[CIG-SHORT] small-scale crack opening convergence problem

[Brad Aagaard]

Scott,

One of the main issues is that your only boundary conditions are two 
Neumann BC, so your model has a null space consisting of translation and 
rotation. You need to constrain your domain with one or more DirichletBC 
to prevent rigid body motion while keeping just one Neumann BC.

Your ksp_atol needs to be less than your snes_atol by at least 1-2 
orders of magnitude.

Sorry for the delay in my response. I have been out of the office.

Regards,
Brad


On 10/29/2012 08:25 AM, Scott Henderson wrote:
> Hello CIG-users,
>
> I've been trying to use Pylith for a "single edge notch in tension
> (SENT)" fracture mechanics simulation. In order to do this I've
> modified step20.cfg to be a 2D plane-stress problem. I've also
> followed what I could from this mailing-list post
> (http://www.geodynamics.org/pipermail/cig-short/2012-September/001120.html).
> I'm still not able to get expected results, which makes me think I'
> doing something incorrectly with the pylith settings (PETSC in
> particular). I'm pretty new to the software, so any tips on the cubit
> mesh or pylith cfg files would be appreciated (zipped folder
> attached).
>
> Essentially, I'm not getting the expected u2 displacements from this
> analysis (but they are not unreasonable), and the following SNES
> output makes me concerned:
>
> 500 SNES Function norm 2.483466683115e-06 Nonlinear solve did not
> converge due to DIVERGED_MAX_IT iterations 500
>
> Increasing snes_max_it doesn't seem to lead to convergence.
>
> Any suggestions appreciated, Scott
>
>
>
>
>
>
>
> _______________________________________________ CIG-SHORT mailing
> list CIG-SHORT at geodynamics.org
> http://geodynamics.org/cgi-bin/mailman/listinfo/cig-short
>