thread799-313846
Finally found a solution to this, for Abaqus 6.10-1 on Ubuntu 10.04 (and perhaps other versions of both)
Salome Meca 2011-2 (and perhaps other versions) uses numpy, scipy and matplotlib in the right versions for Abaqus 6.10-1. What worked for me:
1. copy these versions of...
What is definitely possible is to first run a static analysis (with Standard I mean) to map your SIGINI and then do a *Import, state=yes in a second Explicit run to pick up those stresses from the first model.
I also had a similar message the other day, does this surface belong to a rigid body? I think in my case it had something to do with the fact that I defined the surface in the Part, then again in Constraints.
Anyway, in my case it was quite easy to go into the .inp-file using a text editor...
You can also use the user subbroutine film.f. That way you can program a virtually moving sink (your liquid N2) to gradually move up the tool, as a function of time. You can even change the heat-transfer coefficient as a function of for example temperature, if you want to.
Unfortuately Abaqus...
What can help in such cases:
Edit Step
Incrementation tab
Stable increment estimator: Element-by-element
Time scaling factor: 0.9
If you can attach a zip with your model or place it on some fileshare, I'd be happy to have closer look.
As far as I know, the only way is to use the command line:
abaqus job=<.inp-file> user=<.f-file>
You need to have the intel fortran compiler installed.
If you want to, you can send variables to the UMAT when it's called. These you can enter in the GUI, MATERIAL/GENERAL/USER...
I have a little experience with SHRINK for cold-installing interference fit fasteners, it works like a charm. Often it is recommended to combine it with SURFACE SMOOTHING.
Hard to judge your model without seeing it. At t0, is the ring smaller than the rod? If so, SHRINK seems the option to...
In windows it works by defining the environment variable PYTHONPATH (and making sure the add-on packages such as numpy are for the python version of your Abaqus install). There are a few posts on this forum about that already.
But how to do it on Linux?
Sorry don't follow: you should now have a .inp without *PART etc., and instead one long list of nodes & elements?
If you want to output specific nodal information, it is best to define a node set and request field- and/or history output for that?
Thanks for the help, in the ftplugin file for the Windows version 7.3 I found:
if has("folding")
" Fold all lines that do not begin with *
setlocal foldexpr=getline(v:lnum)[0]!=\"\*\"
setlocal foldmethod=expr
let b:undo_ftplugin .= " foldexpr< foldmethod<"
endif
I'll check what...
That's right, mesa is the only way supported on Linux.
Graphics acceleration is therefore not as good as on windows -- but still workable, on a fast system.
Gvim or vim is great for editing .inp-files because of its speed: large models can be opened quickly.
In Windows, current versions of gvim (I use 7.3) automatically fold the text, i.e. if the .inp-file reads:
*Node
1, 2.4418478, 0., -0.199697748
2, 2.41744566...
I believe UMATHT is the sub-routine best suited for calculating k and cp.
Let us know how it went (share the code if you can...) -- the simulation you want to do is very interesting indeed
In the end I remembered that both SPOOLES and ARPACK need to be compiled with "make lib" rather than just "make"...
This is working for me (Ubuntu 10.04 with proper dev libs installed)
