open TELEMAC-MASCARET The mathematically superior suite of solvers

Dear all,
I'm using telemac 2d to research the salt instruction in the SaiGon-Dong Nai river system (south of Viet Nam). I would like to know the reasonable value of diffusion coefficient used in telemac 2d. Of course, I will calibrate the model with the observed data (hydrodynamic and salty). Thank you for any suggestion.
Yours sincerely
Nguyen Thong
HCMUT, VN

Hello Thong,

When you have large element size the diffusion coefficient is not so important because it is often hidden by the numerical diffusion of advection schemes. A value of 1 to 2 m2/s is reasonable but if in doubt you can use the k-epsilon model, however the latter uses velocity gradients which can be underestimated if the mesh size is too large.

I am just worried that you want to model a salt intrusion with 2D, because it is a typically 3D phenomenon, with a salt wedge going up the river on the bottom, with fresh water above, and this can only be reproduced by Telemac-3D. The only phenomenon that can be taken into account in Telemac-2D with salt is the horizontal gradients of density (sometimes important, like in the North sea in Europe).

With best regards,

Jean-Michel Hervouet

Dear Mr. J-M Hervouet;
Thank you very much for your response. I totally agree with you about the model 2D and 3D for this phenomenon. Unfortunately, I would like also to study this problem in the delta Mekong where the domain is too large. So I can't setup the problem with telemac3d in my machine. However, I don't have the observed data of salty at mouth of river if I use model 1D (other model with telemac). I hope that with telemac2D, I can extend my network where the boundary condition of salty will be resolved (salty of sea with value 33/1000 for example). My problem is that the coefficient diffusion of tracer used in the model is too big (1000m2/s) if I would like to calibrate the salty with certain stations where we have the observed data! What do you think about this problem? One more times, thank you so much for your response.
Best regards
Thong NGUYEN
HCMUT, VN

Hello Thong,

The question is where this value of 1000 m2/s comes from? It looks like an artificial value to prevent numerical oscillations. If it is stable it would be much better to have no diffusion at all!

Regards,

JMH

Dear Mr. J-M Hervouet,
The simulation isn't numerical oscillations with the dissipation value of tracer about 2m2/s as your suggestion. But I found that the salty of certain position where there are the observed data (the value max, min in 24h) is very higher compared to the value calculated by the model. The solution will be improved if we used the coefficient of dissipation of tracer higher. Would you like to give me a suggestion to improve la solution (salty)?
Note that the model has been calibrated good on point of view hydrodynamic.
Best regards
Thong NGUYEN
HCMUT, VN

Hello Thong,

Hum, it should be the contrary, you will have sharper gradients of salinity with a smaller diffusion. Anyway as you cannot reproduce a salt wedge in 2D it is highly probable that there is no solution. Imagine that the salt wedge goes up the river on the bottom and the mean flow goes down, you will see only fresh water in 2D whereas the measurements, even depth-averaged, will give you salt water.

Regards,

JMH

Dear M. J-M-Hervouet,
As you say, we can't study the problem of salt wedge with a model 2D. I totally agree with you. It's true too with the model 1D on this problem, right? In the model 2D & 1D, Are we to assume that the salt (as a passive scalar) is constant in vertical direction when we establish the equation of salinity?
What do you think when I say that with a higher diffusion, the salt of sea (as boundary condition, source of salinity as 33/1000) will have a high effect on the salt of calculate domain. Imagine that we don't have fresh water (or a little!) in upstream? It's the reason I note that the salt at certain position higher when we used the higher diffusion for tracer?
Best regards
Thong NGUYEN
HCMIT, VN

Hello Thong,

Yes, seen like this, it is possible that the salt will go further upstream due to diffusion. Make sure that it is only the diffusion coefficient of tracers that you tune.

Regards,

JMH

Hello JMH,

My case could be the same as the case of Thong, modelling for a tracer released from a riverine estuary in 2D mode and need to be calibrated against the measured data at some points. I have played with the value of COEFFICIENT FOR DIFFUSION OF TRACERS in range of 1.0E+0 to 1.0E+1, and the changes was only the gradient of the tracer concentration at measured points, but the mean magnitude of concentration is still higher compared to measured values, as in my opinion the advective flux is much dominant than the diffusion flux.

Is this situation was due to modelling the tracer's mass transport in 2D mode overestimated the concentration at the measured points as the field data only gave reading of concentrations near the surface?

Kind regards,
Amyrhul

Hello,

Sure, you must never forget that in 2D what you are looking at is the depth-averaged concentration. For salt or heat for example it is very important, especially in estuaries.

Regards,

JMH