[CIG-SHORT] Maxwell bulk rheology behavior in v1.3.1 versus v1.4.2

Brad Aagaard baagaard at usgs.gov
Thu Apr 1 08:20:21 PDT 2010


Ikuo-

The test problem you provided had viscosities as low as 1e+17, which 
corresponds to Maxwell times of 0.14 years. This is consistent with the 
stable time step for MaxwellIsotropic3D of 0.028 years (0.2 times the 
Maxwell time of 0.14 years). Increasing the viscosity to 7.1e+19 results 
in the stresses relaxing over a much longer time scale.

Brad




Ikuo Cho wrote:
> Dear Brad
> 
>> Because the Maxwell time in your problem is so short, the stresses relax 
>> very quickly. In using a time step of 1.0*year with v1.3.1 the stresses 
>> are relaxing artificially slowly because the time step is so large.
> 
>> instead of the elastic constants. The elastic solution at the first time 
>> step (-dt to 0) should match the solution for a purely elastic material. 
>> In v1.4.2 this is true, but in v1.3.1 it is not. This is subtle in most 
> 
> Thank you for your comments. I ran PyLith1.3.1 with a short time step,
> dt=0.028 year, which is the value requested by PyLith 1.4.2. All other
> parameters are the same as those used in the previous calculation (I
> used the cfg file and the spatial database attached to the previous
> email I sent). The calculation results didn't show visible differences,
> however (See the figure). The only exception is the elastic solution at t=0. The elastic
> solution is influenced by some viscoelastic parameters when the time
> step is set to 0.028 year.
> 
> Although I don't fully understand your comments for PyLith1.3.1, I will
> concentrate myself to PyLith1.4.2 from now on because I want to apply
> PyLith 1.4.2 or later to some problems of power-law fluids.
> 
> You said "v1.4.2 looks correct", but it looks incorrect for me. The
> calculation results from v1.4.2 (Figures are attached to my previous
> email) showed that stresses fell two orders of magnitude everywhere in
> just one year. I gave viscosity coefficients ranging from 7.1*10**19 to
> 7.1*10**20 [Pa*s] (relaxation times from 100 to 1000 years) at the rear
> part of the volume (y > 0), and so the stress relaxation seems too short .
> 
> Perhaps I mistook the use of some important parameters for PyLith1.4.2.
> But it is also possible that some problem arises when PyLith1.4.2 reads
> three-dimensional data using a spatial database. I want to solve this
> problem anyway and will write another email soon for the illustration.
> 
> Ikuo
> 
> 
> On Tue, 30 Mar 2010 09:33:34 -0700
> Brad Aagaard <baagaard at usgs.gov> wrote:
> 
>> Ikuo-
>>
>> We found a couple issues related to the test problem you supplied.
>>
>> (1) The time step in your simulation using v1.3.1 is too large so the 
>> simulation isn't accurately modeling the viscoelastic deformation.
>>
>> PyLith v1.3.1 did not have a check to make sure the time step is 
>> sufficiently small relative to the shortest Maxwell time. This was added 
>> in v1.4, which is why if you try to run v1.4.2 with a time step of 
>> 1.0*year you will get an error message, but not in v1.3.1.
>>
>> Because the Maxwell time in your problem is so short, the stresses relax 
>> very quickly. In using a time step of 1.0*year with v1.3.1 the stresses 
>> are relaxing artificially slowly because the time step is so large.
>>
>> (2) The elastic solution in v1.3.1 for the Maxwell bulk rheology is 
>> incorrect.
>>
>> It appears that v1.3.1 is using the viscoelastic elastic constants 
>> instead of the elastic constants. The elastic solution at the first time 
>> step (-dt to 0) should match the solution for a purely elastic material. 
>> In v1.4.2 this is true, but in v1.3.1 it is not. This is subtle in most 
>> problems with large Maxwell times, but your test problem has such a 
>> short Maxwell time that this bug shows up. I don't know the precise 
>> location in the code where this bug occurs in v1.3.1, but v1.4.2 looks 
>> correct.
>>
>> Regards,
>> Brad
>>
>>
>> Ikuo Cho wrote:
>>> Dear PyLith developers
>>>
>>> I calculated linear viscoelastic responses of a medium, under the
>>> setting MaxwellIsotropic3D, while giving three-dimensional
>>> inhomogeneities to its viscosities using the SimpleDB spatial database.
>>> *info.vtk indicates that medium properties have been read correctly, but
>>> the way stresses attenuate apparently doesn't reflect the given
>>> inhomogeneities and absolute values of the viscosity coefficients. An
>>> earlier version (PyLith 1.3.1) produced intuitively expected results,
>>> but PyLith 1.4.2 gave counter intuitive outputs, no matter which one I
>>> used--binary for XP, binary for Linux or the source code to be compiled
>>> by the user (see attached figures). 
>>>
>>> You can validate what I am saying using the  setting files attached to
>>> this email: Here is what I did.
>>> 1) Place four files (dislocation_3d_mat1.4.2.cfg,
>>> dislocation_3d_mat1.4.2.cfg, 3d_mat1.3.1.spatialdb and
>>> 3d_mat1.4.2.spatialdb) in the exercise directory
>>> pylith-1.4.2/examples/3d/hex8.
>>>
>>> 2) Execute PyLith1.4.2 by typing:
>>> cd pylith-1.4.2/examples/3d/hex8
>>> pylith dislocation_3d_mat1.4.2.cfg
>>>
>>> 3) Execute PyLith1.3.1 by typing:
>>> mv pylithapp.cfg pylithapp.cfg.org
>>> pylith dislocation_3d_mat1.3.1.cfg
>>>
>>> The following is the only difference in "spatialdata" that describes material properties.
>>> <   value-units =  kg/m^3  m/s  m/s Pa^s // units
>>> ---
>>>>   value-units =  kg/m**3  m/s  m/s Pa*s // units
>>> I have no idea what the problem is. Did I miss any important changes in
>>> the way input should be given? 
>>> I'd be happy to receive any advise.
>>>
>>> ----------------------------------------------------------
>>> Ikuo Cho  ( ikuo-chou at aist.go.jp )
>>> Geological Survey of Japan,
>>> National Institute of Advanced Industrial Science and Technology
>>> Tsukuba Central 7, Tsukuba 305-8567 Japan
>>> Tel +81-29-861-3891, Fax +81-29-861-3682
>>> ----------------------------------------------------------
>>>
>>>
>>> ------------------------------------------------------------------------
>>>
>>> _______________________________________________
>>> 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
> 



More information about the CIG-SHORT mailing list