XANSYS Message: 39 [Go back to message list]
[bookmark on del.icio.us]
No rating yet
Subject: Re: Hex & Tet elements
Author: Chandresh Shah
Date: 1998-12-15 18:20:00I think the trick is to use a tet mesh with a "sufficient" mesh density so
that it captures your DOF accurately (temperature or displacements). Then,
a submodel can be built in the area of high stress (or temperature). This
submodel can be brick meshed (or alternatively tet meshed with a fine
density). This can result in substantial time savings in the meshing of
the global model which is often very complex (at least the engine
components I analyze are!), without any sacrifice in accuracy. I have been
using this technique (as I am sure several others are) for many years with
success. I am a strong proponent of automatic 10 noded tet meshing
followed by submodeling for accurate results with substantial time
savings.
Chandresh Shah
Principal Engineer,
Department of Engine Design,
Engine & Vehicle Research Division,
Southwest Research Institute,
Phone : 210-522-2628
FAX : 210-522-4673
E-mail : cshah@s...
On Tue, 15 Dec 1998, Martin Liddle wrote:
> From: Martin Liddle
> In article <7D70697C0E38D111B4FF080036B39A03AAE5E1@n...>,
> John Swanson writes
> >From: John Swanson
> > With mid-side node elements (10 node tet, 20 node bricks), the
> >results should be the same (very close) for the same nodal density.
> >The tet mesh will have more nodes because of the nodes in the
> >center of the faces and in the center of the brick, and therefore will
> >be somewhat more accurate.
> > The advantage of fully automatic mesh generation offered by the
> >tet elements normally outweighs the additional time the solution
> >might take (computers are cheaper than people, and time is always
> >critical).
> >John.Swanson
> I agree with what John says as far as it goes. However in practice the
> difficulty with a tet mesh is in controlling the mesh refinement. I do
> a lot of thermal stress analysis with steep temperature gradients near
> the surface and it can be difficult to get a tet mesh that is
> sufficiently refined in the through thickness direction. It can be much
> more difficult to mesh a given geometry with bricks but much easier to
> control the mesh refinement.
> Martin Liddle, Tynemouth Computer Services, 27 Garforth Close,
> Cramlington, Northumberland, England, NE23 6EW.
> Phone and fax: 01670-712624.