[Artur]

Hi,

I'm using sHM to create a mesh for a ship hull and a propeller using AMI. I want to see if it is possible to get reasonable results by assuming that the free surface profile is not affected by the prop and only doing the unsteady calculation for a part of the hull. To do this, I want to mesh the free surface shape I get from a separate run and treat is as an impenetrable wall. Then I want to create a smaller domain bounded by the hull, the free surface and other defined patches.

The approach I'm taking is:
1. run ltsInterFoam to get the free surface shape around the boat
2. create a smaller blockMesh
3. use a few scripts to generate .stl files which will represent the AMI cylinder and the bounding box of the smaller domain
4. run sHM (see attached dict file)
5. run topoSet, createPatch, createBaffles and mergeOrSplitBaffles

It works very well when I'm not running in parallel (see picture) but when I do try to run in parallel the sHM log file is cut at this point:

Introducing zones for interfaces
--------------------------------

Surface : amiCylinder
    faceZone : amiCylinder
    cellZone : amiCylinder
Walking from location-in-mesh (-5 0.1 -3) to assign cellZones - crossing a faceZone face changes cellZone

I can just stick to running on a single core if need be, of course, but it would be nice to know what I'm doing wrong here.

Note: without the freeSurface.stl the meshing goes smoothly in parallel. I tried different decomposition methods and no. cores but none helped.

EDIT: Perhaps I should say how I got the .stl file. I used the contour filter in paraview for the volume fraction field, saved it as a .vrml file and then converted that to an .stl (I know there is a way to do it directly but I couldn't find the correct export option for some reason...).