[CIG-SHORT] Time step

[Brad Aagaard]

Christian,

For linear viscoelastic models in quasi-static simulations, we define 
the stable time step to be 0.2 times the Maxwell time (viscosity / mu). 
This should result in sufficient resolution of the relaxation process. 
For your relatively low viscosity and larger shear modulus, the 
relaxation time is very short (less than 0.3 years).

We seem to be missing a discussion of the stable time step for 
quasi-static simulations in the PyLith manual. We will add in this 
section for the next release.

Regards,
Brad


On 11/11/13 8:47 AM, Christian Baillard wrote:
> Hi Brad,
>
> Here are my .spatialdb for the lithosphere and the asthenopshere. I use a viscosity of 5.0e+17 for the visco-elastic asthenosphere. For the elastic part I use a density of about 3300 kg/m3  and Vp / Vs are taken from the IASP model.
>
>
>
>
>
>
>
>
> 'E' stands for East, The two lithospheres(/astheno) on both side of the trench have the same parameters.
> I run a quasi static simulation, convergence rate is 3.5 cm/yr
> Here is the parameter file for simulation
>
>
>
>
>
> And my mesh looks like that:
>
>
>
>
>
>
> Regards,
>
> Christian
>
> Le 11 nov. 2013 à 17:34, Brad Aagaard a écrit :
>
>> On 11/11/13 4:23 AM, Christian Baillard wrote:
>>> Hi developers,
>>>
>>> I do not want to stuff up the mailing list with all of my problems so that's why I'm sending this message to you personally.
>>> When running pylith I have a problem with the time step:
>>
>> Please keep sending to cig-short at geodynamics.org. As Matt said, all users benefit from hearing the questions and responses.
>>
>>> File "/Users/baillard/pylith/pylith-1.9.0-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   5.0000e+00 exceeds the nondimensionalized stable time step of   4.6202e-02
>>>
>>> I use the same dt (5 yr)  and total_time (100 yr) as in the example given in the 2d/subduction problem.
>>> I really appreciate the support you are giving, it helps me a lot.
>>
>> We need more info to understand what is causing PyLith to calculate a stable time step much smaller than your time step. Are you running a quasi-static or dynamic simulation? What bulk rheology and corresponding parameters are you using?
>>
>> Regards,
>> Brad
>>
>