Beam elements using CCX solver

Mishal

Hi Guys

Why is it that the CCX solver does not support more of the common shapes of beam elements?

I assume that no one in this forum has any influence on the development of the CCX solver however it would be nice if they expanded it further. It is a common issue to build a FEA model for static analysis only to realize later that you would want to use nonlinear analysis to evaluate certain aspects.

Victor

I'm not sure why but @disla here shows how to build up more complex shapes using offset beams: https://mecway.com/forum/discussion/comment/6229/#Comment_6229

disla

Hi Mishal,

Although this method is suggested in the ccx manual there were some concerns abouts it’s suitability. You should test it before applying to check its performance.

https://calculix.discourse.group/t/square-hollow-section-chassis/1246

There is another trick suggested in the ccx manual. You can use the hallow box to build all the U and L sections setting one or two side thickness to zero, respectively.
This has the problem that the section is not modeled (graphically represented) according to the shape and can be difficult to check the proper orientation.

Why is it that the CCX solver does not support more of the common shapes of beam elements?

Beam elements are internally expanded into volumetric elements in ccx. They are not “beam elements” in the classical way. You could see them as an assembly of solid elements with one element per thickness. At the end , one may question why using this custom-made beam and not directly a beam made of shells or solid elements.

Mishal

Hi Guys

Thanks for the answers.

@Victor, would it be possible to have a tool which converts a beam element section defined using the internal solver to a set of plate elements? Thus allowing ease of use when moving from the internal solver to CCX.

Victor

@Mishal I've wondered about this sort of thing, but it gets difficult at the connections between elements when they're not colinear. CCX does this with quite complex and buggy MPCs as well as deforming the expanded elements to fit each other. It's quite a nontrivial problem, especially without access to the rotational DOFs.

To simply convert each element independently, I think you could do it using the API which can read node coordinates, create new nodes and elements, and set material properties (ie. thickness).

My preferred way forward for other section shapes is the two approaches @disla has described.