Problem in calculating plastic stress for hyperelastic material (user defined subroutine)

[whelana5]

I am doing simulations of a cyclic pressure applied to a surface of a circle. The model is a quarter model of a circle, with the pressure applied in the z-direction, the thickness is 0.4 mm and diameter 20 mm, where the material is a user subroutine of a hyperelastic matrix-fibre composite (neo-hookean matrix and an exponential model for the fibres).

The UMAT calculates a plastic stress specifically in the neo-hookean matrix component, according to the peak strain at each cycle, in order to produce a permanent set (i.e. unrecoverable strain) when unloaded. The simulations break down after a few pressure loading cycles, with no obvious reason why- there aren't any major stress/strain concentrations, and the permanent set values are increasing linearly with each cycle. The message file error is just that the solution appears to be diverging. The simulations converge for lower levels of plastic stress but the model inputs are increased to produce more plastic stress per cycle, the simulation breaks down. It appears that its the levels of plastic stress that are causing problems in the solution converging.

Can anyone help with this or suggest why the simulation might be breaking down? Thank you in advance.

 

[FEA way]

Try with refined mesh, use more elements per thickness. Enable automatic stabilization (in step settings). If it keeps diverging use the Job Diagnostics tool available in Visualization module to know more about the reason of nonconvergence.

 

[IceBreakerSours]

Did you try to confirm the UMAT is, in fact, coded correctly by comparing the results with an independent (say, analytical) solution? Have you checked if everything is working as it ought to with simplified models that run efficiently?

*********************************************************
Are you new to this forum? If so, please read these FAQs:

http://www.eng-tips.com/faqs.cfm?fid=376

http://www.eng-tips.com/faqs.cfm?fid=1083

 

[whelana5]

Thank you for your replies- I haven't tried the automatic stabilization so that might work. I've also used up to 4 elements through the thickness and it doesn't help with convergence.
Yes I have checked that the UMAT is correct- the material model is a HGO an it matches the result from using the in-built HGO model (its the permanent set that is the novel part of the subroutine).

 

[IceBreakerSours]

Does the simulation run efficiently without any warning messages - using a supported material formulation?

Also, while comparing your UMAT results with the supported law is a good indicator, I was referring to comparing the entire code with an independent (non-FE) solution. Assuming the code is correct, you could reduce the max. time step around the time you expect the simulation to run into trouble. Alternatively, if you have a VUMAT formulation, give explicit a try. There is no tangent to implement so convergence is a non-issue.

*********************************************************
Are you new to this forum? If so, please read these FAQs:

http://www.eng-tips.com/faqs.cfm?fid=376

http://www.eng-tips.com/faqs.cfm?fid=1083