Contact Penetration
Altair EmployeeHello Prakash,
I am trying to simulate a model with several contacts in Optistruct (Non linear quasi static). The Materials are all linear but with very Little Young Module (40-300Mpa). But there are big Penetrations between sliding parts and the results are not correct.
Contacts with Problem:
ID 16
ID 17
ID 26
ID 27
ID 28
<?xml version="1.0" encoding="UTF-8"?>
I have followed all Solutions in this Forum but it did not work out. I have tried following Solutions:
-Contact type: SLIDE
-Nlparam NINC increase to 200
-CNTSTB load collector and CNSTB Card in Param (EXPERTNL)
- MORIENT= Opengap
-Additional Clearance: 0.1
-String Value>>Adjust>> Auto
-S2S and N2S (Both tried)
-Track INFINIT and Consli (Both tried)
- Symmetric Contact for IP 16 (2 identical contacts with reverse Slave/Master Component)
- PCONT>>STIFF>>HARD
Because the Simulation takes so much time i share the H3D results too.
Thank you so much for your help
P.s: I should mention that this Model with this Material Parameter has been simulated successfully in Abaqus.
Answers
-
0
-
0
-
Thank you for your answer. Yes i tried it too, but results are much worse. Normal displacement acts better than LGDISP.
0 -
OK,
I have modified the model a little and waiting for the analysis to come out. I will update you soon,
0 -
I have tried 2 new Models with LGDISP.
1. NINC 100
2. NINC 300/ HASHASSM=Yes
Both of Runs have resulted Error:
Maximum number of time increment cutbacks reached.
In not converged results the model explods.
0 -
Hi @Farzin
LGDISP is not working. I tried the same. I tried after few modifications and I see no penetration but the model doesn't converge for 1st subcase.
I am trying to fix the same and I will share the updated model soon,
0 -
Thank you so much, Prakash
Currently i am trying a high value of NINC 1000. It takes a week to have the results.
0 -
Altair Forum User said:
Currently i am trying a high value of NINC 1000. It takes a week to have the results.
I think NINC has nothing to do with that.
0 -
Hi @Farzin
Can you try with N2S and provide a stiffness of say 1e5?
You can try with increasing the stiffness, but that may not give reasonable results.
I will update my finding soon with another modeling change...
<?xml version="1.0" encoding="UTF-8"?>
0 -
I am very thankful for your help.
I will try your idea and update you the results.
Also the results from NINC 500 are out. (The loads are a Little different, but the principals are the same)
There is 2 issues that i found:
First. there is a jump in the Situation of the Geometry between load step 1 and load step 2. (With CNTNLSUB in Load step 2)
it resets some parts of the geometry. I dont know if it causes the contact Problem, in first load step the contacts are functional a Little but it gets worse in load step 2.
End of load step 1
<?xml version="1.0" encoding="UTF-8"?>
Start of load step 2
<?xml version="1.0" encoding="UTF-8"?>
Second. As you can see in the first Picture, the pink part follows the blue part and goes upside ( Just like that it is Stick contact but it is actually slide) and i expected that it stays down.
Here is the Result file (Please ignore the movement of parts at the front, because the pressure was defined in incorrect direction)
https://www.dropbox.com/s/kmekskydyey7eie/R464.h3d?dl=0
0 -
New Update: i tried your idea with stiffnes 1e5 (NINC 100) and the results are so similar to NINC 500 (1 week calculation) but in a very shorter time (10 Hours), which it makes your solution more helpful. The Level of penetration is the same. Not ideal but much better than the initial Situation.
0 -
Hi Farzin,
It took 2 hrs on my machine with 1E5 stiffness. But my NINC is 10.
Increase stiffness may give good results.
0 -
Hello @Prakash Pagadala,
New Updates: I have tried a new Setting with LGDISP and the contacts work very good without any Penetration but the Problem is that in Loadcase 2 it does not converge after 60% and the elements beginn to deform incorrectly, i have tried NINC 100 for loadstep 2 and it didn't help. Do you have maybe any Idea about this Problem?
I appreciate your help
FEM:
H3D:
0 -
Hi @Farzin
Unfortunately, I am unable to download the files.
Could you please upload the same to my FTP link?
https://ftam1.altair.com/filedrop/~Ce4god
0 -
I sent the files via your Link again
0 -
0
-
Hi Prakash,
i found 2 wrong values in the model that i sent you. Could please correct them:
1- Node Set (ID= 2) ist redundant, you can delete it
2- The material Parameter for Material Meniscus (ID=3) ist E=59 and Nu=0.49
I am so sorry for the this wrong Settings
With Best Regards
Farzin
0 -
HI @Farzin
Altair Forum User said:1- Node Set (ID= 2) ist redundant, you can delete it
I have deleted this already and ran the model and it converges to 70%
Let me try with new material values and i will share my feedback soon,
0 -
Hi @Farzin
Can you update the solver to the latest?
I am using the file with latest solver update and it is through 70% and still running.
0 -
Do you mean with the new E value for Material (ID=3) it is over 70% now?
Which Version ist the latest? I am currently using Hyperworks 2017.0
0 -
Ok, the University has provided the 2018.0.
I will update you by new results
0 -
Hi @Farzin
The simulation is stuck at 70% convergence since the last couple of days and I have a .nl h3d which is around 4gb which I can't share right now.
I will check with experts on this and update you soon,
0 -
Thanks a lot for your efforts.
I tried with solver 2018 too.
Loadstep 2 converges till 68% and then stops.
0 -
Hi @Farzin
Any update?
If not, try with these settings,
Use S2S instead of N2S
Use CONSLI instead of FINIT
Set S1=1.0e-4 in CNTSTB
Add NLADAPT, NOPCL,0 and run in 2018
0 -
Thank you for your update
I tried your Suggestion, and the model converges to 78% (Load step 2)
did you make it till 100% convergence?
If yes, can you please share the Modell (.FEM).
0 -
Not really,
I am still waiting for the results to come out from 2018 solver.
0 -
@Farzin After few test cases, I couldn't the see the convergence. I remember you said that the same model converges in Abaqus,
Is the OptiSrtruct deck converted from Abaqus deck?
Is it possible to share the Abaqus deck with us?0 -
Thanks a lot for your efforts.
Unfortunately, I have no acces to Abaqus Modell. I just received the Mesh and built the model all by myself in Hyperworks. I read a Research, that used the same Model in Abaqus. I used their boundry conditions for my Modell (same Material_same Forces ...). I have no clue about Abaqus convergence Algorithm.
/emoticons/default_mellow.png' title=':mellow:' />
I will update you, if i find something new.
0 -
OK, I just wanted to check if there is anything missing from the model as I see severe element deformation
If you can find the Abaqus model, please let us know or if you can share the research paper details could help to crack.
0
