[CIG-SHORT] Fault opening error

[Birendra jha]

Hi Brad

I did that. It did not solve the fault opening problem. 

Do you have some time to run this case? It should take 1-2 minutes. The main problem is that when slip is non-zero on the curved fault surface, slip vector on one of these nodes is assigned some non zero value in the normal direction during numerical iteration. Since fault is under compression everywhere, it gives an error from FaultCohesiveDyn::integrateResidual. Is it possible that after the iterations have converged, there is no slip component in the normal direction? If so, then this error check in inetgrateResidual is probably too strict?

Do you have a simple case where spontaneous fault slip is modeled on a curved fault surface?

Thanks and best regards
Birendra


--------------------------------------------
On Sun, 8/4/13, Brad Aagaard <baagaard at usgs.gov> wrote:

 Subject: Re: [CIG-SHORT] Fault opening error
 To: cig-short at geodynamics.org
 Date: Sunday, August 4, 2013, 8:23 PM
 
 Birendra,
 
 Your solver tolerances need adjustment. The FaultCohesiveDyn
 object in a 
 quasi-static simulation requires use of absolute tolerances
 in order to 
 properly detect stick/slip with the iterative solver (see
 the friction 
 session in the CDM2013 tutorial for details). You should
 lower the 
 ksp_rtol and snes_rtol tolerances to 1.0e-20 to force use of
 the 
 absolute tolerances.
 
 Brad
 
 On 8/2/13 2:35 PM, Birendra jha wrote:
 > Hi everyone,
 >
 > I am trying to run a model with a fault
 (FaultCohesiveDyn) in it. The fault surface is not planar
 but is curved. I am getting fault opening error at some
 node, which stops the simulation. Why does this happen? What
 should I do to avoid getting this error?
 >
 > Please see the attached files which should run.
 >
 > Here is what I have tried so far:
 > 1. different fault dips
 > 2. different fault traces (plane to curved)
 > 3. different fault friction coefficients
 >
 > My objective is to get some reverse slip on this curved
 fault surface. The slip does not have to be on the curved
 part, it can be on the part that strikes parallel to the
 y_pos boundary.
 >
 > Thank you very much for any help.
 >
 > Birendra
 >
 >
 >
 > _______________________________________________
 > CIG-SHORT mailing list
 > CIG-SHORT at geodynamics.org
 > http://geodynamics.org/cgi-bin/mailman/listinfo/cig-short
 >
 
 _______________________________________________
 CIG-SHORT mailing list
 CIG-SHORT at geodynamics.org
 http://geodynamics.org/cgi-bin/mailman/listinfo/cig-short