Ply by Ply Stress Validation with NEI Nastran

[LeitenbergerC]

Because of a recent discrepancy in the Strength Ratio's reported in NEI Nastran and the margins found in Hypersizer, I was trying to duplicate the Ply by Ply stresses in NEI Nastran and HyperSizer using the methods found in Jones Second Edition Mechanics of Composite Materials (Sections 2.5, 2.6, and 4.2)

The ply by ply stresses reported by NEI Nastran and hpyersizer don't match, the most troubling problem is that the Shear in hypersizer is about 2x higher than that in Nastran. (Could this be the difference between Engineering Strain and Tensor Strain?) Secondly the 1 and 2 direction stresses are the similar, but the sign and directions don't match.

Ply 1    NEI Nastran  Hypersizer  Hand Clacs
Sigma 1  11.26 ksi      2.62 ksi   2.18 ksi
Sigma 2   2.04 ksi    -10.01 ksi  11.56 ksi
Tau 12   -2.30 ksi     -5.12 ksi  -5.50 ksi

It is likely that I made a sign error, and I will be looking for it. The slight variability in magnitude is understandable, but the discontinuity between NEI Nastran output and the Hypersizer output in load direction, sign and shear magnitude is concerning. Any ideas on how to make them match?

[August]

We need to know a few more things about your comparative analyses before we can have an idea of sources of your discrepancies:

-What are your loads and BCs?
-Which Nastran solution are you using?
-What type of plate theory are you using for your hand calcs?

HyperSizer uses Kirchhoff-Love plate theory as the basis of ply-by-ply strain and stress calculations:

http://hypersizer.com/help_7.0/#StressAnalysis/sa-smeared_stiffness.php

The first step to resolving the discrepancies is to figure out if there are different assumptions in your three comparative methods.

-August

[LeitenbergerC]

The element in question is far away from any load application points or boundary conditions. (Loads are point loads applied though RBE 2's that represent load application fittings. Boundary conditions are similarly applied through RBE2's that represent attachment fittings.)

It is a static run (SOL 101 I think)

My understanding is that the Classical Lamination theory that Jones (Which I referenced) uses is an extension of Kirchhoff. (I don't know how Kirchhoff-Love is different, or if it is the same thing.)

This element is a Honey Comb panel, and the in plane stiffness of the core is very very low, so the use of smeared stiffness shouldn't introduce much error.

[James]

Chris,

I assume you are extracting forces from the FEM elements and typing them into a HyperSizer Non-FEA project and into a spreadsheet to run your verification, is that correct? If so you should be aware of the element direction vs. material directions. By default, Nastran will export the Nx, Ny and Nxy forces in element coordinates. But HyperSizer uses material corrdinates to define the element Nx direction. This could be causing the sigma1/sigma2 dependency between HyperSizer and Nastran.

We would really need to study your verification case to find potential dependencies and provide better feedback. Could you send your FE model and HyperSizer database? Or provide a more detailed PPT showing the element loads, material properties and laminate. Then we could try to repeat the issue here.

-James