Is there some documentation comparing these two for using with Fluent? May somebody has an experience with both? What would you suggest?

Depends upon what you wan to do. Tools are different.

Gambit is more comprehensive in the overall meshing since you can do edge, face, vol mesh etc. You can get different vol mesh as well (map, cooper, tgrid). As far as I understand, TGrid needs to have a surface mesh to begin generating a vol mesh. But if you have complicated geom and want to generate vol mesh, TGrid is a wonderful tool with many features to fix cells etc. So yeah, the bottom line is it depends on what you want to use it for and what resources you have. If you don't have a tool to generate surface grid, I guess having TGrid isn't much use.

Hi. Can you find some example where gambit fails or does a poor job, but tgrid does not. Using gambit is more or less like using cad in building geometry. Is tgrid different? It has a possibility to built a geom and a mesh as well?

Pavs is right. Tgrid is good for volume meshing of high quality surface meshes. However it has no geometry core it can only handle and import meshes. Geometry core of FLUENT package is included in Gambit.

Tgrid is useful for generating boundary layer over 3D geometries.

[hitzhwan]

Originally Posted by Pavs
;149040

Gambit is more comprehensive in the overall meshing since you can do edge, face, vol mesh etc. You can get different vol mesh as well (map, cooper, tgrid). As far as I understand, TGrid needs to have a surface mesh to begin generating a vol mesh. But if you have complicated geom and want to generate vol mesh, TGrid is a wonderful tool with many features to fix cells etc. So yeah, the bottom line is it depends on what you want to use it for and what resources you have. If you don't have a tool to generate surface grid, I guess having TGrid isn't much use.

Does the Tgrid only produce Tetrahedral or tiangle mesh, it cannot produce the quadrilateral or quadrilateral mesh？

[vinerm]

Fluent Meshing cannot generate structured hex cells, however, newer versions do have other options, such as, hex core, poly hex-core, cutcell, polyhedral, etc.

[hitzhwan]

Originally Posted by vinerm

Fluent Meshing cannot generate structured hex cells, however, newer versions do have other options, such as, hex core, poly hex-core, cutcell, polyhedral, etc.

I find there are many unknown problems with the Mesh software, have you met before? I choose the total thickness inflation， I just choose the maximum thickness as 1mm, but it does not reach 1mm after meshing.

I am very distressed， I do not know which mesh software should I know. Do you have any suggestions?

[vinerm]

The choice of meshing tool is only based on one thing; user's comfort, until and unless the objective is only to learn a tool. So, use a tool that you are comfortable with. ICEMCFD and Fluent Meshing are not meant for beginners in CFD until you have some proper guidance and training. Ansys Meshing on the hand is not as sophisticated but easier to use. So, if you don't want to waste a lot of time in meshing, I'd suggest you use Ansys Meshing and generate the mesh. The best approach is not to spend time in meshing but at CAD. Decompose the bodies into smaller bodies and ensure that there is no region that has small faces or sharp cuts. Then, Ansys Meshing does a fair job.

When you specify maximum thickness, then it is maximum but it doesn't mean that inflation should reach that thickness. Don't worry about nitty-gritties of 1 mm and 0.8 mm. It's all good.

[hitzhwan]

Originally Posted by vinerm

The choice of meshing tool is only based on one thing; user's comfort, until and unless the objective is only to learn a tool. So, use a tool that you are comfortable with. ICEMCFD and Fluent Meshing are not meant for beginners in CFD until you have some proper guidance and training. Ansys Meshing on the hand is not as sophisticated but easier to use. So, if you don't want to waste a lot of time in meshing, I'd suggest you use Ansys Meshing and generate the mesh. The best approach is not to spend time in meshing but at CAD. Decompose the bodies into smaller bodies and ensure that there is no region that has small faces or sharp cuts. Then, Ansys Meshing does a fair job.

When you specify maximum thickness, then it is maximum but it doesn't mean that inflation should reach that thickness. Don't worry about nitty-gritties of 1 mm and 0.8 mm. It's all good.

I am trying to use the mesh， but you see the mesh attached the inflation，it is a square shape and has a big jump height， it is not what I want, I just want it to be have a similar height as the inflation layer, how can I deal with it ? I find many ways ,but all does not work.

[vinerm]

I am sorry, did you attach an image of mesh? I couldn't see any. Inflation has to match with the edge sizing. Hence, if you are observing a jump in the inflation, you need to change edge sizing.

[hitzhwan]

Originally Posted by vinerm

I am sorry, did you attach an image of mesh? I couldn't see any. Inflation has to match with the edge sizing. Hence, if you are observing a jump in the inflation, you need to change edge sizing.

Sorry, I forget it, you mean the wall edge sizing, it will increase the mesh elements, I do not want to do like this, I just want to change the mesh beside the inflation.

[vinerm]

The inflation layer look as it should be if you have inflation on two adjacent boundaries. And you are using a lot of layers. Usually, 20 is more than enough.

[hitzhwan]

Originally Posted by vinerm

The inflation layer look as it should be if you have inflation on two adjacent boundaries. And you are using a lot of layers. Usually, 20 is more than enough.

You mean we cannot make inflation layer based more than one adjacent boundaries? As we all know that, the mesh should not increase dramatically, all it would lead to a bad mesh. I do not want to have so much inflation layer,but if I use less ， it will give a bigger difference mesh size between them, it is the problem I have to solve? I have set the min global mesh size，but it does not work, I do not why？

[vinerm]

You can certainly use inflation on the adjacent walls and the results as per that actually look good in your mesh. You are worried too much about the mesh quality. As long as you get skewness less than 0.96 and orthogonal quality higher than 0.1, it doesn't matter that much. Yes, there should not be sudden expansion but that range is quite large. What you see in your mesh is hardly a ratio of 2 while you can go as high as 40-50.

[hitzhwan]

Originally Posted by vinerm

You can certainly use inflation on the adjacent walls and the results as per that actually look good in your mesh. You are worried too much about the mesh quality. As long as you get skewness less than 0.96 and orthogonal quality higher than 0.1, it doesn't matter that much. Yes, there should not be sudden expansion but that range is quite large. What you see in your mesh is hardly a ratio of 2 while you can go as high as 40-50.

You may mistake my main goal， I just want to know is there any ways to adjust the mesh size beside the inflation? Thus I can reduce the inflation layers.

[vinerm]

You can give sizes to the face, to the edges, even to the bodies. For all of these, there are various ways, such as, you can directly specify number of nodes on edge; for faces and bodies you can directly specify sizing or use bodies or sphere of influence, wherein another body or a virtual sphere located at a user provided location and of a user specified size can be used to refine or coarsen a mesh within the vicinity. Inflation is not used for mesh control rather for finer mesh close to the wall to resolve flow boundary layers. Those are not always required. Their requirement is driven by what you want to simulate. E.g., you don't need inflation layers if you want to study mixing or combustion.

[hitzhwan]

Originally Posted by vinerm

You can give sizes to the face, to the edges, even to the bodies. For all of these, there are various ways, such as, you can directly specify number of nodes on edge; for faces and bodies you can directly specify sizing or use bodies or sphere of influence, wherein another body or a virtual sphere located at a user provided location and of a user specified size can be used to refine or coarsen a mesh within the vicinity. Inflation is not used for mesh control rather for finer mesh close to the wall to resolve flow boundary layers. Those are not always required. Their requirement is driven by what you want to simulate. E.g., you don't need inflation layers if you want to study mixing or combustion.

1.My model is 2D, I used the edge and face size ,but does not work, it only can control the line. I do not know how to use the sphere？ But I know the sphere can only refinement a local mesh, it cannot refinement as a line zone.

2. my calculation is laminar film flow, do you think I need layer or not, and how many inflation layer should I need?

[vinerm]

If you are modeling two-phase flow using multiphase model then you need to ensure that you have at least 4-6 layers within the thin film thickness that you expect for the liquid phase. If the simulation is single phase, then you do not need any inflation at all.