SOL 601 - Strange Non-convergence in Non-Linear Contact Problem

[jacksonfem]

Hi everybody!

I 'm trying to perform a contact analysis. As you can see in pics 1 and 2, I 'm trying to pass the ball through the 2 plates. I do the followings:
1) Plates and ball from shell elements
2) Displacement assigned on ball, not force.
3) SOL 601 Non-linear static, not transient
4) ATS method, line search on, iterations 30.
5) Plates are not rigid targets
6) Default contact parameters
7) Large displacements, strains and angles enable
8) Elastic materials (ball E= 10MPa)

At a first analysis, I separated the displacement to 100 steps, with step size 0.01sec (total time 100*0.01= 1sec). The analysis failed to converge at step 34 (time 0.335sec). I thought that may I needed smaller time steps around step 34. Pictures 3 and 4 shows the position of the ball at the final step and picture 5 the nonlinear steps procedure. No strange stresses or strains or displacements.

So, at a second analysis I separated the displacement with very many steps from step 32 and after. 32*0.01+100*0.001+58*0.01= 1sec. Again failed to converge at the same time (0.03348sec).
But, in the second analysis, I saw very strange convergence ratios path near the end of the analysis, as you can see in picture 6.

Could somebody give me an advice of what is happening and a way to proceed?

 

[BlasMolero]

Dear jacksonfem
Very important: you have symmetry in both loads & geometry, then use it!. This is critical, split your model by the X-Y plane (TZ=0) as well as the Z-X plane (TY=0) and study 1/4 model, this way your FE model is stable, the prescribed symmetric boundary conditions will stabilize the FE model and you can play with items like number of steps to use, nonlinear material properties, contact algoritm, solution convergence, etc.. but first you need to start from an stable model, OK?.

Best regards,
Blas.

~~~~~~~~~~~~~~~~~~~~~~
Blas Molero Hidalgo
Ingeniero Industrial
Director

IBERISA
48004 BILBAO (SPAIN)
WEB:

http://www.iberisa.com

Blog de FEMAP & NX Nastran:

http://iberisa.wordpress.com/

 

[jacksonfem]

Dear Blas,

although you are right to your suggestions, the model as I have modeled it is stable.
Some "centric" nodes of the ball are such constrained, so that every movement (x,y,z) or rotation (x,y,z) of the ball is constrained.

The strange thing, is the convergence path...I don 't know what this symmetry no-convergence path does means.

I "played" very much with contact algorithms, solution steps (as I wrote)...nothing is change.