[Kosuke Seto]

Hi,

I'm simulating a incompressible flow around blades.
I wanna do it parallel, so I use decomposePar command but it doesn't work well.
Here's a part of log file of decomposePar.

    Processor 55
    Number of cells = 44285
    Number of faces shared with processor 50 = 1401
    Number of faces shared with processor 51 = 685
    Number of faces shared with processor 52 = 294
    Number of faces shared with processor 53 = 1511
    Number of faces shared with processor 54 = 2569
    Number of processor patches = 5
    Number of processor faces = 6460
    Number of boundary faces = 2488

Number of processor faces = 184549
Max number of cells = 45166 (0.998123% above average 44719.6)
Max number of processor patches = 14 (71.179% above average 8.17857)
Max number of faces between processors = 9366 (42.1021% above average 6591.04)

Time = 0
AMI: Creating addressing and weights between 6612 source faces and 6612 target faces
--> FOAM Warning : 
    From void Foam::advancingFrontAMI::checkPatches() const
    in file AMIInterpolation/AMIInterpolation/advancingFrontAMI/advancingFrontAMI.C at line 71
    Source and target patch bounding boxes are not similar
    source box span     : (0.100001 0.0736783 0)
    target box span     : (0.100001 0.0736783 0)
    source box          : (-0.0632358 0.00793961 3.6765e-06) (0.036765 0.0816179 3.6765e-06)
    target box          : (-0.0632358 0.00793961 -0.162136) (0.036765 0.0816179 -0.162136)
    inflated target box : (-0.0694464 0.001729 -0.168347) (0.0429756 0.0878285 -0.155925)
AMI: Patch source sum(weights) min:0.999716 max:1 average:0.999999
AMI: Patch target sum(weights) min:0.999998 max:1 average:1
AMI: Creating addressing and weights between 25088 source faces and 25088 target faces
--> FOAM Warning : 
    From void Foam::advancingFrontAMI::checkPatches() const
    in file AMIInterpolation/AMIInterpolation/advancingFrontAMI/advancingFrontAMI.C at line 71
    Source and target patch bounding boxes are not similar
    source box span     : (0.0610814 0.0425658 9.69278e-06)
    target box span     : (0.0610814 0.0425658 9.64055e-06)
    source box          : (-0.0279929 -0.00727137 -5.85086e-06) (0.0330885 0.0352944 3.84193e-06)
    target box          : (-0.0279929 -0.00727137 -0.162145) (0.0330885 0.0352944 -0.162136)
    inflated target box : (-0.0317154 -0.0109939 -0.165868) (0.036811 0.0390169 -0.158413)


--> FOAM FATAL ERROR: 
Unable to set target face for source face 25087

I don't understand the meanings of these error well.
What is the "box" in this case?
And what can I do to solve this?

[zordiack]

The box is referring to the bounding box of you AMI patches. I'm not sure if the error is related to decomposePar at all, have you tried running your case without decomposing?


Anyway, I would check the AMI patch geometries carefully, you seem to have some mismatch between target and source patches and AMI is pretty picky about geometric matching. If you have non-matching patches, you need to use ACMI instead, but I'm pretty sure this is just a tolerance issue.

[Kosuke Seto]

Thanks for replying.
I also thought that it may be an independent problem from decomposing, so I tried running without decomposing, and got the same error.

I'll try ACMI from now on, but what does "just a tolerance issue" mean?
Is there some parameter I can modify about AMI?

[zordiack]

What I meant was that try to check if your AMI patches are geometrically in the same place and that you have sufficient mesh density to resolve the AMI patches. The problem is likely caused by too large mismatch between your source and target patches. Check the third coordinate where the difference is. There is a matchTolerance parameter you can set and try to tune for AMI patches, but the first thing to try should be improving patch matching.

[Kosuke Seto]

I do understand now what you mean. I’ll check my mesh.
Thank you for you politeness.