[aspect-devel] A question for setting the depth-dependent viscosity (Nan Zhang)

[Max Rudolph]

Nan,
I am going to CC the ASPECT list on this in case others are interested.

On Tue, Jan 12, 2016 at 6:17 PM, Nan Zhang <Nan.Zhang at colorado.edu> wrote:

> Hi Max,
>
>
> >>* The "depth dependent" material model gets depth directly, so there's no
> *>* need to calculate the radius in this manner:
> *>>*   subsection Depth dependent model
> *>*     set Base model = simple
> *>*     set Depth dependence method = Function
> *>*     subsection Viscosity depth function
> *>*       set Variable names = d
> *>*       set Function expression = if(d>6.70e5,3.0e22,1.0e21);
> *>*     end
> *>*   end
> *>Before I try this, I have several detail questions. when we set "Base model
> = simple" under "Depth dependent model", could we use the "set Density
> differential for composition field 1 = 0.2" to set up the composition field
> in the "Depth dependent model"?
>
>
>  I looked at the code for both the Simple and Depth dependent material
models and see no reason why, in principle, the composition-dependent
density would not work as you suggest. The depth dependent material model
just calls 'evaluate' from the base model. For troubleshooting, you can
always define a constant depth dependence (just specify 1.0 as the
function). Only the viscosity is modified (line 139 in depth_dependent.cc).

> >>>* If you prefer to use a file that specifies the depth-dependent prefactor,
> *>* take a look at these files:
> *>>* https://github.com/geodynamics/aspect/blob/master/tests/depth_dependent_box_file_simple.prm <https://github.com/geodynamics/aspect/blob/master/tests/depth_dependent_box_file_simple.prm>
> *>* https://github.com/geodynamics/aspect/blob/master/tests/visc_depth.txt <https://github.com/geodynamics/aspect/blob/master/tests/visc_depth.txt>
> *
> I read the example, while wonder if this 2D box also works for the
> spherical shell? I guess it is.
>
>
Yes, it gets depth from the geometry model, so it should work in 2D or 3D,
box, annulus, or spherical shell.

>
> >>>* If you read the prefactor from a file, the prefactor is interpolated
> *>* (linearly), whereas if you provide the prefactor using a list of depths and
> *>* prefactors, the prefactor is piecewise constant and each depth provided
> *>* specifies the bottom of a layer. In retrospect, we could have added an
> *>* option to specify the interpolation method for each case.
> *>* -------------- next part --------------
> *>I have a general question. I am trying to repeat our CitcomS model (Zhang
> et al., 2010, JGR; Zhong & Rudolph, 2015, G^3). Have you already try them
> on ASPECT?
>
>
I have not tried to use ASPECT to exactly repeat the CitcomS calculations
that you and I have done previously. Rene may have run some calculations
with time-dependent plate motions. When I last looked at doing this, it was
my impression that getting your plate motion model for 450-120 Ma into the
gPlates format was going to be quite time consuming. I see now that there
is an 'Ascii data model' for velocity boundary conditions, so maybe it
would be straightforward to do this. I will be interested to hear about
your results, especially what refinement options you choose and how long
these calculations take to run.


> Cheers,
> Nan
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.geodynamics.org/pipermail/aspect-devel/attachments/20160112/1970c1fe/attachment.html>