CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   ANSYS Meshing & Geometry (https://www.cfd-online.com/Forums/ansys-meshing/)
-   -   [ANSYS Meshing] Migrating from GAMBIT to ANSYS Meshing (https://www.cfd-online.com/Forums/ansys-meshing/86755-migrating-gambit-ansys-meshing.html)

David-CFD March 31, 2011 23:21

Migrating from GAMBIT to ANSYS Meshing
 
I used to mesh in GAMBIT and I had already gotten acquaintance, but now it seems GAMBIT will be replaced with the pair DesignModeler and ANSYS Meshing and so I have started to mesh in these applications.
The problem is I realized a very different meshing process with these softwares, and also some limitations there are not present in GAMBIT. For instance, the following are some of the problems I have found:
1. There is not possible to mesh section by section, thus the user cannot mesh the parts in a preferred order according to his own criterion. Moreover, big meshes (too many cells) need to be meshed in robust computers since the mesh need to be completely generated in just one step.
2. Only one source/one target parts can be meshed with hexahedrons using the Sweep method. Neither Thin Sweep nor Multizone methods sufficiently help to mesh parts with more than one source (or target, or both) which easily can be meshed using GAMBIT. (A body located between two sweepable bodies cannot be meshed with Thin Sweep method because of the predefined meshing order of the software.)
3. The Mapped Face method is not as user controlled as the methods which are available in GAMBIT, if not it works like if the software decides whether using quadrilaterals or right triangles.
4. Once a geometry have been repaired and cleaned up in DesignModeler, ANSYS Meshing does not read this geometry exactly in some cases, thus forcing the user to review the geometry and use virtual topologies. (In GAMBIT there is not that problem because the both geometry repair and meshing are made therein.)
5. The fact of connecting the bodies by forming a part does not allow seeing exactly in DesignModeler if there exist connection problems and sometimes they have to be noticed through errors when meshing. Furthermore, this issue can alter significantly the geometry and thus the numerical results of the simulation.
6. Perhaps of less importance, I have not found the way to visualize the vertices in DesignModeler and so there is a bit difficult to quickly find details like edges which need to be merged, for instance.
As I am new in DesignModeler and ANSYS Meshing, I do not know if the aforementioned issues are actually problems in these softwares or if they have simple solutions. Therefore, I would thank to everybody who comments about these issues, as well as different issues in these Workbench applications.

sac April 1, 2011 05:22

1. There is not possible to mesh section by section, thus the user cannot mesh the parts in a preferred order according to his own criterion. Moreover, big meshes (too many cells) need to be meshed in robust computers since the mesh need to be completely generated in just one step.

You can do this in v13.0.

2. Only one source/one target parts can be meshed with hexahedrons using the Sweep method. Neither Thin Sweep nor Multizone methods sufficiently help to mesh parts with more than one source (or target, or both) which easily can be meshed using GAMBIT. (A body located between two sweepable bodies cannot be meshed with Thin Sweep method because of the predefined meshing order of the software.)

Multizone can have multiple sources (targets are also sources in the MZ method).

3. The Mapped Face method is not as user controlled as the methods which are available in GAMBIT, if not it works like if the software decides whether using quadrilaterals or right triangles.

Don't have the manual in front of me but I'm pretty sure you can.


4. Once a geometry have been repaired and cleaned up in DesignModeler, ANSYS Meshing does not read this geometry exactly in some cases, thus forcing the user to review the geometry and use virtual topologies. (In GAMBIT there is not that problem because the both geometry repair and meshing are made therein.)

Never seen this.

The only occasion that I can think you are refering to is that you are subtracting volumes and they don't come through exactly as you expect they would if you did the same operations in gambit.

EIther way you can check how the model will be displayed in meshing using the shared topology control in DM.


6. Perhaps of less importance, I have not found the way to visualize the vertices in DesignModeler and so there is a bit difficult to quickly find details like edges which need to be merged, for instance.

Use the automatic repair tools (under tools) to pick up short edges, slithers, voids etc...




All times are GMT -4. The time now is 12:20.