AMI: Minimum patch weight decreases with time
 

Hello all,
I have a problem with using AMI with pimpleDyMFoam. Everything appears to run smoothly at every iteration and converges nicely, however the minimum patch weight creeps lower and lower. At the outset of the simulation the minimum patch weight is at a reasonable 0.97 or so. After each iteration it will go down something like 0.01, and eventually the minimum patch weight is so low tending toward zero that the simulation finally diverges. The average and maximum patch weights stay at reasonable near-1 values.
Any thoughts or suggestions?
Thanks.

Is there a way to procure more detailed information about the patch weights? Like where they occur?

SOLVED!

If you have AMI patch weight problems, the first thing to do is make sure checkMesh is happy via:

My problem was 'underdetermined cells', This occurs if too many faces of a cell have prescribed boundary conditions. My case was 2D, so front and back faces were of the 'empty' BC. I had some cells that were sandwiched in between a stator wall and the AMI interface (on the same cell), so this left only 2 of 6 faces of these cells to be internal. When I refined the cells in this region so that there were at least 3 internal faces to each cell, the checkMesh error went away and so did the patch weight problems.

You can check is this is the case in ParaView by including sets (one of the checkboxes on the main panel when you first open 'paraFoam'), then selecting from 'Mesh Parts' the 'underdeterminedCells' (and any other error sets of your interest that checkMesh may have written out).