Dear all,
Is a good strategy to simulate thick structures using shell elements instead of solid ones. For example I have a cylinder consisting of 3 parts. The parts are connected with flanges with thickness 5cm on each part and mid--width 15 cm. For cylinder shell elements are used. Regarding the flanges I want to simulate them as rings. Is it good approach to use shell elements (S4R elements in abaqus) for the flange? Will the response getting with shell elements is comparable with solids?
Ps. The cylinder will be subjected to bending loads
Hi Spyros Arvanitis
Here is an article I found useful when simulating bending of a tubular, perhaps you will find it helpful too: https://core.ac.uk/download/pdf/191701104.pdf
As ever with element selection, your choice will depend on the type of analysis, sought variables and level of accuracy you are interested in. A radius-to-thickness ratio of R/t=10 is generally regarded as a 'thick' structure, for which solid elements are recommended. I have surveyed many papers on pipe/tube modelling and typically those treated as 'thin' and modelled as a shell will have ratio significant larger than this (modelling a cylinder with ratio approaching 10 using shell elements may be challengeable).
The thick shell element formulation based on First Order Shear Deformation Theory available in Abaqus can give a reasonable solution for deflection/buckling type problems. However, such elements are not expected to yield a precise stress analysis (moreover through-thickness stresses cannot be obtained). If precise stresses are sought, in particular around your flange-to-cylinder region, solid elements may be appropriate. You can apply half-pipe symmetry to reduce your computational size.
Hope this helps.
JCH
James C. Hastie thank you for your response.The cylinder is thin with R/t = 100 , so shell elements are good choice. My concern is about flanges which they have a mid-width of 15 cm and thickness of around 5 cm of each part.
Ideally they should be simulated by using solid elements, however I want to avoid a shell-solid connection so I am asking if it is sufficient to simulate them as shells.
Hello Spyros Arvanitis
in my study:Conference Paper The Biomechanical Performance Investigation of L4-L5 Lumbar ...
I used shell elements for vertebra and solid elements for intervertebral disc. And I didn't face with any problem during analysis. However, when the model is getting bigger, it might be a problem. I think that you might deal with this situation by forming same nodes/node coordinates for shell and solid side in the contact field between them. Or you can divide the analysis as a shell cylinder and solid flanges. You can use the results of flange connection field nodes, which you took from the cylinder analysis (shell model), for flange analysis (solid model).
Regards
- Badges
- Science topic