1. Get CalculiX binary: http://www.dhondt.de/ccx_2.8p2.tar.bz2
2. Untar: tar xf ccx_2.8p2.tar.bz2
3. Rename file CalculiX/ccx_2.8p2 to ccx and move to the same directory as FreeCAD executable. I operate from build directory, so it's build/bin
4. Test if ccx has all libraries:
Code: Select all
$ ldd ccx linux-vdso.so.1 => (0x00007fff9802d000) libpthread.so.0 => /lib64/libpthread.so.0 (0x00000035de800000) libgfortran.so.2 => /lib64/libgfortran.so.2 (0x00007fd70d5a3000) libm.so.6 => /lib64/libm.so.6 (0x00000035df000000) libc.so.6 => /lib64/libc.so.6 (0x00000035de400000) libgomp.so.1 => /lib64/libgomp.so.1 (0x00000035fd600000) libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x00000035df800000) /lib64/ld-linux-x86-64.so.2 (0x00000035de000000)
5. Get libgfortran2 from rpmfind. I use x86_64, so it's
wget ftp://rpmfind.net/linux/Mandriva/offici ... x86_64.rpm
6. Install it: sudo yum install libgfortran2-4.2.3-6mnb2.x86_64.rpm
7. Install brilliant FreeCAD macro provided by psicofil: download/file.php?id=12095
8. Start FreeCAD
9. Make/load a part for FEM analysis and switch to FEM workbench
10. Run the macro by psicofil GMSH mesh (gmsh has to be installed, but it's in fedora official repo)
11. Select "Algorithm: tetgen" - (I get crash on netgen and some other problems with iso)
12. Select the part and click OK in the macro window to create mesh
13. Select the mesh in project tree an click "Create a new mechanical analysis" (icon is a yellow letter A)
14. Click "Material definition" (icon: yellow sphere)
15. Select material and click OK
16. Select the mesh and _hide it_
17. Add constrains and forces to the _part_ (not to the macro-generated mesh) - that part works really nice
18. Click icon "Start calculation" (icon: yellow bending rectangle)
19. Click "Write CalculiX input file" and "Run Calculix"
Proper output in fem console looks like this:
Code: Select all
21. Hide the part and unhide the macro generated mesh
22.Click the rainbow coloured icon to show the results. Select stress type/displacement etc.
The whole process is still not reliable if not started from scratch tower i.e. re-running calculix might not work, even reloading the file with already defined analysis might not allow to run calculix, but I found it very reliable if started from point 9 above.
Any feedback will be very appreciated - I might make a video or an installation script for CalculiX if there is any interest.