Testing a UMAT for biological tissue

[CF]

Hello All:

I've coded a UMAT for a soft biological tissue and it works fine for simple one-element models under uniaxial and biaxial tension. However, when I run the same tests for multi-element models it fails to converge after some time. Up to the point of failure the results are spot on. Abaqus gives some warnings about negative eigenvalues.
I am wondering could this be anything other than the definition of ddsdde? If it works for one element I don't see why it shouldn't work for any number. :? Could it be the element type? I am using C3D8RH. I've tried C3D8R and C3D8H.

Thanks in advance,

Cormac

[Jorgen]

Hello Cormac,

It sounds like a problem with ddsdde (i.e. the Jacobian). If you run a single element simulation with prescribed boundary conditions then you can get convergence even if the Jacobian is not quite right. For multi-element simulations that is not the case.

Also, since you see the right response up till the crash, that further implies that the ddsdde variable is not right.

I recommend that you print out ddsdde from within the UMAT subroutine to make sure that you get the numbers that you expected.

Best of luck,
Jorgen