TELEMAC-MASCARET Forum: PARTEL for global meshes

Welcome, Guest

TOPIC: PARTEL for global meshes - critical implementation

PARTEL for global meshes - critical implementation 10 months 1 week ago #43950

tomsail
OFFLINE
Junior Boarder
Posts: 43
Thank you received: 17

Hi all,

I couldn't find the PARTEL subforum, I hope it is the right place to post this.

I have created a branch for global models:
gitlab.pam-retd.fr/otm/telemac-mascaret/...mmits/support-global

I thought to post here first to get feedback before creating a Merge Request on the git.

I have implemented two commits:

the first one non intrusive, for pretel routines: to generate atm forcings faster and the CLI file for global meshes
My concern is about the second one, which is critical i.e. PARTEL does not work anymore for normal meshes

problem is PARTEL is not something that can be changed on demand - with a keyword in the CAS file - like the other routines in the modules or in BIEF.

So that means that I would have to compile one telemac version for global models and one for normal models.. not ideal

If possible I'd like that have one version that deals with all models.

Any PARTEL experts out here to help me review this ?

I will provide steps to reproduce the mesh and model in the following thread.

The administrator has disabled public write access.

PARTEL for global meshes - critical implementation 10 months 1 week ago #43951

c.coulet OFFLINE Moderator Posts: 3722 Thank you received: 1031	Hi Thomas The problem about partel has been reported 1 month ago... (see incident #957 on git) This will be solved at least partially. Regards
	Christophe The administrator has disabled public write access.

PARTEL for global meshes - critical implementation 10 months 1 week ago #43952

tomsail
OFFLINE
Junior Boarder
Posts: 43
Thank you received: 17

Great thanks Christophe for the answer, this is exactly it.

So the problem actually would not lie in the global / regional configuration of the meshes..

I have been generating my meshes with python (Selafin Class) for some time and encountered this problem only for global meshes - at least I think so.

I'll have a go at Chi-Tuan's suggestion, test it on both on my meshes and the ones from the examples and propose a merge request.

The administrator has disabled public write access.

The following user(s) said Thank You: c.coulet

PARTEL for global meshes - critical implementation 10 months 1 week ago #43953

tomsail
OFFLINE
Junior Boarder
Posts: 43
Thank you received: 17

it works.

the changes in mod_write_solutions.F work on the examples and on global meshes.
the changes in partel.F are necessary for global meshes and have no impact on the examples from telemac.

You can also try for the case you mentioned in issue #957

follow up the MR here : gitlab.pam-retd.fr/otm/telemac-mascaret/-/merge_requests/176

PS: To replicate my global mesh model:

git clone https://github.com/tomsail/pyPoseidon.git
cd pyPoseidon
mamba env create -n pos-dev -f environments/base-p3.11.yml
rm poetry.lock
poetry install
cd Tutorials

to get the meteo and the dem, follow instructions at github.com/tomsail/pyPoseidon/tree/master/Tutorial

stay in ./Tutorial and run

python mesh_model_test_simple.py

(attached)

File Attachment:

File Name: mesh_model_test_simple.py
File Size: 4 KB

or .. easier: here is the telemac files generated after running the script

File Attachment:

File Name: telemac2d.cas
File Size: 7 KB

File Attachment:

File Name: geo.cli
File Size: 211 KB

EDIT: can't seem able to upload the geo.slf (3.4MB)

The administrator has disabled public write access.

PARTEL for global meshes - critical implementation 10 months 1 week ago #43967

tomsail
OFFLINE
Junior Boarder
Posts: 43
Thank you received: 17

More info (to close on this topic)

My meshes were generated with

NPLAN=1

because I was using the old Selafin class (in data_manip.formats.selafin), which was wrong.

After having a look in the latest notebooks, now I generate my meshes with:

res = TelemacFile(outpath, access="w")
res.add_header("pyposeidon generated mesh", date=[1986, 2, 21, 0, 0, 0])
res.add_mesh(X, Y, TRI)
# add BOTTOM variable
res.add_variable("BOTTOM", "M")
res._values[0, 0, :] = Z
res.write()
res.close()

Source: telemac.py on pyposeidon's github

The implementation I proposed for global meshes now only involves the python scripts.

More info about these changes in Merge Request MR #176

The administrator has disabled public write access.