Vous êtes sur la page 1sur 38

Progress in Aerospace Science, 1976. Vol. 17, No. 2, pp, 109-146. Pergamon Press. Printed in Great Britain.

I N T R O D U C T I O N OF C O N C E N T R A T E D LOADS IN
P L A T E S AND S H E L L S

S. A. ~UKASIEWICZ
Warsaw Technical University, Poland
Summary: The deformations and stresses which result from the imposition of concentrated loads are
analysed taking account of stresses normal to the mean surface of the structure and of deformations
due to shear stresses. A brief account of methods for designing optimal structures is given. Special
reference is made throughout to the work of Russian and Polish authors.

NOTATION 6 thickness of external layers of sand-


wich plate
~b stress function A,/.t~, Kj coefficients expressing effects of
E Young's modulus curvature of a shell
E3 Young's modulus in the direction kR 12/R2
normal to the middle surface A 1/2k~
D bending rigidity a, coefficient of thermal expansion
/9.. bending rigidity of sandwich plate
q, Z load normal to middle surface
G shear modulus 1. I N T R O D U C T I O N
G3 shear modulus corresponding to the Thin-walled structures extensively used nowa-
normal direction days in industry and civil engineering are usually
Gc shear modulus for the core of sand- loaded by very complex systems of forces acting at
wich plate their edges or over their surfaces. Calculating the
M;~ bending stress resultants strength of a structure we replace real loads by
N;; membrane stress resultants certain idealized loads, distinguishing between
Qi shear forces typical surface loads distributed over a great area
R radius of spherical shell of the structure and loads acting over a small area.
R; principal radii of curvatures of mid- The latter are called concentrated loads. When the
dle surface, R~ < R2 area under the load is very small in comparison
/~R R2/RI with the other dimensions of the surface of the
30, Jz, Ie0, YI Bessel functions of the first and structure, the load can be considered as a single
second kind force or moment acting on the structure at one
T~ average temperature point only. The real loads which are met in prttctice
T2 difference between temperature of always consist of a combination of components
external and internal surfaces such as forces normal and tangential to the wall as
w deflection well as bending and twisting moments. If we know
v Poisson's ratio the distribution of the stresses in the structure
v~3 Poisson's ratio corresponding to nor- produced by each component we can find it under
mal direction any arbitrary load using the principle of superposi-
h thickness tion.
I characteristic length There are two main reasons for the appearance
g,)7 rectangular dimensional coordinates of the stress concentration in the structure. It can
~,0 polar coordinates be produced by notches, rapid changes of the
x = g / l , y = j / l , r = J/l, dimensionless cross-section, holes, cut-outs, etc. on the one hand
coordinates and by concentrated loads resulting from the
c radius of loaded circular area interaction of the elements of the structure on the
fl; angles of rotations of lateral sides of other. Only the latter problems are the subject of
an element of plate or shell the present paper. It should be noticed here that
o~,/3, 3, variables of integration such loads do not exist in practice. However, if we
3'~i shear angles calculate the stresses at points distant from the
3'0 Eu[er's constant point of application of the load, we can replace the
E,g coefficients expressing effects of real load by an equivalent concentrated load
stress normal to middle surface applied at one point of the surface of the structure.
7/ coefficient expressing effect of trans- This load is only a convenient idealization of the
verse shear deformation real load.
6(x, y ) Dirac's function The problem is different if we are interested in

JPAS Vol. 17, No. 2--C 109


110 S. A. J~UKASIEWICZ

the stresses in the vicinity of the small loaded In the mathematical analysis it is convenient to
region of the structure. Then the correct answer present the concentrated load by the Dirac delta
can be obtained only by taking into account the fact function 6(x). This symmetrical unit-impulse func-
that the load acts on a finite area and considering tion of the real variable x is defined by
the form of the distribution over this surface.
Usually, the surface loads do not involve
important difficulties while determining the stresses
f,) f(l~)8(x - ~) dse
and displacements in thin-walled structures. Analys-
ing the problem by means of infinite Fourier series
0 if x<a or x>b.
we get the result in the form of rapidly convergent
= ~f(x) if x=a or x=b. (I)
series. In order to achieve a sufficiently accurate I f(x) if a<x<b.
result it is enough to take into account only a few
where f(x) is an arbitrary continuous function of x.
terms of the series. On the contrary, a calculation
The delta function is not a "true" function since the
for thin-walled structures loaded by concentrated
definition (1) implies the inconsistent relations
forces runs up against serious difficulties in
computation of the stresses. These difficulties are
brought about by the fact that the equations of the
8(x)=0 if x#0; f[ 8(x)dx=l. (2)
theory of elasticity give singular solutions at the
points of application of the loads. In order to obtain
the stresses in the vicinity of the loading point a
While presenting the concentrated load P acting in
great number of terms of the series must be
the point (x = ~:) we can write
calculated which makes the amount of work
involved prohibitive and decreases the accuracy of
the results. Therefore in the case of concentrated q(x, ~) = P~(x - ~).
loads solutions by means of series should be
avoided. Here solutions in a closed form are
desirable. The formal use of 6(x) function provides a
convenient notation permitting generalization of
many mathematical relations. The common use of
2. METHODS OF SOLUTION
Dirac's function 6(x) and other nonfunctional
The most evident way of defining the state of quantities in physics and technical disciplines
stress and displacement in the plates and shells becomes the reason for a certain kind of computa-
seems to be to solve the problem for a load tional formalism, for example concerning the rules
distributed over a finite small surface of the of differentiation of such quantities. Further formal
structure. Next, calculating the limiting values generalization of the classical notion of the
when this surface tends to zero and keeping a function found its realization in the theory of the
constant value of the resultant of this load, we generalized functions, called the distributionsY~
obtain the solution for the case of a concentrated The definition of the operations, the rules of
load acting on one point of the structure. This differentiation and integration of such functions is
method is convenient in the axi-symmetrical cases, particularly important. The ordinary derivative of
in other cases it is not applied because of certain 6(x) is equal to zero for any x # 0 and is undefined
difficulties met by the solution of the equations of at the point x = 0. The theory of distributions
compatibility at the border of the loaded and gives, for example, the answer to the question of
unloaded regions. what is the derivative of this function. Distributions
A second method often applied consists of the most often are defined as:
presentation of the discontinuous load as the sum (1) a class of equiponderant sequences
of certain continuous forces acting on the whole (Mikusifiski's approach);
area of the structure. In applying this method we (2) linear functionals in appropriate linear spaces
express the load by means of Fourier series. (Schwartz's approach).
However, this method is not very convenient in the Taking as our starting point the functional
case of concentrated loads because of slow approach we can define the delta function as
convergence of the series defining the internal
forces and moments in the structure. In the close
vicinity of the point of application of the load the 8(x) = -1- f [ cos otx dt~. (3)
77"
series are very slowly convergent. This requires
calculation of a very large number of terms and
produces errors in the results. Due to this, we aim If integrating, for example, under the sign of the
at a closed solution in the case of concentrated integral we obtain
loads. These solutions can be found more easily by
1
/.
applying the method of Fourier integral or infinite s i n otx
dr, + C = I(x) + C
f S(x) dx =--" |
Fourier transforms. 'n'Jo ot
Introductionof concentrated loads in plates and shells 111

where I ( x ) is a unit step function defined as differential equation (7). All solutions obtained in
the theory of plates and shells corresponding to
i(x)={l0 for x>0 concentrated loads are fundamental solutions.
for x <0 The third approach to the problem of concen-
trated loads consists of studying the singular
and C is an arbitrary constant. solution of differential equations of plates and
In a similar way we can define the Dirac function shells. Considering the equilibrium of a sector of a
of two variables. We have plate or shell containing the surroundings of the
singular point we come to the conclusion that this
sector can be in equilibrium only if an external
8 (x, y) = ~
f[f[ cos ax cos/3y da d/3. (4)
concentrated force or moment is applied at this
point.
On introducing polar coordinates (r, 0) the function We usually demand from a solution for the
6(r) can be presented as concentrated load:
(1) satisfaction of the differential equations of the
problem;
a(r) = ~ - ~ Jo(3,,r)T dy (5)
(2) satisfaction of the boundary conditions;
(3) regularity, except for the point of application
where J0(% r) is the Bessel function of zero order, of the load; and
r = ?[l is a nondimensional coordinate, and l is an (4) satisfaction of the equilibrium of the concen-
arbitrary length. trated load with the edge forces for an arbitrary
The case of the load distributed over a surface of region surrounding the singular point.
radius c whose centre is at the origin of the The above conditions are not satisfactory. In
coordinates can be obtained by integration of the order to be sure that the solution is "pure", i.e. that
expression (5) over the circular surface r ~< c. On it is not contaminated by self-equilibrating singu-
evaluation we obtain larities of higher order one must consider them as
a singular part of Green's function for the
concentrated load. Then integrating it over an
q (r) = ~ J,(TC)Jo(Tr) d% (6)
arbitrary area of the structure for a distributed load
it should be shown that:
The above integral expresses the discontinuous (5) the solution obtained satisfies the
function satisfying the conditions: nonhomogeneous plate or shell equations;
(6) the displacement w and aw/ax, aw/ay are
P continuous everywhere;
r<c
q rrc'-I"' (7) the stress resultants Nx~, N,., N~y, and Mx,, My,.,
M~y, Q~ and Qy are continuous everywhere.
P
r=c q =~ , and In the following sections of the present paper all
the above-described methods will be illustrated by
r>c q=0 several examples.

The distributions are widely applied in the theory of 3. PLATES UNDER CONCENTRATED LOADS
differential equations. The fundamental solution of
the differential equation 3.1 The circularplate loaded by a concentrated
force at its centre
(7) A solution of the problem of a plate loaded by a
concentrated force can be simply obtained by
is called the distribution h ( x ) satisfying the considering at first a distributed load and then going
equation to the limit that a constant load acts on vanishing
area.
At the beginning let us consider the most simple
(8a) case--the circular plate simply supported at the
edge and loaded at its centre by a concentrated
If we know the fundamental solution h (x) and that force normal to its surface. The solution can be
fix) is the distribution for which exists the derived by means of a solution for the plate in
convolution which the load is uniformly distributed over the
inner portion bounded by a circle of radius ? (Fig.
1). Assuming that the radius ? becomes infinitely
y(x) = f ~ h ( x - ~)f(~) d~, (8b) small, whereas the total load P remains finite, we
obtain a solution for the case of a concentrated
then this convolution is the solution of the force.
112 S. A. EUKASIEWICZ

I- LI - -I

t lrrTtir11i
rr t
-I

FIG. 1. Symmetrically loaded circular plate.

Let us consider the solution of the problem Therefore the results which can be obtained
which can be obtained by means of Reissner's based on the equations presented above are only a
equations of transverse orthotropic plates, taking little better than the results of classical theory and
into account the effects of transverse shear valid only in those regions where the load does not
deformations and transverse normal stresses. change very rapidly. (See Fig. 3)
These equations can be presented as follows, c2. Let us introduce a system of polar non-
dimensional coordinates with the characteristic
D A A w = [ 1 - 12(r/- )A]q, length l = a, where a is the radius of the plate. The
solution for symmetrically loaded circular plate
takes the following form
AAqb=-12gAq, A=12kax2 ay2], (9)
w = wo+C~+C,.r 2+C3r 2 1 n r + C 4 1 n r
where w is the normal deflection of the plate, q is
the external load normal to the plate surface, qb is where C~ are the constants of integration and w0 is
the stress function and 71, , g, are certain the particular solution of eq. (8), r = Ha, c = g/a.
coefficients describing the transverse shear rigidity Let us divide the plate in two portions, loaded (1)
of the plate. The coefficients and g result from the and not loaded (2) and apply the following boundary
effect of the stress normal to the middle surface of conditions
the plate. The bending rigidity of the plate
D =Eh3[12 ( 1 - v : ) , and 1 is here an arbitrary r= 1 w '2~=0, M~'=0, Q)2,_ q c2a
2r '
characteristic length
r = c, w (2, = W '1), M'r2, ) = M."', fl~'=/3~',
rl = D/k3G3hl 2, a = v,3E3h :/lO(1 - v)12E,
r = 0, ~(r I, = 0, O(r' ' = 0, (10)
c: = v,3E3h 2/E12,
where M~i are the bending moments in the plate, Q,
where vt3 is the Poisson's ratio in the normal is the shear force and/3r is the angle of rotation of
direction, G3, E3 are respectively the transverse the lateral side of the element of the plate.
shear modulus and the modulus of elasticity in the (/3, = - ~w/O~ + Q,/k3G3h).
direction normal to the plate, k~ is a constant Owing to the complete symmetry of the plate, the
coefficient which in case of an isotropic plate takes shear forces in the two portions are determined by
the value k3 = 5/6. Then if G3 = G = E/2(I + v) and the equilibrium conditions and are
E3 = E, r/ = h 2 / 5 ( 1 - v ) l 2.
The equations presented above are a simple
Q~,=_qra Q~,= qc2a
generalization of the equations given by E.
2 2r
Reissner for transverse orthotropic plates and
obtained with the assumption that the material of where q = P/zrc2a 2. (ll)
the plate had different elastic properties in the
direction perpendicular to the middle surface. The On solving the above set of equations we find the
effect of transverse normal stress tr= is here taken deflection of the inner portion 0 ~< r <~ c
into account in a very simplified manner i.e. by the
assumption that its distribution across the thickness
is =161rD ~-~c2+r'/ 2(1+v)

1- v - E ) c -2]
1 Z + 2 In c + ~ 8 E - 2(-0

_~ 4 ( 3 + v ) - ( 7 + 3 v ) c : c2 l-v
which is justified only if q is constant. 4(1 + v ) ~- In c - 8 ~ a
Introduction of concentrated loads in plates and shells 113

+ 2(,1 - E)(1 - 2 In c)}. (12) surrounding the point of application of the load
b e c o m e infinitely large. The only reason for the
F o r the outside portion c <~ r ~< 1 we have a p p e a r a n c e of singular terms is the action of the
c o n c e n t r a t e d force. The boundary forces cannot
a) Pa 2 { produce such an effect. The expression (15) below
W = 16~rD [2(3 + v) - (1 - v)c 2 can be considered as the particular solution w0 of
1 -r 2 the general equation of the plate loaded by
- 16(1 - v ) E ] ~ + 2r: In r c o n c e n t r a t e d f o r c e at the point r = 0.

+ [c z - 8('0 - e)] In r}. (13) Pa2 [r" - 4 ( - q


Wo=87rD - E)] In r. (15)
Substituting r = 0 in eq. (12) we find the
To obtain f o r m u l a e for the plate with other
deflection at the centre. The terms proportional to "0
boundary conditions it is enough to add to the
and E represent the correction due to shear and
particular solution (15) the general solution of eq.
normal transverse deformations. W e see f r o m eq.
(l) including a certain n u m b e r of constants of
(12) that this effect is of order (h/c) 2. When ~ is
integration so as to fulfil the new boundary
large in comparison with h it can be neglected.
conditions.
Only in cases where the load acts on a small area
The singularity of the functions defining the state
and t? is small, will this effect be important and
of stress near the point of application of the load
should it be taken into consideration. As c ~ 0 the
does not depend on the boundary conditions of the
outer solution given in eq. (13) gives an expression
plate.
for the deflection due to a c o n c e n t r a t e d force
The bending m o m e n t s in the simply supported
plate can be found on the basis of the solutions (12)
wa~= Pa 2 ~ 3 + v ( 1
167rD ( l - ~ v -r2)+2r:lnr+ and (13). For the inner portion 0 ~< r ~< c

8(11+v-v ) e ( l - r2) - 4('0 - e) In r}. (14) (" P 1-J-~c 2

1 r 2
The singularities in the solution arise both f r o m the - ~ ( 2 ( 1 + u) +- (1 - v))-c-5
classical approach (the second term within the
bracket) and f r o m the i m p r o v e m e n t s taking ac- + 2(1 - v) x (c-2 _ 2)el, (16)
count of shear deformation (the fourth term). The
infinite deformations, of course, imply corres- when c ~ 0, M,~ = Moo ~ ~. The m a x i m u m bending
ponding infinite stresses. The term proportional to m o m e n t s are at the centre r = 0.
In r presents the effect of transverse shear and For the unloaded portion of the plate c ~ r ~ 1.
normal d e f o r m a t i o n s and tends to infinity as r
approaches zero, as a c o n s e q u e n c e of our assump- M~'l(2)= 5 [ - ( l+v)lnr~r 1-v~(l-v)
tion that the load P applied to the surface rr~" is MooJ 2
always finite. Thus, w h e n c a p p r o a c h e s zero, the l-vc2
- + - -(71 4 W-r2)+2(1_ - v ) ( r - : 7 - 2 )
:] . (17)
shearing stresses on the lateral sides of an element

4.0

~ -Thick plol"e "l'heory h = 0,2


=0.3
3.0
~ n p l a t e "theory

2.0
Clessicol "theory

1,0

I I I I
0.05 0.10 0.15 0.20

Z
O

FIG. 2. Comparison of the deflection of a circular plate loaded at its centre by a concentrated force as
obtained by means of various plate equations.
114 S. A. ~UKASIEWICZ

The maximum values of these moments are on the + ~ . However, the stresses produced by these
circle r = c . For c ~ 0 , M , ~ c , M ~ 0 ~ - ~ . It forces are very small in comparison with the
follows from eqs. (16) and (17) that the magnitude bending stresses and except for the immediate
of the moments M~,~ and M~,~,~is different when we vicinity of the point of application of the load they
come up to the radius c from the centre and from can be neglected.
the edge of the plate. These moments differ in the Analysing the previously obtained singular ex-
case of an isotropic plate by vPh"/lO~rc-'-. The pression (15) for c ~ 0 we can prove that it satisfies
moment in the plate is hence expressed as a the eqs. (8) and the condition of equilibrium
discontinuous function whose jump appears on the of the shear forces with the load P for the region
radius c at the position of the rapid change of the containing the point (0, 0). If we separate from it the
load. When c ~ 0 this difference increases to expression
infinity.
Expression (17) can be used to obtain the Po 2
w* = - (rl - e) In r
bending moments M, and Mo,~ in the plate loaded 2rrD
by the concentrated force P. Omitting the terms
containing ( h / a ) : we find for c = 0 we find that the solution w* gives

M,}=~_~[_(l+u)lnr +(l-v)~(l-u) ~ Q, ds = 0 . (21)


M,~o 2

-+2(1 - u ) r - % ] . (18) It represents the singular self-equilibrated expres-


sion which, added to the arbitrary solution, does
We see that the moments for r approaching zero not change the equilibrium of the plate. This
become infinitely large: M , ~ : . M o o ~ - ~ : as solution can be obtained only when taking into
r ---~0. account the conditions[5,6,7] of 1.3. Let us
In order to obtain complete expressions for the analyse the solution w* and find the resulting
stresses it is still necessary to calculate the bending moments.
membrane forces in the middle surface of the plate.
The solution of the eq. (8b) gives for 0 ~< r ~< c Pff l - / ~ ,
M, =- Moo 21r r: ' M,o = Q , = O . (22)
Pga:
p = - 8 r r h c , r" (19) This state of stress can be defined as a centre of
bending analogous to the centre of compression
while for c < ~ r ~ < l given by eqs. (20).
These expressions are the results of taking into
Pga 2 account the effect of transverse normal stress o-..:
4~rh (1 c2)(ln r - ~r:).
perpendicular to the plane of the plate. According
to the simplified theory considered, these stresses
Then we find for 0~< r <~ c appear only in the loaded region. The strains
produced by o-~. in the directions tangential to the
P~ plate surface are proportional to the coefficient v.
N,, = Noo - 4rrhc~., The bending moments (22) and forces (20) are the
resultants of these additional strains.
and for c ~ < r < ~ l The above-derived relations can be used in the
calculation of deflections and stresses in the plate at
N, } P~ - ' - 1). all points except in the immediate vicinity of the
N~o = - 4~rh-(l-- c2) (r " + load. If the radius r is smaller than the thickness of
the plate eqs. (8) and (18) should not be used. In
The value of the membrane force Noo in the case order to obtain more exact results the central part
of an isotropic plate jumps from u h P / 4 7 r c - " to of the plate should be treated as a three-
-z,hP/4~rc -2 at the radius c, i.e. at the position of dimensional body.
the rapid change of load. If c ~ 0 the membrane The problem of a thick circular plate was
forces become for c ~< r <~ 1 considered as early as 1925 by A. Nadai, ~3~and by S.
Woinowsky-Krieger~4~in 1933. The particular solu-
N,, ) P~ tion of the equations of the three dimensional-
= + (r-2~ 1). (20)
Noo j - 4~-h theory of elasticity of axisymmetrical bodies takes
the form
For r - ~ 0 the membrane forces tend to _+:c. But,
calculating these forces at the point r = 0 for the
region O < ~ r < ~ c we find that for c-->0 both
w=
f0 [C~.. C"=J,(ar)+ C~.4z e-':Jo(ar)la da

membrane forces IV,, and Noo are equal and tend to u =


fo [C,., e-':J,(/3r) + C7.~Co=J,(/3r)l/3 d/3 (23)
Introduction of concentrated loads in plates and shells 115

where u, w are c o m p o n e n t s of the displacement plate be considered as thick. The stress distribution
vector, C,-~ integration constants, and a, /3 arbit- across the thickness of the clamped circular plate
rary parameters. The solution of the problem was h / a = 0.4, g / a = 0.1 and v = 0.3 is presented in Fig.
obtained by representing the load by means of the 4. In the s a m e figure the results obtained when
integral (5) and satisfying the boundary conditions taking a c c o u n t of the effects of transverse shear
on the plane z = 0 and z = h , where h is the and normal stresses are presented.
thickness of the plate. The numerical results We o b s e r v e also the differences b e t w e e n the
presented in Figs. 4, 5 and 6 are taken f r o m the assumed and the obtained distribution of the
paper by Woinowski-Krieger. (4) tangential stress z~. calculated at the radius r = c. A
The stresses obtained by means of the a b o v e c o m p a r i s o n of the stresses at the upper surface of
equations are c o m p a r e d with the stresses resulting the thick plate, calculated f r o m the equations of the
f r o m the classical theory of plates. We o b s e r v e that three-dimensional theory of elasticity and the
only if ~ (the radius of the loaded circular area) is classical theory of plates, is presented in Fig. 5. W e
smaller than the thickness of the plate, should the o b s e r v e the j u m p of the calculated g ~ at the radius

C C
%1~
b I0
h
. 8

i pla~e theory
6
2 (7

'2
:='= IO

Improved thin plate theory


X
Clossicol theory

b O,

cu Thick plote theory


E
~ = I00
v
o
/~ Clossicol Cheory

2 -e I I I I
g 0.02 0.04 0.06 0.08 I.O0
m
E
m
(3
FIG. 3. Stresses or. = cry, = O'o on upper and bottom surfaces of a thick plate loaded by a force
distributed uniformly over the circular surface rrc 2. We observe that the improved eqs. (8) give good
results for the stress on the bottom surface for g/a > 0.01 and a/h = lO.
Z
~- #0.1
Improved thin plate theory h
Eqs. (8) ~-=0.4

Thick plote theory~


/ I I I I I I I I[ LcIossicol theory
=c J'Fo092.3 \ /4q.3111111111 /

- _ . ,

/ :E 3
l-2&O -28.1
I --2,5.2
FIG. 4. Distribution of stresses in a thick circular plate under a concentrated force at its centre [4].
116 S. A. J~UKASIEWICZ

_2ol (a)

b
e4

~\ (b)
,~ Thick platetheory
-20 ~- \XX Classicaltheory

b -io \\XX~x"
xx~x~~

->.4.
IO

O I

FIG. 5. Distribution of stresses in a thick circular plate on upper and bottom surfaces under a force
distributed uniformly over surface ~rO2.~'~

r = c. Figure 6 presents the comparison of the more considerable for sandwich and anisotropic
distribution of the stress at the upper surface of the plates.
thick plate with clamped edges and loaded at the
centre by a concentrated normal force. 3.2 Concentrated bending moments
Differences between the results of the classical
and improved theories and calculations based on A concentrated bending moment acting on the
the three-dimensional theory of elasticity appear in plate can be defined as the resultant of oppositely
the immediate vicinity of the loaded region as well directed normal concentrated forces of infinite
as at the edge of the plate. It follows from the above magnitude acting at points at a vanishingly small
comparison that taking into account the effect of distance apart. Applying the theorem of superposi-
transverse normal and shear stress, we obtain a tion we can obtain the deflection due to the
relatively small improvement in the results for the concentrated moment by the addition of the
case in which the load is applied to the plate on a deflections due to two oppositely directed forces P.
circular surface of the radius smaller than the Considering the limit of the deflection when the
thickness of the plate. These effects concern only distance between these forces tends to zero, we
the close vicinity of the point of application of the find that deflection w(M) produced by the
load. The effects of transverse shear deformations concentrated bending moment M is defined by
are more important for the calculation of the
displacements than for the stresses. This effect is w(M)=lim[w(P)]~x_o= M
p a w8x
(P) (24)
Introduction of concentrated loads in plates and shells 117

-I0

I0

20

30
0
7
a

FIG. 6. Distribution of stresses on the upper surface of a thick circular plate with clamped edges, loaded
by a concentrated force at its centre. ''~

where w(P) is the deflection p r o d u c e d by a single


c o n c e n t r a t e d force.
The bending m o m e n t applied to the plate can also
be defined in another way. Into equations of _ 10~2K0(x/(10)h)] (25)
equilibrium of an element of the plate, we can
introduce the distributed m o m e n t s m , mr., applied,
r'
+~310~/(10) sin 0 cos
(
OK, ~/(10) ;
for example, as a pair of two oppositely directed
forces tangential to the plate surfaces. T h e n the O = arccos ( x / r )
c o n c e n t r a t e d m o m e n t can be presented, i.e. as
8(x, y)m~. This new definition does not introduce where K0 and K, are the Bessel functions. The
any difference if we use the classical theory of a b o v e solution differs f r o m the classical solution
plates. H o w e v e r , if we take into account the effects only by the term in brackets multiplied by the ratios
of transverse shear deformations, it leads to new h2/r " and r3/h 3. For small argument x the functions
solutions. Let us notice that while calculating the K0 and K, are K o ( x ) = - I n x l 2 and K , ( x ) =
work of the m o m e n t MS(x, y) realized as a couple (i/2)x In (x/2) and the singularities of the bending
of two normal forces we obtain M(Ow/ax) but the m o m e n t s are l/r In r and l/r 2.
w o r k of the m o m e n t m, is m~/3~, where /3~ is the
angle of the rotation of the lateral side of the plate 3.30rthotropic plates
element;/3~ = -(aw/Ox)+ 3'~:, Y~: is the shear angle
in the x, z directions. Singularities a c c o m p a n y i n g the c o n c e n t r a t e d
The m o m e n t m~ has in the case of a sandwich load in the orthotropic plate are similar to those for
plate, a simple physical interpretation. It is a couple the isotropic plate. The singular part of the
of two forces tangential to the plate surfaces and deflection of the plate has the f o r m
applied to the external layers of the plate.
The singular solution of eq. (8) for an infinitely
w = 8wDxh (h 2 cos z 0 + sin 2 0 ) r 2 In r
large plate under a concentrated bending m o m e n t
has a f o r m identical to the solution of the classical (26)
equations of the theory of plates. H o w e v e r , the
relations for the shear forces and m o m e n t s are where A 4 = Dy/D, and/~, = H/~/(DxDy)>~ 1. Dx, Dy
different. ~4~ We h a v e for e x a m p l e are the bending stiffnesses of the plate in the x, y
directions, H is the shear rigidity.
For the case tz, < I the singular solution takes the
M~x 47rMCs O{ [(l + v ) - 2 ( l - v)sin2
form

PI2A I / 2 \ 2
+52 r:-:v(cos:h20
-
- 3 sin z O) x w = 8--~-~.,~ [ l - - ~ t ) r l n r . (27)

J P A S Vol. 17. No. 2 - - D


118 S. A. J~UKASIEWICZ

The curvatures, the bending moments, and so the w (0, 0) = 2 ~ D s:c [(1 - - ~ - ) In ~/(2K---~) - 1] -
stresses are infinitely large at the point of applica-
(29)
tion of the concentrated load.
This result is interesting because in spite of taking
3.4 Singular solutions [or sandwich plates into account the effect of transverse shear deforma-
tion, the displacement is finite at the point of
Concentrated normal force application of the load. For r ~ 0 the concentrated
A singular solution for the sandwich plate of the force P is equilibrated only by the shear forces in
simplified model, (that is, assuming that the bending the external layers.
rigidity of the external layers is negligible in
comparis.on with the rigidity of the complete plate) Concentrated bending moments
can be obtained immediately from the solution (15) If the concentrated bending moment is defined as
using the analogy of the governing differential the resultant of two concentrated normal forces
equations. The effect of the normal stress o-:: acting at a vanishingly small distance apart the
should be disregarded here because it has in this solution for this case can be obtained by the
case no physical meaning. Then the distribution of differentiation of the solution for the concentrated
the moments in the sandwich plate is the same as in normal force according to eq. (24).
the isotropic plate. The only difference is found in The case when the concentrated moment is
the relation for the deflection. defined as the resultant of two concentrated
Singular solutions for the sandwich three-layer tangential forces N acting on the external layers is
plate have been obtained by R. Ganowicz ~ based more interesting. The singular solution for this case
on the Hoff equations for a sandwich plate. The based on equations of sandwich plates derived by
singular solution for the normal deflection takes the Hoff is
form
N(2h + 6)ifx,,
W -- ~-~ [~tln r2+ 1)
= Pl" r Z l n r + P__~K~(lnr2+2 )
w 8rrD: ,~,~
+_D_~d, X -- rk~K~(rkO]}.
~5[14KcD (30)
PK.D f
(28)
Similar expressions can be derived for the
where K,. = Eh6/G~(I - v2), k~ = D:/2DK,., displacements u, v of the external layers. The
bending moments in the plate are identical to those
D. E6(2h + 6 ) 2 + 2 D obtained by means of the classical theory of plates.
2(1 - v:)

is the rigidity of the sandwich plate and D is the 4. SHELLS UNDER C O N C E N T R A T E D LOADS
rigidity of the external layer (Fig. 7).
The resultant bending moments calculated for
4.1 Solutions of Membrane Theory
the whole sandwich plate are identical to those
obtained by means of the classical theory for the In many technical problems the shells are loaded
isotropic thin plate. Considering the close vicinity in such a way that bending stresses are small and
of the point of application of the load and can be neglected. Then only the membrane stresses
examining the above solution for r - ~ 0 we find that and strains in the middle surface of the shell are
the deflection of the plate at r -- 0 has a finite value. considered. These stresses can be considered as

t z

ht

ink_
h
I
hi

FIG. 7. Dimensions of sandwich plate.


Introduction of concentrated loads in plates and shells 119

uniformly distributed across the thickness. The combinations of concentrated forces acting at
equations based on such assumptions are called the the poles of the spherical shell, see the paper by G.
equations of the membrane theory of shells. The MessmerJ ~
concentrated loads usually produce bending stres- These solutions contain singularities of the first,
ses in the vicinity of the loading points and second and third order and differ from the results
calculations performed on the basis of the mem- obtained by means of the bending theory. The
brane theory give wrong results. However, at a solution based on the membrane theory of shells
certain distance from the loading point, the bending can be found in papers by V. Z. Vlasov, ~8~ W.
stresses vanish and only membrane stresses re- FliJggeJTM and F. Martin. ~7~
main, which can be calculated easily by means of
membrane theory. If we neglect the bending, the
4.2 Solutions for the spherical shell based on
problem of stress analysis is greatly simplified, the general theory
since the resultant moments and resultant shearing
forces vanish in the equation of equilibrium. Hence The problems of spherical shells under concen-
the problem becomes statically determined. trated loads have been widely studied. The
Let us consider as an example singular solutions solutions for a concentrated normal force were first
for the equations of the spherical shell, which can obtained by Reissner in 1946, ~9~using the classical
be presented in the following form theory of shallow shells. This problem was also
considered by Flfigge and Conrad, "~ Kalnins and
Naghdi, "2J Nordgren "3~ and Koiter. "4~ Solutions for
Noo J sin" ~o I_- other concentrated loads, such as a tangential force

+C2(cot~)]+ ~ L C , sinmgF
ltan-~}
/ \"
and a bending moment have been obtained first on
the basis of the membrane approximationJ ~'7~ The
bending stresses and displacements in a spherical
shell subjected to localized loads were determined
by Leckie "s~ for non-shallow shells on the basis of
the simplified theory of spherical shells due to
No,, = - N..
Havers "6~ and working in spherical polar coordi-
The above solution is singular because the factor nates. Leckie obtained an asymptotic solution
sin q~ in the denominator vanishes at the poles of which was valid for all values of the angle ~o. This
the sphere. (Fig. 8.) solution contained constants depending on the
Taking, for example, only the constants C , n = 0 boundary conditions. It is known, however, that
we have N ~ = - Noo = Cdsin" ~o. Considering the bending effects in a spherical shell are highly
equilibrium of the vicinity of the pole we find that localized, except in the case of a shell with free
C~ = P/2rrR where P is a normal concentrated edges. The stresses produced by concentrated
force. forces can consequently be separated from those
Assuming different values of the constants C, resulting from boundary effects and can be
and n we obtain solutions corresponding to various considered separately.
If we consider only stresses and displacements
z due to concentrated loads, simple solutions corres-
ponding to a normal force, tangential force and a
bending moment can be obtained.
Let us present the scope of the improved theory
given by Naghdi "7~ in which the effect of transverse
shear as well as those of rotary inertia are included
was first obtained by J. P. Wilkinson and A.
Kalnins "~ for the case of normal concentrated
force.
Let us present the solution which can be obtained
by means of linear shallow shell equations taking
x
into account the effect of transverse shear deforma-
tions and transverse normal stress. If we adopt the
non-dimensional system of rectangular Cartesian
coordinates x, y defined by x = .~[l, y = ~/l where l
is a certain characteristic length, the shell equations
take the form

FIG. 8. Spherical shell subjected to normal forces at the


poles. = [1 - (t/ - e)12A]Z (32)
120 S. A. JdUKASlEVOCZ

(A+-~)[(A+~-~)*+~w]=-gRAZ where
D(3~) = ("/" - kR )2 + "03,2+ I ;

where in the case of isotropic shell '0 = kR = l']R" = h / R N / [ 1 2 ( l - v2)],


h2/5(1 - v)l 2, E = vh2/lO(l - v)l:, ?: = v h / 2 R , and
"0 = D/k3G3hl 2
l(a2+a2] On evaluation we find
a = 7~kTx~ Tr'-J

is the Laplacian operator, where l = = Pl" [ - k e i r - k R [ ( l + u )


w 2~'D L
"v/(Rh)/'~/(12(1- v2)] is a characteristic length for
the shell.
As, first of all, we are interested in the state of
x ,
Yo(r %/{2kR }) + k e r r + 2 r ker 'r)]
stresses near the loading point, we assume to
+ ('0 - ~) k e r r + ~ r ker' r} (33)
simplify the calculations, that the coefficients of the
first quadratic form of the middle surface are
constant. This simplification enables us to achieve where Yo(r~/{2kR}) is the Bessel's function of the
the solution of the problem in differential equations second kind. Its numerical values can be found in
with constant coefficients. It can be also assumed the tables of special functions.
that the shell extends infinitely in all directions. The first term in (33) corresponds to the Reissner
This assumption is made to avoid the influence of solution obtained by means of the theory of shallow
boundary effects on the stresses and displacements shells, the second term multiplied by the coefficient
near the loading point and enable us to derive the kR appears as the result of using c o m p l e t e
simple particular solution of eq. (32) corresponding non-shallow shell equations (32). The third term in
to the concentrated force. (33) presents the effect of transverse shear and
We consider now a spherical shell loaded by a normal deformations. These terms are significant
concentrated normal force acting at the point x = 0, only for small r. For small r ,~ 1 we can replace the
y = 0 (Fig. 9) in the outwards direction. functions ker r, kei r, by the p o w e r series expan-
sion. Taking only first two terms of the series we
05
obtain an approximate formula for the displace-
ment for r ,~ 1.
~/[3(1 - v-')] P R
w ~. ~--~ far - 4 ( 1 + v)kR In ~/(2kR)
04 Improved theory Eqs,(32)
l
O3
where 3'o = 0.57721566. (34)

v
We o b s e r v e that near the loading point the
O2 deflection of the shell is defined by identical
functions to those obtained for the plate resting on
an elastic foundation. The terms containing the
coefficient kR resulting f r o m the effect of non-
OI
shallowness of the shell are of the order
( h / R ) l n ( R / h ) . They are therefore only a little
larger than the r e j e c t e d terms of the order h / R .
Taking into account the deflection resulting f r o m
0 I 2 3 4 ~ 5 ~
the effect of transverse shear deformation, we
obtain the term containing - In r which increases to
FIG. 9. Deflection of a spherical shell subjected to normal infinity if r--*0. This deflection is of a local
concentrated forces. character and for thin shells the effect of transverse
shear deformations is limited to the area 0 < r < 1
W e apply the Fourier transform. The displace- (Fig. 9). H o w e v e r , eq. (33) cannot be used to cal-
ment w and the stress function 4) are then obtained culate the deflection at the points of the shell that
f r o m eqs. (32) in the f o r m of the integrals are v e r y close to the point of application of the
load ( r < h ) . W h e n r is of the same order of
w = 2 -pI2
-~ fo~ {[1+('0 - E)T2][T2- (1 - v)kR] magnitude as the thickness of the shell, eqs. (32) are
no longer applicable. W e know f r o m the theory of
z) 2, J0(vr)v dT thick plates (see Section 3.1) that the deflection at
- ~ ( 1 + "03' 3' t(TT-L-~2kR)~-y ) the point r = 0, z = 0 is finite. To calculate the
Introduction of concentrated loads in plates and shells 121

maximum deflection we can proceed as previously,


-(~-e) keir-~rkei'r}. (37)
namely, calculate the deflection from formula (34)
for ~ ~- h and the deflection of the central portion of
For the shallow shell the coefficient kR "~ 1 and we
the shell (~ < 2h). This deflection can be estimated
have
approximately as for thick plates resting on elastic
foundations. -~ Yo(r~/[2kR ] ) = In r + yo.

The bending moments are the following:
The membrane forces obtained from (37) are
~_~{ l+v.
M,= k e r r - ( 1 - v) 1 kei' r +--f-/cR
N,~- 1 d~b ~b M,,
r dr R 2 R
X [Tr(Y"(r~/[2kR])+~ Y'(r~/t2ka])
2rrl2 r Y'(r~/[2kR ]) + ker' r-F: ker' r
- 2 ( 1 - v) I k e r ' r -(3 + v)keir-r kei' r/
(~) r
+ -E kei' r +~ k e r r
+ [ - ( 1 - v ) e l ker' r
--kR[r ker r - ( 1 - v) kei' r]}
+ l r / ( 1 + 5v)kei r+~'or kei' r]}, (35)
d2~b q5 M,o
Noo = dr'- R "- R
P { l+v,
Moo = ~ vkerr+(l-v) lkei'r+--f-aR _ PR Y"(rx/[2kR ] ) - kei r - - ker' r
27rl 2 r
x [ ~r(l y'(rx/[2kR l) + vTr Y"(r~/[2k, ]) +
+ F:(lkei' r + k e i r ) - ( ~ n - E ) kerr

- 2(1 - v) 1r ker' r - v kei r - l-~-~v


v r kei' r ]
+ ( ~ - F:)I kei' r + n2 kei' r]

+ (1-v)e ker'r+Tl'O(5+v)keir -kR[kerr+(1-v) 1 kei' r]}.


1
+~nvr kei' r]}. The formulae for the membrane forces for r , ~ l
can be transformed to
For small r ,~ 1,

Y'(r~/[2k,]) = 2/rrr, Y"(r~/[2kR]) =-2/Trr z N,0rr} = ~/[3(1 - v")] 1 + 4(1~.+ v)kR

and finally we obtain x [In r 1 - v] .


~+ y 0 + T J +4(r/ - )
r 1-v
(38)
Mo, j=-~--~ - ( l + v ) I n ~ + y o ~- 2

+ 2 ( 1 - v)E1]. (36)
The first term "1" in the bracket { } corresponds to
the result of the shallow shell equations. The
The bending moments are singular at the point r = 0 remaining terms, are the results of the effect of non-
and approach +o~ and -oo. Two singular terms shallowness of the shell and the transverse normal
appear in the expressions (36). First, classical In r stresses or,. These terms contain singular functions
and second 1/r 2 coming out from the improved ln(r/n) and 1/r 2.
theory which takes also into account the transverse Taking into account the effect of transverse
shear and normal deformations. The terms contain- normal stresses we see that N,~ ~oo, N= ~ - o ~ for
ing i/r 2 contain stronger singularities of the second r ~ 0 . At a large distance from the point of
order. We see that the moments have a singularity application of the load the membrane forces
identical to that obtained for flat plates. When the approach the values which are obtained from the
load is an outward direction, M, and Mnn approach equations of membrane theory (31).
+oo and -oo as r---,0 for improved theory, and for The non-shallow spherical shell loaded at the
the classical theory both approach +oo (Fig. 10). vertex was considered by W. T. KoiterY 4) He
The stress function can be obtained in a similar obtained the following expression for the normal
way to the deflection w. On integration we find deflection at the point of application of the load.

q~ = ~---ffR{ - 2 Yo(r'x/(2kR l) - ker r - F: ker r w(0) = ~/[3(114_- v")] Eh


2PR [ 1+ '2(lrr.~
T - + v)
122 S. A. ~UKASIEWICZ

2 Rh
"[ (1 - c
v) kei' c + ~ ker' c ]
2 = 1/2kR = X/[3(1 -- v " ) ] R / h . + 3(1 + v ) ( 2 - v) P
107r Rh
c ker' c l X/[3(l - v')] P I
which slightly differs f r o m eq. (34). X
k e r c + v~,-TV--~..,/
tL_,jj+ 2~" Rh c 2
Analysing all the above-obtained solutions we
see that in a shell with bending rigidity, the 3(1 + v) P kei' c
O'~M ~ O'OOM
deflection under a normal load remains finite, if we "n" h2 C

neglect the effect of transverse shear deformations.


6 X / [ 3 ( 1 - v2)] P ker' c
But there are still infinite bending m o m e n t s and 10,r VRh c '
shear forces. The strongest singularities those
involving an infinite deflection o c c u r in structures
lacking bending rigidity, (see the case of the where c = ( / l is the non-dimensional value of &
m e m b r a n e shell considered in Section 4.1). This The first terms in the a b o v e relations which
indicates that the bending rigidity is of local correspond to the results of the shallow shell
importance. equations are the most important. The remaining
The stresses on the upper surface of the shell terms appear as the result of taking into account the
caused by bending the m e m b r a n e forces are effects of transverse shear deformations and the
presented graphically in Fig. 10. The case when the complete equations of the spherical shell. T h e s e
concentrated force is distributed o v e r the surface terms are negligible if ( > I.
of the circle of a radius c is interesting f r o m the W h e n the load is distributed o v e r a certain small
technical point of view. The following equations area, the m e m b r a n e and bending stresses at the
can be written for the deflection and the stresses in point r = 0 are equal O-,,~ = O-ooN, O-,,u = o'oou. We
the isotropic shell. At point r = 0 we have obtained previously for the c o n c e n t r a t e d load

w(0) x/t12(_l-v2)]PR[l,+Iker,c This result can be explained in the following way.


~" E h 2 LC- c If the load acts on the surface fro 2 the m e m b r a n e
stresses O-o0N and bending stresses O-0ou change
- ~ ( 1 + v)kR l n v ' ( 2 k R ) - l k R ] + 5 - ~ ( 1 +v) rapidly at ~ = ~. Calculating the stresses in the
circular plate we obtained previously a difference
c ker' c] between the m o m e n t s Moo for r > c and r < c ,
when r o c , equal to v P h 2 / l O z r g '. If c ~ 0 this
difference tends to infinity. The m e m b r a n e f o r c e
X/[3(I- v2'] P~ [ 1 + 1 ker' c] Noo behaves in the similar way, it changes rapidly
O-~N = O'deN = --31"- kC 2 C from v h P / 4 7 r ( 2 to - v h P / 4 1 r ~ 2 for F = (. If c ~ 0

1.5
Improved"theory
~ Classicaltheory
1.0 w
\ ~ ff rrM
\ ~ h
I\ R- =Oll v=0.3
~1..5
b

0.5 ~! 'rrM C~~'~M

.--~ ~ a'rrN
0
o'm~,N

I 2 3
7
l
FIG. 10. Distribution of stresses in a spherical shell subjected to normal forces. "Improved theory"
means the theory taking into account effects of transverse shear and normal stresses.
Introduction of concentrated loads in plates and shells 123

this difference tends to infinity. This peculiar Some numerical results for the deflections and
behaviour of the stresses ~r~ is caused by stresses in the shell for different ratio ~/1 are shown
imperfections of the theory of shells which reduce in Figs. 11 and 12. We see that when the load is
the analysis of the three-dimensional problem of distributed over a small area the logarithmic peak is
the theory of elasticity to a two-dimensional replaced by a sharp finite cusp. The height of this
problem. Considering this problem more accurately cusp depends decisively on the area over which the
we would obtain another, also discontinuous load is distributed. Only precise information on the
distribution of stresses (see the solution for the load distribution can give the real maximum stress.
circular plate). Outside the direct vicinity of the loaded area the

2=0.1

Experiment Theory
0.35

0.30
i/ " Nrr

0.25
Points from experiment
F
[25] 4~)2
N
0.20 u =0.3
c o E/t

0.15

O.JO,F C=2.6

0.05

I I I I "~"~J~--
4 ~ ~ "5 - - ~ l r - - . -
0 I 2 3
Z
l
FIG. I I. Deflection o f a spherical shell under a load distributed u n i f o r m l y over surface ~r~ 2 f o r various
ratios = ~./l.

lllcal
III Experiment Theory
[Cbl
~I~- ~I~-

1.0 o
Poini's from expen"meni" L= ~ )8 -I~
4
M M .-8

o.5 Bending stresses


O'rr
I ~ '~ ~ _ - - - - ? M e m b r a n e stresses ~ ~ cree

area ~ U -~=.=jT'~.~,.j,~,
~' - ~ . m " ' ~ ~-'------~
_.______~ ~-- area-'I ~ ' ~ - ~ ~ A
c On \ ~-.-...._e~---I, ~ ... = 0 ~ ~
-~ "~,----"0

I 2 3 4 5 6 I 2 3 4 5 6
Z 7
7- F
FIG. 12. C o m p a r i s o n o f stresses calculated theoretically and obtained f r o m e x p e r i m e n t f o r t w o values
ofc= ~11.
124 S. A. ZUKASIEWICZ

point load analysis usually gives an accurate the load did not undergo changes with change of
answer if the loaded area is small enough. position of the loaded point on the surface of the
When the ratio ?/2h is very small, the portion of shell. The experiments were performed for loads
the shell near the point of application of the load acting on a circular surface with the radius t~ and
should be treated as a body, all three dimensions of different ratios g/l.
which are of the same order of magnitude. The The results described make it possible to build a
stresses in this area will be different from the solution for a more complicated load. It is possible
results given by eqs. (32). The tensile stresses can to find a solution for the case of arbitrary load by
be obtained approximately from Westergard's using the singular solution as the Green's function
formula ~29~ for the plate resting on an elastic of the problem.
foundation. For the spherical shell the coefficient For example, we can calculate the case of a load
k-~ E h / R 2. Then, adding the effect of membrane acting along a certain line and distributed in a
forces we have the tensile stresses specified way, by superposition of the solution for a
simple concentrated force. The solution for the
load presented in Fig. 13 can be obtained by the
tr,=~-~ 0 . 2 7 5 ( l + v ) l n b~ --

where

b =X/(1.6c2+h2-0.675h) where ?<1.724h

b =6 when ~>1.724h.

The general conclusions concerning the applica-


tion of shallow shell theory resulting from the
above calculations are as follows:
In the case when the load is distributed over a
surface rr~: where ~ is of the order of the
characteristic length I or when the deflection or
stresses are calculated at points lying several l
away from the point of application of the load, the FIG. 13. Spherical shell under a line load.
differences between the results of shallow shell
theory and the theory taking into account the effect integration of results obtained for the concentrated
of transverse shear deformations are negligible. force along the loaded circumference.
In the case when the load is distributed over a
surface 7r~: where h < ~ < I and the stresses are
4.3 Concentrated bending moment
calculated at points lying at a distance of the same
order, the complete equations improved by taking The solution for the spherical shell loaded by a
into account the influence of transverse shear and concentrated bending moment can be easily ob-
normal deformations must be applied. If t7 < h the tained by the differentiation of the result for
exact stress distribution in the vicinity of the loaded concentrated force according to the formula (24).
area can be obtained only by using the three-dimen- We mention here only the formula for the
sional equations of the theory of elasticity. deflection w and stress function ~b based on the
The last two remarks refer to rather thicker shells classical shallow shell equations
when 0.02 < h/R < 0.1. In the case of thinner shells
the non-linear effects of large deflections quickly MI [kei' r - ('0 - E) ker' r
W 2~-D
invalidate the results obtained by means of linear
theory.
The theoretical results have been examined by +-~r kei r} cos O (40)
experiments. Tooth "-~ carried out experiments with
a shallow spherical shell loaded at its top by a
concentrated normal force and obtained data which
_MR[I+
~b= 21rl(r ker'r
confirmed the theoretical calculations. The points
1
(Fig. 12) present results obtained using +~,r/[(3 - 2 v ) kei' r + r k e r r + g l } c o s O.
strain gauges. Good agreement is found between
theory and experiment both for stresses and
4.4 Concentrated tangential .force
displacements.
The displacement of the shell shows greater The case of spherical shell loaded by a concen-
deviation from the theoretical curves than the trated tangential force T is more interesting. In this
stresses. It was also found that the stress and case the effect of transverse shear deformation is of
displacement field near the point of application of small importance. If we solve this problem by
Introduction of concentrated loads in plates and shells 125
-08

-06

-04 ~

-.2

02

(?.4
-: --2 --I 0 I 2 3 4

(2,ro~ z
--~-iw ~:o, ~-
FIG. 14. Membrane and bending stress variation in a spherical shell subjected to a concentrated bending
moment.
means of classical theory assuming, as previously, $2 = - ( 1 + v) T-~/ f0* fo* sin otx cos fly dot dfl (43)
that the shell is shallow and of infinite extent we (ot2+ fl2)2+ 1
obtain the following solution using the infinite
Fourier transform: The integral for ~b, is not convergent and so c a n n o t
be evaluated immediately. However, on calculating
. Tl 3cos 0 (= J,(yr) dy the so-called finite part of the integral (43) we
w=(l+~j g:~--k J0 UTi arrive at the following expression for
. Tl ~ /1 \
=(l+vl-~-~---~r+ker'r} cosO. (41) - T21r
l(r In r c o s O - Or sin O ) + const.
4~, =
Then we obtain the following expressions for the
bending m o m e n t s Finally we obtain

Tl
c# = - ~-~-[r In r cos O - Or sin O
Moo =-(1 + v)~-~- +-(1- r2

+ (1 + v) kei' r] cos O + const.


(l+ker' r) +-(l-v) l-keirr
This yields the following expressions for the
(42) m e m b r a n e forces
- 1 [ ( 1 + v ) - ( l - v)] kei' r } c o s O,

, Tl =-2~'~ l-(l+v) kerr--kei'r -


r r '

+ lr kei r] sin O. Noo = ~-~/ 1 - ( l + v )

Replacing ker r, kei r by the first two terms of


their power series expansions, we obtain from (41) x ( k e r r - 2 k e i ' r - r k e r ' r)] r , (44)
an approximate formula for the displacement of the
shell near the loading point (r ,~ l) as and

l + v Tl ~
W -- - -
16 RD r Cs O" N'e = 2 - - ~ [ l - ( l + v ) ( k e r r -2kei' r ) ] ~ (45)

The m e m b r a n e forces are calculated from The singularities of the m e m b r a n e forces at the
expression (43) for the stress function. We can loading point are the same as in the flat plate
write gb = ~b~+ 4)2, where subject to a concentrated force. The numerical
results for the stresses and displacements in a shell
~b~= T! f0~ f0= sin otx cos fly dot d/3 with Poisson's ratio v = 0.3 are presented graphi-
~. ot(~ 2+ fl2) , cally in Fig. 15.
126 S. A. EUKASIEWlCZ
-0.8

06

-0.4

S, -O2o ;
b

0.2

04
-3 -2 -I 0 I 2 3 4

W\pt 1"

FIG. 15. Membrane and bending stress variation in a spherical shell subjected to a tangential force P.

4.5 T w i s t i n g m o m e n t giving

Let us consider now a spherical shell subjected M, [ ~ / ( R 2 - a 2) ~ / ( R : - r 2)


to a twisting m o m e n t M, acting about its axis of v =~ L Ra 2 Rr 2
revolution. The shell is fixed at the boundary r = a.
1
This case can be solved very simply by considering
" + N/(R 2 _ r .~) j .
the equilibrum of a cup cut out by a plane
perpendicular to the shell axis. We find that there
For R - - * ~ this solution is in a g r e e m e n t with the
exist the shear stresses only.
solution for flat plate. If the twisting m o m e n t is
introduced in the shell by a rigid circular insert the
M, (46) distribution of stress does not change.
r ~ = 2~rrZh .

Since the equilibrium equations and the equation


4.6 A r b i t r a r y shell, l o a d e d b y a n o r m a l
of compatibility are in this case automatically
concentrated force
satisfied, we conclude that the m e m b r a n e and
bending stresses are equal to zero. T h e n the Let us consider now an arbitrary shell loaded by
displacements u = w = 0 and nothing but the shear a concentrated force P normal to the shell surface,
strain 3',~ appears in the shell. introducing the local non-dimensional system of
W e h a v e the following relation coordinates (x, y) with a characteristic length I. The
origin of the coordinates is at the point of
r d{d~ T,~ application of the load. The directions of the axis
~ =~ff/=~. follow the directions of the main curvatures of the
middle surface and are denoted by I/R~ and l/R2.
On substitution of z,~ and integrating we find the As, first of all, we are interested in the state of
displacement stresses in the area surrounding the loading point,
we assume, to simplify the calculations, that in this
area the radii R~ and R2 and the coefficients of the
-r -- 27rR 2hG - ~ cosec ~ cot ~o+ 1 In tan + C. first quadratic f o r m are constant. This simplifica-
tion enables us to achieve the solution of the
problem by means of differential equations with
Since at r = a , v=O constant coefficients.
Let us assume further that the shell extends
infinitely in all directions. In this way we avoid the
boundary effects on the state of stresses and
deflection near the point of application of the load.
W e solve the problem by means of the shallow
shell equations, taking into account the effect of
(47)
transverse shear deformations. T h e s e equations
Introduction of concentrated loads in plates and shells 127

take the following form T h e i n t e g r a n d D(3,, ~0) c a n b e p r e s e n t e d in t h e


f o r m D(3', ~o) = (i + 3,')(1 - - ~ ) w h e r e
DA2w - ( 1 - 1 2 , / A ) A ~ = [1 - 12(~ - e ) A ] Z ,
sin2 ~/'-12A-----2
/~ = 2(1 - ,,R) 7--z-~--
~ ~, sin' ~p)(1 + -03, 2)
1 2 ?:R2Az
ffh a ~ + A,w = - ~ - - - .
3,2
- ~ 7~; AR = R2/R,; R2 < Rz.
where ~ = vh/2R2

A s R is a l w a y s s m a l l e r t h a n u n i t y if 0 < AR < 1
c9z c92 1 02 I 3:
we can develop the fraction 1/0-/~) =
1 + R + R 2 +.... We can then obtain the displace-
m e n t in t h e f o r m of t h e f o l l o w i n g s e r i e s
Z = PS(00, y), Rx<R,

If w e i n t r o d u c e t h e s y s t e m o f c o o r d i n a t e s (r, O),
W ~ 2"--~ m=O n=O
connected with the rectangular coordinates by the
relations x=rcosO, y=rsinO, w e find t h e
x[I.,+(rl -~)I~.+-(l+n)'~I~..+,] 1 . (50)
following solution applying the Fourier transform
to eqs. (48)
w h e r e a~.. a r e t h e coefficients defined by the
recurrent formulae
p l 2 ~= /o "n
w = # - - ~ Jo {1-(n - ~)3"~- #[~ + n3"q
ao.o = 1, a,,: p:'= 1 2hnp4,.
(cos 2 ~o + h~ sin 2 ~)} c o s (3"x c o s ~p)

x c o s (3'y sin ~o) 3" d3' d~p a.o--o ....


(49)
D(3', q~)
T h e sign x m e a n s h e r e m u l t i p l i c a t i o n w h i c h f o r t h e
where
c o e f f i c i e n t s p " " s h o u l d b e p e r f o r m e d p " " x p~'~ =
p'*'"*L T h e c o e f f i c i e n t s p ' ~ a r e f u n c t i o n s o f t h e
D ( % ~o) = 3 4 -.I- (I + "03'2)(c0s 2 ~0 + AR sin 2 ~ ) ,
c o o r d i n a t e s a n d a r e p r e s e n t e d in T a b l e I.
1' = R22h2/3(l - v2), AR = R,/R,. T h e f u n c t i o n s I~.,

T h e a b o v e i n t e g r a l is c o m p l e x a n d its e x a c t
s o l u t i o n is u n k n o w n . H o w e v e r , it c a n b e e v a l u a t e d
,
I,... =
ff 3"+~ J=(Zr)
(3,4 + 1),+. 3'
d3".
a p p r o x i m a t e l y b y t r a n s f o r m i n g it i n t o t h e f o r m o f
series making analytical integration possible. are p r e s e n t e d in T a b l e 2.

TABLE 1.

H2. = fO~/2cos (7x cos ~ ) c o s ( T y sin ~0) sin2" ~0 d~0

~rl" 2,,J,(~'r)+. ...... Jr,- (3,r)


-~ [p ~ p ~ ]
4-... + p2".Jo(vr)j.

p="'"=l.3...(2n-l)cos 2nO
p2 . . . . . = I - 3- (2n - 2 m - I ) [ 2n(2n - l ) - ~ . w(2n
. - - 2 m + 1)

cos 2~"-'J O sin TM 0 - 2n(2n - 1) - (2n - 2m - 1)(m + 1) cos 2~. . . . "O.


(2m + 2)!
x sin 2~-" d + . . + ( - l ) ' 2 n ( 2 n - 1). (2n - 2 m - 2 i + 1)
(2m + 2 0 !
x (m .+i)(m + i - 1). - (i + l)cos~ ~. . . . " O sin2C'+'O + - -
m!

+(-l)"-=n(n-l)'"(n-m + m = l , 2 . . . m <ni=Ol'2""n-+
ml
p 2"." = sin s" 0

p.=l; O=arcsin-Y; r=~/(x2+y 2)


t
128 S. A. EUKASIEW1CZ

TABLE 2. account the way in which the load is distributed


over this surface. This calculation can be done in a
similar way to the case of a force concentrated at
I~.,= f f (y,+l),+.
Y'+~ Jm(yr)
(y)m dy one point of the shell surface, i.e. using the Fourier
transform for eq. (48). Then the problem reduces to
Io.,,= - k e i r: I~.o= 1+ ker' r the solution of a double integral of the type (49).
r
1 2 Let us consider a case when the load is uniformly
l._',,= ~ - r kei' r + kerr
distributed over a small circular area of the radius
c. Then the integral (6) can be used. We obtain a
L.o
o = ~r _~(_2rlOo _ / ~ , ) + k e i , r
solution by simple multiplication of all integrals
previously obtained by [2J,(y, )]/3',.- T h e n on inte-
o _ r':" 2(m-l)[2(mr-2)lO 2o o ]
I~.o- 2._1(m - 1)! r ~- . - l~-~,o gration with respect to q~ we obtain the following
relation for the deflection of an isotropic shell.
+ 2 ( m - 3 ) 1 o ..... - I ~ ,.o; m >>-4
r

9\ o r o I r . PI2 ~ 2 [ 2 ( 1 - Au)l"a.,.
I,~.,= 1-4)Io0-7~Io'o=. . -~keir+~keCr, W ~ ~ m =0 n =0

io...,=( 1 -(n_~_--2))lO.o_r
o,
~I..o x
f f ( T 4 3' J,,,(Tr)J,(YC) dy"
+ 1 ) " " ymrm TC
(52)

-(m -2)].o r o,
I.... = I "4n ] 1 . . . . . - ~n l ...... n~>l
The evaluation gives the following simplified
l?,.o= kerr relation for the deflection at the point r = 0, valid if
2 __
I,.o- kei' r the radius c is not very large.

l~.o = 2 I o - Io = 2 + 2 ker' r + kei r


r ' " r" r w(O'O) = X / [ 1 2v( "l -) ] ~ Eh
P ~/(R,R:) [ 1 , + / k e r ' c]
LC" C
I~.o : [2(mr- l) I., .... - I~ .... ]. m >~2

l g t = ( l _ 4 ) lo.o - ~I0'.o
r 2 = - ~r ker' r
(53)
3 ,
I ~1 = ( I - ~)I;.o 1 (when c < 2.)
- ~ rI , ':o = ~ k1e ( r.- ~ l r k e r r
The bending m o m e n t s calculated at the point r = 0
-(m -4)It 2 r in a similar way are presented by the formulae

[
i~. = i _ - ( 4 : ]4) I~..... r ~nn
. I ..... M,. =~--~ (1 c
--c
48
ker' c

1-x/(AR)( l-v )

In this way we obtain a solution of the equations


(kei' c__l 1 "~
in the form of relatively convergent series of x\ c 2 kerc +~ c ker'c /J AR > 0 .
T h o m s o n functions ker (r), kei (r).
The displacement of the shell at the point of
The twisting m o m e n t M , = 0 at the point r = 0. If c
application of the load (if we neglect the effect of
is small, c ,~ 1 the functions ker c, kei c . . . can be
transverse shear deformations) can be easily
replaced by series expansions. Then the bending
calculated in closed form. We obtain
stresses are defined in the following way
V ' [ 3 ( 1 - v2)] P
w(O,O) = 4 Eh 2X/(RIR'); AR > 0 .
~x 1 = 2 3P
~ - ~ { - (1 + v) ( I n ~c + yo)
(51) or:,,:,, J

The above formula is in agreement with the result + 1 + ,,/5G3-~ l + V ; - l + V - - ? ~ ) '


obtained by E. Reissner for the spherical shellY ~
The bending and twisting moments and the AR > 0, (55)
m e m b r a n e forces can be obtained in a similar way.
The results of these calculations are presented in where c = C/I is the non-dimensional value refer-
Figs. 16 and 17. red to the characteristic length.
From the practical point of view the case of a The m a x i m u m m e m b r a n e stresses do not d e p e n d
load distributed over a small surface is very on the coefficient he if 1/> he >/0. In the case of the
interesting. If we need to calculate the m a x i m u m load distributed over a small surface c ,~ I. these
deflection and stress in the shell we should take into stresses depend to a small degree on he and can be
Introduction of concentrated loads in plates and shells 129

~I (a)
{.0
= , =

ba~ 0.8

0.6

0.4
12XR 0 ~" Trl2

i I I I I I i I I
-0"20 I 2 3 4 5 6 7 8 9 I0

7.
l

{.0
H~I (b)
III

O'S~ / / x :o, ~-= r / 2

~l..._~
o.,l-i\\\
I\\V, x:o,.:o

eL/
o21-
\
--7----
0 I 2 3 4 5 6 7 8 9 IO

7
l
FIG. 6. Bending stress variation in double-curvature shell under a concentrated normal force P for
two values of the parameter AR.

obtained in approximation from cylindrical. In order to solve such problems it is


necessary to evaluate the integrals (49) numerically.
o~, = o'y, = X/[3(I - v2)]PI8h 2. Very often, however these integrals converge only
very slowly. In order to facilitate the computations,
The integrals (49) have been calculated also by the integrals between the limits 0 and co can be
many Russian investigators ~4.3'64'65~ in other ways. calculated in two steps. First between the limits 0
Solutions in the form of various series of Bessel and B, where B is a certain arbitrary value and then
functions were obtained for the case 7/ = E = 0. from B to oo. We have
Also orthotropic shells, have been the subject of
many papers in which a similar method was
applied. L. J. Ernst (9" considered the doubly-curved fox F(3,)d3, = joBF(3')d3' + f ; F(3") d3,.
shell and obtained closed solutions for some I
quantities of stress and strain for elliptic and
hyperbolic paraboloids. He applied the infinite and The closed solution of the integrals between the
finite Fourier transform to the shallow shell limits B to ~ can be found from asymptotic
equations, and obtained interesting results for formulae presenting the integrand in the form of
shells of negative Gauss curvature. The results power series in respect to 113,. The formulae
have been tested by experiment.( 37~ obtained in this way are similar to those for plates
loaded by concentrated forces. (u~
The above method has a physical interpretation.
4.7 Simplified solution The internal forces obtained from the integrals
calculated between B and oo are the results of the
Following the previous Sections, evaluation of self-equilibrated local load ZS(xy). c~
the integrals defining the internal forces and For example, in case of concentrated force we
moments in closed form is possible for spherical have (3),
shells. The series (50) for arbitrary shells become
less and less convergent when the shape of the shell ZS(x) = P f[ PsinBx
cos ax da = P8 (x)
deviates from the sphere and becomes nearly ~" X
130 S . A . EUKASIEWICZ

(o)
1.00 ~ , . / X ~ = O, ~ = 0

075L~X'
~
%
0 I I \~"
,X \ \ X,e=025
" ' ~=.'/2

o.4 ,,

0 I 2 5 4 ,5 6 7 8 9 I0

(b)
I O0

0.75
~~'~"x. ;"R=O'u'=0
%
050

0.25
b

X # = O , ~'= 7 r / 2

I I I I I I I I
0 I 2 3 4 5 6 7 8 9
7"
{

FIG. 17. Membrane stress variation in a double-curvature shell under concentrated normal force P for
various parameter AR.

4.8 Singular solution method equations. Applying the perturbation method, the
The calculation of the stresses in a shell of function [o(r) was assumed to take the form of a
positive Gaussian curvature loaded by a concen- power series
trated normal force was performed by K. Forsberg
and W. Flfigge.(42~ They applied a method of f. = f~, fro. (r)k'~
m
studying singular solutions of the homogeneous
differential shell equations. These solutions can be where k~ = [(1 - AR)/(I + AR)], AR = R J R , is the
interpreted as the effect of concentrated loads. The parameter depending on the curvature of the shell.
environment of the point of application of the load In effect, an infinite set of recurrent ordinary
was approximated by an elliptic paraboloid which differential equations was obtained.
has principal radii of curvature at its vertex equal to The formulae for the concentrated normal load
those of the shell at the loading point. The shallow were obtained by use of the following conditions:
shell equations were applied to a solution which 1. the displacement w normal to the shell surface
was assumed to be in the form of an infinite series to remain finite as r--, 0;
2. slope at the origin: lim [aw/Or] = 0 as r ~ 0 ;
3. hoop strain Eoo to remain finite as r ~ 0 ;
r/(r,O)=n =0,1
~ ._ f . ( r ) cosnO
4. the shell element cut out at the vertex is in
equilibrium under the normal load and the
where ~(r,O) is the function solving the shell stress resultants.
Introductionof concentrated loads in plates and shells 131

The numerical evaluation of the series gives values a radius smaller than the thickness of the shell.
which are in a good agreement with results obtained In zone II, with dimensions of the order of the
by means of integrals (49), (Fig. 18). characteristic length l and surrounding the region I,
A further recent solution is reported by R. Dore the shell can be treated as a plate, in which the
and W. Fliiggefl6~ bending stresses are largest. In this zone we should

05 kx ~'\R,+ez / =0..3
"~ 1 : ' (!~_~_~
\
= rd-- ~-

(degree) 30
oo /

FIG. 18. Distribution of bending moment M, in a double-curvature shell under a concentrated normal
force. (Forsberg and FI/iggem~).

Experimental investigation concerning shells of


positive Gauss curvature +~ proved the correctness
of the assumptions and the theoretical results
obtained by the above-described methods. All the
effects produced by the concentrated loads in shells
of positive curvature are of a local character. If
the edge of the shell is not very close to the point of P
application of the load, the boundary effects have no
influence on the state of stress and displacement at
this point..

4.9 Conclusions
From theoretical and experimental results ob-
tained for shells of positive Gauss curvature we can
draw the following conclusions. If we consider a ' , ~,I k ' l l ~ l , l l l l l ,i/Ill, ' l q l l / I / I / 1 / / / / l i I i ] . . . . .

shell under the action of a concentrated load applied


at a sufficient distance from the edge we can divide
FIG. 19. Zones in a shell of positive Gauss curvature
its surface into the five zones, I, II, III, IV and V (Fig. subjected to a concentrated load.
19).
In zone I which lies in the close vicinity of the
point of application of the load, the shell behaves apply the improved theory of plates or shells, taking
like a flat, thick plate. The dimensions of this zone into account effects of transverse shear and normal
are of the order of the thickness of the shell. If we deformations.
calculate the stresses at points lying at distances In the zone III the membrane and bending
smaller than the thickness of the shell from the point stresses appear comparable. Calculations in this
of application of the load, the stresses can be region can be based on the shallow shell equations.
obtained approximately from the respective three- In zone IV, that is outside zones II and III, there
dimensional solution of the problem of a thick plate appear only the membrane stresses which can be
based on the equations of the theory of elasticity. In calculated by means of the membrane theory of
the same way we would obtain the stresses if the shells.
load was distributed over a small circular surface of The last zone, V, lies in the neighbourhood of the
132 S. A. EUKASIEWICZ

edge. The stresses in this area consist of membrane summation. In this way the value of the deflection
stresses and stresses produced by the edge effects. at the point of application of the load was obtained
Calculation of the stresses in the shell can be as (Eh 2/PR)~,/(h/R) w (0, 0) = 0.740. The equations
performed in a few stages. At first, we calculate the of the Donnell-Vlasov technical theory of shells
membrane stresses, then we take into account the give the value 0.555.
edge effects. Finally we consider the local effects in The same problem was considered by S. W. Yuan
the vicinity of the loaded places. On the boundary and L. Ting~'~ who replaced Donnell's equations by
between zones III and IV the membrane stresses those of Fliigge. Morley ~48~also considered a shell
resulting from shallow shell equations must match loaded in the same way. Bijlaard ~5~presented much
with the results of the equations of the membrane data and curves for bending moments, membrane
theory. forces and radial displacement in a cylinder due to
The dimensions of zone III in the case of doubly radial loads distributed over small areas. A circular
curved shells are between 61 for near-spherical cylinder of finite length was assumed to be simply
shells and 30l for near-cylindrical shells. supported at the ends.
Kempner, Sheng and Pohle ~2~ obtained displace-
4.1 1 Cylindrical shells ments, rotations, membrane and bending stresses
corresponding to line loads on a simply supported
In the case of cylindrical shells, the distortions circular cylinder for the cases:
produced by the concentrated loads are not stifled 1. a radial load uniformly distributed along a line
so soon as in the case of shells of positive Gaussian segment,
curvature. In this situation the real boundary 2. uniformly distributed bending moment acting
conditions must be taken into consideration, in the direction of the axis of the cylinder and
particularly the deflections and stresses of the shell in the direction tangential to the circum-
along the circumference should be presented by ference.
periodic functions. In solving such shell problems it Kleinc66~ presented numerous diagrams for the
is convenient to express the deflection of the shell deflection and stresses in the region of loaded areas
in cylindrical coordinates (x, q~), where x is the of an infinitely long circular cylinder subjected to
distance measured along the generator, ~o--the internal pressure. The series for the stresses in the
angle defining the point on the circumference. Then shell, especially for the bending stresses, converge
the solution usually takes the form of Fourier series. slowly. In order to facilitate numerical computa-
Considering an infinitely long shell we can present tion, a method consisting of the division of the load
two radial concentrated forces P acting along the into a local and a global part can be applied.
diameter, in the following way: This calculation can also be accelerated by
elimination from the series of singular parts. For
Z=P6(x,y)=~ Pfo ~ example the bending moments in the shell for x = 0
are defined by the series
x [ 1 +,= ~
cosmp]cosaxda.,_.4.6 (56)
M.~(O, ) = M.(O, ~)
The deflections of the shell and the stress function P ! + u ~, ( k . + l ) ''2
can be presented in the form of similar series. Then - 2~" ~/(2) .~... nk. cos n~o (58)
we obtain the following solution from Morley's
where
equations~
K 4
kn2 = l -I 4n 4.
=2PR2[l_fo~ cosaxda
w rr2D (2 ( a 2 - 1)2+4K 4
As for large n, k -~ 1 the series (58) converges as
(~ (a z + n 2) COS aX COS n~o da l slowly as the series E~ [(cos n~ )/n ], this con-
+ 2~.4.6 Jo ( a 2 + n 2 ) ' ( a 2 + n : - l)2+4K*a"J ' vergence can be improved by the use of the sum of
(57) the series E~:2.4...[(cos n)/n] = -(1/2) In (2 sin ).
The displacements, bending moments and mem-
where brane forces were calculated in the infinitely long
cylindrical shell by S. W. Yuan in 1946,~49~taking
4K4= 12(i - u2)R2/h2 Donnell's equations as starting point, and by K.
Bieger in 1959~62~based on W. Flfigge's equations.
and a similar expression for the stress function. Results were derived assuming a double Fourier
The above integrals can be easily evaluated with series representation.
the aid of the calculus of residues. Calculation of Holand ~67~ also gave influence diagrams for
the shell deflection is then reduced to numerical moments, forces and displacements for segments
summation of Fourier series. These series converge of circular cylindrical shell. The analysis was based
rapidly and no special techniques are required in on Donnell's equations. In Reference 58 Holand
Introduction of concentrated loads in plates and shells 133

considered the same problem. The cylindrical while adopting a system of non-dimensional rectan-
segment was simply supported at the edges. The gular coordinates x = .~/R, y = ~0 with the origin at
problem was solved by dividing the infinitely long the point of application of the load. The equations
shell into two parts in the loaded cross-section. of the cylindrical shell can be reduced to the one
Stresses and displacements were then obtained as eighth order differential equation
edge disturbances originating from this section.
The deflections of the cylindrical shell with A4 4 a 4w
W+4K ~X 4 =0, where
simply-supported edges can be easily deduced from
the diagram for an infinitely long shell by a method
4K 4 = 12(1 - v2)R2/h 2. (59)
of images. This method was first used by Nadai t3' in
the theory of plates. In the case of cylindrical shell
Factorizing the differential operator we obtain the
simply supported at the edges it was applied by
equation
Yuan and Ting.'ss'-~4~These authors based their work
on Flfigge's equation. I. Kempner, I. Sheng and F.
0
V. Pohle '-~-'~obtained solutions for locally loaded L,L2L3L4w = 0 , where L, = A+/31~-x,
cylindrical shells based on the results of the paper
by N. J. Hoff, J. Kempner and F. V. Pohle. ~-~"This i = 1,2,3,4
consisted of presenting the solution in the form of fl,=-fl4=(l+i)K, /32 = -/33 = (1 - i)K.
the series
It is evident that the solutions of the following
equations
w=~Ce k*cos(nx/R), n =mrrR/L
n =0

Aw +/3,-~-xW= 0 (60)
where ~o is the coordinate measured round the
circumference x along the generator, and L is the
length of the shell. If the cylindrical segment is also satisfy eq. (59). It may be shown by direct
simply supported at all edges, the solution can be substitution that
found easily be means of a double Fourier series.
Such a treatment was first applied by V. Z. w=---~K e-~"l-~'"/2~'Yo( 1 i ) (tZ+yZ) I1' dt
Vlasovfl' If the load is distributed over a rectangu-
lar surface the solution can be also obtained
representing the load by means of Fourier
-+ i(1 t)-~- Yo[(li)s]ds dt. (61)
series; Bijlaard in 1954-1955 calculated in this way
the stresses in cylindrical shells simply supported at Here Y0 is the Bessel function of the zero order and
the edges under concentrated forces and bending the second kind. It follows that
moment distributed over rectangular small surfaces
of various dimensions. Numerical results were
obtained by use of a computer. For the bending
X =12----iKfo~sin(l-i)2tYo[(1-i)2(tZ+Y")"2]dt
moments a great number of terms was calculated + K 2 Y
-7000. In another paper ~s~ more than a million (l-,)--~ f dt fo' Yo[(1-i)s]ds (62)
terms were necessary to obtain satisfactory results.
The problem of a cylindrical shell with free edges
is also a solution of eq. (59). Examination of this
loaded by two radially directed concentrated forces
result indicates that the singular solution corres-
was solved approximately by S. Timoshenko by the
ponding to the action of the concentrated normal
assumption of inextensional deformations of the
force P takes the form
shell, t2~Results of such calculations are in satisfac-
tory agreement with experiment.
In a similar way singular solutions for orthotropic w = C Im)t" [,x',y'-(1-t , (63)
shells can be obtained, see papers by I. P.
Artiukhin,'7~' and I. P. Jikalko a'~ and S. A. Khrus-
'Im' denotes imaginary part of the function X. C is a
tenko. (63~
constant determined from the condition of equilib-
rum of force P with the resultant of the shearing
4.12 Singular solutions for shallow forces for small r
cylindrical shells I" f2~ a
r~-~,
A. Jahanshahi '73~examined singular solutions of
the eighth order equations for shallow cylindrical r - ( x 2 + y2),/2
shells and found that the force singularities were Finally we obtain
expressible in terms of integrals of the products of
cylindrical and circular functions. PR 2
Let us consider a cylindrical shell of radius R
C - 4DK 2.
134 S. A. J~UKASIEWICZ

R. Dor(~ and W. Fliigget76~ show that the above


A4w + 4.K 4c~'w
~ = _ [ ~ R 2 T ~ , A .3.
solution can be presented in the form of real
functions
+ 4K4RT~,A~x:]8(0, 0, xy ). (66)
P 2PRK :
w 8wD(~b,+ ~b,); ~b 87r (~bk-- ~b,) The hot spot is assumed to be located at the origin
(64) of coordinates. The particular solution of eq. (66)
where which constitutes the singularities for the hot spot
is~73~

2 x
( x)co
K K
[- l + v K2
w = _ [r~2-~-ff-re + T , ~-~- Im
]

c s h ( ; x ) sin(;x)kei(-~(2)r)]
c o s ( I - i ) 2 Y 0 [ ( l - i ) 2 r ]. (67)

+ fot'l fo' [ker (~(2)s) ~ kei (~(2)s) ] ds dt.


where Y0 is the Bessel function of the second kind.
Let us note that in the vicinity of the origin the
This solution has been proved to be a pure solution,
deflection and stress function can be presented in
i.e. it is not contained by self-equilibrating singu-
the form
larities of a higher order.
In the vicinity of the origin, w and ~b behave like
(1 + , ) %2 K 4 r 2
W
h 27r
PR 2 ,,
w =~--~r'lnr,
PR ,Ir',
qb=--4--~K k ~ m r + x " r l n r .
)
K' _[l + v\ T~,/r ~" ,\ EhT, I
4~ = T ~ L , ~ - T - ) --~ ~7 + x -) ln r - - - T ~ ln r.
The above solution can be easily generalized for the (68)
case of the force moving along the generator of the
shell.3~ Based on eq. (68), singular expressions for the
R. Dor6 and W. FlOgge6~ obtained in a similar stress resultants can be easily calculated. The
way a solution for various singular displacement above solution was obtained by A. Jahanshahi. ~73~
dislocations. The stress function and the compo-
nents of the displacement vector were given for all
cases considered. 4.14 Shells loaded by concentrated bending
moments and tangential [orces
4.13 Shallow cylindrical shells subjected to If the concentrated bending moment is pre-
local hot spots sented as a resultant of two oppositely directed
The equations of the shallow cylindrical shell normal forces of infinite magnitude acting at points
subjected to a field of variable temperature may be zero-distant apart, all effects produced by it can be
presented in the following form obtained from the solution for the concentrated
normal force. In order to find the deflection
produced by a moment M = lim PAx it is enough
02q"b = _~_~.EDR2A(cl, T,),
D A 2w - R-ff~x, Ax-~0
to differentiate the respective results with respect
to the variable x and multiply the result by M/PI.
l 2 -tgZw If the shell is loaded by a moment distributed
- ~ A ~ + R~x 2 = -- R'-A(a,T,), (65) over a small surface which is for example the
resultant of a variable normal load q(x, y) acting
where Tj is the average temperature of the shell at over this surface the solution can be obtained by
the arbitrary point (x, y), 7'2 defines the difference taking the solution for unit normal force as the
between the temperature of the external and Green's function for the considered problem. In
internal surface of the shell, and a, is the coefficient this way T. E. Hals ~ obtained solutions for
of thermal expansion. concentrated moments applied to a segment of an
Eliminating the stress function and introducing infinitely long shell with simply supported edges.
the intensities T,~ of the plane hot spot and T,,--the The cases considered were, firstly, a concentrated
intensity of the bending hot spot moment distributed along a restricted length of one
of the generators of the surface, and secondly, a
moment distributed over a finite but restricted
T,~ = lim (a,T,S),
s~0
7",2 = lim
s-,O
(a,T2S). rectangular element in the surface. The results were
T~ T~ presented in numerous diagrams.
The solution f o r a double-curved shell subjected
We obtain the equation to a force T tangential to its surface can be
Introduction of concentrated loads in plates and shells 135

obtained in a similar way, using infinite Fourier vicinity of the loading point and extends to infinity
transforms. from the edge in all directions. Let us perform the
The solution of eq. (48) can be presented in the calculations applying the non-dimensional system
form of coordinates x, y with the origin at the point of
application of the load at the edge. (Fig. 20)

.,,..
w = ( 1 + v) 2"~"~-'D f:,: y: 1_]-+-7'-
r,,.,, + ,.
- (1 - AR) sin-' q~] cos q~

sio, x

Tl f "'" (~ [A. (cos" p + )t. sin" ~o)


4' = *:J0 .10

y
+ y4(sin-' ~o - u cos-' p)] sin (yx cos ~o)
R,
cos (yy sin " d y d~p
q~) D(y-)-y:-~os ~o' (69)

where

D ( y ) see (49), An = RdR,.

The first integral (69) can be evaluated by the FIG. 20. Shell loaded at its free edge.
same method as the integrals in Section 4.5. The
Applying a single Fourier transform to the govern-
integral for the stress function can be divided in
ing differential homogeneous equations (48) we find
two parts ~b --- 4~, + ~b,. where the following set of algebraic equations for the
u n k n o w n functions w and qb,'7" in which the effects
~, = T_~/f"'-' f " _____~__1 sin (yx cos ~)
7r-jo Jo y-' cos ~o of transverse shear and normal deformations are
neglected here
COS (yy sin ~0) d y d~o,

qb-' = T! f " n f =(I -I- u)y4 + ( 1 - A . ) [ I - ( 1 - AR) sin-' ~o] D(c& +/32)2w(/3) + ( a : + AR/32)4~(/3) = 0,
7r- Jo J0 y-'D(y)
x cos ~ sin (yx cos ~) cos (yy sin ~ ) d y dq~. ~-ff (~-' +/3 2)% (/3)- Rt~"~(~-' + X,/3-')w(/3) = 0.
(7O) (71)

The first integral in (70) has a closed singular Then we find the solution of eqs. (48) as:
solution

7"/ w = L" ~/ [B~ eh: cos a~x


qbj = - " ( r In r cos 0 - rO sin O) + const.
21r
+ C~ C : cos c~x ] cos/3y d/3, (72)
The second integral can be solved in m a n n e r
analogous to that used earlier for w. where a, b~ and cl, d~ are roots of the characteristic
The above relations allow calculation of the equation
deflections and stresses in the shell. The respective
diagrams can be found in Reference 68. The ( a : +/32) 4 + 4K'(t~-' + AR/32): = 0;
problem of the cylindrical shell under tangential
4 r ' = 12(1 -- v2)R~/h "-.
load was solved by T. E. Hals 'u~ applying the
method proposed by J. H o l a n d : 67~ The result of
The following b o u n d a r y conditions must be
these calculations was a n u m b e r of influence
satisfied at the free edge for
diagrams for the displacements, m e m b r a n e forces
and bending moments.
x=0, N~=0, Mx,=0,

4.15 Edge loads Nx~ + M , IR2=O, Q, + a M , / a , = P S ( 0 , y).

Let us consider now an arbitrary shell loaded by Solving the above set of equations we find values of
concentrated forces at the free edge. This problem the constants Bi and Ci and finally the displace-
can be solved approximately by similar assump- ments and stresses in the shell in the form of
tions to those made previously. That is, we will integrals. The integral (72) defining the deflection
a s s u m e that the shell has c o n s t a n t curvature in the gives on numerical evaluation a value of the
136 S. A. EUKASIEWICZ

deflection which is much bigger than the correct Berry. ~ He obtained an exact solution based on
result. The reason for this is the assumption that the the complete equations for the spherical shell in
shell is of infinite dimensions. The formula (72) terms of Legendre functions. It is very interesting
presents only the deflection of the shell in the that the results of the simplified calculations

R 15

FIG. 21. (a) Deflections of a cylindrical shell under a concentrated normal force at its free edge. (b)
Deflections of a cylindrical shell under a concentrated bending moment at its free edge.
08
vicinity of the point of application of the load with
the accuracy down to the constant value (Fig. 21).
The membrane forces and bending moments can
be, however, obtained in this way to a much closer
approximation. Integrals for the internal forces
decrease rapidly with the distance from the point of
application of the load and depend to a smaller
O6

O4

O2
L
i~ Mr x

IOO

degree on the boundary conditions as x--*~ and 0

___y ~ oo. S. Jukasiewicz m~ obtained the stresses in


the shell while applying the above described v
02

method for various edge loads and a range of the b


ffYYN / /
parameter 1 ~>}tR ~> =0.2. The results were pre-
sented in the form of diagrams. The slow-
O4

O6
GXXM/"
converging integrals were evaluated in two steps as
was explained in article 4.7.
08
This problem was solved by P. Seide c93~in case of
a cylindrical shell loaded at its edge by a system of I0
concentrated and distributed radial forces and
moments. However, he calculated only the dis- 12 [ ] / I I I
0 2 3 4 5 6 7
placement w and the slope awlax because of slow
convergence of the applied series. The hemispheri- T
cal shell under concentrated radial forces and FIG. 22. Stresses in a hemispherical shell under two
moments at the free edge was considered by J. G. concentrated bending moments at the free edge.
Introductionof concentrated loads in plates and shells 137

described here are in complete agreement with his w = w0(1 - r2/b2)[l(l + ~,)r2/(5 + v)b 2]
results in the case of a shell under concentrated
bending moments (Fig. 22). No difference in the where w0 was a free parameter.
diagrams would be observed if we introduce the The same problem was considered by D. G.
notation of Reference 77. Ashwell, who applied a very interesting method of
solution. He solved the non-linear problem of large
displacements of the spherical shell by means of
4.16 Large deflections of shells
the combination of two linear solutions. Namely,
Experiments with shells loaded by concentrated observing the behaviour of the shell during
loads proved that the deflections calculated from deformation he noticed that the dimple appearing
the linear equations formula (51) are compatible on the shell surface has a spherical shape of the
with those coming from experiments only for the opposite curvature to the initial shape. Only in the
initially small values of the forces. In the case of a boundary between the two regions does a bigger
shell of positive curvature an elliptical dimple-zone deviation from the sphere appear. It can be
of reversed curvature was observed on the shell assumed that the dimple, i.e. the inner region, and
surface. This zone spreads outwards as the load the outer region suffer a small displacement from
increases. This particular behaviour, observed first the spherical shell shape. Then for both regions,
in the case of the spherical shell, has aroused the linear equations can be adopted. In this way a large
interest of many scientists. Bieseno (8~ solved this deflection and non-linear behaviour of the shell can
problem as early as 1935, reducing it to two be described in terms of two sets of equations of
simultaneous non-linear differential equations for small deflections, only the compatibility and
two unknown functions, v - - t h e displacement equilibrium conditions must be satisfied at the
normal to the axis of symmetry and ~o = d w / d r - - boundary between the two zones.
where w is the displacement of the shell in the Results of numerical calculations obtained by
direction of this axis. This function was assumed to this method are presented graphically in Fig. 23.
take the form: The thick full line is the variation predicted by
Ashwell. The thin full line results from the linear
~o = C,r/R + C,.(r/R ) In (b /r), theory.
R. Archer (~) reduced the problem of the spherical
which is similar to that for an axisymmetrically shell loaded by concentrated force to the solution
loaded plate. In consequence a relationship be- of a set of algebraic equations by means of
tween the applied load and the deflection of the non-linear Reissner's equations and the method of
shell was obtained, shown in Fig. 23. Chien ~8" finite differences.
treated the spherical cup with a line load uniformly A simplified method of calculation of large
distributed along a circle concentric with the apex deflections proposed recently by Pogorelov (9~ is
and of a radius ~c. Putting a = 0 gives the case of also worth mentioning here. This method is
concentrated load. Two simultaneous non-linear founded on a similar perception to that of Ashwell,
equations for w and Nrr were applied. The solution that the shell deflecting elastically, takes a form
was obtained by an energetic method assuming the resembling one of the forms of its isometric
deflection to take the form deformation. It means that geometrical bending

~ Ashwell

~,~'c~/~/~"
,,P///
"~ Y ~ - Biezeno
. .----

m
Z..J
Q_
./,// ' ~ !~ I

0 0.5 1.0 1.5 2.0 2,5

FIG. 23. Load-deflection curves for spherical shells loaded by a concentrated normal force. Points from
Ashwell's experiment, c~)
138 S. A. EUKASmVnCZ

plays the most important role in deformation of the Assuming that the strains are almost isometrical,
shell, while the strains in its middle surface are of Pogorelov obtained the following relation for the
smaller importance. In fact, assuming that the shell strain energy in the shell referred to unit length of
is made of steel of modulus of elasticity E = the geometrical rib.
2.106kG/cm 3 and yield stress tre = 5.103kG/cm ',
the corresponding strain in the middle surface does U = 0.19E(ha )torT'l:. (75)
not exceed E = tr/E = 2.10 -3 which is a sufficiently
small value. The deformation of the shell by such a where r, is the radius of the rib, 2a is the angle
change of metric tensor of the middle surface, does between the straight lines tangential to shell surface
not differ much from the isometric deformation, by at the internal and external region near the rib and
which the metric tensor does not change at all. It perpendicular to it. Fig. 24. Since for a shallow
means that it is possible to foresee to a certain spherical shell we have
degree the shape of the deflected shell, looking for
it among the isometric transformations. Consider- a =r,/R, 2[=r~/R,
ing a spherical shell, we come to the conclusion that
the simplest isometric deformation corresponds to the strain energy is
a mirror-like reflection (Fig. 24). We can assume
U = 2rr0.19E(2f)312h ~12R-I (76)

Forming the functional W = U - L and equating


the variation of it to zero, we find the following
formula for the deflection of the shell

2[ = R 2P:/(97r:c"E:h -~), (77)


' 1"17

which is presented in Fig. 25. The results of the

10-
3rr Pogorelov

I_

7 ~'~ ..... = O?
FIG. 24. Shape of a deflected spherical shell.
6
therefore that the deflection produced by the
concentrated force, has approximately the shape
presented in Fig. 24 and consists of the increase of Q.

the dimple and the generation of certain strongly 4

curved region at a place where the shell is rapidly


bent, which is called a geometrical rib.
The relation between the deflection and the load
can be found by means of the variational Lagrange
theorem which can be formulated in the following
way. Under the action of the given conservative
load, the shell takes the form for which the I I I I 1 I I
0 5 I0 15 20 25 30 35 40
variation of functional W = U - L is equal to zero;
U being here the total strain energy caused by the h
change in the shape of the shell, L representing the
FIG. 25. Comparison of the results of Pogorelov's theory
work of the external load. So: with results of Penning's experiments. (")

8 ( U - L) = 0. (73) experiment by F. A. Penning (8") are also shown


there. These experiments were performed on a
In our case it is easy to calculate the work of the number of shallow spherical shells with clamped
external load, L = P - 2f, where 2f is the deflection edges made of aluminium and loaded at the apex
of the shell. over a small circular contact area. We observe that
Calculation of the strain energy is more trouble- the deflection given by formula (77) corresponds
some. It can be done approximately. The area of quite well to the curve resulting from experiment
the shell can be divided into three zones: I--central, for the larger value of the ratio b/l, where b is the
I I - - a r e a near the rib, III---external area. The radius of the edge and l is the characteristic shell
energy can be calculated separately in these zones. length I = v'(Rh)~/[12(I - v2)]. The experiments
Introduction of concentrated loads in plates and shells 139

proved that the deflection patterns are symmetrical Pogorelov obtained similar simple results for other
only if the force is smaller than a certain critical cases of shells under bending moments, forces and
value. After surpassing this value the shape of the moments introduced by rigid inserts.
deflection undergoes a change and its magnitude
grows rapidly. The deformed area takes a form
5. D E S I G N O F P L A T E S A N D S H E L L S U N D E R
similar to the triangle, quadrangle, etc. The problem
CONCENTRATED LOADS
of determination of this critical force can be solved
by analysing the stability of the shape of the rib on Results obtained in the previous chapters showed
the shell surface. Pogorelov obtained the critical that concentrated forces causes stress concentra-
force for a near-triangular deformation pattern: tions in thin-walled structures. When the force acts
at a point infinitely large stresses always appear. In
Pc, = 3"rcEh3R -'. (78) a real structure the stresses are limited by the
plasticity of the material. Therefore in the vicinity
In Fig. 25 this force is given by the horizontal of the points of application of loads, plasticized
straight line. areas appear. Sometimes these areas have a local
The non-linear displacements of the shell of character and are not decisive as to the strength of
positive double curvature when R. #R2 can be the whole structure.
calculated similarly to those for the sphere. An However, structures submitted to fatigue loads
elliptical dimple which has the form of the original should be considered differently. There a local
surface but of opposite curvature is observed on concentration of stresses may be the reason for a
the shell surface. Pogorelov obtained the complete local crack, which as it widens in time, leads to
energy integrating it along the rib and in the internal damage. In this case we always try to design a
region, as structure to have the smallest possible concentra-
tion of stresses. We have two methods at our
disposal. The first consists of introducing the load
U =O.19"rrEhS/Z(2f)3r'(--~+l)._ (79)
to the thin-walled structure in such a way as to act
on the greatest possible portion of the surface with
The relation between the deflection of the shell and as little variation as possible. The other consists of a
the concentrated force P is local increase of the rigidity of the structure in the
2a
_ _ p

t
--------...

FIG. 26. Infinite plate joined to a bar of finite length and loaded by a force P applied to the cross-section
of it.

region near the point of application of the load and


P =~ffO.19Eh~'-(-~+~z)(2f)'~2. (80)
shaping the structure in such a way as to make it
transmit the loads without any or with but little
The maximum stresses appear near the rib and are concentration of stresses only.
the effect of bending in the direction perpendicular For example at the loading point we can place a
to the rib. These stresses can be calculated bar or a rib whose task is to distribute the
approximately from concentrated load over the structure in the form of
continuous tangential and normal line loads.
= { _Eh
~ _]~ ( w, ) m x = 0 9 E ( \ 2fh y,2 " (81)
If we assume that at such point of the whole
o'b " \RjR,.] structure the reduced stresses reach a certain
140 S. A, EOKASmWlCZ

constant value then we design it for a uniform e(x) = - -1 fox _ P x (83)


A(x)Eb q ( x ) dx EhA (x-~) a"
strength. Such problems which consist of defining
the shape, thickness and distribution of streng-
The strains in the sheet are
thening elements in the structure, are rather
complex and in the case of concentrated loads they
have not been widely studied. Only a few basic e(x) = - q(~)E(x - ~) dr, (84)
a
solutions will be presented here.

where e(x) = [(3 - u)(l + u)147rhE](l/x) is the influ-


5.1. Design of plates ence function for the strain e,. Equating the strains
in the plate and in the bar we have
As the first example, let us consider the problem
of introducing a load into an infinitely large sheet. x+a
A (x) = Ao - -
In order tO avoid a strong concentration of stress in x--a "
the region near the point of application of the load, In x ~ a a
where (85)
we apply the load through a shaped bar. Then the 21rhaE
load is introduced in form of distributed line forces. A0 =
( 3 - v)(I + v)Eh"
We assume that, for example, the bar is of finite
length. If we demand that between the sheet and
It follows from the above formula that for x = 0 the
the bar there should appear only continuously
cross-section of the bar increases to infinity. But if
distributed tangential line forces q(x), the cross-
the applied load is distributed along a certain
section of the bar should be variable and equal to
distance we obtain a finite cross-section of the bar.
zero at its ends.
At the ends of the bar the cross-section decreases
The solution to that problem can be found after
to zero; it can be proved that even if the force
analysing the equations of compatibility of strains
q ( x ) = const, the stresses in the sheet increase to
in both elastic elements--bar and sheet. Let us first
infinity at the ends of the segment of the joint.
consider a more general case, namely, two elastic
Thus, if we want the sheet to be free of infinitely
elements I and II joined along the line A B . The
large stresses, the ends of the bar should get thinner
forces acting on each other should be equal and
more rapidly than follows from the above formula.
oppositely directed. The equations of compatibility
A good result may be found if we assume that the
of strains for both elements can be written in form
interaction between the bar and the sheet q ( x ) is
of integral equations. For the strain e~, we find
presented by the equation

e,~ = f~ [Kl(x,~)qn(~)+ ' ~)qt (~)] d~


K,.(x,

=
L (KE~t(x,~)qn(~)+K~X(x,~)qt(~)d~ (82) Then the reduced stress in the sheet along the joint
is almost constant. The shape of the bar can be
easily found from eqs. (82).
where Kt~, K~ and K~~ and K~~ are the influence
In the same way various problems of plates and
functions for both elements respectively. K~ is the
sheets strengthened by designed ribs can be easily
strain produced at point x by unit normal load
solved. The design of an attachment introducing a
applied at point ~.K2 is the strain produced at point
tangential load into a spherical shell was solved by
x by unit tangential load applied at point ~. If we
Eukasiewicz and BorajkiewiczY4~ based on the
know the influence functions KI, KI ~, then solving
influence functions given by the relations (33, 41,
the equations (82) together with the equations of
44). The solution for the infinite sheet loaded directly
equilibrum we can define the distribution of forces
at the edge by concentrated normal formal force was
q, and q, acting between the adjacent elastic
found by Kapkowski25~ The thickness of the sheet
elements. However, these equations can be used to
was designed with the conditions of uniform
design a proper shape of both elements. If the
strength. The thickness distribution obtained has the
influence functions, or certain parameters of the
form
influence functions are considered unknown then,
assuming proper and desirable distribution of the 2P cos 0
h (87)
reacting forces q, and q, and solving the equations or0 r
of compatibility, we obtain the geometry of the
structure. If we look only for certain parameters of where r and 0 are polar coordinates and or0 is the
the influence functions, these equations become reduced stress. The same solution may be applied
ordinary algebraic equations. in the case of a wedge loaded at the apex by a
Let us require, for example, that between the concentrated force
sheet and the bar there should appear uniformly
distributed tangential forces q ( x ) = qo. The strains h = P 1 f cos cos O sin sin_ O__]
in the bar of variable cross-section are O.o r L~0 +~ sTn ~.oL,~+ oL0+ ~sin 2a0j (88)
Introduction of concentrated loads in plates and shells 141

where 2a0 and ~o are the angles defined in Fig. 27. It first case is reduced to solving the problem of a
should be noted that this solution is valid if the edge shell which carries both direct and bending stresses
r = const, is far from the point of application of the and is usually rather difficult. In the second case we
load and assures the appearance of radial stresses design the shape of the middle surface and
only. Also the region immediately surrounding the thickness of the wall by assuming that only
point of application of the load should be excluded, membrane forces appear in the shell, and bending is
because h --, ~ as r ~ 0. The load must be applied in eliminated. By designing it in this way we can
the form of distributed radial pressure. achieve the smallest weight for the whole structure.
The solutions obtained by the assumption that both
main stresses are equal and constant at any point of
the shell are well known (see References 102, 103).
However, this assumption determines simultane-
,o
ously the thickness and the shape of the shell. If we
assume satisfaction of the condition of constant
strength it determines only the thickness and we are
still able to choose the shape of the middle surface.
Owing to this the shape of the shell can be found
from additional conditions for example, a minimum
weight, a minimum of costs, etc. As an example let
us consider a shell loaded by a concentrated force
and by a constant external pressure po. Let us
design a shell to the assumption that the bending
moments are zero and that equilibrium is satisfied
by the membrane forces only. Simultaneously let us
demand satisfaction of the condition of minimum
weight for the whole shell and that each point of the
shell fulfils the condition of uniform strength. Such
an optimum design is worked out without regard to
0~0 ~ ~0 difficulties and costs that may arise in its manufac-
ture. Nevertheless it can be the basis of comparison
with any proposed structure. We shall consider this
problem as axisymmetrical.
The four unknowns to be found are the
Fro. 27. Wedge-plate loaded at the apex. following: the internal forces N,~, Noo, the thick-
ness of the shell and the shape of the shell
The optimum design of plates with the assump- determined by the function z(r) (Fig. 28). We have
tion that under a certain load every point of the
plate simultaneously reaches the plastic state was
studied by D. C. Drucker and R. T. Shield, (96~ W.
Prager, ~97~and M. Sz. Mikeladze. ~gs~A circular plate
~P B=Sr~
under concentrated normal force was the subject of
the paper by E. T. Onat, W. Schumann and R. T. , P D

Shield. ~99~In the case of an elastic design, i.e. with


the assumption that the plastic state appears only in
the outermost layers, this problem was solved in
References 100, 101.

FIG. 28. Shell of minimum weight and unifrom strength


5.2. Design of shells subjected to a force P and pressure Po.
If the loads act over a small portion of the shell
at our disposal the following two differential
surface the designers strengthen the loaded regions
equations of equilibrium containing the unknown
locally by welding in strips, ribs or other elements.
functions. "7~
A uniform strength of the shell can be achieved by
designing the thickness of the shell to vary
symmetrically with respect to the middle surface. d-~(rN~, sin ~o) = (Z - X tan ~0)rd-~ ,
To ensure constant strength it is enough to
determine the proper thickness, while the form of N ~ Noo
R, ~ = Z. (89)
the middle surface does not undergo a change. The
second approach to the problem consists in
designing the thickness of the shell together with a where Z and X are the loads normal and tangential
change of its middle surface. The solution in the to the generator respectively.
142 S. A. EUKASIEWICZ

W e apply the condition of constant strength by


assuming that the material of the shell obeys the
T r e s c a yield criterion, and the condition of
which yields
minimum weight, which depends on the v o l u m e of
the shell.
1[ r~+r" )
The function z(r) can be determined f r o m the z' = - ~ / i r o + B + r" (93)
condition of minimum volume. Let us assume that
N~ <~No~ <~O. Then the stresses are represented where B is an arbitrary constant. The equation (93)
by the straight line DE of the hexagon Fig. 29.
can be integrated. We obtain

0"1 N~p~p z = -r2E[gd(arc sinhx/(ro+


rl B))' \(ro+B]
r--~/ w-]J + C

A
where
fro

~ dt

c/
r

gd(r)= cosht"

E(r) is an elliptic function. For r = 0 the angle of


slope depends on the constant B and is z ' =
0"2

/
m- +-(r~/ro+B) m. This angle has real values when
N~- r~, + B > 0. The thickness takes the following form

( h = p__Lr(2r~+n B + 2r:)"-'(r~+ r'-) '/-"


20"fl
The above solution holds valid only when 0 ~< X ~<
1. T h e r e f o r e the value of X must be defined. It can
(94)

E
be obtained f r o m the second equilibrium equation.
On transformation we have

FIG. 29. Yield conditions. r" 4r~,+ B + 4 r "


X = ro+r'2ro+B + 2 r " (95)

We write N . = - o-oh; Noo =-Xooh where X is which satisfies the condition 0~<X ~<1 when 0<~
a coefficient which should be 0 ~< X ~< 1, ~r0 is the r4<~ r~(r~+B/2). The constants B and C make it
tensile yield stress of the material. possible to determine the shape of the shell in
The radii of curvature of the middle surface R,, a c c o r d a n c e with the boundary conditions. An
R2 can be expressed by the function z ( r ) ; we have example has been given in Fig. 28.
Let us assume now that the height of the shell is
R , - (1 + z'") ~f" R : - r ( l + z'-')"" not determined and its edge lies on the line r = r,.
z" z' (90) The condition of minimum v o l u m e gives z' = 1. In
conclusion we o b s e r v e that the minimum value is
where obtained for the conical shape with the slope angle
o~ = 45 . The thickness should change as
z' = dz/dr.
_ p r~+ r 2
We have further; sin ~ = ( z ' / ( l + z'")""). On sub- h N/2o'0 r (96)
stitution into equilibrium equation we obtain the
thickness When the shell is subjected only to the concen-
trated force we have
po r2o+r"(l+z'") m ro- P (91)
2or0 r z' ~'po"
=-- + z' dr. (97)
0%
The v o l u m e of the shell is
The a b o v e functional is minimum w h e n z' = const.
The absolute minimum of v o l u m e is for z ' = 1,
,o (92) i.e. for the conical shell with thickness h =
P /~/ 2 7rcror.
The a b o v e functional has a minimum w h e n the In a similar way we can find solutions if the load
following Euler's equation is satisfied. is uniformly distributed in the horizontal plane.
Introduction of concentrated loads in plates and shells 143

The first equilibrium equation does not change


and therefore the thickness of the shell and v o l u m e
h = C VC d+z'2)exp
' z fJ + z')
are e x p r e s s e d by means of the same formula as
previously. The only change is in the equation for X. C is here an arbitrary constant and can be found by
A f t e r s o m e manipulation we obtain comparing formulae (91) and (99). For z / = 0 we
should obtain a result for the shell loaded only by
X = r2/(ro + r2) (98) the c o n c e n t r a t e d f o r c e P. The v o l u m e of the shell
is, on integration,
W e see that for arbitrary r the condition 0 ~< X <~ 1
is fulfilled. The shape and the thickness of the shell v = f" + z')dr-1] (I00)
are as previously. 3/ L o'oJo

If we consider a case when the concentrated If we assume that z' > 0 when 0<~ r ~< r~, then the
f o r c e P acts in the opposite direction to the
integral in the a b o v e expression is positive. The
distributed load, we assume that the force N,~ is a functional V has its minimum w h e n this integral is
c o m p r e s s i v e force and N0o is a tensile force. Such minimum. That gives z = const., i.e. the conical
state of stresses is presented by the straight line EF shell. The absolute minimum is when z ' = 1. The
in Fig. 29. thickness of the shell z' = 1 is
The shape of the shell is given in Fig. 30. We

h = ~ P exp 2~r. (101)


N/(2)TrCro o'o

The parameter X satisfies the condition 0 ~< X ~< 1 if


for z' = 1, r ~< cr0/'~.
The shape of the shell is similar to that given in
Fig. (27). When P = 0 and the shell is loaded only
by its own weight, we also obtain the minimum
weight for the conical shell. In order to c o m p a r e the
result obtained here with that taken f r o m R e f e r e n c e
104 we can transform expression (99) using the
second equation of equilibrium (89)
FIG. 30. Shell of minimum weight and uniform strength in
the case when a load P is oppositely directed to the h = hoexp f (3'z'-X - 1)dr. (102)
. i \cro r
resultant of the pressure po.
Assuming )( = 1 we find the known solution
o b s e r v e that the thickness of the shell grows to h = hoexp [(-~/tro)Z]; we see that only in this case
infinity when r--* r0. The reason for this p h e n o m e - w h e n P = 0 the thickness of the shell is limited at
non is that the radii of curvature R ~ 0 and R , ~ the top and the shell is not a conical one.
at this point. But the surface of this part of the N o w we consider a case when the force P acts in
cross-section of the shell is limited. We can imagine the opposite direction to the weight. Assuming that
that the shell is strengthened there by a ring of N,~ =Xcroh > 0 and No~ = - ( 1 -x)croh < 0 we find
certain cross-section, which carries the tension in on substitution into eq. (89)., the thickness
the parallel circle.
In the case of a uniformly distributed load in the h=C V,(I+z'2) [ ~ f {l--+z']dr]
horizontal plane we find the shape of the shell in a xr z, exp -~oo \z'
/ X J"
similar way. (103)
On integration the v o l u m e of the shell is

5.3 The shell carrying its own weight V = 2"a'c{l-exp [-~ofor' ( l + z ' ) ~ ] } .
Let us consider now a shell loaded by a (104)
c o n c e n t r a t e d normal f o r c e and by its own weight. If A s s u m i n g that z' > 0 when 0 ~< r ~< r~ we h a v e the
z/ is the specific weight of the material, the minimum of the functional V when the integral in
c o m p o n e n t s of load per unit area of the surface are eq. (104) has its minimum.
x = -~h sin ,p ; Z = - Z/h cos q~. P a r a m e t e r X can be evaluated f r o m the second
W e assume that both internal forces are com- equation of equilibrium. Introducing it into eq. (104)
pressive and N~ <~Noo <~0. T h e n this state of we find the following differential equation
stresses is e x p r e s s e d by the line DE in the
h e x a g o n - - F i g . (29). The N ~ = -her0, No~ = -xhcro,
where 0~<X ~< 1. On substitution into the first z'"[z'2(l+B)-2(l+B)-~-~r-2+tro
B~/r2]
=0fro
equilibrium condition (1) and integration (105)
144 S. A. EUKASIEWlCZ

where B is an arbitrary constant. This equation has 6. FINAL REMARKS


one solution z '2 = 0 which corresponds to infinitely
This paper presents solutions and results con-
thick, flat sheet. Then there remains the algebraic
cerning the distribution of stresses and displace-
equation of second order in respect of z'. On
ments in plates and shells under c o n c e n t r a t e d
solving it and integrating we find the shape of the
loads. The analysis was limited to those problems
shell. Let us consider now the optimum shape for
which can be described by the linear and geometri-
which the weight has absolute minimum.
cally non-linear theory of shells. T h e r e are many
F r o m the variational conditions for the edge
different problems c o n n e c t e d with the action of
r = r~ we find that B = 0. The solution takes the
concentrated loads on shells which are not discus-
form
sed here, though they are very important f r o m the

z' = - - -
o-0
+2 ) (106)
technical point of view. N a m e l y such problems as:
elasto-plastic behaviour of shells, fibered shells,
buckling of shells and optimum design of shells in
the case of more complicated loads, dynamic and
Since z ' > 0 we take the + sign.
contact problems, etc. Many of them still need to be
The shape of the shell can be obtained by solving
solved and others are not mentioned here because
eqs. (106), (103) and is given in Fig. (31). Intro-
of the limited scope of the paper.
ducing z' f r o m eq. (106) into eq. (89) we obtain the

REFERENCES
I. ZEMANIAN, A. H. Distribution Theory and Trans-
form Analysis, McGraw-Hill, New York, 196.
2. TIMOSHENKO, S., WOINOWSKY-KRIEGER. S. (1959)
Theory of Plates and Shells, McGraw-Hill, New
York.
3. NADAI, A. (1925) Elastische Platten, Springer,
Berlin.
4. WOINOWSKY-KRIEGER,S. (1933) Ing. Archly. 4,305.
5. GANOWICZ,R. (1966) Selected problems of Reisner's
theory of plates and sandwich plates, Mechan.
Teoretyczna i Stosowana 3, 4, 1966, O pewnym
rozwi~.zaniu p.l'yty tr6jwarstwowej, Rozprawy In-
ynierskie 3, 14. (In Polish).
6. MESMER, G. (1959) Uber eine Gruppe yon
Singularit~iten im Membranspannungzustand der
Kugelschale, Ing. Arch. 28, 208.
7. MARTIN, F. (1949) Die Membran-Kugelschale unter
Einzellasten, big. Arch. 17, 167.
8. VLASOV, V. Z. (1949) General Theory of Shells,
Moscow-Leningrad. (In Russian).
9. E. REISSNER, E. (1946) Stresses and small displace-
ments of shallow spherical shells, II, J, Math. Phys.
25, 279.
10. FLOGGE, W. and CONRAD, D. A. (1956) Singular
solutions in the theory of shallow shells, Stanford
FIG. 3 I. Shell of minimum weight carrying its own weight. Uniuersity Div. Eng. Mech. Tech. Report No. 101, 10.
II. KALNINS, A. (1966) On fundamental solutions and
coefficient X equal to 1 for r = 0 and )( = 0.4 for Green's function in the theory of elastic plates, J.
r = ~. It follows that the condition 0<~X ~< 1 is appl. Mech., 33, Trans. ASME, 88, ser. E, 31-38.
12. KALNINS,A. and NAHDI, P. M. (1960) Propagation of
always satisfied. axisymmetric waves in an unlimited elastic shell, J.
The results presented have been obtained by appl. Mech., 27, Trans. A S M E 82, ser. E, 690-695.
means of the equations of the m e m b r a n e theory of 13. NORDGREN,R. P. (1963) On the method of Green's
shells. T h e y are exact enough only if the curvature function of the thermoelastic theory of shallow
shells. Int. J. Eng. Sci. 1, 279-308.
changes are not too large nor too rapid. T h e r e f o r e
14. KOITER, W. T. 0963) A spherical shell under point
the results for the area near the top may not be loads at its poles. Progress in Applied Mechanics.
quite accurate. H o w e v e r , they enable us to get an The Prager Anniversary Volume, Macmillan, New
idea of the optimum shape of the shell. Direct York.
application of the results obtained may run across 15. LECKIE, F. A. (1961) Localized loads applied to
spherical shells, J. Mech. Eng. Sci. 3, l l l - l l 8 .
certain difficulties in practice. At large distances 16. HAVERS, A. (1935) Asymptotische Biegetheorie der
f r o m the top the thickness of the shell determined unbelasteten Kugelschale, Ing. Arch. 6. 282-308.
by the given formulae is very small and the 17. NAHDI, P. M. (1957) On the theory of thin elastic
manufacture of such a shell would be impossible shells, Quart. Appl. Math. 14, 331 (1956) and 14,
369-380.
for technical reasons. In these areas the shell may
18. WILKINSON, J. P. and KALNINS, A. (1965) On
b e c o m e buckled, which has been disregarded in non-symmetric dynamics problems of elastic spheri-
designing the shell. cal shells, J. appl. Mech. 32, Trans. ASME, 525-529.
Introduction of concentrated loads in plates and shells 145

19. BIJLAARD, P. P. (1956) On the stresses from local 40. VELICHKO,P. M. and SHEVCHENKO,V. P. (1969) On
loads in spherical pressure vessels, Bul. U.S. Weld. the action of concentrated forces and moments on a
Res. Comm. 34. shell of positive curvature, Izv. A N SSSR, 2. (In
20. CHINN, J. (1958) Influence charts for deflection and Russian.)
stresses in spherical shells, Ph.D. Thesis, Cornell 41. GOLDENVEIZER, A. L. (1954) On the problem of
University. concentrated loads on shells, Prikl. Mat. Mekh. 18.
21. WILKINSON,J. P. and KALNINS, A. (1966) Deforma- (In Russian.)
tion of open spherical shells under arbitrary located 42. FORSBERG,K. and FLUGGE,W. (1966) Point load on a
concentrated loads, J. appl. Mech. Trans. A S M E 6, shallow elliptic paraboloid, J. appl. Mech. 3,575-584.
305-312. 43. CHERNYSHEV, G. N. (1963) On action of concen-
22. EUKASlEWICZ, S. (1967) Concentrated loads on trated loads and moments on thin elastic shells of
shallow spherical shells. Quart. J. Mech. Appl. Math. arbitrary shape, Prikl. Mat. Mekh. 27, I. (In Russian.)
20, 393-305. 44. FEODOSJEV, V. I. and CHERNAKOV,S. N. (1966) On
23. LANGER,R. E. (1935) On the asymptotic solutions of action of concentrated forces on thin walled shell.
ordinary differential equations with reference fo the lnzh. Mekh. Tverd. Tela 6, 57-63, (In Russian).
Stoke's phenomenon about a singular point, Trans. 45. DAREVSKI/,V. M. (1966) Contact problems of theory
Am. Math. Soc. 37. of shells (action of concentrated loads on shells),
24. LECKIE, F. A. (1960) Solutions for the spherical shell Proc. of the VIth All Union Conference on Theory
subjected to axially symmetric loading, Syrup. of Plates and Shells, Izd. Nauka, 927-934, Moscow.
Nuclear Reactor Containment Buildings and Pres- (In Russian.)
sure Vessels, Glasgow. 46. EUKASIEWICZ, S. (1967) On the optimum design of
25. TOOTH, A. S. (1960) An experimental investigation of shells loaded by concentrated forces, Proc. IUTAM
shallow spherical domes subjected to a variety of Syrup. Copenhagen.
load actions, Symp. Nuclear Reactor Containment 47. EOKASlEVaCZ,S. (1967) The influence of the double
Buildings and Pressure Vessels, Glasgow. curvature on the rigidity of the shell of revolution.
26. KENEDI, R. M. (1960) Influence line methods of shell Arch. Bud. Masz. 14, 1.
analysis, Syrup. Nuclear Reactor Containment 48. MORLEY, L. S. D. (1960) The thin-walled circular
Buildings and Pressure Vessels, Glasgow. cylinder subjected to concentrated radial loads.
27. TOOTH A. S. and KENEDI, R. M. (1961) The influence Quart. Mech. Appl. Math. 13, 24-37.
line technique of shell analysis, Proc. Coll. Simpl. 49. YUAN, SHAO WEN (1946) Thin cylindrical shells
Calcul. Meth., Brussels. subjected to concentrated loads. Quart. Appl. Math.
28. MEDICK, M. A. (1962) On the initial response of a 4, 13.
spherical shell to a concentrated force. J. appl. 50. BIJLAARD,P. P. (1954) Stresses from radial loads in
Mech. 29, 689. cylindrical pressure vessels. Weld. J. 3, 615.
29. WESTERGAARD, H. M. (1933) Ingeni6ren, 32, 513, 51. HOFF, N. J., KEMPNER, J. and POHLE, F. V. (1954)
1929, Public Roads, 7, 25, 1926; 10, 65, 1929: 14, 185, Line load applied along generators of thin-walled
1933. circular cylindrical shells of finite length. Quart.
30. AINSO,H. and GOLDBERG, M. A. (1967) Analysis of a Appl. Math. 11,411.
shallow spherical shell under eccentrically applied 52. KEMPNER, J., SHENG, J. and POHLE, W. V. (1957)
concentrated load. Pressure vessel technology, Part Tables and curves for deformation and stresses in
I, Design and Analysis. Proceedings of the First circular cylindrical shells under localized Ioadings. J.
International Conference on Pressure Vessel Tech- Aero. Sci. 2, 119.
nology, Delft. 53. YUAN, S. W. and TING, L. (1957) On radial
31. ~UKASIEWICZ, S. (1969) The effect of transverse deflections of a cylinder subjected to equal and
shear deformations in the locally loaded spherical opposite concentrated radial loads, J. appl. Mech. 24,
shell. Arch. Bud. Maszyn 16, 4. 278.
32. EUKASIEWICZ,S. (1971) Equations of the technical 54. TING, L. and YUAN, S. W. (1958) on radial deflection
theory of shells with effect of transverse shear of a cylinder of finite length with various end
deformations, Quart. J. Appl. Math. conditions. J. Aero. Sci. 25, 230.
33. POLIAKOV,N. A., USTINOV, Ju. A. (1970) Investiga- 55. HOFF, N. J. (1959) The accuracy of Donnell's
tion of asymptotic behaviour of a solution of the approximatiofi for thin-walled circular cylinders. J.
theory of elasticity for a spherical shell under Mech. Appl. Math. 12, 89.
concentrated force, Proc. of the VIIth All-Union 56. HOFFMANN, R. F. and ARIMAN, T. (1971) Thermal
Conference on Theory of Plates and Shells, Nauka, and mechanical stresses in nuclear reactor vessels,
Moscow (In Russian). Symp. Nuclear. React. Containment Buildings and
34. J~UKASIEWICZ,S. (1965) Concentrated loadings on Pressure Vessels, Berlin.
shells. Proc. X I I Int. Congr. Appl. Mech. Springer- 57. HOFF, N. J. (1954) Boundary value problems of the
Verlag, Berlin. thin-walled circular cylinder, 3. appl. Mech. 21,343.
35. FLOGGE, W. (1965) Concentrated force on shells, 58. }-IOLAND, I. (1962) Characteristic equations in the
Proc. of the XII Int. Congr. Appl. Mechn. Springer- theory of circular cylindrical shells. Aero. Quart. 13.
Verlag, Berlin. 59. BIJLAARD,P. P. (1955) Stresses from local Ioadings
36. J~UKASIEWICZ,S. (1967) Local loads on shells, Zesyty in cylindrical pressure vessels. Trans. ASME, 77, 6.
Naukowe Politechniki Warszawskiej, Mechanika 60. BIJLAARD,P. P. (1955) Stresses from radial loads and
21. (In Polish.) external moments in cylindrical pressure vessels.
37. EUKASIEWICZ,S. and STUPNICKI, J. (1971) Effect of Weld. J. 34, 12.
transverse shear deformations in the locally loaded 61. NAUBEREIT,G. (1963) Einflussfl~ichen der Kreiszylin-
shell. Arch. Bud. Maszyn 17, 2. der Schale belastet durch Einzelmomente, Wiss.
38. KWANG-CHIEN, HO and FU, CHEN (1963) A Zeitsch. der Techn. Universit~it Dresden, Heft 1,
simplified method for calculating double curvature 147-156.
shallow shells under the action of concentrated 62. BIEGER, K. W. (1961) Die Kreiszylinderschale unter
loads. Acta Mech. Sinica 6, 19--37. konzentrierten Belastungen, lng. Arch. 30, 57-62.
39. LUKASIEWICZ, S. (1968) The solution for concen- Dissertation "EinflussflS.chen der Kreiszylinder-
trated loads on shells by means of Thomson schalen." T.U., Berlin, 1959.
functions. Z A M M 48, 247-254. 63. KHRUSTENKO,S. A. (1962) On action of concentrated
146 S . A . J~UKASIEWICZ

loads on orthotropic cylindrical shell, Proc. of the and stability of spherical shells subjected to concen-
and All-Union Conference on Theory of Plates trated loads at the apex. Proc. of the Fourth U.S.
and Shells, Kijev. (In Russian.) Nat. Cong. Appl. Mech. ASME, 563-575.
64. SHESTOPAL, K. (1963) Equilibrium of cylindrical 83. Voss, W. C., PEAaODY, D., STALEY, H. R. And
shells under concentrated loads, Ukr. Mat. Zhurnal D1ETZ, A. G. H. (1947) Thin shelled domes loaded
15, 106-114. (In Russian.) eccentrically. Proc. ASCE, 73, 1173-1195.
65. AKHONINA,V. (1962) Application of trigonometrical 84. ASHWELL, D. G. (1960) On large deflections of a
series to solution of cylindrical shells under concen- spherical shell with an inward point load, Proc. of
trated loads, Cbornik Trudov Leningradskogo Vni- IUTAM Symp. Theory Thin Elastic Shells Delft,
versiteta, Inst. Transporta 190 (In Russian) 1959, pp. 43-63. North Holland, Amsterdam.
66. KLEIN, B. (1958) Effect of local Ioadings in 85. THURSTON, G. A. And PENNING, F. A. (1964) Effect
pressurized circular cylindrical shells. Aircraft Eng. of initial imperfections on the stability of shallow
30, 356-36 I. spherical shells, AFOSR, Sci. Report No. 64, 1627,
67. HOLAND, J. (1961) Influence surfaces for bending August, 1964.
moments in circular cylindrical shells or curved 86. ARCHER, R. R. (1962) On the numerical solutions of
plates. Int. Assoc., Bridge and Struct. Eng. Public. the non-linear equations for shells of revolution, J.
21. Math. Phys. 41, 165-178.
68. HALS, T. E. (1966) Circular cylindrical shells 87. PENNING, F. A. And THURSTON, G. A. (1965) The
subjected to concentrated loads, lnstitutt for Statick, stability of shallow spherical shells under concen-
Norges Tekniske Hogeskole, February .Trondheim. trated load. NASA CR-265, July, 1965.
69. LITOVTSEV,Ju. V. (1968) On stability of visco-elastic 88. PENNING, F. A. (1966) Experimental buckling modes
shells under local loads. M.T.T. 5. (In Russian) of clamped shallow shells under concentrated load.
70. RODIONOVA. V. A. (1968) Solution for cylindrical J. appl. Mech. 33, 297-304.
shell under axial compression and concentrated 89. BUSHNELL. D. (1967) Bifurmaction phenomena in
loads. Vestnik Leningradskogo Unit, ersiteta 13. (In spherical shells under concentrated and ring loads,
Russian.) AIAA Jour. vol. 5, No. 11, 2035.
71. SHEVLJAKOV,JU. JA. and SHEVCHENKO,V. P. (1968) 90. POGORELOV, A. V. (1967) Geometrical Methods in
Local stresses in cylindrical shells in the vicinity of Nonlinear Theory of Elastic Shells, Nauka, Moscow.
concentrated loads. Gidroaeromekh. i Teorija Up- (In Russian.)
rugosti 1, 6. (In Russian.) 91. ERNST, L. J. (1973) De draagwerking van dubbelgek-
72. ZHIGALKO. Jv. P. (1966) Action of concentrated rande schalen, belast met een puntlast, Deel I, II
forces on orthotropic and bimetallic cylindrical shell. Rapport 8-73-4 Stevin-Laboratorium Delft.
Proc. of the Vlth All-Union Conference on Theory 92. EUKASlEWlCZ, S. (1969) On shell loaded at its free
of Plates and Shells, Moscow. (In Russian.) edge. Arch. Budowy Mas,.yn 16, 3.
73. JAHANSHAHI,A. (1963) Force singularities of shallow 93. SEIDE, P. (1961) On the bending of circular
cylindrical shells. J. appl. Mech. 30, 343. cylindrical shells by equal and equally spaced end
74. VELICHKO, P. V., SHEVL.IAKOV, T. A. and SHEV- radial shear forces and moments, J. appl. Mech. 28,
CHENKO, V. P. (1970) Stress-deformation state of 1.
plates and shells under concentrated loads, Proc. of 94. EUKASIEWICZ, S. and BORAJKIEWICZ, W. (1971)
the VIIth All-Union Conference on Theory of Plates Optimum design of a bar transmitting a load into a
and Shells, Nauka, Moscow. (In Russian.) sheet. Arch. Bud. Maszyn 18, I.
75. ARTJUKHIN,JU. P. (1966) Solution for homogeneous 95. KAPKOWSKI,J. (1960) Introduction of a concentrated
and sandwich orthotropic: shells on local loads, force in a sheet by the condition of uniform strength,
Sbornik "Issledovanija po Teorii Plastin i Arch. Bud. Masz. 1, 7. (In Polish.)
Obolochek, 4, Izd. Kazanskogo Universiteta. (In 96. DRUCKER,D. C. And SHIELD, R. T. (1957) Bounds on
Russian.) minimum weight design, Quart. Appl. Math. 15,
76. DORI~, R. And FLOGGE, W. (1973) Singular solutions 269-281.
for shallow cylindrical shells. J. Appl. Mech., vol. 97. PRAGER, W. (1958) On a problem of optimal design,
40, Ser. E, No. 1. Proc. Syrup. Non-Homogeneity in Elasticity and
77. BERRY, J. G. (1957) On thin hemispherical shells Plasticity, Warsaw.
subjected to concentrated edge moments and forces, 98. SHCH. MIKELADZE,M. (1957) On minimum weight of
Proc. 3rd midwest Conf. on Solid Mechanics. anisotropic shells, Soobschchenia Gruzinskoi
78. FL(JGGE, W. (1960) Stresses in Shells, Springer- A N SSSR, 19, 1. (In Russion.)
Verlag, Berlin. 99. ONAT, E. J., SCHUMANN,W. and SHIELD, R. T. (1957)
79. NOVOZHILOV,V. V. And CHERNYKH, K. F. (1963) On Design of circular plates for minimum weight.
solution of shell under concentrated loads, Sbornik Z A M P 8, 485-499.
"Issledovanija po uprugosti i plastichnosti'. Lenin- 100. BRZOSKA, Z. (1954) Circular plates of uniform
gradsk. Univ. 2. (In Russian.) strength under axisymmetrical load, Arch. Bud.
80. BIEZENO, C. B. (1935) l]ber die Bestimmung der Masz. 1, I. (In Polish.)
Durchschlagkraft einer schwachgekrummten kreis- 101. EUKASIEWICZ, S. (1960) Circular plates of uniform
formigen Platte, Z. Angew. Math. Mechn. 15, 10, C. strength axisymmetrically loaded. Arch. Bud. Masz.
B. BIEZENO and R. GRAMMEL: Engineering 1, 7. (In Polish).
Dynamics, 484, London (1965). 2. Blackie, London, 102. TOLKE, F. (1939) 0 b e r Rotationsschalen Gleicher
484, (1956). Festigkeit for konstanten Innen oder Aussendruck.
81. WEI-ZONG,CHIEN and CHANG, HU HAl (1957) On the Z A M M 19, 338-343.
shaping of a thin spherical cap, 9th Int. Congr. Appl. 103. MEGAREUS, G. (1939) Die Kuppel gleicher Festig-
Mech. 6, University of Brussels, 309. keit. Bauing. 20, 232-234.
82. EVAN-IVANOWSKI,R. M., CHENG, H. S. And Lop, T. 104. ZIEGLER, H. (1958) Kuppeln gleicher Festigkeit, Ing.
C., (1962) Experimental investigation of deformation Arch. 26, 378-382.

Vous aimerez peut-être aussi