Hi Everyone,
I have been trying to figure out this problem for a while now and I have not been able to find the solution.
What I have:
Particles impacting on a wall.
What I need:
The average forces acting at the nodes of the mesh for a certain number of timesteps.
Alternatively the stress acting on the elements of the mesh for those timesteps.
What I have achieved:
To print the stresses for the full mesh in a txt file.
What I know:
I have read all the Threads in the Forum related to this matter and according to them what I need isn't possible:
"...since the forces are calculated for the whole mesh and not per element, you can't directly dump it."
"You can only calculate the force acting on the whole geometry. You have to split your mesh."
However, when I am post-processing the results in ParaView, I can see the values of the stresses acting on the elements of the mesh. I also can access to this information by opening the dump files ( a bit messy though). This means that LIGGGHTS actually computes these values.
My question: Is there any way to print what I need in a text file and have some control in the way I print it?
Any advise or guidance would be greatly appreciated.
Cheers,
Arnón
PS: Albeit I think the script isn't necessary, I have attached it in case somebody want to use it.
| Attachment | Size |
|---|---|
 drop1.tar_.gz | 11.79 KB |
richti83 | Fri, 09/21/2018 - 09:25
What you are looking for is a
What you are looking for is a new dump-style. I implemented this for the DEM-FEM Coupling Project "SimBA" [1]. You can get the new dump style here:
http://richtisoft.de/transfer/dump_fem.zip
but you have to implement the avaraging function by yourself in stressmodule and store the result in mesh()->prop().getElementProperty >("f_mean")
To identify the mesh-elements one need to extend the stl file and give every FEM Element a unique number, also the stl reader in liggghts needs to be adjusted to read this additional information. the reason for this is, that the internal numbering of mesh elements is not deterministic as every core starts at zero with triangle numbering
A more easier way is to dump in vtk and do all the postprozessing in paraview with a python programmable filter, there is a vtk function to calculate the area of a triangle and when you multiply the dumpes stress with this area you get the forces on the mesh, after that you need to transform the triangle force to the nodes if needed (see [1],[2]).
[1] https://doi.org/10.2314/GBV:871968150
[2] https://www.amazon.de/Kopplung-DEM-Simulationen-Analyse-Gut-Bauteil-Inte...
Arnon | Tue, 09/25/2018 - 11:48
Thanks for your reply
Hi Christian,
Thanks for you reply.
According to your comments, there are two possible ways to obtain what I need. The first one looks very interesting but I don't have the knowledge I need to do it. The second one is less complicated so I will start with that one.
The two links you provided are in German, do you know if there is a English version of this documents. Unfortunatelly, my German is nonexistent. :'(
Regards,
Arnón