Paul, Assuming this element works like ChElementHexaANCF_3843, these dimensions are used to define the straight and undeformed element configuration. This configuration is then mapped back to the bi-normalized configuration (lengths of -1 to 1 for each side) for numerical integration as well as to both the reference configuration (zero stress reference state of the element - potentially a non-straight shape) and the current configuration (defined by the current nodal locations). The code is setup to assume that the initial locations of the nodes define the zero-stress reference configuration.
So again, assuming that this element works like ChElementHexaANCF_3843 as I would expect, you should be able to use reasonable values for these element length dimensions. One solution would be to use approximate average element dimensions. Another solution would be to measure the distance between the centers of the corresponding faces for each of the 3 directions (e.g. Face ADEH to Face CBFG for the "x"/"u" direction). Note that there is a text diagram of the brick element describing the nodal ordering in ChElementHexaANCF_3843.h below the include statements. I hope this helps. Mike On Tuesday, May 21, 2024 at 7:38:44 AM UTC-5 [email protected] wrote: > Hi Mike Taylor, > > Thank you for the reply. The problem is that you seem to have to know the > dimensions of the elements. If I take, for instance, the > "demo_FEA_hexaANCF_3813.cpp" demo and don't specify the InertFlexVec in > line 167, the simulation starts, but nothing happens. > > If I use meshing software to create a hexahedral mesh, the elements don't > necessarily need to have the same edge lengths per dimension. Thus, I can't > set the InertFlexVec vector, can I? > > Kind regards, > > Paul > > Mike Taylor schrieb am Dienstag, 21. Mai 2024 um 04:32:27 UTC+2: > >> Paul, >> >> The "ChElementHexaANCF_3813" element is really a traditional 8 node >> hexahedral element (nodes with only position coordinates). If you Google >> "sphere hexahedral mesh" and look at the images, you'll see some approaches >> to meshing a sphere with only hexahedral elements. If you are trying to >> mesh an ellipsoid, one approach would be to start with the hexahedral mesh >> of a sphere and then scale it along the three axes to match your >> ellipsoid's geometry. A FEA meshing program might help here depending on >> how fine a mesh you need. >> >> As for other "solid" elements, there is code for the >> ChElementHexaANCF_3843 element (ChElementHexaANCF_3843_MR_Damp) that >> implements a 2-term Mooney-Rivlin material law with a non-linear single >> coefficient Kelvin-Voigt damping model in the fork I used for my PhD thesis >> (https://github.com/taylome/chrono/tree/thesis/hyperelastics). This is >> based on Chrono 7 and the code requires updates to use it in Chrono 9. >> Also note that the ChElementHexaANCF_3843 element uses 12 coordinates per >> node, 3 position coordinates and 3 position vector gradient coordinate >> sets. The position vector gradient coordinates can make the element a >> little harder to use from a meshing perspective. Long term, the goal is to >> merge the hyperelastic code developed in this fork into the main Chrono >> repository after some more additional development. >> >> Best Regards, >> >> Mike Taylor >> On Friday, May 3, 2024 at 5:07:33 AM UTC-5 [email protected] wrote: >> >>> Hello everyone, >>> >>> I want to include an ellipsoidal geometry in my model and simulate it >>> with finite elements. My problem now is that I want to use a hyperelastic >>> material model (e.g. Mooney-Rivlin), and only the >>> "ChElementHexaANCF_3813" element implements that. As I take it from the >>> examples, you need a structured mesh where the hexahedral elements have >>> known dimensions, which is difficult for an ellipsoid. >>> >>> I'm quite new to FEA and have only the following ideas: >>> >>> - divide the geometry into multiple simpler boxes (don't know how to do >>> this automatically yet) >>> >>> - add the hyperelastic material model to the other elements >>> >>> - use an external FEM solver and implement an interface to Chrono (can't >>> estimate how difficult this would be) >>> >>> Maybe you can help me find the best option. >>> >>> Thank you, and kind regards, >>> >>> Paul >>> >> -- You received this message because you are subscribed to the Google Groups "ProjectChrono" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/projectchrono/5ff6080b-7cb8-4ad5-b625-be6c23a05bd8n%40googlegroups.com.
