relevance of boundaryTypeMap

[turbulentphil]

Dear Neks,

I am coming from Nek5000, trying to get started with NekRS, and wondering what the use of the parameters boundaryTypeMap in the .par file is.

Here is what I can observe from example conj_ht (as an example):

1. The mesh files fluid.box and solid.box already specify the boundary condition types.
2. The .par file specifies boundaryTypeMap = v, O, W and boundaryTypeMap = t, O, I for the velocity and temperature field, respectively. Such statements are new to me.
3. The .usr file specifies in usrdat2 some integer values for boundaryID and boundaryIDt based on the boundary condition type from the mesh files.
4. Interestingly, usrdat2 seems to associate the values {1, 2, 3} for boundary types {v, O, W} (from the velocity mesh), which are the indices of these letters in boundaryTypeMap. (Q0: Is this a coincidence?)

My questions are the following:
Q1) What is the use of boundaryTypeMap? Is it maybe just adding flexibility for people that didn't use Nek5000/genbox to generate the mesh?
Q2) Is there any reason to use the new boundaryTypeMap parameters in addition to boundary types specified in the mesh files using genbox?
Q3) Assuming the relation between usrdat2 and boundaryTypeMap (as in point 4), could one also skip these steps given a definition of the mesh as done in fluid.box and solid.box?

I would appreciate any help.

Best,
Philipp

[yslan]

Hi Philipp,

boundaryTypeMap is another way to specify the boundary conditions. This is also introduced in recent Nek5000 since this version, but they work slightly differently.

• If your .re2 already has boundary conditions and you are happy with it, you don't need to set anything and NekRS will read Nek5000's CBC arrays for BC. Meaning, you can still modify the CBC in userdat2 in Nek5000 way.
• If you want, you can also set boundary conditions via boundaryID. Basically, you name each boundary an integer (1-base, start from 1, cannot skip numbers), then you can assign boundary conditions in .par via boundaryTypeMap. For example, boundaryTypeMap = v, O, W indicating v (inflow) for boundaryID=1, O (outflow) for boundaryID=2 and W (wall) for boundaryID=3
• boundaryIDt is only used when the case is a conjugate heat transfer case, or nelgt>nelgv. Then, the temperature boundary will use boundaryIDt for t-mesh.

This should answer your Q0 and Q2
Q1. Yes, you can consider it as a new feature. Especially for meshes from gmsh2nek that stores bc id in boundaryID.
Q3. No, you still need to tag BC in box files (or somehow recover it in userdat2) otherwise the code won't be able to distinguish the faces.

Hope this helps,
Yu-Hsiang

[turbulentphil]

Hi Yu-Hsiang,

So I understand that (1) if the .re2 file already contains the cbc arrays (and specifies the boundary conditions sufficiently), then I don't necessarily need to deal with boundaryID or boundaryTypeMap. Otherwise, (2) the integers assigned in usrdat2 are the indices from boundaryTypeMap and can be used alternatively to specify the boundary conditions.
Thanks a lot!

Best,
Philipp