Non-linear punching simulation exploding

Crashphys

Hi,

 

I am trying to model a punching simulation in implicit, but upon contact the model basically explodes like the picture below. I have successfully modeled this in explicit, but I would like to try it in implicit and compare. Can anyone suggest why this is happening? When it happens I get an error:

 

 *** ERROR # 4965 ***
 Maximum number of time increment cutbacks reached,
 analysis aborted.
  
 *** ERROR # 5905 ***
 Severe element distortion detected,
 analysis aborted.

 

I set NCUTS to 50 using NLADAPT card. I suspect something might be wrong with the contact interface perhaps?

 

 

 

Adriano A. Koga

i'm wondering what material are you using and which units.

Looking at your material, the LIMIT1 under MATS1 is only 100. Your Young modulus is extremely high compared to that.

So at the minimum touch, there is plasticity already.

 

Besides checking these values I would recommend you to impose a control in maximum increment DTMAX in NLADAPT.

As there seems to be lots of plasticity, a tighter increment control might be needed.

Also consider changing the NLPARM to NLPARM(LGDISP) if large strains are expected.

 

Crashphys

Hi,

 

I implemented the changes as you suggested and they were very helpful. It now works well, but for some reason the contact interface for the bottom is not working right (see image below). Do you have any ideas why? I have not been able to figure it out.

 

Thanks

 

 

 

Adriano A. Koga

do you have the out file? it is always nice to take a look at it first.

Crashphys

Hi, Sorry yes I do. It does not seem to give indication of any errors, but I do have this warning:

 

 The structure of surface-to-surface contact element is quite different than
 that of node-to-surface contact element. No internal CGAP/CGAPG elements are
 created for surface-to-surface contact.

 

 

Adriano A. Koga

i forgot to mention thath in your case, as there's lots of sliding in the contact region, it is recommended to use TRACK in contact groups, as FINITE or CONTINUOUS SLIDING as there is a lot of changing in contact pairs. Oherwise, contact pairs are not recalculated through the analysis (SMALL SLIDING).

 

Also, when running a NL model for the first tiem, i would recommned using PARAM,NLMON,DISP as it saves a _nl.H3D file that can be opened in HV in real time, and checking critical areas for convergence.

Adriano A. Koga

Maybe in your problem also we could try N2S contact. Your mesh size is too different between master and slaves. It might cause some issues in detecting contact properly.

I believe N2S with FINITE/CONSLI might give better results. But we need to test. 

Crashphys

Your suggestions all worked. Thank you for the help!