Академический Документы
Профессиональный Документы
Культура Документы
.I
_'o
THEORY OF ANISOT_IOPIC SHELLS
By S. A. Ambartsumyan
..
I
b
Translation of "Teoriya anizotropnykh obolochek"
State Publishing House for Physical and Mathematical Literature
;_ Moscow, 1961
i,
.!
t_
"1
,_ NATIONAL AERONAUTICS AND SPACE ADMINISTRATION
:4
.._ For sale by the Office of Technical Services, Department of Commerce,
Washington, D.C. 20230 -- Price $6.03
J;
1964012887-002
I
Section 10. Boundary Conditions .......................... 38
Section 11. Additional Remarks Concerning the Conditions of Contact
of Adjacent Layers and the Conditions at the Olter
Surfaces of a Shell ........................... 40
Section 12. Special Cases of Anisotropy of the Material of the
Shell Layers ................................ 43
Section 13. Shells Consisting of an Odd Number of Layers Sym-
metrically Arranged Reiatiw to the Coordinate Surface . . . 46
Section 14. Single-Layer Anisotropic Shells .................. 50
Section 15. Further Remarks Concerning Elasticity Relationships .... 55
Section 16. Calculatiou of Stiffnesses for Arbitrary Directions ...... 58
CH2, PTER IH
MEMBRANE THEORY t)F ANISOTROPIC SHELLS
t Section 1. General t'remises and Initial Relationships in the
Membraue Theory of Single-Layer lsotropic Shells ...... 61"
Section 2. Boualdary Conditions ........................... 64
Section 3. Area of Applicability of the Membrane Theory ......... 65
,, Section 4. Fundamental Equations of the Membrane Theory of
Symmetrically loaded Shells of Revolution ............ 66
Section 5. Examples of Calculation of Symmetrically Loaded Shells
of Revolution ............................... _3
Section 6. Evaluation of Results Obtained in the Preceding Section , . . 86
Section 7. Continuation of Section 5 ....................... 87
Section 8. An Arbitrarily Loaded Cylindrical Shell of Arbitrary
Shape ................................... 95
Section 9. Some Remarks Concerning the Membrane Theory of
Anisotropic Laminar Shells ...................... 104
CHAPTER IV
SYMMETRICALLY LOADED ANISOTROPIC SHELLS OF REVOLUTION
(
Section 1. Basic Premises. Initial Relationships and Equations ..... 109 /
Section 2. Equation_ of Solution and Design Formulas ........... 113
Section 3. Shells of Revolution Consisting of an Odd Number of
Layers Symmetrically Arranged Relative to the Median
Surface of the Shell ........................... 117
Section 4. Single-Layer Shells of Revolution ................. 120
Section 5. Reduction of the System of Equations in (3.16) and
,, (3.17) to a Single Equation. A Particular Solution of the
Inhomogeneous Equation ........................ 121
Section 6. Asymptotic Integration of the Equation of Solution (5.9) . . . 124
t
1964012887-004
Sect,'on 7. Internal Forces, Moments, Stresses and
Displacements ............................. 132
Section 8. Edge Effect in Anisotropic Shel !_ . ................ 136
Section 9. Long Shells of Revolution ...................... 139
Section 10. Examples of Calculation of Long Shells of Revolution .... 143
Section 11. Solution of a Few Problems of Shells of Revolution
of Zero Gaussian Curvature Consisting of an Arbitrary
Number of Layers ........................... 166
Section 12. Anisotropic Cylindrical Shells of Revolution Reinforced
by Lateral Ribs ............................. 177
CHAPTER V
ANISOTROPIC CYLINDRICAL SIIELLS
Section 1. Basic Premises. InitialRelationshipsand Equations.... 195 J
Section 2. System of DifferentialEquations ofSolc, tionin Displace-
ments ................................... 198
Section 3. CylindricalShellsConsistingofan Arbitrary Number
ofOrthotropicLayers ........................ 201
Section 4. Fngineering Theory of CylindricalShellsConsisting
of an Arbitrary Number of AnisotropicLayers ........ 208
Section 5. Continuationof Section4 ....................... 214
Section 6. CylindricalShellsConsistingofan Odd Number of
Homogeneous Anisotropic Layers Symmetrically
Arranged Relative to the Median Surface ............ 224
Section 7. Engineering Theory of Cylindrical Shells Consisting
of an Odd Number of Homogeneous Aniso_ropic Layers
Symmetrica!ly Arranged Relative to the Median Surface.. 228
Section 8. Continuation of Section 7 ..................... 232
Section 9. Integration of the Eq,za_lons of Engineering Theory
of a Cylindrical Shell by the Method of Double
Trigornetr'; c. Series .......................... 240
Section 10. Intega'ation of Equations in the Engineering Theory of
Cylir, drical Shells by the Method of Single Trigono-
metric Series .............................. 265
CHAPTER VI
SHALLOW ANISOTROPIC SHELLS
/
Section 1. Basic Premises. Initial Relationships and Equations .... 277 j
Section 2. Equations of Solution and Design Formulas .......... 280
Section 3. Shells Consisting of an Odd Number of Homogeneous
Anisotropic Layers Symmetrically Arranged Relative
to the Median Surface ........................ 285
ill
1964012887-005
Section 4. Extremely Shallow Shells. Basic Pr-uises, initial
Relationships and Equations ..................... 288
Section 5. Equations of Solution and Design Formulas in the
Theory of Extremely Shallow Shells Consisting of an
Arbitrary Number of Anisotropic Layers ............. 292
Section 6. Equations of Solution and Design Formulg..q in the
Theory of Extremely Shallow Shells Consisting of an Odd
Number of Homogeneous Anisotropic Layers Symmetrically
Arranged Relative to the Median Surface ............. 300
Section 7. Integration of the Equations of Solution in the Theory of
Extremely Shallow Orthotropic Shells ............... 306
CHAPTER VII
NEW THEORIES OF ANISOTROPIC SHELLS AND PLATES
/
Section I. Basic Premises and Hypotheses .................. 315
Section 2. Theory ofOrthotropicPlates..................... 316
Section 3. Theory ofBending ofa Plate Possessing Cylindrical
Anisotropy ................................ 328
Section 4. Approximate Theory of an AnisotropicPlate Considering
Transverse Shear ............................ 332
Section 5. Another Approximate Theory of Anisot_opic Plates ...... 336
Section 6. Examples of Plate Calculations ................... 342
Section 7. Theory of Extremely Shallow O,'_hotropic Shells ........ 358
Section 8. Approximate Theory of an Extreme]y Shallow
Shell Considering Transverse Shear ............... 365
Section 9. Another Approximate Theory for an Extremely Shallow
Shell .................................... 368
Section 10. Example of Calculation of a Shell ................. 372
Section 11. Theory of Extremely Shallow Laminar Orthotropic SheLls.. 376
Section 12. Examples of Calculation of Laminar Shells and Plates .... 386
iv
1964012887-006
FOREWORD
Shells are widely used as structurel elements in modem construction
engineering, aircraft construction, ship building, rocket cons._ruction, etc.
A careful study of the shells used in engineering leads to the conclusion
that they are most often anisotropic (naturally or structurally) and in many cases
are anisotropic and laminar.
The last few years have seen the appearance of a number of thorough
monographs: V.Z. Vlasov, Obshchaya teoriya obolochek (General Theory of
Shells); A.S. Vol'mir, Gibkiye plastJnkl i obolochki (Elastic plates and shells);
A.L. Gol'denveyzer, Teoriya uprugikh tonkikh obolochek (Theory of thin elastic
shells); A.I. Lurtye, Statika tonkostennykh uprugikh obolochek (Statics of thin-
walled elastic sh_11s); Kh. M. Mushtar' and K.Z. Galimov, Nelineynaya teoriya
uprugikh obolochek (Nonlinear theory of elastic shells); V.V. Novozhilov,
Teoriya tonkikh obolochek (Theory of thin shells). These monographs shed
light on many fundamental problems in the theory of shells in general. How_.ver,
there is almost nothing therein dealing with the theory of anisotropic laminar
shells.
Despite the large number of articles appearing in journals, there is as
yet not one book devoted to the theory of anisotropic laminar shells. I_Lthe
present book the author partially fills this gap.
The text is based on the author's investigations over the last few years.
It consists of the following divisions: (a) fundamental equations of the theory of
elasticity of an anisotropic body in curvilinear coordinates; (b) general theory
of anisotropic laminar shells; (c) membrane theory of anisotropic shells; (d)
theory of symmetrically loaded anisotropic shells of revolution; (e) anisotropic
cylindrical shells; (f) shallow anisotropic shells; (g) new theories of nniso-
tropic shells and plates.
In distinction from the first divisions of the book, which are based on the
hypothesis of nondeformable normals as given for the stack of the shell as a
V
1964012887-007
whole, the iast chapter attempts to construct a theory of essentially anisotropic
shells and plates without the hypothesis of nondeformable normals.
The book does not deal with the undeniablyimportant problems of non-
linearthe')ry,the theoriesof stabilitya_.dvibration,as well as temperature
problems of anisotropiclaminar shells. Nor does itdealwith problems associ--
atedwithplasticand elastic-plastic deformations of the material of the shell
layers,sincethese problems have not been adequatelyinvestigated.
Within each chapter the formulas have a two-part enumeration. Where
reference is made to the formulas of preceding chapters a three-part enumer-
ation is used (the first digit referring to the chapter).
In conclusion I wish to express my deep gratitude to my colleagues at
the Institute of Mathematics and Mechanics AN of the Armenian SSR, D.V.
Peshtmaldzhyan, A.A. Khachatryan and L.A. Movsesyan, who were of
great assistance in the preparation of this book.
S.A. Ambartsumyan
May 1959, Yerevan
vt
1964012887-008
CHAPTER I
FUNDAMENTAL EQUATIONS OF THE THEORY OF ELASTICITY
OF AN ANISOTROPIC BODY IN CURVILINEAR COORDINATES
SECTION 1. SOME REMARKS ON CURVILINEAR
COORDINATES IN SPACE
Here we 3halldiscuss in briefand withoutproof those aspects of the
theoryof curvilinearcoordinatesin space which willbe used in subsequentdis-
cussion.
As is known, the positionof any pointM in space may be uniquelydefined
by itsradiusvector r relativeto a certainfixedpointO. In rectangularCartesian
coordinatesfor r we have
r --_xi-4-yj-t.-zk, (1.1)
where, as usual, l, j. k represent the corresponding unit vectors.
In problems of the theory of shells the location of any point M with radius-
vector r is conveniently defined not by three cartesian coordinates x, y and z
but by any three other numbers a, It, _. The quantities a, _, .f.which uniquely de-
fine the location of point M in space are known as the curvilinear coordinates of
point M. Each of these coordinates is a function of radius vector r or a function
of the components x, y and z of this radius-vector in Cartesian coordinates:
a (r) :- a Cx. y. z). /
(r)= {_(x,y,z).I (I.2)
7[(r) = 7[Cx, y,z).
Conversely, since the radiu9 vector r of any point M in space _.s completely de-
fined when we are given _, B and y, it is a function of these independent
1964012887-010
2
variables, and, consequently, the components of this radius vector will be func-
tions of curvilinear coordinates
x-= xta, _. "I), Y_-Y(_, d,'l), z-=z(a, _, "I). (1.3)
In (1.2), assumir.g - (r) -- tonst. _(r) -- const, and _ (r) =--const, we obtain three fam-
ilies of surfaces. Through each point M in space there passes a surface of
each of these families. These surfaces are known as the coordinate surfaces.
The lines of intersection of the coordinate surfaces are referred ;o as coordin-
ate l'mes.
It is evident that in our case there will be three such coordinate lines:
coordinate line _ (with current ccord_nate _), which is formed by the intersec-
tion of coordinate surfaces ;_: co,st and-_ co_t; coordinate line _ (with current
coordinate 8), wb_ieh is formed by the intersection of coordinate surfaces
a-_ constand "_-:- const; coordinate line _ (with current coordinate y), which is
formed by the intersection of coordinate surfaces _ - const and -, =- cow, st (Figure 1).
y
/
I/
Figure i
In subsequentdiscussionwe shallbe interestedonlyin orthogonalc .rvi-
linearcoordinates,thatis, those curvilinearcoordinatesof which allcoordinate
lines_, B and _,are mutually perpendicularat each pointM in space. Inthe
giventriorthogonalsystem of curvilinearcoordinateswe have for the square of
a linearelement in space
ds2= H_ #d" -4- H_ d__-q-!:t_d'l", (1.4)
where
l). t3,"0,
1964012887-011
in the general case of orthogonal curvilinear coordinates, are functions of in-
dependent variables o_, B and y and are known as tSe Lame coefficients.
With the given relations of (1.3} for the given coordinate system the
Lame coefficients are determined by means of the following formulas:
Hi=lxl t')Y'? I
H:=\o_] _ _ _ .)2; (1.5)
" _ 06 ] )2,
/ 0x _ 0v 2 0z
For example, in cylindrical coordinates with _ := R. _ :,_. _ ----- :. when the re-
lations in (1.3) have the form
x := R ,:o_ ?, y _- R ,,in 9, z -= z,
from (1.5) we obtain for the Lame coefficients
H1:=1, H2=r, H:_=I.
In spherical coordinates for a =/_, 3 -_ _. _ --:-._. when the relations in (1.3) have
the form
x ---_R sin 0 cos ?, y -_- R sin rjsin 9, z --- R cos 0,
from (1.5) we bave for the Lame coefficients
H, --_ 1, H2=R, H3: R sin rj.
Finally, in Cartesian rectangular coordinates x, y, z, as is known, the formula
for the square of a linear element has the furm
ds 2 .= dx 2 _- dy 2 -}- dz 2,
2 2"
_ /..," ,/zd
Figure 2 Figure 3
1964012887-012
4
in virtue of which we have for the Lame coefficients
141= 1t2-= H3 = 1.
Coefficients Hi, as functions of the three independent variables c_, _ and
3/, ir_ any system of curvilinear orthogonal coordin ctes must satisfy the following
six differential equations:
( ) ( ) ]
2 , ,w, -t- , ,w. _)_ /I x a_ a,( II: a,',
( ) P (I ,)tl,) I ,)t12 #,t ,
__a i ,2/.,., + 4- --,- ........ o; (1.6)
()[', -/-I-, a: _1.( Jt_ a; II i d_ u_ "
a:tt, ! _)1t_ dlt I I dtt_ #ttl __ O;
2,j_,__ ___ eta, OH, __L o/,', oH, o: (1.7)
d_ 07 ti 3 O_ O'[ ttl "-d'_- -_)_- ----:-
d21t3 1 off, dH 3 I dt12 dli3 O.
Proof of the validity of these equations may be found in texts on differential
geometry. *
SECTION 2. DEFORMATION COMPG.ENTS AND DIFFERENTIAL
EQUATIONS OF EQUILIBRIU,.I IN THE TRIORTHOGONAL SYSTEM
OF CURVILINEAR COORDINATES
Let a solid body expressed ir, ehe triorthogona| system of curvilinear co-
ordinates eL, B, and 3/ undergo deformation m,der the influence of any forces.
Then any point M belonging to the body and having the coordinates or, _ and 3/will
acquire a displacement which may be represented b:,, the following three pro-
jections of the total displacement vector onto the directions of the tangents t_
coordinate lines _, _ and 3/.
", = ", (=, _, 7); }
% = -_(=, _, "f); J (2.1)
% -= "I(=' _' _r).
*S.P. Finikov, Teoriya poverkhnostey (Theory of surfaces),ONTI, Gos-
tekhizdat,1934, pp. 185.-186.
1964012887-013
5
Henceforth, we shall refer to all these quantities as the displacements of point
M. For the positive displacements let us useu,, uvand uv directed toward
positive changes of the corresponding independent variables _, _ and y.
The deformed state of a solid three-dimensional body- in the vicinity of
point M is characterized by six deformation components. Of these components
three, which are designated by e. e+. e_, are the relative deformations of
elongation in three mutually perpend_,cular directions _, _ and y, respectively,
while the remaining three, designated by e,_. e+_. G,' are the corresponding de-
formations of shear occurring in three mutually perpendicular planes, which
planes are tangent at point M to three mutually perpendicular coordinate
surfaces T= const. ==const, I_-_ const.
The deformation components e,. e_..... e_, are associated with dis-
placements u, u_. and u_ of point M by means of the following formulas:
| Ou= 1 dH_ 1 _.Hi I
ea_-ill da -_ HIH2 #_ u;+_-HIH3 OT ul;
f
I c)u3 I OH2 1 c)H2
"'--Hz O_ + 142143 O'l U.f-.{-It, H= a= U,,; [ (2.2)
1
Ou.t I OHa 1 OH3
/t, a't + HsH, a,, u=-+-H3H, o_ u:,;
I
(.1.
kit 0 H3 0
H, 0 0 ]
e,,--._ -_-g( _--_u+)+ "
-
As is known, the stress at any point M of a solid three-dimensional
body is characterized by a stress tensor which is defined by nine compovents.
Of these components three are 9rmal stresses which act in three mutually
perpendicular directions of coordinate lines _, B and y and six are shearing
stresses acting in three mutually perpendicular planes which are tangent at
point M to three mutually perpendicular coordinate surfaces -----onst, _-----,onst.
and _ = const. In virtue of the duadicity of the shearing stresses the number of
independent stresses is not nine, but only six.
% a_, % are normal stresses with the subscripts indicating the direction
of the externa_ normal to that area to which the given stresses, let.
1964012887-014
6
_.__- _.' _.T_ zT.' _T-_ _ are shearing stresses; the first subscript
indicates the direction in which the _ven shearing stress acts and the second
subscript indicates the direction of the external normal to the area to which
the given stress is applied.
All stresses, both normal and shearing, are considered positive if,
upon being applied to the areas with positive external normals, they act along
the direction of the corresponding positive external normals (Figure 4).
Y
P
Figuce 4.
If the 3olid three-dimensional body under consideration is in equilibrium,
the equilibrium conditions of a differential element of the body in an arbitrarily
chosen triorthogonal system of curvilinear coordinates are represented by the
following three differential equations:
8 d
a H dH2 a H dHs OH_
Jr- _,H2-_ -}-P,HIH2H3 --_0; (2,. 4 v)
(HsH,.%) _ d
0"-_ + "-_ (H2HI_) -]- _ (H_H3_") --
#Hs
a.Hs _ +
OHs
+ ",_""_.;_ _dllt ..}_p_HiHoHa==O; (2 .4")
1964012887-015
) c) H._H2,c_.jr. d
#H, _H2 _4__._H20Hs
-- %H2-_ _:H1 a'_ -- _ -.-4.-
0H3
_. ,cpTi-ll _ "_. p1HIH2Ha _ (j, (2.4 '_)
where P.= P.(_, P, _), P_ := P_(_, P, _),and PT= P_(_, P' 7) are tLe corresponding
projections of the volumetric force onto the directions of the t_,:_ents to the
coordinate lines _, /_ and _/.
The reader will find a detailed discussion of the results _.resented in the
first two sections of this chapter in the work by V.Z. Vlasov. * These subjects
are also discussed in many courses in the theory of elasticity (for example,
L.S. Leybenzon** or A.I. Lur'ye). ***
SECTION 3. CURVILINE#_R ANISOTROPY. GENERALIZED
HOOKE 'S LAW,****,'****
Let us assume that the discuseed uniform elastic body is cu__vi, _uearly aniso-
tropic (that is, from the standpoint of elastic properties the equ_,. _, ent direc-
tions are not the parallel directions through various points of tbc _;_ly but
directions which conform to other rules). In choosing a system of curvilinear
coordinates so that at each point the elastically equivalent directions coincide
with coordivate directions, we note that infinitely small elements defined at
different points of the body by three pairs of coordinate surfaces, being anti-
sotropic, pos se s s identical elastic propertie s.
We shall assume that the discussed solid elastic body undergoes small
deformations and obeys the generalized Hooke's Law. Ip the general case of
a uniform curvilinearly anisotropic body the general_.zed Hooke's Law in the
discussed system of _.riorthogonal coordinates ol, .Band y has the following
form:
*V.Z. Vlasov, Obshchaya teoriya obolochek, Gostekhizdat, 1949, pp. 181-
182.
**L.S. Leybenzon, Kurs teorti unrugosti (Course in the theory of elas-
ticity), Gostekhizdat, 1947, pp. 125, 138.
***A. I. Lur'ye, Prostranstvennyye zadachi teorii uprugosti (Spatial prob-
lems in the theory of elasticity), Gostekhizdata 1955, pp. 29-49.
****S. G. Lekhnitskiy, Teoriya uprugosti anizotropnogo tela (Theory of
elasticity c/an anisotropic 1_9dy), Gostekhizdat, 1950, pp. 15-33.
*****S.G. Lekhnitskiy, Anizotropnyye plastinki (Anisotropic plates), Gos-
tekhizdat, 1957, pp. 15-23.
1964012887-016
8
e, -_ ann,, -+-al2% --{-al3al -+-- vt'c_,+ alsx,_ --{-ate_,_;
an. +- + ......... + (3.1)
, * . , o .
e,_ --_alsa, -t-"a_% -4- ......... "-t-a_x,_,
where ajk are the elastic constants (coefficients of deformation); there is a total
of 21 independent elastic constants; in terms of engineering constants, they are
as follows:
1 _ !
l" all "_- "_a' a_ : --_, d33 = W"
I!. a]2:_------ E_' a2s:--_: E.t'
'va. [ _,la
1 ! I
I!1 a 0, ass ac_ .
O_._ -- G_q -- O,_
IV. a_-- (Jc,t ti_o a46-_- _ '
--_ ' GEt O_
a4s-- G_ J,t
V. a14:---_- U_t ' a_s--_--_-_ -- O,_ ' aa_..... E1 -- O,_ "
where e,. E_. and E_ are Young's moduli in directions or, _ and _,, respectively;
O_v O,v O,_are the shear moduli for the --_--- . v,,-,_s parallel at each point to the
coordinate surfaces a -= const. [_= const, and _ = const; _. v_. _,v _,_' "i,v '_a are
Poisson's ratios characterizing the transverse compression in the presence
of tension in the directioi_ of the coordinate axes (the first subscript indicates
the chrection of transverse compression and the second subscript indicates
the direction of the force); }h,._v _',._.,_..... _.,_ are Chentsov coefficients char-
acterizing the shear in planes tangent to the coordinate su','faces and caused by
shearing stresses acting in. planes tangent to the other coordinate surfaces.
For example, _,,,_ characterizes the shear in a plane tangex.at to the coordin-
ate surface _ = const, and caused by stress 'nr %_.,' _,., ..... _,._ are the co-
efficients of interaction of the first kind and _a. ;m _. _t ..... _, ,., are the coefficients
1964012887-017
of interaction of the second kind. The first of these coefficients characterize the
shear in planes tangent to the coordinate surfaces and caused by normal
stresses; the second coefficients characterize elongation in the directions of
the coordinates and caused by shearing stresses. The coefficients of elasticity
aik given above are grouped according to the classification proposed by
p. Bekhterev* (the Roman numerals indicate the group numbers).
For the elastic potential referred to a unit volume of the body, we have
-- I 2+%
1
_- _-an_ + a_(a_v 4- a_4_1-F a2s_, + a_%_)_-
1 2
+ _ aa3a_+ nI (a34":_.f -_ a3s'C_, + a_ O) +
1 2
! 2
"4-"_ a_% + _,a_,_ 4-
1 2 (3.2)
On the b_sis of {3.1) V may also be represented in bilinear form
-_-(&e, --__e_ 4-- n_e_4- _e_v 4-':_,e_, + %_e,_). (3.3)
The potenti_al energy of det'ormatlon for the entire body is determined by inte-
gration of V over the entire volume of the body.
f Y
If there exists any symmetry in the internal structure of the material
of an anisotropic body, then m its elastic properties there is detected a cer-
tain elastic symmetry, th_.t _s, there exist symmetrical directions relative to
which the elastic properties of the material are identical; in this case the
symmetrical directions are known as equivalent directions.
When an anisotropic body possesses elastic symmetry the equations
ot Hooke's Law in gentral form are simplified. Let us discuss a few of the
more important cases of elastic symmetry.
1. Plane of Elastic Symmetry. Assume that at each point of the body
there is a plane possessing the property _hat any two directions which are
symmetrical relative to this plane are equival_nt with respect to their etastic
* P. Bekhterev, Analiticheskoye issledovaniye obobshchcnnogo zakona Guka
(Analytical investigation c_ generalized Hooke's Law), Author's lithograph,
Leningrad, 1925.
1964012887-018
10
properties. Assuming that coordinate 7 at each point of a curvilinearly
anisotropic body is perpendicular to the plane of elastic symmetry (that is,
the plane of elastic symmetry at each point is parallel to the coordinate
surface _= cot.st), we obtain the following equatigns for the generalized Hooke's
Law:
G= all:, + a12%+ a13%+ al_:o_;
e;m -=- a45_ _ -,4- as5"_.l;
/
e.:,: al6_. -F-a2sz,_ + a3601-F" a_'c,_. J
In thi._ t.r_se the number of independent elastic constants aik is reduced to 13.
Equations (3.5) may also be represented in the following form:
l _,2 _J3 -4-- r'l' t_
_'at "_a2 i r,3, _
(a.6)
i _._, _,
Fat, :_ 1
_,, , r_, t v_,_, a i
Here and henceforth the literal subscripts of the engineering constants
are replaced by numeral subscripts; t_ is replaced by 1, _ by 2 and _ by 3.
Directions perp_.ndieular to the plane of elastic symmetry are lmowxa
as the principal dire_ tions of elasticity. In the discussed case of elastic
; symmetry one principal direction passes through each u_.int of the body.
2. Three Planes of Elamic Syrametry. Through each point of the body
let there pass three mutually perpendicular planes of elastic symmetry.
i Assuming that at each point of a curvilinear anisotropic body these planes are
l perpendicular to the corresponding orthogonal coordinate directions a, _,
(that is, all three planes of elastic symmetry at each point of the body are
parallel to the coordinate surfaces a = const. [t= const, and -f_ const), we ob-
[ tain the follow*rig equations for the generalized Hooke's Law:
9
1964012887-019
11
e, == allaa-+- a!2_ _ -+- al3al; e_l _ aoix_7; /
e_ = at2_, + ar_%-+-a, _; el,-= a_'_,; I (3. '7)
e_ _ alaaa -+- a23 _) -_ a3aal; ea) _ aGG-ta).
In this case the number of independent elastic constants aik is nine
Equatio, s (3.7) may alto be l:resented in the following form:
_,, _, 1 ] (3.8)
1
el _ -- E"]- "-- -_t- % Ea _i,z "_,0"
Here, in virtue of the symmetry of equations O. 7), w,' have the relations
E2_21= Etvt2, Es_s2= E2_s, EF, s-- Es_s,. (3.9)
A body in which at each point there are three mutually p_rlmndicular
planes of elastic symmetry is known as an orthogcnally mzisotropic or ortho-
tropic body.
3. Plane of Isotropy. Through each point of a body let there pass a plane
in which all directions are elastically equivalent. Assuming that in a curvilin-
early anisotropic body coordinate _ is at each point perpendicular to the plane
of isotropy (that is, the plane of isotropy is at each poin_ parallel to the co-
ordinate surface _ -- const), we obtain the following equations for Hooke's Law
in general form:
In this case the number of independent elastic constar, _. ajk is five.
Equations (3.10) may also be given in the following form:
_' |
! (a, _ v%)-- at;
e_= -E"(a;_-- va,)-- _ as: e_,= -Or .c1,;
_' 1 2(I _- _)
e_-- -- -B O, + %) + -_ %; e,_= _ 't,r
O
1964012887-020
12
Here E is Yotmg's modulus for directions in the plane of isotropy; E' is Young's
modulus for directions perpendicular to the plane of isotropy; v is the Poisson'a
ratio characterizing contraction in the plane of isotropy in the presence of
tension in this same plane; v t is the Poisson's ratio characterizing contrac-
tion in the plane of isotropy in the presence of tension in the directiora per-
pendicular to this plane; G' is the shear modulus for planes normal to the
E
plane of isotropy; O _-_ is the shear modulus for planes parallel to the
plane of isotropy.
A body possessing the mentioned elastic properties is known as a trans-
versely isotropic body.
In the discussed case of elastic symmetry the direction perpendicular
to the plane of isotropy and all directions in this plane are principal directions.
4, Complete Symmetry -- Isotropic Body. Here all directions are
equivalent and any plane at any point of the body is a plane of elastic symmetry.
In thls case the equations for the generalized Hooke's Law have the form
1 /
1 _(%-or _.f)l; e_ _- _: e, -_- -_-[%_
e_= --K[a_-- _(_ -4- %)1; e_, = "U"_" (3.12)
I 1
e_:T[a :- *(o,+-a_)l; e,_=-U'... _ j
Here 1_.is YoungVs modulus; l_ i_ tke Poisson:s ratio; ,.__-- E
2(1 _-,) is the shear
moduLus. The number of indepevdent elastic constants is two.
From (3.2) Jt is easy to derive the expressions for potential energy
in any special case of anisotropy_ In view of their elementary nature we
shall not present these constructions here,.
SECTION 4. TRANSFORMAx'TION OF ELASTIC CONSTANTS VqITIt
ROTATION OF THE COORDINATE SYSTEIVI__, **
In the theory of elasticity of an anisotropic body we encounter problems
in which the solutions require the values of elastic constants for a certain
*S. G. I_khuJtskly, see pp. 33-48 of work cited in footnote **** of p.
7.
**S.G. Leldmttskly, see pp. 44-48 of work cited in footnote ***** of p.
7.
1964012887-021
13
system of coordinates a', _' and 7' when the elastic constant_ for another
system of coordinates a, _ and 7 are known.
Let us discuss the generalized plane stress of an anisotropic plate.
At each point of the material of the plate let there be only one plane of elastic
symmetry parallel to the median plane of the plate, which coincides with the
combined coordinate planes _ and _'_'. Let the coordinate _ystem (_, 6. _)
and (_'. 3', T') be obtained one from the other by rotation through a certain
angle _ around the common axis Y-
Proceeding in the known manner, * we obtain the following transforma-
tion formulas of the elastic constants:
a'n = an cs_? +(2an +- aee)sin__ cs2 _ + (4.1)
-+-a22sm4_ -t- (ale cos2_ -4- a2esin__) sin 2_,
a_ -= a n sin4_ -'l-(2an + ass)sin2_ cos2? -'1- (4.2)
--t-a_ cos4_ -- (a;s sin2? -'l- a2ecos_ T)sin 2?.
/
a n ----- (an -Jr-a_ -- 2an -- ass) sin2? cos2_ +
1 (4.3)
-'l- an-k- _- (a_s-- als) sin 2_ cos 2?,
I
aee= 4 (al; -t- a_ -- 2an -- a_s)sin2_ocos2_ -I-
-_- ace -4- 2 (a_s -- ale ) sin 2cpcos 2% (4.4)
' Ia22 sin2 Cos2 ,.
a.16 =- ? -- _qll
L
1 (2ax 2 -I- a66) COS
27] sin 2:p -+- al_ cos 2 ? (cos 2? -- 3 sin 27) _-
+
2
(4.5)
-t- a26 sin2 ? (3 cos 2 '_ -- sin _?),
' [ " 1 (2a,2_L_ 27]
a26___ a2_cos.?__ansin2?-- _- ac_);os sin2?-_--
4- a_6 sin2? (3 cos 2? -- sin2 ?) + a2_ cos _? (;os 2 ? -- 3 sin2?). (4.6)
Let us give the following invariant relations:
I I I
a, t --t- a_ --{- 2a_2 = a_ + a_ -4- 2at2,
a'_ -- 4a',_----%6-- "a;r (4.7)
in the special case where we have an orthotropic plate for which the principal
directions of elasticity coincide with the directions of coordinate lines a, _,
and for them the principal elastic constants are known, from (4.1)-(4.6) by
virtue of (3.8), we obtain for the elastic constants atjk
(__ ) sin*? (4.8)
all_' cos*,_ T _+_ 2_,E, sin2 T COS__P't- Es '
, sin* T [ I '
a_ --__ Jr __- -- _'_e, ! _in__Pcs_ _0+- _cos'/_l _ ' (4.9)
1964012887-022
, ! { 14- '1 I-*-_2 i)4n 227_ " ,
aL,-----4- _---K- + E, (s _,- (4.1O)
{._l_-._ "q 4- 1-i"',_ 1 ) sin22? 4 1
a,. =_ E, F-2 O O' (4.11)
, [ (_,,,,,_ c,,.,, _ )+ alb -- 2 t: 2 El
+(,f (4.12)
E, ) (cs2? ?
-- sit,- ?_j sin cos?.
, i (cn_0 s,n'_) a2t' =- ."2)\ /=2 PZt
From (4.7) we find the corresponding invariant expressions for an or-
thotropic material
!
1 I 2h 1 1 2"q
e-i-+., , -- --t
12z E 1 Et E2 E l
!
l 4_ t 1 4h
l- _ -- (4.14)
G' _ G ' /-",
In formulas (4.8)-(4.14) the following new notation is introduced:
v,2= v2, v2,= _,. (Ep2= E2h). 0,2:=0 (4.15)
The notationgiven in (4.15)willalso be used subsequently.
SECTION 5. ELASTIC CONSTANTS FOR CERTAIN ANISOTROPIC
MATERIALS
Of the various types of anisotropicbodies we shalldiscuss onlynon-
crystallineanisotropicmaterials, which are widelyused in modern engineering
for shells. These materials include,for example, naturalwood, concrete,
rolledsheet metal, paper, deltawood and plywood, fabrics,and others. In
thefirstapproximationwe shallassume thatthe d_scussed materials are homo-
geneous, ortlaotropic and subjectedto a generalizedplane stress. We shall
assume thattheprincipalplane xOy is the center plane of an elementary plate
cutfrom the discussed anisotropiematerial.
1. Natural Wood. Due to the presence of annual layers, naturalwood
isregarded as an anisotropicbody. Disregarding nonuniformityand curva-
tureof the layers innaturalwood, itis easy to detectthree planes ofelastic
symmetry: one of them (yOz)is normal to the grain of tha wood, the second
(xOy)is parallelto theplane of the annual layers and the third(xOy)is ortho-
gonalto the firsttwo (Figure5).
1964012887-023
15
?
Figure 5
By way of example let us present the numerical values of the elastic
constants for pine:*
EI=E_r= I . 10Skgic.a2; E2=E>=O,042. 10Skg.'c_2:
(5
1)
_xy _ _a _-_ 0.01 ; Oi2 _ Oxy = 0.075 10s kg 'c._ 2. _ "
A detailed discussion of the physical characteristics of wood may be found in
the work by A.N. Mitinskiy. **
2. Delta Wood. Delta wood is a laminated wood material obtail,ed by
hot pressing of a wood stack consisting of a large number of layers (ph_s)
impregnated with resin. At intervals of every ten plies with longitudinally
directed grain one layer with transversely directed gxain is inserted; delta
wot;d is essentially laminar and anisotropic. However, to the first approxi-
mation we shall assume that delta wood is a homogeneous orthotropic ma-
terial. Assuming that the x axis coincides with the direction of the prevailing
grain (Figure 6), for the averaged elastic constants of delta wood we have the
following numerical values:***
E1=-:3,05 10Skg/c_2; E2= 0,467 10Skg/c_2; ]
'2 =-:0,02; O2_-=-0,22, 105kg/c ._. I (5.2)
*A. L. Rabinovich, Concerning elasticconstantsand strengthof anisotropic
materials,Trudy TsAGI, No. 582, 1946.
**A. N. Mitinskiy,Elasticconstantsofwood as an orthotropicmaterial,
Trudy Lesotekhnicheskoyakademii ira.S.M. Kirova, No. 63, 1948.
***A L. Rabinovich, see footnote* on thispage.
1964012887-024
16
Iz
Figure 6
3. Plywood. Let us discuss birch plywood made of an odd number of
layers (plies) bound by bakelite film. The plies are symmetrically arranged
relative to the center ply and the directions of the grain of two adjacent
layers are mutually perpendicular (Figure 7). A sheet of plywood is laminar
and anisotropic. However, in the first approximation we may assume that
in the case of plywood we also have a homogeneous orthotropic material.
For the averaged elastic constants of three-layer plywood with thick-
nesses of 1.0, 1.5, 2.0, 2.5, 3.0, 4.0 and 5.0 mm we have the following
numerical values:*
E1-- 1,2. 10skg/c_2; E2_-0,6. 10Skg/c_2; [
(5.3)
v2 --_ 0,036; O12 _ 0,07 105kglc._ _.
!
Here it is assumed that the x and y axes are parallel and perpendicular, re-
spectively, to the grains of the outer layers of the plywood sheet.
Shells made of delta wood or of plywood will be dealt with in subsequent
discussion as laminated structures of orthotropic layers of wood (plies).
4. Concrete. By virtue of the technological characteristics in manu-
facture and placement of concrete this material acquires the properties of
*Spravoclmik aviakonstruktora, t. III, Prochnost' samoleta (Aircraft
designers handbook, Vol. III, Aircraft strength), Izd. TsAGI, 1939, p.
325.
1964012887-025
17
anisotropy. We present the numerical values of the modulus of elasticity of
anisotropic (orthotropic) lightweight concrete:*
E,:= 1,08. 105kg/c.u2; E2_---0.81 105lrg/ca4_. (5.4)
5. Paper. The technological aspects of paper manufacture are such
that paper generally possesses the properties of anisotropy.
The numerical values of the elastic constants of semi-Whatman paper**
(produced by the Goznak plant) having a density of 160 g/m 3 are:
El_---30,1" 103kg/c_2; E2=22,6" 103kg/._2; I
v2: 0,23; O12= 9,96 103kg/c.u2. I (5.5)
Iz
b
Figure 7.
*K.S. Karapetyan, Intluence of anisotropy on creep of concrete, Izves-
tiya AN ArmSSR, FM nauki, Vol. X, No. 6, 1957.
**V. M. Chebanov, Investigation of stability of thin-walled shells by
means of paper models, Inzhenernyy sbornik, Vol. XXII, 19_5.
1964012887-026
D
18
CHAPTER II
FUNDAMENTAL EQUATIONS OF THE THFORY OF SI_'ELLS CONSISTING
OF AN ARBITRARY NUMBEB OF ANISOTROPIC LAYERS
SECTION 1. BASIC CONCEPTS, INITIAL RELATIONSHIPS AND
H_ POTHESES
Let us examine a multi-layer thin shell of uniform thickness h consist-
ing of mu arbitrary number of homogeneous anisotropic layers of uniform
thickness ti (Figure 8).
fY
J
Figure 8
It is assumed that at each point of each layer of the shell there is only
one plane of axial symmetry parallel to the coordinate surface of the shell;*
the coordinate surface is parallel to the outer surfaces of the shell and lies
*Here and henceforth we shall assume that the term "coordinate surface
of the shell" refers to the surface _ = 0.
i
1964012887-028
19
within any i-th layer. In special cases, the coordinate surface of the shell
may also be any of the surfaces of contact of the layers or any of the boundary
surfaces of the shell*,**,***, ****
The c, and/3 be curvilinear orthogonal c9ordinate_ coinciding wii.h the
lines of principal curvatu.re of the coordinate surface of the shell while 3/, be-
"rig normal to the coordinate lines _ _ cons+, and _ =: const, is rectilinear and rep-
resents the distance along the normal from point (a, _) of the coordinate
surface to point (a.._. 7_ of the shell (Figure 9). We assume that all layers of
the shell remain elastic in the presence of deformation, that is, they obey the
generalized Hooke's Law (1.3.5) and function simultaneously without slip.
In the chosen triorthogonal system of coordinates for the Lame coef-
ficients (1.1.5) we ",,,,illhave*****
Hi=A(1 _t-kl'{). H2=B(I _-k_.y),
11/3 --: l, (1.1)
/
,ql
Figure 9
where A --- ,4(a, It) and B --:-B (2, _l are coefficients of the first quadratic form of
the coordinate surface, k_ ---k, (a..3). /_2= k2(_. _)-- are the principal curvatures
of the coordinate surface of the shell at lines _ --- const, and a = cons,, respectively.
'_D. Yu. Panov, Heat resistance of bimetallic shells (toward a theory of
a heat _;,_tch). PMM, Vol. XI, No. 6, 1947.
**8.A. Ambartsumyan, Anisotropic laminar shells (problems of strength).
Dissertation, 1951.
***S.A. Ambartsumyan, Concerning design calculations for laminar
anisotropic shells. Izvestiya AN Arm SSR (FM'YeiT nauki), Vol. VI, No. 3, 1953.
****E. I. Grigolyuk, Thin bimetallic shells and plates. Inzhenernyy sbornlk,
Vol. XVII, 1953.
*****V.Z. Vlasov, Obshchaya teoriya obolocnek, Gostekhizdat, 1949, pp.
212-214,
1964012887-029
2O
Inserting the values of Hj from (1.1) into (1 4.3) and _tssuming T= 0.
we obtain the well-known Gauss-Codazzi relationships for the coordinate sur-
face of a sheU
0(__0B_ , 0/! OA
- _j+_)=--k,k_._B; i
0_
(A_,) : k__ _ I 1.,.)
Invirtueof relationsIxips (I.2) we also have
1 OH_ 1 OU I OH, I OA (1.3)
wldch will be used in subsequent discussion.
For the deformation compownts of the i-th layer of the shell of (1.2.2)
and (1.2.3) we obtain (for the sake of simplicity we have omitted the i subscript_
of H1 and H2)
, 10u _- ! OH_u_-a- _ 0_ t. (1-. 4)
10u_ 1 OH2 t 10H2 I (1.5)
eJ = -ff2-_ _f- H,Ha _ U"+-A_ -_- UT;
l __ Oulr.
e_---_-, (1.6)
HI O H2 1 l
e_ := H 0
O 1
.,:n,_(_.)+_, ' (z.9)
B]_.-,_.
The equilibrium equations in (1.2.4) are also simplified, which in the
chosen system of coordinates for the 8-th layer of the shell have the following
form:
- o ., _(n%)+
jr. 1_ O(H?H_x___4_H_Hap_" ._ O; (1.10)
1964012887-030
21
(n,o_) , o,, i o..._,.
-- a _ t 74%_ (n2%-) +
I 0
-r -kl-a _ (HLII_!_) + HiH._P_= 0; (1.11)
(n, ne',)- %_'-o'n a-, __
0
+ _.CM_, )+_(,4,) +H,H_P' =0. (z. 12)
The initial, equationsof the generalizedHooke's Iaw (1.3.5)for the
i-th layer are written as follows:
The basic premise for construction of a theory of thin ,qnisotropie lam-
inar shells is the wel!--kitown hypothesis of nondeformable nurmals. It is formu-
lated as follows: after deformation a rectilinear element normal to the coordin-
ate surface of a shell remains rectilinear, normal to the deformed coordinate
surface of the sheil and preserves its length. Usually this geometric assumption
ie supplemented by the static assumption that the normal stresses _' at areas
parallel to the coordinate surface of a thin shell may be disregarded in compari-
son with other stresses.
As is .known, the classical formulation adopted here for the hypothesis
of nondeformable normals is not, strictly speaking, a fortunate choice and re-
quires refinement* ,**, the more so since we are discussing anisotropic
laminar shells. ***
In applying the hypothesis of nondeformable normals we introduce into
the theory of shells a certain unavoidable error which, as investigations
*A. L. Gol'deaveyzer, Teoriya uprugikh tonktkh cbolochek, Gostekhizdat,
1953, pp. 64-66.
"*V.V. Novozhilov, Teoriya tonkikh obolochek, Sudpromgiz, 195!. p. 33.
***S. A. Ambartsumyan, Toward calculation of two-layer orthotropic shells,
Izvestiya OTN, AN SSSR, No. 7, 1957.
1964012887-031
22
reveal*,**,***, will be of the order of h/Ri. however, there are cases in
which this error is considerably greater. **** In the present work we shall
a.asume that tke. error introduced into the theory of shells by the t,ypothesis
of nondeformable normals is at least as great as h'R,.
Finally, with the usual engineering assumption of a relative error of
1
5%, those shells vail be considered thin in which***** max (hkj) _< _ and at the
same time max (h,'a)_ s,where a is the minimum linear dimension of the shell
at the c,_ordinate surface and is a small quantity (for example, for an
isotropic shell---0.1), which will be established subsequently. The second
condition (borrowed from the theory of plates) is a necessary condition, since
if a thiu shell is defined only frem the point of view -_f the ratio of the thick-
ness of the shell to the minimum radius of curvature of the coor'.ilnate surface
(the first condition), from the point of view of the theory of 9rotes (second
condition) it may be thick and the adopted basic assumption becomes un-
aeceptable.
The definition of a thin shell as given here is somewhat conditional
in nature, since if the thickness of the shell is regarded from the stand-
point of the possibility of applying the hypothesis of nondeformable normals,
the geometric definition given here for a thin shell will, in the case of artiso-
tropic laminar shells, require substantial nmdification.
We shall return to the problem of formulation of the hypothesis of
nondeformable normals and, accordingly, to the definition of a thin anisotropic
shell (particularly in the last chap:zr).
SECTION 2. DISPLACEMENTS AND DEFORMATIONS
The geometric hypothesis of nondeformable normals, given for the
entire stack of the shell as a whole, relieves us to the necessity of investigat-
ing the displzeements and aeformations of each layer individually.
*B. G. Galerkin, Equilibrium of 0, spherical elastic shell, PMM, Vol
VI, No. 6, 1942.
**A.I. Lur'ye, Equilibrium of a symmetrically loaded, spherical,
elastic shell, PMM, Vol. VII, No. 6, 1943.
***V.V. Novozhilov, R.M. Finkel'shteyn, concerning error in the
Ki='khoff's hypothesis in the taeory of shells, PMM, Vol. VII, No. 5, 1943.
****A. L. Goi'denveyzer, see p. 66 and 236 of the work cited in reference
1 above.
*****V.V. Novozhilov, see p. 6 of the work cited in reference 2 above.
1964012887-032
23
Having the deformations of elongationand shear as well as the par-
arneterscharacterizingthe change in curvature and torsicnof the coordinate
surfaceof the shell,we may determine the deformations and displacements of
any layer of the shellby elementary means. Moreover, as is easilyshown,
allthe characteristicsof deformation and displacement of each layer are ob-
tainedfrom thedeformations and displacements of a certainhomogeneous
shell. Using thefundamental hypothesis,we may consider* thatofthe six
relationshipsin (1.3.5)the third,fourthand fifthmay be replaced by the
followingapproxih_ateequalities-
eT--O. e_--_O, e,_=O (2.1)
or, for an _ndividual layer of the shell
ei=O. e_---0.d =0. (2.2)
T _'T _T
which are formally equivalent to the assumption that tl-.e deformation of the
shell as a whole occurs without shear deformations e._. and e_ in the planes
of normal sections and without elongation e_ in the direction of the thickness
of the shell.
In virtue of (2.1) and (2.2), from (1.6) we have
du' T _
O_ O. u_=- wi (2. _) = _v(_, _), (2.3)
that is, normal displacement ui of any point of any layer of the shell does not
T
depend on coordinate ),. Normal displacement of all points of a given normal
element have constant value and are equal to the normal displacement w = dv,_. _)
of that point of the coordinate surface of the shell, which point is formed upon
intersection of the given normal with the coordinate surface of the shell.
On the basis of (2.1) and (2.2), from (1.7) and (1.8), after the known
transformations**, for tangential displacements of the 8-th layer of the shall
we obtain
0w. (2.4)
ut = (1 -_-/h7) u -- _
tt I
Ow (2.5)
u_=(l+k27 )v-B O_'
*A. L. GoUdenveyzer, see footnote * on p. 21.
**V_ Z. Vlasov, see pp. 218-220 of work cited in footnote **'** on p. 19.
1964012887-033
24
where u = u (a. {}). v = v(a. _) are the tangential displacements of the corre-
sponding point of the coordinate surface of the envelope, that is, the displace-
ments of this point are in the directions of the positive tangents to the lines
of curvature of the coordinate surface _=const. and a = const, respectively.
Thus, formulas (2.3)-(2.5) establish the geometric model of the deformed
state of the shell.
On the basis of (2.3)-(2.5) deformations e:. e_ and e_, may be repre-
sented in the form of power series of variable % wherein, in virtue of the
fundamental hypothesis, we may limit ourselves tc the first two terms of the
expansions, namely,
' (2.6)
e. = _! Jr- _xl;
(2.7)
e_=__2+ _x2;
c_= _+ T_. (2.8)
Inserting the values of u:. u_ and u'; respectively, from (2.4), (2.5)
and (2.3) into expressions (1.4), (1.5), (1.6) and equating tbe results to (2.6),
(2.7) and (2.8), respectively, for the coefficients of expansions (2.6)-(2.8)
we obtain
I ou 1 0A (2.9)
l or, I on ' - (2 10)
n2-- B O_ -t- _---B_ u -I- R2w;
/
t=-Ba_ + AO,,\BI' (2.11)
,o(_ ) , . v k,+:
OW OA Ow Ok I Ok t
xa-- AOa Ta --AB 20,_ 0_ +_A q 0_ B (2.12)
1 O(l 0w) ) OBOw.t. Ol_tu.a Oktv ,,
(2.13)
2 [ O*w l OAOtv 1 OBOw'
-t-tk, h,)[A u b_(X) -8
which are the same as the corresponding coefficients of the expansions derived
in the case of a homogeneous isotr,_pic shell. *
*V. Z. Vlasov, see p. 221 of the work cited in footnote ***** on p. 19.
1964012887-034
25
The expansion coefficients _. _2, _. which are known as the components
of tangential deformation, represent the relative deformations of elongation and
shear of the coordinate surface of a shell. The expassion coefficients x_. z2. and
whl.ch are known as the changes in curvature of the coordinate surface of the
shell, represent the varia'" -_ in deformations e., e_, e,_ over the thickness of
the shell (Linearly). x_and _ describe the bending deformation, r the torsional
deformation.
_stead of expressions (2.12)-(2.14), most authors use the following
relationships for changes in curvature:
1 0(10. kl_) ! c)B[I Ow__k,u):
2 [ O_o 1 OAO_ ! OBOw]
(,o_, , oa ,_ (._0_, , o, ) + 2ka S 0_ A-B_ u] -'t- 2k2 O.. AB 0.. v , (2.17)
which, strictly speaking, cannot be _dentified with changes in curvature of the
coordinate surface of the shell, since in deriving relationships (2.15)-(2.17)
it is assumed that the tangential deformations of the central surface do not
correspondingly affect the changes in curvature of the coordinate surface of the
shell.
Analyzing the above derived variant_ of (2.12)-(2.14) and (2.15)-(2.17),
for evaluation of the change in curvatures it is not difficult to see* that they
differ only within the limits of accuracy of the initial assumptions of the theory
of thin-walled shells. For the purpose of simplification of calculations in each
special case in the present work we shall use either the first or second variant
of the relationships for the change in curvature.
SECTION 3. EQUATIONS OF CONTINUITY OF DEFORMATIONS
OF THE COORDINATE SURFACE
Among the six parameters_l, _2, w, ,_. x2, ,,which characterize deforma-
tion of the coordir_ate surface of a shell there are three differential relation-
ships which are valid for any values of displacements u, v and w.
*V.V. Novozhilov, see p. 27 of work cited in footnote ** en p. 21.
1964012887-035
26
Eliminating displacements u, v, _. from equations (2.9)-(2.11) and
(2.15)-(2.17_ and using the Gauss-Codazzi equations (1.2), we obtain*,**,***
three differential relationships in terms of the six components of deforma-
tion of the coordinate surface of the shell:
0,2 #B ,10_ OA __k20A
r 0,,_ c)A aE_ OB 1 4- k1[A_ + 3_ _ --B _- a,, (%--el)-_0; (3.1)
D'____OA B _: )B OB
[n 9'" OB OE, OA ] (3.2)
4,2z1____kl_.2.___-___BI o I [B dE2 " OB
A a,,, OA ] O , [ .0_,_4-
'2 0,_ 0_ +_-B- AO__-
04 B 0,o OB ]} (3.3)
+ -_-(E,--%) 2 0_ -&--w ----0.
The functions _,... _, s_ :isfying these equations characterize that de-
formed state of the shell in which the coordinate surface remains continuous,
without undergoing rupture. For this reason (3.1)-(3.3) are known as the con-
ditions of nondiscontinuity of the coordinate surface.
These conditions for a shell of arbitrary shape were first derived
by A. L. GoUdenveyzer.****
SECTION 4. STRESSES IN LAYERS
Using the fundamental hypothesis and disregarding stresses o_,we ob-
tain from the generalized Hooke's Law (1.13)
e' = fio!. + , t a' "
= r-, ._ %6%_' (4.I)
l
_ --a_ -_-a2_o _H-aeon.
*V. Z. Vlasov, see pp. 287-292 of work cited in footnote ***** on p. 19.
**V.V. Novozhilov, see pp. 28-29 of work cited in footnote ** on p. 21.
***A. L. Gol'denveyzer, see pp. t}5-59 of work cited in footnote * on p. 21.
**** A. L. Gol'denveyzer, Additions and corrections to Love's theory of
thin shells, Sb. Plastinki i obolochki (Symposium on plates and shells), Gos-
stroyizdat, 1939.
1964012887-036
27
Solving these _quations for the components of the stress tensor and
considering (2.6)-(2.8), we obtain*
' B_l=l + t i
B12_,2 + _6=+ 1 + q-
(B,_, B_,_ B_); (4.2)
OI =
-i'- T(S._,z_.,a_B.2x, -t- B2s_); (4.3)
_13 l
= s,6,, + s_,-.t-- 8'_=+ .f(s_, + B'_+ '_k,_). (4.4)
where for coefficients Bjk we have
Having the values of stresses _=, o_, and _,_,the remaining three com-
ponents of the stress tensor are easily determined from the eqt_ilibrium equa-
tions (I. i0)--(I. 12). Restricting ourselves to the general form, for the men-
tioned stresses we obtain
/[ OH, j
' ' _ _ (U_o:)-- U, +
_"= ._/ u, Wo,
,{/[ o.,,
4, =- ug-'-_, u, (u,4)- u, -_ o.+
o
+_-
. .(/[ o...
_- #Tui #= "+ H, _ o,--
1'
_.(",_.,)--l(H,_)]'_+_,(,. P)}. (_'_)
Here _ (=. _), St (=, _) and X_(=, _) are functions resulting from integration for
7. They may be determined from the conditions at the contact surfaces of
*S. G. Lekhnitskiy, Anizotropnyye plastinki, Gostekbizdat, 1957, pp.
243 -244.
1964012887-037
28
adjacent layers and at the external surfaces of the shell *,**
SECTION 5. CONDITIONS OF CONTACT OF ADJACENT LAYERS
Previously, in formulating the basic concepts, it was stated tilat the
1,'tyers of the shell function concurrently without slippage. In virtue of this
the stresses and displacements of individual layers at the bonded surface
must satisfy the following contact conditions (Figures 8 and 10):
When _ = _l
[Y
'Z Figure 10
In addition, the stresses of the extreme layers must also satisfy the
following ,_onditions at the outer surfaces of the shell:
When "r= --
I 1 _X-,
0T=: -- Z -, '_=T
, (5.3)
when _ --=(/=-- 4)
*S.G. Lekhnitskiy, Bending of heterogeneous, anisotropic, thin plates
of symmetrical structure, PMM, Vol V, No. 1, 1941.
**S, A. Ambartsumyan, Some fundamental equations ill the theory of a
laminar, thin shell, DAN ArmSSR, VIII, No. 5, 1948.
1964012887-038
29
ffl n _ t/I n ;
a_ _ Z , _1 _ X
v+, (5.4)
where X-. .... Z corresponding components of the force vectors of
the surface loads; 1 and (m + n) are the numbers of the outer layers (Figure
11).
(_ .... -dj tsm.,-,_J J'
Figure I1
In virtue of the fact that the hypothesis of nondeformable normals was
given for the entire stack of the shell, the contact conditions of (5.1) are auto-
matically satisfied. This may be seen from an e.-.amination of the initial
relationships (2.3)-(2.5).
SECTION 6. INTERNAL FORCES AND MOMENTS
Section 4 of the present chapter established the laws of change in
stresses directed along the thickness of the shell; however, as in the
theory of homogeneous isotropic shells, instead of the stresses it is con-
venient to introduce their statically equivalent internal forces and moments
acting on the areas of the principal normal cross-sections of the shell.
From the conditions of static equivalence for internal tangential
(T_. 7"2, S_2, $20 and transverse (N_. _J2)forces as well as for the bending (M_. /_)
and torsional (H_2, H2_) moments referred to a unit length of arc of the cor-
responding coordinate lines, we have (Figures 8, 11, 12, 13)
1964012887-039
30
'X f o:"_v r,='X f '
r_ : _ -X _H_ d'I;
, _-I (_,_.-_) s-t (_,_,-A) (6. I)
,,,., (_-A) m+,, (8,-A)
s,__,,)2 f ,._,_. s,,: X f 4,-,<; {6.2)
.,., (,-A) ,,,+. (_.-A)
, f , y., f
N1 = -B X '_,qH._ d"T, N2 -= '_,H, a'_; (6.3)
s=l (_,_,-_) ,-1 (S,_I-A)
m,n (_s -p') m+n (_$-a)
l f, , f (6.4)
M, : _ _ a='{H,d"l, A'i,=-_- _, a_'iH, d'_;
,=1 (a,_.-a) _.1 (__,-_)
m+n {_$-fl) ra+, (85-.A) (6.5)
H_2= -_ _,_'IH2a'_. Ha = -X
Here (m + n) is the number of layers of the shell, m is the number of layers
below the coordinate surface of the shell, n is the number of remaining layers.
(If the coordinate surface of the shell is located within any layer, n is under-
stood to be the number of layers above the coordinate surface plus one, and if
the coordinate surface of the shell coincides with any surface of contact, n
is understood to be the number of layers above the coordinate surface).
Ir
I
t._g.
Figure 12
Tangential forces T1, T 2 and S12 , $21 are, respectively, the normal
and shear forces. The normal forces are considered positive if they are
tensile forces. The shear forces are considered positive if, being applied to
areas with positive external normals, they act in the directions of the cor-
responding positive tangents to the coordinate hnes a ==const, _ = coast. Trans-
verse forces N1 and N2 which, upon being applied to areas with positive
1964012887-040
31
3xterr.al normals, are directec' alr'lg the positive normal to the coordinate
surface are assumed to be posi_.ive. The positive directions of internal
forces are shown in Figure 12.
lY
I
Fi_,__lre 13
Bending moments M1 and M2 are positive if they increase the curva-
ture of the coordinate surface of a shell of poMtive Gaussian c,_rvature. The
torsional moments H12 and H21 are positive if the vectors of these moments,
when laid off according to the-right hand screw rule, have directions opposed
to the directions of the tangents to coordinate lines _=const. and a _-const.
respectively. The positive directions of the moments are shown in Figure 13.
Replacing the stresses with their statically equivalent stresses,
henceforth, instead of an a.'bitrary three-dimensional element of the shell,
we may examine the corresponding two--dimensional element of the coordin-
ate surface of a shell under the action of internal forces and moments. Herein
the examined element of the coordinate surface of the shell will be considered
to possess the derived physical-mechanical characteristics of the correspond-
ing three--dimensional element of the shell.
SECTION 7. EQUILIBRIUM EQUATIONS
In Section 1 of the present chapter we presented the equilibrium equa-
tions for a volumetric element of the i-th layer of a shell; however, in sub-
sequent discussion, as in the case of a homogeneous isotropic shell, we
shall also be interested in the integral equilibrium conditions of the shell.
Using four sections normal to the coordinate surface of the shell we
distinguis_ a small curvilinear rectangle whose sides coL:cide with the coordin-
ate lines_= const, and _= const.
The indicatedshellelement willbe acted upon by internalforces in the
form of stressesactingon the edges of the shell,externalforces inthe form
ofvolumetric forces appliedtoeach layer of the discussed shellelement, and
surfaceforces appliedto thecorresponding externaland internalsurfaces
delimitingthe discussed shellelement.
1964012887-041
32
As was stated in the previous section, all the internal forces may be
reduced to their statically equivalent forces and moments which v4ll be ap-
plied to _he contour lines of the corresponding elem_',_ of the coordinate
surface of the shell. In similar fashion, the external lo_c_s may also be
reduced to their statically equivalent forces applied to the surface of the
corresponding element of the coordinate surface of the shell. ,_ioreover,
as usual, we shall disregard moments which appear upon achieving the indi-
cated process of reduction.
In the case of multi-layer shells, for the given external forces applied
to the coordinate surface of the sLell we have
T,j -_j+x- --_,,(,---_)
m+n (% -*)
; (7.1)
_.1 (%_,.-A)
z /
_+,, (%-':)
; %,-_ 1"
+ --_ z__ , _H, H2d.r: (7.2,)
IJ
_-I (_ _,-_)
.,+. (h-A)
5 J"
s-I (_j_i-_)
Thus, the problem of equilibrium of a small element of the shell has been re-
duced to the problem of equilibrium of the corresponding small element of
the coordinate surface of the shell under the action of i_.ternal (6.1)-(6.5)
and external (7.1)-(7.3) forces (Figurel4).
/
S.
I,e
N,
,%
T,.. s.... s_,
Figure 14
1964012887-042
33
Henceforth, without further reference thereto, within .he accuraoy of
the fundamental hypothesis in expressions (7.1)-(7.3), we sball disregard
quantities of the order of h/Rj in comparison with unity.
The equilibrium equations _f an anisotropic laminar shell, as is known,
cannot differ from the corresuonding equations for a homogeneous isotropic
shell. In this connection, without going into a detailed derlvation, we present
the final expressions for the equilibrium equations
OBTI OB 0AS2, OA
O_ 0,, T2 _-_-r--_ SI2"+- ABkINI : - AB)(;
OAT 2 OA . 0BS,2 OB S
0'_ d_ TI -_'_-+--0-_- 21-_- ABk_N2 =--. ABY"
1 I OPN t OANs_
- (k_r_+ _r_) +--_ t-o-;--+ --o3--i = -- z;
(7.4)
OBM_ OAH v OA OB
da -{- ----O_ -= --t- _ Ht2 -- _ M2 = ABNt;
OAM_ OBHt, OB /./21 OA
H,2 H2, = O.
Sl2 - S,214 R, R,,
The derivation of equations (7.4) may be found in tlm monographs by A. L.
Gol'denveyzer* and V.V. Novozhilov. **
The sixth equilibritLm equation
n,, "_" =- 0 U. 5)
$12-- S214 _, Rz
is an identity. This is easily seen by inserting into (7.5) the values of Sik and
Hik from (6.2) and (6.5) respectively.
SECTION 8. POTENTIAL ENERGY OF DEFORMATION
By virtueofthe fundamental hypothesis,from (1.3.3):._d(1.... 4) we
obtainfor the potentialenergy of deformation of a multi-layershell
*A. L. Gol'denveyzer, see pp. 33-36 of work cited imfootnote * on
p. 2l,
**V. V. Novozhilov, see p_. _4-38 of work cited in footnote ** on p. 21.
1964012887-043
34
v=_L.1ff|--L _ . (o:e:+ o_4
(_,- ;-_)
3
qLx_e_'_), (i Jr- kiT) (1 -Jr-k2T) dT] AB da d_.
(8. 1)
J
Inserting the values of stresses from (4.2)-(4.4) and of deformations from
(2.6)-(2.8) it)to (8.1) and integrating with respect to variable y, we obtain
the following expression for the ootential energy of deformation of an aniso-
tropic laminar shell:
! l f C 2
= . " E ,'_._m2 V _ ./ . IC,;-'_+ 2C;.:., 2_- "'.,2%+ _-
f
I i c L
'2-C26_s21
AB
d2d3
._L 2Ci6tu.i-4- Jr-,-- ,1 iKnel'/I _-
.4- /('12 (_l't2 _- E2Xl) -_- K22"c2x2 + K{_6'o= -F
"/_lfi(_'l _i _ tU'_I) _1--11"26(E27 -F_%)l ABdad_ F
,f/
+ ff [Duxi + 21)l:,hx. + D,.,',_- D_Y: 4-
Jr 2D_6xx_-Jr 2D26:%]AB da d_. (8.2)
where coefficients Cjk, Kjk and Djk have the following form:
tt/,h
(-/k = ,_ Bjk(_, -- _, ,); (8.3)
PTIII
; X" 1(_ o, o,
h'/, _ _B,_ - _)- 25(_,-- ,)]; (8.4)
.:1%_"_, ;, i(?_: 5: i) 3A(_: o_ ,1 _-3A;(% _,. t)] (8.5)
/);_ J/# - - -- ' " -
, I
In the special case in which the coordinate surface of the shell coin
c_des with the lower "-urface of the shell, A becomes zero (Figures 8 and 11),
in conjunction with which the expressions for Cjk, Kjk and Djk are consider-
If
ably simplified. Assuming in formulas (8.3)-(8.5) that _ = 0, we obtain*
m4tl
c_ :; _ _, (_,--_,_,); (8.6)
$-I
*S.A. Amo,,rtsumyan, see footnote *** p, _9.
)
1964012887-044
35
1
(8.7)
$=|
ffl_.R
(8.8)
-- -- o,_,).
,Tffil
In all three components of formul_ (8.2) both integr_.ls extend over the
entire coordinate surface of the shelf. In (8.2) the first component is the
potential energy of elongation _md shear, the third component is the potential
energy of bending and torsion and, finally, the s_ ond component is the poten-
tial energy which the shell acquires due to interaction of bending, torsion,
elongation and shear.
By analogy with the theory uf bending of isotropic shells and anisotropic
plates for coefficients Cjk, Kjk Djk let us introduce the concept of stiffness.
Let Cll and C22 be the tension and compression stiffnesses along coordinate
lines _ = const.and _-_ const: let C66 be the sl-._.r stiffness in the plane tangent
to the coordinate stiffness, characterizing the influence of elongation along
the coordinate lines _ _ const, and _ _ const on shear in the tangential plane.
Let Dll and D22 represent the ben_ing stiffnesses about the axes tangent
to _he coordinate lines a _ const, and_ = const;let D66 represent the torsional
stiffness; let D16 and D26 represent the _econdary stiffnesses of bending and
tGrsloh characterizing the infiuence of bending about the ax_s tangent to
c_rdinate lines _ = const, and _ _ con_t, on torsion. Let Kll and K12 represent
the stiffnesses of interaction of tension and bending characterizing the im'er -
action of tension and bending along lines _ _ const, and _ :_ const; let K66 represent
the stiffness of interaction of torsion and shear; let K16 and K66 represent
the sec_d_.ry stiffnesses of interaction characterizing the mutual effect of
bending on shear and tension on torsion.
Formula (8.2) did not take into account those components of the potential
energy of deformation which in comparison with the other co_.ponents were of
the orderof h/R.
SECq'ION 9. _LASTICITY RELATIONSHIPS
In th_ theory of s_e-tsthe equationsestablishingthe reiationshipbe-
tween internalstresse_ _,d deformations of _.hecoordinatesurf_,ceof a
1964012887-045
36
she_i are known a_ elasticity relationships. These relationships may t,_.
derived by elementary means. Inserting the values of stresses from (4.2)-(4.4)
into formulas (6.1)-(6.5) and integrating, thereby retaining only the lowest
powers of _, (the zero and first powers), we obtain the simplest elasticity
relationships for the general case of a multi-layer anisotropie _.hell
1"1----- Cn_1J,-Cl2_2Jr C16_+Knx ! Jr-K12_-I-K_6_; (9. l;
7"2=- Cr2_+ Cl2EI+ C2__ -F Kr;'_ + Kl2xl + K26"q (9.2)
312: C_u)Jr"CI6EIJr-C26_-2 Jr-K_ -t- Kl6Xl-JrK26x2; (9.3)
S_l= C_ + Cl6_I + C_%+ K_ -{- KlexsJr K_; (9.4)
M, = DI,xI + D,2_2-_-DI_ +Kl[e I + Kl2_ -{- K_; (9.5)
_ = D._. _ + D_x_Jr D_ + K_a_+ K_ + K_; (9.6)
H_ = D6_'- t D_x_4- D_'_--{- Kr_) -{- K_ -4-K_; (9.7)
H2_: Dr_ + D_x_-d-D_ -{- K_ Jr-K_s _LK_%, (9.8)
where for stiffnesses C.]k, Kjk and Djk we have the known expressions (8.3)-
(8.8).
Here and :'n subsequent text t'ae formulas for determining N1 and N2 will
not be derived, since in the presence of the well-known elastici_ relationships
for Mj and Hjk the transverse forces N1 and N2 may be derived from the fourth
and fifth equilibrium equations of (7.4).
In examining the elasticity relationships of (9.3), (9.4), (9.7), (9.8)
and equation (7.5) it is not diff, cult to see that between them there is an inter-
nal contradiction, namely: the mentioned relationships do not satisfy the
sixth equilibrium equation, that is, the sixth equation is not an identity.*
I_awever, we n_ay derive other sufficiently simple elasticity relation-
ships which do not contain the above-mentioned contradiction. In obtaining
_ relationships (9.1)-(9. _) we have everywhere disregarded quantities of
. the order of kin in comparison with unity. However, if we deviate from strict
-,
: logic and in all formulas in (6. I)-(6. _), w'_th the exception of (6.2), neglect
._ quantities of the order of kjh in comparison with unity, while in formulas
4,
(6.2) we leave quantities of this same order of magnitude, we obtain new
*A. L. GoUdenveyzer, Appltcability of general theorems of elasticity
theory to thin shells, PMM, Vot. VIII, No. 1, 1944.
_'
1964012887-046
37
elasticity relationships which identically satisfy the sixth equilibrium equation
T_----Cnh -I-"C12_2-+ C16_+Knx, q'- K,2h 4- K,r._; (9.9)
7"2= Cmt2 -Jr-C12_1 -Jr-C26_Jr-K_x2-_ Kl,,xl -Jr-K2(t; (9.10)
S, _.--- C_t_ + Cl6_1+ C2d-2+ K_': + Kl6xI + K_x2+
+/._(K_o + Kl6-:l + K_6_2+ Dc_ + D16xl+ D2_x2); (9.11)
S2I -- Cf_o"iA-C,6=_: -_C26E2,-_.-K66'c-FKI6xI + K26x2 +
+ kl (K66m+ Kl6Sl Jr-K26=2+ D66'_+ Dl6xl -[--D26x2); (9.12)
M, - D]Ix,Jr-D,2x2+ D,n:Jr-Kn_l-'-K_2-:2 -Jr- K,6_: (9.13)
342= D_.2x2 Jr- D,2x, Jr-D26": -_-K_s2Ji-Kl2SlJr K2@); (9.14)
tt,a_= D_ -Jr-D_6x] Jr-D2__2-F K6_o-JrKI6_-! Jr- K2@2, (9.16)
where, as before, for Cjk, Kjk and Djk we have formulas (8.3)-(8.8).
Elasticity relationships (9.9)-(9.16) may be obtained if we use the ex-
pression for potential energy (8.2) and apply the method proposed by V.V.
Novozhilov* for the case of a homogeneous isotropic sb,'ll; this is not derived
here, since the resulting elasticity relationships are in complete agreement with
relationships (9.9)-(9.16).
By adopting a purely formal approach to the problem it would be possi-
ble _o achieve further refinement of the elasticity relationships ;** however, we
shall not do this here, since, in accordance witl_ the well-founded remark of
A.L. Gol'denveyzer,*** an attempt to refine, the elasticity relatior_ships
without proceeding beyond the bounds of the hypothesis of nondeformable nor-
mals has no chance of success and hence the best of all of the approximate
variants of the elasticity relationships must be considered to be t_at which
does not contain formal contradictions and leads to the simplest calculations.
Finally, we vail note that, as in the theory of homogeneous isotropic
shells, there is not and, regrettably, cannot be a single approach to the
problem of choosing one or another elasticity relationship. In this connec-
tion, in subsequent discussion, in showing a certain preference for relation-
ships (9.9)-(9.16) containing no formal contradictions, we sh_ll, wherever
possible, find wide use for the simplest relationships in (9.9)-(9, 8).
*V.V. Novozhilov, see pp. 45-48 of work cited in footnote ** p. 21.
**A.I. Lur'ye, General theory of elastic thin shells, PMM, Vol. IV, No.
2, 1940.
***A. L. GoPdenveyzer, see p. 71 of work cited in footnote * p. 21.
1964012887-047
J
38
SECTION 10. BOUNDARY CONDITIONS
As is known, the arbitrary functions contained in the general integral
of differential equations in the theory of shells must b3 determined from the
boundary, conditions. The boundary conditions of an anisotropic laminar
shell, when the hypothesis of nondeformable normals is given for the entire
stack of the shell as a whole, do not differ from the corresponding boundary
conditions given for a homogeneous isotropic shell, given the hypothesis of
nondeformable normals.
In actual designs we may encounter extremely varied types of shell
supports and this great variety of structural solutions for the supports of
real shells, regrettably, cannot be represented in the form of mathematical
models -- boundary conditions. In this connection we shall here present only
a few of the possible variants of boundary conditions.
In subsequent discussion we shall be interested in those sh_iis which
have an edge or edges coinciding with the lines of curvature of the coordinate
surface of the shell.
For the sake of brevity, the boundary conditions will be given only
for the edge defined by coordinate line _ =-%----const.
I. Homogeneous boundary conditions:
a) Free edge
Hi2 1 OHm_
T2----O' S_2--t---_-_---0' N'+ B O_--0, M_=Q; (10.1) (10.1)
b) Hinged edge
._4_0, :_=0, v--:0. _o=0; (10.2)
c) Hinged edge free in the tangential direction
T_=_0, .:Iz=0, -_o=0, v=0, (10.3)
or
h',, (10.4)
d) Hinged edge _tree in the normal direction
" 1 (6.2)
A = CI_-DII -- ho '
where a and b are c, rtain constants depending on the elastic properties of the
layer and on the ratios of the thicknesses of the individual layers to the half-
thickness of the entire shell (h =- 2h0). It is evident that these constants have the
dimension of Bj_ (see (3.1) and (3.2)). Introducing the notation
k2= A. k = /--ff ho'/', (6.3)
the integrals of equation (6.3) will be sought in the form
a = _ (s, k) ek/(s), (6.4)
where _(s. k) is a function of intensity and/(s) ls a function of variability***.
The intensity function may be represented in the form an asymptotic
series
_,s) _ -I- =/:0). (6.5)
4_ _, (s) + --g- + + . . (%
(s. k)_%(s)_ k "'" ,
*Ye. F. Burmistrov, Symmetrical deformation of a near-cylindrical
shell, PMM, VoI. XIII, No. 4, 1949.
**A. L. Gol'denveyzer, see pp. 282-326 of work cited in footnote *** on p. 123.
***L. L. Gol'denveyzer, see p. 287 of work cited in footnote ***, on p. 123.
1964012887-137
126
in which :,,. :, ..... %.... do not depend on k. Inserting the value of cr from
(6.4) into (6.1), considering {6.3) and (6.5) and canceling e_', we obtain,
, l
V I [(,;I_" i X. I fd!f_ ._,tf.l_ ._.,_d1,,l +
..... o (6.6)
]
_,r { ds_ -- r- d,_ -- t, r2- % =0.
n=lJ
Requiring that the coefficients in (6.6) for all powers of k vanish, we
obtain an infinite system of recursion formulas for determination of 1. %. _ .....
(d! t
,,,) + -_ = o:
- as 9-( -_ = O;
' _,1,_2 -r ds 1%
o d.f d: n [d__2f___m_ df', (6.7)
__ [d2:n_t _ina d_n-i ) .,m2_ ) \ da i as "--r _-- an-I
(n=l. 2. 3 .... ).
Limiting ourselves to the first approximation of asymptotic integration (that is,
examining only those equations which are obtained when the coefficients of k 2
and k vanish), we obtain
; (6. s)
(a(/"
ds I -- R2 ;
2 _tsds "d_ 7 cls/% = O. (6.9)
1" m (6.8) for the variability function we find
(1- i)_/ ds (6 10)
1
I"_= -+- iG __V_"
$n
Inserting the values of/(s) from (6.10) into (6.9), In order to determine
% (s) we obtain the following differential first-order equation:
d% sin 0 { 1 A?2
i_s :17 k -Jr R;) ao= O. (6.11)
1964012887-138
127
From (6.11) for % we have
%_ ff,"r-"_. (6.12)
Inserting the values of _,, and f from (6.12) and (6.10), respectively, into (6.4)
and considering (6.3), we obtain for a
" _/_ [_e -II-i' :;-F:C',e-'):_I. (6.13}
o-- _rr
where
3 __ (_4" f ds (6.14,
'-- VR;:
_n
(; and .'2._are arbitrary complex constants.
Inserting the value of r in terms of C2 into (6.13), we finally obtain for r
- ' + (6.15)
_R2 cos2,_
The solution (6.15) derived in this manner is the first approximation of
the general solution of differential equation (6.1).
Examining the procedure for obtaining solution (6.15), it is easily seen
that with the same accuracy the solution may be obtained directly from the initial
homogeneous equations (3.16) and (3.17) if we disregard therein the third and
tourth terms (that is, those terms which contain ratios C_/C_ and D_'D,,. By
virtue of this it may be asserted that the limiting assumption in (5.1) loses its
force; that is, if we limit ourselves to the accuracy of the first approximation
of the asymptotic integration, condition (5.1) wil! not have any effect on the sub-
sequent calculations for a shell of revolution in the general case of symmetrical
lamination and orthotropic anisotropy.
By elementary reasoning it may be shown that the first approximation in as-
ymptotic integration of the equations of solution of s.vmmetrically loaded orthotropie
saells of revolution has an error of the order of l/h/-Rjin comparison with unity*, **.
*V. V. Novozhilov, see pp. 255-267 of work cited in footnote ** on
p. 123.
**A. I. Lur'ye, see pp. 147-148 of work cited in footnote on p. 111.
1964012887-139
128
In the present work we shall limit ourselves only to the first approximation,
since ._or purposes of engineering calculatior, its accuracy is quite adequate*.
With these remarks in mind, it is easily sho_n that with at_ accuracy of
] . upon differentia-
the first approximation the factor in formula {6.15), _/R, cos2
tion, behaves as a constant. Then, including the mentioned factor in the inte-
gration constants, we obtain from (6.15)
a-_C,e-O-O__4__2e(,-O _. (6.16)
Considering that
C,-_-E, +IF,; C2=:E2q--IF2;
e-('-0._" = e-,_(cos 0+ l sin _); e('-O ___ e_ (cos _ -- t sin_).
we finally obtain for
o -_-(E, cos _ -- F, sin _) e- _-+-(E2cos _ -t---/",sin 3) e,_--}-
-_I[(Elsin_+ FIcos6)e-_--(E2sin_--F_cos_)#I. (6.17)
Since the general solutionofthe equationof solution(5.9)is the sum of
solutions(5.14)and (6.17),on the basis of (5.10)we obtain
$V-- IA -_,'-V = (E, cos _ -- Ft sin _)e-,' -+-(Ei cos _ -[- F, sin _)e_+
+ 1 [(E l sin _ -_- F t cos 6) e-_ -- (E 2 ._in _ -- F 2 cos _) e_l -+-
___tC,,i_ (s_) (618)
_2 "" COS _}
Hence, separatingthe imaginary and real parts, we obtain
W -_- (El cos _ -- Ft sin _) e-_ + (E_ cos _ + F_sin _)e_. (6.19)
V ---_-- _ [(El sin_ -}- Ft cos 6) e-B--
-- (E2sin _-- F2cos B)a_I t-, (s) (6.20)
" COS_' "
*If desired, proceeding from the system of equations in (3.16) and (3.17),
we may obtain the second approximation, the accuracy of which will not be less
than that of the initial system of equations.
1964012887-140
129
Following A. I. Lur'ye*, *.he solution of equation (6.1) may be repre-
sented with the aid of
a = (C1-- lC3)(91 -- 219_),_-(C2 -- tC4)(91 -- 2/94). (6.21)
whereC_ ... C4are arbitrary constants, 9](_) are A. N. Krylov's functions (See
Table 1), which are linear combinations of e-O -o_ and e(!-o_. From (6.21) and
(5.14), on the basis of (5.10), we obtain
W --- lA-_ V = (C__ lCs)(o _-- 2lO.s)Jr- (6.22)
+ (C2-- :C,) (92-- 2:24) + t _ A
fl
ccs0 "
hence, by analogy with (6.19) .',_',.d(6.20), we have
W -- Cz9zJr C2_2_ 2Cs93-- 2C494; (6.23
V = ADn (Cs91-F C492-Jr"2Czga-_- 2C2Q4) fz (S) (6.24)
COS 0 '
Table 1
A. N. Krylovts _unctions
dgil d*2 k da121 d'O k
] ch _cos _ -- 4_4 -- 4(2s -- 49z -- 4_-)-1
2 2"1 (oh (_ sin _ -t- sh _ cos _) I ....4t)4 4QS 4'.t2
l
3 -_-shp sin p _2z ._ -- 4o. _ 4_.__3
1
4 _- (ch I_sSnp-- sh _cos [_) o_ _ _, -- 4_4
The solution of equation (5.9) may be presented in slightly different
form**. Let us introduce the new variable (Figures 18 and 28)
t/f f"f
= _ _--_, (6.25)
$_ $o
*A. I. Lur_ye, see pp. 68-69 of work cited in footnote on p. 111.
**A. I. Lur'ye, see pp. 59-61 of work cited in footnote on p. 111.
1964012887-141
130
considering (6.14) and introducing for the constant the notation
$;
Su
Table 2
0 1,0000 1,0000 1,0000 0
O,1 0,9_07 0,81O0 0,9(X)3 0,0903
0,2 0,9651 0,6398 0.8024 O,1627
0.3 0,9267 O,4888 0,7077 0,2189
0,4 0.8784 0,3,564 0,6174 0,2610
0,5 0,8231 0,2415 0,5323 0,2908
0,6 0,7628 0.1431 0,4530 0,3099
0,7 0,6997 0,0599 0,3798 0,3199
0,8 0,6354 --0,(X)93 0,3131 0,3223
0,9 0,5712 ---0,0657 0,2527 0,3185
1,0 0,5083 ---0,1108 0,1988 0,3096
I,I 0,4467 ---0,1457 0,1510 0,2967
1.2 0,3899 ---0.1715 0,1091 0,2807
1.3 0,3355 ---0,1807 0.0729 0,2626
1,4 0,2849 ----0,2911 0,0419 0,2430
1,5 0.2384 ---0,2068 0,0158 0.2226
1,6 O,1959 ---0,2677 ---0.0059 0,2018
1,7 0,1576 ---0.2047 --0,0235 0,i812
1,8 0,1234 --0,1985 --0,0576 0,1610
1,9 0,0932 --0,1899 --0,0484 O,1415
2,0 0.0667 ---0.1794 --0,0563 f 1230
2,1 0,0439 --0,1675 --0.0618 O,1057
2,2 0,0244 --0,I 548 --0,0652 0,0895
2,3 0,0880 ---0,1416 ---0,0508 0,0748
2,4 ---0,0(0_ ---0,1282 ---0.0569 0,0613
2,5 --0,0166 --0. ! 149 --0,0858 0,0492
2,6 --0,0254 ---0,1019 --0,0536 0,0383
2 7 --0,0320 --0.0895 ---0,05_ 0,0287
2,8 --0,0369 --0,0777 ---0.0573 0,0204
2,9 --0,0403 ----00666 --0,0534 0,0132
3,0 ----0,0423 --0,0563 --0,0493 0,0871
3, I --0,0431 --0.0469 --0,0450 0,0019
3.2 ---0,0431 --0,0383 -0,0407 -0,0024 i
3,3 -0,0422 ---0,0306 -0,0364 -0,0058
3,4 -0.0408 --0.0237 --0,0323 ---0,0085 I
3,5 --0,0389 --0,0177 -0,0283 -0,0106
3,6 -0,0366 --0.0124 -0.0245 -0,0121 I
3,7 --.-0,0341 --0. 0079 -0.0210 ---0,0131
3,8 -0,0314 --0,0040 -- 0,0177 ---0,0137
3,9 -0,0286 -0,0008 --0.0147 ---0,0140
4,0 ---0,0258 0,0019 -0,0120 ---0,0139
4,1 -0,0204 0,0057 -0,0074 -0.013 t
4,2 -0,0'204 0,0057 --0,0074 ---0,0131
4 3 -0,0179 0,0070 -0,0054 --0,0125
s,4 -0,0155 0,0879 -0,0038 --0,0117
4,5 -0,0132 0,0085 --0,0523 --0,0108
4,6 -0,0111 0,0089 --0,0811 --0,0100
4,7 --0,0092 0,0090 ---4),0001 ----0,0091
4,8 --0,0875 0.0089 0,0007 --0,0982
4,9 -0,0059 0,0087 0,0014 --0,0073
I
L
!
Note: Comma represents decimal point.
1964012887-142
131
Table 2 Continued
5.0 --O,(D4(i 0.00hi4 0.0019 --0,(}(Y6.5
5, I --0,0033 0.O08U 0,0023 --0.0057
5,2 --0,0023 0/_)75 0,G02(_ --0.0049
5,3 --0,0014 0,0069 0,0023 ---0,0042
5,4 ---0,0005 0,_1o4 0,(X)29 ---0,0035
5,5 O.O(X_ 0.0058 0,0029 --0,0029
5,6 0,0005 0,0052 0,0029 --0,0029
5,7 0,0010 0,0046 0,0028 --0,0018
5,_ 0,0013 0,(X)41 0,0027 --(}.0014
5,9 0,0015 0,0036 0.002fi --0,00l 0
6,0 0,0013 0,003| 0,00'24 --0.0007
Note: Comma represents &ecimal point.
for the independent variable we obtain
_ %-- B_. (b. 27)
Inserting the value of f] from (6.27) into (6.19), we ob 'tain
W _- (El cos _ -- Fl sin _) e--_-+-I(E_cos %+ F_ sin%) cos _14-
r
-_- _E2sin % -- F_cos %) sm _l] e'_- ,.
Introducing the new constants
A_._---El: A2_ (E2cos % _t..F2sin :Zo) e_.; ]
BI -_- -- fl; B2--- (E2sin ao_ F2cos ao) e_, j _6.28)
for the sought W we obtain
W .-_ Ale-_ cos _-_- Ble-$ sin _-b A,2e-_cos _l -_- B2e-_' sin )l. (6.29)
Similarly, from (6.20), for the sought V we obtain
V -_ -- _ (Ale-" sin Ble-_ cos _ -+-A,_e-_, sin _1
(6.3O)
--B_t-_,cos_,) --f_(s_____.}
COS}'
For the tabulated functions let us introduce the following notation:
0 ([,_) .-_ e-_ cos _, C,O) ---e-';_ sin,'_, |
(6.31)
q,O) = oCD--r.O).f
1964012887-143
132
Table 2 liststhe values of these functionsfor values of arguments 0..<__ (i (The
tablewas taken from A. I. Lur'ye. See his frequentlycitedwork.) Between
these functionsthere existsthe followingrelationships:
d_(_)=__(_), d
,_). _-_:_) = _,(_}.
(6.32)
d_,- O)= 2'-_,_). -- :.(_)----- -- 20{_). d[_2
Considering (6.31), from (6.29) and (6.60) we finally obtain for the
sought flmctions
IV= A,0 (_)-JrB,r(_)4- A20 (_,)_ B2_ (_i); (6.33)
O
V _ A-C_, [- A[.(_)4-R,O(_)--.A2_ (_,) 4--B_0(_,)] --/'2_os(S)_. (6.34)
SECTION 7. INTERNAL FORCES, MOMENTS,
STRESSES AND DISPLACEMENTS
Inserting the value of the homogeneous part of V from (6.24) into homo-
geneous formulas (2.1) and adding the corresponding particular solutions (5.15)
and (5.1S), we obtain for internal forces T 1 and T2
sln@
Tl -= --ADn [Cagl qLC492 4- 2C1_3 Jr- 2C241 -_- Jr-
, {p:__ ./;' ) (7.1) F R.,cos2_ _2= rk.', ds ;
So
yrA,
T. _= D,, 2_ 1C4' -L 2C,O 2 -Jr- 2(:29_-- 4C3941--
R_cos2_) _2_---. rE, ds _ R2Z" (7.2)
$,p
inserting the value of V from (6.24) into (2.2), we obtain for the lateral
force
c,,,__ (7.3)
N='4DnICa_21 "_C4_2 4-2CI9_Jr-2C2_1] r
Inserting the value of W front (6.23) into (3.14) and (3.15), and disre-
Vrr
garding the terms of the order of #7' we obtain for the moments
1964012887-144
133
V/--.A-
Ml = Dt_ _ 1-- C2O_-,- 2CaO2_:- 2C4,_3_ 4Cr941; (7.4)
M2= D12 _ [-- C291-I- 2C3Q2-Jr2C4_a+- 4C1941-+-
sin 8
"Jr"D_ IC1.(21 --.}- C292-- 2Ca93-- 2C49.11 "7" (7.5)
From (3.8) and (3.9), for the normal stresses we have
1
, _, dV _2{VSln__ _j) ,, _/W (7.6)
J_ -- _-" d,_ @ _2\-- / T B] -j; -- B_2-7- W . (7.7)
Co.nsidering (1.16), (1.19); (3.6) and (3.7), from (3.10) and (3.11), with
the accuracy of calculations adopted here, we obtain for the shearing stresser,
in each layer of the shell:
for the outer layers (t = m + t, m + 2..... 2m Jr 1)
-t=_:,. . :_, ,l_W C_lB_ "'_-'
_/= X _Jr- _ _,_ -E_ _ :: X
["_: "-' c_ d"v h_ . d2_"
X \B-_-_._-? ) _,,,2 2 P_':-'-- 4- (7.8)
12
.+ hoCtlRi r_ 4/R ='m+2-' Ct_I cl2V
for the interna; layer (t--= 1, 2. 3..... m Jr- 1)
_,_ d2U/' CHB'I_, B'_ C_'\ a_V
Inserting the values of _ and _ from (3.6) and (3.7) into formulas (2.26),
we obtain for the two components of displacement
et .... _'v d-_<l,,i -t'---_,!I,"sin it.. -- -=.'C!' /=,(s); {7.10)
r '$
' : f. [ C,= O___._vC.{v=,., _-9)_]ds.
rz -:- e Jr- . [["sinI) -- -_ cos ds [_-_-- Ra" RI ]J
(7.11)
i The components of displacement may be represented in slightly differ-
i ent form. From (3.4) for the deformatlous _e have
I
I
1964012887-145
134
C_2T, -- CtL7"z . C_,T2- C,2T,
_'- L, . _ = ,, (7.12)
Inserting the values of _, and E_from (7.12) into (2.26), for the displacements
we obtain
c,,r_- c,2r, . (7.13)
er _ t.2 '
2
o J ( C22T,--CI,T, )
e, -= e, + Wsint, -t ,.2, cos 0 as. (7.14)
The values of functions V and W in design formulas (7.6)-(7.14) are not
represented here due to unwieldiness of the formulas derived thereform. In
order to facilitate calculations associated with the indicated design quantities
let us present the final expressions for the required derivatives of the sought
functions V and W
( / )
R, cos 2_ _- -- rE. ds -4--RLZ;
$o
cl2V Di I - _
ds_ -- 2t?--2 (2C,_, + 2C292-- 4C3_-- 4C,.q,t);
d--;-- --= _ (C29_-- 2C39-_ -- 2C'49a-- 4C'94):
d2W / --A o _
d3 2 -- -- _ (_{'3Q1 --{-- 2C4Q 2 _- 2C1Q 3 -_- 4C294).
d
In deriving these formulas terms of the order of _ were discarded in com-
parison with unity.
Proceeding from (6.33) and (6.34), then following the procedure given
at the beginning of this section, for the design quantities of the problem we
obtain the following formulas:
sin 8
rt = ADn IA,_.(t_)-- e_0(_) +- A2r. (_) -- B20(_])!--7- "+
, [p_o_ / ) (7.15)
$o
/,/ )
R, cos, _ _,2,_ . :E, as -R2z: (7.16)
$o
1964012887-146
135
cos___., (7.17)
N -_ ADI_[Bl0 (_) -- Al_ (_) -+-B20(_l) -- "12_ (_1)! r
V/--_ ,,,
M, ----D,, ___2 IA ._(_q) _ B,.. (_) ._ 4,.o(_,) _FBi_,(,3t)l; (7.18)
./'7,- ,.. _/_ :,( ) j.
---- [A,_(_) -Bf_tP)--A2"-?(_l) , 2', _ ]
...... sin 0
-F D_2IA_0(_) %-B_5(_) -F A20(_) -_- v2- t?_)l --F-" (7.19)
The formulas for determining the stresses and displacements (7.6)-
(7.14)remain unchanged. As before, letus present the finalexpressions for
therequiredderivafivesof the sought functionsV and W:
dV ___-- Dn ,1/- -_-
d_ _;,l-- &q' (_) -- B,_ (_) + A2_(_,_+ 8_'_@)l --
] -- rE zds 4- R2Z;
RI cos_
d_V _
d_, --c,,,_, [4_o@)+ B_r.(_) F A_0(_,) P B.,'.. (?_)1;
as -- | Z_IA,_(,_)-- B_,_(_)-- A_',_(_,)-FB_,_(,_,)I;
d_W ,4
_ = 7q_l/x:, (_) -- 8_o(_ A_ (_) -- B_0(_:)I
/--L-
In deriving these formulas we have discarded terms of the order of V_in
comparison with unity.
We will note that we cannot expect that from the design formulas de-
rived here all points of the shell will have an accuracy of the first approxirna
tion of as_unptotic integration. At singular points of equation (6.1) the design
formulas may yield unacceptable results. The radii of curvature R1 and R2
of the discussed shells must be sufficiently smooth functions of coordinata s.
In the special case of a cylindrical shell, wherein I/R 1 = 0, R2 = R =
const and :_= 0, the derived design formulas in the first approximation of
asymptotic integration are precise, that is, they have the accuracy of the
initial equations. This is easily seen by introducing appropriate changes in the
course of deriving all of the equations and design formulas of the present
chapter.
1964012887-147
136
SECTION 8. EDGE EFFECT IN ANISOTROPIC SHELLS*
If the point of reckoning of coordinate s coincides with one edge of the
shell and the length of the shell over the meridian arc is designated as L, for
values of E corresponding to the edges of the shell we will have (6.14) and (6.26)
L
= o. _= _ _% -- _- 18.1)
o
Then, as is easily seen from (6.27), the coordinate of the current cross-section
of the shell reckoned from edge s :-- T. (Figure 31) will be expressed in terms of
E1 = _'u-B.
z _.,_A :,_,,_ rB,:oJ
I i s__o./.y:o//._: _o)
Figure 31
Examining the fmlctions entering into solutions (6.33) and (6. P,4), it is
easily seen that functions e (B) and _ (_) decrease with disW.nce from edge
s = 0 and functions O (B1) _,nd _ (El) decrease with distance from edge s = L.
As is seen from Table 2, Li_ese flmctions are of the order of 0.04 even _,-ith
E = _r. Hence, if the inve3tigated point lies on the periphery of a cross-section
of the shell which is sufficiently removed from the edge of the shell that with
the given elastic constants E _ lr, the influence ol this edge on the stress of
points of the investigated periphery or more remote points, within the accuracy
of el_'lneering calculations, may be disregarded. This condition may be ex-
pressed in slightly different terms: if the elastic constan*_ (Cll, C22, C12,
Dll ) of the material of the shell are such that _ _, _r with given s, then the
influence of the edge on the stress of the part of the shell corresponding to s
may be disregarded.
*S.A. Ambartsumyen, Long anlsotropic shells, Izvestiya AN ArmSSR
(FMYeiT nauki), Vol. IRr, No. 6, 1951.
1964012887-148
137
If in expression (6.14) we disregard the variability of R2, for _ we obtain
/-
-7-
= s. (8.2)
Inserting _ instead of _, we obtaiv the length (over the meridian arc) of the zone
of extension of the edge effect
(8.3)
S'=V, F -A-"
Hence, if the length of shell over a meridian arc L _ s*, the interaction of the
edges may be disregarded. This means that _ith L _ s* the load applied to the
edge B = 0 has practically no effect on the s_ess of the other edge (B = o4)) and
vice versa.
For homogeneous anisotropic shells, on the basis of (6.2), (2.1]L (4.1)
and (2.14.9), we have
A___-E
Inserting the value of A from (8.4) into (8.3), we obtain Me length of the zone of
edge effect in the ease of a homogeneous orthotropic shell of thickness h:
I (8.5)
In the special case of an anisotropic shell, considering that E_0= Es and t_2 =
1 for the zone of edge effect we obtain the well-imown formula uf A.I. Lur'ye*
m'
VFT Vf m2
s"= _,R2 _ 3_,m'- 1) (8.6)
Although formulas (8.3)-(8.6) are precise for a circular cylindrical
shell, they elsewhere introduce a certain error for reasons of simplicity and
in order to obtain general formulas for all types of shetls of revolution. How-
ever, where necessary, in each special case the quadrature in formula (6.14)
may be calculated precisely and we may find the required value of length of
the edge effect zone.
*A. I. Lur_ye, see p. 61 of work cited in footnote on p. 111.
1964012887-149
138
Comparing formulas (8.3) and (8.6), we note that in distinction from
isotropic shells, by appropriate selection of the elastic constants in the case of
amsotropic shells we may substantially change the effect of the edge conditions
(with a given length of shell) on the overall stress of the shell.
Let us examine two examples.
First example:
_ h L
E-_-_--12, "2-_-0. R-_= 0.04, R- = 0.4;
from formula (8.5) we obtain for s*
3.14.0.2
s"_ _ R_ ----0,256R_.
}r3_
Second example:
E_ h L
_-___ 0.083, _2= 0, _-__ 0,04, _-2-_ 0,8;
from formula (8.5) we obtain for s*
s" _ 3,144"0,2 R2 _-- 0,88R2.
It is seen from these examples that in the first case the i:ngth of the edge effect
is less than the length of the shell L = 0.4 R2 > _.256R2, while in the second
case, conversely, the length of the edge effect is grea_er than me length of the
shell L -- 0.8R 2 < 0.88R2. Hence, it is ev_.dent :,,at in the first case the edges
of the shell will have no effect on one another and interaction may be disregarded,
whereas in the second case the edges of the shell will effect one another and
there will be no justification for disregarding interaction of the edges.
In shell theory long shells are considered to be those shells for which
interaction of the edges may be disregarded and short shells are those for
which the interaction of the edges cannot be disregarded. From this stand-
point, in discussing anisotroptc shells we notice a certain contradiction be-
tween the geometric and mechanical concepts of length of a shell.
If from the geometric standpoint an anisotropic shell is considered long
(or short), this still does not mean that from the point of view of mechanics it
may be regarded as a long (or short) shell; this is also true of the converse
condition.
1964012887-150
139
This is easily seen by examination of the examples given above. Despite
the fact that from the geometric point of view the length of the shell in the first
case is half the length of the shell in the second case, we must consider the
shell in the first case to be long and in the second case to be short.
The results derived in the previoua sections are wholly acceptable and
sufficient for solution of the various problems or long shells as well as for
short shells. However, in subsotuent discussion we shall deal only with long
shells as being most important from the standpoint of application. An under-
standing of the procedures for calculation of short shells may be obtained from
an examination of the work by A.I. Lur rye. *
SECTION 9. LONG SHELLS OF REVOLUTION**
The anisotropic shells of revolution used in engineering practice are
largely long shells (for which the interaction of the edges may be disregarded).
Let us discuss a long shell of revolution along the edges (s = 0, s = L) of
0 and L and the intersecting forces NO
which there act the bending moments M1 M1
0 and L
and NL (Figure 32). We will note that M1 M 2 are positive if they increase
the curvature of the coordinate surface of a shell with positive Gaussian
curvature or if the stresses caused by them at the outer grain of the shell are
tensile forces and on the inner grain of the shell are compressive; the inter-
secting forces NO and NL are positive if, being applied to areas with positive
external normals, they are directed toward the positive normal to the coordinate
surface. The positive directions of the forces and moments are sho_ ia.
Figul _ 32.
In consideration of the basic property of long shells, we shall discard
functions B1 in calculating quantities referring to edge s = 0 and functions 8 in
calculating quantities referring to edge s = L.
With s = 0 (which, on the basis c. _6.14) and (8.1), is equivalent to
assuming 8 = 0) for the transverse forces and moments we obtain from (7.17)
and (7.18) (Figure 33)
N _---NO: AD11 cos __._._o Bl __ AD_l
ro - (9.1)
*A. I. Lur'ye, see pp. 68-80 of work cited in footnote on p. 111.
**S.A. Ambartsumyan, see footnote on p. 136.
1964012887-151
140
A/A /_/_ /V
Figure 32
Ml:-M:Du V_(Az-- Bz). (9.2)
With s = L (which is equivalent to _1 = 0) for the transverse forces and mo-
ments we obtain from (7.17) and (7.18)
NL_ AD,, (9.3)
/W (B_.- ,42).
M 1 = ML = Dll
(9.4)
By simultaneous solutionof (9.1), (9.2) and(93), (9.4) for constants
A1, B1, A2 and B2 we obtain new values which are represented by means of the
initial parameters
NO; Ax _ NO ] V[_/Vl0; (9.5)
Bx-- AD,, _ _;;
R__.NL 1
B2-- AD,, NL; A2-- D,,A D,, M_. (9.6)
Inserting the values of A1 . . . B2 into (6.33), (6.34) and (7.15)-(7.19),
for the principle required functions and internal forces we obtain the following
formulas:
1 [R_NO_([_)Ac- ]/-2---_o ..,' ,_0([3)+
W = _-DT,
-+-R2LNL_ (_z)-- V'2--_2L A,Vl_0([_,)]; (9.7)
1964012887-152
141
p
I 5/I
Y
=0
Figure 33
+ n_N'_(P,) + i/_AM_: 0,)] _,o). (9.8)
COS8 '
T, [ R_No_(,_)+Vo . = -- . 2R2AM,.O)--
- R_N'-_ (p.) - V2n_AM'F(Ih) ] '_"e
---r-+
( ; )
! po re. ds ; (9.9)
+ R_cos_e 2_ ,
$(,
1 -- rE. ds (9.10)
Rj cost O
$o
N [n__:*, 0)- _r_--g,,., ..o, : v'_,,.'_"h.Q)+
+ R_Nt_ (_,) -+-V2-RL2AMfr,(p,)] cos8. (9.11)
M,=2n_ _-_ N:O_ + v _, _,"_,0)--
(9.19. )
D,, . De, sin8[RON09(p)__ -
M_ = -D-_'% D,,A
-+"V 2R A/$Io(P)-+-R_NZ_(P,)-- I/2"g_AM_O(P0]. (9.1 3)
1964012887-153
142
The formulas for determining the stresses and displacements (7.6)-
(7.14) remain unchanged. In determining the design stresses and displacements
we shall also be interested in the following quantities:
4V _/r-_- o 0
--2RbvL0 01)--V_,_AMI_('_1)] + R_z--
( /
I P_ (9.14)
Rtco s2B 2n o rEzds ,
$o
+ 2_ 2-YOGA NL:0,)-% V R,
As in the general case, the boundary conditions do not differ from those for an
isotropic shell of reTolution. * Let us present the three basic variants of
boundary conditions (see Chapter II, Section 10):
edge/3 = 0 is free (2.10.1)
NO= o. Mo= o: (9.16)
edge _ = 0 is M_ged (2.10.3)
M = 0. w = O; (9.17)
edge/3 = 0 is fixed (2.10.6)
R_N+ V2R_AM_ = 0, _= 0. (9.iS)
The correspondingboundary conditionsfor edge fll= 0 have the form
N L= 0. M_ ---- O; (9.19)
M(- = 0. w = O; (9.20)
R_NL--V-2-R._AM_=O, w=O. (9.9-1)
For the normal displacement _, by virtueof (9..9.7), (7.13),and (7.14),
we have
*A.I. Lur'ye, see p. 64 of work cited in footnote on p. 111.
1964012887-154
143
C,17,- Ct_T,
w- ,_-- rcos0+
+I.o : c.r,-c,,r. ) }
, . ,, cos_ (is sin 0. (9.22)
$j
SECTION 10. I XAMPLES OF CALCULATION OF LONG
SHELLS OF REVOLUTION
In all the examples we shall limit ourselves to only the first approxima-
tion of the asymptotic integration.
1. A long cylindrical shell possesses two nondeformable end plates
which may be displaced in the axial direction. '.The edges of the shell are fixed
in the end plates. The shell is loaded by a uniformly distributed normal load
of intensity q. The radius of curvature of the shell is R. Using (2.3.-(2.7),
for the load terms we have
E, -: q, E_ :- O. t_z: r,R2q. |
_._2 R2 / (10.1)
Fl ---=_- q = _ q. F2-_ -- qRs.
since for a cylindrical shell 0 = 0 and r = R; for the lower limit of integration
we take s 0 = 0.
From (9.9) and (9.10), considering that the edges of the sheel are fixed
in the end plates (that is, with consideration of (9.18) and (9.21)) and that for a
cylindrical shell sin a = 0, for the internal tangential forces we obtain
R (10.2)
T,: 2_R-- :2q;
T2=-- _/_R RINcP(_)-- Nt_ (_')]-_- Rq. (10.3)
Then from (9.22) we have
_v --_ R CILT_ _ Ci2Ti __-- R2q (Ci, _0.5C,2)
u. t.,
I/' (lO.S4
Assuming a shell length L, for the values of fl corresponding to the edges of the
shell we obtain from (8.1)
0.
1964012887-155
144
Since with these values of _ the normal displacement 00 becomes zero, we obtain
for the initial parameLers NO and NL
if i/
No - C,, -- (),5C,_ tR C,,--0,5C1_ 2R
- - C;,--- A q' N' ....... C._--- -A-V. (I0.5)
0 and L
Further, from (9.18) and (9.19), for the initial parameters M 1 M1 we have
M,:= kq c,,-osc,, (lo.6)
A C1, '
Rq Cj] -- 0,5C,2
M_-= A C,, (i0.7)
Using the values of N and M1 in (10.5) and (10.6), from (9.7) and (9.8)
we obtain for the principal required functions
IX/ __ -A--_Rq vlf_-A C,l- 0,5C_,c_,[:(_)--r_(_x)l; (I0.8)
F
V :qR|/ _-_C,,--o.sc,_lO 0) - O(_,)l-FRqs. (10.9)
Inserting the values of W and V from (10.8) and (10. b) into (7.6) and (7.7), we
obtain for the normal stresses
a, = Rq C,, i -L;- ["P(_) Jr ?(t9_)] --
i (,0,0, ADll _ I
C,,--0.5C,_ ( "_I
=t = _ Rq .......... I'P(r,j)+ :p0x)l Jr
YB1,
' '!Cq.- ('"-_-_;' 1 -+ AD,; I'.I, 0) F_, (10.11)
From (9.22) we finally obtain for the normal displacement
_,= _-q(c,, -- 0.5q2) {1- l_(P+,P0_)1}.
(lO.
12)
Let us discuss an example of a three-layer shell. Let the outer layers
1964012887-156
145
by isotroplc (Es 1] f = E [1] = E_lfa _- E[31 = E), while the inner layer is orthotropic
(_.[2] E[2]} For simplicity we shal! assume that t)._ Poisson ratios of all
u s , _.'
layersare equal to zero. We shallcon_ _lerthathI = h3 = h0, h2 = kh0 (Figures
15, 27, 34). From (3.1)and (3.2)we obtainfor the stiffnesses
Cn = 2Eho(I __ t_ __ rnsk); C2_ -- 2Fho(I --It _-m_k);,
with the following notation:
e!71
ms=-_-; m,_= E
For thestresses in each layer we obtain
E i
0 i =.: "_
-+",_(_q)l,
Rqc_; [1 --[V(,_) .- ", '
Comparing these formulas with the corresponding formu!as in the membrane
theory, it is easily seen that changes in the normal stresses, associated with
torsional phenomena (the second terms in the formulas), are local in nature
and at a certain distance from the edges these cl_nges may be disregarded.
In view of this it will be of interest _o examine the stresses at the edge of the
shell, where in addition to stresses of the membrane state we have greater
stresses from bending.
Considering that with _ ----- 0. d_(D --- _ (_) = 1.0. _ (_,) _ _ (_,) _ 0. we obtain
i
for crs
a'-qR Ets [ 1 , 21"3 1
s 4n0 E (1--It+m,k} h o '[/'(1 :-_3-_-m]k_)(l- /t--_-m-_'k} "
With R = n stresscsoI are equal to zero.
)
For the sake ofclaritywe shalldiscuss a numerical example. Let all
the layersofthe shellbe isotroplcand for the ratiosof the elasticrnodul!of
"helayerswe have m s = n_ = 0.1. Let ue furtherassume thatk = 0.8.
1964012887-157
146
Figure 34
4
Representing the normal stress _s
in the form
oL _ Rq
,--Xl-4_,
for individual points of the end section of tile shell (Figure 34) and for Ai we
obtain the following values:
Layer I H m
Points 1 2 3 4 5 6
Xi 12.45 10.67 1.07 - 0.35 - 3.53 - 5.31
I
The results of these calculations are represented graphically in Figure 34.
2. An infinitely long cylindrical shell is loaded by forces and mo-
ments uniformly distributed over the periphery of the cross-section. The
load intensity is q kg/cm and the intensity of moment m kg (Figure 35).
Let us first examine the case wherein the shell is loaded only by uni-
form transversely peripheral distributed forces q. Let the origin for the
reckoning of coordh-xates s arxdB coincide with the plane of the loaded cross-
section.
1964012887-158
1_7
Figure 35.
Let us examine the part of the shell to the right of the loaded section.
Then, with _= 0, for the right portion of the shell we have
W:=0, No= 1
_q; (10.13)
further, from (9.19), we obtain
Rq (10.14)
M'_: 2fnA"
Inserting the values of NOand M_ into formulas (9.7)-(9. _13), for the design
quantities we obtain (0 _ _ _ _)
v - _ o(_); w-- 2AD,,: (_);
M,- (10.15)
Similarly, for the left portion of the shell we obtain (0 _ _ >/-- co)
Rq Rq _
v=--_ o_([_);w:.E_-bT ' (_);
M, -- 2_-_ (_); N ==-- q 0, (_}; (10.16)
Rq ,/'_ R,qC,,v/'-_"
T2=-- _]/ _FI([B); w-- 2 ,2 _7,(,_).
where we introduce the notation:
0,(_)= e: *cos_; r (_)___ e_sin[3; }
ml(_)= 01([3) -- q ({_); +1([_)= 0,([_)+-_., (_). (10.17)
1964012887-159
148
Among these functionswe have tllefollowingrelationships:
a o_(_)= v_(_). _ "
d3 _ -,(8) ----- _ (_),
(10.18)
d2 d 2
_ 01(_)= -- 2:1(_). d_-_ q (_)= 20, (6)"
In the case where the shell is loaded only by moments with uniform dis-
tribution over the transverse periphery, all the design quantities of the problem
are derived in elementary fashion on the basis of the solutions obtained in the
first case (that is, when the shell is loaded only by uniformly distributed
forces q).
Assuming that the shell is loaded by two systems of uniformly distri-
buted annular forces (Figure 36) (each of intensity q) which act in opposite
directions and as 17approaches zero form a distributed pair with inte.-.sity
m =q _. from (10.15) we obtaia* for the right portion of the shell
,.,.. ]
-: 2ADll "q 2ADI_ dS "
or, finally,
V-- p(rq), W _(_). (10.19)
2D,__2-EA
Similarly, from (10.15) we obtain for the remaining quantities
m ,aV/- _
M_= _ o(_); N= -- _ _ (_);
(lO. 2o_
m ___@ T2 : -_- A_, (_); _ = -- _ A_ (_).
*S. P. Tlmeshenko, Soprotivleniye materialov, Vol. II, Gostekhizdat,
1946, p. 17.
1964012887-160
149
I a I
i iI,
Figure 36.
In like manner, from (10.1,;) we obtain for the left part of the shell:
Rm /-_ Rm
V--
2 |__fi ,_, (_); w -- 'hfib;
2D,, 1/'2-R-A
m ,_V/-x
MI = -- _ 0_(_); B = -- 2 _-_+_(_); (10.21)
m R,'nC_,
Now let the shell be loaded both by an annular force q and distributed
moment m. We also assume that the origin of coordinate s (as of _) lies not in
the plane of the loaded cross-section, but to the left by a distance b from the
loaded section* (r-'igure 35).
Assuming, that in the coordinate system (01 -- _) to the quantity b there
corresponds t = b _A_. for the sought W and V, from (10.15), (10.16),
(I0.19) and (10.21), we obtain:
in t_ sector 0_t
_/ q_ :(t_D4 mR
-- 2AD,, 2D,, 1/'2#-A_(t -- p); (10.22)
in the sector t _ ff_ o0
mR
w:- qR , (_ t) _(_- t_:
2-AD,---;" -- -I-2D,,Y_-fi3
(10.23)
*S. A. Ambartsumyan, Toward calculation of anisotropic cylindrical
shells of revolution fastened with transverrJe ribs, Izvestiya OTN AN SSSR,
No. 12, 1955.
1964012887-161
150
Inserting the values of W and V from (10.22) and (10.23) into iormulas
(2.1), (2.2), (3.14), (3.15) and (7.10) or by direct use of formulas (10.15)-
(10.21), we obtain for the design quantities:
in the sector 0 _ _ .< t
qR_C,, V/---A mR C,, A: (t _ _); (10.24)
34 i _ qR m
2 _r2R-A_ (t -- [3)--_ 0 (t -- _); (10.25)
v 0(t -- _)-- m VFW (10.26) N -- 2 Y _ _(t -- _):
T2-- qRV/--A2_?(t--_)--_, m A:(t--_); (10.27)
in the sector t _< _ _< :o
w= q_' C" d_2AR mRC" "4""" (10.28)
u _0--t) --'t_-t);
AqI _ qR '_(_ t) m
2}_2RA. -- -F--_-0(_--t); (10.29)
m /-W
N : -_0 (1_-- t) -- _ }, _ _ (1_-- t); (10.30)
qR VF"A m
r2= -- -_- _ _(_-- t)+ _ A: 0 -- t). (10.31)
Let us analyze the stress of the shell in the case where the shell is
loaded only by a uniformly distributed annular force of intensity q. From (7.6)
and (7.7), in virtue of (10.15), (2.24) and (1.12.2), we obtain for the normal
stresses in the shell layers:
[?< ]
o' e_, --A Ci_ "_ Rq (10.33)
"= _- _I,', _ _--_c,,]'_(_)+'_-i_,, _'0) -_
In the vicinity of the sections where the load q is applied we have for the normal
stresses
([_( C,l_ Rq. I
,= i i _.]_ c,,S 2-_,, ' I
I -- h _ (i0.34)
t
:rL +
1964012887-162
151
Let the orthotropic shell be a single-layer shell and the Poisson ratio of
the shell material be zero. Then, for the stresses entering into (10.34), we
obtain from (4. I)
Cll -- Esh, C_z==E_h. Dn Esh_
-= - I-2-' C12: O,
and further, from (6.2), for A we obtain
2
On the basis of the above relationships for the maximum normal stresses %
and % we obtain from (10.34)
a_ 2h II' -_-s "
By means of these formulas it is easy to establish the nature of the
change in normal stresses % and % in accordance with e=5_/5_, that ,s, de-
pending on the ratio of the elastic moduli of the principal directions. Modifying
the formulas, we obtain
q 1/'_-_ q]/_
% h2 1,14e-'/, -- h_ _'s.
hz 0,66e'/,= h_ k_.
The coefficients of these formulas Xs and X_, are represented in the following
form:
0,I 0,25 0,5 1,0 2.0 3,0 I 4,0 8,0 IO,v
_s 2,03 1,61 1,36 1,14 0,96 0.871 0,81 0,67 0,64
,. 0,37 0,47 0,55 0.66 0,79 0.87 I 0,93[ 1.12 !.17
For the sake of vonvenience the data of this table are represented in graphical
form (Figure 37), whence it is easily established that for the discussed problem
1964012887-163
152
from the standpoint of strength it is more convenient to have greater stiff-
nesses in the annular direction of the shell. It is also of interest that with a
sufficiently wide range of change in e (from two to five) the normal stresses
% and % have sufficiently close values.
3. A semi-inifinite cylindrical shell is loaded by forces and moments
uniformly distributed over the periphery of a loaded cross-section at a distance
b (t := b V _t-/-2R)from the free end of the shell. The origin of coordinates s and f_
is tocated in the plane of the cross-section of the free end of the shell (Figure 38).
The solution of this problem may be constructed by the method of super-
position. That is to say, on the solution of the problem of an infinitely long
cylindrical shell loaded at any intermediate cross-section by forces and mo-
ments uniformly distributed over the periphery of the loaded section (Figure 35)
we superpose the solution of a semi-infinite cylindrical shell loaded by the fol-
lowing uniformly distributed (over the periphery of the end section) bending
moments and intersecting forces (Figure 39):
q_ '7 m 0(t); (10.35)
m i/ A (10.36)
No -- _ 0 (t) 4- 2- v _- ? (t),
,1
, r i
ZC --_ i I I
t f t I i _
0 .:J gd d.Y 4.0 3,q dO 7.(] 8./) gO No
Figure 37.
which with opposite sign are the values (10.25) and (10.26) when
q
;_.ZZA _'L::: z._,:f." .; "..;z.. ";.... -'-.-'_ /_;1 _;__._;_,4._(.4_'_
ms,../