open TELEMAC-MASCARET The mathematically superior suite of solvers

When running my first TELEMAC simulation I've faced some issues related with boundary conditions. Since I have some part of the ocean in my model, I set two open boundaries with free depth and velocity. I also set one open boundary with free velocity and prescribed depth (which is between the other two boundaries)...

The set up is something like this (see picture):

North: free open boundary
West:Land/ East: tide
South: free open boundary



However, after a few time steps the simulation crashed and also I got a warning message saying that the problem was ill posed due to supercritical flow at some nodes of the boundary. I checked the very few results that were printed and water depths vary from -6E7 to 6E7, even for the first time step printed; it was the same for all the variables (U,V,Q). I realized that the problem were the free open boundaries, so I decided to close them just to see if the simulation was running. Yes, it was running but I need the north and south boundaries to be open, because otherwise I think it'll behave like a pool.

I found that one option is to use Thompson BC, so I did it. But I get very high velocities at both free open boundaries and also the direction of the velocity does not make sense to me. I want to be able to represent that the water reaching those boundaries, simply leaves the domain without changing the patter inside the domain.

I know that the boundary can also be closed with slip or non slip condition, but this would mean that there is no normal flow across the boundary.

What is the best way to represent this case, since the Thompson BC doesn't seem to work properly for my case?

Thanks

Hello,

I think you should try to prescribe tidal conditions on your north and south boundaries too, by defining 5 4 4 conditions on all your maritime boundary nodes, until land points.
Then you can play with Thompson option to see the effects on your results.

Concerning your post about negative depths, it should not happen, depending on your settings on tidal flats, but at least not as much as few millimeters.
Be careful to adapt your time step so as to have a Courant Number around 1.
And try TYPE OF ADVECTION = 14;5 with associated tidal flat options described in the user manual.

I hope it helps.

Regards,
Laurent

Thank you, I've got the simulation to be completed by using these options:

/

---

/ EQUATIONS, BOUNDARY CONDITIONS
/

---

OPTION FOR LIQUID BOUNDARIES =1;2;2;2

PRESCRIBED ELEVATIONS =1.0;0.0;0.0;0.0

PRESCRIBED FLOWRATES =0.40;0.0;0.0;0.0
/

---

/ EQUATIONS, ADVECTION
/

---

SCHEME FOR ADVECTION OF VELOCITIES =14

/

---

/ NUMERICAL PARAMETERS
/

---

TREATMENT OF NEGATIVE DEPTHS =2

CONTINUITY CORRECTION =true

TIME STEP =2

SUPG OPTION =2;0;2;2

/

---

/ NUMERICAL PARAMETERS, VELOCITY-CELERITY-HIGHT
/

---

MASS-LUMPING ON H =1


However, I get some high velocities at the free open boundaries (much better than using the default velocity advection option, but still high). Inside the lagoon, I get -what seems to be- reasonable results, but I'm still concerned about the boundaries because I think this can influence the velocity at the mouth of the lagoon. Can you please take a look, when you have some time, at my results and advise me if this looks correct at the boundaries?

I believe the velocities at the boundaries should be smooth, however the issue seems to be only at the connection between the closed (land) and the free open boundary. What option might improve this?

I see a bit difficult to prescribe tide at the other two boundaries because of its variability in time and space.

Ps. You can find my results in this link:
www.dropbox.com/s/fi1rbk5sf0ajf5d/Results_Lagoon.slf?dl=0

Thanks

Thank you, I've got the simulation to be completed by using these options:

/

---

/ EQUATIONS, BOUNDARY CONDITIONS
/

---

OPTION FOR LIQUID BOUNDARIES =1;2;2;2

PRESCRIBED ELEVATIONS =1.0;0.0;0.0;0.0

PRESCRIBED FLOWRATES =0.40;0.0;0.0;0.0
/

---

/ EQUATIONS, ADVECTION
/

---

SCHEME FOR ADVECTION OF VELOCITIES =14

/

---

/ NUMERICAL PARAMETERS
/

---

TREATMENT OF NEGATIVE DEPTHS =2

CONTINUITY CORRECTION =true

TIME STEP =2

SUPG OPTION =2;0;2;2

/

---

/ NUMERICAL PARAMETERS, VELOCITY-CELERITY-HIGHT
/

---

MASS-LUMPING ON H =1


However, I get some high velocities at the free open boundaries (much better than using the default velocity advection option, but still high). Inside the lagoon, I get -what seems to be- reasonable results, but I'm still concerned about the boundaries because I think this can influence the velocity at the mouth of the lagoon. Can you please take a look, when you have some time, at my results and advise me if this looks correct at the boundaries?

I believe the velocities at the boundaries should be smooth, however the issue seems to be only at the connection between the closed (land) and the free open boundary. What option might improve this?

I see a bit difficult to prescribe tide at the other two boundaries because of its variability in time.

Ps. You can find my results in the attachments.


Thanks

Hi,

Indeed, you have side effects on your north/south boundaries with high current values entering your domain and disturbing the circulation, due to high free surface slope between land and sea points.
As I told you, you can try to prescribe tidal elevations also on these boundaries (with or without Thompson option). The variability in time should not be a problem as you already prescribe a time-varying signal on the east boundary, you can prescribe the same or with a decreasing free surface gradient towards the coast.

I saw you also have a river with prescribed flowrate. Maybe you can refine the mesh around this boundary.

Regards,
Laurent

I've tried to modify the BC fortran file, but I've not achieved it yet. Is the Bord.f the only file used to prescribe the depth (when this is variable in time)?
Because I've checked the bord file and I couldn't see something like a time interpolation there, but maybe I just couldn't identify it.

If I change the bord.f and what I prescribe there is different from the liquid boundary file; the fortran file replaces all the other options, right?


Thanks

I've tried all the possibilities now and none of them seem to work

First, I managed to set an imposed depth at the boundaries I used to have free. I also have an imposed depth at the boundary where the tide comes from. I tested the Thompson boundary condition and also the free option; after a few time steps the discharge at certain control section goes to Nan (it's faster when the Thompson BC is not used, but at any case seem to work).

Second, when I simply use the Thompson BC and no depth is imposed at the north and south boundaries, the simulation is completed but some non-sense values come up and there is kind of circulation affecting the velocity field!

When I close the boundaries with no friction, the simulation is not completed either!

What else can I do?


Thanks

Hello,

I'm not sure to really understand how you're prescribing tidal signal on your east and north/south boundaries, it shouldn't differ. May I see your steering file and potentially your fortran file, liquid boundary file?

Did you check your elevation on north/south boundaries?
Can you locate where the NaN values occur?

Regards,
Laurent

Hi,

The bathymetry goes from deeper to zero, from east to west. I mean, north and south boundaries vary gradually from -15 m to zero, in a distance of about 1 km. Because the tide comes from the east, it should have a small gradient when imposed in the north and south boundaries. Since I didn't have enough data, I simply tried to prescribe a sinus function, with similar amplitude to the tide I have in the east boundary. So I modified the bord.f file.

Although I was not able to run the simulation with the modified bord, because it starts exceeding the maximum number of iterations (after 900 TS). I couldn't identify where they come up, but I've noticed that the problem is (mainly) the interception between the flow and the solid boundary.

I'm attaching two cases:

1) Lagoon.rar: with north and south boundaries set as free and Thompson BC. This works but some wrong circulation patterns come up.
www.dropbox.com/s/cmj9477rfgsklro/Lagoon.rar?dl=0

2) Lagoonprescribed.rar with north, south and east boundaries set as prescribed depth with Thompson BC as well. This doesn't work.
www.dropbox.com/s/qd3hna8q4fpa2wf/Lagoon_prescribed.rar?dl=0

Thanks for taking some time to reply to my post.


Best,

JR

Hi,

The bathymetry goes from deeper to zero, from east to west. I mean, north and south boundaries vary gradually from -15 m to zero, in a distance of about 1 km. Because the tide comes from the east, it should have a small gradient when imposed in the north and south boundaries. Since I didn't have enough data, I simply tried to prescribe a sinus function, with similar amplitude to the tide I have in the east boundary. So I modified the bord.f file.

Although I was not able to run the simulation with the modified bord, because it starts exceeding the maximum number of iterations (after 900 TS). I couldn't identify where they come up, but I've noticed that the problem is (mainly) the interception between the flow and the solid boundary.

I'm attaching two cases:

1) Lagoon.rar: with north and south boundaries set as free and Thompson BC. This works but some wrong circulation patterns come up.
2) Lagoonprescribed.rar with north, south and east boundaries set as prescribed depth with Thompson BC as well. This doesn't work.


Thanks for taking some time to reply to my post.


Best,

JR