Problem of Singularity

[christian.steinfals]

Hi,

i'm currently working on the case of a small device and i'd like to perform a static stress analysis with simulation multiphysics.

Unfortunately the stress-results are almost linearly dependend upon the mesh size.

Using percentaged mesh-size of 100% leads to a maximum stress after van mises of approximately 100MPa under a load of 1800N.

Using percentaged mesh-size of 40% leads to a maximum stress of 1300MPa.

Using e.g. 2.5mm absolut mesh-size even exceeds 1300MPa stress.

Somehow both results don't seem realistic to me...

Well, according to some articles i found this problem ist known as "Singularity", right?

Due to the fact that I lack the necessary experience in FEM-Analysy I'm wondering, how to get reliable results.

I hope you can help!

Thank you for your effort in advance!

Greets,

Christian

P.S.: The Geometry is displayed in the attachment. Its dimensions are 77x41x16mm with a minimal thickness of 1.5mm. Material is stainless steel, aisi 405, if relevant. Load magnitude is 1800N.

[John_Holtz]

Hi Christian,

I agree that it looks like a singularity, but without seeing the mesh and the load/constraints, it is hard to know for sure. Did you use a nodal force? And is the maximum stress at the force?

Here are some options:

1) Ignore the maximum stress. Unless the load is applied with a needle in real life (that is, an infinitely small point), the model is only an approximation, so the stress is exaggerated.

2) Make the model more realistic in regards to how the load is applied in real life.

3) Use plate elements instead of brick elements. (I'm not sure what you used from the images. Maybe you already are using plate elements.) 

[Anonymous]

if you can archieve the model and upload it.....you may get some solutions......

[christian.steinfals]

First of all, thanks for your answers.

Well in fact i used a nodal force, in order to approach the real load, which is applied by a globular component with a diameter of 2mm and yes, the maximum stress occurs at that nodal point.

In the first attempts yesterday i used brick elements. This morning i simplified the model to an ordinary Coboid and chose Shell as the mesh type, using 1.5mm for the thickness of the walls in the element definition, which lead to a similar model in the end. The results were, that the stresses still varied with the mesh size but it seems to converge to 1400MPa 🙂

My current adjustments and model are in the attachment. Would you consider this a realistic simulation-result with regard to geometry and force?

Again, thank you for your efforts, i really appreciate this!