The analysis consists of a modeling of the tube and the well’s wall with D_PLAN elements. The topic of this modeling was to study the effect of ovalization and its influence in relation to collapse by applying a uniform load along the outer edge of the tube. The tube was modeled with a certain ovalization (two concentric ellipses) in order to favor the collapse.
Is expected that, at the end of the loading application, the contact between the tube wall with the well wall occurs first and then a self-contact of the inner region of the tube, as previously obtained in another simulation using the software ABAQUS. There's one image attached that shows the simulation on ABAQUS.
Based on this, a simulation was first carried out without defining the contacts to verify the behavior of the material and identify the instant when the contact would occur and also if the contact of the tube with the well would actually occur first.
Then, after achieving the desired behavior, the contact configuration was defined through operator DEFI_CONTACT. First I tried the DISCRETE mode, but I didn't achieve convergence. Then I tried the CONTINUE mode and it was possible to get a result, however, completely non-physical. At the exact moment the contact occurs, the displacement field takes on absurdly high values and the model itself does not appear on the plot. I'm attaching some pictures with the sequence of timesteps that show this behavior.
The simulations were carried out with the DYNA_NON_LINE solver. First, simulations with PETIT as behavior were performed to check if the contact converged. It was possible to achieve convergence, however, the self-contact occurred before the contact between the tube and the well. Due to this results, I tried to perform simulations using SIMO_MIEHE behavior with VMIS_ISOT_LINE (which requires the definition of the ECRO_LINE parameter within the DEFI_MATERIAU operator), to include the hypothesis of large deformations and rotations, however the results obtained were those previously exposed.
I also tried to change the behavior to GROT_GDEP (that's why there is the TRACTION part in the definition of the material), but for this case it was not possible to achieve convergence.
I've tried to refine the mesh of the well, the mesh of the tube (which is already the most refined, as it is the slave mesh), the time vector and even tried to fix the edges of the well ( before that i was fixing only the faces), however the result was the same.
I would like to know if someone has gone through something similar or has any recommendation or hint of what could be causing this behavior which is very strange once the simulation does not present any "error". It is completed "successfully", but the result is completely non-physical.
I am attaching the .COMM and the .MED that I used in this simulation to help with the analysis.
Thank you for attention,
Daniel Rodrigues.
Attached file:
Forum.zip, 559153kb
