Nasa - cr-1457 Manual For Structural Stability Analysis of Sandwitch Plates and Shells

https://ntrs.nasa.gov/search.jsp?
R=19700004831 2018-12-12T16:27:07+00:00Z
CONTRACTOR NASA CR-1457
_N A S A REPORT
_- i_ B
I
!z CASE Fi LE
MANUAL FOR STRUCTURAL

STABILITY ANALYSIS OF
I SANDWICH PLATES AND SHELLS
I by R. T. Sulli,is, G. IFC Smith, a,id E. E. Spier
i _Prepared b),
E7 :
_ GENERAL DYNAMICS CORPORATION
_:San Diego, Calif.

_r Manned Spacecraf Ce,zter
E NATIONAL AERONAUTICS AND SPACE ADMINISTRATION • WASHINGTON, D. C. • DECEMBER-1969
E_
k
_k
_____ ". _ -........ =..... -- •!
!
|
E
!
NASA CR-1457
MANUAL FOR STRUCTURAL STABILITY ANALYSIS
OF SANDWICH PLATES AND SHELLS
By R. T. Sullins, G. W. Smith, and E. E. Spier
Distribution of this report is provided in the interest of

information exchange. Responsibility for the contents
resides in the author or organization that prepared it.
Prepared under Contract No. NAS 9-8244 by

GENERAL DYNAMICS CORPORATION
San Diego, Calif.
for Manned Spacecraft Center
NATIONAL AERONAUTICS AND SPACE ADMINISTRATION
For sale by the Clearinghouse for Federal Scientific and Technical Information
Springfield, Virginia 22151 - Price $3.00
ACE
I
/
/ ttONEYCOMB..-'_'III Iil tli I _ //" //

/ _o_ _q-ll II tIL-_ / // /4
_ / /_D 1Rp_l
IBCB_]N);
CORE i
/
{
DEPTH ! / _ _t l--
RIBBON
DIRECTION
HONEYCOMB SANDWICH CONSTRUCTION
iii
ABSTRACT
The basic objective of this study was to develop and compile a manual which
would include practical and up-to-date methods for analyzing the structural
stabilityof sandwich plates and she]Is for typical loading conditions which
might be encountered in aerospace applications. The methods proposed for
use would include known analytical approaches as modified for correlation
with applicable test data.
The data presented here covers recommended design equations and curves
for a wide range of structural configurations and loading conditions, includ-
ing combined loads. In a number of cases, actual test data points are in-
cluded on the design curves to substantiate the recommendations made. For
those items where little or no test data exists the basic analytical approach
is presented along with the notation that this represented the 'best available"
data and should be used with some caution and judgment until substantiated
by test.
The following subjects are among those covered in the manual:
Local Instability
General Instability of Flat Panels
General Instability of Circular Cylinders
General Instability of Truncated Circular Cones
General Instability of Dome-Shaped Shells
Instability of Sandwich Shell Segments
Effects of Cutouts on the General Instability of Sandwich Shells
Inelastic Behavior of Sandwich Plates and Shells

TABLE OF CONTENTS
Section Page
INTRODUCTION ................. I-i
i. 1 GE 2:ERA_" ................. i-i
I. 2 FAILURE MODES ............. 1-4
LOCAL INSTABILITY ............... 2-1
2.1 INTRACELLULAR BUCKLING (Face Dimpling) 2-1
2. i. 1 Sandwich with Honeycomb Core ......... 2-1
2.1.2 Sandwich with Corrugated Core ......... 2-8
2.2 FACE WRINKLING ............. 2-21
2.2.1 Sandwich with Solid or Foam Core (Antisymmetric

Wrinkling) ................ 2-21
2.2.2 Sandwich with Honeycomb Core (Symmetric

Wrinkling) ................ 2-28
2.3 SHEAR CRIMPING ............. 2-37
2.3.1 Basic Principles ............. 2-37
2.3.2 Design Equations ............. 2 -40
GENERAL INSTABILITY OF FLAT PANELS ....... 3-1
3.1 RECTANGULAR PLATES .......... 3-1
3.1 1 General ................ 3-1
3.1 2 ITniaxial Edgewise Compression ........ 3-5
3.1 3 Edgewise Shear .............. 3-25
3.1 4 Edgewise Bending Moment .......... 3-37
3.1 5 Other Single Loading Conditions ........ 3-46
3.1 6 Combined Loading Conditions ......... 3-47
3.2 CIRCULAR PLATES ............ 3-73
3.2.1 Available Single Loading Conditions ....... 3-73
3.2.2 Available Combined Loading Conditions ...... 3-73
3.3 PLATES WITH CUTOUTS .......... 3-74
3.3.1 Framed Cutouts .............. 3-74
3.3.2 Unframed Cutouts ............. 3-75
4 GENERAL INSTABILITY OF CIRCULAR CYLINDERS .... 4-1
4.1 GENERAL ................ 4-1
4.2 AXIAL COMPRESSION ........... 4-5
4.2.2 Design Equations and Curves ......... 4-17
vii
TABLE OF CONTENTS, Cont'd.
Section Page
4.3 PURE BENDING .............. 4-25

4.4 EXTERNAL LATERAL PRESSURE ....... 4-32
4.5 TORSION ................ 4-47
4.6 TRANSVERSE SHEAR ............ 4-62
4.7 COMBINED LOADING CONDITIONS ....... 4-65
4.7.1 General ................ 4-65
4,7.2 Axial Compression Plus Bending ........ 4-67
4.7.3 Axial Compression Plus External Lateral Pressure. 4-71
4.7.4 Axial Compression Plus Torsion ........ 4-89
4.7.5 Other Loading Combinations .......... 4-95
GENERAL INSTABILITY OF TRUNCATED CIRCULAR

CONES .................... 5-1
5.1 AXIAL COMPRESSION ........... 5-1

5.2 PURE BENDING .............. 5-6
5.3 EXTERNAL LATERAL PRESSURE ....... 5-9
5,3.2 Design Equations and Curves ......... 5-12
5.4 TORSION ................ 5-14
5.5 TRANSVERSE SHEAR ............ 5-18
5.6 COMBINED LOADING CONDITIONS ....... 5-21
5.6.1 General ................ 5-21
5.6.2 Axial Compression Plus Bending ........ 5-23
viii
TABLE OF CONTENTS, Cont'd.
Section Page
5.6.3 Uniform External Hydrostatic Pressure ...... 5-27

5.6.4 Axial Compression Plus Torsion ........ 5-34
5.6.5 Other Loading Combinations .......... 5-39
GENERAL INSTABILITY OF DOME-SHAPED SHELLS .... 6-1
6.1 GENERAL ................ 6-1

6.2 EXTERNAL PRESSURE ........... 6-3

6.3 OTHER LOADING CONDITIONS ........ 6-15
INSTABILITY OF SANDWICH SHELL SEGMENTS ..... 7-1
7.1 CYLINDRICAL CURVED PANELS ........ 7-1

7. i. 1 Axial Compression ............. 7-1
7.1.2 Other Loading Conditions ........... 7-11
7.2 OTHER PANEL CONFIGURATIONS ....... 7-11
EFFECTS OF CUTOUTS ON THE GENERAL INSTABILITY

OF SANDWICH SHELLS .............. 8-1
INELASTIC BEHAVIOR OF SANDWICH PLATES AND

SHE L LS 9-1
9.1 SINGLE LOADING CONDITIONS ........ 9-1

9.1.2 Design Equations ............. 9-3
9.2 COMBINED LOADING CONDITIONS ....... 9-i0
9.2.2 Suggested Method ............. 9-13
ix
LIST OF FIGURES
Figure P age
1.1-1 Typical Sandwich Construction ............ 1-2
1.2-1 Localized Instability Modes ............. 1-5
1.2-2 Ultimate Failures Precipitated by Face Wrinkling ..... 1-5
1.2-3 Non-Localized Instability Modes ........... 1-6
2.1-1 Critical Stresses for Intracellular Buckling Under

Unia.xi al Compression ............... 2-3
Definition of Dimensions .............. 2-5
Chart for Determination of Core Cell Size Such That

Intracellular Buckling Will Not Occur ......... 2-6
2.1-4 Corrugation Configurations ............. 2-8
2.1-5 Buckling Modes ................. 2-10
2.1-6 Local Buckling Coefficient for Single-Truss-Core

Sandwich ................... 2-14

Sandwich ................... 2-15

Sandwich .................. 2-16
2.1-9 Local Buckling Coefficient for Double-Truss-Core

Sandwich ................... 2-17

Sandwich ................... 2-18

Sandwich ................... 2-19
2.1-12 Local Buckling Coefficient for Truss-Core

San&rich ................... 2-20
Typical Variation of Q vs. q ............. 2-24
Comparison of Theory vs Test Results for Face Wrinkling

in Sanchvich Constructions Having Solid or Foam Cores. 2-25
2.2-3 Parameters for Determination of Face Wrinkling in

Sandwich Constructions Having Solid or Foam Cores .... 2-27
xi
LIST OF FIGURES,Cont'd.
Figure Page
2.2-4 T_3_ical
DesignCurvesfor FaceWrinklingin Sanchvich
ConstructionsHavingHoneycomb
Cores......... 2-30
2.2-5 Comparisonof Theoryvs Test Resultsfor FaceWrinkling

in SandwichConstructionsllaving HoneycombCores ..... 2-31
2.2-6 Relationship of K6 to Honeycomb Core Properties (Fc/Ee)

and Facing Waviness Parameter (6/te) .......... 2 -34
2.2-7 Graphs of Equation (2.2-12) for the Wrinkling Stress of

Facings in San_vieh Constructions ttaving ttoncycomb
Cores .................... 2-35
2.3-1 Uniaxial Compression ............... 2 -40
2.3-2 Pure Shear .................. 2-41
3.1-1 Elastic Properties and Dimensional Notations for a

Typical Sandwich Panel .............. 3-4
3.1-2 K M for a Sandwich Panel with Ends and Sides Simply

Supported, Isotropie Facings, and Orthotropie Core,
(R = 0.40) ................... 3-10
3.1-3 K M for Sanchvich Panel with Ends and Sides Simply

Supported, Isotropic Facings, and Isotropic Core,
(R = 1.00) ................... 3-11
3.1-4 K M for Sandwich Panel with Ends and Sides Simply

Supported, Isotropic Facings, and Orthotropic Core,
(R = 2.50) ................... 3-12
3.1-5 K M for Sandwich Panel with Ends Simply Supported and

Sides Clamped, Isotropic Facings, and Orthotropic Core,
(R = 0.40) ................... 3-13

Sides Clamped, Isotropic Facings, and Isotropic Core,
(R = 1.00) ................... 3-14

Sides Clamped, Isotropic Facings, and Orthotropic Core,
(R = 2.50) ................... 3-15
xii
LIST OF FIGURES, Cont'd.
Figure Page
3.1-8 K M for Sandwich Panel with Ends Clamped and Sides Simply
(R = 0.40) ................... 3-16
Supported, Isotropic Facings, and Isotropic Core,
(R = 1.00) ................... 3-17
(R = 2.50) ................... 3-18
3.1-11 K M for Sandwich Panel with Ends and Sides Clamped,

Isotropic Facings, and Orthotropic Core, (R = 0.40) .... 3-19

Isotropic Facings, and Isotropic Core, (R = 1.00) ..... 3-20

Isotropic Facings, and Orthotropic Core, (R = 2.50) .... 3-21
3.1-14 K M for Simply Supported Sandwich Panel Having a

Corrugated Core. Core Corrugation Flutes are
Perpendicular to the Load Direction .......... 3-22
3.1-15 K M for Simply Supported Sandwich Panel Having a

Corrugated Core. Core Corrugation Flutes are
Parallel to the Load Direction ............ 3-23
3.1-16 KM o for Sandwich Panel With Isotropic Facings in

Edgewise Compression .............. 3-24
3.1-17 K M for a Sandwich Panel with All Edges Simply Supported,

and an Isotropic Core, (R = 1.00) ........... 3-29

and an Orthotropic Core, (R = 2.50) .......... 3-30

and with an Orthotropic Core, (R = 0.40) ........ 3-31

Isotropic Facings and Corrugated Core. Core Corrugation
Flutes are Parallel to Side a ............ 3-32
xiii
Figure Page
3.1-21 K M for a Sandwich Panel with All Edges Simt)ly Supported,

Isotropic Facings and Corrugated Core. Core Corrugation
Flutes are Parallel to Side b ............ 3-33
3.1-2'2 K M for a Sanchvich Panel with All Edges Clamped, Isotropic

Facings and Isotropic Core, (It = 1.00) ......... 3-34
"5.1-23 K M for a Sandwich Panel with All Edges Clamped, Isotropic

Facings and Orthotropie Core, (R = 2.50) ........ 3-35
3.1-24 K M for a Sanchvich Panel with All Edges Clamped, Isotropie

Facings and Orthotropic Core, (R = 0.40) ........ 3-36
3.1-25 K M for a Simply Supported Sanchvich Panel with an Isotropic

Core, (R - 1.00) ................ 3-42
3.1-26 K M for a Simply Supporled Sandwich Panel with an Orthotropic

Core, (H = 2.50) ................ 3-43
3 1-27 K M [or a Simply Supported San(kvich Panel with an Orthotropie

Core, (R = 0.40) ................ 3 -44
3 1-28 K M for a Simply Supported Sanchvieh Panel with Corrugated

Core. Core Corrugation Flutes Parallel In Side a ..... 3-45
3 1-29 Interacti,m Curve for a Honeycomb Core Sanchvich Panel
Subjected to Bimxial Compression ........... 3-59
3 1-30 Interaction Curve for a tloneyeomb Core Sandwich Panel
Subjected to Bending and Compression ......... 3-60
3 1-31 Interaction Curve for a Itoneyeomb Core Sand_vich Panel

Subjected to Compression and Shear .......... 3-61
3 1-32 Interaction Curve for a Honeycomb Core Sandwich Panel

Subjected to Bending and Shear ............ 3-62
,,'). 1-33 Buckling Coefficients for Corrugated Core Sandwich Panels

in Biaxial Compression (a/b = 1/2) .......... 3-63
3 1-34 Buckling Coefficients for Corrugated Core Sandwich Panels

in Biaxial Compression (a/b = 1.0) .......... 3-64
3.1-35 Buckling Coefficients for Corrugated Core Sandwich Panels

in Biaxial Compression (a/b = 2.0) .......... 3-65
xiv
LISTOF FIGURES,Cont'd.
Figure Page

Under Combined Longitudinal Compresmon and Shear with
3-66
Longitudinal Core (a/b = 1/2) ..........

Under Combined Longitudinal Compression and Shear with
Longitudinal Core (a/b = 1.0) .......... 3-67

Longitudinal Core (a/b = 2.0) .......... 3-68

Transverse Core (a/b = 1/2) .......... 3-69

Under Combined Longitudinal Compressloh and Shear with
Transverse Core (a/b = 1.0) .......... 3-70

Under Combined Longitudinal Compression and Shear with
3-71
Transverse Core (a/b = 2.0) ..........

Under Combined Longitudinal Compresmon, Transverse
3-72
Compression, and Shear with Longitudinal Core ......
4.1-1 Equilibrium Paths for Axially Compressed Circular

Cylinders ...................
Typical Equilibrium Paths for Circular Cylinders .....
Schematic Representation of Relationship Between K c and

V for 0 _ 1 .................. 4-7
C
4.2-2 Semi-Logarithmic Plot of _/c vs R/t' for Isotropic (Non-

4-9
Sandwich) Cylinders Under Axial Compression ......
4.2-3 Knock-Down Factor 3Zc for Circular Sandwich Cylinders

Subjected to Axial Compression ........... 4-10
4.2-4 Comparison of Proposed Design Criterion Against Test

Data for Weak-Core Circular Sandwich Cylinders Sub-
jected to Axial Compression ............. 4-12
XV
Figure Page
4.2-5 Comparison of Proposed Design Criterion Against a Test

Result for a Weak-Core Circular Sandwich Cylinder
4-13
Subjected to Axial Compression ...........
Stresses Involved in Interpretation of Test Data ...... 4-15
Buckling Coefficient for Axially Compressed Circular

Sandwich Cylinders ................ _-19
4.2-8 Design Knock-Down Factor for Circular Sandwich

4-20
Cylinders Subjected to Axial Compression ........
4.2-9 Buckling Coefficient for Short Simply-Supported Sandwich

Cylinders Subjected to Axial Compression (0 = 1) ...... 4-24
4.3-1 Knock-Down Factor _o for Circular Sandwich Cylinders

4-28
Subjected to Pure Bending ...........
4.3-2 Design Knock-Down Factor Mo for Circular Sandwich

4-30
Cylinders Subjected to Pure Bending ..........
4.4-1 Circular Sandwich Cylinder Subjected to External

Lateral Pressure .............. 4-32
4.4-2 Schematic Representation of Log-Log Plot of Cp Versus

L/R for Circular Sandwich Cylinders Subjected to
External Lateral Pressure ............. 4-36
4.4-3 Buckling Coefficients Cp for Circular Sandwich Cylinders

Subjected to External Lateral Pressure; Isotropic Facings;
Transverse Shear Properties of Core Isotropie or Ortho-
4-44
tropic; Vp = 0 .................

Transverse Shear Properties of Core Isotropic or Ortho-
4-45
tropic; Vp = 0.05 ................

4-46
tropic; Vp = 0.10 ...............
4.5-1 Circular Sandwich Cylinder Subjected to Torsion ...... 4-47
xvi
Page
Figure
4.5-2 Typical Log-Log Plot of the Buckling Coefficient K s for

4-50
Circular Sandwich Cylinders Subjected to Torsion .....
4.5-3 Buckling Coefficients for Circular Sandwich Cylinders

4-56
Subjected to Torsion ...............

4-57

4-58

4-59

4-60

4-61
4-66
Sample Interaction Curve ..............
Design Interaction Curve for Circular Sandwich Cylinders

4-68
Subjected to Axial Compression Plus Bending .......
4.7-3 Design Interaction Curve for Circular Sandwich Cylinders

4-70
Subjected to Axial Compression Plus Bending .......
4.7-4 Circular Sandwich Cylinder Subjected to Axial Compression

Plus External Lateral Pressure .......... 4-71
4.7-5 Typical Interaction Curves for Circular Sandwich Cylinders

Subjected to Axial Compression Plus External Lateral
4-74
Pressure ...................
4.7-6 Interaction Curves for Circular Sandwich Cylinders Subjected

4-79
to Axial Compression Plus External Lateral Pressure ....

to Axial Compression Plus External Lateral Pressure .... 4-80
4.7-8 Interaction Curve for Circular Sandwich Cylinders Subjected

4-81

4-82
xvii
Figure Page

to Axial Compresmon Plus External Lateral Pressure .... 4-83

to Axial Compress:on Plus External Lateral Pressure .... 4-84

to Axial Con:press:on Plus External Lateral Pressure .... 4-85

to Axial Con:press:on Plus External Lateral Pressure .... 4-86

to Axial Compresmon Plus External Lateral Pressure .... 4-87
4.7-15 Interaction C_rves for Circular Sandwich Cylinders Subjected

to Axial Compresswn Plus External Lateral Pressure .... 4-88
4.7-16 Circular Sandwich Cylinder Subjected to Axial Compression

Plus Torsion .................. 4-89
4.7-17 Conditional Interaction Curve for Circular Sandwich Cylinders

Subjected to Axial Compression Plus Torsion ....... 4-93
4.7-18 Conservative Interaction Curve for Circular Sandwich Cylinders
Subjected to Axial Compression Plus Torsion ....... 4-94
5.1-1 Empirical Knock-Down Factors ........... 5-2
5.1-2 Truncated Sandwich Cone Subjected to Axial Compression. 5-4
5.2-1 Truncated Sandwich Cone Subjected to Pure Bending ..... 5-7
5.3-1 Truncated Cone Subjected to Uniform External Lateral
Pressure • , • . • o . , . . . , . . . o . • . 5-9
5.3-2 Truncated Sandwich Cone .............. 5-12
5.4-1 Truncated Sandwich Cone Subjected to Torsion ...... 5-17
5.5-1 Truncated Cone Subjected to Transverse Shear ...... 5-18
5.6-1 Sample Interaction Curve .............. 5-21
5.6-2 Design Interaction Curve for Truncated Sandwich Cones
Subjected to Axial Compression Plus Bending ....... 5-26
XVIII
Figure P age
5.6-3 Truncated Cone Subjected to Uniform External Hydrostatic

Pressure ................... 5-27
Truncated Sandwich Cone .............. 5-31
Truncated Cone Subjected to Axial Compression Plus

Torsion ................... 5-34
5.6-6 Conditional Interaction Curve for Truncated Sandwich

Cones Subjected to Axial Compression Plus Torsion ..... 5-38
5.6-7 Conservative Interaction Curve for Truncated Sandwich

Cones Subjected to Axial Compression Plus Torsion ..... 5-39
5.6-8 Truncated Cone Subjected to Axial Compression Plus

5--40
Bending Plus Torsion .............
6.1-1 Structural Dome Shapes .............. 6-1
6.2-1 Sanchvich Dome Subjected to External Pressure ...... 6-3
6.2-2 Schematic Representation of Relationship Between Kc and V c. 6-7
6.2-3 Knock-Down Factor 7d for Sandwich Domes Subjected to

Uniform External Pressure ............. 6-9
6.2-4 Buckling Coefficient for Sandwich Domes Subjected to

External Pressure ............... 6-13
Cylindrical Panel and Associated Flat-Plate Configuration 7-2
Schematic Logarithmic Plot of Schapitz Criterion for Non-

Sandwich Cylindrical Skin Panels ........... 7-3
7.1-3 Schematic Logarithmic Plot of Test Data for Cylindrical

Isotropic (Non-Sandwich) Skin Panels Under Axial
Compression ................ 7-6
7.1--4 Graphical Representation of Equations (7.1-5) through

7-10
(7,1-8) ....................
xix
LIST OF TABLES
Table Page
2-1 Summaryof DesignEquationsfor Local Instability Modes
of Failure ................... 2-42
3'1 Summaryof DesignEquationsfor GeneralInstability of
Flat SandwichPanels ............... 3-76
4-1 Summaryof DesignEquationsfor Instability of Circular
Cylinders ................... 4-98
5-1 Summaryof DesignEquationsfor Instability of Truncated
Circular Cones................. 5-42
6-1 Summaryof DesignEquationsfor Instability of Dome-
ShapedShells.................. 6-16
7-1 Summaryof DesignEquationsfor Instability of Cylindrical,
CurvedPanels ................. 7-12
9-1 Recommended Plasticity ReductionFactorsfor Local
Instability Modes ................ 9-7
9-2 Recommended Plasticity ReductionFactorsfor the General
Instability of Flat SanchvichPlates........... 9-8
9-3 Reeommended Plasticity ReductionFactorsfor the General
Instability of Circular SandwichCylinders, Truncated
Circular SandwichCones,andAxisymmetricSandwich
Domes.................... 9-9
XX
LIST OF SYMBOLS
Panel length, inches. Major semi-axis of an ellipse, inches.
a Axial length of a cylindrical panel, inches.

R
a Length of the fiat panel shown in Figure 7. I-i, inches.

P
b Panel width, inches. Minor semi-axis of an ellipse, inches.
b Circumferential width of a cylindrical panel, inches.

R
bf Pitch of corrugated core, inches.
b Width of the flat panel shown in Figure 7. i-i, inches.

P
C Length parameter defined by Equation (4.7-25), dimensionless.

L
C Parameter defined by Equations (4.2-21) and (6.2-19), dimensionless.

O
C Buckling coefficient for sandwich cylinders subjected to external

P
lateral pressure, dimensionless.
_(Elt 1) (E2t2)h2
D Bending stiffness of sandwich wall or panel =
inch-lbs. )'(Eltl + E2t2)
D Shear stiffness of sandwich wall or panel = h2(G )/t , lb/inch.

q XZ C
d
Total thickness of sandwich wall or panel (d = t 1 + t2 + tc), inches.
Young's modulus, psi.
E Young's modulus of the core in the direction normal to the facings,

C
psi.
Young's modulus of facing, psi.

Ef
E Secant modulus of facing, psi.

S
Et Tangent modulus of facing, psi.
xxi
E 1, E 2 Young's moduli for facings I and 2 respectively, psi.
e Strain intensity defined by Equation (9.2-2), in./in.

i
F Transverse shear force, Ibs.

V
(Fv) Critical transverse shear force, ibs.

cr
F Flatwise sandwich strength (the 10wcr of flatwise core compressive,

C
flatwise core tensile, and flatwise core-to-facing bond strengths),
psi.
G Transverse shear modulus of core, psi.

e
G Core shear modulus associated with the plane perpendicular to the

ca
facings and parallel to side a of panel, psi.
Gcb Core shear modulus associated with the plane perpendicular to the
facings and parallel to side b of panel, psi.
Gf Elastic shear modulus of facing, psi.
G.. Core shear modulus associated with the plane perpendicular to the
lj
facings and parallel to the direction of loading, psi.
G Secant shear modulus of facing, psi.

S

XZ
facings and parallel to the axis of revolution of a cylinder, psi.

yz
axis of revolution of a cylinder, psi.
h Distance between middle surfaces of the two facings of a sandwich

construction, inches.
K Buckling coefficient for an isotropic (non-sandwich) flat plate,
dimensionless. Buckling coefficient for fiat rectangular sandwich
pane] under edgewise compression (Kc) , edgewise shear (Ks) , or
edgewise bending (Eb). K = KF + KM.
K Theoretical flat panel buckling coefficient which is dependent on

F
facing stiffness and panel aspect ratio, dimensionless.
xxii
K Theoretical fiat panel buckling coefficient which is dependent on
M sandwich bending and shear rigidities, panel aspect ralio, and
applied loading, dimensionless.
K Buckling coefficient for sandwich cylinders under axial compression

C
and sandwich domes under external pressure, dimenionlcss.
K I Buckling coefficient for short sandwich cylinders under axial com-

c
pression, dimensionless.
K Parameter defined by Equation (4.4-2), dimensionless.

P
K Buckling coefficient for sandwich cylinder subjected to torsion,

S
dimensionless.
K Parameter defined by Equation (2.2-4), dimensionless.

5
Buckling coefficient, dimensionless.
k Buckling coefficient associated with compressive stress acting in

X
the x direction, dimensionless.
Loading coefficient for applied compressive stress which is acting

X
in the x direction, dimensionless.
k Buckling coefficient associated with compressive stress acting in the

Y
y direction, dimensionless.
k I Loading coefficient for applied compressive stress which is acting in

Y
the y direction, dimensionless.
L Over-all length, inches.
L Effective length, inches.

e
M Applied bending moment, in-lbs.
M Critical bending moment, in-lbs.

cr
M. S. Margin of safety, dimensionless.
N Critical compressive running load, lbs/inch.

cr
Number of circumferential full-waves in the buckle pattern,

dimensionless.
xxiii
P Axial load, lbs.
pl Equivalent axial load defined by Equation (5.6-32), lbs.
p Critical axial load, lbs.

cr
Empirical lower-bound value for critical axial load when acting

(Per) Empirical alone, ibs.
P External pressure, psi.
P Critical value for external pressure, psi.

cr
Experimental value for critical external pressure, psi.

(Pc r ) Te st
Classical theoretical critical pressure for a cylinder subjected to

(P×tCL external pressure acting only on the end closures, psi.
External pressure acting only on the lateral surface of a cylinder,

Py
psi.
Classical theoretical critical pressure for a cylinder subjected to

(PY)cL
external pressure acting only on the lateral surface, psi.
Q The relative minimum, with respect to {, of expression (2.2-2),

dimensionless.
q Quantity defined by Equation (2.2-3), dimensionless.
R Degree of core shear modulus orthotropicity = G /G _,

dimensionless. Radius to middle surface, inehee, a CD
Stress ratio defined by Equation (4.7-9), dimensionless.

(Rb)cL
R Load, stress, or pressure ratios as defined in appropriate sections

C
of this handbook, dimensionless.
(Re) Stress ratios as defined in appropriate sections of this handbook,

CL dimensionless.
R Effective radius, inches.

e
xxiv
R.
Stress or load ratio for the particular type of loading associated with
1
the subscript i, dimensionless.
R Stress or load ratio for the particular type of loading associated with
J
the subscript j, dimensionless.
Radius to middle surface at the large end of a truncated conical shell,

Rlarge
measured perpendicular to the axis of revolution, inches.
R Maxium radius of curvature for middle surface of a dome-shaped

Max
shell, inches.
R Pressure ratios as defined in appropriate sections of this handbook,

P dimensionless.
Pressure ratio defined by Equation (4.7-15), dimensionless.

CL
R Load or stress ratios as defined in appropriate sections of this
s
handbook, dimensionless.

(Rs)cL
R Radius to middle surface at the small end of a truncated conical shell,

small
measured perpendicular to the axis of revolution, inches.
R Stress or load ratio corresponding to the x direction, dimensionless.

x
R stress or load ratio corresponding to the y direction, dimensionless.

Y
R Middle-surface radius of curvature in the plane perpendicular to the

2
meridian, inches.
r Parameter defined by Equation (4.2-37), dimensionless.

a
Cell size of honeycomb core, inches.
T External torque, in-lbs.
T Critical external torque, in-lbs.

er
(¥cr) Empirical lower-bound value for critical torque when acting alone,
Empirical in-lbs.
t Thickness, inches.
xxv
t Total thickness of the cylindrical panel shown in Figure 7.1-1, inches.
R
t Thickness of core (measured in the direction narmal to the facings),

e
inches.
t Thickness of a single facing, inches.

f
t Total thickness of the fiat panel shown in Figure 7.1-1, inches.

P
t Thickness of material from which corrugated core is formed, inches.

0
Thicknesses of the respective facings of a sandwich construction

t 1, t 2
(there is no preference as to which facing is denoted by the subscript
1 or 2), inches.
h2
U Sandwich transverse shear stiffness, defined as U = T-- Ge _ hGe'
lbs. per inch. e
2
lr D
_7 Bending and shear rigidity parameter which is defined as V -
dimensionless, b2U
Parameter defined in Sections 4.2 and 6.2, dimensionless.

C
_T Parameter defined by Equation (4.4-4), dimensionless.

P
_7 Parameter defined by Equation (4.5-4), dimensionless.

S
V Parameter defined by Equation (4.7-13), dimensionless.

XZ
V Parameter defined by Equation (4.7-14), dimensionless.

yz
W Bending and shear rigidity parameter for fiat sandwich panels with
?T2tc (Eltl)(E2t2) T?
corrugated core which is defined as W
dimensionless. Xb 2 Gcb(Eltl+E2t2 )
W Running compression load, lbs/inch.

C
z Length parameter defined by Equation (4.2-33), dimensionless.
z Length parameter defined by Equation (4.5-3), dimensionless.

S
xxvi
Angle of rotation at appropriate joint in corrugated-core sanchvich
construction (sec Figure 2.1-5), degrees. Vertex half-angle of
conical shell, degrees.
Angle of rotation at appropriate joint in corrugated-core sandwich

construction (see Figure 2.1-5), degrees.
7 Knock-down factor, dimensionless. Ratio = _y/a x, dimensionless.
Knock-down /actor associated with general instability under pure

bending, dimensionless.
Knock-down factor associated with general instability under axial

"Ye
compression, dimensionless.
Knock-down factor associated with the general instability of a dome-

Td
shaped shell under external pressure, dimensionless.
Knock-down factor determined from a test specimen subjected to the

(Ti)Test
loading condition corresponding to the subscript i, dimensionless.
Knock-down factor associated with general instability of a cylinder

under uniform external lateral pressure, dimensionless.
Knock-down factor associated with general instability under pure

torsion, dimensionless.
Amplitude of initial waviness in facing, inches.
E Normal strain in the x direction, in/in.

x
E Normal strain in the y direction, in/in.

Y
E Shear strain in the xy plane, in/in.

xy
Parameter involving the core elastic moduli, core thickness, and

buckle wavelength, dimensionless.
Plasticity reduction factor, dimensionless.
Plasticity reduction factor corresponding to an experimental critical

_Test
stress value, dimensionless.
(1 - PaPb ) = (1 - p_) for isotropic facings, dimensionless. Ratio =
r/_×, dimensionless.
xxvii
Ratio of transverse shear moduli of core Esee Equation (4.2-1)],
dimensionless.
Actual Poisson's ratio of facing, dimensionless.
v Elastic Poisson's ratio of facing, dimensionless.

e
Radius of gyration for shell wall of sandwich and non-sandwich

constructions (p _ h/2 for sandwich constructions whose two
facings are of equal thickness), inches.
(l Stress, psi.
_b Peak compressive stress due solely to an applied bending moment,

psi.
Classical theoretical value for the critical peak compressive stress

((_b)c L
under a bending moment acting alone, psi.
_CL Classical value of critical stress, psi.
c Uniform compressive stress due solely to an applied axial load,

psi.
(_!
e Effective compressive stress defined by Equation (4.7-38), psi.
Peak axial compressive stress due solely to an applied bending

((_c)b
moment, psi.
(%) Uniform axial compressive stress due solely to an applied axial

c load, psi.
(_c)CL Classical theoretical value for the critical uniform compressive

stress under an axial load acting alone, psi.
(Y Critical stress, psi.

cr
cr Experimental critical stress obtained from a particular test speci-

test men, psi.
!
(I
cr Ex_perimental critical stress which would have been attained had
test the test specimen remained elastic, psi.
xxviii
ff Critical value for the compressive stress acting in the x direction,
cr
x psi.
Compressive stresses in facings 1 and 2, respectively, in the pres-

ence of the critical loading for general instability (there is no prefer-
ence as to which facing is denoted by the subscript 1 or 2), psi.
Uniaxial compressive stress at which shear crimping occurs in sand-

crimp
wich constructions, psi.
Stress intensity defined by Equation (9.2-1), psi.

1
a ttoop membrane stress, psi.

H
Meridional membrane stress, psi.

M
Maximum possible critical stress corresponding to a particular

{YMax
material, psi.
Minimum value of stress for the post-buckling equilibrium path, psi.

O'MI N
Predicted value for critical stress, psi.

predicted
o" Critical buckling stress for a flat plate, psi.

P
Critical buckling stress for a complete cylinder, psi.

a R
(y Facing wrinkling stress, psi.

wr
(y Stress acting in the x direction, psi. Uniform axial compressive

X
stress due to an applied axial load, psi.
0 -I Effective compressive stress defined by Equation (4.7-37), psi.

X
(ax) Peak axial compressive stress due solely to an applied bending

b moment, psi.
Classical theoretical value for critical uniform axial compressive

CL stress when acting alone, psi.
(z x) Uniform axial compressive stress due solely to an applied axial

C load, psi.
Stress acting in the y direction, psi.

Y
xxix
T Shear stress, psi.
I
T Effective shear stress defined by Equation (4.7-39), psi.
Classical theoretical value for critical uniform shear stress when

(T)C L
acting alone, psi.
T Critical shear stress, psi.

cr
I
T Critical shear stress for an equivalent cylinder subjected to an
cr
applied torque, psi.
T Pure shear stress, acting coplanar with the facings, at which shear
crimp
crimping occurs in sandwich constructions, psi.
Uniform shear stress due solely to an applied torque, psi.

TT
Peak shear stress due solely to an applied transverse shear force,

T V
psi.
Angular dimension of corrugated core (see Figure 2.1-4), degrees.

Quantity defined by Equation (4.2-10), dimensionless.
Angle of rotation at appropriate joint in corrugated-core sandwich

construction (see Figure 2.1-5), degrees. Parameter defined by
Equation (4.4-3), dimensionless.
XXX
CONVERSION OF U.S. CUSTOMARY UNITS TO THE
INTERNATIONAL SYSTEM OF UNITS 1
(Reference: MIL-HDBK-23)
U.S. Customary Conversion

Quantity Unit Factor 2 SI Unit
lbm/in. 27.68 × 103 kilograms/meter3 (kg/m33)

Density
lbm/ff 3 16.02 kilograms/meter ;_ (kg/m)
meters (m)
Length ft 0.304854
in. 0.02 meters (m)
Stress psi 6. 895 x 103 newtons/meter 2 (N/m 2)
lb/in. 6. 895 x 103 newtons/meter 2 (N/m 2)

Pressure
lb/ft 2 47.88 newtons/meter 2 (N/m 2)
3
IElastieity psi 6. 895 x 10 newtons/meter 2 (N/m 2)
Moduli _ Rigidity
Tempe r atu re (° F + 460) 5/9 degrees Kelvin (°K)
The rmal conductivity Btu in./hr ft 2 ° F 0. 1240 kg cal/hr m °C
Prefixes to indicate multiples of units are as follows:
Prefix Multiple
giga (G) 109
mcga (M) 10 6
kilo (k) 103

-3
milli (m) 10
micro _) 10 -6
1
The International System of Units [Syste'me International (S1)] was adopted by the
Eleventh General Conference on Weights and Measures, Paris, October 1960, in
Resolution No. 12.
2
Multiply value given in U.S. Customary Unit by conversion factor to obtain
equivalent value in SI unit.
xxxi
1
INTRODUCTION
]. 1 GENERAL
This handbook presents practical methods for the structural stabilityanalysis of
sandwich plates and shells. The configurations and loading conditions covered here
are those which are like]y to be encountered in aerospace applications. Basic equa-
tions, design curves, and comparisons of theory against test data are included.
For the purposes of this handbook, a structural sandwich is defined as a layered
construction formed by bonding two thin facings to a comparatively thick core as
depicted in Figure 1.1-1. The facings provide practically all of the over-all bending
and in-plane extensional rigidity to the sandwich. The core serves to position the
faces at locations removed from the neutral axis, provides virtually all of the trans-
verse shear rigidity of the sandwich, and stabilizes the facings against local buckling.
Thus the structural sandwich concept is quite similar to that of a conventional I
beam. The sandwich core plays a role which is analogous to that of the I beam web
while the sandwich facings perform a function very much like that of the I beam
flanges. The primary difference between these two types of construction lies in the
Numbers in brackets [ ] in the text denote references listed at end of each major
section (1; 2; etc.).
1-1
fact that the transverse shear deflections are usually significant to the sandwich
behavior; whereas, for I beams, these deflections are only important for the special
case of relatively short, deep beams.
FACING"
J
Figure 1.1-1. Typical Sandwich Construction
The sandwich is an attractive structural design concept since, by the proper choice
of materials and geometry, constructions having high ratios of stiffness-to-weight
can be achieved. Since rigidity is required to prevent structural instability, the
sandwich is particularly well suited to applications where the loading conditions are
conducive to buckling.
The use of sandwich construction in aerospace vehicles is certainly not a recent
innovation. The British de Havilland Mosquito bomber of World War II employed
1-2
structural sandwich throughout the airframe. In this case, the sandwich was in the
form of birch face sheets bonded to a balsa wood core. Many other airplanes, includ-
ing the B-58, B-70, F-Ill, C-5A, etc., have taken advantage of the high strength-to-
weight ratio enjoyed by sandwich construction. Space vehicle applications have
included the Apollo spacecraft, the Spacecraft LM Adapter (SLA) fairings on the
Centaur and other launch vehicles, as well as propellant tank bulkheads.
In view of the ever increasing application of structural sandwich, it has become desir-
able to assemble a handbook which presents latest design and analysis criteria for the
stability of such construction. The practicing designer and stress analyst need this
information in a form suitable for easy, rapid use. This document is meant to fulfill
that need. However, it should be kept in mind that, in many areas, all practical
problems have not yet been fully resolved and one can only employ what might be re-
ferred to as a "best-available" approach. In these cases it is advisable to supplement
numerical computations with suitable testing. Such areas of uncertainty are identified
in this handbook in the sections dealing with the appropriate configurations and loading
conditions.
In the sections to follow a discussion is given of the basic principles behind the design
equations along with conclusions derived from an analysis of available test data. This
is followed by the design equations along with any limitations on their use. Also, to
facilitate their use, a table of these equations and restrictions immediately precedes
the list of references in Sections 2, 3, 4, and 5 since these sections cover a wide
range of loading conditions and considerations.
1-3
1.2 FAILURE MODES
Structuralinstability of a sandwichconstructioncanmanifestitself in a numberof
different modes. Thevariouspossibilities are as describedbelowandas shownin
Figures1.2-1 through1.2-3.
Intracellular Buckling (Face Dimpling) - This is a localized mode of instability
which occurs only when the core is not continuous. As depicted in Figure 1.2-1, in
the regions directly above core cells (such as those of a honeycomb core), the
facings buckle in plate-like fashion with the cell walls acting as edge supports. The
progressive growth of these buckles can eventually precipitate the buckling mode
identified below as face wrinkling.
Face Wrinkling - This is a localized mode of instability which manifests itself in the
form of short wavelengths in the facings, is not confined to individual cells of
cellular-type cores, and involves the transverse (normal to facings) straining of the
core material. As shown in Figure 1.2-1, one must consider the possible occurrence
of wrinkles which may be either symmetrical or antisymmetrical with respect to the
middle surface of the original undeformed sandwich. As shown in Figure 1.2-2,
final failure from wrinkling will usually result either from crushing of the core,
tensile rupture of the core, or tensile rupture of the core-to-facing bond. However,
if proper care is exercised in the selection of the adhesive system, one can reason-
ably assume that the tensile bond strength will exceed both the tensile and com-
pressive strengths of the core proper.
1-4
A - Intraceilular Buckling (Face Dimpling)
SYMMETRIC ANTISYMMETRIC
B - Face Wrinkling
C - Shear Crimping
Figure 1°2-1, Localized Instability Modes
A - Core Crushing
B - TensileRupture C - Tensile Rupture

of Bond of Core Proper
Figure 1.2-2. Ultimate Failures Precipitated by Face Wrinkling
1-5
i
,ut|
,.,_|
"_Sl
I
.mm_'
..--m_
D"--'4
L_
C
o=
Z
4_
1-6
Shear Crimping - Shear crimping is often referred to as a local mode of failure but
is actually a special form of general instability for which the buckle wavelength is
very short due to a low transverse shear modulus for the core. This phenomenon
occurs quite suddenly and usually causes the core to fail in shear; however, it may
also cause a shear failure in the core-to-facing bond. Crimping will sometimes
occur in cases where relatively long-wave general instability first develops. In such
instances the crimp appears because of severe local transverse shear stresses at
the ends of buckle patterns. As the crimp develops, the general buckle may dis-
appear and a post-test examination would then lead to an erroneous conclusion as to
the mechanism which initiated failure.
General Instability - For configurations having no supplementary stiffening (such as
rings) except at the boundaries, the general instability mode is depicted in Figure
1.2-3A. The phenomenon involves over-all bending of the composite wall coupled
with transverse (normal to facings) shear deformations. Usually, transverse exten-
sional strains do not play a significant role in this behavior. Whereas intracellular
buckling and wrinkling are localized phenomena, general instability is of a more
gross nature. Except for the special case cited under the identification "Shear
Crimping", the wavelengths associated with general instability are normally con-
siderably larger than those encountered in intracellular buckling and face wrinkling.
For configurations having supplementary stiffening at locations other than the bound-
aries, the term general instability takes on new significance and reference is also
made to an additional mode identified as panel instability. For this case, general
1-7
instability is as definedabovebut with the addedprovisionthat thebucklepattern
involvessimultaneousradial displacementof boththe sandwichwall andthe inter-
mediatestiffeners. As shownin Figure 1.2-3B, the appropriatehalf-wavelength

of
thebucklepatternmustthereforeexceedthe spacingbetweenintermediatestiffeners.
Theexampleusedin Figure 1.2-3B is thatof a sandwichcylinder stiffenedby a
seriesof rings whichhaveinsufficientstiffnessto enforcenodalpointsat their re-
spectivelocations.
Panel Instability - This mode of instability applies only to configurations which have
supplementary stiffening at locations other than the boundaries. Figure 1.2-3C
depicts this mode by again using the example of a sandwich cylinder stiffened by a
series of rings. However, in this case the rings have sufficient stiffness to enforce
nodal points at their respective locations. The rings experience no radial deforma-
tion. Therefore, the half-wavelength of the buckle pattern cannot exceed the spacing
between rings. As in the case of general instability, this mode involves over-all
bending of the composite wall coupled with transverse shear deformations. Here
again, transverse extensional strains do not play a significant role in the behavior.
1-8
2
LOCAL INSTABILITY
2.1 INTRACELLULAR BUCKLING (Face Dimpling)
2.1.1 Sandwich with Honeycomb Core
2.1.1.1 Basic Principles
From a practical viewpoint, intracellular buckling can be regarded as flat-plate
behavior. Even where curvature is present, as in the cases of cylinders and spheres,
the honeycomb core cell size will normally be sufficiently small to justify such an
assumption. As noted from Reference 2-1, the critical stress for flat plates can be
expressed in the form
(2.1-1)
Crcr - 12(1-v0) _s
whe re
%r = Critical compressive stress, psi.
k = Coefficient which depends on the plate geometry, boundary

conditions, and type of loading, dimensionless.
= Plasticity reduction factor, dimensionless.
Ef = Young's modulus (for facing material in the case of intra-

cellular buckling), psi.
ve = Elastic Poisson's ratio (for facing material in the case of

intracellular buckling), dimensionless.
2-1
tf = Thicknessof plate (Facingthicknessin the caseof intra-
cellular buckling), inches.
s - A selectedcharacteristic dimensionof the plate, inches.
It is convenienthereto combineseveralof the constantsin Equation(2.1-1) to obtain
r/Ef _tf_ _ (2.1-2 )

O_r K (1- Vea ) \s/
or
%r (l-re)
= K (2.1-3)
_TEf
To apply these equations to the case of intracellular buckling, it is only necessary to
define the dimension s and establish a corresponding value for K. In Reference
2-2, Norris took s to be equal to the honeycomb core cell size. By convention,
this is taken equal to the diameter of the largest circle that can be inscribed within
the cell. Based on the analysis of test data, Norris then chose K = 2.0 for the
case of uniaxial compression. This provides a reasonably good fit to the test results
as shown in Figure 2.1-1 which was taken directly from Reference 2-2. It should be
noted that the choice of K = 2.0 does not provide a lower bound to the data. Six of
the test results fall significantly below the values predicted by the recommended
formula. This situation can be tolerated since the dimpling of several cells in a
honeycomb sandwich construction will not lead to catastrophic failure so long as a
sufficiently large number of cells remain unbuckled. As indicated by the scatter in
Figure 2.1-1, one could reasonably expect the majority of unbuckled cells to possess
considerably greater buckling strengths than would be indicated by the proposed
design curve. Under these conditions, some redistribution of stress would occur
2-2
0.1
Theoryfor Simply
_Suppor_d SquarePla_
0.01
Legend:
Dimplingin Elastic Range
• Test Datafor SpruceCorewith
SingleCircular Hole
• Test Data for Honeycomb Core
Dimpling Beyond Elastic Range
O Test Data for Spruce Core with
Single Circular Hole
[2 Test Data for Honeycomb Core
0.001
0.0005
0.01 1.0
Figure 2.1-1. Critical Stresses for Intracellular Buckling

Under Uniaxial Compression
2-3
but the structure couldcontinueto supportthe appliedload. In addition, it is pointed
out that thedimpledregionsretain significantpost-bucklingload-carryingcapability
sincetheybehaveessentiallyas flat plates. This doesnot mean,however,that one
canpermit the dimplesto growwithoutbound. The pointcanbe reachedwherethese
deformationsprecipitatewrinkling andthis cannotbe tolerated.
It is alsoof importanceto noteherethat mostof the test datashownin Figure 2.1-1
wereobtainedfrom sandwichplateshavinga solid sprucecore throughwhicha
singlecircular holewasdrilled to representa core cell. It is questionable

that
suchspecimenstruly simulatethe cell edgesupportlikely to be encounteredin
practical honeycomb
configurations. Onlythree datapointswereobtainedfor speci-
mensactuallyhavinghoneycomb
coresand, as shownin Figure 2.1-1, thesepoints
lie in the lower regionof the total bandof scatter.
In viewof theforegoingdiscussion,it is evidentthat the useof Equation2.1-3
togetherwith the selectionof K = 2.0 is certainly not a rigorous approach to the
analysis of intracellular buckling. However, until further work is accomplished in
this area, it is recommended that this criterion be employed as a 'best-available",
approximate design tool.
2-4
2.1.1.2 DesignEquationsandCurves
Thefacingstress atwhich intracellular bucklingwill occur underuniaxialcompres-
sionis givenby thefollowing semi-empiricalformula:
_cr = 2.0 _Ef

(l_Pe_) (_)_ (2.1-4)
The dimension s is the diameter of the largest circle that can be inscribed within
the cell shape. For example, in the cases of hexagonal and square cells, s is
measured as shown below.
@
Figure 2.1-2. Definition of Dimension s
Solving Equation (2.1-4) for s gives the result
(2.1-5)
s = tf ,,_ [_cr _(1- Ve_)]"j -g
This equation may be used to determine the maximum permissible cell size corre-
sponding to particular facing materials and thicknesses. Figure 2.1-3 presents a
family of plots of Equation (2.1-5) for selected values of tf ranging from tf =
0.001 to tf = 0.100.
For elastic cases, use _? = 1. Whenever the behavior is inelastic, the methods of
Section 9 must be employed.
2-5
2.00
1.00
0.80
0.60
0.40
0.30
0.04
0.03
0.02
1 2
"0Ef
Figure 2.1-'3. Chart for Determination of Core Cell Size Such That
Intracellular Buckling Will Not Occur
2-6
When the facings are subjected to biaxial compression, it is recommended that one
use the interaction formula
R x + Ry = 1 (2.1-6)
whe re
Applied Compressive Loading]

in Subscript Direction J
Ri = [Critical Compressive Loading (when] (2.1-7)
[acting alone) in Subscript DirectionJ
This straight-line interaction relationship is based on the information provided in
Reference 2-1 for square flat plates. For cases involving shearing stresses which
are coplanar with the facings, it is recommended that the principal stresses first be
computed and that these values then be used in the above interaction equation. When-
ever one of the principal stresses is tensile and the behavior is elastic, the analysis
should be based on the assumption that the compressive principal stress is acting
alone.
2-7
2.1.2 Sandwich
With CorrugatedCore
2.1.2.1 BasicPrinciples
This sectiondealswith corrugated-coresandwichconstructionswhosecross sections
maybe idealizedas shownin Figure 2.1-4. For cylinders, the only case treated
here is that where the axis of the corrugations is parallel to the axis of revolution.
For flat plates, however, the corrugations can be oriented in either the longitudinal
or transverse directions.
Single-Truss Doub le-T russ
Figure 2.1-4. Corrugation Configurations
Each of the following loading conditions is considered:
a. Uniaxial compression acting parallel to the axis of the corrugations.
b. Uniaxial compression acting parallel to the facings but normal to the

axis of the corrugations.
c. Biaxial compression resulting from combinations of a and b above.
The design curves presented here are taken directly from Reference 2-3 and are
based entirely on theoretical considerations. No comparisons are made against test
2-8
datato confirm the validity of thesesolutions. Until suchsubstantiationis obtained,
the recommended
designcurvescanonlybe consideredas a 'best-available"criterion.
It is pointedout, however,thatthere doesnot appearto be anyreasonto suspectthat
test datawoulddisagreewith the curves.
AlthoughReference2-3 is devotedsolelyto flat plates, the results are consideredto
be applicableto thecylindrical configurationsshownin Figure 2.1-4 sincethe dimen-
sions bf will usuallybe smallwith respectto the radius. Undersuchconditions,
curvatureinfluenceswill be negligible.
Thetheoreticaldevelopmentincludesconsiderationof eachof the bucklingmodes
shownin Figure 2.1-5. Both of thefollowing possibilities are covered:
a. Theface sheetsare the unstableelementsandare restrainedby the core.
b. Thecore is the unstableelementandis restrainedby theface sheets.
Bucklingis assumedto be accompanied

by rotationof the joints but with nodeflection
of the joints. Theanglesbetweenthe various elementsat anyonejoint are takento
remain unchanged
duringbuckling. It is alsoassumedthatthe over-all sandwich
dimensionsare sufficientlylarge suchthatendeffectsare negligible.
2-9
cl -oL Q ci 01
Clomped
oL o_ cl ol ol
,gl o. -el. i::i -oI

-cl ci -_ Cl -cl cl
rz _1 i1 ol cl
-_ cl -cl ol -cI,
-el - rx rl -_ _1 cl
__ply t
Single-T russ-Core Double-Truss-Core
(a, fl, and q_ denote angles of rotations at the appropriate joints)
Figure 2.1-5. Buckling Modes
2-10
Thetheoreticalstress at whichintracellular bucklingof thefacingsor bucklingof the
corrugatedcore will occuris givenby the followingformula:
kl zru 77E /tf_ _ (2.1-8)

°_cr - 12(1-V0) \bf]
where
%r = Critical compressive stress, psi.
ki = Coefficient which depends upon the geometry and loading

conditions, dimensionless.
_7 = Plasticity reduction factor, dimensionless.
E = Young's modulus of facings and core, psi.
re = Elastic Poisson's ratio of facings and core, psi.
tf = Facing thickness, inches.
bf = Pitch of corrugated core (see Figure 2.1-4), inches.
The only case considered here is that where the two facings are of the same thickness
and the entire sandwich construction (facings and core) is made of a single material.
Figures 2.1-6 through 2.1-12 give values for k i for each of the following loading
combinations:
a. kx when k_ = 0
b. kx when k_ = 0.5
c. kx when k_ = 1.0
d. ky when kx = 0
The coefficients k_ and k_ are defined as follows:
2-11
12(1-Ve _ ) (bf._ _
kx - rr _rIE _-f_ (Applied Compressive ¢rx) (2. I-9)
ky - 12(1- re2 ) (Applied Compressive O'y) (2.1-10)

_ 17E
The subscript x (for k and k') is used to identify cases where the loading is
directed along the axis of the corrugations (x direction). The subscript y (for k
and k') is used to identify cases where the loading is acting in the y direction which
is parallel to the facings but normal to the axis of the corrugations. For combinations
a through c, separate plots are furnished for single-truss-core and double-truss-
core configurations. For combination d, a single family of curves covers both
arrangements since all of the corresponding applied load is transferred through the
facings. The dashed lines in Figures 2.1-6 through 2.1-11 divide the charts into two
regions. Above the dashed lines, the face sheets are the unstable elements and are
restrained by the core. Below the dashed lines, the core is unstable and is restrained
by the face sheets.
To clarify the design charts given in Figures 2.1-6 through 2.1-12, the following
additional definitions are provided:
to = Thickness of material from which the corrugations are formed

(see Figure 2.1-4), inches.
_b = Angle shown in Figure 2.1-4, degrees.
In addition, the sample problem given below should be helpful to the user of this
handbook.
2-12
Given: SampleProblem Data for Single-TrussCore Type SandwichPanel
E = 30 x 106p si to = .016" bf = .700"
v e = .30 tf = .020" ¢ = 65 °
Proportional Limit a= 90,000 psi (ry = 16,300 psi (Compression)
Required: Find acr x ; Assuming _? = 1, one obtains
12ay(1-re2) _k_2 = 12 x 16,300 x ,910 _.70012

k_ = 7r2_E \tf / 9.87 x 1 x 30 x I0' \.020/ = 0.736
%=.o16 = .800
tf .020
Using linear interpolation between values given on Figures 2.1-7 and 2.1-8 one
obtains k x = 2.68.
Hence, the critical stress in the x direction (parallel to the corrugation axis) is
kx _2_ E /tf_ 2
aCrx - 12(1-Ve2 ) _bf/
and, assuming _ = 1, one obtains
_Crx 2,68 x 9,87 x 1 x 30 x 106f.O20h 2

= 12 x . 910 \. 700/ = 59,300 psi (Compression)
The stress intensity ai (See Section 9) can now be computed as follows:
o"i = _/O-x2 + O'y2 - ax(_y + 31"2
= 103_ f (59.3) 2 + (16.3) 2 - (59.3 x 16.3) + 0 = 53,100 psi
Since this value is below the proportional limit, the assumption _ = 1 is valid.
In cases where the qi value exceeds the proportional limit, the methods of Section
9 must be employed.
2-13
o
E_
I
_d
!
_4
2-14
0
hL
.E
2-15
r..
<.i.-.i
H
2
r,D
I.-i
.,.-I
,,,.,
2-16
0
= 2
_9
_D t_ re) _xj 0
2-17
=1
,g
11
eo
".>
.,,-i
2-18
[
ll[i
:iiiiiii
,.
I 1ii
1_]
;I
iiiitiii
:_!!!!!!!
:::::::::
iiiiiiiii
.'__!!!!!!!
!N!!iiJi
iiNiiiii
H
Mili_iii
J[l{I {1ill
'_ILI ] I IAI I I
I%J I I ]/I_l I I
ll_lllll
[ I I_I I Ii I
t I i_i_ll I [nlL[
i [l I kl l_l
t'_ll 1 IN ]l
I_ I I |k] i
U I I_1 II Ill
i 1 i ll_i 1 111
I I I • II III
! [_1 ] /!/
t i lil.l i i i i
!11]_1|1
l_Jl 1 I Illi 1
_ i i 1 IRl.Jll i I
flllll_ I ]
'WL] I ] I II{ J {
iiiiiiiii
iiiiiii;.:
,,!!144_
2-19
_:..i
.......l....
I .... H
__._-_-- --
.... -+ : ....... ZZ<01 _....
C_
I
r_
i::::b::: .... .... "X
..... + ............... _. _
i_ f_J c<J
2-20
2.2 FACE WRINKLING
2.2.1 Sandwich With Solid or Foam Core (Antisymmetric Wrinkling)
The problem of face wrinkling has been treated by many investigators dating back as
far as 1940. The most important publications on this subject are listed as References
2-4 through 2-14. For the purposes of this handbook, it was decided that the results
in References 2-7 and 2-9 would be the most useful. The latter applies only to sand-
wich configurations which have solid or foam cores. The development there includes
consideration of both the symmetric and antisymmetric modes along with the influences
from initial waviness of the facings. It is pointed out that, when the core is sufficiently
thick, the wrinkle patterns of the two facings will be independent of each other and the
same critical load is obtained for the symmetric and antisymmetric modes. However,
for sandwiches having thinner cores, the core strains introduced by one facing influ-
ence the wave pattern in the other facing. Under these conditions, it was found that
sandwiches having solid or foam cores can be expected to wrinkle antisymmetrically.
The following governing equation was derived to predict this form of wrinkling for
isotropic facings subjected to uniaxial compression:
c_ _r]Ef Ec Gc_ _
{rwr : _L (1-Pea)_ (2.2-1)
whe re,
Crwr = Facing wrinkling stress, psi.
17 = Plasticity reduction factor, dimensionless.
Ef = Young's modulus of facing, psi.
2-21
E c = Young's modulus of the core in the direction normal to the facings, psi.
Gc Core shear modulus associated with the plane perpendicular to the

facings and parallel to the direction of the applied load, psi.
_e - Elastic Poisson's ratio of facings, dimensionless.
The quantity Q is the relative minimum, with respect to _ , of the expression
_
30V ÷ 16q ( cosh
sinh_- 1
5)
(2.2-2)
1 4 6.4 KS_ ( 11cosh_-

sinh _4 1 5 )
whe re
tc I (1-Ve_)13
q = _f Gc LrlEf Ec Gc] (2.2-3)
6Ec (2.2-4)
K8 - tc Fc
and
- Parameter involving the core elastic moduli, core thickness, and

buckle wavelength, dimensionless.
tc Thickness of core, inches.
tf : Thickness of facing, inches.
6 : Amplitude of initial waviness in facing, inches.
Fc : Flatwise sandwich strength (the lower of flatwise core compressive,

flatwise core tensile, and flatwise core-to-facing bond strengths),
psi.
The initial waviness plays an important role in the wrinkling phenomenon since it
causes transverse facing deflections to develop even when the applied loading is very
small. As the load increases, these deflections grow at steadily increasing rates and
lead to transverse tensile or compressive failure of the core or tensile rupture of the
2-22
core-to-facingbond. Thesefailures occur, of course, at loadvaluesbelowthe pre-
dictionsfrom classical theoryin whichit is assumedthat thefacingsare initially
perfect (K6 = 0).
The results from Reference 2-9 can be summarized in the form of Equation (2.2-1)
accompanied by plots of Q vs q with K8 as a parameter. A family of such curves
is given in Reference 2-9 and they are of the general shape shown in Figure 2.2-1.
The limiting values established by the straight line 0A correspond to the shear
crimping mode of failure (see Section 2.3). All other points on the curves al"e for
antisymmetric wrinkling. In actual practice, curves of this type do not prove to be
very helpful since the K 8 values appropriate to particular structures are rarely
known. Therefore, in order to provide a practical means for the prediction of face
wrinkling in sandwich constructions having solid or foam cores, it has become com-
mon practice to select a single conservative lower-bound Q based on available test
data. This approach is followed here. Elastic test data selected from Reference 2-9
are plotted in Figure 2.2-2 from which the value Q = 0.50 has been selected as a
safe design value. Additional data are given in Reference 2-6 which are not shown
here but lead to the same value for a lower-bound Q. This is in conformance with
the observation made by Plantema in Reference 2-15 that the value Q = 0.50 has
often been recommended for practical design purposes. However, since much of the
existing test data were obtained from specimens that were not very representative of
configurations likely to be encountered in realistic structures, the selection of Q = o. 50
can only be regarded as a "best-available" approach. In view of the uncertainties
2-23
K6 = 0
Q K 8 = Constant
K 6 = Constant
0
q
Figure 2.2-1. Typical Variation of Q vs. q
involved, it is recommended that for the verification of final designs, wrinkling tests
be performed on specimens which are truly representative of the actual configuration.
The method presented here for the prediction of wrinkling should only be regarded as
an approximate guideline.
2-24
o
.i
0
O
°,-_
0 ¢.j
Qo ¢.0 5'q
t
2-25
2.2.1.2 Design Equations and Culwes
The following equation may be used to compute the approximate uniaxial compressive
stress at which face wrinkling will occur in sandwich constructions having solid or
foam cores :
awr = Q L (1-pe 2 (2.2-5)
In cases where the amplitude of initial waviness is known, one can use the curves of
Figure 2.2-3 to establish Q. Whenever such information is unavailable, it is recom-
mended that the value Q = 0.50 be used to obtain a lower-bound prediction.
For elastic cases, use }7= 1. Whenever the behavior is inelastic, the methods of
When the facings are subjected to biaxial compression, it is recommended that one use
the interaction formula
Rxa + Ry = 1 (2.2-6)
whe re
[Applied Compressive Loading in Subscript Direction]
Ri= iCritieal Compressive Loading (when acting alone) in ? (2.2-7)

Subscript Direction
and the y direction corresponds to the direction of maximum compression. This inter-
action relationship is based on the information provided in Reference 2-1 for rectangular
flat plates having very large aspect ratios. For cases involving shearing stresses which
are coplanar with the facings, it is recommended that the principal stresses first be
computed and that these values then be used in the above interaction equation. When-
ever one of the principal stresses is tensile and the behavior is elastic, the analysis
should be based on the assumption that the compressive principal stress is acting alone.
2-26
I
_o
/ 2_
//
\
d
L"-
O0
(y
2-27
2.2.2 Sandwich With Honeycomb Core (Symmetric Wrinkling)
As noted in Section 2.2.1.1, the results of Reference 2-9 apply only to sandwich con-
figurations which have solid or foam cores. However, the basic theory of that report
is capable of extension to constructions having honeycomb cores and this is accomplished
in Reference 2-7. The extension is achieved by incorporating conditions which recog-
nize that the honeycomb core elastic moduli in the plane parallel to the facings are
very small in comparison with the core elastic moduli in the direction normal to the
facings. Full consideration was given to both the symmetric and antisymmetric wrin-
kling modes along with the influences from initial waviness of the facings. However,
in this case it was found that, except for the region controlled by shear crimping (low q),
symmetric wrinkling develops at stress levels which are lower than those at which the
antisymmetric mode will occur. Based on this observation, the development of Refer-
ence 2-7 resulted in the following equation for the prediction of wrinkling for isotropic
facings in sandwich constructions having honeycomb cores and subjected to uniaxial
compression:
1
082(Ec ( Ef,
awr = . \r/Ef te/
1 + 0.64 K8 (2.2-8)
where
8 Ec
K8 - tc Fc (2.2-9)
2-28
and
awr = Facingwrinkling stress, psi.
Ec = Young'smodulusof the core in thedirection normal to

the facings, psi.
tf = Thickness of facing, inches.
77 = Plasticity reduction factor, dimensionless.
Ef = YoungTs modulus of facing, psi.
tc = Thickness of core, inches.
8 = Amplitude of initial waviness in facing, inches.
Fc = Flatwise sandwich strength (the lower of flatwise core compres-

sive, flatwise core tensile, and flatwise core-to-facing bond
strengths), psi.
Equation (2.2-8) can be used to plot a family of design curves of the form shown in
Figure 2.2-4. It should be noted that the curve for K8 = 0 is an upper-bound classi-
cal value which is based on the assumption that the facings are initially perfect. This
particular curve agrees very closely with the following symmetrical wrinkling equation
recently obtained by Bartelds and Mayers [2-14] :
awr = 0.86 L_TEf tc j (r/Ef) (2.2-10)
Comparison of Equations (2.2-8) and (2.2-10) shows that, when K8 = 0, the former
gives critical stresses which are approximately 5 percent less than those obtained by
Bartelds and Mayers [2-14].
Numbers in brackets [ ] in the text denote references listed at end of each major
section (1; 2; etc.)
2-29
K(_ = 0
= Constant
K 6 = Constant
_TEf t c ]
Figure 2.2-4. Typical Design Curves for Face Wrinkling" in Sandwich

Constructions Having Honeycomb Cores
In actual practice, curves of the type shown in Figure 2.2-4 do not prove to be very
helpful since the K5 values appropriate to particular structures are rarely known.
Therefore, in order to provide a practical means for the prediction of face wrinkling
in sandwich constructions having honeycomb cores, a lower-bound approach is taken
in this handbook. For this purpose, test data selected from References 2-7 and 2-10
are plotted in Figure 2.2-5. All of the specimens from Reference 2-7 failed within the
elastic range. Several of these failures occurred by means of shear crimping and
these data were discarded. For the remaining tests reported in Reference 2-7, three
data points are plotted in Figure 2.2-5 for each group of nominally identical specimens.
One point is plotted for the maximum test value for the group, one point for the mini-
mum, and one point for the average. The data from Reference 2-10 were selected in
a similar manner with several added restrictions. A number of these specimens
wrinkled under highly inelastic conditions. Since rather crude plasticity reduction
2-30
°1°l
o
0
0 o
o
o
o _-' 0
_._
_=
e_
o
o
r/3
•C_'_
_._
o
00
o
0
2-31
factors (77 Et/Ef) wereusedin thedatareduction, it wasdecidedto plot dataonly
for thosespecimenswhichwrinkledat stress levels where (E t/Ef ) /> 0.85. In addi-
tion, many of the test specimens of Reference 2-10 had very poor eore-to-faeingbonds
as measured by flatwise tensile strengths. It was therefor(, decided to plot data only
for those specimens whose flatwise tensile strengths were at least equal to the flatwise
compressive strengths. Adhesive technology has now advanced to .the point where, with
proper care, one can usually select an adhesive system which satisfies such a require-
ment.
Based on the plot of Figure 2.2-5, the relationship
/Eetf _
O'w r = 0.33\r/E f te / (_TSf) (2.2-11)
has been selected here to provide safe design values. This implies that a knock-down
factor of approximately 0.4 is applicable to this wrinkling phenomenon. Obviously,
this is not a rigorous approach to the problem and it would be advisable to base the
design equation on a much wider selection of test data of specimens which were truly
representative of contemporary practical designs. Therefore, Equation (2.2-11) can
only be regarded as a '_est-aw_ilable" approach and it is recommended that, for veri-
fication of final designs, wrinkling tests be performed on specimens that actually dup-
licate the selected sandwich configuration. The method presented here should only be
regarded as an approximate guideline.
2-32
2.2.2.2 Design Equations and Curves
The following equation may be used to compute the approximate uniaxial compressive
stress at which face wrinkling will occur in sandwich constructions having honeycomb
cores:
z/Ec
0.8 tc/ (,El)
°'wr- 1 + 0.64 K(5 (2.2-12)
whe re
(]Ec
K6 - tc Fc (2.2-13)
In cases where the amplitude of initial waviness is known, one can either use these
equations or the curves given in Figures 2.2-6 and 2.2-7 to establish the wrinkling
stress. Both of these figures are taken directly from MIL-HDBK-23 [2-16]. When-
ever the initial waviness is unknown, it is recommended that the following equation be
used to obtain a lower-bound prediction:
awr = 0.33 \_--_f tc/ (UEf) (2.2-14)
For elastic cases, use 77 = 1. Whenever the behavior is inelastic, the methods of
When the facings are subjected to biaxial compression, it is recommended that one use
the interaction formula

3
R x + Ry = 1 (2.2-15)
where
[Applied Compressive Loading in Subscript Direction]

Ri= (2.2-16)
Critical Compressive Loading (when acting alone) in]
Subscript Direction
2-33
CD 0 0 0 0 0 0 0 0 O0
i
4hl/,/l
ll " _o
CD
/ LO
//
/ _I
_
, /// CO
0 o
z.//, /
0
I>
0
C_
jjz//
f
&
2-34
0
?]
II 11
¢£:
.....
oi 0
b b _J
and the y direction corresponds to the direction of maximum compression. This
interaction relationship is based on the information provided in Reference 2-1 for
rectangular flat plates having very large aspect ratios. For cases involving shearing
stresses which are coplanar with the facings, it is recommended that the principal
stresses first be computed and that these values then be used in the above interaction
equation. Whenever one of the principal stresses is tensile and the behavior is elastic,
the analysis should be based on the assumption that the compressive principal stress
is acting alone.
2-36
2.3 SHEARCRIMPING
2.3.1 BasicPrinciples
To understandthe phenomenon
of shearcrimping, onemustkeepin mindthatthis
modeof failure is simplya limiting caseof generalinstability. Theequationsfor
predictingshearcrimpingemergefrom generalinstability theorywhenthe analytical
treatmentextendsinto the regionof lowshearmodulifor the core. For example,the
theoreticalderivationof Reference2-17, as reformulatedin Section4.2. I. 1 of this
handbook,yields the result that, whenthetwo facingsare of the samematerial, shear
crimpingwill occurin axially compressedsandwichcylinderswhenever
Vc _ 2 (2.3-1)
where
_o
V c = (2.3-2)
Crerimp
h 2 xJtz t2
(2.3-3)
(ro = *)Ef_ l___pe2(tl + re)
h_
(2.3-4)
°'crimp - (tl + t2) tc Gxz
Ef = Young's modulus of facings, psi.
h = Distance between middle surfaces of facings, inches.
R = Radius to middle surface of cylindrical sandwich, inches.
t_ and t_ = Thicknesses of the facings (There is no preference as to which

facing is denoted by the subscript 1 or 2.), inches.
Pe = Elastic Poisson's ratio of facings, dimensionless.
2-37
t C 'l'hickn_,ss of core, inches.
Gxz ,_. _r-< shear modulus associate,I with the plane perpendicular to the
facings and crienled in the axial direction, psi.
The critical stress c:m t)(_ dt, termined from the eq_mtion
% r' Kc % (2.3-5)
and, when the Inequality (2. :{-1) holds true, Kc: can be computed as follows:
1
(2.3-6)
Hence,
'{; rimp
"i?, ....... _'_-7- _r° : 'rcri'nP (2.3-7)
Therefore, when the two facings ",re made of the same material, the following equation
can be written for the c_'itk_'t! stres_ for _h¢,ar crimpir_g in a circular sandwich cylinder
under axial compression:
h _
_r_r : (r . _ixz (2.3-8)

crimp 7t: ! t>_) t c
I ' 4 .) • .
An equlvalen_ rest)It can be obtained from Referen¢'e :_-t _ for sandwich cylinders sub-
je,"led [(_ _mif,__,'m e>:t,",_i !g_!,,ra! pressure. Tim! ,_, ,.:i_,,c_, the two facings are made
of t}l(_ .'a{!til(' Illa[eFi:'J, _}l i, ( ;_n ',vl'Jte
rrcr :: (r crimp
. - (t t + re)t e O.vz (2 3-9)
x\' h e 9"e
Gyz ;'-,_',, sl_( :z_' ul _d_!t s us,;¢)( iated with the tqane perpendicular to the
'l'<i';
',rrcvol,!lion, psi.
In addition, the development of Reference 2-19 leads one to the following formula for
circular sandwich cylinders under pure torsion and having both facings made of the
same material :
- h2 JGxz (2.3-10)
-rcr = "rcrim p (t 1 + t2) tc GY z
It should be noted that, although Equations (2.3-8) through (2.3-10) were derived for
sandwich cylinders, all of these final expressions are independent of curvature. Thus,
these equations have a general applicability which is not limited to the cylindrical con-
figuration.
2-39
2.3.2 Design Equations
The following equations may be used to compute the facing stresses at which shear
crimping will occur in sandwich constructions having both facings made of the same
mate rial:
ao For uniaxial compression acting coplanar with the facings (see Figure 2.3-1),
use
h _
_crimp = (t_ + t_) tc Gij (2.3-11)
where
Gij = Core shear modulus associated with the plane perpendicular to

the facings and parallel to the direction of loading, psi.
O', psi
o-, ps .i___
o-, psi _
Figure 2.3-1. Unia×ial Compression
2-40
bo For pure shear acting coplanar with the facings (see Figure 2.3-2), use
h_ (2 3-12)
Tcrimp - (t_+ t2) tc _GxzGyz
Figure 2.3-2. Pure Shear
The foregoing equations are valid regardless of the overall dimensions of the structure.
m addition, no knock-down factors are required since shear crimping is insensitive to
initialimperfections. The predictions from these equations will be somewhat conserva-
tive since their derivations neglect bending of the facings about their own middle sur-
faces. Although such bending is of negligible importance to most sandwich buckling
phenomena, in the case of shear crimping thls influence can be considerable.
Further mention of the shear crimping mode of failure is made in the various sections
on general instability included in this handbook.
2-41
_4
= i-
ECJ
o
2_
• ,_ _
©
O
<
_ g _
_D • b_
m _
.Z
b£.
o
c-i ,"
._ °
b _ r,9 ¢-/ _ m ._ b ._ c'd
o_ o e r2_
la., el _-) L:I
Z ;
i i#_ _
i_ iII ¢',1
.-A
i i
©
i,-ii
)-'( I _,,._
e_
_'i
,--i
2 -42
= 4o
7
¢o
o _
°_, °} S_
_'_ _ • I
_ m
a _
"0 0
_ ._ ._f
•.,_,_-_
o _, ._
i ._ _ .
i ,°
_ .
:i 0
"_ b _ _ _ ._ _ be
%
0
8
g g g
1ii
c,|
2-43
RE FERENC ES
2-1 Gerard, George and Becker, Herbert, "Handbook of Structural Stability,
Part I - Buckling of Flat Plates," NACA Technical Note 3781, July 1957.
2-2 Norris, C. B., "Short-Column Compressive Strength of Sandwich Construc-
tions as Affected by Size of Cells of Honeycomb Core Materials," U. S.
Forest Service Research Note, FPL-026, January 1964.
2-3 Anderson, M. S., "Local Instability of the Elements of a Truss-Core
Sandwich Plate," NASA Technical Report R-30, 1959.
2-4 Gough, C. S., Elam, C. F., and deBruyne, N. A., "The Stabilization of a
Thin Sheet By a Continuous Supporting Medium," Journal of the Royal Aero-
nautical Society, January 1940.
2-5 Williams, D., Leggett, D.M.A., and Hopkins, H. G., "Flat Sandwich Panels
Under Compressive End Loads," Royal Aircraft Establishment Report No.
A.D. 3174, June 1941.
2-6 Hoff, N. J. and Mautner, S. E., '_rhe Buckling of Sandwich Type Panels,"
Journal of the Aeronautical Sciences, Vol. 12, No. 3, July 1945.
2-7 Norris, C. B., Boller, K. H., and Voss, A. W., 'WVrinkling of the Facings
of Sandwich Construction Subjected to Edgewise Compression," FPL Report
No. 1810-A, June 1953.
2-8 Yusuff, S., "Theory of Wrinkling in Sandwich Construction," Journal of the
Royal Aeronautical Society, Vol. 59, January 1955.
2-9 Norris, C, B., Ericksen, W. S., March, H. W., Smith, C. B., and Boller,
K. H., "Wz2nkling of the Facings of Sandwich Constructions Subjected to
Edgewise Compression," FPL Report No. 1810, March 1956.
2-45
2-10 Jenklnson, P. M. and Kuenzi, E. W,, "Wrinkling of the Facings of Aluminum
and Stainless Steel Sandwich Subjected to Edgewise Compression, " FPL
Report No. 2171, December 1959.
2-11 Yusuff, S., "Face Wrinkling and Core Strength in Sandwich Construction,"
Journal of the Royal Aeronautical Society, Vol. 64, March 1960.
2-12 Harris, B. J. and Crisman, W. C., "Face-Wrinkling Mode of Buckling of
Sandwich Panels," Proceedings of the American Society of Civil Engineers,
Journal of the Engineering Mechanics Division, June 1965.
2-13 Benson, A. S. m_d Mayers, J., "General Instability and Face Wrinkling of
Sandwich Plates - Unified Theory and Applications," AIAA Paper No. 66-138
Presented in New York, New York, January 1966.
2-14 Bartelds, G. and Mayers, J., "Unified Theory for the Bending and Buckling of
Sandwich Shells - Application to Axially Compressed Circular Cylindrical
Shells," Department of Aeronautics and Astronautics, Stanford University
Report No. SUDAAR No. 287, November 1966.
2-15 Plantema, F. J., Sandwich Construction, John Wiley & Sons, Inc., New York,
Copyright 1966.
2-16
U. S. Department of Defense, Structural Sandwich Composites, MIL-HDBK-23,
30 December 1968.
2-17
Zatm, J. J. and Kuenzi, E. W., "Classical Buckling of Cylinders of Sandwich
Construction in Axial Compression - Orthotropic Cores," U. S. Forest Service
Research Note FPL-018, November 1963.
2-18
Kuenzi, E. W., Bohannan, B., and Stevens, G. H., "Buckling Coefficients for
Sandwich Cylinders of Finite Length Under Uniform External Lateral Pressure,
U. S. Forest Service Research Note FPL-0104, September 1965.
2-46
2-19 March, H. W. and Kuenzi, E. W., "Buckling of Sandwich Cylinders in
Torsion," FPL Report No. 1840, January 1958.
2-47
3
GENERAL INSTABILITY OF FLAT PANELS
3.1 RECTANGULAR PLATES
3. i. 1 General
As previously noted, one of the potential modes of failure for sandwich panels is that
of general instability. This occurs when the panel becomes elastically unstable under
the application of certain types of in-plane loads. Further, it should be noted, the
loads which are critical for instability may or may not be of such magnitude as to
cause a failure of the basic materials.
The fiat, rectangular sandwich panel represents that configuration for which the vast
majority of fabrication and test data has been accumulated over the past decade. This
is probably due to the fact that this configuration was best adapted to the structural
needs for a number of applications and that it represented the minimum in fabrication
problems and costs as far as this type of construction is concerned. By the same
token, analytical solutions have been developed for a wide range of loading applications
for flat panels, and an appreciable amount of testing for correlation with these solu-
tions has been accomplished.
As a consequence of this past work, it is now possible to employ the analytical solu-
tions for fiat panels, as given in MIL-HDBK-23, [3-1], with a high degree of con-
fidence. This view is supported by recommendations given in References 3-2 through
3-7, inclusive, for basic panel design. Therefore, with this background in mind, the
3-1
buckling coefficients, K, which will be given in this section for the various plate loading
conditions will be those taken from the applicable sections of Reference 3-1, with no
"knock down" factor to be applied to them.
The development of plate buckling coefficients for sandwich construction requires the
consideration of a ntmlber of factors, some are: 1) the degree of orthotropicitsr of the
face plates, 2) the use of the same or of dissimilar materials for the face plates and,
3) the degree of orthotropicity of the core material. The general equations given in
the following sections account for these possibilities; however, the curves showing K
as a function of (a/b), V, the type of loading and edge support conditions will assume
the use of isotropic faceplate materials since this is largely typical of aerospace
vehicle design practices.
In all cases, the final design of the sandwich panel must comply with the following four
basic design principles, Reference 3-1;
a. The sandwich facings shall be at least thick enough to withstand the chosen
design stresses under the application of the ultimate design loads.
b. The core shall be thick enough and have sufficient shear rigidity and
strength so that over-all sandwich buckling, excessive deflection, and
shear failure will not occur under the design loads.
c. The core shall have high enough moduli of elasticity, and the sandwich
shall have great enough flatwise tensile and compressive strength such
that wrinkling of either facing will not occur under the design loads.
d. If the core is a cellular honeycomb or constructed of corrugated material

and dimpling of the facings is not permissible, the cell size or corrugation
spacing shall be small enough so that dimpling of either facing into the
core spaces will not occur under the design loads.
3-2
Otherrequirementsincludethe useof moduli of elasticity andstressvaluesrepre-
sentativeof thosevalueswhichprevail underthe conditionsof use. Also, wherethe
stressesare beyondtheproportionallimit, the appropriatereducedmodulusof elas-
ticity shouldbeused.
The followingsectionson specifictypesof panelloadsdefinethe appropriateequations
for eachparticular situationanddiscussusefullimits andother considerations,as
applicable. A summarytable, (Table3-1), listing the panelinstability equationsgiven
in the variousparts of this section, alongwith a definitionof terms, equationlimitations
if any, andreferencesfor the appropriatebucklingcurvesimmediatelyprecedesthe list
of referencesto facilitate useof the manualfor specificproblemsolution.
Figure3.1-1 showselasticproperties anddimensionsfor the typical sandwichpanel

underconsiderationin this section.
3-3
_,_ b[ __ _tf
--_ b__..t _
ttf W-bf --q \¢ •/ t¢ _-_-'t

(TYPICAL) f
SINGLE -TRuss DOUBLE -TRUSS
CORRUGATED CORE CONFIGURATIONS
_h-_
N
v
CORE --,
E b, Gcb
4-t4 _--t
tl-_ C 2
.-t-o
ItONEYCOMB CORE
-e_ b
---.--]_ y
Figure 3.1-1. Elastic Properties and Dimensional Notations

for a Typical Sandwich Panel
3-4
3.1.2 Uniaxial Edgewise Compression
The buckling coefficient equations and curves given here for uniaxial edgewise com-
pression arc those originally developed by Ericksen and March, _3-8], and are in-
cluded in the MIL-HDBK-23 documents, issued since then. The basic principles and
assumptions employed in the development of these general instability equations are
noted in the references and are not repeated here except where required to limit their
use because of the original restrictions imposed.
The basic equations for calculation of the allowable sandwich panel edgewise com-
pression loads are given in the following section. Curves for panel buckling coeffi-
cients for panels having isotropic facepk-ttes and both orthotropic and isotropic cores
for various panel edge support conditions follow the equations.
As previously noted, the equations presented in this section arc those developed by
Ericksen and March, and presented in MIL-HDBK-23, as well as in other documents.
Supporting data such as pertinent assumptions and definition of terms are also in-
cluded along with the equations.
Sandwich Panels With Honeycomb Cores
One of the basic assumptions used in the design and analysis of sandwich panels is
that the face plates carry the inplane loads applied and that the core provides that
shear support to the face plates required for them to act as a unit in preventing early
3-5
individual buckling. From this, the edgewise compression capability of the panel is
given by the following equations, which are taken from Section 5.3, Reference 3-1:
N = (_2/b_)(K)(D) (3.1-1)
cr
where D is the sandwich bending stiffness. Solving this equation for the facing sires-
ses gives the following:
(E_t_)(E_t_) (h_) (_ (3.1-2)

l'c_, = = _7_K (E_t_ _ E't2)2 (b _) k
For equal facings:
E I
_K (h)_ f
F - (3.1-3)
e 4 (b) 2 )_
w he re
K = buckling coefficient = K + K (see definitions in following

F M
work).
1
E J
= (,'a = effective modulus ()f elasticity for orthotropic
facings.
X = (1 - _ta_b)
_a'_b = Poisson's ratio as measured parallel to the subscript direction.
f, 1, 2 = subscripts denoting facings.
h,b = see Figure 3.1-1.
Since the buckling coefficient curves to be presented here are being limited to the case
of isotropic face plates, which is representative of the large majority of structural
sanchvich applications, the affected equations given previously arc revised below for
this situation.
3-6
For isotropic facings:
Eai' = E' bi = E.'1 = 77iEi; and _ai = _bi = _i
where 77i = plasticity correction factor (see Section 9.0).
As noted above the buckling coefficient for the panel under this loading condition is
given by the equation
K = KF+K M
where
KF - 12 E_t a (E/t2) h_ KM (3.1-4)

0
KM = KM for the case where V=0 Esee Figure (3.1-16)1 (3.1-5)

O
Values of K F are generally quite small relative to KM, thus a safe first approximation
is to assume it is equal to zero until a final panel check is made. On this basis, K
= K M may be used to develop initial face plate and core thicknesses for the panel.
K M is a theoretical coefficient which is dependent on the sandwich panel bending and
shear rigidities and panel aspect ratio. Other factors which influence the magnitude
of this coefficient include the panel edge support conditions and the orthotropicity of
the core. A discussion of these considerations along with development of the equations
for calculation of this coefficient are given in References 3-1 and 3-8. This manual
does not propose to repeat these equations here; however, the curves shown in Figures
3.1-2 through 3.1-15 give values of K M as a function of edge support condition, panel
aspect ratio, and the bending-shear rigidity parameter, V which is defined as follows
y_D
V - b2U (3.1-6)
3-7
which further canbewritten as:
-It
C
V = (3.1-7)
kb; Ge (E_t,+E;t:)
_2 t t
c Eftf
V - (for equal facings) (:3.1-7a)
2). b 2 G
C
where U is sandwich shear stiffness; Gc is the core shear modulus associated with the
axes parallel to direction of loading (also parallel to panel side of length a) and per-
pendicular to the plane of the panel.
An indication of the influence and importance of the core shear modulus may be obtained
from inspection of the above equations for V m_d the curves giving values of K M given
later. Holding all terms constant except G , an increase in its value reduces the value
e
of V to be used with the buckling coefficient curves, this reduced value then calls for
an increased value of K
M"
Sandwich Panels With Corrugated Core
The equations and fornmlas previously given are for sandwich panels with honeycomb
cores; however, they may be adapted to cover the case of panels with corrugated cores
by means of the following modifications:
a. For the case where the corrugation flutes are oriented normal to the direc-
tion of the load application, the shear modulus in the direction parallel to
the flutes, Geb, is very high with respect to the shear modulus parallel to
the direction 2f andl°ading'R Ge :}; thus,= 0.the previous curves may be used by
letting Gcb = = Gca/Gcb
b. For the case where the corrugation flutes are parallel to the direction of
loading, the corrugations may be assumed to carry load in a direct pro-
poYtion to their area and elastic modulus. The parameter V for this case
is replaced by the parameter W, which is defined as
3-8
e
= i (3.1-8)
W ).b2Gcb (E_t_ + E_t,)
Or, for equal facings,
W = _t cE'[t/2xb 2 Gab (3.1-8a)
Values of K M as a function of (b/a), R = (Gea/Gcb), and V, or W, are given for various
edge support conditions in Figures 3.1-2 through 3.1-15, with Figures 3.1-14 and
3.1-15 representing the case of panels having corrugated cores.
Figure 3.1-16 gives values of KMo as a function of panel aspect ratio and edge support
conditions for use in determining values of K F in order that final values for K may be
obtained for specific designs.
The curves and equations just given may be used in developing a panel design in addi-
tion to checking the adequacy of an existing design; however, this is a slow iterative
process. As a consequence, this manual recommends the use of the design-procedures
approach described in Reference 3-1 since it was specifically developed to expedite the
new design process.
3-9
]
10
2.5 Gc
END
V
05
0.15
0.30
1
FOR V>0.40 tC =--
M V
o .__ _k .... 1 1
0 0.2 0.4 0.6 0.8 1.0 0.8 0.6 0.4 0.2 0
a b
b a
Figure 3.1-2. K M for a Sandwich Panel with Ends and Sides Simply Supported,
Isotropic Facings, and Orthotropic Core, (R = 0.40)
3-10
14
12
10
.20
2 .40
0.60
1 -- 1.00
FORV > 1.00 _.'=
M
0 i i .
0 0.2 0.4 0.6 0.8 1.0 0.8 0.6 0.4 0.2 0
a b
b a
Figure 3. i-3. K M for Sandwich Panel with Ends and Sides Simply Supported,
Isotropic Facings, and Isotropic Core, (R = 1.00)
3-11
14
.... .......... [ l-i
I I
12 I 2
II
1_ _ !
10 V - rr D
b2U
0.4G i
8 ° I
END 'r
I_ _1
I- b -I
KM X I
0.10
0.20
0.30
1
FOR V >2.50 I( = 1 O.5O
M V
o t i I 2.50
0 0.2 0.4 0.6 0.8 1.0 0.8 0.6 0.4 0.2 0
a b
b
a
Figure 3.1-4. K M for Sandwich Panel with Ends and Sides Simply Supported,
Isotropic Facings, and Orthotropic Core, (R = 2.50)
3-12
16
14
2
?r D
12 V m
b2 U
10
_0.05
.10
.15
0.20
0.25
0.30
0
0 0.2 0.4 0.6 0.8 1.0 0.8 0.6 0.4 0.2 0
b
a
a
b
Figure 3.1-5. K M for Sandwich Panel with Ends Simply Supported and Sides
Clamped, Isotropic Facings, and Orthotropic Core, (R=0.40)
3-13
16 T I
14"I ....
12
10
1 V m
2
?r D
b2 U
G
END
c
_,
S
\ v
0
0.I0
0.20
0.30
0.40
0.60
1 0.75
0 FOR V a>0.75 KM=__ V

0 0.2 0.4 0.6 0.8 1.0 0.8 0.6 0.4 0.2
.-_ b
a
Clamped, Isotropic Facings, and Isotropic Core, (R=I.00)
3-14
16 i T
i i
14 I l
L_o
2
D
12 V m
b2U 0.4G
C
END
10 L_ -- -- -
] I
V
8
0
6
\
\
0.10
0.20
0.30
0.60
1 1.00
.l, =_77:1-- 1. 875
FOR V > i1.875 lK_"
0 J
0 0.2 0.4 0.6 0.8 1.0 0.8 0.6 0.4 0.2 0
b
a
a
b
Clamped, Isotropic Facings, and Orthotropic Core, (R= 2.50)
3-15
16
14
12
.30
O. 40
1
FOR V >0.40 K - V
0
o 0.2 0.4 0.6 0.8 1.0 0.8 0.6 0.4 0.2 0
a b
b a
Figure 3.1-8. K M for Sandwich Panel with Ends Clamped and Sides Simply
Supported, Isotropic Facings, and Orthotropic Core, (R=0.40)
3-16
16
14
2
D
12 V
b2U
G
C
END
1
0 FORVi > i.?0 K M=_ V
0 0.2 0.4 0.6 0.8 1.0 0.8 0.6 0.4 0.2
a b
b a
Supported, Isotropic Facings, and Isotropic Core, (R = 1.00)
3-17
16--
I
14 ...... l,
\ 2
D
\
12
b2 U
0.4G
C
\ END
K
M
6
t
0
0 0.2 0.4
Supported, Isotropic Facings, and Orthotropic Core, (R= 2.50)
3-18
;_iA
14
2 0°
V = __
rr D 1'
12 \ b2U
\, 2.5G
END
e
2 ?
1
FoR v > 0.30 % - V
o 1
0 0.2 0.4 0.6 0.8 1.0 0.8 0.6 0.4 0.2 0
a b
b a
Figure 3.1-11. K M for Sandwich Panel with Ends and Sides Clamped, Isotropic
Facings, and Orthotroptc Core, (R = 0.40)
3-19
16
.... _.....] 1 i i
14
b2U
12
END
I _'_
10
0.i0
O. 20
0.30
0.40
0.50
I 1 0.75
FOR V >0.75 K =--
M \7
0 / i •
0 0.2 0.4 0.6 0.8 1.0 0.8 0 6 0.4 0.2 0
a b
b a
Figure 3.1-12. K M for Sandwich Panel with Ends and Sides Clamped, Isotropic
Facings, and Isotropic Core, (R = 1.00)
3-20
16
14 2
D
V =
b2 U
12
0.4G
e
END _'
V
8
0
O. 10
0.20
0.30
0.60
1 1.00
FORV > 1.875 K.o -- --_ 1.875
0 1 J M v
0 0.2 0.4 0.6 0.8 1.0 0.8 0.6 0.4 0.2 0
a b
b a
Figure 3.1-13. KM for Sandwich Panel with Ends and Sides Clamped, Isotropic
Facings, and Orthotropic Core, (R = 2.50}
3-21
14
I
V =0
12
lp
10
=0.1
2
Tr Dt
c
+I t
V _t: Gb=_
b2h2 G
ca _L
I t? _,
\
°v0 \
[ l
V_0.4
0
0 0.2 0.4 0.6 0.8 1.0 0.8 0.6 0.4 0.2 0
a b
b a
Figure 3.1-14. K M for Simply Supported Sandwich Panel Having a Corrugated

Core. Core Corrugation Flutes are Perpendicular to the
Load Direction
3-22
14
12
2 Dt
C
]0 W
b 2 h 2 Gob
0.1
_'--0o2
_- o.4
0.4 0.6 0.8 1.0 0.8 0.6 0.4 0.2 0

a b
b a
Figure 3.1-15. K M for Simply Supported Sandwich Panel Having a Corrugated

Core. Core Corrugation Flutes are Parallel to the Load
Direction
3-23
END t
\
\
K
Mo \
0.i 0.2
a
Figure 3.1-16. KMo for Sandwich Panel with Isotropie Facings

in Edgewise Compression
3-24
3.1.3 Edgewise Shear
As noted earlier in Section 3.1.1, sufficient analysis, design, and testing of flat sand-
wich panels has been accom,lished to demonstrate the adequacy of the analytical
approaches presently in use. Thus, the panel buckling coefficient equations and
curves given in the following paragraphs for edgewise shear are those taken from the
MIL-ttDBK-23 documents presently in use. These equations were originally developed
by Kuenzi and Ericksen E3-13] and employ the same general assumptions as those
described in Section 3.1.1. Specific limitations or restrictions on the use of these
equations will be noted where these require consideration.
The basic equations for use in calculation of the allowable sandwich panel edgewise
shear loads are given in the following section along with applicable background data
and assumptions. Design curves and buckling coefficients for panels having isotropic
faceplates and both orthotropic and isotropic cores for both simply supported and
clamped edge conditions follow the equations.
The design equations presented here are taken from Reference 3-1 and 3-13. Support-
ing data and design constraints are also noted and discussed as required.
The edgewise shear load carrying capability of a sandwich panel is given by the follow-
ing equation:
Nser = (Tr_/b_) (Ks) (D) (3.1-9)
3-25
where
Nscr = critical edgewise shear load, lb per inch
D = sandwich bending stiffness
Solving this for the facing st_'esses gives the following equation:
/ I f
(E_tI)(E_t_)(he)E z,
Fsz,_ = _K , + i (3.1-10)
s (Elt 1 E_t_) (be) x
Or, for equal facings
y2K s (h e) Ef'
F - (3.1-10a)
s 4 (b_)k
where
E' is the effective modulus of elasticity of facing at stress F s = 7/E
r_ = plasticity correction factor (Section 9.0)
X = 1-D 2
= _ = p_ = Poisson's ratio of facings
h = distance between facing centroids (Figure 3.1-1)
b = panel width (<a) (Figure 3.1-1)
Ks = KF + KM (Note: These terms differ from those of Section 3.1.2)
whe re
i 3 t
(E_t_3 + E_t2 ) (E'_t_+ E_t_)KMo
KF = 12(E_t_)(E,2t_
) h2 (3.1-11)
(tf_KMo
KF = 3h 2 (3.l-lla)
KMo = value ofK M for V=W=0
3-26
The equation defining the value of K M is quite complex and involved, being dependent
on panel aspect ratio, (a/b), the number of half-waves, (n), for the minimum energy
buckle pattern, and the panel bending and shear rigidityparameter, (V, or W). This
manual proposes to follow general practice in the literature and provide curves only
for the definition of this buckling coefficient. Those interested in the basic equation
and its development will find this in Reference 3-13.
Values of K M are given in Figures 3.1-17 through 3.1-24 as a function of the panel
aspect ratio and the parameter V, or W, for various panel edge support conditions.
These figures cover panels with isotropic faceplates and both isotropic and orthotropie
core, including panels using corrugated flutes for cores. Values of the buckling coeffi-
cient, KMo, may also be obtained from the same set of figures.
The equations defining the parameters V and W are the same as those given in the
previous section for edgewise compression; however, they are repeated below to
facilitate their use. The equation numbers previously assigned to them are retained
below
(Y% tc
(E_tl)(E'_t2) (3.1-7)
V = k(E[t_ +E_t 2) (b 2) G
ca
(3.1-7a)
V = _r_tcE'ft/2),b2 Gca (equal facings)
For a sandwich panel with a corrugated core in which the corrugation flutes are parallel
to the edge of length a, the parameter V is replaced by the parameter W which is de-
fined as follows:
3-27
W
)_b2 %b {EIlt_ + E_ta) (3.1-8)
W = rr_t c E_t/2kb e Gcb (3.1-8a)
In checking a particular design for the critical buckling stress, Fscr, Figures 3.1-17
through 3.1-21 should be used for those panels having all edges simply supported.
Curves of K M for sandwich panels having all edges clamped are given in Figures
3.1-22 through 3.1-24. These curves may be interpolated in order to obtain the
buckling coefficients for other values of core orthotropicity, (R = Gea/Gcb), and inter-
mediate values of V or W°
It should be noted that if the resulting value of F is above the proportional limit
set
value, the value of E' shall be an effective value based on that stress level, and this
effective value shall be used in computing the value of V, Equation (3.1-7) or (3.1-7a)
or W, Equation (3.1-8) or (3.1-8a), as the case may be. This same effective value
for E I shall also be used in Equation (3.1-10), or (3.1-10a) when calculating the criti-
cal panel buckling stress. Thus, several interations will be required to establish the
actual value of F scr in those cases where it exceeds the proportional limit.
The equations and curves just given may be used in the development of panel designs
as well as in checking an existing design; however, as was the case for uniaxial com-
pression, this is a lengthy iterative process. Thus, this manual recommends the use
of the design-procedures approach described in Reference 3-1 for those cases where
the initiation of new designs is required.
3-28
10
0
l
2
7r 1) /
b2U /
0.05
0.10
KM
---- 0.40
[ 0.2 0.4 0.6 0.8 1.0

0
b
a
K M for a Sandwich Panel with All Edges Simply Supported,

Figure 3.1-17.
and an Isotropie Core, (R = 1.00)
3-29
10
V
0.4 G
0.40
b
a
Figure 3.1-18.
and an Orthotropic Core, (R = 2.50)
3-30
J
ff
10
t_
2
D
/
t_L
b2 U /
/
/
0.05
_J" J
O. 20
0.40
C
0 0.2 0.4 b 0.6 0.8 1.0
Figure 3.1-19. K M for a Sandwich Panel with All Edges Simply Supported,
and with an Orthotropic Core, (R = 0.40)
3-31
10
K
M 5
i
0
0 0.4
b
a
Figure3.1-20.
Isotropic Facings and Corrugated Core. Core
Corrugation Flutes are Parallel to Side a
3-32
10
V
9- -r 2
_r t D
I '
c
_L V
b2 h2 G
ca -
/ 0.05
_b_ / " /
/ /
7 1
i /
1
f
// 0.20
J J
_J
i
i
i
I
KM O. 50
11
I
jl
4
0
0 0.2 0.4 0.6 0.8 1.0
b
a
Figure 3.1-2 i. K M for a Sandwich Panel with All Edges Simply Supported,
Isotropic Facings and Corrugated Core. Core
Corrugation Flutes are Parallel to Side b
3-33
16
V
14
l Gc
12
10
0.05
K M
O. i0
O. 20
0
0 0.2 0.4 0.6 0.8 1.0
b
Figure 3.1-22. K M for a Sandwich Panel with All Edges Clamped,

Isotropic Facings and Isotropic Core, (R = 1.00)
3-34
16
2 0
14 i
I
V
77 D
----"--
b2 U
/
0.4G
c
If /
/
12 _-b ---4
10 | i r_
I jj
J
0.05
5_ 7
J
J
_..I
6 v
I
_J
J 0.10
I
I
0.20
o I
0 0.2 0.4 0.6 0.8 1.0
b
a

Isotropic Facings and Orthotropic Core, (R=2.50)
3-35
16
V
14
2
_r D
/
0
Il 2.s C
II
"_----- _
12
10 0.05
0.i0
K
M
.2O
0 0.2 0.4 0.6 0.8 1.0

b
a

Isotropic Facings and Orthotropic Core, (R= 0.40)
3-36
3.1.4 Edgewise Bending Moment
The application of an edgewise bending moment to a flat, rectangular sandwich panel
produces a loading condition such as that shown in Figure 3.1-25. This represents a
somewhat different situation from the ones previously covered, since the tension loading
on one half of the panel represents a stabilizing effect. The edge compression load
on the other half of the panel varies linearly from zero at the neutral axis to a maxi-
mum value, N, at the panel edge. It is this compression loading which can produce
panel buckling in the same fashion as the uniaxial compression case; however, the
presence of the panel edge support along the line of maximum loading forces consider-
ation of a more complex failure mode.
These failure mode considerations for this type of loading have been covered in the
development of analytical techniques for the evaluation of flat plates (Reference 3-17).
hlso, as has been previously noted, sufficient analytical development and testing has
been accomplished on flat, rectangular sandwich panels to enable the use of the buckling
coefficients given in Reference 3-1 for this loading condition with complete confidence.
The general equations for the behavior of flat, rectangular honeycomb sandwich panels
under this loading condition were developed by Kimel E3-15] while whose applicable to
panels with a corrugated core were developed by Harris and Auelman, E3-14] and E3-16].
The assumptions employed in the development of the basic equation for the panel sta-
bility coefficient for this loading condition are generally the same as those described
3-37
in Section 3.1.1, with one particulal' exception. This exception requires that the
critical design faeeplate stress, Fcr, shall not exct,ed the elastic buckling stress for
the faceplates. This requirement stresses the fact that the analysis is based on a
linear loading variation across the edge of the panel. Once the elastic buckling stress
is exeeede.d this variation is no longer linear, and extrapolation to a buckling stress
beyond the elastic range of facing stresses cannot be done by using an effective elastic
modulus such as the tangent modulus, in the buckling formulas. Since the proper
extrapolation to stresses beyond the elastic range must consider the variation of
effective elastic modulus across the panel width associated with the stress variation,
the equations and buckling coefficier_ts given here are thus strictly applicable only to
buckling at facing stresses within the elastic range.
The basic equations to be used in the calculation of the allowable sandwich panel edge
loading are given in the following section. Design curves and buckling coefficients
for panels having isotropic faceplates and both isolropic and orthotropic cores based
on simply supported edge conditions follow these equations.
The design equations presented here arc those taken from Reference 3-1. Background
assumptions and any applicable design constraints are also covered.
Using a linear stress variation as previously discussed, the value of N at the panel
edge is given by the equation:
N = 6M/b _ (3.1-12)
3-38
where
N _-_
load per unit width of edge
M = edgewise bending moment
b = panel width (Figure 3.1-25)
The edgewise bending load capability of a sandwich panel is given by the following
equation, taken from Reference 3-1:
N (3.1-13)
er = (_,/be) (%) (D)
where
N = critical edgewise loading, lb per inch

er
D = sandwiehbending stiffness
The critical faceplate stresses are obtained by solution of the previous equation and
are as follows:
(E1t_)(E2t_) (h_) (El,s)

Fe_,_ = Y2Kb (Eli
a +E_t_) 2 (be) 1 (3.1-14)
Or, for equal facings,
F = (3.1-14a)
c 4 (b _) ),
whe re
E = modulus of elasticity of facing
I = (1 - _)
= Poisson's ratio of facings: /_a = _b assumed above
h = distance between facing centroids
length of loaded edge of panel
K F + K M (Note: The values for these buckling terms differ

from those given in Sections 3.1.2 and 3.1.3)
3-39
(E_tl_+ E_t2
3) (Elt1+ Eat)
(3.1-15)
KF = 12 (E1t_)(E_t
2) (he) KMo
(tf 2) KM o
(3.1-15a)
KF = 2 h2
where
KM = value ofK M for V=W=0

o
for panel buckling are given in Figures 3.1-25 through 3.1-28 as a func-
Values of K M
These cover panels having

tion of the parameter \T or W, and the panel aspect ratio.
isotropic faceplates using both isotropic and orthotropic cores, including those using
corrugated flute-type cores.
The equations defining the parameters V and W are the same as those given in the
previous section for edgewise compression; however, they are repeated below to
facilitate their use. The equation numbers previously assigned to them are retained
below; however, values of E _ are replaced by those of E for this case.
(Elt _ ) (E2t_) (z) tc

V = (3.1-7)
k (Eit_ + E_t_)(b2 ) G ca
V = 1r_t e Eftf/2k b 2 Gca (equal facings) (3.1-7a)
For a sandwich panel with a corrugated core in which the corrugation flutes are parallel
to the edge of length a, the parameter V is replaced by the parameter W, which is de-
fined as follows :
3-4O
C
w = (3.1-8)
I b_Gcb (Elh +E2t _)
Or, for equalfacings,
W = _r_tc Eft/2X b 2 Geb (3.1-8a)
A particular design may be checked by using the graphs given in Figures 3.1-25 through
3.1-28 to determine the appropriate value of the buckling coefficient to use in Equation
(3.1-15), or (3.1-15a) to compute the critical buckling stress, Fer. This approach,
which involves trial and error solutions by iteration, may also be employed to develop
new panel designs; however, this manual recommends that the design-procedures
approach described in Reference 3-1 be considered since it was set up to facilitate
such design caleulations.
3-41
36 ......
_-
°°°I
C _-- _|_T
2
32 ?r D
V --
V--0
b2 U
28 .......
I
r
/
"_ n--2/ n--3
24
16
V=0.215
i
8
FOR v => 0.21_ KM = 1.ssG/v
0
o 0.2 0.4 0.6 0.8 1.0 0.8 0.6 0.4 0.2 0
a b
b a
Figure 3.1-25. K M for a Simply Supported Sandwich Panel with an

Isotropic Core, (R = 1.00)
3-42
4O
a -------_
G _
e c, ,, |
3_
y2 D
V -
b2 U
32 !V=0
A
/
_4 ! ", n=l S
f
n=2vxn=3
,o "X v=o.osl
16 \ \_.
\
\
12
-'_..L--
o.2
8
V--0 .4
I I----
FOR V _-
> 0.4 K M -- 1. ss6/v
0
0 0.2 0.4 0.6
{ I
0.8 1.0 0.8 0.6 0.4 0.2 0
a b
b a
Figure 3.1-26. K M for a Simply Supported Sandwich Panel with an

Orthotropic Core, (R = 2.50)
3-43
4O
36
2 D
32
V=0 V --
2
b U
28
K
M
0.15
FOR V >0
15 KM= 1.886/V
0
0 0.2 0.4 0.6 0.8 1.0 0.8 0.6 0.4 0.2 0
a b
b a
Figure 3.1-27. KM for a Simply Supported Sandwich Panel with an

Orthotropic Core, (R = 0.40)
3-44
4O
v cr rl a-----_
36 G --_ 'i
±
y2t D
e
32
b2h 2 Gcb
28
Ill/ \
"_.. n=21 n=3
24---
W=O.05
2O
W=0.20
16
_W=0.50
12
............ L .......................
C
0 0.2 0.4 0.6 0.8 1.0 0.8 0.6 0.4 0.2 0
a b
b a
Figure 3.1-28. K M for a Simply Supported Sandwich Panel with Corrugated Core.
Core Corrugation Flutes Parallel to Side a
3-45
3.1.5 OtherSingleLoadingConditions
A searchof the literature, aswell as contactswith a numberofpeoplewhohavebeen
activein the analyticalmethodsfield for this type of construction,revealednoother
singleloadingconditionswhichmight leadto panelinstability problems. Consequently,
the previously'describedloadingconditionsrepresentthe extentof the fiat panelsta-
bility datawhichwill begivenherefor individualloadingcases.
3-46
3.1.6 Combined Loading Conditions
A study of the effects of combined loadings on the buckling of fiat sandwich panels
requires the consideration of a number of factors. Some are:
a. The mode of failure of the panel under each of the applied loads.
b. The interaction between different modes for precipitation of panel
buckling or failure.
c. The influence of variations in the core shear rigidity values on the
interaction equations for panel instability failure under combined
loadings.
Since little specific testing for biaxial instability modes has been accomplished for
fiat sandwich panels, this part of the manual will provide analytically developed
equations for combination of the stress ratios which are conservative for most appli-
cations. Additionally, some discussion of the considerations involved is included
along with appropriate references in case more specific solutions or background is
needed.
The equations given on the following pages cover the interaction relationships between
the stress ratios, (R i = Ni/Nicr) , for each of the separate loadings which produce
failure by overall panel instability under the action of the combined loads. For the
stress ratio relationships which produce panel failure by local instability only, refer
to Section 2. These latter equations and pertinent discussion are not repeated here
3--47
although the specific equation number and report page are listed below for each of the
local instability modes:
a. Intracellular Buckling: Equation (2.1-6), Page 2-7.
b. Face Wrinkling (Asymmetric): Equation (2.2-6), Page 2-26.
c. Face Wrinkling {Symmetric): Equation (2.2-15), Page 2-35.
It should be noted that there are no known data available for potential panel failures
which might occur as a result of interaction between a local instability situation
arising from the loading applied along one edge in conjunction with a general insta-
bility problem arising from the loading applied along the panel edge perpendicular
to the first one. This situation might occur for panels having very high aspect
ratios; however, most of these would also indicate a potential local instability failure
under the action of the combined loads. In all cases, however, as has been previously
noted, tests should be run to substantiate a final design in all cases where there is
some question as to the structural adequacy of the sandwich component.
The effects of plasticity must be accounted for in calculating the stress ratios, R.,i to
be used in the interaction equations which are given in later paragraphs. Reference is
herewith made to the discussion and recommendations given in Section 9.2, COMBINED
LOADING CONDITIONS, in this report and in particular to Equation (9.2-1) or (9.2-1a).
Either of these equations define an effective uniaxial stress, cr., for use in determining
1
an effective plasticity reduction factor which accounts for the effects of the biaxial
stress field. Once the value of (y. is known, the plasticity reduction factor, _7, may
1
be calculated by means of Equation (9.2-3).
3-48
The design equations and curves for combined loading conditions are separated into
those which should be used for sandwich panels having honeycomb cores and those to
be uscd with panels having corrugated cores. Supporting references are given for
each type and loading condition along with any limitations or restrictions on the use
of the interaction equation.
Sandwich Panels with Honeycomb Cores
The interaction relationships between the stress ratios which define the onset of general
instability buckling of honeycomb core panels under combined loadings are complex
functions of a number of factors. Some of these will be covered briefly here. One of
the prerequisites for the development of the interaction equation is the determination
of the number of half-waves in both the x and y directions for minimum energy plate
buckling. Since each of these is a function of not only its relationship with the other
but is also dependent upon the core shear rigidity parameter, V, the panel aspect
ratio, panel edge support considerations, etc., the establishment of general equations
covering all of these influences presents a formidable problem.
In view of the complexity involved in an exact definition of combined load interactions,
the writers of this manual propose the use of the following simplified stress ratio
relationships for panel buckling. These give somewhat conservative results over the
typical range of aerospace application and have been recommended for general use,
[3-I].
3 -49
A° Biaxial Compression. The following formula is recommended for estimating
buckling of a panel subjected to biaxial compression:
R +R =1 (3.1-16)
cx cy
where
R = N/N
e cr
N = Loading along panel edge, lbs/inch.
N = Critical loading along panel edge, lbs/inch. (See Equation

cr
3.1-1,)
x, y = Subscripts denoting direction of loading. (See Figure 3.1-1.)
A plot of Equation (3.1-16) is given in Figure 3.1-29 to facilitiate its use in
making design checks.
As noted in References 3-1 and 3-23, the above equation is correct for square,
isotropic sandwich panels for which V _ 0. It becomes appreciably conservative
for panels of large aspect ratio, (a/b _ 3.0) and for panels bordering on the weak
core regime (V _> 0.3). For panels with aspect ratios of 2.0 or less, and which
have reasonably stiff honeycomb cores, Equation (3.1-16) provides a satisfactory
method for prediction of the onset of panel buckling.
Bo Bending and Compression. Equation (3.1-17) provides a sufficiently reliable
method for the estimation of panel buckling under the action of combined bending
and compression loads.
R + (RBx)3/2 = 1 (3.1-17)
CX
3-50
where
(See definition of Terms for Equation 3.1-16.)

Rc = N/Ncr
RB = (N/Ncr)bending
N = Load per unit width of edge due to edgewise bending, lbs/in.
N Critical edgewise loading on panel due to bending moment,

cr
lbs/in. (See Equation 3.1-13.)
Figure (3.1-30) plots the interaction relationship given by Equation (3.1-17) to
enable its ready use.
References 3-19 and 3-23 are recommended, in case more accurate analysis
of this loading combination is desired.
C. Compression and Shear. The following interaction formula furnishes a depend-
able method for the prediction of panel buckling under this particular combination
of loads:
Rc + (Rs)2 = 1 (3.1-18)
whe re
Re = N/Ncr (See definition of terms for Equation 3.1-16.)
RS = (Ns/Nsc r)
N = Shear loading per unit width of panel edge, lbs/in.

s
N = Critical edgewise shear loading, lbs/in. (See Equation 3.1-9.)

ser
Equation (3.1-18) is plotted in Figure 3.1-31 to enable it to be more easily used
in the solution of specific problems. References 3-21 and 3-23 develop this
interaction relationship in greater depth for those needing this information.
3-51
O_ Bending and Shear. The following interaction equation represents a close
approximation of tile buckling behavior of panels under combined edgewise
bending and shear loads.
(3.1-19)
(RB)2+ (Rs)2 = 1
w he r c
R These terms are defined as before for

B = (N/Ner)bending
Equation (3.1-1 7).
Rs = (Ns/Nscr) As previously defined for Equation (3.1-18).
Again, as for the previous combined loadings, Equation (3.1-19) is plotted in
Figure 3.1-32 to make it more easily and readily usable. Reference 3-19 pro-
vides additional background information on the development of this interaction
equation.
Sandwich Panels with Corrugated Cores
The interaction equations for predicting the onset of general instability failure for
sandwich panels with corrugated cores involve the consideration of a number of com-
plex relationships also, as for the honeycomb core case. The same influences prevail
for fluted corrugations as before, with the additional consideration that the core shear
modulus normal to the direction of flute orientation is negligible in comparison to the
shear modulus measured parallel to the flutes. Also, the ability of the corrugations
to carry axial loading when it is applied along the axis of the flutes, further compli-
cates the problem since the distribution of this loading between the faceplates and the
flutes depends on the geometry and material thicknesses.
3-52
In view of the magnitude of the problem involved in developing specific equations for
the interaction relationships, this manual will take advantage of the extensive studies
in this area performed by Harris and Auelman, [3-14 and 3-16]. The latter reference
presents interaction equations for the prediction of the onset of panel buckling in the
form'of curves relating the buckling coefficients to each other as a function of panel
aspect ratio, the core bending-shear rigidity parameter, W, the _!ation between load
direction and flute orientation, and the ratio of the loading carried by the flutes with
respect to that carried by the faceplates. These interaction curves are repeated here
in Figures 3.1-33 through 3.1-42 for several values of the shear rigidity parameter,
W, and for the following additional relationships: 1) Panel aspect ratio, a/b = 1/2,
1.0, and 2.0, and 2) Amount of axial load carried by the core corrugations is negligible
(D = bending stiffness
with respect to that carried by the faceplates, i.e., De/D = 0. c
of corrugations, and D = bending stiffness of sandwich panel.)
A discussion of each of the sets of interaction curves follows.
Ao Biaxial Compression. Interaction curves relating the buckling coefficients for
this combined load condition are given in Figures 3.1-33 through 3.1-35.
Buckling coefficients for other panel aspeo* ratios and different values of W
may be obtained by interpolation.
The following example problem is offered to demonstrate how these curves may
be used to predict the onset of panel buckling.
Given: Panel withN =2000 lbs/in, N =400 lbs/in, a=30 in, b-- 60 in,
x y
D = 3.0 × 10 _ lbs/in e, use W = 0 for example problem.
3-53
Figure3.1-33 is usedfor this casesince(a/b) = 1/2. The top line of this figure
applies since W = 0. The interaction equation takes the following general form:
R + R <_1 (3.1-20)
cx cy
or
N N
x y
+ __ 1 (3.1-21)
N N
xcr ycr
where
= Stress ratios for loads in subscript directions, dimensionless.

Rcx'Rcy
N = Critical panel loading for loading applied in the x direction, =

xcr
(rr_/b _) (Kx) (D), lbs per inch.
N = Critical panel loading for the y direction, = (rr_/b _) (%) (D),

ycr
lbs per inch.
K,K Buckling coefficients for loading parallel to the subscript

x y direction.
E; (tf)(t e + tf) 2
D = Sandwich bending stiffness = for equal facings.
2 (1 _Nfe)
W = _r_ (t c) (EO (tf)/2 (1 -_Zfe) (b 2) Gcb for equal facings.
Ef ' = Effective Young's modulus for faces, psi.
Gcb -- Core shear modulus in the direction parallel to the flutes, psi.
t = Thickness of core, inches.

C
tf = Thickness of faceplates, inches.
_Zf = Poisson's ratio of faceplates.
a,b = Panel dimensions, inches.
3-54
Substitutingin Equation(3.1-21):
N N
X Y
+ 1.0 (3.1-22)
(K x) (_) D/b _) (%) (_'_D/b2)
and, letting
r = N /N (3.1-23)
y x
then
(3.1-24)
Nxb_ [ 1 r]
Since D, b, N , and r will be known for the design in question, and K and K
x x y
may be obtained from the appropriate curve, Equation (3.1-24) can be used in
checking the panel stability on the basis that the panel margin of safety is the
same for each loading direction. Thus,
(M.S.)x = (M.S.) Y (3.1-25)
From which
(_) 1o:(_yc4
.Nx.
lO \Ny/
or
= r (3.1-26)
Nxer/ =
the n
(Ky) (y2D/b 2) KY
-----r - (3.1-27)
K
(Kx) (_'_D/b _) x
3-55
Returningto the datagivenfor the exampleproblemto demonstratethe method
for checkingpanelstability:
r = N /N = (400/2000) = 0.20
y x
K /K = 0.20, from Equation (3.1-27)

y x
Using Figure 3.1-33, erect a line passing through the origin and having a slope
of K /K = 0.20 and extend it until it intersects the line for W = 0. The coor-
y x
dinates of this intersection point, as taken from the figure, are: K = 6.0,
X
and K = 1.2.
Y
Then,
N = K (_2D/bZ) = (6.0) (_r_ × 3.0 × 102/602 )

xcr X
N = (6.0)(822.0) = 4930 lbs/inch

xer
N = K (_r_D/b 2) = (1.2)(822.0) = 986 lbs/inch

ycr y
Solving Equation (3.1-21) for a panel stability check:
2000 400
- 0.406 + 0.406 =: 0.812
4930 986
Since the total is less than 1.0, the panel is stable under the applied loads. The
margin of safety for panel buckling is: M.S. = (1.0/0.812) - 1.0 = +0.232.
3-56
B°
Combined Compression Along Core Flutes and Shear. Figures 3.1-36 through
3.1-38 give eurves showing the interaction relationships between the buckling
coefficients for panels loaded in this manner. Curves for other panel aspect
ratios and values of the shear rigidity parameter may be developed by inter-
polation from those given. Panel stability checks for this combined loading
condition are made in the same manner as for the biaxial compression ease.
This is accomplished by handling the calculations for the t2 term in the inter-
s
action equation in the same way as was done for the I_ term in the example
cy
given on page 3-53.
C, Combined Compression Normal to Core Flutes and Shear. Interaction curves
for the buckling coefficients covering this particular combination of loads are
given in Figures 3.1-39 through 3.1-41. These curves may be interpolated to
obtain values for the specific design under study and the stability checks may
be made in a similar fashion to those for the biaxial compression case. The
method to be used in performing design checks on panels loaded in this manner
is the same as that noted in item (B) above.
D°
Combined Biaxial Compression and Shear. Figure 3.1-42 shows the relation-
ships for the compression and shear buckling coefficients for this loading con-
dition. These curves are for a square panel only, however, as may be noted
from the small change in the values of K between the various values of the
Y
shear rigidity parameter, W, approximate interpolations may be made on the
basis of ratios obtained from the curves of Figures 3.1-36 through 3.1-38.
3-57
Panelstability checksare madein basicallythe samemanneras for the example
problemgivenonpage3-53, exceptthatthe stress ratio, Rcy, is handleddiffer-
ently. Thebasic interactionequationfor this conditiontakesthe following
generalform:
R +R +R _1
cx cy s
where
R andR are as definedonpage3-54.
cx cy
Rs = (N xy/Nscr ) = [Nxy /(_/b 2) (Ks) (D)]
K = buckling coefficient for shear

s
Since, as may be seen in Figure 3.1-42, K is a function of W only for this case
Y
and is independent of the values of Kx and Ks, the value for Rcy may be calcu-
lated imm :diately and the interaction equation put in the following form:
Rcx + Rs = (1.0- Rcy ) : Or, Rcx +Rs = C
The design check may now be performed in the same way as for the example
problem on page 3-53, if the R term and calculations are handled in the same
s
way as the R term and calculations were handled for the example. It is to be
cy
noted, however, that the term on the right side of the equation, C, has a value
which is less than 1.0 and this value should be used in place of the 1.0 used in
the example. Thus, assumingR = 0.10, then C = 1.0 -0.1 =0.9, and the
cy
margin of safety for panel buckling as calculated on page 3-57 for the example
would now become:
M.S. = (0.90/0.812)- 1.0 = +0.109
3-58
R
cx
0.6 0.8
0 0.2 0.4 R
cy
Figure 3.1-29. Interaction Curve for a Honeycomb Core Sandwich

Panel Subjected to Biaxial Compression
3-59
1.0_::: hi !': ft+.
....
....
I
+ +
::
,
t;+
'++ ....
.... _it_
+P+_ :i!: if!! _'_ |
i.
+++.
Itt _:
: ' +i ' +_+ P [ il +r + ;i,-
'::;::.t ....... T++[ "I :'
r!','_T !_ !! !'
;I fl ::::I::::
i,+!t
: ;:{!:..
t_+:
....
: ,:
: ,:--!:
,_!_ _ _
!_ilii!:
,,_i,],,i
+i::
•
:::
: ::"
'
_!i+ i I i! .i.. :fill+: !:i :':i ;:. ::+_lli-'ii_

:+'i: +- i_ _i i!
,,, i:+
:::: til + +_ :,-:::
" "+ :
!i"
o.8i!.!!!...ii::i
:ii i_ i '
_+fl .... i+ .
![;:
:iiii:i: il!i ::-t:l '+li_li:i: :_
II + --.+v'":=;!: ;:,- tit, , + +:;
++' 111 i':,: '+
::i 1! ?} r :::-::::,:ii !]ii _ !_!il!i[i ++:_

_t t[
Ittt]t::iit :: _!:iii!L
+;++I+ ......
:+++: _i_ +,+
+.
+P_
+!t+l+++:
:'.'.i +:': +!+ ;+++t!G:_
',',+
...... ::--i , ,+_ +_+ N_:
0.6 lll.+'.l; _'.| ',t:
.rg_!!i!!:Fi:i i;I H. N_!! (Rcx

ii "
ii[" + RBx 3/2 : 1.0):_,+.
It !!! _i: .,+.
,_ i!i lii:
CX _; | ' I ::"' _' _t_'l][t JJl.
+_I :....... +4,)
1+ ?f Irt++'P_+l?': !:++ fff+l_+'f ....
144 ;4 _++ ] ,+| :::
0.4 ":r !!
.la
:'I!
I.
I+++++N
+:
.......
: : _bJ::!+`:!ti:ifill+ +::"
++:I++++I++++:N
: : +++:+_,,,
_,++If+1+ +
',_tt++,
ttTT
0.2 tit tf: !t r! T[

....... + +!+_
+ !+I++++
F_F!i!!!i:_ TTC_
+4++t++++t+!i+
+1++FiHtN
. ...... ;Ut
iii tr _
t!t:Ir _ +|:tr [[i[ ' '
_t_;t:'+:: itt_ "_]_ It l_tt I i!::
!-F i:
U4ttt!iiilii!t
0 Hf+H_ii_t+xl _ ]PI tt 11 Pt ii _t!,_ii!l
I _++_!d+:]l+_fi..
o 0.2 0°4 0. 6 o. 8 1. o
R
Bx

Panel Subjected to Bending and Compression
3-60
I!+t:: : +!:iii _!< i! !;;
- [![ ' !1 ....

_ ,, ii:
![![i[I
+':t!tlii! !i'. t!'.: ;;:: [b!:

_iiil
_;t ii :_i ilti !!!: tr_r,i; ,,_+.:
tt _i::
:_ i_ tR + R = 1 O) "TrY
i[ _:i _ c s "
ii! I[ :t_l liT! T.... :: ,lii:*+
gt
++ _!!1
+,+l
. _ : ;+ t_; !_
R
2 .: .- , ]q[ _I_ + , ii
1I....
tti_
Jill
C !! iTii
+÷,. _':_it:_l}_
l;ii _.XiitiliaS:.-!i I ¸ !_i
';' t;t: tell :::?='¢tii _!" i
i: !i!!
' !111
::;ilf!+ ::',. :::;+,+u m;:

:I1iI2:; ::;; +.1!!Ill: '_+ +:i.... 1tit
4 4;;;
r!t !!! r i}_+ _iii!!iT:!!!+! L +:_i
+ , I i , , + i ' _ ' 4 =1 " : l r t! f1 t _ ] i , '
i]4!ti!i::;;i:_i;I +t_;liilli) t_
Lilt
t;:i
L+ '.1+
!+'.,.J,i
........ ;++_ ++++i;+++++l++ 11
!+,+:t;;_p.+!:_i-]-Ll ++ L !![
;i 1%I:
,+++i;+11++1+Pl_+_
+ H _ t_
0.6 0.8 1,0

R
S

Panel Subjected to Compression and Shear
3-61
Panel Subjected to Bending and Shear
3-62
20.0 Y
N ___ +. b
x
16.0
W=0
L...×
-tttt _
N
12.0
\ i
Y
Nb
2
X
W = 0.05 K
X 2
_r D
K
Y
N b2
Y
. Z = --
8.0 y 2
D
I
W = 0.20-
4.0
(a/b = 1/2)
I
W 0.50-
\
4.0 6.0 8.0
0 2.0
K
X
Figure 3.1-33. Buckling Coefficients for Corrugated Core Sandwich

Panels in Biaxial Compression (a/b = 1/2)
3-63
5.0

Panels in Biaxial Compression (a/b = 1.0)
3-64
2.0
Y l
L a _1
I- -I
N
X
b
1.6
N
_'_ Y N b2
K
X
D
2
1.2
K - 2
K
Y
\ I
0.8
w \ "-,g_,. I
"_...... -.., \'_
O. 50 --_
0.4
(a/b = 2. O) _ _ \_
o
0 1.0 2.0 3.0 4.0
K
X

Panels in Biaxial Compression (a/b = 2.0)
3-65
10.0
y_
"k- 4
N '-_" b 4.._ b
X
-"_---4
8.0
(a/b = 1/2)
xy 2
Nb
I K
X
x2D
6.o __ 2
N b
xy
K
K _ \ S
2D
× \ "_,
4.0
\
\ \\ \
W = 0
2.0
W=0.20 - ..W = O. 05
0
0 8.0 16.0 24.0 32.0
K
S

Panels Under Combined Longitudinal Compression
and Shear with Longitudinal Core (a/b = 1/2)
3-66
4.0
Y a _1
N _ b
3.0
x ' "K = x
_ x 2D2
K
X
2.0
_
\ g1>w\_--io.
o_
w_-o._o
\ jA%
1.0
w=o. o_l \ ,w:o

(a/b = 1.0)
0 2.0 4.0 6.0 8.0 10. O
K
S

and Shear with Longitudinal Core (a/b = 1.0)
3-67
5.0
Y,I
a _1
vI
Nx_'_ *-b
(a/b = 2. o) : ...... _ "-
4.0
_ xy
_-_'_ _ ! Nxb2
_'_'\ ff D
_' _ Kx - 2
3.0
Ks _rxy2D
K
X
__ N b2
2.0
.....
1.0
o,,
0
0 2.0 4.0
, 6.0 8.0
K
S

and Shear with Longitudinal Core (a/b = 2.0)
3-68
10.0 •
(a/b = 1/2)
N _
x.... F I t q b
8.0 _t t_
"" ¢-N _" "-32"_" x

xy
i K - N xb 2
6.0 "_ x 2
KX _ 2
_ N b
_, \ Ks xy
4.0
2.0
PW = 0.05 i
I _,
w_o._o- \ '
0 8.0 16. 0 24.0 32.0
K
S
Figure 3.1-39. Buckling C(_fficients for Corrugated Core Sandwich

and Shear with Transverse Core (a/b = 1/2)
3-69
--N f I I I /o__ ' b
Kx
Nbx2
3.0
W= 0. Ir D
N b2
2.0 = 0.05
K
X
_-W=O
1.0
W
(a/b = 1.0) \
0 2.0 4,0 6.0 8.0 10.0

K
S

and Shear with Transverse Core (a/b = 1.0)
3-70
5.0 Y ii
a _1
N x -_ l I l I l _- b
(a/b -- 2.0) ""t.,_
,,,_
, i i i t"-AL..
,,_ X
4.0 \N
xy
I L K -
_x_
x 2
_ D
3.0
N b2
K
\ S
xy
2
17 D
X
\
2.0
05
1.0
w=o._\ _
0 2.0 4.0 6.0 8.0

K
S

and Shear with Transverse Core (a/b = 2.0)
3-71
2.5 Y 1
._e_q_
(a/b : 1. O) Nx ----- = b
2.0
_N_ _ N b2
1.5 Nxy rr D
rr D N b2
K X
N N_ __ Ks =- 2
2_ ' KY - Y
Ky = 1. 7r 2D
1.0 \
K =1.7. W=0
-W = 0.05
0.5
_ 0 50
\i " ()'20 _ __ =_
K 1"9y
o ,_,o 4o oo =o
Ks
Figure 3.1-42. Buckling Coefficients for Corrugated Core Sandwich Panels

Under Combined Longitudinal Compression, Transverse
Compression, and Shear with Longitudinal Core
3-72
3.2 CIRCULARPLATES
3.2.1 AvailableSingleLoadingConditions
A searchof the availableliterature as well as contactswith otherswhoare familiar
with sandwichpanelstability referencesandstudiesin progressuncoveredno stability
solutionsfor anysingleloadingcondition. This result might havebeenanticipated
sincethe flat, circular sandwichplate hasvery few applicationsin aerospacevehicle
structures in whichit mustbestableunderthe appliedloads. Consequently,

this
manualmakesno recommendations
for techniquesto be usedin design, andstrongly
suggeststhatall final configurationsbetestedas requiredto demonstratetheir ade-
quacystructurally.
3.2.2 AvailableCombinedLoadingConditions
Nopanelstability solutionswere foundfor anycombinedloadingconditionsapplicable
to flat, circular platesin the courseof the literature searchnotedin Section3.2.1.
Consequently,
this manualmakesno recommendations
for possibleanalyticalap-
proacheswhichwoulddescribeanystability limits for circular, flat sandwichplates.
3-73
3.3 PLATESWITH CUTOUTS
3.3.1 FramedCutouts
Whileit is highly desirableto avoidcutoutsin aerospacestructuresbecauseof the
attendantweightproblemsas well as uncertaintiesaboutloadpile-up andredistri-
bution, theseare a practical necessitybecauseof accessandother requirementsand
every effort shouldbe madeto derive reliable designapproaches

which minimizethese
drawbacks.
Mostgeneralizedsolutionsfor plateswith cutoutsemployframing membersandbase
the analysisonthe assumptionof buckledskinpanelswhichcarry only shearloads.
Obviously,the solutionbecomesmuchmore complexwhenskin bucklingdoesnot
occur, as wouldbe the casefor a framedcutoutin a sandwichpanel. Despitethe
increasedcomplexity,however,solutionsfor the load distributionaroundthe cutout
canbe obtainedfor variousload applicationsawayfrom the opening. Knowingthe
load distributionadjacentto the cutoutdoesnot necessarilyprovideananswerto all
questionsregardingthe adequacyof the design, however,particularly in the caseof
sandwichconstruction.
In the caseof monocoque

or semi-monocoque
panels, the lateral momentsof inertia
of the framingmembersare generallysufficientlygreaterthanthoseof the skin such
thatthey maybe consideredto providelateral supportfor the paneledge. This is not
necessarilythe casefor sandwichpanels,thus settingupthe caseof a free, or nearly
free, edgefor thepanelandfor whichconditionnogeneralstability solutionsor data
were foundin the courseof this study.
3-74
It maybepossiblefor specificdesignsto be assessed,onthe basisof goodengineering
judgment,to be critical in local instability rather thanfor generalinstability. This
beingthe case, designchecksmaybe madeonthe basisof the equationsgivenin
Section2. This manualmakesno recommendations

for thosecaseswherethe general
instability modeappearsto controlbeyondthe exerciseof goodjudgmentin the devel-
opmentof the design, andsufficient testingas neededto insure its integrity.
3.3.2 UnframedCutouts
Unframedcutoutsin sandwichpanelshaveall of the disadvantages

notedfor framed
cutoutsandrepresenta muchmore seriousdesignproblemlocally, insofar as the
free edgeis concerned. Thewriters of this manualencounteredno instancesin which
sucha designapproachwasusedin primary or secondarystructure and, in general,
recommendavoidanceof this practice. This recommendation

is basednot only onthe
lack of anyanalyticalor test databut alsoonpotentialproblemsof faceplate-corebond
separationalongthe free edgedueto damagewhile in use, adhesivedeterioration,
load cycling, etc.
3-75
+ o
h9
Z
I o .._
,.m _ _.5
0
i
[I r] ._ "_ ,._
,--+
%-i _u o_ +a
0
_ +a oa
.,4
+ii!
,-..i
o,,-I
+..o
,B,
ff
ro o "_ m+ d6
Z
p,-+ ©
,+.9
o
+ +.+B _ Z
[,¢] ,+_ --+ O "+
o+o:_
0
o,,-i
<
Z
8+
c ---...
¢a9
B , B _
0 _ 0 B
o
0 o
% e_ o o b+
b 0 +
o
0
0 0 I
B
©
C.l] _t_ _ "_
Eo_.S
I
o.9
0
m
ryj
3-76
o
fF
o o
E,.b
p.
¢,
tm
©
U H II II II
M m
_i ._ r_
_ +
M E
'1=
H
II
Z _
r..,q I._
o
0 ° 0 _'
o
<:
u II II II [I
,M
0 i-,].
3-77
bJo o
0 01
0
09
.-o_ o_ >
c_
b_
0
_7
7 '-_, °_
o"
E _o °_
0 o o=
.o _ +
tn
r_
o
o r_ Ir
o _
0 ,_ o_ _9
0 !
o
L_
!
Z
0 0
tn
3-78
_9
b_
b_
_o
o
o"
0
0
II
L_ z
0
_o
3-79
...... lr J
g_
_r
_ ,_
god
c,l
t
g
o _ f'--I
•_ ..
gl _
N
o
rj
N J_ -_ _ •
0
L_
I.
o
Z° Z L_._J
Ca
_ _._
II
v
i
I m ×
°
,_,_= ._,
g
°_,1 ,
r/?
k -T
8
,-_ ..o i 0 ,
b'_ _ , 0
__j
'ii £ _:
3-80
_ _ _ o
o _ o_
._ _ o_ _
_ , _.
o _ _ o_
_
0
, _'_
_ o_ _ _._
o _ _ _ _._
°_ _ _ _.-_ o o.
.o _ _ . _
b._ • I o _ _ "_
,._ _ _ _._
0
o _ ._ _ _.
-t
.._ _ _ _ _o
II
o _ ,, __ _4
• _ _ ,__ _ _ ._._
_,_,4 I
L! iJ
l/ \,II
D-
3-81
._..E
_._ o
r_
o
_.
0
L_ e- *') _1
_ _°
_._ _ _._
o_ _ _ _ _._
r._ o
_d
_ z_ ._ _
_ z
m
o o • _ _ "_
Z _ ._ _ _
_' _ z _'
_ °_' _ ° _
M I!
_,_ _ _ ,, _ _ _ m
_ • _ M_ 2_..
_._ _ hi)' i
_._ _:
e_
o
,,,4 8
o - I I o
<
r_
m
m
I I IJ
e_
>._ir-'J , , ' <
u o
z
I
8_
3-82
0
2
0 _ "_
z
_ o.
o _ ._ "___
C
° o_ ,, °_o ._
o_ _ z_ _
._ o° o
o_
_ _ z_
+
_.o C_
Z _ o_ o
c_
II
t_ _ _ 0 2
_o _
o
o xj-F__
I I i I I I_ _
<i
3-83
REFERENCES
3-1 U. S. Department of Defense, "Structural San_vich Composites, " MIL-
HDBK-23, 30 December 1968.
3-2 General Dynamics Fort Wo,__th Division, Structures Manual - Volume 1,
Methods of Analysis, November 1959.
3-3 Konishi, D., et a], Honeycomb Sandwich Structures Manual, North
American Aviation Corp. Los Angeles Division, Report NA58-899.
3-4 Structures Manual - Section 11.00: Honeycomb Sandwich Structures,
North American Aviation, Inc. Space and Information Systems Division,
May 1964.
3-5 Lockheed-Georgia Company, A Division of Lockheed Aircraft Corp.,
"Sandwich Panel Tests for C-5A Airplane," GELAC Report No. ER 8976,
22 February 1967.
3-6 General Dynamics Corporation Forth Worth Division, "Design Allowables
and Methods of Analysis, F-111 Airplane," Report No. FZS-12-141,
1 August 1965.
3-7 Hexcel Corp., "Design Handbook for Honeycomb Sandwich Structures,"
Report No. TSB 123, October 1967.
3-8 Ericksen, W. S. and March, H. W., "Fffects of Shear Deformation in the
Core of a Flat Rectangular Sandwich Panel - Compressive Buckling of
Sandwich Panels Having Dissimilar Facings of Unequal Thickness," Forest
Products Laboratory Report 1583-B, Revised 1958.
3-9 Norris, C. B., "Compressive Buckling Curves for Flat Sandwich Panels
with Isotropic Facings and Isotropic or Orthotropic Cores, " Forest
Products Laboratory Report 1854, Revised 1958.
3-85
3-10 Norris, C. B., "Compressive Buckling Curves for Flat Sandwich Panels
with Dissimilar Facings, " Forest Products Laboratory Report 1875,
September 1960.
3-11 Kuenzi, E. W., Norris, C. B. and Jenkinson, P. M., "Buckling Coeffi-
cients for Simply Supported and Clamped Flat, Rectangular Sandwich Panels
Under Edgewise Compression, " Forest Products Laboratory Research Note
FPL-070, December 1964.
3-12 Jenkinson, P. M. and Kuenzi, E. W., "Buckling Coefficients for Flat,
Rectangular Sandwich Panels with Corrugated Cores Under Edgewise
Compression, " Forest Products Laboratory Research Paper FPL 25,
May 1965.
3-13 Kuenzi, E. W., and Ericksen, W. S., "Shear Stability of Flat Panels of
Sandwich Construction, " Forest Products Laboratory Report 1560, 1951.
3-14 Harris, L. A. and Auelman, R. R., "Stability of Flat, Simply Supported
Corrugated Core Sandwich Plate Under Combined Longitudinal Compression
and Bending, Transverse Compression and Bending, and Shear," North
American Aviation, Inc., Missile Division, Report STR 67, 1959.
3-15 Kimel, W. R., "Elastic Buckling of a Simply Supported Rectangular
Sandwich Panel Subjected to Combined Edgewise Bending and Compression, "
Forest Products Laboratory Report 1857A, 1956.
3-16 Harris, L. A., and Auelman, R. R., "Stability of Flat, Simply Supported
Corrugated Core Sandwich Rectangular Plates Under Combined Loadings,"
Journal of Aerospace Sciences, Vol. 27, No. 7, p. 525-534, 1960.
3-17 Timoshenko, S. P., and Gere, J. M., Theory of Elastic Stability, McGraw-
Hill Book Company, Inc., 1961, p. 373-379.
3-18 Gerard, G., and Becker, H., Handbook of Structural Stability, Part I-
Buckling of Flat Plates, National Advisory Committee for Aeronautics
Technical Note 3781, 1957.
3-86
3-19 Kimel, W. R., "Elastic Bucklingof a SimplySupportedRectangularSand-
wich PanelSubjectedto CombinedEdgewiseBending,Compression,and
Shear," ForestProductsLaboratoryReport1859,1956.
3-20 Noel, R. G., "Elastic Stabilityof SimplySupportedFlat RectangularPlates

UnderCritical Combinationsof LongitudinalBending,LongitudinalCom-
pression, andLateral Compression,"Journalof the AeronauticalSciences,
Vol. 19, No. 12, p. 829-834, 1952.
3-21 Norris, C. B., and Kommers, W. J., "Critical Loads of a Rectangular,
Flat Sandwich Panel Subjected to Two Direct Loads Combined with a Shear
Load," Forest Products Laboratory Report 1833, 1952.
3-22 Norris, C. B., and Kommers, W. J., "Stresses within a Rectangular,
Flat Sandwich Panel Subjected to a Uniformly Distributed Normal Load and
Edgewise, Direct, and Shear Loads, " Forest Products Laboratory Report
1838, 1953.
3-23 Plantema, F. J., Sandwich Construction, John Wiley and Sons, Inc., New
York, 1966.
3-24 U. S. Department of Defense, "Plastics for Flight Vehicles: Part I, Rein-
forced Plastics, " MIL-HDBK-17, Armed Forces Supply Support Center,
1959, Available from U. S. Government Printing Office, Washington, D. C.
3-25 U. S. Department of Defense, "Metallic Materials and Elements for Aero-
space Vehicle Structures," MIL-HDBK-5A, February 1966, Available from
U. S. Government Printing Office, Washington, D. C.
3-87
4
GENERAL INSTABILITY OF CIRCULAR CYLINDERS
4.1 GENERAL
In the case of axially compressed, thin-walled, isotropic (non-sandwich) cylinders, it
has long been recognized that test results usually fall far below the predictions from
classical small-deflection theory [4-1]. These discrepancies are usually attributed
primarily to
a. the shape of the post-buckling equilibrium path coupled with the presence
of initial imperfections
and
b. the fact that classical small-deflection theory does not account for pre-
buckling discontinuity distortions in the neighborhood of the boundaries.
Neglecting the discontinuity distortions, the equilibrium path for an axially com-
pressed perfect cylinder is of the general shape shown by the solid curve in Figure
4.1-1. This path is linear until point A is reached and general instability occurs at
a stress level crCL equal to the result from classical small-deflection theory
[ °'eL - R _/3(1-Ve _) for elastic, isotropic (non-sandwich) cylinders However, if
the cylinder is initially imperfect and the discontinuity distortions are considered, the
behavior will be as shown by curve 0B and buckling will occur at the stress O'cr. The
rrtio (Crcr/erCL) will be dependent upon the magnitude of the initial imperfections pres-
ent in the cylinder. However, since this information is not normally available, one
4-1
usuallyfinds it necessaryto resort to either of the followingpracticesto obtainpracti-
cal designvalues:
a. Setthe allowablecompressivestress equalto the valuecrMi N shown in
Figu_e 4. i-I.
b. Use the classical small-deflection value o-CL in conjunction with a suitable
knock-down factor )'e which is based on the results from a large array of
test data. The allowable compressive stress is then obtained from
(rcr = )'c_CL (4.1-I)
Perfect
Cylinder
A Imperfect
crCL Cylinder
Axial
Compressive
Stress
O'cr
C
O'MiN
0
End Shortening
Figure 4.1-1. Equilibrium Paths for Axially

Compressed Circular Cylinders
For isotropic (non-sandwich) cylinders it is common practice to follow the second of
these approaches and, for such cylinders, the test data shows that _'c is a function of
the radius-to-thickness ratio (R/t).
4-2
In the caseof sandwichcylindershavingrelatively rigid cores, the behavioris similar
to that of the isotropic (non-sandwich)

cylinder andonecanexpectimperfectionsand
boundarydisturbancesto precipitategeneralinstability at compressivestressesbelow
the predictionsfrom classical small-deflectionsandwichtheory, ttowever, in most
practical applications,the sandwichwall will provide aneffectiverelatively thick shell
so thatthe discrepancieswill not be as large as thosenormally encountered

in thin-
walledisotropic (non-sandwich)
cylinders. In addition,as the coretransverseshear
rigidity decreases,the differencesbetweentest results andclassicalpredictionswill
diminish. In the extremecasewhereshearcrimping occurs, initial imperfectionsdo
not appearto haveanyinfluence.
Oneof the mostprominentof the early designcriteria developedfor axially compressed
circular sandwichcylindersis that of Reference4-2. This solutionemployedlarge-
deflectiontheorytogetherwith approach(a) cited above(Crcr= °'MIN). However, it is
now rather generally agreed that this criterion often provides design values which are
too conservative. In addition, the theoretical development of Reference 4-3 indicates
that CrMi N can be decreased to essentially zero by including a sufficient number of
terms in the large-deflection displacement functions. Therefore, in recent years, it
has become common practice to design sandwich cylinders by method (b) cited above
[4-4 and 4-5]. This approach, which employs small-deflection theory in conjunction
with an empirical knock-down factor, is likewise followed in this handbook.
In the treatment of various types of external loading, it is important to note that the
characteristics of the equilibrium paths are not identical for cases of axial compres-
sion, torsion, or external radial pressure. For purposes of comparison, Figure 4.1-2
4-3
depictsthe generalshapesof thesepathsfor eachloadingcondition[4-6] assuming
that the cylindersare initially perfect andthatnodiscontinuitydistortions are present.
Deflection Deflection Deflection

Axial Compression Torsion ExternalPressure
Figure 4.1-2. Typical Equilibrium Pathsfor Circular Cylinders
Basedonthe relative shapesof thesecurves, onewouldexpectthat, undertorsion or
externalpressure, the cylinderswouldbemuchless sensitiveto initial imperfections
thanin the caseof axial compression. This has beenborneout by the availabletest
datafrom thin-walledisotropic (non-sandwich)

cylinders.
4-4
4.2 AXIAL COMPRESSION
4.2.1 Basic Principles
4.2.1.1 Theoretical Considerations
The theoretical basis used here is the classical small-deflection solution of Zahn and
Kuenzi [4-7] which includes the follox_ing assumptions:
a. The facings are isotropic but the core may have orthotropic transverse
shear properties.
b. Bending of the facings about their own middle surfaces can be neglected.
c. The core has infinite extensional stiffness in the direction normal to the
facings.
d. The core e×_tensional and shear rigidities are negligible in directions
parallel to the facings.
e. The cylinder is not extremely short (a quantitative limit is specified in
Section 4.2.2).
f. The approximations of Donnell [4-8] can be applied without introducing sig-
nific ant error.
In this handbook, the final equations of Reference 4-7 have been transformed into
equivalent formulations which should be more meaningful to the user. For those cases
where the core shear moduli satisfy the condition
Gxz
O- _;1 (4.2-1)
Gyz
the following expression is obtained:
_cr = Kc(ro (4.2-2)
4-5
where
h 2 _r}-_l t 2
(4.2-3)
_o = _Ef--_ l_:-_ee (tl + t_)
and
1
When Vc _ 2 (4.2-4)
Kc = 1 - -_- V c
1
When Vc > 2
Kc = V--c (4.2-5)
where
o-o
Vc - (4.2-6)
°'crimp
5 2
(4.2-7)
_rcrim p - (t 1 + t2)t c Gxz
z
Plasticity reduction factor, dimensionless.
E f= Young' s modulus of facings, psi.
h= Distance between middle surfaces of facings, inches.
R= Radius to middle surface of cylindrical sandwich, inches.
t andt_ = Thicknesses of the facings (There is no preference as to which

facing is denoted by the subscript 1 or 2.), inches.
ue = Elastic Poisson' s ratio of facings, dimensionless.
tc = Thickness of core, inches.
Gxz --
Core shear modulus associated with the plane perpendicular to
the facings and oriented in the axial direction, psi.
Gy z = Core shear modulus associated with the plane perpendicular to

the axis of revolution, psi.
The relationship between Kc and Vc can be plotted as shown in Figure 4.2-1. It is
important to note that the value V c = 2.0 establishes a dividing line between two
different types of behavior. The region where V c _ 2.0 covers the so-called stiff-core
4-6
1.0
K c
I
I
I
2.0
Vc
Figure 4.2-1. Schematic Representation of Relationship

Between Kc and Vc for 0_< 1
and moderately-stiff-core sandwich constructions When Ve is in the neighborhood of
zero, the core shear stiffness is high and the sandwich exhibits maximum sensitivity to
initial imperfections. Hence, for any given radius-to-thickness ratio, the knock-down
factors applicable to such constructions are of maximum severity. As Ve increases
from zero to a value of 2.0, the sensitivity to Imperfections beeomes progressively
less. The domain where V c > 2.0 is the so-called weak-core region where shear
crimping occurs. Sandwich constructions which fall within this category are not influ-
enced by the presence of initial imperfections, and a knock-down factor of unity ean be
applied to such structures. It should be possible to develop a continuous transitional
knock-down relationship which recognizes the variable influence of the core rigidity but
this is beyond the scope of the present handbook.
4-7
4.2.1.2 Empirical Knock-DownFactor
Asnotedin Section4.1, the allowablestress intensitiesfor axially compressed,thin-
walled, isotropic (non-sandwich)

cylinders areusually computedusingthefollo_ving
equation:
_cr = _c °-cL (4.2-8)
The quantity Yc is referred to as the knock-down factor and this value is generally
recognized to be a function of the radius-to-thickness ratio (R/t). Various investi-
gators have proposed different relationships in this regard. The differences arise out
of the chosen statisticalcriteria and/or out of the particular test data selected as the
empirical basis. One of the most widely used of the relationships proposed to date is
the lower-bound criterion of Seide, et al. [4-9] which can be expressed as follows:
}c = 1 - 0.901(1 - e -¢) (4.2-9)
where
1 R
(4.2-10)
This gives a knock-down curve of the general shape depicted in Figure 4.2-2. For the
purposes of this handbook, it is desired that an empirical means of this type also be
provided for the design of sandwieh cylinders. One of the major obstacles to the
achievement of this objeetivo, is the lack of sufficient sandwich test data for a thorough
empirieal determination. Faced with this deficiency, one finds it expedient to employ
the data from isotropie (non-sandwich) cylinders in conjunction with an effective thick-
ness concept and eorrection factors which are based on the few available sandwich test
points. Toward this end, it is usually assumed that, when V c _< 2.0, equal sensitivity
4-8
1.0
Yc
Log Scale
Figure 4.2-2. Semi-Logarithmic Plot of YC vs R/t for Isotropie (Non-

Sandwich) Cylinders Under Axial Compression
to imperfections results from equivalence of the shell-wall radii of gyration p
h
(_ _ for sandwich constructions whose two facings are of equal thickness). There-
fore, the approach taken here is to rewrite Equations (4.2-9) and (4.2-10) in terms
of P. The revised formulations give the plot shown as a dashed curve in Figure
4.2-3. Also sho_ in this figure are the appropriate test points obtained from
axially compressed sandwich cylinders [4-2, 4-10, 4-28] which did not fall into
the weak-core category. Eleven such data points are shown. In addition, two
test points are shown for axially compressed conical sandwich constructions
[4-10] which likewise did not lie in the weak-core region. The conical data are
included in Figure 4.2-3 in view of the scarcity of available test results and
also because the cones were analyzed as equivalent cylinders whose radii were
taken equal to the Rz (finite principal radius of curvature) values at the small end
of the specimens. Based on this limit_i amount of sandwich test data, it is recom-
mended that the solid curve of Figul"e 4.2-3 be used for design purposes. This
4-9
1.0
.8
.7
.6
.3
.2
.1
10 103
Figure 4.2-3. Knock-Down Factor Yc for Circular Sandwich

Cylinders Subjected to Axial Compression
4-10
gives Ycvaluesthat are 75percentof thoseobtainedfrom the dashedcurve whichwas
basedonthe empirical formulaof ,':_eide,

eta]. [4-9].
In additionto the test results describedabove,a considerablenumberof test points
are availablefrom cylindrical sandwichconstructionswhichfall into the weak-core
classification. Asnotedin Section4.2.1.1, the methodsrecommended

in this hand-
bookare suchthat, in the weak-coreregion, no empirical reductionwill be applied
to the theoreticalresults of Reference4-7. In order to explorethe validity of this
approach,plots are furnishedin Figures4.2-4 and4.2-5 whichcomparethe weak-
core test results of References4-2 and4-11againstpredictionsfrom the recom-
mendeddesigncriterion. It canbe seenthat all but oneof the test results exceed
the predictedstrengths,andthatthe singleexceptionfailed at 86percentof the pre-
dictedvalue. In manyof the caseswhere (O-CrTest/O-Predicted)

> 1.0, althoughthe
discrepanciesmeasuredin units of psi werenot very great, thepercentagediffer-
enceswerequitelarge. This behaviorcanbe explainedby the fact that the theoreti-
cal basis [4-7] proposedin this handbookassumesthat bendingof the facingsabout
their ownmiddle surfacescanbe neglected. As shownin Reference4-12, this
assumptioncanbe very conservativein the weak-coreregion. However,in the
interest of simplicity, the methodsof this handbookretain this assumptionespecially
sinceit is a conservativepractice andmostpractical sandwichconstructionswill not
bedesignedas weak-corestructures.
In view of the meager compressive test data available from stiff-core and moderately-
stiff-core sandwich cylinders, the method proposed here is not very reliable when
4-11
Vc < 2.0. Therefore, in suchcasesthe methodcanonlybe consideredas a "best-
available"approach. On the other hand, where the failure is by shear crimping
(V c _>2.0), the method is quite reliable and will, in fact, usually give conservative
predictions.
2O
15
• _,_b.
10 8
_D
Test Data from Reference 4-2
I I
5 10 15 20
Predicted O'cr, ksi
(Neglecting bending stiffness of individual facings)
Figure 4.2-4. Comparison of Proposed Design Criterion Against Test Data for Weak-
Core Circular Sandwich Cylinders Subjected to Axial Compression
4-12
2OO
Test Point from

Reference 4-11
lOO
100 200
Predicted Crcr , ksi
(Neglecting bending stiffness

of individual facings)
Figure 4.2-5. Comparison of Proposed Design Criterion Against

a Test Result for a Weak-Core Circular Sandwich
Cylinder Subjected to Axial Compression
4.2.1.2.1 Interpretation of Test Data
As indicated in the preceding paragraphs, appropriate test data must be used in order
to arrive at practical values for the knock-down factor. However, one can be easily
misled in this endeavor when the test data and/or the classical theoretical predictions
lie in the inelastic region. To demonstrate this point as simply as possible, the pre-
sent discussion is limited to the case of axially compressed circular sandwich cylin-
ders for which V c = 0. Then the recommended design value for the critical stress
can be expressed as follows:
h 2 _ t_ (4.2-11)
°'cr = Yc _ Ef R I_Z-Y-Ye2 (tl + t2)
4-13
For anyparticular test specimen,the relatedvaluefor the knock-downfactor shouldbe
computedfrom the followingexpressionwhichis obtainedby a simpletranspositionof
Equation (4.2-11):
C(rCrTest_
\ "_Test /
= (4.2-12}
(YC)Test [E fRh 2vf_-lt_(t_

_ t_) ]
The plasticity reduction factor DTest is evaluated at the actual experimental buckling
stress. By inspection of the numerator and denominator of Equation (4.2-12), one can
conclude that this formula may be rewritten in the following more meaningful form:
wEXperimental critical stress value]

hich would have been attained had]
the material remained elastic ]
(Yc)Test _lassical theoretical critical" (4.2-13)
| stress value assuming the
L behavior to be elastic
The example illustrated in Figure 4.2-6 should help to clarify this concept. In this
figure, the solid line represents the stress-strain curve for the test specimen mate-
rial. Suppose that this particular specimen buckled at a stress equal to O-CrTest . As
indicated in the figure, it is assumed here that this stress lies in the inelastic region
so that NTest will be less than unity. For the purposes of this discussion, further
assume that _?Test - 0.80. If the material had remained elastic, the experimental
critical stress would have been somewhat higher than OcrTest . This greater value
will be denoted as CrcrTest . Then it follows that
(4.2-14)
°'CrTest °-CrTest
- 1.25 (rCrTest
CrcrTest - _)Test 0.80
4-14
¢rCL
(r_rTest
O'Ma x
_rcr
Test
Strain
Figure 4.2-6. Stresses Involved in Interpretation of Test Data
Now let it also be assumed that, using elastic material properties, the classical thee-
retical critical stress equals the value o-CL indicated in Figure 4.2-6. The following
formula would then give the proper value for the experimental knock-down factor:
O'CrTest 1.25 ¢rcrTe st

( ¥C)Test - (4.2-15)
¢rCL oCL
where
h 2v/t _ tz
(rCL = Ef _ _pe z {tl +t2) (4.2-16)
4-15
The abovediscussionis givenhere sincesomeof the results presentedin the literature
canbe quitemisleading. That is, comparisonsare oftenshownbetweenthe actualtest
value (rcrTest (withoutregard as to whetherelastic or not) andthe inelastic classical

theoreticalprediction. For the caseshownin Figure4.2-6, the latter valuecannot
exceed _Max andthis type of comparisonmight lead oneto believethat the appropriate
knock-downfactor is very closeto unity. However,use of the correct approachas
expressedby Equations(4.2-13)and(4.2-15)givesa muchlower Yc value. For any
given geometry, one could always show very close agreement between O'crTest and
O-Ma x simply by choosing a material with a sufficiently low yield strength and having
a flat post-yield stress-strain curve.
4-16
4.2.2 DesignEquationsandCurves
For simply supportedcircular sandwichcylinders subjectedto axial compression,the
critical stressesmaybe computedfrom the relationshipsgivenonpage4-18wherethe
subscripts1 and 2 refer to the separatefacings. "I2aere is no preference as to which
facing is denoted by either subscript. These equations were obtained by a simple ex-
tension of the formulas developed in Reference 4-7 which only considered the case
where the behavior is elastic and the moduli of elasticity are identical for both facings.
The extended versions given in this handbook were derived through the use of equivalent-
thickness concepts based on the ratios of the moduli of the two facings. For cases
where the two facings are not made of the same material, these equations are valid
only when the behavior is elastic (7 = 1). Application to inelastic cases (_? _ 1) can
only be made when both facings are made of the same material. For such configura-
tions, El and E_ will, of course, be equal.
The buckling coefficients Kc can be obtained from Figure 4.2-7. Curves are given
Gxz
there for both 0 _< i and 0 = 5 where 0 - Since these two plots are not very
Gyz
different from each other, one may use Figure 4.2-7 to obtain rather accurate esti-
mates of Kc when 1< 0 <5.
Whenever V c < 2.0, the knock-down factor Yc can be obtained from Figure 4.2-8.
When V c _> 2.0, use Yc = 1.0.
For elastic cases, use _) = 1. Whenever the behavior is inelastic, the methods of
4-17
! I I I I
cq L'_
+ IJ
j
O
oJ
17 A
O
v
¢--..J
v-4
I I I I
I O,1 o,1
-.>
O
O
¢,J
+
o
¢7 L)
g-
°,.._
¢7
_>
4-18
JiJ, i¸ il
........ :!:i!
_9
0
_9
<
i
}.
1
_9
4-19
1.o _i:il
..... :::i_/_'!!I
:i!_
..... ii_i
!_....' e ...............
::i
.9 i:, x/(EI tl) (E2 t2) h
•8 'ki_ i_L[ii_4.!i_
i T i_
.............
'!t
i i,I
_'C
T:t:T!_-T, ,, jklt....
, F3 dii_t
_:ii11 i!ii
.2 : J'_' [71_i!:7
_ _ti3 : lit, " 3!7!t7['_i
I0 10 _ 10 a
Figure 4.2-8• Design Knock-Down Factor for Circular Sandwich

4-20
The critical axial load (in units of pounds) can be computed as follows:
Pcr = 27rR [_cr_ tl + _cre t_] (4.2-26)
In the special case where t_ = t2 --- tf and both facings are made of the same material,
Equations (4.2-17) through (4.2-26) can be simplified to the following:
(rcr = Yc Kc ¢ro (4.2-27)
(_Ef) h (4.2-28)
5 2
°'crimp Gxz (4.2-29)

2 tf t c
o-o
Vc - (4.2-30)
arcrimp
Pcr = 41rRtf o-cr (4.2-31)
Equations (4.2-17) through (4.2-31) and Figure 4.2-7 are valid only when the length L
of the cylinder is greater than the length of a single axial half-wave in the buckle pat-
Gxz
tern for the corresponding infinite-length cylinder. For the case where _ - Gy z - 1,
one can apply the following test to determine if the cylinder length is sufficiently large:
When V < 2
el When Vcl > 2
Equations (4.2-17) through (4.2-31) Equations (4.2-17) through (4.2-31)
and Figure 4.2-7 are valid only where and Figure 4.2-7 are valid for any
value of L.
4-21
For constructionswhere e _ 1, nocorrespondingnumericalcriterion is presently
available. In suchcases, onecanonly usethe abovetest in conjunctionwith engineer-
ingjudgement. It is helpfulto pointout, however,thatmostpractical sandwichcylin-
ders for aerospaceapplicationswill besufficiently longfor Equations(4.2-17)through
(4.2-31) and Figure 4.2-7 to be valid. In addition, it is comforting to note that the
use of these relationships for shorter cylinders results in conservatism.
Cylinders which fail to meet the foregoing length requirement are usually referred to
as short cylinders. The only means available for the analysis of such sandwich cylin-
ders under axial compression is the solution of Stein and Mayers [4-13] which is only
valid
a. when 0 = 1
and
b. when both ends of the cylinder are simply supported
and
c. when both facings are made of the same material
and
d, the thickness of one facing is not more than twice the thickness of the
other facing.
For short sandwich cylinders which satisfy these conditions, one can use the design
curves of Figure 4.2-9 which involves the following parameters:
Z - 2L_ (4.2-32)
Rh 1-_/_e2
4-22
2
rr D
r a - (4.2-33)
L 2 Dq
• _cr (t_ + t_) Le

(4.2-34)
Kc = ycrr 2 D
where
(_ h)(E2 ts)h2
(4.2-35)
D = r](l_Pe_) [(F__tl) + (F__t_i]
h _
Dq = _-c Gxz (4.2-36)
and
L = Over-all length of cylinder, inches.
During the preparation of this handbook, no solutions were uncovered for axially com-
pressed sandwich cylinders having any degree of rotational restraint at the boundaries.
However, in most practical aerospace applications, the cylinders will be sufficiently
long for such fixity to have negligible effects on the buckling loads.
4-23
...... ]
70 ---- -_
1
4o- - + ....
•_ _
30-: _ I
I '
20-=' i
::ii4' :!
1
10_ _, :
7- --'_-, :
!
4- ..... i
K
C
• i:
.5- _ ::
.4- ,!
• 9
Figure 4.2-9. Buckling Coefficient for Short Simply-Supl_orted Sandwich Cylinders

Subjected to Axial Compression (0 :: 1)
4-24
4.3 PUIIE BEN1)ING
Based on small-deflection theory, investigations were made in References 4-14, 4-15,
and 4-16 of elastic instability in thin-walled, isotropic (non-sandwich) cylinders sub-
jeeted to axial compressive stresses which vary in the circumferential direction. From
the results of these references, it can be concluded that, regardless of the nature of
the circumferential stress distribution, classical instability is reached when the peak
axial compressive stress satisfies the condition
Et
o" _.6 R (4.3-1)
It should be recalled that the value . 6 Et/R is also obtained from the small-deflection
solution for thin-walled, isotropic (non-sandwich) cylinders subjected to uniform axial
compression. In view of this result, one might reasonably expect that small-deflection
sandwich theory would also indicate that only the peak axial compressive stress need
be considered in cases of pure bending or combined bending and axial compression. It
has been shown in References 4-17, 4-18, and 4-19 that this is indeed the case. Ref-
erences 4-17 and 4-18 demonstrate this for weak-core sandwich cylinders while Ref-
erence 4-19 deals with infinitely long cylinders which fall in the stiff-core and moder-
ately-stiff-core categories. Therefore, for the purposes of this handbook, it is assumed
that the theoretical considerations of Section 4.2 (axial compression) apply equally well
to sandwich cylinders which are subjected to pure bending if the analysis considers only
the peak value of the applied compressive stress. The only differences lie in the em-
pirical knock-dow_l factors recommended for the two cases.

4-25
4.3.1.2 Empirical Knock-I)ownFactor
In the caseof pure bending, only a relatively small portion of the cylinder's circumfer-
ence experiences stress levels which initiate the buckling process. Because of the
consequent reduced probability for peak stresses to coincide with the location of an
imperfection, it is to be expected that the knock-down factors for pure bending will not
be as severe as the corresponding factors for axial load. For thin-walled, isotropic
(non-sandwich) cylinders under pure bending, Seide, et al. [4-9] have proposed the
following lower-bound relationships:
Yb = 1 - 0.731 (1 - e -¢) (4.3-2)
where
= _ (4.3-3)
Comparison against Equations (4.2-9) and (4.2-10) shows that this bending criterion
does indeed give Yb values of lesser severity than those which apply to the axially
compressed cylinders. Following the same approach as that taken in Section 4.2, the
h
above equations are rewritten in terms of the shell-wall radius of gyration p(_ for
sandwich constructions whose two facings are of equal thickness). The revised formu-
lations then give the plot shown as a dashed curve in Figure 4.3-1. Also shown in this
figure are the appropriate test points from stiff-core sandwich cylinders subjected to
pure bending [4-201. Since only three such data points are available, it was thought to
be helpful to include the axial compression sandwich data points previously shown in
Figure 4.2-3. To fully understand the information given in Figure 4.3-1, it is im-
portant for the reader to be aware of the data reduction techniques used here. For an
4-26
explanation of the procedures used in this handbook, reference should be made to the
related discussion in Section 4.2.1.2.1.
Based on the limited amount of available test data, it is recommended here that the
solid curve shown in Figure 4.3-1 be used for the design of sandwich cylinders sub-
jected to pure bending. This gives T b values that are 75 percent of those obtained
from the dashed curve which is based on the empirical formula of Seide, et al. [4-9].
This is consistent with the practice followed in Section 4.2 for the case of axial com-
pression where the design knock-down factor was likewise taken to be 75 percent of
the value obtained from the corresponding curve derived from Reference 4-9.
In view of the meager test data available from sandwich cylinders under pure bending,
the method proposed in Section 4.3.2 is not very reliable when V c < 2.0. Therefore,
in such cases, the method can only be regarded as a "best-available" approach. On
the other hand, when the failure is by shear crimping (V c >_ 2.0), the method is quite
reliable and will, in fact, usually give conservative predictions.
4-27
• !
(_) Data from Ref. 4-20 (Cylindrical; Pure Bend.)

+ Data from Ref. 4-2 (Cylindrical; Axial Compr.) t
• Data from Ref. 4-10 (Cylindrical; Axial Compr.)_
10 a
10
Figure 4.3-1. Knock-Down Factor _ for Circular Sandwich
Cylinders Subjected to Pure Bending
4-28
4.3.2 Design Equations and Curves
For simply supported sandwich cylinders subjected to pure bending, one may use the
same design equations and curves as are given in Section 4.2.2 (for axial compression)
except for the following:
a. For the ease of pure bending, use Figure 4.3.2 to obtain the knock-down
factor Yb whenever V e < 2.0 (When Vc >_2.0, use Yb = 1.0).
bo For the case of pure bending, the critical stresses obtained from the
equations and curves of Section 4.2.2 correspond to the circumferential
location which lies on the compressive side of the neutral axis and is
furthest removed from that axis. Hence the computed stresses are the
peak values within the variable circumferential distribution. Therefore,
when the behavior is elastic, the critical bending moment Mcr can be
computed from the following:
Mcr = rrR _ [¢rcr 1 t_ + _cr2 t_] (4.3-4)
where
Mcr = Critical bending moment, in.-lbs.
R = Radius to middle surface of sandwich cylinder, inches.
(rcr: and O-cra = Critical compressive stresses in facings 1 and 2, respec-
tively, which result in general instability of the cylinder, psi.
tl and ts = Thicknesses of the facings 1 and 2, respectively, inches.
Note: There is no preference as to which facing is denoted by the
subscripts 1 and 2.
4-29
1.0
.9
6
_b
5
.2
.I
I0 io_
(s)
P
Figure 4.3-2. Design Knock-Down Factor "¢b for Circular Sandwich
Cylinders Subjected to Pure Bending
4-30
To computel_cr whenthe behavioris inelastic, onemust resort to numericalintegra-
tion techniques.
Sincethe procedurerecommended
heremakesuseof the methodsof Section4.2.2, all
of the limitations of that sectionare equallyapplicableto the presentcase. Thatis,
onlysimply supportedboundariesare consideredandthe primary solutionis excess-
ively conservativefor the so--calledshort-cylinderconstructions. In addition,only
very limited meansare availableto facilitate a quantitativeassessmentof whetheror
not a particular constructionfalls within the short-cylinderclassification. Further-
more, the computationof critical stressesfor short-cylinder constructionscanonly
be accomplished
for rather specialcasesas cited in Section4.2.2.
Asnotedin Section4.2.2, duringthe preparationof this handbook,no solutionswere
uncoveredfor axially compressedsandwichcylindershavinganydegreeof rotational
restraint at the boundaries. However,it was alsonotedthat, in mostpractical aero-
spaceapplications,the cylinders will be sufficiently longfor suchfixity to havenegli-
gible effectsonthe critical stresses. The samesituationexists for the caseof pure
bending.
4-31
4.4 EXTERNAL LATERAL PRESSURE
This section deals with the loading condition depicted in Figure 4.4-1. Note that the
sandwich cylinder is subjected to external pressure only over the cylindrical surface.
No axial loading is applied. In addition, it is specified that the ends are simply sup-
ported. That is, during buckling, both ends of the cylinder experience no radial dis-
placements and no bending moments.
p, psi
ends
s imply s uppo
Both rt_
p, psi
Figure 4.4-1. Circular Sandwich Cylinder Subjected to

External Lateral Pressure
The theoretical basis used here is the classical small-deflection solution of Kuenzi,
et al. [4-21] which includes the following assumptions:
a. The facings are isotropic.
b. The facings may be of equal or unequal thicknesses.
c. The facings may be of the same or different materials.
d. Poisson's ratio is the same for both facings.
e. Bending of the facings about their own middle surfaces can be neglected.
f. The core has infinite extensional stiffness m the direction normal to the
facings.
g. The core extensional and shear rigidities are negligible in directions

4-32
h. The transverse shear properties of the core may be either isotropic or
o rthotropic.
2R
i. The inequality _- >> 1 is satisfied.
j. Several additional order-of-magnitude assumptions are valid, as noted

below in connection with Equation (4.4-2).
The solution of Kuenzi, et al. [4-21] draws upon the earlier groundwork laid by
Raville in References 4-22, 4-23, and 4-24. Norris and Zahn used these reports to
develop design curves which are published in References 4-25 and 4-26. The work of
Kuenzi, et al. [4-21] constitutes the latest revision to this series of reports and is
the most up-to-date treatment of the subject. However, the format of their results
has been slightly modified in Reference 4-5 in order to reduce the scope of interpola-
tion required in practical applications. The revised format is used here. However,
the need for interpolation has not been entirely eliminated since separate families are
still required for each of the selected values for Vp [see Equation (4.4-4)].
The final theoretical relationships used in this handbook are as follows:
Cp [(Eltl) + (E_)] (4.4-1)

Pcr = R (l-re _)
where
Cp = Minimum value (with respect to n) of Kp , dimensionless.
and
(4.4-2)
_2 (n 2 _ 1) (3 + --_--_/L\_-'_R
n2 L2 _[/n2 L22 1)(nL 7v:'R2
1+ _-)-2]+ 98 [1+ (n 2 + _]Vp]
+
4-33
(Ez tl) (E2 t_) h2
L_/2
(4.4-3)
[(Eltl) * (E_t_)laW
(Eltl) (Ezt2) h
Vp =
[(El tl) + (Ez t2)] (1;,seg)R_-Gi, z (4.4-4)
whe re
Per = Critical value of external lateral pressure, psi.
It = Radius to middle surface of cylindrical sandwich, inches.
e - Elastic Poisson's ratio of facings, dimensionless.
Et and E2 = Young's modtdi of facings 1 and 2, respectively, psi.
tx and te = Thicknesses of facings 1 and 2, respectively, inches.
n
= Number of circumferential full-waves in the buckle pattern,
dimensionless.
L - Over-all length of cylinder, inches.
h - Distance between middle surfaces of facings, inches.
Gyz - Core shear modulus associated with the plane perpendicular

to the axis of revolution, psi.
Note: There is no preference as to which facing is denoted by the

subscript 1 or 2.
For cases where the two facings are not made of the same material, the foregoing
formulas are valid only when the behavior is elastic (_ = 1). Application to inelastic
cases ( E _ 1) can only be made when both facings are made of the same material.
For such configurations, E_ and E_ will, of course, be equal.
4-34
Equation(4.4-2) constitutesanapproximateexpressionfor Kp sinceit embodiesthe
assumptionscited earlier in this sectionin additionto the following:
a. Terms containingKp and 4 R2 wereneglected.
b. It wasassumedthat (1 + m2_) = 1, where m is a small whole number.
By using Equation (4.4-2), plots can be generated of the form shown in Figure 4.4-2.
The design curves of this handbook are of this type and were taken directly from Ref-
erence 4-5. It is helpful to note here that lower and upper limits exist for the coeffi-
cient Cp and these are identified in Figure 4.4-2. The lower limit is associated with
long-cylinder behavior. Such configurations are unaffected by the end constraints and
the related critical pressures are equal to those for rings which are subjected to
external pressure. For portions of the cylinders that do not lie in the neighborhoods
of the boundaries, the buckle patterns will be the same as are obtained from such rings.
In this connection, it should be noted that application of the Donnell approximations
[4-8] to non-sandwich rings leads to critical pressures which are 33 percent higher
than the predictions from accurate ring formulations. This is due to the fact that the
related number of circumferential full-waves (n = 2} is not sufficiently high to justify
Donnell's [4-8] assumptions. It is important to observe that the theory of Reference
4-21 retains a sufficient number of terms to accurately predict the buckling of long
cylinders. That is, when Gy z -* o0 {Vp -_ 0) and L/R is large, the critical pressure
is equal to the value obtained from that ring theory which is capable of properly de-
scribing the behavior where n = 2. The upper limit to the curve of Figure 4.4-2 is
associated with the shear crimping mode of failure which involves extremely short
4-35
circumferential wavelengths (n--oo). Specialization of Equations (4.4-i) through
(4.4-4) to this case gives the following formula for the critical compressive running
load Ncr measured in units of lbs/inch:
Ncr = h Gy z (4.4-5)
where
Ncr = Per It (4.4-6)
By using the approximation h _ tc , it can easily be shown that Equation (4.4-5) is
equivalent to the crimping formula presented earlier as Equation (2.3-9).
• " Crimping)
Cp
Lower Limit
(n = 2)
Note: Vp = Constant
0 a :- Constant
Figure 4.4-2. Schematic Representation of Log-Log Plot of Cp

Versus L/R for Circular Sandwich Cylinders
Subjected to External Lateral Pressure
Another important point which should be noted is that the approximate formula for Kp
[Equation (4.4-2)] does not contain the core shear modulus associated with the plane
perpendicular to the facings and oriented in the axial direction (Gxz). This modulus
has very little influence on cylinders longer than approximately one diameter and has
4-36
therefore disappeared through the approximations made in the development of Refer-
ence 4-21. Thus the theory and design curves presented in this section (Section 4.4)
of the handbook can be considered applicable to sandwich cylinders having cores with
either isotropic or orthotropic transverse shear moduli.
4-37
4.4.1.2 Empirical Knock-DownFactor
In Section4.1 it is pointedoutthat, for circular cylinderssubjectedto external
lateral pressure,the shapeol 0m post-bucklingequilibriumpathis suchthatone
wouldnot expectstrongsensitivity to the presenceof initial imlxerfections. This has
indeedbeenshownto be the casefor isotropic (non-sandwich)

cylinderswherethe
availabletestdatashowrather goodagreementwith the predictionsfrom classical
small-deflectiontheory. In viewof this, it hasbecomewidespreadpracticeto either
acceptuncorrectedsmall-deflectiontheoreticalresults asdesignvaluesor to applya
uniformknock-downfactor _p of 0.90 regardlessof the radius-to-thicknessratio.

In Reference4-4 the latter practiceis also recommended
for sandwichcylindersand
this approachhaslikewisebeenselectedas the criterion for this handbook.
Theonly availabletest datafor sandwichcylinderssubjectedto externallateral pres-
sltre are thosegivenin References4-27and 4-28. In the first of thesedocuments,
Kazimi reports the results from two specimenswhichwereidenticalexceptlor the use
of normal-expanded
core in onecylinder while the other incorporatedover-expanded
core. Thefollowingresults wereobtained:
Comparison of Theoretical Predictions Versus

Test Results of Kazimi _4-27_
@ @ @ @
Theoretical Pcr ('(P)Test
Core Type Test Pcr Based on Ref. (Test Pcr)

m
(psi) 4-5 and "/p : 1.0 (Theo. Pcr)

(psi) : ®-®
Normal-Expanded 17 30.5 .56
Over- Expanded ! 27 30.5 .88
4-38
Kazimi [4-27] attributes the scatter in his test results to the circumstance whereby
the over-expanded condition gives more uniform core properties than are obtained
from normal-expanded honeycomb. The argument put forth on behalf of this viewpoint
rests on the fact that the over-expanded core exhibits less anticlastic (saddle-type)
deformation in forming the core to the shape of the cylinder.
In Reference 4-28 Jenkinson and Kuenzi report the results obtained from five test
cylinders of nominally identical construction. These cylinders all had glass-reinforced
plastic facings. Each facing was composed of three layers of glass fabric with their
individual orientations controlled to provide a laminate having in-plane properties
which were essentially isotropic. The following results were obtained from these
cylinders :
Comparison of Theoretical Predictions Versus

Test Results of Reference 4-28
® ® ® ®
Theoretical Per (YP)Test
Cylinder Test Pcr Based on Ref. (Test Pcr)

No. (psi) 4-5 and 7p = 1.0 (Theo. Pcr)
(psi) ®+®
60 55.2 1.09
52.5 45.2 1.16
52.5 52.6 1.00
52.5 45.2 1.16
52.5 47.6 1.10
For specimens 2 through 5 it was reported that initial buckling occurred at external
lateral pressures which ranged from 50 to 55 psi. Therefore, in the above tabulation
4-39
it wasassumedthateachof thesefour cylindersbuckledat 52.5 psi. In general, the
test valuesare somewhathigherthanthe theoreticalpredictions. This is probably
due to
a. the absence of precise data on the material properties
b. inaccuracies due to interpolation between the theoretical curves

and
C. the fact that the facings were relatively thick in comparison with the
sandwich thickness (_ _ .25).
The foregoing test results from References 4-27 and 4-28 seem to provide added
justification for the use of '{1) = 0.90 as a lower-bound knock-down factor, ltowever,
it would certainly be desirable to supplement these data with additional tests on speci-
mens having small tf/h ratios which would be truly representative of configurations
usually found in realistic full-size sandwich cylinders.
An additional point of interest concerning the use of a unifornl value of Fp = 0.90 is
the fact that shear crimping failures will be insensitive to the presence of initial im-
perfections. Hence, in the region where this mode of failure prevails, one could
safely use the value ,(p = 1.0, especially since the theoretical basis used here neglects
the bending stiffnesses of the facings about their own middle surfaces. However, in-
spection of the design cuz_ves of Section 4.4.2 shows that this type of failure will only
occur for extremely low L/R values. This fact, coupled with considerations of
simplicity and the moderate nature of the value "_p - 0.90, led to the selection here
of a uniform knock-down factor.
4-40
In viewof the meagertest dataavailablefrom sandwichcylinderssubjectedto external
lateral pressure, the methodrecommended

herecanpresentlybe regardedas onlya
"best-available"approach. However,there appearsto be little reasonto doubtthat
f_'ther testingwouldshowtheseproceduresto bequitereliable.
4-41
For simply supportedcircular sandwichcylinderssubjectedto externallateral pres-
sure, the critical pressuremaybe computedfrom the equation
7p T, Cp
(4.4-7)
Pcr - R(l__e _) [(Eltz) _ (Eet2)]
where
"_p = 0.90
In order to use these curves,

and Cp is obtained from 1,'igures 4.4-3 through 4.4-5.
one must compute the following values:
(E _t_ )(Ea te) he

,2
(4.4-8)
uJ : [(Eltl) + (E2ta)] 2 112
(E 1 tl )(Ea ta ) h
(4.4-9)
Vp = TI [(Eltt) +(Eata)] (l-re _) He Gyz
For elastic cases, use _] = 1. Whenever the behavior is inelastic, the methods of
For cases where the two facings are not made of the same material, the foregoing
formuIas are valid only when the behavior is elastic (,Z - 1). Application to inelastic
cases (_ _ 1) can only be made when both facings are made of the same material.
For such configurations, E_ and Ee will, of course, be equal.
Since separate families of design curves (Cp vs L/R) are provided for only three
values of Vp, one will usually find it necessary to use graphical interpolation or
extrapolation to establish Cp for the configuration of interest. Where desired,
4-42
improved accuracy can be obtained by minimizing Equation (4.4-2) with respect to n
in order to obtain Cp.
The results given by the procedures specified here apply to sandwich cylinders having
cores with either isotropic or orthotropic transverse shear moduli.
4-43
0.01
Yp rtC
Per '2 [(I lt]) + (E2t,2)]
I_:] - v )
\5
9O
P
\ ",v
I 2 defined by Eq. (4.,t-8)
\.
•] Vp defh]ed by Eq. (,i.4-'J)

'
CIo00] ----*,-
_x
NN N < 2
= O. 0001
N-<
/oll,-
i _-ll oJo!ol
0.000] -
C
p
......
w..:
---:oTioii
\ I
O. 00001 ---
\
X_
0.000001--.-- ---
1 10 100
Figure 4.4-3. Buckling Coefficients Cp for Circular Sandwich Cylinders

tropic; Vp : 0
4-44
0.0 :,.-
...... _p ,7c
2
--
..... Per - R(1 -v eP ) [(Eltl) + (E2t2)]
_2 defined by Eq. (4.4-8)
V P defined by,Eq. (4.4-9)
0.00]
O. O00l
C
P
0.00001
0,000001
10
Figure 4.4-4.
Buckling Coefficients Cp for Circular Sandwich Cylinders
tropic; V = 0.05
P
4-45
0.01
¢' yz 1
0.001
0.0001
C
P
0.0000]
0.000001 -- 100
1 10
Figure 4.4-5. Buckling Coefficients Cp for Circular Sandwich Cylinders

tropic;V = 0. i0
P
4-46
4.5 TORSION
4.5.1 BasicPrinciples
4.5. I. 1 TheoreticalConsiderations
This sectiondealswith the loadingconditiondepictedin Figure 4.5-1. Notethat the
onlyconsiderationgivento boundaryconditionsis that, during buckling, it is assumed
that no radial displacementsoccurat either end. Further conditionsat theseboundar-
ies are completelydisregarded. This approachshouldbesufficiently accuratefor all
simply supportedcylinders exceptthosewhichare very short.
T, In.-Lbs. Torque
__ T, In.-Lbs. Torque
It is assumed that, during buckling,

no radial displacements occur at
either end.
Figure 4.5-1. Circular Sandwich Cylinder Subjected to Torsion
The buckling of isotropic (non-sandwich), circular cylinders subjected to torsion was
treated by Donnell in Reference 4-8 which has become a standard source of information
concerning reasonable approximations which can be employed in practical thin-shell
theory. Using the Donnell approximations, Gerard [4-29] has investigated the buck-
ling of long circular sandwich cylinders subjected to torsion. This solution gives no
consideration whatsoever to the boundary conditions. Such an approach is valid in
4-47
view of the qssumed extremely long configuration. On the other hand, in Reference
4-30, March and Kuenzi develop small-deflection solutions for sandwich cylinders of
both finite and infinite lengths. The boundary conditions taken lot the finite-length
cylinders are as indicated in Figure 4.5-1. For the purposes of this handbook, Refer-
ence 4-30 is considered to provide the most up-to-date treatment of the subject. The
theoretical design curves given in Section 4.5.2 were taken directly from that report
and embody the following assumptions:
a. The facings are isotropic.
b. The facings are of equal thickness. However, the curves are reasonably
accurate for sandwich cylinders having unequal facings, provided that the
thickness of one facing is not more than twice the other.
c. Young's modulus is the same for both facings.
d. Poissonts ratio is the same for both facings.
e. The core has infinite extensional stiffness in the direction normal to the
facings.
f. The core extensional and shearing stiffncsses are negligible in directions
g. The transverse shear properties of the core may be either isotropie or
o rthotropic.
h. The approximations of Donnell [4-81 can be applied.
The design curves include separate families which respectively neglect and include
bending of the facings about their own middle surfaces. However, for both of these
situations, it is assumed that the facings are thin.
4-48
The theoreticalbucklingrelationshipusedhereis
d
rcr = K s ]] ErR (4.5-1)
which is based on the further assumption that both facings are made of the same mate-
rial. The notation used here is as follows:
rcr = Critical value of facing shear stress, psi.
Ks : Torsional buckling coefficient, dimensionless.
Ef - Young's modulus of facings, psi.
d = Total thickness of sandwich wall.
d t c +tz + te (4.5-2)
t c = Thickness of core, inches.
t l and te : Thicknesses of the facings (There is no preference as to which facing

is denoted by the subscript 1 or 2.), inches.
R - Radius to middle surface of sandwich cylinder, inches.
The buckling coefficient Ks is arrived at by the minimization of a complicated ex-
pression given in Reference 4-30. This formulation is not reproduced here. However,
it should be noted that the indicated minimization leads to Ks values which can be
plotted in the general form shown in Figure 4.5-2 where
L_
Zs - dR (4.5-3)
16 t c tlt_ Ef
- (4.5-4)
Vs 15 (tl +t2) Rd Gxz
4-49
Gxz
(4.5-5)
Gyz
and
L - Over-all length of cylinder, inches.
Gxz = Core shear modulus associated with the plane perpendicular to the
facings and oriented m the axial direction, psi.
Gy z = Core shear modulus associated with the plane perpendicular to the
axis of revolution, psi.
n Number of circumferential full-waves in the buckle pattern,
d imens ionl es s.
t c/d - Constant I
_ Constant '/
K s
Upper Limit (Shear Crimping)
: Constant ]
Long Cylinder
(,1 : 2)
ZS
Figure 4.5-2. Typical Log-Log Plot of the Buckling Coefficient K_

for Circular Sandwich Cylinders Subjected to Torsion
The curves given in Section 4.5.2 are of this type. Note that the upper limit for the
buckling coefficient K s correspends to the shear crimping mode of failure which
involves extremely short circumferential wave-lengths (n -. _o). Specialization of
the buckling equations to this case leads to the following result when it is assumed
that tc/d _ I :
4-50
5 2
Tcr = rcrimp - (tz + t_) tc v/Gxz GY z (4.5-6)
where
h =
Distance between middle surfaces of facings, inches.
In connection with sandwich constructions having large values for the parameter Zs
(long cylinders), it is pointed out that the cylinder will buckle into an oval shape (n = 2)
for which the Donnell approximations [4-8] are no longer valid. To illustrate this
point, attention is drawn to the results obtained for isotropic (non-sandwich), circular
cylinders subjected to torsion. By using the Donnell approximations, Gerard [4-31]
obtains the following result for the critical shear stress of such cylinders:
rcr - (l_Ve_)a/_ E (4.5-7)
In Reference 4-32, Timoshenko presents the following result from a more rigorous
solution which does not invoke the Donnell approximations:
: E (.___.) 3/_
rcr 3 _]2 (1-Ve2)a/a (4.5-8)
The more exact result gives a critical stress which is only 87 percent of that given by
the Donnekl approach, This is similar to the situation encountered in the case of exter-
nal lateral pressure (see Section 4.4) where the difference is even more pronounced.
Since the torsional design curves of Section 4.5.2 incorporate the Donnell approxima-
tions, they must be used with caution in the case of long cylinders (n = 2).
4-51
4.5.1.2 Empirical Knock-Down Factor
In Section 4.1 it is pointed out that, for circtdar cylinders subjected to torsion, the
shape of tile post-buckling equilibrium path is such that one would not expect the sensi-
tivity to initial imperfections to be as strong as that encountered in tile case of axial
compression. On the other h:md, tile sensitivity iil torsion would be expected to be
somewhat more severe than is exhibited by circular cylinders under external lateral
pressure, hi the case of isotropic (non-sandwich), circtdar cylinders loaded in torsion
I{eference 4-8 indicates that, over an enormous range of sizes, proportions, and
materials, a lower-bound curve to the available test data can be obtained by taking
60 percent of the values obtained from classical small-deflection theory {_ - 0.60).
Average values of the test data can be approximated by using 80 percent of the classi-
cal theoretical predictions (3's = 0.80)
To date, no test data has been ptd)lished for sandwich cylinders which are of the types
consider_t in this hmld!yook and are subjected to torsion. Therefore, no empirical
basis exists lor the determination of reliable knock-down lactors in such cases.
Based on the moderate drop-off of the post-buclding equilibrium path, some sources
[4-51 recommend that no reduction be employed ( "_s :: 1.0). ttowever, Reference 4-4
takes a more cautious approach in recommending the use of _s .80 for the sand-
wich configuration. This selection was made largely on the basis of the isotropic
(non-sandwich) data. Although this value did not furnish a lower-bound to the isotropic
test points, it is reasonable to expect that the usually greater thicknesses of sandwich
cylinders should lead to more moderate reductions than apply to the isotropic (non-
4-52
sandwich) conligurations. In addition, it should be noted that cylinders under torsion
will continue to supt_)rt considerable torque well into the postbuckled region, ttence
the torsional buckling mechanism should not be nearly so catastropic as the general
instability of axially compressed cylinders. With these several factors in mind, the
value "is - O. 80 has been selected for use in this handbook. In view of the lack of
sandwich test data to substantiate this selection, the methods proposed here can only
be regarded as a "best-available" approach.
4-53
For simply supported circular sandwich cylinders subjected to torsion, the critical
shear stress may be computed from the equation
d
rcr = Xs Ks _ Ef _ (4.5-9)
where
"¢s = 0.80 (4.5-10)
d = tc + t_ + t_ (4.5-11)
and Ks is obtained from Figures 4.5-3 through 4.5-8. border to use these cu_,es,
one must first compute each of the following values:
L2
Zs - (4.5-12)
dR
V = 16 tc t_ t_ TiEr (4.5-13)
s 15 (tl + t_) R d Gxz
Gxz
@ - (4.5-14)
Gyz
It is required here that both facings be made of the same material.
For elastic cases, use r1 = 1. Whenever the behavior is inelastic, the methods of
The critical torque Tcr , measured in units of in.-lbs, can be computed from the
following for both elastic and inelastic cases:
Tcr = 2rrR 2 (tl + ts) rcr (4.5-15)
4 -54
Culwesfor Ks aregiven for values of @ -0.4; 1.0; and 2.5. Estimates of Ks
for other values of @ can be obtained by interpolation.
In addition, curves for Ks are given for values of t c/d = 1.0 and 0.7. The former
neglect the contribution from bending of the facings about their own middle surfaces.
The latter may be used to obtain numerical estimates of the conservatism introduced
by neglecting these stiffnesses.
As noted in Section 4.5.1.1, the design curves are somewhat inaccurate in the region
where Zs is large (long cylinders). Some caution should be exercised in the appli-
cation of the curves in this region.
Strictly speaking, Figures 4.5-3 through 4.5-8 apply only when the facings are equal.
However, the curves are reasonably accurate for sandwich cylinders having unequal
facings, provided that the thickness of one facing is not more than twice the other.
4-55
e_
r_
.!
m
El
4-56
0
¢)
II II
_J
¢)
0
C)
iZ/
4-57
0
b
or,,i
°,,-I
r..)
II
4-58
S
E)
4-59
.2
_D
"0
.2
e_
.2
I
t_
O(3
o
S
4-60
©
,.,,-.i
%
©
ca
od
I
4-61
4.6 TRANSVERSE SI[EAR
In Reference 4-33, Lundquist reports the results from a series of tests on isotropic
(non-sandwich), circular cylinders subjected to combined transverse shear and bend-
ing. The same type of data is published in Reference ,I-34 for elliptical cylinders.
Both sets of data were obtained from cantilevered cylinders of varied lengths. Extrapo-
lation of these results to the condition of zero bending stress permits a determination
of critical stresses for pure transversc shear loading. It has proven useful to com-
pare these stress values against the theoretical results obtained from small-deflection
theory for isotropic (non-sand_vich), circular cylinder,_ loaded in torsion. Gerard and
Becker [4-35j report that, for nominally identical specimens, such comparisons yield
the following ratios where the theoretical predictions are obtained by using Reference
4-36:
Average of "rer
Shear Test Values f¢)r t
Transverse Loading
Small-Deflection Theoretical Ter
[ Values for Torsional Loading i a, 1.(; (4.6-1)
Lower-BoundTcr Shear Test Values for]

Transverse Loading
_ 1.25 (4.6-2)
Small-Deflection Theoretical Vc r ]
Values for Torsional Loading j
To properly interpret these ratios, it is pointed out that, for torsional loading, the
shear stress Tcr is uniformly distributed around the circumference. On the other
hand, under transverse shear loading, the shear stress is non-uniform and the value
Tcr then corresponds to the peak intensity which occurs at the neutral axis.
4-62
For the lack of a better approach,it is recommonded
that Equation(4._;-2)beused[or
the designandanalysisof circular sandwichcylindersthat are subjectedto transverse
shearforces. In suchcases,the requiredsmall-deflectiontheoreticalrcr valuesfor
torsional loadingshouldbeobtainedas specifiedin Section4.5 of this handbook

with
the exceptionthat 7s = 1.0 shouldbeusedhere. Notest dataare availableto sub-
stantiatethe reliability of this practice. Until suchdatadobecomeavailable, onecan
onlyregardthis procedureas a "best-available"approach.
4-63
For simply supportedcircular sandwichcylinderssubjectedto a transverseshear
force andhavingboth facingsmadeofthe samematerial, the critical shearstress
maybe computedfrom the equation

d
rcr = 1.25KsrlEf -_ (4.6-3)
where thebucklingcoefficientKs is obtainedfrom Figures4.5-3 through4.5-8 and
the notationis the sameas thatemployedthroughoutSeetion4.5. As notedin Section
4.5.1.1, thesefigures are somewhatinaccuratein the regionwhere Zs is large (long
cylinders)andoneshouldexercisesomecautionwhendealingwith suchconfigurations.
Strictly speaking,Figures4.5-3 through4.5-8 applyonlywhenthe facingsare of equal
thickness. Itowever,the curvesare reasonablyaccuratefor san_viehcylinders having
unequalfacings,providedthatthe thicknessof onefacingis not more thantwice the
other.
For elastic eases,use r? = 1. Whenever the behavi'n" is inelastic, the methods of
For elastic eylinders the critical transverse shear force (Fv)cr, measured in units of
pounds, can be computed from the following:
(Fv)er = rrR (t 1 + ta) rer (4.6-4)
To compute (Fv)cr when the behavior is inelastic, one must resort to numerical inte-
gration techniques.
4-64
4.7 COMBINED
LOADINGCONDITIONS
4.7.1 General
For structural memberssubjectedto combinedloads, it is customary to represent
critical loading conditions by means of so-called interaction curves. Figure 4.7-1
shows the graphic format usually used for this purpose. The quantity R. is the ratio
1
of an applied load or stress to the critical value for that type of loading when acting
alone. The quantity R. is similarly defined for a second type of loading. Curves of
J
this form give a very clear picture as to the structural integrity of particular con-
figurations. All computed points which fall within the area bounded by the interaction
curve and the coordinate axes correspond to stable structures. All points lying on or
outside of the interaction curve indicate that buckling will occur. Furthermore, as
shown in Figure 4.7-1, a measure of the margin of safety is given by the ratio of
distances from the actual loading point to the curve and to the origin. For example,
assume that a particular structure is subjected to the combined loading condition
corresponding to point B of Figure 4.7-1.
Then, for proportional increases in R. and R., the margin of safety (M.S.) can be
1 j
computed from the following:
(Rj)D
M.S. - 1 (4.7-I)
(Rj)B
As an alternative procedure, one might choose to compute a minimum margin of safety
which is based on the assumption that loading beyond point B follows the path BM.
Point M is located in such a position that BM is the shortest line that can be drawn
4-65
between point B and the interaction curve. The minimum margin of safety can then be
calculated as follows:
OB _ BM
Minimum M.S. OB 1 (4.7-2)
1.0
(Rj) D _M
(Rj)B
R
J
iiiiii
0 R 1.0
i
Figure 4.7-1. Smnple Interaction Curve
4-66
4.7.2 Axial Compression Plus Bending
i.7.2.1 Basic Principles
In References 4-17, 4-18, and 4-19 it has been shown that, for circular sandwich
cylinders subjected to axial compression plus bending, the classical theoretical
interaction curve may be accurately described by the equation
(4.7-3)
(Re)c L + (Rb) C L = 1
where
U
e
(4.7-4)
(Rc)cL -
(5c)CZ
_b
(.i. 7-_)
{Rb)cL- (Sb)CL
= (4. v-6)
((_b) CL ((_e) C L
and
(Y = Uniform compressive stress due solely to applied axial load,

e
psi.
= Peak compressive stress due solely to applied bending mo-

_b
ment, psi.
= Classical theoretical value for critical uniform compressive

(_c)CL
stress under an axial load acting alone, psi.
= Classical theoretical value for critical peak compressive

((_b) C L
stress under a bending moment acting alone, psi.
References 4-17 and 4-18 develop the foregoing result for weak-core constructions
which fail in the shear crimping mode. On the other hand, Reference 4-19 deals
with infinitely long cylinders which fall in the stiff-core and the moderately-stiff-
core categories. Since Equation (4.7-3) is written in terms of classical theoretical
4-f;7
allowables,it doesnot includeanyconsiderationof the detrimentalinfluencesfrom
initial imperfections. For the pullmsesof this handbook,theseinfluencesare treated
by'introducingtheknock-dt_vnfactors ye and Tb (see Figures 4.2-8 and 4.3-2, re-
spectively) to obtain
Rc +Rb = 1 (4.7-7)
where
e
R - (4.7-8)
c 'Yc ((_c) C L
(_b
% Yb ((_c) CL (4.7-9)
Therefore, the design interaction curve can be drawn as shown in Figure 4.7-2. Since
no test data is available for sandwich cylinders subjected to combined axial load and
bending, the general validity of this curve has not been experimentally verified. Some
degree of empirical correlation is inherent in the approach since the knock-down fae-
tors 3/c and 7b were established, in part, from sandwich test data (see Sections 4.2
and 4.3). ttc_vever, even these data were few in number. Therefore, until further
experimental substantiation is obtained, the recommended interaction relationship
can only be considered a "best-available" method.
lo0 -
Rc 1.0
Figure 4.7-2. Design Interaction Curve for Circular Sandwich Cylinders

Subjected to Axial Compression Plus Bending
4-68
For simply suppol_ed, circular, sand_vich cylinders subjected to axial compression
plus bending, the following interaction equation may be employed:
Re + Rb = 1 (4.7-10)
w h e re
(y
c
R - (4.7-11)
c Tc ((7c) C L
au - _b (_c)CL
(4.7-12)
A plot of Equation (4.7-10) is given in Figure 4.7-3.
In Equations (4.7-11) and (4.7-12), the knock-down factors Tc and _b are those ob-
tained from Figures 4.2-8 and 4.3-2, respectively.
The quantity (ffc)CL is simply the result obtained by using Tc = 1.0 in the method of
Section 4.2.2.
Plasticity considerations should be handled as specified in Section 9.2 except that,
in this case, one may use
ao
IlVel Et
r7 = [ 1---2-_1 _ff for short cylinders, and
t s
b. _ --
for moderate-length through long cylinders.
tl-u_] Ef
Equation (4.7-10) may be applied to sandwich cylinders of any length. However, length
considerations should be included in the computation of ((_c)CL when the structure falls
into the short-cylinder range (see Section 4.2.2).
4-69
0.8
!i!ii
0.6 _' j
i!il
Rb
0 0. '2 0.4 0.6 0.8 1.0

R
C
Figure 4.7-3. I)esigm Interaction Curve for Circular Sandwich Cylinders

Subjected to Axial Compression Plus tk, nding
4-70
4.7.3 Axial Compression Plus External Lateral Pressure
4.7. :5.1 Basic Principles
This section deals with the loading condition depicted in Figure 4.7-4. The sandwich
cylinder is subjected to uniform external pressure over the cylindrical surface. Axial
loading is imposed as indicated by the [orces P. These forces can originate from any
source including external pressures which are uniformly distributed over the end elos-
ures. In addition, it is specified that the ends of the eylinder are simply supported.
This is, during buckling, the ends are constrained such that they experience no radial
or circumferential displacements and they are free of bending moments.

p , psi
Y
p , psi
Y I
lbs
P, lbs__
/ lllttIlIIlt'(
Both Ends Simply Supported
Figure 4.7--4. Circular Sandwich Cylinder Subjected to Axial

Compression Plus External Lateral Pressure
The theoretical basis used here is the classical small--deflection solution of Maki
[4-37]. The design curves given in this handbook were taken directly from that source
and embody the following assumptions:
ao The facings are isotropic.
b. Both facings are of the same thickness.
e. Both facings have identical material properties.
4-71
d.
Poisson's ratio for the facings is equal to 0.33.
e° Bending of the facings about their own middle surfaces can be neglected.
f. The core has infinite extensional stiffness in the direction normal to the
facings.
g* The core extensional and shear rigidities arc negligible in directions

ho The transverse shear moduli of the core are the same in the circum-
ferential and longitudinal directions (G = G ).
xz yz
io
The mean radius of the cylinder is large in comparison with the sandwich
thickness.
The theoretical relationship derived by Maki [4-37_ is in the form of a complicated
sixth order determinant and no significant advantage would be gained by reproducing
that formulation in this handbook. However, it is important to note that a sufficient
number of terms were retained throughout the derivation to obtain accurate results
when the number of circumferential full-waves equals two (n = 2). If the derivation
had been based on the well-known Donnell approximations [4-81, the results would
not be applicable to structures which buckle in this manner.
The interaction curves given in Reference 4-37 are of the two different types depicted
in Figure 4.7-5 where
Ef tf h
V - (4.7-13)
xz 2 (1-.33e) RaG
XZ
Eftfh
Vyz = 2 (1- (4.7-14)
.33_) RaG
yz
P
_ Y
(4.7-15)
(Rp) CL _y) CL
4-72
cr
x
(Rc)CL (4.7-16)
(C_x)C L
and
Ef : Young's modulus of facings, psi.
tf = Thickness of single facing, inches.
h = I)istancc between middle surfaces of facings, inches.
R = Radius to middle surface of cylindrical sandwich, inches.
G = Core shear modulus associated with the plane perpendicular

XZ
to the facings and oriented in the axial direction, psi.
G = Core shear modulus associated with the plane perpendicular

yz
to the axis of revolution, psi.
p = Applied external lateral pressure, psi.

Y
(1-iy)CL = Classical theoretical value for critical external lateral pressure

when acting alone, psi.
g = Uniform axial compressive stress due to applied axial load, psi.

X
(C_x)C L = Classical theoretical value for critical tmiform axial compres-

sive stress when acting alone, psi.
L = Over-all length of cylinder, inches.
Note : The value .33 appearing in Equations (4.7-13) and (4.7-14) is an

assumed representative value for the elastic Poissonts ratio of
the facings.
Since the curves of Reference 4-37 were developed from a classical, small-deflection
approach, they do not include any consideration of the detrimental effects from initial
imperfections. This is evident from the fact that classical theoretical allowables are
used in the ratios (Rp)cL and (Rc)CL. For the purposes of this handbook, the effects
4-73
1.0 1.0 : 4> -._I
% ..
o
(Rc)CL
(Rc)CL
Jill/ = Constant \
\XZ - \VZ - O/ _
[-_ : Constant] _ .
1.0 1.0
(tlp)CL (Rp)CL
Figure 4.7-5. Typical Interaction Curves for Circular Sanchvieh Cylinders Subjected
to Axial Compression Plus External Lateral Pressure
from initial imperfections are introduced through the replacement of (%)CL and
and R which are defined as follows:

(Re)CL by the ratios Rp c
Py
R = - (4.7-17)
P Tp (Py) CL
(y
R = x (4.7-18)
e Yc ((_x) C L
The quantities yp and Yc are the knock-down factors discussed in Sections 4.4 and 4.2,
respectively. Values for yc can be obtained from Figure 4.2-8 while yp may be taken
equal to 0.90.
No test data are available for sandwich cylinders which are of the types considered
here and are subjected to axial compression plus external lateral .ressure. Therefore,
the general validity of the design curves recommended here has not been experimentally
verified. Some degree o[ empirical correlation is inherent in the approach since the
4-74
knock-down factors Te and Tp were established, in part, from sandwich test data (see
Sections 4.2 and 4.4). ttowever, even these data were few in nmnber. Therefore,
until further eN)erimental substantiation is obtained, the recommended interaction
curves can only be considered as "best-available" criteria.
4-75
For simply supported,circular, sandwichcylinderssubjectedto axial compression
plus externallateral pressure, onemayemploythe interactioncurvesof Figures
4.7-6 through4.7-15where
Eftf h
(4.7-19)
xz 2(1-. 33z')R_ Gxz
Eftf h
(4.7-20)
Vyz 2(1-.332)R _G
yz
P
Y (4.7-21)
R =
P Yp (Py)CL
R = x (4.7-22)
c Tc (fix) C L
In Equations (4.7-21) and (4.7-22), the knock-down factor Tc is that obtained from
Figure 4.2-8 while Tp may be taken equal to 0.90.
The quantity (15y)CL is simply the result obtained by using 7p = 1.0 in the methods of
Section 4.4.
The quantity ((_x)CL is simply the result obtained by using Tc = 1.0 in the methods of
Section 4.2.
Plasticity considerations should be handled as specified in Section 9.2.
Figures 4.7-6 through 4.7-12 give interaction curves only for cases where Vxz =
V = 0 (G = G - co). Separate families are provided for each of three selected

yz xz yz
values for the parameter -_ = 50; 160 and 500 . Graphical interpolation may be
4-76
used to obtain results for intermediate values of this parameter. Each family includes
separate curves for ten different values of the ratio --L-- -- = 0.1; 0.'2; --- 1. . In
view of the restrietions on V and V , these curves can only be used to describe the
xz yz
behavior of stiff-core eonstruetions. For the purposes of praetieal design and analysis,
it is proposed here that Figures 4.7-6 through 4.7-12 be considered applicable only
when
Rt
C
-- V ± 0.05 (4.7-23)
h_ xz
Rt
C
V _ 0.05 (4.7-24)
h* yz
\vhcre

C
It is ex_pected that many realistic sandwich configurations will satisfy these requirements.
Fi_ires 4.7-13 through 4.7-15 present a partial picture of the effects which variations
in V (= V ) will have on the interaction relationships. These figures only treat

XZ yZ
cases for which --rrR _= 0.1. }tc_vever, the trends displayed furnish some basis for one
I,
to conjecture that the curves given for V = V = 0 would result in conservative pre-
xz yz
dictions if they were applied to sandwich configurations which do not satisfy the In-
equalities (4.7-23) and (4.7-24). However, one should be cautioned againstmaking
sweeping application of this observation in view of the limited scope of the information
shown in Figures 4.7-13 through 4.7-15.
4-77
It shouldbe kept in mindthatthe interactioncurvesgivenin Figures4.7-6 through
4.7-12 include C L values ranging only from 0.1 through 1.0. Since
_R
C - (4.7-25)
L L
it follows that these curves only embrace the range where
L
3.14 _ -_- <- 31.4 (4.7-2(;)
4-78
h= 50
O.
8 _. Vxz = Vyz : 0,0
__C rrI1
0.6
O. ,1
0,2
0 0.2 0.4 0.6 0°8 1,0

R
P
Figure 4.7-6. Interaction Curves for Circular Sandwich Cylinders

Subjected to Axial Compression Plus External
Lateral Pressure
4-79
1
I1
50
11
V = V = 0.0
xz yz
'gR
C
I_ L
0.6
R
C
0.4
CL 0.7"
CL 0.8-
C O. 9_
0.2
0
0 0.2 0.4 0.6 0.8 1.0
R
P

I,ateral Pressure
4-80
1.0
R
N = 50
h
V = V = 0.0
0.8 xz yz
_R
C =
L L
0,6
'0
R
c
\
0.4
0.2
0 0.2 0.4 0.6 0.8 1.0

R
P
Figure 4.7-8. Interaction Curve for Circular Sandwich Cylinders

Lateral Pressure
4-81
I
H
-- 160
h
V :-: V :: 0.0
- xz yz --
=0.]
0 0.2 0.4 0.6 0.8 1.0

R
P

Lateral Pressure
4-82
I
R
-- 160
h
0.8 = V 0.0
C LC_ 0_ 8(_ KC L = 0.6 Vxz ?rRYZ

) CL _[[
0.6 _ I
0.4 C
L
0.2
0
0 0.2 0.4 0.6 0.8 1.0
R
P
Figure 4.7-10. Interaction Curves for Circular Sandwich

Plus External Lateral Pressure
4-83
1.0 I
R
CLI0. 4 5O0
h
_ jt/ICL = 0. V = V = 0.0
0.8 xz yz -
_R
C
C L = 0.5'__ L L
0.6
R
c
0.4
C L = 0.3
C =0.1
L
0.2
0
0 0.2 0.4 0.6 0.8 1.0
R
P

Lateral Pressure
4-84
= 0o0
R
C
0.4
=0.8
CL=0.
0.2
0
0 0.2 0.4 0.6 0.8 1.0
R
P

Lateral Pressure
4-85
l°01
R
C
R
P

Lateral Pressure
4-86
R
C
0.2
0
o 0.2 0.4 0.6 0.8 1.0
R
P

Lateral Pressure
4-87
1.0
0.8
t 5
0.6 % kk --
R %
C
0.4 \
h = 500
0.2 --
CL = _R
I--_- =0.1
0 0.2 0.4 0,6 0.8 1.0

R
P

Lateral Pressure
4-88
4.7.4 Axial CompressionPlus Torsion
4.7.4.1 BasicPrinciples
This sectiondealswith the loadingconditiondepictedin Figure4.7-1(;. Thesanchvich
cylinder is subjectedto endtorqueT plus axial loadingindicatedby the forces P.
T, in-lbs Torque T, in-lbs Torque
BothEnds
SimplySupported
Figure 4.7-16. Circular Sandwich Cylinder Subjected to

Axial Compression Plus Torsion
In Reference 4-18 Wang, et al. treat this type of problem but only consider the case
of weak-core configurations which fail in the shear crimping mode. In addition they
assume that the cylinder is long so that the boundary conditions can be ignored. This
small-deflection analysis makes use of the Donnell approximations [4-8] to arrive at
the following interaction relationship:
(4.7-2 7)
(Rc)CL + (Rs) CL
2 = 1
where
C
(4.7-28)
(Rc)CL - _
(_c)CL
4-89
I- (4.7-29)
(Rs)CL- (_)CL
and
(I = Uniform axial compressive stress due to applied axial load, psi.

c
= Classical theoretical value for crilical uniform axial eompres-

((_e)CL
sire stress when acting alone, psi.
_- = Uniform shear stress due to applied torque, psi.
(T)CL = Classical theoretical value for critical uniform shear stress

duc to torque acting alone, psi.
Because Equation (4.7-27) was developed from a classical, small-deflection approach,
it does not include any consideration of the detrimental effects from initial imperfee-
tions. That is evident from the fact that classical theoretical allc_vables are used in
the ratios (Rc)cL and (Rs)CL. For the purposes of this handbook, the effects from
initial imperfections are introduced through the replacement of (Rc)cL and (Rs)CL by
the ratios R and 1R which are defined as follows:

c s
(y
c
R -
(4.7-30)
c "/c(ffc)C L
R _ 1- (4.7-31)
s
7s (_)CL
The quantities _c and _s are the knock-down factors discussed in Sections 4.2 and 4.5,
respectively. Values for Yc can be obtained from Figure 4.2-8 while :Ys may be taken
equal to 0.80. Incorporation of the foregoing substitutions into Equation (4.7-27) then
gives the following interaction relationship for weak-core constructions:
R + R 2 = 1 (4.7-32)
c s
4-90
In Reference 4-38, Batdorf, et el. deal with the subject loading condition for thin-
walled, isotropic (non-smldwich), circular cylinders. Since, for such constructions,
transverse shear deformations of the shell wall are of negligible importance, one
might conjecture thai this work could be applied to sandwich cylinders which fall into
the stiff-core category. Based on theoretical considerations modified by test results,
Batdorf, et el. c4-38n: arrived at the same interaction expression as that given above
as Equation (4.7-32). In view of this, one might choose to view Equation (4.7-32) as
a comprehensive interaction formula for sandwich cylinders, ttowever, some caution
should be observed in implementing this viewpoint, partially because of the fact that
only the extremes of transverse shear stiffness of the core have been considered. In
addition, although the interaction relationship for the subject loading condition should
probably be dependent upon a length parameter, no investigations were made to estab-
lish the sandwich cylinder lengths over which Equation (4.7-32) is a reasonable repre-
sentation of the actual behavior. Furthermore, no test data are available for sandwich
cylinders which are of the types considered in this handbook and are subjected to axial
compression plus torsion. Therefore the general validity of Equation (4.7-32) has not
been experimentally verified. Some degree of empirical correlation is inherent in the
approach since the knock-down factor _'e was established, in part, from sandwich test
data (see Section 4.2). ttowever, even these data were few in number. Therefore,
until further theoretical and experimental investigations are accomplished, the inter-
action relationship cited here can only be considered as a "best-available" criterion.
4-91
For simply supported, circular, sandwich cylinders subjected to axial compression
plus torsion, one might choose to employ the interaction formula
R +R _ = 1 (4.7-33)
c s
which is plotted in Figure 4.7-17 and where
a
R - c (4.7-34)
c _'c (_c) CL
R - _" (4.7-35)
s
Ys (_) C L
In Equations (4.7-34) and (4.7-35), the knock-down factor _c is that obtained from
Figure 4.2-8 while _s may be taken equal to 0.80.
The quantity ((_c)CL is simply the result obtained by using )'c = 1.0 in the methods of
Section 4.2.
The quantity (_) is simply the result obtained by using _s = 1.0 in the methods of
CL
Section 4.5.
Plasticity considerations should be handled as specified in Section 9.2.
Attention is drawn to the fact that, in Section 4.7.4.1, several factors are cited which
shed considerable doubt upon the reliability of results obtained from the indiscriminate
use of Equation (4.7-33) and Figure 4.7-17. In view of these uncertainties, one might
often choose to employ the straight-line interaction formula
R + R = 1 (4.7-36)
c s
4-92
which is plotted in Figure 4.7-18. This relationship can be used with confidence for
any length of cylinder and for any region of transverse shear rigidity of the core since
experience has shown that the linear interaction formula is never unconservative for
shell stability problems, ttowever, in many cases it will, of course, introduce execs-
sive conservatism.
(}. (!
CO 0.2 0.4 0. (; 0. _ I . 0
Figure 4.7-17. Conditional Interaction Curve for Circular Sandwich Cylinders

Subjected to Axial Compression Plus Torsion
4-93
0.6
R
s
0 0.2 0.4 0.6 0.8 1.0

R
c
Figure 4.7-18. Conservative Interaction Curve for Circular Sandwich Cylinders

4-94
4.7.5 Other Loading Combinations
In Sections 4.7.3 and 4.7.4 the following combined loading conditions are treated:
a. Axial Compression plus External Lateral Pressure.
b. Axial Compression plus Torsion.
The corresponding interaction relationships can be used for certain additional com-
binations by recognizing that
a. the peak axial stress due to an applied bending moment can be converted
inio an equivalent uniform a.xial stress, and
b. the peak shear stress due to a transverse shear force can be converted
into an equivalent tmiform torsional shear stress.
With this in mind, the design equations and curves of Section 4.7.3.2 can be used for
the combination o[ AXIAL COMPRESSION PLUS BENDLNG PLUS EXTERNAL LATERAL
PRESSURE if one simply substitutes the quantity g_ for (_ whe re

X X
, = ((:rx)c +()'c_ ((rx)b (4.7-37)

% \rb/
and
= Uniform axial compressive stress due solely to applied axial

((_X) C
load, psi.
= Peak axial compressive stress due solely to applied bending

(Crx) b
moment, psi.
= Knock-down factor associated with axial compression and as

%
given in Figure 4.2-8, dimensionless.
= Knock-down [actor associated with pure bending and as given in

_zb
Figure 4. ;/-2, dimensionless.
4-95
This formula is based on the findings reported in Section 4.3.
In addition, the design equations and curves of Section 4.7.4.2 can be used for the
combination of AXIAL COMPRESSION PLUS BENDING PLUS TORSION PLUS TRANS-
VERSE SHEAR FORCE if one simply substitutes the quantities _' and T _ for (_ and 1",
C C
respectively, where
(4.7-38)
(r'e = (_c)c + (_c)b
, O. 80
T = T T + 1.2----55rV = 7"T + 0"641"V (4.7-39)
and
(gc) c = Uniform axial compressive stress due solely to applied axial

load, psi.
(gc) b = Peak axial compressive stress due solely to applied bending

moment, psi.
_'T = Uniform shear stress due solely to applied torque, psi.
TV = Peak shear stress due solely to applied transverse shear force,

psi.
7c and :_b = Knock-down factors specified above.
Equation (4.7-38) is based on the findings reported in Section 4.3 while Equation
(4.7-39) stems from a comparison of Equations (4.5-9) and (4.6-3).
Since no sandwich test data are available to substantiate the foregoing procedures, they
can only be regarded as "best-available" criteria.
4-9G
4.7.5.2 DesignEquationsandCmwes
For the combinationof AXIAL COMPRESSION

PLUSBENDINGPLUSEXTERNAL
LATERALPRESSURE,
substitute(_X for(_X andusethe designequationsandcurves
follows:
given in Section 4.7.3.2. The quantity (i X is defined as
(4.7-4O)
0"X = (ax) c + ((Ix) b
used in Section 4.7.3.2 remains as defined in thai

Itowever, the quantity (CTx)CL
sectior_.
For the combination of AXIAL COMPRESSION PLUS BENDING PLUS TORSION PLUS
TRANSVERSE SIIEAR FORCE, substitute cr I for a and T _ for "r in the design equations
e c
and curves given in Section 4.7.4.2. The quantities a t and _-_ are defined as follows:
e
e = (<:'c)e\ 'bl
u (4.7-41)
0.80
7" = T T + 1.2-----_o
"rV = TT + 0.64"r V (4.7-42)
However, the quantities ((Tc)CL and (_)CL used in Section 4.7.4.2 remain as defined
in that section.
The foregoing criteria will still apply, of course, where one or more of the specified
applied loads equal zero.
4-97
^ o
I . t
ao
i
.# ul
1
,--t
_t_ _ _ O H
_ m
"tJ
o
.o
L_
o b
"8
e_
•_ ,, ,, o_ -_ _ o
__ -_ .
N
"o Ii N
% L) ° el# °
r_
r_
,2 d d
c,1 _,1 > 2®_ o
..... _ "_'_K_ V AI II e_ o o.
_2
d ° d'
II > > & 0
% b b > o _1 o
II % "_
_._o _, _ o
e_
._
o
N O [._ II
_ "_,_ _ _o o
Q q_ _ o
(_ "_ _ 0
NI.._
h vl
£zJi
o
,-1
o _.-i I
o_ °"K v _ _ _.
z
4-98
©
<
o
_q
t_
O)
@ "a o
8
o .,2.If
o
,x2,
e_ I
d'
rll o © c,,! ,.g

v_
r-,
If _,
.P
o
o
_,o II I1
H
o_.oK
dz o _D
H
V H _
"_o_,_.
H
°_
K
×V uoTlnloAa}:I jo sl.xv
©
/ _ r_ ° ,_
-a
v_ _ >_]
o _4
,-4
4-99
r_
d _
I
d N5
I
4
d 4 N
0
-ip _ '_: _,
[._ ml
o I
o 1-__. ¢1
Q
c,_ °
_D 4 rl
_
°
b
O
0
©
_D
:- r_ _= o ,'f
d 0
o
r_ b
o o
O _ _)
d'
o; 0; _- u d_
© IP PI II
V
II
II
c_
C_ II vo ,\
0
% e_
Ii
bl
r/l
©
o
_d
t_
I
_2
0 AI _I
,.Q
V _ _
.1:2_ AI _
II
4-100
0
o
a _
o _ g
o E
.a o
.0 ._
o
0
2 _
fll
,, _ _ o -_
o
_ b o "_
II
. _ _ V AI fl II _ _
_ ° ,__1-.__¢_ _
0
./.E I 1 H----_ ..... _
_.5
Ca m
4-101
o _
e_
0
o
i
u,l '_
0
o _
_-_
o
d_
_._ . ._
o 0
_._•
T_ _
0 2 _
o
[._
_ x
2_ _
m
_ +
II
_.__ ,_
+
5_
v I _e°i
i
.o o• _0
Cl "5
% II ,_
0 _ H _
rr
g)
0
>,
e_
o
! o
,.0
o
©
ff_2___ "- o
>,
ca_
ca _ z
4-102
©
L_
_,_ b_
0 o
"_ ,
r_
>
°_ "_ c_ CJ
H o
%
C)
_)
_._ _
> N o © .o o
>_ b,8
_,_
_' 0
_.,_
* o ._
_
V1 m
.j, ..1:2,
io
I
i_.
> >
o ©
07
-¢o to
II ._ _ o
% _ •
h,1
>
__) i_ c_
rn,
e_
0 >'
ca_
< m
ca_
_-103
• _ _ _ _
.,-4
II
r0 _
o_
÷ H
c_
3
o
_o _
_m ._ _ _ _ , ,
li
tn i._ o
o_ (d
°v-q 0
o "_ "o o _
_0
.r-#
?
c_ c-
r./2 .o
t3
O ×
I - u°!ln[°AaH .P) ,qPcV --
•,,,r,i
L_ rn
4-104
©
,,2
o
_ o_
I_ _', _u
©
o ll
._ .-=
_>
I
©
I-
(- _-i
-g
_4
©
_ s
>, o
>
,.4
_f i-
t-
V_
g
©
m
g _;< -e _e
a2 _
_ m ,/')
,._ o _
_ _ ,
_>' i_ _'
c_ ,_
el o
rF [_
2)
•_ _
2_ o
M
N r_ m _
: _ = _ UoTlnlOAO}{ JO s!x_ "o

o
t_q
(J I ;_1
t-
4-105
©
0
"0
(0
[/3
P_
0S
4-106
r_
"b
_ o o 0.;
I, [
b_
,,_, ,c ._
d
2
m
o o
o
_' o _ _ ,_ .2.
r_
r_ 0
,,-1 r_
¢, o
_ o_ 0
Ct _ © ,
-a <
'l:J
_ _ _''_< •
r_ o
._ _ _ _ o _ _4 ._ o
0 '_ II _'
m £ • m
_14
i>-, ..o _Q
_m
T ¸ __ o_1
_0 M
_ ca_
4-107
RE FERENCES
4-1 Donnell, L. H. and Wan, C. C., "Effect of Imperfections on Buckling of
Thin Cylinders and Columns Under Axial Compression, " Journal of Applied
Mechanics, March 1950.
4-2 March, H. W. and Kuenzi, E. W., "Buckling of Cylinders of Sandwich
Construction in Axial Compression, " FPL Report No. 1830, Revised
December 1957.
4-3 Hoff, N. J., Madsen, W. A., and Maycrs, J., "The Postbuckling Equilibritm_
of Axially Compressed Circular Cylindrical Shells, " Stanford University
Department of Aeronautics and Astronautics, Report SUDAER No. 221,
February 1965.
4-4
Anonymous, "Buckling of Thin-Walled Circular Cylinders, " NASA Space
Vehicle Design Criteria, NASA SP-8007, September 1965.
4-5 U. S. Department of Defense, Structural Sandwich Composites, MIL-HDBK-23,
30 December 1968.
4-6 Thielemann, W. F., "New Developments in the Nonlinear Theories of the
Buckling of Thin Cylindrical Shells, " Proceedings of the Durand Centennial
Conference, Held at Stanford University, 5-8 August 1969, Pergamon Press,
New York, Copyright 1960.
4-7 Zahn, J. J. and Kuenzi, E. W., "Classical Buckling of Cylinders of Sandwich
Construction in Axial Compression --- Orthotropic Cores, " U. S. Forest
Service Research Note, FPL-018, November 1963.
4-8 Donnell, L. H., "Stability of Thin-Walled Tubes Under Torsion, " NACA
Technical Report No. 479, 1934.
4-109
4-9 Scide, P., Weingarten, V. I., and Morgan, E. J., "Final Report on the
Development of Design Criteria for Elastic Stability of Thin Shell Structures, "
STL-TR-60-0000-19425, 31 December 1960.
4-10 Baker, E. H., "Ex]aerimenta] Investigation of Sanchvich Cylinders and Cones
Subjected to Axial Compression, " AIAA Journal, Volume 6, No. 9, September
1968.
4-11 Anonymous, "Status Summary Report for R&D Project 6002, " Hexce] Products,
Inc., Advanced Stm_clures Group, 25 July 1963.
4-12 Plantema, F. J., Sandwich Construction, John Wiley & Sons, Inc., New York,
Copyright 1906.
4-13 Stein, M. and Mayers, J., "Compressive Buckling of Simply Supported Curved
Plates and Cylinders of Sandwich Construction," NACA Technical Note 2601,
January 1952.
4-14 Bijlaard, P. P. and Gallagher, R. H., "Elastic Instability of a Cylindrical
Shell Under Arbitrary Circumferential Variation of Axial Stress, " Journal
of the Aerospace Sciences, Vol. 27, 1960.
4-15 Abir, l). and Nard(_, S. V., "Thermal Buckling of Circular Cylindrical Shells
Under Circumferential Temperature Gradients, " JcJurnal of the Aerospace
Sciences, Vol. 26, 1959.
4-16 Seide, P. and Weingarten, V. I., "On the Buckling of Circular Cylindrical
Shells Under Pure Bending, " Transactions of the A.S.M.E., Journal of
Applied Mechanics, March 1961.
4-17 Wang, C. T. and Sullivan, D. P., "Buckling of San_vich Cylinders Under
Bending and Combined Bending and Axial Compression, " Journal of the
Aeronautical Sciences, Vol. 19, July 1952.
4-110
4-18 Wang, C. T., Vaccaro, R. J., and DeSanto, I). F., "Buckling of Sandwich
Cylinders Under Combined Compression, Torsion, and Bending Loads, "
Journal of Applied Mechanics, Vol. 22, September 1955.
4-19 C<.llatly, R. A. and Gallagher, R. It., "Sandwich Cylinder Instability Under
Nonuniform Axial Stress, " AIAA Journal, February 1964.
4-20 Peterson, J. P. and Andersen, J. K., "Structural Behavior and Buckling
Strength of Itoneyeomb Sandwich Cylinders Subjected to Bending, " NASA
TN D-2926, August 1965.
4-21 Kucnzi, E. W., Bohannan, B., and Stevens, G. H., "Buckling Coefficients
for Sandwich Cylinders of Finite Length Under Uniform External Lateral
Pressure," U. S. Foresl Service Research Note FPL-0104, September 1965.
4-22 Raville, M. E., "Analysis of Long Cylinders of Sandwich Construction Under
Uniform External Lateral Pressure, " FPL Report No. 1844, November 1954.
4-23 Ravillc, M. E., "Supplement to Analysis of Long Cylinders of Sandwich
Construction Under Uniform External Lateral Pressure, " FPL Report No.
1844-A, February 1955.
4-24 Raville, M. E., "Buckling of Sandwich Cylinders of Finite Length Under
Uniform External Lateral Pressure, " FPL Report No. 1844-B, May 1955.
4-25 Norris, C. B. and Zahn, J. J., "Design Curves for the Buckling of Sandwich
Cylinders of Finite Length Under Uniform External Lateral Pressure, " FPL
Report No. 1869, May 1959.
4-26 Norris, C. B. and Zahn, J. J., "Design Curves for the Buckling of Sandwich
Cylinders of Finite Length Under Uniform External Lateral Pressure,"
U. S. Forest Service Research Note FPL-07, 1963.
4-27 Kazimi, M. I., "Sandwich Cylinders Part II - Uniformity of the Mechanical
Properties of the Core, " Aerospace Engineering, September 1960.
4-iii
4-28 Jenkinson, P. M. and Kuenzi, E. W., "The Buckling Under External Radial
Pressure and Buckling Under Axial Compression of Sanchvich Cylindrical
Shells, " FPL Report PE-227, January 1962.
4-29 C_rard, G., "Torsional Instability of a Long San(hvich Cylinder, " Proc. 1st
U. S. Natl. Congr. Appl. Mech., Copyright 1951.
4-30 March, H. W. and Kuenzi, E. W., "Buckling of Sandwich Cylinders in
Torsion, " FPL Report No. 1840 Revised, January 1958.
4-31 Gerard, G., Introduction to Structural Stability Theory, McGraw-Itill Book
Company, Inc., New York, Copyright 19(;2.
4-32 Timoshenko, S. P. and Gere, J. M., Theory of Elastic Stability, McGraw-
Hill Book Company, Inc., New York, Copyright 1961.
4-33 Lundquist, E. E., "Strength Tests of Thin-Walled Duralumin Cylinders in
Combined Transverse Shear and Bending, " NACA TN 523, 1935.
4 -34 Lundquist, E. E. and Burke, W. F., "StrenKth Tests of Thin-Walled
Duralumin Cylinders of Elliptic Section, " NACA TN 527, 1935.
4-35 Gerard, G. and Becker, H., "Handbook of Structural Stability, Part II1 -
Buckling of Curved Plates and Shells, " NACA TN 3783, August 1957.
,t-36 Batdorf, S. B., Stein, M., and Schildcrout, M., "Critical Stress of Thin-
Walled Cylinders in Torsion, " NACA TN 1.244, 194_.
4-37 Maki, A. C., "Elastic Stability of Cylindrical Sandwich Shells Under Axial
and Lateral Load, " U. S. Forest Service Research Note FPL-0173, October
1967.
4-38 Batdorf, S. B., Stein, M., and Schildcrout, M., "Critical Combinations of
Torsion and Direct Axial Stress for Thin-Walled Cylinders, " NACA Technical
Note 1345, June 1947.
4-112
5
GENERAL INSTABILITY OF TRUNCATED CIRCULAR CONES
5.1 AXIAL COMPRESSION
It appears that no significant theoretical solutions have been published for axially com-
pressed sandwich cones. Therefore, for the purposes of this handbook, the equivalent-
cylinder concept of Seide, eta!. [5-11 has been adopted as a practical expediency.
Based on a large array of test data from thin-walled, isotropie (non-sandwich), trun-
cated cones, Seide, et al. concluded that the critical stresses for such cones can be
taken equal to the values for circular cylinders which satisfy the following conditions:
a. The wall thickness of the equivalent cylinder is equal to that of the cone.
In the ease of sandwich constructions, the logical extension of this con-
dition is that the equivalent cylinder have the same facing and core thick-
nesses found in the cone.
b. The radius of the equivalent cylinder is equal to the finite principal radius
of curvature at the small end of the cone.
c. The length of the equivalent cylinder is equal to the slant length of the
e one.
In Reference 5-2, Baker presents test data from two axially compressed, truncated
sandwich cones having vertex half-angles equal to 15 degrees. These data were used
in conjunction with the foregoing equivalent-cylinder concept to arrive at knock-down
factors _c" The results are shown in Figure 5.1-i, along with data obtained from
5-1
!!
i i
+ t
_t
t !
2_
I
4 J ,
-ill
i!:
I I
5-2
axially compressedsandwichcylinders. This figure alsoincludesthe designcurve
recommended
in Section4.2.2 for suchcylinders. It canbe seenthat the datafrom
the conesare in favorableagreementwith the results obtainedfrom cylinders. This
providesat least a small degreeof experimentalsubstantiationfor application(>_

the
equivalent-cylinderapproachto sand_vich
cones. However,in view of the scarcity of
test points from conicalspecimens,this methodcanpresentlybe consideredas only
a "best-available"criterion.
5-3
For simply supported, truncated, right-circular, sandwich cones subjected to axial
compression, the critical stresses a and a (for facings 1 and 2 respectively)

crl cG
may be computed from the equations and curves of Section 4.2.2, provided that the
following substitutions are made:
So
The values
• t_, t_, tC , and h are measured as shown in Figure 5.1-2.
(There is no preference as to which facing is denoted by the subscripts
1 or2.)
6
The radius R is replaced by the effective radius R shown in Figure 5.1-2.
e
C°
The length L is replaced by the effective length L shown in Figure 5.1-2.
e
Both Ends
Simply Supported
Axis of
P, lbs
P, lbs Revolution
R
small
View A
N()F}:
t 1, 12 t , h, I_"lll a] r ' let ' a_ld
[ c art all o_t a,l_rttl ul IllrJlq (i I
it1( hes w]li}t ¸ _ I_ r41ca_url'd irl
Figure 5.1-2. Truncated Sandwich Cone Subjected to Axial Compression
5-4
The applied axial load P and the computed stresses are associated with the directions
indicated in Figure 5.1-2. In addition, since the maximum stresses occur at the small
end of the cone, the critical values are associated with this location. For both elastic
and inelastic cases, one can therefore write
Pcr = 2_Re ((rcr It_ +_cr_t_) cos2_ (5.1-I)
where
Rsmall
R - (5. i-2)
e cos o_
It is recommended that the approach specified here be applied only to cases where
_ 30 degrees.
Plasticity reduction factors should always be based on the stress at the small end of
the cone (see Section 9).
5-5
5.2 PURE BENDING
It appears that no significant theoretical solutions have been published for sandwich
cones subjected to pure bending. Therefore, for the purposes of this handbook, the
equivalent-cylinder concept of Seide, et al. has been adopted as a practical expediency.
Based on a large array of test data from thin-walled, isotropic (non-sandwich), trun-
cated cones, Seide, et al. concluded that the critical peak stresses for such cones can
be taken equal to the corresponding values for circular cylinders which satisfy the
following conditions:
a° The wall thickness of the equivalent cylinder is equal to that of the cone.
In the case of sandwich constructions, the logical extension of this condi-
tion is that the equivalent cylinder have the same facing and core thick-
nesses as are found in the cone.
b. The radius of the equivalent cylinder is equal to the finite principal radius
of curvature at the small end of the cone.
c. The length of the equivalent cylinder is equal to the slant length of the cone.
No test data are available for sandwich cones which are of the types considered in this
handbook and are subjected to pure bending. Therefore, the validity of the method
recommended here has not been experimentally verified and can only be considered as
a "best-available" approach.
5-6
5.2.2 Design Equations and Cu_wes
For simply supported, truncated, right-circular, san_vich cones subjected to pure
bending, the critical peak stresses _ and ff (for facings 1 and 2, respectively)
e r 1 c r>
may be computed from the equations and curves of Section 4.:'.2, provided that the
folh_ving substitutions are made:
a. The values t_, t.z, re, and h are measured as shown in Figure 5.2-1.
(There is no preference as to which facing is denoted by the subseripts
1 or 2.)
b. The radius R is replaced by the effective radius R e shown in Figure 5.2-1.
c. The length L is replaced by the effeetive length L e shown in Figure 5.2-1.
[
1
"r C - Axis
a_nraJll \ r<evoIut ;;_n
View A
Rsmall
L
Note:
tl, t2, t c, h, R small' Re, and 1, e
are all measured in units of inches
while Olis measured in degrees.
Figure 5.2-1. Truncated Sandwich Cone Subjected to Pure Bending
5-7
The applied bending moment M and the computed stresses are associated with the
directions indicated in Figure 5.2-1. In addition, since the maximum stresses occur
at the small end of the cone, the critical values are associated with this location.
When the behavior is elastic, one can therefore write
Mcr = _r R 2e ((Ycr 1 tl +Crcr_t2) cos 3(_ (5.2-1)
whe re
Rsmall
R = (5.2-2)
e cos (_
To compute M cr when the behavior is inelastic, one must resort to numerical inte-
gration techniques.
o_ _ 30 degrees.
Plasticity reduction factors should always be based on the peak compressive stress at
the small end of the cone (see Section 9).
5-8
5.3 EXTERNAL LATERAL PRESSURE
The loading condition considered here is depicted in Figure 5.3-1. As shown, the cone
is subjected to a uniform external lateral pressure. The axial component of this loading
o_ R R
i
-I-
R
smal 1 F
%
R
large
!
w c, lbs/in
Figure 5.3-1. Truncated Cone Subjected to Uniform External Lateral Pressure
is reacted by a uniform compressive running load at the large end of the cone. This
results in principal membrane stresses which may be computed as follows, when the
core has a relatively high extensional stiffness in the direction normal to the facings:
P Ram (5.3-!)
(YH - (t_ + t_)
(TM (t_ + tp) 2c-_s _ R (5.3-2)
5-9
where
R (5.3-3)
R_ - cos (_
and
(YH = Hoop membrane stress, psi.
(rM = Meridional membrane stress, psi.
p = Uniform external lateral pressure, psi.
= Finite principal radius of curvature of middle surface, inches.
= Thicknesses of the facings, inches. {There is no preference as

to which facing is denoted by the subscripts 1 or 2.)
R = Radius of middle surface measured perpendicular to the axis of

revolution, inches.
Radius of middle surIace, at small end of cone, measured per-

Rsmal 1 =
pendicular to the axis of revolution, inches.
Radius of middle surface, at large end of cone, measured per-

Rlarg e =
pendicular to the axis of revolution, inches.
= Vertex half-angle of cone, degrees.
Since the radii R and I_ vary with the axial location, the stresses ¢YH and a M are non-
uniform over the conical surface. The maximum values for each of these quantities
occur at the large end of the cone.
It appears that no significant theoretical solutions have been published for the stability
of truncated sandwich cones which are subjected to uniform external hydrostatic pres-
sure. Therefore, for the purposes of this handbook, the equivalent-cylinder approach
suggested in Reference 5-11 has been adopted as a practical expediency. Based on
this method, the critical lateral pressure for the truncated cone may be taken equal
5-10
to that for anequivalentcircular sandwichcylinderwhichsatisfies the following
conditions:
a. Thefacingandcorethicknessesof the equivalentcylinder are the same

as thosefoundin the cone.
b. Thelengthof the equivalentcylinder is equalto the slant lengthof thc conc.
c. The radius ofthe equivalentcylinder is equalto the averagefinite principal

radius of curvatureof the cone. Thatis,
Rsmal 1 + Rlarg e
R = (5.3-4)
e 2 cos
The critical lateral pressure for the equivalent cylinder can be obtained by using the
equations and curves of Section 4.4.2.
Since no test data are available from truncated sandwich cones subjected to external
lateral pressure, the reliability of the foregoing approach has not been experimentally
verified and can only be considered as a "best-available" technique.
5-11
For a simply supported, truncated, right-circular, sandwich cone subjected to uni-
form, external, lateral pressure, the critical pressure may be taken equal to the
critical lateral pressure for an equivalent sandwich cylinder which satisfies the
following:
a, The values t_, t:., tc, and h are measured as shown in Figure 5.3-2.
b. The length is taken equal to the slant length L .

e
c. The radius is denoted R and is computed from the formula

e
Rsmal 1 + Rlarg e
R = (5.3-5)
e 2 cos_
where Rsmal 1, Rlarg e, and _ are as shown in Figure 5.3-2.
t k_\
,..--- BOTH ENDS VIEW A
SIMPLY SUPPORTED
tl, t 2, Ic, h, R, R2, Rsmall,, Rlarge'

a_d Lu aR: all measured it] units of
in{'hcs while _ is measured in degrees,

NOTE:
Figure 5.3-2. Truncated Sandwich Cone
5-12
The critical lateral pressurefor the equivalentsandwichcylinder canbe obtainedby
usingthe equation,_
andcurvesof Section4.4.2.
Plastieily c_)nsiderations
shouldbehandledas specified in Section 9.2. Tile plasticity
roduclion factor 77 should always be based on theprincit)al nlembrane stresses at the
lat-ge end of the cone where
P R1 arge
= in. 3-6)
_I! (L _ t:.) cos q
P Re ( R_m_ll /
_M - (L t_) \1 -- (5.3-7)
. ' _ Rlarg e /
It is ruconm_(,nded thai the at)l)roach specified here be applied only to cases where
c_ _ 30 degrees.
5-13
5.4 TORSION
5.i. 1 BasicPrinciples
It appearsthat no significantlheoreticalsolutionshavebeen1)ublished
for sandwich
conessubjectedto torsion. Therefore, for thepurl)(,s(,s,)1this handbook,the
equivalent-cylinderconcepl()t'Seide[5-3i hasbeenadopledas a practical expediency.
Basedonthe analysisof his numericalcomputationsfor lhin-walled, isotropie {non-
sand_vich),
truncatedcones,Scideconcludedthat the critical l¢)rquesfor suchshells
canbe takenequallo the valuesfor circular cylinderswhichsalisfy the following
conditions:
In the case of sandwich constructions, the logical extension of this condi-
tion is that the equivalent cylinder have the same facing and core thick-
nesses as arc found in the cone.
[?. The leng-th of the equivalent cylinder is equal to the axial length of the cone.
c, The radius of the equivalent cylinder is coml)uted from the relationshit)
'lc :: (Rsmall cosc_)1 I[( R _. _ 1 _ R---_---_maH/j- 1 1 )I/larg_eRsmall >]1 ,,5.4-1)
where
R = Radius of equivalent cylinder, inches.

C
It = Radius at small end of cone, inches (measured perpendicular

small
to the axis of revolution).
Rlarg c :: Radius at large end of cone, inch(_s (measured perpendicular

to the axis of revolution).
cy :_ Vertex half-angle of cone, degrees.
5-14
In R(_ference 5-1, Seide, et al. present test results from ten isotropic (non-sandwich),
truncated cones which were subjected to torsion. These tests included specim¢,ns
having vertex half-angles (c_ of both 30 and (;0 degrees. The agreement of these,
results with equivalent-cylinder predictions was similar to that obtained from com-
parisons of test data from isotropic (non-sandwich) cylinders against the corresponding
small-deflection theoretical solutions. For conical sandwich constructions it was
therefore decided to use the same knock-down factor (7 s = 0.80) as was selected in
Section 4.5 for sandwich cylinders under torsion.
No test data are available for sandwich cones which are of the t33)es considered in this
handbook and are subjected to torsion. Therefore the method recommended here has
not been ex-perimentally verified and can only be considered as a "best-available"
approach.
5-15
5.,1.2 Design Equations and Curves
For simply supported, truncated, right-circular sanck_'ich cones subjected to torsion,
the critical torque may be computed from the equation
T = 2vR z (t t + t._) y_ (5.4-2)

er e • cr
where
T Critical torque for sandwich cone subjected to torsion, in-lbs.

cr
R =: Radius of equivalent sandwic:h cylinder, inches r see Equation

o
(5.,,_-3)].
t 1 and t2 = Ttticknesses of the facings, inches. (There is no preference

as to which faeing is denoted by the subscript 1 or 2.)
/
T = Critieal shear stress for equivalent sandwich cylinder when
cr
subjected to torsion, psi. (it should be noted that this value is
not equal to the critical sh,-ar stress of the conical sandwich
construction. )
The radius R is coml)uted from

e
1 1
(,_. 4-:_')
Re = (Rsmal 1 c_)scs) 1 + [2 (1 + Rlarge_]'- [1(I + Rlar_e_] -_
Rsmall/] Rsmall/l
where Rsmal 1, Rlarg e, and c_ are as shown in Figure 5.4-1.
The stress 7_ may be computed from the equations and cur_'es of Section 4.5.2 pro-
or
vided that
a. The values t_, t_, te, and d are measured as shown in Figure 5.4-1.
bg The radius R is replaced by the effective radius R .

e
co The length L is taken equal to the axial length of the cone (see Figure
5.4-1).
5-16
Both Ends
Simply Supported /
7 d .!
( R
i
small
Axis of
Revolution
l, Hl-/bs
T, itl-lbs View A
N _Tt- i1, 12, it, l/, tl, I, P'srl_al[,
aud Rlarg c art al_ illtasclr_ tl

ill II[lllS O[ il_C lies wllilt' t+_ iS
rneastlrt d i_ (t('_rt cs,
Figure 5.4-1. Truncated Sandwich Cone Subjected to Torsion
In a truncated cone which is subjected to torsion, the maximum shear stress will occur
at the small end. Hence, for sandwich constructions of this type, the critical stress
value is associated with that same location. One can therefore write
T
cr
rcr = 2rr R _ (t_+t 2) (5.4-4)

small
where
r = Critical shear stress for truncated sandwich cone when subjected

cr
to torsion, psi.
_ 30 degrees.
Plasticity reduction factors should always be based on the stress at the small end of
the cone (see Section 9).
5-17
:).;_ TRANSVERSE SIIEAR
Th(, case considered here is that of a truncated sandwich cone which is subjected only
to transverse shear forces a,_ shown in Fig_,re 5.5-1. No_c that all transverse see-
tions, such as A-A, are subjected to the same magnitude <)f shear load.
--,.-- A
J l.i
____ --
_.--A
Figxlre 5.5-1. Truncated Cone Subjected to Transverse Shear
This, of course, is a pur_qy h3_pothetical loading condition since it does not result in
over-all static equilibrium of the structure. To obtain the necessary balance of forces
and moments, it is required that an external bending moment also be present. Never-
theless, the hypothetical unbalanced loading system does prove to be of interest since
the combined effects of transverse shear and its associated bending are usually analy-
zed by using an interaction equation. Such a relationship invulves both the critical
peak meridional stress under a bending moment acting alone and the critical peak
shear stress corresponding to the subject artificial loading condition.
It appears that no significant theoretical solutions have been published for sandwich
cones subjected to transverse shear. Therefore, for the purposes of this handbook,
5-18
the conceptusedfor sandwichcylinders (see Section 4.6) will also be adopted here as
a practical expediency. As noted in Section 4.6, the results from a series of tests
[5-4 and 5-5] on isotropic (non-sanchvieh), circular and elliptic cylinders led to the
conclusion [5-6] that
Lower-Bonndrcr TestValues for 1

Transverse Shear Loading _ 1.25 (5.5-1)
Small-Deflection Theoretical ]
rcr Values for Torsional Loading]
To properly understand this ratio, it should be observed that for torsional loading of a
thin-walled circular cross section the shear stress Tcr is uniformly distributed around
the circumference. On the other hand, under transverse shear loading, the shear
stress is nonuniform and the value _'cr then corresponds to the peak intensity which
occurs at the neutral axis.
For the lack of a better approach, it was reconm_ended in Section 4.6 that Equation
(5.5-1) be used for the design and analysis of sandwich cylinders that are subjected to
transverse shear forces. For the same reason, it is recommended here that Equa-
tion (5.5-1) also be used for truncated sandwich cones. In the latter ease, the re-
quired small-deflection theoretical Tcr values for torsional loading should be obtained
as specified in Section 5.4, with the exception that 7s must now be taken equal to unity.
No sanchvich test data are available to substantiate the reliability of this practice.
Until such data do become available, one can only regard this proeedm_e as a "best-
available" approach.
5-19
For simply supported,right-circular, truncatedsandwichconessubjectedto trans-
verse shearforces, the critical peakshearstress maybe comtmtedfrom theequation
= 1.25 (T) (5.5-2)

rcr cr Torsion
Ys 1.0
where
(_r) = The critical torsional shear stress obtained by substituting

Torsion Ys = 1.0 throughout the meth_gls cited in Section 5.4, psi.
Ys = 1.0
In a truncated cone which is subjected to transverse shear, the maximum shear stress
will occur at the small end. ttence, the critical stress value is associated with that
1ocation.
Plasticity reduction factors should always be based on the stress at the small end of the
cone (see Section 9).
When the behavior is elastic, the critical transverse shear force .._Fv)cr can be corn-
puted from the following:
(F v) cr = _ Rsmall (t_ +%)r (5.5-3)

cr
To compute (Fv)cr when the behavior is inelastic, one must resort to numerical inte-
gration techniques.
5-20
5.6 COMBINED LOADING CONDITIONS
5.6.1 General
For structural members subjected to combined loads, it is customary to represent
critical loading conditions by means of so-called interaction curves. Figure 5.6-1
shows the graphic format usually used for this purpose. The quantity R i is the ratio
of an applied load or stress to the critical value for that type of loading when acting
alone. The quantity" Rj is similarly defined for a second type of loading. Curves of
this form give a very clear picture as to the structural integrity of particular con-
figurations. All computed points which fall within the area bounded by the interaction
curve and the coordinate axes correspond to stable structures. All points lying on
or outside of the interaction curve indicate that buckling will occur. Furthermore,
as shown in Figure 5.6-1, a measure of the margin of safety is given by the ratio of
distances from the actual loading point to the curve and to the origin. For example,
assume that a particular structure is subjected to the combined loading condition
corresponding to point B of Figure 5.6-1.
(Rj)
D -- -- -- M
I
1.0
(Rj) B --__
Rj 1/111
0 ].0
Ri
Figure 5.6-I. Sample Interaction Curve
5-21
Then, for proportional increases in R. and R., the margin of safety (MS) can be com-
1 j
puted from the following:
(Rj) D
MS - i
(Rj) B (5.6-1)
As an alternative procedure, one might choose to compute a minimum margin of safety
which is based on the assumption that loading beyond point B follows the path BM.
Point M is located in such a position that BM is the shoo-test line that can be drawn
between point B and the interaction curve. The minimum margin of safety can then
be calculated as follows:
Minimum MS - OB + BM 1
OB (5.6-2)
5-22
5.6.2 Axial Compression Plus Bending
In Section 4.7.2 this loading condition is treated for the case of circular sandwich
cylinders. For such configurations, it was concluded that one may use the following
interaction relationship:
(5.6-3)
R c +%=1
where
c
R = (5.6-4)
c "/c ((_c) C L
ab
(5.6-5)
%= Tb ((_c)
CL
and
= Uniform compressive stress due solely to applied axial load,
(Y
c
psi.
= Pcak compressive stress due solely to applied bending moment,

_b
psi.
= Classical theoretical value for critical uniform compressive

(_c)C L stress under an axial load acting alone, psi.
Tc = Knock-down factor given by Figure 4.2-8, dimensionless.
Tb = Knock-down factor given by Figure 4.3-2, dimensionless.
In this handbook it is proposed that for truncated sandwich cones the cases of pure
bending and of axial load acting alone both be treated by means of an equivalent-
cylinder concept (see Sections 5.1 and 5.2). For both types of loading, the radius
of the equivalent cylinder is taken equal to the finite principal radius of curvature at
the small end of the cone. It should be noted that the maximum stresses from both
5-23
bendingandaxial compressionoccurat this samelocation. In viewof theseseveral
considerations,it is assumedherethatEquations(5.6-2,)through(5.6-5) canbe

appliedto truncatedsanchvich
conesif
a,
o"c and gb are bcCh computed for the meridional direction and at the small
end of the cone, and
b.
the values for ;¢e' and Yb' and ((Te)r_T are those which apply to the equivalent
sandwich cylinder described in Sect_i_ns 5.1 and 5.2. (It is important to
keel) in mind that 7c must be taken equal to 1.0 when computing the value
((_c) C L • )
Since no test data have been published for truncated, sandwich cones subjected to axial
compression plus bending, the recommended approach has not been experimentally
verified and can only be regarded as a "best-available,, method.
5-24
For simply supported,truncated, right-circular sandwichconessubjectedto axial
compressionplusbending,the followinginteractionequationmaybeemployed:
Re+ Rb = 1 (5.6-6)
where
c
R - (5.6-7)
c 7e (6c)CL
_b
ab 7t) (5c) (,_.(i-s)

CL
Equation (5.6-6) may be used for cones of any length. A plot of this equation is given
in Figure 5.6-2.
The quantity gc is the uniform meridional compressive stress, at the small end of the
cone, due to the axial force acting alone.
The quantity Crb is the peak meridional compressive stress, at the small end of the
cone, due to the bending moment acting alone.
The quantities Ye' _b' and (Cre)CL- are those which apply to the equivalent sandwich
cylinder described in Sections 5.1 and 5.2.
In Equations (5.6-7) and (5.6-8), the knock-down factors )'e and_b are those obtained
from Figures 4.2-8 and 4.3-2, respectively.
The quantity (_c)CL is simply the result obtained by using Yc = 1.0 in the method of
Section 4.2.2.
5-25
Plasticity considerationsshouldbehandledas specificdin Section9.2 except,that in
this case, one may use
(a)
[ l
r? -: [ _-]--_
1 1 EZ for short cones, and
(b) r7 =
Ii- VEtEs
o l for moderate-length through long cones.
1---_ l Ef
The plasticity reduction factor 77 should always be based on the peak compressive stress
at the small end of the cone.
1.0t
I 'iii
0.8
\ ........ T
0.6
It b
O. 4 ................... 4-.......... ' ' ' _ ' ' _ -
..... r t
!
0.2 1 _
i I '
0 i .
0 0.2 0.4 0.6 0.8 .0
R
e
Figure 5.6-2. Design Interaction Curve for Truncated Sandwich Cones

Subjected to Axial Compression Plus Bending
5-2 6
5.6.3 Uniform External Itydrostatic Pressure
The loading condition considered here is depicted in Figure 5.6-3. As shcavn, the cone
is subjected to a uniform external pressure over the lateral surface and both cnd closures.
Both Ends
Simply SuppoEtLd
p, psi
c_ R Rz
-I
p, psi
..q-_
_t_
Rlarg c
Figure 5.6-3. Truncated Cone Subjected to Uniform

External Itydrostatic Pressure
This results in principal membrane stresses which may be computed as follows when
the core has a relatively high extensional stiffness in the direction normal to the
facings:
p R_
CrH - (t_ + t_) (5.6-9)
pR._ (5.6-10)
cr M - 2 (t_ + t_)
where
R
R - (5.6-11)
2 cos
5-27
and
=: Itoop membrane stress, psi.
- Meridional membrane stress, psi.

crM
p - Uniform external hydrostatic pressure, psi.
Finite principal radius of curvature of middle surface, inches.
tI andt -- Thicknesses of the facings, inches. (There is no preference

as to which facing is denoted 1)3, the subscripts 1 or 2.)
R : Radius of middle surface measured perpendicular to the axis

of revolution, inches.
(_ = Vertex half-angle of cone, degrees.
and r; are non-

Since the radii [/ and R vary with the axial location, the stresses (Ytt M
uniform over the conical surface. The maximum values for each of these quantities
occur at the large end of the cone.
of truncated sandwich c(mes which are subjected lo uniform external hydrostatic pres-
sure. Therefore, i'or lh(, purposes of this handbook, _he equivalent-cylinder at)preach
of Seide, et al. [5-11 has been adopted as apractical expediency. Based on alarge
array of test data from thin-walled, isotropic (non-sandwich), cylinders and truncated
cones, Scide, et al. concluded that the critical hydrostatic pressures for such cones
can be taken equal to the values for equivalent circular cylinders which satisfy" the
following conditions:
In the case of sanokvich constructions, the logical extension of this condi-
tion is thai the equivalent cylinder have the same facing and core thick-
nesses as are found in the cone.
5-28
b° The length of the equivalent cylinder is equal to the slant length of the cone.
C. The radius of the equivalent cylinder is equal to the average finite principal
radius of eurvature of the cone. That is,
Rsmall + Rlarge (5.6-12)

R -:
e 2cos
where
R = Radius of middle surface for equivalent cylinder, inches.

e
R = Radius of middle surface at small end of cone (measured

small
perpendicular to the axis of revolution), inches.
= Radius of middle surface at large end of cone (measured

Rlargc
perpendicular to the taxis of revolution), inches.
The critical hydrostatic pressure for the equivalent cylinder can be obtained by using
the equations and curves of Section 4.7.3.
The only available eN)erimental results for conical san&vieh shells under uniform
external hydrostatic pressure are the data from two tests eondueted by North American
Rock, veil, Corp. _5-7 and 5-8_ in conjunction with the Navajo missile program. To
assist in the preparation of this handbook, an analysis was made of the result published
in Reference 5-7. The other specimen was not studied since it was stressed too deeply
into the plastic region. The specimen of Reference 5-7 was also inelastic but the
stresses in this instance were low enough to permit reliable computations. Using the
approach of the present section in conjunction with the plasticity reduction criteria of
Sect{on 9, the design critieal pressure was computed to bc 36.4 psi. This is in satis-
faclot3' agreement with the experimental value of 43.6 psi.
5 ')9
The foregoingsubstantiates,to a very small degree,the reliability of the equivalent-
cylinder conceptrecommended
here. However,in viewof the lack of a sufficient
numberof test results, this approachcanpresentlybe consideredas only a "best-

available"method.
5-30
I,'or a simply supported, truncated, right-circular sandwich cone subjected 1o uniform,
external, hydrostatic pressure, the critical pressure may be taken equal to that for tin
equivalent sandwich cylinder for which
a. The values t_, t:., t c, and h are measured as shown in Figure 5.6-4.
of the cone as shown in

b. The length is taken equal to the slant length Le
Figx_re 5.6-<1.
C0 The radius is denoted R and is computed from the fornmla

e
Rsmal 1 + Rlarg e
R = (5.6-13)
e 2cos (_
where Rsmal 1, Rlarg e, and a are as shown in Figure 5.6-4.
, Rlargc
R R2
L_L\I
_-T
VIEW A
P'smail
BOTH ENDS
SIMPLY SUPPORTED
NOTE: t 1, l,. i c , h, R, R,2 Rslnall' Rlarge'

and Le are all measured in units of
inches while o/is measured in degrees.
Figure 5.6-4. Truncated Sandwich Cone
The critical hydrostatic pressure for the equivalent sandwich cylinder can be obtained
from the equations and curves of Section 4.7.3 if the ratios R and R are now defined
c p
as follows :
5-31
R = P
c ye (l_x)C L (5.6-14)
P
R =
(5. (;-15)
p Wp (_y)CL
w he re
p
Uniform, external, hydrostatic pressure applied to lateral
surfaces and end closures of the equivalent sandwich cylinder,
psi.
In Equations (5.6-14) and (5.6-15), the knoek-d(r_cn factor _/e is that obtained from
Figure 4.2-8, while v may be taken equal to 0.90.

'p
It should be noted that
(Px)CL Re
((rx) CL - 2 (t_ * t:_) (5.6-16)
or
2 (gx)CL(t ! _ t_)
(ISx)CL = R (5.6-17)
e
whe re
(C_x)CL = Classical theoretieal value for critical uniform axial com-

pressive stress when acting alone on the equivalent sandwich
cylinder. This value can be obtained by using Yc = 1.0 in the

equations and curves of Section 4.2.
The value (tSy)c L can be obtained by using Tp = 1.0 in the equations and curves of
Section 4.4.
Plasticity considerations should be handled as specified in Section 9.2. The plasticity
reduction factor r7 should always be based on the principal membrane stresses at the
large end of the cone where,
5-32
pR large
(5.G-18)
(YH= (t1 _ t2) (cos _)
P Rlarg e
(5. (;-19)
_M = 2 (t 1 + t_) (cos (_)
_ 30 degrees.
5-33
5.6.4 Axial CompressionPlus Torsion
Theloadingconditionconsideredhereis depictedin Figure 5.t;-5. Theaxial load P
canoriginatefrom anysourceincludingexternalpressureswhichare distributed
uniformly over the endclosures.
T, m-lbs torque
], Hl-lbs (orqtle_
f_ / \\
P. I bs
_.__________ ____ _ _ _ ___ _ P, l bs
Bol:h Ends
Simply Supported
Figure 5.6-5. Truncated Cone Subjected to Axial

Compression Plus Torsion
of truncated sanchvich cones under this combination of loads, ltowevcr, MacCalden
and Matthiesen _5-9] have arrived at certain conclusions for non-sandwich shells under
such loading and, for the purposes of this handbook, these results provide the basis for
an expedient engineering approach to the case of conical sandwich constructions. Based
on a large array of test data from Mylar specimens, MacCalden and Matthiesen con-
cluded that the following interaction relationship could be applied to thin-walled, iso-
tropic (non-sanchvich), truncated cones:
R +R _ = 1
c S (5.6-20)
5-34
where
R :2
P (5.6-21)
c m
(Pcr)Empirical
R T (5.6-22)
s
(Ter) Empirical
and
= Empirical lcnver-bound value for the critical axial load

(Pcr) Empirical
when acting alone, lbs.
= Empirical lower-bound value for the critical torque when

(Tcr) Empirical
acting alone, in-lbs.
This result is identical to that given in Reference 5-10 for thin-walled, isotropic (non-
sandwich) cylinders subjected to axial compression plus torsion. One might, therefore,
conjecture that in the case of sandwich constructions the interaction curves for trun-
cated cones under the subject loading condition are of the same shape as those pre-
sented in Section 4.7.4.2 for circular cylinders. The design equations and curves
recommended here arc based on this premise. That is, one might choose to view the
formula,
R + R2 = 1 (5.6-23)
c s
as a comprehensive interaction equation for truncated cones of both isotropic (non-
sanchvich) and sanchvich construction. However, it is important to note here that
MacCalden and Matthicsen observed that the presence of even a very small axial load
made the torsionally-loaded conical shell much more sensitive to imperfections than
was the case when no axial load was applied at all. They, therefore, recommended
that whenever R c is non-zero, the same knock-down factor be employed in computing
5-35
m
(Tcr)Empirica 1 as is used in the calculation of (Pcr)Empirieal" It was further speci-
fied that this single knock-down factor should be taken equal to that which applies for
the case of axial compression acting alone. The same practice is adopted here.
Caution should be exercised in implementing the foregoing recommendations, partially
because only the extremes o[ transverse shear rigidity of the core have been consid-
ered (see Section 4.7.4.1). In addition, although the interaction relationship for the
subject loading condition should probably be dependent upon a length parameter, no
investigations were made to establish the sandwich lengths over which Equation (5.6-23)
is a reasonable representation of the actual behavior. Furthermore, no test data have
been obtained for sandwich cones subjected to axial compression plus torsion. There-
fore, the general validity of Equation (5.6-23) has not been experimentally verified
and can only be regarded as a "best-available" approach.
5-36
For simply supported,truncated,right-circular sandwichconessubjectedto axial
compressionplustorsion, onemight chooseto employthe interactionformLda,
(5.6-24)
R +R e =1
C S
which is plotted in Figure 5.6-6 and where,

P
(5.6-25)
c (Per)Empirical
T (5.6-26)
- 3,c
p = Applied axial load, lbs.
T = Applied torque, in-lbs.
Lower-bound value for the critical axial load when

(Pcr)Empirical acting alone. This value can be obtained by using the
equations and curves of Section 5.1.2, lbs.
Lower-bound value for the critical torque when acting

(Tcr)Empirica 1= alone. Tkis value can be obtained by using the equa-
tions and curves of Section 5.4.2, in-lbs.
The knock-down factor obtained from Figure 4.2-8

_C
(dimensionless). For the purposes of the present
case, the quantity R (see Figure 4.2-8) must be set
equal to the equivalent radius R e which is computed
as follows:
R
small (5.6-27)
Re = cosc_
R = Radius at small end of cone, inches (measured per-

small pendicular to the axis of revolution).
= Vertex half-angle of cone, degrees.

5-37
0 O. 2 0.4
Figure 5.6-6.
Conditional Interaction Curve for Truncated Sandwich Cones
Subjected to Axial Compression :Plus Torsion
The factor _ should be introduced into the demoninator of the ratio R s only when
R c is non-zero. For the special case where no axial load is present (R c 0), Rs
should be taken equal to T + (Ter)Empirical"
Attention is drawn to the fact that in Section 5.6.4.1, several factors are cited which
shed considerable doubt upon the reliability of results obtained from the indiscriminate
use of Equation (5.6-24) mid Figure 5.6-6. In view of these uncertainties, one might
often choose to employ the straight-line interaction formula,
R +It =1
c s (5.6-28)
which is plotted in Figure 5.6-7. This relationship can be used with confidence for
any length of cone and for any region of transverse shear rigidity of the core, since
5-3 8
experiencehas shownthat the linear interactionformulais neveruneonservative
for
shell stability problems, ttowever, in manyeasesit will, of course, introduce
excessiveconservatism.
Plasticity considerationsshouldbehandledas specifiedin Section9.2. Theplasticity
reductionfactor r{ should always be based on the stresses at the small end of the cone.
0o
0°
0.4
li
(Ii 0.2 0.4 0.6 0.8 1.0
R e
Figure 5.6-7. Conservative Interaction Curve for Truncated Sandwich Cones

5.6.5 Other Loading Combinations
In Section 5.6.4, the combined loading condition of axial compression plus torsion is
treated. The interaction relationships presented there can be used for an additional
5-39
loadingcombination1)5, recognizing that at any given axial location on the cone the
peak meridional stress due to an applied bending moment can be converted into an
equivalent uniform meridional stress. With this in mind, the design equations and
era'yes of Section 5.6.4.2 can be used for the combination of axial compression plus
bending plus torsion which is depicted in Figure 5.6-S.
Fig_lre 5, 6-8.
Truncated Cone Subjected to Axial Compression
Plus Bending Plus Torsion
To accomplish this it is simply required tlmt the quantity Rc be redefined as follows:
p#
R
c (tScr) .... (o. 6-29)
r_mplmcal
whcre
(5, 6-30)
and
P Applied axial load, lbs.
M Applied bending moment, in_-lbs.
Yc = Axial comt)ression tmock-down factor from Figure 4.2-8

dimensionless.
Note: Vor the purposes of the present ease, the quantity R (see
Figure 4.2-8) must be set equal to the equivalent radius Re which
is comt)uted as follows:
5-40
Rsmall
(5.6-31)
Re - cosa
Tb Bending lo]ock-down factor from Figxlre 4.3-2, dimensionless.
Note: For the pro'poses of the present case, the quantity R (see Fignre
4.3-2) must be set equal to the equivalent radius R e which is computed
as follows:
Rsmall (5.6-32)
Re - coset
R :: Radius at small end of cone (measured perpendicular to the

small
axis of revolution), inches.
=: Vertex tmlf-angle of cone, degrees.
The foregoing formula for P_ is based on the principles cited in Section 5.2
Since no sandwich test data are available to substantiate the recommendations made
here, they can only be regarded as a "best-available" criterion.
For simply supported, truncated, right-circular sandwich cones subjected to the
loading condition depicted in Figure 5.6-8, one may use the design equations and
curves of Section 5.6.4.2, except that the quantity R c must now be defined as follows:
p'
Rc = - (5.6-33)
(Pcr)Empirical
where
(5.6-34)
_,Tb/ a11
5-41
c_
_1o
_4
", :a£
c_
II
©
o
,:D
c_
r_
©
t_
%[, b d ' .
_, HIh
d
4_ _ _' D
ii_ Q_
_q
_'_ _ > r, _"
q_
_q
b
c4
_-. ,
C_O
©
I
L_
o
C_
b_
e_
©
0
tO'
5-42
0
< o
0 "_
u.,
0
:Z ._ • o:
0a
b _
Zo _
%
0
c_
b ,.--i
o _ _
.m la: _
0
D
_. o o_
01
At
11
II II
0 _
b _
/ o
0
m
V,
5-43
o
0
0
,_a
C
II
©
,Jl
¢1
E ":'1
o
r,..) 4
_1 o
'gs H
0
0 xz
, ,-..,i E
r,..)
k., ©
.fo _
o
c
j, p t_ ° ,_0
[...4
e-, ,JI -_ r--i, ..... _ _ '_
° II
o
E °
0 N _Dt"
F_ N
_ "
o ,j
r_ r_
H
r/l ,i
II
eo -_ zy
C , g C
,..t=
,,-i
-5 .2 ._
C,
2,
j m
.a
e-,
© z, ; d
I ht_
LQ
,-..i r"r" I
©
,7--,..,i _.1=
2 _ ,_..J,
5-44
r_
+
a ,_ c
g _
_ # _ .._
0
2 _o
u 4
©
b
"_ II ©
_ _ .
,_ _ • _
_g .... II
_ v
'C d'
d'
b _
__ _
Z
._._jo_ ,v _ .-, _
_7
,;'1
5-45
:4
',9
Ir rn
©
-d
°
o_
-i T
._ _.,'_
o
LJ
_ 4 _ b
r.q , ©
o
A ,-' 2
c_
'_ N0
C u ^', 2 c "_ D.21 _

<n
C_ r/,
--5
_ •
©
b_
_ m
II
r/l
>___. _?
I ° II
r/?
5 o iz 2 N e- I_' o . :,
._f
. 0
I _o -
_9
°2 O
C-
..--, © © ' > m '_
5-46
_q
,.%
i© o
o I
.,_
oo
4
• ©
o
o
o g
7
o
ba _._
c0 o $_ o
O"
o ,.c: _
© r_
mm ,'
II M bo m '_d
,4
II % + • I_ _ _,._ _ g cq
2 _i _ _*
LJ
+
,--4 r_ _ _..o
v
% o _
II g--
II
r,o o.) e_ II
T_
C3
II g .o _ 2:
N1 o
:>
d'
v1
d
r
o
5-47
+
M=
-j 2 "S,
.,% '?
o
U3
m-
.c
e_
o
_J
_d
r_
2,
E
D
Lq
C '_ +i
£,+ +f+I:
,.,Q
0 , 4 "f.
+4
,.£ _-._ Z ,_ ,.,_ ,_

+4
II
[[
o
r_
c"J
,. :.;
'B _3
fi
+ +i + -
B
B r_
0
O'3 r_
hq
I _Ju
Lm+
,.-+ +
.q
+_
+.,-3
++
_i _ ,:j
5-48
---- ...... .................
r_
C _ cO
c_ o
2o
"EI I-_ _ I:,C
0Y
_ _ _ >
o ,'_
._ _ _ ,-,
©
II
o
..-, +
c) 0 _,
e_
_ _°" s _
o "r,
_ 2
;a_ =
L_ ,al
e_
VI
_3
J_l
I
t
I c_
_ E
I
5-49
r_ _ ._
.o
o t_
t_
©
_8
4 .._ _ _ _ _
_ _ e _3e
©
_ m _ NH_
0 0
O
0
"_ ._ _. _'_o
o
g
._ °
>
_o = g
&
t_
5-51
o
II
5-52
o
5-53
o
G_
c_
O ao
0 o_
f_
oiig
_S
QJ
o I
0
o
olml
c_
o
O
c_
f_ c_
O
c_
_J o
Z
c_ ©
o_
o
c8_
c_
c_
5-54
REFERENCES
5-1 Seide, P., Weingarten, V. I. , and Morgan, E. J., "Final Report on the Devel-
opment of Design Criteria for Elastic Stability of Thin Shell Structures", STL-
TR-60-0000-19425, 31 December 1960.
5-2 Baker, E. If., "Experimental Investigatim of Sandwich Cylinders and Cones
Subjected to Axial Compression", AIAA Journal, Volume 6, No. 9, September
1968.
5-3 Seide, P., "On the Buckling of Truncated Conical Shells in Torsion", Journal
of Applied Mechanics, June 1962.
5-4 Lundquist, E. E., "Strength Tests of Thin-Walled Duralumin Cylinders in Com-
bined Transverse Shear and Bending", NACA TN 523, 1935.
5-5 Lundquist, E. E. and Burke, W. F., "Strength Tests of Thin-Walled Duralumin
Cylinders of Elliptic Section", NACA TN 527, 1935.
5-6 Gerard, G. and Becker, tt., "Handbook of Structural Stability, Part HI - Buckling
of Curved Plates and Shells", NACA TN 3783, August 1957.
5-7 Anonymous, "Test of XSM-64A Booster Fuel Tank Conical Bulkhead No. 3,"
North American Rockwell, Corp. Report No. AL-2622, 15 September 1957.
5-8 Anonymous, "Test of XSM-64.A Booster Fuel Tank Conical Bulkhead No. 4,"
North American Rockwell, Corp. Report No. AL-2623, 16 September 1957.
5-9 MacCalden, P. B. and Matthiesen, R. B., "Combination Torsion and Axial
Compression Tests of Conical Shells", AIAA Journal, February 1967.
5-10 Batdorf, S. B., Stein, M., and Schildcrout, M., "Critical Combinations of
Torsion and Direct Axial Stress for Thin-Walled Cylinders," NACA Technical
Note 1345, June 1947.
5-55
5-11
Baker, E. tt., Cappelli, A. P., Kovalevsky, L., Rish, F. L., andVerette,
R. M. , Shell Analysis Mmmal, Prepared by North American Rockwell Corp.
for the National Aeronautics and Space Administration, Manned Spacecraft
Center, [[ouston, Texas, June 1966.
5-56
6
GENERAL INSTABILITY OF DOME-SHAPED SHELLS
6.1 GENERAL
This section deals with dome-shaped shells whose contours are surfaces of revolution.
Figure 6. i--i shows the shapes considered here, all of which are truncated at the equa-
tor. Note that the torispherical shape consists of a lower toroidal segment which
blends into a spherical cap. It is expected that the configurations shown here will
cover the large majority of the dome structures likely to be encountered in aerospace
applications. One should observe that for each of
,) ,? 2 2
x y
_2 b2 Y
_...._Spheri ca 1
_---x
I.
, -
Boundary Simply- t-Boundary Simply-

Supported or
Clamped Supported or Supported or
Clamped Clamped
(b) Ellipsoidal (c) Torispherical

(a) ttemispherical
Figure 6.1-1. Structural Dome Shapes
these domes the maximum radius of curvature RMa x occurs at the apex. As a practi-
cal engineering expediency, analysis of all the illustrated configurations will be based
on this radius.
6-1
In the case of externally pressurized, thin-walled, isotropic (non-sandwich) domes, it
has long been recognized that the test results normally fall far below the predictions
from classical small-deflection theory for the axisymmelric buckling of complete
spheres. The discrepancies are usually attributed to,
a. the shape of the postbuckling equilibrium path coupled with the presence of
initial imperfections,
b. the fact that large-deflection analyses of asymmetric behavior yield criti-

cal stresses approximately 20 percent lower than the small-deflection axi-
symmetric values, and
c. the fact that classical small-deflection theory does not account for pre-
buckling discontinuity distortions in the neighborhood of the boundary.
This is analogous to the situation described earlier in this handbook (see Section 4.1)
for the case of circular cylinders. For the latter, it has become common practice to
base stability analyses and design procedures on the use of classical small-deflection
theory modified by empirical knock-down factors. This approach was selected in
Section 4.1 for sandwich cylinders and is also adopted here for sandwich domes.
6-2
6.2 EXTERNAL PRESSURE
This section deals with the loading condition depicted in Figure 6.2-1. That is, a uni-
form external pressure acts over the entire surface of the sandwich dome. The net
p, psi
+
lbs l
l lbs
w j --
w -- c in
_' iI1
Figure 6.2-1. Sandwich Dome Subjected to External Pressure
vertical component of this loading is reacted by a uniform rtmning load on the boundary.
From Figure 6.1-1, note that the domes can have either simply-supported or clamped
edges. That is, during buckling the boundary is constrained such that no radial dis-
placements occur. In the simply-supported case, the shell wall is free to rotate along
the boundary whereas for clamped edges such rotations are completely suppressed. It
follows, of course, that intermediate restraints to edge rotation are also acceptable.
The theoretical basis used here is the classical, small--deflection solution by Yao [6-1]
as reformulated by Plantema [6-2]. This result embodies the following assumptions:
ao The facings are isotropie.
b. Both facings are of the same thickness.
C. Both facings have identical material properties.
6-3
d. Bending of the facings about their own middle surfaces can be neglected.
e. The core has infinite extensional stiffness in the direction normal to the
facings.
f. The eore extensional and shear stiffnesses are negligible in directions

pacallel to the facings.
g. The transverse shear properties of the core are isotropie.
R
11. The inequality 7-- >> 1 is satisfied,
C
W hc re
It _: Radius to middle surface of sandwich sphere, inches.

C
i. Approximations equivalent to those of I)omlell TG-a] can be applied.
Strictly speaking, this solution was derived for complete sandwich spheres which
exhibit small buckles that are axisymmetric with respect to a radius of the sphere.
The development isolated one such buckle as a free body so that shallow-shell theory
could then be employed. Yao presented his results in a form which is not conducive
to a ready physical interpretation of the phenomena involved. Therefore, Plantema
undertook to ex-press the final relationships in a manner which would foster some
insight in tkis regard, tie was able to sh(rvv that, when the core has isotropie trans-
verse shear stiffness, Yao's solution is identical to the equations given earlier in this
handbook for axially compressed circular sandwich cylinders rsee Equations (4.2-27)
through (4.2-30) and Equations (4.2-4) and (4.2-5)]. TEat is, when the knock-down
factor, Y'd' is included,
(_er = 3/d Kc°'o (6.2-1)
6--4
where
_TEf h
(6.2-2)
O" o R
and
When V ---2
C
1
K = !---V (6.2-3)
c 4 e
When V >-2
C
1
K - (6.2-4)
c V
C
where
0
V (6.2-5)
C
crimp
5 2
(7 - G (6.2-6)
crimp 2 tft c c
(I = Critical compressive stress for sandwich sphere, psi.

cr
= Young's modulus of facings, psi.

Ef
h = Distance between' middle surfaces of facings, inches.
v = Elastic Poisson's ratio of facings, dimensionless.

e
R = Radius to middle surface of sandwich sphere, inches.
tf = Thickness of a single facing, inches.

C
G = Transverse shear modulus of core, psi.

C
6-5
Theequivalence
betweenanaxially compressedsandwichcylinderandanexternally
pressurizedsandwichspherehasbeenanalyticallydemonstratedonlyfor the case
wherethe two facingshaveidenticalmaterial propertiesandare of the samethickness.
If oneassumesthatthis equivalencestill holdstrue whenthe facingsare of different
thicknesses,Equations(4.2-2) through(4.2-7) canthenbe usedhereif Gxzis replaced
by Gc so that, whenthe knock-downfactor Ydis included,
o"cr = yd K c o o (6.2-7)
where
h 2_/tl_
_7o = rtEf R (6.2-8)
and
When V 22
C
1
K = 1---V (6.2-9)
c 4 c
When V '2
C
1
K - (6.2-10)
c V
C
whe re
(7
0
V - (6.2-11)
e (7
crimp
5 2
- G (6.2-12)
crimp (t 1 + t2) t c c
t and t = Thicknesses of the facings (There is no preference as to

1 2
which facing is denoted by the subscript 1 or 2.), inches.
6-6
The relationshipbetweenKe andVe canbeplottedas shownin Figure6.2-2. It is
importantto notethat the valueVc = 2.0 establishes a dividing line between two
different types of behavior. The region where Vc < 2.0 covers the so-called stiff-
core and moderately-stiff-core sandwich constructions. When V is in the neighbor-

C
hood of zero, the core transverse shear stiffness is high and the sandwich exhibits
maximum sensitivity to initial imperfections. As V increases from

e
1.0
K e
L I
2.0
Vc
Figure 6.2-2. Schematic Representation of Relationship Between K and V

e --e
zero to a value of 2.0, this sensitivity becomes progressively less. The domain
where V _ 2.0 is the so-called weak-core region where shear crimping occurs.
c
Sandwich constructions which fall within this category are not influenced by the pres-
ence of initial imperfections and a knock-down factor of unity can be applied to such
structures. It should be possible to develop a continuous transitional knock-down
relationship which recognizes the variable influence of the core rigidity but this is
beyond the scope of the present handbook.
6-7
6.2.1.2 Empirical Kmock-D(_vn
Factor
As notedin Section6.1, for the purposesof this handbook,the allowablestressesfor
externallypressurizedsandwichdomesare establishedby applyinganempirical knock-
downfactor (_d)to the results from classical small-deflectiontheory, ttowever, since

the availabletest datafrom sandwichdomeconstructionsare very scarce, onecannot
yet determineYdvalueswith a highdegreeof reliability. Theonlyusefuldataun-

coveredduringthepreparationof this handbook
are thosewhichwere obtainedby
NorthAmericanRockwelli 6-4_in conjunctionwith the SaturnS-II developmentpro-
gram. Theseresults givethe yd values shown in Figure 6.2-3 which inehides two data
points from hemispheres and six data points from domes that were approximately
ellipsoidal. Reference (;-4 includes specimens whose membrane stresses at failure
ranged all the way from the elastic to the deeply plastic zones. In three cases it was
felt that these stresses were too high to permit the computation of reliable plasticity
reduction factors. Therefore, these particular data were discarded and they do not
appear in Figure 6.2-/t. Still another experimental point was discarded because of a
faulty edge condition in the test. In addition, as noted in Figure 6.2-3, two specimens
were subjected to a thernml gradient along with the external pressure. For each of
these domes, the inner facing was at roughly +280°F while the outer facing was at
apt)roximately +10 °F. This gradient was completely neglected in the analysis per-
formed to arrive at the related Yd values. Nevertheless, these results are retained
in Figure 6.2-3 since they fall within the scatter band displayed by the other speci-
mens having the same basic contour.
6-8
1°0
i L ..... Lt • I io,
LEGEND
• ttEMISPHERES AT ROOM TEMPERATURE
,, APPROXIMATE EI_LIPSOIDS AT ROOM
TEMPE RATURE
APPROXIMATE EI, LIPSOIDS WITH TItERMAI_
0,
--[]
GRADIENT
p = SHE]_I,-WAI,I, RADIUS OF GYRATION
(p _ h/2 FOR SANDWICtt CONSTRUCTIONS i [ ' '
WtIOSE TWO FACINGS ARr._EOF EQUAL THICKNESS) I ....

0o
• ' ' ' t i i i RECOMMENDED DESIGN. '

• ' ........ . _ IVAI,UE FOR EI,I_IPSOIDS] i
_d , . .... AND TORISt tlERICAI, +
..... '' _ '•: _.. 'liiDOMES '
0o
0._
. • : ...... i ...... RECOMMENDED DESIGN

' _' '. i _' _ '.. '. i• ' _[ _"i VALUE FOR HEMISPttERES
, r ' • " " •
, I I i
0 100 200 300 400 aO0
(->)
Figure 6.2-3. Knock-Down Factor _/d for Sandwich Domes

Subjected to Uniform External Pressure
6-9
To fully understandthe informationgivenin Figure6.2-3, it is importantfor the
readerto beawareofthe datareductiontechniquesemployedhere. For anexplanation
of theseprocedures,relbrencemaybemadeto the discussionin Section4.2.1.2.1.
Althoughthat sectionis concernedwith sandwichcylinders, the samebasicapproach

wasusedin analyzingthe domes.
Basedon Figure6.2-3, it is recommended

that, exceptwhereshearcrimping occurs,
the followingvaluesma3be usedfor Td:
)'d = 0.20 for hemispheres (6.2-13)
"Yd = 0.35 for ellipsoids and torispherical domes (6.2-14)
Insufficient data are available to discern any dependence of the knock-down factor on
the ratio RMax/P. ttowever it is quite possible that even a large array of data would
lead to the same conclusion. This would be consistent with the practice usually accepted
for isotropic (non-san&rich) domes.
It is thought that there is physical justification for the hse of a 7d value for hemispheres
which is lower than that for ellipsoids and torispherical domes. This justification lies
in the fact that, for the latter two configurations, the maximum membrane stresses
occur at the apex which is well-removed from the boundary disturbances. On the other
hand, the membrane s_resses in a hemisphere are uniform over the entire surface.
Discontinuity distortions at the boundaries are ignored in classical small-deflection
stability theory but, in reality, these deformations can act somewhat like initial im-
perfections and precipitate buckling. This fact, coupled with the uniform membrane
6-10
stress in the hemisphere,canlead to earlier failure thanwouldbe encounteredfor
shapeswherethe peakmembranestressesdonot extendinto the boundaryregions.
Sincethe recommended
valuesfor Tdare basedonmeagertest results, the method
proposedhereis not very reliable andcanonlybe regardedas a "best-available"
technique. It shouldonlybeusedas a roughguidelineandfinal designsmustbe
substantiated
by test.
6-11
For sandwichdomesof the typesshownin Figure6.1-1 andsubjectedto uniform
externalpressure, the critical apex stresses may be computed from the relationships
given in the equations on page (;-14 where the subscripts 1 and 2 refer to the separate
facings. There is no preference as to which facing is denoted by the subscript 1 or 2.
The equations on page 6-14 were obtained by a simple extension of the formulas pre-
sented in Section 6.2.1.1. The extension was accomplished in order to cover some
situations where the two facings arc not made of the same material. This was achieved
through the use of equivalent-thiekness concepts based on the ratios of the moduli for
the respective facings. For cases where the two facings are not made of the same
material, the resulting equations are wtlid only when the behavior is elastie (r_= 1).
Application to inelastic cases (7)/1) can only be made when both facings are made of
the same material. For such configxlrations, E_ and E_, will, of course, be equal.
The buckling coefficients KC can be obtained from I"ig_rc (;.2-4.
The knock-down factor Yd may be chosen as follows:
When V 2.0 When V _ 2.0

C
e
Use ?/d :: 0.20 for hemisl)heres , Use )Jd = 1.0 for hemispheres,
ellipsoids, and torl-
USe_d _ 0.35 forellips_dds and spherical domes.
torispherical domes.
The quantity RMa x is the maximum principal radius of curvature for the dome and is
measured in units of inches. For all of the shapes shown in Figure 6.1-1, this value
occurs at the apex.
6-12
t_
[ • .
.... i':T
.!
ttL!
T T
!!
6-13
The formulations given here are based on the assumption that the transverse shear
stiffness of the core is isotropic. However, in most practical sandwich constructions,
this stiffness will vary with direction. In order to apply the given criteria to such
structures, one must select a single effective G c value. Whenever the shear crimping
modc is critical (V c : 2.0), G c must be taken equal to the minimum value for the core.
In all other cases one must rely on engineering judgment in making an appropriate
selection.
The plasticity reduction factor should always be based on the stress at the apex of the
dome. For elastic cases, use r?= 1. Whenever the behavior is inelastic, the methods
of Section 9 must be employed.
Facing 1
Facing 2
Apex a = (i {6.2-15) Apex

er 1 Yd Ke ! oi (6.2-16)
cr_ =: Yd Kcr_o_
< .:
o" = r?E 1 C (6 2-17) cr :r/E C

01 0 " (6.2-18)
% o
h 2 _ (Eztl) (Estp)
C _
o R (6.2-19)
Max
5 2
= G (6.2-20)
h _
crimp_ G (6.2-21)
C
Its+ E(_(.)t_] tc c ta+t t
(7
_ O 1 (7
o
(7 (6.2-22) V =V
(7 (6.2-23)
crimp_ C 1 C_
crimp_
6-14
Thecritical pressurePer (in units of psi) maybe computedas follows:
2 t2] (6.2-24)
Pcr - RMax ECrcrl
tl +acr2
In the specialcasewheret1= t2_tf andbothfacingsare madeofthe samematerial,
Equations(6.2-15)through(6.2-24)canbe simplifiedto the following:

(6.2-25)
Apex _cr = _d Kc(Y o
(fiEf) h
(6.2-26)
(7
O
ffe_ RMax
5 2
- G (6.2-27)
Crcrimp 2 tft c c
(7
o (6.2-28)
V - ---
e (7
c rimp
4 (6.2-29)
Pcr - R ((_crtf)
Max
6.3 OTHER LOADING CONDITIONS
No information is available concerning the general instability of dome-shaped sandwich
shells under loading conditions other than that of uniform external pressure which is
covered in Section 6.2.
6-15
r
r_
i c_
_J C,
b-, " '_ "b L_
o o _'-go
c,i H
'_@ AI
p d-' l /I,
c,_
3 _D
o
m
o -_ o
Ph _:o
0 "_ ,_
d_
b P'- ._ _._
o
U +
g._
M
0 cq lI II
;,.I, '-'
<: H ,¢
-,,11--
--0 _q
6-15
L;
2,
o
'-c
II
,-,o
b _
d_
el
0 o
_._ _ ,_ It
b
II o o
:I
li
h
rd'' o
o
p- b o @
6-17
REFERENCES
6-1 Yao, J. C., "Bucklingof Sandwich

SphereUnderNormalPressure", Journal
of the AerospaceSciences,March 1962.
6-2 Plantema,F. J., SandwichConstruction, John Wiley & Sons, Inc., New York,
Copyright 1966.
6-3 Donnell, L. H., "Stability of Thin-Walled Tubes Under Torsion", NACA
Technical Report No. 479, 1934.
6-4 Gonzalez, H. M. and Patton, R. J., "Development and Fabrication of 55-Inch
Diameter Scale Model S-II Common Bulkheads", North American Rockwell,
Corp. Report No. SDL 468, May 1964.
(_-19
7
INSTA BILITY OF SANDWI CH S HE LL SE GME N TS
7.1 CYLINDRICAL CURVED PANELS
7.1.1 Axial Compression
It will be helpful here to first consider the case of axially compressed, isotropic
(non-sandwich) skin panels for which all four boundaries are simply supported. In
such cases, the Schapitz criterion _7-1] furnishes a practical means for the corn-
putation of critical stresses. This criterion accounts for the effects of skin-panel
geometry as the transition is made from wide panels, which behave essentially as
flfll cylinders, to narrow panels which approach the behavior of fiat plates. In par-
ticular, Schapitz proposed that one use the following relationships which have been
verified by the rederivation of Reference 7-2:
When
(7.1-1)
then
2
c_R
(l = U (7.1-2)
er p 4_
P
when
>2_ (7.1-3)
(YR P
then
U (7.1-4)
cr = crR
7-1
where,
(Y = Critical stress for buckling of a simply supported flat plate of the con-
P
figuration shown in Figure 7.1-1, psi.
(rfl Critical stress (in units of psi) for buckling of a simply supported complete
cylinder of radius R, length aR, and thickness t R (see Figure 7.1-1). The
quantities R, aiR, and t R are all measured in units of inches. An empirical
knock-down factor should be incorporated here to account for the detri-
mental effects from initial imperfections.
(Y, psi / a, psi
a
p
g, psi
t
P
a =a _a
R p
R
b =b ::b
R p
t =t t
R p
Figure 7.1-1. Cylindrical Panel and Associated Flat-Plate Configuration
7-2
b. For sandwichpanelswhichfall in the moderately-stiffor weak-corecate-
gories, gcr shouldbetakenequalto thehigher of the two valuesut)and(_R"
In the courseof preparingthis handbook,noanalysiswasmadeof test datafrom sand-
wich panels. Therefore,the reliability of this approachhasnot beenestablished,and,
until ex_perimental
substantiationis obtained,onecanonlyregardthe methodas a "best-
available"teebxlique.
In view of the lack of sandwichdatacomparisons,it is informativeto notethata large
collectionof test results from isotropic (non-sandwich)

specimensis evaluatedin
Reference7-3 andit is shownthere thatthe Sehapitzcriterion is a reliable approach
for suchpanels. Thetest configurationsembraceda wide rangeof , -_ , and
ratios. Narrow, wide, and intermediate panels were included. The K values fell
between those for the ease where all four boundaries are simply supported and the case
where all four boundaries are fully clamped. The results are summarized in the qual-
itative presentation of Figure 7.1-3. This figure shows the general characteristics
and relative positioning for each of the following when displayed in a nondimensional
logarithmic format :
a. The theoretical buckling relationship for fiat plates.
b. The classical, small-deflection, theoretical buckling relationship for

complete cylinders.
c. A lower-bound buckling relationship for complete cylinders. This is

obtained by multiplying the values from b; above, by the empirical
knock-down factor of Reference 7-4.
d. The design curve based on the Sehapitz criterion.
7-5
\
\
, * _ O"
er t
• .\\
% \
'\
De sign Cum'e Based
on Schapitz Criterion
Figure 7.1-3. Schematic Logarithmic Plot of Test Data for Cylindrical Isotropic
(Non-Sanchvich) Skin Panels Under Axial Compression
7-6
Althoughderivedspecificallyfor the easeof simple support,this criterion hasbeen
successfullyemployed[7-3_ wheretheboundariesprovidevariousdegreesof rotational
restraint alongwith the conditionof no radial displacement. This wasaccomplishedby
simply adjustingthe val_,efor ap to correspondwith the appropriateedgerestraints.
For the case under immediate discussion (non-sandwich skin panels), the Schapitz
criterion can be graphically represented as shown in Figure 7.1-2. A series of design
curves of this type are given in Reference 7-3. The transition curve defined by Equation
(7.1-2) becomes tangent to the full-cylinder curve when gR = 2gp. For (R/t) values
greater than that of the tangency point, the skin panel behaves as a complete cylinder.
For all other (R/t) values, the transitional relationship applies. Note that the transition
curve asymptotically approaches the line for Crp. The quantity K denoted in Figure 7.1-2
(b) = C°nstat_
Figure 7.1-2. Schematic Logarithmic Plot of Schapitz Criterion

for Non-Sandwich Cylindrical Skin Panels
7-3
is the conventionallist-plate bucldingcoefficientwhichis oh,pendent upon the aspect
ratio (a/b), 1)oundary c(mditions, and t33)e of loading. Fr,.m_ this fig-ere, it can 1)e
seen that, if the critical stress weco taken equal to the higl_er of the two values cr
P
and,7 R, one would only be neglecting the tnulsitional strenglh associated with the
cross-hatched region. When (_-_R / =: 1, neglect of this ccmtribution would restllt in a

\ P/
design value which is sl) pc, rccnt of the Schapitz prediction. For all ()/her values of the
ratio {_f{ /, the differences would be less sig_,ificanl. Indeed, for most ranges oi(;tl t
\Crp/ \ P,
the conservatism introduced hy neglecting the cross-hatched area would be quite small.
Since lhe Schapitz criterion is: dependent solely on lhe values (71) and cr R, the speculation
is nmde here that on(, might extend its application to cylindrical sandwich panels merely
by computing crp and cJt{ from the sandwich design equations and curves which are pro-
vided in Sections :; al?d .I. tfowever, in making such an extension, one inust recognize
that the behavior of a sandwich panel is dependent upon the (:ore stiffness. For stiff-
core constructions (see So.ellen 4.2), it should be possible to makt. direct application
of Equations (7.1-1) lhr_ugh {7.1-.t). On the other hand, in the weak-core region, the
sandwich panel will lail by shear crimping, and curvature will not contribute to the
buckling strength, in such cases, Equations (7.1-1) through (7.1-4) would yield uneon-
servative predictions. The situation for sandwich constructions having moderately-
stiff cores would, of course, fall somewhere between the foregoing limiting eases.
Consequently it is recommended here that,
a. For stiff-core sandwich panels, Equations (7.1-1) through (7.1-4) can be

applied.
7-4
Also sh(_vn in Figure 7.1-3 are the approximate locations of the test data fr()m the
non-sandwich cylindrical panels of References 7-5 through 7-8. During the course
of the study reported in Reference 7-3, quantitative plots were made for each of these
specimens and the corresponding test points were accurately located on the appropriate
graph. Based on these many different plots, the test points were inserted in Figure
7.1-3 in approximation to their actual positions relative to the several basic curves
and regions of behavior. This figure shows that all but four of the test points which
fall below the design curve lie within the region where the panel beltaves essentially
as a flat plate. Except for those four points, all of the test data for the regions of
transitional and full-cylinder behavior fall between the following t_vo bounds:
a. The recommended design curve.
b. The values which would have been predicted if _R did not incorporate
an empirical knock-down factor.
It is concluded that Figure 7.1-3 verifies the reliability of the Schapitz criterion
for the case of isotropic (non-san&rich) skin panels, even where the boundary con-
ditions include some rotational restraint in addition to the requirement of no radial
displacement. This conclusion is based partly on the fact that the character of flat-
plate buckling is quite different from that exhibited by wide cylindrical panels and
complete cylinders. The fiat plate can continue to support steadily increasing in-
plane loading well into the postbuckling region. This is in contrast to the sudden
drop-off in load usually observed for wide panels and full cylinders. Consequently
the Schapitz criterion utilizes full theoretical predictions as the limiting case of a
7-7
flat plate is approached. One might, therefore, expecl that within this region test data
will display some small degFee of scatter on both sides of th(, design curve, ttowever,
bucause of tile physical behavior cited above, this generally will not lead to any seFious
st ructur,,tl de fieieneies.
7-8
For cylindrical sandwichpanelssubjectedto axial compression,the critical stress
maybe computedfrom the following:

Weak-Core and Moderately-Stiff-
Core Constructions
Stiff-Core Constructions
When The higher of the two] (7.1-9)

(7. i-5) (y
cr = |values _ and gR !
[ P J
then
(_R
(3" = (7.1-6)
+ 4--_-- ' and
cr p
P
when
(7.1-7)
then
U = U (7.1-8)
cr R
whe re,
= Critical axial compressive stress (in units of psi) for the buckling of a
u
P flat sandwich plate which has the same boundary conditions as the cylindri-
cal panel and, except for curvature, is of the same geometry as the cylin-
drical panel (see Figure 7.1-1). No knock-down factor is required in com-
puting this value.
= Critical axial compressive stress (in units of psi) for the buckling of a
(;R
complete sandwich cylinder which, except for the circumferential dimen-
sion, is identical to the curved panel. An appropriate empirical knock-
down factor should be incorporated here to account for the detrimental
effects from initial imperfections.
As a rule-of-thumb, one may assume that stiff-core constructions are those which
satisfy the inequality
V <0.25 (7. i-i0)

C
where V c is computed as specified in Section 4.2.
7-9
The quantity _p should be computed by using the design equations and curves given in
Section 3.
The quantity (_R should be computed by using the design equations and curves given in
Section 4.
A graphical representation of Equations (7.1-5) through (7.1-8) is provided in Figure
7.1-4.
The method given here applies only where all four boundaries are completely restrained
against radial displacement. Therefore, no free edges are permitted. Any or all of
the four boundaries may include rotational restraint of any degree ranging all the way
from a hinged condition to fully clamped.
4.0
+ 4_-
-T.H
}--p_
3.0 4.0
(ffR/(_p)
Figure 7.1-4.
Graphical Representation of Equations (7.1-5) through (7.1-8)
7-10
7.1.2 OtherLoadingConditions
7.].2.1 BasicPrinciples
In the preparationof this handbook,almostno considerationwasgivento the buckling
of cylindrical sandwichpanelssubjectedto loadingsotherthanaxial compression.
Therefore, nofirm recommendations

canbe madehere concerningdesignequations
andcurves, llowever, the suggestionis offeredthat, for suchcases, onemight con-
sider an extensionof the conceptspresentedin Section7.1.1. Inparticular, for all
regionsof core stiffness, it mightbepossibleto applythe equation
[The higherof the two] (7.1-11)

Crcr = [values Crp and (_R J
if one simply computes the values _p and o R for the loading condition of interest.
In conformance with the restrictions of Section 7.1.1, the foregoing suggestion applies
only when all four boundaries of the panel are completely restrained against radial dis-
placement. Therefore, no free edges are permitted. Any or all of the four boundaries
may include rotational restraint of any degree ranging all the way from a hinged con-
dition to fully clamped.
No rcconm]endations are made here.
7.2 OTHER PANEL CONFIGURATIONS
No information is available concerning the instability of sandwich shell segments of
shapes other than the cylindrical configurations considered in Section 7.1.
7-11
©
b b
II
t, b
-<
0
Z
7-12
_L _
_ °
c_
q_
_Q
m_
r!
_o
7-13
REFERENCES
7-1 Schapitz, E., Festigkeitslehre f_r den Leichtbau, 2 Aufl., VDI-Verlag GmbtI
D_isseldorf, Copyright 1963.
7-2 Spier, E. E. and Smith, G. W., "The Sehapitz Criterion for the Elaslic Buckling
of Isotropie Simply Supported Cylindrical Skin Panels Subjected to Axial Com-
ln'ession", Contract NAS8-11181, General Dynamic Convair division Memo AS-
D-1029, 31 January 1967.
7-3 Smith, G. W., Spier, E. E., and Fossum, L. S., "The Stability of Eccentri-
cally Stiffened Circular Cylinders, Volume II - Buckling of Curved Isotropic
Skin Panels; Axial Compression", Contract NAS8-11181, General Dynamics
Convair division Report No. GDC-DDG-67-006, 20 June 1967.
7-4 Seide, P., Weingarten, V. I., and Morgan, E. J., "Final Report on the Devel-
opment of Design Criteria for Elastic Stability of Thin Shell Structures", STL-
TR-60-0000-19425, 31 December 1960.
7-5 Petcrson, J. P., Whitley, R. O., and Deaton, J. W., "Structural Behavior
and Compressive Strength of Circular Cylinders with Longitudinal Stiffening",
TN-D-1251, May 1962.
7-6 Cox, It. L. and Clenshaw, W. J., "Compression Tests on Curved Piates of
Thin Sheet Duralumin", British A.R.C. Technical Report R&M No. 1894,
November 1941.
7-7 Crate, It. and Levin, L. R., "Data on Buckling Strength of Curved Sheet in
Compression", NACA WR L-557, 1943.
7-8 Jackson, K. B. and Hall, A. H., "Curved Plates in Compression", National
Research Council of Canada Aeronautical Report AR-1, 1947.
7-15
8
EFFECTS OF CUTOUTS ON TttE GENERAL INSTABILITY OF
SAND_\rICtt SttELLS
In many practical aerospace shell structures, it is required that cutouts be incorporated
for purposes of access, lightening, venting, etc. However, no theoretical solutions or
experimental data have been published [or the general instability of sandwich shells
having such penetrations. Even in the case of isotropic (non-sandwich) shell struc-
rares, this problem tins received little attention. Some theoretical solutions have been
accomplished concerning the stress distributions around cutouts in isotropic shells
but the authors of this handbook are aware of only one paper (8-1) dealing with the
general instability problem, and this paper is not sufficiently comprehensive to pro
vide a practical design criterion.
An obvious need exists for further theoretical and experimental work to be accomplished
in this area, and, in view of this situation, no related design recommendations can be
made at the present tirne.
REFERENCE
8-1 Snyder, R. E. , "An Experimental Investigation of the Effect of Circular Cutouts
on the Buckling Strength of Circular Cylindrical Shells Loaded in Axial Com-
pression," Thesis submitted to the Graduate Faculty of the Virginia Polyteclmic
Institute in candidacy for the degTee of blaster of Science in Engineering Mechan-
ics, May 1965.

8-i
9
INELASTIC BEHAVIOR OF SANDWICH PLATES AND SItE_LLS
9.1 SINGLE LOADING CONDITIONS
For structural members stressed beyond the proportional limit of the material, it is
customary to compute critical loads or stresses through the use of so-called plasticity
reduction factors. In this handbook, such factors are denoted by the symbol r7 • In
many cases, appropriate formulas for _7 are established by theoretical derivations
based on plasticity theory but, when this approach proves impractical, one must some-
times resort to empirical expressions. Section 9.1.2 gives the formulations for
which are recommended in this handbook for various sandwich configurations, types
of loading, and modes of instability. These equations are based on the information
provided in References 9-1 through 9-5 for isotropic (non-sandwich) plates and shells.
Application of these reduction factors involves the trial-and-error procedure outlined
below:
a. First, assume 7) = 1 and compute the critical stress for the appropriate
configuration, loading condition, and mode of failure.
b. If the critical stress computed in a, above, is less than the proportional

limit of the facing material, no further computations are required, ttow-
ever, if the computed critical stress exceeds the proportional limit, one
must continue as specified below.
c.
Assume a new value for the critical stress which is in excess of the pro-
portional limit but less than the value computed in a, above.
9-1
dp
Based on the stress level assumed in c, above, and the stress-strain curve
for the facing material, compute a value for the appropriate plasticity re-
duction factor. The formulas of Tables 9. i-1 through 9.1-3 can be used
1ol- this purpose.
_° Using lh(. r_ value computed in d above, recalculate the critical stress.
f. If the critical stress calculated in e, above, is in reasonable agreement

with the value assumed in e, no further computations are required. How-
ever, if such agreement is not aehieved, one must then repeat the eompu-
ration cycle starting with e. This iterative procedure must be continued
until acceptable agr¢_ement is attained between the assumed and the ealeu-
lated critical stlvsses.
A numerical example of the fc,regoing procedure is provided in Section 9.1.2.
9-2
9.1.2 DesignEquations
Recommended
formulasfor plasticity reductionfactors are givenin Tables9.1-1
through9.1-3 where
Ef = CompressiveYoung'smodulusof facings,psi.
E = Compressivesecantmodulusof facings, psi.
s
E = Compressivetangentmodulusof facings, psi.
t
Gf = Elastic shearmodulusof facings,psi.

G = Secantshearmodulusof facings, psi.
s
v = Elastic Poisson'sratio of facings,dimensionless.
e
v - Actual Poisson's ratio of facings, dimensionless.
Values for v can be obtained by using
v = 0.50- (0_. 50 - re) (9.1-1)
or
(9.1-2)
= o. o -\ El/
The technique for applying the plasticity reduction factors is demonstrated below by
means of a numerical example for an axially compressed sandwich cylinder which is
It is further
assumed to be of sufficient length to fall outside the short-cylinder range.
assumed that
al both facings are of the same thickness,
b. both facings are made of the same material, and
e. the transverse shear properties of the core are isotropic so that
0= (Gxz/Gyz) = 1.
9-3
For such cylinders, Section 4.2.2 specifies that the critical stress for general in-
stabilit}, may be computed [rom
(Tcr = Yc Kc(Yo (9.1-3)
where
(71E f) h
cr
o (9.1-4)
_/c is obtained from Figure 4.2-8. K e is obtained from Figure 4.2-7 where
Gr
o
V - (9.1-5)
c (Ycrimp
and
5 2
_crimp - 2tft c Gxz (9.1-6)
For the purposes of the present sample problem, assume that
Ef = i0 ×106 psi
v = 0.30
e
R = :_2.0"
h = .320"
tf = .020"
t = .300"
C
G = 20,000 psi
XZ
h
f) - 2 . 160"
R 32.0"
- - 200
p .160"
Facing Proportional Limit = 25,000 psi
9-4
By using these values and assuming that _ = 1, it is found that
7c = 0.49
cr = 104,900
0
= 170,800
crimp
V = 104,900/170,800 = .614
C
K = 0.85
C
The re fore,
gcr 7cKcCro = 49 x.85x104,900 43,600
Note that the computed critical stress (43,600 psi) is higher than the proportional limit
(25,000 psi) of the facings. Hence the use of _ = 1 cannot be valid and one must now
proceed on a trial-and-error basis. That is, one must select an assumed critical
stress value which exceeds the proportional limit. For the purposes of this sample
problem, suppose that the value (_cr = 30,000 is selected. By using the stress-strain
curve for the facing material, the corresponding plasticity reduction factor can then
be computed from the following formula which is taken from Table 9.1-3:
(9.1-7)
1--=71 Ef
Suppose that this gives the result that
= 0. 900
so that one now obtains
7c = 0.49 (remains unchanged)
= .900 ×104,900 = 94,400

O
9-5
g = 170,800 (remains unchanged)
erimt)
V = 9,t, !t00/170,800 = .553

C
K = 0.8G
e
There fore,
(rcr =YcKc cro = .49 ×.86 ×94,400 = :_9,800
Note that the eon_puted critical stress (39,800 psi) does not agree very closely with
the assumed value 1:_0,000 psi). Therefore, another iteration will be performed by
selecting a new assumed critical stress, say 35,000 psi. Suppose that by using
Equation (9.1-7) the corresponding plasticity reduction factor is found to be
r? = 0. 790
so that one now obtains
:/c = 0.,i9 (remains unchanged)
(Yo = "790 ×104,900 = 82,900
cr = 170,800 (remains unchanged)

crimp
V = 82,900/170,800 = .486
C
K = 0.87
C
"rile re fore,
cr = ,/cKcffo = .49 ×.87 ×82,900 = 35,400
Note that the computed critical stress (35,400 psi) is now in reasonable agreement
with the assumed value (35,000 psi). Therefore, no further iterations are required
and the design value for the critical stress is 35,000 psi.
9-6
=o._
"_ _ 0
_2J
L,-_ _ j
M
r_ _--
_ o
.,..4
[ i o
0
0
0
O_
..-i +
0
¢) _ 0
_o ,--_I cq 0
+
o I
,-_ I c"q
|
0
U9
r_
0
I.-..4
0
0
I I
0 I J
0 II
0 r 1
_'_ I_ _ _L
r_ 0 +
•r_ 3
'tJ
c_ +
0
I
0 o _
r..) _
0
0 r--¢
0
0
b_
I l .a
li
I
_b
N
r..)
t:_
b_
r_
,-_
0
0 [-,
©
2;
9-7
r
_I_ c'o I '._-t._
_1'_
+ IJ
_1_1
L J
_1_
f ]
i
I
1
L J L""* J
][
*3
©
r_ p q3
l i[
0 Ir
ir
,....,
c_
E_
9-8
Table 9-3. Recommended plasticity Reduction Factors for the General Instability
of Circular Sandwich Cylinders, Truncated Circular Sandwich Cones,
and Axisymmctric Sandwich Domes
PlasiicJly Heducliol_ i.'zLclol's
Mo(h:ralc l,englh
Short (:ylinders Through l,ong ('ylinders
_tnd Cones ail(I (:(rues
I
............... 1 - Lit_ Et I -u(:' _
] ,:: L,-:V ] , : L,_-77] Ef
External Laieral I)rcssure _ rl- ll_--_J Ef _4 + ,I Es)
T ors ion
Transverse Shear
ile, mispherical, Ellipsoidal, and
Torispherical Domes (All trun-

cated at the equator)
External Pressure
*This formula for r l is not valid when the cylinder or cone is so short that it
behaves essentially as a long, flat plate. However, it is unlikely that such
configuralions will be encountered in aerospace applications. Furthermore,

it is informative to note that, for such constructions, the given formula for
rl is conservative.
**This formula for rl is not valid when the cylinder or cone is so short that it
behaves essentially as a long, flat plale, ttowever, it is unlikely that such
configurations will he encountered in aerospace applications. Furthermore,

it is informative to note that, for such constructions, the given formul.a is
approximately 13-1)ercent unconscrvative.
9-9
9.2 COMBINED LOADING CONDITIONS
As noted in llefe_'cnce 9-(;, only limited ilfformation is available on the inelastic sta-
l)ility of shell structures subjected to comt)ined loading conditions. A similar situation
exists for flat-plate constructions. Very little theoretical work has been done in these
fields due to the complexity of the problem and, in general, related phtsticity reduc-
tion criteria have not been esklblished, t[owever, in many practical engineering
applications, one is co_ffronted with this type of problem and it becomes necessary to
determine at least a rough estimate of the critical loading conditions. Toward this
end, one should note a fundamental hypothesis of plasticity theory which specifies that,
for a given material and when the stress intensity is increasing (loading condition), the
stress intensity cri is a tmiquely defined, single-valued ftmction of the strain intensity,
e i. When cri is decreasing (unloading condition), the relationship between cr i and e i
is linear as in a purely elastic case. Based on the oetahedral shear law for plane
stress conditions, the stress and strain intensities gi and c i can be defined as follows
[ 9-]] :
,. _<, a:-, _ +3r 2 (9.2-1)

1 3' xy
ei ,/,_ ¢2 _ ¢2 + ¢ ¢ +'¢:_ /4 (9.2-2)

x y x y xy
It sJ_o'_ld be noted that Eq'_tation (9.2-1) is sometimes written in ti,c following form to
facilitate its use:
[
cri (Crx)y I - 7 + )Z + 3),:: (9.2-1a)
9-10
where
cr - Normal stress in the x direction, psi.

x
Normal stress in the y dh'ection, psi.

Y
_- : Shear stress in the xy plane, psi.
- Normal strain in the x direction, in/in.

x
£ = Normal strain in the y direction, in/in.

Y
E = Shear strain in the xy plane, in/in.

xy
y x
From the foregoing discussion it can be concluded that, for the case of increasing
cy (loading condition), the relationship between (yi and e i is identical to the conven-
i
tional stress-strain curve obtained from a uniaxial loading test. It should therefore
be evident that although each individual stress component may be less than the propor-
tional limit of the material, the combination of these stresses can give a ffi value which
lies above the proportional limit so that the behavior is actually inelastic. It is im-
portant to keep this phenomenon in mind when deciding whether or not plasticity effects
must be considered.
Lacking a rigorous approach to the subject stability problem, it is conjectured here
that the foregoing generalization of the stress-strain relationship might be used in
conjunction with the plasticity reduction factor
Et
(9.2-3)
9-11
to obtain conservative predictions of inelastic instabilit%_ under combined loadings. The
quantities E t and Ef are as h)llows:
Et Tangent modulus of facing material obtained from the curve of

(_i vs e i at a prescribed value of gi' psi.
Ef Young's modulus of facing material, psi.
The above formula for _ was selected in view of its conservative nature. Since the
overall procedure suggested here is based purely on an engineering estimation, it is
thought that tttis conservatism is well justified.
The details of the suggested approach are outlined in Section 9.2.2. It is important
to keep in mind that this method does not give a rigorous solution, and its reliability
has not been evaluated by comparisons against test data. Therefore, this can only be
regarded as a "best-available" technique and one should be cautious in its application.
9-12
9.2.2 Suggested Method
The method suggested here for analysis of the inelastic stability of sandwich plates
and shells first requires that the conventional stress-strain curve for the facing ma-
terial have the stress coordinates relabeled as cYi and the strain coordinates relabeled
as e. By completely ignoring all plasticity considerations (_ = 1), one should then

1
proceed to establish a first-estimate for the critical combined stress condition. This
can be achieved by using the appropriate interaction relationships provided in earlier
sections of tiffs handbook. In performing tiffs computation, the assumption should be
made that for the critical combined stress condition the individual stress components
are in the same ratios to each other as exist for the actual applied loading condition.
That is, during loading, proportionality between the several individual stress compo-
nents is maintained. The stresses from the elastic first-estimate computation must
then be inserted into the equation
cYi = 4a; + (y2 _ cy a + 3-r 2 (9.2-4)

y xy
to determine the associated stress intensity value. If this value does not exceed the
proportio_ml limit of the cYi versus e i curve, the first-estimate stress values are in
fact the critical combination. However, if the related a i value exceeds the propor-
tional limit of the cYi versus e. curve, the first-estimate results are not valid and one
must then resort to the following trial-and-error procedure which is similar to that
outlined in Section 9.1:
a. Assume a new value for cri which is in excess of the proportional limit
for the (Yi versus e i curve.
9-13
b °
For the cri value assumed in a, above, compute the plasticity reduction
factor
(9.2-5)
who re
Et Tangent modulus of the cri versus e i curve, psi.
Ef Elastic modulus of the gi versus e i curve, psi.
co Using the 7?, value from b, above, recalculate the critical stress intensity
_.1 This is accomplished by simply multiplying the first-estimate a i value

by _.
do
If the new value for cri computed in c, above, is in reasonable agreement
with the c_i value assumed in a, above, the related plasticity reduction
factor rl is valid. Then the critical combination of stresses is obtained by
multiplying each of the first-estimate stress components by this rl value.
If the vMue of cyi computed in c, above, is not in reasonable agreement

with the cri value assumed in a, the related plasticity reduction factor is
not valid. One must then repeat the computation cycle starting with a.
This itcratlve procedure must be continued until acceptable agreement is
attained between the assumed and the computed or. values.
1
9 -14
REFERENCES
9-1 Gerard, G., "Plastic Stability Theory of Thin Shells," Journal of the Aeronautical
Sciences, April 1957.
9-2 Gerard, G. and Becker, tt., "Handbook of Structural Stability, Part I - Buckling
of Flat Plates," NACA Technical Note 3781, July 1957.
9-3 Gerard, G. and Becket, H., "Handbook of Structural Stability, Part III - Buckl-
ing of Curved PIates and Shells," NACA Technical Note 3783, August 1957.
9-4 Gerard, G., "Compressive and Torsional Buelding of Thin-Wall Cylinders in
Yield Region," NACA Technical Note 3726, August 1956.
9-5 Gerard, G., Introduction to Strueturai Stability Theory, McGraw-Hill Book
Company, Inc., 1962.
9-6 Baker, E. H., Cappelli, A. P., Kovalevsky, L., Rish, F. L., and Verette,
R. M., "Shell Analysis Manual," Prepared by North Ameriean Roe_vell, Corp.
for the National Aeronautics and Space Administration, Manned Spaceeraft
Center, Itouston, Texas, June 1966.
NASA-I.aIt!:I"v, 1969 -- 32 CR-!45_ 9-15

NATIONAL
AERONAUTICS AND SPACE ADMINISTI_ATION _
WASHINGTON, D.C. 20546

OFFICIAL BUS{NESS FIRST CLASS MAIL
POSTAGE AND FEES PAID ,
NATIONAL AERONAUTICS AN
SPACE ADMINISTRATION
If Undeliverable ( Section 158

POSTMASTER:
Postal Manual) Do Not Rettm-
"'The aeronautical and space activities of the United States rhall be

conducted so as to contribute . , to the expansion of human knowl-
edge of phenomen,t in the atmosphere and space. The Administration
shall provide ;t_r the widest practicahle and appropriate dissemhtation
o/ information concerninq its activities and the result_ thereof."
--NATIONAL AERONAUTICS AND SPACE ACT OF 1958
NASA SCIENTIFIC AND TECHNICAL PUBLICATIONS
TECHNICAL REPORTS: Scientific and TECHNICAl. TRANSLATIONS: Information

technical information considered important, published in a foreign language considered
complete, and a lasting contribution to existing to merit NASA distribution in English.
knowledge.
SPECIAl. PUBLICATIONS: Information
TECHNICAL NOTES: Information less broad derived from or of value to NASA activities.
in scope but nevertheless of importance as a Publications include conference proceedings,
contribution to existing knowledge. monographs, data compilations, handbooks,
sourcebooks, and special bibliographies.
TECHNICAL MEMORANDUMS:
Information receiving limited distribution TECHNOLOGY UTILIZATION
because of preliminary data, security classifica- PUBLICATIONS: Information on technology
tion, or other reasons. used by NASA that may be of particular
interest in commercial and other non-aerospace
CONTRACTOR REPORTS: Scientific and
applications. Publications include Tech Briefs,
technical information generated under a NASA
Technology Utilization Reports and Notes,
contract or grant and considered an important and Teehnolog, 5, Surveys.
contribution to existing knmvledge.
Details on the availability of these publications may be obtained from:
SCIENTIFIC-AND TECHNICAL INFORMATION DIVISION
NATIONAL AERONAUTICS AND SPACE ADMINISTRATION

Washington,D.C. 20546 ::
.... _ z

Nasa - cr-1457 Manual For Structural Stability Analysis of Sandwitch Plates and Shells

Uploaded by

Copyright:

Available Formats

Nasa - cr-1457 Manual For Structural Stability Analysis of Sandwitch Plates and Shells

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Nasa - cr-1457 Manual For Structural Stability Analysis of Sandwitch Plates and Shells

Uploaded by

Copyright:

Available Formats

https://ntrs.nasa.gov/search.jsp?

CONTRACTOR NASA CR-1457

MANUAL FOR STRUCTURAL

I SANDWICH PLATES AND SHELLS

I by R. T. Sulli,is, G. IFC Smith, a,id E. E. Spier

_ GENERAL DYNAMICS CORPORATION

_:San Diego, Calif.

E NATIONAL AERONAUTICS AND SPACE ADMINISTRATION • WASHINGTON, D. C. • DECEMBER-1969

_____ ". _ -........ =..... -- •!

MANUAL FOR STRUCTURAL STABILITY ANALYSIS

OF SANDWICH PLATES AND SHELLS

By R. T. Sullins, G. W. Smith, and E. E. Spier

Distribution of this report is provided in the interest of

Prepared under Contract No. NAS 9-8244 by

for Manned Spacecraft Center

NATIONAL AERONAUTICS AND SPACE ADMINISTRATION

/ ttONEYCOMB..-'_'III Iil tli I _ //" //

HONEYCOMB SANDWICH CONSTRUCTION

The following subjects are among those covered in the manual:

General Instability of Flat Panels

General Instability of Circular Cylinders

General Instability of Truncated Circular Cones

General Instability of Dome-Shaped Shells

Instability of Sandwich Shell Segments

Effects of Cutouts on the General Instability of Sandwich Shells

Inelastic Behavior of Sandwich Plates and Shells

INTRODUCTION ................. I-i

i. 1 GE 2:ERA_" ................. i-i

I. 2 FAILURE MODES ............. 1-4

LOCAL INSTABILITY ............... 2-1

2.1 INTRACELLULAR BUCKLING (Face Dimpling) 2-1

2. i. 1 Sandwich with Honeycomb Core ......... 2-1

2.1.2 Sandwich with Corrugated Core ......... 2-8

2.2 FACE WRINKLING ............. 2-21

2.2.1 Sandwich with Solid or Foam Core (Antisymmetric

2.2.2 Sandwich with Honeycomb Core (Symmetric

2.3 SHEAR CRIMPING ............. 2-37

2.3.1 Basic Principles ............. 2-37

2.3.2 Design Equations ............. 2 -40

GENERAL INSTABILITY OF FLAT PANELS ....... 3-1

3.1 RECTANGULAR PLATES .......... 3-1

3.1 1 General ................ 3-1

3.1 2 ITniaxial Edgewise Compression ........ 3-5

3.1 3 Edgewise Shear .............. 3-25

3.1 4 Edgewise Bending Moment .......... 3-37

3.1 5 Other Single Loading Conditions ........ 3-46

3.1 6 Combined Loading Conditions ......... 3-47

3.2 CIRCULAR PLATES ............ 3-73

3.2.1 Available Single Loading Conditions ....... 3-73

3.2.2 Available Combined Loading Conditions ...... 3-73

3.3 PLATES WITH CUTOUTS .......... 3-74

3.3.1 Framed Cutouts .............. 3-74

3.3.2 Unframed Cutouts ............. 3-75

4 GENERAL INSTABILITY OF CIRCULAR CYLINDERS .... 4-1

4.1 GENERAL ................ 4-1

4.2 AXIAL COMPRESSION ........... 4-5

4.2.1 Basic Principles ............. 4-5

4.2.2 Design Equations and Curves ......... 4-17

4.3 PURE BENDING .............. 4-25