I’m working on a script where I could see the stresses and deformation caused by hydrostatic pressure on a cylinder. I would like to eventually get into more complex shapes, but I want to see if it works on a simpler case first, in this case a cylinder. The future shapes will still be symmetrical on two axis. So I would like to see if a can get a consistent result on a quarter of a cylinder, taking advantage of the symmetries. While I’m able to get logical looking results on a flat plane (which was my previous step), quarter of a cylinder is not really working. I’m not sure what’s wrong. I would like someone to explain where I’m going wrong with this script. Thanks in advance!
Hello @idilgumruk,
I found two issues in the definition:
The supports were not entirely correct. They should be minimized to avoid disrupting the uniform stress state.
The unit of the input force vector in the MeshLoad component is already force per area. Therefore, there is no need to divide the pressure by the area of the mesh faces. You selected a globally oriented mesh load and calculated the force components via Grasshopper. If switching to “local,” you would only need to specify the pressure in the local z-direction.
I’ve adjusted the support conditions in this script. Warnings appear due to duplicate supports at certain nodes, which could be resolved by refining the node selection.
Additionally, when using a material with a non-zero lateral contraction coefficient (ν), the sides of the cylindrical section are not perfectly straight. I think this is due to the slightly anisotropic mesh and the relatively large mesh faces. Chosing a material with ν = 0 improves the situation, though minor deviations may still occur due to numerical inaccuracies and the flat nature of the triangular elements in Karamba3D.
Here is the script: HW_Hydro_cp.gh (92.5 KB). Thanks for this very interesting example!
Thank you for your quick reply. This was very helpful. I wouldn’t have figured the issues on my own. Going forward as I’m planning to use more complex shapes for the next step, would you recommend using a custom meshing algorithm to avoid anisotropic mesh?
Hello @idilgumruk,
when the mesh is fine enough and the faces are not very distorted the results should be fine. One should always test different mesh densities to check whether the results have converged.
– Clemens
