Learning Forum

[Ian Whitby]

Hi,

It has been over a week of trying to solve this issue myself without much success so I decided to seek for help on the forum. 

I am working on a friction brake model which is very similar to the in-depth tutorial provided by Ansys: (link to courses.ansys.com) Thermo-Structural Analysis of a Brake

I have noticed that the exact problems are present even if I download the tutorial file and try to run it on the HPC cluster without a single modification. 

The convergence/ deformation problems:

1. At first, the brake pads were moving far away from the rotor causing a convergence issue. This is likely due to lack of force/ restriction applied to the pads during initial spin-up of rotor. (See fig. below, hand annotations in purple colour).

2. The rotor tends to deformate radially as soon as spun up. Counterintuitively, it scales down along radius. The stress values on rotor increase irrationally. 

Procedure:

1. I downloaded the .wbpj file from ansys.courses and opened it in 2023R1
2. I written the .dat input file, migrated the files (.dat, .sub and work directory) to our HPC running on RHEL and scheduled the work with PBS .sub script for MAPDL solver.
3. At first, the brake pads displaced far away from the rotor causing convergence issues. ''Error: The value of UY at node 12941 is 1000331.42''
4. I applied the stops on both translational joints (see fig. below). It helped, the simulation converged for longer, but brake pads still displaced beyond the stops (-0.3,0.3 mm).

Troubleshooting:

1. I double checked, the large deflection is on. I notice that I did not modify the simulation file apart from the stops applied to the brake pads' translational joints.
2. Possible legacy issues? I am using ANSYS 2023R1 whilst the tutorial was created in 2020R2.
3. Perhaps the released tutorial file was not the latest version used in the recorded lectures?

Edit r01:

Another possibility regarding troubleshooting: Is there possibility for a mismatch of units between my workstation's Ansys and the HPC cluster?

Any insight into this matter would be invaluable. 

Many thanks and regards,
Ian

[Ian Whitby]

After a few days I came to a solution myself and am sharing in case somebody will face similar issues in future.

The first problem with displacement of brake pads was linked to excessive penetration causing problems. Playing with frictional contact parameters resolved the issue completely. 

The second problem regarding the radial deformation of rotor is linked strictly to post-processing tool. If results are saved for n-th time step, the animation goes crazy. As soon as I saved all time steps to results, animation looks 100% accurate.

There is a last missing piece of puzzles: stress around revolute joint. I am reaching >9000 MPa. I noticed that the stress is correlated strictly to the angular motion of the disc (the disc is accelerated 0-0.1s, no force on pads).

Info about revolute joint:

Could someone advise me please what am I doing wrong?
Many thanks,

Ian

[Ashish Khemka]

Hi Ian,

Thanks for updating the post. The high stess looks to be because of singularity. The stress is seen on a sharp edge.

Regards,

Ashish Khemka