Tubular Frame Analysis

[B0ND]

Hi, I'm in need of some advice and help. Currently I'm trying to design and analyze a tubular race car frame. My first design is shown below:View attachment 2248

After I finished this design I tried to analyze it using Mechanica to find the torsional rigidity. When I started my Design Study, Mechanica gave me an error stating that my geometry was invalid, and then proceeded to take a while doing (apparently) nothing; I say this because when I tried to view the results later there wasn't anything there.

I believe that the reason for this error was the way I modeled it, which was to first create datum planes, then datum points, then sketches between the points, then swept protrusions along sketches. This method gave me lots of overlap of my tubes (which I didn't think would be a problem at first....guess I was wrong).

So I was wondering if there is any way for me to fix my model easily and quickly so that I can perform analysis? Also since I will be creating more/updated models, is there a method you have found best for doing these sorts of designs? (I posted about this in the Modeling forum as well)

Also of note, I have access to: Mechanica, ANSYS 10, MSC Visual Nastran 4D and MSC Patran. Which of these programs would you recommend for my analysis? I'm most familiar with Mechanica, barely touched Nastran, and never used ANSYS or Patran.

Thanks in advance!

Eric

 

[Israr]

3D frame structure using beams idealizations. Check boundary conditions.


Pro/Mechanica is ideal for this.


Israr

 

[indigo2]

Another way of saying what Israr mentioned is to use the 'sketches between points', and apply beam properties to them. You get the same results, and the net result is a much shorter solution time. There is (typically) no need to analyze the behavior in a given cross section of the tube, as contrasted to determining the behavior of the frame structure.

I can't really provide comments about other solvers for this, but Mechanica has some pretty good functionality for doing this type of analysis. I would suspect that other tools would work well, too, since the basic premise of applying beam properties to line segments / curves should be readily available in most analysis packages.

If you *really* had to have results showing the behavior through the frame tubes, you could consider using shell pairs, and merginging the assembly of tubes into a single part. The shell pair concept is another idealization that improved solution time while maintaining accuracy. Essentially, it computes the solution at the midplane of the given geometry, rather than through the entire wall.

The merge model concept would eliminate the problem of the tubes overlapping, and allow the meshing tool to create the elements.

The beam model solution is fast and very accurate, so if it were my job to run this analysis, that's where I'd start.

Hope that helps-
-tsl

 

[CaptnPea]

I'm with Indigo.... but, want to add a little more.


What is the material you are considering, and how are you putting the tubes together at the nodes? I learned the hard way to be really careful how the beams are connected to each other. If you are welding you need to be careful about the heat effected zone and watch your stresses there (especially true if using Al alloy tubing). If you are really going for optimal strength/weight (F1 style), you may want to do some breakout modeling and really get into the design at the nodes.


If your just going to pile the corners full of steel weld (NASCAR style) this may not be too much of an issue, but I'm willing to bet the first mode of failure will be a weld breaking, not a tube buckling.