PF not converging

[Francis Oussou]

Dear all,

Please suggest what to do if the model is not converging.

I increased the iteration and reduced the total residual but it still is not. 

Kindly find attached the run script and share your view.

Best,

Francis

#-----------------------------------------------------------------------------

# Set solver parameters

#-----------------------------------------------------------------------------

nalohou_run.Solver.TerrainFollowingGrid = True

nalohou_run.Solver = 'Richards'

nalohou_run.Solver.MaxIter = 25000

nalohou_run.OverlandFlowDiffusive = 0

nalohou_run.Solver.Nonlinear.MaxIter = 800

nalohou_run.Solver.Nonlinear.ResidualTol = #1e-10

nalohou_run.Solver.Nonlinear.EtaChoice = 'Walker1'

nalohou_run.Solver.Nonlinear.EtaChoice = 'EtaConstant'

nalohou_run.Solver.Nonlinear.EtaValue = 0.001

nalohou_run.Solver.Nonlinear.UseJacobian = False

nalohou_run.Solver.Nonlinear.DerivativeEpsilon = 1e-16

nalohou_run.Solver.Nonlinear.StepTol = 1e-30

nalohou_run.Solver.Nonlinear.Globalization = 'LineSearch'

nalohou_run.Solver.Linear.KrylovDimension = 50

nalohou_run.Solver.Linear.MaxRestart = 3

nalohou_run.Solver.Linear.Preconditioner = 'PFMG'

nalohou_run.Solver.Linear.Preconditioner.PFMG.MaxIter = 5

nalohou_run.Solver.Linear.Preconditioner.PFMG.Smoother = 'RBGaussSeidelNonSymmetric'

nalohou_run.Solver.Linear.Preconditioner.PFMG.NumPreRelax = 1

nalohou_run.Solver.Linear.Preconditioner.PFMG.NumPostRelax = 1

nalohou_run.Solver.Drop = 1E-20

nalohou_run.Solver.AbsTol = 1E-9

#nalohou_run.Solver.WriteSiloSubsurfData = True

#nalohou_run.Solver.WriteSiloPressure = True

#nalohou_run.Solver.WriteSiloSaturation = True

#nalohou_run.Solver.WriteSiloConcentration = True

#spinup key  ## Read Readme washita

#nalohou_run.OverlandFlowSpinUp = 1

#nalohou_run.OverlandSpinupDampP1 = 1.0

#nalohou_run.OverlandSpinupDampP2 = 0.00001

[Reed M. Maxwell]

Please send the kinsol log for the timesteps that aren’t converging.

<dot>py (python scripts) are blocked by email FYI and don’t come through.

 

Run with only these solver settings, comment the rest out:

nalohou_run.Solver.TerrainFollowingGrid = True

nalohou_run.Solver = 'Richards'

nalohou_run.Solver.Nonlinear.MaxIter = 100

nalohou_run.Solver.Nonlinear.ResidualTol = 1e-5

nalohou_run.Solver.Nonlinear.EtaChoice = 'EtaConstant'

nalohou_run.Solver.Nonlinear.EtaValue = 0.001

nalohou_run.Solver.Nonlinear.UseJacobian = True

nalohou_run.Solver.Linear.Preconditioner = 'PFMG'

 

 

 

 

--
You received this message because you are subscribed to the Google Groups "ParFlow" group.
To unsubscribe from this group and stop receiving emails from it, send an email to parflow+u...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/parflow/CAE5G0iqXBTn_QU8pta0QomeZOH5MAqQS%2BbR1jzv9PJ_k5%3D-CJA%40mail.gmail.com.

[Francis Oussou]

Ok. Thanks.

Kindly find attached the kinsol log and a txt file of the run script.

The solver setting is changed accordingly. I will share the outcome when the simulation is done.

Best,

Francis

To view this discussion visit https://groups.google.com/d/msgid/parflow/DM6PR04MB6575AF2B4F23E6017EB73B51A6FA2%40DM6PR04MB6575.namprd04.prod.outlook.com.

[Francis Oussou]

Kindly find attached the kinsol log of the new solver settings.

Francis