You are on page 1of 14

ARTICLE IN PRESS

Thin-Walled Structures 47 (2009) 10781091

Contents lists available at ScienceDirect

Thin-Walled Structures
journal homepage: www.elsevier.com/locate/tws

Strengthening metallic cylindrical shells against elephants foot buckling


with FRP
M. Batikha a, J.F. Chen a,, J.M. Rotter a, J.G. Teng b
a
b

Institute for Infrastructure & Environment, The University of Edinburgh, Edinburgh EH9 3JL, UK
Department of Civil and Structural Engineering, The Hong Kong Polytechnic University, Hong Kong, China

a r t i c l e in f o

a b s t r a c t

Available online 21 December 2008

Thin metal cylindrical shell structures such as silos and tanks are susceptible to an elasticplastic
instability failure at the base boundary known as elephants foot buckling, due to its characteristic
deformed shape. This form of buckling occurs under high internal pressure accompanied by axial
compression in the shell structure. This is a common failure mode for tanks under earthquake loading.
Another common situation is in a silo where the silo wall is subjected to both normal pressures from the
stored granular solid and vertical compressive forces developed from the friction between the stored
solid and the silo wall. This paper presents a novel method of strengthening cylindrical shells against
elephants foot buckling in which a small amount of bre-reinforced polymer (FRP) composite, used at a
critical location, can effectively eliminate the problem and increase the buckling strength. The
strengthened shell is analysed using linear elastic bending theory in this preliminary study. Within the
scope of this research, the strengthening effect is shown to be sensitive to the thickness, height and
location of the FRP sheet. The issue of optimal FRP strengthening to allow the shell to attain pure
membrane-state deformation is examined in detail as strengthening with too much and too little FRP
are both undesirable. Both pinned-based and xed-based shells are examined and their responses are
compared.
& 2008 Elsevier Ltd. All rights reserved.

Keywords:
Metallic shells
Cylindrical shells
Silos
Elephants foot buckling
FRP
Strengthening

1. Introduction
Thin metal cylindrical shells are widely used as containers,
such as silos and tanks. These shells are sensitive to the
magnitude of the imperfections, which can cause elastic buckling,
but under high internal pressure, this sensitivity is much reduced
[15]. Existing research [4,5] has demonstrated that cylindrical
shells fail near local imperfections by elastic buckling if the
internal pressure is small. By increasing the internal pressure,
yielding of the wall near the base boundary leads to local
reductions in exural stiffness and local amplications of
displacements. The circumferential membrane stress resultants
are raised [6] and elasticplastic buckling occurs [1]. This
elasticplastic instability failure near the base boundary is known
as elephants foot buckling. Further, Rotter [1] showed that, for
thin cylinders, a clamped base is considerably stronger than a
pinned base, whilst in thicker cylinders, clamped and pinned
bases have similar strengths.
Chen et al. [7,8] proposed to strengthen the shell against
elephants foot buckling by using a small ring stiffener. The

 Corresponding author. Tel.: +44 131 650 6768; fax: +44 131 650 6781.

E-mail address: j.f.chen@ed.ac.uk (J.F. Chen).


0263-8231/$ - see front matter & 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tws.2008.10.012

optimal dimensions and location for this stiffener were derived. In


this paper, an alternative method of strengthening cylindrical
shells against elephants foot buckling using bre-reinforced
polymer (FRP) composites is proposed. FRP composites have
superior properties, including a high strength-to-density ratio and
superior durability to many traditional materials. Applying FRP
composites to the external surface of the shell is particularly
advantageous in many cases because it avoids the need to access
the interior of the storage structure and even to take the structure
out of service, thus minimising health and safety concerns and
operational disruptions. Bonding FRP to the shell is also easier
than welding reinforcement to it, since the latter would not only
result in residual stresses and imperfections in the shell but might
also have signicant safety implications when sensitive or
ammable materials such as petroleum products have been
stored in the structure.
Extensive research on the use of FRP for strengthening concrete
structures has been undertaken since the 1990s (e.g. [911]). FRP
strengthening research has been extended to metallic beams and
columns (e.g. [9,1215]), masonry (e.g. [16,17]) and timber
structures (e.g. [18]). In the vast majority of these cases, it is
the strength rather than the stability of the structure that is
the main concern. Teng and Hu [19,20] explored the behaviour
of circular steel tubes subjected to axial load and conned by

ARTICLE IN PRESS
M. Batikha et al. / Thin-Walled Structures 47 (2009) 10781091

Nomenclature
[A]
{B}
{C}
Br
D
Db
Dfz
Ds
Eb
Efz
Efy
Es
h
hf
Mz
My
Nz
Nzf
Nzs

coefcient matrix (Eq. (48))


coefcient vector (Eq. (48))
constant vector (Eq. (48))
extensional stiffness of the composite FRPmetal shell
in the circumferential direction
shell exural rigidity
exural rigidity of the composite FRPmetal shell in
the meridional direction
exural rigidity of the FRP shell in the meridional
direction
exural rigidity of the metal shell
Youngs modulus of the composite FRPmetal shell
Youngs modulus of the FRP shell in the meridional
direction
Youngs modulus of the FRP shell in the circumferential direction
Youngs modulus of the metal shell
height of cylindrical shell
height of FRP sheet
meridional bending moment
circumferential bending moment
vertical load per unit circumference
axial force in the FRP shell
axial force in the metal shell

FRP jackets, and also studied elephants foot buckling of


cylindrical shells under combined axial compression and internal
pressure when the whole shell is wrapped with FRP. Both
experiments and nite element (FE) analysis demonstrated that
the ductility of the tubes was greatly enhanced by the FRP. Their
numerical investigation also showed that the elephants foot
buckling strength of a cylindrical shell under combined axial
compression and internal pressure can be signicantly enhanced
by the FRP wraps.
The aim of this paper is to investigate the strengthening of
thin metallic cylindrical shells by local application of FRP to
increase the elephants foot buckling strength. The elephants foot
buckling strength falls below the von Mises failure criterion
applied to the membrane stress resultants in the shell wall
because of the effect of local bending near the boundary
condition, which is exaggerated by geometrically nonlinear effects
[6]. In this paper, a linear shell bending analysis is used to obtain a
rst estimate of the effect of local FRP strengthening on increasing
the elephants foot buckling strength. Equations are developed for
the behaviour of the cylindrical shell with FRP strengthening, for
both simply supported and xed boundary conditions. The
optimal dimensions and location of the FRP strengthening are
examined.

Ny
p
pf
Qz
r
tb
ts
w
wm
wmb
xf

a
b
l
lb

nb
nfzy
nfyz
ns
$b
$c

1079

circumferential stress resultant


uniform internal pressure
conning pressure induced by the FRP shell
meridional shear stress resultant
radius of cylindrical shell
effective thickness of the composite FRPmetal shell
thickness of the metal cylinder
radial displacement
membrane theory normal deection
membrane theory normal deection of the composite
FRPmetal shell
starting position of FRP sheet above the base
extensional stiffness ratio (Eq. (8))
rotation around circumference
bending half-wavelength
meridional bending half-wavelength of the composite
FRPmetal shell
Poissons ratio of the composite FRPmetal shell
Poissons ratio of the FRP shell in the meridional
direction
Poissons ratio of the FRP shell in the circumferential
direction
Poisons ratio of the metal cylinder
normalised FRP height (Eq. (25e))
normalised FRP starting position (Eq. (29e))

In the following analysis, the boundary condition at the base of


the shell is considered to be either simply supported or xed. The
simply supported case (radial, axial and circumferential displacements restrained, but not the meridional rotation) is considered
rst, but a xed support is also later included. The shell is
assumed to be relatively long, so that the top boundary condition
does not inuence the local behaviour at the location of the
elephants foot buckle.
An FRP sheet with a height of hf is bonded to the external
surface of the cylindrical shell starting at a distance xf above the
base. It is assumed that the FRP sheet consists of uniaxial bres in
the circumferential direction. The FRP sheet is therefore treated as
orthotropic with Youngs modulus Efy in the circumferential
direction and Efz in the meridional direction, with Poissons ratios
nfyz and nfzy (nfyz/Efy nfzy/Ez).
It is assumed that the elephants foot buckling phenomenon
occurs over such a short length of shell that the internal pressure
and vertical load can both be treated as invariant with height z.
Thus a uniform internal pressure p and a vertical load per unit
circumference Nz at a distant top boundary are applied to the shell
as shown in Fig. 1.

2.2. Governing equations

2. Stress resultants in cylindrical shells strengthened with an


FRP sheet
2.1. Geometry, boundary conditions, loading and material properties
An isotropic elastic cylindrical shell of height h, radius r,
and thickness ts is strengthened by bonding FRP sheets to
its exterior. The wall thickness of the cylindrical shell is
assumed to be constant over the whole structure. The Youngs
modulus and the Poissons ratio of the shell are Es and ns,
respectively.

For the purposes of the analysis, the shell is divided into three
sections. In Sections A and C, above and beneath the FRP sheet, the
shell is isotropic and unstrengthened. In Section B, FRP jacketing is
applied, and the equations must capture the interaction between
the base shell and the FRP.
In Sections A and C, where the cylindrical isotropic shell has
uniform thickness t, the radial displacement w under a uniform
internal pressure p and vertical load per unit circumference Nz is
governed [21] by
4

d w Es t
vs N z
2 wp
r
dz4
r

(1)

ARTICLE IN PRESS
1080

M. Batikha et al. / Thin-Walled Structures 47 (2009) 10781091

in which

Nz

Db Ds Dfz

(7a)

Eb t b Es t s Ef y t f Es t s 1 a

(7b)

Shell section A

nb ns

Za

a
Qb2

Zb

hf

Mb2

Mc
p

xf

Qc

Zc

(8)

Eb t 3b =121  n2b
Db

Eb t b
Eb t b

Shell section C

(9)

Substituting Eqs. (7a) and (7b) into the left hand side of Eq. (9)
and rearranging the resulting equation gives
s
12Db
(10)
tb
Br

Fig. 1. Coordinate systems and stress resultants.

Steel shell

Ef y t f
Es t s

Section B may be treated as an equivalent metal shell.


The effective thickness tb of the transformed Section B may be
derived by equating the ratio of the exural to the extensional
rigidity of the equivalent shell to that of the metalFRP composite
shell:

Shell section B
Qb1

Mb1

(7c)

in which the extensional rigidity ratio of the FRP to the metal shell
is characterised by

Qa

Ma

N zf
N zs
nfzy
Nz
Nz

FRP shell

where Br is the extensional rigidity

pf

pf

Br

Fig. 2. Interaction pressure between the FRP and the metal shells within Section B.

in which D is the shell exural rigidity given by


D

Es t 3
121  n2s

(2)

In Section B, the metal shell expands as a result of both the


applied internal pressure p and (through Poisson effects) the
vertical loading Nz. The FRP sheet then induces a conning
pressure pf on the metal shell as shown in Fig. 2. The governing
equation for the metal shell in Section B is thus given by
d w Es t s
vs N zs
2 w p  pf
r
r
dz4

(3)

and that for the FRP sheet is


4
vfzy Nzf
d w Ef y t f
2 w pf
Dfz
r
r
dz4

(4)

where Nzs is the axial force in the cylindrical metal shell and Nzf is
the value in the FRP shell in Section B.
In Eqs. (3) and (4), Ds and Dfz are the exural rigidity of the
metal and FRP shells, respectively. They can be obtained by
substituting the thickness and respective properties for the z
direction into Eq. (2).
Eliminating pf from Eqs. (3) and (4) gives


4
Nzf N z
d w Es t s Ef y t f
Nzs
(5)
w

n
Ds Dfz 4
s
fz
y
Nz
Nz
r
r2
dz
By comparing Eq. (5) with Eq. (1), the differential equation for
Section B can be written as follows:
4

Db

d w Eb t b
v Nz
2 wp b
r
r
dz4

(6)

(11)

The same expression for tb can be found in Baker et al. [22] and
Zienkiewicz [23] for an orthotropic material.
If the FRP bres run in the circumferential direction, the
Youngs modulus of FRP in the meridional direction is very small
compared with that in the circumferential direction and that of
the metal shell. In addition, it is anticipated that only thin FRP
sheets would be required to achieve the strengthening purpose.
Therefore, in the meridional direction, both the exural and
extensional rigidities of the FRP are much smaller than those of
the metal shell and they can be neglected. In such a case, Eq. (5)
may be rewritten ignoring the terms relating to the meridional
properties of FRP
4

Ds

Ds

Eb t b
Es t s 1 a

1  n2b
1  n2b

d w Es t s 1 a
Nz

w p ns
r
dz4
r2

(12)

Similarly, the properties for an equivalent metal shell for


Section B can be obtained from Eqs. (7a)(7c) and (10) as
Db D s

Es t 3s
121  n2s

(13a)

nb ns

(13b)

and
s
121  n2s Ds
tb
Es t s 1 a

(13c)

2.3. General solution


The general solution of Eq. (1) may be written as [24]
nh
pz
pzi pz=l
w C 1 cos
C 2 sin
e
l
l i
o
h
pz
pz pz=l
C 3 cos
C 4 sin
1 wm
e

(14)

ARTICLE IN PRESS
M. Batikha et al. / Thin-Walled Structures 47 (2009) 10781091

in which
given by

is

the

meridional

bending

half-wavelength

p
rt s

p
31  v2s 1=4

(15)

and wm is the membrane theory normal deection


wm


vs N z r 2
r
Es t s

xb

lb

31  v2b 

wmb

p ns

p
rt b
1=4


C a1 C a2 

(21d)

pzb
lb

z  xf

(22)

lb

From Eq. (14), the deformations for Section B are:


w C b1 cos xb C b2 sin xb exb C b3 cos xb C b4 sin xb exb 1wmb

(23a)

(17a)

Nz
r
r Es t s 1 a

(17b)

The stress resultants in the shell may be expressed in terms of


displacements as
Et
Ny w vN z
R

(18a)

d w
M z D 2
dz

(18b)

M y vMz

(18c)
3

Q z D

p3

2.3.2. Shell section B


Section B is a composite shell with different properties as
described earlier. It also needs to be treated as a short shell.
Writing

(16)

For Section B, the meridional bending half-wavelength, lb,


and the membrane theory normal deection, wmb, can be obtained
as

Q a 2Dwm

1081

d w
dz3

(18d)

dw
p n
wmb
C b1 sin xb cos xb C b2 cos xb  sin xb exb
lb
dzb
o
C b3  sin xb cos xb C b4 cos xb sin xb exb
(23b)

 2 n
2
d w
p

2w
C b1 sin xb  C b2 cos xb exb
mb
lb
dz2b
o
C b3 sin xb C b4 cos xb exb

(23c)

 3 n
3
d w
p

2w
C b1  sin xb cos xb C b2 cos xb sin xb exb
mb
lb
dz3b
o
(23d)
C b3 sin xb cos xb C b4 cos xb  sin xb exb

The deformation and internal forces at zb 0 can be found from


Eqs. (18a)(18d) and (23a)(23d):
wb1 C b1 C b3 1wmb

2.3.1. Shell Section A


It is assumed that Section A of the metal cylindrical shell is so
long that the upper boundary does not inuence the deformations. Writing
xa

pza
l

pz  xf  hf
l

(19)

From Eq. (14):

bb1 wmb

p
lb

C b1 C b2 C b3 C b4 


Mb1 2Db wmb

2

p
lb

Q b1 2Db wmb

w C a1 cos xa C a2 sin xa exa 1wm

(24a)

3

lb

(20a)

(24b)

C b2 C b4 

(24c)

C b1 C b2  C b3 C b4 

(24d)

Similarly, those at zb hf are:


dw
p
b
wm C a1 sin xa cos xa C a2 cos xa  sin xa exa
l
dza
(20b)
2
p2
d w
2wm
C a1 sin xa  C a2 cos xa exa
2
l
dza

(20c)

3
p3
d w
2wm
C a1  sin xa cos xa C a2 cos xa sin xa exa
l
dz3a

(20d)
The deformation and internal forces at the junction between
shell Sections A and B (za 0 in Fig. 1) can be found from
Eqs. (18a)(18d) and (20a)(20d):
wa C a1 1wm

ba wm C a1 C a2 
l
M a 2Dwm

p2

C a2

wb2 fC b1 cos $b C b2 sin $b e$b


C b3 cos $b C b4 sin $b e$b 1gwmb

(21c)

p

C b1 sin $b cos $b C b2 cos $b  sin $b e$b



C b3  sin $b cos $b C b4 cos $b sin $b e$b

bb2 wmb

lb

(25b)

C b3

2

C b1 sin $b  C b2 cos $b e$b



sin $b C b4 cos $b e$b

Mb2  2Db wmb

Q b2  2Db wmb

(21a)
(21b)

(25a)

lb

lb

3

(25c)


C b1  sin $b cos $b C b2 cos $b sin $b e$b

C b3 sin $b cos $b C b4 cos $b  sin $b e$b

(25d)

where

$b

phf
lb

(25e)

ARTICLE IN PRESS
1082

M. Batikha et al. / Thin-Walled Structures 47 (2009) 10781091

2.3.3. Shell Section C


Section C, between the base and the reinforced section, is also
taken to be a short shell. Writing
xc

pzc
l

pz

(26)

Substituting Eqs. (28a)(28d) into Eqs. (30a) and (30b) gives:


C c3 C c1 1

(31)

C c4 C c2

(32)

The deformations in Section C can be obtained from Eq. (14):


2.4.2. Continuity at Joint 1
The deformation continuity at Joint 1 requires that

w fC c1 cos xc C c2 sin xc exc


C c3 cos xc C c4 sin xc exc 1gwm

(27a)

pn

dw
wm
C c1 sin xc cos xc C c2 cos xc  sin xc exc
l
dzc
o
(27b)
C c3  sin xc cos xc C c4 cos xc sin xc exc

p2 n
d w
2wm
C c1 sin xc  C c2 cos xc exc
2
l
dzc
o
C c3 sin xc C c4 cos xc exc

(27c)

p3 n
d w
2wm
C c1  sin xc cos xc C c2 cos xc sin xc exc
3
l
dzc
o
(27d)
C c3 sin xc cos xc C c4 cos xc  sin xc exc
3

The deformation and internal forces at zc 0 can be found from


Eqs. (18a)(18d) and (27a)(27d):

bc0 wm C c1 C c2 C c3 C c4 
l
M c0 2Dwm

Q c0 2Dwm

p2

p3

C c2 C c4 

C c1 C c2  C c3 C c4 

(28b)

(28c)

(28d)

(34)

C c1 sin $c cos $c e$c  C c2 cos $c  sin $c e$c 


C c3  sin $c cos $c e$c  C c4 cos $c sin $c e$c 
C b1 F 1  C b2 F 1  C b3 F 1  C b4 F 1 0

(35)

where
F1

wmb l
wm lb

(36)

2.4.3. Equilibrium of Joint 1


Equilibrium at Joint 1 requires
Mc M b1

(37a)

Q c Q b1

(37b)

Substituting Eqs. (24c), (24d), (29c) and (29d) into Eqs. (37a)
and (37b) gives
C c4 cos $c e$c C b2 F 2  C b4 F 2 0

(38)

(29a)
$c

C c1 sin $c cos $c C c2 cos $c  sin $c e



(29b)
C c3  sin $c cos $c C c4 cos $c sin $c e$c

C c3 sin $c cos $c e$c  C c4 cos $c  sin $c e$c 


 F 3 C b1  C b2 F 3 C b3 F 3  C b4 F 3 0

(39)

where

p2 

C c1 sin $c  C c2 cos $c e$c



sin $c C c4 cos $c e$c

M c  2Dwm

F2
(29c)

wmb l Db
wm l2 D

(40)

and

p3 

C c1  sin $c cos $c C c2 cos $c sin $c e$c


l

(29d)
C c3 sin $c cos $c C c4 cos $c  sin $c e$c

Q c  2Dwm

C c3 cos $c e$c C c4 sin $c e$c


w
w
w
 C b1 mb  C b3 mb 1  mb 0
wm
wm
wm

C c1  sin $c cos $c e$c  C c2 cos $c sin $c e$c 

p

C c3

(33b)

Substituting Eqs. (24a), (24b), (29a) and (29b) into Eqs. (33a)
and (33b) gives

C c1 sin $c e$c  C c2 cos $c e$c  C c3 sin $c e$c

wc fC c1 cos $c C c2 sin $c e$c

bc wm

bc bb1

(28a)

Those at z xf are:
C c3 cos $c C c4 sin $c e$c 1gwm

(33a)

C c1 cos $c e$c C c2 sin $c e$c

wc0 C c1 C c3 1wm

wc wb1

F3

wmb l Db
wm l3 D

(41)

where

$c

pxf
l

(29e)

2.4. Solution for a pinned-based shell


2.4.1. Boundary conditions at the base
The rst boundary condition examined here is a simply
supported base. The boundary conditions can be expressed as
wc0 0

(30a)

M c0 0

(30b)

2.4.4. Continuity at Joint 2


Continuity at Joint 2 requires
wa wb2

(42a)

ba bb2

(42b)

Substituting Eqs. (21a), (21b), (25a) and (25b) into Eqs. (42a)
and (42b) gives
C b1 cos $b e$b C b2 sin $b e$b C b3 cos $b e$b


wm
wm
1
0
C b4 sin $b e$b C a1 
wmb
wmb

(43)

ARTICLE IN PRESS
M. Batikha et al. / Thin-Walled Structures 47 (2009) 10781091

C b1 sin $b cos $b e$b  C b2 cos $b  sin $b e$b 


C b3  sin $b cos $b e$b  C b4 cos $b sin $b e$b 
 


1
1
C a2 
0
(44)
C a1
F1
F1

2.4.5. Equilibrium of Joint 2


The equilibrium of Joint 2 requires that
M a M b2

(45a)

Q a Q b2

(45b)

Substituting Eqs. (21c), (21d), (25c) and (25d) into Eqs. (45a)
and (45b) gives
C b1  sin $b e$b C b2 cos $b e$b C b3 sin $b e$b


1
0
C b4  cos $b e$b C a2 
F2

(46)

C b1  sin $b cos $b e$b  C b2 cos $b sin $b e$b 


C b3 sin $b cos $b e$b 




1
1
C a2 
0
C a1 
F3
F3

C b4 cos $b  sin $b e$b 


(47)

Substituting Eqs. (31) and (32)


rearranging the resulting equations in
2
A11
0
A13 A14 A15 A16
0
6
0
6 A21 A22 A23 A24 A25 A26
6
6 0
A
A
A
A
0
A
32
33
34
35
36
6
6
0
6 A41 A42 A43 A44 A45 A46
6
6 0
0
A
0
A
0
A
55
57
53
6
6
6 0
0
0
A64
0
A66 A67
6
6 0
0
A73 A74 A75 A76 A77
4
0

A83

A84

A85

A86

A87

into Eqs. (34)(47) and


matrix form gives
3 8
9 8 9
C a1 >
0
B1 >
>
>
>
>
>
>
>
>
>
>
7 >
>
>
>
>
>
0 7 >
C
0 >
>
>
>
>
a2
>
>
>
7 >
>
>
>
>
> >
>
>
>
>
>
>
0 7
C
0
>
>
>
>
b1
7 >
>
>
>
>
>
>
>
> >
> <
>
7 <
=
0 7
C b2
0 =
7

7
A58 7 >
>
> C b3 >
> B5 >
>
>
>
>
> >
> >
>
7 >
>
>
>
>
>
A68 7 >
C b4 >
B6 >
>
>
>
>
>
>
7 >
>
> >
> >
>
>
>
>
7
>
>
>
A78 5 >
C
B
7
c1
>
>
>
>
> >
>
>
>
> >
>
:
:
;
A88
C c2
B8 ;
(48)

which may be expressed as


A  fCg fBg

(49)

The coefcients Aij and Bi in Eq. (48) are presented in Appendix


A for a pinned base. The solution {C} is
fCg A1  fBg

(50)

Note that coefcients Cc3 and Cc4 are not included in


Eqs. (48)(50) but are given in Eqs. (31) and (32).
2.5. Radial displacement of the strengthened shell
Having obtained the coefcients from Eq. (50), the radial
displacement of the strengthened shell can be calculated from

and thickness ts 5 mm was studied here. This shell has a radiusto-thickness ratio of 1000, which is a medium length cylinder
according to Eurocode 3 Part 1.6 [25] for the purposes of buckling
strength evaluation. The Youngs modulus Es was taken as 200 GPa
and the Poissons ratio ns as 0.3. An FRP sheet was bonded onto the
shell at a distance xf above the base. The FRP sheet had a height hf
and thickness tf. The elastic moduli Efy, Efz in circumferential and
vertical directions and the Poissons ratio nfyz were taken as 230 GPa,
3 GPa and 0.35, respectively. An internal pressure of 0.25 N/mm2
and a vertical load of 500 N/mm were applied. The effects of the FRP
sheet rigidity, height and position were explored.
Fig. 3 shows the deformed shape of the shell near the base, for
different quantities of FRP. The maximum deection is closely
related to the peak circumferential stress, so a peak in this
deformed shape is strongly coupled with the formation of the
elephants foot buckle. The effect of changing the quantity of FRP
can be seen in Fig. 3 for the condition where hf/l 1 and the FRP
starts from the base of the shell (xf 0). The amount of FRP is
represented by the normalised FRP rigidity parameter a (Eq. (8)).
When no FRP sheet is used, the maximum radial displacement
occurs at a height of 0.75l above the base with the maximum
deection equal to 1.067wm.
A small amount of FRP (e.g. a 0.05) can reduce the maximum
deection but it cannot eliminate the peak. Whilst a larger amount
of FRP (e.g. aX0.2) can eliminate the peak within the region with
FRP, it produces a peak above the FRP zone and the maximum
deection above the FRP zone increases as the amount of FRP
increases. Using a very large FRP thickness (a 50) reduces the
deection in the strengthened zone close to zero and the maximum
deection occurs instead at another position above the FRP sheet.
Clearly, using a large amount of FRP does not help in preventing the
elephants foot buckling from occurring, but simply moves the
location of the buckle to another location. There is an optimal
amount of FRP, with a value of a around 0.1, which can reduce the
maximum deection near the base to the membrane theory
deection, without introducing a new peak elsewhere.
Fig. 4 shows the effect of the normalised FRP height, hf/l, when
the thickness and the starting position of FRP remain constant
with a 0.1 and xf/l 0. When the FRP is only attached to a small
section at the bottom of the shell (e.g. hf/l 0.5), the strengthening is not very effective because a large peak value remains. If the
FRP height is increased, the peak deection reduces. However, if
the FRP height is too large the cylinder starts to behave as if it
were two separate sections of shell with the lower section
behaving as a metalFRP composite shell and the upper as the
original metal shell. The lower composite shell results in a peak at
a distance of 0.75lb from the base. A peak also appears above the
position where the FRP ends due to bending arising from
discontinuity at the joint. This repeats and reinforces the
observation made above that both too little and too much FRP
will reduce the elephants buckling strength compared with the
case with an optimal amount of FRP.

8
fC cos xc C c2 sin xc exc C c3 cos xc C c4 sin xc exc 1gwm
>
>
< c1
w fC b1 cos xb C b2 sin xb exb C b3 cos xb C b4 sin xb exb 1gwmb
>
>
: fC a1 cos x C a2 sin x ex 1gwm

3. Patterns of deformation in the shell


The radial displacements and internal forces of the cylindrical
shell, with an axisymmetric FRP sheet of any size bonded at any
location, can be obtained from the equations above. An example
cylindrical shell with a height h 5000 mm, radius r 5000 mm

1083

for 0pzoxf
for xf pzoxf hf

(51)

for hf xf pz

Fig. 5 demonstrates the effect of the starting position of the


FRP, xf/l, when both the thickness and the height of the FRP are
kept constant (a 0.1 and hf/l 1.0). It appears that the
maximum radial displacement is minimised when xf is about
0.25, in which case the FRP sheet is centred at a height of 0.75l
above the base.

ARTICLE IN PRESS
1084

M. Batikha et al. / Thin-Walled Structures 47 (2009) 10781091

Normalized distance above base z/

2.50
Without FRP
Alpha = 0.05
Alpha = 0.1
Alpha = 0.2
Alpha = 0.5
Alpha = 1
Alpha = 50

2.25
2.00
1.75
1.50
1.25
1.00
0.75
0.50
0.25
0.00
0

0.1

0.2

0.3
0.4
0.5
0.6
0.7
0.8
Normalized radial displacement w/wm

0.9

1.1

Fig. 3. Effect of FRP rigidity on shell deformation (hf/l 1 and xf/l 0).

Normalized distance above base z/

6.00
Without FRP
Normalized FRP height = 0.5
Normalized FRP height = 0.75
Normalized FRP height = 1
Normalized FRP height = 1.25
Normalized FRP height = 4

5.25
4.50
3.75
3.00
2.25
1.50
0.75
0.00

0.9 0.92 0.94 0.96 0.98 1 1.02 1.04 1.06 1.08 1.1 1.12 1.14 1.16 1.18 1.2
Normalized radial displacement w/wm

Normalized distance above base z/

Fig. 4. Effect of normalised FRP height on shell deformation (a 0.1 and xf/l 0).

3.00
2.75
2.50
2.25
2.00
1.75
1.50
1.25
1.00
0.75
0.50
0.25
0.00

Without FRP
xf/lamda = 0
xf/lamda = 0.25
xf/lamda = 0.5
xf/lamda = 0.75

0.9

0.92

0.94

0.96 0.98
1
1.02 1.04 1.06
Normalized radial displacement w/wm

1.08

1.1

Fig. 5. Effect of FRP start position on shell deformation (a 0.1 and hf/l 1).

4. Optimal FRP strengthening

with the local value of the circumferential stress resultant Ny to


produce failure according to von Mises criterion

4.1. Failure criterion


In elephant foot failures, the membrane stress resultants
dominate the failure of the shell when it is isotropic and
unstiffened [12]. The axial membrane stress resultant Nz interacts

N2eq N2z Nz Ny N 2y

(52)

Since Nz is unchanged by deformation, the highest von Mises


stress resultant arises when Ny is at its peak value. Further,

ARTICLE IN PRESS
M. Batikha et al. / Thin-Walled Structures 47 (2009) 10781091

because Ny is linearly related to the radial displacement w


according to Eq. (18a), the peak value Ny is located at the point
where w has its maximum value. From Eq. (14) for a long
cylindrical shell, it can be shown that the maximum w is located
at z 0.75l when the cylindrical shell is not strengthened. The
peak value of deection is then 1.067wm as shown in Fig. 3.

4.2. Estimation of optimal FRP strengthening


It is clear that the extensional rigidity of the FRP (related to a),
the normalised height of the FRP sheet (hf/l) and its normalised
lower edge position (xf/l) affect the deformation of the strengthened shell. Fig. 6 shows the effect of the normalised FRP height, hf/
l, and the normalised start position, xf/l, on the maximum
displacement when a 0.1. It is seen that each curve has a
minimum, at which the position of the FRP is optimal. A close
analysis shows that the minimum of the maximum displacement
is achieved if the FRP is centred at a distance of 0.75l above the
base, i.e.
xf 0:75l 

hf
2

Based on the failure criterion of Eq. (52), the deformation of the


cylindrical shell strengthened with optimal FRP may be expected
to look like Fig. 7, which satises the boundary condition at the
base whilst the radial deformation is not greater than the
membrane theory solution anywhere in the shell.
For the deformed shape of the shell as shown in Fig. 7, the
radial displacement and the rotation at the joint between shell
Sections A and B (Fig. 1) should be wm and 0, respectively. This
leads to the condition that Ca1 Ca2 0 according to Eqs. (21a)
and (21b). Substituting this condition into Eqs. (43)(47) gives


cos $b
wm
 e$b 
1
(54a)
C b1
2
wmb
C b2



sin $b
wm
 e$b 
1
2
wmb

(54b)

C b3



cos $b
wm
 e$b 
1
2
wmb

(54c)

C b4



sin $b
wm
 e$b 
1
2
wmb

(54d)

(53)

Further analysis shows that the same relationship applies for


different values of a.

1085

where $b is given in Eq. (25e).


Further, if it is assumed that the deformations match the membrane theory values (w wm, b 0) at z 0.75l above the base

1.08

Normalized maximum radial


displacement wmax/wm

Minimize the maximum displacement: xf = 0.75-hf/2


1.06

1.04
hf/lamda = 0.1

1.02

hf/lamda = 0.25
hf/lamda = 0.35

hf/lamda = 0.5
0.98
0.25

0.5

0.75

1.25

1.5

1.75

Normalized FRP start position xf /

Normalized distance above base z/

Fig. 6. Effects of hf/l and xf/l on the maximum radial displacement (a 0.1).

3.00
2.75
2.50
2.25
2.00
1.75
1.50
1.25
1.00
0.75
0.50
0.25
0.00
0

0.1

0.2

0.3
0.4
0.5
0.6
0.7
0.8
Normalized radial displacement w/wm

0.9

Fig. 7. The ideal deformation of a shell with optimal FRP strengthening.

1.1

ARTICLE IN PRESS
1086

M. Batikha et al. / Thin-Walled Structures 47 (2009) 10781091

Normalized distance above base z/

3.00
2.75
2.50

alpha = 0.1, xf/Lamda = 0.53, hf/Lamda = 0.45

2.25

alpha = 0.15, xf/Lamda = 0.605, hf/Lamda = 0.289

2.00

alpha = 0.2, Xf/Lamda = 0.64, hf/Lamda = 0.217

1.75

alpha = 0.25, Xf/Lamda = 0.662, hf/Lamda = 0.175

1.50

Without FRP

1.25
1.00
0.75
0.50
0.25
0.00
0

0.1

0.2

0.3
0.4
0.5
0.6
0.7
0.8
Normalized radial displacement w/wm

0.9

1.1

Fig. 8. Deformation of the example shell strengthened with different optimal FRP sheets.

Normalized distance above base z/

3.00
2.75

alpha = 0.1, xf/Lamda = 0.53, hf/Lamda = 0.45

2.50

alpha = 0.15, xf/Lamda = 0.605, hf/Lamda = 0.289

2.25

alpha = 0.2, Xf/Lamda = 0.64, hf/Lamda = 0.217

2.00

alpha = 0.25, Xf/Lamda = 0.662, hf/Lamda = 0.175

1.75

Without FRP

1.50
1.25
1.00
0.75
0.50
0.25
0.00
0.97

0.98

0.99

1
1.01 1.02 1.03 1.04 1.05
Normalized radial displacement w/wm

1.06

1.07

Fig. 9. Detailed deformation of the example shell strengthened with different optimal FRP sheets.

where the displacement reaches its peak value for the unstrengthened
shell, then from Eqs. (27a) and (27b) together with the boundary
condition at the base (Eqs. (31) and (32)), the following are found

for $b. From Eqs. (25e) and (57), xf and hf can then be obtained
directly.

C c1 sin 2$  1  e2$ =e2$  2 sin 2$  e2$ 

4.3. A worked example

C c2

C c1 e$  e$ e$
tan $e$ e$

(55a)
(55b)

C c3 C c1 1

(55c)

C c4 C c2

(55d)

where $ 0U75p.
Substituting Eqs. (54a)(54d) and (55a)(55d) into Eq. (38)
results in:
C c1 sin $c e$c  C c2 cos $c e$c  C c3 sin $c e$c C c4 cos $c e$c


sin $b wm
F2 
 1  ewb  ewb 0
(56)
2
wmb
where F2 is given in Eq. (40) and

$c

pxf
l

0:75p  0:5$b

lb
l

The cylindrical shell described in Section 3 with the same


properties for both the cylindrical shell and the FRP sheet was
studied here to verify the above procedure in estimating an
optimal FRP strengthening scheme. The shell was assumed to be
under an internal pressure of 0.25 N/mm2 and vertical force
per unit circumference of 500 N/mm. For each value of the
circumferential extensional rigidity for the FRP, a, the above
procedure can be used to estimate optimal values for the FRP base
height xf and the height of the FRP strip hf. The results are shown
in Figs. 8 and 9.

5. Cylindrical shells with a xed base


5.1. Solution

(57)

Substituting Eq. (57) into Eq. (56), the resulting equation


contains only a single unknown $b. The equation can be solved

To study the effect of boundary conditions, a xed base was


also investigated. Compared with the shell with a pinned base, the
only difference in this case is that the rotational boundary

ARTICLE IN PRESS
M. Batikha et al. / Thin-Walled Structures 47 (2009) 10781091

Normalized distance above base z/

2.50

1087

Alpha = 0
Alpha = 0.05
Alpha = 0.1
Alpha = 0.2
Alpha = 0.5
Alpha = 1
Alpha = 50

2.00

1.50

1.00

0.50

0.00
0

0.1

0.2

0.3
0.4
0.5
0.6
0.7
0.8
Normalized radial displacement w/wm

0.9

1.1

Fig. 10. Effect of FRP rigidity on the deformation of a shell with xed base (hf/l 1 and xf/l 0).

Normalized distance above base z/

6
Normalized FRP height = 0

Normalized FRP height = 0.5


Normalized FRP height = 0.75

Normalized FRP height = 1


Normalized FRP height = 1.25

3
Normalized FRP height = 4

2
1
0
0.9

0.92

0.94

0.96
0.98
1
1.02
1.04
Normalized radial displacement w/wm

1.06

1.08

1.1

Fig. 11. Effect of normalised FRP height on the deformation of a shell with xed base (a 0.1 and xf/l 0).

condition at the base becomes bc0 0 at z 0 (instead of Mc0 0


as in Eq. (30b)). Applying this condition to Eqs. (28a)(28d) gives
C c4 2C c1  C c2 1

(58)

From Eqs. (31), (33a), (33b)(47) and (58), the solution can be
obtained in the same form as Eq. (48), but with different
coefcients Aij and Bi. These coefcients for a xed base are given
in Appendix B.
The same example as in Section 3 but with a xed base was
studied here. Fig. 10 shows the effect of the governing rigidity
parameter a (Eq. (8)) when the normalised FRP height is hf/l 1
and the FRP starts at the base (xf/l 0). For the unstrengthened
shell, the maximum radial displacement occurs at a height of 1.0l
above the base (instead of 0.75l for a pinned base) and has a value
of 1.043wm (c.f. 1.067wm for a pinned base). The addition of a small
amount of FRP can signicantly reduce the peak value. The
optimal value of a for a xed base is about 0.2 (c.f. a0.1 for a
pinned base). Again if a very large thickness of FRP is used (e.g.
a 50), the deection within the strengthened zone is mostly
eliminated and a new peak develops above the FRP.
Fig. 11 shows the effect of the normalised FRP height hf/l when
a 0.1 and xf/l 0. It shows that an increase of normalised FRP
height reduces the peak radial displacement initially but a further
increase introduces a peak above the FRP sheet which is less

effective than the optimal case. Fig. 12 demonstrates that the FRP
lower edge location (normalised starting position xf/l) can also
signicantly affect the shell deformation.
5.2. Optimal FRP strengthening
As was the case for the pinned-base cylinder, the parameters a,
hf/l and xf/l all inuence the deformation pattern in the
strengthened shell. Figs. 1012 show the effects of changing the
thickness (Fig. 10) and the height of the FRP sheet (Fig. 11), and its
start position (Fig. 12). All three gures use the unstrengthened
xed base cylinder as a reference case, where it can be seen that
the maximum radial displacement w, located at z l, attains the
peak value of 1.043wm.
The critical design information is contained in Fig. 13, where
the peak deection is plotted against the normalised start position
xf/l. As the height of the FRP sheet increases, the peak deection
can be steadily reduced towards wmax wm, but there is clearly a
narrowly dened critical height that should be used: the optimal
position for the FRP centres it on the level z l, which indicates
that
xf ;opt l 

hf
2

(59)

ARTICLE IN PRESS
1088

M. Batikha et al. / Thin-Walled Structures 47 (2009) 10781091

Normalized distance above base z/

3.00
Without FRP
2.50

xf/Lamda = 0.5
xf/Lamda = 0.75

2.00

xf/Lamda = 1

1.50
1.00
0.50
0.00
0.92

0.9

0.94

0.96 0.98
1
1.02 1.04 1.06
Normalized radial displacement w/wm

1.08

1.1

Fig. 12. Effect of FRP start position on the deformation of a shell with xed base (a 0.1 and hf/l 1.25).

Normalized maximum radial


displacement wmax/wm

1.06

Minimize the maximum displacement: xf = -hf/2


1.04

hf/lamda = 0.1

1.02

hf/lamda = 0.15
hf/lamda = 0.2
hf/lamda = 0.25
1
0

0.25

0.5
0.75
1
1.25
1.5
Normalized FRP start position xf /

1.75

Fig. 13. Effects of hf/l and xf/l on the maximum radial displacement of a xed-base shell.

Normalized distance above base z/

3.00
2.75
2.50
2.25
2.00
1.75
1.50
1.25
1.00
0.75
0.50
0.25
0.00
0

0.1

0.2

0.3
0.4
0.5
0.6
0.7
0.8
Normalized radial displacement w/wm

0.9

1.1

Fig. 14. The ideal deformation of a xed-base shell with optimal FRP strengthening.

A similar procedure to that for a shell with a pinned base is


developed here for estimating the optimal FRP parameters.
Assuming that the deformation of the strengthened shell with

optimal FRP strengthening is as shown in Fig. 14. Applying w wm


and b 0 at the joint between shell sections A and B leads to the
same outcome as for a pinned base (Eqs. (54a)(54d)). Further

ARTICLE IN PRESS
M. Batikha et al. / Thin-Walled Structures 47 (2009) 10781091

1089

Normalized distance above base Z/

3.00
2.75

alpha = 0.1, Xf/Lamda = 0.86, hf/Lamda = 0.28


alpha = 0.15, Xf/Lamda = 0.91, hf/Lamda = 0.184

2.50

alpha = 0.2, Xf/Lamda = 0.93, hf/Lamda = 0.138

2.25

alpha = 0.25, Xf/Lamda = 0.945, hf/Lamda = 0.11

2.00

Without FRP

1.75
1.50
1.25
1.00
0.75
0.50
0.98

0.99

1
1.01
1.02
1.03
Normalized radial displacement w/wm

1.04

1.05

Fig. 15. Deformation of a xed-base shell strengthened with different optimal FRP sheets.

applying w wm and b 0 to Eqs. (27a) and (27b) respectively, at


z l together with the boundary conditions Eqs. (31) and (58)
gives
C c1 

ep

ep
C c2
 ep

(60)

C c3 1 C c1

(61)

C c4 C c1 1

(62)

reducing the radial deformation of the shell, which would


increase the elephants foot buckling strength.

Acknowledgements
The authors are most grateful to Damascus University for
providing the rst authors studentship. Financial support from
the Royal Society Kan Tong Po Visiting Professorship for the third
author is also gratefully acknowledged.

Substituting Eqs. (6062) into Eq. (56) and considering

$c

pxf
l

p  0:5$b

lb
l

(63)

the optimal FRP sheet can be derived as for the pinned base from
Eq. (56).
To verify the practical procedure, the worked example
of Section 4.3 for the pinned base were studied again for
the xed base. Fig. 15 shows the radial displacement after
adding different optimal FRP sheets by using the equations
derived here.
Using the practical procedure, the radial displacement of the
shell can be seen to have decreased signicantly (Fig. 15). This
radial displacement is very close to the membrane theory uniform
displacement wm at the mid-point of the FRP sheet, located at l
above the base, and thus it satises the assumptions, which were
made for this design procedure in Section 5.2.

6. Conclusions
Cylindrical shells under internal pressure accompanied by
axial loads are susceptible to elephants foot buckling near the
base. This paper has presented a preliminary study of the
possibility of strengthening these shells against elephants foot
buckling by using externally bonded FRP. The linear elastic
bending equations for the strengthened cylindrical shell have
been derived and illustrated through exploratory examples. Both
pinned and xed base boundary conditions for the cylindrical
shell have been considered. It has been shown that a small
amount of FRP sheet, placed in the critical location, is optimal in

Appendix A. Matrices [A] and [B] in Eq. (48): pinned base


For an FRP strengthened cylindrical shell with a pinned base,
the coefcients in matrices [A] and [B] in Eq. (48) are given as
follows:
9
wm
>
>
A11 
>
>
wmb
>
>
$b >
>
=
A13 cos $b e
(A.1)
A14 sin $b e$b >
>
>
$
>
A15 cos $b e b >
>
>
>
A16 sin $b e$b ;
A21

9
>
>
>
>
>
>
>
>
>
>
>
>
>
=

1
F1

1
A22 
F1
A23 sin $b cos $b e$b >
>
>
A24 cos $b  sin $b e$b >
>
>
>
>
>
A25  sin $b cos $b e$b >
>
>
>
;
$
b
A cos $ sin $ e
26

1
A32 
F2
A33  sin $b e$b
A34 cos $b e$b
A35 sin $b

e$b

A36  cos $b e$b

(A.2)

9
>
>
>
>
>
>
>
>
=
>
>
>
>
>
>
>
>
;

(A.3)

ARTICLE IN PRESS
1090

M. Batikha et al. / Thin-Walled Structures 47 (2009) 10781091

1
A41 
F3
1
A42 
F3
A43  sin $b cos $b e$b
A44 cos $b sin $b e$b
A45 sin $b cos $b e$b
A46 cos $b  sin $b e$b

w
A53  mb
wm
w
A55  mb
wm
A57 cos $c e$c  cos $c e$c
A58

9
>
>
>
>
>
>
>
>
>
>
>
>
>
=

A21

(A.4)

>
>
>
>
>
>
>
>
>
>
>
>
>
;

A23
A24
A25
A26

9
>
>
>
>
>
>
>
=

(A.5)

A33

A34 cos $b e$b


A35 sin $b e$b
A36  cos $b e$b

9
>
>
>
=

(A.6)

A42

9
>
>
>
>
>
>
>
>
>
=

A73 F 3
A74 F 3
A75 F 3
A76 F 3

>
>
>
>
>
>
>
>
>
sin $c cos $c e$c  sin $c cos $c e$c ;

A44
A45
A46

A85 F 1
A86 F 1
A87
A88

>
>
>
>
>
>
sin $c cos $c e
sin $c  cos $c
>
>
>
$c
$
c ;
 sin $c cos $c e
sin $c cos $c e

wm
1
wmb
w
B5 mb  1 cos $c e$c
wm
B6  sin $c e$c
B1

>
>
>
>
>
B7 cos $c sin $c e$c >
>
>
>
;
$
c
B8 cos $c  sin $c e

16

(B.4)

wmb
wm
wmb

wm
cos $c e$c  cos $c e$c 2 sin $c e$c

9
>
>
>
>
>
>
>
=

A58 sin $c e$c  e$c

A64 F 2
A66 F 2

9
>
>
>
>
=

(B.5)

(B.6)

A67 sin $c e$c sin $c e$c 2 cos $c e$c >


>
>
>
;
A68  cos $c e$c e$c

(A.9)

A73 F 3
A74 F 3
A75 F 3

9
>
>
>
>
>
>
>
>
>
>
>
=

A76 F 3
>
A77  sin $c cos $c e$c sin $c cos $c e$c >
>
>
>
>
$
>
>
2 sin $c cos $c e c
>
>
>
;
A78 sin $c cos $c e$c   sin $c cos $c e$c
(B.7)

For an FRP strengthened cylindrical shell with a xed base,


the coefcients in matrices [A] and [B] in Eq. (48) are given as
follows:

A14 sin $b e$b >


>
>
>
A15 cos $b e$b >
>
>
>
A sin $ e$b ;

(A.8)

Appendix B. Matrices [A] and [B] in Eq. (48): xed base

9
wm
>
>
A11 
>
>
wmb
>
>
>
=
A13 cos $b e$b >

9
1
>
>
>
>
>
F3
>
>
>
>
1
>
>
>

>
=
F3
$b
 sin $b cos $b e
>
>
>
cos $b sin $b e$b >
>
>
>
>
$
>
b
>
sin $b cos $b e
>
>
>
;
$
b
cos $  sin $ e

>
>
>
>
>
>
>
;

A57

e$c

9
>
>
>
>
>
>
>
>
>
=

(B.3)

>
>
>
>
>
>
>
>
;

A53 
A55

A83 F 1
A84 F 1

9
>
>
>
>
>
>
>
>
=

(A.7)

A77  sin $c cos $c e$c

9
>
>
>
>
>
>
>
>
>
=

(B.2)

A41 

A43

$c

1
F2
 sin $b e$b

A32 

A67 sin $c e$c sin $c e$c >


>
>
;
A68 cos $c e$c  cos $c e$c

A78

1
F1
sin $b cos $b e$b >
>
>
cos $b  sin $b e$b >
>
>
>
>
$
>
 sin $b cos $b e b >
>
>
>
;
$
cos $b sin $b e b

A22 

>
>
>
>
>
>
>
$c
$
sin $c e
sin $c e c ;

A64 F 2
A66 F 2

9
>
>
>
>
>
>
>
>
>
>
>
>
>
=

1
F1

A83 F 1
A84 F 1
A85 F 1

(B.1)

A86 F 1
A87 sin $c cos $c e$c sin $c  cos $c e$c
2sin $c cos $c e$c
A88  sin $c cos $c e$c  sin $c cos $c e$c

9
>
>
>
>
>
>
>
>
>
>
>
=
>
>
>
>
>
>
>
>
>
>
>
;
(B.8)

ARTICLE IN PRESS
M. Batikha et al. / Thin-Walled Structures 47 (2009) 10781091

wm
1
wmb
w
B5 mb  1 cos $c  sin $c e$c
wm
B6 sin $c cos $c e$c
B1

B7 2 cos $c e$c
B8 2 sin $c e$c

9
>
>
>
>
>
>
>
>
>
=
>
>
>
>
>
>
>
>
>
;

(B.9)

References
[1] Rotter JM. Local collapse of axially compressed pressurized thin steel
cylinders. Journal of Structural Engineering, ASCE 1990;116(7):195570.
[2] Rotter JM. Elastic plastic buckling and collapse in internally pressurised
axially compressed silo cylinders with measured axisymmetric imperfections: interactions between imperfections, residual stresses and collapse. In:
Proceedings of the international workshop on imperfections in metal silos.
Lyon: CA-Silo; 1996. p. 11940.
[3] Rotter JM. Buckling of cylindrical shells under axial compression. In: Teng JG,
Rotter JM, editors. Buckling of thin metal shells. London: Spon; 2004. p.
4287.
[4] Rotter JM, Teng JG. Elastic stability of cylindrical shells with weld depressions.
Journal of Structural Engineering, ASCE 1989;115(5):124463.
[5] Teng JG, Rotter JM. Buckling of pressurized axisymmetrically imperfect
cylinders under axial loads. Journal of Engineering Mechanics, ASCE 1992;
118(2):22947.
[6] Rotter JM. Stress amplication in unstiffened cylindrical steel silos and tanks.
Civil Engineering Transactions, Institution of Engineers, Australia 1989;
CE31(3):1428.
[7] Chen JF, Rotter JM, Teng JG. Strengthening silos and tanks against elephants
foot buckling. In: Proceedings of the fourth conference on advances in steel
structures. Shanghai, China, 2005. p. 45966.
[8] Chen JF, Rotter JM, Teng JG. A simple remedy for elephants foot buckling in
cylindrical silos and tanks. Advances in Structural Engineering 2006;9(3):
40920.
[9] Hollaway LC, Teng JG, editors. Strengthening and rehabilitation of civil
infrastructures using FRP composites. Cambridge, UK: Woodhead Publishing
Limited; 2008.

1091

[10] Teng JG, Chen JF, Smith ST, Lam L. FRP strengthened RC structures. Chichester,
UK: Wiley; 2002.
[11] Teng JG, Chen JF, Smith ST, Lam L. Behaviour and strength of FRP-strengthened
RC structures: a state-of-the-art review. Proceedings of the Institution of Civil
EngineersStructures and Buildings 2003;156(SB1):5162.
[12] Hollaway LC, Zhang L, Photiou NK, Teng JG, Zhang SS. Advances in adhesive
joining of carbon bre/polymer composites to steel members for repair and
rehabilitation of bridge structures. Advances in Structural Engineering
2006;9(6):791803.
[13] Teng JG. Combination of bre-reinforced polymer and steel in structures. In:
Young B, editor. Proceedings of the international symposium on innovative
design of steel structures. Hong Kong, China: University of Hong Kong; 2006.
p. 1324.
[14] Zhao XL, Zhang L. State-of-the-art review on FRP strengthened steel
structures. Engineering Structures 2007;29(8):180823.
[15] Elchalakani, M., Bambach M. Plastic mechanism analysis of CHS stub columns
strengthened using CFRP. In: Xie M, editor. Proceedings of the fourth
international structural engineering and construction conference. Melbourne,
Australia; 2007. p. 71320.
[16] Chen JF. Load-bearing capacity of masonry arch bridges strengthened with
FRPs. Advances in Structural Engineering 2002;5(1):3744.
[17] Triantallou TC. Strengthening of masonry using epoxy-bonded FRP laminates. Journal of Composites for Construction, ASCE 1998;2(2):96104.
[18] Gilllan JR, Gilbert SG, Patrick GRH. The use of FRP composites in enhancing
the structural behavior of timber beams. Journal of Reinforced Plastics and
Composites 2003;22(15):137388.
[19] Teng JG, Hu YM. Suppression of local buckling in steel tubes by FRP jacketing.
In: Proceedings of the second international conference on FRP composites in
civil engineering. Adelaide, Australia; 2004. p. 74953.
[20] Teng JG, Hu YM. Behaviour of FRP-jacketed circular steel tubes and cylindrical
shells under axial comparison. Construction and Building Materials
2007;21(4):82738.
[21] Timoshenko SP, Woinowsky-Krieger S. Theory of plates and shells. New York:
McGraw-Hill; 1959.
[22] Baker EH, Kovalevsky L, Rish FL. Structural analysis of shell. 3rd ed. FL, USA:
Robert E. Krieger Publishing Company; 1986.
[23] Zienkiewicz OC. The nite element method in engineering science. 2nd ed.
London, UK: McGraw-Hill Publishing Company Limited; 1971.
[24] Rotter JM. Bending theory of shells for bins and silos. In: Design of steel bins
for the storage of bulk solids. School of Civil and Mining Engineering,
University of Sydney; 1985. p. 7181.
[25] EN 1993-1-6. Eurocode 3: design of steel structuresPart 16: strength and
stability of shell structures. Brussels: CEN; 2007.

You might also like