Instability fixes for NiTi stent

[ryankb]

I've made a whole lot of progress this past week on my stent analysis. I have it effectively going through a three-step expansion/annealing procedure before a crimp - but now the best I can do is about 61% of the way through the crimp step. Abaqus standard procedure. I think that I'm seeing some type of instability issues, but all I've managed to do is increase from 31% to 61% before failure - but alas it still fails and I've run out of ideas.

I have nlgeom turned on, as well as automatic stabilization and adaptive stabilization (using defaults for both - playing around with the numbers didn't help mush at all that I was able to find at least).

I tried altering solution controls Io and Ir to 8 an 10 respectively, but this job is waiting in line. I'm not incredibly hopeful on these variables though.

I also tried replacing the crimp static step with an implicit quasi-static step. This also required me changing some symmetry in my model. My job is currently waiting in line to run this one so I hope my new definitions are correct and the dynamic can help.

Last thought for now: could I be having problems due to working "outside" the material model? At 61%, I'm showing 21% strain which is exceptionally high. At 53% I'm only at 7.6%, so it's clear the last two steps are somehow messed up. My material model doesn't account for going this high, but my attempts to date with trying to include plasticity in the nitinol model haven't been successful. Despite defining (or believing I'm defining) the material correctly per the UMAT instructions, the material seems to act as though the second half of the model (plasticity data points) aren't there.

Any other thoughts? I have my cae file attached in-case anyone is really curious.

 

[Dave442]

I've had problems with C3D8I elements in analyses like these before. In the users manual it also states: "For trapezoidal element shapes the performance (of C3D8I elements) is not much better than the performance of regular displacement elements." If you look at the shape of the elements along the inner radius of your stent they are predominately trapezoidal. I opened your model, switched to reduced-integration elements with enhanced hourglass control (C3D8R) and the job ran to completion.

I also noticed that you redefine the model change and contact interactions for each individual step. You don't have to do this. for a model change, there two boxes "deactivated in this step" and "reactivated in this step". You can use these to control activation/deactivation of the stent elements during each successive anneal step. Similarly for contact interactions - there is a box in the bottom left of the Edit Interaction window "active in this step". you can tick/untick this box to control activation/deactivation of the contact interaction during each successive anneal step.

Also, you don't need stabilization or dynamic implicit for these analysoes.

 

[ryankb]

Thanks for the tips on the elements. I'll check that out.

As far as the model change definitions - it appears that after an anneal step I cannot simply reactivate as it would seem. The box on the bottom left of the interaction for "activate in this step" is grayed-out, not letting me select it. Also, It looks to me that if I deactivate any interaction, I am not allowed to reactivate it if I "miss" a step. That's why my work-around was to redefine multiple contact interactions. I assumed what you mentioned should be the most efficient method, and it seemed odd to me that I wasn't "allowed". I'm sure it's because I'm not doing the correct order of operations. My work-around is clunky, but it seemed to actually work - for now.

My overall plan is to work to get this one running fully to completion, then explore some of the alternatives/extras, such as nitinol material plasticity, summing all nodal forces (still haven't got this to work yet), and streamlining some of these contact definitions you noted.

 

[Dave442]

you misunderstood me - you don't make changes in the Interaction Manager. You edit the interactions themselves. You should have one interaction for your model change (anneal), one for contact with the expander and one for your contact with the crimper. When you double click on each interaction in the Interaction Manager, you should open an "Edit Interaction" dialog box.

For the model change (anneal) you use the options "Deactivate in this step" and "Reactivate in this step" to control the activation/deactivation of elements throughout the entire analysis. So for each anneal step you would tick "Deactiate in this step". Then for the following Expansion step you would tick "Reactivate for this step". The interaction remains active throughout the entire analysis and should show as "modified" in the Interaction Manager.

Similarly, for the expander contact, you should open the "Edit Interaction" dialog box. In the bottom left is a box you can tick "Active in this step" that you use to control the activation/deactivation of the contact constraint during the anneal steps. So for each anneal step you would untick "Active in this step". Then for the following Expansion step you would tick "Active in this step". The interaction remains active throughout the analysis and should show as "modified" in the Interaction Manager.