Mesh misalignment

[rakyuv]

I understand the importance of aligning the mesh when it comes to between different resolutions, as mentioned in the FDS User guide 6.3.4.

I have the following mesh

&MESH ID='mesh1', IJK=125,125,40, XB=5250,5375,5250,5375,0,100,     TRNZ_ID='my trnz 1', MPI_PROCESS=0 /
&MESH ID='mesh2', IJK=125,125,40, XB=5375,5500,5250,5375,0,100,    TRNZ_ID='my trnz 1', MPI_PROCESS=1 /
&MESH ID='mesh3', IJK=125,125,40, XB=5250,5375,5375,5500,0,100,    TRNZ_ID='my trnz 1', MPI_PROCESS=2 /
&MESH ID='mesh4', IJK=125,125,40, XB=5375,5500,5375,5500,0,100,    TRNZ_ID='my trnz 1', MPI_PROCESS=3 /

&MESH ID='mesh5', IJK=125,125,40, XB=5250,5500,5250,5500,100,2180, TRNZ_ID='my trnz 2', MPI_PROCESS=4 /

&TRNZ ID='my trnz 1', CC=50,  PC=20 /
&TRNZ ID='my trnz 2', CC=620, PC=260 /

Mesh 5 has a two times coarser grid compared to Meshes 1, 2, 3 and 4, and the former (Mesh5) is on top of the latter (Meshes 1, 2, 3 and 4). I believe I am satisfying 'integral number of fine cells abutting a coarse cell' rule here. However I get the error stating the misalignment between 'Mesh 5' and 'Mesh 2'.

When I tried removing 'Mesh 2' and ran the code, I got the error that 'Mesh 5' is misaligned with 'Mesh 4'. So I am assuming that the 'Mesh 5' isn't aligned with any of the Meshes 1, 2, 3, 4.

Also I went through similar problem from someone else discussing how to solve the problem by visualising the grid in smoke view. In my situation, I can only see one of the grids on smoke view for xz plane. It would be helpful if you could explain how to check for my errors visually.

I have attached my input FDS file.
test_fds.txt

Any help or inputs are kindly appreciated

[drjfloyd]

The case you attached only has four meshes and not five.

In smokeview, when you are displaying the grid typing x,y,or z will goggle the corresponding planes. See Appendix C of the Smokeview User's Guide.

Your TRNZ lines generate extremely large aspect ratios. This is not recommended (see FDS User's Guide 6.3.5).

[rakyuv]

Thank you @drjfloyd for your answer. My apologies for the wrong input FDS file. I attach the correct file here
test_fds.txt

Thank you for the keyboard shortcut for looking at multiple meshes. I found that the keys 'M/m' helps to toggle between multiple meshes if present. However I tried typing m or M, and seem to have no change in the display of the meshes

I understand the problem with my mesh stretching. I will try to maintain my aspect ratio less than 3.

Could you please help me with the Mesh misalignment ?

[rmcdermo]

It is helpful to get handy with Smokeview to fix these issues. In this case, the first step is to add CHECK_MESH_ALIGNMENT=T on your MESH lines and then set T_END=0. Run the case. Then you can zoom in on your mesh boundaries in Smokeview. See below.

&HEAD CHID='step_methane_emission_multi_mesh',  TITLE='Test multi-mesh in methane emissions' /

&MESH ID='mesh1', IJK=125,125,40, XB=5250,5375,5250,5375,0,100,    TRNZ_ID='my trnz 1', MPI_PROCESS=0, CHECK_MESH_ALIGNMENT=T /
&MESH ID='mesh2', IJK=125,125,40, XB=5375,5500,5250,5375,0,100,    TRNZ_ID='my trnz 1', MPI_PROCESS=1, CHECK_MESH_ALIGNMENT=T /
&MESH ID='mesh3', IJK=125,125,40, XB=5250,5375,5375,5500,0,100,    TRNZ_ID='my trnz 1', MPI_PROCESS=2, CHECK_MESH_ALIGNMENT=T /
&MESH ID='mesh4', IJK=125,125,40, XB=5375,5500,5375,5500,0,100,    TRNZ_ID='my trnz 1', MPI_PROCESS=3, CHECK_MESH_ALIGNMENT=T /

&MESH ID='mesh5', IJK=125,125,40, XB=5250,5500,5250,5500,100,2180, TRNZ_ID='my trnz 2', MPI_PROCESS=4, CHECK_MESH_ALIGNMENT=T /

&TRNZ ID='my trnz 1', CC=50,  PC=20 /
&TRNZ ID='my trnz 2', CC=620, PC=260 /

&TIME T_END=0.0 /
...

[drjfloyd]

@gforney run the case thrid post and load the smv file into smokeview. If turn on mesh outlines and type m, I cycle through the meshes. If I turn on the grids by typing g, then m no longer works. Is this intentional?

[drjfloyd]

Your coarse mesh is half the resolution of your fine meshes; however, you have an odd number of cells for the x and y direction of your fine meshes. This means that the midpoint cell of your coarse mesh sees cells from two meshes and this is not allowed.

[drjfloyd]

Looking at the user's guide, this requirement isn't very clear in Figure 6.2.

[rakyuv]

Thank you @drjfloyd and @rmcdermo for your inputs and clarifications regarding the mesh alignment and aspect ratios. I have incorporated everything and the code seems to run well.

Somehow, the option 'm/M' never worked for me in Smokeview. I could circle through 'g' option.

[gforney]

should be fixed in smokeview's posted below (using 'm' hilight meshes while in 'grid view' mode )

https://drive.google.com/drive/folders/0B_wB1pJL2bFQc1F4cjJWY2duWTA?resourcekey=0-J_yHat7iJoBp1fk6QS8gMA&usp=sharing

[rakyuv]

Hello everyone,

I am reopening this issue. I followed the recommendations, and I constructed the grid. I still get the mesh misalignment. However, with Smokeview, I don't see any sort of misalignment. I have attached the image here

This is my fds file
test_fds.txt

Any help or input would be kindly appreciated

[drjfloyd]

In the image the meshes have different Z ranges. When I look at the file you attached, all the Z ranges are the same. Did you attach the correct file?

[rakyuv]

I think the way I had taken the screenshot (in the previous comment) might give the impression that it is the Z grid. However, the image shows the XY- grid. And the image only shows one half of the grid, because I wanted to show the two refinements in a single image. Full XY- grid is

The central grid is

The Z ranges are the same for the entire domain. I have attached the correct FDS file corresponding to the image

[drjfloyd]

You have the same problem. Coarse cells can only see fine cells in one mesh. It says mesh 4 + 8 and 5 +9 are not aligned.

The boundary of mesh 4 is from x=5400 to x=5580 with a 9 m resolution.
The boundary of mesh 8 is from x=5220 to x=5760 with a 27 m resolution.

What are the cell edges of mesh 8 at lower x for mesh 4? The cell edges of mesh 8 are at 5382 and 5409. So the mesh 8 cell is seeing cells from two meshes.

[rakyuv]

Thank you @drjfloyd for your answer. Could you please explain what you mean by cell edges of a mesh ? How did you get the numbers 5382 and 5409 ? I am not able to understand how mesh 8 cell is seeing cell from two meshes.

If by seeing cell from two meshes, you mean that size of mesh 8 is long enough that it meets different meshes, then the y value of mesh 6 goes from y=4680 to y=6300, and therefore it meets mesh 2, mesh 8 and mesh 9. I don't see any error from this direction.

Could you please tell me where I am going wrong ?

[drjfloyd]

In the input file you attached an hour ago your mesh lines for meshes 2, 4 ,and 8 are:

&MESH ID='mesh2', IJK=20,60,40, XB=5220,5400,5220,5760,100,1100, TRNZ_ID='my trnz 2', MPI_PROCESS=1, CHECK_MESH_ALIGNMENT=T /
&MESH ID='mesh4', IJK=20,20,40, XB=5400,5580,5580,5760,100,1100, TRNZ_ID='my trnz 2', MPI_PROCESS=1, CHECK_MESH_ALIGNMENT=T /
&MESH ID='mesh8', IJK=20,20,40, XB=5220,5760,5760,6300,100,1100, TRNZ_ID='my trnz 2', MPI_PROCESS=2, CHECK_MESH_ALIGNMENT=T /

This means the grid cell edges for mesh 8 are 5220, 5247, ... 5382, 5409, ... 5760.
Mesh 4 starts at 5400 and the next cell edge is 5409.
Mesh 2 starts at 5220 and ends at 5400 making the last two cells 5382 to 5391 and 5391 to 5400.

So the mesh 8 cell that goes from 5382 to 5409 sees two cells from mesh 2 (5382 to 5391 and 5391 to 5400) one cell from mesh 4 (5400 to 5409). This is not allowed. The mesh 8 cell from 5382 to 5409 can only be adjacent to cells from a single mesh.

[rakyuv]

Thank you @drjfloyd for your patience and for making it completely clear. I certainly understand where I went wrong