M21 FM Boundaries: Constant water level

Hi there!

When I'm setting up a boundary in my Mike 21 FM model with a constant water level, is this an open boundary which works in both directions (subtracting water from the domain when the water level within is higher and adding water when it is lower in the domain)?
Because in my test model there is no flow entering the domain when I set up a boundary with constant water level, but it sucks out all the flow coming towards it. The boundary is an internal (enclosed in the domain) boundary. See attached picture.

Looking forward to a response :)

Best regards, Sam

Hi Sam,

I would assume, that the boundary works in both directions: inflow AND outflow.

You can stable your boundary condition, if you apply initial conditions identical to the intial state of the boundary condition. MIKE will simulate faster and more stable, if there are already wetted cells in the model.
Alternatively, you can try to apply a soft start.

Hope that helps
Christian.

Hey Christian

that's what DHI told me too. So I created initial conditions (water level = 0.3 around open boundary surroundings and 0 elsewhere, 0 for both velocity components) and fed them into my model.
When I then started the simulation, it kept initializing for hours and didn't start with the calculations at all. DHI said the initial conditions file should be fine (which I don't believe since it is still not working). Running the same simulation without initial conditions yields a normal run with the known instabilities near the boundaries.
I'm starting to lose patience when even DHI cannot give an explanation to this  :-\

Hi Sam,

I could imagine two more issues, which could raise your difficulties:

1. the mesh resolution around the boundary could be to coarse.
2. Some of your boundary nodes are above your assigned boundary water level.

Aah - yes - please keep in mind, that you have to assign absolute surface water levels and not water depths.

Bests
Christian

Hi Christian

to your answers:
1. I created a cfl output file and checked the critical regions. Some mesh refinement somehow solved some of the problems.
2. This would be very difficult to solve since the mesh/boundary nodes level is based on interpolation of the DEM. Theoretically the addition of 0.15m height to the DEM at and 1m around the boundary should avoid this in my opinion..

Perfect, with the absolute surface water levels the simulation starts without hesitation. I don't know why the support told me that my file is fine  :-X
The only issue is now that water is added all over the domain due to the initial conditions (dfs2 grid, filled with DEM values plus boundary raise). Is it possible to use the mesh to interpolate from the dfs2 grid to have the exact same elevation on the mesh where there is no water to be added?

Best, Sam

I have a similar opinion to you.