Your Feedback on Code Aster !

Laetitia

Hi Code Aster users,

We are launching a survey to better understand your use of the code, the challenges you are facing and how we can support and train you to better reach your goals.

By giving your feedback you can get in touch with us for further discussion and apply to get in our beta support and training test. You can also stay anonymous, it’s entirely up to you.

If you feel like contributing to our survey, give it a go !

[https://forms.office.com/e/JG3hCwv8ef]

EDF Open-Source Code Valorization Team

apalazzi

Laetitia Can you publish the results of the survey? This will also provide some insight for any community effort to document or contribute in any way to the CA open source ecosystem.

Andrea

lIlIl

As cacciatorino mentionned, Code_Aster desperately needs a Wiki to ease the learning curve.

benico

My native language is Afrikaans - so I VERY well I understand the "thing" between French and English ... but continuously bumping in to French-ONLY documentation is quite frustrating for me.

Is there and english version of this:
https://codeaster.gitlab.io/doc/docaster/manuals/man_u/other_pages/home/index.html

It was a HUGE learning curve trying to make sense out of the 8-chararter-ony FORTRAN keywords combined with words like POUT, etc. So one day perhaps the code can changed to have "better" (more intuitive) keywords / function names ?

Also - I find the number of Finite elements, conditions, etc. quite bewildering ... for example: [u4 - 245 documents] ... I guess I will not use 90% of them. So I would like a "dumbed down" set of drop-down-options in AsterStudy (not a list of 60 - since I don't have the time to go and learn what each one does and think about if I should use it or not.)

And then - look at the user-friendliness of the page below for example ... I just want to know what INFO_MODE does .... quickly and simply ... but MOST of the documentation seems like a PhD Thesis - and then I still don't know what it does:
https://codeaster.gitlab.io/doc/docaster/manuals/man_u/u4/u4.52.01/Op_randes.html#principes

Problem is ... I am an engineer ... NOT a PhD ... so I use FEA once or twice a year - and I need something that is understandable and workable, without taking weeks to re-learn every time.

Sound_Spinning

benico
If you open that help docs link in a modern browser, it'll offer to translate into English on the fly, and works brilliantly, at least in Chrome & Edge latest versions.

As for the content, I see not much difference with commercial FEA help docs in terms of clarity. FEA speaks a certain language, not necessary theory in-depth, that needs to be learnt, as it is the same in all (good) FEA solvers.
In fact, the notes half-hidden in code_aster help docs are brilliant, as they explain practical stuff. All FEA solvers take a while to get used to the jargon and commands.
Keep trying!

k_zurawski

benico If you are using FEA from time to time, try these books in following order:
https://engineering.moonish.biz/fea-using-open-source-software-4th-windows-edition/
https://engineering.moonish.biz/product/intermediate-fea-using-open-source-software-2nd-windows-edition/
https://www.jean-pierre-aubry.ouvaton.org/files/beginning_with_code_aster-jp_aubry-20190129.pdf

apalazzi

Hi,

From my point of view the biggest problems of using Code Aster are the following:

available information is not up to date and coherent even on official sites: for example, at https://code-aster.org/spip.php?rubrique21 the latest stable version is 14.6, at gitlab https://gitlab.com/codeaster/src the readme says main, V15 and V14, and https://gitlab.com/codeaster-opensource-documentation/opensource-installation-development/-/blob/main/devel/changelog.md there is v15, v16 and v17.... like this even understanding which version is the latest stable is complicated
if the Singluarity image is the preferred/supported way of using Code Aster there should be some easy to find documentation explaining how to use it, and how to overcome some common issues, like config.txt not found
a more or less official tutorial would also be very helpful to set people up and running with limited effort, at least for the most common cases, and no need to search the forum on how to do this and that

Even though it's open source and I thank you for making CA available to everyone, I have to say that using it is still a frustrating experience with a lot little obstacles to overcome before being able to obtain some results.

Bye
Andrea

Nidish

It would be great if a build-able version could be released along with the singularity containers (just like the pre-2021 version).
Specifically, releasing the AsterStudy as a standalone plugin for Salome would be very convenient. I would then be able to use AsterStudy with my existing installation of Salome that has other plugins too.

Overwan

I think a proper tutorial with videos for all the modules of Salome-Meca would be useful.

basile-simvia

Overwan

Hello @Overwan
You can find some tutorials on Salome-Meca on the Simvia YouTube channel:
https://www.youtube.com/@Simvia-tech

There are also tutorials about Salome on the official SALOME Platform YouTube channel:
https://www.youtube.com/@SALOMEPlatformOfficial

And also about Salome there is a dedicated channel https://forum.code-aster.org/public/d/27425-your-feedback-on-salome/5

Basile

Overwan

basile-simvia Thank you

k_zurawski

Look for Cyprien Rusu on YouTube, he did a lot of videos regarding Salome_Meca.

TTH12

I have error and could you pls advise.

johan

The call for feedback on the software is nearly 18 months old, so I trust that my feedback will still help to make Salome-Meca more useable. I am giving my comments only now because my recent experience (and frustrations) are still fresh.
I am a practicing mechanical engineer and I am using Salome-Meca to solve real practical problems 3 to 6 times a year, and this is usually done under time pressure. I am using SM on a standalone workstation, bought as Linux compatable, with a Nvidia graphics card. I say this because I am possibly one of the typical SM users.

I see most of the posts talk about training and related issues. For myself I have accepted that SM is a very, very powerful FREE software and the long learning curve is the price I have to pay to use it. I can anyway not afford anything commercially available and comparable with SM. There are excellent training videos from Cyprian Russu and others. Maybe a link to this training material on the website can help.

So my feedback and recommendations from my recent experience is as below.

1 PLEASE DEBUG THE SOFTWARE BEFORE MAKING IT AVAILABLE

I upgraded from Ubuntu 20.04 to 24.04 to find that the neither 2023 nor the 2024 version of SM is working. I understood that everything needed to run SM is packed into the singularity file, but apparently not, it needed some Debian 11 files. The help I could find on the forum (like deleting some .so files) did not work for me.

Next I thought I will do the WSL2 installation. The installation, following the instructions, went reasonably well. Creating a small model worked well but opening a previous large model, the graphics was so slow that it was virtually impossible to manoeuvre the model. Not a solution for me.

The only solution left for me was to re-install Ubuntu 20.04, which brings me to my next point.

2 PLEASE GIVE UPDATED INSTALLATION INSTRUCTIONS

The installation instructions on the download website is still for SM 2021. It is easy enough to change the file name, but

what about the prerequisites?
should I still use singularity or rather apptainer?
is it OS dependent and what is additionally required?
are there any hardware issues?

3 OTHER ISSUES

Related to my first point:

Running a simulations the first time on SM 2024 seems to work fine. Opening the hdf file again, it hangs at "Wait while AsterStudy finishes loading". SM 2023 opens it fine. (I have not checked this on the WSL version.)
Post processing a results file from Aster study (sometimes) shuts down SM. Opening the file through "Post process an external MED results file" does not give any problems.
Memory violation errors have been with me for a long time and still occur at the most inconvenient moments. I can be 90% sure that I will get one after using Homard.

Things I will find very useful:
- I usually work with large thin walled models and keeping track with the orientation of the flat elements can be challenging. If Code Aster knows the orientation is wrong, why not just fix it?
- Being able to connect shell elements to solid elements will be fantastic. The documents have said for a long time now that this feature is under development.
And on a positive note:
- As a non-English speaker and one that needs Google translate to understand a word of French, the document translations on github are absolutely fantastic. Keep up the good work.
- And thank you for a GREAT software.

k_zurawski

johan check if Strand7 will be in your budget.

AsterO'dactyle

Hello,

1/ The container has been thoroughly tested, but on dedicated systems for internal use. Unfortunately, there are still dependencies on the system in virtualisation systems, particularly in all graphical aspects. The problem is therefore generally Salome, not code_aster. 95% of the errors you have come from Salome and not from code_aster.

2/About orientation. This means that you have not fully understood how code_aster works (don't worry, I'll explain!).
A command generally modifies only one concept (one data structure), the one found to the left of the equal sign. As a result, code_aster can detect cases where the mesh is incorrectly oriented when pressure or contact is applied (via the skin elements), but these commands are AFFE_CHAR_MECA and DEFI_CONTACT, which are not allowed to modify the mesh.
This cannot be known in advance because a 2D cell (TRIA3, QUAD4, etc.) can be either a skin mesh if the model is 3D or a plate element. This information is provided by AFFE_MODELE (distinguishing between mesh/geometric support and model/dof description, based on Ciarlet's principle).
You can connect 3D and shell (LIAISON_ELEM).

k_zurawski

AsterO'dactyle (...) Unfortunately, there are still dependencies on the system in virtualisation systems, particularly in all graphical aspects. The problem is therefore generally Salome, not code_aster. 95% of the errors you have come from Salome and not from code_aster.

Do you know if Simvia is working on resolving those issues?

johan

AsterO'dactyle

Thanks for the reply. I see Salome -Meca as an integrated suite of programs and certainly hope that the people responsible for Salome would read my post.

Thanks for the orientation clarification.

Regarding linking elements, I found what I was looking for in LIAISON_MAIL. (I should have read the new U4.44.01 gitlab before I made the comment.) Having found this I thought I will test it on a simple model, just to be confronted with a new surprise: AFFE_CHAR_MECA does not apply the load where I intended it to be. The message is:

MAILLE : M97
MAILLE : M98
MAILLE : M99
MAILLE : M100

║ <EXCEPTION> <MODELISA6_96>
║
║ les 4 mailles imprimées ci-dessus n'appartiennent pas au modèle
║ et pourtant elles ont été affectées dans le mot-clé facteur : FORCE_ARETE

while my intention was to apply it to edge group "Load" elements 335 to 338.

My first question always is "What am I doing wrong." and in a real simulation it takes a lot of time and effort to find an alternative, if there is one, to get past the error. The alternative is that it confirms my first point.

I have used Salome_Meca 2023 with the default version of Code Aster. I attach my mesh and command files and will appreciate comments on it . (I can't upload the file, I will post a link to it.)

k_zurawski

When it comes to solver itself there's not much to talk about. It's a great piece of software, period.
Syntax could be easier, for example requesting field outputs in Calculix looks like this:
** Field outputs +++++++++++++++++++++++++++++++++++++++++++ ** *Node file RF, U *El file S, E, NOE ** ** End step ++++++++++++++++++++++++++++++++++++++++++++++++

I'd like to see QUAD4 shells capable of doing the same as COQUE_3D QUAD9 (elasto-plastic, large displacment and rotations, etc.).

I still don't know what to do with 'COMPOR1_47' alarm in shells, except ignoring it.

The documentation could describe better what the coefficients of arc-length are for and maybe show on examples how they influence the outcome.

I'd like to see contact with arc-length. This is somehow solved in other software.

The rest, like stability issues, lack of some functionality and other as you said are related to Salome Platform. Pity there's no commercial pre/post for Code_Aster.

k_zurawski

BTW, could we see the results of our feedback?

johan

k_zurawski

It is a long time since I have last looked at Strand7,but will do so again. I am also using Mecway, which I find very good for small models and it is very user friendly.