[CIG-LONG] Fwd: CIG-LONG Digest, Vol 47, Issue 2

[Walter Landry]

"Julian Giordani (Sci)" <julian.giordani at monash.edu> wrote:
> On 21 October 2010 05:59, Walter Landry <walter at geodynamics.org> wrote:
> 
>> Laetitia Le Pourhiet <laetitia.le_pourhiet at upmc.fr> wrote:
>> > Hi all,
>> >
>> > I was intrigued by all this problems with thermal diffusion...
>> >
>> > I don't understand why this limitation for the time step is not
>> > included in gale, I guess it is historical since diffusion is
>> > negligible for convection problem and gale is based on underworld but
>> > as Charmaine showed, it is not negligible for post-rift simulation.
>>
>> It turns out that there is some machinery in AdvectionDiffusionSLE to
>> use a maximum diffusivity to compute a step size.  However, the
>> maximum diffusivity is hard coded and can not be set from an input
>> file.  I do not know why that is.
>>
> 
> just on this note and assuming the GALE implementation is the same as
> Underworld...
> 
> Maximum diffusivity is initially hard-coded in AdvectionDiffusionSLE.c -
> some legacy condition I suspect.
> Then the value of 'self->maxDiffusivity' is dynamically calculated using the
> all material diffusivity values or the value of 'defaultDiffusivity', this
> is done in the file UpwindParameter.c
> 
> the diffusive timestep = self->courantFactor * minSeparation * minSeparation
> / self->maxDiffusivity
> 
> so max diffusivity should be a reasonable value.

To be exact, this is the parameter defaultDiffusivity in the
AdvDiffResidualForceTerm component, usually named
defaultResidualForceTerm.  If that is set, then it seems to take
reasonable time steps.

Cheers,
Walter Landry
walter at geodynamics.org