open TELEMAC-MASCARET The mathematically superior suite of solvers

I have run telemac2d and telemac3d of a coast in UK. When the computation is run on telemac2d the results are fine. When telemac3d is used (5 horizontal layers), I get negative E-Coli and velocity inceases seemingly to be a result of numerical instability. Please help.

The attached files are 1. .cas file for 3d simulation; 2. .cas file for 2d simulation; 3. figure showing numerical instability in velocity in 3d run; 4. negative E Coli concentration in 3d run. None of these problems are observed in telemac2d runs.

Hello,

Hard to read a steering file with so many keywords, in particular if useless or with default values. I would recommend you to clean it a little bit, help would be easier to give.

Perhaps the enclosed steering file gives you the same results?

Which release do you use to run your computations?

I would suggest you to read the TELEMAC-3D user manual and apply some recommendations (you may find other tips in other chapters):
- use a mesh transformation = 3 for the 2 elevations where you want to put your source points,
- use NON-HYDROSTATIC VERSION = YES (default and recommended option),
- read the appendix E of the user manual.

Can you specify what are the axes and units in your velocity figure, I am not sure to understand it. If you can also add the E-coli figure, I cannot see it. If you can also add the 2 sames figures with TELEMAC-2D to compare, I would be interested in. What are the negative values magnitudes?

You can try to see the effect of the FREE SURFACE GRADIENT COMPATIBILITY (rather for free surface elevation, but change velocities also).

If the problem still occurs (and not before changing my first recommendations), can you try to change (they are just tests to circumscribe the possible issue):
- limit the bottom elevation to the minimum free surface elevation value over the whole domain -1 m to delete any dry areas (it may be a problem with tidal flats),
- try VERTICAL TURBULENCE MODEL = HORIZONTAL TURBULENCE MODEL = 1 (constant viscosity in all directions)
+ COEFFICIENT FOR HORIZONTAL DIFFUSION OF VELOCITIES = 0.
+ COEFFICIENT FOR VERTICAL DIFFUSION OF VELOCITIES = 0.
+ COEFFICIENT FOR HORIZONTAL DIFFUSION OF TRACERS = 0.
+ COEFFICIENT FOR VERTICAL DIFFUSION OF TRACERS = 0.
(it may be a problem when solving the diffusion step).

Hope this helps,

Chi-Tuan

Thank you.

The problem turns out to be a tidal flat scheme problem. I put TREATMENT OF NEGATIVE DEPTHS=1 and SCHEME FOR ADVECTION OF VELOCITIES=1 then problem with velocity solved.

Thanks for your answer Arthur.

Can you tell me if you also see this issue if using scheme 13 rather than 14 (they are not so different, but it would be interesting to know for your model).

Another question: are the MAXIMUM NUMBER OF ITERATIONS FOR ADVECTION SCHEMES = 200 reached during your simulation or not? This parameter is used e.g. for advection schemes = 13 or 14 but also when TREATMENT OF NEGATIVE DEPTHS = 2 or 3.
It could be seen in the listing, e.g.:
CVTRVF (BIEF): 200 SUB-ITERATIONS
or
POSITIVE DEPTHS OBTAINED IN 200 ITERATIONS

If it occurs, sometimes or often, this parameter should be increased.

FYI, there will be a warning message if this maximum is reached in next release (v8p2), already available in the trunk.

Best regards,

Chi-Tuan

I went back to advective scheme 14 since scheme 14 guarantee tracer-mass conversation. However, the negative tracer concentration is not solved. Some details during my attempt is written below:

POSITIVE DEPTH can be obtained in 10 ITERATIONS for my case with advective scheme for both velocity and tracer = 14 and SCHEME OPTION FOR ADVECTION OF VELOCITIES =1 and SCHEME OPTION FOR ADVECTION OF TRACERS =1.

I have attempted the SCHEME OPTION FOR ADVECTION OF VELOCITIES =4 and SCHEME OPTION FOR ADVECTION OF TRACERS =4 following the user guide, but the program says that such option does not exist. Any idea about the problem?

I read the TELEMAC user guide but I don't see an option TREATMENT OF NEGATIVE DEPTHS = 3. Can you describe it more? In addition, do you know any documents that describe the actual method used in TREATMENT OF NEGATIVE DEPTHS =2.

Hello Arthur,

Partial answers as I do not have so much time at the moment.

Do you have the same behaviour with scheme 13 rather than 14?

If POSTIVE DEPTH can be obtained in 10 iterations, you should then increase it, you can try 200 or more for example and still check if this limit is reached or not. Do you still have the same behaviour. If not, is it better anyway (less negative values)?

SCHEME OPTION FOR ADVECTION OF VELOCITIES = 4 and SCHEME OPTION FOR ADVECTION OF TRACERS = 4 are only available in combo with SCHEME FOR ADVECTION OF VELOCITIES = 4 or 5 and SCHEME FOR ADVECTION OF TRACERS = 4 or 5 (to give LIPS schemes). It is normal if you get an error message while reading the keywords at the beginning.

Sorry for TREATMENT OF NEGATIVE DEPTHS = 3: it is available in 2D, e.g. for ERIA scheme (sorry for the confusion). I do not remember exactly when the feature = 2 appeared, I have to search longer.

Do your negative concentrations appear in the whole domain? Near dry areas or only on dry areas?

Chi-Tuan

I have tried MAXIMUM NUMBER OF ITERATIONS FOR PROPAGATION = 200 and the program returns

---

PROPAGATION AND DIFFUSION WITH WAVE EQUATION
GRACJG(BIEF): 22 ITERATIONS, RELATIVE PRECISION: 0.8338294E-08
POSSITIVE DEPTHS OBTAINED IN 10 ITERATIONS

What should I do?

My negative concentration occurs in near dry and dry areas, not the whole domain.

I also do not see the point of trying scheme 13 - why can it improve?

Hello Arthur,

MAXIMUM NUMBER OF ITERATIONS FOR PROPAGATION is not the same keyword as MAXIMUM NUMBER OF ITERATIONS FOR ADVECTION SCHEMES. If you have not set this last keyword, the default value is kept i.e. 10 for release v8p1, and it is reached during your computation.
So please try the right keyword to test.

Scheme 13 is obviously different from 14, not so much, but not exactly the same. I just want to check its behaviour with your model.

Chi-Tuan

However, the problem of negative E-Coli is not solved. In fact, the negative E-Coli problem get worse because MOC does not guarantee tracer mass-conservation.