[CIG-SHORT] Nonlinear solution did not converge

[Charles Williams]

Dear Chunyu,

It’s hard to tell for certain what the problem might be without seeing an output log.  You might want to double-check your mesh quality again.  For many problems, especially those involving fault friction, you will want the best possible quality.  Do as much mesh smoothing as possible (probably using condition number smoothing), and also check the mesh quality using condition number.  Ideally, you would like to get the values all below 2, especially near the fault.

Some other questions I have:

1.  Is your linear solution also converging slowly, or just the nonlinear solution?
2.  Have you tried the solution settings in share/settings?  I would suggest first trying the solver_fault_fieldsplit.cfg settings, and then possibly the exact settings if that doesn’t work.
3.  Send us your log file for a run that did not converge.  We may be able to tell something from that.

Cheers,
Charles



> On 28/07/2015, at 4:34 am, Chunyu Liu (cliu5) <cliu5 at memphis.edu> wrote:
> 
> Hi all, 
> 
> I try to lock a fault shoulder and keep the central part of the fault free to displace by changing friction coefficients and cohesions. 
> 
> Like the following database, keep lower cohesion in the central part but linear changing values in the shoulder of the fault. 
> 
> But when the max cohesion is bigger than 1000MPa, It could not converge. 
> 
> The mesh aspect ratio( from Trelis) :  most 1~3 , some of them are 4.5 ( ideally 1~3, but I think the mesh quality is ok).
> 
> How can I turn on inner solves to further diagnose things?
>  Are there any thing I can do to make it converge or even converge faster ? 
> 
> Thanks for your time
> 
> Solution parameters
> 
> # Preconditioner settings.
> pc_type = asm
> sub_pc_factor_shift_type = nonzero
> 
> # Convergence parameters.
> ksp_rtol = 1.0e-10
> ksp_atol = 1.0e-9
> ksp_max_it = 5000
> ksp_gmres_restart = 50
> 
> # Linear solver monitoring options.
> ksp_monitor = true
> ksp_view = true
> ksp_converged_reason = true
> ksp_error_if_not_converged = true
> 
> # nonlinear convergence parameter
> snes_rtol = 1.0e-10
> snes_atol = 1.0e-3 ( Just save time) 
> snes_max_it = 5000
> 
> # Nonlinear solver monitoring options.
> snes_monitor = true
> snes_linesearch_monitor = true
> #snes_view = true
> snes_converged_reason = true
> snes_error_if_not_converged = true
> 
> 
> Fault properties spatial database 
> 
> #SPATIAL.ascii 1
> SimpleDB {
>   num-values = 2 // number of stress components
> 
>   // names of the fault coeffients
>   value-names = friction-coefficient cohesion
> 
>   value-units = None MPa  // units
> 
>   num-locs = 33
>   data-dim = 2
>   space-dim = 3
> 
>   cs-data = cartesian {
>     to-meters = 1.0
>     space-dim = 3
>   }
> }
> // Columns are
> // (1) x coordinates(m)
> // (2) y coordinates(m)
> // (3) z coordinates(m)
> // (4) friction_coefficient
> // (5) cohension(MPa)
> -163000 300000  0           0.6     1000
> -163000 300000  -60000      0.6     1000
> -163000 300000  -120000      0.6    1000
> -111000 222000  0           0.6     978.8375
> -111000 222000  -60000       0.6    978.8375
> -111000 222000  -120000      0.6    978.8375
> -74000  166500  0           0.6     555
> -74000  166500  -60000      0.6     555
> -74000  166500  -120000      0.6    555
> -47000  126000  0           0.6     420
> -47000  126000  -60000      0.6     420
> -47000  126000  -120000     0.6     420
> -37000  111000  0           0.6     0
> -37000  111000  -60000      0.6     0
> -37000  111000  -120000      0.6    0
> 37000        0  0           0.6     0
> 37000        0  -60000      0.6     0
> 37000        0  -120000      0.6    0
> 111000 -111000  0           0.6     0
> 111000 -111000  -60000      0.6     0
> 111000 -111000  -120000      0.6    0
> 121000 -126000  0           0.6     420
> 121000 -126000  -60000      0.6     420
> 121000 -126000  -120000     0.6     420
> 148000 -166500  0           0.6     555
> 148000 -166500  -60000      0.6     555
> 148000 -166500  -120000      0.6    555
> 185000 -222000  0           0.6     978.8375
> 185000 -222000  -60000       0.6    978.8375
> 185000 -222000  -120000      0.6    978.8375
> 237000 -300000  0           0.6     1000
> 237000 -300000  -60000      0.6     1000
> 237000 -300000  -120000      0.6    1000
> _______________________________________________
> CIG-SHORT mailing list
> CIG-SHORT at geodynamics.org
> http://lists.geodynamics.org/cgi-bin/mailman/listinfo/cig-short

Charles A. Williams
Scientist
GNS Science
1 Fairway Drive, Avalon
PO Box 30368
Lower Hutt  5040
New Zealand
ph (office): 0064-4570-4566
fax (office): 0064-4570-4600
C.Williams at gns.cri.nz <mailto:C.Williams at gns.cri.nz>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.geodynamics.org/pipermail/cig-short/attachments/20150729/868d1b83/attachment.html>