Hi everyone! I'm a beginer in Fluent and Gambit. Now I'm making the external flow over spheroid. But, I cannot control the minximum mesh space at the wall by structured or unstructured. Someone can help me, How to control it?

Thanks you in advance. Hung

Easiest way is using sizing function.

Another option is using a boundary layer mesh (but this tends to be a huge problem in 3D).

Otherwise, decompose the volume into smaller volumes. Make sure there is at least one edge that protrudes from the spheroid. Control the mesh on this edge and you should control the volume mesh. But this ONLY works in structured meshes! For an unstructured mesh you have to use sizing functions. The unstructured volume mesh algorith is questionable at best if you don't use sizing functions.

Hope this helps and good luck, Jason

Thanks Jason!

I try to use the sizing function and Tet/Hybrid+Tgrid, But the smallest size cannot be reduced to 1e-4 or 1e-5. Could you advance to me about that!

Thanks more! Hung

What is actually happening when you try to create a mesh that small? Are you getting an error, are you hitting the face mesh limit? How big is your geometry? What were the steps you were performing that didn't allow you to mesh that small?

Without getting some more info from you about what you're doing or what's actually going wrong, the only suggestions I can think of are that if you're hitting the 100,000 face element limit, then up that in your defaults. Or if it's because the mesh is too small for the tolerance in Gambit (which is 1e-6), then scale up the geometry and you can scale it back down when you import it into Fluent (Grid->Scale).

Jason

Thanks to Jason!

I try to make grid of the Pheroid, which has the lage axis: 1, small axis: 1/6 And the smallest size is 1e-4 or 1e-5. But I cannnot use the sizing function with Start size is 1e-5. So, could you advance to me the way to solve it. Thanks!

Hung

You still haven't said why you can't use the sizing function with a start size of 1e-5, so I'm guessing that it's giving an error saying the start size is too small which leads me to believe that the mesh size is too close to the tolerance level built into Gambit. I've never tried using a mesh smaller than 1e-3, so that's why I have to guess what's going wrong. The other thing is that you say "and the smallest size is 1e-4 or 1e-5"... did you mean the smallest mesh size you want to use is 1e-4 or 1e-5, or is there something in the geometry that is actually that small?

If the only issue is that the mesh size is too close to the tolerance in Gambit, then you can scale the model up by say 1000 in Gambit, do your meshing, and then bring it into Fluent and scale it back down (Grid->Scale). But I don't think that's the real problem.

I think the real problem is that your mesh is just too small for your geometry. Quick estimate of your face mesh... I assume its a 1 by 1/6 rectangle... then using a 1e-5 mesh on the long edge means there is 1e5 elements on that edge, and (1/6)e5 elements on the short edge. Assuming a quad mesh, that means (1/6)e10 total face meshes!!! A tri mesh will use about 2 to 3 times as many elements, and a spheroid actually has much more surface area than the rectangular plane I just estimated (remember a spheroid is 3d... I basically estimated the mesh on one face of a box... if I estimated the mesh you're going to use, it would probably be about 4 to 6 times the (1/6)e10 estimate I came up with, or (2/3)e10 to 1e10). The default limit is 1e5 on any single face. So your options are to up the default limit for any face, break your spheroid so it's made of smaller faces, or use a larger mesh. Another option is to try the boundary layer mesh. You can use a larger mesh on the surface, and then use a boundary layer mesh that grows off of this face with a start height of 1e-5. I've heard Tgrid (it's a tool that comes with Fluent) works better for creating boundary layer meshes in 3d, but I've never actually used it so I couldn't tell you how to do it.

The next issue is that you're going to have a rediculous volume mesh if you don't up your mesh size. Those numbers of (1/6)e10 is just for the face. That means you have about 1.67 billion face elements. Even if you don't worry about the volume mesh and just look at this many face elements you're going to need a rediculous amount of CPU power to handle that. Once you start dealing with a volume mesh, you're probably going to need at least 50 elements perpendicular to the surface, so you're talking about 83billion volume elements (assuming you can use a Structured mesh or Cooper scheme to grow the mesh in 3D... a tet mesh will of course be even more). Now, I don't know what you have for resources, but I don't even think I have enough resources within the company (counting about 15 batch machines and about 10 local machines I can network together) to open an 80billion element mesh.

Looking at this leads me to believe that your only hope is to get the boundary layer option to work. This way you can use a larger mesh on the surface, and then the height from the surface is controlled by the boundary layer mesh and you can try to use the 1e-5 for that height. That's also assuming you need such a refined mesh. Maybe you can get away with a coarser mesh. Why did you choose 1e-5 as a mesh size?

Hope this helps, and let me know about why the mesh size was chosen. Maybe a larger mesh can be used and I would be willing to help you estimate that mesh size. Jason

use wall function on the boundary

Thansks to all!

I used the multi-domain method around the body. And in each domain, I used the Map or Cooper scheme! Now I try to use Sizing function and Boundary for complex geommetry, hoping your help!

Thanks one more!

PAH