I am trying to set up a morphodynamic river model with multiple
upstream boundaries/tributaries specified as total discharge
boundaries. The total discharge at each of the boundaries is
an unsteady hydrograph. However, at the upstream main-stem boundary
(and a couple tributaries) the discharge drops off to zero after the
discharge increases for some time. I can delay the failure of the
boundary discharge by reducing the time step but it always eventually
fails. The channel gradient at the boundary is fairly steep (~1%). I
can specify a water level time-series boundary condition and it seems
to work fine. But this type of boundary is not desired as the bed
level will not necessarily be constant at the boundary.
Is there a better way to introduce flow at a steep boundary? Is there
some Courant number requirement for a total discharge boundary?
I have had similar problem at total discharge boundaries before in
steep rivers - no quick fix but a few things to try/check could be:
Good luck finding a solution!
Thanks for your suggestions, Richard!
The threshold depth for wetting/drying seemed to work for a while but
just increased the discharge threshold where failure occurs. The
elevations at the cell edges also seem to be interpolated correctly.
Although the end goal is a full morphodynamic model, this test
includes only hydrodynamics.
We *tentatively* solved the issue by adding a "flume" of
several cell lengths to the upstream portion of the boundary with a
milder slope. However, we have since concluded that for stability (and
computational time limitations) it is best to move our upstream
boundary a short distance downstream to exclude this steepest part of