FEA - Modeling Aluminum Honeycomb Sandwich
Christopher Wright
<adk...@ix.netcom.com> wrote:
> Antone willing to offer some help regarding fea modeling of
> aluminum honeycomb?
> ... how should I define the material properties?
You need a laminar element to model a sandwich material. The stiffness
isn't uniform, so a single elastic modulus won't tell you enough. You
might be able to fake it, using equivalent values of elastic and shear
moduli and thickness, so you get the right strain distribution, but it's
more complicated than just doing it right the first time.
--
Christopher Wright P.E. |"They couldn't hit an elephant from
chr...@skypoint.com | this distance" (last words of Gen.
Voice phone (612)933-6182 | John Sedgwick, Spotsylvania 1864)
James Conrad Pope Smith
: Antone willing to offer some help regarding fea modeling of
: aluminum honeycomb?
: I have a simple 2" sandwich with thin aluminum facers and some
: standard commercial-grade aluminum honeycomb.
: If I model the core with 8-noded bricks, how should I define
: the material properties?
Why use bricks? You can do this with plate elements. I assume that
the honeycomb is used only for bending strength (by moving the face
sheets away from the neutral axis). You can compute the A, B, and
D matrices by hand for this. In this case the B matrix will be zero.
The A matrix relates line loads to strains: N=A*strain. The D matrix
relates moments to curvatures: M=D*curveature. These A and D matrices
are used for in-plane and plate bending solutions. These matrices are
traditionally used in composites, where layer orientation must be
accounted for, but they are equally suitable for stiffened panels.
Check out any composites book to see how to form the matrices.
James
--
----------------------------------------------------------------------
James P. Smith
NASA/JSC, Mail Code ES
Houston, Texas 77058
smi...@smithj.jsc.nasa.gov
smi...@smd4.jsc.nasa.gov
Thomas Reiter
>Antone willing to offer some help regarding fea modeling of
>aluminum honeycomb?
>I have a simple 2" sandwich with thin aluminum facers and some
>standard commercial-grade aluminum honeycomb.
>If I model the core with 8-noded bricks, how should I define
>the material properties?
>Thanks for any help from out there!
>-Brian M. Adkins
As pointed out in other postings you don't have to use 3D-Solids FE
to model a sandwich structure - mostly a layered plate or shell
element will be more effective. Anyway, in both cases you have to
find the effective material parameters of the core-layer, which happen
to be orthotropic (or transversly isotropic) This is usually done by
smearing out (or homogenizing) of a suitable unit-cell of the honeycomb.
Such procederes and results should be given in any "good" book dealing
with sandwich and honeycomb structures. Try e.g. Gibson, Ashby: Cellular
Solids (can't remember the publisher) or (if you are familiar with
German) Wiedemann: Leichtbau I, II Springer Verlag.
Hope that helps
Thomas
--
Thomas J. Reiter e-Mail: rei...@ilfb03.tuwien.ac.at
Institute of Light Weight Structures (E317) Tel.: +43 1 58801 3715
Vienna University of Technology \ Austria Fax: +43 1 505 44 68
A-1040 Wien, Gusshausstrasse 27-29
Brian M. Adkins
Bob Brockman
Brian,
I'm not out to flame anyone, but a certain amount of the advice
you've been offered is crap. There is more than one way to approach
modeling sandwich with a flexible core; which method you use depends
on your objectives in performing the analysis, the software you have
at your disposal, and how much you know about material properties,
among other factors.
Method 1. Model the constituents (core, face sheets) separately.
This approach is often appropriate when you're just modeling the
sandwich component (more detail), or when you need precise control
over exactly how constraints are applied. There may be close-outs
or reinforcements present which wouldn't be easy to represent in a
plate model. The usual approach here would be to use solid elements
for the core, and plates (with appropriate offsets as necessary) for
the face sheets. If the face sheets are extremely thin and act
essentially as membranes, the offsets may not be important. For the
core properties, I'd suggest MIL-HDBK-23A, Structural Sandwich Com-
posites, which has effective properties for many core configurations
with different ribbon thicknesses, cell sizes, etc. For most
aluminum-core sandwich, the core properties will be orthotropic. If
you're trying to look at core crushing, face sheet wrinkling, or
other highly localized effects, even this level of detail may be too
crude.
Method 2. Model the whole sandwich with layered plate or shell
elements. This makes for a simpler model, but there are dangers
lurking which can give you really disappointing results. DO NOT
just compute A, B, and D matrices using the moduli and thicknesses
of each of the layers, and assign these to plate elements. The
laminate stiffness approach is based upon the assumption of linear
variations of the displacement through the shell thickness, which
for most sandwich laminates is not even remotely realistic. Using
plate elements for sandwich panels requires one of two devices
(assuming you want results which are within an order of magnitude or
so of being right):
(a) specify an appropriate shear flexibility correction
(this is that factor of 0.83333333 that we always use for
isotropic plates). The value 5/6 is an upper-bound, and the
shear correction factor appropriate for most sandwich is at
least an order of magnitude less. These can be computed but
there are relatively few tools around for doing so.
(b) compute effective stiffnesses by one of the special-
purpose methods cooked up over the years by people who do
sandwich panel analysis. As a reference for this kind of
procedure, I'd recommend the books on sandwich panel analysis
by Plantema or Allen.
I realize this isn't the simple answer you're hoping for, but it
might get you started in the right direction.
Good luck,
Bob Brockman (broc...@udayton.edu)
Marc M. DeBower
-*snip*-
>If I model the core with 8-noded bricks, how should I define
>the material properties?
I think that I would use plate elements, at least for the first
few runs, to find hotspots and such. Then when you get a good model,
change to bricks if you need to.
-mmd