Strange Friction Contact Behaviour

Numerical methods and mathematical models of Elmer
Post Reply
KrishanSingh0706
Posts: 3
Joined: 10 Nov 2022, 17:12
Antispam: Yes

Strange Friction Contact Behaviour

Post by KrishanSingh0706 »

Hi everyone,

I was trying to run a contact analysis with friction but encountered some strange behaviour. I have simplified the problem to two bodies, one inside another with a positive y-force applied to the inner body. The Mechanical.sif file contains the setup I would like to run but the solution is having some trouble converging. When I look at the output vtu I see that there is a lot of motion in the x-direction. I have increased the friction coefficients for the contact surfaces hoping this would restrict x-motion but changing this does not seem to have any effect. I have also tried reducing the body force but still the solution fails to converge.
Mechanical.png
Mechanical.png (8.17 KiB) Viewed 488 times

The Mechanical_constrained_in_x.sif is the same setup but BodyB's displacement 1 is set to 0 and this is showing more sensible results.
Mechanical_constrained_in_x.png
Mechanical_constrained_in_x.png (29.61 KiB) Viewed 488 times

However, I need to get a setup without this constraint working so I can transfer it over to the actual contact analysis I would like to run. This has different geometry with more bodies and forces are applied in a direction not aligned with X/Y axes.

Any guidance would be much appreciated.

Thank you very much in advance. :)
Krishan
Attachments
Mechanical Test Case.zip
(20.54 KiB) Downloaded 46 times
kevinarden
Posts: 2237
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: Strange Friction Contact Behaviour

Post by kevinarden »

In order to have a friction force you have to have a normal load, in this case your downward Y force, but you also have to have a load trying to push the compressed block orthogonal to the normal load, I don't see any orthogonal load.
kevinarden
Posts: 2237
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: Strange Friction Contact Behaviour

Post by kevinarden »

displacement.png
(20.45 KiB) Not downloaded yet
Mechanical.sif
(3.39 KiB) Downloaded 49 times
KrishanSingh0706
Posts: 3
Joined: 10 Nov 2022, 17:12
Antispam: Yes

Re: Strange Friction Contact Behaviour

Post by KrishanSingh0706 »

Hello,

Thank you for your prompt response.

However, I think my issue has been misunderstood. Taking the first case again where force is only applied in the y-direction, I would expect minimal displacement in the x-direction. The solution is struggling to converge hence the result we are seeing:
Convergence1.png
Convergence1.png (15.2 KiB) Viewed 460 times

I ran the SIF file you sent back and running that also fails to converge on a solution:
Convergence2.png
Convergence2.png (16.87 KiB) Viewed 460 times

I am wondering what is wrong with the setup in the Mechanical.sif file I sent to cause these convergence issues and how I can go about resolving them? When I swap out the friction contacts with tie contacts it is OK but I need to use friction contact for these interfaces.

Thank you.
Krishan
kevinarden
Posts: 2237
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: Strange Friction Contact Behaviour

Post by kevinarden »

The elastic solvers have problems converging when rigid body motion is allowed. If the is no x restraint, or any other rigid body direction restraint. Then rigid body motion is allowed, (i.e infinite displacement with no force applied). This mathematically causes convergence issues. Solutions are to ensure there is some sort of restraint in every possible rigid body motion. Not always possible with friction or contact. In the cases of friction or contact you have to ensure that there are initial contacts or forces to oppose motion. In your original sif you had no X restraint and no X force, rigid body translation allowed which the solver can't handle. The sif I posted ran fine for me with no issues.

In real life with only Y force you would suspect no X motion, but in linear elastic numerical analysis, unrestrained directions with no load can have infinite motion., because it is a numerical solution, at least in steady state. Maybe try transient solution instead (which would include acceleration forces)
kevinarden
Posts: 2237
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: Strange Friction Contact Behaviour

Post by kevinarden »

It works fine in Steady State and Transient without doing those 100 nonlinear iterations and not using the relaxation factor

Nonlinear System Max Iterations = 1
! Nonlinear System Relaxation Factor = 1.0
Post Reply