[CIG-SHORT] material model for landers/hector mine benchmark
baagaard at usgs.gov
Thu Jul 13 11:40:03 PDT 2006
Yes, it is possible to setup the codes to work with the material model
that Brad originally proposed. For PyLith-0.8 (and I imagine GeoFEST
as well), it would require some significant recoding of how material
properties are assigned. This is why I proposed a quick and dirty
approach with 10-12 layers that would come close to matching the
structure Brad proposed but not require any recoding.
On Thursday 13 July 2006 11:27 am, Eric Andreas Hetland wrote:
> > That is why I thought breaking it up into 10-12 layers
> > with only small jumps in elastic properties between layers
> > without remeshing would probably work okay.
> Forgive the interjection (that may be way off base), but can't it
> be possible to just figure out the depth of the center of each
> element, then "look-up" what the material properties for that
> element should be based on an interpolation of the material
> property model Brad H built? Of course, this would only work for
> depths with smooth gradients in the material property model. And as
> pointed out, property discontinuities should be preserved in the FE
> mesh by forcing element boundaries to fall onto the discontinuity
> in material properties.
> It does not seem necessary to re-generate the material property
> model nor the FE mesh (unless the discontinuities in the FE mesh
> and material model do not correspond). I am probably missing
> something, but it just does not seem that FE meshes should have to
> follow layers in a material property model, unless there are
> discontinuities in property across those layers.
> - Eric.
> CIG-SHORT mailing list
> CIG-SHORT at geodynamics.org
More information about the CIG-SHORT