Finite
Element
Modeling
of the Buckling
Cheryl
NASA
Research
Hampton,
VA
David
Hampton,
for
sandwich
predicting
described.
The
anisotropic
VA 20151
Charles
C. Rankin 4
from
conventional
panel
overall
modes
are
panel
compared
with
different
finite
element
element
solutions
element
models,
flexibility,
elements
layered shell/solid
for the face sheets
with
of
the
approaches
models
element.
shell/solid
and
with
respect
thickness
discretization,
the
indicate
study
provides
an accurate
that
solutions
sandwich
obtained
using
without
specialty
effective
and these
effects
may
analytical
solutions.
shear
aeronautical
Results
increased
increased
efficient
along
with
the
buckling
affect
not
be represented
Modeling
performance
and
aerospace
demand
for
aeronautical
concepts
concept
modeling
modeling
moduli,
panel
indicate
the
that
ratio
of
response
accurately
by
recommendations
are
Introduction
blends
of
the
passenger
type
wings
Due
pressurized
approach
air
of
and
travel,
suggest
aerospace
future
the projected
that
flight
more
vehicle
are needed.
An example
of a revolutionary
for an efficient,
large
transport
aircraft
is a
surface.
element
requirements
vehicles,
and
blended-wing-body
through-the-
sandwich
sheets,
elastic
The
element models,
with shell
and solid elements
for the
demonstrated.
face
principle
shell
using a recently
developed
Convergence
characteristics
are
of the
sandwich
results
using
transverse
and
and localized
Furthermore,
Finite
layered
element
overall
also provided.
type
approaches.
to in-plane
and
for
solutions
sandwich
the
with
Results
face-sheet-wrinkling
obtained
and
core.
both
response.
anisotropy
is
panels
thick
modeling
are
response
sandwich
analytical
and
Center
94304
buckling
approaches
buckling
and a very
buckling
modeling
considers
sheets
Technology
CA
for predicting
of different
study
face
Jr. 3
Division
Chantilly,
Advanced
Company
23681
F. Knight,
Abstract
study
and Sciences
Systems
Palo Alto,
A comparative
Center
23681
VA
Norman
Lockheed-Martin
Panels
F. Moore 2
Engineering
Veridian
of Sandwich
A. Rose 1
Langley
Lockheed-Martin
Response
and
to the
centerbody
area
(BWB)
fuselage
shape
region,
of
into
of the
which
a single
lifting
BWB
which
area,
aircraft,
airplane,
includes
is non-circular.
both
the
the
The
1
American
Institute
of Aeronautics
and Astronautics
non-circular
centerbody
regionis challenging
fromthe strains.
They
present
linear
stress
solutions
for
standpoint
of structural
designsincethebasiccover thermally loaded, simply supported sandwich plates
panelstructure
carries
bothinternal
transverse
pressurewith laminated face sheets using an analytic, closedloadandnormalwingbending
andtorsionloads.In form solution.
Vonanch
and
Rammerstorfer
[9]
order to satisfy the performanceand weight presented
a Rayleigh-Ritz
solution
for face-sheet
requirements
fortheBWBaircraft,
andotheradvancedwrinkling of general unsymmetric sandwich panels with
concepts
thataresubjected
to bendingor pressureorthotropic face sheets. Comparison of their analytical
finite element
loadings,
advanced
sandwich-type
constructions
with results with unit-cell three-dimensional
composite
material
facesheets
andrelatively
thickcores results is used to verify their approach.
(seeFigure1) offera potential
design
advantage
over
conventional
metallicmaterialsandstiffenedskin Bert [10] summarized different theories for sandwich
construction
[1].
plates with laminated
composite
face sheets that account
for both
transverse
shear
and transverse
normal
effects.
Mostsandwich
structures
aredefined
usingathree-layerNoor et al. [11] presented an exhaustive reference list
typeof construction,
asillustrated
in Figure1. The (over 1300 citations) of analytical and computational
outerlayers
arethin,stift high-strength
material;
while proceduJces for sandwich structures. Librescu and Hause
themiddlelayerisa thick,weak,low-density
material.[12] presented
a further
survey
and extended
the
Initialanalytical
workonsandwich
structures
treated
the formulation
to include
buckling
and
postbuckling
three-dimensional
sandwich
structureas a pair of response of flat and curved sandwich
structures
membrane
facesheets
heldapartbyacorematerial
with subjected to mechanical and thermal loads. A Rayleigharelatively
largetransverse
shear
stiffness.
ThebendingRitz procedure for simply supported sandwich plates
stiffness
of thefacesheets
is ignored
andthecoreis was developed by Rao [13] where the bending stiffness
assumed
tobeinextensional
inthetransverse
direction of the face sheets was ignored. Kim and Hong [14]
andhasnegligible
stiffness
in thein-plane
directions.extended Rao's work to account for the face-sheet
Thistypeofmodel,
calleda sandwich
ofthefirstkind, bending stiffness. Hadi and Matthews [15] presented a
hasbeenappliedsuccessfully
in manyapplications.Rayleigh-Ritz
procedure
based on a zigzag
theory
and
in the face
sheets.
However,
a morerobustformulation
withadditionalaccounts for shear deformation
fidelityisneeded
tomodelcomplex
nonlinear
structuralComparisons
with other Rayleigh-Ritz
solutions
were
behavior
includinglocalfailuressuchasface-sheetpresented for sandwich panels with thin face sheets.
buckling
andface-sheet
disbond.A sandwich
of the Results are reported for different face-sheet stacking
second
kindaccounts
fortheout-of-plane
response
of sequences. Dawe and Yuan [16, 17] presented a finite
thefacesheets
andthefullthree-dimensional
behaviorstrip formulation using B-splines for sandwich panels
ofthecorematerial.
with anisotropic
face sheets.
However,
no comparative
numerical
studies
of different
finite
element
modeling
Earlyanalysis
workformetallicsandwich
structuresstrategies
for predicting
the buckling
response
of
includesPlantema[2], anddesignguidelinesfor sandwich panels have been identified in the literature.
sandwich
structures
aregivenin Ref.[3]. General
instabilityandface-sheet
wrinklingaredescribed
by The present paper describes the basic buckling behavior
Benson
andMayers
[4]. Oneaspect
oftheirworkwas and response of sandwich panels loaded in axial
related
to defininga stabilityboundary
between
face- compression
and compares
buckling
predictions
for
sheetwrinklingandgeneral
instability.Researchers
various
levels
of finite
element
modeling
fidelity.
havealsostudied
theanalysis
of sandwich
structuresBuckling results obtained from approximate analytical
withemphasis
ontheuseof composite
materialface expressions
are also compared
with buckling
results
sheets
andfoamcores.Finiteelement
formulations
for obtained from the finite element analyses. Different
sandwichpanelsare reviewedby Ha [5, 6]. finite element models of the sandwich panel are
Displacement-based
formulationsand hybrid considered including layered shell models, specialty
formulations
areconsidered
aswellasdifferent
through-sandwich element models, and layered shell/solid
the-thickness
kinematic
models.Nonumerical
studiesmodels. Numerical results obtained using the STAGS
arepresented.
Frostig[7] investigated
sandwich
panel (STructural Analysis of General Shells) nonlinear finite
buckling
usingahigher-order
theorywhichaccounts
for element code [18] for the three different finite element
different
boundary
conditions
ontheupperandlower modeling
approaches
are
presented
for
selected
facesheets.
Usingaclosed-form
solution,
hestudied
the sandwich panel design parameters. Parameters varied in
influence
onbuckling
of different
boundary
conditionsthe study include the face-sheet thickness and the core
forvarious
panelaspect
ratiosandforbothsoftandstiff thickness as well as modeling fidelity.
Particular
cores.Tessler
et al. [8] present
a {1,2}-order
theory
attention
is given to examining
the buckling
behavior
for
accounting
for transverse
shear
and normal
stresses
and
a specific
panel
aspect
2
American
Institute
of Aeronautics
and Astronautics
ratio,
as
the
core
thickness
becomes
large.Theresultspresented
demonstrate
the parameters
interplay
between
finiteelement
models
withdifferent modeling
levelsof fidelityandthemodeling
requirements
for
accuratepredictionof sandwichpanelbuckling
response.
General
modeling
guidelines
areprovided. Sandwich
taken
Sandwich
Sandwich
panel
modes.
Three
study.
One
general
Panel
buckling
modes
is an overall
instability
mode
together
in the sense
of
the
planar
Another
into long
wavelength
of
the
the
wavelength
referred
short
to as an asymmetrical
material
is either
a short
wavelength
referred
panel
dimpling,
included
often
and assume
face sheets
and
of
with
loaded
supported
Brush
and
an analytical
model
deformation
for
of composite
face
sheets.
This
[21],
is used
herein
instability
loads
for
to determine
critical
that
face-sheet
with
buckling
mode
having
panel
into
given
shape
the
panel
such
non-dimensional
structural
response
planar
as Allen
for the
parameters
of
the
to
[22],
for
The
as its computational
modeling
is a full
of
three-
model,
in which solid threeused to model both the face
models
Recently,
as
This
sandwich
are
referred
an additional
the
in the STAGS
modeling
element
to as
kinematics
and
Sandwich
analysis
approach
for large-scale
as well
as for detailed
modeling
approaches
than
sandwich
models
layered
provide
sandwich
structure
analyses.
are described
to
uses
the
for capturing
sandwich
local
of the
cost
element
23],
stiflhess
models.
referred
[18,
specifically
Sandwich
computational
nonlinear
approach,
models
for less
panel.
Layered
Shell
Layered
shell
simulations
Details
for these
finite
tj; the
Gc as
element
codes.
two-dimensional
b.
groups
derived
the
These
element
Models
models
of layers.
to the
laminate
layers
corresponds
group
of layers
Institute
of Aeronautics
exploit
analysis
the existing
features
plate
available
and shell
in most
The sandwich
panel is modeled
shell
elements
with
at least
The
first
of one
represents
and Astronautics
group
face
to the
3
American
layered
predictions.
approach
These
code.
on the
and
have
characterize
sandwich
load
sandwich
influence
dimensions
and Feng
as
approach
sandwich
buckling
response
are the face-sheet
thickness,
core thickness,
he, and the core shear stiffness,
Researchers,
element
behavior
buckling
a significant
core.
cost.
cost-effective
face-sheet-wrinkling
to
modeling
shell/solid
for general
referred
structure
predictions.
minimum
the
provide
The second
elements
for the face
finite elements
for
as well
finite element
elements
are
as layered
may
a modeling
third
finite
computational
embody
PANDA2
to herein
to the through-the-thickness
The
The
shell
panels
response
specialty
element
developed
of sandwich
structures.
effects
instabilities
typically
of their
herein
is
local
material.
and
local
three-dimensional
solid models
and they are
reserved
for detailed
local modeling
because
finite
orthotropic
model
for
the
the
core
approach
[20]
overall
panels
and
of
are
of this approach,
is related
sheets
face
Vinson
and
dimensional
dimensional
core
all edges
includes
wrinkling
computed
Parameters
as
cost,
panel
buckling
general
accuracy
are not
of the
on
incorporated
general
are
well
both
compression
predicting
face-sheet
configuration
panel.
Other
isotropic
as the analytical
instabilities
and the
These
models
[19].
sandwich
model,
and local
Buckling
core.
shell/solid
by in-plane
Almroth
of
buckling
the
is
panel
conditions
developed
shear
is
and
models
sandwich
behavior.
mode
stiflhess
overall
sandwich
This
as
homogeneous,
are referred
mode,
sandwich
the bending
models
of the global
modes
the
standard
and provide
Accurate
modeling.
exploits
approach
uses standard
shell finite
sheets
and solid three-dimensional
disbond,
overall
of
the
may
instability
instability
of
the core
the
for predicting
panel
presented
Another
mode.
such
These
models,
a first approximation
study.
ignore
approach
with
shell
overall
prediction
approach
on
and
two.
through-the-thickness
modeling
elements.
a sandwich
sheet
modes,
predictions
buckling
simple
wrinkling
and face-sheet
in the present
Analytical
mode.
is one where
or compressed.
failure
core crushing,
first
is commonly
Jitce-sheet-wrinkling
to as a symmetrical
sandwich
and
wrinkling
mode
stretched
core
detailed
panel
buckling
a local
the
representation
whereas
requires
on
dependent
mode,
or
generally
properties
between
general
has
The
is
and
instability
adequate
stiflhesses
short
and the
of
requires
panel.
of
panel
geometry
an interaction
prediction
is equal to one
sheets
mode,
buckling
- long
or
analyses.
depending
materials.
a sandwich
a general
mode,
modeling
approaches
constituent
panel
involve
extension
through
the
This
mode
is a short
panel-buckling
wavelength
the
buckles
consists
face
do not exhibit
any transverse
thickness
of the sandwich.
or
and
sandwich
mode
where
mode
sheets
of a half-wave
buckling
buckles
face
of
sandwich
in selecting
Modeling
element
three
and
an analyst
panel buckling
Element
finite
of
behavior
in this
buckling
the
wavelength
dimensions
possible
possible
considered
panel
where
core buckle
several
are
Finite
one
to guide
for sandwich
panel
geometry
involves
buckling
mode
Buckling
can be used
fidelity
of layers
sheet.
core
the
The
material,
laminate
finite
using
three
corresponds
next
and
group
the
of the other
of
last
face
sheet.Thisapproach
givesanequivalent
single-layer
resultusingclassical
lamination
theorytocompute
the Sandwich Element Models
sandwich
stiffness
coefficients.
Inthismodelalllayers A specialty finite element for sandwich panel analysis
havea common,
uniquerotationthroughthecross has been formulated by Riks and Rankin [23] and is
sectionof thesandwich.Because
thecorematerial described in the STAGS manual [18]. This specialty
generally
onlyoffersshearstiffness,
shear-flexible
Co element exploits the existing shell finite element
shellelements
aretypically
usedinlayered
shellmodels.technology available in the finite element code itself.
Shellelements
basedontheclassical
Kirchhoff-Love
The deformation
of the face sheets is modeled
by using
theory(C1shellelements)
canbeused;however,
such individual shell elements for each face sheet. Coupling
anapproach
ignoresthetransverse
shearflexibility between the two shell elements is carried out by
offered
bythesandwich
structure.
WithinSTAGS,
the applying an appropriate penalty function, consistent with
4-node
C1shellelement
is calledthe410element
(see the material behavior of the core, to enforce the
[24])andthe9-node
ANSCoshellelement
iscalled
the kinematics of the core as a function of the kinematics of
480element
(see[25,26]).Fullintegration
isusedfor the face sheets. The core is assumed to have generally
anisotropic
three-dimensional
elastic properties
whose
bothshellelements
(i.e.,2x2and3x3,respectively).
is defined
by the change
in distance
Finiteelement
formulations
forsandwich
panels
based deformation
between
adjacent
face sheets.
For this special
sandwich
onhigh-order
theories,
such
asFrostig
[7]andTessler
et
al. [8],
are not
available
included
in this study.
Layered
Shell/Solid
within
STAGS
and hence
not
element,
the face sheets are intrinsically
the standard
STAGS
410 quadrilateral
Multiple
of modeling
face sheet
These
shell/solid
multi-layer
models
models.
laminate
elements,
while
elements.
Multiple
thickness
may
core
solid
be
deformation
Each
and
the
are referred
face
may
be a
modeled
using
shell
is
modeled
using
solid
elements
required
between
sheet
as
is
to
shell
element
intermediate
face
sheets
sheets
zero
thickness.
with
solid
is triggered
sheets.
Stifler
computationally
on a C o
analysis
Lagrangian
same
order,
give
with
shell
elements
standard
displacement-field
the
based
solid
elements
compatibility
translational
degrees
of freedom.
Combinations
4-node,
8-node
or 9-node
C o shell element
with
node,
20-node
or 27-node
give
translational
solid
element,
displacement
the 27-node
integration
(i.e.,
3x3
shell/solid
defined
adjacent
ANS
C o solid
is used
and
models,
solid
for both
3x3x3,
to coincide
element
the
shell
with
element.
sheets
are constrained
using
elements
layered
reference
surface
bounding
surface
In
addition,
the
associated
elements
detailed
is defined
in Figure
are
layered
element
rotational
with the
of the face
finite
the
STAGS
quadratic
also
a
panel
provides
characteristics
for
[21].
predictions
Finite
element
element
9-node
for
are
[20]
as
analyses,
analysis
code
shell models,
layered
element
models.
The
cubic
in
and lower
Vinson
finite
approximation
the
analyzed
upper
of
using layered
and sandwich
is
panels
The
Analytical
nonlinear
shell
models
and the 9-node
element
2.
approach
in PANDA2
displacement
to zero.
and
sandwich
identical.
following
is
this
element
and Discussion
study
implemented
of the
Results
this
the
840
strip model.
of the
using
two
provides
response
geometry
sheets
the
for sandwich
structures
assess
when
STAGS,
approach
approach
basic
made
obtained
is called
The
face
in the face
Within
modeling
attractive
Numerical
Full
the
a local
of the
the 8-
For
the
degrees
of freedom
of grid points
core and not connected
to the shell
for the
Within
used.
element
This
to
a spurious
conditions
are generally
are
for large-scale
capability
element).
results
sandwich
23].
of the
and solid
respectively).
[18,
with the
strain is
on whether
by the boundary
points
specialty
respectively,
compatibility.
STAGS,
the compatible
set of shell
are the 9-node ANS C o shell element
linearly
normal
depending
one or two
mode
normal
used
face
to be either
leads
formulation,
face
adjacent
core
element
displacements.
as "phantom"
Between
the
integration
for
are treated
the
of integration
points through
the
sandwich
element
can be chosen
and the
incompatibility
through
through-thewherein the
constant.
The number
core thickness
of each
must
be considered
during
the modeling
For example,
the use of 4-node
C 1 shell
for the face sheets
with
an 8-node
solid
an
be stacked
core
element
process.
elements
to
may
to provide
a refined
of the core material
Displacement-field
the
elements
of the core
discretization
the
through
represent
accurately.
compatibility
to model
to model
to herein
sandwich
thickness
thickness
Models
modeled
using
shell elements.
for
the
element.
the
4-node
out-of-plane
element
For
shell
The
the
and
layered
shell/solid
models, the face sheets are modeled
using the
9-node
ANS 480 shell element
and the core is modeled
4
American
Institute
of Aeronautics
and Astronautics
usingthe 27-node
ANS 883solidbrick elementsidentical single-layer orthotropic face sheets of a given
throughits thickness.This approach
providesa fiber orientation angle and isotropic core. The face
compatible
displacement
field betweenthe finite sheets have thickness tf equal to 0.2 mm and the core has
elements
inthefacesheets
andthose
inthecore.Forthe thickness hc equal to 10 mm (hJa 0.044). The
sandwich
element
models,
the840sandwich
elementsmechanical properties of the face-sheet material are:
[18,23]areused
through
thecorethickness.
E1 229.0
GPa,
E2 13.35
GPa,
G12 5.25
GPa and
v12 0.315
(i.e.,
E2/E1
0.058).
Core
material
data
are
Fullpanelmodels
aredefined
asfiniteelement
modelsgiven in Table 1. The fiber orientation of the face-sheet
oftheentiresandwich
panel.Spatial
discretization
ofthe material is varied from zero to ninety degrees where the
full panelis relatedtotheanticipated
buckling
mode. zero-degree
orientation
is parallel
to the loading
Thus,
some
knowledge
ofthepanelbuckling
response
is direction. For fiber orientation angles other than 0- and
needed
todevelop
thesefullpanelanalysis
models.
For 90-degrees, the face sheets exhibit anisotropic behavior
arectangular
panel,
thenumber
ofhalf-waves,
m, along
and the D16 and D26 bending
stiffness
terms become
the panel
length
mode
is
assuming
in a general-instability
function
of
a single
panel.
the
half-wave
Generally,
five
needs
ratio
the
refined
a/b,
width
points
buckling
mode,
to be sufficiently
wavelength
aspect
across
to six grid
are required
for accurate
capture
a face-sheet-wrinkling
mesh
panel-buckling
panel
per
of the
nonzero
and large
terms.
Consequently
stiffness
half-wave
terms
shapes
may
with
to represent
of half-waves
buckles.
These
of
strip
models
a localized
are defined
region
thickness
modeling
wrinkling
modes
model
wrinkling
panel
behavior
models.
the present
study
mm
through
used
and
the
length
of
wavelength
local
has
strip
25
models
away
layers
of
thickness
local
of the
buckling
with
edge
The
or x
a thin
state
equal
elements
integration
point
depends
layer.
are
In
the
to
zero.
results
case considers
for two
a square
orthotropic
face
on
studies
influence
and
the
boundary
predictions
loaded
in
considers
of different
condition
response
to
uniaxial
a rectangular
sheets
and an isotropic
the
influence
of different
on
the
The
core.
modeling
two
To
1: Square
Sandwich
researchers
case
buckling,
under
supported
uniaxial
225-mm
square
compressive
sandwich
stress
this
z=0
Edge
3 (x
z=0
along
and
a).
In addition,
2 and
element
are:
1 (x
4 (y
= 0y
a point
to
the
edge
Edge
displacement
buckling
Having
established
boundary
second
defined
case
a uniform
conditions
next.
for
All nodes
the
0);
b and
= 0z = 0
at the center
of
v set equal
to
These
boundary
state in the face
four
This
case
examines
approaches
on the
with
(x
edges
of the
0 andx
a,
Edges
analyzed
stress,
study,
the
straightforward
of a simply
(a/b
shell
these
However,
1) with
sandwich
modeling
simply
5
Institute
of Aeronautics
and Astronautics
state,
are simply
are
along
the
supported
These buckling
boundary
= 0 along Edges
1 and3
and,
b and y
approach
supported
to impose
stress
analyses
the thickness
panel
respectively);
2 and 4 (y
uniaxial
buckling
through
isotropic
American
finite
w = 0x
and
equal
state,
each
along
Edges
with
panel
the
along
stress
are
stress
in
points
end
(Edge
direction,
components
applied
grid
0, respectively);
edge
pre-buckling
pre-buckling
panel
have
of a panel
a uniform
loaded
in-plane
x=Oy=O
x=0y=0
the layered
17]
analysis
sheets.
Panel
[13-15,
all
by
in the longitudinal
conditions
to
configuration
response
predictions
for sandwich
panels
core thickness
and face-sheet
thickness.
Several
achieve
loaded
The
in-plane
oandw=O
along
Case
impose
the
approaches
sandwich
sandwich
This
is
zero to remove
rigid-body
motion.
conditions
result in a uniform
stress
single-layer
core.
modeling
compression.
increases.
shape
analysis
to the entire
stress
along
first
panel
has uniform
other
w=0
several
The
with
applications
for a given
face
buckling
different
panel
and an isotropic
the number
mode
for the
axis
as the
effects.
2).
u=u
the
analyzed
are presented.
sandwich
sheets
increases,
on
required
mode
In addition,
requirements
0 in Figure
models
load is determined.
cases
lines.
mode
Uo, applied
the panel
Numerical
those
sandwich
boundary
is
The
Therefore,
lengths,
over
anisotropic
buckling
in the buckling
modeling
without
full
sandwich
mode.
node
stiffness
to the loading
general-instability
complexities
models
the
relative
orientation
in the
shortening,
effects.
model
buckling
the
represent
core
different
with
of each
strip
strip
face-sheet-
has a width
One
nonzero,
modal
bending
wavelength
A local
from
model
thickness.
the
models,
ensure
this
short
obtained
strip
of a panel
the
that
be detected.
predictions
core
through
so
models
through-the-
short-wavelength
Local
slice
element
significant
readily
to verify
longitudinal
2.54
detail
can
is used
as finite
with
the
skewed,
fiber-angle
additional
Local
become
non-straight
face-sheet
to the other
when
become
predictions.
To
the finite element
short-
relative
u = w =
used
boundary
on the
shell/solid
0y
0, respectively).
finite
0
For
in the present
conditions
element
modeling
are
model.
approach
when
using
one
integration
point,
it
is
Results
of a convergence
studyconducted
usingthe possible in some situations to trigger a spurious mode.
sandwich
element
modeling
approach
to predictthe In such cases, two integration points should be used.
buckling
response
ofasquare
sandwich
panel
witha30degree
face-sheet
fiberorientation
angleareshownin Results of the present analysis approach, of Hadi and
Figure3. Theopensymbols
ondashed
linesdenote Matthews [15], and of Yuan and Dawe [17], expressed
results
obtained
usingmodels
withoneintegration
point in terms of the in-plane stress resultant, are shown in
(IP)through
thethickness
ofthesandwich
element,
and Figure 4 as a function of the fiber orientation in the face
thefilledsymbols
onsolidlinesdenote
results
obtainedsheet. The present results were generated using a full
usingmodels
withtwointegration
points.One,two, panel model with 25 grid points in each planar direction.
four,andeightlayersof sandwich
elements
were These 625 grid points were used to define a finite
analyzed
formeshes
withdifferentlevelsof in-planeelement mesh of either 576 4-node quadrilateral shell
discretization.Thepredictedbucklingloadsare elements or 169 9-node quadrilateral
shell elements.
normalized
by thesolution
obtained
usingthelayered Results of the present analysis are presented for layered
shell/solid
approach
withfoursolidelements
through shell models with and without shear deformation,
thecorethickness
anda 25x25planarmeshof nodes layered shell/solid
models,
and
sandwich
element
(i.e.,thebucklingloadusedin thenormalization
is models. The results shown in Figure 4 indicate that the
340.0N/mm).Thissolution
obtained
usingthehighest layered shell finite element models are stifler and
fidelitymodelis referred
to hereinasthereferencepredict higher buckling loads than the layered shell/solid
solution
andisdenoted
inFigure
3 bythe"x" symbol
on finite element model (reference solution). The results
asolidline.HadiandMatthews
[15]reported
avalueof shown in Figure 4 indicate that the layered shell model
(410
element
models)
467.8
N/mm,andYuanandDawe[17]reported
avalue without shear deformation
of 382.6N/mm. HadiandMatthews
[15] useda provides a much stifler solution (higher buckling loads)
Rayleigh-Ritz solution with trigonometric than the layered shell model with shear deformation
approximations
forthein-plane
variation
anda zigzag (480 element models) and the buckling load decreases as
theorythroughthe thickness.
Theirresultsgivea the fiber angle increases from zero (results indicated by
buckling
solution
stiflerthanthepresent
solution
dueto filled symbols). Results for the layered shell model with
shear deformation
(480 element
models)
indicate
a more
implicitboundary
conditions
fromthedisplacement
flexible
solution
(open
symbol
results)
than
that
fieldapproximations.
YuanandDawe[17]used
a finite
stripsolution
withB-spline
approximations
forthein- obtained using the 410 element. Results obtained for the
sandwich
element
model
(840 element
models)
with
plane variation and the through-the-thickness
approximations
for the in-planedisplacements
are rigid links along the boundary edges, denoted by the "x"
quadratic
andtheout-of-plane
displacements
arelinear. symbol, indicate a response similar to that predicted
The
Theirresults
aremuchimproved
duetotheirtreatmentusing a layered shell model with shear deformation.
of theboundary
conditions,
yetstill stiflerthanthe rigid links impose classical plate theory kinematics
6
American
Institute
of Aeronautics
and Astronautics
alongthe boundaries
while internallythe shear thickness
witheachsandwich
element
having
oneortwo
flexibilityof the corematerialis includedin the integration
pointsin thethickness
direction
(openand
simulation.
Thebuckling
loadisnearlyconstant
uptoa filledsymbols,
respectively).
Thefiniteelement
models
fiberangleof30degrees
andthendecreases
asthefiber of thefull panelhave25gridpointsin eachplanar
angles
increases
fuxther.
direction.Thepresent
resultsbracket
the solutions
reported
by YuanandDawe[17]andappear
to be
Theresultsshownin Figure4 indicate
thatall finite converging
tothereference
solution.Asthenumber
of
element
models
of thepresent
studyandtheanalysissandwich
elements
through
thecorethickness
increases,
approach
ofYuanandDawe[17]predicta maximumthe sensitivity
to thenumberof integration
points
These
resultsillustrate
theeffectiveness
of
buckling
loadwhentheface-sheet
fiberorientation
angle decreases.
sandwich
element
formulation.
equalszerodegrees
anda progressive
decrease
in theSTAGS
bucklingloadoccursas thefiber orientation
angle
increases.HadiandMatthews
[15]andadditional Dueto thenatureof thisproblem
andthediffering
earlier
analyses
[13,14]usingaRayleigh-Ritz
approachresults
obtained
byvarious
researchers,
twoadditional
withtrigonometric
approximations,
however,
predict studies
wereperformed
to studytheeffectof theface
thatthebucklingloadincreases
withincrease
in the sheet
moduliratioandpanelaspect
ratioonthegeneral
face-sheet
fiberanglefrom0-degrees
upto40-degrees.
instability
response
of asandwich
panel.Results
were
Thenthebuckling
loaddecreases
rapidly
tothebuckling obtained
usingthesandwich
element
modeling
approach
load value for the 90-degreecase which is withfourelements
through
thecorethickness,
andtwo
approximately
halfthebuckling
loadvalueforthe0- integration
points
through
thethickness
ofeach
element.
degree
case.Thistrendis similartothetrendobtainedResults
forthepresent
square
panelwithE2/E1 0.058
for a laminate
composite
platewith anaspectratio are shown in Figure 7 along with results for a
greaterthanone(e.g.,AshtonandWhitney[27]). rectangular panel with an aspect ratio, a/b, equal to two,
However,for a squareplatewith thesematerial and face-sheet moduli ratio E2/E1 0.058, and for a
properties,
thebuckling
loaddecreases
asthefiberangle square panel with the transverse modulus, E2, doubled
increases.
YuanandDawe[17]reporta decrease
in _2/E1
0.116).
Results
for the square
panels
with
buckling
loadwithanincrease
in fiberangleforthe E2/E1
0.058 and 0.116 are shown
by the solid and
square
sandwich
panelunderconsideration
andindicate dashed lines, respectively, and results for the rectangular
thatRayleigh-Ritz
solutions
basedontrigonometricpanel are shown by the dotted line. As shown in Figure
series
overlyconstrain
thepanelasmaterial
anisotropy7, doubling the ratio of the face-sheet moduli ratio in the
increases.
Double
sine-series
solutions
forfinitepanels square panel causes an overall increase in the panel
withsimplysupported
boundary
conditions
andhaving buckling load. Increasing the panel aspect ratio to two,
symmetric
laminates
giveriseto artificialderivativewhile holding the face-sheet moduli ratio constant,
boundary
constraints
thatprevent
convergence
to the results in a variation in the buckling load with face-sheet
correctsolution[28,29]. StoneandChandler
[29] fiber angle similar to the variation obtained for a
report
thatbuckling
loadswillbeoverestimated
andmay composite plate with similar material properties and an
contribute
tomisleading
trends
orconclusions.
aspect ratio greater than one. For this case, the buckling
load increases
with
an increase
in face-sheet
fiber
angle
All finiteelement
results
fromthepresent
studyindicate up to 25 degrees, and then declines with increase in
thesame
buckling
modeshape
(general
instability)
for face-sheet fiber angle.
anyface-sheet
fiberorientation.Initially (0-degree
case)themodeshape
involves
onlyonehalf-wave
in Case 2: Rectangular Sandwich Panel
bothdirections.Astheangleincreases,
thebuckling The panel considered has a rectangular planform and is
modeshape
becomes
skewed
andtendsto followthe 508-mm long and 254-mm wide (a/b 2). Identical
fiberangle.At approximately
60degrees,
theskewedaluminum face sheets of thickness tf on an aluminum
modeshape
hastwolongitudinal
half-waves
andone honeycomb core of thickness hc define the sandwich
transverse
half-wave.
Thispattern
becomes
lessskewedpanel. Two face-sheet thicknesses are considered in this
astheface-sheet
fiberangleapproaches
90degrees.
In study: a thin face sheet with thickness equal to 0.508
Figure
5,contour
plotsoftheout-of-plane
displacement
mm and a thick face sheet with thickness
equal to 2.794
component
ofthebuckling
modeshape
areshownto mm. The core thickness is varied from very thin, equal
illustrate
thiseffectfordifferent
face-sheet
fiberangles. to the face-sheet thickness, to very thick, approaching
half
the
panel
width.
As
the core
thickness
increases,
Theeffectoffiberangle
onthebuckling
results
obtainedthe buckling response transitions
from
a general
usingthe sandwich
elementmodelingapproach
is instability mode to a short wavelength
face-sheetshownin Figure6. Results
areshownforallmodels wrinkling
mode.
Young's
modulus,
E, for the
usingoneandfoursandwich
elements
through
thecore aluminum alloy is equal to 68.95 GPa and Poisson's
7
American
Institute
of Aeronautics
and Astronautics
obtained
using
Inthepre-stress
condition,
thesandwich
panelisloaded expected.
Good
byauniform
endshortening,
Uo, applied to the sandwich
model results
and
face
sheets
Figure
on the ends
2).
For the
conditions
u=u
Edges
and
a).
panel has v
calculations,
z=0
along
3 (x
at x 0 and x
respectively);
Edge
buckling
applied
0andw=0
w=Oy=0
of the panel
u =-u
0 to remove
along
each
Edge
0 andw=0
In addition,
rigid
body
(y b
at the
(x
y
the finite
given
0,
panel
finite
present
element
study
and 21 grid
has
buckling
panel
across
(1,701
parametric
with
and
=0
along
core
center
of the
flexibility.
used
in the
PANDA2,
length
the
spatial
compared
during
are
(800
approach
is
anticipated
because
and low
computational
as compared
elements).
to 2.54
core
thickness
cost
One
used
the STAGS
however,
boundary
response.
wrinkling
local
strip
the PANDA2
those
the
through-the-thickness
predicted
PANDA2
are
quite
PANDA2
predictions,
modeling
model
face-sheet
is used
through
element.
The
state in the face
associated
For
values
small
correlate
For larger
STAGS
by
thin
core.
on the layered
from
for a
of hJa)
due
for modeling
to
the
or four
27-node
solid
(open
results
are
from
wrinkling
buckling
core
thickness
these
results
to panel
thickness
less than
with
thickness
hc/a,
the
PANDA2
from
a result
of hJa,
through
lower
allows
Apparently
made
in
on
the
thickness
that
thickness
of an
results
assumed
the thickness
a piecewise
the
face-sheet
the
Institute
of Aeronautics
core
models
and Astronautics
just
thick
to face-sheet
with
a ratio
greater
of core
than
50,
to be conservative,
as
The
response
local
using
strip model
25 sandwich
In addition,
in Figure
results
the
results
for moderately
thickness.
case
8
American
wrinkling
distribution
summarized
when
In addition,
displacement
of the core.
shell/solid
results
linear
linear
through
appear
to a
of the
between
of core thickness
For panels
results
analytical
the
in
that the
value
(hJa).
model
panel
curve
mode
agreement
50).
wrinkling
a solid
buckling
face-sheet
buckling
thin
strip
(for ratio
symbols,
overall
at a small
and
core
8 indicate
ratio
results
the local
elements
layered
a general-instability
to
length
excellent
one
the
filled
in Figure
occurs
either
through
as
panel
mode
panels
shown
an overall
wrinkling
shell/solid
using
PANDA2
shown
show
predictions
and
also
8. The results
wrinkling
line, and the
model
elements
8).
are
layered
is discretized
Figure
model
STAGS
In the
symbols
in
transition
the
shell/solid
brick
and
predictions
by the dashed
symbols.
models,
wrinkling
in
model,
shell/solid
line,
layered
model
wrinkling
PANDA2
dotted
the
with
Based
core
from
PANDA2
analytical
strip
are shown
full-panel
PANDA2
the
are much
model.
conservative.
the
shell
through-the-thickness
layered
8.
the
sandwich
values
loads
by
obtained
of
assumptions
to a transition
based
models
the
local
panel
local strip
sheets and
wrinkling
full
in Figure
shown
Figure
the
using
predictions
buckling
a width
through
investigate
model.
a very
values
shell
and
STAGS
respectively,
approach
with
the
thickness
is
of modeling
elements
point
by the
are
modeling
instability
conditions
loads
than
corresponds
shell
ease
strip
to
integration
STAGS
the
width.
obtained
shown
along
approaches.
mm and 25 sandwich
is
layered
behavior
STAGS
model
the panel
general
of this
the local
from
with
results
as
layered
results
structuJces
(increasing
in the
shear
represent
(3,200 elements)
has four
as the mesh
of 9-node
when
thickness
of each sandwich
model has a uniform
stress
PANDA2
half-waves
across
of the simplicity,
study,
equal
symmetric
to
In the present
buckling
constant
A layered
attractive
This
is held
up to sixteen
elements
the panel
width.
adequate
panel
along
is
The mesh
times the
wrinkling.
full
grid points)
element,
the
motion.
the panel
studies
modes
length
for the
81 grid points
points
discretization
the
mesh
analytical
element
length
limitations
Results
The
shear-flexible
between
shell modeling
approach
quickly
tend to deviate
the analytical
results as the core thickness
increases
0);
and
the
for sandwich
However,
edge are:
x =0_
a point
is obtained
the boundary
along
and
a (see
the 480
correlation
8 show
obtained
correlate
a single
well
solid
that
the
for
using
the
with
the
element
through
thecorethickness
isused.Goodcorrelation
is Finally,
alsoshown
between
theresults
obtained
usingthelocal associated
stripmodelandresultsobtained
usingthe layered shell/solid
shell/solid
modelwhenfoursolidelements
through
the for the
corethickness
areused.Forthethickface-sheet
case sandwich
element
withtj-2.974 mm and hJa 0.4, the layered shell/solid
models
predict
higher
a buckling
model.
indicate
The
that
shell/solid
models
convergent
sheet-wrinkling
even
for
behavior
between
the buckling
element
Similar
results
to
of the layered
is evident
by the local
tj-2.974
mm
load predicted
is converging
behavior
approach
load predicted
results
for
the buckling
for
4-node
material
obtained
element
elements
face
quadrilaterals.
through
sheets
Figure
the
for the
and filled
element
well with
using
sandwich
results
results
increases
from
six
eight
predict
435.31
kN,
convergence.
Use
in a very
the
buckling
predicted
buckling
respectively,
with
were
performed
with
predicted
kN,
respectively.
of two
stiff
load
loads
Again
integration
points
per
powerful
general
for
eight
times
results
indicate
structures
exhibiting
strip models
the
single
layer
and
the
a single
model
as
in a large-scale
buckling
be
shell/solid
accuracy.
scale
quite
a
to achieve
panels
aspect
element
structures
modes
determine
of Aeronautics
ratio
modeling
provides
its general-instability
9
Institute
and Astronautics
four
by using
shell/solid
layers
of
approach
equal
the
levels
of
layered
solution
approach
an
to large-
to apply
However,
for
effective
buckling
the
cost
is over
of
expected
limits.
of
a solution
the computational
cost
60%
Approximately
are
with
only
is achieved
modeling
model
Extrapolation
are used.
American
to obtain
using
either
thickness
obtained
in the core.
element
at the
using
the
required
by
in
expended
solutions
accuracy
increases,
proportionally
require
computational
sandwich
elements
elements
the
sandwich
a layered
in the core.
obtained
In this case,
sense
stiffness
sandwich
obtained
element
for
element
through
of solution
structural
of solid
values
stiffness
global
of that
solid
effort
the
layers
of
increases
element
sandwich
sandwich
layer
of sandwich
effort
elements.
times
cost
of
in the
60%
layers
element
is
element
of the global
a single
Buckling
level
number
size
a single
of sandwich
solid
modeling
computational
the
with
four
solid
converges
The computational
effort
element
model is defined in
However,
a single
the same
four
computational
computational
with
that
large-scale
local
equal.
of two
for modeling
the
nearly
for the
to
the sandwich
equations
with
elements
while
face-sheet-wrinkling
level.
approach
kN and 461.93
approach,
is not effective
element
sandwich
modeling
instability,
to
77.7 kN to
approach
of equations
with
of
approach
from
of using
the
of layers
number
sandwich
case
modeling
of through-the-thickness
increases
model
layers
respectively.
element
as
the
element
sandwich
is approximately
addition
value
appear
four,
number
from
number
per
layer
for the
the
the
a model
the
model,
the
as
number
The
analyses
results
These
per
to
that change
increases.
a sandwich
As the number
a slow
layer
the
for
Additional
of 479.28
the
layers
with
number
panel,
52.1 kN to 67.6 kN to
modeling
than
For
conducted
points
kN
decreases
shell/solid
of layers
kN
(same
sandwich
shell/solid
loads
approach
elements
0.4, indicate
element
predicting
unless
matrix
point
432.99
four
slowly
layer.
matrix.
and higher
indicating
model.
six and
extremely
two
evident.
two
The advantage
element
associated
each
through
for a single
using
but
sandwich
of
integration
buckling
converging
the
loads
is nearly
layer
of
number
elements
solution
a four
in
shell/solid
sandwich
the
from
to
shell/solid
quickly
perspective
the
of
analyses
thereby
layered
in turn
and hc/a
sandwich
73.7
of
one integration
Further
kN
which
response
as the
75.2
terms
Use
of the
for
layered
buckling
to
The
C 1 shell
of through-the-thickness
one
critical
point
is less
case
load
one to four.
and
thickness
and
face-sheet
the
0.1.
layered
discretization
changes
gives
approach.
are observed
for the thick
as the core
thickness
using
Similarly,
performed
hJa
the 4-node
the
plane
load
approach
thickness
behavior
in buckling
integration
model.
the
in a stifler
obtained
an increase
results
strip
this convergent
example,
using
local
through
loads.
Similar trends
cases, However,
increases,
layer
the
element
buckling
face-sheet
one
results
element
as the number
modeling
9). Results
while
in the
from
more
symbols
case,
finite
increases
exploits
layered
buckling
kN
the
core
"phantom"
(open
approach
was
with
cost
and
approaches
case
formulation
computational
element
thickness
layers
increases
from one
respectively.
Clearly
for this wrinkling
sandwich
face-sheet
with
the analytical
points
the
with
in Figure
are
models
models
models,
layers
thin
models
one
models
element
element
thickness
for the
with
one or four
respectively,
correlate
integration
core
that
sandwich
obtained
either
intermediate
symbols,
9 show
In these
using
change
modeling
approach
C o shell
a fixed
76.8
of the
sandwich
face-sheet
element
modeling
sandwich
is discretized
2%
the
element
thin
critical
face-
review
with
of grid points)
modeling
summarized
in Figure 9. The sandwich
have the same discretization
as the finite
with
For
The
wavelength
only
through
strip
kN.
shell/solid
load predictions
element
9-node
and hJa
0.4
by the layered
430.03
short
with
8%
a limited
because
use of the
large-scale
means
response.
to
Conclusions
A comparative
studyofdifferent
modeling
approaches
1.
for predicting
sandwich
panelbuckling
response
has
beenpresented.
Thestudyconsidered
sandwich
panels
with anisotropic
facesheets
anda verythickcore.
Resultsfromconventional
analyticalsolutionsfor
sandwichpaneloverallbucklingand face-sheetwrinklingtypemodes
werecompared
withsolutions
obtainedusingdifferentfinite elementmodeling 2.
approaches.
Thenumerical
analyses
wereconducted
usingtheSTAGS
nonlinear
finiteelement
code,
andthe 3.
analyticalresultswerecomputed
usingPANDA2.
Finiteelement
solutions
wereobtained
usinglayered
shellelement
models,
withandwithout
transverse
shear 4.
flexibility,layered
shell/solid
element
models,
thatuse
shellelements
forthefacesheets
andsolidelements
for
the core,and sandwich
modelsusinga recently
developed
specialty
sandwich
element.Convergence
characteristics
of theshell/solid
andsandwich
element5.
modelingapproaches
with respectto in-planeand
through-the-thickness
discretization,
weredemonstrated.
Results
ofthestudyindicate
thatthespecialty
sandwich
element
formulation
implemented
in STAGS
provides
an accurate
andeffectivemodelingapproach
for
predicting
bothoverallandlocalized
behavior.This 6.
modeling
approach
provides
theflexibilityformodeling
anentirepanelorformodeling
localdetailthrough
the
corethickness.Resultsof the studyindicatethat 7.
transition
froma general
instabilitymodeto a short
wavelength
face-sheet-wrinkling
modeoccursasthe
corethickness
isincreased.
Modeling
fidelitytocapture
bothpossiblemodesmustbe providednearthe 8.
transition
region. Furthermore,
resultsindicatethat
anisotropy
ofthefacesheets,
alongwiththeratioof
principle
elastic
moduli,significantly
affectthebuckling
response,
andtheseeffectsmaynot be accurately
represented
byanalytical
solutions.
9.
References
Allen,
Howard
Axel
F.
12.
New York,
Anon.,
Structural
HDBK-23A,
A.
Theory
S. and
June
Construction,
John
Composites,
of Defense,
Mayers,
Wrinkling
Mil-
Washington,
J. "General
of Sandwich
and Applications,"
AIAA
No. 4, April
Ha, K. H.,
Construction:
"Finite
Element
A Critical
Construction
(editors),
Constructions,
1989,
Instability
Plates
Unified
Journal,
Vol.
Analysis
Review,"
1, Karl-Axel
Reichard
International
5,
of Sandwich
in Sandwich
Olsson
and
Proceedings
Conference
Stockholm,
of
on
Sweden,
Ronnal
P.
the
First
Sandwich
June
19-21,
pp. 69-85.
K. H.,
Plates:
"Finite
An
Element
Overview,"
Analysis
of
Computers
and
Sandwich
Structures,
Y.,
Flexible
Core
High-Order
Journal
of Solid
and Structures,
1998,
pp. 183-204.
Tessler,
"Buckling
A.,
and
Vonach,
M.
W.
52, 2001,
K.
and
Approach
Composite
Rammerstorfer,
Noor,
Chapman
A.
K.,
G. J. Turvey
Models
Shells,"
Applied
Mechanics
3, March
1996,
pp. 155-199.
Librescu,
the
L. and Hause,
Modeling
and
Sandwich
Constructions:
Structures,
Vol.
and Astronautics
48, Nos.
of
S.,
for
Sandwich
Postbuckling
of
pp. 157-189.
and
Bert,
Sandwich
T., "Recent
Behavior
C.
W.,
Panels
Reviews,
and
and I. H. Marshall
1995,
W.
"Computational
and
and
and Hall,
"A
pp. 363-376.
Deformation
Burton,
G.,
Instability
Engineering
in Buckling
Plates,
F.
Wrinkling
Structural
"Shear
G., "A
Laminates,"
Mechanics,
Composite
Gendron,
3-4,
Plates,"
C. W.,
35, Nos.
pp. 67-84.
to the
Configuration,"
Accounting
Deformations
Sandwich
Bert,
with
International
Vol.
S., and
Sandwich
Vol.
Panels
Theory,"
Plate Theory
Thermoelastic
Composites
Structures,
of Sandwich
Annett,
{1,2}-Order
Dimensional
10
of Aeronautics
on
Sweden,
1966.
Sandwich
pp. 1-17.
Institute
(editors),
Conference
Stockholm,
Department
Face
Ha,
Reichard
1968.
Benson,
in
American
P.
International
J., Sandwich
Wiley,
and
Today
1, Karl-
Plantema,
D.C.,
Construction
Construction
Ronnal
Constructions,
(editors),
11.
and
of the First
Sandwich
19-21,
"Sandwich
in Sandwich
Olsson
Proceedings
General
Acknowledgement
Theworkofthesecond,
thirdandfourthauthors
was
supported
bytheNASALangley
Research
Center
under 10.
contracts
NAS1-00135,
GS-35F-0038J
(Task1730),
and
NAS1-99069,
respectively.
G.,
and Tomorrow,"
Vol.
and
49, No.
Developments
of
Advanced
Survey,"
1-3, January-March
Composite
2000,
pp. 1247-1253.
Kim,
C.
G.
Finite
15.
Bonding
Hadi,
Stiffness,"
1988,
Load
Approach
1998,
pp. 245-255.
D. J. and
Part I: Analysis,"
Sandwich
Mechanics
Vol.
of
26,
Deformation
Vol.
27.
An
of
and
with
the
28.
in Applied
Rankin,
C.
Laminated
in
Vol.
190,
19.
Brush,
Plates'
20.
Vinson,
Mechanics
C.,
Brogan,
29.
F. A.,
Local
David
W.,
4.0,
and
Sandwich
Panels
Subjected
Vol.
an
Analysis
Panel,"
AIAA
Rankin,
C.
Mechanics
and
December
1989,
Stone,
M.
Sine
Peshkam,
A.
M.,
Composite
to Uniaxial
24,
No.
10,
of
Honeycomb
or
Paper
No.
97-1142,
A.
C.
NASA
360
France,
Theory
V.,
Diaphragm
"Buckling
Ends,
Vol.
and
Series
Chandler,
H.
Solutions
Laminated
Part
13-15
77,
Nos.
D.,
Properties
Ecy, MPa
Ecz, MPa
Plates,"
Sciences,
to
the
38,
Composite
2001-1323,
and
CR-4358,
April
Brogan,
0.265
82.74
Gcyz, MPa
90.4
49.64
Vcxy
0.3
0.3
0.3
0.01
Voyz
pp.
2001.
The
Mechanics
Testbed
ES5: STAGS
Shell
1991.
11
American
Institute
of Aeronautics
Case 2
146
146
Strength
A.,
Gcxy, MPa
G .... MPa
Compression
F.
properties
0.6895
68.95
Modeling
Residual
Case
and Astronautics
in
Supported
International
Vol.
200
200
1997,
1998,
1-2,
"Errors
for Simply
1. Elastic mechanical
for core materials
April
May
I: Finite
in Applied
pp. 517-526.
(editor),
Dordrecht,
and
Prismatic
Methods'
Engineering,
of
pp. 1-30.
Colloquium
C. C., "Sandwich
a Damaged
Paper
Vautrin
EUROMECH
Rankin,
Vol.
Co., 1970.
Composite
Computer
of Mechanical
Ecx
of Bars,
1975.
via PANDA2
Computational
Structural
Structural
Element
Processor
Element,
with
J.
Publishing
Alto,
with
Application
of
Design
Publishers,
NASA
Mechanics,
Lockheed
Panels
Saint-Etienne,
E. and
Formulation,"
Voxz
Structures,
Academic
with
Structures
Double
Whitney,
Technomic
and
Feng,
Z., "Classification
of
Panel Behavior,"
in Mechanics
of the
Riks,
Elements',
of Applied
Finite-Length
Strip
Table
Shells'
Palo
of
Journal,
AIAA
Allen,
H. G. and
Structural
Sandwich
in
Shell
Engineering,
Loden,
Design
"Optimum
Sandwich
Cores,"
Kluwer
1-12.
J.
of
Plate
1996,
Center,
"Optimum
AIAA
Proceedings
and
Symmetrically
Foam
1997.
E.
D.
Computer
of General
Version
Technology
Sandwich
Compression,"
Composite
Journal
Plates',
Material
J. R.,
Bushnell,
J.
Dawe,
Journal
D. O. and Almroth,
B. O., Buckling
and Shells', McGraw-Hill,
New York,
October
ASME
Ashton,
Laminated
2000.
Honeycomb
held
and
Analysis
Manual,
Advanced
CA, June
ANS
pp. 5215-5231.
User
Martin
and
with
Applications,"
H., Structural
- STAGS
and
Structural
Processor
Vibration
Local
Methods'
D. J., "Overall
Plates
II:
SRI
The Computational
Structural
Element
1990.
Laminated
Engineering,
Sandwich
190, 2001,
Basic
Strains,"
the
Panels:
_bmputer
and
Part
Cabiness,
24.
Vol.
"Predicting
Sandwich
Plates
W. X. and Dawe,
Methods'
23.
26.
with
pp. 5197-5213.
Yuan,
of
Journal,
W. X., "Overall
of
Faceplates,
22.
Yuan,
Faceplates,
Buckling
21.
of
Plates
M.,
Testbed
Structures,
Buckling
2001,
"Buckling
F. L.,
Shear
Composite
Applied
18.
AIAA
of Anlsotropic
Dawe,
S.,
pp. 982-988.
Including
Faces,"
17.
C.
Sandwich
B. K. and Matthews,
Buckling
16.
Hong,
Anisotropic
No. 8, August
ESI:
G.
CR-4357,
and
Unbalanced
Stanley,
Mechanics
No.
5,
Figure
1. Photograph
ofthicksandwich
panel
segment.
Face
sheet
tf
Core
Face
sheet
_dge
Edge
Edge
Edge
3
]
Figure
2. Sandwich
panel
geometry.
12
American
Institute
of Aeronautics
and Astronautics
Layered
One
shell/solid
sandwich
- 1, 2, 4 layers
layer
- 2 IP
Two sandwich
layers
- 2 IP
Four
sandwich
layers
- 2 IP
Eight
sandwich
.- _- - One
sandwich
layers
layer
- 2 IP
N x
- 1 IP
.- ,_ - Two sandwich
layers
- 1 IP
.- _- - Four
sandwich
layers
- 1 IP
.- 43 - Eight
sandwich
layers
- 1 IP
1.6
ttttl
Z
1.3
Z
1.0
)[
Z
0.7
10
20
30
Number
Figure
3. Effect
degree
fiber angle.
of mesh
refinement
for sandwich
element
models
13
American
Institute
of Aeronautics
and Astronautics
sandwich
panel
with
a 30-
--
Reference
-
.....
800 --
Hadi
solution
(layered
& Matthews
shell/solid)
[15]
[ 17]
Layered
shell model
using
STAGS
410
elements
----O--
Layered
shell model
using
STAGS
480
elements
-----)b-
Single-layer
with
rigid
sandwich
element
model
links
_-y
0
0
30
60
90
Ply layup
Figure
4.
Effect
of the square
of face-sheet
sandwich
fiber angle
angle, 0, degrees
on buckling
load
for layered
shell/solid
panel.
14
American
Institute
of Aeronautics
and Astronautics
and layered
shell
models
>>>>>>>>_
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
ilililililililililililililililll
N
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
-1.0 iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
iiiiiiii iiiiiiiii
Figure
different
5.
a)0-degree
case
b) 40-degree
case
c) 60-degree
d) 90-degree
case
buckling
mode
Contour
face-sheet
plots
case
of the
out-of-plane
displacement
component
of the
fiber angles.
15
American
Institute
of Aeronautics
and Astronautics
shape
for
--
Reference
-
Hadi
.....
solution
800 --
(layered
& Matthews
Four-layer
[ 17]
sandwich
sandwich
----O--
Single-layer
----V--
Four-layer
shell/solid)
[15]
sandwich
sandwich
element
element
model
model
element
element
model
model
- 2 IP
- 2 IP
N x
- 1 IP
- 1 IP
600 -Z
200 --
0
0
30
60
90
Figure
element
6.
Effect
models
of face-sheet
of the square
fiber
sandwich
angle
0, degrees
on buckling
load
for layered
shell/solid
panel.
16
American
Institute
of Aeronautics
and Astronautics
model
and
sandwich
E2/E
1 = 0.058,
a/b = 1
E2/E 1 = 0.116,
a/b = 1
E2/E
a/b = 2
1 = 0.058,
8OO
600
Z
..)
400
"'_
sss
SS--
200
30
60
90
Ply layup
Figure
buckling
7.
Effect
of face-sheet
fiber
element
angle,
model
aspect
angle, 0, degrees
ratio,
of the square
and face-sheet
sandwich
elastic
panel.
17
American
Institute
of Aeronautics
and Astronautics
modulus
ratio
(E2/E1)
on
[]
-.....
-
1000
STAGS
STAGS
STAGS
STAGS
solid/shell
solid/shell
solid/shell
solid/shell
model
model
model
model
- 1
-4
- 1
-4
800
600
[]
400
tf
2.794
mm
200
0 _.C___
-_ ......
_.0_
'-'
o
-._--4-"7 .............
o o }
-...04
[tf=
.15
Ratio
of core
0.508
mm
.30
thickness
to panel
from analytical
model,
18
American Institute of Aeronautics
length,
[
.45
hc/a
and Astronautics
element model
and full
[]
1000
800
z
600
_
400
tf = 2.794 mm
200
= 0.508 mm
..4--....
O,
"-0
..o-t_
-
-o--
--O-.o)[tfj,
_D
- - "0- - - - ..,0,
.15
Ratio of core thickness
.30
to panel length,
.45
hc/a
Figure 9. Buckling load predictions from analytical model, local strip sandwich element model and full
panel sandwich element model.
19
American Institute of Aeronautics and Astronautics