Learning Forum

[puh69]

I have a 3D mesh, the symmetry plane of which is shown in the image below. The Overset mesh is flush with the symmetry plane. As a result I previously set this face of the overset block as a symmetry boundary as well. However this led to issues trying to run the simulation. On the otherhand I can set this face touching the symmetry plane to be an overset interface, this runs, however, I'm worried that this is also incorrect as the area where the overset mesh lies no longer uses the symmetry plane conditions here. How am I best to approach this problem?

[Karthik Remella]

Hi,nA solution would be to not flush the overset mesh on to the symmetry planes. I'm assuming that these are the front and back planes (out of and into the screen). Since your computational domain is going to have to be pretty large in these directions anyway, you might want to shorten the overset region in the wake of the wing. The wake is not likely to extend too far in the Y direction away from the wing. If you do this, you can use the overset interface condition here.nKarthikn

[puh69]

Karthik,nnThank you for your response. To clarify I have an O-grid (Semisphere) and this is the 'Front Plane'. The reason I want to have the grid flush with the symmetry plane is so that I can peform a farfield analysis here, likewise why I'm stretching my overset quite far downstream. Can you tell me what's going on so that I cannot have the overset lying flush on the symmetry plane? I suppose a possibility would be to instead mirror the wing on the other side of the symmetry plane, though not ideal for computational processing.nThanks,nPiercen

[puh69]

Still hoping for an update on this. Thanks.nPiercen

[Karthik Remella]

nGenerally, overset is used in Fluent to overlay the component mesh on top of the background mesh with you have a flow feature you wish to capture. Based on your problem, say you had a flap just downstream of this wing and you want to have a mesh around this flap, I'd use overset in such scenarios. Overset in Fluent is, however, not a tool to obtain localized mesh refinement. In your case, you are attempting to refine the mesh to capture the wake of the wing. In this case, you are better off refining your background mesh.nIf you had a flap downstream, then you can extend its solid boundary beyond the plane of symmetry. Your component mesh will also have to be extended beyond this symmetry plane. But, because there is a hole, you might be able to prescribe an overset interface in such a situation. nAlso. my apologies for this delayed response. It seems to have fallen through the cracks.nThank you!nKarthikn

[puh69]

Karthik, While I agree the ideal circumstance is to refine the background mesh, the grid I have for this Onera M6 case is an O mesh, their natrue leads to nice surface data but struggles with the downstream. While you could attempt to refine the mesh ( I even looked at using fluents mesh refinement tool in 2D), the way the Ogrid expands makes this pretty impossible. As a result I'm trying to use this feature to provide mesh refinement in the wake region, which has proved successful in 2D.n

[puh69]

Karthik,I'd just like to add that while I'm not using an overset mesh for what its specifically designed for, one would hope that the boundary conditions still hold as this impacts the overall flow.nnThanks.nnPiercen

[Rob]

Given the inefficiencies of switching between the two meshes I'd consider using a pave mesh and adaption. Same flexibility (assuming you want to move the wake refinement) but much easier to manage and control. n

[puh69]

Given the inefficiencies of switching between the two meshes I'd consider using a pave mesh and adaption. Same flexibility (assuming you want to move the wake refinement) but much easier to manage and control./forum/discussion/comment/104065#Comment_104065

Hi Rob,nnCould you please pass on information, whether a link or detail as to using a pave mesh, I'm not familiar with this. One thing with what I've done with the overset is it lies parallel to the wake with and meshes along the trefftz plane at every downstream station (if that makes sense).n

[Rob]

Now I've looked up a Trefftz Plane: I'm a Chemical Engineer and leave all the simple airflow stuff for the Aeronautical Engineers! Pave is one of the other meshing methods, and gives an unstructured hex mesh so you can grow in two directions. You may lose the perfect alignment you get with the O-grid but gain in the mesh resolution. You also need to be careful with mesh in the extruded direction as you want a good quality mesh to pick up the tip vortex. I'd actually look at Fluent Meshing and all polyhedrals for this and adapt those as necessary. n

[Karthik Remella]

& nI was going to suggest Fluent Meshing - Watertight Workflow. You won't be able to obtain a structured mesh of course, but you will be able to obtain a really good mesh (and that too very fast). You can add a body of influence in the wake region to obtain a good mesh refinement. And as Rob mentioned, you should be able to adapt the mesh in the wake region to capture the details better. nKarthikn