Depending on what exactly the purpose of this study is, it may be sufficient to model this geometry as a beam (one-dimensional discretization with the given cross-section) or even a shell.

If it has to be a 3D model, the safest way would be to use a highly regular mesh of hexahedral elements ('brick elements') to avoid the issues related to poor mesh quality.

Not the best case study to start learning fem. By the way, note that your problem can be solved by hand in a few steps (torsion of thin-wall sections), so fem is not necessary

IT would be very helpful if you have posted graphical representation of the mesh. My guess is that your element size/thickness ratio is too high, which makes elements too elongated for SOLID element. If it is required for you to use SOLID element, use sizing of the element equal to thickness of the section.

Check option "sizing" on video: https://youtu.be/ODJ11YDC6ac

If not, try with "Treat as beam" mode.

Depending on what exactly the purpose of this study is, it may be sufficient to model this geometry as a beam (one-dimensional discretization with the given cross-section) or even a shell.

If it has to be a 3D model, the safest way would be to use a highly regular mesh of hexahedral elements ('brick elements') to avoid the issues related to poor mesh quality.

Are you using free mesh? If yes, try mapped meshing -you need to break down your volume to smaller cubical volumes and map-mesh them one-by-one.

I totally agree with Hauke, it is best to partition the beam into three parts, i.e., the two vertical legs and a horizontal one. Then meshing is very straightforward. However, since you are already using FEM, make sure that enough element over the thickness of the structure are used. For linear (8-node) brick elements 5 elements is a minimum I would not get below. If you use quadratic (20-node, 27-node) brick elements 2 elements would be already good enough for a solution with "engineering" accuracy.

You can then easily compare your results with the theory mentioned by Denis to make sure that you modelled everything correctly.

Cheers,

Sascha

Hauke Gravenkamp

Sascha Eisenträger

I was at the beginning stage of this project and now have a (I think) better way indeed to model this problem. I implemented it as followed:

- create a z-section

- create 2 keypoints

- create a line between the 2 keypoints

- use the mesh tool to set the line to the z-section and mesh it

- set all DOF's at one side at a constant of 0

- apply a bending moment Mx of 800Nm to the second keypoint

However, if I look at the deflections, there are none. It looks like the bending moment is not applied to the structure correctly. Any way I can solve this?

PS: the goal of this assignment is to calculate the deflections, stresses and strains manually first and then use ANSYS to look if the results are the same.

If I understand your boundary value problem correctly, I would only expect rotations but no deflection. Are the x-rotations also zero? I'm not sure whether torsion is visualized graphically in Ansys Classic as the beam consists of 1D elements.