numerical instability in domain decomposition with inclined bathymetrynumerical instability in domain decomposition with inclined bathymetryhttps://oss.deltares.nl/c/message_boards/find_thread?p_l_id=1806765&threadId=19396402024-03-29T06:16:49Z2024-03-29T06:16:49Znumerical instability in domain decomposition with inclined bathymetryMarkus Grosshttps://oss.deltares.nl/c/message_boards/find_message?p_l_id=1806765&messageId=19396392018-08-31T22:40:47Z2018-08-31T22:36:22Z<p>Hi,</p>
<p> </p>
<p>We are trying to run a domain decomposition model but run into
several difficulties. Attached one of our problems as Smallest
Possible Case.</p>
<p> </p>
<p>It is a cut out rectangle and inside it is a higher resolution grid
with a 1:3 ratio.</p>
<p> </p>
<p>It is currently setup as spherical, but I converted it into Cartesian
at some point and it did not change anything. We also tried the recent
trunk version.</p>
<p> </p>
<p>In this case there are no open boundaries. So everything should stay
within the numerical noise of the solver. The bathymetry is inclined,
with a constant gradient.</p>
<p> </p>
<p>In a relatively short period it spins up a significant instability,
originating from the domain decomposition boundary, which eventually
leads to model failure. For me:</p>
<p> </p>
<p> Time to finish 2m 14s, 33.5% completed, time steps left
766<br /> Flow exited abnormally</p>
<p> </p>
<p>We must be missing something obvious, but with now two weeks of
searching I cannot find it.</p>
<p> </p>
<p>Does anybody have any advice? Is there something really silly in the setup?</p>
<p> </p>
<p>I feel that without resolving this, going to more complex
bathymetries and boundary conditions would not make any sense.</p>
<p> </p>
<p>Any help/advise would be greatly appreciated!!</p>
<p> </p>
<p>Best wishes,</p>
<p>Markus</p>
<p> </p>Markus Gross2018-08-31T22:36:22Z