Problem: negative volumes appearing when trying to adapt or smooth an originally sound grid.

A ship is modelled using 'utility tmerge' to combine 3 parts of the domain:

1. A rectangular block containing the ship geometry as a gap (subtracted volumes). This has been meshed and appears to be sound.

2. An 'inner cylinder' less the rectangular shape in part 1 above. This has been meshed and appears to be sound.

3. An 'outer rectangular block' less the inner cylinder in part 2. This has been meshed and appears to be sound.

All three grids were combined using 'utility tmerge -3d', ensuring correct positioning and alignment. This method allows for rotation of part 2 to model wind-over-ship at different angles, and has been successfully implemented on a different ship previously.

The four 'interface' boundaries defined in Gambit were used to create two grid interfaces in Fluent. The complete combined grid was checked in Fluent and reported no errors and no negative volumes.

A calculation ran without errors or divergence until convergence.

However, after running, any attempt to adapt the grid or smooth it immediately results in 'non-positive' volumes which appear to correspond to the size of the grid interface boundaries, i.e. quite large. Progression of the solution is impossible, any attempt to continue iterations after the adaption fails at the first iteration, with the following error message:

Error: divergence detected in AMG solver: x-momentum Error Object: ()

The whole process was started again (for a different wind angle) in Gambit but the eventual result was identical. Initial grid was good but any adaption led to negative volumes.

As mentioned, the same process was used in the past successfully, so the reason for these errors is unclear. Any help or suggestions would be gratefully received. Thanks.

I think you should consider deleting the grid interfaces before the adaption process is started, than make the adaption (which should still work considering the fact that deleting an interface does not change any solution values stored in the cells adiacent to the interface boundary). After obtaining the desired result, recreate the interface with the new grid boundaries.

I bet this would work.

Best whishes, Razvan