though my problem is not yet strictly related to Calculix I do assume that there are many Users with huge experience in GMSH and maybe could me give some hints how to solve my problem.
I’m working on a mesh for a ventricular assist device which in general will consist of the following parts:
inner cylinder which will act as a pump – the bottom surface will move up and down and the mesh will remesh
outer torus domain will be stationary
2 connectors
In the beginning I’ve made the geometry in FreeCAD, but after getting meshing errors I’ve decided to build the geometry fully in GMSH.
In order to get a possible good mesh I’ve decided to mesh the axisymmetric geometry figures with the extrude technique.
For the reason that the cylinder mesh will be a dynamic mesh I think I can’t make a stationary mesh with solid connections between the individual domains, but the data exchange between the torus and the connectors should be in the form of interfaces.
I’ve tried a lot of solutions but it seems neither have worked - actually only the gemoetry shown on the picture (with connectors) or the axissymetric (disk+torus) can be meshed without errors.
If I start any boolen operations I get error messages on both: torus ad connector.
I do not ask for a final solution but just a hint in which direction I should search for possible solution – because maybe I’m looking for something which is not possible with the present version of GMSH, though I rather think that the reason is my missing experience and knowledge.
PS
Do apologize if I should not create a new topic - but the last time I logged-in some long time ago and simply forgot how to exactly use the platform.
So it will be FSI simulation ? With what software ?
It seems that there’s an intersection between the parts. Is it necesary ? Can you show some examples of actual components of this kind ? It resembles peristaltic pump but without the rotary motion, right ?
Hello,
many thanks for quick replay.
Exactly, the present goal is to prepare the mesh for a CFD simulation with OpenFOAM.
The disk seen in the middle should deform like a normal piston pump and simultaneously with piston movement the mesh should deform with layering technic.
At a later modelling stage the 2 connectors should be extended and inside 6DOF disk should be installed - just to model the “valve” in the heart.
That’s the plan but I faced problem already in the beginning - i.e. with the proper domain geometry - i.e. attach the connectors without any moving parts to the torus / disk.
OpenFoam works with the INTERFACE technic in case of different meshes exchanging data - I was able to identify the connection surfaces between the connector and the outer surface of the torus - the so called imprint / interface command, but when I try to mesh the connector and the torus + cylinder I get always an error.
I have also tried to unite the connector and the 1/4 of the torus by using the GMSH Fragments commands - also without success. but the interface ate more general and flexible.u
Actually the only domain that I can mesh without getting any errors from GMSH is the one shown on the pictures above, but here I share your opinion that we have intersecting parts any by that the mesh can not be used for any CFD simulation.
I’m aware that I’m here on a Calculix discourse but my question refers to GMSH and not OpenFOAM, and GMSH is being used by some Calculix users, so I allowed myself to post the question here with the hope that maybe someone could give me a hint in which direction to go.
It seems that most such devices use centrifugal/rotary pumps.
More detailed CAD model with separate parts would make it easier to understand exactly how this device is supposed to work here, allowing proper preparation of the FE model.
If you encounter errors that seem to be a fault of Gmsh (not invalid geometry) then I would suggest asking here (but of course you should provide the details about the error and files, if possible): Issues · gmsh / gmsh · GitLab
Exactly, but due to the fact that my problem is actually an academic one (mainly to learn different dynamic meshes technics), without any ambition to be implemented in real live, my drive unit is much easier - no pneumatic one like shown on the picture but a simply mechanic one (without even solving the problem how to move the piston - I will simply make the assumption in the simulation that the piston makes 1 movement in ca. 1s) - but the blood amount in 1 stroke should be similar to an real adult.
The case with the pneumatic drive unit is much more complicated and I assume requires an FSI e.g. CalculiX - OpenFOAM simulation - by that it is also much more ambitious, and also gives much more perspective for future (even the possibility to work on real world problems) - but requires more background knowledge and experience in conducting software simulations.
You are also fully right that the present left ventricular assist devices are mainly axial / rotary pumps but as said though my system seems to be the easiest one I have problems to do it.
So in general your advices are fully right and I do thank you for that - but as one said “before you start to run learn to walk” and this is what I’m currently trying to do, with the hope that one day I will be able to make a full FSI simulation incl. pneumatic drive unit, just as you proposed.
Thank you very much for the hint about the GitLab repo - this is exactly what I was looking for - I thought that there are some GMSH discourse group like e.g. CalculiX or PrePoMax , but could not find any and that is the reason why I decided to look for help from the CalculiX community.
Once more Calc_em, thank you very much for your help - appreciate it very much and will follow your advice.
