open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello All,

I am wondering if some one has played with programmed tracer degradation in TELEMAC-3D. I have been trying to tackle this problem for last couple of days and now running out of ideas and patience

I am attaching my steering file (ChannelSim.txt) , fortran code (FortranCode.txt) and a plot which compares various simulation that I conducted.

I have turned off the hydrodynamic solver, made the bed friction zero and imposed a constant u velocity of 0.03 m/s in a channel of about 10 KM length. The analytical and numerical solution are compared at x = 2.0 KM from the upstream end in the middle of the channel and at 5 m from the bed. The channel is 10 m deep. Strangely if you do the same in HEC-RAS you can capture the analytical results in a very nice manner.

Any and all help is really appreciated.

Kind regards,
Sumit

Forgot 2 attach the plot

Hello Sumit,

I read your steering and Fortran files and a few options are questionable.
You seem to freeze the advection of velocities and the horizontal diffusion of velocities, but not the vertical diffusion. So, for the velocities, only the vertical diffusion and the numerical diffusion may change the results.
Anyway, you consider the advection and diffusion of your tracer (in particular in the horizontal directions) with an hydrodynamic that roughly does not change a lot, this is strange to me. Moreover, you do not use the non-hydrostatic version of TELEMAC-3D, so that the vertical velocity should not be taken into account. I wonder if you did the same with HECRAS or you also deactivate some parts of the resolution of the Navier-Stokes equation (which ones in that case?).

By the way, you can simplify your steering file and Fortran file. For example, your initial condition is a CONSTANT ELEVATION with the INITIAL ELEVATION = 10. not a particular one. With your Fortran file, you can manage the velocities as you did.
In your Fortran file:
- line 153: IF(LIWBOL%I(IBORD).EQ.KENT) WBORL%R(IBORD) = 0.D0 is useless with the previous line WBORL%R(IBORD) = 0.D0
- lines 948 and 949: V%R((I-1)*NPOIN2+J)=0.00D0 and W%R((I-1)*NPOIN2+J)=0.00D0
should be useless with the previous and following lines:
CALL OS( 'X=0 ' , X=V )
CALL OS( 'X=0 ' , X=W )
and you can use a similar call for U, something like
CALL OS( 'X=C ' , X=U, C=0.03D0).

Regards,

Chi-Tuan

Hello Chi-Tuan,

Thanks a ton for your reply. As far as HEC-RAS is considered I setup the channel simulation with constant upstream discharge (15m3/s) and constant downstream elevation. After conducting the steady state simulation I switched on the tracer degradation and without any further struggle I was able to capture the analytical result.

For TE3D when I posted this thread I was keeping the hydrodynamics frozen but after that I have made several more attempts and have switched on the hydrodynamics with no bed-friction, but my results have not changed from the plot that I presented before.

As of now I am running another simulation with SCHEME FOR ADVECTION OF TRACER = 2, according to the manual this is suppose to be the least dissipative. I have further reduced the time step also.

Do you think using non-hydrostatic solver will improve the tracer results ?
I am also attaching my latest steering file, kindly let me know if you have further ideas, I really appreciate.

Best regards,
Sumit