Design and Analysis of Shell Structures (3-4-5) PDF
Design and Analysis of Shell Structures (3-4-5) PDF
Chapter3
3.1 - Introduction
Cylindrical shell fOnDS are used in water and gas retaining structures, circular silos, pipes,
pressure vessels, and cylindrical vaulted shell roofs of various kinds. This chapter studies
the membrane behavior of cylindrical shells. First, we derive the governing membrane
equations. Then, we apply these equations to the analysis of various types of cylindrical
shells, including vessels, pipes, and vaults. The results of the present chapter will also be
used in more comprehensive analysis and design of cylindrical shells treated in future
chapters. Some basic design considerations related to such shells will also be presented.
26 Design and Analysis of Shell Structures
A cylindrical surface is generated by sliding aplane generating curve (the plane of which
remains parallel to a fixed plane) along a straight line called the directrix of the
cylinder.The generating curve of a cylindrical shell may be a c10sed profile (pipes) or open
(vaults). In the particular case of a circular (i.e., constant curvature) profile, the resulting
shell is called a circular cylindrical shell.
The mid-surface of a cylindrical shell is a surface with single curvature. The curvature of a
generating curve may vary along the profile curve. Any point on the mid-surface of a
cylindrical shell can be identified by two coordinate parameters. These two parameters are
the longitudinal distance from some arbitrarily chosen origin, x, and the angle <p between the
normal to the shell at that point and a reference normal at some chosen origin, figure (3-1).
Consider an element of a cylindrical shell, such as the one shown in figure(3-1). Let (x, <p)
designate the coordinates of a corner point of this element. The free body diagram of this
element is shown in figure (3-2). In this figure, the r direction identifies the normal to the
shell surface at the point (x,<p). The symbols Nx, Ncp. Ncpx, and Nxcp denote the unknown
membrane forces in the shell; these have units of force per length. The known quantities
Px, P" and Pr, having units of force per area, represent the intensity of the applied
distributed forces in the x, <p, and r directions, respectively. All these parameters are
functions of the position coordinates x and <p.
Membrane Behavior oE Cylindrical Shells 27
7fx = 0
~
aNx
()x dx. o
Equilibrium along the tangent to the shell, the <I> direction, yields:
H<j> = 0
3N4> dN 4>
~ d<j> • dx + ax
X
dx. rd.p + P<D • dx. rd~ = 0
Finally, by writing the equilibrium relation in the r-direction, i.e., along the normal to the
shell, we obtain:
Lf = 0
r
o
Dividing both sides of these equations by the arbitrarily small, but nonzero, length
parameters ds and r d<l>, we finally obtain the following membrane equations of equilibrium
for cylindrical shells.
28 Design and Analysis of Shell Structures
N
q, = Pr r (5-1)
3Nxep 1 dN~
3):(= - P<j> ---- (5-2)
r "'1>
dN
x
3x (5-3)
These equations can be integrated in a sequential manner to yield the membrane force field
functions Nq,. Nq,x, and Nx. The general integral expressions are,
N = rP (5-4a)
4> r
aN
N
xep
=- J (P4> +.l..r _.~
34>
)dx + fl (ep) (5-4b)
N
x
= - J(P x + 1 dNXt!l
--~-)dx + f2(ep)
r oGl (5-4c)
The functions f 1 (<I» and f2 (<I», resulting from integration, depend on the boundary
conditions of the particular shell. They can be uniquely obtained for each individual shell
problem. In the following seetions, we will apply these expressions to some shell analysis
problems.
Pep = PSincp
Hence, using the expressions (3-4), and using the condition of Nx = 0 at the ends we find
the following membrane field:
N.(> =- Pa Cost>
N
c;,x =- 2P x Sin:;. (3-5)
N
x
=- ~ {~2
4a
- 4x 2 )Cos4>
Figure (3-3) shows the variations of internal membrane forces in an arbitrary section,
located at a distance x.
Membrane Behavior of Cylindrical Shells 29
Referring to these graphs, and expressions, we can make the following observations:
(a) At the longitudinal edges of the shell, i.e., at <11= ± 1t /2, we have: Ncp=O. This means that
this shell has no need of longitudinal normal support.
(b) At the longitudinal edges (<11= ±1t /2), the membrane shear force in the shell is Nxcp =
2px. This means that there is a need for a longitudinal edge member to absorb this force
and to transfer it to the end diaphragms. This edge member is shown in figure (3-4).
The longitudinal edge member required by the membrane theory is called an edge beam.
Figure (3-3) shows the mechanism of shear load transfer from the shell to the edge beam.
The shear force in the shell, upon its transfer to the edge beam, becomes an axial tension in
the beam. The magnitude ofaxial tension force in the edge beam at any section is obtained
by integration of edge shears along the edge. 1ts expression is:
H =J~.\!, N
xlji
dx =- 2PJX xdx
R.
= !4 p(.\!,2 - 4x 2 ) (3-6)
"2 2"
At the ends of the shell,Le., at x=+ I /2, we have Nx=O, but NX<jl= -pI sin cp. This means that
part of the applied load is transferred to the end diaphragm by the shear force mechanism.
30 Design and Analysis of Shell Structures
Px = 0 , P~ = 0 P = y(H - x) (3-7)
r
N = 0 (3-8a)
x
xcp = 0
N (3-8b)
,·t[- ., . F,~\-
tI
+
L@ __
I - _ l _ _'_x-.-, I
L[W--
a~ N.
L 1 I ~
The pressure head, measured at the level of cylinder axis, is assumed to be Po. Thus the
components of applied loading are
(3-lOa)
Membrane Behavior of Cylindrical Shells 31
'(2
X2
Cos~ - ax df 1 (CP)
d<p + f 2 (CP)
The unknown functions f 1(</» and f 2 (</» are to be detennined from the boundary conditions:
(N )
x x=0
= 0 (N)
x x="
n =0
The constant C, representing the non-existing resultant end torsional couple may be set
equal to zero. Therefore, the solution to this problem is
N = P ä - ya 2 Cos<p (3-11a)
<p 0
Nx = - ~ x (~ - x)Cos<P
(3-11c)
If this cylinder were viewed as a horizontal beam, then the derived expressions for Nx and
NXq> could be interpreted as the longitudinal bending stress and shear stress, respectively.
Figure (3-5) shows that Nx is linear in the sectional height.
In order to obtain the membrane displacement field in a cylindrical shell, with already
known internal forces, we use the constitutive and the shell kinematic relations. We assume
that the shell is made of a linearly elastic and isotropie material which obeys Hooke's law;
the two elastic constants are Young's modulus E and Poisson's ratio v.
32 Design and Analysis of Shell Structures
For a linearly elastie and isotropie cylindrieal shell of thiekness t the constitutive relations
are:
EX ~t lN x - vN;j»
E\jJ it-C
1
N<j> - vN) (3-12)
20+'J) N
YXtj! = Et x,p
To derive the kinematie relations we eonsider a displaeed configuration of the shell element,
as shown in figure (3-6b). If we denote the axial strain by ex, the transverse strain by E</>,
and the shear strain by Y</>x, then we can write the linear strain-displacement relations as:
dU
EX dX
1 dV
E\jJ = r(d\jJ + w) (3-13)
_ dV + 1 dU
Yx\jJ - dX r ä(j)
dU = ..!-_ (N - \JN )
3x rt x ~
dV 1 dU 2 (1 + v) '1 (3-14)
dX + r a~ = ~-t- "x'~
.:::!.... + ~ 3v = l... (Ci - \l}; )
r r dtP Et 4> x
Assuming that the membrane forces Nx, N</>, Nx</> are already detennined, we Can integrate
the relations (3-14) in a sequential manner, to obtain
Membrane Behavior of Cylindrical Shells 33
The functions f3 ($) and f4 ($) in the integration process can be determined for each shell
from specific boundary conditions.
Return to the circu1ar cy1indrica1 shell of figure (3-4). We wou1d 1ike to determine the
displacement fie1d in the shell under a dead weight 10ading of intensity p. As specific
boundary conditions, we assurne that at x=+ 1/2 we have v=Ü and w=Ü.
If we substitute the membrane force expressions (3-5) into (3-15) we obtain the following:
(3-16)
Note that the constraint on the radial displacement, w, at x=+l /2 vio1ates the requirements
of the membrane action of the shell at that section. As a resu1t, some ben ding will develop
around that region, and can be predicted by the bending theory of cylindrical shells.
Solution
Evaluating the expressions (3-16), for the shell displacements, and substituting the assumed
nurnerical values in the resulting expressions, we obtain
34 Design and Analysis of Shell Structures
Uo = Ik (0) = 0
vo = 0
=0.06 cm = O.6mm
As we observe, the resulting displacement is very small.It would be interesting to compare
this value with a relatively high lateral deflection of a beam having the same span and
loading.
Whether the cylindrical shell has mostly the beam action or the arch action depends on the
shell geometry and the edge conditions. For example, longitudinal supports transfer the
applied forces in the longitudinal direction.
neutral plane
Long cylindrical shells resting on end supports act like simply supported beams. A long
shell acts like a beam, having as its section the shell profile, and as its span the span of the
shell. The extern al bending moment at each section is supported by the resultant of
compressive longitudinal membrane forces acting as the compression flange. and the
tension in the edge beam acting as the tensionflange, figure (3-7).
Membrane Behavior of Cylindrical Shells 35
As we have seen in our membrane analysis of long vaults, the applied forces are transferred
longitudinally to the end supports. The transfer of internaiload to the end diaphragms or
end arches takes place by the internal shear force mechanism, as shown in figure (3-8).
I
i
I
.I
~
Figure (3-8) Shear force mechanism of applied fOI·ces to the end
arches in a cy lindrical vault
In an overall beam action of the long shell, part of the applied load could be transferred by
internal shear forces to the longitudinal beams, as demonstrated in figure (3-9). However, if
there are no edge beams, then because the free edge situation is incompatible with the
requirements of the membrane theory, some bending field of forces will develop along the
longitudinal edges
the shell
The Ion ger the shell is, the stronger is the beam action of the shell. On the other hand, in a
short shell, the so-called arch action is more effective and the applied load is mostly
transferred by the trans verse arches.
Figure (3-10) shows the longitudinal internal stresses in two end suppOited cylinders, one
long and the other short. As we see, the behavior of the longer shell is c10ser to the action
of a simple beam, whereas that of the shorter shell is different.
36 Design and Analysis of Shell Structures
p p
tens\J~Jll
Il@
tension tension
So far, in our qualitative presentation of cylindrical shells behavior, we have discussed the
membrane action of the shell. As we pointed out earlier, in chapter 2, some bending fields
can develop in shells. For example, end supports (diaphragms or arches) give rise to
internal bending forces. In these cases the ben ding force field is needed to satisfy
compatibility. The region of influence of the bending field depend on the end conditions,
the loading, the thickness, and the length of the cylindrical shell.
Figure (3-11) shows how the behavior of long and short shells are influenced by the
bending field. The effect of the ben ding field, caused by the end diaphragms, is local and
confmed to the ends of a long shell. In shorter shells, the influence of bending field is more
pronounced. For very short shells, the bending field penetrates throughout the shell and
affects the "whoie system".
Figures (3-12) summarize the cylindrical shell behavioral patterns. These figures
demonstrate c1early the beam action, figure(3-12a) and (3-12b); the arch action, figure (3-
12c); the bending fields developed by the stiffening longitudinal beam, figure (3-12a), and
end arches, figure (3-12c).
Membrane Behavior of Cylindrical Shells 37
The qualitative understanding of the cylindrical shell behavior, arrived at in this seetion, is
extremely useful in cylindrical shell design and analysis. The quantitative analysis of
cylindrical shells for bending effects will be carried out in the next chapter.
38 Design and Analysis of Shell Structures
Problems
P 3.1 - Consider a cylindrical vessel with elliptical profile as shown in figure (P 3.1). The
vessel is in a horizontal position and is simply supported at its two ends by means of end
rings or end diaphragms. Determine the internal forces in this shell for Px =0 and arbitrary
distributions of pr and Pr. Show that the resulting force field has the following form:
1 ClNcp
Nx ,.
'f'
=- x(P,.
'f'
+ -_)
a Clcp
N = - ...!. 4X2)~ (P ClNp
x 8a
(R,2 _
dcp cP + a1 Clcp )
Figure (P 3-1)
P 3.2 - Consider a cylindrical vault having a parabolic profile as shown in figure (P 3-2).
The equation of the profile curve is
r;;2 = 2r E;.
o
Figure (P 3-2)
Membrane Behavior of Cylindrical Shells 39
wherein, ro is the radius of curvature of the parabola at the origin. At any other point, the
radius of curvature can be expressed as
r = [1 + (d~/ds)2]3/2
(d2~)/(dsL)
and since
so rO
tanljJ = ds r = ---,,-
d~ Sin 3 Q
P = 0
x
r Px = 0
I PA = - P SintCbsljJ
<;J
1 Pr = - P Sin 2 1jJ
(3) Lateral wind load of intensity p:
Show that the internal force field for these loading systems are as folIows:
Px
Nx - 2r (9. - x)Sin~.j;
o
40 Design and Analysis of Shell Structures
r
o
Nq, p SinljJ
N = 0
x
Interpret these results and note that for snow loading, parabolic arches are funicular and
hence the load is carried mainly in the transverse direction. By plotting the membrane force
field for dead loading, reproduce the following figure:
T -
"
N,p., • =l
,
,-1 .....
"
Note that the effect of shell on the end diaphragm is in apparent contradiction to intuition.
Verify this result and carry out a physical interpretation with the help of the following
diagrams.
Membrane Behavior of Cylindrical Shells 41
3.4 - A vertical chimney of circular cross section and radius a is subjected to a wind load
defined by Px = 0, P~ = 0, Pr = q Cos 8. Here 8 is the horizontal angle and q is the
intensity. of wind load. Find the membrane stress resultants in this chimney shell. Compare
the results with those obtained by classical cantilever beam theory.
3.5 - Consider a simply supported shell vault with a catenary cross section. The equation of
the middle surface in the cross-sectional coordinate system y-z is given by,
z = - a ( 1 - 85 h y / a)
Obtain the membrane stress resultants in the shell under uniformly distributed dead loading
of intensity p.
42 Design and Analysis of Shell Structures
3.2 - A. M. Haas, Design o/Thin Concrete Shells, John Wiley & Sons, New York, 1962
3.4 - A. Pflüger, Elementary Statics 0/ Shells, 2nd Edition, McGraw-Hill Book Co., N.Y.,
1961
3.5 - M. Salvadori and R. Heller, Structure in Architecture, Prentice-Hall, INe., N.J., 1963
Bending Analysis of Circular Cylindrical Shells 43
Chapter 4
4.1 - Introduction
A complete analysis of various cylindrical shell forms, for silos, press ure vessels,
containment shells, containers, and shell roofs, would require an appropriate bending
theory. A general bending theory would embody the membrane theory of shells. It would
also predict the ben ding action of the shell. This chapter introduces a general bending
theory of circular cylindrical shells. The theoretical basis of the present chapter forms the
foundation of approximate bending theories for cylindrical shells. The special theories for
cylinders, such as axisymmetric cylindrical shell theory and the bending theories of
cylindrical vaulted roofs, developed in future chapters, can be considered as the offspring of
the general theory presented in this chapter. The theoretical developments in this chapter
will lead to a set of useful relations for treating a variety of practical shell analyses and
design problems.
44 Design and Analysis of Shell Structures
a () () .
~=
M ••
The equations of equilibrium of internal and external forces, along the x, <1>, and r axes, are
as follows:
u'X + NA,
'+'x
+ aP
X
= 0
(4-1a)
N + N'
<jl x<jl
- Q + aP
<jl <jl
=0
(4-1b)
Q<jl + ~ + N.p - aPr = 0 (4-1c)
Bending Analysis of Circular Cylindrical Shells 45
Compared with the membrane equations for cylindrical shells, these equations contain extra
terms emanating from the bending effects. When the bending field comes into play they are
needed for providing the equilibrium and / or compatibility to the shell elements.
M' + M
x lIx
- aO
-X
=0
(4-2b)
aN -aN +M =0
x<j> ~x q,x
(4-2c)
Note that, due to the presence of the twisting couple M<jlx. the membrane shear forces NX<jl
and N<jlx need no longer be equal.
If we eliminate the two shear force functions ox and 0<j> from the two sets of equations (4-
1) and (4-2), we obtain the following equilibrium equations:
(4-3a)
N~ + Nq,x + aPx = 0
aN
~
+ aN'
x(j,
- t-l
~
- M'
x~
+ a2p
<P
=0 (4-3b)
M + M"
ep xep + M"+x + ll"x + aN
ep - a 2p
r
= 0 (4-3c)
o
(4-3d)
There are ten unknown internal force components (N x • N<jl, NX<jl. N<jlx. Mx, M<!>. M<jlx. Mx<!>. Ox,
O<!», on the one hand, and six equilibrium equations, on the other. Therefore, the shell is
statically indeterminate and hence requires extra relations for its analysis.
To derive the kinematical relations for circular cylinders, we use the basic kinematic
assumptions outlined in section 2.5. If we denote the components of displacement field of
the shell middle-plane by u,v,w, then we can express the displacement of an arbitrary point
located at a distance z from the middle-surface in terms of these components. An outward
radial displacement, w, is considered positive.
Figures (4-2a) and (4-2b) show the undeformed and the deformed longitudinal and
transverse sections of the shell element. In these figures, points Ao and A represent typical
material points originally located on the shell middle-surface, and a distance z from the
middle-surface, respectively. The components of displacement of the material point Aare
denoted by UA, v A, W A.
46 Design and Analysis of Shell Structures
~I-u" -l I undeformed
a I--u-l
t (a) longitudinal section (b) transverse section
2
Using the basic kinematic assumptions of section 2.5, we can express the displacement
components of an arbitrary material point located at a distance z in tenns of the middle-
surface displacement components as folIows:
u A= U - -a
w'
:z (4-4)
a + z w
l.=-a- v --a z (4-5)
w*
A
=w (4-6)
Having found the displacement relationships, we can now fonnulate the strain-displacement
relations. Tbe components of strain in cylindrical coordinates can be expressed as
(4-7a)
v* + w*
A A
a + z (4-7b)
Using relations (4-4), (4-5), and (4-6) we ean write these as finally:
w" (4-8a)
E = ~- z~
x a
z Iv W
E p =~-
a
---+--
aa+z a+z
(4-8b)
Yx4J - u
- +a -+a-2zv ,
a + z
\V'
-(
a
Z
a
+ Z
a + z
) (4-8e)
These relations eonstitute the kinematie relations for eireular eylindrieal shells.
Throughout this book, we shall assume that the shells under eonsideration are eomposed of
isotropie materials having linearly elastie behavior. Therefore, assuming a generalized plane
state of stress, we may write the stress-strain relations, at any point of the shell as folIows:
(4-9a)
I; = -E- (E + VE,j)
K l-v 2 X 'T'
\j = _E_ (E + VE ) (4-9b)
,~ 1 - V2 4> x
G" _ E
KejI - 2 (1 + v) Yxq, (4-ge)
We can express these eonstitutive relations in terms of internal force resultants and the
displaeement eomponents of the middle surfaee. In order to do so, we use the set of
relations (2-1), (4-8), and (4-9). The eombined eonstitutive and kinematic relations are:
K
N
4>
=.-2.
a
(v + w + V u') + - 3 (Iv + w)
a (4-lOa)
N
x
= ~(u' + ,v + VVl) - -a~- w"
:1 (4-lOb)
(4-lOe)
(4-lOe)
K
MijJx = az (1 - v) (w'· + ~ - ~')
48 Design and Analysis of Shell Structures
N
q,x
= J2...
a
1 - v (u· +
2
V ') + K 1 - v (.
aT -2- u - w ' ·) (4-lOg)
Kxq, = aJ 1 -v
-2- (u + v') + aK1-,)
3- - 2 - (v' - w ' ·) (4-lOh)
K = [t 3
(4-11)
12(1 - v 2 )
From a physical point of view, the parameter K signifies the bending stiffness and D the
membrane stiffness of the shell. Therefore, all terms containing K represent contributions
of the bending field to the total field. So, if we assume no ben ding stiffness for the shell,
i.e., if we disregard the terms containing K, in the above relations, we shall obtain the
constitutive relations for a membrane elastic cylindrical shell.
The shell constitutive relations (4-10) can be cast in a more appealing and yet compact
form. In looking back to relations (4-8) we recognize the middle-surface strain field, as
defined by Ex, Ecp, and 'Yxcp to be related to middle-surface displacement components as
follows:
v+w (4-12)
E:<p = - a -
Also, the terms containing thickness variable, z, are recognized to be contributions from the
sectional curvature change and twist. Hence, using relations (4-8) we can write the change
in the shelliocal curvature as weIl as the local twist in terms of parameters 'l'x. 'l'cp. and 'l'xcp
as:
w" (4-13a)
= a2
Now, if we insert the expressions (4-12) and (4-13) in relations (4-10), we obtain the
following constitutive relations:
(4-14a)
(4-14b)
N
x
= D (EX + vE..,)
'I'
- JS....~,
a x
(4-14c)
N = D(l - v) - K(l - v) (1jJ + "Y'x<P )
~x 2 Yx<p + 2a x<p 2a
(4-14e)
(4-14f)
(4-14g)
M = K(l - v) 1jJ
cjJx xq,
- (4-14h)
yx~
Mx</> = K(l - v) (iJix<P - 2a )
These relations, are a set of fairly complete but useful forms of shell constitutive relations
expressed in terms of internal force resultants. They also embody the kinematic relations.
In these expressions, the terms containing K signify the contributions arising from the
bending action of the shell.
50 Design and Anal ysis of Shell Structures
The eoIlection of equilibrium, kinematie, and eonstitutive equations, (relations (4-3) and (4-
10», eonstitute the field equations of bending theory of cireular eylindrical shells. In
passing, we note that by using the expressions for Nxcp , Ncpx, and Mcpx we can satisfy the last
equation of (4-3) identieally. Hence, the total of 11 equations eorrespond weIl with the total
of 11 unknowns (internal forees and displacements).
By eombining the relations (4-3) and (4-10), we can obtain the displacement equations
governing the bending behavior of circular eylindrieal shells. So, by substituting (4-10) into
(4-3), and making some elementary manipulations, we obtain
11 1 - v .• 1 + \J ,'. , k [1 - v .. '"
u +-2-u+-2-v "".1W + --2-u-w (4-15a)
a2
"'J+ -0---
P
+I
--2v- w x - 0
1 + \J .
-2- u '+v + 1 ; \ \." _ ,.:, + k [ ; (l - \1)\1"
(4-15b)
P~a~
3 - \. ".1
- 2 - \.' - --- 0
:J
The parameter k in these equations represents the ratio of bending to membrane stiffness. It
has the following definition:
(4-16)
The displacement equations (4-15) governing the ben ding behavior of eircular cylindrieal
shells are quite general. These equations ean be solved by numerical sehern es or, in some
special cases, by analytieal means. However, eertain simplifieations ean be made 10 make
them more useful for praetical analysis and design. If we assurne that the shell is relatively
thin, then we may only retain those terms containing the parameter k which are of higher
gradients, i.e., higher order of differentiation with respeet to the x and <l> variables. If we do
so, we obtain the more simplified equations which read:
Bending Analysis of Circular Cylindrical Shells 51
P a2 (4-17a)
u" +l--2
v" l+v "
- u +-2- v + \JW' +~
0
0
P a2 (4-17b)
1 ... v " - v v " +W
+v +1- 2 + 4>
0
- 2 - -u - ])
P a2 (4-17c)
\u· + v+ W+ r
k (w'''' + 2w"" + w) - - 0 - = 0
These equations can be simplified further. We shall discuss special theories based on these
equations in the forthcoming sections and chapters.
The equilibrium equations for the axisymmetric theory of cylinders are thus:
fi' + aP = 0
x x
(4-18a)
:-J + 0' - aP = 0
'.j> -x r (4-18b)
1'1' - aO = 0 (4-18c)
x -x
52 Design and Analysis of Shell Structures
Tbe last two equations, representing the axisymmetric bending (now decoupled from the
axial deformation) may be combined to yield the single equation:
aN
I, + M"x - a 2 Pr = 0
1 1 d"!-1 x 1 (4-19)
~=> Pr = -. M"
a":< + -N.
a . , = dx 2 + a N'"
'P
(4-20a)
N<jJ = a1 D (w + \IU')
(4-20b)
N
x
= 1..-
a
D(u' + \'w)
(4·20c)
1 (4-20d)
K(w + vw")
a2
(4-20e)
M
x
= 1_
a2
K(w" - u')
(4-20t)
M = M = 0
4Jx x4J
In order to derive a single differential equation on the radial displacement, we decouple the
axial effects from radial bending by assuming that Px = O. By assuming a constant
thickness shell, we obtain
(4-21)
N
<p
= - a1 Dw(l' v )
2
(4-22)
M
Ij>
= vr1
x
M
a
x = 12 Kw"
Bending Analysis of Circular Cylindrical Shells 53
(4-23)
(4-24)
This fourth order differential equation resembles the equation of a beam resting on an
elastic foundation and laterally loaded by the distributed force Pr. From a physical
viewpoint, the equivalent elastic foundation effect comes from the inclined hoop force, Ncp.
acting on a longitudinal" strip" of the shell in its radial beam-type behavior.
In this chapter, as in other chapters, we have treated static problems for shells. To deal with
dynamic problems we must include the inertia effects in the equilibrium equations. The
theory governing the axisymmetric vibrations of circular cylindrical shells is quite simple.
In fact, all we have to do is to add a lateral inertia term to the right hand side of the equation
(4-24).
In this expression, the function f(x) represents a particular solution to (4-24). To obtain the
general solution to nonhomogeneous equation this particular solution is to be
superimposed on the general solution to the corresponding homogeneous equation. Being a
fourth order homogeneous differential equation, it would contain four integration constants
Cl. C2. C3. <4.
The general solution to the homogeneous equation (i.e., the differential equation without
the right hand side ) gives the correction to the particular, i.e., membrane solution so that the
prescribed boundary conditions can be satisfied. The integration constants Cl. C2. C3. C4
are determined by the four boundary conditions (two at each edge) prescribed for each
case.
Consider the circular cylinder of figure (4-3) subjected to a distributed edge moment Mo
and a distributed edge shear force Qo.
Mo
f----,--,.
Since the shell of figure (4-3) is acted upon by the edge forces alone, the particular solution
to this problem is identically zero, i.e., f (x) = O. The edge effects should decay out as we
move further away from the loaded edge. So that Cl =C2 =O. The general solution to this
problem is therefore
(4-26)
Two remaining constants of integration can be determined from the prescribed force
boundary conditions:
(4-27)
By substituting the expression (4-26) into these end conditions and solving two resulting
simultaneous equations for C3 and C4, we obtain
-Mo (4-28)
C4 = 2ß2K
This gives the general solution, (4-26), for this problem as follows:
(4-29)
Bending Analysis of Circular Cylindrical Shells 55
This is the expression for the radial displacement of a circular cylinder under edge loading.
The resulting hoop force and bending moment at any seetion can be determined from the
relations (4-22). As we expect, all these field functions decay character with distance from
the loaded edge.
Q.
Before entering into the analysis of this problem, we offer a physical observation on the
expected behavior of this shell. If the ends of this cylinder were free from all restrietions,
the shell under interna! pressure would act as a membrane. The membrane hoop stress
obtained from the membrane theory of cylindrica! shells would be
(4-30)
(4-31)
However, the imposed edge constraints inhibit the free membrane expansion of the shell.
To compensate for the incompatibility of membrane action, a bending field develops at the
edges and influences the shell behavior.
We assume that the cylinder is long enough so that the influence of the constraint at one
end would not be feit at the other. Therefore, the edge effect solution, (4-26), can be applied
to each end region. If we did not assume this, then for Sh0l1 cylinders we would have to
obtain all four constants Cl, C2, C3, C4 simultaneously.
56 Design and Analysis of Shell Structures
The end conditions, being the same for both ends in this problem, are
o o (4-32)
Substituting the sum of the fundamental edge effect solution (4-29) and the membrane
solution (4-31) into these relations, we obtain
MO -- 2S 2 1(D -- 262
P (4-33)
,. 0
Having found the quantities Mo and Qo, we can easily write down the expressions for the
radial displacement, the bending moment, and the shear force at each section of the shell.
p = Y (H - xl (4-34)
r
d~w ~ Y (H - xl (4-35)
dx 4 + 4ß w = y
From a physical point of view, this solution is the membrane displacement fjeld of the same
container having unconstrained edges. The general solution of the governing differential
equation, with edge constraints, is
Px -Px
W = e~ (C 1cosßx + C2Sinßx) + e ~ (C 3 Cosßx
(4-37)
+ C4 Sinßx) + Y (! IEt
- x) a 2
If we assume the container to be high enough so that the effect of the constraint at the base
would not be feIt at the top, then we may set Cl = C2 = 0; then
(4-38)
The remaining integration constants can be obtained through the use of the boundary
conditions at x =0, which are the same as (4-32); we obtain
Y 2 1
C = - aCH - - )
4 Et :3
ya" ( -ßx [ 1
w = Et 1H - x - e IICosßx + (H - 13) Sin6x ]} (4-39)
Having determined the radial displacement function, w (x), we can also write down the
expressions for the hoop stress and the bending moment, at each section of the cylinder, as
wall as the value of constraining forces at the base. These expressions are
-Px 1 -;lX ]
;~, ya [ H - x - He I-' Cos6x + (-6 - H)e ~ Sinsx
<,:
(4-40a)
Yat [1 _P -IOx ]
M ( - - H) e ,0 x Cosßx + He ~ Sinßx
x 112(1 _ v") ß
Solution:
First, we calculate the value of the parameter 13:
e= [ 3~;t~ V
2
)J!.o = 1.59 l/m
Q:2.7 Sm
r-:~~-:--~
I Q:2.75 m
l
~c::_ m
t : 2S c m I H:3.65
~J
Figure (4-6) The cylindrical water tank of ex am pie 4.1
Referring to relations (4-40), we find the hoop force and the bending moment
N
~
= :750 [3.65 - x - 3. 65e-1.59xCos1. 59x - 3 .02e -1. 59xSinl . 59x i.
Mx =-198.1 [_3.02e-l.S0xCOS1.59x+3.65e-1.59XSinl.59X]
l'-1 \
+5~8 .:2 6 k,g -mim
x x= 0
Bending Analysis of Circular Cylindrical Shells 59
Because of our sign convention, the positive value for the bending moment at the base
shows that the outer longitudinal fibers are in compression while the inner ones are in
tension.
+ 5 98.
Figure (4-7) Variations of ben ding moment and hoop force along
the height of the water container of example 4.1
Figure (4-7) shows the hoop stress and the bending moment variations with height. We
observe that the maximum hoop stress occurs somewhere along the cylinder wall, and not at
the base of the shell, as the membrane theory would have had predicted. In fact, the hoop
force at the base, being proportional to w, is equal to zero.
60 Design and Analysis of Shell Structures
Problems
P 4.1 - Consider a long pipe, of radius a, with a radialline load P (per unit width) applied
around the circumference at the mid-Iength of the pipe, figure (P 4-1). Show that the
longitudinal bending moment under the load is,
What would the expression for the bending moment under the load P be if the pipe had a
relatively short length of 2L while being free at the both ends ?
<
- - - - - - -I--- - - - - - f -
Figure (P4-1)
P 4.2 - Consider a long pipe with stiffening rings spaced at 2L centers and subjected to a
uniform external pressure p, figure (P4-2). Outline the general approach needed to analyze
such a structure.(Hint: there are two redundant forces at the pipe ring junction which, due 10
symmetry, are the same on the two sides of the ring-pipe intersection. These redundant
forces can be determined by writing the appropriate compatibility relations).
p p p
4.2 - S. Timoshenko and S. Woinowsky-Krieger, Theory of Plates and Shells, 2nd edition,
McGraw-Hill Book Co., N.Y., 1959
4.3 - G. P. Manning, Reinforced Concrete Design, Third edition, Longmans, Green and
Co. Ltd.,England, 1967
4.4 - Krishna and Jain, Reinforced Concrete, Vol.l,2, McGraw-Hill, New Delhi
4.5 - J. Faber and F. Mead, Reinforced Concrete, Third edition, E. and F.M. Spon Ltd.,
London, 1967
Design of Concrete Cylindrical Shell Roofs 63
Chapter5
5.1 - Introduction
Thin concrete cylindrical shells can cover the roofs of various buildings efficiently and
aesthetically. Large roof spans of bus, railroad, and air terminals, sport stadia, and aircraft
hangars have been effectively covered with reinforced concrete shells, many of which have
been cylindrical. Cylindrical shell forms can be easily shored and easily reinforced.
Cylindrical concrete shell roofs can also be constructed from the precast shell elements.
In previous chapters, we have discussed the membrane and the bending theories of
cylindrical shells. Those theories, and their simplified versions, can be used in the design of
concrete cylindrical shells. In the present chapter, we will present fundamentals of thin
concrete shell roof design. we also present a number of design recommendations and
practical analysis schemes. These design-analysis methodologies will be accompanied by a
set of detailed design and analysis examples.
64 Design and Analysis of Shell Structures
The first step in the structural design of a cylindrical shell roof is the design of the
geometrical configuration of the shell and the structural elements associated with the shell
body. Figures (5-1) show a number of choices which can be made for such a purpose.
These simple shell modules include a single shell simply supported on end arches (or
diaphragms), figure(5-la); a group of simply supported shells, figure (5-1 b); a continuous
cylindrical shell with intermediate stiffening arches and supports, figure (5-lc).
Cylindrical shell roofs are generally composed of the main shell together with some
stiffening and / or supporting elements. These elements could form a collection consisting
of edge beams, stiffening arches, and end members. Figure (5-2) shows a typical
continuous cylindrical shell roof having these elements.
The end edge members themselves can have various forms. They can be in the form of a
solid diaphragm, figure (5-3a); a tied arch (with or without verticals), figure (5-3b); a simple
arch, figure (5-3c); an elevated grid, figure (5-3d); or an arched truss, figure (5-3e). Figures
(5-3) show varieties of end-stiffening and supporting members associated with a shell roof
structure.
I~
(a) solid diaphragm (b) tied arch with verticals (c) simple arch
HJD1
-.~.>
- - ..
-. -- --=-- --=
A shell roof could be placed on vertical supports or on its foundation in a variety of ways.
Figures (5-4) show some possible vertical support arrangements for eylindrieal shell roof
structures.
(4)1 (3)W
(1) (2)
(6)
We propose some suitable relative dimensions for cylindrical shell roofs in figure (5-5)
and in its accompanying table. Figure (5-5a) relates to a group of cylindrical shells with an
edge beam located at the longitudinal edge of the last shelllocated in the shell group.
The shell dimensions given in figure (5-5b) are for a particular type of cylindrical roof
called the Northlight shells. This shell type is used to provide natural light through the
roof. These shell types are particularly suitable for factory and manufacturing installations.
The information presented in figure-table (5-5) serves as a useful guide in proportioning
the cylindrical shell roofs.
Large shell spans, up to 120 meters can be constructed using plain or prestressed
reinforced concrete material. These large spans are used in terminals, airplane hangars,
sport stadia, and unobstructed storage places. In such cases, a number of transverse arches
are normally constructed with the shell. These arches stiffen the shell and carry its load to
the foundations. Shelliengths up to 30 meters can be made of non-prestressed reinforced
concrete. For longer shell lengths, the shell should be prestressed. The longitudinal
expansion joints are usually provided every 45 meters of the shelllength.
I I
i I: ! r t
(m) 1(_)
3
;"')
I
I (mml i~~;~~·(~i
:0 60 ;70
-"
30 I :5 :1 70 300
';0 I :) :5 I 80 350
I
3 :5 :8
I 100 I I
390
'~~.'~;j"
\
,.. support
b;O.6d-\.Od
choice of the shell thickness. As a general guideline, a thickness to chord width ratio of
11300 to 11200 could be considered asound choice for concrete shell thickness.
The thickness of concrete shells is usually increased near the edge members, and gradually
reduced to a constant design value. This simplifies the placing of reinforcement: It also
prevents stress concentrations, and facilitates flow of forces to the edges and supporting
members.
Concrete materials have high compressive strength but low tensile resistance. Therefore,
thin concrete shell roofs must be reinforced to acquire sufficient tensile strength.
Reinforcing of the concrete shells is usually provided by means of steel reinforcing bars.
The existing codes of practice for shells normally require that the reinforcing bars be
placed along the general directions of principal tensile stresses. However, deviations from
these directions are also conditionally allowed by these codes. Figure (5-7) shows the
stress trajectories (i.e., the principal stress directions) in a simply supported single
cylindrical shell subjected to its own dead weight. These types of results will be obtained in
the later sections of this chapter.
tensile trajectories
compression trajectories
As we know from the elementary courses on solid mechanics, the stress trajectories
consist of two families of curves which are mutually orthogonal. In cylindlical shell roofs,
one family of these curves represents the directions of principal tensile stresses while the
other family of trajectories gives the directions of principal compressive stresses. In this
sense, the behavior of the shell can be conceived to consist of a mesh of eables and arehes.
As we noted earlier, thin shells are usually accompanied by stiffening and / or supporting
members. These elements must be monolithic with the shell so that they can transfer the
internal forces. On the other hand, these members produce a bending field. This bending
field would require extra reinforcement which must be provided in addition to the
membrane field reinforcements.
Design of Concrete Cylindrical Shell Roofs 69
The knowledge of the bending field in the shell, is crucial in the design of the extra
reinforcements. We presented a qualitative discussion of the cylindrical vaults behavior in
chapter 3; that is useful in devising the general pattern of shell reinforcements, but a
detailed design must be based on a quantitative analysis of the shell.
Tensile stresses in concrete shells tend to cause cracks in the concrete. Figure (5-8) shows
the most probable cracking pattern in a single cylindrical shell. This figure shows the
positions as weIl as the directions of the probable cracks.
The transverse cracks appearing in the longitudinal edges of figure (5-8) are due to
longitudinal tensile force. The longitudinal cracks designated by solid and dotted lines can
be caused by transverse bending moment. Finally, the inclined cracks at four corners of the
shell are produced by the diagonal tension. these latter cracks are called shear cracks. In
addition to these global crack field, local cracks may occur at the junctions of the sheIl with
its stiffening members.
The pattern of shell reinforcement must naturally follow the probable cracking. The steel
reinforcement should , in principle, be placed in the directions perpendicular to the possible
cracks. Figure (5-9) shows a reinforcing pattern for the shell of figure (5-8). A review of
figure (5-8) would easily lead to the reinforcement design strategy adopted in figure (5-9).
Based on this knowledge and the eurrent practiee, we offer a praetical eonstruetion tip for
placing of reinforcements. The shear reinforeement whieh is to absorb the diagonal tension
field is normally plaeed in the lower layer. The longitudinal reinforcement oeeupies the
mid-Iayer; the transverse reinforeement is usually placed on the top. These struetural
reinforcements also serve as shrinkage, distribution, and temperature reinforeements.
The total amount of steel reinforeement in a eylindrieal shell roof depends on the individual
designs. However, as an initial emde estimate of steel eonsumption, the following formula
could be used.
q= t(t + B) + 6
20f (5-1)
In this approximate formula, I is the shelliength, B is the ehord width, fis the rise of the
shell, and q (in kilograms per square meters of shell surfaee) is the amount of the
eonsumed reinforeement.
The values of principal stresses at each point of the shell ean be obtained from the
following relation:
N+N /N-N 2
~ = x 2 cP ± I ( x 2 $) 2 + NX $ (5-2)
and the principal directions ean also be determined from the following formula:
2NXp
tan29 (5-3)
Nx - N$
Assuming that the internal forces Nx. N</I, and NX</I are determined through the analysis of
the shell, we can determine the principal stresses and the principal directions at each point
of the shell, and plot the stress trajeetories for each case.
Figure (5-10) shows the plan of stress trajectories for a unifonnly loaded simply supported
cylindrical vault. In this figure, the solid curves represent the directions of compressive
principal stresses while the dashed lines represent the curves that are tangent to the
directions of maximum tensile stresses.
According to the American Concrete Institute (ACI) code of reinforced concrete shell
design, the reinforcement is to be placed in the general direction of principal tensile
stresses. This code allows for slight deviations of reinforcement direction from principal
direction. In this code, a directional error of 5 degrees or less is considered acceptable.
However, due to constructional considerations, it is usually convenient to place the
reinforcing bars in perpendicular directions. The codes have provisions for these cases. In
these cases, the reinforcing network must be designed so that it can absorb the principal
tensile stress at each point. This situation is shown in figure (5-11).
(a)
Figure (5-11) (a) Plincipal tensile force acting in some elemental
direction, (b) contributions of perpendicular steel reinforcement
to absorb the principal tensile stress
Referring to figure (5-llb), we can write the internal equilibrium equation can be written as
(5-4)
T ds
P
= ASI f SI dx Cos 0 + A
S2
f
S2
dy SinD
In this relation, As 1 and AS2 are the cross sectional areas (per unit shell width) of
perpendicularly placed reinforcing bars, fS l and fS 2 are the stresses in reinforcing bars, Tp
is the value of principal tensile force, and Ö is its angle of inc1ination. Using the geometrie al
relations between ds, dx, and dy, we may rewrite the relation (5-4) as,
(5-5)
(5-7)
This relation ean be effectively used in designing reinforeement, i.e., in detennining the area
of reinforcing bars.
72 Design and Analysis of Shell Structures
One of the simple methods for preliminary analysis and design of cylindrical shell roofs is
called the beam-arch method. The "beam-arch" method makes use of combined beam and
arch actions of cylindrical shells. In this method, beam analysis and arch analysis of the
shell are carried out alm ost independently by the well-known simple theories of the
structural analysis. The results of these analyses are, then, combined appropriately to yield
the spatial behavior of the cylindrical shells.
In the following, we shall outline the principles of the "beam-arch" method for cylindrical
vaults having an arbitrary profile. Later on, we shall also apply the method to the analysis of
a vaulted roof problem.
The beam-arch method of vaulted roof analysis is composed of two steps: (1) the beam
analysis, (2) the arch analysis.
In this stage, the shell is conceived as a beam resting on the end supports and behaving
according to the cIassical beam theory. The beam has the profile of the cylindrical shell as
its cross seetion and the length of the shell as its longitudinal span.
To perform the beam analysis of a circular cylindrical shell, we consider the profile of the
shell to be as shown in figure (5-12). In this figure, <l>k is the half central angle, a is its
radius, and t is the shell wall thickness. The origin of cross-sectional coordinate system y-z
coincides with the centroid of the section, designated by the letter G. This centroid is
assumed to be located at a vertical distance ofz from the crown of the shell. For a circular
sector shown in figure (5-12), the following formula for z in terms of cross-sectional
parameters (a and <1>0 can easily be derived:
(5-8)
The longitudinal membrane stress (expressed as force per unit width, N x) can be calculated
by the flexure theory of beams. The appropriate bending stress formula is
M
N
x
= (...Y:l
I
z)t (5-9)
yy
In this relation, M yy is the bending moment of applied loads calculated about the y-y axis,
~y is the second moment of cross sectional area with respect to y-y axis. A general formula
for Iyy for a circular sector is
2t J <P=4>
<p=O
k ad<jl(aCos<P
.
_ aSl.IlQ')
tP k
2
(5-10)
According to classical beam theory, the relations (5-8) to (5-10) are valid in a principal
coordinate system for the cross section. For non-principal coordinates, we must use a more
general beam formula
The transverse shear stress (expressed by the membrane shear force, NX $) can also be
determined by a formula given in the classical beam theory. So, using the formula for the
shear stresses in beams, we find
(5-11)
In this relation, V is the total vertical shear force at any arbitrary section of the shell, the
symbol Q represents the value of the first moment of the part of the areal section lying
beyond the layer at which the magnitude of shear stress is to be ca1culated. A general
formula for Q for a circular sector is
= 2a 2 t(Sinrp - ~ Sinrp )
<Pk k
The longitudinal and shear membrane forces Nx and NX $, determined through beam
analysis, are functions of x and <». This completes the beam analysis.
74 Design and Analysis of Shell Structures
The second stage of beam-arch method, consists of analysis of transverse arches. In this
stage, a transverse arch having an arbitrarily smalliongitudinal width, dx, is taken out from
the shell. Figure (5-13) shows the free-body diagram of such an arch slice cut out from
some arbitrary location along the shell.
The free body diagram of figure (5-13) shows the applied loading as weIl as the effect of
the rest of the system; this includes the longitudinally varying shear force which has
already been determined. The longitudinal edge effects consist of horizontal shear forces
and the bending moments, both distributed on the edges of the arch.
VQ
P
I Tb
In the vertical equilibrium of transverse arches, the applied vertical fm"Ces are supported by
the vertical component of the internal shear stresses. In this sense, one may conceive of this
arch to have been hanging from the rest of the system, i.e., to have been vertically supported
by the adjacent arches through which the applied load is uitimately transfelTed to the end
supports. Figure (5-14) shows an arbitrary transverse arch in the shell and its role in
supporting its share of the load.
(a) profile of the arch (b) lateral view of arch (c) internal stress in beam
Figure (5-14) The side and the end view of an arbitraJY transverse
arch in a cylindlical shell
Design of Concrete Cylindrical Shell Roofs 75
For a uniformly distributed applied loading, the change of internal shear force, in the
longitudinal direction, is
= ~(~
= n yy
dNX <1> t) Q dv Q (5-13)
dx dx 2It' dx =~yy
w
(5-14)
Having defined the transverse arch and its loading, we can proceed with its analysis.
Depending on the number of unknown boundary forces (M and / or H), the arch could be
statically determinate or indeterminate; in any case, it can be analyzed by standard methods
of structural analysis. The transverse arch analysis gives the transverse membrane force,
Nq,. and the transverse bending moment, Mq,. are determined.
By combining the results of the beam and arch analyses we can find the internal force
system in the shell, consisting of Nx. Nxq,. Nq" and M<I1. The strength of the shell roof can
now be verified against the applied loads. The calculated internal fOl·ces can be used to
design the proper shell reinforcement to support the prescribed loading.
The beam-arch method for cylindrical vaults can be displayed in a tabular form. Table (5-
1) shows coefficients of internal force system for a circular cylindrical shell subjected to its
dead weight and snow type loading. This table gives the internal forces for various values
of the half central angle <Pk.
(a) Single shells with no edge beams provided that I / a >5 (1 is the length and a is the radius
of the shell )
(c) Shell units in a shell group with small edge bearn, if 1/ a > 1.67
Modified version of the beam-arch method mayaiso be used for the analysis of folded
plates.; the topie which will be discussed in chapter 11.
76 Design and Analysis of Shell Structures
,jllllllll g 11 1 1 I I I 11
'~~F4
• I.--- , ----I ~:~-~
"--,-_
.e.
tL
t ".
Cil
"t
[21
5
(l)
tlt
[.)
".
(5)
"t
(6)
5
(7)
"+
(81
N.
~,
• N
x N. S
". N
x
5 M.
.,
Je& (1) ( 2) (l) 14) 15) 16) ,7) (8)
.,
0
0.75+,
4.146
-\.853
-1.498
0.260
-I. 335
-1.127
0.000
0.000
1. 211
-0.01509
-0.00742
-0 00277
4 458
-2 013
-I 617
0.311
-1.404
-1. 176
o
o
1.315
000
000
-0 01796
-0 00897
-0.00337
<0 0 O. SO" -0.443 -0.592 1 929 o 00595 -0 482 -0. S92 2 095 0.00719
0.25 .. 1. 279 0.005 1 67S o 00578 1.389 0 055 1 819 o 00705
0 3 616 0.241 o 000 -0 01641 .3 9~8 o 297 o 000 -0 02006
shell thickness: t = 10 cm
half central angle: <Pk =40 0
Solution:
Location of cross sectional centroid:
Sin~'k
Z= r(l - - - ) 10 (1 _ 0 - 643 )
0_698
~
Design of Concrete Cylindrical Shell Roofs 79
2Sin<l>k ]
I yy = r 3 t [ <l>k + Sin<l>k (CostPk - if>k ) +
2 x 0.643 )]
= (10)3 (..:!:Q. .}0.698 + 0.643 (0.766 0.698
100 L
4
= 0.587+0.10 = 0.687m
To perform the numerieal analysis of this symmetrie problem, we eonsider one half of the
cross section. Then, we divide it arbitrarily into 8 segments, figure (5-16), and from now
ealeulate everything for these discrete segments.
Table (5-2)
0 40 O.
7
I-
1 35 i 0.223
2 30 0.358
3 25 0.418
4 20 0.413
'"
5 15 0.355
6 10 O. 259 0:10 m
7
8
5
0
0.136
O. .
~.,.,,100
Beam Analysis:
The longitudinal bending moment, M yy ' and the longitudinal stress, N xx , are ealeulated at the
mid-Iength of the shell, where they are maximum, while the shear force, V, and related shear
stress are ealeulated at the ends of the shell. The related ealculations are shown below. The
numerieal results for discrete segments are given in table (5-3).
80 Design and Analysis of Shell Structures
M
N = (...nI Z) t
x yy
Wf =(2x250xS + 120B+60)~
2
~ =
,.
=(250 x 10 x 2 x 0.698 + 150 x 2 x 10 x 0.623+60) (2~)
N = ~ _ 54190 _
x4> 2Iyy Q - 2 x 0.687 Q - 39418Q
Toble (5-3)
I Q
pOint! 4>. I
'?
N NX $
'" x
from top m m 3 kg/m kg/m
0 40 1.547 0 60972 0
8 0 -0.793 0 -31248 0
Design of Concrete Cylindrical Shell Roofs 81
Arch Analysis
Since the applied uniform loading is constant along the shell the shear force varies linearly
in the longitudinal direction. Therefore, the longitudinal change of shear force, Nxq,. i.e., its
derivative with respect to x, is obtained by dividing the maximum values, calculated at the
end supports, by 11 2. These values as weIl as the vertical and horizontal components of
shear force, for each discrete segment, are given in table (5-4).
Figure (5-17) shows the side view of half of the transverse arch together with its applied
loading, as extracted from table (5-4). As we see, for this half arch, the vertical as weIl as the
horizontal equilibrium equations are satisfied.
Tab/e (5-4)
tI
'174.5
270.1
144.1 61.'
214.5 I7B',
"'.1 13017"- 11'1.1
IIOJ 1067. . IU1.S .
627.,
,
0= 10 m
The longitudinal edges of the shell in figure (5-15) are fixed against rotation and horizontal
translation. Therefore, the transverse arch can be considered a fixed arch. The redundant
forces are taken to be the bending moment, Mo, and horizontal shear force, H, at the crown
of the arch, as shown in the figure (5-18). To determine the redundant forces, we use the
energy theorems of structural analysis, and in particular the method of least work .
m8 = 0.0
x ( 0.2985 +
2 0.3688,) -- -....
4'3] . --
I:J
,._,g-rn
Jo
s M~ds
U=2 l.
2EI
where
M.
l.
= M0 +HZ' +rn.
l.
Z' = a(l - cos~}
dU au
t:.H = '"
oH = 0 , t:.e:: dM = 0
o
Therefore
S
dU - 2 \
ClH-
~
)0 EI (M0 + HZ' + rn.l. }ds 0
dU
dMo
=2 Jo
S ,
~I(Mo + HZ' + rni)ds =0
Since the bending stiffness of the arch cross section,EI, has been assumed constant, these
relations can be rewritten as
MO J(os Z 'ds + H JS
0 Z2'ds Z'rn.ds
l.
o
Numerical calculations related to these integrals can be perfonned in a routine fashion; the
final results are shown in table (5-5).
84 Design and Analysis of Shell Structures
Table (5-5)
. I
seetlon Z' da m. Z'· Z'da Z"da miZ'ds mida
1.
8 0 0.436 0 0 0 0 0 0
MO - -350.15 Kq-.!m
H .. 5498.24 kq/m
Onee the quantities Mo and H have been determined, the intern al fOJ·ces ean be easily
obtained using onIy the equations of statics. These ea1culations and the resulting values of
internal forces are presented in tables (5-6) and (5-7).
Design of Concrete Cylindrical Shell Roofs 85
Table (5-6)
8 0 0 -350.15 0 -350.15
Table (5-7)
1356.8
~
1039.4
rvs~ N... 5+6
Xqfin
102.9 1142.3
1 35 41 729.2 597.3 26.4 623.7
2 30 -238.5 -337.8 -292.5 -119.3 -411.8
3 25 -508.7 -1639.5
I -1485.9 -215 -1700.9
4 20 -652.8 -2973 II -2793.7 -223.3 -3017
5 15 -624.7 -4154.1
I -4012.6 -161.7 -4174.3
6 10 -432.3 -5031.6 -4955.2
.-
-75.1 -5030.3
-
7 5 -124.5 -5498.2 -5477.3 -10.9 -5488.2
8 0 50 -5498.2 -5498.2 0 -5498.2
The results of these ca1culations are summarized in the set of figures (5-19); these show the
variations of internal forces across the shell profile.
86 Design and Analysis of Shell Structures
Finally, we calculate the horizontal thrust, R, along the longitudinal edges of this inner shell;
it is
aN
R; H - roxX~dY; 5498.2 - 6855.0; -1356.8 Kg~
The lateral thrust, R, is negative; this means that the direction of R is outwards and not
inward, as would have been the case for arches and frames. In other words, the inner shell
has a tendency to shrink inwards along its longitudinal edges, so that the restraining effect
of adjacent shells is to constrain this inward movement. This quality differs from our
experience in the analysis of framed structures; and has its own structural implications in
the design of such shells.
I 1
(a) (b)
Nx ,,/m
.•
-, -1 a .. I .. 2 .. 1....1o
The analysis of this problem could have also been carried out with the help of table (5-1).
Design of Concrete Cylindrical Shell Roofs 87
Calculations in the ASCE manual are based on simplified versions of the exact cylindrical
shell equations (Flügge equations). These calculations relate to the so-ealled "long shells"
(for which I / a>1.67) and short shells (for which I / a <1.67).
The ASCE tables are based on the force method of structural analysis which has been
outlined in previous chapters; the forees at the longitudinal edge of the shell are chosen as
the redundant forces.
The ASCE tables use Fourier se ries expansions. All distributed applied loads, as weIl as
distributed edge effects, are expanded into Fourier si ne or eosine series in the longitudinal
variable, x. Figures (5-20) show the first terms of the Fourier expansion of the redundant
edge force distributions. Similar type of distributions are also assumed for the edge beams.
Note that the ASCE tables choose the origin of the transverse angular coordinate, <1>, at the
longitudinal edge of the shell. Therefore, the angle defining the normal to the surfaee at
each point is measured from the normal to the shell at its longitudinal boundary.
In the analysis of cylindrical shells by the ASCE tables the following procedure should be
followed:
(4) Superposition
Having obtained the redundant forces, we superpose the results of calculations performed
in steps (1) and (2) and find the total force field in the shell.
The ASCE tables provide a set of eoefficients for the interna! forces in a variety of circular
cylindrical vaults, including single shells with or without edge beams,and shell groups with
or without edge members. The ASCE tables are included in the appendix.
88 Design and Analysis of Shell Structures
(al (bI
Positive dlrectlons
of N)(~
b J cos T
(cl
(dl
In this section, we present some design examples of reinforced concrete shell roofs. We
assurne that the analysis of the shells under design has been calTied out by the beam-arch
method, the ASCE tables, or some computer program. In practical designs, differences in
solutions obtained by different means of analysis are usually of secondary importance.
Design:
The overall dimensions of the shell are given in figure (5-21). The objective is to design the
steel reinforcement for given shape.
Tables (5-8) and (5-9) summarize the results of analysis of this shell performed by using
the ACI tables. We would have obtained similar results if we had analyzed the shell by
other methods.
90 Design and Analysis of Shell Structures
Table (5-8) Internal forces in the single simply supported shell of tigure (5-
21)
.0 ~
T
". (J<q/ll» ".. (J<q/ll»
111,
N. 1I<q/ll»
11' I 3/8
M4 II<q....vmj
" .....,
1/8 1/' 3/8 1/2 0 1/8 1/' 3/8 1/8 1/2 0 1/8 1/, 3/8 1/2
...
'0 o -2U3 ,",331 -6853 0 0 0 0 0 -13'9 -2493 1-3257 -3525 o -50S -933 -1220 -1320
30 o -2564 ... 738 ,",190 -67QO -2013 -1860 -1423 -770 0 -5513 -10186!-13308 o --457 -
.....
-14405 -1102 -1193
20 o -2115 -3907 -5105 -55216 -84<&' -7804 -5913 -]233 0 o -11591 -2H1S!-27984 -30290 . 0 -267 .... S -698
10 o -942 -1740 -2214 -2461 -1-4798 -13672 -1046" -S663 0 -1353 -2500 1-3267 -3536 o -34 "'2 -81 -OB
0 0 0 0 0 0 0 0 0 0 0 .52641 +972671+127086 +137557 0 0 0 0 0
~
Table (5-9) Principal tensile stresses and directions in shell oftigure (5-21)
N., ..... • ...., ..... • .... , .... , 8 ...., ..... • ..... ..". •
40 0 0 0 -2623 -1349 0 -4946 -2493 0 -6331 -3257 0 -6853 -3525 0
30 -2013 2013 4S -6-412 -1665 26 -10535 -4J89 ,. -13390 -6108 6 -6700 -14405 0
.
20 -8477 B477 45 -15983 "2277 29 -23261 -2064 17 -28432\ -4657 8 -5526 -)0290 0
10 -1479B 14798 4S -14821 +12590 45 . -12526 +-8351 -.'SS I ·2914 42 -2461 -)536 0
- -
0 0 0 0 0 +52641 0 0 +97267 0 0 ! ·'27ca, 0 0 +-137557 0
Figure (5-22) shows variations of internal forces Nx. N<I>. and M<I> at the mid-length and the
variation of NX<jI at the end support. For Nx• the numbers in this tigure should be multiplied
by a factor of 10000 (kg Im). For other force quantities. the multiplication coefticient of
1000 should be applied.
crown "0 , ..
I Cl)
•0 Nt M•
I
,
.
2
3.'
..
u Nx
u.
'lO
c: "- I
(\j ?' i I
iii
,.
I
'5 '0
~ I
,
~ I '\....
........ .
€
~
0
I I
10 12 •
U
Once the internal forces are known, the reinforcement design, for each subdivided region of
the shell, can be carried out according to the standard Working Stress Design (WSD)
method.
Figure (5-23) shows the plan of reinforcement pattern for the shell of figure (5-21). Due to
the double axial symmetry, only a quarter of the shell is shown. Figure (5-24) shows a
detailed section of the reinforcement at the corner region. The reinforcement of the shell in
this region consists of longitudinal, transverse, and diagonal (shear) reinforcements.
• ., .. AI JO cm '11 al 20 cm
end diaphragm
~3SaE _ _Rlm_ _mm~~
~12at20cm~----~~~~
section
,.10.0 m
To design an inner shell of the shell group, we take it out and analyze it using one of the
methods discussed before. Figures (5-26) show the physically reasonable assumptions
which can be made. These assumptions are also useful in the analysis of shells by
computer programs.
The results of analysis of an inner shell are summarized in tables (5-10) to (5-14).
Table (5-10)
~I 0
I 1/8 I 1/4 I 3/8 1/2 I
I
~o
!
I
0 1-2601 -4807 I -6281 -6798
30
20
i
I
0
J
1-2237 -2293
-4134 -5401
-2996
-5846
-3243
10
0
I 0
I 0
1--241241
507
-44
937
-57
1224
-62
+1325
Table (5-11)
Table (5-12)
40 0 0 0 0 0
30 -7224 -6674 -5108 -2765 0
20 -12496 -11545 -8836 -4782 0
10 -11992 -11079 -8480 -4589 0
0 0 0 0 i 0 0
Table (5-13)
.
M, (Kg-m!m)
40
30
0
0
-121
-53
-223
-98
-292
-128
I -316
-139
20 0 +110 204 266 +288
216 282 +305
I
10 0 117
0 0 -329 -608 -795 -860
94 Design and Analysis of Shell Structures
Table (5-14)
~o
~, ~2 ~, I ~2 ~, ~2 .
lt., , ~2 "P, ~2
30 -7224 +7224 -13352 1770 -19021 1-2381 , -22996 -4966 -24920 -5846
20 -12496 +12496 -14525 8792 -14629 4036 , -13106 -734 -11738 -3243
10 -11992 +11992 -8035 15299 -4131 1755] i -1179 18720 +19044 -62
0 0 0 507 27635 937 51063 i 1224 66717 +72214 +1325
Figure (5-27) shows variations of internal forces across the shell cross section. For the
transverse moment, M4I, the values of the table should be multiplied by 1000 (kg mim).
Also, for Nx a factor of 10000 (kg Im), and for Nell and Nxell a multiplication factor of 1000
(kg I m) should be applied.
crown
I I I
-
HO I '
E ...-
.>- t-r \ 11 1 I
( ])
200 ........ "l \ V-1~ I
Cl
"0
(])
./
,/
N~~ \ \ I
I
-E '- \1 o
" o.
C
ro r-- t--. /l i\ "-
üi r- r- JJ i \
'ö - 6 _4
_I" -12 -10 -8 -2 8
Figure (5-28) shows the quarter plan of reinforcing scheme of an inner shell from the shell
group in figure (5-26).
"lIa.15 tm
C rown
-........ ""- I
E
Cl
""''-....
r--- "-
~
.c 1"-
a.
ctl
'ö " mid-Iength
"C
c
Cl) " ~
f--- \2.81 m .1
Table (5-15)
N
x
Table (5-16)
Table (5-17)
K 40
0
0
1/8
0
1/4
0
3/8
0
1/2
0
I
Table (5-18)
4) Plan of reinforcement:
4l12C1t)] tm ~
crown ./'
1'2 Cl 2S c".,
~~~
""
I I /1
I I
11
end diaphrag rn
I
!
I
I
i
I
, I
fj~+~i
,,,,-, i I
!
i
,~ i I I I
t'0 rx: '" i
iI ~
i f
~ ['--,,~<~ I ~ I I
m/
f0
~"
~ l\:~
f>'~\:
0~ N
[}::
~~,,.
I
I
j
-*H
1 I
~ 11~
cJlI2 0115 tm
edge L.4> 1111']0 tft'\
Table (5-19)
I
10 0 -1677 -lB15
0 I 0
\-695
i :' 32 :
-1283
I
42B-----L 559 60S
I
Table (5-20)
~xz: i
Nx
0 1/8 1/. 3/8
Table (5-21)
~I 0
I
1/8 1/' I 3/8 1 1/2
I
I f)(g1Vrn )
I 40 I
0 -92 -170 -223
I -241
I 30
20
0
0
-33
55
-60
102
-79
133 1-::'
I ~o
0 67 124 162 175
I 0 -191 -354 -462 -500
Table (5-22)
1'~
;:<'('1 0 1/8 1/4 3/8 1/2 i "xl>
I [l'.g-'T,j
'0 0 I 0 i 0
!
3D
I -27':'2 -1484 I
20 I I
1J
Design of Concrete Cylindrical Shell Roofs 99
Table (5-23)
3) Reinforcement system:
'"
!~ t". .c
I"", ,,\ Öl
c
~"""'~"
,," " ~
-0
"", ",'~ I 'E
~"",
,>,,>~ ~,\, "
~,,'\ ,~
,
, I
E .,
~'"
~'\'\" ~"''''~
U o
'\' ~,\" I'} "
;1,8
~ I'
LI
~I'..
I I
edge.l 1 1
t/2
° '/8 '/4
,12'@30cm
31/8
12'ill~ • 12 '@ 20 cm
- = , ,4 "-t: 10·
"to 1'cil
12 r@ 15 cm
Problems
P 5.1 - Analyze the inner shell of the multi-cylindrical shell roof shown in figure (P 5-1).
The profiles of shells are semicirc1es. The shells are semicircular,of thickness t; the width
of the inner and the outer edge beams is 3 t, and the height of the edge beams is 5 t. The
width of the shells is one half of the roof length. The system is subjected to its own dead
load.
Edge beam
V /
Length of the shell
P 5.2 - Figure (P 5-2) shows a shell roof over a bus or train station. The length of the shell
is 10.0 m, and it is supported by arched diaphragms which cantilever out from the columns.
The radius of the circular shell section is 2.4 m. The intensity of applied load is 300 kg /
m 2 distributed over the shell area. Analyze this shell by using the beam-arch method, and
present a design of shell reinforcement.
Shell
I 10 cm I
DIaphragm
arches
~~
(a)
\ (b) I
5.4 -J. E. Gibson, The Design of Shell Roofs, E. and F. N. Spon Ltd., London 1968
5.5 - Design ofCylindrical Concrete Shell Roofs, Manual of Engineering Practice, No.31,
American Society ofCivil Engineers, New York, 1952
5.7 - C. B. Wilby, Design Graphs for Concrete Shell Roofs. Applied Scicnce Publishers,
Ltd., Barking, Essex, England, 1980
5.8 - Coefficients for design of Cylindrical Concrete Shell Roofs. Portland Cement
Association, Skokie, Illinois, 1959
5.9 - J. Chinn, Cylindrical Shell Analysis Simplified by the Beam Method, ACI J., Vol. 55,
May 1959
5.10 - V. N. Baikov, (ed.), Reinforced Concrete Structures, MIR Publishers, Moscow,
English Translation 1978