[aspect-devel] Tracking time-dependent properties through compositional fields (general discussion)
Wolfgang Bangerth
bangerth at tamu.edu
Fri May 27 16:39:52 PDT 2016
On 05/26/2016 12:39 PM, John Naliboff wrote:
>
> 1. In finite_strain.cc, each component of the strain tensor is assigned to
> and advected with a unique compositional field via the reaction term. If one
> wishes to track additional properties, say CPO and melt, this will require
> additional unique compositional fields for each component. At some point,
> will the advantages of using compositional fields over tracers be negated if
> too many compositional fields are present?
Yes, that is likely going to be the case. I don't know where that point is,
but I suspect it will be the case with *a lot* of compositional fields.
Especially in 3d, where one needs lots of particles per cell.
As a guide, recall that solving each advection equation costs only a few per
cent at most of the cost of solving the Stokes equations, in particular if you
have a model with highly variable viscosity. My best guess would be that you'd
need at the very least 10 compositional fields (maybe 20 or 30) for the same
cost of the Stokes equations. I don't recall how much it costs to track
particles, but would expect it to be on the same order of magnitude as the
Stokes equations. Rene will be able to say more about this.
(Of course it depends on how many particles you have, which depends on what
you want to do with them: for active particles, you'd need ~100 per cell in
3d; for tracing origin, a few will be plenty.)
> 2. In finite_strain.prm the compositional fields for each component (4 total)
> of the 2-D strain tensor are assigned a value of 0 under the initial
> conditions section. My interpretation is that this procedure allows the
> fields to track the strain through the reaction term, but the "volume
> fraction” of these fields in each cell is still effectively 0. Is this
> interpretation correct? If it is correct, that means composition dependent
> material models (ex: multicomponent.c) following this approach will not be
> affected by the compositional fields tracking strain unless they specifically
> use the reaction terms.
I think you already found your answer, but for my own education: what is the
"volume fraction" you refer to?
Best
W.
--
------------------------------------------------------------------------
Wolfgang Bangerth email: bangerth at math.tamu.edu
www: http://www.math.tamu.edu/~bangerth/
More information about the Aspect-devel
mailing list