Elmer keeps changing boundary indices when loading
Posted: 20 Oct 2019, 04:08
Hello,
I am currently fighting to get a working ElmerFEM toolchain on my Ubuntu 18.04 home PC. I installed the elmerfem-csc-eg package using sudo apt-get install. It works ok, when I setup a new model, no problem. The problem arises, when I exit Elmer, and reopen the program, reload the previously solved project by "load project" in the GUI, and try to calculate once more: Elmer has changed the used boundary indices for the defined boundary conditions while loading the project. The SIF is regenerated with a seemingly random assigment of boundary indices to boundary conditions. The new simulation then produces garbage for sure.
I suspect there is something with permissions, or the install locations of Elmer, but I cannot track it down. Can someone provide some useful hints?
I attach two ZIP files of a small cantilever beam test case project directory. The file "original.zip" contains the state of this working directory after the first successful solve. The file "2nd open.zip" is the contents just after loading the directory contents again with "load project" in ElmerGUI.
When more complex models are used (with more boundary conditions or bodies), everything is screwed up even more, equations get assigned to other bodies and such stuff. The only thing that Elmer seems to remember, is the number of boundary indices a certain boundary condition is linked to. But the index values are completely set by rolling the dice.
This has kept me up all night long, but now I realized I cannot correct this on my own. Any help is greatly appreciated.
Thanks, Thomas.
I am currently fighting to get a working ElmerFEM toolchain on my Ubuntu 18.04 home PC. I installed the elmerfem-csc-eg package using sudo apt-get install. It works ok, when I setup a new model, no problem. The problem arises, when I exit Elmer, and reopen the program, reload the previously solved project by "load project" in the GUI, and try to calculate once more: Elmer has changed the used boundary indices for the defined boundary conditions while loading the project. The SIF is regenerated with a seemingly random assigment of boundary indices to boundary conditions. The new simulation then produces garbage for sure.
I suspect there is something with permissions, or the install locations of Elmer, but I cannot track it down. Can someone provide some useful hints?
I attach two ZIP files of a small cantilever beam test case project directory. The file "original.zip" contains the state of this working directory after the first successful solve. The file "2nd open.zip" is the contents just after loading the directory contents again with "load project" in ElmerGUI.
When more complex models are used (with more boundary conditions or bodies), everything is screwed up even more, equations get assigned to other bodies and such stuff. The only thing that Elmer seems to remember, is the number of boundary indices a certain boundary condition is linked to. But the index values are completely set by rolling the dice.
This has kept me up all night long, but now I realized I cannot correct this on my own. Any help is greatly appreciated.
Thanks, Thomas.