Sie sind auf Seite 1von 10

119

A simplied method for calculating the


stress of a large storage tank wall
Z P Chen
1
, YY Duan
1
, J MShen
2
, and J L Jiang
1
1
Institute of Process Equipment and Control Engineering, Zhejiang University, Hangzhou, Peoples Republic of China
2
Ningbo Special Equipment Inspection Center, Ningbo, Peoples Republic of China
The manuscript was received on 7 February 2007 and was accepted after revision for publication on 21 June 2007.
DOI: 10.1243/09544089JPME125
Abstract: With consideration of the inuences fromthe constraint reaction of tank bottomplate
toeachshell course, a simpliedlongshort shell methodwas proposedtocalculate the stress of a
large storage tank wall. The rst shell course was regarded as a short cylindrical shell while all the
others as long cylindrical shells, and the analytic solution equations of shell stress were achieved
by theory of plates and shells. With resistance stressstrainmethod, a eld stress test was done on
a 1510
4
m
3
oating-roof oil tank during its water lling test, and the measured stress data are in
good agreement with the calculation results obtained by this simplied calculation method for
shell stress. Therefore, the simplied method can be adopted to calculate the shell stress of large
storage tanks accurately.
Keywords: storage tank, mechanical analysis, stress test, theory of plates and shells
1 INTRODUCTION
Large storage tanks are the key equipment of strategic
petroleum reserve for every country. Including shells,
bottom plate, and tank roof, they are mainly vertical
metallic cylinders. They withstand hydrostatic pres-
sure of the stored liquid, which reduces from bottom
to top. To comply with the uniform-strength theory,
almost all large tanks are made by welding unequal-
thickness cylindrical shell courses together [1, 2]. To
avoid buckling due to wind load, wind girders, or
stiffening rings are welded on the upper part of the
cylinder. Bottom plate is composed of annular plate
and centre former. Considering the cost and the dif-
culty of construction, many tanks have oating roof
and sit directly on elastic foundation with concrete
ring beam. There are no anchor bolts betweenannular
plate and concrete ring beam.
For the lower part of the tank, the rst and the
second courses (from bottom to top, the same below)

Corresponding author: Institute of Process Equipment and Con-


trol Engineering, Zhejiang University, 38 Zheda Road, Hangzhou,
Zhejiang 310027, Peoples Republic of China. email: zhip-
ing@zju.edu.cn
have the most complicated stress condition of the
whole tank. Furthermore, maximum stress appears
here. In addition to hydrostatic stress, constraint reac-
tion from foundation and concrete ring beam also
make the calculation more difcult. At present, two
typical methods for calculating shell stress are used,
short shell method and long shell method.
In short shell method, unequal-thickness shell
courses are considered as cylindrical shells, both sides
of which are under local stress. This method is accu-
rate, but involves complicated procedure. In long
shell method, each shell plate is considered as semi-
innite cylinder shell, which means that only single
side of each shell is under local stress. This method
is simple but it neglects the constraint reaction from
bottom plate to the second course, which causes big
errors especially in large storage tank whose diam-
eter is over 80 m. Therefore, to calculate the shell
stress accurately and briey, this paper proposed a
model in such a way that, the rst shell was regarded
as a short cylindrical shell while the others as long
cylindrical shells, bottom plate is seen as rigid-elastic
coupling grade beam, and the analytical solution
of shell stress was derived from the simultaneous
equations of bottom plate, the rst and the second
courses.
JPME125 IMechE 2007 Proc. IMechE Vol. 221 Part E: J. Process Mechanical Engineering
120 Z P Chen, YY Duan, J MShen, and J L Jiang
2 PRESENT CALCULATIONMETHODS
2.1 Short shell method
According to the stress features of large tanks,
unequal-thickness shell courses are considered as
cylindrical shells, both sides of which are under local
stress. The mechanical model of the ith shell course is
shown in Fig. 1.
In accordance with theory of plates and shells [3],
the deection equation of cylindrical shell in presence
of fringe force and bending moment is given by
y
i
= C
1i

1
(F
i
x
i
) + C
2i

2
(F
i
x
i
) + C
3i

3
(F
i
x
i
)
+ C
4i

4
(F
i
x
i
) +
g(H
i1
x
i
)
K
i
(1)
where y
i
is the deection of the ith shell course; C
1i
,
C
2i
, C
3i
, and C
4i
are indeterminate constants; H
i1
is
the calculated water level at the (i 1)-th to ith course
junction; K
i
= Et
i
/R
2
; R is shell radius; t
i
is the thick-
ness of the ith shell course; E is the steel elasticity
modulus; is the water density; g is the gravity accel-
eration. To simplify the calculation, function
is applied

1
(X) = cosh X cos X

2
(X) =
1
2
(cosh X sin X + sinh X cos X)

3
(X) =
1
2
sinh X sin X

4
(X) =
1
4
(cosh X sin X sinh X cos X)
Whenx
i
= 0 andx
i
= l
i
, there are boundary conditions
as follows
D
i
_
d
2
y
i
dx
2
i
_
x
i
=0
= M
i1
D
i
_
d
3
y
i
dx
3
i
_
x
i
=0
= Q
i1
D
i
_
d
2
y
i
dx
2
i
_
x
i
=l
i
= M
i
D
i
_
d
3
y
i
dx
3
i
_
x
i
=l
i
= Q
i

(2)
where, D
i
is the stiffness coefcient of the ith shell
course, D
i
= Et
3
i
/12(1
2
).
Furthermore, there are deformation compatibility
formulae at the junction of neighbouring courses
y
i
|
x
i
=l
i
= y
i+1
|
x
i+1
=0
dy
i
dx
i

x
i
=l
i
=
dy
i+1
dx
i+1

x
i+1
=0

(3)
Fig. 1 Mechanical model of the ith shell course in short
shell method
As shown in Fig. 1, the number of unidentied fringe
forces is 2(n + 1) whenthe shell number is n. For oat-
ing roof tank, M
n
= Q
n
= 0 happens on the top of the
shell. M
0
and Q
0
can be derived by solving the calcu-
lation equations of bottom plate stress. According to
equation (3), the 2(n 1) deformation compatibility
formulae can be obtained for the 2(n 1) unidenti-
ed fringe forces at the (n 1) course junctions. Then,
the four coefcients C
1i
, C
2i
, C
3i
, and C
4i
are obtained
by equation (2). Substituting them into equation (1),
leads to the deection equations of shell plates.
For the exure theory of rotation shell, the longitu-
dinal stress on the outside and inner side of a cylinder
wall is given by

xi
=
6D
1
y

i
t
2
i
(4)
The hoop stress on the outside and inner side of
cylinder wall is given by

i
=
E
R
y
i

6D
1
y

i
t
2
i
(5)
In equations (4) and (5),
xi
is the longitudinal stress of
the ith course.
i
is the hoop stress of the ith course.
Symbol + represents the stress on the outside of the
cylinder wall, and symbol represents the stress on
the inner side of the wall.
The short shell method is quite accurate since inter-
actions of neighbouring shell courses are considered.
But for large tanks, whose unequal-thickness shells are
more than6 andunknowns of eachshell are more than
six, ten or even more equations are needed to obtain
the theoretical stress. It is so complicated that it is not
often used in engineering.
2.2 Long shell method
For convenience, in engineering, tank walls are gen-
erally considered as semi-innite cylindrical shells.
Local stress is assumed to exist on single side of each
Proc. IMechE Vol. 221 Part E: J. Process Mechanical Engineering JPME125 IMechE 2007
A simplied method for calculating the stress of a large storage tank wall 121
Fig. 2 Mechanical model of the ith shell course in long
shell method
tankwall. The mechanical model of the ithshell course
is shown in Fig. 2.
When innite cylindrical shells are under local
stress, C
3i
= C
4i
= 0 in equation (1). Compared with
short shell method, two undetermined coefcients
are omitted for each shell. Based on deformation
compatibilityformulaeof bothdisplacement androta-
tion angle at the course junction, fringe forces and
moments at junctions can be derived.
In reality, shell courses of tanks are not innite, so
superpositionmethodis oftenadoptedtocalculatethe
deectionof eachshell withconsiderationof thefringe
forces at the course junctions.
Long shell method has the advantages of being sim-
ple but it is not accurate. Especially for the rst hoop
weld, calculation results differ greatly from measured
values of the shell stress. Actually, the shear force and
bending moment are rather big at the shellbottom
junction. It cannot be neglected because it has a great
inuence onthe stress distributionof the secondshell.
3 COMBINEDCYLINDRICAL SHELL METHOD
Based on the advantages and disadvantages of short
and long shell methods, this paper proposes a
mechanical analysis model of combined cylindrical
shell. The rst shell was regarded as a short cylindri-
cal shell while the others as long cylindrical shells,
and analytical solution of shell stress was derived by
simultaneous equations of bottom plate, the rst and
the second courses. Figure 3 shows the mechanical
analysis model of combined cylindrical shell under
hydraulic pressure and local stress.
In this analysis, effects of openings and differential
settlement of tank foundation are ignored. The entire
tank is seen as a cylinder under hydrostatic pressure.
3.1 Stress analysis of the second and the above
courses of a tank wall
Second and all the above courses of tank shell are
regarded as long cylindrical shells, which is shown in
Fig. 3(a). Under single-side fringe force and hydraulic
Fig. 3 Bottomplate and shell model: (a) second and the
above courses and (b) bottom plate and the rst
course
pressure, deection equations are as follows
y
A
=
2F
i
K
i
e
F
i
x
A
[Q
i
cos F
i
x
A
F
i
M
i
(cos F
i
x
A
sinF
i
x
A
)] +
g(H
i
+ x
A
)
K
i
(6)
where, y
A
is the deection of the ith course; x
A
is the
distance from the ith course to the weld joint; Q
i
,
M
i
are the fringe shear force and moment at the
ith to (i + 1)-th course junction, respectively; K
i
=
Et
i
/R
2
, F
i
=
4

K
i
/4D
i
, D
i
= Et
3
i
/12(1
2
); H
i
is the cal-
culated water height at the ith to (i + 1)-th course
junction
y
B
=
2F
i+1
K
i+1
e
F
i+1
x
B
[Q
i
cos F
i+1
x
B
F
i+1
M
i
(cos F
i+1
x
B
sinF
i+1
x
B
)] +
g(H
i
x
B
)
K
i+1
(7)
where y
B
is the deection of the (i + 1)-th course; x
B
is the distance from the (i + 1)-th course to the weld
joint.
The compatible conditions of displacement and
rotation angle at the junction, when x
A
= x
B
= 0,
JPME125 IMechE 2007 Proc. IMechE Vol. 221 Part E: J. Process Mechanical Engineering
122 Z P Chen, YY Duan, J MShen, and J L Jiang
y
A
= y
B
, y

A
= y

2F
i+1
K
i+1
+
2F
i
K
i
2F
2
i
K
i

2F
2
i+1
K
i+1
2F
2
i+1
K
i+1

2F
2
i
K
i

4F
3
i+1
K
i+1

2F
3
i
K
i

_
Q
i
M
i
_
=

_
1
K
i

1
K
i+1
_
H
i
g
_
1
K
i

1
K
i+1
_
g

(8)
The local stress can be obtained according to
equation (8).
Each tank wall undergoes two groups of local stress
and hydraulic pressure. The longitudinal stress and
hoopstress of theithcoursearecalculatedby adopting
superposition method

xi
=
6
t
2
i
e
F
i
x
i
_

Q
i
sin(F
i
x
i
)
F
i
+ M
i
[sin(F
i
x
i
) + cos(F
i
x
i
)]
_
+
6
t
2
i
e
F
i
(l
i
x
i
)
_
M
i+1
[sin(F
i
(l
i
x
i
))
+ cos(F
i
(l
i
x
i
))] +
Q
i+1
sin(F
i
(l
i
x
i
))
F
i
_
(9)
where l
i
is the height of the ith course; 0< x
i
< l
i
,
i = 2, 3, 4, 5, 6. Symbol + represents the stress on the
outside of the cylinder wall, and symbol represents
the stress on the inner side of the wall.

i
=
E
R
_
2F
i
K
i
e
F
i
x
i
[Q
i
cos F
i
x
i
F
i
M
i
(cos F
i
x
i
sinF
i
x
i
)] +
2F
i
K
i
e
F
i
(l
i
x
i
)
[Q
i+1
cos F
i
(l
i
x
i
)
F
i
M
i+1
(cos F
i
(l
i
x
i
) sinF
i
(l
i
x
i
))]
+
g(H
i
x
i
)
K
i
_
+
xi
(10)
whereif
i
represents thestress ontheoutsideor inner
side of the cylinder wall depends on
xi
of equation(9).
When the longitudinal stress on the outside of the
cylinder wall (symbol is + in equation (9)) is sub-
stituted into equation (10),
i
is determined to be the
hoop stress on the outside of the cylinder wall. And it
works both ways.
The stress of the secondcourse cannot be calculated
by equations (9) and (10) until M
1
and Q
1
are obtained
by simultaneous equations of bottom plate and the
rst course.
3.2 Stress analysis of the rst course
The rst shell course is considered as a short cylindri-
cal shell. Its mechanical model is shown in Fig. 3(b).
Deection of the rst course is described by the
following equation
y
1
= C
1

1
(F
1
x
1
) + C
2

2
(F
1
x
1
) + C
3

3
(F
1
x
1
)
+ C
4

4
(F
1
x
1
) +
g(H x
1
)
K
1
(11)
There are six unknowns or indeterminate con-
stants (C
1
, C
2
, C
3
, C
4
, M
1
, Q
1
) in equations (11) and
(7). There are four equations for compatible dis-
placements, rotation angles, bending moments, and
shear forces at the junctions; and two equations
since displacements are zero and bending moments
equal at the shell-bottom junction. Six equations
for six unknowns C
1
, C
2
, C
3
, C
4
, M
1
, and Q
1
can
be obtained to determine the deection equation
y
1
of the rst course. Then y
1
will be substituted
into equations (4) and (5) to obtain the theoretical
stress.
Meanwhile, the rotation angle at the shell-bottom
junction can be derived by equation (11)

w
= C
2
F
1

g
K
1
(12)
where K
1
= Et
1
/R
2
, F
1
=
4

K
1
/4D
1
, D
1
= Et
3
1
/12
(1
2
)
Equation (12) is used for stress calculation of the
bottom plate.
3.3 Stress analysis of bottomplate
Inmost cases, largeoil tanks sit directlyonelasticfoun-
dation. Therefore, a coupling method of rigid-elastic
grade beam is used, as shown in Fig. 3(b). After the
tank is lled, the bottom plate adjacent to the rigid
grade beamwill tilt, and the uplift distance is L. As ver-
ied in references [4] and [5], calculation formulae of
bending moment, bearing pressure and rotating angle
are as follows
M
0
=
0.25P(2A
1
A
4
A
2
A
3
) + GC(A
2
C
2
A
1
)
A
1
3A
2
C
2
M
p
=
0.5PA
4
GC M
0
A
2
R
2
=
0.5P(L
2
C
2
) + GC + M
p
+ M
0
L

b
=
1
D
0
_
R
2
2
(L C)
2

P
6
(L C)
3
M
p
(L C)
1
2F
2
0
(R
2
+ 2F
0
M
p
)
_

(13)
where C is the extended length from tank wall centre
to outer edge of annular plate; L is the uplift distance
Proc. IMechE Vol. 221 Part E: J. Process Mechanical Engineering JPME125 IMechE 2007
A simplied method for calculating the stress of a large storage tank wall 123
up to which the bottom plate lifts the foundation;
t
0
is the thickness of annular plate; G is the tank
wall weight per length of circumference, in this paper
all forces and bending moments are represented in
per length of circumference, and per length of cir-
cumference will be omitted in the sequel; R
1
is the
supporting force from foundation to the bottom plate
periphery; R
2
is the shear force of bottom plate at
the bottomfoundation junction; M
0
is the bending
moment at the shellbottomjunction; M
p
is the bend-
ingmoment of bottomplateat thebottomfoundation
junction;
b
is the rotation angle of bottomplate at the
shellbottom junction; P is the tank bottom hydraulic
pressure; K
0
is the bedding value of tank founda-
tion; is Poissons ratio; H is water-lling height;
D
0
= Et
3
0
/12(1
2
); F
0
=
4

K
0
/4D
0
; A
1
= 3L
2
+ F
0
L
3
+
3L/F
0
; A
2
= 1 + F
0
L; A
3
= C
4
L
4
2L
3
/F
0
+ 6L/F
3
0
;
A
4
= C
2
L
2
L/F
0
.
The simultaneous deection equations of the rst
and the second courses are adopted in calculation
by using trial-error. By assuming the uplift distance
of bottom plate L, the rotation angles
b
and
w
are
calculated by equations (13) and (12), respectively.
Comparing
b
with
w
, L should be re-specied and
thus calculation is repeated until
b
and
w
have
the same magnitude and opposite directions. There-
fore, the radial bending stress of upper surface is
presented by
When x

< C

br
=
6
t
2
0
[G + P(L C) R
2
]x

When C < x

< L

br
=
6
t
2
0
_
M
p
+
P
2
(L x

)
2
R
2
(L x

)
_
When x

> L

br
=
6
t
2
0
e
F
0
(x

L)
{M
p
{sin[F
0
(x

L)]+
cos[F
0
(x

L)]} + R
2
sin[F
0
(x

L)]}

(14)
The hoop bending stress of upper surface is pre-
sented by

b
=
br
(15)
Membrane stress of the bottomplate is rather small,
which can be ignored. In this case, bending stress is
specied as the surface stress of bottom plate. And
lower-surface stress has the opposite direction against
upper-surface stress.
4 EXAMPLE
Measured data is used to check the accuracy of
analytical method of combined cylindrical shell.
A new 15 10
4
m
3
oating-roof oil tank is built
somewhere in China, whose building construction
photo is shown in Fig. 4. The author did experimental
stress test on the foundation and shells during water
lling test after foundation.
4.1 Technical parameters of the oil tank
A tank is designed according to American Standard
API 650. Its diameter is 100 000 mm, total height
is 21 800 mm, water-lling height is 20 180 mm, the
extended length from tank wall centre to the outer
edge of annular plate is C = 120 mm, and other tech-
nical parameters are listed in Table 1.
About 400 piles that are 8 m long were built before
construction of tank foundation. It makes the founda-
tion quite reliable. Meanwhile, foundation settlement
was measuredduringwater-llingtest. The result indi-
cates that the differential settlement of tank is much
less than tolerance of design standard. Therefore, it
can be neglected in stress analysis.
4.2 Stress testing
In addition to certifying the strength of tank, water-
lling test also compacts the foundation. That is why
the process of water lling should not be too fast. The
process of water lling lasted 13 days and the level has
kept constant at its maximum for 3 days. The entire
test period is 16 days.
Stress test adopts traditional stressstrain test
method based on resistance and uses biaxial strain
gauges. Several strain gauges are installed on the out-
side of the two shell courses adjacent to the bottom
plate. The arrangement of the strain gauges on the
outside of a cylinder wall is shown in Fig. 5. The strain
gauges are waterproof and thus are prevented from
being damped during test.
In order to reduce the inuence from solar heat-
ing and temperature changes, initial balance and data
acquisition of strain gauges were all scheduled at
9 a.m. everyday. When the water level kept constant at
its maximum, three series of data were recorded every
24 h. The result shows that the three series of data sta-
bilized quite well and almost have nothing to do with
the weather and temperature.
After amendment the data of the strain gauges are
converted into stresses by using generalized Hookes
law. Table 2 lists the measured hoop stress and lon-
gitudinal stress on the outside of 15 10
4
m
3
tank
wall at the maximum water level 20 180 mm. In Fig. 2,
for x = 25 mm, there are two series sharing the same
JPME125 IMechE 2007 Proc. IMechE Vol. 221 Part E: J. Process Mechanical Engineering
124 Z P Chen, YY Duan, J MShen, and J L Jiang
Fig. 4 A new 15 10
4
m
3
oating-roof oil tank built somewhere in China
Table 1 Technical parameters of 15 10
4
m
3
oil tank
Circle number (from
bottom to top) l
i
(mm) t
i
(mm) Material
Tank shell 1 2980 40 SPV 490 Q
2 2680 33 SPV 490 Q
3 2680 26 SPV 490 Q
4 2680 22 SPV 490 Q
5 2680 17 SPV 490 Q
6 2680 12 SPV 490 Q
7 2660 12 16MnR
8 2660 12 Q235-B
Top wrapping rolled
steel
100 10 mm Q235-B
Bottom plate Annular plate Width of annular plate SPV490 Q
2000 mm, t
0
= 23 mm
Centre plate Thickness is 11 mm Q235-B
x-coordinates but different angles in circumference.
So do x = 3097 mm and x = 3102 mm. These data are
set in order to evaluate the inuence from foundation
to shell stress and the repeatability of test data.
4.3 Calculation of theoretical stress
With combined cylindrical shell method, the shell
stress of 15 10
4
m
3
oating roof tank is obtained. For
it, K
0
= 0.1 N/mm
3
, = 1 10
3
kg/m
3
, g = 9.8 m/s
2
,
= 0.3, E = 2.06 10
5
MPa, other parameters are
listed in Table 1. By trialerror, requirements
are met when the uplift distance is L = 730 mm,
and R
1
= 175.53 N/mm, R
2
= 15.165 N/mm, M
0
=
62 328 N, M
p
= 4780.3 N,
b
= 0.022 564 rad,
w
=
0.022 564 rad. Calculation results show that bending
moments and shear forces (bearing pressure) are neg-
ative, which means their directions are opposite to the
directions assumed in Fig. 3. Table 3 shows bending
moments and shear forces obtained by the combined
cylindrical shell method at the shellbottom junction
and course junctions.
Fig. 5 Arrangement of the strain gauges on the outside
of the tank wall
Proc. IMechE Vol. 221 Part E: J. Process Mechanical Engineering JPME125 IMechE 2007
A simplied method for calculating the stress of a large storage tank wall 125
Table 2

and
x
when reaching the maximum water
level 20 180 mm
x


x
x


x
(mm) (MPa) (MPa) (mm) (MPa) (MPa)
25 58.0 266.0 3097 237.5 81.8
25 60.6 270.1 3102 243.7 46.7
126 18.7 164.0 3390 264.6 46.4
277 5.2 129.1 4082 249.9 28.8
424 36.1 66.4 4467 192.6 1.3
801 88.8 12.4 4812 216.4 20.1
1023 132.4 19.7 5159 205.6 17.8
1328 164.4 72.8 5465 218.4 19.8
1630 201.0 78.6 5629 214.9 19.2
1920 178.3 54.5 5703 220.9 12.8
2216 208.7 51.7 5808 239.1 9.5
2530 237.7 55.4 5933 222.6 3.7
2732 221.5 42.8 6111 237.0 10.5
2950 251.3 33.0 6304 241.3 13.3
3006 246.5 58.7 6604 241.4 7.5
x is the distance from the measuring points to the upper surface
of bottom plate.
Table 3 t
i
, l
i
, M, and Q at course junctions
Q
i1
Level l
i
(mm) t
i
(mm) M
i1
(N) (N/mm)
1 2980 40 61 554
2 2680 33 4374 2.131
3 2680 26 637.9 8.038
4 2680 22 164.9 4.128
5 2680 17 287.2 4.453
6 8000 12 280.9 3.679
4.4 Comparison
Calculated results and measured values of hoop
stresses on the outside of a cylinder wall are shown
in Fig. 6. The dash-dot-line represents the curves
obtained by long shell method, and the broken curve
represents the membrane stress achieved by shell
theory.
It is noted that, theoretical curves of the proposed
method and long shell method agree very well on
the rst shell course below the rst hoop weld. How-
ever, there are so many differences between the two
Fig. 6 Hoop stresses on the outside surface
methods on the second shell course, especially for the
lower part of it. The results fromthe proposed method
are much closer to the measured values. Accord-
ing to localization of boundary stress, constraint
reaction from bottom plate to shell stress decreases
with the increasing distance. When x 2.5

Rt =
2.5

50 0.042 = 3.63 (m), effect from distance can


almost be omitted. Calculation curves from the two
methods all match well with the measured values.
The measured hoop stresses and the calculation
results of the proposed method and long shell method
at the six measuring points adjacent to the rst shell
second course junction are shown in Table 4. By com-
parison, the results obtained by the proposed method
are closer to the measured values. The maximumerror
between measured values and the results obtained
by the proposed method is
max
= (
c

m
)/
m
=
6.02 per cent (
c
is the calculationresult and
m
is the
Table 4 Hoop stresses of six measuring points on the outside of a cylinder
wall

/(MPa) (%)
Measured Proposed Long shell Proposed Long shell
N x (mm) value (
m
) method (
c
) method (
l
) method method
1 2950 251.3 249.9 249.8 0.56 0.60
2 3006 246.5 251.5 220.3 2.03 10.63
3 3097 237.5 251.8 222.7 6.02 6.23
4 3102 243.7 251.8 222.8 3.32 8.58
5 3390 264.6 250.9 228.7 5.18 13.57
6 4082 249.9 241.8 233.8 3.24 6.44
x is the distance from the measuring points to the upper surface of bottom plate.
JPME125 IMechE 2007 Proc. IMechE Vol. 221 Part E: J. Process Mechanical Engineering
126 Z P Chen, YY Duan, J MShen, and J L Jiang
Fig. 7 Longitudinal stresses on the outside surface
measuredvalue), whichcanmeet withtheengineering
demand.
Calculation results of longitudinal stresses by the
proposed method and long shell method on the out-
side of a cylinder wall are shown in Fig. 7 together with
measuredvalues. Thebrokencurveshows that thelon-
gitudinal membrane stress achieved by shell theory
is zero. In the mass, measured longitudinal stresses
uctuate around zero, which means that the results
from proposed method are closer to the measured
values.
The longitudinal andhoopstresses onthe inner side
of a cylinder wall can also be derived by the proposed
method.
5 CONCLUSIONS
1. For the two present methods for calculating the
stress of large storage tanks, short shell method
is more accurate, but it is not reliable in engi-
neering because of its complicate procedure. Long
shell method is simple. But maximumstress always
appears at the lower part of the second shell course
induced by fringe force of bottom plate, which is
not consideredby long shell method. As a result, the
calculation results deviate a lot from the measured
values.
2. The paper proposes a mechanical model of com-
bined cylindrical shell. The rst shell course is
regarded as a short cylindrical shell while all the
others as long cylindrical shells, and the analytic
solutionequations of shell stress canbeobtainedby
theory of plates and shells. The method takes into
account the constraint from bottom plate to the
second course, which makes the calculation of the
maximum hoop stress feasible and accurate. The
proposed method has higher accuracy and much
simpler procedure than short shell method.
3. With the development of economy, most highly-
industrialized countries, which are dependent on
importing petroleum, have to increase their raw
oil storage because there is shortage of petroleum
almost in every country. As a result, more and more
largeoil storagetanks havebeenbuilt inthesecoun-
tries. Meanwhile, toreduce the cost, oil storage tank
capacity becomes bigger and bigger. The analytical
calculation method can derive the stress distribu-
tion of large tanks accurately and quickly, and the
safety of storage tanks is guaranteed.
ACKNOWLEDGEMENTS
This work was supported by the National High Tech-
nology Research and Development Program of China
(863 Program) (No. 2007AA04Z427).
REFERENCES
1 American Petroleum Institute. API standard 650. Steel
tanks for oil storage, 10th edition, 1998.
2 Japanese Industrial Standard. Welded steel tanks for oil
storage. JIS B 8501, Tokyo 107, Japan, 1987.
3 Timoshenko, S. and Woinowsky-Krieger, S. Theory of
plates and shells, 1959, pp. 466487 (McGraw-Hill Book
Company Inc., NewYork).
4 Wu, T. Y. More accurate method devised for tank-bottom
annular plate design. J. Oil Gas, 1996, 94(21), 8183.
5 Wu, T. Y. and Liu, G. R. Comparison of design methods for
tank-bottom annular plate and concrete ring wall. Int. J.
Press. Vessels Pip., 2000, 77(3), 511517.
APPENDIX
Notation
C distance from shell plate centre to
outer edge of bottom annular plate
C
1i
, C
2i
,
C
3i
, C
4i
indeterminate constants
D
0
stiffness coefcient of the bottom
annular plate
D
1
stiffness coefcient of the bottom
shell course
D
i
stiffness coefcient of the ith shell
course, i = 2, 3, 4, . . . (from bottom to
top)
E modulus of elasticity of the steel
g gravity acceleration
Proc. IMechE Vol. 221 Part E: J. Process Mechanical Engineering JPME125 IMechE 2007
A simplied method for calculating the stress of a large storage tank wall 127
G tank shell weight per length of
circumference
H water-lling height
H
i
calculated water height at the ith to
(i + 1)-th course junction
K
0
bedding value of tank foundation
K
1
elastic coefcient of the bottom shell
course
K
i
elastic coefcient of the ith shell
course
l
i
height of the ith course of the
tank shell
L uplift distance up to which the
bottom plate lifts the foundation
M
0
bending moment at the shellbottom
junction
M
i
edge moment at the ith to (i + 1)-th
course junction
M
p
bending moment of bottom plate at
the bottomfoundation junction
P tank bottom hydraulic pressure
Q
i
edge shear force at the ithto(i + 1)-th
course junction
R shell radius
R
1
supporting force from foundation to
the bottom plate periphery
R
2
shear force of bottom plate at the
bottomfoundation junction
t
0
thickness of annular plate
t
i
thickness of the ith course of the
tank shell
x distance from the measuring point
to the upper surface of tank bottom
plate
x
A
distance from the ith course to the
welded joint
x
B
distance from the ith course to the
weld joint
y
i
deection of the ith shell course
y
A
deectionof the ithcourse of the tank
shell
y
B
deection of the (i + 1)-th course of
the tank shell
error between measured values and
theoretical calculation values

b
rotation angle of bottom plate at the
shellbottom junction

w
rotationangle of the lower node of the
rst course
Poissons ratio
water density

xi
longitudinal stress on the outside and
inner side of the ith course

i
hoop stress on the outside and inner
side of the ith course
JPME125 IMechE 2007 Proc. IMechE Vol. 221 Part E: J. Process Mechanical Engineering

Das könnte Ihnen auch gefallen