Hi Adam, 

The most probable cause of such a problem is an incorrect enclosure file. The fact that you are using domain decomposition further complicates the matter. In order to solve the problem, I would suggest to first open the enclosure files of the subdomains using a text editor and check that it is all logical. Then, try running each subdomain individually. Also, please have in mind the limitations of domain decomposition. In particular, the domain interfaces must have no overlap and no holes. Check that the grid orientation is the same in all subdomains (i.e., the M and N directions do not change from domain to domain). The interfaces must be straight. Finally, the coupling must be aligned (i.e., a cell edge perpendicular to the domains interface must be continued by the cell edge of the next domain). 

V

Thank you for that. Maybe it is the limitation of the DD boundary because i derefined grids of flood plains to reduce the computation time as my main interest is the river bed. however, I kept the interface straight and aligned coupling.  Maybe I should have the same grid size on the whole cross-section. 

https://drive.google.com/file/d/1SaCqo0SvHR3f7l8HgMEZ1J24z9t0BmBD/view?usp=sharing

Hi Adam, 

I do not see anything strange in the figure you sent. However, I do not see the domain boundaries. Remember that you cannot have an open boundary at a DD boundary. If you are using different grids for main channel and floodplain, it seems that you have an open boundary at a DD boundary. 

If you have derefined a grid, check that the enclosure is still correct. Also, check that flag 'MNKMax' in the mdu-file is consistent with the new grid. 

Last but not least, easy grid refinement and variability is the key strength of Delft3D Flexible Mesh. You could use the same grid you have now without being restricted by the location of open boundaries etcetera. Maybe you want to try it. 

Regards,

 V