Torsional Rigidity - FEA issues in Creo

[Alex J.]

Hello,

I'm attempting to analyze the torsional rigidity of my frame in Creo; however, despite my analyses converging, I'm getting obviously incorrect displacements.



The rear is constrained to restrict translation in all directions.
The torsional loads at the front-upper suspension points are set to 130 lbf, acting 9.0125 inches from the CG. This equates to a torque of 195.3 lb-ft.

Now, here's a (rather hilarious) fringe plot of the displacements in the system:



As you can see from the picture, my frame is crumpling up like a soda can, with a maximum displacement of 36 inches, which happens to be the smallest I've achieved so far.

I'm assuming it's constrained improperly, but after attempting several different schemes, I've decided to turn to you folks for some help.

What are your thoughts?

[Z]

Originally posted by Alex J.:
The rear is constrained to restrict translation in all directions.
...
What are your thoughts?

Alex,

Your rear constraints will give you artificially high torsional rigidity numbers.

Don't know about your 36" deflection problems though???

Z

[Alex J.]

I found the issue. I had set Young's modulus for AISI 1020 Steel to 29700 psi rather than ksi. Guess I was more tired than I thought, as I checked the material properties a couple of times last night.

Z: Can you recommend any other constraint systems? Unfortunately, we don't have a car to test FEA results, so we're flying pretty blind. My intention is to design a frame that is sufficiently rigid, based on common sense and experience. Then, when the car is built, we'll devise a test for the torsional rigidity. Such data should go a long way to helping out next year's team.

Thanks.

[MCoach]

Hopefully these will help give you a compass to fly by.

Torsional rigidity papers:


From Bill Riley
http://www.theoryinpracticeeng...20testing%20fsae.pdf

From North Carolina State
http://www.mae.ncsu.edu/klang/.../Steel%20Chassis.pdf

There are quite a few more if you check SAE.org that will give you a good idea for what is needed for chassis stiffness based on the other requirements that you designed your car on (lateral Gs, power, weight, etc.) It's just that these were on the first page of Google. :P


Good luck!

[Alex J.]

Thanks for trying to help, but I have actually read these papers.

My issues come from inexperience. Using the scheme from the Cornell paper, I achieved a torsional stiffness around 8000 lb-ft/deg, which I have a hard time accepting. If I had a pre-existing car to work from, I could test it and verify using FEA. Unfortunately, my team is the first to compete since 2006, and we are starting from scratch.

Thus, I'm hoping someone wouldn't mind sharing the constraint setups that they've verified by experiment. It'd go a long way towards helping me design the lightest frame possible. Right now, I'm sitting at 80-lbs, which I know I can cut down. But, without knowing whether my torsional stiffness is even close to accurate, I'm not comfortable making significant changes to an adequate design.

Alex J

[Z]

Originally posted by Alex J.:
Z: Can you recommend any other constraint systems?

Alex (and anyone else making this same mistake),

Your constraint system in the image above is that of an "encastre" cantilever beam. A torsionally very flexible beam, such as an "H" or "C" section, will become much more torsionally stiff when it has one end "cast into" a wall as you have shown. The difference can be hundreds or thousands of times for initially very flexible sections. Try it with a sheet of paper bent into a "C" shape...

I guess a suitable constraint system depends on your particular "box-of-grief" (ie. FEA software ). If you can apply an equal but opposite system of two forces at the rear of the frame, as you have at the front, then that should be enough (ie. two diagonally opposite forces acting downward, and the other two diagonally opposite forces acting upward).

But if your BoG demands more positional constraints, then a maximum of six is all that is needed (ie. because the frame as a whole has 6 DoF wrt ground). Perhaps 3 constraints (xyz) at one node, two at a second node, and a single constraint at a third node, with all constraints being "orthogonal" (ie. not pointing at each other).

It would be very helpful to other students if you could post here the torsional stiffness numbers for the SAME frame, but with different constraint conditions (ie. yours above, and some less "overconstrained" ones). I would not be surprised to see a 100+% difference, or more if the frame is quite flexible to begin with.

Z