[aspect-devel] Is the correct Boussinesq approximation now the default in ASPECT?
sdk at vt.edu
Sat Feb 4 06:02:28 PST 2017
For what it is worth. We repeated the 2D cylindrical cases described in the manual as from Davies et al benchmark using the prm files in the distribution. We should probably try with the bousinessq formulation too, but I suspect it's been tested. We get the results quoted in the table in the manual with the prm in the distribution. We still can not reproduce the Zhong et al 3D results. With the new Bousinesq formulation (below)
set Formulation = boussinesq approximation # default: custom
we actually get a worse result compared with the Zhong et al results than we did when setting alpha to be a small number and gravity to be Ra*(1/alpha) with custom. Trick used in Davies et al and Blankenbach et al results.
If someone has made a change since early/mid January, we have not picked that up.
On Feb 3, 2017, at 6:53 PM, Elbridge Gerry Puckett <egpuckett at ucdavis.edu> wrote:
> Timo, Wolfgang Juliane, Rene, Max Rudolf and CIG researchers know what I am referring to.
> My understanding is that Timo and Juliane have fixed the problem with the original - incorrect - Boussinesq approximation.
> How does one make certain that we are approximating solutions of the correct incompressible Mantle Convection Equations when running the current ASPECT master branch.
> Also John Naliboff suggested that I also ask if the defaults are still no internal heating, adiabatic heating, etc. etc. In other words, the only source of heat is the boundary conditions on the top and bottom walls (in 2D) assuming no heat flux BCs on the side walls.
> The genesis of this particular post is that we are getting startling different results with the same parameter file and presumably, the same code for a problem we're currently working on.
> - Gerry
> Aspect-devel mailing list
> Aspect-devel at geodynamics.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Aspect-devel