by

Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    Gerasimos Chourdakis
    @MakisH
    Regarding your second error (CalculiX) I have no clue
    Yuxiang Lin
    @yuxianglin
    The error is straight forward, the calculix cannot find mesh file, right?
    check your path of mesh file all.msh
    neko2650
    @neko2650
    @yuxianglin the example is taken from the tutorial-precice https://github.com/precice/tutorials/tree/master/FSI/cylinderFlap/OpenFOAM-CalculiX/Solid it has no all.msh file.

    I assume that the FSI part needs some configuration update for OpenFOAM 7, I am not sure if I have seen this before. I would need to check the release notes of OpenFOAM 7.

    Thank you

    I look forward to it.
    seung-eon-jang
    @seung-eon-jang
    @MakisH The Force and Displacement data confirmed by the watch point function is different.
    (calculating up to 2 seconds without stopping
    estarting after stopping at 1 second)
    Benjamin Uekermann
    @uekerman
    @seung-eon-jang How much different? Please note that preCICE does not do a "proper full restart", e.g. the IQN columns are constructed from scratch again. Thus, differences in the size of the coupling convergence measure threshold can be expected.
    seung-eon-jang
    @seung-eon-jang
    Up to 5% error is occured.
    What does mean "proper full restart" ?
    Benjamin Uekermann
    @uekerman
    That preCICE does not restart all components, but rather starts from scratch again. 5% is a lot though. I guess sth in OpenFOAM or CalculiX then goes wrong. Can you successfully restart both solvers separately?
    seung-eon-jang
    @seung-eon-jang
    I will restart each solver separately and compare.
    Yuxiang Lin
    @yuxianglin
    Hi, I wonder in Calculix does anyone tried interpolating the pressure (not concentrated force)and pass it to xload rather than xforc? Currently I use qu8r as structure element but preCICE does not support the quad element to date, and split the quad into triangles and use conservative interpolate scheme does poor in getting the concentrated force in my implementation
    Kyle Davis
    @KyleDavisSA

    the example is taken from the tutorial-precice https://github.com/precice/tutorials/tree/master/FSI/cylinderFlap/OpenFOAM-CalculiX/Solid it has no all.msh file.

    If there is no all.msh file, did you run the pre_flap.fbd file?

    David Schneider
    @DavidSCN
    @yuxianglin In case you are using Openfoam for coupling you might want to checkout precice/openfoam-adapter#125
    Gerasimos Chourdakis
    @MakisH
    @seung-eon-jang the residualControl issue you found is actually a precice/tutorials#40, depending on the OpenFOAM version (behavior changed in OpenFOAM 6). The precice/tutorials#78 is to change to the new format, but you can also just remove the residualControl for now. The tutorial is not exactly the same as the Turek case anyway.
    neko2650
    @neko2650

    the example is taken from the tutorial-precice https://github.com/precice/tutorials/tree/master/FSI/cylinderFlap/OpenFOAM-CalculiX/Solid it has no all.msh file.

    If there is no all.msh file, did you run the pre_flap.fbd file?

    Form previous gitter message I found
    ```
    cgx -bg pre_flap.fbd

    ``` is it the required code I need to write in the terminal. I am new with Calxulix environment, it would be a big help if someone shares location for Calculix solver setup. As I have seen the Setup for fem simulation https://github.com/precice/precice/wiki/CalculiX_setup and it has prebuilt all.msh . But I have to create my own file I don't know how it's done in Calculix. Thanks in advance.

    Gerasimos Chourdakis
    @MakisH
    @neko2650 how did you start the simulation? This should already be handled by the Allrun script (as well as the runSolid, if you prefer to run the separately).
    ajaust
    @ajaust
    Is there a way to let preCICE stop the simulation if the coupling diverges (in a certain number of time steps)? I am running a parameter study and I would expect that not all parameter combinations will lead to a stable coupling.
    Benjamin Uekermann
    @uekerman
    no, not yet. one of the solvers needs to crash. but sounds like a helpful feature 👍. feel free to open an issue.
    ajaust
    @ajaust
    I will do that.
    Also another question regarding IQN-IMVJ. When I run into a segmentation fault, which parameter is too large? It runs for some time steps and iterations and ends with a segmentation fault. That happens for parallel-implicit and serial-implicit coupling. The same test case with IQN-ILS runs normally.
    ajaust
    @ajaust
    Or does it rather mean that my simulation blew up such that I have nans in my solution?
    Frédéric Simonis
    @fsimonis
    @ajaust Segfaults should never occur in preCICE. Do you mind opening an issue with additional information, so that we can track down this issue?
    ajaust
    @ajaust
    I can do that. This could be caused by my code then. I use FEniCS via the Python interface (not the preCICE-FEniCS adapter) and it runs in parallel, too. So plenty of potential pitfalls. :S
    Yuxiang Lin
    @yuxianglin
    Hi, preCICE usrs, I wonder does anyone familiar with dload in Caluclix, it seems it can only apply with normal pressure , I wonder can it apply shear traction loading?
    Benjamin Uekermann
    @uekerman
    @ajaust With logging as in the first example here it should easy to find out whether the segfault happens within preCICE or not.
    Yuxiang Lin
    @yuxianglin
    Does anyone know the Calculix discussion group?
    the yahoo group it provides seems not working properly
    Martin Saravia
    @martinsaravia
    Hi guys, I am trying to compile the Calculix adapter. The adapter was realease for version 2.15 of Calculix, but the currenct version is 2.16 (I cannot find a repository with old versions). When I compile the adapter I get an undefined reference error:
    bin/ccx_2.15.a(nonlingeo_precice.o): In function nonlingeo_precice': nonlingeo_precice.c:(.text+0xf6b9): undefined reference toprecfd_'
    nonlingeoprecice.c:(.text+0x13480): undefined reference to `calcmatwavspeed'
    Can you help me whit this? Thank you in advance.
    Yuxiang Lin
    @yuxianglin
    I think you have problem in compiling the calculix
    first trying to compile calculix alone
    then trying to compile the calculix with preCICE
    make sure the path in Makefile is correct
    Benjamin Uekermann
    @uekerman
    @martinsaravia
    There is already an adapter version for v2.16: https://github.com/precice/calculix-adapter/tree/v2.16
    You can also get older CCX versions by inverse engineering the path: http://www.dhondt.de/ccx_2.15.src.tar.bz2
    The error you get look indeed like a version conflict.
    @yuxianglin
    The Yahoo CalculiX group still exists, but Yahoo shut down some of the services some months ago. They are thinking about moving to another service, but for the moment you can still ask questions through the Yahoo group.
    I got your message there.
    ajaust
    @ajaust
    @fsimonis @uekerman I will debug a bit more and see if I can figure out what causes the segmentation fault. If I trace it down to be preCICE's fault, I will open an issue.
    I guess for the debug log level I need a debug build of preCICE?
    Benjamin Uekermann
    @uekerman
    yes
    ajaust
    @ajaust
    I think the next time I will just compile everything as ReleaseWithDebugInfo :D
    Benjamin Uekermann
    @uekerman

    I think the next time I will just compile everything as ReleaseWithDebugInfo :D

    That unfortunately does not give you the Debug logging of preCICE :/

    Frédéric Simonis
    @fsimonis
    @ajaust you can overwrite the RelWithDebInfo Cxx flags with -o2 -g (Omit -DNDEBUG) That should do the trick.
    ajaust
    @ajaust
    Ok. I will recompile stuff and see what happens. Maybe FEniCS also gives some hints regarding what might be wrong.
    neko2650
    @neko2650

    @neko2650 how did you start the simulation? This should already be handled by the Allrun script (as well as the runSolid, if you prefer to run the separately).

    I gave the path to cxx_precice in Allrun file. It worked. Now I am looking in preCICE env for my case setup. Previously I tried to couple Matlab with OpenFOAM but I was stuck in linking libraries and compiler compatibility (gcc-6.3x required for Matlab 2020). From https://www.cfd-online.com/Forums/openfoam-programming-development/226781-openfoam-coupling-matlab.html it got to know about preCICE. I have a query is it advisable to use Matlab instead of CalculiX to couple with OpenFOAM using preCICE.

    Martin Saravia
    @martinsaravia
    Great @uekerman, I did not see there was a version for 2.16. Thank you very much!
    Martin Saravia
    @martinsaravia
    Hi guys, I have everything working well with version 2, thank you very much!. I have a question regarding the solid side using Calculix, is the CLOAD set name Nsurface hardcoded? I cannot find where it is specified, so I thought it may be fixed in the adapter code.
    venkcy
    @venkcy
    Hi! I am trying to couple OpenFOAM with Calculix for FSI problems and was successfully solve few cases in Clear Linux OS with the your help. Now I am trying to replicate the same in Mac OS. As I can find the installation procedure for Calculix and preCICE for Mac, the problem is with OpenFOAM. Home-brew did not work (displacing DARWIN not supported). So, I am currently using docker. But, docker for OpenFOAM uses builtin GCC not the system GCC. So, if I install calculix and preCICE using brew, will I be able to couple it? Any input will be really helpful. Thanks
    Kyle Davis
    @KyleDavisSA

    Hi guys, I have everything working well with version 2, thank you very much!. I have a question regarding the solid side using Calculix, is the CLOAD set name Nsurface hardcoded? I cannot find where it is specified, so I thought it may be fixed in the adapter code.

    Hi, no the name is not hardcoded. The name of the surface in the .inp file must match the name in the config.yml file under the patch name. This name must also be at the top of the .nam file. The tutorials available for Calculix use other names, so you can use them as a reference

    Benjamin Uekermann
    @uekerman
    @venkcy If you use TCP/IP you can couple through docker containers, see also here: https://precice.discourse.group/t/run-fenics-adapter-of-precice-when-fenics-runs-with-docker/187/3
    Alternative would be to have everything in one docker container?