hydraulic fracturing example in the basic examples fails to converge

[zhishuaizhang]

I'm trying to run the hydraulic fracturing example in the basic example of the tutorial: https://geosx-geosx.readthedocs-hosted.com/en/latest/docs/sphinx/basicExamples/hydraulicFracturing/Example.html

It turns out that GEOS stops converging after about ~6 minutes out of a total of 20 minutes of modeling. I didn't change any parameters in the input files.

I tested with different nonlinear solver parameters but was not able to solve the problem... I am attaching the log file from running the original example for reference.

Thanks a lot! heterogeneousInSitu_benchmark_6_2_3.log

[frankfeifan]

Hi @zhishuaizhang. It looks like a linear solver problem to me. Maybe a quick check is to try the direct solver if your problem is not too large.

[zhishuaizhang]

Hi @frankfeifan Thank you very much for your suggestion! I changed the linear solver in Hydrofracture to a direct solver as shown below but was still not able to have the modeling finish running.

      <LinearSolverParameters
        solverType="direct"
        preconditionerType="none"/>

[paveltomin]

anyone is looking into that?

[paveltomin]

Two changes in the input seems to help to at least finish the case:

1. Set preconditionerType to mgr here:

     <LinearSolverParameters
       solverType="gmres"
       preconditionerType="mgr"/>

2. Set meanPermCoefficient to 0 here:

     <TwoPointFluxApproximation
       name="singlePhaseTPFA"
       meanPermCoefficient="0.0"/>