[CIG-SHORT] Sign of shear stress changes from one fault node to another

[Romain Jolivet]

Dear Brad,

Thank you for the precisions. I checked in my log file and there is one time step that diverged, although it does not corresponds to the time step when the first of the nodes concerned sees its friction dropping to negative values.
I will check how my tolerances are set up and I will get back to you (I perfectly remember reading this one order of magnitude thing between tolerance and zero slip rate in the manual).
Could it also have to do with the time step (too long maybe)?

I'll keep you in touch.
R

Sent with my phone, sorry for typos...

> Le 5 nov. 2015 à 17:26, Brad Aagaard <baagaard at usgs.gov> a écrit :
> 
> Romain,
> 
> The first thing to check is the behavior of the linear and nonlinear solves. Are they converging at every time step? Are the absolute and relative tolerances set to appropriately relative to the friction zero sliding tolerance?
> 
> The relative sizes of the absolute KSP and SNES tolerances are very important. This is what works in most of our examples. They may need to be tuned slightly for your particularly problem, such as greater relative separation (here there is one order of magnitude difference).
> 
> [pylithapp.problem.interfaces.fault]
> zero_tolerance = 1.0e-11
> [pylithapp.petsc]
> # Linear solver tolerances
> ksp_rtol = 1.0e-20
> ksp_atol = 1.0e-12
> # Nonlinear solver tolerances
> snes_rtol = 1.0e-20
> snes_atol = 1.0e-10
> 
> Regards,
> Brad
> 
> 
>> On 11/05/2015 05:20 AM, Romain Jolivet wrote:
>> Hello,
>> 
>> I am running a 2d, fully elastic experiment in which a fault
>> (pylith.faults.FaultCohesiveDyn) with Rate and State friction
>> (pylith.friction.RateStateAgeing) cuts through the entire rectangular domain.
>> Gravity is on.
>> Left boundary of the model has a pylith.bc.DirichletBC dirichlet BC set so that
>> it moves horizontally, with the entire domain in compression.
>> Surface is stress free while bottom boundary is set to compensate gravity with a
>> pylith.bc.Neumann BC. One point at the surface is fixed so the problem can be
>> solved.
>> Initial pylith.faults.TractPerturbation on the fault is set so that shear and
>> normal stresses are 0.
>> The fault is entirely rate-strengthening a>>b and I just wait until the fault
>> slips at steady state (simulation runs 10000 years with 10 year time steps).
>> 
>> The final distribution of stresses along the fault is very awkward (see
>> attached) as, for some reason, Shear Stress is negative for certain nodes, while
>> it should be positive everywhere.
>> I looked at the orientation of the normalDir, but it is constant along the
>> fault, so it does not seem related.
>> Any thoughts?
>> 
>> Cheers,
>> R
>> 
>> PyLith version: git repository downloaded on sept 30th 2015
>> PETSc version: git repository downloaded on sept 30th 2015
>> 
>> 
>> —————————————————————————————————————
>> —————————————————————————————————————
>> Romain Jolivet
>> Postdoctoral Fellow
>> 
>> University of Cambridge
>> Department of Earth Sciences
>> Bullard Labs
>> Madingley Rise
>> Madingley Road
>> Cambridge CB3 0EZ
>> United Kingdom
>> 
>> email: rpj29 at cam.ac.uk <mailto:rpj29 at cam.ac.uk>
>> Phone: +44 1223 748 938
>> Mobile: +44 7596 703 148
>> 
>> France: +33 6 52 91 76 39
>> US: +1 (626) 560 6356
>> —————————————————————————————————————
>> !!!! MOVING TO ENS, PARIS, FRANCE IN NOVEMBER !!!!
>> —————————————————————————————————————
>> 
>> 
>> 
>> _______________________________________________
>> CIG-SHORT mailing list
>> CIG-SHORT at geodynamics.org
>> http://lists.geodynamics.org/cgi-bin/mailman/listinfo/cig-short
> 
> _______________________________________________
> CIG-SHORT mailing list
> CIG-SHORT at geodynamics.org
> http://lists.geodynamics.org/cgi-bin/mailman/listinfo/cig-short