Abaqus Deep Drawing Stability

[FODDD]

Hi, I am modelling a deep drawing process of a thin sheet and am having some issues with the stability of the model. I started my analysis using a step time approx. equal to the lowest mode of the blank (frequency = 290.50s-1 giving a period = 0.0034s and I used 0.003s). This gives reasonable results but if I try to carry out a mesh convergence study the model produces oscillatory type deformation in the deformed blank near the die entrance radius as the mesh is refined. I also get this issue if I try to increase the time step above 0.003, with some elements stretching by a huge amount and the model eventually failing.

Has anyone come across this before or what can be done to rectify it? I've attached an image of these oscillations in case it helps.

Thank you.

 

[FODDD]

Apologies, if I coarsen the mesh also to get rid of these oscillations then the ratio of ALLKE to ALLIE increases near the start of the analysis and can become noisy for a longer period of time into the step. I believe I should be minimising this noise but unsure of what is an acceptable limit. Abaqus state 5-10% but is this at any point during the analysis or is it highest ALLKE compared to highest ALLIE?

Thank you

 

[TERIO]

What kind of elements are you using?

 

[FODDD]

I'm using linear axisymmetric CAX4R elements

 

[FODDD]

I should have mentioned from the outset that I'm conducting the analysis in explicit, because of the large non-linearities associated with the plastic strain and contact.
Also, the sheet is approx. 0.14mm thick and I'm deforming it 10mm deep which has proven experimentally.

I would be happy to take any ideas that come to mind as it is getting quite urgent to sort this issue out now.

Thank you

 

[TERIO]

Have you tried adjusting or using different options for hourglass control?

 

[FODDD]

Hi, yes I've tried all the available options and there does not seem to be any variation in the results whatsoever.

 

[TERIO]

For explicit dynamics smaller elements means smaller stable time steps. If the problem is quasi-static you could use mass scaling to increase the stable time step some. Fewer ties steps may help since there will be less steps for the hourglass displacements to accumulate.