open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi Laurent,

I did some simulations tests without tracer and i notice that the velocity don't stabilize with a constant set up values, it has ocilation during the time, even with a long time of simulation. And i don't find yet who to fix this maximum iteration error.

So i want to know with it is normal with a small flow rate, the velocity don't get a fixed values. I'm sending a photo with a velocity of 1 point.


Another point is if possible to add a sediment set up for a tracer like coliforms, and how to do it at the steering file.

Regards,
Fernando

Hi Fernando,

Did you first try to set-up a 2D model with exactly the same boundary and physical settings, and is it running fine without oscillations?
If yes, can you tell us more about what you modified from the previous posts? (steering file, liquid boundary file...)
Are you starting your model from zero conditions or from a previous (2D/3D) computation?

Regards,
Laurent

Yes, with the 2D simulation, the velocity doesn't oscillate.
And i use the 2D velocity as initial condition of the 3D simulation.
The only thing that i change it was the time of simulation, i was doing with 2 days of simulation and now is with 12 days, and without the tracer, for what i remember. I'm sending the case that i simulate, so could you see and try to find some thing to fixed.

I'm not sending the SLF file because is to heavy to send it from here, so if you want to see the results maybe you could decrease the number of time steps

The water level don't oscillate so much as the velocity



Thank you again to your help
Best regards
Fernando

Hi Fernando,

Are you sure you sent me the right case ? Tracer and coupling with waqtel are activated, there is also a meteo data file...
Can you explain why you have modified bord3d.f ?

Your liquid boundary file contains Q(1) and SL(1) with 555 code for the first boundary. I'm not sure it's a good solution and it may be an over-constrained system which can cause instabilities with such as small domain. Prescribing only SL(1) with 544 code should be enough (and keeping Q(2) for the other boundary).
What did you use for your 2D model? Please, try first to keep the same boundary settings as your 2D case (cli file and qsl file).

If it's not sufficient, you can try the following:
- reset the coefficient for horizontal diffusion of velocities to its default value (0 diffusion for velocities doesn't help to stabilise the model)
- decrease the value of the free surface gradient compatibility
- increase the implicitation coefficient for depth and velocity.

I strongly advise you to have a look at the numerous test cases which deal with this kind of configuration.

Regards,
Laurent

Hi Laurent,

Yes, it is the same configuration of the last one, i had already added Tracer and coupling with waqtel, at the other simulation that i sent to you. At bord3d.f, i only modify the transformation of the mesh as you suggested before.
My results are quite similar with the real. The only thing now is the oscillation. I had some simulation with your modifications suggested, and the boundary condition works better with 555 at the boundary 1
Thank you for the advice, I've been following the forum a few months, but in my case, the flow velocity is very low, so i just ask to be sure if this oscillation is normal.
The expected velocity is approximately 0.02 m / s, and the model ranges from 0.01 to 0.05

Regards,
Fernando

Hi Fernando,

As I already told you, try to go step by step and I advised you not to include tracer and coupling with waqtel first please.

No, bord3d.f doesn't deal with the transformation of the coordinates, you're confusing with corrxy.f. I think you need no fortran file for your case.

No, the oscillations aren't normal.

Can you send me you're 2D case please?

Regards,
Laurent

I'm sending the files, that i use to analyse the velocity

The 2d case




the bord3d.f subroutine i don't change, as i remember

Fernando,

It's a mess in your 2D/3D steering files, there are a lot of inappropriate or useless options. Try to work more rigorously!

I'm not sure you really tested the advices I gave you... Please, before posting, try it!
By setting the same options (numerical and physical) in both 2D/3D steering files, your case runs fine. Oscillations disappear quit rapidly after the initialization (as in 2D). If you want they vanish more rapidly, try to progressively prescribe the flowrate (with a ramp function for instance).

Here is your 3D steering file modified:



-same time step than in 2D (2s but I think you can increase it a bit)
-same horizontal diffusion than in 2D
-increaase in implicitation coefficients (should be lower or equal than 1)
-no need of tidal flats here
-defaut settings for solvers

Regards,
Laurent

Hi Laurent,

Sorry for the mess, it is my first experience if this type of software, so i'm kind of lost. I tried to follow the manual, read some examples at the forum. But yours advice, and yours modification were very useful for me.

Thank you for your help and patience.
Sorry again for take your time.
Best regards,
Fernando.