Scripting in Abaqus

[3JC]

Dear collages,

In this moment I have designed a Python script of a model. How could I define a parametric study, modifying one of the parameters of the problem (for example, imagine a cantilever rectangular beam of length L with a distributed load q, and I would like to see the behaviour by modifying the Length from 5000 to 6000).

Thank you very much for your attention.

Best regards,

Juan José

 

[mechfeeney]

I've done this but I modeled more complicated geometries. I interfaced matlab with abaqus as a means of generating .inp models because I too wanted to see the effects of changing geometries. If you are simply testig the length of a beam..my interface technique is not necessary. You can use ngen and elgen commands to code. The .inp directly. Let me know some more details and I can surely help you.

 

[3JC]

Dear mechfeeney,

Thank you very much for your answer. My model is too complex (is the model of a T-stub, whit contact and plasticity, for example). I think that I can't work in the *.inp file because if I made changes in the geometry, the *.inp file has information also about nodes, and I have to remesh, do not? For this reason I have develop a *.py script, and I want to study the influence of variables as diameter of bolts, distances between bolts, thickness of profiles, etc.

Model of a T-stub you can see in next link:

http://www.youtube.com/watch?v=8mbm2RFfrjc

Thank you very much. Regards,

Juan José

 

[DanStro]

Though I can't remember which ones, I believe there are a couple of example files that use scripting to run a parametric study.

 

[mechfeeney]

Yes the .inp was easier to work with for my case as I was using beam elements. Remeshing is not necessary. The seed size (which is embedded in the .inp) is all that is needed for meshing. I'm not entirely sure if this is true for 3d elements (solids). Are you using 3d elements? The model you have seems relatively easy to change manually. Linear bolt pattern for the two bolt holes?

 

[mechfeeney]

Well now that I think about it, you do have quite a few variables, so automation might be necessary. Have you solved the problem? Or are you looking for more help?

 

[3JC]

Dear mechfeeney:

My model is a 3d solid model, with contact and considering non-linearity of material (elastic-plastic behaviour).

About solve the problem, I have only one question: when I apply a little load the problem run well, but when I apply a high load the problem runs with error. I think it is due to plasticity and from the load when it is not a structure (it is a mechanism) it fails. So, how can I apply a criteria to finish the calculation and do not have any error?

Finally, I have about 7-8 variables that I want to study. For this reason, I want to develop a parametric model (first calibrate with experimental data from tests, and then to develop a parametric study to study the influence of modifying variables).

Thank you very much.

Regards,

Juan José

 

[mechfeeney]

Did you enter the plasticity parameters in tabular form? If so you should make sure that you have enough points defined: Yield Stress per unit Plastic Strain. If you only have one point defined...obviously this won't work for high deformations. So make sure you have enough points to describe the Yield Stress vs. Plastic Strain curve. Because of the fact that it works for small loads and fails for high loads, my guess is that it could be that you don't have enough points defined.
As for creating a parametric model:
I did a very similar procedure for a truss structure. Like I said, I wrote a MATLAB file(s) that described all of the geometry in parametric form. My code can generate all of the nodes and elements (properly numbered) and eventually can generate an input file (.inp) to send to ABAQUS. The nice thing about MATLAB is that you can run ABAQUS inside of it and submit batch files for multiple iterations of your model (i.e. multiple material or geometry variations). I can send you my thesis describing how this works if you'd like.
As you probably know (and are doing) get that first model working for any loads then tackle the design iteration problem. This sounds like a very fun project.

 

[3JC]

Dear mechfeeney:

Yes, I have enter the curve with only two points (the elastic and plastic regimen) because in this moment I haven`t got the results of the steel coupon test. I would be very interesting in see your thesis. And what you have done in Matlab, I think it is also possible in other plataforms, do not? I have developed my analytical models in Mathematica.

Thank you very much for your attention. I'm very gratefull.

Regards,

Juan José