Strange deformation in bearing dinamic analisys (2D)

[Fabi0]

Look at the video in the link:

https://vimeo.com/40404888

Analisys stopped because of this error:
===================================================================
***WARNING: In element 230715 of instance ESTERNO-1 the ratio of deformation speed to wave speed is 0.31876 at increment 4134657. See the message file for further information.


***WARNING: Step 2, Increment******: 1 material point(s) failed to converge in the material constitutive routines.


***ERROR: The ratio of deformation speed to wave speed exceeds 1.0000 in at least one element. This usually indicates an error with the model definition. Additional diagnostic information may be found in the message file.


The following checklist may be helpful in diagnosing the error:

1. Check contact definitions for problems such as excessive initial
overclosure or unrealistic tied definition between contact pairs.
A vector plot of velocities or accelerations will usually help to identify contact problems.

2. Check stiffness (elastic modulus) and mass (density) definitions for consistent units and verify that the combination is reasonable.

3. Check for poor mesh definition.

4. Check the boundary conditions for an excessive loading rate. The
*DIAGNOSTICS, DEFORMATION SPEED CHECK=DETAIL option may be used to obtain detailed diagnostics information.

5. Check the current status of the structure to see if it has totally failed.

6. A dashpot or a very stiff spring may cause the analysis to go unstable. The *DYNAMIC, DIRECT option may be used to control the time increment directly.

***ERROR: The fatal error will generate a new field output in the last increment. All variables applicable to the current procedure and material type(s) will be written to the output database.

===================================================================
===================================================================
Mesh, density&units, boundary condition are ok.
The only thing is the friction coefficient.
I used:
1- 0.11 for roller vs roller contacts
2- 0.11 for roller vs raceway contacts
3- 1 for outer ring vs lamination cylinder contact
Maybe that the tangetial force is too high beacause of a too high friction coefficient (3) ?
And this gave a stramge outer ring shape like in the last increment?

I's my first bearing dynamic analisys, so i'm not very skilled

 

[rstupplebeen]

To me it looks like it was running fine then numerical noise increased and crashed your analysis. Can you run your analysis in less time? It looks like the dynamic step is 1.5seconds. Double preceision may also improve the robustness. I hope this helps.

Rob Stupplebeen

https://sites.google.com/site/robertkstupplebeen/

 

[Fabi0]

Thx for the tips Rob, i'll try with double precision because i need a long step time if i want to have a complete revolution of the outer ring.

Just another question: usually I work on 3d analysis. This is my first in 2d, and the only reason of a 2d approach is the computational cost (with our machine, a 3d dynamic analysis is impossible).
The 2d model has a thickness of 1mm. The "real" roller is 20mm long, so in the 2d model i impose a force that is "total3DLoad/20" : is it a correct approach?
I ask this because the stress that i found in the first static step is higher than what i give from the same condition in the 3d analysis: it looks like if the 2d analysis take into account the stress gradient that is usually on the border of the cylindrical roller, near the plane face, but why, if i'm in a 2d model?

I've just done a test with a 3d cube with a pressure on its top face and an encanstre on the opposite face, then i've take a square that is a 2d section of the cube and i've applyed the same load and boundary condition and....the stress is the 2d section is not the same than a middle section (far away from borders) of the 3d. I can't manage it out

 

[Fabi0]

On the left the 2d, on the right the "corresponding" middle section of the 3d

 

[rstupplebeen]

Your 1/20th loading is appropriate. On one of the cut faces constrain it in that plane. On the other constrain the faces to remain parallel to the cut plane by using a tie constraint. When comparing stresses make sure to look at the center section of your 3d model.

A plane strain model could be more appropriate and computationally lighter. Basically this is a similar assumption to axisymmetric but for long aspect ratio structures such as I-beams. I have not tried using plane strain for implicit or explicit dynamics though so that will require some help manual searching.

I hope this helps.

Rob Stupplebeen

https://sites.google.com/site/robertkstupplebeen/

 

[Fabi0]

Thx again for your answer Rob.

But it's not clear for me: what do you mean with "cut faces"? i.e for the roller the cut faces are the "two circle" along z axis?
If so, they are just costrained in that plane...if i choose bc the z direction doesn't appear because 2d analysis.

 

[Fabi0]

up?