open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi all,

I'm hitting a most perplexing wall in attempting to continue a GAIA computation from a previous file. The error message is consistently this and it occurs at the beginning of the computation, which for reasons I will explain below makes no sense to me:
In essence, for my project, I am trying to combine sedimentological data (stratification etc) from a previous computation (file A) with new bathymetry data (file . To do this, I adapted a Python script that will take values of BOTTOM from file B and all the other variables from file A, outputting a file C containing both. Thus, I set GAIA to continue computation from file C. Once this is initialized, a user_bed_init subroutine is run to define certain non-erodible zones.

Additionally, the hydrodynamic (T2D) computation is also continued from a previous file (file D) to avoid cold-start issues. In this previous simulation as well as the current one, depth is defined at both boundaries and a velocity profile is defined upstream. For this reason, flow criticality should not be problematic and anyway, in file D, every node on the boundary has Fr < 0.7 (high, but not supercritical). The same boundary conditions are used between files C and D and in all the following simulations.

Here is what I have tried to narrow down and mitigate the cold crashing issue:

1. Run the simulation in hydrodynamic mode only (T2D) - successful
2. Run the simulation coupled T2D & GAIA, without a previous computation file - successful
3. Run the simulation coupled T2D & GAIA, with the previous bathymetry (file A) as previous sediment computation file - successful
4. Reinitialize (set to 0) variables in file A pertaining to bedload rates for each class and cumulative bed evolution - unsuccessful
5. Copy bed shear stress from file B to file A - unsuccessful
6. Verify that bathymetry is identical between file C and file D
7. Reduce computation timestep - unsuccessful
8. Not running user_bed_init - unsuccessful

Evidently, the problem lies with the introduction of file C, as every other simulation "leading up" to this runs successfully. I cannot see for the life of me what could be going wrong, as file C is not corrupted in any visible manner and in principle, it should be a drop-in replacement. I have attached the .cas file in case anyone sees something strange, but these settings all work in other simulations with no issues.

Thanks,
André Renault

I just noticed that Kunena maps B and a parenthesis to the emoji. So "(file " should read (file B )". Apologies if this is confusing.

Hi again,

My issue is solved, as far as I can tell. I reran the previous sedimentological computation file, changing some of the settings, such as reducing the number of layers and adding more variables to the graphic printout. Crucially, I made sure to add the RIGID BED variable to the printout. These simulations were successful and did not give the boundary problem above.

However, it seemed that even if I correctly changed the BOTTOM variable for the new bathymetry, it would somehow reinitialize to the old bathymetry. My guess is that when GAIA reads the previous computation file, the RIGID BED variable overrides the BOTTOM variable, such that BOTTOM is recalculated to be RIGID BED + Σ (LAYER i THICKNESS).

My solution was therefore to update the RIGID BED variable in the Python script by performing the reverse of the above calculation: taking the new BOTTOM and subtracting all the LAYER THICKNESS variables. This, after nearly a week of bashing my head against the wall, made everything work.

I have not done further testing on the pertinence of copying over BED SHEAR STRESS and resetting the solid flux variables. When I do, I will provide the updated Python script to the forum, probably in the thread linked above as well as here.

Hello again André,

FYI, I pushed a new feature for GAIA a few days ago on the main: RESTART MODE as it was done for TELEMAC-2D last year (for release v8p4) and for TELEMAC-3D for a few years.

This feature enables 2 main things:
- automatically saving of every necessary variable needed to continue a GAIA computation coupled with TELEMAC-2D or TELEMAC-3D in a specific file (RESTART FILE) during the 1st computation,
- reading of a previous computation file with every variable necessary for a good continuation without loss of accuracy in 2D (perfect for most of the examples and nearly perfect in 3D). If the variables are not in the previous computation file (e.g. not a RESTART FILE), some variables can be initialised by 0 or other ways (but there will be loss of accuracy).

Anyway, I think it is a good basis for you to start from and you can try your additional modifications after.

For more descriptions, you can read:
- issue #899: gitlab.pam-retd.fr/otm/telemac-mascaret/-/issues/899
- MR !121: gitlab.pam-retd.fr/otm/telemac-mascaret/-/merge_requests/121

Hope this helps,

Chi-Tuan