NL-Quasi-Static Analysis with Friction Contacts + Pretension not converging
Altair EmployeeHello,
I am working on an non-linear-quasi static model with pretensioned bolts and friction contacts.
I set up the model with first order elements first, this model was converging fast and results were ok.
in order to get the correct tensions i changed the elements to second order.
After this i recreated the contacts and pretension to make sure everything is correctly linked to the second order elements.
If i run this model it is taking days and doing a few iterations until it starts all over again and again.
I use surface to surface contacts and a value of friction of 0,14.
I am not sure why this is happening because everything is the same as in the first order elements.
Could you please help? I uploaded the out. file and did a few screenshots.
Thanks a lot!
<?xml version="1.0" encoding="UTF-8"?>
<?xml version="1.0" encoding="UTF-8"?>
Answers
-
Hi,
Contacts with first order elements perform much better with second order elements due to oscillations. It is recommended to use 1st order elements for contacts instead of 2nd order elements.
0 -
Okay i understand, but how i should implement this in my model?
Do you mean i should use both types, first and second order, in one model?
Should i transform the elements around the contact only and keep the others second order?
Do i have to make a clear cut trough the solid model in a speific distance to the contact surfaces?
Or just transform the elements near the contact 'by hand'?
0 -
Hi,
Avoid midside node elements at contacts and check if this helps.
Or,
Refine mesh and have only 1st order elements and check the results.
0 -
0
-
I avoided second order elements near contacts this time but it leaded to the same issue as before.
Don't you have any other hint for me? Shouldn't it be possible to do an analysis with friction and second order elements?
And by refining you mean that i should refine the mesh as much until i reach the distance between midside nodes and edge nodes?
I need reliable results for the tensions.
0 -
I found this https://pveng.com/why-use-2nd-order-integration-elements-post/
According to it the error is getting very small when i reduce the element size.
But still its not what i am looking for.
Could i change the convergence criteria somehow?
0 -
Hi,
Parabolic elements with contacts will take more time because calculating time will be more for 2nd order equations.
My idea is having 1st order elements all over with finer mesh at contacts and coarse mesh on rest of the structure.
Did this help a all?
0 -
no sorry this did not help.
I build a another model today and i have the same problem as before.
But this time the problem also comes up when i use first order elements for the entire model.
Its not about the time, the models do just not convert. After a few iterations the calculation starts again and again as shown below.
Is there maybe another reason for this happening?
Please help.
<?xml version="1.0" encoding="UTF-8"?>
0 -
@Prakash Pagadala i sent you one of my models, so maybe you can have a look?
0 -
Thank you,
I will check and update to you soon.
0 -
@Prakash Pagadala any update on this`?
0 -
Hi,
Maybe Restart with statsub(Preload)?
0 -
Do you mean statsub(PRETENS) or something else? Where to find this`?
I am already using statsub(PRETENS)...
0 -
Hi @User123
I am having the same problem too. I know your post is dated almost 1 year ago but...did you happened to find something? If you don't maybe we can try to get a solution...
0 -
0
-
Hi Prakash,
Sorry for the late reply.
In short, I have a problem during non linear analysis of the structure contained in the file below.
As you can see, there are 2 subcases. For the first one (1.2mm of imposed displacement) there are seemingly contradictory (and wrong) results with what happens in subcase number 2 (2 mm of imposed diplacement). This happens when the model is made of second order elements. Not big problems are encontered when I use only first elements instead.I need to use second order elements for this analysis, so it is my priority to make it run.
Find attached my file below.
Best Regards
Matteo
0
