was edited by Erich Leonard on November 20, 2002 at 12:47 PM.]

My chassis is about 110 lbs compleate with all mounting tabs. Almost everything mounts directly to the chassis. Most tubing was 3/4 x 3/4 16 gauge or 1 x 1 16 gauge.

Andy

I've got a similar sort of query...

I've developed a SpaceGASS structural model of my car (I don't have access to solid modelling/FEA software at work). I am confident that my structural model is accurate, and that all suspension members have the correct freedoms/restraints, etc. The spring/damper units are modelled as solid links and there is no facility in the model for an ARB setup.
I have set up a loading/support case similar in nature to that awhittle described and is shown in Milliken RCVD. I've supported the rear and front left stub axles and applied a vertical point load at the unrestrained stub axle.
Statics say that a 2766N force with a 1.46m lever is ~4040Nm (3000ft-lb) applied torque. However this only results in wait for it... 0.00118deg twist measured at the stub axles.
Now my query is this... you can take Millikens calculation of applied torque in two ways. The first is using the vectorial difference in the front end applied force and supporting force x tw/2 which leads to the same torque as mentioned earlier. The other way is to take the difference in the *magnitudes* of the applied force and support reaction. This yields a more sensible 4135 ft-lb/deg torsional stiffness - but makes zero physical sense.
Can anyone who has performed an FEA or structural analysis using this testing methodology shed some light?
I can contacted off the list on scot.costello_at_mitek.com.au.

TIA

Scot
non-fsae privateer

Stiffness is the only way to go!!

Got is sorted. When switching units from kN-m to N-mm my E value for steel got left behind so it was a factor of 1000 out. Oops

Hi everybody...

Our frame team just did the FEA on our new frame and we got a torsionnal stiffness of only 700 lbf-ft/degree... since people were speaking about number around 2000 lbf-ft/degree we did some minor change but couldn't fing any way to change it so drasticaly. Also we have test back our last year frame and ended up with a even lower stiffness. (we do know it did the job quite well).

Maybe we didn`t do it the good way... We fixed the back left a-arm point in xyz, the back right in xz, the front left in z and putted a load in front right. We then calculat the angle with the width of our frame at the front and the vertical displacement of the point. And we found the torque by multiplying the load by the same width of the frame in the front. Dividing the torque by the angle gave us the 700 value...

Anything wrong with that?

Thanks for your help...

David Lemire
Team Leader
McGill Racing Team

I've done the FEA on our chassis in I-DEAS. Currently, it weighs in at a hefty 50kg with a torsional rigidity of 1800 Nm/deg. There's a good possiblility I can get another 5kg off that without losing much strength. For testing, I've used solid links on a simple suspension mock-up. I hold the rf in fixed x,y,z translation and from rotation around the y axis (up). then the lf and rr are held only from y translation.
The force is applied at the lr and the y translation of that point with the width of the suspension gives me the angle.

Ben Beacock
Chassis Design and Technical Coordinator
Gryphon Racing - University of Guelph

Seems to me the best way to test the chassis would be to actually load the outer lower ball joint on one suspension arm (and restrain the other three) while replacing the shocks with rigid links. Since I don't know how to model pivots for FEA to simulate bellcranks (if it's even possible) it would be tough to come up with a good number in CAD...

Lehigh Formula SAE

www.lehigh.edu/~insae/formula

David

My guess is that some tube is in sheer or a bay is not transfering the loads. DAH but how do you find it???

Try viewing the model at about 50x from the side and top view looking for any one bay that seams to be taking more than it's fair share of the total twist. Now at least you can figure what bay is causing the problem. Normaly it tends to be the bay that the driver enters. In my example named sample2 above, the removing of the two tubes right next to the drivers hands drops the strength of the chassis by 1/2. E-mail me a jpg of the chassis and I may be able to steer you in the right direction.

Andy