[CIG-SHORT] pylith error in model with big time step and low viscosity
Charles Williams
willic3 at gmail.com
Sun Jun 12 16:07:21 PDT 2016
Brad is correct. Using a high viscosity will not change the size of the stable time step. If you really want to use a much larger time step, you can do as Brad suggested earlier — change the stability factor. To insure that you are getting an accurate solution, you could run for a shorter time period with a smaller time step, and compare the results at that time. That will at least make sure your solution to that point is accurate using the larger time step.
Cheers,
Charles
> On 11/06/2016, at 9:24 AM, Brad Aagaard <baagaard at usgs.gov> wrote:
>
> David,
>
> I am not sure what you mean by adding a second viscous layer with a high viscosity. The rate of the viscoelastic deformation is controlled by the material with the smallest viscosity, so as long as a low viscosity material is present, the appropriate time step to resolve the deformation will be small.
>
> Regards,
> Brad
>
>
> On 06/10/2016 02:18 PM, David Rodrigo Mora Cofré wrote:
>> Mr.Brad Aagaard:
>>
>> Thank you very much. I will check the manual. I think that adding a
>> second viscous layer (with high viscosity) will fix the problem.
>>
>> best regards,
>> David.
>>
>> 2016-06-10 12:28 GMT-03:00 Brad Aagaard <baagaard at usgs.gov
>> <mailto:baagaard at usgs.gov>>:
>>
>> David,
>>
>> The error message you are getting is associated with a built-in
>> check that compares the Maxwell relaxation time with the time step.
>> PyLith generates an error if the time step is greater than 1/5 of
>> the Maxwell time. This insures that the simulation will be able to
>> resolve the viscoelastic deformation with a reasonable accuracy.
>>
>> It is possible to circumvent this by using the TimeStepAdapt time
>> stepper with a stability factor much less than 1.0, but you should
>> be aware that your simulations will have significant errors in
>> resolving the viscoelastic relaxation.
>>
>> Regards,
>> Brad
>>
>>
>> On 06/10/2016 08:07 AM, David Rodrigo Mora Cofré wrote:
>>
>> Mr.Brad Aagaard:
>>
>> I'm testing a surface load model with 2 elastic layers + 1
>> viscoelastic
>> layer for study a postglacial rebound. When i use a viscosity
>> 1.0e21
>> Pas (dt=100*year), pylith computing my model very well. But,
>> when a use
>> a viscosity =<1.0e20 Pas (dt=100*year , total_time=10000*year),
>> pylith
>> not computing my model because my dt > stable time step. The stable
>> time step suggested by pylith is < 10*year. As this dt (<
>> 10*year) is
>> very small, pylith will take long time to computing the model.
>> How i can
>> force to pylith to computing the model with dt>=100*year and a low
>> viscosity?
>>
>> The following is the error message:
>> -------------------------------------------------------------------------------------------------
>>
>> File
>> "/Users/rodrigo/fem/pylith-2.1.1-darwin-10.6.8/lib/python2.7/site-packages/pylith/problems/TimeStepUniform.py",
>> line 87, in timeStep
>>
>> "step of %12.4e." % (self.dtN, dtStable))
>>
>> RuntimeError: Current nondimensionalized time step of 1.0000e+02
>> exceeds
>> the nondimensionalized stable time step of 9.2050e+00.
>>
>> application called MPI_Abort(MPI_COMM_WORLD, -1) - process 0
>>
>> /Users/rodrigo/fem/pylith-2.1.1-darwin-10.6.8/bin/nemesis:
>> mpirun: exit 255
>> ---------------------------------------------------------------------------------------------------
>>
>> best regards,
>>
>> PhD(c) David Mora
>> Doctorado en Ciencias De La Tierra
>> Universidad de Concepción
>> Chile
>>
>>
>>
>
> _______________________________________________
> 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
More information about the CIG-SHORT
mailing list