open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

I seem to be having some difficulty in writing results files. To give you some background, the project I'm running was originally a 2D one of a estuary. It included a tracer and surface level input at the estuary end.

Validation file run fine, however I think I'm having some issues with my steering file. When imputing my 3d and 2d results file the system does not seem to recognise them.

le mot <3D Results file> n'existe pas dans le dictionnaire

my steering file is attached, any help would be greatly appreciated.

Hello,

It is just that it is :

3D RESULT FILE

and

2D RESULT FILE

You can also remove the line giving the steering file, it is not useful and can drive you mad if you copy the file to another name.

With best regards,

Jean-Michel Hervouet

Thank you I knew it would be something silly like that.

Just another question, My computation has run, it see's the tracer and is computing it, but when looking at the two files in Fudaa the tracer doesn't seem to be there. The only thing which seems to be a problem on the run is an error on the mass during the time steps.

Any ideas what could be my problem?

Thanks Emma

Hello,

Make sure you ask for the tracers in the result files :

VARIABLES FOR 3D GRAPHIC PRINTOUTS : should contain TA1 or TA* if you have several

and

VARIABLES FOR 2D GRAPHIC PRINTOUTS : same thing

JMH

Thanks again,

I have now managed to record the tracers. However, I am having trouble with my surface level at the end of the estuary. I would like the surface level (tides) to be the influencing factor for the velocities. I have set the scheme for advection of velocities to 0. This seems to have stopped the problem of the surface level fluctuation. I am still getting a large amount of error on the mass during the time steps though.

Could there be any particular reason for this?

I have attached my latest log.

Thanks again

Emma

Ok, so I've managed to sort the error on the mass to a better level (mass error is still over 100) by changing the solvers for the tracers to the generalised minimum RESidual method option 7. With this I have changed the vertical advection scheme to option 2 (explicit scheme + SUPG)

Unfortunately with these improved mass changes the advection step for the diffusion of the tracers is causing me trouble. The MURD3D option (whether it be scheme 4 or 5) is stating the maximum iteration no. reached is 100.

I have tried changing the maximum iterations with no change. Can anyone tell me if I'm going down the right lines for improving the mass balance?

Also is there any way of changing the maximum iterations for MURD3D, which I assume is to do with the diffusion?

Thanks

Emma

Hello,

Schemes 4 or 5, which are advection schemes, will not work with tidal flats. If it is your case use scheme 14. It should work and generally better than SUPG. If the maximum number of iterations is reached it means that you have a Courant number which is too large, try to reduce the time step or look to see if you have a very large velocity somewhere.

It is not normal to have such a mass error, as we can now have the accuracy of the machine.

If you do not manage your way send your steering file so that we can have a look at it.

With best regards,

Jean-Michel Hervouet

Thank you Jean,

Unfortunately I am still struggling. I have managed to reduce the error on the water mass balance to an acceptable level. However, the tracer is still causing trouble with the error going up into the thousands.


I have attached my steering file. Any idea where I may be going wrong?

Kind regards

Emma

Hello,

try the following:

remove the keyword "STEERING FILE" (it is useless)
remove the keyword "NUMBER OF SUB ITERATIONS FOR NON LINEARITIES=2" (1 is OK, we'll see after)

put :

SOLVER FOR DIFFUSION OF TRACERS : 1 (instead of 7 which is GMRES, GMRES is meant for non symmetric systems, it works also for symmetric systems, but may crah if given a mere diagonal, which is the case when diffusion is 0.)

NUMBER OF HORIZONTAL LEVELS : at least 2 or 3, 1 is impossible.

test the non-hydrostatic option :

NON-HYDROSTATIC VERSION : YES

and clearly state that you have tidal flats :

TIDAL FLATS : YES
OPTION FOR THE TREATMENT OF TIDAL FLATS : 1

I would also use : SUPG OPTION : 0;0;0

and choose instead :

COMPATIBILITY OF FREE SURFACE GRADIENT : 0.9 or even lower, down to 0, if there are wiggles of the free surface.

Then you should decide what you do for diffusion (horizontal and vertical : turbulence model, etc.)

I hope this helps,

Jean-Michel

Jean-Michel,

I have applied your suggestions with varying success. I occured to me that my mesh may need refining. This has helped, however I am still getting errors on the tracer mass of over 100.

I have tried varying turbulence options, I have gone with the mixing length option along the vertical. With the PRANDTL mixing length model. The horizontal turbulence I have left as 1.

I have attached my log and correlating steering file.

Any ideas on how I could further improve the mass balance?

Thanks again

Emma