[CIG-MC] Removing Tracer Advection

Tue Jul 2 20:54:27 PDT 2013

Just to clarify, the tracers behave like Eric said, but they become
"active" whenever an actual chemical buoyancy number is assigned to them,
as per the common usage of these terms. Moreover, they can be assigned a
different viscosity.

If you want a static lateral viscosity variation field, the way to go is
indeed to modify the viscosity routines so that the pre-factor is based on
some geometric formulae. Alternative, the ggrd extension can read in
GMT/Netcdf grid files of prefactor fields, though this is not well
documented.

Cheers

T

Thorsten W Becker - http://geodynamics.usc.edu/~becker

On Tue, Jul 2, 2013 at 7:15 PM, Scott Tarlow <scott_tarlow at yahoo.com> wrote:

> Hey Eric,
>
> Thanks for the quick response. I think you are right, in general it would
> be better to hardcode in a low viscosity region. If I were to create this
> region, I do have a few questions:
>
> In the low_viscosity_channel_factor function it finds the indices for the
> minimum and maximum radius:
>
> for(m=1; m<=E->sphere.caps_per_proc; m++) {
>         /* find index of radius corresponding to lv_min_radius */
>         for(e=1; e<=E->lmesh.elz; e++) {
>             rad_mean = 0.5 * (E->sx[m][3][E->ien[m][e].node[1]] +
>                               E->sx[m][3][E->ien[m][e].node[8]]);
>             if(rad_mean >= E->viscosity.lv_min_radius) break;
>         }
>         nz_min[m] = e;
>
> To modify this for the x and y directions can i just change lmesh.elz to lmesh.elx
> along with changing the looping parameter and if statement parameters? Does
> sx[m][3] refer to the radial direction, meaning sx[m][2] would be y and
> sx[m][1] would be x?
>
> How I think the best way to implement it is by creating a new static void,
> finding the start and ending indices for the x,y, and z direction, and then
> looping over that volume and applying the viscosity reduction factor. Doing
> this, where would the best place be to call the function in the
> Viscosity_structures.c file?
>
> Thank you for your time,
>
> Scott Tarlow
>
>
> On Jul 2, 2013, at 6:15 PM, Eric Heien <emheien at ucdavis.edu> wrote:
>
> Hi Scott,
>
> Keep in mind the tracers in themselves do not impart any properties on the
> simulation (they are passive).  Even if you stop them from moving, you have
> to couple their existence (or flavor) to whatever viscosity property you
> want.  Since this change in viscosity is just placed on certain nodes, it
> might be simpler to define a specified viscosity on whatever nodes you want
> to be a weak zone.  This way there isn't any confusion about how the
> tracers are interacting with the simulation and it's clear what the size
> and properties of the weak zone are.
>
> I don't know the details of what you're doing so your current solution
> might work fine, I just want to make sure there's no confusion about it.
>  In any case let us (or the mailing list) know if you have any more
> questions.  Good luck,
>
> -Eric
>
> On Jul 2, 2013, at 12:22 PM, scott tarlow wrote:
>
> I'd be happy to.
>
> One of the necessary requirements for the project I'm working on is that
> the overriding and subducting plate are decoupled. I think there are a few
> ways of doing this: One is to write some sort of extension to the
> Viscosity_structures.C file and change a specific volume of nodes to a much
> reduced viscosity, ie "weak" nodes. The other solution is using the tracers
> in conjunction with a Low Viscosity Channel. This worked great, and seems
> much easier to use with the exception that because the tracers move, the
> Low Viscosity Channel also moves. This causes the slab to move
> horizontally, I think. If I am able to stop the tracers from moving, I am
> essentially reproducing "weak" nodes without significant coding changes to
> the file.
>
> Thank you for your time,
>
> Scott Tarlow
>
>
> From: Louise Kellogg <kellogg at ucdavis.edu>
> To: scott tarlow <scott_tarlow at yahoo.com>
> Cc: Eric Heien <emheien at ucdavis.edu>
> Sent: Tuesday, July 2, 2013 12:24 PM
> Subject: Re: [CIG-MC] Removing Tracer Advection
>
> Hi Scott,
>
> I'm curious what scientific problem you're working on that would need the
> tracers to stay in one place. The tracers are passive and massless, so
> locking them in place will not influence the flow - but it also defeats the
> usual purpose of the tracers.
>
> Can you tell us a bit more about what you're working on?
>
> Best wishes,
>
> Louise
>
>
> On Jul 1, 2013, at 2:01 PM, scott tarlow <scott_tarlow at yahoo.com> wrote:
>
> Hello,
>
> I am looking to modify the of the files : Tracer_setup.c and
> Regional_tracer_advection.c (in CitcomS) in order to force the tracers to
> stay at their original positions instead of moving with the flow field.
>
> My original approach was to find in the code where where the nodal
> velocities are pulled and then assigned to a tracer and set them to 0
> instead of Vx,Vy,Vz respectably. This did not work, so my next solution was
> deleting Regional_tracer_advection.c from the lib file. This hasn't worked
> either.
>
> If Regional_tracer_advection.c is on my desktop instead of the lib folder,
> CitcomS wont read it correct?
>
> I'm not really sure where to go from here. Perhaps someone has a
> suggestion ?
>
>
> Thank you for your time,
>
> Scott Tarlow
> _______________________________________________
> CIG-MC mailing list
> CIG-MC at geodynamics.org
> http://geodynamics.org/cgi-bin/mailman/listinfo/cig-mc
>
>
> *********************************************
> Louise Kellogg
> Professor, Department of Geology
> One Shields Avenue, University of California, Davis, CA 95616
> Phone: (530)752-3690
> http://mygeologypage.ucdavis.edu/kellogg/
> http://www.keckcaves.org
> *********************************************
>
>
>
>
>
>
>
>
> _______________________________________________
> CIG-MC mailing list
> CIG-MC at geodynamics.org
> http://geodynamics.org/cgi-bin/mailman/listinfo/cig-mc
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://geodynamics.org/pipermail/cig-mc/attachments/20130702/6d7b413d/attachment-0001.html>