I Have chosen PPM/calculix since it is easy to learn and best of free cae software . I hope that next versions will be better and better …
The OP is using Pardiso. And yeah, PaStiX may still have some issues with modal analyses.
I’m using Pardiso. A completely free model should not deform like this.Even if one contact has fail. Strain energy was suspiciously high so i checked the animation. There is something wrong going on ¿isn’t it? .
I can think of four possible sources of the problem.
-Wrong eigen value extraction. I mean that is not the fifth and sixth modes but higher ones.
-One BC is still active.
-There is some limitation regarding a part being Master and Slave at the same time.
-I have done some mistake or wrong consideration.
I think that reduce the options to two:
-Wrong eigen value extraction (solver or Prepomax reading the dat file).
-There is some limitation regarding a part being Master and Slave at the same time and Abaqus is able to detect it and change it.
I have computed with other software (MW) + exactly the same ccx solver.
I have adjust master and slave Tie tolerances and respect same surface never be defined as master and slave in different contacts and I got same result as Abaqus.
Those last two modes shown on Prepomax are modes 7th and 8th and do not corresond to the free Rigid body modes.
By the way. This is called Grounding Check Technique which I recommend as good practice to check contacts performance. It focusses on the first 6 modes and Strain Energy of the free-free frequency analysis.
1 Like
Then we know that it’s because of the difference in how Abaqus and CalculiX treat overconstraints.
Is that a civil engineering nomenclature ? I know this approach but I’ve never heard this name.
I knew about it recently.
I used to perform the frequency analysis but typically to check Boundary conditions and discard rigid body motions. This one allows to check and clean contacts.
There are some tutorials on You tube where it calls it Grounding check.
Seems Nastran implemented it with that name but I have ear Ansys users to use that name too. Not sure where it originally comes sorry.
In my experience, a single surface may be a master surface only in one of the constraints. That is why PrePoMax tries to swap master/slave surfaces when using Search Contact Pairs in such a way to fulfill this requirement (one surface is master only in one constraint). This limitation does not affect the contact pairs.
First, I tried creating the constraints using the Search Contact pairs tool that fulfills the aforementioned condition. I ran the free frequency analysis and the results were NOT OK since I did not get 6 almost zero frequencies.
Then I merged the constraints provided in the shared model in a way that one constraint is created to connect one surface to multiple other surfaces (select multiple constraints → Right click → Merge). Four separate constraints remained like this one:
I ran the simulation and the 6th frequency is rotational and equals 0.2 Hz.
The modified frequency analysis file:
arek2_MB.pmx (3.4 MB)
I see I am a little late with the response but the file might help someone 
But maybe I can try to add a feature to merge the contact pairs in a way that the surfaces are not repeated in more than one tied constraint.
1 Like
Same strategy here. As few masters as possible. I got only two contact definitions. That’s good to know for users.
I think it comes from the aerospace community- NASTRAN-based, if I recall correctly.
They are and will be better, just be patient and contribute to the community by submitting bugs and having these questions that you have in here. It is the only way we can make the software better !
2 Likes
Yep, that looks just like some of my models in the past with modes that are not realistic.
This is exactly why the community is important for open-source developers who usually do not have the resources to completely check/debug the code.
2 Likes