element choice - contact, high stresses
element choice - contact, high stresses
(OP)
Dear all,
I am modelling a structure including regions of high stresses and strains (weld, notch)and al lot of contact (3 plates on top of each other - all in contact). The investigated model is a steel connection with bolts. It is modelled 3D. I have modelled the problem with various element types (hexahedral and tetrahedral) plus.... the obtained results vary quite a bit.
Questions
- What is a good/precise/reasonable ement type? Hex is always quite ok, tetra (e.g. C3D4 or C3D10M) is good in contact. Tetra always are bad with peak stresses (notch). Mixing element types is quite impossoble with the model (problems arising). So what?
- If I use hex (eg C3D8R)for the area where two steel plates are pressed together (prestressed ba bolts) the contact areas are quite jagged/craggy after loading (doesnt happpen for C3D4). Why? Is this ok?
thanks for any coments,
regards,
Anton
I am modelling a structure including regions of high stresses and strains (weld, notch)and al lot of contact (3 plates on top of each other - all in contact). The investigated model is a steel connection with bolts. It is modelled 3D. I have modelled the problem with various element types (hexahedral and tetrahedral) plus.... the obtained results vary quite a bit.
Questions
- What is a good/precise/reasonable ement type? Hex is always quite ok, tetra (e.g. C3D4 or C3D10M) is good in contact. Tetra always are bad with peak stresses (notch). Mixing element types is quite impossoble with the model (problems arising). So what?
- If I use hex (eg C3D8R)for the area where two steel plates are pressed together (prestressed ba bolts) the contact areas are quite jagged/craggy after loading (doesnt happpen for C3D4). Why? Is this ok?
thanks for any coments,
regards,
Anton





RE: element choice - contact, high stresses
I always use linear elements in contact rather than quadrilateral elements (20 noded brick) as there are too many problems with the mid-side nodes in contact when Abaqus calculates tries to calculate the contact forces. I would avoid tetrahedral elements if you can as again the midside nodes present a problem in contact and obviously you can't use linear tet elements.
If you're trying to include the effects of a stress concentration then try and sub-model that region rather than include it in the overall contact problem.