Contact problem

[highlander3]

Hi all,

I have an interesting problem with contact simulation.
I'm simulating contact between ball and plate (Hertz theory) and that works quite nice. But when i changed position of nodes ( tring to include ball's and plate's wear surface) in 3 direction (1 and 2 stays same) i always stuck with convergence (***ERROR: TOO MANY ATTEMPTS MADE FOR THIS INCREMENT)

Does anyone have some
advice for me?

Thanks in advance!

 

[corus]

It could be anything. Use contact controls to help. Have you used axi-symmetric geometry? With fewer degrees of freedom you'll have fewer problems.

corus

 

[highlander3]

Hi corus,
thx for ur replay but,

I'm using axi-symmetric geometry (half ball and half plate)in abaqus standard mode.
In the message file i have NUMERICAL SINGULARITY on some nodes too. Maybe that can play some role,I am not sure.
Also I included contact controls with automatic overcloser tolerances and automatic stabilization.

I can sand you my msg file,of course if u have some free time to look at it.

 

[xerf]

I assume you use automatic time incrementation:
When solving contact problems, you should set the minimum time increment to a very small value (e.g., 10^-12,the default is 10^-5). Also, you should set the initial time increment to a small value (e.g., 10^-2, the default is 1). You might need to set the maximum number of increments to higher value than the default (which is 100).

You might want to consider using the adaptive meshing capabilities in order to avoid excesive distortion of the elements which may lead to non-convergence.

 

[highlander3]

No I'm not using automatic time incrementation.

Interesting is that i have model that works but when i've changed node positions in input file (new geometry with wear surface) than i lost convergence in solution.
This procedure is very good way to change geomerty without need to mesh part again and i have an model which works well.

So procedure is good but i don't know why doesn't work with contact analysis.