Shell Analysis Manual
Shell Analysis Manual
Shell Analysis Manual
R TEPORT
LonJ
Prehayr't! b,
For sale by the Clearinghouse for Federal Scientific and Technical Information
Springfield, Virginia 22151 - CFSTI price $3.00
FOREWORD
Center, Houston, Texas. Mr. Herbert C. Kavanaugh, Jr., was the NASA
The program was performed between May 1965 and June 1966.
iii
publications. Particular credit is given to the following publishers
technical assistance.
NOTE
Texas 77058.
iv
ABSTRACT
from concepts of the linear theory of elasticity and includes the basic
V
development is based on variational principles and the concept of
safety calculation under uniaxial and biaxial loading conditions are also
presented.
methods for obtaining the design allowable buckling loads for unstiffened
loading conditions. Also included are procedures for the stability analysis
for inelastic buckling and combined loading conditions are also presented.
vi
Chapter 4. 00 presents methods of analysis to be
vii
CONTENTS
Page
NOMENCLATURE . . . . . . . . . xxv
1. 01 GENERAL . .. * 0 1i
1. 10 LINEAR SHELL THEORY . . . . . 3
1.11 Introduction . . . . 3
Thin Shells . . . . . . 6
Shell . . . . . . . 25
Theory . . . . . . 37
Theories . . . . . . 54
ix
Page
of Revolution . . . . 65
1.13.5 Membrane Theory of Shells .
1. 21 Introduction . . . . . . 93
1. 22 General Strain-Displacement Relations 95
Resultants . . . . . . 100
1. 24 Principle of Potential Energy . , 102
1. 7 Summary . . . . . . 118
1. 31 Introduction . . . 119
x
Page
1. 36 Discrepancy Between Theoretical and
Experimental Results ..... 142
General . . . . . . . . . 166
Structures . . . . . 174
Shells . . . . . . 178
xi
Page
Theory . .. . 179
Geometries . . . . . . . 182
Geometry . . . . . 182
Elements . . . . . . 183
2. 24 Conclusion . . . . . . . 193
Forces . . . . . 195
Displacements . . . . 198
xii
Page
Discussion . . . . . 246
xiii
Page
2. 41 Introduction . 296
continuities in Geometry or
Loading . . . . . . 521
Cylinders .4 33
xiv
Page
2.53.3 Conclusion • . . .. 38
2.61 Introduction . . . . . . .4
xv
Page
Elements . . . . .. 415
xvi
I IIi
Page
2. 65 Conclusion. a449
. . . . .
2.71 Introduction . . . . . . 4
450
Shell . . . . . . . 469
2. 74 Summary . .a . . 482
xvii
Page
2. 81 General . . 483..
2. 8Z Definitions 4185
(Mariotte) . .187 4
(Beltramni) * *189 4
REFERENCES * . .505*** * . o
Xviii
Page
3. 10 GENERAL . . .. 508
3.20 UNSTIFFENED SHELLS . 513
Panels . . . . . . . 514
Cylinders . . . . . . 532
Cylinders . . . . . . 540
3. 23.5 Combined Loading, Unstiffened
Cylinders . . . . . . 544
xix
Page
Cones . . . . . . .. 550
C ones . Cones564
. .. . . .
. 575
3. 32. 2 Orthotropic Layered Shells . .
xx
Page
Cylinders . . . . . 588
Cones . . . . . . 610
xxi
Page
Cylinders . . .. 637.
3. 50 SANDWICH SHELLS . . . .. 647
Cylinders . . . . . . 6571
Cylinders . . . . . . 661,
xxii
Page
Bending .. . . . . . 743
4.31 General . . . . . . . 7
REFERENCES . . 0 750-
UTILIZATION o o 0 0 757
xxiii
Page
TECHNIQUES . . . . . . . 768
xxiv
NOMENCLATURE
Af Frame area
Ax e, A
Et
B Extensional stiffness (rigidity),- for homo-
bw Web depth
xxv
C Rigidity Et ; buckling coefficient for stiffened
4(1 - V )
cylinder; plasticity curve
(Section 3. 32)
(Paragraph 3.42. 2)
3
Et
D Bending stiffness (rigidity) for homo-
xxvi
.D Dxe/Dx +4
(Section 3.32)
equilibrium configuration
,1' e, '
e12 equilibrium position
xxvii
elz, ezz Non-linear'transverse shear strains
curve
G0 B( 12Gx8 e) "F
(Section 3. 50)
I Moment of inertia
xxviii
le Effective moment of inertia
X0,
number
K bp K C K p K8 Buckling coefficient
V'Rt
xxix
k2 Negative of slope of Cc versus Vc curve
k~w
2 Height of cone
xxx
M 10Bending moment per unit length acting at section
xxxi
m Number of layers or parts
for sphere
sphere
xxxii
N2 Inplane force per unit length acting in circum-
coordinate in a direction of 92 (N 1 2 = N 2 1 )
(Fig. 1. 12-14)
n Number of buckles
(Fig. 1. 12-14)
(Fig. 1. 12-14)
xxxiii
Qb Dimensionless parameter for buckling analysis
subjected to bending
and cylinder
surface
xxxiv
R. Radii of curvature (i = 1, 2 for and
Re Equivalent radius
curvature of stringer.
0Base
r° radius of spherical cap
XXXV
- 1/2 1/2 - 1/2 1/2
S 3 , S4 , 55 (Dx B0)/(De Bx), D Dx / G Bx o
respectively
s Arc length
T Torque
t Shell thickness
configurations
xxxvi
U, UM, U1 , Orthotropic cylinder parameters
. U2 , U3 (Paragraph 3.33. 1)
(Section 3. 53)
coordinate
loads)
W Weight of sandwich
function
xxxvii
X Loading component in meridional direction
a cone surface
the cone
dis continuity
bulkheads
xxxvi ii
Y Loading component in tangent-direction to the
circumference (psi)
to the surface
parameter Z+
/R)
xxxix
Pik Forced rotation of tangent on meridian at edge i
position respectively
parameter
"YI2' '13' Y23 Linear shear strains (y13 : Ylz and y 2 3 YZz are
strains
Ar Change of radius
xl
Arik Forced change of radius at upper bulkhead i for
(Eq. 1. 13-22c)
respectively
11 1 TG
KAF Plasticity correction factor
xli
I$' K2 , 9I2 Change in curvature and twist of the middle
(Paragraph 3. 25. 1)
(Fig. 3.43-5)
(Paragraph 3. 32. 1)
II'X
1 Ile, Poisson's ratios for an orthotropic layer
(Section 3. 32)
xlii
x/L (05 g:5 1), meridional coordinate
•I •Curvilinear coordinates
xliii
a.0' -25 T1 2 ° Components of stress corresponding to the
equilibrium configuration
T 1,
ITV T 1 2, Associated stress of admissible variations
T Waffle angle
1 at edge i
12 $ at edge k
xliv
nL
,ZR
(Paragraph 3.33. 1)
0
configurations, respectively
xlv
1.00 INTRODUCTION TO THE THEORY OF SHELLS
1.01 GENERAL
concerned with the study of deformations of thin elastic bodies under the
The most common shell theories are those based on linear elasticity
The nonlinear theory of elasticity forms the basis for the finite
often required when dealing with shallow shells, highly elastic membranes,
and buckling problems. The nonlinear shell equations are considerably more
difficult to solve and for this reason are more limited in use.
of shell theory.
with the shell theories based on linear elasticity. The linear theory of
shell theory from elasticity considerations are outlined along with a variety
of simplifications.
calculus of variations.
is included.
the intent is to supply the analyst with the theoretical background necessary
2
1. 10 LINEAR SHELL THEORY
1. 11 INTRODUCTION
The relationships governing the behavior of thin elastic shells are based
The geometry of shells (i. e., one dimension much smaller than the
to reduce the shell problem to the study of the deformations of the middle
(or reference) surface of the shell. In all cases, one begins with the
ables, to a new system involving only two space variables. These two
3
Shell theories of varying degrees of accuracy may be derived
depending upon the degree to which the elasticity equations are simplified.
Aron (Ref. 1-1). The first apparently successful approximate theory was
of its popularity, the development given by Love is not free from inadequacies
including Love (Ref. 1-3) himself, have attempted to improve on the approxi-
tance in the practical analysis of most shell problems. The simplicity and
shell theory.
4
To better understand the theory of shells, the subsequent development
of the theory will start from consideration of the general elasticity equations.
The various shell theories will be classified and the assumptions and simpli-
and their effects assessed. A linear shell theory based on Love's first
as they appear.
5
1. 12 BASIC RELATIONSHIPS FOR THE THEORY OF THIN SHELLS
These are given for reference and serve as a basis for the subsequent
A. Arbitrary Shell
by specifying the form of the middle surface and the thickness of the shell
on tensor analysis and differential geometry (Refs. 1-6, 1-7, and 1-8).
terms of two independent parameters (If' 62). ithe range of these param-
6
e1 -=C3 el =C4 el--C5
el1 =C el =C2
e'2 CI'
' /
e2=c
e =C5'\
The square of the linear element (ds) of any curve traced on the
7
where X is the angle between the tangents to the coordinate lines l and 2
at any point. Eq. 1. 12. 1-1 is called the first quadratic form of arc length in
equal to 90 and Eq. 1. 12. 1-1 is considerably simpler because the last term
ý2 and represent the first fundamental magnitudes of the surface (for ortho-
parameters or coefficients.
linear elements along constant coordinate lines of the surface when the
W2
8
The quadratic form for a given surface will have different expressions
ds 2 = drZ + rZdeZ
is chosen to simplify the formulas for the surface. Such a system is that
9
in which the two families of coordinate curves are simultaneously lines of
surface which possesses the property that normals to the surface at consecu-
tive points on the curve intersect. In the general case, the curvature of
this line varies with its orientation on the surface. The directions at which
the curvature reaches extremes are the directions of the lines of principal
__ __
Lt2 "C1
at that section drawn through the normal at the point (P) which contains the
10
thickness, t, can be related by three parameters. Two l 2) vary on
the middle surface of the shell, and the third (z) varies along the normal to
f2 c2
dS 2 = Al 2
dýl 2
+ A2 2
dg 2 2
+ A 3 dz 2
(1.12-2)
A1 = +
+ Z
A2 =02 1 + (1.12-3)
A3 = 1
11
The geometrical significance of Eq. 1. 12-3 can be seen by exami-
nation of Fig. 1. 12-5 which shows the cross section of a shell element
MIDDLE SURFACE
Cl' a2 ' Rio R 2 in Eq. 1. 12-3 are functions of 1 and and must satisfy
the three relationships from the theory of surfaces. Two of these relation-
(1. 12-4a)
Rl d -, U1 R
12
The other relation, called the condition of Gauss, is
Equations 1. 12-4a and -4b are presented for reference. For a more
B. Shells of Revolution
about an axis lying in the plane of the curve. This curve is called the
meridian, and its plane is the meridian plane. The intersection. of the
For such shells the lines of principal curvature are its meridians
middle surface are the angle 4 (between the normal to the middle sur-
face and the axis of rotation) and the angle 0, determining the position
13
MERIDIAN
R=R 2 SINO
R2 III AS
PARALLELMERIDIAN
RI2
normal to the middle surface between the surface and the axis of the
shell since, considering two adjacent points on the same circle, the
dsI = RIdq
ds 2 = R 2 sink dO
14
In this case, the Lame parameters are
a 1 = R1
a 2 = R 2 sin (1.12-5)
where
e S=
alamely *.
symmetry.
15
TABLE 1. 12-1. GEOMETRY OF SHELLS OF REVOLUTION
Increment of Arc
d)2 +z d 2
=a 2Idd
ds 2 d2 + 0 zd0"
22 22
form:
a1 =R2
a2 = R? sin *= R
Shell
Shape R1 R2
Sphere Y= 0
Degenerate Cases
Increment of Arc
ds 2 =adX +dZdO
1 2
a 2 =R 2
.R1 R2
Cylinder c R
Cone co R
sin (90" -a)
16
1. 12. 2 Geometry of Strain
this section.
requires relating the position of points in the shell before deformation with
ships relating the six dependent components of strain must exist. These
relationships are the compatibility conditions of the state of strain (Ref. 1-5).
they can be linked by virtue of this state. The relating of strains to dis-
17
The strain displacement relationships appropriate for linear shell
1 aw + v A3 U 8A 3 (1. 12-6c)
3 - W3_ T3 A 3A2 2 AIA3 *Tl
_A
21 / V A 1 a (1. 12-6d)
A-1 a~l A2 A27._a2
A, a . +-A l
13A 3 a• A 1
-,
\ 3 (1. 12-6e)
A3 a __ A2 a
.A2 a _ (1.12-6f)
A3 3A-
where (Al, A 2 , A 3 ) are the Lame parameters which are basic quantities in
18
In describing the geometry of shells, in Sec. 1. 12. 1, the coordinates
within the shell wall, and z was taken to correspond to ý3, where z is a
coordinate measured along the normal to this surface such that (•1 • z)
3 =Z
19
I. Strains are small in comparison with unity, i. e., c << 1
aw
c)F F
au etc... << 1
with unity. These assumptions permit higher order terms in the strain
In the next two subsections, the basic relationships for shells obtained
from the law of statics are presented. The concept of stress resultants will
distortion, and the effect of the forces is transmitted throughout the body.
Across any small internal plane area of the body, forces are exerted by the
part of the body on one side of the area upon the part of the body on the other
side. The term "stress" denotes this internal force per unit area. (See
Fig. 1. 12-9).
20
AF
&F
FIG. 1. 12-9
TK
21
The internal stresses shown in this figure are described as follows:
r1, a2Z are normal stresses, acting on the faces of the element.
T
12 , T
21 are in-plane shear stresses acting parallel to the middle
surface.
surface.
normal stresses acting on the faces which coincide with positive directions
are positive.
acting on a faceof an element of shell; this stress, which will be called (rI1 ,
-oil
N f Oild, ... MN dr
22
The resultants (forces per unit length of given arc length) acting on a
t/ 2
N1 2 T 12 1 + .-- d z
t /Z R?
t/2
Ql f T 1 3(1 + 4 .dz
t/2
N2 =f t/2 T2l(1 + dz
21 ft/?-2 RI d
t/2
Q2 .- / T 2 3 1 + dz
.1-t/2 R
and for the stress resultants for moments,
t/2
/= z 0-11I + Z)dz
t/2
M12 = - TlZ 1 2 /1+z
ft/2 + )z (I.12-8b)
J-t/2
M2 = z Or
2 1 + Z dz
-t/2
21f= t/2 Z T2 1 ( +
23
In the above expression, the term + present to account for the
trapezoidal shape of the shell element resulting from the curvature-of the
may, in the future, consider instead of the space element cut from the shell
the corresponding element of the middle surface on the sides of which act
N1I N1 N 21
C'2
24
1. 12.4 Equilibrium Equations for a Shell
section will now be considered and the conditions for its equilibrium
under the influence of all external and internal loads will be determined.
The external loads are comprised of body forces that act on the
element and surface forces (stresses) that act on the upper and lower
bounding the shell. The internal forces will be stress resultants acting
sponding to the considered element of the shell and they were replaced
25
Now, instead of considering the equilibrium of an element of a
shell one may study the equilibrium of the corresponding element of the
middle surface. The stresses, in general, vary from point to point in the
shell and as a result the stress resultants will also vary.
N2
Nf2 eN1 N21 ÷21
#N2 de
0S,-
N 1 +1~
dC
N.1 ÷'_2- doj N +dd
N1 +o d•f, 12 a •
o
02 01 ml '2 + 2 12
m,+;•
*M j M12 +OM2d
of,- f,
26
The equilibrium of the shell in the 1, , and z coordinate
8 alN 2 aN 802
12 8a1 1l 2
Ba 2 Q1 a Cy
1 Q2 N1IN
g1 +Z -l +
\R I R2)a l&2 q = 0
per unit area applied to the middle surface of the shell. (The details
N-zN2 M1 2 M21
Nl2"
N1 - ---=0
Rl R2
to warping of the faces and result from in-plane shears and twisting
moments.
27
In the equilibrium equations presented here, changes in the
follow:
parallel
a1 = ,a 2 =R = Ro +x sin&
R Ro
RI io
== D, R R= R + x tana
R1=R=coR 2 = osa Cosaa
where
28
R R 2
8Ne a
aE -l (Nx X) sina + Ro-5x (Nx 0 ) + Nex sine + Q0 cos a + p R =0
a a a
8 +0-- (MxOx) sina + Ro T (Mxe) +Mex sn
sino' -
(
= 0
Nx Nox M = 0
29
Eqs. 1. 12-10 become the equilibrium equations for a cylindrical
a, = R, a = R sin 4)
of curvature are \
R4, = R 8 = R = const.
become:
30
(N aNe
ao (No sin o) - N 8 cos +Q
+sin6+Rp
+- sin4
aNO a N
cos 0 + Qe sin 0 + Rpe sin4 0
-88 + 5 (Noe sin 0) + No.
N$ in
No sin0++ Nosinc~-
O sn • -aQae bo (Qp sin 0) - Rq sin 0. (1. 12-11)
C3 aoc
that is that stresses are linear functions of strains and thus the propor-
31
This section is devoted to a brief description of the basic ideas
A. Anisotropic Bodies
body whose physical properties may vary in any direction, the generalized
72 +
gI
eZ2
1 I "-cI A
+ A1 2 (Z + Ar 3 43 + A 14 yz 3 + A2 5 + A1 6
T3 1 = A3 1 E1 + A5 2 f2 + A 5 3 E3 + A 5 4 y 23 + A 5 5 y 13 + A 5 6 Y12
T 12 A6 1 EI +A 6 2 E2 +A 6 3 Q3 -rA 6 4 Y 2 3 +A 6 5 Y 13 +A 6 6 Y1 2
32
As shown in Ref. 1-12, the 36 coefficients reduce to 21 when
of stresses are desired. The Aij's are material constants. Space and
simplified.
B. Orthotropic Bodies
33
C. Isotropic Bodies
Many bodies have elastic properties which do not vary with respect
to orientation in the body. Metals such as steel and aluminum very closely
011 = 2 EI + )e
22= 2 •2+ Xe
(1. 12-15)
'33 = 2 4(3 + X e
where
k- •E
(1 + i0) (1 - 2 R)
E G(1. 12-16)
2 (1 +
e = I + + C3
and E, G and } are called the engineering technical constants. The number
follows:
1= EE + . pC)
Ir - R•2 (' I + F2 )
E +
E2( i (1. 12-17)
2 - 2
pA 2
34
1. 13 CLASSIFICATION OF SHELL THEORIES
how either the stresses or strains vary across the shell thickness.
This situation arises from attempts to reduce the shell problem from
can be established.
35
The selection of the proper form of these approximations has
classify them according to the assumptions for which they are based.
order of the terms in the thickness coordinate that are retained in the
36
modifications thereof and as either first- or second-order approximate
distinguished from those which include the shear effect. Linear mem-
37
2. Linear elements normal to the unstrained middle surface
A. Strain-Displacement Relations
z = aW
8Z"= 0 (1. 13-18)
coordinate.
38
Assumption (2) of Love's first approximation is analogous to
the form
U u +z1
V = v + zP 2 (1. 13-19)
W=w
It can be seen that the displacement functions at any point in the shell
(Ref. 1-13):
39
1 + -'Z
12 (= + z•K 2 ) (1.13-20)
1 + z 1
Y1 2 -= + 2 2
Z z
1 + 1 +
Rl R
where
1 a 1 + 2 80l
•1 - al -5 l + &16•2 ag2
(1. 1 3 -22a)
2 1O 2~
2 a72 0g2 '1 '2 8a1
are changes in curvature of the middle surface directions 911 and %2,
respectively.
0 0
Contributions Y and Y to the inplane shear strain and 6 1 62
40
1/ av U a1F
o =2 = aao80 2 )
o,(1.13-22b)
1, apP
2 p1 3ar 1 1 2ao3~
1- 62 a (1. 13-22c)
UNDEFORMED DEFORMED
FIG. 1. 13-1.
straight but remain normal and suffer the same rotation as the middle
surface. The angle change between the middle surface and normal to
41
given by the transverse shear strains, .ylz'"y2z (evaluated at z = 0).
Since the angle between normal and middle surface does not change,
RI+
0
lz +1 Rz ( at,
R1I
(1. 13-23)
'2z 1 + R(2)
R2
of Eq 1. 13-23 become
P= u - -1 aw
(1. 13-24)
v I 'aw
R2 22 Z2
continuous loads and local areas around a shells edge, shear deforma-
42
surface forces, having flexibly supported edges, are usually
Paragraph 1. 13.3.
z
From the thinness assumption (1), terms -R are small in compari-
son with unity, and can be neglected in the strain and stress resultant
linearly across the thickness of the shell. For this case, from
C2 = O+ZK
2= (1. 13-5
€l = Y10+ Z K 12
0
where the shear strain, Y 1 2 , of the middle surface from Eq. 1. 13-22b
is given by
Y1
o =~0Y++-o a (3_ aU
~ ) IauL
iA v 89
f (1. 13-26a)
2 1 2 1 2-l g2 -2+ a/2 1 1
43
K - 6 + 6 122 P 1 1 2 2 (1. 13-26b)
12 1 2 a l \atl 0'2 at 2) 2 \a, ~ 1
do not vanish for small rigid body rotations.. Other versions of these
theory.
tions for stresses shows that, unless the surface loads are highly
44
concentrated, the traverse normal stress, a-z, is generally of smaller
order of magnitude than the in-plane stresses o-,, and T." Consequently,
orl1 2 f I + ýL(2
E i
E
-2 (1 +L) YI 2
terms which are linear and quadratic in z. (See Naghdi Ref. 1-13.) If
these additional terms are small in comparison with the leading term, W.
45
Thus, to obtain a first approximation theory the additional terms may
and Y12
t/2 t/2
N1 = f /2 0'i dz, NZf= a-Z dz, ... etc. (I. 13-28)
For an isotropic shell, utilizing Eqs. 1. 13-25, 27, and 28, the
N1 - Et€ + o]c
N2 -= - ,[E1 + EI
Et o
N12 = N2 1 Et
2 (1+ 0
M =
= D I + (KI. 13-29)
46
M2 = DIK2 IL
+ KI
(1.13-29 cont)
M 12 = M2 1 DKI2
where
3
Et
12 (1- FZ)
0 0 0
equations 1. 13-21 and 26a and change in curvature and twist terms
follow.
47
neglected. Thus it might be thought that one should neglect shear
forces QI, Q 2 , and consequently assume that the shearing forces are
zero. However, this is not correct, since the shear forces play an
aa2N1 1 + N1 2 - N2 aaZ + 1 +
M M M1 2
da+ l C1 PI = 0
Rl1 a Rl )2 R1 dE2 p
(1. 13-30)
a N2+c&-N +N ae-N a + a2M 2
N 1 2 + N2 1 . N1 +_R 2 +a
N ag
48
+( a&? 12- M 2 +Ml?.
a ["( I ae,2 12+ a, Mn2MMI -l
M 2A]
1 - N, ca 2 -NN 2 '2+ a, 2 q =0
M21 a~/ ) R, RZ
displacement relations (Eqs. 1. 13-21, 22, 24, and 29) into the
forces.
49
D. Boundary Conditions
shell and, hence, do not have solutions as long as they are subject to
ment comes about due to the fact that in neglecting shear strains the
The five resultants are reduced to four by noting that the distribu-
50
distribution of shear forces. This leads to shell boundary conditions
Love has derived the required shell edge conditions for an arbitrary
Ný or u
M 12
, -- or v
R2
(1. 13-31)
aM 2 1
Q2 - or w
aw
M2 or -a are prescribed.
simultaneously.)
The theory presented was first given by Love (Ref. 1-2)andis referred
order theory because the strain Eq. 1. 13-25 and constitutive relationships,
51
Eq. 1. 13-29, include terms up to the first order in the thickness
Love's theory. Since then Sanders (Ref. 1-14) has developed an improved
for small rigid body rotations ot the shell whereas for Love's theory
Eq. 1. 12-9 with modified forms for the in-plane force and twisting
problem contains all the essential facts necessary for the treatment
stresses.
52
the nonhomogeneous differential equations consists of a particular solution
equation.
generally of the mixed type, involving both boundary and middle plane stresses,
but for some problems, such as a shell under concentrated loads, the homo-
Thus, there are two extreme cases possible within the first approxi-
mation; (1) the inextensional or pure bending case in which middle plane
strains are neglected compared to flexural strains and (2) the extensional or
membrane case in which only middle plane strains are considered. The
general or mixed case lies between these two extremes. The significance of
53
1.13.2 Second-Order Approximation Shell Theories
1 + z2
1 -+ +lR
R1
0
0
-1 + z - + Z2 -
2 2 + R R R2/
(1. 13-32)
12~~ ~ YI 1 R R
+R2?/ R2 \2 )
54
If these expressions are substituted into the stress resultant
expressions, retaining second-order terms in z, the constitutive
1 1 2 R1
t 0 0 ) 1 2 (1 3
22 =- 2[ + 1) 12) -( 1 1 z "
SR- R R2/
N12 --Gt 0 t1 1 1
12 RZ/J R\i
12 (l - •z) )C
Gt3
12 Rl R7.
M2 1 =- 12- R j 2
where
2K 1 2 = T + 20 + 0
55
The preceding equations are characteristic of the Fligge-Byrne
small terms in the wall thickness leads to relations that contain terms
56
Hence, these formulas are regarded as inconsistent and, for shells of
approximation.
57
distributed or, as before, terms up to and including the second power
W = w+(i,
+2 z) (1.13-34)
_ 1
_ - -- (1.13-35)
R1
E = +I~ Z R gi +I+R•l
il. 13-36)
58+
58
and
• Z+
- ~zE [ z(' 1
)+-- 2 (CK+ K) 1 (1.13-37)
(l -)EzL 2 J
equations become
t2 zE 1 'L ) ( 1 + K
12 (1 - f)E (1. 13-38)
K1 + 1.1K 2 2 +P
KC DC 11
+ + Ii
R, R2
neglected). More important, however, is the fact that new terms are
59
As can be seen, Love's second approximation contains some
Bassett (Ref. 1-23), Treffty (Ref. 1-24), and others (e.g., Ref. 1-15).
60
displacements u, v, w, and their derivatives. When the effects of
t
2
(1.13-39)
t
Q2/ Q2 r 2z dz
t
2
equations.
61
done in Paragraph 1. 12. 4. Thus, when shear effects are included,
laaaN N 2 1 + N 2 1 - 8 2N
+ atz 2P+ + N12 a6 N2--+
at Ql
1 Rl-- + &l02 pl =0
a' 1 N2 aaZN 1 2 a + =z 02
+ N 2 1 9a NI-- + Q2 -- + al 2
a2 1+ 89 NI a1R2 N2 a
R2 +C12 12 q = 0 (1.13-40)
a& M1 + iM 2 8o + M
---,2
2 2 1
M Q2 " 2= 0
8M
at+ - M1 21
~a 1 Mzj + aZMI at - z+ M
. 892-a42 _1 - Q lala2 =0
N, or u
U 2 or v
M, or PI
M 2 or P2
Oorw
62
Reissner (Ref. 1-25) was one of the first investigators to incor-
this section, some of the simplified shell theories resulting from con-
63
Eq. 1. 12-5 into relationships developed in section 1.13. For the particular
1. 13-26a reduce to
1R dR R1
(1. 13-41a)
0
•2 = eoO u cote + u dR
-+2 .. _
w_
2 + RIR 2 dO R2
'12 =y8 = 0
and the curvature (Eq. 1.13-22a) and twist (Eq. 1. 13-26b) expressions
become
K K d (.1 dW u'•
1 1 2
40
62
64
dR
For a general surface of revolution, the expressions d-and
dR 2
- are as follows for
R = R 2 sin 4
dO
R 1 cosý
dý
S=
(1.13-42)
dR2
dR2 = (R 1 - R 2 )cot+
u cot+ w
R2 R2
(1.13-43)
2- cot'O[dw
2 R1 R2 [d0
ý4( vanish and the equilibrium relations from Eq. 1. 12-9 become
d (N~R)N
dNo -
0 R 1 cosý + QOR + RlRp = 0
d(MOR)
d_ M 0 R, cosO - RRQo = 0
65
where the second, fourth, and sixth equations of Eq. 1. 12-9 have been
identically satisfied.
and -43, two second order ordinary differential equations in the two
Meissner (Ref. 1-29) shows that the equations for a general shell of
condition. ")
66
The transformation to the Reissner-Meissner equations is
A
dw\ (1.13-45)
U = R2 Q
or, in fact, for any shell of revolution satisfying the Meissner con-
form
(1. 13-46)
L (V).Z :..
R1I D
67
where the operator
L(
L( R d 2()
-1
•+-cot~ rdo (R 2 \ Cot 1o
RZ dO R 1 [d R1 do
RI cot 2 q•
(R. 13.-47)
R2 R 1
form
where
F4 Et FL2
r 4=-Et-L-
R21
(1.13-49)
D
L(0) * ir U = 0 (1.13-50)
68
spherical, and toroidal shells of uniform thickness. Furthermore,
they follow directly from Love's equations in the more general case,
are satisfied.
B. Spherical Shells
69
For axisymrmetrically loaded spherical shells of constant thick-
70
d3Q' d2Qý dQ4
d4 A3 3 A 2-+ + A, - + A004 + 4(4Q0 0 (1.13-52)
dc 4 dO d 2 1 dO
where
A0 = 1 - 3 csc4* -
A = cot 0 (2 + 3 csc 2
A2 = -3 csc 2
A3 = 2 cot
and X4 = 3 (1-F 2) R
t
d4
dQ'- 4xQ =0o (1. 13-53)
angles ý ; that is, for thin, nonshallow spherical shells. This can be
seen from the fact that Q01 is a rapidly damped function of the form
71
coefficients A 0 . .. , A3 are small for high angles •, all the terms
transformation
Es slinger (Ref. 1-36). The angle being small, only the highest power
72
3
A0 u -_-T
3
A?= --
3
A2 T
2
A 3 x--
d4 d3 0 3 d2Q 3 dQ 3 4
+ ",2+ - - - Qo+4x QO=O (1.13-55)
d* 4 0 dO3 2 dO 2 o3 d 4
[d2o
which may be rewritten as
equation:
d2QO 1 dQo 1
+ -- Q iX
Zi- 40= 0 (1.13-57)
73
measure of their relative accuracy at a particular angle 0 can be
to be constant over short segments of the shell (Refs. 1-26 and 1-41).
FIG. 1. 13-2
74
analysis is applicable to shells that are not shallow when the stresses
are effectively restricted to a shallow zone.
2= e
01 = 1 (1.13-59)
&2 = r
RI =R2 = R = Const.
82
r 8rr-2
of plates.
The observation can be made that for shallow shells the effect of
75
equilibrium (Eq. 1. 13-40) in the meridional and circumferential
setting
2F
NO-8r2
N 2 F2
aF
Br 2
N --F----+-- (1.13-61)
r r 8r r 2 a e2
re 8r re
For convenience, the load potential has been neglected. The complete
2 2 tE 2
F W (1. 13-62)
where
v2=a 2 + Ia-+
ar 2 r 8r r 2 ae 2
76
The second differential equation involving F and w is obtained by
2 2 1 2
DV Vw R F =0 (1.13-63)
set of shallow spherical shell equations for the case when the effects of
with coordinate axes (s, e) measured along the generator of the cylinder
yields
77
E! = a"'u
1 av w
=v 1 au
12 ax+R 8 1
2
K -= (1. 13-64)
ax2
X= a2W+2R_I_ av__
+ a8
R ae R ae
2 a~ +. I av
K1 2 R axae 8~x
I aIM + amxe -
aMx 1 aMxe -Q =0
ax R ae
78
The final system of three partial differential equations in the three
obtained using Eq. 1. 13-64. The resulting three equations contain cer-
du w d2 w
1i 0 -Z K~ 12 dx(1.13-66)
*1z = K2 K12 = 0
and
dNxX+qx = 0
dx X
dQ N0 (1.13-67)
dx R q
dMx Q 0
dx
"79
If only pressure is considered, Px = 0 and q = p, and the above
where
4 3 (1 •L2) Et 3
R2t2 , D= 2 -2)
R t 12 (1 -F)
Particular solutions are given in Refs. 1-26 and 1-43, and are
1 a2 w 2 a 2w
2 R 2 8a2 ' 12 R aOax (1.13-69)
ture and twist become the same as for flat plates. A similar simplifi-
cation was made in the previous section for shallow spherical shells.
80
Following Donnell's approximation, the terms R are also
a2F 2F a2F
Nx - asx Ns = a Nxs xas (1.13-70)
where
DV w + -1 a X2F
x2 - q = 0 (1.13-72)
V8w Et84w
4
14
DR2 8x D v1q (1. 13-73)
complex form of the above equation was obtained by Naghdi (Ref. 1-33)
81
when transverse shear distortion was considered. Donnell also pro-
form of the above equations. The details of this will be discussed in the
Byrne (Ref. 1-19) retain the z terms in comparison to unity in the stress
R
resultant equations and in the strain-displacement relations. Fl~igge-
case solutions are obtained in Refs. 1-19 and 1-45. In the latter
compared.
82
zero transverse shear strains permits a rapid transition from the
of resisting bending, or (2) shells that are flexurally stiff but loaded
83
resistance to bending, only a momentless state of stress is possible.
For shells with finite stiffness, such a state of stress is only one of the
cases using less material and, thus, resulting in less weight. The
theory and, for this reason, historically preceded the latter theory.
moments could exist in the shell, the loading could only produce normal
84
determined" (i. e., the analysis could be performed solely with the help
relations).
A. Equilibrium Equations
0
1 1
12 = 0 (1.13-74)
12
the moment terms inthe equations of equilibrium for the shell element
85
M = M2 = M 12 = 0 (1. 13-75)
which implies the neglection of transverse shear forces from Eq. 1. 12-9b
Q1 = Q 2 = 0 (1. 13-76)
N 1 2 = N2 1 (1. 13-77)
8a N1 1N2 1 a1 aa 2
2 + + N12 -- N 2 - +Cl 1 2Pl = 0
ag1 ag1
2la 2 Ia142
1 R N2 R 2 +a 1 le 2 q 0
86
analogous situation occurs in the problem of the bending of a beamn where
displacements.
al
=011 au
- + 1 a&,
8•2
01I0I2 V+- w1I
1 v + "2 +w (1.13-79)
2 2 "20l12 891 R+2
02 8 1+ a
1 i 22 2 (
87
The indeterminacy of displacement magnitude and necessary bounds
B. Boundary Conditions
QI =Q 2 = M = M 1 2 = M 2 1 = M2 = 0 (1.13-80)
which hold at all points in the middle surface and, hence, on the
must be free from external edge loadings in the form of normal shearing
88
stresses and bending moments. As a result, membranes can only
would affect the corresponding general forces, Q, and Ml (e. g., the
membrane theory results from the fact that the solutions of this theory
89
Another contradiction in membrane theory is the fact that its
several conditions concerning the shape of the shell, the character Cif
symmetry as the shell itself. Forces are then independent of 2 or8, andall
derivatives, with respect toe, disappear from Eq. 1.13-78. There results
N) No
R+IR ý r (1. 13 -81 b)
d (rN¢)
d sin q) + rN 4 cos 4 = RIR Pr cos4
2 sine - RIR 2 pO sin2 o (1. 13-82)
90
In combining we obtain, for one term on the left-hand side of equation
(1. 13-82).
d (rNoý sin )d (Zp)sn
d-O d-O(z0 i2
hence
FIG. 1.13-3
N4 = -, N8 = pR
91
For a conical shell under uniform internal pressure,
pR pR
N4 - 2 sin4 . Ne =sino
1.13.6 Summary
92
1.20 NONLINEAR SHELL THEORY
1. 21 INTRODUCTION
these equations, which are based on Hooke's law and the omission of
nonlinear terms both in the equations for strain components and the
sions for strains and equilibrium in order that the equations could be
93
"physically nonlinear" with respect to the stress-strain relations. This
latter type of nonlinearity forms the basis of inelastic shell theory and
deflection shell theories form the basis for the investigation of the
paragraphs.
94
1.22 GENERAL STRAIN-DISPLACEMENT RELATIONS
rotations were made. It was assumed that strains were small (i. e.
(<< 1) and that rotations were of the same order of magnitude as the strains
ble to restrict the rotations, and it becomes necessary to deal with the
1 1
aW +2aV au (1.22-1b)
e2 z Y2z + P?. "-z + + Y2 "89
95
where the terms c1 E2 , 'Yl 2 Ylzo andYzz are the linear strain expressions
(Eq. 12. 2-6) which in terms of displacements and their derivations, are
given by:
- + V + -W
= _1 11 av u 2It
a W
2 2 -5 2 8t R2
+R
2
I fi U + 1 aU V C1112
Y +- - - - +1
, (1. 22-2)
R1 R2
Ou 1 /18W U\ /a
Ylz +
R1
Cv 1 1 w v,
0-
1+ a22 R2
2
and the additional terms PV, P2.' Y1 and YZ are given by:
I I0lal u U)2 1 a+ u
1 +1+R ;1
2 -
1a1
U V a 1 . w V (1 . 2 2 - 3 )
v a2 I +W V
21 +1 z '2 - 2 1 1 + IT?-( 2 RZ)
96
It is noted that Eqs. 1.22-la and b are a set of nonlinear differen-
Ylz =0
Y ~z0
z=0
and Kempner (Ref. 1 -22) and result in nonzero strain expressions of the
form
97
2
e 1l= •1 +'7-
e=2+ "2"(1.22-4)
e =
e12 = Y1 2 + P1 P
2
U Uo + z l
V vo + zP 2 (1.22-5)
W =wo = w
and Eqs. 1.22-4 and 5), there results the general form of the strains as
eo
el e0 + Z KI
1 1 1
e2 =e0 + ZK2
=e
2 (1.22-6)
e -e0+ZK
12 12 12
o 0o
where e1, e2, and el 2 are the nonlinear middle surface extensional and
in-plane shear strains given by Eq. 1.22-4 (superscripted with o) and
where
98
1 ap1 1 1
Kl•1 - r 0 1 a
+cl~ 2 a2[2
i 1 8+2
1C + + p7
12 02 i a2 at2
(Ref. 1-10) for small strains, the correction introduced by these terms
the strains (e) were of the order of the magnitude of the rotations (p) or
0=
O() << 1. By relaxing the constraint on the middle surface
rotations, the nonlinear strains of the middle surface are of the order
99
1.23 STRESS-STRAIN RELATIONS AND STRESS RESULTANTS
written as follows:
T1 2 = y1 2
Yl
N{ = IJdz, {z = zdz
NIZ2T z 12) 12)Z
where the integrals are taken over the shells thickness (t). If
Eq. 1.23-8 and, subsequently, Eqs. 1.22-4 and -6 are inserted into
follows:
N1 B (e 0 + •00) + C ( K +
1 K1 2 )
N2 B (eo + lieO) +C (K 2 + •)
=
j 2 +=
N2 = =0 (eo +01 + N
N(21 (13-0
12 = (e12 +1 +
M1 = C (eI10 + ýLe.) + D ( Ki + ýLK2)
100
wh e re
'
B =
1
-fEdz C=
-
fEzdz
t 1
D= fEzzdz
j' (1.23-11)
B = f Gdz, J Gzdz,
f- fGzZdz
ft
These expressions are presented for reference and will be utilized later
this simplification was possible since the reference surface was selected
101
1. 24 PRINCIPLE OF POTENTIAL ENERGY
since strains and rotations were negligibly small, the deformed and
called calculus of variations; for this reason, these methods are some-
102
The potential energy, V, of a shell is given by the expression
V = Oj-
- (1. 24-12)
internal forces, and (-W) is the potential energy of forces acting on the
shell if the potential energy of these forces for the unstressed state is
sion can be simplified by substituting Eq. 1.23-8 into the above, which
relations (Eqs. 1.22-4, -6, -7) into the above, the strain energy
103
conditions, those which satisfy equilibrium conditions make the potential
first variation and the one analogous to the second derivative corre-
sponds the second variation, resolving the problem to that of finding the
to give the engineer not familiar with variational methods some idea as
104
stationary. In this connection, it will be assumed that the displacement
parameter X.
X = 0 yields
, -X d IdV(X)
==0 (I,.24-16)
105
the following section. A similar operation using the second variation of
Section 1. 30.
work, in which the terms X-, OV, and XW can be thought of as virtual
under action of given forces are those that lead to zero variation of the
total energy of the system for any virtual displacement from the position
that
106
1.25 NONLINEAR EQUILIBRIUM EQUATIONS
method was introduced in the previous section and will now be utilized to
basis.
2=
e 0oe +X; +÷X2
12 12 12 12
0
where elf e 20 P e 0Iare
2 given by Eq. 1. 22-4 with appropriate superscript
o and
e= eo +I p 1
1 j2 I I
= 2 +21P0 (1.25-19)
2 22 222
10
-2
e(2 - ) (1.25-20)
107
where * and P are as defined previously provided these quantities are
the form
02+=0•0 +x •
T 1 1
0-2
= + X(2 + (1.25-21)
S2=
T 12 12 +XT 1 2 +X T21 2
where neglecting thermal terms the To oI, --- ,T12 expressions are
Inserting the right-hand side of Eqs. 1.25-18 and 1.25-21 into the
U0 +Xi3 +X ii 2 (1.25-22)
where
U0 = o2 T 01 e°0
1 2e 2 + T 102 e1 221
+ o"2 °1
ola 2 dtldgdz (1.25-23)
V
-U = f [e. +oT 2
+l. e +. roe &la 2 ddg~dz (1.25-24)
TU
2 = Tfj 16 + T 2. 2 + 7ý12 71 Zj
0 0 - 0
+2 TIe 1 + 2e2 + T 2 e1 2 a 1 a 2 dgId21dz (1.25-25)
108
An expression similar to Eq. 1.25-22 can be formed for the potential
energy expression (-W) due to applied external forces. For large deflec-
~2
where the superscript, o, and roof, (A), on the integration limits repre-
sent initial and final boundary coordinates, and the f's are a function of
Eqs. 1.24-15 and 1.25-23 -25. The potential energy expressed similar
V =v +XV + X V2 (1.25.27)
dv - - = (1.25-28)
X=0
109
Thus, the first variation expression described by Eqs. 1. 24-16
and its equivalent, Eq. 1. 25-28, are used to establish the equilibrium
equations. The insertion of Eqs. 1. 23-8 and 1.25-19 into Eq. 1.25-24
and these into 1. 25-28 and integrating by parts noting arbitrariness of
admissible displacements U, V, W leads to the large deflection equili-
brium equations of the form
aa
N° N
No0 aa 2
__2 1 1
___12 + _1 1 N + "1 o
S12 ag- 2 R Q1
01 a2 0g
aI aZ (P 1 ~o o ) + alapl =0
ClaaN
N
12Oz
a2 2 9 t_1 N
12
o9a 'l 1
N0 + 2 o
1 2c
1 1 2 1 2 2-
a1 0-2N 0 +N ) +&l aQ
R2 1N 2 2 1 2 (1.25-29b)
2~1 ~2 ~ ~3N
1 (1P2No+c PN
+ ~ 1 ~q=O(1.25-29c)
8ý11 22 R R
82M 021 22 0
1M12
8aM 8I 10 82120
8~1 + MI 2 - -- • M2 - a 1 a 2 Q1 0 (i.25-29d)
110
2a21 2 + 0o 8a1 0 o
0
C3 12 a- 1 1 2 2 (1.25-29e)
where it is noted NI0= N1; Ml° = M 0
The details of the above operations can be found in Ref. 1-22. The
may be realized. Solving for the last two equations for shear forces
QI, Q2, in terms of the moments, and substituting the resulting expres-
sions into the first three equations results in the following "large"
(N
0 a (N 0o 2) -N 00o 8e 2 +I [•_ oc 2)
1a
+(NO2) a1Z 1 R
) 71((M2 21)?2) - N- P M1
11 1 .2
+• 1 '• 211' 2 • 1 12
NTAZ 0 P
M0-N 1 'a
12 2 1 . I + p 0
a102= 0 1. 25(30a)
1 a 1o 2 ) 00l 1aI o o
+ a2 (M 1 2 2 1 a- N 22z0Iaz
-N? 001P2=0
N1 1 1 , 2 + = 0 1. 25(30b)
111
/N 0 N2°\ a [ a 1 a 2
"•la2 RI
+ -Z)+=" o<[ a~ (MI°%2) +7 Z • M2 I
a 1
M 2 0 a0 2 0 0 a
01I - Nl°l a2 + N1 2 P2 ° 2 I+ at 2 (M2 Ol)
S0 2) M al o0 0 0 10
+ --
C22 ý2 (M 1I 2a ) cr2 a ý 22= N2 P 2 oI- N 21
0 1 o ++q &, 2 = o
(1. 25-30c)
0
u° or azN?+-
or a? o(N 1 +- - )
1R2
(1. 25-31)
oo [r (a 2 M) + 3(M 0 ) 2 a&2
+ a ( r M o) - N1 o0 1a2 - N120
(a ) 23 100a
0 0
1 aw orM M
1 l
112
A
2. And at t2 = to and 2, initial and final boundary curves for
M21
0
u or a, (N 2 1 ° +
0
0 M2
vor a 1 (N 2 +--- ) (1.25-32)
0 P2° a, Opl ]N
"_a1 N
or -M1
a 2 a 2z
to Eqs. 1.25-29. This fact could also be seen if 0 and Po are of same
Po would vanish from the above and the classical equilibrium equations
would result.
113
1. 26 NONLINEAR EQUATIONS FOR CYLINDRICAL SHELLS
twist become the same as for flat plates. Although the simplifications
have formed the basis of the nonlinear analyses that appear in the
ax a
BN 0 8
ax a.
114
a Mx+ 2 2m NO
X+.2 M~+
-x a _
NO (Nx
__
e) a(Nx (1.26-33)
3XZ ax as asZ ) R ax ax
as as
s =Re
as = Ra0
The elastic relations for the Donnell form of the equations are
Et
Nx=- (Ex+~+ )
Et
NO - ( Ce+.p •x)
Et
Nx 2x(1 + p) Yxe (1. 26-34)
Mx D(Kx +• KO )
Me = D(K 8 + K x)
•
MMxo=ex• D(M
- 2 x
x ax
Bu ()z
2 ax
8v w 1 )2
' -s +R-+7-
115 (0_wss
115
Yx= -au + av
-- +-
+w aw
as ax ax as
4x-aw
lx=as
(1.26-35)
0 = aw
a%x
4,=- O~w
K
x x?.
82w
O2
282w
FIXO
o~ axa
(D/RZ)V~w
24 +N + Nx---
a +wZNxe aa--ea•T
+ + NN aZw
O 2
(D/)V axaG(N 0 ----
where
4 a4() 2 a4( ) 4( )
;x4 + 2 +x0
+ 4
116
Eqs. 1. 26-36 may. of course, be explicitly expressed in terms of
Within the limitation of their accuracy they determine all linear and non-
117
1.27 SUMMARY
energy was introduced and the use of variational techniques for obtaining
118
1. 30 STABILITY THEORY OF SHELLS
1.31 INTRODUCTION
119
shells of arbitrary shape were presented in the previous section. Other
for the buckling of shells have for the most part been restricted to only two
shell shapes, the cylinder and the shallow spherical cap. The problem of the
plotting all equilibrium paths given by the solutions to the nonlinear equi-
librium equations, and observing the lowest load at which large lateral
only those particular points of the paths at which the equilibrium changes
That is, assume that some load on the shell is the critical load; then,
120
equilibrium exist. Two equivalent methods of the theory of elastic stability
have been developed for determining the critical loads: the adjacent equi-
librium theory and the minimum potential energy theorem (Ref. 1-9). The
use of these methods for obtaining the shell stability equations will be
121
1.32 CONCEPT OF STABILITY
32
of illustration, consider the shallow spherical shell, shown in Fig. 1. -la,
clamped at the outer edge and subjected to uniform external pressure. The
load displacement curves for this model are given in Fig. 1.32-lb. The
represent the actual physical solutions. Each position plotted in Fig. 1.32-b
librium path.
and the load displacement curve is a straight line, as shown by dashed line
OL in Fig. 1.32-lb. Since yielding and large rotations are excluded, the line
122
FIG. 1.32-1
VIEW ()
LC
PA
&
VIEW (b)
exist for nonzero values of the load. A bifurcation point is illustrated by
D' in Fig. 1, 2-lb. The second path can slope upward, downward, or
In general, the rotations of the elements of the shell effect the con-
ditions of equilibrium. When this effect is taken into account, the conditions
in Fig. 1.32-lb.
load P is not unique unless P > PA or P < PB" The behavior of the shell
123
displacement A (if this were possible), the equilibrium path is followed over
its entire length, with the load P undergoing an appropriate decrease from A
the value PA causes the shell to jump from point A to point C without follow-
ing the path. This large increase in displacement from a small increase in
applied load is termed buckling. The load PA is called the critical load.
For some shell configurations it is possible that there are two (or
two branches are called branch points or bifurcation points. They are
During buckling the shell can move along a nonequilibrium path from D to
of the curve at the point is positive, zero, or negative. For example, points
between 0 and A and between B and C on the line OABC represent stable
configurations.
124
1. 33 ADJACENT EQUILIBRIUM METHOD
stability of shells.
the initial equilibrium position which becomes unstable when the critical or
u = u + k)i
v= vo .+ VV (1.33-1)
w= w* + X
pendent of shell coordinates. Thus, M!, XV, Mv are the incremental displace-
ments to which positions on a shell shift from the initial configuration to the
125
Now apply the nonlinear equations of equilibrium derived in the
shell. With the observation that u*, v°, we are solutions to the nonlinear
Eq. 1.25-29 with the omission of terms containing the factor x to a degree
higher than the second results in the following stability equations for the
shell:
a 2 2 _I _
+ -- ' ; N - F
N17 - i N; - Jz N; 2 ) + a•m•,p -0
-1 ____a -2 O
(N + 1 12 2
T42 _Nalz 14 82 ]
-- + --- I2a2)
a --. 8 (.3-b
+ R t at a at, M -
Gal 12 t2-
--- 2 (-NP;."• - N2 - N`.,) + a a o
126
1a +[;+
+ T, +
aar
222
at 2[ 2a at•2 a2 2a 2a 3 l
theory discussed earlier (Section 1.26), the above stability equations for
8-+ = 0
Ox 88
(1.33-3)
ONe8 -87 o
8x 8s
127
D V4 W +-+\N
gx--z ax2 (1.33-3 cont)
+~~
ez ax?-- axas
0 )E;
+2 INo +1(N* -+
-+ -8w.% -Nw =0
xaxs ax 8s O Gas 2 as2
where
(1. 33 -4a)
N Et_ •+ (awx--)
w"+ u__
N•~
=-lý - -- X)\.2\x
Et
E [Bu* ave IWaw*)w.
Nxe= + - +2
+
2(1 + F) I ax as ax as
displacements are
Et + aw- 2+" Ow
x l 8
ox2 x ax +71:7 R as Os
N Et r •+OV
a vw w° aW .O 1
128
When Eq. 1.33-4 is substituted into Eq. 1.33-3, the latter represents
on u°, v°, and w° the stability equations permit the determination of critical
ever, in general, u°, v*, w are solutions of the original nonlinear set;
theory to the initial position of equilibrium such that u°, v°, w- are
equilibrium are of the same order as strains. Thus, the coefficients u°,
v , w" are given by the solution of the linear equilibrium equations instead
of the nonlinear ones. This restriction of u°, v°, w° to points along the
buckled state, then for consistency, all terms explicitly containing prebuckled
129
If these terms are neglected, the stability equations (1.33-3) reduce
to the form
aNx aX0
-- + - 0
aNx aW
where
Et 91 + IiI- W
xN 1 a'x as
.S
SL g+ P' (1.33-6)
=Et [a1 - ]
xe Z(1+L) s +IOx
130
A more familiar uncoupled form is obtained by suitable differentiation
3- 3-
RV4 +
0 3 2
44-3- Ow Ow•
3ý-.
RV v = -(2+p.) -- - (1.33-7)
ax2Cs 6s3
2- 2-1
Et a 4 w r-
a 2-w 8w a NV
DV8 N + +V4 Ix -
N" + 2NOx0 - + N Z =0
Ox4 Ox2 0xas Os
R12
Thus, for linear stability theory, the prebuckled displacements u°, v°,
These quantities are known functions of shell coordinates x and 0 and of the
magnitude of the applied load. By substituting these functions into Eq. 1. 33-7,
and W.
The resulting system has a nontrivial solution only for certain definite
131
The boundary conditions prescribed must be satisfied at both the
elastic stability and reduces the problem of finding critical loads to that of
132
1.34 ENERGY METHOD
relative minimum for stable configurations along the equilibrium paths but
acterizing a stable configuration. The critical load on the shell can thus
be defined as the lowest load at which the total potential energy ceases to
is stable only if the second variation of the total potential energy is positive
definite.
133
The second variation expression using results of Section 1. 24 is
given by
A A
91 •z t/2
V2
=
•1910
i 2 -t/2
1/1~ e1 +2 2+ 12 121
e + r" 0 e- + T e 1 aa td
+2 2 121a
12d d•, (1.34-9)
The strain and stress equations (1.25-18 and -21, respectively) can
ment functions uO, v°, and w° and admissible increments U, 7, and W. The
134
appropriate integration by parts of Eq. 1.34-9, noting arbitrariness of
the increments"i, ', and ;, would result in the identical stability equations
shells can be obtained by substituting strain terms (Eq. 1.25-19) into the
1° and P20' in the expression (Eq. 1.34-9) results in the second variation
of the potential energy of the form (integrating over the thickness):
V2 = EtR 2+
2. I~a _L
+F + -F+J iC
0 0
2-87 2 /V2
2 ýa
ýi
)x) Ja
80x aa/
x 80 \XOU/J
0X .Ž + N 0( - + 2N O. -J dx dO
Et ax /;
) 0 56)
exO x8
where
135
N 0 Et o
x + * 'O) etc;
According to the Treffty condition (Refs. 1-24 and 1-55), the limit of
the Euler equations for the integral, where the variations are taken with
in Eq. 1.34-10 again yields the Donnell stability equations (Eq. 1.33-3).
In summary, all linear and nonlinear equilibrium paths for shells can
mination of only those particular points along the path at which the equi-
determine such points along the linear elasticity path. To illustrate the
significance of these various equations, two kinds of analyses for the axially
136
nonlinear equilibrium paths are determined by approximate solutions of the
value s.
137
1. 35 "CLASSICAL" BUCKLING ANALYSIS
Previously, the shell stability equations were obtained from the non-
solutions for the initial or prebuckled state with loading magnitudes implicit
in the solution. The critical or buckling loads can be obtained from these
o 0 u (x)
0 (1.35-11)
w° w° (x)
138
, 1 I I 1
a1x axQ
-3 + - =0
ax )s
8Nx aNO
+ -- = 0 (1.35-12)
ax as
4N 2.... o o 2'
DV ,+ + (No + a 2 Nw
I)+ No = 0
x ax 2 ax X s2
0
aN x
= 0
ax
N0
4 -•w+--R-+
40o N 0 2 0
aw (1.35-13)
D N ---- = 0
x4 x 2
applied only at the end of the cylinder, the nonlinear equations become
0
N = constant = + P (applied compressive end load) (1. 35-14a)
and
0
N0 - Et-- - ýP (1. 35-14b)
R1
139
In the classical buckling problem, it is assumed that the cylinder
w = +--- R (1.35-15)
Et
and
NO 0- 0
special case, the solution to the nonlinear and linear equilibrium paths
8- 2 E 4v
the associated bifurcation points along the nonlinear and linear equilibrium
path.
140
A solution of Eq. 1. 35-16 that satisfies the boundary conditions of
nwrx
w = A sin f---cos m-K
s(I35 (1.35-17)
7
Substitution of Eq. 1. 35-17 into the stability equation (Eq. 1. 35-16), yields
the following stability criteria for the nontrivial solution, after appropriate
P(Nx) - Et(
c- 2 (1.35-18)
especially for lohg cylinders. If the cylinder shell has edge restraint
throughout its loaded configuration, solution of Eq. 1. 35-13 will not result
can be used in obtaining classical linear buckling loads for spherical and
141
1. 36 DISCREPANCY BETWEEN THEORETICAL AND EXPERIMENTAL
RESULTS
The solution of the linear stability equation Eq. (1. 35-16) for a
(1.36-19)
1, o. 606 E-
This value is called the "classical buckling stress. " It represents the
test data (Ref. 1-56) for axially compressed cylinders in Fig. 1.36-2. As
shells. Similar discrepancies can be observed for other shell shapes and
discrepancy.
by Von Kgrm~n and Tsien (Ref. 1-57) who first used nonlinear shell
equilibrium path that branches off at the linear theory bifurcation point.
142
1.0
CL
0 I I
0 1.0 2.0 .3.0 4.0
ACL
CL =0.606
1.0
0 BUCHY
4b- CLARK & HOLT
0.6- <<, DONNELL
A
S*
CR ¶ 10 FUNG & tECHLER
"UL HARRIS, ET AL
0.4 - "0 KANEMITSU & NOJIMA
00
A .0 10 LUNDQUIST
0.2 A ROBERTSON
143
the potential energy expression and a Raleigh-Ritz approach rather than by
w : a00 + a11 cos (mvrx) cos (niTQ) + a20 cos (2mvx) + a 0 2 cos (2nrQ)
(1. 36-20)
where m and n are wavelength parameters and the aij's are constants.
aV
8- = 0 (1. 36-21)
Baij
buckling load occurred. The results obtained by Von K.rman and Tsien are
in Fig. 1.36-4 is the result of critical load obtained by Kempner (Ref. 1-58);
was hoped that a minimum postbuckling load could be established from such
144
1.0
C ,
A
ACL
1.0
CURVE A
P
CURVE II
145
Stanford University (Ref. 1-60), postbuckling equilibrium configurations
this approach.
it did not change the bifurcation point load. Tests show that the cylinder
without passing through the linear theory bifurcation point. There has
been much speculation concerning the cause of this cylinder jump and the
cause of the jump. (Ref. 1-57.) Von K~rman and Tsien showed that the
conducted since the work of Von Karm<n and Tsien have strongly
indicated that accidental disturbances are not the major reason for the
146
Von Kdrm~n and Tsien also suggested that imperfections whichwere
cylinder, might cause a roundoff of the sharp peak between the linear
aN X NX9
ax +x 0
N+ aN - 0
ax C) (1.36-22)
DV w+.--
DN + N a2(w* +2 w) + 2Nx 2(w* + w) + No 82(w* + w)
R x 2 xQ axOg 9 a
A systematic study of various initial shapes will firmly establish the role
147
of initial imperfections in buckling analysis. A general evaluation of various
values of w" has not been performed to date due to the mathematical diffi-
In 1950, Donnell and Wan (Ref. 1-62) presented analysis with con-
w* = K-1
2 (1.36-23)
particular initial shape but probably does represent one of the most
attempting to solve Eqs. 1. 36-22, Donnell and Wan used the correspond-
ing potential energy expression and the Raleigh-Ritz procedure for their
148
/
GEOMETRICALLY PERFECT CYLINDER
01C L ! .0
initial shape.
shape was reported by Koiter in Ref. 1-63. The solution of the nonlinear
149
differential equations with constant coefficients. Eq. 1. 36-22, with
the faorm
.loiter analysis are shown in Fig. 1. 36-6 for the case where p was
per cent of classical) have been obtained when initial imperfections were care-
fully minimized. Brush and Almroth (Ref. 1 -59) were able to repeat buckling
loads in their tests for a given cylinder, but the results were quite different for
ditferent cylinders. The results of these tests and the analytical results
150
PCR 10
IMPERFECTION PARAMETER, o
the influence of edge effects was suggested by Stein (Ref. 1-64) and
buckling and axial forces in the cylinder at the ends are eccentric relative
151
considered, the theoretical prebuckling equilibrium becomes nonlinear.
Stein and Fischer sought a bifurcation point along this nonlinear path.
The approaches were quite similar except for differences in the in-plane
the boundary, while Fischer assumed the perhaps more likely restraint
(Ref. 1-67) used boundary conditions similar to Stein and found a bifurca-
tion point at a relatively low load along the linear elasticity path. His
results are also shown in Fig. 1. 36-7. Hoff also reported lower buckling
152
loads based on linear stability theory in Ref. 1-68. All these results
show that the edge effect can be a significant factor in reducing the
been observed for other loading conditions and shell shapes. For
values (e. g., Ref. 1-7Z), for the case of a spherical cap under uniform
"CLASSICAL- VALUE,
1.0 LINEAR STABILITY THEORY
FISCHER (65)
P OHIRA (67)
STEIN (64)
I CL
153
1.5
,BUDIANSKY
1.0 n-
pero 3 4 5 6 7 8 9 10 11 12
0.5 A A 0
&A 0 &A
A 0
EXPERIMENT £0 0 0
A KAPLAN AND FUNG 0
0 HOMEWOOD st al 0
0 I I
5 10 1 15 20
values and test data. The theoretical results shown for axisymmetric
buckling were obtained by Budiansky (Ref. 1-69); and results shown for
154
1.37 SUMMARY AND CONCLUSIONS
The static nonlinear stability theory of thin shells has been discussed
has been little agreement between theoretical and experimental results for
boundary conditions and in the shape of the shell yield drastic reduction in
critical loads. For the researcher, these results indicate that he must
tion of a problem in terms of this theory and exact solution of the equations
155
linear stability theory, but at present it is not a design tool for direct
analysis and experimentation in order that the shell structures for which
the results were obtained apply to the structures for which he is designing.
156
REFERENCES
1-? Love, A.E. H. Phil. Trans. Royal Soc., Vol. 17A (1888) pp. 491-546.
1-4 NMighdj, P.M. "A Survey of Recent Progress in the Theory of Elastic
pp. 365-368.
(1959).
157
1-11 Reissner, E. "A New Derivation of the Equations for the Deforma-
1-13 Naghdi, P.M. "On the Theory of Thin Elastic Shells," Quar. Jour.
(March 1949).
1-17 Budiansky, B. and J.L. Sanders, Jr. "On the Best First Order
1-18 FlUgge, W., "Statik und Dynamik der Schalen, " Berlin, Germany:
Springer-Verlag (1957).
pp. 103-152.
158
1-20 Biezene, C. B. and R. Grammel. Technische Dynamik. Berlin:
1-21 Lure, A.I. "The General Theory of Thin Elastic Shell, " Publ. Mat.
Bending of Elastic Plates, " Jour. Appl. Mech., Vol. 12, No. 2
(June 1945).
Festechr. (1912).
159
1-29 Meissner, E. "Uber Elastizit•t und Festigkeit dunner Schalen,"
Viertelschr. d. natur. Ges. , Bd 60, Zurich (1915).
1-30 Reissner, E. On the Theory of Thin Elastic Shells. H. Reissner
Anniversary Volume (1949), pp. 231-247.
1-31 Clark, R.A. "On the Theory of Thin Elastic Toroidal Shells,"
Journal of Mathematics and Physics, Vol. 29 (1950) pp. 146-178.
1-32 Naghdi, P.M., and C. N. De Silva. "beformation of Elastic
Ellipsoidal Shells of Revolution, " Proc. 2nd U.S. National Congress
pp. 396-398.
160
1-40 Hetenyi, M. "Spherical Shells Subjected to Axial Symmetrical
Bending, " Intern. Assoc. of Bridge and Struct. Engr., Pub., Vol. 5,
Press (1946).
Press (1961).
Verlag (1962).
161
1-50 Donnell, L.H. "A New Theory for the Buckling of Thin Cylindrical
(March 1960).
1-53 Sanders, Lyell J., Jr. Nonlinear Theories for Thin Shells.
pp. 115-168.
162
1-57 Von K~rm~n, Th and H.S. Tsien. "The Buckling of Thin Cylindrical
Shells Under Axial Compression, " J. Aero. Sci. , Vol. 8 (June 1941),
pp. 303-312.
163
1-65 Fischer, G. "Uber den Einfluss der Gelenkigen Lagerung auf die
pp. 111-119.
(Jan. 1966).
Tokyo (1961).
164
1-73 Parameter, R. R. "The Buckling of Clamped Shallow Spherical
1-76 Cox, H. L. The Buckling of Plates and Shells, The Macmillan Co.,
165
2.00 PROCEDURES FOR STATIC ANALYSIS
OF SHELL STRUCTURES
2. 10 INTRODUCTION
GENERAL
solve shell problems. Chapter 1. 00 defined the structural shell and several
shell theories, with their limitations and ramifications. It was pointed out
loadings, and other factors all play a role in establishing which theory is
different approaches even though they fell into the same thin shell theory.
are certain restraining conditions, called edge restraints, that the solution
must satisfy. The edge restraints are reduced to unit loads and, by making
the solution of the differential equations satisfy these unit edge restraints.
the influence coefficients for the geometry are obtained. These influence
166
coefficients, etc., are then used to solve problems that involve determining
induced loads are included in the equilibrium equations and will be part of the
"primary solutions" and solutions for the unit edge restraints are called
"secondary solutions."
shown in Fig. 2.10-1, having the characteristics of (a) nonshallow thin shell
theory is that a normal plane section before bending remains a normal plane
that the bending moment exists only in the neighborhood of the edge of the
167
SEGMENT I Novozhilov (Ref. 2-1)
does not exist, the so-called "thick" shell theory may have to be used. The
division into thin and thick shells is still artificial and arbitrary unless
those values which are negligible in comparison to unity are defined. For
then the range of thin monocoque shells will generally be dictated by the
relation t/R < 1/20. The great majority of monocoque shells commonly
used in practice are in the 1/1000 < t/R < 1/50 range which means that they
belong to the thin shell family. However, as was noted above, the division
into thin and thick shells is arbitrary and depends on degree of accuracy that
permissible, the theory of thin shells can be used with caution even where
168
be small in comparison to the wall thickness). Loads and material
restrictions are such that the laws of the linear theory of elasticity are
applicable.
ensure that the deflection is small in comparison with the wall thickness.
MEMBRANE SOLUTION
The membrane theory for shells assumes that the basic resistance of
the loading; with a change or redistribution of the loading, its shape immedi-
ately will change to allow response to the loading with tension stresses only.
which found its equilibrium position under the illustrated loading condition.
If this deflected membrane could be made rigid by some technique and the
loading reversed, as shown in Fig. 2.10-3, then the loading will be resisted
169
LB/IN.2
(TYPICAL FOR ENTIRE
RFACE)
LB/IN.
(TYPICAL FOR ENTIRE
CATEGORY)
2
LB/IN. (TYPICAL FOR ENTIRE SURFACE)
170
The thin shell must be capable of resisting both tension and compres-
sion. In reality, the similarity between the thin shell and membrane is
shells, the effects of edge moments and shears are usually localized in the
region immediately adjacent to the boundary. (In shallow shells the edge
effects and concentrated loads are felt throughout the entire shell. )
Consequently, the shell, unlike the membrane, will also be able to resist
some moments, but this resistance is not the prime function of the ideal
shell.
analogies can be given: A plate supported along the edges and loaded
lent of a beam supported at the ends and loaded perpendicularly to the beam
axis. In this case the plate, like the beam, resists loads by two-dimensional
manner to the plate, except that for the plate, bending is the main
171
UNIT EDGE LOADING METHOD OF SOLUTION
sequent sections, are unit moment, unit shear, and unit in-plane edge loads
Fig. 2. 10-1), to get the solution of a complicated shell. This method has
for which solutions are generally known, and then piecing the solutions
172
2. 20 GENERAL APPROACH OF UNIT-LOADING METHOD
for application in the unit loading method; then, the interaction process
173
2.21 INTRODUCTION
internal forces of the shell are determined from six equations of equilibrium,
which are derived from the three force and three moment equilibrium
conditions.
There are ten unknowns that make the problem internally statically
If reactions to the applied loading can be found with the help of known
joints), all joints are welded or riveted together. This introduces the
simplified but accurate procedure called the unit loading method. This is
174
displacement, and rotations at the junctions. The following paragraphs
175
2. 22 MEMBRANE AND BENDING THEORIES
shown how both theories can be combined to achieve a simpler and more
bypassed with the help of an approximation method that can lead to useful
results for most cases in practice. This method is called the membrane
theory. Its justification and success are closely connected with the interplay
(two bending moments, two torsional moments, two normal shears, two
in-plane shears, and two in-plane loads), only four are of any significance.
The membrane theory is based on the assumption that only two in-
plane shears and two in-plane loads are significant. This theory neglects
shown in Fig. 2.22-1. It is noted in Fig. 2. 22-2 that the concentrated loads
176
FIG. 2.22-1. Boundary Conditions FIG. 2.22-2. Disequilibrium Due to
Compatible With the Membrane Concentrated Load
Theory
normal to the middle surface are not compatible with the membrane theory
bending, and twisting are considered, is more general and exact than the
177
2. 22. 3 Comparison of Membrane and Bending Theories for Nonshallow
Shells
compare the two theories, assume a nonshallow spherical shell with some
axisymmetrical loading and built in along the edges. When the results are
2. Except for the strip along the boundary, all bending moments,
membrane solution.
178
2. 22. 4 Combined Bending and Membrane Theory
Since the bending and membrane theories give practically the same
results except for a strip adjacent to the boundary, the simple membrane
theory can be used; then, at the edges, the influences (moments and shears)
can be applied to bring the displaced - Ige of the shell into the position pre-
scribed by boundary conditions. The bending theory is used for this operation
leading to final formulas. Consequently, once the solutions are obtained, they
can be used later without any special derivation. The results obtained from
application of both theories can be superimposed, which will lead to the final
results being almost identical to those obtained by using the exact bending
theory.
179
2. Apply the following edge loadings:
following:
or as secondary solutions.
180
determine the correct amount of corrective loadings M, Q, and
determined.
181
2.23 INTERACTION BETWEEN SHELLS OF VARIOUS GEOMETRIES
that mechanically interact with each other. A shell can be regarded as one
vehicles contain bulkhead and cylinder combinations; both shells are built
ship exists for each of these shell elements. Analytical methods are
inte raction.
simple elements. Very often the combination of shells and rings must be
transition and conical shell. By analysis of such a shell, the analyst must
required.
182
NOTE: SHELL THEORY DOES
NOT APPLY HERE.
CONICAL
'BULKHEAD
SPHERICAL OR
___TOROIDAL
CYLINDRICAL
The loading can vary considerably too. The most frequently used loadings
183
of interaction of several elements, as is usually the case if the combined
will not be subdivided into separate portions in the great majority of cases.
separated into two main parts: the cylindrical shell and dome, as shown
6c 6d 4d
c
184
Consequently, there exists the discontinuity:
(a) in displacement 6 - 6d
(b) in slope PC - d
6 I 6 and 6,C MC
c' QC Mc M
The corresponding values for the dome for the same unit loadings will be:
in Fig. 2.23-3.
c+ Q+ mc + M(Pd)M P
Assume that all coefficients 6 and 0 are known. For any shell geometry and
185
0 6 d M6d
C> Q= 1
~Pd
-
Q1
j>M~
MP
(A LL
A ROUND)
QPC MPC
Q6 + Qdd =6 Q
Qc P + Q Pd dQ
; QC =P
8 (2.23-2)
M6 c + M 6 d = 6
M ; MPc + MPd M
6c-6d =6 c" Pd P
case of interaction of a cylindrical shell and a dome (not toroidal) with the
because
PC =0 and Pd = 0
186
Finally, Eq. 2. 23-1 is reduced to a system of two equations with the two
unknowns,Q and M:
PQQ + OM M = 6
f3QQ+ 8BMM=.6
6
6Q M 6M 6 6
D- ,D 1 = 2=
PQ M PPM PQP
D D
It is noted that one cut through the shell lead to two algebraic equations .
of the shell.
adopted if it does not conflict with the logic and is used consistently.
187
It is noted that in addition to M and Q, there is an axial force distri-
buted around the junction between the cylinder and dome (reaction of
bulkhead), but the effect of this force on the displacement, due to M and Q,
is negligible.
further explanation: If the shape of the bulkhead is such that its meridian
of a basic shape. This is shown in Fig. 2.23-4, which has two imaginary
cuts and the compound shell consists of three elementary shells: circular,
toroidal, and cylindrical. Fig. 2. 23-4 illustrates also the loading and
188
SPHERICA L
TOROIDAL SHELL SHELL
M 2 M2 SELL
""MI
Q _., . p G9
M pQ
r(INTERNAL
pTYP. ELEMENT
PRESSURIZATION LB/IN. 2)
CYLINDRICAL SHELL
The symbols used for the two successive cuts m and n are also shown
in Fig. 2.23-4.
MMnn
6n Q6nn = horizontal displacement due to the same
respectively
189
Indicating n = 1 and m = 2, the above nomenclature can be considered
Fig. 2.23-4.
in Fig. 2.23-4.
68
Ms Q6 s - horizontal displacements due to the same conditions
as above.
s (sphere).
subscripts c and s refer to the cylinder and sphere. The subscripts lt and
Now the equations for the total rotation and displacement can be
formed.
Spherical Shell:
6s M s M2 + Q6s Q2+ A sP
s = BS M2 + QPs Q ++s p
190
Toroidal Shell:
6 = 6 M + 6 0Q+ 6 M + 6 Q A
2t M622 2 Q 22 2 M621 1 Q 21 1 + 2tp
Cylindrical Shell:
C M.6cC M1 + QC + A p
3c =Mc M++
6s = 6 Pr
s 2t s = P2t
6 c 6it
i6 PcC Pitl5t
finally be obtained.
191
It is noted that two imaginary cuts lead to four equations with four
unknown s:
Ml, M 2 , Q 1 and Q.
can be obtained.
cients may be negative and, as such, would have been introduced in the
preceding equations.
2.23.4 Summary
set of simple elements and one way to perform the interaction process.
structural elements due to the primary loadings and the secondary loadings
192
2.24 CONCLUSION
the unit loadings method. The following section will be limited to mono-
193
2.30 MONOCOQUE SHELLS
2.31 INTRODUCTION
monocoque shells of revolution. Thin shells are considered and all loadings
made to collect and present as many existing solutions as possible within the
tions or unit edge loading solutions. The geometry of every shell being
194
2.32 PRIMARY SOLUTIONS
solutions.
shell problems.
The forces acting on the sides of shell element are denoted with
R 2 sin dý
0REVOLUTION R2 sin
C1,
R1k
FIG. 2.32-1. Shell Element Forces
of rotation
195
Z =Pr
direction is zero)
NO'
an = loads on meridional and circumferential side of differential
N
element
vertex or around the hole at the vertex. A solution of the following shape
can be given for such shells as mentioned above (axial symmetry), loaded
N No
_- +. = r . . . . (2.32-1)
where
N = 0
1 2zsin [fRIR2
1 (Pr cos(l•~ -p, sin ) sin0 dO + Cl
where the constant C represents the effect of loads which may be applied
defines the opening in the shell of the revolution. The resultant of these
196
forces will be 2rrC. If the shell were closed, such loading would degenerate
2rC = - P
P P
N =P and N + 2
2rR2 sin 0 2TR I sinf
These loads may always be treated as additive loads due to the loaded
opehiing at the vertex of the shell. If the shell is closed, other influences
197
2. 32. 2 Determination of Membrane Displacements
shells.
0
eo N0 - R 1NQ)
(2.32-2)
(N9 - VN6 )
where
E = Young's modulus
i. = Poisson's ratio
198
where
Sis the angle which locates any point on shell-middle surface along
u = sin • sin
w =u cot • - RZE0
S• w Ar = w sin @ + u cos@
F •U Y = -w cos @ +U sin@
hr "• ---
199
2. 32. 3 Any Shape of Meridian
200
0) 0
0 z-a >0 g
0 +3
N0 N NN
0H
W -
(/~0 +
N zi N
<~ ± ±
E-4
~~]I
A 0.
za- .~
020
2. 32. 4 Spherical Shells
Listed below are the loading cases under consideration, which are
to each loading case. The spherical shells, which satisfy the relation
cot Cos
sin
are not in the scope of this section and fall into the category of shallow shells.
In order that a membrane state of stress exist, the boundaries of the shell
must be free to rotate and deflect normal to the shell middle surface.
X =q sin 0. Y 0,
Z a q cos 0 -
202
2. Uniformly Distributed Loading over the Base (Fig. 2. 32-5)
ap
X p cos sin
Y 0
Z P Cos z P COS2
X p Cos and
P Cos
P
p specific weight
X Y = 0.
Z P if + R (1 - cos R (1-cos
Z= lf+R-
(1 -Cos P
203
4. Uniform Loading in z-direction (pressurization; Fig. 2. 32-7)
N-= Z
X =Y = 0
Z p
form for each loading case indicated for the closed and open spherical
204
w
z
X4:
to 0
Q 0
U 3I
w I
Fo le A
- 5 t~S N S S205
4
*S
* c Nj*
0
- 4.
20l
z- -6
ZO 0
1-4N
S 207
2.32.5 Conical Shells
considered.
of the shell it must be free to rotate and deflect normal to the middle shell
X = q sin a
So~
Y=0
Z q cos ao
g g
208
(b) Uniformly Distributed Loading over the Base (Fig. 2. 32-10)
X p cos a sin a
0 0
$1 z
2
000
2
is in lbs/in
p COS ao-p x
, -- p = specific weight of
f£ liquid
209
(d) Uniform Normal Pressure p (psi) (Fig. 2. 32-12)
z 0 00 Y
X
(e) Equally Distributed Loading Along the Opening Edge (Lantern Load)
p &0
a0 / 1,
FIG. 2. 32-13. Equally Distributed Loading
Along the Opening Edge
210
(f) Hydrostatic Pressure Over Portion of Shell (Fig. Z. 32-14)
h .....x•\ j xx 0
p = specific weight liquid
-z Y 0
hh
for (B)
tabular form (Table 2.32-4) for each loading case as indicated for
211
TABLE 2.32-4. MEMBRANE SOLUTIONS FOR CONICAL SHELL (REF. 2-2)
(SHEET 1 OF 4)
7/
(a) DEADWEIGHT
q o 0t€o20 COzo
si as
N lqxZ qj
Ac x qcot%
r0 Cos % cot C 0. -
a ia
At -p coo.cot.. (.%. . 1 coeo% cos%..T
""cot& [oa
+ .)cosa.*. 0°
P-, €Cot°o* coa. i1
(C:) HYDROSTATIC PRESSURE LOADIDR
NPo
5 .. .... m.)
_o.(W+.. [.(-..( -
.,.fo.(•.. Ii"-."'"[I.{').)
"( Il] .,..,•
,-,a"""I~[ •.
I'(•' ••
Et ,sin% sline, Tt
212
TABLE 2.32-4 (SHEET 2 OF 4)
NO p x cot . -p x cot e
"a
Cr 1 -p cot I
-3h cot 0)
213
/ TABLE 2.32-4 (SHEET 3 OF 4)
/f'
Open Conical Sholl (Supported) Open Conical Sh eu(hangod)
(a) DEADWEIGHT
Ng .q 0 0q-
in,%.in%
1 - 0,- 0
nz 12 aI\1
rxz
ZCto z / I
Y 1t cotI . - ki- ) b cot Z 1
NO
qxco
,a 2 J)OM& -I-P x -j1 Z(Z+I&ce q oos C002 I+
q.
- I ,1
-
01-
(b) UNIFORMLY DISTRIBUTED LOADING OVER THE BASE
No
N0
I ae
1a
a[I cot%21
CON -
Cola ] Com%C
0 '(na co es s 1b.
Ng -xca.. f x~
Cos (. L.
I
#02~~~~~~~ r f
aI1 I' F 0\'*~ ... L 0 0 21rr/\1 t .1 2 1
At.I X x\1;t~;f] x()"ICo.Zo~
0 .(.] +i).
214
TABLE 2.32-4 (SHEET 4 OF 4)
(e) EQUALLY DISTRIBUTED LOADING ALONG THE OPENING EDGE (LANTERN LOAD)
No 0 0
0 0
aim.
-I-s siniso
Nx •x I ft•
0 a Xi 1Pi
Xcoe 0Ict
215
2. 32. 6 Cylindrical Shells
stress may exist, the boundaries of the shell must be free to rotate and
0= CONSTRUCTION
tu
tI tX xd
t t WALL
I THICKNESS Ns
I x
I X(9) X
gx
-ZR L
216
x where o -5
L 1
B -~ Et
2
•/Rt
The following loadings will be considered:
pv
(Fig. 2.32-17)
217
p P0
P0
Cos
40
Pokp
Z = p( ) = -po exp(-ag)
FIG. 2. 32-18.
L
Pi pi ex
-iPiea-)
2.i•
Exponential
. 1 obtainable as in previous cases.
Loading Case
WALL THICKNESS X = x ( (1 -
L X DISTRIBUTION
of X
IL
218
5. The constant function Y in circumferential direction (Fig. 2.32-20).
e
Y = Py
Y"•__
*W
Table 2. 32-5.
219
N -
0.
V
--. 0. N .
a -IN
0 N
0 N
•.:z. o. •,,.1-
- 0 0
CL>
00
-I V
0 t04
U 220
+ 0
a
.4 S
o Ni
L- .. r e;N
fn -L
0 0
w -u Lj +1
± I,~.-.
4 .+ .
- -~ a NC4
N Zo .-.
0 .
uC
IX.
221
2. 32. 7 Elliptical Shell
Table 2. 32-6 and are obtained with membrane theory (small deflections)
222
0
-rU)- +
-1L
A
2 U•N , *.I ..-: -
,, ... +
I
a.• .
S• z= * N
0 NNN a I +
0 + N IN4
NIO.
S NN[,
N3
NjN a
C6 -
CN NN
CL
N NNN
++ 0
N NA N
'.A.
.0 0
E-
Nd
IN N4
Dicayor
223*
CD'o 00
D.
'd . ý 4
0 0
00 00-
00_
- -Hti -
0NN
1-4-
0 0;0?
ýD c.:)224
FIG. 2. 32-23 DISPLACEMENT AND ROTATION PARAMETERS FOR
ELLIPSOIDAL SHELLS UNDER UNIFORM PRESSURE
- r0.2 0.4
Aral 0 0r iAI;:L
AA
4A -0.4
ap 2 a/b 11.0
0r Ar -1.2
~~0 0.2 0.4 . 0.8 1.6. .
r/a
225
FIG. 2. 32-24. DISPLACEMENT PARAMETER FOR ELLIPSOIDAL
SHELLS UNDER UNIFORM PRESSURE
5.0
4. 5
3.5
I a I/b =Z. 0
a 0
3.0 --
a/b =1. 6 \
1.5 -
Sab =1Io4 • \
1.0 \
,a/b =1.2 - -
0.5 l "-"\
226
2. 32. 8 Cassini Shells
hoop forces of the boiler and the boiler drum may be avoided by choosing
where n is a number > 1, r and z are variables along r and z lines, and
shell thickness must not be present. The only loading that will be con-
sidered is pressurization.
227
The second loading case is plotted (Fig. 2. 32-26) in nondimensional
2. 32-28.
228
TABLE 2.32-7. CASSINI SHELLS-MEMBRANE SOLUTIONS
Case 0) Case 0
and n>l
Shape b 0. 743a
2 2 2z2 2 2 1/ 2
0 a2 2 +nz 2 z
F a +r
0
r
3n2 a4
e
- N 2 2- R2P -
(a2 - r2 + n Z2) 2 R
Fr2(a2 2
+ n zzZ)+
4
n .z(a
2
Z
r-2j
r pR 2sin 2 t
2Et
D
aZ Ar cot -fR 1 (N, - INe) - RZ(N, - ILN4,)
p Et sin* d
t
e mr 2Z 2 s in 4
2160
Z-
a K(3+4K)4
1/2
K I I -0.94(,)
229
FIG. 2.32-26 MEMBRANE FORCE PARAMETERS FOR CASSINI SHELLS
UNDER UNIFORM PRESSURE (CASE:)
1.0 - ------
0.9
0..8 •"
0.7- - - - - - - .........
0.6-
0.5 4 - - - - --. -
0.4
0.3
0.2
0.1-
0
0 0.2 0.4 0.6 0.8 1.0
rla
230
FIG. 2.32-27 DISPLACEMENT AND ROTATION PARAMETERS
FOR CASSINI DOMES UNDER UNIFORM PRESSURE (CASE 1)
3.0-P 1
Et
ap -
2.5-
f /Of
1.0- -
a p / --
- -
0.5k5 - I
-0.5 - /"
-- ,, ---- -- -- -
-1-- -• - -. +
¢ /a
-z
.
0: - - - - - - - -
S1231
FIG. Z. 32-28 DISPLACEMENT PARAMETER FOR CASSINIAN DOMES UNDER
UNIFORM PRESSURE (CASE 1)
0.35 ------------
0.30- -
0.25 ----
0 0...0.4.6 ' 0
a p
o0.15 /\
0 0. 0.:. . .
r/a
/ 232
23
2. 32. 9 Toroidal Shells
This section presents some known solutions for closed and open
indicated in the Tables 2. 32-8 and 2. 32-9. The solutions are based on
free to rotate and deflect normal to the shell middle surface. Abrupt
233
o o0
0"o
01u
u.10
+ +
So +
.F..
0 -
e a0 .0
2341 - 10
TABLE Z. 32-9. TOROIDAL SHELL (Ref. 2-4)
Loading N+ N6
Shell
Condition
"P".%
p Cos R (sin#o - mint 0 ) sin - (coo 0
- co. *) 4
4 (sin 4 - sin so) sin co. #1
+
~?~YYYI1Y p,
0 sin *Cos 4 R /so) '(cON
24 4 Z min# stn%
2
I' % .l~~P p=
P, p corn 4 -o p Z lb,#,
at.
sin? 4/
P. P. pcmin*
p Cos 6 OR co Cl-co.
aI -coo* +4 sino [
I~ *iin
2•:p~!hiiŽ
Z v(R
p 40 inf~ *0pR sin +(sin++nsin + sin *ot
(0.*.
04
2 sin,
tzp
pp p ,sin cog
R sin 4 +Zsi~n 0
nR0
07 ps p cogn *p 2 si#fsn# (cognZ. 2sin*4
2 Z
r p. p sin #
(b + R in52 ) min
bo - R 0 - co 0)
235
4
. , *0• -S004 -
OLI . WI . .0,
00
N
v 00 4 4,0 + 4
0 .0~. u +
NI IL
0 0 oo
S.. .. . . ...-
-jS1~I
X X SS + . @
*7 0 o
S.3{.
236. ,% , ,
2. 32. 10 Other Geometries of Shells
The formulas for the internal forces are obtained with linear mem-
defle-ct normal to the shell middle surface. The thickness of the shell is
237
It!~
.-. _ L. -. 4
- * N -- "
a v
m No. N
,-e1
0.0,+0 aa *I- F
0 X N - N
A a-
K- -' KIL
~ -. ~ 238
QAq4dh N4
.~ e0e)
I.0 IN I jN
2I - .J I
1 ~
Mj3 0. a. 0
+ - ..-
0
0.Iw
*239
2. 32. 11 Irregular Shell
purpose.
curve y (x) around the y-axis, as is shown on Fig. 2. 32-29. The following
I Y(
y
'0
AXIS OF EQUATOR
_0 1 ROTATION
240
The shell is loaded by a distributed load, p, per unit of horizontal pro-
p(x) = P0 x)
a
Ar - Et s
where
r = equator radius = a
E = modulus of elasticity
241
Poisson ratio, p., is assumed to be zero. For the load distribution not
2 n
Sp = 2 P(Xi)xiZx
por i=1
PO
xi
I~e. X
FIG. Z.32-30. Numerical Evaluation of Sp
Fig. 2. 32-31 is
242
ro=R b
ro =r7a
t L r
is a pressing element.
243
TABLE 2.32-11. SP FACTORS
Distribution of Equation of
Loading Intensity Sp Factor
Loading
ty
P-= P
lp Po
I
z
0P "P O 2]
SXP Po ( 1 ) +/
ty x P O
1o
y T
311
S
T
, --Po a-
)
ty
p 2
P =Po 33
2
K5
244
2. 32. 12 Conclusion
that any analyst may face in actual practice. The exact and approximate
following section.
245
2. 33 SECONDARY SOLUTIONS
M = 1 lb-in/in
Q = 1 lb/in
unit loadings. Influences of this nature are of local character and do not
progress very far into the shell from the disturbed edge.
ized with týma = 9O*; such bulkheads are very common in air space
vehicles and pressure vessels. The bulkhead shells are tangent to the
When the values of deformations due to the unit loadings are avail-
able, the deformations can, be entered together with the primary defor-
be determined.
246
introductory sections. Some conclusions will be discussed and the final
It has been mentioned that deflections and internal loads due to unit
due to edge-unit loadings will disappear completely for a 2' 20° and will
cal shell.
unit-edge loadings, practically all parts of the shell, satisfying the con-
dition oa 200, will remain unstressed and undisturbed. These parts will
N N M M0 r 6
ZOO _0__
L
,
M
247
not be needed for satisfying equilibrium. They do not effect the stresses
and deformations in disturbed zone 0 < a < Z0° in any way. We can delete
all material above a = 20° because this material does not contribute to the
stresses or strains, which are computed for the zone defined with o<a<20°
0< a<20° if we replace the removed material with any shape of shell, as
(A) M~ (C)
it is indifferent what shape the rest of the shell has (Fig. 2. 33-2).
248
which is loaded with V = Q = 1 all around the edges (Fig. 2. 33-3)
place of a = 20.
a 0 20
finding unit-influences.
249
with unit loadings) in equilibrium is shown in Fig. 2. 33-4. The nomen-
Fig. 2. 33-4. Due to deformation, point A was displaced to the position A'.
Ar = horizontal displacement ar r
250
The circumferential deformations are not independent. They are related
to the unit loading action. This will be done in the following presentation.
251
2. 33. 2 Spherical Shells (Open, Closed)
1
cot = (Ref. 2-2)
Physically, it means that for shallow shells, the disturbances due to unit
edge loadings will not die before reaching the apex. Consequently, from
in the shell thickness must not be present. Thickness of the shell must
The formulas are listed for closed and open spherical shells. Open
at the apex. The spherical segment must have such meridional length
that the disturbances due to the unit-edge loading will die or become
Unit-edge loadings may act at the lower or upper edge of the open shell.
For derivation of the formulas presented, linear bending theory was used.
252
The following designations will be used:
k= 4(-)2 3(1-FL); a= 1 -4
The table can be used for open shells, however, if the segment is such
that the influences due to the unit loadings will die before reaching the
used, for open shells too provided that a o - ZO'. Otherwise, the
253
TABLE 2.33-2. SPHERICAL SHELL (REF. 2-9)1
-2ke_•-w2k k•
+ P Mo
DEFORMATIONS
4 k$
2
Eto ZvZ" k min 4 1 • a "ka sin (kl + -) ) - -ka cog ka
4 R
FOR a 0 AND 4- 41
2s4ks
EtP
Etp --2k sin 41k
For 41 - 90*
t . ....
R
Et(Ar) 1 + 2k2
254
A. Open Spherical Shell, Unit Loading at Upper Edge
on the upper edge of an open shell. These cases can be reduced easily
to the previous case, as shown in Fig. 2. 33-7, and the same formulas
can be used, except that 41 1 >90*" The actual shell (Fig. 2. 33-6) will
be imagined turned at 180 degrees (Fig. 2. 33-7), and the shell can be
sed with functions Fi and Fi (a). These functions are tabulated in the
OPENING OPENING
R S R Q Qa iM
a
255
x
y Y
OPENING OPENING
256
6F,
0 1,7
P141
0 E4- ".•' 11. I
•.•_" • ,- .- - -. i .- -
25'7
TABLE 2. 33-4. OPEN SPHERICAL SHELL-UNIT EDGE LOADINGS
AT UPPER BOUNDARY 1
• ki
f -0 D k
Boundary Conditions
a (0 0 M+ 0
00=
(4=) 2 ) M,=0
i F9 ZF8 F
ki 2[FI Fu)]
kik 1 8 F1 9
Ar Me -H•2n•4[-
Hk mkin 20 [ F9' F +IF)•F W-F
cokF F a F
--
o1+)
(a) (a) "Ft F +) 91
4
Hki ZInE 2k F9 11 F F8uF 1l 10
r Hki Et IF 1 9s •-IF 10
1
Foc F factors, see paragraph 2.42.6
IFor k factors, see paragraph 2.42..6
258
TABLE 2. 33-4 (CONT)1
Boundary Conditions
Q= 0
Q0
M2 Mk F8 ()-F10 (a,)
+ - MkiI.F, a (I F 19 F1
259
,..
41 a
I IAIC"
z - -1
0 - - I'
<OOO -"I-
--
A ha IC
o a
al_.
.I
.h
- "h
--:-I, ;.
' .
1- -- --
AIC _ ±-__+
0 -6
B. Unit of Distortions
P ik = unit rotation
I k unit rotation
261
TABLE 2.33-6. CLOSED SPHERICAL SHELL SOLUTIONS DUE TO UNIT EDGE
LOADING/DEFORMATIONSI (REF. 2-2)
t . t
SikAPik
Boundary Conditions
Et coto Et2
N# rik Rk sin, 1 F 1 7 (a) Pik 2k cot+ F1 9 (a)
Et
Et
2
Q0, arik Et F1 () Plk 2k F1 9 (a)
Rk sin +I
Et
Mt )REt
M + Arik -2k sin #1 F2 0 (O -Pik 3 F17
smin F R sin+la
Hi
Boundary Conditions
M. -Mik Q-O M#
o-O #*
*1)
HikwQiksin*l +Nikcos#l
2k cot 'r) H i
N4 Mik R d1 cot 4 F0(
2
No Mik - 2k 20 (2 ) Hik 2k sin Oo F1? (o)
2k
00 Mik FIs
1 (a) Hik sin 41 F20 (a)
R siin *#
M Mik F19 (0) Hik FIB (a)
k
2 k2 sin+ 2Rk
Ar - Mik Et F20 (o) Hik "- sin
5 s'n#F17 (0)
3
0 k4k k
a- 7 (40) - Hjk 2k2 sin * I1F9 (a)
EtR Et
1
For F factors, see paragraph 2.42.6
For k factors, see paragraph 2. 33. 2:
263
TABLE 2.33-7. OPEN SPHERICAL SHELL-UNIT EDGE
DEFORMATIONS AT UPPER BOUNDARY
k "k
I
S... r .......
zp 2
J
]•onandary Condition
a = r %: •ki
(÷ : Z'): o=o0(• = z): p=0 •=o( 1: •~
ElcoIC Cole P.
R l (l fF •)in/alCF 9 aJE
Ar 1 R
3 -kcosCzFI
0 F7 C)-F
Pk
2
2ko4[F L-•- F7 (oj •-F F9 ~).
N V Ar= 0 1 al =F Ar = 8 -4( 0- (0
F1 tF R F r(1ces+ .) ln "k cOn Fn 9 F1 8
SRsin L 3 2
+ "PF
No0 r _ E El
_ Fl.F
_ [Fl
0 7ro
(* 8 F.E90 1 AVki t [F1 F9 (.)
[-Flo 0F 7(o) F- 28 F,
F9(.)]Fo(a)t Ok -l [ ,8 F9'
l7(&* (a)-
'in 4 [F V Fa F FIr
9) a - 2 F aFa.)IRsi FS F
A Ark1 in 7,1 0 9 42s F FSa) s~~ viI +0p) FR-k.o
8in 83 2 O ink* 7 F9 . T FS a
P rP1o F8 +.
4- Fl.(.)] 2k ki 1- 1 1 0 FS (a) 4 F8 Flo (.)I e a + F9 1F
-A (14+P) si.+2, F., - k Co.+,2 r 1 jr-F a8
rk1 1
i+
L0 AkR 7T 8 71
l~rF factors, see paragraph 2.42. 6
Fc, k factors, see paragraph 2. 33. 3(b)
264
2.33.3 Conical Shells
the case of the sphere, the disturbances due to unit-edge loadings will die
an undisturbed edge (or apex) due to unit loading influences on the respective
opposite edge.
of shell must be uniform in the range in which the stresses are present.
The formulas are assembled for closed and open conical shells.
Open conical shells are characterized by removal of the upper part above
If the height of the segment is less than V¶t-, the analyst is practically
dealing with a circular ring instead of the shell. The, following constants
are important:'
265
k • 3(1 2)i
4k- Sine)V3(
Et3
1Z (1 - L )
Additional designations are indicated on Fig. 2.33-8.
z6
I
XXr
stant. Table 2.33-8 presents the formulas for a closed conical shell.
266
TABLE 2.33-8. CONICAL SHELL- UNIT EDGE LOADING SOLUTIONS1
I , a
2
No 2RZ ksin + e-k co R2-R kZsin Z.e'k*coa (k +4 )
# k cos ka i2
DEFORMATIONS
FOR * 0
Ar PI cot1
3
ZDk In ZRk in#J at#1a.
2 Dksin*
2Dk sin0
1
For k factors, see paragraph 2. 33. 3
267
(a) Open Conical Shell-Unit Loading at Lower Edge
Since unit influences are not progressing very far from the edge
into the cone, the formulas presented in Table 2. 33-8 can be used for
FIG. 2. 33-9.
4M >
replaced with shell as per Fig 2. 33-10B, the result is a conical shell
M M
(A) (B)
268
loaded with unit loading at the lower edge. The same formulas are used,
also used for closed cone) is presented in Table 2. 33-9. These formulas
are expressed with the functions Fi and Fi Q), which are tabulated in
k- 3 (1 - ;
tx cot O
m 0
269
TABLE 2. 33-9. OPEN CONICAL SHELL - UNIT EDGE
LOADING SOLUTIONS' (REF. 2-2)
M (D
(t
M 2k cota[6 F 5()ot ]i M
A0 2 FI01
+ I FI6 Fl F
F Fo1 16 1
N+ F F F3 -F
F13 ()..1F10)] 10
M Zk[F F5 4 1 MiZk(
A()10 F)i F
1 0 6 8F1 (F
+ -Fr F 3 F 0 I0 ()
mki S" *F6 F14 () m 1- -0 F
+ - A--- 2DkF 20 F1 7
ZDk FI1
- F 1 3( -- •F 10 - - F 10
I F1 10
"F5 F3 ] "io (0
M id[ F6m Ik [~~ F9 (
DkF 6AD F F
F5 F 4 3 F l
71 is F1 7 J
1
For F factors, see paragraph 2.42.8
For k factors, see paragraph 2.33.3(b)
270
TABLE 2.33-9. (CONT)l
a =" Hki sin o N -H. cos a
I H
F9
Hik 0
0 Qx =H sio N =H
r2 [ F
2
"Hkix 2kcose 0 1 9 (9 os F -
() 7
N0 ZF 4 F2 HFF xm Cos ] CI F9
F F1 (7 F 1F1 F(1
-F 1(
Hki 2 0 1 () HA k[ 8F()
Mx
2F 4 +F F 2
F8
. 1 1 99 .28
-.--
Hs i 1 0 a0 +--F(l
F9 (F0 HF si8 ' l 7]
F1 10 2 'F8 8
.2sin 2 ar F
3 . ()H Li
Hki 4Dk 3I9 2Dk3
2F 4 F2 1 F
+-- 8 F 0 (4)
+ F 7 (o'- I F o (4)
Hki siH 2k
0 i -F [ 8 (•) 0 +•
- Hik~k sin(o 2,+ F
F9 )
ki2Dk L i 2Dk F19
- F9 ( 4 7 +
F 7
1
For F factors, see paragraph 2.42.6
For k factors, see paragraph 2.33.3(b)
271
In connection with some problems, it may be of interest to know
AArik
i unit displacement in horizontal direction
At lower boundary i
272
TABLE 2.33-10. OPEN CONICAL SHELL-UNIT EDGE
LOADING SOLUTIONS '(REF. 2-2)
3
4D1, cot *o Ark Fs
-F7 (V
3 is -- F 14 (s F
I-Fl +~ FS F (
5
+ r~
16-F6~ 1F 6
A? ~ ~ _ IU F* (t)
F53 (0n[F14
+-16 F13W
_F1 F, Fg F
sin % 1 Jl F3 _
1
For ~
F1facors seei
F paagap 2.2.
~
For ~se7 paa(ph2
~ r facors
3.)b +
273
TABLE 2.33-10. (CONT)'.
Pik IL®
÷74
I F (U].1
l)+ 9F (.)]
[1
2Dk
3
xm cot . O.ki [ r.2i rq (t)
3
4Dk xrn cot V. Pik [ .• '9 1(1
N*
274
+z ( r,)1
F 0
()91 + I
(Fa
[
Z jo It,
Dk Pj,.i -...A. ZDkp
Pik .2. rio (o)
+ Z-.F7
÷ It) + ]
(FF-( r9 r? (It
+ F1
2Dk Ok
2
2
i
11
FZ 7 It) 20kZ Pik Ir
i
-.rzP, 'q , r 9
FI F
.,..0[ 1 j FI
,. •
+ -2r a(. r, ()9 FS (9I
Phr
÷Z4IF (V) Pl0{'k {gF F. (I)I
"A z rs M ikr re
F, Fj (v + Fr (t)
0
(U]
274
2. 33. 4 Cylindrical Shell
This section presents the solutions for long and short cylinders,
loaded along the boundary with the unit-edge loadings (moment, shear,
the cylinder is less than x, then the analyst is dealing with a circular ring
S4 = 3(i - 2) / R2 t2
the length of the cylinder. (See paragraph 2. 32. 6.) The boundaries must
be free to rotate and deflect because of the action of the unit edge loadings.
The shell thickness must be uniform in the range where the stresses are
pre sent.
275
(a). Long Cylinders
L 4
k - 3(1l
D = Et 3
12 (1 - )
The formulas for the disturbances caused by the unit-edge loadings are
276
a +
z Ica
0A
00
JIC.
M --
u0
z
o0 4.
94
IC.
u. -
00.
U
aa
000
z". :5
27
(b) Short Cylinders with Uniform Wall Thickness Without Abrupt
Dis continuity
The following constants are used for tables 2. 33-12 and 2. 33-13:
4 = 3 (1 - 2L /RZ t
p = sinh 2L -sin 2L
278
TABLE 2.33-12. SHORT CYLINDRICAL SHELLS-UNIT EDGE
LOADING SOLUTIONS 1
Re-a~rks
0 0 0( k
Ix
Oi,(ihp.-mnO)cspxcs-2snz
.i'Lcop
inbpL
p. MkK
2 =D+ ihcopl2
hp.
coshP. coosP.+ &inox -K6 (CoshP
box ino
(I
;PT1 .Rfhop . c iZL - oahz 2
2
*i cosS p Kihk p c on hoh~
Pz-ai 2P i sco
*ibo ethos -IpK3 ceohpx cosop * Kj (coghpx. pohx ~ thxc~
Z -*
D o - fifthoi(cosoaII + Mn, Pl. coosh P saftn'x,
3
*il o 1'K 1 I2P D M ikK I ZZPD
me 3gw/a Etw.ft
Me Am i
2
(c o h(0hxconPx + K, in h Px aln O KI (aIn hpx P (KS pulnhpx slnop . 2 .lnh OL coo hox sinpu
t oop
Ph+coabox StAPul .I stlnpl. *inbox cosop)
1For
k factces. see paragraph 2. 33. 4(b)
279
TABLE 2.33-13. K-COEFFICIENTS (SHEET I OF 4)
PL K1 K2 K_
3 K4
K K5 K6
0. 100 20.0000 10.0000 300.0010 299.9993 6000.0726 5999.9726
0.102 19.6078 9.8038 288.3502 288.3484 5653.9797 5653.8777
0.104 19.2307 9.6153 277.3661 277.3643 5334.0202 5333.9163
0.106 18.8680 9.4340 267.0011 266.9992 5037.8137 5037.7076
0.108 18.5185 9.2592 257.2024 257.2004 4763.0654 4762.9574
0. 110 18.1818 9.0909 247.9354 247.9334 4507.9753 4507.8654
0.112 17.8572 8.9286 239.1601 239.1580 4270.7759 4270.6639
0.114 17.5438 8.7718 230.8407 230.8385 4049.8977 4049.7837
0. 116 17.2414 8.6206 222.9497 222.9474 3844.0227 3843.9067
0.118 16.9492 8.4745 215.4565 215.4542 3651.8686 3651.7506
0.120 16.6667 8.3333 208.3347 208.3323 3472.3100 3472.1900
0.122 16.3934 8.1967 201.5600 201.5576 3304.3290 3304.2070
0.124 16.1291 8.0645 195.1109 195.1083 3147.0173 3146.8933
0.126 15.8730 7.9364 188.9657 188.9631 2999.5240 2999.3979
0.128 15.6250 7.8124 183.1071 183.1044 2861.1180 2860.9900
0.130 15.3847 7.6923 177.5168 177.5140 2731.0977 2730.9677
0.135 14.8148 7.4074 164.6109 164.6079 2438.7537 2438.6187
0.140 14.2858 7.1428 153.0632 153.0600 2186.6934 2186.5533
0.145 13.7932 6.8965 142.6899 142.6864 1968.2148 1968.0698
0.150 13.3334 6.6666 133.3360 133.3323 1777.8949 1777.7449
0.155 12.9033 6.4515 124.8725 124.8685 1611.3412 1611.1862
0.160 12.5001 6.2499 117.1905 117.1862 1464.9675 1464.8076
0. 165 12.1213 6.0605 110.1957 110.1912 1335.7945 1335.6295
0.170 11.7648 5.8823 103.8095 103.8047 1221.3795 1221.2095
.0.175 11.4287 5.7142 97.9624 97.9573 1119.6639 1119.4889
0.180 11.1112 5.5555 92.5962 92.5908 1028.9436 1028.7636
0.185 10.8109 5.4053 87.6591 87.6534 947.7650 947.5800
0.190 10.5265 5.2631 83.1066 83.1005 874.9079 874.7179
0.195 10.2566 5.1281 78.8997 78.8933 809.3321 809.1371
0.20 10.0002 4.9999 75.0043 74.9976 750.1503 749.9503
0.21 9.5240 4.7618 68.0319 68.0246 648.0355 647.8255
0.22 9.0911 4.5453 61.9887 61.9806 563.6514 563.4314
0.23 8.6959 4.3477 56.7164 56.7076 493.3093 493.0794
0.24 8.3336 4.1665 52.0895 52.0799 434.2073 433.9673
0.25 8.0003 3.9998 48.0066 47.9962 384.1866 383.9366
280
TABLE 2.33-13. (CONT) (SHEET 2 OF 4)
PL KI K2 K3 K4 K K
281
TABLE 2.33-13. (CONT) (SHEET 3 OF 4)
PL KI K K3 K4 K5 K6
282
TABLE 2.33-13. (CONT) (SHEET 4 OF 4)
PL KI2 2 K3 K4 K5 K6
283
(c) Special Formulas
Hampe (Ref. 2-2) gives exact solutions for cylinders, based on linear
the edge ik = 1 and forced displacement of the edge Arik = 1. All for-
Tables 2.33-16 and 2.33-17. The functions, F. 1 and F.i (t), are from
Paragraph 2.42. 6.
284
UP4
a %
:D4 - . I-+
UP uLn
j~uLn
0IX4 %0-
14 -4~ Lr~-
N +
41L1
u ~ ~ t - 4 i .
0 0
44 44 0
I + Pj4
o< "-4 OD tS
zZ N1
-4 .- 4
+ N
IAP285
,
4A.0
44~
Ln~
~ALVA
In
-4 wO - up u
i AA
-E0
SN
ca.
Ca co Ca
Go~ ufi
4U44
U~114
UP4
W P4
c- -
*+
134NL
1 en
'U 1 '44C
'A .
++
N N.
A - - -
0l
- - i -S
tb
z cc.
286
H'-40 :pa~jsies aq Isn~~
oL
rz4-
4JI
Z UP
'.4n
41~P ,IA_ -4
N C;
-N 0 3c
C4 C
H ~ cito
287
UU
z Ln
1-4
co4
0 UP
4 CI t- P4
NN N P4
0- 31J
Pul
su ,4eTuz'a C0
Nalo 'vlau cmN I0
1
.
U.
'U
2881 1 . 1
2.33.5 Salvadori's Approximate Method for Irregular Shell
graph 2.32.11.
A. Displacements
to
Qo Mo
h •ri
Xi F3 D, Z PZ D,"
00 Mo
289
Ratio ts/a is assumed to be small enough for displacements,
ratio, ts/a, is less than 1/50 and whose thickness near the edge does
B. Interaction
MO = m. P ats (-)Sp
where
2
1 (1 - X )
0 4 2 2 3/2
24" (1 + X) + 2X (I + X)
x = t /tc
whe re
529
t is thickness of cylinder
C
290
C. The Maximum Values
q0
tan P xs = o '4
This distance is measured along the meridian, starting from the discon-
defined as follows:
tan _V =
q0 0
This work will be performed using Fig. 2.33-11 and the formulas
that are given on this graph for M and Q. To obtain M. or Mc, use the
formula.
291
mc m)
M' ,11po at
1_4 r0 LSL
-- - t /a \
0. 4X.03
1.7
1.6. .0.10.
1. 5.o.3. -• V-' - •= " •" -
1.4. 3.02 '--"• •
1.3.
0.9 0.1 1
0.8.
0.7.N
0. 6 0 0 0.
0 _
0.5 1.5
292
2.33.6 Conclusion
follows:
293
2. 34 SUMMARY
process between shells was discussed at the beginning, and then the pri-
within the range of 20 degrees, the shell at junction with the cylinder can-
294
/0"
ORIGINAL SHELL
SPHERICAL /
SPHE.RICAL OR
TOROIDAL
*, 20-
into the interaction process for the most used simple shells with different
295
2.40 SPECIAL SOLUTIONS
2.41 INTRODUCTION
shells has been discussed. Problems can be solved in the same manner,
The "fixed" (or "built-in") boundary does not permit any displace-
direction.
by this method.
296
In addition, the formulas will be presented for circular plates (with
and without a circular hole at the center) under different loading con-
plates and rings usually are the common interacting elements in the
297
2.42 HAMPE METHOD
Hampe (Ref. 2-2) derived a set of general formulas for the stresses
ing theory) are presented in tabular form in this section. The following
Paragraph 2.42.6
follows:
The scope of this manual does not permit a breakdown of the deri-
mation refer to Hampe (Ref. 2-2). The general formulas are given in the
298
The stresses and deformations of the cylinder with any fixity at the
2. 42. 1 Stresses
A. General
w' %N)
SIF 9(N
% S2F14%•) " V 3SF%() + S4F0() - k =0
The moment is
M =D
x 1() + 2k 2 IS
w~p1 1F8M + S 2 F 1 6 (0 - S3 F 1 5 (0 + S4 F 7 (o I
S F (10M)
- S2F13(gM) + S3F14(tM) + S4F9(tM) 2k3 0
299
The shear is
No =- -- wp(t) + S 1 F 1 7 () + S FI 8 (o)]
ww(9N)
S1FI 9(•N) - SZFZ20(N) k -0
The moment is
The shear is
300
2. 42. 2 Distortions
A. General
For hinged edges of the shell (rotation but not displacement permitted,
301
Table 2.42-6 presents the coefficients Si for cylinders with both
bending will be involved, and the membrane theory will not be adequate
to solve the shell problem even if the edges are free to rotate and
These cases are assembled for the free boundary conditions, however
Si factors are determined with linear bending theory in most of the cases.
Dash over Si factors (Table 2.42-9) indicates that the length of the
cylinder is such that the supporting condition of one boundary does not
302
NS
00
JU
4.4
a r4
+*+
aa
NP f
ow 0
303
00
00
0
ti X
+t + oj
+
U,,-
. l..l .- •(+
[mQ. +
,
N Na
0 adI0 M-
N
zo U,
0
++ +
++
4+
u!
304
o A I l i l.IJ
IL• •
its•,
NO
I- .- ÷~
u S
Zia.
H305
305.
TABLE 2. 42-2. (GONT)
"'odC . .....
P.Rz 4 (k04
4
~ 3
+ P.RZ 4 (kL)
4
F3 P.It' 4 (kLI
4
1A
El L)*~~.4(k E .4k4 iT 1 *.(0 I,
2
P.1l 4 (kLY4 - _____2 /F 2 F
*+(L) 1L),(k I(F--- Z(1L)2(i -F..Z 2 1.L)
2
Vz4
r7 )l *L v1
kLkL. 1.Li kL -7
I , 2 (kLl)Zk4
2
~ * 7
r?7 * r7 +~ & F an
kL F 1 +a /J Y17 I k. F, 1 0+0 1 1
4
PtR 4 (kL)
4(kL)44 F3 (kL)4 Po
A 4 3
U .l4 4+ CkL
0.
V .. 4 + I L L FJ4 21 El-j *1.4.4(kLI4 [FI4Z. ± rj4
2
P.R 1 4 (1L()4+ * +I F, F 5 .) I
-kI ---kL ,-+ r
\F,+2 ILL1.2
r7 .PI. F7 *. 1
kL r, +z ) I2(1.04 kLFTj /(I.L)
z -!I . I
2 (kL)l
t
4 ,- - Fj -L r, +I + 0
R, F3 r-F I-ILI 1 * F
kCL .R n.
306
N N
(I) N N
+ +
NN
SI• +"
z
0
N +
NI in N N
NIl N
Uq
a.
Cr4
307
TABLE 2. 42-3. Sl AND S2 COEFFICIENTS, kL ý4. 0 (Ref. 2-2)1
Load Condition
2
0P 0R P. RZ
Et Et
P(4 - POl-ons
PvEt Et -;LI
PM~*p, (14)
0 -Ft- kL 0 + 4(kL)4
*<kL 0PR
Et 7
P(W Po Miot
1M=-io 0 0tE
1
For sez paragrap 2.2. factrs 4
308
TABLE 2. 42-4. Si - VALUES, kL44.O0 (REF. 2-2)
PC)
P * o~t.
El DS T Et 74
4 4
.I 4 (1kL) .2 PoR2 4 .
-2 5
4 4 4 4
Et . + 4(kLC 2 (kL)Z Et . +4 (kL) 2 (kL)
5
1P.K2 .2 pP. l .2
5
P. kLi~ (kL)2 Et 2 (1d4
7L4 / F4 kL 4/
.Iz 4 (kL)
4
4
r[r4Ts PoR
5
4
4 (kL)
4
r6
+EIr . C+4(k L) Et 0 + 4(kL 174
4
.Itz 4 (kLl .* P+1.2 + .5z5 7- I .5 Pn5 .2 4 (kL)4
5 4
7(- (kL)
4+(i + 2 (kL) =45 (kL)
4 z Z U4 lkL)Z + 4 (kL)
+ *,s 1 + pOiL . 5
,.jit
+ -1 Co EI (
*k~ ICI F4 IkLF 1.. 4
*I
74
309
* "'I•. .~
0A
A 4
o -L
Z10
v 0
•• . iiIi I I
3100
TABLE 2. 42 -5. S IAND S 2COEFFICIENTS, KL > 4. 0 (REF. -~2)1
Load Condition
a.-. consi
HoR Et 0
4 4 2
Et 4 + 4 (kL) Et a4 + 4 (kL), Z (kL)
2
.0P.R p0 R
2
,2
p1)= ,cosof Et Et 2 (kL)2
2
peR 4 (lcL)4 OZ 4(L4 0
4 4 2
Et 4~ + 4 (kL) Et 04 + 4 (kL) Z (kL)
<kL p 0 RZ + .
P(91J-p. excEg" 0
Et Z (kL)2
0 0
311
TABLE 2.42-6. Si-GOEFFICIENTS (REF. 2-2)1
V/0 I 3 I pA
4
Et .+4(k) Et . +4IkLS4
Et __O EF-E F 1 1- / t 74 F l
-CC.. (OCos
Et[1, RFj Fi
SFopine- r F sin-
10
[ZF
s ine- ".
p(C)~~F 0 kL Po on) k.( .Lcs)
.~I 7F,
argah24.
F~~~~~F k (a74LatrI
3 1 2+ - 0.-
TABLE 2. 42-6. (CONT)
4
p.R" 4(kL)4 poR 4(kL) poRZ 4(kL)4
Et 04+4 (kL)
4
Et '
4
+ 4(kL)4 ~Et . + 4(kL)'
PoR R2p
L) (F Flo F 0 / F3 Flo - O*
a ~ F ~ I Fe 8
- F9 in
1,L F l F kL F r F
2
3 Flo*aa.,[ Flo paR" - 2F cg
-t- &
Cop.o -- --. x (
P( POcom *<kL PR l -F Cs7
Et
* F6 1 8 5F F8 F i F'
and)
fatos see) pararap 2.42.
k.R paor 4
3137+4(L4 t0 k)40 k)
N
Eli
N
040
.0r
'-.. kjh1•
..... - .
0 i C4
r31 4
3141
i
•0 e0•
• j 0 ISI."S.•• 0 • • .. . :.
- I -
.
=, - I
lU •l •,,
1315
TABLE 2.42-8. Si--COEFFICIENTS, kL<4.0 (Ref. 2-2)
(rI r I r? / 7 . r.
• 4 |L7
F, k, 7
TS /I
7-
b
T
I /
d ..,.,.'" . • , .o , '
/ 44L +_oL A4 4
t
E' I DOg VF, 1t z7iLr~ rIL
k r, Ti
316
,i= . .4o
0 I I
uK ".A"1- . k " • .1
4
0 - _ _1_
.3 A. li
I0•1
03 0
A A,
NN
N bib. :lb•- . ~ wI
SJ, S, ,,÷-÷
S!,
II
317
31'7
TABLE 2. 42-9. S 1 AND S2 COEFFICIENTS kL >4. 0 (Ref. 2-2)
Load Condition S1 S2
0 ýoI
M. 00
Symmetrical
About Centerline
PoR 2 4 (kL)4 4 3 0
Et .4 + 4 (kL) 2 (kL)3
P o
R9a
• k< L Et 2 , (k L)3
A 0o =(P=
p $in . at < kL + P R a3 0
318
TABLE 2.42-9 (CONT) ________
Lo~ad Condition Si S2
osC9 oW kL
a< + P0 R 2 02 PORZ a2
Et z (kL)2 Et 2 (kL)2
IA( (
PO 24 (kL)4 aZ RZ0 2
Et a4 + 4 (kL)4 2 (kL)Z + 0R 4 (kL)4
OA. 2
~ 2 p0 R2 a?
2 2 2k
Rp k a
Et E't
+ -Z2k 0
0.Et
319
TABLE 2.42-9 (CONT) 1 ______
LOAD CONDITION
0 0
PC4p(2
pk.` ( pConR 02
p0 R
+ PQ Z a
in - 0 o
4 *o2
+POR *2Z 4 (k L) P.Rz
2 4
Et 2 (kL) 0 + 4(kL)4 El Zd)r
4
060
min4 +0 (kL)
4 4
kL 0~ + 4 (kL) 0&
p0 R p*z2 *
P C.06
*<kLEt 2(kL) t 2(L o
P(4) t
Ck kL
2
Et2 (cL) 04 k)Et 2 Z(k)
*
+kL -
3200
2.42.4 Analysis of Cylinders with Rotationally Symmetric Discontinuities
in Geometry or Loading.
especially if they are located not far from the edges of the cylinder.
occur. Fig. 2.42-1 shows the designations and the system of coordi-
I
M.
1W T-!
be obtained:
321
M=.?P!Dk4
k, w + 4) Aw')
QiDk2 ký Aw + 4 Aw)
k Q2 p M p
where
2 M1 2
The values 4M
M, 0, are given in the Table 2.42-10. Some special
2.42-11.
322
NIen
N
4. +
=.I,
N i.=i.=
IN
• . N
r~)N N
'-Z4 -___
-III 323
H a
I
0 - N N
++
IN N ++ NN
N + S+
N +
Nm
fQ fjq
I--it
o IIl
-I- •-
., I .
I-
I.,
- - -I
S• ~F4- ÷
I IIIIII4
i 'N
N - N
0 23 4.44
MV 40. N+
++C
+
I&
0le+
H~~ LZA I
1ff..~g325
+
- -
14-4 - ÷
+. 0.
+
• + + + A
AA
. +I .. + +
++ -I +
'
326
2.42.5 Spherical Shell, Any Fixity at the Lower Boundary
herein in the same manner as for the cylindrical shell. The boundary
Et ( Arm 2F F2
1 Rk
t .m F5-3+
1 P R ikF
1
/Ar Fl F4
6 F1
S Et
EIA/r ,
F6
212 + P
R F4
F2
3 2Rk~s' F 1 +2 rnkm +2
5 Et PR
4 2Rk rnk
For the special case of "closed" spherical shells, the only constants needed
are:
I2k2 M sin R
327
S= -~ •
2k2
Ar EtF1
m1 1
1 2 sin 01 RkF 4
Ar Et
Sz = m
2 2 sin 1 Rk
=
S3
Ar m Et
3 2 sin 4)Rk
ArmEt F
Sm
4 2 sin O1 Rk F4
For the special case of the "closed" spherical shell, the only constants
needed are
Ar
m Et
1 "sin
2 1 Rk
Ar
m Et
2 2 sin Rk
328
With these indications the general formulas can be given.
Longitudinal Load:
N4) = N ,membr
+ cot 4 [S 1 F 7 (a) + S2 F 1 5 (a) + S3 F 1 6 ((a) + S 4 F 8((a)
Circumferential Load:
N N membr -k [-S 1 Fg
9 (a) + S2F1 4 (a) + S3 F 1 3 (a) + S4 FI 0 (Cl)I
Shear:
Moment M :
Moment M 0 :
329
Deformations:
Horizontal Movements:
Rotation of Meridian:
2k
=- 2[ S F 8 (a) +S 2 FI 6 (a) -S 3 F 1 5 (a)+S 4 F 7 (F(a)
Edge
Longitudinal Load:
Circumferential Load:
Shear:
Q- S1F 1 a) + S2 F 1 8 ()
330
Moment M
M =-L-SIF0(a)
R + SzFg(a)
- 2k 1Z '0 2 19j
Moment Me:
Deformations:
Horizontal Displacement:
II 4
w + 4kw
331
can be represented with the following combination of trigonometrical
332
Definition of the F-factors is given in Table 2.42-12. If the
333
TABLE 2.42-12. Fi(g) AND Fi FACTORS, (Ref. 2-2)
Fi{F
3 sinh kLE cosh kLE + sin kLE cas kLE sinh kL cosh kL + sin kL cos kL
4 sinh kLE cosh kLE - sin kLE cos kLE sinh kL cosh kL - sin kL cos kL
9 cosh kLE sin kL - sinh kLE cos kL cosh kL sin kL- sinhlLcos kL
10 cosh kLt sin kLk + sinh kL cos kLE cosh kL sin kL + sinh kL cos kL
13 cosh kLE cos kL" - sinh kLt sin kLA cosh kL cas kL - sinh kL sin kL
14 cosh kLk cos kLE + sinh kLt sin kLt cosh kL cos kL + sinh kL sin kL
19 exp [-kLe (cos kLe + sin kLf)] exp[-kL (cos kL + sin kL)]
20 exp [-kLt (cos kLU - sin kLU)] exp[-kL (cos kL - sin kL)]
334
FIG. 2.42-Z. Fi FACTORS (i = 1. 2, 3. 4, 12)
(SHEET I OF 3)
;i;;: ;:4
;tj.
qq ....
....
.. .
......
m: ...
..
1 4:
. ........ . .......... T -1 .:w
.. ... . . .......
....
....
....
TT ...
.... ......
... .......
[ w.:as
4
1JU "if,
.. . .
....... ... ........
....
7:-
....
....
...
Fi i,,1-7 7,7
m;:: . .........
HE , T.-An=
-1-:1=1=1 r
....
.....
....
=j ;4-
........
....
. .... . ...
......
.......
.. .. .......
=::I.':77
........I 7L . -7n
....
............. 7ý; ......
7. imuU!
lin r.:-
.............
............
H:-
............
.. ..........
tr-. M
:7ý
.......
.....
.............
335
FIG. 2.42-Z.. (SHEET 2 OF 3)
34
24
F. I
to-
2(~+ -4
244;
;-
Z6 - - -
F -T-'
Fo-
.51.01.
336
FIG. 2. 42 - 2. (SHEET 3 OF 3)
31;
... . ...... ...... . .. iým 'rul
. .... ... . I. .
.. .... A ..... . . .....
. .... ... ... .
=TJ
Ti:
....
....
..... i:!;j:!ý!:..
.... -af
...... .......... .... ..................
I i: 4
........
I.. ....
i:: w: :7:
............
t7
..........
. .
.........
. ........
.........
IF
H;.-
it T:r.
uFi
ii. 1IR Im" j: 1 1;4. 7:
nit
it
OFT
"77-.
i2i i-.!!
'ir
til
ýT
=Jiiii
7it zz
I M .14
NI,
rM .. ... .... .11
177
4:1
337
FIG. 2.42-3. Fi FACTORS (i = 7. 8. 9. 10, 13, 14, 15.* 16)
(SHEET 1 OF 3)
T
f
T
7ij
ý17
-A
Tij t t
-T7-
+
0 21.
1:
Fi :7
!:tA:j
ENLARGEMEN
ELOW 1-71 T
ýj 71,: f
0-
4.0- .1 T.:.
1
j k-A
0-
-1ý L-4 L
4
4;7
41 7
0.6
I L I
0.4
T
7fý- -;-ý7-
T '91
338
FIG. 2.42-3. (SHEET 2 OF 3)
.....
. ......... -7'77:
4 ý77.
..........
1: T1 . 71 M;
in .
......... .... .... ln:
... ..........
... ..... .. .......
q -7.- ME :jjj
. .... ....
...
....... .........
....... :1 fj
7: 7 14=
.............
.
. .......... 7 _7-T-1
....
.............
E:
T :r
....
.........
...
:14411!
1-.:: Mdý :ill!:
. ................
Z=::::!7 7ý=
t4
.... ........
... .........
... ....
JA
I ..
. .. .... .
.........
--- --------
71
....
....... I.-
...
..... ..
... UY....
i -I M.
.. . ......... 7', -1
.. ... ... ..
it lip -
S E EENLARGEMENT N
t7 NKý-ý
S H E ET 3 0 F 3
it:
R.H
... 7:1:.
:1ý;
-4.... ..
44
jHE .......
...
....... .....
1:11.
'"'I '= 4
......
....
.
...............
T:;m: n ... .. ....
. ... ...
....
::T in
-- iH-
T1
339
...
....
...
1. Z . ...
...
........
i n :... 4 '
'77
ýz4 !:t1 .
......
r:r
340. . .. 4
FIG. 2.42-4. Fi FACTORS (i = 5, 11, 17, 18, 19. 20)
(SHEET 1 OF 2)
0.9
0.
'T' 4-
T
1'7
.
.l~ TI~
A':
4-[4 7'_ 774
.0.7 7
-0.4 .
a -ýI'1
341A
...
.....
7 1 .
. ... . ...
...
....
t.
wr,
*~..................
- .
PCi7T
.... .
434
2. 43 APPROXIMATE METHOD FOR DETERMINATION OF LOCATION
AND MAXIMUM STRESSES IN CYLINDERS (REF. 2-2)
the location of
ýN and 9M are nondimensional values that represent
maximum circumferential stress and maximum moment due to any linear
loading, characterized with PV and Xp. (See Paragraph 2.32. 6.) Nmax is
along the meridian. QF and MF are the reactionary forces at the boundary
are two similar sets of graphs: one for the fixed lower boundary and the
Nmax.
These graphs were plotted by Hampe and are based on his methed which
343
FIG. 2.43-1. DETERMINATION OF LOCATION AND VALUE OF
MAXIMUM CIRCUMFERENTIAL LOAD IN CYLINDERS LOADED
LINEARLY, WITH FIXED BOUNDARY
r = xpPv Pv xp
4II-
0.9 0
0.9
Pv
.fl Pv( +,, p
0. 8
rnmax
P R
0 = location of
Nmax
xmx
0.3 1F p Fixed
(d . 5 !4 boundary
3 5 7 9 1 13, 15
344
FIG. 2.43-2. DETERMINATION OF LOCATION AND VALUE OF
MAXIMUM CIRCUMFERENTIAL LOAD IN CYLINDERS LOADED
LINEARLY, SIMPLY SUPPORTED AT LOWER BOUNDARY
Po = vP p vP
1.0
0.9 - Nmax
p R
0.8
0.7
o. 6 9N-= L
X 0.5- k = "
kLL
kL:
345
FIG. 2.43-3. DETERMINATION OF LOCATION FOR MAXIMUM MOMENT
FOR CYLINDER LOADED LINEARLY WITH FIXED LOWER BOUNDARY
0.55 t __L _
x ,, o.z
p
0.50
0.45 __
,-0.0
0. 40 •
0.35
0.30 , _ _
0. 25
0.20 3-
0. CPS,:
I7 IL I'S
kL
346
FIG. 2.43-4. DETERMINATION OF LOCATION FOR MAXIMUM MOMENT
FOR CYLINDERS LOADED LINEARLY WITH SIMPLY SUPPORTED
LOWER BOUNDARY
+_
00pv _p
O..3
2
0.Lxp,.<
I
0.25
0.20
em R
0.15
L
x
Simply Supported
3 57 9 'It 13 1
kL
347
FIG. 2.43-5. DETERMINATION OF MAXIMUM MOMENT FOR LINEARLY
LOADED CYLINDERS, FIXED AT LOWER BOUNDARY
0. 020
Mmax
PvL 2 k 2
0.016
0.010 ,,
o.
0. 004 004 Fixed.. end-•
0. 002 11
0
3 5 7 9 11 13 15
kL
348
FIG. 2.43-6. DETERMINATION OF MAXIMUM MOMENT FOR
LINEARLY LOADED CYLINDERS, SIMPLY SUPPORTED
AT LOWER BOUNDARY
Pvkp
411
- LPv ",ox Po -P--),+
0. 020
Mmax
2
PvL
0.016
0. 014
0.012
0.010 0,.0 _. t_ R
0.008
0. 008
0. 002
03 5 7 9 11 13 is
kL
349
FIG. 2.43-7. DETERMINATION OF SHEAR AT LOWER FIXED BOUNDARY
OF CYLINDERS LOADED WITH LINEARLY DISTRIBUTED PRESSURE
iL
O R
0. 16
0.08
0. 10
0.041_____
57 9 11 13 15
kL
350
FIG. 2.43-8. DETERMINATION OF SHEAR AT LOWER SIMPLY
SUPPORTED BOUNDARY OF CYLINDERS LOADED WITH
LINEARLY DISTRIBUTED PRESSURE
•Pvkl,
0.20 I
Or __ _011
0.16__ _ _ _
Xpt
0.12 R
0.12.,O. • k'k ~~0..2kO'
-- "
0.10 - L
0.08 0.08
• Simply Supported
0. 06 (No Movement)
0.04
0.02 __ __13 7 is
kL
351
FIG. 2.43-9. DETERMINATION OF THE FIXTY MOMENT AT THE
BOTTOM OF THE FIXED CYLINDERS FOR LINEAR LOADING
MXP
0. 05 I____ ( )p
pvL 2 l-T
0.05l
0.04 .-1
1i~ ..
_ _......_ __
RD
0.1
0.02
0. 03 .0•,
kL
352
2. 44 CIRCULAR PLATES
plates with and without a central circular hole are considered. These
w = deflection
P = rotation
E = Young's modulus
p. = Poisson's ratio
t = thickness of plate
3
D - Et
12 (1 - )
M = radial moment
r
Mt = tangential moment
Q = radial shear
r
solutions are presented in the same way as for the shells. Finally, spe-
353
2. 44. 1 Primary Solutions
354
U- ' ! .,--.
o1 No.
I ; 0. q I * aI I N• 4i
0- S
/C I . -
P4.
a . 0
LIa
LE _ I It
355
N• T
*--
o. --
I* N
2 ". __ __ _ __ __
I-
"Z A
K -w
U a - a "-
- -1• =.K• , •1 •- -•
K - . .
,, . -'L'-a a a a=
o : '- *
IlkN
a ~~. +
I-- + .+ + +.
A
z ~+ .
Is
'a a- +4
00
- 0 +
g - -all.
+ +~
d .*
.44 P 0 N +9
II
4
~ - 'L IL
IN
-M . + I
+l .8I
+
La. I
- .357
A A
o N
H
9-4
-.-.
A
4.
.
C
-
-
a
-
a
-- .Ia A
Al: - - £
a *
- . -Ia A
rz -
-
*
H 0 .4
- * 9-
"IN - .
A
N a a -
* A
A .. J
K * a - *
or.J
a.
-
0r4
+
H.
eO
a a
A A a
k * - Na A
14 .4
9' a
-
A
jN
-. RI,
I- T
- *
N a -
A - a A
.I
*a .4 ja -
* -
- - A - 4 .4
* V * - .4 .4
("".1 P.4 - - -a
.Z 4J* . Z
* .2* J I.
'1 *:14 I . a - - .4 4 -
U - -
a ZI - N .4a
rJ * A
.4
- -
A
. A
.4
RD A .. #4 -
a
-.
U * 14 -
- .4
a.-. - - ____ _____
-
a * .4
a
* a -
a h - - -
H -
ZI *-.-
ia
ea -
a
a
JN
A
.4
NJ -I: *
.4
A - .- -
I 9.I
358
2. 44. 2 Secondary Solutions
along the edges (Fig. 2.44-1). Table 2.44-3 presents solutions for this
loading for different cases of circular plate with and without the circular
= Ma 2 ( )lrpap
2 D(1 + t) a I
Ma
D(l+p) -P
w
Mr = Mt = M
Qr= 0
359
0 -
N Z
o o
ILI
0360
2. 44. 3 Special Cases
Special cases and solutions for circular plates that occur com-
boundary conditions, and loadings for special circular plates (with and
without a central hole) are shown in Tables 2.44-4 and 2.44-5. The
information for these tables was obtained from Ref. 2-2. and 2-8.
361
N I
x ~ -I I I
"- r .
0N , -, .- -- .-
uU
I-I---- _I___
_ N NNot
0 I a -
- . i a %
N•, _ - _ ÷
o +
- - , .5 -
4.4
,.. - "-- - ,
TA,
i I# 0 it 'I.15.4
S~~~-I. I •tf rId
I• •-.3-
~36W
TABLE 2.44-4. (CONT)
m M Ibm-in M
in
+
-4.• " - Mt -
x2
ks 1
Ma&Z 2_
,- 2
w kS(-1+ I.n p)
ZD
Mt -- Mks 11 +
Or 830
363
LU~IL.... - - .
-10 7
u a
E- - I
- +
+
- IL
4364
A --
~l le--
-F-: j -I
I i- i•• iIi 0
7T7
o 0. ,"
o .A' )+I
II
I-
, 1 - ±-=
_
A . ...
5:
365I
365
H ~z NL 4
zz
P4.
44 -. 4
Ii2 J4
~zI.14
z . -3
u+
0 N -
0
Y.
-~ ~I366
2.45 CIRCULAR RINGS
act with shells. The theory of shells would not be complete without infor-
367
j. - H
0 LI
N 4 Ed g~r
W W
Ia
-w 0 IQ a
00 000
Z 74 Z- I- Z
01 0~ 0~
wBU
7- 0-
00
1 00
14 E
N - aI - ~
a a Ljt-J 0
a 14 C r--368
2.46 CONCLUSION
method was presented and special attention was given to the cylinders and
the necessary information for circular plates and rings was provided.
With all the information on numerous shell plates and ring elements, two
369
2.50 MULTISHELL STRUCTURES
(Ref. 2-Z)
2.51 INTRODUCTION
the introductory sections of this chapter, instruction was given for the
chapter, two methods, (1) the force-method and (2) the displacement-
can be determined and, vice versa, if the forces are known, the
370
displacement can be determined. Consequently, there is no basic
371
2.52 FORCE METHOD
2. 52. 1 Introduction
solving rigid frames, which is called the "force method" and which
introduces the cuts in the system under consideration and then, in order
indeterminate force.
2.52.2 Analysis
_*1
372
Each element (Fig. 2.52-1(b)) is statically determinate, and is
loaded with the primary loading (pressurization) and with the secondary
x1 is adapted here:
SYMMETRICAL A. Shell-Element
ABOUTIj
"CL
5 r (1) Upper Edge of a Shell Element:
X4 X6
X7• Moment in the clockwise direc-
x•1 tion.
Xj7 - X
• x-0
- 19 Horizontal load directed outside.
X11
373
Horizontal load directed in inside direction.
2. 23.2 and 2. 23. 3), a system of linear equations can be written for
Two cases which usually are considered when 'the system is built
A. Case 1
The length of each element is such that unit influences due to unit
loading will overlap. In other words, stresses due to unit loadin~g will
depend in every joint on both edge unit influence. For such a case, the
374
i X1 X2 X3 X4 X5 X6 X7 X8 X9 X10 X11 X12 jio
1 6
11 - 810
2 - - 20
3 - 30
4 - __ 640
5 650
6 660
7-- - - - - ----- . . . . . 670
8--- 680
9 9
g-
10 1______ __ ____ __ 10,0
6 10- 1
11 -'6
12 6 o
B. Case 2
by the unit loadings will die at short distance from the edge and will not
overlap with the disturbances due to the unit loading on opposite edge of
In this case the matrix will have the shape shown in Fig. 2. 52-4.
375
X X X X X X X X X X X X 6.
1 2 3 4 5 6 17 8 9 10 11 12 6io
___6io0
2 ,. 6 0
3 - 630
4 ,640
5 650
6 . . . .- - 66 0
7 670
8 ---- 680
9 __ 69o
10 ... - -- 60 0-
'11
611,0
12 - 61z 0
Elements
376
but then the elements are of a
O
X3 secondary loadings, because fewer
elem ents w ill be present.
T heseSX4
A. Case 1
The length of the element is such that the unit influences due to
shell, on both opposite edge unit-loadings. For this case the corres-
377
i X1 X2 X3 X4 io
1 - 61
2 -- = - -- 61
3 6
4 6 14
B. Case 2
The lengths of the elements are such that the unit influences due to
the unit loading does not depend on unit loading of the opposite edge.
i X1 X2 X3 X 4 16io
61
2 - 6
3 6 30
4 -640
378
In case 2, the system of four linear equations will be split into
two independent systems of only two linear equations to make the solution
greatly reduces the number of the statically unknowns, thus making the
2. 52. 5 Conclusion
of this chapter.
The formulas in this section are shorter and simpler than the general
loads and displacements at the edge only. Use of these formulas simplifies
the procedure.
379
2.53 DEFLECTION METHOD
2. 53. 1 Introduction
previous method and this method, the same multishell system will be
2. 53. 2 Analysis
Designate the edge rotations with Pi and the end horizontal move-
determinate shell elements are shown in Fig. 2. 53-1. Note that 6vi
(vertical movements) are negligible at junctions (3) and (4). Also note
6 6 6
Horizontal displacements: 2H' 3H' 4H
380
THE VALUES P1 = 6HI = 6vl = 0.
o P 6 H3 (Nv3 - NEGLIGIBLE)
©P2' 6H2' 6 vz
0
FIG. 2. 53-1. Multishell With Statically Indeterminate Deformations
will not be zero, but will be of such an amount to make M° and N° zero.
x x
For example, junction (2) will be rotated by some angle P , resulting in
M2 1 (62H1). M2 3 (6zH)
(moment M 2 " (62H) was neglected here, assuming the curvature of shell
2-2 to be small).
381
Also, junction (2) will be moved vertically to the amount 62v,
will influence thejoint (Z) by delivering M 2 3 (P 3 ) to it. The joint (3) also
Mi MO + M + M
2 23 22 2
then
presented).
382
A corresponding matrix is given in Fig. 2.53-2 for case 1, and
in Fig. 2. 53-3 for case 2. Again, it can be noted that in case 2, the
8 3H 6 2H 6 2V
P4 $4H P3 P2 6io
1 = 610~
2 820
- - 630
4 -40
-5 - 650
6 - "60
71 670
P4 6 4H P3 63H
P2 52H 52V 6io
7Z T- - -o 61
2 -
. 0
3 = 30
4 6 40
5 __ 50
6 . . . . .66 0
7 . . . . 6 70
383
Similarly, as before, the use of statically indeterminate elements
2.53.3 Conclusion
found, the stresses and deformations in any point of the structure can
384
2.54 TABLES FOR MULTISHELL ANALYSIS
Tables 2. 54-1 through -12 contain the formulas for the edge-
The formulas are shorter and simpler than those given in Section 2. 30
because they cover the special cases of loads only at the boundary.
conditions.
385
TABLE 2.54-1. CYLINDRICAL SHELL WITH FREE EDGES. EDGE
DISTORTIONS DUE TO PRIMARY LOADINGS (Ref. 2-2) 1
ah*" l -Illrty 50nd-vy di.lstorto
""d:.ion l AH£2 P
Oh.
2
*'x pit
p 0
0
p(e) =pv(t-)
4
""7 _---•__,__
1 " J- "° 7"•
" '1,7". '•
__
___ IF
" ",/... "
P(t4)
4
)1
. r 4. 32 I
PQ(h) WIk -
2 / F8 F -(2k0 r FkL[ 23
I- r
S e 20 3 2 .2/ 2
+o4 fatos rl .4 +.42.6
L.
4+1(-
1
P(O Po 4
* 012 38
NU) N~
*u) 0J
lk
N
1% 1
NP
0
0
0 G.s
Uc
38
TABLE Z.54-3. CONICAL SHELL WITH FREE EDGES. EDGE
DISTORTIONS DUE TO PRIMARY LOADINGS
q 0q
PI.s 1 ]
C..* .- ,.*C a
P
L--d--, CondiS.
vI-•b- V
388
** *
W 44
-
o
H
0
0
9..
3 89a
oIra o ~ 0
a
o C -- a;C
I-=-- 0
4t
a
390
o90
+ +
40P
00
z:DI
0 n
rx1
zI
t
ht
391
TABLE 2.54-6. SPHERICAL SHELL WITH FREE EDGES. EDGE
DISTORTIONS DUE TO PRIMARY LOADING (Ref. 2-Z)
(2 +l0 sin 6
+ -.- (-cos6 1 + 1)1
sin o
S- (I -PR-. a,
Z Et
.InZ*I
" RE"T C $inn #1t
LoEin Cin#ditionW
- -ljt Rq[in*
sinZ- • (1Cos$I r
PO-cos*
29
ý -. 6 coo2 (co 3 + let
I +p ) sin#,l - le Z
+ sin 4 1 mi)
A I-z 1&
in #, --- i*
Et
392
.
.4 0
0 -4
Z co
P4
o4 .
In
* - S.~ ~393
-.
'4
S4, •1
O4
- 4. .,= -•
.0 4'.4 .-- -
0 "
o. .
0 , .~ .- .
C)÷
S* '
- ,, .,.-.__- .,, ,,
4. - - 4°-'
* S I ~~
I. - .-.- I I 4 - I I I.
I..i I.. ..
NNI
I~
o N
I- 4-, f
0-N
, 4 4 N
-I a.-
"I..I
3951
TABLE 2.54-9. CYLINDRICAL SHELL WITH ONE EDGE FREE,
OTHER EDGE FIXED. EDGE DISTORTIONS
DUE TO PRIMARY LOADING (Ref. 2-2) 1
JH2
:• • L
- Po
p(P)-p 0-const
F ZF 71x22k7
R0 Fl 7 v ZF 9 _ _
Vk 2 EtL F
2
+ p+kL
-- F "Z - kL F,FR+ - coo 2 •F--•,
2F3+ i ... .... F +
F
4+4(L)14
F +Z Et 4 + 4
[kL)kL• F +Z
Co Z(L i(
112 -, L ZF, k)2 2 F rL ZF,
R2 4 (kL) exp (- a) - ` + -PR . 4 (kL)4 exp (-) n) -
Et 4 + 4 (kL)4 + kL Et 4 (kL)4111, FI + z kL
~
10 ~ x Zi~ 7__Z~:!+
exp a)
up (t ,) - p. e x p ( -.0 ) Z( k ) FF1+
+ kL F) 1 + )Z FI + kL
( k .)
z k " kL"+7 T+2
396
TABLE 2. 54-10. CYLINDRICAL SHELL WITH ONE BOUNDARY SIMPLY
SUPPORTED OTHER FREE. EDGE INFLUENCE DUE
TO PRIMARY LOADING (Ref. 2-2)1
All.,1l.
,n
I,'( k) P, 0 •nc)
4 '4
+ ,12 F 4 pIV(]L ,, 2 8 L
F4(
p(i) i',.
•. t F4 444 ,r F4 k-
p11
2
4(kL)
4
)'9 2 F1 p 2 4 4(kL)4 L 2 zi.7
P({,) pu ,nin ,
4 [o4 F4 2( F 2 4 [T Zin&2 ZF
2
• 0 + F4( LLL4 F4 -- .42(kL) (k [4 .4 4 2k'-
4
__ _) 2 40:
]•t2 O= F~£Z kL , F4
* ( in
F
P(g) Po. 'O
4
"o"R i 4L F 9 .I2 " 4(kL) z[ a
SEt 4+
+ 4(kL) 4
x pI( - F).
F4
. ..
Z(kL) 2 Et 04
a --+ -4(kL)
4 .p - .F
P(0) p (-cc)
o,.xp
1
For F and k factors, see paragraph 2.42.6
397
TABLE 2.54-11. DISTORTIONS DUE TO SECONDARY LOADINGS (Ref. 2-2.)
I.,I;I,111• ~ n o~ ;olvr
n.dtoS*'t
Ar A
2
Z2Rk 2 2k
3
2k
2 41
m Et -EtR
2 3
M RZ k2 ZF R k 4F 3
I --
Et F + 2 Et FI + 2
R2 k 2F4 R2 k2 ZF2
Et F+ Et F+ 2
m m
2 3 4(F +11
R 2 2 2FR
Et F4 2 Et F 4
2 2 2
R k 2F2 R k 2F73
1114 1 Et F4
398
0
coj
0y I
N L
Z 91 ILI
INN
onTTTnm
V, 0
399
+
I
I +
I IAI
pq
+ I J
Itt,
Li9L
___________,________,_________ ,___ , U
400
+- -
o.iN
I I
-4
.4N N
0 NN
401
LCL
I I
eq
1z,1
/c. ....... . ..
401
2. 55 SUMMARY
shells were described, and tables were presented for simple and fast
402
FIG. 2. 55-2. Discontinuities in the Wall Thickness
tion that the material must be isotropic. In the logical process of gen-
403
2.60 COMPOSITE SHELLS
2.61 INTRODUCTION
approach will only be mentioned here. The minimum weight design should
some cases, the minimum cost design actually corresponds to the mini-
404
If the section is stressed axially, excluding the stability consid-
erations (Chapter 3. 00), the distribution of material does not affect the
sideration should be given to the material in the two flanges, which will
405
Based on the principle explained, the stiffened structures were
406
2.62 STIFFENED SHELLS
2.62.1 General
ferential stiffeners provide most of the lateral support for the meridional
frame. The plates between the stiffeners distribute and transfer loading
407
approach will be given later. The geometry included is for cylindrical,
outs are needed, they will usually be located between the stiffeners.
408
the interaction will be performed as shown in Section 2. 50. If only
F ig. 2.62-2. _
SHELL
STIFFENER
S(CIRCUM-
FERENTIAL)
-0PANEL
STIFFENER
(LONGITUDINAL)
409
If both circumferential and longitudinal stiffeners are present, the
distributed over the plate surface and perpendicular to the middle sur-
face. Plates with the ratio of side equal to one are most effective; they
transmit one half of the load in one direction and the other half in
than two, for all practical purposes, the whole loading can be considered
Markus (Ref. 2-10) method for analysis of such plates. Curved panels
more from the plates. The curved span under pressurization usually
deflects less than the noncurved span; consequently, more loading will
410
There are no fixed formulas in existence for the case previously
tioned, if the stiffeners are positioned close together, the structure can
the section. This method shows how the combined section can be sub-
411
certain corrections are entered into the previous formulas. The
analysis for the shells where the shear distortions cannot be neglected
sections.
ti
t4
elasticity, Ei, and a cross-sectional area, A.. First select one con-
412
the thickness, ti, of individual layers was not changed, but areas A,
/AI
_______ _ itl.
-tz
*t4
t5
"A5 *
the neutral axis, the thickness, t, can be found for the new monocoque
413
TABLE 2. 62-1. TRANSFORMED SECTION METHOD
3 Yl
4 Y3 Y2
-i y " 2~
12 1 1
414
2.63 SANDWICH SHELLS
2. 63. 1 Introduction
more than with geometry; the second type is dependent on-the geometry
struction are considered to be the core, the facings, and the bonding
415
properties of these three independent elements. Table 2. 63-1 presents
three materials.
Shear strength
Shear modulus
Flatwise compressive strength
BOND MEDIA
Shear strength
Shear modulus
Tensile strength
Peel strength
416
In general, there are properties which are unaffected by the con-
core, and bond media. This class of failure modes may be further
and those that are localized to a portion of the structure but still limit
The general modes of failure generally are associated with the buckling
417
strength of sandwich structural elements. Excessive deflections or
flexibility of the element might govern for specific designs, but these
exact analysis of all these many types of "foreign" elements that must
analytical approximations.
be identified and analyzed. Then the applied stresses (or loads) can be
compared with the allowable stresses (or loads) and a margin (or factor)
418
to-weight ratio. From a structural standpoint, the minimum weight
to bend about a common neutral axis. Although this Utopia has not been
place as much material as far away as possible from the neutral axis.
The weight of the web member is reduced further in a tension field beam
nothing with a core and bond media. It should be pointed out that the
the core and bond may not add materially to the tensile strength. In a
(although the bending rigidity of the facings about their own neutral axis
may become an important factor for some sandwich designs), and the
419
In practice, the achievement of a high bending rigidity is used
with sandwich.
might be used, in spite of the fact that it may weigh more or cost more
be able to design into the structure large areas or volumes which are
420
2.63.4 Types of Sandwich Cores
properties on these two planes are equal are termed isotropic cores,
whereas those with different properties in the two planes are termed
type cellular cores are generally nearly isotropic with respect to shear
which efforts are made to increase properties in one of the two planes.
erties both with respect to strength, bending rigidity, and shear rigidity.
commercially available. There are the solid cores, the cellular cores,
421
and the corrugated cores. Solid cores are typified by the balsa wood or
member. Such cores generally have bending and shear rigidities and
sandwich may be greater with the same facing sheets than one in which
Actually, there are two types of cellular cores. The first type
bubbles has been solidified. These cores, of course, were widely used
in the inception of the very low density core materials. The more
case of at least one common product, has the appearance of the cross-
with the axis of the cell perpendicular to the surface could be considered
of this type. There are a multitude of such cores, each with somewhat
different properties. It is pointed out that when the core has bending
rigidity in either plane, it accepts bending loads and stresses and must
422
The corrugated cores are orthotropic, both with respect to bending
and to shear properties. Both the bending and the shear properties of
the corrugated core are many times greater in the plane parallel to the
the corrugated core sandwiches than for the cellular core sandwich.
those of the facing sheets. Some of these properties are included in the
basic analytical parameters of the entire sandwich and govern the behavior
Unfortunately, all is not known about the local modes of failure and
has been indicated, the best the analyst can do is work with experi-
mental data which give some of the basic allowables. In the case of
wrinkling, for instance, neither theory nor test has yet yielded a
423
the design margins are low, the analyst should request a structural
test. Although the structural test may not be of any significance for the
local detail.
The similarity between the sandwich and stiff .ied shells was
extensive, and the core cannot take the shear. In such a case
424
deformation of such a core will be usually small and not
2. 63. 7 Conclusion
Section 2.63 basically was taken without change from Ref. 2-11.
difficulties that are connected with the exact analysis of sandwich shefls
the following sections will continue to put forth the theory for the
425
sandwiches but also other materials, such as stiffened shells (if
this manual. This will make clear what kind of modifications can be
made, in order to apply the same formulas (that were derived for
426
2.64 ORTHOTROPIC SHELLS
2. 64. 1 Introduction
are not the same in two mutually orthogonal directions (two dimensional
space). Such material has different values for E, G, and p. for each
negligible but in order to distinguish one from another ti' will be desig-
nated for the Poisson ratio which corresponds to the bending, and 11 for
are the same). In order to see the dependence of stresses and deforma-
definitions hold.
427
Strain rigidity
B= Et
Bending rigidity
3
Et
12D(= -
N= B (eo + Eel
No = B (Ee + LE€)
Me = -D •+
d cot
=
MDA COS +111 dP\
M8 D Re ~ fR d)
a.,= E-'-(2.
E No M 64-1)
T E(INe D/
428
For monocoque section of rectangular shape
N + MO
t- t 3
(2. 64-2)
No Me
0 -t 3
TZ
The physical meaning of D and B is obvious if Eqs. (2. 64-1) and
EN E *N-
=
N__
2 No- N
10 l%(i - 2) B i -• * 0 Et I • t A
ENe E 1- L N N
I O G 2
(1-FL)B i-FL 0N Et 1 • t A
where
A= lxt
It is convenient to chose the width of the section strip that is equal to unity.
The componental stresses due to bending:
Mez M•.E
ezt = E 120 - 2z)_
2 1 M0z
t3Z = MOZ
2' 0 D 1E 2 (1-I2) E t3
2 0 Cr M 9z
D 1 E 2 =Mz
0z E2 t3 =-Mo)
12(1- t342 )- 12 M 9z 1
429
Evidently, if stiffened or sandwich shell is being dealt with, a modified
equations for monocoque shells may be used for stiffened and sandwich
shells. If the values from the "transformed section" are used, then
BA*ED I* E*
2 - 2
1-p.
Now the orthotropy is defined if, for two mutual orthogonal main-
case) for the analysis of the orthotropic structures, the formulas must
tion in order to make possible the use of the unit edge loading method
430
orthotropic analyses which neglect the shear distortions will be
A. Primary Solutions
revolution).
surface.
431
When N and N are obtained, u and w can be obtained in the
1
= 1 (N -N
0 Et 0 N)
1
to = 1 (N "N
C0 = B (1-. 'P¢ (N 0 - p 0N 0)
Note: D B• t
equation:
du (call)
dT - ucot0 = RIe0 - R2f = f (0)
u = sin0( ffn d0 + C
432
where C is the constant of integration to be determined fromthe condi-
following equation:
U w
fecotes-
e 2 R2
for the isotropic case can be used and then, using the substitution of
in the form of
x (Function of Geometry)
conical and cylindrical. Any other geometry in the majority of the cases
can be approximated with these three types, as was shown in Section 2.34.
k, X, B and D
433
These constants, in order to transform the "isotropic" formulas into
reference to the analysis for the isotropic case. To make the trans-
k4 = 3 (1 - 2L)M)
434
where C and C2 are the constants. These constants are different for
MG = IlaM0
=. lgR12
D•(I" ••}tzsi($1s- a •k +Clek
12 R 2sin(- a)k %r2C1 -ka
C =2kM/R ; C = 0
435
The correspondent deformations at the edge are
4k3 M 48 k3 M
ERt t 2 R D (1 1•)
2 2
2 Hsin01
C 4
1 f- 2 4
E t2 t 2 D( - -2
2 Rk sin2 01 -24Rksin 2 0 1
Et Et tD
2 01 - i
436
za
00
0 ea
00
+ a
Hl +
0
w +
'-4 4
a ~ 0
a; +0 + x
P4
0 -
0 ~ a ±
0 *ao
0 ~C.~" ~.-.j
'-4'4 O93 -4 X
cy U -'
a N 04I.M 1
a
0
CDC
a z 4
p 4 U437
Conical Shell. In the same manner the formulas previously given
for the orthotropic case. The following constants shall replace those
Et 3 Ex t3
D===D - 'I ,
12(1- x 1.) 1-
12
k 3/,(1. zL T3
4(1I- x'L
tx cot i /tx cota
m 0 m 0
S 4
( 2
Vf-tsin0
The modification in this case is the same (F2 is replaced with IxL.).
In this way, all formulas given in Tables 2. 33-8 through 2. 33- 10 can
438
-44
N N~
0 C
UP
4LO
Ao - 1
L~ j - 2 1.
N
~ .; ~ L
o Mb
3439
0
0
00
0 zc lu
4,-
- ~ .
w_
. _ _ _ _ _ _ __w
UP ~_ _ __ _ _ _ _ _
o%
0m.0
qv
c ~ r.
0 rQ
e]-~~ N
C4
"I£A
44
Cylindrical Shell. All formulas given in Tables 2. 33- 11 and
E t3
E t3
12(1- pL 12(1 - L2.Lp.)
12
441
15.
4
-
+ U
+ A
w
0
* *
00 0
* + A NI
x CL
zaI
4425
2. 64. 5 Orthotropic Analysis, If Shear Distortions are Included
considered herein.
A. Cylindrical Shell
relatively low traverse shear rigidity, the shear distortion may not
respectively, in-lb.
443
= Poisson's ratio associated with bending in x and y
directions, respectively.
S S
directions, respectively.
B. Half-Spheres
derived for cylinders can be adapted for the spherical shell, too.
444
0.
z
0 0
C.I.N
0 0.
, - 0.N
ICA6
4.9
cncr
N 445
0 So
0: - 0 a A$
U 0 OL
00
SO' ao a
0
0 N
-. N3
U
U
0
-
A
Z
IL
0
N4464
2. 64. 6 Influence of Axial Forces on Bending in Cylinder
extended by the same author (Ref. 2- 12) to include the effect of the
2 rBy 2 +
1 -RR
D No
Nx
2
4 0I+.._0
D0x
4 = B (1 - jx yL
xy
4D R .[
B2_
y V24Y4 4Y2&2 + 4
4 DQxR
447
TABLE 2.64-7. MODIFICATION OF TABLES 2. 64-5 AND -6 TO INCLUDE
THE EFFECTS OF AXIAL FORCES ON BENDING (REF. 6-12)
Qua ntitie s:
Formula Substitute
(Formulas in tables 6. 64-5 and -6)
(4•02 A) (4.y2 A2 )
(Az. 4&2 ) (A22 4y 2)
,00 , ,O v. M0
9 V S
vI S
10 Whole formula w4 (M/v.D) (2a 2 + p2 _ 2zZ)
19 V S
20 V S
448
2. 65 CONCLUSION
sented for the analysis of stiffened and sandwich shells. Such shells
are analyzed in the same manner as the monocoque shells with minor
must be considered. For this case, the additional formulas are given.
for the stability analysis of shells. Also in Chapter 3:00 Di, Bi, and
Ii and i'. values are given for the most common types of sandwich
section, and this information is useful for stress design also. (See
Section 3.32.)
449
2.70 UNSYMMETRICALLY LOADED SHELLS
2.71 INTRODUCTION
to the geometry, material, and loading. The "unit load method" was
exclusively used for the solution. In Chapter 1:00, the most commonly
used procedures for solving different shell systems have been dis-
cussed. It was shown that the most complex solutions are applied to
450
2.72 SHELLS OF REVOLUTION
Table 2. 72-2 presents the solutions for spherical shell with non-
symmetrical boundaries.
451
31
I --- 31
4 ... A..,°).
,31
a ,,, ,,
, ,,, , ... .
C) - N 315
•,, 331 i V
0
* .-
C N 0
E
k0 4. 5 +
+ 0
00 - * *0
- 4 N 4
* N4 5
0
*
0 0
0 0
0
* 0 o
- z I 4
+ 4 *4 04
4
* - -
-.01 -
*
0I.
"I. e* *o .0
0
00
I
Z4LY
Cl) 0 U - 0 -
f.o 171 uN U - __ N
0.
- 0 0. * U
0 S 5 0.
U 0
-___ _____
____ ______
I-
I 0
7 0 0
- 0 0
S
e
.5
4 A S S
*o
S
0
0"
9*
:
0
I.
-. .2 015
* N 0 4
A -
+
* - 0
0 4 0. 4 0 4
0 N 4 h A
g A
z 7. *
' . *
* N S
___ 0 0.
0. 0
F-' U
0.
-
*
S
0
o 0.
Cl) 0 _______
- I..
0
.5 * *- 4 0
N N
1) S
'-S * 0 *0
0 3 -
5 * ** * N
-.
A 0.
* :i'o A. * * *
U) C I $ 0 0 U
U 0.
0 0. U
N * 0.
I.-
4: 0*
F-' A
0. 0.
14 N
453
2.73 SHELLS OF BEAM SYSTEMS
Presented herein are some solutions collected for the thin shells
loads, etc. The shells are cylindrical, conical, and curved panels
454
*4.0
o"
T I
o <A :
!444
n •
Z 4
.. L ""94
Ucc
, > x Z" -CM
I 4
7 z
4 *.
Z- ZZ +
÷ ,.
.t-
•oo 4
455
TABLE 2.73-2. CYLINDER LOADED BY WIND LOADING (REF. 2-9)
Wind t
Direction t
2R 1i x No NxG Nx U wW
Stresses
pn 2
No=-P RcosnO; N= 2RZ (L-x)? cos nO
Deformations
Etv = p
"-P
R2
4R 1
j48 (1 +
4 JR-L (2+ 1)
22- 12 in2 Rx + n~x(6L' 4Lx+ x2)] shin n
-IZR2n2[I + 2(
L2+2 ) Lx - 2x2 +
456
2. 73. 2 Cantilevered Conical Shell
Fig. 2.73-1. A set of such thin conical shells was analyzed, and the
parameters.
LOADING:
P8 = p cos e
(UNIFORM ALONG
MERIDIAN)
457
Extrapolation beyond indicated ranges is not recommended. The
of interest.
Suppose the loading is of the type shown in Fig. 2.73-1 with p = 2 psi
8 1
P 6.67 x 10- in.
Et
Since the case described does not correspond exactly to any given in
results.
458
TABLE 2.73-3. METHOD OF INTERPOLATION TO SOLVE
EXAMPLE PROBLEM
459
FIG. 2.73-2. MAXIMUM MERIDIONAL STRESS FOR =30'
2 max~~1~-~1f
10
11 7t 17
-trill
L=1011
3
pit (b/in
460
FIG. 2.73-3. MAXIMUM CIRCUMFERENTIAL STRESS FOR 30*
17 1 1 1
L=1011 "I", 14 t
T7
.. -JOE) ax:L
10 2 ....
......
111!ij
4 7 7
Tt + iL 4f j-41
t ol
L=10011-
L
14
R=501'
1c, 4i# 1 ::E
ý+7
CD
b N1, R=10011
3 L=50011
10 ........--- -- .....
..
N.
....
k 114 Ht
R=5011 -----
R=2 '0011 2
R=1 0011
lit
-44-
. . ....
R=200"
..........
R sott
=00,#t
-E I K=20ort
10
0. z CL 4 1.0 2-0 4ý 0 10.0
P/t (1b/ tU3 y
461
FIG. 2.7 3 -4. MAXIMUM NORMAL DEFLECTION FOR =300
10-
1 1 j 0
. ..... ...
10 s10
L=10(i11
... 462
FIG. 2.73-5. MAXIMUM MERIDIONAL STRESS FOR t~=60*
ma
o 4 ~~_1 E 0 . 1.
pit 3 44 -1n 4
146
FIG. 2. 73-6. MAXIMUM CIRCUMFERENTIAL STRESS FOR a'=600
Imax 4:
4 -
L' AA
o 41 0~ 3 O 4. 1q
pit (hfin
464l
FIG. 2. 73-7. MAXIMUM NORMAL DEFLECTION FOR a =600
10-4
L=10
i _W 6
i~t
(w4)
_1J 46L
FIG. 2.73-8. MAXIMUM MERIDIONAL STRESS FOR a = 900
10.....I.....
q ....
...
..- .; . .G1
pit ~bku'ij)
466f .. ý.
FIG. Z. 73-9. MAXIMUM CIRCUMFERENTIA L STRESS FOR i=0
-J
-~~ - t k -
L=10" -FF4
0T-0
4674
FIG. 2.73-10. MAXIMUM NORMAL DEFLECTION FOR =90*
10-4 ~ -T+"
A::r
::, axtý : 2
4-I
4~ A
+ 'I
114,
T- I
-3-
10- 2=O"
ME R=101
1468
2. 73. 3 Simple and Fixed Beam Cylindrical Shell
simple beam and fixed beam system under different loading conditions
(Ref. 2-9).
469
f.J
.k7
t
NI
0li h
J.kP'Lk
j N
S. I z -I
.3
4 7
'I'70
o 0
44
00
A e •|.J
K• ' a,, at
, .," *
'aI~
a. a.
2.73.4 Continuous Cylindrical Shell Under Dead Weight (Ref. 2-9)
For the statically indeterminate value, select the moment XI above the
loaded with dead weight (call case 1) and simple beam loaded with the
end moment (call case 2). For reference, see Table 2. 73-4.
"
7 =il~i 11 1iI1 11I 1,J i N'
1 I !4II IJJI 1
472
For X = 0 (case 1), Etu0 _ qL - 6 R ) Cos
where u is deflection.
u = u0 + X1 u1 0
Consequently,
__q~iL U
2 L 2 -6.Rz
u0 qRLZ - 6
x 1 = =
)R 2
u1 4 L +6(1+
finally,
1= 2pL
8 1 + 6 (R/L)z
for
R/L - 0 ... X - L8
and for
R/L -o ... X =0
I
473
The null-point of shear corresponds to
2
L 5L2 +6 (4 +34)R
x0 8 L2 +6(1 + 2)R2
474
2. 73. 5 Curved Panels (Barrel Vaults)
2. 73-5 to -7. The shells under consideration are thin, and linear
475
we
* 4
N 1*4
Z Z Z
U --
@0z
S.476
z.x
dd
'44
P4 U
1
N4
k
>) 4
Z Zr•
0
.u u
C44
C4 4)
aDa 4) u 0
Or r
N
UN 0. .0
0 .. 0
* ,, ,, -
0 0v
e IN A N
+ 111
v~ Nal N0
N . 0
'ZI - .0 S -e- Et f
1. . -N Iwo.. N
N. o 41 .0
a0 0 x x
- ~ ~ I 0.j
II CN WN
.4477
iiHim
I. U -
o Aw
-. 478
N 0
x a .
*0 *0
0 0Na
IIN
0.M ill
IL -4..
479
q06
N 0
A 41 k1
N 06
IL +
Z U ff1
04 U
CS. 16I
LI
- -x
A.o2
z z q-l N
A.
M
I6
xC
44 a
NI
No
.
a. ÷.
MN
, . * " . *
S0
o * • . a- - '
q)4 0 + --. * *+-" a.
=
o,, * • * -.
• •
iu
,4,
H"- 0 -- .0
Ulm L) +
K 1
. -~ ,.481
2. 74 SUMMARY
482
2.80 MARGIN OF SAFETY
2.81 GENERAL
(rupture, buckle)
actual stress)
structure
-l
Margin of safety= failing stress
ultimate stress
- failing stress
limit stress x factor of safety
allowable stress
calculated stress
483
The material will fail if a certain combination of stresses reaches
a certain level, which is called failing stress. The uniaxial failing stress
484
2.82 DEFINITIONS
such that the maximum actual stress a- max s a- allow. Instead of using
defined as
Tallow n '
where
485
some amount, which is satisfactory to the design require-
T
T allow
n
present any difficulty for the uniaxial case, as was evident from the
made, then not the stresses, but certain combinations of such may be
486
2.83 THREE-DIMENSIONAL FIELD OF STRESSES
because then it is not clear what shall be used for allowable stresses.
multiaxial and uniaxial stresses, but only theories that are justified
multiaxial stress system (b) is a-,, for equivalency mustbe or' = (rl. Then
a-I - aallow, where a-1ow is the allowable stress for uniaxial system (a).
E = E
max max
E- -
max E
487
For a multistress system,
max E 1 *3)
Cons equently,
According to this theory, two states of stress (a) and (b) are
TImax r max
Since
0" - O
T 1 3
max 2
for uniaxial
Sur1
max 2
then
T1 1 3
Consequently,
488
2. 83.4 Maximum Strain Energy Theory (Beltrami)
U' = ( l /ZE
stress field)
where o1, a2, and `3 are principal stresses.
Unfortunately, the experimental data do not agree with this theory.
V I+ 02 + (3 1c " Tý,3
2 31 allow
or
I j• •Z2 •3211
+ (0"3 T <-
<1 o
"" 2r" + ( 02 0- 3
This theory holds for the ductile materials only, as will be shown later.
489
2.83.5 Mohr's Theory
two Mohr circles can be drawn as shown in Fig. Z. 80-2. The first two
Consequently, the third circle also represents the same for the
al -~ m ;
all Tc1 a-I m "3- 'all "t
a ll a-t = 1 - - a
all' Ic a'3 - allow. (for uniaxial case)
all (rc
490
SDEENVELOPE
DEGENERATIVE CIRCLE
- 2 -3)
S-T3
491
2.84 CONDITIONS OF PLASTICITY (HENCKY-VON MISES, AND TRESCA)
exceed the elastic limit for uniaxial stress. According to Huber (see
T2- 2-
+()2+( 3 )2 +(T3a -T 1 )2 =TS
or
r2 2 2
r-2 12 + T2 3 +T 31 = CTS
2 + T22+ = 2 2
(C I T2)
1 2 2 1 T
492
It is important to note that the inter-
and later San-Venant, the plastic stage begins when maximum shear is
2 2'1 2
o-1 o"2
If Tl and a-, have the same sign, the max T -- or -2 This
493
the first or third quadrant of the ellipse, the condition Tma = will
T1 = *a- ; r =
1 s 2 °'s
T max= will be represented with the straight lines, that are parallel
to the bisectrix between (rI and a'2 . These lines can be found in
Fig. 2. 80-3. The hexagon inscribed into von Mises' ellipse is known as
= x ± x y + T 2
1,2 =(-2 XY
494
2.85 TWO-DIMENSIONAL FIELD OF STRESSES
2 2
1,3 T/y zy
__ ( y)2 2
2 y zy all.
495
Maximum~hoy Shear y)
Theory: o-z + 4+T . <all
zy al
Maximum Strain I 2 4
Energy Theory: Vrz+ cy z y zy all tens
1-rn
Mohr's Theory: 2 (z + ay ) +
1+rn
F z .-a/( 2 ý+4 T 5a<
2oz y + y allow
tension.
496
2.86 COMPARISON OF RESULTS
However, if the bar is loaded with pure shear (torsional loading), then
The first two theories do not lead to results that tests can
to the tests. The fifth theory gives the following satisfactory result:
497
2. 87 FAILURE
corresponding margin of safety), the structure is safe and will not fail.
elastic limit.
plastic range. This is the case of all pressurized membranes. Since the
i- Ei for uniaxial and multiaxial case are identical in elastic range and
498
only slightly different in the plastic range. The meaning of these symbols
is as follows:
2))
6
- _ f,• . )x +(o(
)2 "-i 2 +(o- -T) 2+ (T 2 +T 2 +T
V +y z x yz zx xy
NF2
= -- 3 (c
ix -E y )++( y - z )+(6- z )+6(c
x yz + zx +C xy )
stress a-u . This does not mean that the same stress will be stress of
the element which is stressed with the uniform tension in all three
directions:
T1 = 2 = ar3 = 0
499
then
'. = 0
zero, and the point of failure on Ei-a- i diagram will correspond to the
the points of failure (several) can be discussed. The test shows that
500
2.88 CONCLUSION
recommended for ductile materials and Mohr's theory for brittle materials.
2
= + 2
y
-
z0
+3
zy - wallow (uniaxial)
p Vz
Margin of Safety:
the point (a-l, or) must be within the areas limited by ellipse or hexagon.
501
The use of presented theory of thin shells would be incorrect in
For more detailed study of this subject Refs. 2-17 to 2-21 are
recommended.
502
2.90 SUMMARY AND CONCLUSION
In this chapter a method was outlined for the analysis of shells and
At the end of the chapter the collection of known formulas for other
503
TABLE 2. 90-1. FLOW DIAGRAM FOR STATIC ANALYSIS OF SHELLS
Chckif.ready
fontlo
foM
Constants. D,I,k, tc. Stefion 2.70 i
250San foces.
disontinity
bf S~cio ee computer
pr.m cm)0 lp
iW isto
i o sr~sret oe s to tns ifdo
l i afne
r doiqt
f onoti
rs e rett io ben e gl e cte d
[ i' to obt i nF or ad ic
SetinhwanndwhchuntS
.2. sabis
ee Chp 5.er00 ate
0 ofidtentload
coeffffnt
wilntb needed
th o inr cas.atwe
IfYe etian
written With fran bar ® odfyttioresonin
Ifboitiin r oo
6. onsricoýIfeadiso.frtins tot
r benn~o forth.o
For e secondary~~~
2.3 solutions r1s1led
Secto odfe onisgvsI
AlTo
fied rectilondcefiiehnee I.4 .64.
Winthvlumfr (1) mnod
).in thiswa
6ax09
Fy c+oom
urýOcofrm te Fmud the mononicaloues
form. cost as wasmit e
colecio co ofm~~u
Frrula teinlSerctioneee 2.64.4 e
seffect orbninon
FoArnoysltions 2.32o siee efftondbadingution be Sctrionred3
Feor
To bton oit-nfiload0To
te noneica obtin
tire nunmerical
ie o uniticd
h
504
REFERENCES
Springer (1960).
505
2-10. Marcus, H. , Die Theorie Elastischer Gewebe. (In German)
Berlin (1932).
2-11. Harris, Dr. L.A., and R.V. Spencer, "Design and Analysis
Astronautics.
2-13. F6ppl, A., and L. F6ppl, Drang and Zwang, I and II. (In
McGraw-Hill (1940).
506
2-18. Levshin, V.A. , Strength of Materials. (In Russian)
ROSTEHIZDAT (1961).
Hill (1957).
507
3.00 PROCEDURES FOR STABILITY ANALYSIS
3. 10 GENERAL
large increase in the deflections of the shell, which may or may not be
accompanied by a change in the basic shape of the shell from the pre-
shell may be gradual, and the actual buckling point is rather arbitrary.
However, for most types of shells and loading conditions, the buckling
The load carrying capability of the shell may or may not decrease
after buckling. This depends on the type of loading, the geometry of the
shell, the stress levels of the buckled shell, etc. Only the buckling
collapse load are nearly the same and, if they are different, the defor-
508
For columns and flat plates, the classical small deflection
theory predicts the buckling load quite well and, in general, the
not be used for shell structures. The buckling load for some types of
shells and loadings may be much less than the load predicted by classi-
cal small deflection theory and, in addition, the scatter of the test data
particular metal, none of the cylinders would fail at the same axial
compressive load. In fact, the scatter of results may range to 500 per-
cent at a given time, and the average buckling load may be one eighth
able buckling load. This method has been used to determine most of
the design curves for unstiffened curved panels and cylinders preaented
509
A best fit curve is determined for a given set of data, and the
probability level. The probability level used for the statistical design
shell is subjected to the design allowable buckling load, the chances are
nine out of ten that the shell will not buckle. The load at which the shell
may be expected to buckle is the load which corresponds to the best fit
curve. Best fit curves have not been presented in this chapter because,
in design analysis, the load at which the shell will not buckle is the
loads less than the load corresponding to the best fit curve.
design curves is that the test specimens and boundary conditions used
structure which the design curves are being used to analysis. However,
curves.
510
available information. In general, this involved recommending cor-
lack of data for some types of shells and loading as well as the question
edges should be assumed unless test results are obtained which indi-
rotation in the axial direction. For clamped edges, the rotation of the
edge is zero.
so that the design allowable buckling loads may be obtained from hand
computations and graphs. The analyses which have been presented are
511
general, results can be obtained quickly with a few simple computations.
warranted.
abreast with changes in the state of the art because significant changes
512
3.20 UNSTIFFENED SHELLS
3.21 GENERAL
lished only for the more common loading and edge support conditions.
The design curves that have been presented in this section for unstif-
have, in most cases, been obtained statistically from test data. The
load of the cone is obtained from the buckling stress or load of an equiv-;
alent cylinder. The definition of the length and radius of the equivalent
cylinders depends on the types of loading. The available test data for
513
3.22 CURVED PANELS
A. Unpressurized
a' 2. .2
cr nT _E_ t2
1 Kc 1Z(l - V
parameter Z > 30 and for fixed-edge curved panels having a Z > 50.
Fig. 3.22-2 may be used instead in Fig. 3.22-1 to compute the critical
rcr Et
ni- c R
buckling, 71 = 1.0. For inelastic buckling, the critical stress, 0 cr, may
Note that the design curves in Fig. 3. 22 -1 or 3. 22-2 are valid only
for a/b>0. 5.
514
B. Pressurized
tensile load in the skin. Note that the design curves in Fig. 3. 22r3
are valid only for a/b > 0. 5. In addition, the curved panels must fall
shown in Fig. 3.22-1. For inelastic buckling, the critical stress may
515
LUU
C4
0*
0~ u
C44
C;I ONNN
'-b4
4 a4 0 (44
16a * q 44 a *'*C4 % co 4
-0 V
516
., . 4 4 ~
p.4 - - ~ ~-
- 77 -- 7..-.-. -
.. ._
...............
.1
-u . u
4
CIA
Tj
0 HU
+ +:44 :2
-17
FIG. 3.22-3. INCR.EASE IN AXIAL-COMPRESSIVE BUCKLING-STRESS
COEFFICIENT OF CURVED PANELS DUE TO
INTERNAL PRESSURE
10
8
6
2 - - - -- - - -I I-
0.1 1 1. ]0
111, 1
0.01 2 - 6- 8 26
0.01 0 0 1 1
2 68 2 4 68 24 68 2 46 8
0.01 0.10 1.0 10 102
51
5181
3.22.2 Shear, Curved Panels
A. Unpressurized
Tc- K
is12(1 -
[tf
in which the buckling-stress coefficient K. is given in Figs. 3.22-4
relation between Tcr/rI and Tcr may be determined from the (rcr/Il
T -0
xy %r3 Y
519
B. Pressurized
3. 22-5. The curves in Fig. 3. 22-6 allow the calculation of the increase
must first be computed from the design curves in Fig. 3. 22-4 or 3. 22-5.
which the axial tensile load caused by internal pressure is not balanced.
For inelastic buckling, the critical stress may be found by the procedure
is determined.
520
FIG. 3.22-4. BUCKLING STRESS COEFFICIENT, K., FOR UNPRESSURIZED
CURVED PANELS SUBJECTED TO SHEAR
6 6 Icr = Kt. 2
42H
3
i0 i
2 -
1.2
IL -
1.0 10 to 10
z
521
FIG. 3.22-5. BUCKLING STRESS COEFFICIENT, K 8 , FOR UNPRESSURIZED
CURVED PANELS SUBJECTED TO SHEAR
8
6
12
aR
10
a>~
6
2 2 6520
'00i~
102
3.
22
- -Y p41 i -
~Z4
j I
77§ 4T -- '1
1. 4 ~.::
rz
0 14
0~~iT'l IV
14.
r.4 . . . . . .
.4.
-1 -4-
- I
+ ': ~ $
11:
.. V4
I'I
*4
140 VI_4
- 4-
g ; C; C; C4;.
N .... .523
3. 22. 3 Bending, Curved Panels
approach that for a long, flat plate in bending and, at high values of Z,
These extremes are plotted in Fig. 3. 22-7 with smooth curves faired
Tcr 2 2
I b 1b2(1- 2) b
524
FIG. 3. ZZ-7. CRITICAL BUCKLING STRESS COEFFICIENTS FOR
LONG CURVED PANELS SUBJECTED TO BENDING
i, i oo I_ i iiII1 I i1 ll / loo -
1,000 00
tlit
ioo [ -=
Kb
I0-
SIMM EDO1S
L AYY M SUPPORTED ES :
_C
.
1L
1 10 100 1,000
z
525
3. 22. 4 External Pressure, Curved Panels
torsionally weak and the circumferential load in the skin is not applied
use the curve given in Fig. 3. 22-8. a straight line is drawn through the
526
FIG. 3.22-8. BUCKLING STRESS INTERACTION CURVE FOR
RECTANGULAR CURVED PLATES UNDER COMBINED
SHEAR AND COMPRESSION
l ~e l fi l l IiI
I~ I I I. . . T%I I J l l 6c
1.0 : ' S. . . .I I I I I Ji l l
: : : : I I I Jl l
0.8
0.6
2f I
oiii:
R0
0 0.2'• oi06
III5il2 . .
527
3.23 CYLINDERS
A. Unpressurized
O'c r Et
c R
For simply supported cylinders with the curvature parameter Z >25 and
T1= I is used. In the inelastic range, the critical stress, a-cr' may be
B. Pressurized
tion with Fig. 3.23-1. Fig. 3.23-2 presents a curve that allows the
528
The design allowable buckling stress is
•rcr Et
r= (Cc+ +ACc)t
71 c c R
Fig. 3. 23-2. For inelastic buckling, the critical stress may be found
P = 2wR o t + 1TRp
cr cr
valid only for long cylinders. Very long cylinders must be checked for
529
i'• ,.tI I.•
jlI-' '"-L:
I I1 4
..
41 . .. . 41 ....
" .....
U2
Z2Y "T--- •-, .4 -1
.
u•
1 . *...............................
". .......
u
";"•
.•
',•.
........... .... .... . . .
0tt tit,- -
i .. . .. . . . . . . . . -
- 4 +1.... ..
44__
03
44 . . '. ...
- -" "' .
44 -4 N a
1:4t
530
0U20
C4
- Ib I
- N-
z 40
>4
0.0
-44
.0
531
3. 23. 2 Shear of Torsion, Unstiffened Cylinders
A. Unpressurized
Tcr Et
1 - s RZl1/4"
curvature parameter Z > 100. For elastic buckling, the plasticity cor-
rection term i = 1.0 is used. For inelastic buckling, the critical shear
B. Pressurized
in conjunction with Fig. 3.23-3. Fig. 3.23-4 presents curves that allow
T
- crr= (C + AC)Et Et
RZ
Fig. 3. 23-4.
532
Two curves are presented in Fig. 3.23-4 for calculating the
labeled "No External Axial Load," should be used for calculating the
pressurized design curves of Fig. 3.23-4 are valid only for long
total stress field should be taken into consideration when the plasticity
correction is determined.
533
4. 4 Ii;;-
.4 1
IzIt
+- --- - 4
0 cc - IiL
- -_T
tfMT. ..........
4
t . I~4j
V - .1 .... ...
jS41
t.. 4 . .. ...
4 11
4'u
4ill -4:
-1:- 1
o a C .1 0 0 0
-1 -4
534
r34
0 0 (0
ue
114
--- 4
x C4
~CJ4
"0~
a _ ga
N , 0
UiI •
I NN
535
3.23.3 Bending, Unstiffened Cylinders
A. Unpressurized
Tcr
Et
T1 bR
where the buckling-stress coefficient, Cb, is given in Fig. 3. 23-5 for
for clamped edge cylinders with Z > 80. 0 cr is the maximum stress
due to the bending moment (e. g. , the outer fiber stress). For elastic
M =nR Z or t
cr cr
B. Pressurized
with Fig. 3. 23-5. Fig. 3. 23-6 presents curves that allow the calcu-
Fig. 3.23-6.
536
Two curves for calculating the increment in critical stress caused
equal to the internal pressure load, 1T R-p, acting on the heads of the
cylinder.
If the curve for no axial load is used and the stresses are elastic,
Mcr TR R [rcrt + j
It should be noted that the pressurized design curves in Fig. 3. 23-6 are
valid only for long cylinders. For inelastic buckling, the critical stress,
acr' may be found by using curves E 1 in Section 3.62. The total stress
correction.
537
'4 A'-
.lf '44 44
-4*
41
1,
4, _+
0 j 2 ~ 4 ~ -
44.
14 -
aU~
LA~) 14 441 --- f
0 C4
A5A8
H10~
u w0
Lo0
C4
A 4
zE
0'
000
53
3. 23. 4 External Pressure, Unstiffened Cylinders
q p 12(1-1 2 ) QL
where
ES = secant modulus
Et = tangent modulus
curves does not vary appreciably from the value obtained from the
540
sufficiently accurate. For long cylinders, (e. g. , L Z/R >11 R/t) the
(rc r Y Et)
n 4 (l-t±2)
E
E(s + Et)
Sufficiently accurate values of 0 cr may be obtained by using the
rcrt
Pcr- R
buckling of the shell (e. g. , when buckles have formed all the way around
the cylinder). For some values of the parameters (large R/t and/or
than pcr' but complete buckling will occur at higher pressures. There-
fore, for some applications these results should be used with caution.
541
The plasticity correction factors recommended in this section
were obtained primarily for the case of lateral pressure, but they are
probably sufficiently accurate for the case of lateral and axial pressure.
542
T-T
04 0
t i -
£0 21
>4 Cd
0
044-
543
3. 23. 5 Combined Loading, Unstiffened Cylinders
a combined loading condition to the allowable stress for the same kind
to allowable stress.
combined torsion and axial loading is given in Fig. 3. 23-8. rcr is found
544
from Paragraph 3.23. 1 and Tcr from Paragraph 3. 23. 2. In Fig. 3.23-8,
the curves for R/t ratios of 600, 800, and 1000 were determined by test
data in Ref. 3-8. Curves-for R/t of 1500 and 2000 were drawn by extra-
polation.
The test data presented in Ref. 3-7 and 3-9 indicate that the
linear interaction for the case of cylinders under combined axial com-
buckling stress due to bending alone maybe found from Paragraph 3.23.3,
and the buckling stress under axial compression alone may be found in
Paragraph 3. 23. 1.
The limited test data from Ref. 3-9 for cylinders subjected to
axial compression and external lateral and axial pressure indicate that
the linear interaction curve presented in Fig. 3.23-11 may be used for
design. 0 cr is found from Paragraph 3. 23. 1 and Pcr from Paragraph 3.23.4.
545
oo0
U *0
0'
00
0.4 IF
0 0
(fIf
dosd
- - - - - - -0
a54
1-L
00
Cl) 0
0 0
E 0
0 -44
00
UU
t547
u 0
4-
00
z '0
00
0-4
0 94 toN Elf
- 0 C 0 0Il
v4 1 U to (
0bo -- 548
z b
-- o
P4 c0
0'
00 0
P4 I U 4Ic
(34
0 CL
0O N
0 04
U u 0
454
3.24 CONES
A. Unpressurized
be used for determining the buckling stress for a circular right cone sub-
cone data presented in Ref. 3-10 indicates that the equivalent cylinder
the results of Ref. 3-10 would be unconservative for small cone angles
(e. g., cones that are almost cylinders). The design-allowable buckling
-r-
ircr Et
a- is the stress at the small end of the cone. The buckling stress
cr
coefficient, C c, and a definition of the geometrical parameters is given
curve of C versus Re/t given in Fig. 3.24-1 for cones is the same
P 22•R (r t coo2a
cr e cr
550
B. Pressurized
cr =( + Et
Y -C ( + c
e
Section 3.62. The total stress field should be taken into consideration
551
the buckling load of cylinders. However, test data for pressurized
552
iti
N u p... "
N
zl Z -7 v ...-
E-_4
--
-
-
~Z
~ .......................... , . .... .....................
N0
B . ,,• . .1 .. . •: 1i'
P4t l !]]Ua
, U!!• ;,, , ..... .....
-
u ..tit- - ttffI - t *1
i 4 4 'y
0 o 0 " 0 0 €0
553
L)
00
0 4
+ IOU
0'
i P4
o 0 go 0Vc 0 't
554
3. 24. 2 Shear or Torsion, Unstiffened Cones
A. Unpressurized
R2
T"rcr e Et
S2 R1 s RZ Z1/4
1 e
T cr is the shear stress at the small end of the cone. The buckling stress
T - 2R t tr
cr cr
B. Pressurized
The theoretical results and the test results of Ref. 3-12 show
that internal pressure will increase the torsional buckling load of cones,
but simple design formulas for computing this increase are not available.
555
41i ýt i W Tl
- •0 .. ... . . _
0 d N I- N + - v I
.4....... . --
o + E : .. .. . .. .....
oo 0n .. . . . . -. .. . -.-
-
0
+. . . . . . . . . ..
. .
S-...-.. .. .. ...... . .
".,,... . .
t iF . ....
*-,4
4j 4 ; 1# y ' .. K : > :
....
,- z . .... ... .. .... .
I t 1ý t-t ;t
~~41
556
3. 24. 3 Bending, Unstiffened Cones
A. Unpressurized
Orcr GEt
S- b --
e
where (rcr is the maximum stress at the small end of the cone. The
Fig. 3.24-4 for cones is the same curve as that givenin Fig. 3.23-5 for
the critical stress, a-¢r, may be found by using curves El in Section 3. 62.
If the stresses are elastic, the allowable moment may be obtained from
the formula
2
M : rR a- t conar
cr 1 Cr
B. Pressurized
557
stress as a function of pressure and geometry only. The design-
Orcr Et
I (b b) R'
e
Section 3. 62. The total stress field should be taken into consideration
elastic and no external axial load is applied, the allowable moment may
3
M -= Rt ta- coson+& p--
I cr
558
. . .. . . . .
j *.-4.* -4
:IL. ..
--
4.4.4.. ~ 4.
00 -
- -j :
.. Iv . .... ..... 41
. . . . . . . . . . . . . . . . . . .
.I
.......
U> N Ni44 -
-. . . . . . . .
44, IT T- -14- -
-4 .. ;:441 it1 ~, 4
4 14 4. 4
-t-44
o4 - , - +. . .4. . . .
*i*+4~ 4~
4 4
d ~~ R
559
0
z
COff
cjUj
0 0fz
op.
I I I
a."
~L) 4
U 1
cl 0 V0ta 0 f a .0 V C4C
d c;C;
560
3. 24.4 Lateral and Axial External Pressure, Unstiffened Cones
•cr K ITj->)(E_)R 2
i P 12(l - ecoso
0cr is the circumferential membrane stress at the large end of the cone
Fig. 3. 24-6 is for simply supported edges and will be conservative for
the critical stress, a- cr, may be found by using the method discussed
C" tcosao
cr
Pcr RZ
when buckles have formed all the way around the cone).
of the shell (e. g.,
R
For some values of the parameters (such as large - or large initial
t
561
imperfections), single buckles will occur at pressures less than p cr
function of the quantity (1-R 1 /Rz), but the effect is generally small and
the available information has not been reduced for design purposes.
562
FIG. 3.24-6. BUCKLING PRESSURE COEFFICIENTS FOR CIRCULAR
CONES SUBJECTED TO EXTERNAL RADIAL
AND AXIAL PRESSURE
103
10 +K2 R2"
4L/2 Re = RI +-R
(rcr t cos a
f 2z pc R?
Ree
R -,
102
IR2 a < 75"--
-- 4---€--I-+--4
• .. .. -4 - - -4 - -
===
22
Kp : -i ii- ' - H
--4+-
- -63
If l
1I1 f i llI
563
3. 24. 5 Combined Loading, Unstiffened Cones
Ref. 3-13 has shown that for unstiffened right conical shells, the
curve given in Fig. 3. 23-8 may be used for predicting the interaction
The very limited test data in Ref. 3-9 indicate that the linear
interaction equation shown in Fig. 3. 23-10 may be used for right cir-
buckling stress due to bending alone may be found from Paragraph 3.24.3,
and the buckling stress under axial compression alone may be found
in Paragraph 3. 24. 1.
The limited test data from Ref 3-9 indicatedthat the curve given in
Fig. 3.23-11 may be used for right circular cones subjected to axial
564
3.25 SPHERICAL CAPS
The following figure shows the type of shell and load that will be
r = C9-Et for
4< X < 24
C = 0.175
P
565
where I 1
2t
cr Urcr R
shown in Ref. 3-14 that for 6< X < 24, the increase in the theoretical
for 4< X < Z4. The quantity C P = 0. 175 was obtained by a statistical
reduction of the test data from Refs. 3-15, 3-16, 3-17, and 3-18. The
test specimens in Refs. 3-15, 3-16, 3-17 and 3-18 failed considerably
below the theoretical curve given in Ref. 3-14. One of the primary
caps with very small imperfections, as reported in Refs. 3-19 and 3-20,
566
was approximately Cp = 0.39. which indicates that the buckling stress
not felt that the shells tested in Refs. 3-19 and 3-20 are typical of a
production part.
Spherical caps for which H /Zr >1/8 (see Section 3.25. 1) have
not been discussed in the literature as often as spherical caps for which
H p/2r < 1/8 due to the added complexity of the problem. Theoretical
buckling pressure for deep spherical caps would be greater than for
shallow spherical caps of the same radius and thickness, but this has not
been verified experimentally. The test data for deep spherical caps sub-
small area of the shell. Therefore, the area of the shell that participates
allowable buckling stress for spherical caps withX > 24 and Hp/2ro >118
567
The test data from Ref. 3-21 indicate that the inelastic properties
of the material influences the buckling load even though the nominal
obtained, the analysis which has been presented may be used for hemi-
presents test results for spherical caps with various boundary con-
568
3.30 ORTHOTROPIC SHELLS
3.31 GENERAL
ties of the shell wall in the two directions may be completely independent
stiffness, and twisting stiffness of the walls of the shell are not neces-
shell. For the case of buckling, a buckle must include several stiffeners
buckling load.
orthotropic shell wall are given in Section 3. 32, and approximate for-
mulas are given for computing the stiffness properties for several
types of construction,
569
The buckling formulas presented in this section are based on the
and theoretical studies indicate that the discrepancy between test and
theory as well as the scatter of the test data may be much smaller for
The analysis presented assumes that the centroid planes of the orthotropic
wall in the axial direction coincide with the centroid planes of the wall
the analysis. Although the effects of coupling can be large for some
types of construction, they are usually small for the types of construc-
homogeneous orthotropic).
570
3.32 ELASTIC CONSTANTS
3. 32. 1 Definitions
formulas are given for computing the elastic constants for several types
chapter (Chapter 3.00) for stability analysis. The formulas for the
constants have been derived for flat orthotropic plates but are suffi-
571
ILxp' = Poisson's ratios associated with bending in the x
(lb/in.)
From the reciprocity theorem, it has been shown that the follow-
Bxwfe = Befi'.
The elastic constants required for Chapter 2. 00 are Be, Dx, ILxp
I'0' ILlx$ IL1e and DQx. All of the elastic constants with the exception of
DQ0 , are not needed in Section 3.33 because transverse shear deflec-
used for the stability analysis of sandwich shells. However, if Section 3.50
made from a different material), and if the core is very stiff in trans-
572
shear stiffness is large), the design buckling load can be estimated using
the formulas from Section 3.33 and the elastic constants given in this
section.
in Fig. 3.32-1. It can be seen that there are m layers. The inner
layer is 61 thick, the next layer is (62 - 61) thick, etc., and the total
respectively
1
4x i I I') = Poisson's ratios associated with stretching in the
573
Let
E
X.
. iL
x I'e
= I~L e1i
i T-- X Fl
o0 =0
Then
z c (62-6 1~
x m (
2E0 e xi - i1
Ae = i=1
Zn 2 *Z
zc(6
EL i-
574I
m 1
(62
2EG x 6 i 6 1-)
i=
B = xC (6. -i
m
Be ZG E(6 i -i
i=1 l1
In3 3(6
D E ~c[( _6 1)>A 3&(2 ,?-
x 3 i=I xi[6 -i I
) - A6
3- -
D .E 2) 6
Gx[(6 3 - b Xe, (6l x0(6 , 6i1?
G =e ZG .6-
1
= D
5,75
Im
x i=l i"
B
x
DQx and DQ are effectively infinite for most layered shells (if
B = t
x
Ix
1
B9 = C0 t
D -
x 12
D
0 12
DO = G 12 t 3
xe I
Dxe = 12
Gxe =G t
576
576
x xI x
B x
= Be 1 oEt 2
D =Et 3
Dx
DX =D
x GtGt 112
x e
D = tGt=
G t = G, =t=,, - w
e 12
Xe
577
3. 32. 3 Sandwich Shells
(such as honeycomb core) and thin facing sheets relative to the overall
Facing sheet #2
centroid
of facing
respectively
578
Sub-subscript i represents the facing sheet to which the material
Gxz and Go..z which are the transverse shear moduli of the core in the
E
X.
C - 1
ES0.11
C0 . 1
i 1l- x.1 i0
B x.
CE .
1. 1
B =Ct C
Bx xlt+ CxtZ
Be = C0 +C 0 t2
G = Gx It 1 Gx e 2t 2
579
G xeIG 0zt Ith 2
D X G X0 t1 +
e2 2
G
D
xcxO 1 x e2t
0 1 02
x I t+ +t
xl
1 2
x C t 1+ C 2t
D 1z
Q B6
xO
GBz
DQC) x
C5C80h
B. Sandwich with Isotropic Facings
BB - E _t t1 )
x - (1 2 1 2
E tlthZ
D :Do E ) I z
xL 1 - 2)t
G = G(tI +tZ)
t It2h2
Dxe =G tI + t2
I
X=
x 'e =fix = ýe = F
G h
xz
DQ - cx
X
Gzh
- Oz
DG
Qe
Ref. 3-26. Fig. 3. 32-3 shows the type of construction being considered
581
The definitions of the material properties are:
E = Young's modulus
G = Shear modulus
ý = Poisson's ratio
B = E-
x x (1 - IL)(y)
-yI y/
B EH (As)
B ,:•
EH s
I A A
xx 3s x - 2•
Dx = EH3 Ix 2 (kx ks
D =EH 3(1ep)i)
D = EH 3
18"2
DQ Go
x
D =
582
Gx - EH(Ax,)
-iz
Is
IAxA AO -k),
s 0
A
s A 0
s sO
l--2
where
II
s2 Is 6 + A sA x A0 (k x/
A :s(k
k" k 6s
A~ = A t w
AS snT o2
A s = A I s+ s cos4T
x 2 H H
58
A A= -AH + H-2
s sin4T
A tS A/b
H
sin T cos T
A8 Z(+
D= ±H
A lb
1 ts W 4
k k cosnT
x A HW S w
x _ S
A /b
k __ sin T
k0 Ae H w
A /h-
Ws a - 2 2
kxk H
sinIcou
w
A /b
-W S is2
-- H w sin TcosT
Ws
W S
ik sn1 cos2 T
Sl
k -cos T+s
A /b"
+ H ws xj
1 ibin4
+)iI T+ I HFke)
12(1-u (IH H32H
A /b
4T
+ w s sin2T-os
584
A\ 1 w/b t
I i sin2T cos2T +- -- L ((k)
A/b
Hw
A 3 I/bb
+4 + 2T
sin 2TcoBT --
A~
~bs lb 2 s ts t ws ts
+4 ts ) Bin2 cOB 2
A/b T r .4t
w w I w t
Ss VWS S
Iw/b s 3 + t A 3
w B w
H -
585
The special case of T= 90 degrees corresponds to a shell with
For more exact formulas of the elastic constants for this type of
586
FIG. 3.32-3. WAFFLE CONSTRUCTION
\4/ta
(b)Waffle Section
587
3.33 CYLINDERS
Refs. 3-27 and 3-28. This analysis may be used for layered construc-
togehter. The definition of the elastic constants used in this section and
the formulas for computing the elastic constants for typical types of
load per unit width, Nx, for moderate length orthotropic cylinders is
Nx = y[ B Dx (0- W'We) U
Pcr= N X ZiR
~cr xN w
constants
G-- ZGxe
BX0 1+J(
±'±e)
D +
B8
BPG2
588
BD
Sxe
2 BeDx
w3
For o2_> 1, w3 a 1, U 1.
- .
+2[ 2+ (Be /Bx ]
1
for
2 =
~=sel Si+2
22
of 42 are greater than zero, then U 1 must be computed for each 412 and
589
The parameter U? can be obtained from the formula
where
L 2 Bn(I
2X4 ) L4
12 D R 2
x
found. The quantity nis the number of half waves of the buckles in the cir-
from Fig. 3. 33- 6, where P is the radius of gyration of the cylinder wall.
direction, Ref. 3-29 recommends using the geometric mean of the two
4D D
BB
x 6
590
The term Y /0. 49 was introduced to make Y a function of ratio
R/p as suggested in Ref. 3-29. The curve given in Fig. 3. 33-6 forY 1
sistent with this chapter and is-not the same as in Ref. 3-29. The term
the ratio of the postbuckling load to the classical buckling load for
If the results of this section are reduced to the special case of a homo-
verified by only a limited amount of test data and caution should be used
described in Ref. 3-28 and 3-31, indicate that this method may be con-
servative but the specimens tested were not typical of large production
parts.
U3 -- I/ + + -'O,4 I+ 2 2 +. 2
1/,z
591
The parameter Z is as previously defined
ZZ = B- (-• .- L4
12Dx R2
For axial-stiffened cylinders (Dx > 300 Do) in the short cylinder
2 7y
cN D+ R ý/1 _--DO
Nx = L2 1
of this formula has been limited and the range of validity is not well
consistent with this section and is not the same as in Ref. 3-32.
mended that Section 3. 42 be used unless the stringers and frames are
very close together. The test results of Ref. 3-33 for cylinders with
light frames and heavy stringers indicate that the analysis presented in
spacing.
592
FIG. 3. 33-1. CORRECTION COEFFICIENTS FOR ORTHOTROPIC
CYLINDERS SUBJECTED TO COMPRESSION
s3 =
De)t
0.8B
0.6-
NNO
0.2
54 =0 54 =0.2
593
FIG. 3. 33-Z. CORRECTION COEFFICIENTS FOR ORTHOTROPIC
CYLINDERS SUBJECTED TO COMPRESSION
0.2
0.4 * - - D
-
0. 2. II 4. 8. 0. . . . I.
594
FIG. 3.33-3. CORRECTION COEFFICIENTS FOR ORTHOTROPIC
CYLINDERS SUBJECTED TO COMPRESSION
1.0 B8 D_,
Dx Be - (8
S3 - _iS4
-- Do
De B -
0.8
0.6
Y2
0.2
595
FIG. 3.33-4. CORRECTION COEFFICIENTS FOR ORTHOTROPIC
CYLINDERS SUBJECTED TO COMPRESSION
I T TT Dx Be
3S3 =
De B
0.23
53s
III
59
0. 4 0.
02
0.33
$4 o:4
0.1 T I SI 0.I 5
0.2 . ....... 1 ,X
t *-
7..q.-•,•--. t
÷+÷N+I•-J,
4;-
0010
S3
S4 = 0.5
596
FIG. 3.33-5. CORRECTION COEFFICIENTS FOR ORTHOTROPIC
CYLINDERS SUBJECTED TO COMPRESSION /
Be.s = D lZB
s.•D_
D ,S5=
0.3 Do Bx
. . .. .
0 .1
444
410-
,.-- ! 1 0
F 10i
0.1 S5 3
0.3
S3
2
S4 = 1.0
;4
5 : 8 .0 •
444 1 0
1
S3
$4 = 2.0
59'7
-. X
01
0 $4 II
11i
.: 0 0 ~ -7 , lH
-- l H
Ing2 d . 0
LE I.- t
0'~ 0 ,, l
0I 0
i a :. pi
]I~1i1j 11
en 14
zx -; -; - -
598I~
3.33.2 Torsion, Orthotropic Cylinders
The curves in Ref. 3-34 will be used to determine the buckling load
YK wr2D
N - x
A L2
T cr = N x0 2irR 2
for the elastic constants as given in Table 3.33-1, andY may be obtained
from Fig. 3. 33-8. The definition of the elastic constants used in this
section and formulas for computing the elastic constants for typical
It can be seen from Fig. 3. 33-7 that for large values of Z., all curves
merge into one line. The equation of this line as given in Ref. 3-34. is
K =0.89 Zs 3/4
599
where
' y')L4
2 Be (I-p
z = x
12 D R 2
x
estimate the critical torque for geometries other than the ones given
in Table 3.33-1.
This section should not be used for any type of cylinder with skin
The test data presented in Ref. 3-33 indicate that the results may be
1 0 1/8 4 1
2 0 1/2 1 1
3 0 2 1/4 1
4 1 1/8 4 4
5 1 1/2 1 4
6 1 2 1/4 4
7 1 4 1/8 4
8 1 8 1/16 4
600
FIG. 3.33-7. BUCKLING COEFFICIENT FOR ORTHOTROPIC CYLINDERS
SUBJECTED TO TORSION
2 3 4
104
T -0 41
L 71EH i , ýý ýTT
YK8 Dx
Nxo
LZ
10
3 Tcr= Nxe 2 wR 17
3/4
102
0. 89 Z
F,- ;7:ý -
Tit:
T
MITI
T ... .
....
3 4
3 h g LEH 11:ý-
T-1
-4-
IA 4 1
4"-, -4ý
10 ILL
10
4 .10
10 2 103
601
wA
.. ... ... ..
. ... ... ..
... ........
... .. ....
Y='2 +
S(Y + 20. 062)
2/0.64 Yi
obtained from Fig. 3.33-9. Fig. 3.33-9 was obtained from Fig. 3.22-5
due to the bending moment (e.g., outer fiber load). If the stresses are
603
FIG. 3.33-9. DESIGN CORRECTION COEFFICIENT FOR CYLINDERS
SUBJECT TO BENDING
0.65 of cylinder
n wl
P = radiusofgrt
0.60 • ~ ~for
orthotropic yidr
0.60D 4
(tl + t21
0.50
0.45
0.40
0.20
0.10
0.40
1I 3 456 78 91 2 3 4 5 67 8 91 1.5
xl 02 xl 0 x104
604
3.33.4 Lateral External Pressure, Orthotropic Cylinders
is
YK W2 D
Pcr 2
RL
of the elastic constants and formulas for computing the elastic constants
Fig. 3.23-7.
Therefore,
It can be seen from Fig. 3. 33-11 that for large values of Z p all
605
and the design allowable pressure is
Pcr =LR3/2
estimate the critical pressure of geometries other than the ones given
in Table 3.33-1.
K= - [1+2D + ,4 - - -2Z 14
p i2 Dx I + 2 4G2 + (B /Bx)
nL
I4L4
Be (I- ') L4
1Z D R2
x
than or equal to 4.
3Y DO
Pcr =R3
606
The point at which the buckling pressure becomes independent of
shell (e. g., when buckles have formed all the way around the cylinder).
If single buckles are not allowable for a particular design, the pressure
607
FIG. 3.33-10. BUCKLING COEFFICIENT FOR SHORT ORTHOTROPIC
CYLINDERS SUBJECTED TO LATERAL EXTERNAL PRESSURE
Pcr Kir2 D
RL
ZP (Do/D,)' (Bx/Be)144,
K~~1 D. R211I
160
lit~
1: -i
Oz z
INN
00
0 0 Q-
060
3.34 CONES
The design allowable load per inch, Nx, for the equivalent cylinder can
2
P cr = ZTR e Nx cos 2
610
FIG. 3.34-1. CONES SUBJECTED TO VARIOUS LOADINGS
RI R, •
RLe
I>
i1 -. I
R1 R
I'R
Re
(c)
(()
611
3. 34. 2 Torsion, Orthotropic Cones
tropic cones subjected to torsion. The cone shown in Fig. 3. 34-lb can
R + RCosa~
The design allowable shear per unit length, NxO, for the equivalent
tropic cones subjected to bending. The cone shown in Fig. 3. 34-1c can
The design allowable load per inch, N , for the equivalent cylinder can
612
be obtained from Paragraph 3, 33. 3. If the stresses are elastic, the
M = nR N cos a
cr 1 x
R 1+ R2
e Z cos a
613
3.40 STIFFENED SHELLS
3.41 GENERAL
for three modes of failure: (1) material failure, (2) buckling between
analysis, but the range of applicability of the method is not well defined.
614
3.42 FRAME AND STRINGER STIFFENED CYLINDERS
it may fail in one of three distinct ways. The types of failure are
A. Material Failure
615
properties. The stress caused by the applied moment should be
heavy frames and light stringers; the cylinder tends to act as a number of
modified by the effect of the attached sheet. The frames will remain
will be to determine the end fixity coefficient of the stringers. The four
are as follows:
Although there are four distinct instability modes, the ultimate buckling
Buckling of the sheet between the stringers and frames does not
616
buckling stresses, acs of the sheet must be known to determine the
on curved panels.
sections of the stringers are distorted in their own plane but not
figure.
617
Crippling generally occurs in stringers having wide thin flanges.
________ bl -j"'--b -
1ti
b3 b3.
618
The allowable crippling stress for the total stringer is
T T Ai)/ Ai
Ai = Area of part
equal to T-cyiO The amount of sheet to include with the part of the
619
TABLE 3.40-1
'Tb1
620
TABLE 3.40-1 (CONT)
rotates but does not distort or translate in its own plane. Typical
b$ t
bf
(a) Antisymmetric
621
(b) Symmetrical
the case of cylinders with typical ring spacing, Id>7r(EIG r/k) 1/41 ,
the allowable torsional instability stress, tct' for the mode shown in
a' [-IJIzLIlr E nk
where
rotation (in. 4)
p p
two commonly used types of stringers
622
J = Torsion constant of the stringer
(GJ = torque
twist per unit length
1 1 + 1
-k k
k kweb kshe et
Et 3
w
k -
web 4b +6bf
3
Et
ksheet 1I b
(%~t S)
- Mco
1/3 1 1 + 0. 6 for the antisyrnmetric mode
a=
cs Compressive buckling stress of the sheet
623
Correct X if required, and repeat procedure until desired convergence
is
ct IA(AG + I ) ITIr) +
p p p
where
p P
and J/I and %'fTI may be obtained from Figs. 3.42-1 or 3. 42-2.
p p
The formulas which have been presented may be used for stringers
with sections other than those shown in Figs. 3.42-1 and 3. 42-2 if the
624
FIG. 3.42-1. TORSIONAL SECTION PROPERTIES FOR LIPPED
Z STRINGER - SHEET PANELS
0.16i II I i
0 .0 1 0. 0
0.0 .7 - -0
0.6 0.- : i
A I I I - -L
0.1 0
0.8- - .3
"•-p
I/• • I -------- If
lh-. Ii I if
stt
'II AiI
Fr Fbwt] I h II
"xI"'
1"
I I II I
0.14
0.6b0
Ip
NI
Xj
V
v I
0.3
0.5 i
I
o
0.4--
625
0.3
0"0.9 O.3 0.4 a5S bf 04 0.7 0.0
625
FIG. 3.42-2. TORSIONAL SECTION PROPERTIES FOR
J STRINGER-SHEET PANELS
.06
ip
.046
.02w 4 5
3 1 3r 11
Ip~b J
tI I0 II - -
D4 -- --
2-------- --
-
62
in alternate in-and-out waves in which the frames are nodes. In such
case, the panel will act as a pinned end column and the effective end
certain restraint to the outward buckle mode will exist because that
deflection mode will involve some stretching of the sheet in the hoop
present time and the use of fLxity coefficients greater than 1. 0 must
be substantiated by tests,
effective width of skin, we, which is acting with the stringer. Two
following illustrations:
ve ve e Ve ve. ,o ,,
627
For calculating the effective width of sheet acting with a stiffener.
the following equation has been found to give results consistent with tests:
we = .85 t5 •cr'
E
where
a-ccr = @-
cp
When a-cr is known, a-cr/ri may be found using curves C in Section 3. 62.
general, the difference will be small if the skin does not carry a large
interaction between the lateral buckling of the panel and the torsional
stringer panel.
628
1. ' Determine the radius of gyration of the stiffener cross-
stiffener alone
Lt d
__ _ \ 4 -z U r C ZE
\ - Tangent % (L /p)
Modulus \
\\ ~w Et•
"6p
629
4. Using the column curve and the slenderness ratio determined
plus effective sheet. Ast and pst are the area and radius
8. Enter the column curve a second time with the new L'/p
10. The allowable load, Pr0 for the skin stringer combination with
P = a- (A +2 W t )+ a- (b -2 w ) t
cr cp st e s cs e a
630
z%
]JA 43
uH- oC
U)CZ c
H I fI f1 1i
fI4 II0I i
z It,
E0
'-W
IHf II11
If
1 11 1, I
If +4
0t I
zL
4+Go t
F-4
~14
m - - - - - .
631L
be used to account for the effects of curvature. If c n ?/(d/p)2 is
3t 1
important.
C. General Instability
wavelength less than the length of the cylinder but greater than the
632
frame spacing. The buckle form corresponds to the usual "diamond-
and stringers over the entire cylinder and then uses orthotropic shell
Section 3.30; however, the test results of Ref. 3-33 show that this
Ref. 3-33 which was done when there was little in the way of theory to
this method should be used for the analysis of frame stringer stiffened
method is established.
633
crc
T- CEQb
=1 4• P5 3 Pf3
Qb IR bd
where
C = 4.2
six tests reported more recently in Ref. 3-38 and was conservative for
w = 0.447 -Ic
while for the frames, the total width, d, should be used. Fig. 3. 42-3
634
necessary if the full width of skin cannot be used to compute P.. For
be seen that the preceding analysis does not depend on whether the
gations (Refs. 3-39 and 3-40) have shown that there can be an
stiffeners on the inside and the buckling load for the same cylinder with
635
rather than panel instability can be estimated from the following
formula
M (ZR)2
Ef =C
f f d
where
10-
Cf = 6.84x
636
3.43 FRAME-STIFFENED CYLINDERS
A. Material Failure
Ref. 3-43. The actual stresses must be compared against the material
frames. The sheet will buckle between frames and the frames will
637
remain circular in cross-section. The allowable buckling stress for
C. General Instability
with the sheet at the critical load. The design allowable general insta-
where
Refs. 3-45 and 3-46, which are based on Ref. 3-47. Reference 3-47
638
analyzes simply supported cylinders. The test results of Ref. 3-48
but a method of including these effects for design purposes has not been
developed.
cylinder. For the test data given in Refs. 3-48 and 3-49, Y = I is
adequate. However, the cylinders in Refs. 3-48 and 3-49 were machined
lated from
Afe2 d t3
I+ +e s
e A f 12
e s
where
Aft If = area of the frame and moment of inertia of the frame about
639
d = (d - tw) +t
Z = (1 - ±2 ) /2 L 2/Rt . If the frames are on the inside and Z > 500, the
results will be slightly conservative because the curves presented are for
servative if the frames are located on the outside because most stiffened
(given in Ref. 3-37) can be used to estimate the design allowable external
pressure
3/4
5.5 V d(•' tso/
Pcr R 3/2
LR 3
640
if
1/21 3L L
e) - -R > 4 x 10
\dt I t5. 4 R
design purposes, and ithas been found that the theory of Ref. 3-47
buckling pressure.
641
FIG. 3.43-1. SHELL PRESSURE FACTOR (ps) AS A FUNCTION OF
BULKHEAD SPACING (L/R) FOR I e 3 x 106 <10
d
10 4 ...........
.
4114 M1 MIR
= 2In p
4 p
44-P
R
I'll: x.
M-
IS
n 3.1 d
-ý::4*ý:-
1. M14.
L
VE
Pcr 30 x 10 P, + PfJ
-4 f may be obtained from Fig. 3.43-3.
3 -4 ý44
10 44,.
till mnli
1L.1 11"j,
'141
1 4
n '4 it d L!
H
41
ti f ii t I
!1:1 it It
144
9) CD
5 44
Ln ,11, 1
14
4
fill
2
10
IN,
14 RIM
M11 :1111H
tt
4f'
t
1 11ý fit
; 11
It 111. lilt 1111il
I 1ý H
j1f4
10 ills
0 1 2 3 4 s 6 7 8 9 10
L/R
642
FIG. 3.43-2. SHELL PRESSURE FACTOR (p,,) AS A FUNCTION OF
BULKHEAD SPACING (L/R) FOR xe 106 >10
10s
L/R
643
FIG. 3.43-3. FRAME PRESSURE FACTOR (ps) AS A FUNCTION OF
FRAME STIFFNESS (Ie/dR )
fl Ii
ft- 4 4.
:I l. li t ! l :
at.
10 1
" r644
FIG. 3.43-4. FRAME PRESSURE FACTOR (pf) AS A FUNCTION OF FRAME
3
STIFFNESS (Ie/dR )
. .
10 . .
VI 31
Li V -t
..... M H1 i
104
10i IozI
I r~
3R
164
FIG. 3.43-5. STRESS FUNCTION FOR DETERMINING EFFECTIVE LENGTH
1.0
0.8
X3=• (I ) (d'tw)
.Rt.
0. 6
0 2E
AT_ e = (d-tw) F, + tw
0.4
F1
0. 41 11 6
0.6
6464
:~ ~~ . i ,i . .. ~
3. 50 SANDWICH SHELLS
3.51 GENERAL
of three layers bonded together. The middle layer of the sandwich is the
core; the outer two layers are the facing sheets. Generally, the facing
sheets are very thin relative to the overall thickness of the sandwich
and the elastic modulus of the facing sheet material is much larger than
analysis.
in which the whole shell fails with the core and facings acting together;
and (3) local instability failure, in which the facing sheet fails because
buckling and wrinkling. Design formulas for these two modes of local
647
Although a considerable amount of theoretical information is
not enough test data available to obtain design curves directly. There-
fore, the design curves for homogeneous isotropic shells are used to
648
3. 52 LOCAL INSTABILITY
spaces between core walls. Dimpling of the facings may not lead to
failure unless the amplitude of the dimples becomes large and causes
the dimples to grow across the core cell walls, resulting in wrinkling
of the facings. Dimpling that does not cause total structural failure
KLE
I
-I±
1i
where
2E
t
E+E
t
649
For most materials, curves E of Section 3. 62 are a sufficiently
It should be noted that the formula for obtaining Tcr is based pri-
marily on test data from brazed flat honeycomb sandwich panels with
PH 15-7 Mo core and facings. Limited test results indicate that the
formula may be used for other types of materials and bonding methods.
of the foil thickness of the core and does not include any interaction
buckling.
0 E
'cr KL
650
where
Dimples may occur at stress levels less than %cr,but the sand-
wich will carry more load. An estimate of the stress, o' , at which
Cr..9 = E Zisl 2
used. In the inelastic range, the stress, rroI can be found by using curves
G in Section 3.62.
with very thin facing sheets, dimples in the facings can sometimes be
observed before the sandwich is loaded. These dimples are the result
flat panels, they are adquate for curved panels because the cell size
651
FIG. 3.52-1. INTRACELL BUCKLING COEFFICIENT
SANDWICH WITH
HONEYCOMB CORE
Tcr =KLE
'11 V
s core cell size (diameter of inscribed circle)
.006
652 1 i 1 ,1lj f 1
3. 52. 2 Face Sheet Wrinkling
•i.• • ir
-
• ,r
-Z
.•.._ .191
SYMME TRICAL
ASYMMETRIC
strength of the core relative to the flatwise tensile strength of the bond
between the facing core. Typical wrinkling failures are shown below.
separation from core
core crushing
653
If the bond between the facing and the core is strong, the facing
flat, the wrinkling load will also depend on the initial facing eccentricity
bles is given in Ref. 3-54. One method of determining the initial wavi-
ever, until the analysis in Ref. 3-54 is developed, for design purposes.
ily to flat panels. However, it is adequate for shells because the wave-
O'cr K 3 E E G
T1 L EE G
where
K = 0.43
654
E = Young's compressive modulus of elasticity of the facing sheet
3E
+ E
4Et
will occur. Theoretically, it has been shown that the wrinkling stress
principal stress. However, Ref. 3-52 has indicated that if the princi-
necessary:
,.43
L Vl_+S 3
655
where
656
3. 53 CYLINDERS
unequal facing sheets with orthotropic core that does not carry in-plane
°'chrh~-
- Y1 CcER
I
%/I 72~
-1 (t tz)
where
that can be used for design purposes. The factor N1 can be obtained from
P= tlt 2 h 2 /(t I+ t 2 )
657
Existing test data show that the value of Yl may be conservative for
stiffness of the core cis large (i. e. , when Vc is small). Due to man-
thin facing sheets (c/h = 1). For small values of c/h, the results of
Fig. 3. 53-1 may yield excessive error for certain values of the
function of the stress level for stresses above the proportional limit.
658
By assuming V. is independent of the stress level, the results will be
small.
compression.
659
FIG. 3.53-1. CLASSICAL BUCKLING COEFFICIENT FOR SANDWICH
CYLINDERS SUBJECTED TO AXIAL COMPRESSION
1.-2 P I
C,
VR 7j
2R +RG
RbfVý-
llif,1"1 t111L 7 ý1
cc
t~~~~ Go z 25 nirnply .pporte.d .dg-n
HIl:11111 l 7x 80 1-arpnd edges
5 tH.It l
0.4P
0..0.
ýI , 11 ll ' 1111 vc1
66
3. 53. 2 Torsion of Shear, Sandwich Cylinders
facing sheets and an orthotropic core which does not carry in-plane
Tcr
-h-: s R
h1
where
to reduce the theoretical results of Ref. 3-57 to values that can be used
where
p = ýtl tzhZ/(tl + t z )
graph 3. 33. 2 and this section would be consistent if Vs is small (i. e.,
661
a core with relatively large shear stiffness). Due to manufacturing
higher than the results of Ref. 3-58 for isotropic sandwich cylinders with
supported edges.
be used. The parameter, V5, is also a function of the stress level for
of the stress level, the results will be conservative. For most practical
References 3-62 and 3-63 can be used to obtain the buckling load
662
-4 t0:
0u)
1-4
f-v4
L)
663
CO0
ZN
ul x
'-14
PL
4
4E-
0U
6664
z
x 1
0U
z ... ....
zI ..
...
44 z......
. .....
......
-. 4 4.
UI
60
665
w
0Z_
z 11
-o
1:00
LLU)
UJA,
ýXU
0 z666
00 '
-o)
U11)
66..
7 .. .....
U X
(j)
0 mi
-4
U)0
668
3. 53. 3 Bending, Sandwich Cylinders
P = 4tt h'_/ (t I + t 2 )
For bending, rcr is the maximum compressive stress due to the bending
for some values of the parameters, but there has not been a sufficient
buckling stress for sandwich cylinders with simply supported edges sub-
isotropic but may not carry in-plane loads. The design allowable
"669
buckling pressure is
YCp (E1 tI + E2 t2)
Pcr = (1- 2 R
where
respectively
= (GL E 1 tI + I 2 E 2 t 2 ) / (EIt 1 + E 2 t 2 )
respectively
P 12 E1 t1 E2 t2 Rh
670
L
of L
For large values of L/R, it can be seen that C P becomes independent
3 YDe
where
E t E1 t2
(- 1 1
(E2t + E2t2
plete buckling of the cylinder (e. g., when buckles have formed all the
way around the cylinder). For some values of the parameters, single
buckles are not allowable for a particular design, the results of this
pressures less than p.r, the resulting stresses may fail the core,
shells are generally relatively stiff with only small initial imperfection;
likely to have single isolated buckles at pressures less than the design
671
The curves presented in Figs. 3. 53-8 through 3. 53-23 are for a
sandwich construction with very thin facing sheets ( c = 1). For small
h
values of , the results of Figs. 3 .53-8 through 3.53-23 may yield
above the proportional limit and both facing sheets are made of the
used. The parameter, Vp, is also a function of the stress level for
dent of the stress level, the results will be conservative. For most
pressure.
672
E t
p p Et2
(j.,.Z)
L
am,. I..*
2
- -(h/ZR) 0.0001
cp ot
673I
XL/
E t
FIG. 3. 53 -9. VALUES OFGC FOR V =0, AND FOR I 1
p p E ,t
ac~ 2
4........
.... t )
E (El' L;t)
+ ~~~~Lb.aa,lbk.( 2
tt
L/R
674....
FIG. 3.53-10. VALUES OF C FOR V =0, AND FORR-t =3
p p E2,
..........
- v tj
G. 3G~z pzz~
LII
. ...... 675. .....
Et
FIG. 3. 53 -11. VALUES OFGC FORYV0, AND FOR 4
p p E
jj
4~~C ,r jE ________ 4 t,. L :1
A I tA
71~¾.71 ~V
......... _ _._
- * (E4t X, t
II
LIR
677 t I~t
FIG. 3.53 -13. VALUES OF CFOR VP 0. 5 AND FO RI
. p 2.
~ ~30
I oZI ;tz lt ti R h:
LIR
678
FIG. 3. 53-14. VALUES OF C FOR V =0. 5, AND FOR 1 3.
p p E 2t
-. vi .(l +i: x
4~~
Pit-Eit 4- -2
3j Go(I
~ ...........
Em i
'-rI-tIEZz o
.2..
........'.
LIR
679
E t
FIG. 3.53-15. VALUES OF C FORYV0. 5, AND FOR -= 4
*1p p t
v C,(K ti z-tGX
I~ 101f
LIR
680,
FIG. 3.53-16.
VALUES OFC
FOR V =1, AND FOR
p It -
p
Etz
.. .Eu ......
......
)
C~~q** -j * ZEEti U~
......... I.IPa)
(Eiti * E~u)
Gs.
CI
j 1 ~ ~ p
.......
.. f. L...
FIG. 3.5 3-17. VA.LUES OF C FOR V =1, AND FOREt
p p E
V-, z 11 Z2Za
k~t.z T G)
zLil rzt A
jt p
.K h
4 ......
. L1 :
t (h/4R 2 0061,1
4 h
LIR
682
FIG. 3.53-18. VALUES OF C FOR V =1, AND FOR- 3
p p E t
V7:A (I-IncI~EatZ) Go
.1 h
LIR
683
FIG. 3.53-19. VALUES OF C FOR V =1, AND FOR t - 4
p p E
L/ R
684 zt
E 1t
FIG. 3.53 -20. VALUES OF C FOR V =1. 5, AND FOR 11
p p E 2t
v cp ([j I4
t r 2)G
ItI.
L/R
685t jzt
FIG. 3. 53 - 21. VALUES O F C FOR V =1. 5, AND FOR 112
p pEt
Ia;, -
ci( -pi~l
(Elt1
L
pj
L/R
68
FIG. 3. 53 - 22. VALUES OF C FOR V =1. 5, AND FOR = 3
p p E zt
ap
~~~I~ L. Gpj
L
(hIZR 2 oQO
c "N,AII ta
~~
sIR
68000
E t
FIG. 3.53 -23. VALUES OF C FOR V =1. 5. AND FOR E = 4
i-tc 11*~
-Pa) Itil * Eztzl e
...
....
~'C.. Gi 3Gt%
+.I 4211
L/tR
6884
3.54 CONES
Re = Rj/coso and length L. The design allowable stress, Ocr, for the equiva-
lent cylinder can be obtained from Paragraph 3. 53. 1. The design allow-
able total compressive load for the cone may be obtained from
P = ZnR
R (t, + t2) cos2&
cr e cr
Re = 11 + [ I+R 2
1 -/R
[ ] 1/ RRC/R
689
torque for the cone can be obtained from the equation
T = 2iR2 T (t +t)
cr e cr1 2
The design allowable shear stress for the cone should be based on t cr°
smaller end of the cone and not on the stress of the equivalent cylinder.
If the stresses are elastic, the design allowable moment for the cone
690
3. 54. 4 Lateral External Pressure, Sandwich Cones
should be based on the stresses at the larger end of the cone and not on
used with caution and should be limited to cone with a< 30 degrees.
691
3.60 INELASTIC BUCKLING
3.61 GENERAL
are based primarily on theoretical results which have assumed that the
if the stresses are in the inelastic range, however, the modulus of the
the material. The agreement between the predicted buckling stress and
test data has been quite good. It is considerably more difficult'to include
the effects of plasticity for shells. Methods have been developed but, in
general, they are quite complicated, and computer programs are needed
of loadings and shells are briefly discussed in the next section. Until
simple way to account for the effects of plasticity on the buckling load.
692
3. 62 PLASTICITY CORRECTION FACTOR
Wcr
We
whe re
level)
0'e wc r
T1
loading, the type of shell, the boundary conditions, and the type of
E 69
7
693
where Et, E.9 and R are the tangent modulus, secant modulus and
stress, 0-cr//,, the actual buckling stress o-cr can be read from the chart
defined in Refs. 3-7 and 3-60 will be presented because these curves have
within the accuracy of which the actual 11 is known. The curve that
694
Figs. 3.62-1 through 3. 62-Z7 present curves of (r cr i/Tversus
Curve q1
E
S
A
E E 2
[0.330 + 0.E670 ?-
c+0.7 g)- i 1
EE
Es Et
F 0.046 -- + 0.954 -- =0.33
E E
E6 9/
695
The formulas for 71 for curves A through G were obtained from
Ref. 3-60, which is based on Ref. 3-61. However, Ref. 3-61 assumes
i = 1/2. The constants outside of the radical for curve B differ from
Ref. 3-61 due to a correction that was made. Although the value of I
E . It can be seen that curve A and curve G bound the range of 11.
696
FIG. 3.62-1. 2014-T68 -T651 ALUMINUM ALLOY SHEET AND PLATE
SPECIFICATION MB0170-OOZ
PLASTICITY CORRECTION CURVES (-423 F, -300 F)
t-
260
Fcy~Ec - 24x16pt4w1.x16p
~~~ - 74k¾F
76
FO
- 6 k- 2-
4 ~ ~ .4 .. .
...
I4 I-
4 .-. +4 . .. . .. ..
C t fi . --
'-c
4--4-.
1--
1--- t .,
.. .A . .
669
FIG. 3.62-2. 2014-T6, -T651 ALUMINUM ALLOY SHEET AND PLATE
SPECIFICATION MB0170-002
- , -f'
F02 a 4 i .o .L..,2 - 2
26 T
--:- -711 A:
Li ~±431
4j4ý4U4
ISO.
war (kt) oc (k4t
6981
FIG. 3.62-3. 2014-T6, -T651 ALUMINUM ALLOY SHEET AND PLATE
SPECIFICATION MB0170-002
HOUR)
CURVES (R.T.; 200 F, 1/2
PLASTICITY CORRECTION
Minimu
Gufi~
-, f. M~inim' .f.. .•
GwarowU
.M.n.u . 4
G
, FG --
-. 200 F,1/2 Ho u ..
..
Room Tenip erctu ,e
Eo. - 10.
in6061i
x si0 -lT•rI 1 -
Fjcy in5S 1
4
10-.- 16 p . . ."••
F6 .2 60 101 F0 .2 55 61
240 -
T 1 ------ . ....
41t74--
-. 4
- 4-- -
. ;.. . . . . .. .
"20 T
-T!H .
-T "J!. . ,
.t. V t .- * . _.* . .....
t:,..o,.j....~
,
4 4
4-- ij-4
BC,
120
.: ~........
..... . . . . . ....
-A ... .... .
4
•:4------------------..._. -. ••;... ....
,4 ,.....4 .4
0i
40 50607040556
699
FIG. 3.62-4. 2014-T6, -T651 ALUMINUM ALLOY SHEET AND PLATE
SPECIFICATION MB0170-002
PLASTICITY CORRECTION CURVES (30 F, 1/2 HOUR; 400 F.. 1/2 HOUR)
A~iu
wr*WE,F,G hinimm uGuaran't.. - * E,F,G
280
24-013 0 j Fc *- 96 0 6psi .-
.i...
..
Fi.~ 2 9ill.2406
404
2W 54 552 54
700-
FIG. 3.62-5. 2014-T6, -T651 ALUMINUM ALLOY SHEET AND PLATE
SPECIFICATION MB0170-00Z
PLASTICITY CORRECTION CURVES (500 F, 1/2 HOUR; 600 F, 1/2 HOUR)
F4-5 F,t4
lg 1/2 Hour
i - *: ~st. "600
-" Fy, F, 1/'2 How .-. . _- .. -1.-
._,
. . ....
""120 06 u8
9x...... Ec
"F- I ...... F'T 7.4.x 106Pui
-O.2l!ks -cnO 0 1 i ;-
- F6
- .. -6- . .- 1
. ... . - t 4
t 4~.+.
1000
ib1 4 . 4~. ~ .. 4
470
90% Probabilit Lou itudinal Grain In
Ll..
Ka I3
LA
3K3
U t
94
'00
N*p.
IsIf
702
90% Probability Lon itudinal Graini
1. .
H NIT
I-It
UQ
LA703
90% Probability
I -T
17%
o _
Zcc
o0 0 0 0 0
o i 0 In 00
70
90% Probability
9
'-4M
0-1K
00
HJ
w 0
0, C
V)
0 0k
0 00A
0705
90% Probability
Hcu
U.
1
0
HH HH
" !111 1 '
ZN
706
FIG. 3.62-11. PH 15-7 Mo STAINLESS STEEL SHEET AND PLATE -
RH'1050, RH 1075 - SPECIFICATIONS LBO16O-100, LBO16O-129.
LB016O-130, HBO160-010. LAO111-022,.MAO1O7-023
PLASTICITY CORRECTION CURVES (-200 F)
4--4-4f
62000
Ft215 ks 2=
~~~tI .......... 4
-~ +4+
.. .. .. ::
707
FIG. 3.62-12. PH 15-7 Mo STAINLESS STEEL SHEET AND PLATE
RH 1050. RH 1075 - SPECIFICATIONS LB0160-100, LBO16O-129,
LBO16O-130, HBO16O-O10, LAOLL1-022, MA0107-023
PLASTICITY CORRECTION CURVES (-100 F)
. ..........., . .~I * 1*
t,.
500*- 2~. 7
'B, C, Di
300,
708
FIG. 3.62-13. PH 15-7 Mo STAINLESS STEEL SHEET AND PLATE
RH 1050, RH 1075 - SPECIFICATIONS LB0160-100, LBO16O-129,
LBO160-130, HBO16O-O1O, LAOJ11-022, MA0107-023
4 Minimum Guarantood
- Room Tempemture
6 psi
E, = 30. 0 X10
F 185 ksi . ..
i'0!.2 =188 ksit
500 4-
-~I
. ~ 4L-. t
-.--
I- 4
104 120
709+
FIG. 3.62-14. PH 15-7 Mo STAINLESS STEEL SHEET AND PLATE-
RH 1050, RH 1075 - SPECIFICATIONS LB0160-100, LB0160-129o
LBO160-130, HBO16O-010, LAO1LL-022, MAOLO7-0Z3
Itt
-I., ~ C 1 4.
D
300 -
100 4
710
FIG. 3.62-15. PH 15-7 Mo STAINLESS STEEL SHEET AND PLATE -
RH 1050, RH 1075 - SPECIFICATIONS LB0160-100, LB0160-129,
LB0160-130, HBO160-010, LA0111-022. MA0107-023
PLASTICITY CORRECTION CURVES (300 F)
Minimum Guaranteed
E, F, G
600
300 F
Ec = 28.8 X 106 psi
F = 168 ksi
F0 2 = 172 ksi
500
400 El
300 C, D
200
100
0-
100 120 140 160 1s0
rcr (ksi)
711
FIG. 3.62-16. PH 15-7 Mo STAINLESS STEEL SHEET AND PLATE -
RH 1050, RH 1075 - SPECIFICATIONS LB0160-100, LB0160-129,
LB0160-130, HBO160-010, LA0111,022, MA0107-023
PLASTICITY CORRECTION CURVES (400 F)
J I I
. .. .. . . . .. ..i . : 2 :2
E,F,G', -' 4 4 ,~
. • •..... • ... -. -- .- . ..
.. t 4-. . .. . .... ,
I00
712
FIG. 3.62-17. PH 15-7 Mo STAINLESS STEEL SHEET AND PLATE
RH 1050, RH 1075 - SPECIFICATIONS LBO16O-100, LBO16O-129,
LB0160-130, HBO16O-010, LAO1L1-022, MA0107-023
PLASTICITY CORRECTION CURVES (500 F)
'.F -~.154' 4s
4w~ I l.....4.'.
3W 1
4,,S
.--. . .. . .. a .4 f
I 2it
14-
oWcr (ksl)
713'
FIG. 3.62-18. PH 15-7 Mo STAINLESS STEEL SHEET AND PLATE
RH 1050, RH 1075 - SPECIFICATIONS LBO16O-100, LBO16O-129,
LBO160-130, HBO16O-010, LAOL11-022, MA0107-023
PLASTICITY CORRECTION CURVES (600 F)
Minimumn Garanteed ~ T
600..............
200 f~
AA
acr (kit)
714
FIG. 3.62-19. PH 15-7 Mo STAINLESS STEEL SHEET AND PLATE
RH 1050, RH 1075 - SPECIFICATIONS LBQ16O-100, LBO1,6O-129 1
LBO16O-130, HBO16O-O1O, LAO111-022, MA0107-023
PLASTICITY CORRECTION CURVES (700 F)
Minimum Guanto
EFG .4-i I - .
.4,.44
400 --------
17,45kf I 2
4 4,4
-I 1 I
10 441... ......
wcr (ksO)
715
FIG. 3.62-20. PH 15-7 Mo STAINLESS STEEL SHEET AND* PLATE'-
RH 1050. RH 1075 - SPECIFICATIONS LBO160-100, LBO16O-129,
LBO16O-130, HBO16O-010, LAO111-022, MA0107-023
PLASTICITY CORRECTION CURVES (800 F, 900 F)
4-44 +.*
*-4
.1L 7 7
716
FIG. 3.62-21. PLASTICITY CORRECTION TITANIUM ALLOY SHEET <. 25
6AL-4V ANNEALED LB0170-113
MINIMUM GUARANTEED
ROOM TEMP
=15.8 X I psi
4001 11 1F 0 -i120ksi
300
0100~~~ 2
70-1015
1717
FIG. 3.62-22. PLASTICITY CORRECTION TITANIUM ALLOY SHEET.<. 2S
6AL-4V ANNEALED LB0170-113
M00INIMUM GUARANTEED
600
-
-i{ t-
i 'lll
500510
300OF
7180
MINIMUM GUARANTEED
>4
0-~z
7191
FIG. 3.62-24. PLASTICITY CORRECTION TITANIUM ALLOY SHEET<. 25
6AL-4V ANNEALED LB0170-113
MINIMUM GUARANTEED
400 - - -- I- -
300
F 71rkks
2000
FIG. 3.62-25. PLASTICITY CORRECTION TITANIUM ALLOY SHEET<.ZS5
6AL-4V CONDITION S. T. A.
MINIMUM GUARANTEED
IfI
If,
'72
FIG. 3.62-26. PLASTICITY CORRECTION TITANIUM ALLOY SHEET <. 25
6AL- 4V CONDITION S. T. A.
MINIMUM GUARANTEED
' i 1
T 1 '
:-71 --
,70 i
I- x '4,
IL
'l- :-P I
I I H.,I
I ' _ ~
off•- -7-T", ;) -
T. .. • •I•
LA.i iii i ,
& 72,
li i : :: .-
FIG. 3.62-27. PLASTICITY CORRECTION TITANIUM ALLOY SHEET<. .2S
6AL-4V CONDITION S. T. A. LB0170-113
MINIMUM GUARANTEED
400
00O
68-7-72-4-7 78 - 80F
ksis
3723
3.63 COMBINED LOADINGS
work has been done in this field due to the complexity of the problem
range are discussed in Section 2.80. The method that can be used for
The stress intensity a-. 1 and strain intensity.e.1 are obtained from
the formulas
e. IE (+ 40 4
where
Y+ 9 = shear strain
For a ductile material, the a-.1 versus e.1 curve for a biaxial stress
724
Therefore, if a. is greater than the uniaxial p'roportional limits of the
It can be seen from the formula for o-. that each of the individual
stresses may be less than the proportional limits of the material, but
to modify the elastic buckling load. This method is useful if the plas-
the same.
therefore,
a -
%723
725
The elastic critical circumferential buckling stress a-cr/i may be
the refore
a".
a" '[N3 'c r
are inelastic, ci may be obtained from Section 3.62 using a1/9 and
Pcr = tcrt/R.
726
REFERENCES
DZ-23835-1, (1965).
(1966).
TN 3783 (1957).
727
3-8 Batdorf, Stein, and Schildcrout. Critical Combinations of
STL/TR-60-0000-19425 (1960).
728
3-15 Kaplan, A. and Y. C. Fung. A Nonlinear Theory of the Bending
TN 3212 (1954).
SESA Paper No. 547, Vol. 18, No. 1 (1961), pp. 88-96.
SM-38938, (1962).
3-18 Tsien, Hsue-Shen, "A Theory for the Buckling of Thin Shells,"
(1963).
No. 12 (1963).
(July 1964).
729
3-22 Modeer, J. Buckling Coefficients for Hemispheres and Oblate
3-26 Dow, N., C. Libove, and R. E. Hubka. Formulas for the Elastic
730
3-30 Almroth, B. "Postbuckling Behavior of Orthotropic Cylinders
(May 1962).
731
3-38 Card, M. F. Bending Tests of Large Diameter Stiffened
732
3-46 Ball, W. Formulas and Curves for Determining the Elastic
(June 1959).
Report 1587.
733
3-52 Konishi, D. et al, Honeycomb Sandwich Structures Manual. NAA
(March 1960).
(Jan. 1958).
734
3-59 Kuenzi, E., B. Bohannan, and G. Stevens. Buckling
(1962).
735
4.00 MINIMUM WEIGHT SHELL DESIGN
4. 10 INTRODUCTION
Whereas use of these chapters will achieve designs which will preclude
also,
this statement have been presented (Refs. 4-1 and 4-2). An additional
assumption is that the margins of safety are zero for all buckling modes.
736
Some of the existing minimum-weight analyses are based on
weight analyses should be used with caution unless the proportions are
examined to ensure that they are realistic. The weight of the final
components.
737
4.20 STIFFENED SHELLS
4.21 GENERAL
dimension and found the ideal value of the optimum stress as well as
bending. Shanley (Refs. 4-6 and 4-7) also considered this problem by
design index for aircraft fuselages, the diameter was far beyond the
Joyce and Mitchell (Refs. 4-8 and 4-9) determined stiffener and
the frame weight, Shanley (Refs. 4-6 and 4-7) also investigated
738
extended this analysis by a numerical procedure which included the
Wenk (Refs. 4-12 and 4-13) has established the pressure (p) as the design
index and has presented a series of minimum weight design charts for
orthotropic shell theory and obtained results for the optimum configu-
739
longitudinal ring-stiffened cylinders under compression based on ortho-
The advent of thin aircraft wing and tail structures in the late
of parameters and also by Kolom (Ref. 4-19). Hubka et al., (Ref. 4-20)
of multicell boxes with steel sandwich covers and channel and corrugated
740
flat, integrally stiffened and sandwich cover plates employing plate and
grid stiffening systems for plates used as compression covers for multi-
configurations for sandwich plates. Crawford and Burns (Refs. 4-15 and
They also treated multicell boxes with flat and corrugated core sand-
741
and longitudinally stiffened cylinders of the same length under bending.
Crawford and Burns (Refs. 4-15 and 4-16) compared unstiffened and
theory, Gerard and Papirno (Ref. 4-17) compared ring and longitudinal-
742
4.22 STIFFENED CYLINDRICAL SHELLS IN PURE BENDING
failure: (1) buckling of the panel between frames, and (2) general
stiffened shell will buckle between the frames. Buckling of the skin-
If, however, the calculated buckling stress of the sheet is higher than
that of the stringer plus its effective width sheet, the entire width is
assumed to act at the column buckling stress (Ref. 4-29). If the frames
are not sufficiently stiff to resist buckling, the stiffened cylindrical shell
The frame stiffness for which panel instability and general instability
743
are equally critical may be estimated from the following formula
where
d = frame spacing
If the frame stiffness is greater than (El)f, the cylinder will fail in
panel instability, and i! the frame stiffness is less than (El)f the
justification. For frames with flixural stiffness greater than (El)f the
744
4.30 SANDWICH SHELLS
4.31 GENERAL
that transmits shear between the facing sheets. The facings are designed
transfer shear between the facings and allow the facings to bend about a
common neutral axis. For most cases of interest for space vehicles,
the facing sheets are so thin that their flexural rigidites about their
section.
most common cores are the open cell, honeycomb configurations, the
corrugated core, and the closed-cell foam cores. The honeycomb cores
745
desirable, yet in reality its weight is a substantial part of the total
weight analysis: the facings, the core, and the bonding medium.
various face materials and core densities, and concluded that there is a
definite optimum core rigidity for each face material and design index
(N/R).
746
The nondimensional feature accounts for material properties at various
temperatures.
unless the sandwich proportions are examined to ensure that they are
realistic.
747
4.32 BUCKLING OF SANDWICH CYLINDERS UNDER AXIAL
COMPRESSION
C C
=crcr~ ZYE-t2 (4.32-1)
IR 2 (tlI + t2)
CcY
= Li--- (4.32-2)
whe re
Et
Vc= 2 R Gxz (approximately)
748
Minimizing the sandwich weight with respect to the facing thick-
ness (Ref. 4-33) and excluding local failures (wrinkling, dimpling, etc.)
leads to the equation for the weight of the core to the sandwich weight
where
W c = weight of core
Wb weight of bond
compression.
749
REFERENCES
750
4-8. Joyce, N. B. and L. H. Mitchell. Panel Shape for Least Weight
pp. 272S-288S.
pp. 373-393.
751
4-15. Crawford, R. F. and A.B. Burns. Strength, Efficiency and
(Feb 1962).
for Stiffened Plates and Shells, " AIAA Journal, Vol. 1, No. 4
752
4-22. Semonian, J. W., and R. F. Crawford. "Some Methods for the
(Oct.' 1957).
"753
4-30. Leggett, D. M. A., and H. G. Hopkins. Sandwich Panels and
754
5.00 OPTIMUM USE OF COMPUTER PROGRAMS
5. 10 INTRODUCTION
Chapter 1.00, procedures for the static analysis of shells were given in
Chapter 2. 00, and procedures for the stability analysis of shells were
lyst may encounter a problem not considered in any of the previous dis-
unsymmetric load is best accomplished using a digital computer (e. g., see
755
AXIS OF REVOLUTION and division functions in minute
756
5.20 FUNDAMENTALS OF COMPUTER UTILIZATION
computers.
Some small computers are not much larger than an engineer's desk.
whereas the latest large computers possess over a dozen units and occupy
an entire room. All computers used today are superior to desk calcula-
tors in three respects: they operate faster, possess memories, and make
standing and using the basic rules governing the computer being used.
The engineer is aware that he can perform operations and obtain accuracy
with some slide rules that cannot be obtained using other (usually smaller)
slide rules. In the case of computers, not only do some possess greater
many significant figures) but some are also capable of working several
757
To harness the capabilities of the computer and effectively utilize it
in use: source (or user's) language and object (or machine) language. In
the case of object language, the program is written in a manner so that the
machine can act on it; a source program is very similar to typical analyti-
machine uses part of its memory to translate from the source language to
are more familiar to engineers, they are recommended over object lan-
guages. In either case, the engineer should be aware that all computers
necessary in nearly every case. Briefly, these statements are (1) control
758
statements, which tell the machine to add A to B, for example; and (3) input -
output statements, which tell the machine to read in data, plot curves,
Often, when dealing with the larger computer systems, the engineer
rarely, if ever, sees the computer itself. Instead, he deals with the input -
output receiving station that may be located several miles from the com-
puter he intends to use. Also, for reasons of efficiency, the input provided
by the engineer may be given to a subordinate machine (e. g., the IBM 1401)
for reduction to a form which the main computer can more readily accept.
machines) and cards (IBM and CDC). If it is not desired to produce the
tape or cards themselves, coding sheets can be prepared from which the
collecting similar quantities into arrays and dealing with entire arrays
rather than with the individual elements of the arrays. This capability
following section.
759
5.30 INTRODUCTION TO MATRIX ALGEBRA
a XI +a x .. ...... + ann x ny
n
I a 2ixi Y2
j=1
n
Yn (.02
I a ni Xi
ji=
760
Consider the following representation of these equations
a11 a12
al a1 .... ain
al 1
xll Y
a21 a22 . a2n x2 Y2
ann Xn n (5.30-3)
anl an2 ......
The kth equation of Eq. 5. 30-1 is obtained by multiplying the kth row of the
x column and setting it equal to Yk" For example, the second equation is
obtained by multiplying the second row of the aij rectangular array by the
x column, term by term, and adding, then setting the resulting equation
equal to V2:
Eq. 5. 30-3 is a matrix equation that represents the set of linear algebraic
761
members of a matrix are called the matrix coefficients. Eq. 5. 30-3 can be
Some matrixes are given special names. For example, the n x 1 matrix
(i. e., the matrix consisting of n rows and 1 column) is called a column
vector. The 1 x n matrix which consists of 1 row and n columns (e. g.,
n columns is called a square matrix. The matrix coefficient in the ith row
and jth column of a matrix A is denoted by aij. A square matrix in which all
its diagonal terms, aii, are equal to 1, while all other coefficients are
all its coefficients equal to zero is called a null matrix. Identity and null
762
denoted by [A] T. For example, consider the 2 x 3 matrix A = 7
1st column of [A] T is the same as the first row of matrix A, and its 2nd
matrixes can only be equal if they are of the same order (i. e., if they both
n
CJ aiR bRj
R=I
763
[A] [B] is not equal to [B] [A] . In fact, if matrix A is a mx n matrix
has m rows, the multiplication is not even defined (if p # m). For example,
A - 2 and B ]
then
[A) [B) -- 2 46 C
n
where the coefficients of C are given by cij I aiR bRj (i. e..
R=l
cll = allbll + al2 b2 l (1) (1) + (3) (0) = 1, c12 = allb1 2 + al2 b2
(1) (1)+ (3) (1) = 4 etc.). In the preceding example, the premultiplication
and results in the matrix A (i.e. , [A] [I] = [I] [A] = A). Suppose a
matrix B, and the result in either case is the identity matrix 1. Then
764
matrix B is called the inverse matrix of A, and is denoted by [A] 1i For
[B] [A] =
=
[
-1/5 2/5] l1
-1/5
1- =[ 1
[A] [ B]
1/5 2/51=i 10
then B (- [A] -1)is the inverse of A, and A (- [B] -1) is the inverse of B.
The result is
2x + x2 10
x1 +3x= 5 (5.30-6)
[2 :] H [I=3
765
Applying Eq. 5.30-5
we have
x= 5 and x 2 = 0 (5.30-7)
The validity of the results of Eq. 5. 30-7 can easily be verified by con-
matrixes (i. e., matrixes having nonzero elements only within a band
766
efficiency over the inversion of the entire A matrix makes problems
X X
X X X
x x • 0
XX
[AJ= Kx
X X
X X
X X X .
X +X
767
5.40 TECHNIQUES FOR SOLVING SHELL PROBLEMS AND THE
USE OF THE COMPUTER IN THESE TECHNIQUES
presented for the static analysis of shells. The computer could be used as
problems are in use today that utilize the computer in this central capacity.
These analyses all share the property that the continuous shell is replaced
by a shell with a finite number of points on it, and the computer is then
768
5.41 THE FINITE DIFFERENCE METHOD
ential equations are replaced by a set of linear algebraic equations that have
a readily obtainable solution. The basis for this method is the replacement
769
and the backward difference
-A k k-i (5.40-2)
is used since it involves values of w on both sides of xk, and it results that
the order of h 2 , whereas, the errors in Eqs. 5.40-1 and 5.40-2 are of the
manner. For instance, the central finite difference expression for d 2 w/dx 2
at x = xk is given by
Finite difference expressions for all orders of derivatives are easily derived
770
conditions, the set of finite difference equations form a set of n linear
matrix form and the solution easily follows. It is seen that the exact con-
the method and will be discussed further. An example of the finite differ-
Fig. 5.40-2.
771
Y 23 16 7 15
22
7 77
17 8 2 6 14
2a- 9 3 1 5
x 0
10 4 12 13
24 19 11 20 21
The square plate is partitioned into 16 segments as shown. That is, the
given as follows:
2 A f
21x
2x = 7-2"f1
2oY 1(f 2f 0f+f f3
Bx X0 9 YO h h
2 2
a Xo'O Af
hf f = I 1 2fo + f 4 ) (5.40-6)
by~ ~~ 22 'Y
a2f + -0if
Then at Point 0 the partial differential equation- 8x + -2 is
ey
772
replaced by the following algebraic equation
where [A] is the 25 x 25 coefficient matrix, and [f) is the solution column
vector. If the matrix [A] has an inverse, the (ff may be obtained by
It can be seen from Fig. 5.40-2 that the method of finite differences
replaces the continuous plate with a gridwork of discrete points. When this
method is applied to shells, grid systems of the type shown in Fig. 5.40-3
occur.
773
FIG. 5.40-3. Finite Difference for Shells
These three vectors will have as many elements as there are grid points
on the shell.
774
For the general case of unsymmetrical shells loaded unsymmetrically,
times on the computer and loss in accuracy, will result. For the case of
775
5.42 THE NUMERICAL INTEGRATION METHOD
equations.
equation
2f +2
8f+ --. = -2 (5.42-1)
ex2 8y2
valid in the interior of the square shown in Fig. 5.42-1. On the boundary,
f = 0.
finite differences. Along any y = constant line, say y = Yk. the following
holds true:
y2 y=k h2 k
Iy 2 If
[~+ (x) - Zfk (x) + fk- (x)] (5.42-2)
776
Substituting Eq. 5.42-2 into
obtained:
a
h dd 2 fk 1
Yk+1
Yk dx 2 + 2 x
a -Z
f2 1 -2 10 .. . 0 f2 -2
• 0 1 -2 10 0 -2
* ~ * * 1 -2 1
0 0 00 . 0 1 -Z -2
777
with boundary conditions fk = o at x = o, a (where fk is equal to f along
reduce the order of all pertinent equations to the first order. This can be
•(5.42-6)
fk=
778
In matrix form, Eq. 5.42-6 and 5.42-7 can be written
g1 9l -2
II
g2 I g2 -2
*I
0 1 gn- -2
k0 ;4 [A]--
[
gn- -
I f1 0
d f
d fx + 1 f (5.42-8)
f2 f2
I 0
* I
fk -Iii 0 "f
k fn- 0
*I
. I
f I
n-i
Eq. 5.42-8 can be written more compactly if the following symbols are
used:
'gl'
IGI
JGJ ; F H (5.42-9)
7gn-7
779
r [0] 12z[A]'
1]
C
h
Note that Eq. 5.42-10 represents a set of (Zn -2) first-order ordinary
HI H0 + h f (x, H)
where
CURVE h = x1 -x 0
APPROXIMATIONx •
HI , H0 = H matrix at
H0 H1 = H matrix at
x
x1
and
f (X, H) = K- JH
= slope at x = x0
X0
XI
780
It can be seen that this procedure merely assumes that the first
H2 = H1 + h f I(x, H)
able. One of the most popular is the Runge-Kutta method. Because of its
grams available. As in the Euler method, the interval size may be changed
initial value problems. Thus, the Euler procedure yields the results.
function, f, and its first derivative at x = o for all yk" However, the
was a boundary value problem (as are all shell problems). To use
781
start the process. When the boundary, x = a, is reached, the values,
f(a, y), are compared to the given boundary conditions. The assumed
slopes are then adjusted and the numerical integration performed again.
After a series of trials, the boundary values from the numerical integra-
tion should compare favorably with the given boundary conditions, and
hence the initial assumed slopes are acceptable and the interior results
are valid.
cated not only by the necessity of having to convert a boundary value prob-
lem to an initial value problem, but more importantly, by the fact that
782
5.43 THE BUDIANSKY - RADKOWSKI METHOD
where
where t is the independent variable and the Fourier coefficients for the
783
Using Hooke's law and the conventional strain-displacement relations, the
posed of the Fourier coefficients for the three displacements and the
A zI + B zz0 = o
B nBnz n+ C n Zn-i-=
Zn+n 9n
784
the Gaussian elimination procedure. In this case, it is necessary to invert
"zi+1 , zi+1 in terms of zi+2 , etc., until the boundary is reached. Then
"zN can be explicitly obtained because the boundary conditions supply the
necessary additional information. Having zN, all other z's are system-
down the principal diagonal. Once the z matrix is known, all other quan-
addition, the method has been extended slightly to treat shells in which
Because of the efficient running times resulting from the use of the
which it is applicable.
785
5.44 THE FINITE ELEMENT METHOD
given shell. For example, to find the stresses and deflections existing in
(b) Finite Element Approximation to Spherical Shell the behavior of a typical element
is studied and the elements are
FIG. 5. 44-1 Spherical Shell and then tied together to produce the
Finite Element Approximation required shell.
Two methods are commonly used when dealing with finite elements.
The se methods are compatibility (or force) and equilibrium (or displacement).
786
In the compatibility method, the requirement of geometric compat-
ibility is used to tie the finite elements together. That is, the redundant
forces and moments which exist at the interfaces of the several elements
existing (both applied external forces and redundant forces) by the following
equations:
U = a F + a F + " + a F + . ".+ a F
1 21 1 12 2 1k k 1, k+n k4-n
u 2 =a 2 1 F 1 + a 2 2 Fz +. " ". + ak Fk +." •+ a2, k+n F kn
'* 0 0
. (5.44-.1)
uk aakl F1 + akZ F2 + + akk Fk + + ak, k+n Fk+n
787
In matrix form these equations are
whe re
{ u*_
uo
ra**,
[aO* a
a*O
00
F0
and
788
3• - C• • ._
Solving the first of these equations for the redundant forces [FO] yields
Thus, the redundant forces (F°l and the displacements at the k coordinates
(uý3 are obtained using Eqs. 5.44-2 and 5.44-3. The only matrix inver-
Once the redundant forces are known the forces on each element are
the shell flexibility matrix A. It can easily be shown that this matrix may
789
where
and [a]i is the flexibility matrix of the ith element which is known. • is
by the relationship
(6] = [)()
zeroes except for a single term of unity, the position of which identifies
other words, the function of the matrix P is to select the appropriate dis-
element chosen. For example, for the conical shell frusta elements
mentioned above, the flexibility matrix is easily obtained. From what has
790
manipulations and inversions required can be handled quite efficiently on
the computer.
unknown displacements and rotations are dealt with. The unknown dis-
[[K]2
and since the force -displacement behavior of the elements is known. the
791
ful = [K]-.1 (F)
Once the shell displacements are known, the internal forces in the ele-
The P matrix in Eq. 5.44-4 is the same as was used in the compati-
bility method. The element stiffness matrixes can be obtained from the
strain energy in an element, as shown in Ref. 5-2. Refs. 5-3 and 5-4
use the conical shell element discussed in conjunction with the equilibrium
similar to the one shown in Fig. 5.44-1. The results agree wel with a
ally not strictly banded; hence, it is generally not possible to use Gaussian
792
5. 50 ASSOCIATED PROBLEMS AND FEATURES
RELATED TO COMPUTER USAGE
793
5.51 ACCURACY
puter can be exceeded by using a number too large or too small for the
that the number system used is discrete; that is, finite gaps exist between
794
5. 52 TIME
aspect of the work, which involves filling out coding sheets, punching cards,
etc., is called programming time. After the given problem has been
required for this phase may vary from several days to a month or more.
After the program has been completely checked out, the compiled deck
of the program, the sophistication of the computer facility, and the num-
795
5. 53 COMPATIBILITY PROBLEMS
Cylinder Analysis
796
are required but the standard library routines from tape A.
797
5. 54 THE DYNAMIC NATURE OF THE COMPUTER
entirely possible that some of the statements made in this chapter will
destined to replace FORTRAN II. The time will come when it will no
798
L
5.60 A TYPICAL COMPUTER PROGRAM
and its associated input and output. A schematic illustration of a deck that
The main program contains the heart of the analysis. The subroutines may
SUBROUTINES
CONTROL CARDS
II III
I I III
I I
III
799
are available for obtaining the results graphically. A typical plot from
:.i ~ ~
.0.
.
*~ I 4
.m .........
....
~R"i1WECE SSW I=3?
zc•sThICKWm
RZFW3CE s
ErIDIAS
N.O...0 POINTS ALONG
..
-. 0
-- . .... .. . . . _. . .l... ..............
..... I CE
800
o a
ama
0 f" 0
0 0
04 0
o
th~
LU
a
0 -' 0
1 000000000000000000
zj mi2
0 0 0 ooooooooooooooooooooooaacoccoo
in in in
WA AIA
4A in AA
nALn in A IA I% tA W%
9%inIA in in Anin In 6nInA
Il
U) IL 49
z a z r
000000000000000oco00 0000000000000
0;- oo0o000ooo00o0o0ooo00000000000
'00 inwN0i4nnoL n on on m mn n onin w
n An n M
*N 0 4a.. 40 4V
an
0 0 00o0o000000o0o00000o000000a00
LU Lu
j~- -~
0 0
j 14
4 0c9 0
-ý (7 cy OlcýO ,V -a -V ý0 .C.a ýV ý0 PO , a
o 0. Ck- ýV y ,Q N0 ýo .0 ýa ýa ,o ) 1 ,0 ,4
0. LU 0ý V ýO ýa ,0 4a ,Q lcý NV %V %0 ha o %0 l0
801
FIG. 5.60-4. TYPICAL COMPUTER PLOT
3200C) X
26000
isr
rU(w
44-
IWAX)
4+
4-f 4--
12C
f
'4+ T.
I jr -4
-4-i 4 .4
-4-
1.44.
OU00 Ht
t I
Eý1, it J
4000 1 1 1 1 1 11 1- -1.
a 20 40 Gu" OU IDU 120 140 160 ISO ZU0 22U 240 no too
802
5.70 CONCLUSION
From the brief description of the computer and its use in shell
analysis given in this chapter, it is obvious that the computer has a strong
803
REFERENCES
pp. 2138-2145.