RASA82
Mechanical
- Aug 11, 2009
- 1
Dear readers,
My question is how to handle stress peaks which are related to the discretization of the FEM model, in combination with Required safty factors mentioned in Norms such as the British Standard, DIN or Eurocode.
Say two closed rectangular profiles are connected as an L-shaped structure. A shell model of the structure is created and the analysis shows that there is a stress concentration aloong the corner line corner of the L. At the top and bottem of the corner the stress values are lets say 200MPa, but in the entire region between the tio abd bottem the stress is about 60 MPa. When the mesh is refined the stress value at the top and bottem increases (it seems to be unbounded), while the stress allong the rest of the corner stays constant.
It seems a nummerical effect that if stretched to its limits would endup in a singularity similar case. (infinite stress in an infinitly small region).
Radii and Roundings would solve the singularity. But is there any other way arguing that the mentioned region suffers from numerical problems caused by c0 continuity in the model whereas c1 continuity would be required for bounded solutions of the stress peaks in the structure?
Are there any standards that mention the posibility of excluding or separating such regions from the standard treatment?
Thanks for any helpfull sugestions
My question is how to handle stress peaks which are related to the discretization of the FEM model, in combination with Required safty factors mentioned in Norms such as the British Standard, DIN or Eurocode.
Say two closed rectangular profiles are connected as an L-shaped structure. A shell model of the structure is created and the analysis shows that there is a stress concentration aloong the corner line corner of the L. At the top and bottem of the corner the stress values are lets say 200MPa, but in the entire region between the tio abd bottem the stress is about 60 MPa. When the mesh is refined the stress value at the top and bottem increases (it seems to be unbounded), while the stress allong the rest of the corner stays constant.
It seems a nummerical effect that if stretched to its limits would endup in a singularity similar case. (infinite stress in an infinitly small region).
Radii and Roundings would solve the singularity. But is there any other way arguing that the mentioned region suffers from numerical problems caused by c0 continuity in the model whereas c1 continuity would be required for bounded solutions of the stress peaks in the structure?
Are there any standards that mention the posibility of excluding or separating such regions from the standard treatment?
Thanks for any helpfull sugestions