Principal stresses at surface nodes

[fabio84]

Dear all, I am a new user of forum. In my job I use Nastran/Patran. I noticed that FEM analysis leads to incorrect values ​​of the principal stresses at nodes of a surface. For example, suppose you have a meshed volume consists of HEXA elements (8 nodes). Well, Nastran calculates value of stress at the centroid and at the corners of the element. In Patran I want to see the values of the principal stresses at nodes, but on surface it gives me the output values ​​of all 3 principal stresses (this is against the classical theory: on the surface of a solid there are only 2 values ​​of principal stresses if on surface there are no forces). My question is: how do I calculate the exact values ​​of principal stresses at a node on the surface? I use Nastran/Patran but the problem exists in any solver.
Thank you all

 

[ESPcomposites]

What is the magnitude of the component that is theoretically zero? It may be relatively small compared to the other components (effectively zero for engineering purposes). You will probably never get a true "0.0" because you the FEM always has some amount of discretization.

Brian

http://www.espcomposites.com

 

[flash3780]

The FE solver is solving for stresses at the element integration points, and then linearly interpolating out to the nodes. So, you would expect some induced error when reading nodal values - the stress field likely isn't linear. If you have a reasonably refined mesh, these errors should be relatively small.

When it comes to FE modeling, nodal values (nodal forces, displacements) are always more accurate than derived values (strains, stresses). You're essentially solving a system of equations based on Hook's Law: -{f} = [k]{d}.

What's the amplitude of your phantom stress? If your model is, in fact, correct then it should be small when compared to your other principal stresses and should reduce when you refine your mesh or use higher ordered elements.

//signed//
Christopher K. Hubley
Mechanical Engineer
Sunpower Incorporated
Athens, Ohio
--

http://engineeringliberty.wordpress.com

 

[kellnerp]

Post processors I have seen don't handle surface stresses when using 3D elements. If you are using Patran it would seem you could calculate what you need and plot that.

TOP
CSWP, BSSE

http://www.engtran.com

http://www.niswug.org

http://www.linkedin.com/in/engineeringtransport

Phenom IIx6 1100T = 8GB = FX1400 = XP64SP2 = SW2009SP3
"Node news is good news."