open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi,

I defined a control section in my setup and run the model in finite volume mode. Unfortunately the control section output is controlled by the LISTING PRINTOUT PERIOD. That means, due to the variable timestep lengths in FV mode, the output is written only after a fixed number of model iterations but with differing resolution.

Is it possible to get control section output in FV mode at specific pre-defined timesteps?

Regards,
Tobias

I just noticed that with PRINTING CUMULATED FLOWRATES = YES the output flux (not the cumulated volume!) in SECTIONS OUTPUT FILE is given after every model iteration. Only in the listing the cumulated volume is provided. This kind of solves the issue above (although this generates lots of superfluous output). Still, this behaviour is rather unexpected

Hello Tobias,

First, thank you for your feedback.

Concerning your first message, changing the listing printout period in the FV module to give a better control to the user is considered but it is not decided yet if we will do it or not. Maybe a solution can be to synchronize the control section prints with the graphical printout period by replacing ENTET by LEO in a fortran file using telemac2d.F (at line 1690 if you are working in V8P2).

Concerning your second message, this seems to be a bug and I will take some time to look at it.

Regards,
Florent

Thanks, Florent, for consideration.

So far the "bug" serves my needs best (even though the output is a bit large but this is not a major problem)

Hi,

I have to add another observation that I made: the flow through the control section is considerable lower than flow through an open boundary just close by.

See also the figure attached: in the lower left the red line is the control section and depicted in blue the open boundary (LIHBOR, LIUBOR, and LIVBOR set to 4); the rest of the boundary is closed (LI*BOR = 2). For clarification: I use TELEMAC-2D here with the rainfall-runoff module and apply some rainfall to the initially dry catchment. I use FV scheme with default settings.

In my application the mass balance output in the listing is:
However, the flow obtained from the control section output is (in units as above) 0.126E+08 M3 (compare with VOLUME THAT ENTERED THE DOMAIN above). Note that I converted the control section flow for each timestep into M3 by multiplying with the current time step length before adding it up. As the time steps are small the error should be negligible. This value differs a bit depending on how COMPATIBLE COMPUTATION OF FLUXES is set, but overall the discrepancy remains.

What could be the reason for this? Shouldn't the values be (at least) similar?

Regards,
Tobias

Hi,

The difference between the results comes from the fact that the listing is printing the mass fluxes of the mass conservation equation computed at the boundaries and the control sections are printing the variables h*v (where v is the velocity magnitude and h the water depth). This difference can also be seen in the test case rain_runoff.

This velocity underestimation comes from the discretization of the slope source term in case of high gradient slopes and low water depth (see attached file for more details). You can try to improve your results by using OPTION OF THE HYDROSTATIC RECONSTRUCTION = 2 which change the treatment of this source term. Again, the rain_runoff test case illustrates the improvments this option can make.

Regards,
Florent.

Thanks, Florent, for the clarification. Indeed I was not aware of that problem and that the impact could be that high.

However, even with OPTION OF THE HYDROSTATIC RECONSTRUCTION = 2 it only got slightly better. Also with other numerical parameters such as 2nd order solutions. I thought then maybe the resolution is too small but also with much higher resolution (much more mesh points) there is still a gap (still a lack of 4 million out of 18 million m3). Quite difficult to run such rainfall-runoff problems with TELEMAC.