Convergence problem with a non-linear elastic material model

[sagar]

Hi everyone,
I am using a non-linear elastic model to predict the response of soft tissues with single collagen fiber family (collagen fibers oriented in only one direction).

I have my lab data which is almost linear elastic in nature. I am trying to verify my model with that data. I have implemented my model in fortran (attached file).

Element used is simple C3D8. When I try to run the model for any orientation of fibers other than 0 deg it is giving me an error saying: ***NOTE: THE RATE OF CONVERGENCE IS VERY SLOW. CONVERGENCE IS JUDGED UNLIKELY.

***ERROR: TOO MANY ATTEMPTS MADE FOR THIS INCREMENT

I am using symmetric boundary conditions on 3 faces in 3 directions and stretching in the X-dir. But when I tracked the deformation gradient (F) I observed that instead of having F(2,2) and F(3,3) less than 1, they were going beyond 1 even though I was stretching only in the X-dir.

I have already simulated non-linear elastic and viscoelastic models by extending this model, so I know that it should work. But I am just not able to locate where the error is occuring. Please help me out in this.

Thanks
Sagar

[Jorgen]

Have you tried to use the "unsymm" option to the user material specification. I did not study your UMAT in detail, but for anisotropic materials the Jacobian is often unsymmetric.

Jorgen

[sagar]

Hi Dr.Bergstrom,

Thanks for your quick reply.
I am defining my Jacobian in the material model itself. I am defining the Jacobian as constant symmetric. So it does not make a difference if I use the UNSYMM option. But do you think it is fine to use a constant symm Jacobian in this case? Or should i just let abaqus calculate the jacobian and then use the UNSYMM option?
I have used this constant jacobian previously for my other simulations which involved viscoelasticty as well as anisotropy and it had worked fine. So I am not sure where the problem lies.
If you have a look at the code...I have defined I4 as the strain in the fiber direction with D4 as the orientation tensor. The strange thing happening is when I have my fiber say at 45 deg to X-dir (anything other than 0 deg) and if I apply a displacement in the X dir then the I4 goes less than 0, which does not make sense. Do you have any suggestions on this?

Thanks
Sagar

[Jorgen]

I am afraid that I did not get a chance to verify your code. Your error message is likely caused by an inaccurate jacobian. I recommend that you redirive your jacobian and try again ...

- Jorgen