|
[Sponsors] |
May 7, 2022, 03:02 |
Best practice: Meshing for MRF-Approach
|
#1 |
Member
Wolfram Schneider
Join Date: Jan 2018
Location: Germany
Posts: 57
Rep Power: 8 |
Hello everyone,
I slowly become acquinted with OpenFoam and cfMesh since I am fascinated by the opensource philosophy. My next step is the MRF approach, for what I started with a tutorial. It is a rotating impeller in a vessel as shown below. Screenshot 2022-05-06 111036.jpg Approach 1: - Meshing with cfMesh as single domain, using the impeller.stl and vessel.stl as boundaries - Implement the MRF-Zone by means of { name cellMRFzone; type cellSet; action new; source cylinderToCell; sourceInfo { radius 0.175; p1 (0 0.06 0); p2 (0 -0.06 0); } } { name cellMRFzone; type cellZoneSet; action new; source setToCellZone; sourceInfo { set cellMRFzone; } } within the topoSetDict - From my understanding it creates a cylinder within the domain, where the MRF approach should be applied. - Now my question: ○ What happens with the cells, that are located on the boundary of the cylinder? How are they treated regarding rotating and stationary reference of frame? Screenshot 2022-05-07 083439.jpg Approach2: - Meshing two domains with cfMesh and merge them afterwards (not succesful yet) ○ First domain: Vessel surface to "MRF Interface" ○ Second domain: "MRF Interface" to Impeller surface ○ ( the "MRF Interface" is a cylinder as .stl) - Implement the MRF Zone by means of topoSetDict and use the cells inside the "MRF Interface" Screenshot 2022-05-07 085146.jpg - Now my question: ○ Is this approach more preferable since there are no "cutted" cells? ○ So far, I have not managed to let it run. The mesh is merged, but the "MRF Interface" is still recognized as boundary and is not "ignored" ---> Does anyone have an idea how to "glue" the mesh together? The Nodes are not conformal Additional questions How is the interface between arbitrary rotating mesh and stationary mesh modeled? Is it some kind of a "mixing plane" oder "frozen rotor"? Best Wishes Wolfram |
|
July 21, 2022, 02:39 |
|
#2 |
Senior Member
Join Date: Dec 2019
Location: Cologne, Germany
Posts: 355
Rep Power: 8 |
comment for approach 1:
if you execute topoSet on this geometry, you will not get a perfect cylinder, but that is because of your cells. the ones, that are within the defined "cylinder" will be in your MRF-domain. the others will not. you can visualize your cells by executing $foamToVTK -cellSet cellMRFzone comment for approach 2: if you mesh two bodies independently, the boundaries will not match in terms of cell nodes, as you already show. such a mesh could be good for the sliding mesh method where the rotator zone is rotating for the simulation. for the mrf approach you can not use this mesh though, you need one region in total. you could, but not do that for this type of simulation purposes, use "stitchmesh". it will create polyhedral cells with probably high skew and high non-orthogonality. but that is what mean by "glueing" these two zones together. |
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
[Gmsh] gmshToFoam generates patches with 0 faces and 0 points | Simurgh | OpenFOAM Meshing & Mesh Conversion | 4 | August 25, 2023 07:58 |
[Other] Good learning materials for robust meshing techniques in 2021 | FoxInCFD | ANSYS Meshing & Geometry | 0 | November 19, 2021 05:52 |
[ANSYS Meshing] Help about meshing well | berkmm | ANSYS Meshing & Geometry | 1 | May 11, 2020 15:32 |
Map of this Meshing (sub-)Forum | wyldckat | OpenFOAM Meshing & Mesh Conversion | 0 | April 27, 2019 09:33 |
[ANSYS Meshing] Parallel meshing utilisation failing | crc1622 | ANSYS Meshing & Geometry | 0 | February 14, 2019 02:42 |