Issue when using grid stretching
 

Hi everyone, hope you're all well.

I've been optimising a model by changing the resolution of the grid in the area where I intend to place the solid of interest. Once I add grid stretching with option 8 - "three zones with constant d[...]" - the execution of Reef3D fails with the following message:

ERROR detected by Hypre ... BEGIN
ERROR -- hypre_BiCGSTABSolve: INFs and/or NaNs detected in input.
User probably placed non-numerics in supplied b.
Returning error flag += 101. Program not terminated.
ERROR detected by Hypre ... END

It runs without issues when I use a cubic cells size and a rectangular domain.

Does anyone have the same issue when running the input files?

Thank you in advance.

My system details for reference:

Model Name: MacBook Pro
Model Identifier: MacBookPro15,1
Processor Name: 6-Core Intel Core i7
Processor Speed: 2.6 GHz
Number of Processors: 1
Total Number of Cores: 6
Memory: 16 GB

Use B 2 to define the total number of cells. Otherwise the maximum number of cells is undefined and thus the decomposition incomplete.


Also you can use the output DIVEMesh-Non-Uniform-Mesh.txt to look at the max and min sizes and their locations.

Include B 130 1 and you get gridSpacing.vtk which shows you the grid in Paraview.

Hi Arun, thank you, that certainly helped test a bit more. The B130 is very handy for troubleshooting.

Edit: found the issue that was remaining in my interpretation of how the code in B 123 is used. The z1 and z2 are the absolute coordinate where to start the following section - I though they were additive to each other.

Thank you once again. Good stuff this. :)

Hi, jfbrazao, have you solved this problem? I recently encountered a similar issue while running a Fortran program but don't know where the bug is.

Hi there wangya, I did solve the problem. The problem was misinterpretation of how to provide inputs to the variable grid sections. Testing it, by using B130, and then looking at the gridSpacing file on Paraview, was really helpful. I advise you do a couple tests, print with B130 and inspect your grid - you'll probably find what input you are providing that might be wrong. You just need to run the diveMESH, no need for a full simulation. Hope this helps!

Hello REEF3D team
I am experiencing the same error while using spacing.dat file for generating the grid mesh.
My simulation runs ok while using the given options other than B 101 10, B 102 10 and B 103 10. However, I am getting the similar error while using the B 102 10 option.
I am attaching all the necessary files for your consideration.

Thank you

Regards
KP

Your y-spacing file contains "0.500" twice, so dy becomes zero there - resulting in divide by zero errors.