Sticky contact error termination

[Rakefet]

Hi,
I would like to simulate a viscoelastic multi material performing an elastic deformation to an elastic material and then releasing it.
As a start I gave all parts neo Hookean material because I have read in the forum they work well with contacts. I started with a cylinder and a box created in FEBio geometry. Between the parts I choose a Sticky contact because they are not in contact at the beginning only after they reach each other they should be connected. I have a running simulation that works with 2 steps; one pushing and another releasing. I have added my results in file: “cylinder_2_steps_StickyContact1_Runs_No_Penetrati on_To_Slave”
I tried performing the same simulation with geometries imported from isomash matlab.
As long as the part do not touch the slave (prescribed displacement > -6) the application runs properly. When I change the prescribed displacement so that it touches the slave I get and error.
File: “StickyContact1_Error_termination_if_BC_prescribed _displacment_is_lower_than_minos6”
If I use the same simulation but with tied elastic contact (which should not work, they do not touch at the beginning) the simulation runs but after contacting the slave it doesn’t perform elastic deformation instead it goes through the substrate. I played with the penalty factor and augmented values; nothing helps.
File: “TiedElastic3_Runs_But_penetrates_the_slave”

This problem is before I change the materials to viscoelastic.

Do you have any suggestions?
Thank and Regards,
Rakefet

[maas]

Hi Rakefet,

It looks like the model is underconstrained. Only a displacement in Z is prescribed, but the X, Y degrees of freedom remain open. This allows parts of the model to move in ways that do not affect the stresses in the model. This results in numerical ill-conditioning and a chance that the model will invert elements and fail to converge. I think adding some additional constraints will help. Regarding the specific models:
- cylinder_2_steps_StickyContact1_Runs_No_Penetratio n_To_Slave: even though it is under-constrained, this model is highly symmetrical, which is probably why it runs.
- StickyContact1_Error_termination_if_BC_prescribed_ displacment_is_lower_than_minos6: This one fails, most likely because it is under-constrained.
- TiedElastic3_Runs_But_penetrates_the_slave: The tied contact ignores nodes that are outside of the contact search radius. Due to the initial gap, none of the nodes are considered in contact. Even though under-constrained, the model runs because there are no lateral forces in X, Y.

Cheers,

Steve

[Rakefet]

Hi,
Thank you for your answer it helped and now it works.
I added BC and changed the contact to sticky and now the part goes down and creates strains at the surface below it without penetrating.
I would now like to add another step. In the second step I would like to see what happens to the surface when the sub-parts of the main part that went down move. I have a simulation that works but only if the sub-parts are connected with a tied elastic contact and only if I have the BC displacement of x,y fixed. It works but penetrates, does not create strains within the sub-parts.
I would like to give the sub-parts opportunity to contact new surfaces (sticky, right?) and I would also like them to be free from x and y fixed displacement.
Because the part is now sicked with the surface I thought it might be possible.
Do you have a suggestion for me?
Thanks a lot,
Rakefet

[maas]

Hi Rakefet,

If your parts are penetrating, then that indicates a problem with your contact definitions, or possibly, a conflicting boundary condition. It's difficult to say without seeing the file. If you still have problems, perhaps upload your new model, and I can take a look at it.

Cheers,

Steve

[Rakefet]

Hi,
I have attached 2 smaller versions of my problem.
The one that ends with "sticky" is defined with a sticky contact for the sub-parts and does not converge.
The one that ends with "works" is defined with tied contact and converges but the movements within the sub-parts penetrate.
Thanks a lot,
Rakefet