open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi all!

I am currently working on a spillway modeling issue, the model includes the upstream reservoir, the weir, the spillway and the stilling basin.
First, T2D has been implemented with imposed water levels as both US and DS boundaries. The model gave rather good results, especially using the Thompson option.
As the course of the spillway has one bend, we also wanted to implement T3D in non-hydrostatic mode to see if dynamic pressure has an effect on results, and if yes, with which magnitude.

I then set up the 3D model using the same mesh, same conlim file, and same LIQUID BOUNDARIES FILE (imposed water levels, identical in 2D/3D):

# Fichier des frontières liquides transitoires
# Généré le Feb 8, 2012
T SL(1) SL(2)
s m m
0.0 566.5 566.5
600.0 566.5 543.4
18000.0 566.5 543.4

The steering file only refers to the LIQUID BOUNDARIES FILE (2D/3D) as BC.
The initial elevation was set to CONSTANT, of 566.5.

T3D does not maintain these elevations at the boundaries, the water surface decreasing quickly until the bottom elevation is reached.

Does anyone have any advice ? Is such a BC type allowed by T3D ?
I have attached my steering file so you can see if this can be due to my settings.

Thank you a lot in advance for your help!
PL

Hello,

For the boundary conditions, T3D should behave like T2D (but Thompson in 3D will be possible only in version 6.2), but I do not think that prescribing elevations upstream is correct, as you will not control the discharge, then non-hydrostatic equations may give a different discharge, and even no discharge in the end, as your problem is not well posed, this is also a solution. It seems that you were just lucky with 2D. Another possibility is to have a supercritical flow upstream and then you can prescribe both discharge and elevation.

With best regards,

Jean-Michel Hervouet

Thank you Jean-Michel.

It was indeed a bit tricky to have pseudo-steady results with T2D, but the Thompson option helped a lot...
As I don't think that setting a supercretical condition upstream will be of significant help, I will simply try to set the discharge computed by T2D.

A last word to say that on this issue, T2D has been compared with MIKE 21. I will post something later on the forum in the T2D section about it. A first comment is that MIKE 21 seems very performant in reaching stability with supercritical flows (F=3-5).

Best regards,
PL

Hello,

The good question is whether the discharge is correct or not if you do not prescribe it. It is sure that Thompson helps but you may have an infinite number of solutions if your problem is not well posed. In this case there is a risk in comparing two programs that may find solutions at random.

I am also surprised with your small time step.

Regards,

JMH

For this case, I simply set up T2D exactely the same way than a previous MIKE 21 model was set, using these imposed water levels US and DS.
I do agree with you, a better way would have been to impose discharge instead, and then have had a look at the reservoir level computed.

Anyway, the discharge computed by T2D and MIKE 21 is very similar (0,8% difference). Furthermore, a verification using the weir formula, given the discharge computed and the imposed US level, indicates a weir coefficient of 0,37, which matches pretty well with the weir geometry.

Once more, i will post a bit more on that in the T2D section...

For the time step, yes, it is a bit small since it gives maximum Courant Numbers of 0,25 in T2D. I could have increased it a little bit, but did not, due to a lack of time (I couldn't wait for a new T2D calculation to run, and wanted to be sure to have CN < 0,8 in T3D), and due to the fact that the model was not stable in some parts (hydraulic jump).
This brings a question: can the stability be improved by lowering the Courant Numbers even when the results are not stable for Courant Numbers < 0,8-1,0 ?

Thanks for your comments,
Regards
PL

OK. To increase the numerical stability you can set all the implicitation coefficients to 1 (I noticed they are at 0.6). As you look for steady state this should not change the results.

It is also possible to decrease "compatibility of free surface gradient" down to 0 (but with possible differences).

Please note that shallow water equations do not give the correct weir coefficients, they have a unique coefficient for any shape of weir, and it is underestimated. This fact is often overlooked, so again you may be lucky as your weir coefficient is rather low.

With best regards,

JMH

Thank you for these comments.

I will try to launch a simulation with those numerical settings.
Regarding the weir coefficient approach in shallow water equations, do you have any litterature tips ? Is this point described in your book that I will order soon ?

Thank you,
PL

Hello,

This weir issue is not in my book, it is in lectures that will be put on the website soon. Actually the conservation of head, if you assume that the velocity upstream is 0, which is the case if it is very deep upstream, combined with the fact that the flow above the weir is critical (hence a relation between velocity and depth), gives you directly the discharge above the weir, regardless of its shape, and it corresponds to a coefficient equal to 0.3849, you understand now that you are lucky with your weir geometry. Literature gives coefficients between 0.4 and 0.5 for real weirs.

Regards,

JMH

Hello,

The T2D and T3D calculations performed with the settings you suggested on implicitation coefficients and on "compatibility of free surface gradient" gave good results (without significant differences due to comp. free surface gradient 0,9 => 0,0).

Another question: how does telemac deals with the Smagorinski coefficient in this turbulence model? In MIKE 21 it has to be user-defined.

Thanks in advance,
PL

Hello,

Yes, there is a parameter that can be tuned in the Smagorinski model, you can change it in subroutines SMAGOR in 2D and SMAGO and SMAGO3D in 3D.

Regards,

JMH