open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello!

I am currently working on a model simulating rising sea levels and the resulting water depth/risk for lives.

The model consists of around 8 800 000 elements with a mesh density of between 2m and 100 m.

I have simulated one scenario with extreme water levels (peak of +2,85 m sea level) which ran fine and I was able to get good results both movies and maps.

In the second scenario I am simulating a less extreme situation (peak of +1,85 m sea level)and I am getting some errors.

I have 3 error warnings for the water depth and 189 warnings for the velocities. The velocity is forming lines in places the water has not reached for that particular time step. The lines "blink" during one time step and then dissappear. I have attached a file showing the temporal maximum. Where the lines and the outline of the model is visible.

Has anyone else experienced this?

Best regards

Jonas

Hello,

What do you call "error warning", especially on depth ?

The lines look like some post-processor problem, but otherwise it could be due to dry zones, or dry points near a shoreline, but I wonder what are those straight lines in your mesh, and why it would happen only one time step. Very strange.

Regards,

JMH

Hello!

I am currently using Blue Kenue for the post processing.

When opening the .res-file in Blue Kenue i get a warning saying "Warning 189 bad values found in object VELOCITY UV", that's what I meant with error warning.

I ran the simulation one more time with a smaller time step and less frequent printout which seemed to fix the problem as i had no warnings or "lines" appearing. Could it be something that has gone wrong when storing the values?

I ran another simulation now for the same project where the DEM is slightly altered. I had the same smaller time step and the same less frequent printout and now the problem has resurfaced and I am seeing these lines in the results again.

There is nothing in my mesh that would indicate straight lines for any of the variables in the spots where the lines appear. In some cases the water is not even close to reaching the spots where the error occurs.

Very strange indeed, seeing as the model has run fine for some scenarios and with small altercations in the DEM or hydrograph the error occurs and then dissappears when running the model again.

Best regards

Jonas

Hello

These strange values seem to be some NaN that could appear when there was no convergence of the solver.
Check the listing in order to detect any sign of divergence of your computation.

with my best regards

Riadh

Hello!

I think I will have to run the model again since I am making some movies for the client of this specific flooding scenario. However, this model requires quite a bit of processing capacity which at the moment is a problem due to many models being run at the same time. Ideally it would be great if i could work in a little "fail-safe" to the model in case the divergence would appear again.

Therefore i am wondering if there is a way to assign a different value to these sporadic NaN-values, more specifically 0. Say the water depth is above 5 m (which it should never be able to reach with the current hydrograph)then the value 0 is automatically assigned rather than x*10^36. Is there anyone who might know how to achieve this?

Best regards

Jonas

Have a look on this [url=http://http://opentelemac.org/index.php/kunena/blue-kenue/10171-cutoff-at-a-certain-height-bulk-selection-of-nodes#23358]post[/url]

kind regards

Riadh

Hi,

The Blue Kenue message: "Warning 189 bad values found in object VELOCITY UV"
is displayed because NaN or Infinite values were found in the result file.

This indicates that your simulation has "blown up" somewhere.
I would not show a client results from such a simulation even if I "doctored"
the data.

You must fix the problem.

Just my 2 centimes worth... Martin