You are on page 1of 118

17 19

(B)

17360278

Tank Model and Set-up of


High-speed Micro Cameras

Floating Roof Model


(Single-deck Type)

Resonance at the
natural frequencies of
the fundamental and
2nd-order modes

Time Histories and Fourier Amplitude


Spectra of Displacement at the Wave Front
(Single-deck Type Roof of Model L)

Bi-harmonic
resonance at twice
the fundamental
natural frequency

Time Histories and Fourier Amplitude


Spectra of Displacement at the Roof Center
(Single-deck Type Roof of Model S)

20 5

3D Plot of Bi-harmonic
Resonance Oscillation

()

-2-

1.

1.1

1.2

1.3

9
14

2.

15

2.1

15

2.2

16

2.3

21

2.4

28

28

28

3.

30

3.1

30

3.2

30

3.3

36

3.4

42

42

43

4.

45

4.1

45

4.2

45

4.3

53

4.4

56

56

57

5. FEM

58

5.1

58

5.2 FEM

58
-3-

5.3

62

5.4

67

68

68

6.

70

6.1

70

6.2

70

6.3

72

6.4

73

6.5

75

6.6

75

6.7

80

80

7.

81

8.

93

9.

111

-4-

1964 1983 1999

2003

2003
7 2 1

312

km

2003 EW 76cm/s2NS 69cm/s2


58

58 119 100cm/s

2005
17 30
2

-1-

9
1

5
FEM

1983
2003

-2-

7
780mm
100,000kl 1/100

9 3

-3-

20 3

17
18
19

6,700,000
1,200,000
1,000,000
8,900,000

0
0
300000
300000

6,700,000
1,200,000
1,300,000
9,200,000

(1)
1)

594 pp.167-1732005.8

2)

607 pp.101-1082006.9

3)

12
pp.1470-14732006.11

4)

612 pp.87-942007.2

5)

615 pp.119-1262007.5

6)

FEM
619 pp.81-882007.9

7)

T. Matsui: Sloshing in a Cylindrical Liquid Storage Tank with a Floating Roof under Seismic
Excitation, Journal of Pressure Vessel Technology, Transactions of the ASME, Vol. 129, Issue 4,
pp.557-566, November 2007

8)

627 pp.741-7482008.5

9)

T. Matsui: Sloshing in a Cylindrical Liquid Storage Tank with a Single-deck Type Floating Roof
under Seismic Excitation, Journal of Pressure Vessel Technology, Transactions of the ASME, to be
published

(2)
1)

55
-4-


Vol.55pp.1-62006.1
2)

1 A-2pp.327-3282006.9

3)

2 A-2pp.329-3302006.9

4)

1
45 pp.189-1922007.2

5)

2
45 pp.193-1962007.2

6)

56 Vol.56
pp.347-3482007.3

7)

47 pp.142-1492007.3

8)

FEM
6 pp.87-962007.3

9)

T. Matsui: Sloshing and Dynamic Interaction between Liquid and Floating Roof in a Cylindrical
Liquid Storage Tank Subjected to Seismic Excitation, Proceedings of ECCOMAS Thematic
Conference on Computational Methods in Structural Dynamics and Earthquake Engineering,
Rethymno, Crete, Greece, June 2007

10) T. Matsui: Sloshing in a Cylindrical Liquid Storage Tank with a Single-deck Type Floating Roof
under Seismic Excitation, Proceedings of the ASME Pressure Vessels and Piping Division
Conference, PVP2007-26249, San Antonio, Texas, U.S.A, July 2007
11) 1
FEM A-2pp.321-3222007.9
12) 2

A-2pp.323-3242007.9
13)
3

A-2

pp.325-3262007.9
14)
A-2pp.327-3282007.9
15) T. Matsui: Prediction of Sloshing and Seismic Response of a Floating Roof in a Cylindrical Liquid
Storage Tank by the Response Spectrum Method, Proceedings of the Australian Earthquake
Engineering Society Conference, Wollongong, Australia, Paper No.36, November 2007
-5-

16) T. Nagaya, T. Matsui, T. Wakasa: Model Tests on Sloshing of a Floating Roof in a Cylindrical Liquid
Storage Tank under Seismic Excitation, Proceedings of the ASME Pressure Vessels and Piping
Division Conference, Chicago, Illinois, U.S.A, PVP2008-61675, July 2008, to be published
17) T. Matsui, Y. Uematsu, K. Kondo, T. Wakasa, T. Nagaya: Wind Effects on Sloshing of a Floating
Roof in a Cylindrical Liquid Storage Tank, Proceedings of the ASME Pressure Vessels and Piping
Division Conference, Chicago, Illinois, U.S.A, PVP2008-61688, July 2008, to be published
18)
1
A-22008.9
19)
2
A-22008.9
(3)
1)

2007.12 408pp.272-294

-6-

1.
1.1

[1]

CRT DRT
CFRT

40,000kl

1-1

1-1 [2]

1-1 1
2 1-2
2

1-1 [2]
-7-

(b)

(a)
1-2

1.2

2
(1) 0.11.0 (Bulging)

z
z

(2) 310 (Sloshing)


z
z
z
1-2

1964

1964


1978

1983

1985

1989

1971

1999

1999

2003

-8-

z ()

1-2

1.3

1964
1983
2003 [1~8]
1.3.1

1964

1964 39 6 16 1 02 40 km
M7.5 6

360
15

38 26
1-3 A BCD

5 10
30,000kl 51.5m 14.5m
34 4
1
45,000kl 62m 16.5m
2 30,000kl 2 1-4(a)

6 29 17

-9-

1-3 [4]

z
1,000kl 964kl

14 50cm

332
1,400 1-4(b)

(b)[3]

(a)[4]

1-4
- 10 -

1.3.2

1983

1983 ( 58 ) 5 26 11 59 80 km
M7.7
5 10

z
12 01 50 m 20
m 1-5(a)

2 30 2 29

1-3

1-5(b)

(a)[5]

(b)[7]

1-5
- 11 -

1-3 [7]

1.3.3

2.00 m
0.60 m
0.40 m
0.73 m
0.57 m
3.50 m
4.50 m

2003

2003 2003 ( 15 ) 9 26 4 50
80 km M8.0

1-6
z
2003 K-NET HKD129030926450
NS&EW 1-7 NS

(b)[2]

(a)[2]

1-6

(a)

(b)

1-7 K-NET
- 12 -

1-8 [8]
87cm/s2EW 73cm/s2 58 sec

[9] 100 cm/s


z
7
1-6(a)
1-8

z
2 1
2
1
4 51 42.7m 24.9m
7
12 9

2 2 9 28 10 45
42.7m 24.39m
2 9 30 6 55
- 13 -

1-6(a)

[10]

[1]

2005

[2]


http://www.idemitsu.co.jp/factory/hokkaido/safety/acc/index.html

[3]

. 1964

[4]

1965

[5]

, 58 (1983 )
, 1983

[6]

1982 1983 1984

[7]

1983 , 1986

[8]

2005

[9]

99 1974

[10] 30
2005

- 14 -

2.

2.1

[1, 2]

1954 Senda
and Nakagawa [3]

[4]

Nakagawa [5][6]

Sakai et al. [7]

[8][9]

[10, 11]

- 15 -

2.2
2.2.1

2-1
R H (r , , z ) z

= 0 xg (t ) = g (t )

2 = 0

= xg (t ) cos
r

=0
z

(1a)

r = R

(1b)

z = 0

+g
=0
2
t
z

(1c)

z = H

(1d)

t g
(1a)(1b)(1c)

i =1

= xg (t )r + Ai (t ) cosh i

z r
J1 i
cos
R R

J1 1 1 i J1 ( i ) = 0

2-1
- 16 -

(2)

(1d)(2)

g (t )r + Ai (t ) + i2 Ai (t ) cosh i
i =1

H
R

r
J1 i R = 0

(3)

i i

g
H
i tanh i
R
R

i =

(4)

(3) ( r R ) J1 ( i r R ) 0 r R = u 1
(A1) Ai
Ai (t ) + i2 Ai (t ) = g (t )

2
1
1 J1 ( i )
2
i

1
R
H
cosh i
R

(5)

(2)(5)

z r
cosh i J1 i

2
R R R cos
= xg (t )r + xi (t ) 2
H J ( )
1

i =1
i
cosh i 1 i

(6)

xi i
xi (t ) + i2 xi (t ) = x g (t )

(7)

(7)

xi (t ) + 2 i i xi (t ) + i2 xi (t ) = x g (t )

(8)

i i
(6)
1

g t z=H

r
J1 i

1
2
R R cos
= xg (t )r + xi (t ) 2
g
J
1

i =1
i
1 ( i )

(9)

2.2.2


2 = 0

(10a)

= xg (t ) cos
r

r = R

(10b)

- 17 -


=0
z

z = 0

= w(t )
z

(10c)

z = H

(10d)

w(t )
p


gw
p =
t z=H

(11)

mw + D 4 w = p

(12)

m D

w(t ) Z 0 cos Z n cos ( n = 1, 2, , )

w(t ) = n (t ) Z n cos

(13)

n=0

Z0 =

r
r
r
, Z n = n J1 n + n I1 n (n 1)
R
R

(14)

I1 1 1 n
n n

n4 =

mR 4 2
n
D

(15)

n n n

c11 c12 n 0
c
=
21 c22 n 0

(16)

c11 = n2 J1( n ) + n J1 ( n ) J1 ( n )
c12 = n2 I1( n ) + n I1 ( n ) I1 ( n )

- 18 -

c21 = n3 J1( kn ) + n2 J1( n ) n ( 3 ) J1 ( n ) + ( 3 ) J1 ( n )


c22 = n3 I1( kn ) + n2 I1( n ) n ( 3 ) I1 ( n ) + ( 3 ) I1 ( n )

(17)


(10a)(10b)(10c)(1a-c)
(2)

(2)(13)(10d)

A (t )
i =1

i2
H r
cosh i J1 i = n (t ) Z n
g
R R n=0

(18)

(18) ( r R ) J1 ( i r R ) 0 r R = u 1
(A1) Ai

Ai (t ) =

2
2
i 1

1
1
g
H J ( ) i2
cosh i 1 i
R

a
n=0

in n

(t )

(19)

ain =

1
i2
uZ n J1 ( i u )du

J1 ( i ) 0

(14)(A2)~(A4)

ai 0 = 1

ain = i2 n 2 n 2 J1 ( n ) + 2 n 2 I1 ( n ) (n 1)
i + n
i n

(20)

(19)(2)

z r
cosh i J1 i
2

g
R R a (t ) cos
= x g (t ) r + 2

in
n
H J ( )
i =1 i 1 i2

cosh i 1 i n = 0

(21)

(11)(13)(21)

r
J1 i

2 g
R a (t ) + g Z (t ) cos
p = xg (t )r + 2

in n
n n
2
J1 ( i ) n = 0

i =1 i 1 i
n=0

(12)(13)(22)

- 19 -

(22)

r
J1 i

2 g
R a (t ) g Z (t )
m n (t ) + n2 n (t ) Z n = xg (t )r + 2

in n
n n
2
J1 ( i ) n = 0

n=0
i =1 i 1 i
n=0

(23)

(23) ( r R ) Z n 0 r R = u 1

(
l =0

nl

n + nl )l (t ) + n2 +
n n (t ) = n xg (t )
m

(24)

nl
1

n = uZ n2 du nl =
0

2 g

1
R 1 2
1
u Z n du
ail uZ n J1 ( i u ) du n =
2
2
0
m 0
i =1
i 1 i J1 ( i )

(14)(20)[12] n nl n

1
4
2

1
n = n2 { J1 ( n )} J 0 ( n ) J 2 ( n ) + n n J1 ( n ) I 2 ( n ) + J 2 ( n ) I1 ( n )

2
n

0 =

nl =

0 =

2
1 2
n {I1 ( n )} I 0 ( n ) I 2 ( n ) (n 1)

m
i =1

R
4m

2
i

g
a a
2 in il
1 i

(26)

n = 0 (n 1)

(27)

(25)

2
i

(24) nl g m
m n

(24) 0

(24)
n (t ) n (t ) (13)

(21)(22) w p

r =

2 w
,
r 2

1 w 1 2 w
,
+
r r r 2 2

r =

1 2 w
1 w
2
r r r

(28)(13)(14)

- 20 -

(28)

n
r
r
n J1 n + n I1 n cos

R
R
R

r = n (t )
n =1

= n (t )
n =1

r
r
n J 2 n + n I 2 n cos
Rr
R
R

r = n (t )
n =1

n
r
r
n J 2 n n I 2 n sin
Rr
R
R

(29)

M r = D( r + ), M = D( + r ), M r = D(1 ) r

(30)

2.2.3

(24)Nakagawa [3][4]

r
J

2 g 1 i R
gr
cos
(
)
(
)
p = xg (t )r + 2
t
t

+
0
0
2
J1 ( i )
R

i =1 i 1 i

(31)

D 4 w = p

(32)

(32)(13) ( r R ) Z n 0 r R = u 1
n (n 1)

n =

R n 0
0
m n n 2

(33)

(33)(29)(30)

2.3
2.3.1

2-1 40,000kl

100,000kl

2-2

- 21 -

2-1
Small tank
25 m
15 m
3
850 kg/m
700 mm*
500 mm**
4.5 mm

Large tank

2.3.2

0.3

0.3

2.46410 kN-m* 3.22310 kN-m*


5
5
1.25110 kN-m** 1.80610 kN-m**
2
2
120 kg/m
120 kg/m
2
2
180 kg/m
180 kg/m

Small tank

Large tank
40 m
20 m
3
850 kg/m
800 mm*
600 mm**
4.5 mm

0.001
0.01
6
5

0.001
0.01
6
5

2-2

2
2
120 kg/m
180 kg/m
5
DRL-25
DRH-25
2.46410 kN-m
5
DFL-25
DFH-25
1.25110 kN-m
5
DRL-40
DRH-40
3.22310 kN-m
5
DFL-40
DFH-40
1.80610 kN-m

2-2 2-3
1 1
2

Sakai et al. [7]


Sakai et al.

2-2 (R=25m, H=15m)

2-3 (R=40m, H=20m)

- 22 -

3
2

2.3.3

2003 K-NET
HKD1290309260450EW
300s 0.01s =1/4
1 2-1

DRL-25 DRL-40 2-4 2-5


2-6 2-7
58s
DRL-25 1
8.25s2 4.35sDRL-40 1 10.97s2 5.52s
DRL-25 1 8.25s2 0.77sDRL-40 1
10.95s2 2.04s 1
DRL-40 2

2-4 DRL-25

2-5 DRL-40

- 23 -

DRL-25 1

DRL-40 2
1
DRL-25 1
DRL-40 2

2-6 DRL-25

2-7 DRL-40

- 24 -

1
2

2-8 =0
(Small tank: R=25m, H=15m)

2-9 =0
(Large tank: R=40m, H=20m)

- 25 -

= 0
2-8 2-9
DFL-40 DFH-40

Small tank Large tank


r R = 0.3 ~ 0.4 r
r R = 0.8 r R = 1
2-8 2-9
r R = 0.4
M r M M r M 1.5

2.3.4

M r t d
t 3 Z e
Z e

td 3 r = M r Z e = M Z e

Z e 2-3
2-3 *

DRL-25DRH-25
4.5 mm
700 mm
1050 mm2
DFL-25DFH-25
4.5 mm
500 mm
750 mm2
DRL-40DRH-40
4.5 mm
800 mm
1200 mm2
DFL-40DFH-40
4.5 mm
600 mm
900 mm2
* 1/3

2-10 =0
(Small tank: R=25m, H=15m)

2-11 =0
(Large tank: R=40m, H=20m)

- 26 -

r 2-10 2-11

2.3.5

DRL-25 DRL-40 2-12 2-15

2 DRL-25

0.77sDRL-40 2.04s

2-12
DRL-25

2-13
DRL-40

2-14
DRL-25

2-15
DRL-40

- 27 -

2.4

40,000kl 100,000kl

[1]

594 pp.167-1732005

[2]

T. Matsui: Sloshing in a Cylindrical Liquid Storage Tank with a Floating Roof under Seismic
Excitation, Journal of Pressure Vessel Technology, Transactions of the ASME, Vol. 129, Issue 4,
pp.557-566, November 2007

[3]

K. Senda and K. Nakagawa: On the Vibration of an Elevated Water Tank (I), Technical Reports of
Osaka University, Vol.4, No.117, 1954

[4]

1996

[5]

K. Nakagawa: On the Vibration of an Elevated Water Tank (II), Technical Reports of Osaka
University, Vol.5, No.170, 1955

[6]

3 1 pp.2-8

1965
[7]

F. Sakai, M. Nishimura and H. Ogawa: Sloshing Behavior of Floating-Roof Oil Storage Tanks,
Computers and Structures, Vol.19, No.1-2, pp.183-1921984

[8]

76 pp.59-661980

[9]

2003 JSSCNo.52pp.20-25

2004
[10]
Vol.56No. 5812005

[11] 2005
[12] 1960
[13] 1957

i J1 ( i ) = 0 ([13]p.168(8.10))

- 28 -

1
2
1
1 2 { J1 ( i )}
=
uJ

u
J

u
du
2 i
0 1 ( i ) 1 j

( )

i= j

(A1)

i j

[12]

1
1
1
u J ( u ) du = J ( ) = J ( ) J ( )
1

2
i

2
i

1
m I 0 ( m ) J1 ( i ) i I1 ( m ) J 0 ( i )
+ m2

uI ( u )J ( u ) du =
1

1
m J 0 ( m ) J1 ( i ) i J1 ( m ) J 0 ( i )
m2

2
i

uJ ( u )J ( u ) du =
0

J1 ( i ) = J 0 ( i )

J1 ( i )

= 0 , i.e., J 0 ( i ) =

J1 ( i )

1
u J ( u ) du = J ( )
1

2
i

2
i

1
m J 0 ( m ) J1 ( m ) J1 ( i ) = 2 m 2 J1 ( m ) J1 ( i )
i m
m2

(A3)

2
i

1
m I 0 ( m ) I1 ( m ) J1 ( i ) = 2 m 2 I1 ( m ) J1 ( i )
i + m
+ m2

(A4)

uJ ( u )J ( u ) du =
1

uI ( u )J ( u ) du =
1

(A2)

- 29 -

3.

[1-3]

3.1

Nakagawa[4][5]

[6-8]

[9, 10]

3.2
3.2.1

3-1
R H

Rd
m D
R p M
T EI GJ
(r , , z ) z
= 0 xg (t )

2 = 0

(1a)

= xg (t ) cos
r

(1b)

r = R

- 30 -

3-1

=0
z

=w
z

z = 0

(1c)

z = H r Rd

(1d)

= W + r Rp
z

z = H Rd r R

(1e)

w W 3-2

t
p


r Rd
gw
p =

t z = H

g W + r R p
p =

t z = H

(2a)
Rd r R

(2b)

w W
mw + D 4 w = p

r Rd

(3)

Deck

Pontoon

3-2

- 31 -

EI 2
1 2W GJ 2
W
+
+ 3

3
2
2
2
R p
R p R p
R p
R 3 w 1 2 w 1 w
1
3w 3 2 w
2

D d 3 +

=
(
)

r 2 r 3 2 r = R
Rp r
r r 2 r 2 r
Rp
MW +

(4)

prdr

Rd

T
+eD

R
1 2W
GJ 2
W EI

+
+
+D d

2
2
2
2

R p
Rp Rp
R p
Rp

2 w
1 w 1 2 w
+ 2
2 +
2
r
r r r r = Rd

Rd 3 w 1 2 w 1 w
1
3 w 3 2 w
2

3
=
(
)

3 +
2
2
2
2
r r r = R
Rp r
r r
r r
Rp
d

Rd

p r R p rdr

(5)

e = R p Rd

w W

( w )r = R

3.2.2

= W R p Rd

w
=

r r = Rd

(6)

w W

w(r , , t ) = l (t ) Z l (r ) cos

(7)

l =0

l =0

l =0

W ( , t ) = l l (t ) cos ( , t ) = l l (t ) cos

(8)

Z 0 = r R Z l = l J1 ( l r Rd ) + l I1 ( l r Rd )

(9)

(l 1)

l = ( dZ l dr )r = R
l = Z l + e ( dZ l dr )
r = Rd
d

(10)

J 1 1 1 I1 1 1 l
l l

l4 =

mRd 4 2
l
D

(11)

l l l (4)(5) 0
(7)~(10)

c11
c
21

c12 l 0
=
c22 l 0

(12)

- 32 -

c11 =

D 2
EI + GJ
l + 1 l J1 ( l ) (1 ) J1 ( l ) +
l J 2 ( l )
2

Rd
Rp3

l 2 MR p J1 ( l ) + e l J1 ( l )
Rd

c12 =

D
EI + GJ
l2 + 1 l I1 ( l ) (1 ) I1 ( l )
l I 2 ( l )
2

Rd
Rp3

l 2 MR p I1 ( l ) + e l I1 ( l )
Rd

c21 =

D
EI + GJ
l J 2 ( l )
(1 ) l J1 ( l ) + l 2 + 1 J1 ( l )
2

Rd
Rp3

l 2

c22 =

D
Rd 2

Rp

l
2
J1 ( l ) + eMJ1 ( l )
T +e M
Rd
Rd

(1 ) I ( ) + 2 + 1 I ( ) + EI + GJ I ( )
1
l 1
l
l
l
l 2
l

Rp3

l 2

3.2.3

Rp

l
2
I1 ( l ) + eMI1 ( l )
T +e M
Rd
Rd

(13)

(1a)(1b)(1c)

i =1

= xg (t )r + Ai (t ) cosh ( i z R )J1 ( i r R ) cos

(14)

i J1 ( i )

(7)(8)(14)(1d)(1e)

Ai (t )
i =1

Ai (t )
i =1

i2
cosh ( i H R ) J1 ( i r R ) = Z l (r )l (t )
g
l =0

r Rd

i2
cosh ( i H R )J1 ( i r R ) = l + r R p l l (t )
g
l =0

(15a)

Rd r R

(15b)

i i
i =

g
i tanh ( i H R )
R

(16)

(15) ( r R ) J1 ( i r R ) 0 r R = u 1
[11] Ai (t )
Ai (t ) =

g
2
1
1
ail l (t )

2
2
i i 1 cosh ( i H R ) J1 ( i ) l = 0

- 33 -

(17)

ai 0 =

i2 1
Z 0 J1 ( i u )udu = 1
J1 ( i ) 0

ail =

i2 R
J1 ( i ) 0

Z l J1 ( i u ) udu +

Rd R

{ + ( r R ) }J ( u)udu
l

i2 l
R
J J i J1 ( l ) J 0 ( i )} d
2
2 { l 0( l) 1( i)
J1 ( i ) i l
R
2

i2 + l2

+ l

{ I ( ) J ( ) I ( ) J ( )} RR + (
d

l 0

i 1

R p l Si

2

R
Rd
J
J 2 ( i )

2( i)

i
R

(18)

i = i Rd R
Si =

Rd R

(19)

u J1 ( i u ) du

(20)

Si
(17)(14)

i =1

= xg (t )r +

g
2 cosh ( i z R ) J1 ( i r R )
ail l (t ) cos

2
2
i i 1 cosh ( i H R ) J1 ( i ) l = 0

(21)

(2)(7)(8)(21)

2 J1 ( i r R )
g
p = xg (t )r + 2 2
ail l (t ) + g Z l l (t ) cos

J1 ( i ) l = 0
i =1 i i 1
l =0

r Rd

(22a)

2 J1 ( i r R )
g
(
)
+
l + r R p l l (t ) cos

p = xg (t )r + 2 2
a
t
g

il l
J1 ( i ) l = 0
i =1 i i 1
l =0

) }

Rd r R

(22b)

3.2.4

(3)(4)(5)

d
d mw + D 4 w p wrdr

MR pW +

EI 2
1 2W GJ 2
+

+
R p 2 2
R p 2 R p 2 2

R p

R
3 w 1 2 w 1 w
3 w 3 2 w
DRd 3 +

3
prdr Wd

2
(
)

2
2
2
2
R
d
r r r = R
r r
r r
r
d

TR p

GJ 2
W EI
1 2W

+
+
+ DRd
2
R p
R p R p
R p 2

- 34 -

2w
1 w 1 2 w
+ 2
2 +
2
r r r r = Rd
r

R
3 w 1 2 w 1 w
3 w 3 2 w
2
+eDRd 3 +

3
p r R p rdr d = 0

(
)

2
2
2
2
R
d
r r r = R
r r
r r
r
d
(23)

(3)(4)(5)
[12]

(23)(7)(8)(22) r
Rd
m Z n Z l rdr + MR p n l + TR p n l = nl M n

(24)

n (n = 0, 1,

, ) (23) n

l =0

nl

M n + nl l (t ) + nl M nn2 + K nl l (t ) = n xg (t )

(n = 0, 1,

, )

(25)

nl
M n = mRd 2 n + MR p n 2 + TR p n 2

0 =

1
Rd 2

Rd

n =

1
Rd 2

Rd

Z 0 2 rdr =

(26)

1
2
( Rd R )
4

2
1
Z n 2 rdr = n 2 { J1 ( n )} J 0 ( n ) J 2 ( n )

2
n n
1
J1 ( n ) I 2 ( n ) + J 2 ( n ) I1 ( n ) + n 2 { I1 ( n )} I 0 ( n ) I 2 ( n ) (n 1)

2
n

nl = R 2
i =1

g
2
ain ail
2
2
i i i2 1

(28)

) }{ + ( r R ) } rdr
m ) + ( 2 R + R TR

K nl = g Z n Z l rdr + n + r R p n
Rd
0
Rd

= g nl M n m + 2 MR p

(27)

m nl

(29)

0 = Z 0 r 2 dr = R3 4
0

n = Z n r 2 dr +

) }

Rd
R
n + r R p n r 2 dr
Rd
0

= Rd 3 n { n J 2 ( n ) + n I 2 ( n )} + 3 n R p n + 4 n

2 =

(n 1)

R 2 Rd 2
R 3 Rd 3
R 4 Rd 4
3 =
4 =
2
3
4

(30)
(31)

(25) nl K nl

(25)
n (t ) (n = 1, 2,
- 35 -

, )

3.2.5

n (t ) (7)(8)(21)(22)
w W p

r =

2 w
1 w 1 2 w
1 2w
1 w

=
+
r

r r r 2
r2
r r r 2 2

(32)

(32)(7)(9)
2

r
r
r = n (t ) n n J 1 n
+ n I1 n
cos
n =1
Rd
Rd
Rd

= n (t )
n =1

r
r
n J 2 n
+ n I2 n
cos
Rd r
Rd
Rd

n
r
r
n J 2 n
n I2 n
sin

r = n (t )
n =1

Rd r

Rd

(33)

Rd

mr = D( r + ), m = D( + r ), mr = D(1 ) r

(34)

(A7)(A8)
M =

EI
R p2

(t ) (
n =1

R p l cos

Mt =

GJ
R p2

(t ) (
n =1

R p l sin

(35)

3.3
3.3.1

3-1 3-2 40,000kl


100,000kl
D-25D-40

DRL-25, DRL-40 S-25S-40


D-25 D-40 1/n

D1/n-25D1/n-40 S-25S-40 1/n S1/n-25


S1/n-40

3-1 3-2

- 36 -

3-1
D-25
D-40

25
m
40 m

15
m
20 m

3
850 kg/m3
850 kg/m

700 mm
800 mm

1200 mm3
1050 mm3
2.464105 kN-m 3.223105 kN-m

120 kg/m2
120 kg/m2

0.3
0.3

0.001
0.001

0.010
0.010

6
6

5
5

3-2
S-25
S-40

25 m
40 m

15 m
20 m

850 kg/m3
850 kg/m3

4.5 mm
4.5 mm

3m
5m

700 mm
800 mm

4.5 mm
4.5 mm

12 mm
12 mm

5717 mm3
3930 mm3
1.719 kN-m
1.719 kN-m

6
2
6
kN-m
1.67110
kN-m2
0.80910

1.036106 kN-m2 2.259106 kN-m2


58 kg/m2
58 kg/m2

558 kg/m
819 kg/m

783 kg-m
2815 kg-m

0.3
0.3

0.001
0.001

0.005
0.005

30
30

29
29

3.3.2

3-3
3-4 1 1
2

Sakai et al. [6]

- 37 -

3-3 (R=25m, H=15m)

3.3.3

3-4 (R=40m, H=20m)

2003 K-NET
HKD1290309260450EW
300s 0.01s =1/4
1 3-1 3-2

S-25
S-40 3-5 3-6 3-7
3-8 3-1 3-2

S-25 1 8.25s2
4.35sS-40 1 10.97s2 5.52s

S-25 1 8.28s2 4.39sS-40 1 10.99s2 5.56s

1
S-40 2
2
1
S-25 1
S-40
2 1
2
= 0
3-9 3-10
D-25D1/2-25D1/5-25 D-40

D1/2-40D1/5-40

- 38 -

3-5 S-25

3-6 S-40

r R = 0.3 ~ 0.4 r R = 1 r R = 0.4

r r 1.5
D-40D1/2-40

S1/5-25 S1/5-40
10%

1983 2003
- 39 -

3-7 S-25

3-8 S-40
[8,

13][9]

- 40 -

3-9 =0
(Small tank: R=25m, H=15m)

3-10 =0
(Large tank: R=40m, H=20m)

- 41 -

3.4

40,000kl 100,000kl

1983
2003

[1]

607 pp.101-1082006.9

[2]

612 pp.87-942007.2
[3]

T. Matsui: Sloshing in a Cylindrical Liquid Storage Tank with a Single-deck Type Floating Roof
under Seismic Excitation, Journal of Pressure Vessel Technology, Transactions of the ASME, (to be
published)

[4]

K. Nakagawa: On the Vibration of an Elevated Water Tank (II), Technical Reports of Osaka
University, Vol.5, No.170, pp.317-336, 1955

[5]

3 1 pp.2-8

1965.1
[6]

F. Sakai, M. Nishimura and H. Ogawa: Sloshing Behavior of Floating-Roof Oil Storage Tanks,
Computers and Structures, Vol.19, No.1-2, pp.183-192, 1984

[7]

76 pp.59-661980.10

[8]

2003 JSSCNo.52pp.20-25

2004.4
[9]

2005.1

[10]
Vol.56No. 581pp.89-942005.2

[11] 1957
[12] Y.C.)1980
[13]
24 6 pp.12-161984.11

[14] 1953

- 42 -


R p d
3-11 Q M M t
Fz M r

M t
d

Mt +

M +

Q +

Q
d

Fz
Mr

M
d

M
Q

Mt

3-11

( MR )W Q

Fz R p = 0

(A1)

=
Q =

1 M M t

R p
Rp

(A2)

(A2)(A1)
MW

1 2 M
1 M t
+ 2
Fz = 0
2
2
R p
R p

(A3)

r =
T

1 M t M

Mr = 0
R p
Rp

(A4)

Vr M r p r = Rd
Fz M r
Fz =

Mr =

Rd
1
Vr +
Rp
Rp

Rd

(A5)

prdr

Rd
1
( M r + eVr ) +
Rp
Rp

Rd

p r R p rdr

- 43 -

(A6)

W B
M M t [14]
M =

Mt =

EI
Rp

1 2W
+

R p 2

(A7)

GJ
W

R p
R p

(A8)

M r Vr
w
2 w
1 w 1 2 w
+ 2
M r = D 2 +
2
r
r r r r = Rd
3 w 1 2 w 1 w
3 w 3 2 w
2

Vr = D 3 +

(
)

r 2 r 3 2 r = R
r r 2 r 2 r
r
d

(A9)

(A10)

(A5)(A10)(A3)(A4) (4)(5)

- 44 -

4.

[1]

4.1

100,000kl 80m

4~5m

4.2
4.2.1

4-1
R H

Rd m D
R p M
T EI GJ
Rc Tc

(r , , z ) z
= 0 xg (t )

2 = 0

= xg (t ) cos
r

=0
z

(1a)

r = R

(1b)

z = 0

(1c)

= r c cos z = H r Rc
z

(1d)

=w
z

(1e)

z = H Rc r Rd

- 45 -

4-1

= W + r Rp
z

z = H Rd r R

(1f)

w c W
2 t
p


p =
t z = H

g c r cos

gw

g[W + (r R p ) ]

r Rc
Rc r Rd

(2)

Rd r R

g
w W c

mw + D 4 w = p

Rc r Rd

(3)

EI 2
1 2W GJ 2
W
+
+ 3

3
2
2
2
R p
R p R p
R p
R 3 w 1 2 w 1 w
1
3 w 3 2 w
2

D d 3 +

3
=
(
)
2
2
2
2
r
R p r
r r
r r
r r = R
Rp
d
MW +

4-2

- 46 -

Rd

prdr

(4)

+eD

R
1 2W
GJ 2
W EI
+D d

+ 2 +
2
2
2

R p
Rp Rp
R p
Rp

2 w
1 w 1 2 w
+ 2
2 +
2
r r r r = Rd
r

3 w 1 2 w 1 w
1
3 w 3 2 w
2

3
=
(
)
3 +

2
2
2
2
r
r r
r r
r r = R
Rp
r
d

Rd
Rp

Rd

p r R p rdr

(5)

2
2 w
1 w 1 2 w
Tc c D 2 +
cos Rc d
+ 2
2
0
r r r r = Rc
r
3
2 w
1 2 w 1 w
3 w 3 2 w

2
cos Rc d
+ DRc 3 +

(
)

0
r r2 r 2 r
r 2 r 3 2 r = R
r
c

Rc

pr cos rdrd

(6)

e = R p Rd
w W c

(w

r )r = R = ( w) r = Rc Rc = c cos

( w )r = R

4.2.2

(7)

= W R p Rd

(w

r )r = R =

(8)

w W c

w(r , , t ) = Z l (r )l (t ) cos

(9)

l =0

l =0

l =0

W ( , t ) = l l (t ) cos ( , t ) = l l (t ) cos

(10)

c (t ) = l l (t )

(11)

l =0

Z0 =

r
r
r
r
Z l = l J1 l + l Y1 l + l I1 l + l K1 l
Rd
Rd
Rd
Rd

(l 1)

(12)

dZ

dZ
l = Z l + e l
l = l
dr r = Rd

dr r = Rd

(13)

Z
dZ
l = l
= l
dr r = Rc Rc r = Rc

(14)

J 1 Y1 1 1 2 I1 K1 1 1
2 l
l l

- 47 -

l4 =

mRd 4 2
l
D

(15)

l l l l l (4)(6) 0
(7)(9)~(14)

c14 l 0
c24 l 0
=
c34 l 0

c44 l 0

c11

c21
c31

c41

c12
c22
c32
c42

c11 =

D 2
EI + GJ
l + 1 l J1 ( l ) (1 ) J1 ( l ) +
l J 2 ( l )

Rd 2
Rp3

c13
c23
c33
c43

(16)

l 2 MR p J1 ( l ) + e l J1 ( l )
Rd

D 2
EI + GJ
c12 = 2 l + 1 l Y1 ( l ) (1 ) Y1 ( l ) +
l Y2 ( l )

Rd
Rp3

l 2 MR p Y1 ( l ) + e l Y1 ( l )
Rd

D
EI + GJ
c13 = 2 l2 + 1 l I1 ( l ) (1 ) I1 ( l )
l I 2 ( l )

Rd
Rp3

l 2 MR p I1 ( l ) + e l I1 ( l )
Rd

D
EI + GJ
c14 = 2 l2 + 1 l K1 ( l ) (1 ) K1 ( l ) +
l K 2 ( l )

Rd
Rp3

l 2 MR p K1 ( l ) + e l K1 ( l )
Rd

D
EI + GJ
l J 2 ( l )
c21 = 2 (1 ) l J1 ( l ) + l 2 + 1 J1 ( l )

Rd
Rp3

l 2
c22 =

Rp

l
2
J1 ( l ) + eMJ1 ( l )
T +e M
Rd
Rd

D
EI + GJ
l Y2 ( l )
(1 ) l Y1 ( l ) + l 2 + 1 Y1 ( l )
2

Rd
Rp3

l 2

c23 =

Rp

l
2
Y1 ( l ) + eMY1 ( l )
T +e M
Rd
Rd

D
EI + GJ
(1 ) l I1 ( l ) + l 2 + 1 I1 ( l ) +
l I 2 ( l )
2

Rd
Rp3

l 2

Rp

l
2
I1 ( l ) + eMI1 ( l )
T +e M
Rd
Rd

- 48 -

c24 =

D
EI + GJ
(1 ) l K1 ( l ) + l 2 + 1 K1 ( l )
l K 2 ( l )
2

Rd
Rp3

Rp

l
2
K1 ( l ) + eMK1 ( l )
T +e M
Rd
Rd

2
D l
T
c31 =
J l ) l J1 ( l ) l 2 c 2 J1 ( l )
2 1(

Rc
Rd

l 2

c32 =

D l 2
Rd 2

Y ( ) Y ( ) 2 Tc Y ( )
l
l 1
l
l
l
1

Rc 2 1

c33 =

D l 2
Rd 2

2 Tc

I1 ( l ) l I1 ( l ) l R 2 I1 ( l )
c

c34 =

D l 2
Rd 2

K ( ) K ( ) 2 Tc K ( )
l
l 1
l
l
l
1

Rc 2 1

c41 = J1 ( l ) l J1 ( l )
c42 = Y1 ( l ) l Y1 ( l )
c43 = I1 ( l ) l I1 ( l )

4.2.3

c44 = K1 ( l ) l K1 ( l )

(17)

= Rc R d

(18)

(1a)(1b)(1c)

i =1

= xg (t )r + Ai (t ) cosh ( i z R )J1 ( i r R ) cos

(19)

i J1 ( i )
(9)(11)(19)(1d)(1f)

r Rc
r l l (t )
l =0

2
Ai (t ) i cosh ( i H R ) J1 ( i r R ) = Z l (r )l (t )
Rc r Rd

g
i =1
l =0

l + r R p l l (t ) Rd r R
l =0

(20)

i i
i =

g
i tanh ( i H R )
R

(21)

(20) ( r R ) J1 ( i r R ) 0 r R = u 1
[2] Ai (t )

- 49 -


g
2
1
1
ail l (t )

i2 i2 1 cosh ( i H R ) J1 ( i ) l = 0

Ai (t ) =

(22)

ai 0 =

i2 1
Z 0 J1 ( i u )udu = 1
J1 ( i ) 0

ail =

R
i2 R R
r l J1 ( i u )udu +

R
0

J1 ( i )

R
R

Z l J1 ( i u ) udu +

Rd R

{ + ( r R ) }J ( u)udu
l

i2 l
{ l J 0 ( l ) J1 ( i ) i J1 ( l ) J 0 ( i )}

J1 ( i ) i2 l2
l
{ J ( l ) J1 ( i ) i J1 ( l ) J 0 ( i )}
i l2 l 0

+ 2 l 2 { l Y0 ( l ) J1 ( i ) iY1 ( l ) J 0 ( i )}
i l

2 l 2 { l Y0 ( l ) J1 ( i ) iY1 ( l ) J 0 ( i )}
i l

+ 2 l 2 { l I 0 ( l ) J1 ( i ) i I1 ( l ) J 0 ( i )}
i + l

2 l 2 { l I 0 ( l ) J1 ( i ) i I1 ( l ) J 0 ( i )}
i + l

2 l 2 { l K 0 ( l ) J1 ( i ) + i K1 ( l ) J 0 ( i )}
i + l

l
K l ) J1 ( i ) + i K1 ( l ) J 0 ( i )}
2 { l 0(
i + l

( i )2
i2
R
R
R
( l R p l ) Si + l J 2 ( i ) d J 2 ( i ) +
l J 2 ( i )
J1 ( i )
R
J
i

1( i)
i

i = i Rd R
Si =

Rd R

(23)

(24)

u J1 ( i u ) du

(25)

Si
(22)(19)

i =1

= xg (t )r +

g
2 cosh ( i z R ) J1 ( i r R )
ail l (t ) cos

2
2
i i 1 cosh ( i H R ) J1 ( i ) l = 0

(2)(9)(11)(26) p

2 J1 ( i r R )
g
p = xg (t )r + 2 2
ail l (t ) cos

J1 ( i ) l = 0
i =1 i i 1

- 50 -

(26)

g
r l l (t ) cos
r Rc

l =0

Rc r Rd
g Z l l (t ) cos
l =0

g l + r R p l l (t ) cos Rd r R
l =0

(27)

4.2.4

(3)(4)(5)(6)

d
d mw + D 4 w p wrdr

Rc

MR pW +

EI 2
1 2W GJ 2
+
+
2
2
R p
R p 2 R p 2 2

R p

R
3 w 1 2 w 1 w
3 w 3 2 w
DRd 3 +
2
( 2 )
3
prdr Wd
2
2
2
Rd
r r r = R
r r
r r
r
d

TR p

GJ 2
W EI
1 2W

+
+

+ DRd
R p 2
R p R p
R p 2

2w
1 w 1 2 w
+
+ 2

2
r r r r = Rd
r

R
3 w 1 2 w 1 w
3 w 3 2 w
2
+eDRd 3 +

3
p r R p rdr d

(
)
2
2
2
2
Rd
r r
r r
r r r = R
r
d

2
2 w
1 w 1 2 w
+ Tc c D 2 +
+ 2
cos Rc d
2
0
r r r r = Rc
r
3
2 w
1 2 w 1 w
3 w 3 2 w
2
cos Rc d
+ DRc 3 +

(
)

3
0
r 2
r r 2 r 2 r
r 2 r = R
r
c

Rc

pr cos rdrd c = 0

(28)

(3)(4)(5)(6)
[3]
(28)(9)(14)(27) r
Rd
m Z n Z l rdr + MR p n l + TR p n l + (Tc ) n l = nl M n

(29)

Rc

n (28) n (n = 0, 1,

, )

l =0

nl

M n + nl l (t ) + nl M nn2 + K nl l (t ) = n xg (t )

(n = 0, 1,

, )

(30)

nl
M n = mRd 2 n + MR p n 2 + TR p n 2 + (Tc ) n 2

- 51 -

(31)

0 =

1
Rd 2

Rd

n =

1
Rd 2

Rd

Rc

Rc

Z 0 2 rdr =

1
1 4
4

)( R

R)

Z n 2 rdr

2
2
1
1
= n2 { J1 ( n )} J 0 ( n ) J 2 ( n ) n2 2 { J1 ( n )} J 0 ( n ) J 2 ( n )

2
1
+ n n 2 J1 ( n ) Y1 ( n ) J 0 ( n ) Y2 ( n ) J 2 ( n ) Y0 ( n )
2
1
n n 2 2 J1 ( n ) Y1 ( n ) J 0 ( n ) Y2 ( n ) J 2 ( n ) Y0 ( n )
2

n n
2
J1 ( n ) I 2 ( n ) + J 2 ( n ) I1 ( n ) n n J1 ( n ) I 2 ( n ) + J 2 ( n ) I1 ( n )
n
n

n n
2
J1 ( n ) K 2 ( n ) + J 2 ( n ) K1 ( n ) + n n J1 ( n ) K 2 ( n ) + J 2 ( n ) K1 ( n )
n
n

2
2
1 2
1
n {Y1 ( n )} Y0 ( n ) Y2 ( n ) n2 2 {Y1 ( n )} Y0 ( n ) Y2 ( n )

2
2

n n
2
Y1 ( n ) I 2 ( n ) + Y2 ( n ) I1 ( n ) n n Y1 ( n ) I 2 ( n ) + Y2 ( n ) I1 ( n )
n
n

n n
2
Y1 ( n ) K 2 ( n ) Y2 ( n ) K1 ( n ) + n n Y1 ( n ) K 2 ( n ) Y2 ( n ) K1 ( n )
n
n

2
2
1
1
+ n2 { I1 ( n )} I 0 ( n ) I 2 ( n ) n2 2 { I1 ( n )} I 0 ( n ) I 2 ( n )
2

n n

2 I1 ( n ) K1 ( n ) + I 0 ( n ) K 2 ( n ) + I 2 ( n ) K 0 ( n )
2

n n 2
2 I ( n ) K1 ( n ) + I 0 ( n ) K 2 ( n ) + I 2 ( n ) K 0 ( n )
n 1

2
2
1
1
+ n2 { K1 ( n )} K 0 ( n ) K 2 ( n ) n2 2 { K1 ( n )} K 0 ( n ) K 2 ( n )
2

(n 1)
(32)

nl = R 2
i =1

g
2
ain ail
2
2
i i i2 1

(33)

) }{

) }

Rc
Rd
R
K nl = g r 2 n l rdr + Z n Z l rdr + n + r R p n l + r R p l rdr
Rc
Rd

0
= g nl M n m + 2 MR p m n l + 4 2 R p 3 + R p 2 2 TR p m n l

+ Rc 4 4 Tc (m ) n l

(34)

0 = Z 0 r 2 dr = R3 4
0

) }

Rc
Rd
R
n = r n r 2 dr + Z n r 2 dr + n + r R p n r 2 dr
Rc
Rd
0

= n J 2 ( n ) 2 J 2 ( n ) Rd 3 n + n Y2 ( n ) 2Y2 ( n ) Rd 3 n

}(
) {
}(
)
+ { I ( ) I ( )}( R ) + { K ( ) K ( )}( R )
+ ( R ) + + ( R 4 ) (n 1)

- 52 -

(35)

R 2 Rd 2
R 3 Rd 3
R 4 Rd 4
3 =
4 =
2
3
4

2 =

(36)

(30) nl K nl
(30) nl K nl

4.2.5

(30) n (t ) (9)(11)(26)(27)
w W c p

r =

2w
1 w 1 2 w
1 2w
1 w

=
+

2
=
r

2
2
2
r r r
r
r r r

(37)

(37)(9)

d 2Zl
1 dZ l 1
2 Z l cos
cos = l (t )
2
dr
r dr r

l =1
n =1

dZ
1
1

l
= l (t )
+ 2 Z l sin
r
dr
r

l =1

r = n (t )

(38)

mr = D( r + ), m = D( + r ),

mr = D(1 ) r

(39)

M =

EI
R p2

(t ) (
n =1

R p l cos M t =

GJ
R p2

(t ) (
n =1

R p l sin

(40)

4.3
4.3.1

1 100,000kl

S-40CP
S-40 S-40

- 53 -

4-1
S-40
S-40CP

40 m
40 m

20 m
20 m

850 kg/m
850 kg/m3
4.5 mm
4.5 mm

5m
5m

800 mm
800 mm

4.5 mm
4.5 mm

12 mm
12 mm

5717 mm3
5717 mm3
5m

800 mm

4.5 mm

12 mm

1.719 kN-m
1.719 kN-m

1.671106 kN-m2 1.671106 kN-m2

2.259106 kN-m2 2.259106 kN-m2


2
2

58 kg/m
58 kg/m

819 kg/m
819 kg/m

2815 kg-m2/m
2815 kg-m2/m

104,090 kg-m2

0.3
0.3

0.001
0.001

0.005
0.005

30
30

29
29
4.3.2

1
2
3
4
5
6

4.3.3

4-2
S-40 S-40CP

10.97 s
11.00 s
11.00 s
5.52
5.56
5.56
4.34
4.37
4.41
3.71
3.72
3.80
3.29
3.28
3.42
2.99
2.96
3.11

2003 K-NET
HKD1290309260450EW

- 54 -

4-3 (S-40CP )

4-5 =0
S-40 CP S-40

4-4
(S-40CP )
- 55 -

0.01s 300s =1/4


1 1

S-40CP 4-3 4-4

1 2
2

S-40CP S-40
= 0 4-5

4.4

100,000kl

[1]

615 pp.119-1262007.5

[2]

1957

[3]

Y.C.1980

- 56 -


4-6
Tc c +

( M r Vr Rc ) cos Rc d 0 0

Rc

pr cos rdrd = 0

(A1)

M r Vr
w
2 w
1 w 1 2 w
M r = D 2 +
+ 2
2
r
r r r r = Rd

(A2)

3 w 1 2 w 1 w
3 w 3 2 w

Vr = D 3 +
2

(
)

r r 2 r 2 r
r 2 r 3 2 r = R
r
d

(A3)

(A2)(A3)(A1)(6)

4-6

- 57 -

5.

FEM
[1]

5.1

2003

FEM

FEM
Sakai et al. [2] FEM

5.2
5.2.1

FEM

5-1
R H
(r , , z ) z
= 0
xg (t )

5-1
- 58 -

2 = 0

(1a)

= xg (t ) cos
r

r = R

(1b)

=0
z

z = 0

(1c)

=w
z

z = H

(1d)

w t
w Z n (r ) cos

w(r , , t ) = n (t ) Z n (r ) cos

(2)

n =0

Z 0 Z n (n 1) n (t ) N

m n (t ) + n 2 n (t ) Z n cos = p

(3)

n =0

m n 0 = 0

p

p =
gw
t z = H

(4)

g
5.2.2

FEM
FEM
M r , M M r w
Dr , D [3]
2 w
1 w 1 2 w
1 w 1 2 w
2 w
M = D
M r = Dr 2 + D1
+ 2
+ 2
+ D1 2
2
2
r
r r r
r r r
r
2
1 w 1 w
M r = Dr
2

r r r

- 59 -

(5)

(a)

(b)
5-2
Dr = D =

Eh3
= D, D1 = D, Dr = (1 ) D
12(1 2 )

(6)

E h
5-2
Dr = EI r Br ,

D = E I B ,

D1

(7)

Br , B r I r , I 1
5-2 2
Dr Dr , D

4.5mm

2a
5-2b[4]

5.2.3

(1a)(1b)(1c)

= xg (t )r cos + Ai (t ) cosh ( i z R )J1 ( i r R ) cos

(8)

i =1

J1 1 1 i J1 ( i ) I

- 60 -


(2)(8)(1d)
I

Ai (t )
i =1

N
i2
cosh ( i H R )J1 ( i r R ) = Z n n (t )
g
n=0

(9)

i i
g
i tanh ( i H R )
R

i =

(10)

(9) ( r R ) J1 ( i r R ) 0 r R = u 1
[5] Ai (t )
Ai (t ) =

2
1
1
g
2
i 1 cosh ( i H R ) J1 ( i ) i2

a
n=0

in n

(t )

(i = 1, 2,

, I)

(11)

ain =

i2 1
Z n J1 ( i u )udu
J1 ( i ) 0

(12)

Z n FEM FEM
Z n
(12)

(11)(8)
I

= xg (t )r cos +
i =1

g 2 cosh ( i z R ) J1 ( i r R ) N
ainn (t ) cos
i2 i2 1 cosh ( i H R ) J1 ( i ) n = 0

(13)

(4)(2)(13) p
I

2 J1 ( i r R ) N
g
p = xg (t )r + 2 2
ain n (t ) + g Z n n (t ) cos

J1 ( i ) n =0

i =1 i i 1
n =0

(14)

(14)(3)
N
I
N

2 J1 ( i r R ) g N
m Z n n (t ) + n2 n (t ) = xg (t )r + 2
a (t ) + g Z n n (t )
2 in n
J1 ( i ) i n = 0
n =0
i =1 i 1
n=0

(15)

(15) rZ n 0 r R
n (t )

(
N

l =0

nl

M n + nl l (t ) + nl M nn2 + K nl l (t ) = n xg (t )

- 61 -

(n = 0, 1,

, N)

(16)

nl
R
M n = mZ n 2 rdr

(17)

nl = R 2
i =1

g
2
ain ail
2
2
i i i2 1

(18)

K nl = g Z n Z l rdr

(19)

n = Z n r 2 dr

(20)

(12)
(16) M n nl K nl
n (16)
nl K nl

(16) n (t ) (2)(13)(14)
w p
M r , M M r (2)(5)

d 2 Zn
1 dZ n Z n
+ D1
2
M r = n (t ) Dr
2
dr
r
r dr
n =1

cos

1 dZ n Z n
d 2 Zn
2 + D1
M = n (t ) D
cos
r
dr 2
n =1
r dr

1 dZ n Z n
2 sin
M r = n (t ) Dr
r
r dr
n =1

(21)

FEM

1. FEM
2. n Z n
3. M n , nl , K nl , n (17)~(20)(12)

4. (16) n (t )
5. (2)(14)(21) n (t ) n (t ) w p
M r , M M r

5.3
5.3.1

FEM
5-1 5-2 5-3

- 62 -

5-1 U-40
40 m

20 m

850 kg/m3
80 cm

3.223105 kN-m

120 kg/m2
206 Gpa

0.3

235 MPa
5-2 S-40CP
40 m

20 m

850 kg/m3
4.5 mm

5m

80 cm

4.5 mm

12 mm

5717 mm3
5m

80 cm

4.5 mm

12 mm

1.719 kN-m
1.671106 kN-m2
2.259106 kN-m2
58 kg/m2

2815 kg-m2/m

104090 kg-m2

5-3

5717 mm3

819 kg/m

206 Gpa
0.3
235 Mpa

5-4 mm

100,000kl
U-40 S-40CP
S-40CP 5-4 T
T-40
D-40

- 63 -

5-3


U-40
8, 12
6
5
#1
#2
40 , 46
40
39
S-40CP
T-40
8
6
5
D-40
8
6
5
#1 40
#2 40
4 46

FEM
cos r 3
3 5-3 S-40CP

U-40 T-40 D-40


2003 K-NET
HKD1290309260450EW

0.01s 163.84s =1/4


1 0.5%S-40CP 1.0%U-40
T-40 D-40
5.3.2

FEM U-40
S-40CP 5-5
w p M r , M
= 0
5-6 5-7 U-40 S-40CP

5-5 FEM

- 64 -

5-6 = 0 U-40

5-7 = 0 S-40CP

- 65 -

5-2aS-40CP
5-2b
S-40CP

100

5.3.3

S-40CP T
T-40 D-40
5-8 w p
= 0 9

S-40CP S-40CP T-40


D-40
2b
5-8 S-40CP
T-40 D-40 2 S-40CP
9
S-40CP T-40 D-40
1
1983 2003

5-8

- 66 -

5-9 = 0

D-40 T-40
D-40 T-40
D-40 T-40

5.4

FEM

2003

- 67 -


[1]

615 pp.119-1262007.5

[2]

F. Sakai, M. Nishimura and H. Ogawa: Sloshing Behavior of Floating-Roof Oil Storage Tanks,
Computers and Structures, Vol.19, No.1-2, pp.183-192, 1984

[3]

S.P. Timoshenko and P. Woinowsky-Krieger: Theory of Plates and Shells, 2nd ed., McGraw-Hill,
1959

[4]

2005.9

[5]

1957

5-10 j
j j + 1 rj rj +1

5-10

FEM j Z jn cos r jn cos


j Z n (r ) [ L ]
Z n (r ( )) = [ L( ) ]{ j }

(A1)

{ } = Z
j

nj

nj

Z n , j +1

[ L( )] = 1 3 2 + 2 3
r ( ) = rj (1 ) + rj +1

n, j +1

(A2)

( 2 2 + 3 ) 3 2 2 3
(0 1), = rj +1 rj

( 2 3 )

(A3)
(A4)

(A1)(2)(21) j

- 68 -

w = w cos , w = n (t ) [ L ]{ j }

(A5)

M r M r cos


M = M cos ,
M M sin
r r

(A6)

n=0

Mr

N
M = n (t ) [ D ][ B ]{ j }
M n=0
r

[ B] =

( 6 12 ) 2

2
( 6 6 ) (r )

2
3
2
+ (1 3 + 2 ) r

2
2 ( 6 + 6 ) (r )

2 (1 3 2 + 2 3 ) r 2

Dr
[ D ] = D1
0

D1
D
0

( 4 6 )

( 6 + 12 )

( 1 + 4 3 ) r ( 6 + 6 ) (r)
+ ( 2 + ) r
+ ( 3 2 ) r
2 (1 4 + 3 ) r
2 ( 6 6 ) (r )
2 ( 2 + ) r 2 ( 3 2 ) r
2

0
0
Dr

( 2 6 )

( 2 3 ) r

+ ( 2 + 3 ) r 2

2 ( 2 + 3 2 ) r

2 ( 2 + 3 ) r 2
2

(A7)

(A8)

- 69 -

6.
6.1

[1]

FEM

6.2

6-1
R H
(r , , z ) z

6-1

- 70 -

= 0
xg (t )

2 = 0

(1a)

= xg (t ) cos
r

r = R

(1b)

=0
z

z = 0

(1c)

=w
z

z = H

(1d)

w t
w Z n (r ) cos

w(r , , t ) = n (t ) Z n (r ) cos

(2)

n =0

Z 0 Z n (n 1) n (t ) N

m n (t ) + n 2 n (t ) Z n cos = p

(3)

n =0

m n 0 = 0


p =
gw
t z = H

(4)

g
M r , M M r

w [2]
2 w
1 w 1 2 w

M r = Dr 2 + D1
+ 2
2
r r r
r
1 2 w 1 w
M r = Dr
2

r r r

1 w 1 2 w
2 w
M = D
D
+ 2
+

1
2
r 2
r r r

(5)

Dr , D , D1 Dr (5)(2)

- 71 -


d 2 Zn
1 dZ n Z n
M r = n (t ) Dr
+ D1
2 cos
2
dr
r
n =1
r dr

1 dZ n Z n
d 2 Zn
M = n (t ) D
2 + D1
cos
r
dr 2
n =1
r dr

1 dZ n Z n
M r = n (t ) Dr
2 sin
r
r dr
n =1

6.3

(6)

(1a)(1b) (1c)(1d)

I

= xg (t )r cos +
i =1

g 2 cosh ( i z R ) J1 ( i r R ) N
ainn (t ) cos
i2 i2 1 cosh ( i H R ) J1 ( i ) n = 0

(7)

ain =

i2 1
Z n J1 ( i u )udu
J1 ( i ) 0

(8)

J1 1 1 i J1 ( i ) I i
i
i =

g
i tanh ( i H R )
R

(9)

(2)(7)(4) (3)
I
N
N

2 J1 ( i r R ) g N
+
m Z n n (t ) + n2 n (t ) = xg (t )r + 2
a
t
g
Z n n (t ) (10)
(
)

in n
2
J1 ( i ) i n =0
n =0

i =1 i 1
n =0

(10) rZ n 0 r R
n (t )
N

(
l =0

M n + nl ) l (t ) + nl M nn2 + K nl l (t ) = n xg (t )

nl

(n = 0, 1,

, N)

(11)

nl
R

M n = mZ n 2 rdr

(12)

nl = R 2
i =1

g
2
ain ail
2
2
i i i2 1

(13)

K nl = g Z n Z l rdr

(14)

n = Z n r 2 dr

(15)

- 72 -

(8)(12)(14)(15)

(11) M n nl K nl
n (11)
nl K nl

(11) n (t ) (2)(6)

w M r M M r

6.4

(11)

(11) n (t )

n (t )

(11) 0 j j

jn jn
N

(
l =0

nl

M n + nl ) j 2 + nl M nn2 + K nl jl = 0

(n = 0, 1,

, N)

(16)

(
jn

n=0 l =0

nl

(
jn

n=0 l =0

nl

0
M n + nl ) kl =

M j + j

jk
j=k

jk
0
M nn 2 + K nl kl = 2

j ( M j + j ) j = k

(17)

(18)

M j = jn M n jn

(19)

n=0

j = jn nl jl

(20)

n =0 l =0

[3]
- 73 -


n (t ) jn
J

n (t ) = jn j (t )

(21)

j =1

j (t ) J ( N )
(21)(11) jn n (17)
(18)(11) J

j (t ) + j 2 j (t ) = j xg (t )

( j = 1, 2,

, J)

(22)

j =

1
M j + j

n =0

(23)

jn n

(22)

j (t ) + 2h j j j (t ) + j 2 j (t ) = j xg (t )

( j = 1, 2,

, J)

(24)

h j j

(24) j (t ) (21)(2)(6)
w M r M M r

w(r , , t ) = j (t ) Z j (r ) cos

(25)

d 2 Z
J
1 dZ j Z j
j
M r = j (t ) Dr
+ D1
2 cos
2
r dr
dr
r

j =1

1 dZ
J
Z j
d 2 Z j
j
cos
M = j (t ) D
2 + D1
r
dr 2
r dr
j =1

J
1 dZ j Z j
M r = j (t ) Dr
2 sin
r dr
r
j =1

(26)

j =1

Z j (r ) = jn Z n (r )

(27)

n =0

Z j (r ) cos

- 74 -

6.5

(24)

w M r M M r j

( w) j = j Z j cos S D (Tj , h j )

(28)

d 2 Z
1 dZ j Z j
j

( M r ) j = j Dr
+
2 cos S D (Tj , h j )
D
1
r dr
dr 2
r

1 dZ
Z j
d 2 Z j
j
cos S D (Tj , h j )
2 + D1
( M ) j = j D
r
dr 2
r dr

1 dZ j Z j
( M r ) j = j Dr
2 sin S D (Tj , h j )
r dr
r

(29)

S D (T , h) T h
T = 2 j
j

2 SRSS
2 CQC [4] w
SRSS
w |max =

(w)
j =1

(30)

CQC
w |max =

c
j =1 k =1

jk

(31)

( w) j ( w)k

c jk =

8 h j hk (h j + jk hk ) 3jk 2
(1 jk2 ) 2 + 4h j hk jk (1 + jk2 ) + 4(h 2j + hk2 ) jk2

(32)

jk = j k

6.6

6-1 6-2 40,000kl 100,000kl

- 75 -

6-1
U-25
U-40

25
m
40 m

15
m
20 m

3
850 kg/m3
850 kg/m

700 mm
800 mm

1050 mm3
1200 mm3
2.464105 kN-m 3.223105 kN-m

120 kg/m2
120 kg/m2

0.3
0.3

1.0%
1.0%

6-2
S-25
S-40

25
m
40
m

15 m
20 m

850 kg/m3
850 kg/m3

4.5 mm
4.5 mm

3m
5m

700 mm
800 mm

4.5 mm
4.5 mm

12
mm
12 mm

3930 mm
5717 mm3
1.719 kN-m
1.719 kN-m

0.809106 kN-m2 1.671106 kN-m2

1.036106 kN-m2 2.259106 kN-m2


58 kg/m2
58 kg/m2

558 kg/m
819 kg/m

783 kg-m
2815 kg-m

0.3
0.3

0.5%
0.5%

U-25 U-40
S-25 , S-40
2003 K-NET
HKD1290309260450EW0.5%
6-2 0.01s 300s
=1/41

6-2

- 76 -

6-3 U-25
(a) (r = R) (b) (r = 0.4 R) (c) (r = 0.4 R)

6-4 U-40
(a) (r = R) (b) (r = 0.4 R) (c) (r = 0.4 R)

6-5 S-25
(a) (r = R) (b)(c)

6-6 S-40
(a) (r = R) (b)(c)

- 77 -

6-7 S-25
(a) (r = R) (b)(c)

6-8 S-40
(a) (r = R) (b)(c)

1.0%U-25/U-40 0.5%S-25/S-40

SRSS CQC

6-3 6-6 S-25/ S-40 6-7 6-8


6-1 6-2
U-25/U-40 2
S-25/S-40 3 S-25/S-40
30
U-25/U-40
S-25/S-40

S-25/S-40 24 30
6-7 6-8

= 0
6-9 6-12 U-25/U-40 S-25/ S-40
SRSS CQC
S-25/S-40 SRSS

- 78 -

6-9 = 0 U-25
(a)(b)(c)

6-10 = 0 U-40
(a) (b)(c)

6-11 = 0 S-25
(a)(b)(c)

6-12 = 0 S-40
(a)(b)(c)
SRSSCQC

- 79 -

CQC
S-25/S-40
CQC

6.7

30

SRSS CQC

SRSS

CQC

[1]
615pp.119-1262007.5

[2] S.P. Timoshenko and P. Woinowsky-Krieger: Theory of Plates and Shells, 2nd ed., McGraw-Hill,
1959
[3] 1981
[4] Der Kiureghian, A.: A Response Spectrum Method for Random Vibration Analysis of MDF Systems,
Earthquake Engineering and Structural Dynamics, Vol.9, pp.419-435, 1981

- 80 -

7.

[1]

MODEL TESTS ON SLOSHING OF A FLOATING ROOF IN A CYLINDRICAL


LIQUID STORAGE TANK UNDER SEISMIC EXCITATION
ABSTRACT

Shaking table tests are carried out to validate the analytical solutions for the sloshing of a floating roof
in a cylindrical liquid storage tank under seismic excitation. The experimental tank is a 1/100 scaled model
of typical oil-storage tank of 100,000m3 capacity, made of acrylic tube of 800mm in diameter. The tests are
performed using three types of floating roof model: (1) a roof composed of a pontoon ring only, (2) a roof
composed of uniform isotropic plate, and (3) a single-deck type roof composed of an inner deck and an
outer pontoon. The motion capture system using high-speed micro cameras is employed to measure the
roof displacement over the whole roof surface. The test results are compared with the analytical solutions
based on linear potential theory. Overall agreement is confirmed between theory and experiment, while
nonlinear bi-harmonic resonance oscillation is observed to occur in certain cases.
Keywords: shaking table test, liquid storage tank, floating roof, sloshing, fluid-structure interaction,

potential theory
INTRODUCTION

Sloshing of contained liquid is one of the major considerations in the design of liquid storage tanks. In
the past major earthquakes many tanks have been subjected to serious damages which may be attributed to
liquid sloshing. Especially during the September 2003 Tokachi-oki earthquake, seven oil-storage tanks of
floating-roof type located at Tomakomai, Hokkaido, Japan were seriously damaged (Hatayama et al. [2]).
Damages include the sinking of the floating roof, which led one of the seven tanks to a whole surface fire,
as observed also in the 1999 Kocaeli earthquake, Turkey (The Japan Society of Civil Engineers [3]).
Although the failure of the floating roof and the fire of oil-storage tanks have been observed frequently,
e.g., during the 1964 Niigata earthquake and the 1983 Nihonkai-chubu earthquake, the sinking of the
floating roof caused by sloshing has never been experienced so far in Japan. It was a very dangerous
situation that the oil surface was directly exposed to the air.
After the 2003 Tokachi-oki earthquake, the Fire and Disaster Management Agency of Japan [4] has
issued the amended Notification of the Fire Defense Law, in which the design spectrum for sloshing was
increased to almost twice the spectrum in the past Notification. In addition, the standard for the seismic
design of floating roofs under long-period ground motion, which has never been included in the past
Notification, has been newly regulated. This requires the evaluation of earthquake-resistance capacity of
the floating roof of many existing as well as newly designed tanks and raising it up to the level requested
by the amended Notification. Thus, there is a rapidly increasing demand for predicting the sloshing
response of floating roofs under long-period seismic excitation.

- 81 -

In the earlier papers by Matsui [5, 6], analytical solutions have been presented for the sloshing of a
floating roof in a cylindrical liquid storage tank under seismic excitation. Explicit solutions have been
obtained for two types of floating roof: (a) a floating roof composed of uniform isotropic plate, which may
be a rational approximation to a double-deck type floating roof, and (b) a single-deck type floating roof
composed of an inner deck with relatively small bending stiffness and an outer pontoon with relatively
large stiffness. The dynamic interaction between the liquid and the floating roof was taken into account
exactly within the framework of linear potential theory.
In the present paper, shaking table tests are carried out to confirm the validity of the analytical
solutions based on the linear potential theory. The experimental tank is a 1/100 scaled model of typical
oil-storage tank of 100,000m3 capacity, made of acrylic tube of 800mm in diameter. The motion capture
system using high-speed micro cameras is employed to measure the roof displacement over the whole roof
surface. The test results are compared with the analytical solutions to be validated.
TEST PROCEDURE
Experimental Model

The principal dimensions of the experimental tank are shown in Table 1 and Fig. 1. The tank is made of
acrylic tube of 800mm in diameter, 10mm thick and 800mm high and filled with water to a depth of
400mm. The tests are carried out using three types of floating roof model: (1) a roof composed of a
pontoon ring only (which is referred to as free surface hereafter), (2) a roof composed of uniform
isotropic plate (which is referred to as double-deck type roof hereafter), and (3) a single-deck type roof
composed of an inner deck and an outer pontoon. The floating roofs are made of acrylic plate of 2mm thick
except the inner deck of the single-deck type roof made of polyvinyl chloride sheet of 0.1mm thick. To
provide buoyancy a number of Styrofoam blocks of 1mm thick are attached to the lower surface of the
double-deck type roof and of the pontoon of the single-deck type roof, as shown in Photo 1. Due to their
light weight and low bending stiffness compared with those of acrylic plates the effect of the Styrofoam
blocks on the response is ignored in the analysis later.
800 mm

770 mm
2 mm

780 mm

Double-deck Type Roof

2 mm

400 mm

770 mm

0.1 mm
45 mm

Single-deck Type Roof

Tank

Fig. 1 Experimental Tank

- 82 -

Table 1 Principal Dimensions of Tank Model


Outer diameter of tank
800 mm
Inside diameter of tank
780 mm
Liquid depth
400 mm
Mass density of liquid
1000 kg/m3
Diameter of floating roof
770 mm
Thickness of deck (double-deck)
2 mm
Thickness of inner deck (single-deck)
0.1 mm
Thickness of pontoon (single-deck)
2 mm
Width of pontoon (single-deck)
45 mm
1244 kg/m3
Mass density of deck (double-deck)
Mass density of inner deck (single-deck) 2836 kg/m3
1206 kg/m3
Mass density of pontoon (single-deck)
Young's modulus of acrylite
2.90 Gpa
Measurement System of Roof Displacements

Three high-speed micro cameras are set up at the top of tank as shown in Photo 2 to catch the 3D
movement of the targets marked by white-colored small circles on the black-painted surface of the floating
roof. The motion capture system is employed to obtain the digitized time-series data of roof displacement
over the whole roof surface, measured simultaneously at the frame rate of an integral sub-multiple of 60fps
(10, 12, 15, 20, 30 and 60fps) with the restriction of 262 frames in storage capacity. The arrangement of the
measuring points is shown in Fig.2.

Double-deck Type Roof

Single-deck Type

Photo 1 Floating Roof Models

Photo 2 Set Up of High-speed


Micro Cameras

Ground

Free Surface

Double-deck Type

Single-deck Type Roof

Fig. 2 Arrangement of Measuring Points

- 83 -

FREE-OSCILLATION TESTS

Free-oscillation tests have been carried out to measure the natural period, damping ratio and
free-oscillation mode of the floating roof models in liquid. The free-oscillation was generated by pulling up
and down the string attached to an edge of the roof and releasing it after the oscillation was fully
developed. Six tests have been performed for each floating roof type under the test conditions shown in
Table 2. The frame rate was determined to be an integral sub-multiple of 60fps by taking into consideration
the range of natural frequencies to be measured and the storage capacity of the motion capture system.
The measured time histories of free-oscillation y (t ) were fitted by the least-square method to
y (t ) = exp( t ) { A cos(d t ) + B sin(d t )}

(1)

where t denotes the time, and d denotes the damped natural circular frequency of the roof. The
coefficients A , B , and d can be determined to minimize the error norms between the measured
and fitted time histories of oscillation. Once these coefficients have been determined, the natural period T
and damping ratio h can be evaluated from
T = 2 d ,

h=

d 2 + 2

(2)

Table 2 Test Conditions for Free-oscillation Tests


Frame rate (fps)
Duration time (s)
17.5
Test 1 / Test 4
15
21.8
Test 2 / Test 5
12
Test 3 / Test 6
10
26.2
Table 3 Natural Periods Evaluated from Free-oscillation Tests (s)
Free surface Double-deck Single-deck

Test 1
Test 2
Test 3
Test 4
Test 5
Test 6
Average
Analysis

0.950
0.950
0.951
0.951
0.950
0.950
0.950
0.945

0.951
0.947
0.950
0.953
Failed
0.950
0.950
0.946

0.951
0.952
0.951
0.953
0.951
0.951
0.952
0.947

Table 4 Damping Ratios Evaluated from Free-oscillation Tests


Free surface Double-deck Single-deck

Test 1
Test 2
Test 3
Test 4
Test 5
Test 6
Average

0.00229
0.00234
0.00243
0.00198
0.00245
0.00199
0.00225

0.01130
0.01060
0.01070
0.01180
Failed
0.01120
0.01110

- 84 -

0.00323
0.00336
0.00340
0.00319
0.00368
0.00342
0.00338

Free Surface

Double-deck Type Roof

Single-deck Type Roof


Fig. 3 Measured and Least-square Fitted Time
Histories of Free-oscillation

The corresponding free-oscillation mode can be obtained from the amplitude ratio of least-square fitted
free-oscillation curve at each measuring point.
Figure 3 shows examples of the measured and least-square fitted time histories of free oscillation.
Complete agreement is observed between the measured and fitted oscillation curves, illustrating the validity
and high accuracy of the least-square fitting.
The natural periods and damping ratios evaluated from free-oscillation tests are shown in Tables 3 and
4. These are the averages of those values obtained from the free-oscillation curve of every measuring point.
In Table 3 the fundamental natural periods predicted by the analytical solutions (Matsui [4, 5]) are also
shown for comparison. It can be observed that the variance of the natural period of each test is quite small
and that the measured natural periods are very close to the analytical solutions, confirming the validity of
the linear potential theory. The variance of damping ratios is also found to be small. It can be noted that the
damping values depend strongly on the type of floating roof. They are lowest in the case of free surface,
and the single-deck type roof provides lower damping than the double-deck type roof.
The free-oscillation modes evaluated from free-oscillation tests are plotted and compared with the
analytical solutions in Figs. 4 and 5, where r R and denote the radial coordinate normalized by the
radius R of the floating roof and the circumferential coordinate, respectively. Again satisfactory
agreement is observed between theory and experiment, confirming the validity of the linear potential
theory.
- 85 -

Double-deck Type Roof

Single-deck Type Roof

Fig. 4 Radial Free-oscillation Modes

Double-deck Type Roof

Single-deck Type Roof

Fig. 5 Circumferential Free-oscillation Modes


SEISMIC OSCILLATION TESTS

Seismic oscillation tests were carried out to measure the sloshing response of the floating roof models
under seismic ground motion. The earthquake wave recorded at the K-NET Tomakomai station during the
2003 Tokachi-oki earthquake (HKD1290309260450EW) was selected as an input ground motion. Three
tests have been performed for each floating roof type under the test conditions shown in Table 5.
The experimental tank was designed to be a 1/100 scaled model of typical oil-storage tank of
100,000m3 capacity (80m in diameter), but it is possible to consider tanks of other size by adjusting the
time scale of ground motion. It is well-known that the law of similarity is satisfied by t ' t = L ' L , where
t and L denote the time and characteristic dimension in real scale, while t ' and L ' denote the

corresponding quantities in model scale. In the present study two tank sizes, referred to as model L and
model S hereafter, are considered by reducing the time scale as shown in Table 6.
Figure 6 shows an example of time histories and corresponding velocity response spectra of the target
and generated ground motions in the case of free surface of model L. It can be observed that the target
wave is generated on the shaking table with high accuracy over the whole range of periods.
Earthquake response analysis was performed by using the table motions generated on the shaking table
as input ground motions. Newmarks method (the average acceleration method) was adopted with the
sampling period and the duration time shown in Table 7. The Rayleigh damping as well as the stiffness-

- 86 -

Table 5 Test Conditions for Seismic Oscillation Tests

Test 1
Test 2
Test 3

Frame rate (fps)


15
12
10

Duration time (s)


17.5
21.8
26.2

Table 6 Model Scales Corresponding to Real Tank

Model

Time scale

Model L
Model S

1/10
1/7.91

Dimension
scale
1/100
1/62.5

Diameter/Liquid
depth of real tank
80m/40m
50m/25m

Table 7 Parameters in Time History Analysis

Model

Sampling period (s)

Duration time (s)

Model L
Model S

1/160
1/187.5

51.2
43.7

proportional damping were assumed for each mode using the measured damping ratios shown in Table 4
for the fundamental and second modes in liquid. 6 free surface modes and 5 elastic vibration modes in air
were adopted for the double-deck type roof, while 30 free surface modes and 29 elastic vibration modes in
air for the single-deck type roof. It should be noted that a larger number of modes are needed to express the
pressure and the bending moment in the deck of the single-deck type floating roof, which increase rapidly
as approaching to the connection between the deck and the pontoon. For more details, see Matsui [5, 6].
Figures 7 and 8 show the comparison between the measured and simulated maximum amplitudes of
roof displacement along the radius parallel to the direction of ground motion. Quite good agreement is
observed between theory and experiment except the slight discrepancies along the outer circumference.

Fig. 6 Time Histories and Velocity Response Spectra of Target


and Generated Ground Motions (Free Surface of Model L)

- 87 -

Double-deck Type Roof

Double-deck Type Roof

Single-deck Type
Type Roof
Roof
Single-deck

Single-deck Type Roof

Fig. 7 Maximum Amplitude of Displacement


along the Radius Parallel to the Direction of
Ground Motion (Model L)

Fig. 8 Maximum Amplitude of Displacement


along the Radius Parallel to the Direction of
Ground Motion (Model S)

Figures 9 - 14 show the time histories and corresponding Fourier amplitude spectra of roof
displacement at the wave front, together with the natural frequencies evaluated from the analytical

Fig. 9 Time Histories and Fourier Amplitude


Spectra of Displacement at the Wave Front
(Free Surface of Model L)

Fig. 10 Time Histories and Fourier Amplitude


Spectra of Displacement at the Wave Front
(Free Surface of Model S)

- 88 -

Fig.11 Time Histories and Fourier Amplitude


Spectra of Displacement at the Wave Front
(Double-deck Type Roof of Model L)

Fig. 12 Time Histories and Fourier Amplitude


Spectra of Displacement at the Wave Front
(Double-deck Type Roof of Model S)

solutions. In the case of model L quite good agreement is observed between analysis and experiment. On
the other hand, in the case of Model S it can be observed that the analysis gives somewhat higher

Fig. 13 Time Histories and Fourier Amplitude Fig. 14 Time Histories and Fourier Amplitude
Spectra of Displacement at the Wave Front
Spectra of Displacement at the Wave Front
(Single-deck Type Roof of Model S)
(Single-deck Type Roof of Model L)

- 89 -

Fig. 16 Fourier Coefficients of Displacement


Expanded around the Circumference
(Single-deck Type Roof of Model S)

Fig. 15 Time Histories and Fourier Amplitude


Spectra of Displacement at the Center of Roof
(Single-deck Type Roof of Model S)

Fig. 17 3D Plot of Measured Bi-harmonic


Resonance Oscillation Mode
(Single-deck Type Roof of Model S)

predictions than the experiment. The oscillation after the main earthquake motion has passed decays more
slowly in the analytical curve than in the experimental one. As a possible reason for these discrepancies it
can be considered that the damping ratio for model S is higher than that evaluated by the free-oscillation
tests due to its amplitude-dependent nature. This may be justified by the fact that the roof displacement of
model S amounts to 50mm, which is beyond the range of displacement in the free-oscillation tests (see Fig.
3).
From the figures of the Fourier amplitude spectra shown in Figs. 9 14, it can be observed that the
contribution of the fundamental mode is dominant in most cases. In these cases the difference between the
Rayleigh damping and the stiffness-proportional damping cannot be recognized. However, in the case of
the free surface and the single-deck type roof of model L, it can be noted that the contribution of the second
mode can never be ignored. In these cases the Rayleigh damping gives better predictions than the
stiffness-proportional damping.
Figure 15 show the time histories and corresponding Fourier amplitude spectra of displacement
measured at the center of the single-deck type roof of Model S. According to the linear theory, the roof
oscillates with only a component with the number of circumferential waves 1, resulting in zero
displacement at the center of the roof. However, the measured displacement at the center of the roof is
found to be as high as 7mm which can never be ignored, and has the peak of Fourier amplitude spectrum at
2.09Hz which is just twice the fundamental natural frequency (1.05Hz). This implies the occurrence of
nonlinear bi-harmonic resonance oscillation, as discussed by Ohmori, et al. [7] for the case of free surface.
It is well-known that the bi-harmonic resonance oscillation occurs when the natural frequencies of modes
- 90 -

Table 8 Computed Natural Frequencies of Single-deck Type


Floating Roof Model (Hz)

Number of radial half


waves
1
2
3

Number of circumferential waves


n =0
n =1
n =2
1.56
1.06
1.39
2.11
1.84
2.07
2.55
2.33
2.52

with the number of circumferential waves 0 and/or 2 coincide with twice the fundamental natural frequency
with the number of circumferential waves 1. This is just the case in the singe-deck type floating roof model
as shown in Table 8, where the natural frequencies of modes with the number of circumferential waves 0
and 2 and the number of radial half waves 2 are close to twice the fundamental natural frequency of mode
with the number of circumferential waves 1 and the number of radial half waves 1. In order to confirm the
occurrence of bi-harmonic resonance oscillation, Fig. 16 shows the Fourier coefficients of the roof
displacement expanded around the circumference. In addition to the linear component with the number of
circumferential waves 1 which is dominant, the components with the number of circumferential waves 0
and 2 can be recognized. The bi-harmonic resonance oscillation mode can be observed clearly in Fig. 17,
showing the 3D plot of the measured roof displacement band-pass filtered between the frequencies of 2.0
and 2.2Hz.Such a nonlinear oscillation mode can also be observed in model L although not so significantly
as in model S.
CONCLUSIONS

Shaking table tests using a small-scale model have been carried out to validate the analytical solutions
for the sloshing of a floating roof in a cylindrical liquid storage tank under seismic excitation. The tests
have been performed using three types of floating roof model: (1) a roof composed of a pontoon ring only,
(2) a roof composed of uniform isotropic plate, and (3) a single-deck type roof composed of an inner deck
and an outer pontoon. The test results were compared with the analytical solutions based on linear potential
theory. Overall agreement was confirmed between theory and experiment, while nonlinear bi-harmonic
resonance oscillation was observed to occur in certain cases. The theoretical prediction of such nonlinear
oscillations will be a matter of further research.
Finally, the conclusions obtained from the present experimental study can be summarized as follows:
The analytical solutions for the fundamental natural periods and the corresponding free-oscillation
modes coincide well with those measured by the free-oscillation tests.
The damping ratios evaluated from the free-oscillation tests depend strongly on the type of floating roof.
They are lowest in the case of free surface, and the single-deck type roof provides lower damping than
the double-deck type roof.
The analytical solutions for the roof displacement also coincide well with those measured by the seismic
oscillation tests if the damping ratios are properly assumed.
To the roof displacement the contribution of the fundamental mode is dominant in most cases. However,
in the case of the free surface and the single-deck type roof of larger diameter, the contribution of the
- 91 -

second mode can never be ignored. In these cases the Rayleigh damping gives better predictions than
the stiffness-proportional damping.
As the roof displacement increases, the nonlinear bi-harmonic resonance oscillation with the number of
circumferential waves 0 and 2 can occur at twice the fundamental natural frequency with the number of
circumferential waves 1.
ACKNOWLEDGMENTS

This work has been carried out as a part of research project in the Advanced Research Center for
Seismic Experiments and Computations, Meijo University. The financial support by the Grant-in-Aid for
Scientific Research from the Japan Society of Promotion of Science (through Grant No. 17360278) and
from Meijo University Research Institute is also greatly acknowledged. The strong motion data was
provided by the K-NET of National Institute for Earth Science and Disaster Prevention, Japan.
REFERENCES

[1] T. Nagaya, T. Matsui, T. Wakasa: Model Tests on Sloshing of a Floating Roof in a Cylindrical Liquid
Storage Tank under Seismic Excitation, Proceedings of the ASME Pressure Vessels and Piping
Division Conference, Chicago, Illinois, U.S.A, PVP2008-61675, July 2008, to be published
[2] Hatayama, K., Zama, S., Nishi, H., Yamada, M., Hirokawa, M., and Inoue, R.: The Damages of Oil
Storage Tanks during the 2003 Tokachi-oki Earthquake and the Long Period Ground Motions,
Proceedings of the JSCE-AIJ Joint Symposium on Huge Subduction Earthquakes -Wide Area Strong
Ground Motion Prediction-, pp.7-18, 2005, in Japanese
[3] The Japan Society of Civil Engineers: The 1999 Kocaeli Earthquake, Turkey-Investigation into
Damage to Civil Engineering Structures-, the Japan Society of Civil Engineers, Tokyo, 1999
[4] The Fire and Disaster Management Agency: On the Enforcement of the Ministerial Ordinance Which
Amends a Part of the Rule Concerning the Control of Hazardous Materials, Notification 14, 2005, in
Japanese
[5] Matsui, T.: Sloshing in a Cylindrical Liquid Storage Tank With a Floating Roof Under Seismic
Excitation, Journal of Pressure Vessel Technology, Transactions of the ASME, Vol. 129, No.4,
pp.557-566, November 2007
[6] Matsui, T.: Sloshing in a Cylindrical Liquid Storage Tank With a Single-deck Type Floating Roof
Under Seismic Excitation, Proceedings of the ASME Pressure Vessels and Piping Division Conference,
San Antonio, Texas, PVP2007-26249, July 2007
[7] Ohmori, H., Hibino, H., Kato, K., and Matsui, T.: Nonlinear Sloshing of Cylindrical Liquid Storages,
Proceedings of the IASS Symposium on Shells, Membranes and Space Frames, Osaka, Vol.1, pp.
97-104, 1986

- 92 -

8.

[1]

WIND EFFECTS ON SLOSHING OF A FLOATING ROOF IN AN OPEN-TOPPED


CYLINDRICAL LIQUID STORAGE TANK
ABSTRACT

Sloshing of a floating roof in an open-topped cylindrical liquid storage tank under wind loads is
investigated analytically. Wind tunnel test in a turbulent boundary layer is carried out to measure the wind
pressure distributing over the roof surface. The measured data for the wind pressure is then utilized to
predict the wind-induced dynamic response of the floating roof, which is idealized herein as an isotropic
elastic plate of uniform stiffness and mass. The dynamic interaction between the liquid and the floating roof
is taken into account exactly within the framework of linear potential theory. Numerical results are
presented which illustrate the significant effect of wind loads on the sloshing response of the floating roof.
Keywords: liquid storage tank, floating roof, wind load, sloshing, fluid-structure interaction, wind tunnel

test
INTRODUCTION

Sloshing of contained liquid is one of the major considerations in the design of liquid storage tanks. In
the past major earthquakes many tanks have been subjected to serious damages which may be attributed to
liquid sloshing. Damages include the failure of floating roofs followed by the whole-surface fire or the ring
fire of tanks as observed frequently, e.g., during the 1964 Niigata earthquake and the 1983 Nihonkai-chubu
earthquake. What is most dangerous is the sinking of floating roofs followed by the whole surface fire
observed at the first time during the 1999 Kocaeli earthquake and the 2003 Tokachi-oki earthquake. Learnt
from these experiences, extensive studies have been performed to predict the sloshing of floating-roof type
liquid storage tanks under earthquake excitation (e.g., Nakagawa [2], Yamamoto [3], Sakai et al. [4],
Nishiguchi et al. [5], Yamauchi et al. [6], Matsui [7, 8]). However, there has been scarcely any information
on the wind loads and their effects on the sloshing of the floating roof in a storage tank.
Recently, wind tunnel test in a turbulent boundary layer has been carried out by Uematsu et al. [9] to
measure the wind pressure distributing over the floating roof in an open-topped cylindrical liquid storage
tank. With these data in hand, the present paper aims to investigate the effect of wind loads on the sloshing
of the floating roof analytically. The measured data for the wind pressure is utilized to predict the
wind-induced dynamic response of the floating roof, which is idealized herein as an isotropic elastic plate
of uniform stiffness and mass. The dynamic interaction between the liquid and the floating roof is taken
into account exactly within the framework of linear potential theory (Matsui [7, 8]). Numerical results are
presented to investigate the effect of wind loads on the sloshing response of the floating roof.

- 93 -

MATHEMATICAL SOLUTION
Boundary Value Problem

Sloshing in an open-topped cylindrical liquid storage tank of radius R with flat bottom is considered
here. The tank is partially filled with liquid to a height H . The liquid surface is covered by a floating roof
composed of an isotropic elastic plate with uniform stiffness and mass. The tank wall may be assumed to be
rigid with reasonable accuracy because the natural periods of shell vibration modes are much shorter than
the natural periods of sloshing modes. A cylindrical coordinate system (r , , z ) is defined as shown in Fig.
1 with an origin at the center of the tank bottom and the z -axis vertically upwards. The tank is subjected
to wind pressure pw (t ) over the floating roof due to the wind blowing in the direction = 0 .
The liquid is assumed to be inviscid, incompressible and irrotational. Then the liquid motion may be
completely defined by a velocity potential function , for which the boundary-value problem can be
defined as
2 = 0 in a liquid region

(1a)

= 0 at the tank wall r = R


r

(1b)

= 0 at the tank bottom z = 0


z

(1c)

= w at the liquid surface z = H


z

(1d)

where w denotes the vertical displacement of the floating

roof, and the dot denotes the derivative with

respect to time t .

Fig. 1 Tank Geometry and Coordinate System

- 94 -

The hydrodynamic pressure p on the floating roof can be evaluated from linearized Belnoullis
equation

p =
gw
t z=H

(2)

where denotes the mass density of liquid, and g denotes the acceleration of gravity.
Neglecting the damping terms, the equation of motion of the floating roof can be written as
mw + D 4 w = p pw

(3)

where m denotes the mass density of the floating roof per unit surface area, and D denotes the bending
rigidity of the floating roof.
Modal Decomposition of Displacements

The validity of the superposition principle is assumed for both the displacement of the floating roof and
the liquid motion. Then the vertical displacement w(t ) of the floating roof can be represented as a linear
superposition of the rigid-body modes Z 0 0 , Z 01 cos and the free elastic vibration modes Z ln cos n
(l = 1, 2, , ; n = 0, 1, , ) in air

w(t ) = ln (t ) Z ln cos n

(4)

n=0 l =0

where ln (t ) denotes the modal displacement corresponding to the mode Z ln cos n with the number of
radial half waves l and the number of circumferential waves n .

The radial mode shape Z ln can be

given by
Z 00 = 1 , Z 01 =

r
r
r

, Z ln = ln J n ln + ln I n ln (l 1)
R
R
R

(5)

where J n denotes the Bessel function of the first kind of order n , I n denotes the modified Bessel
function of the first kind of order n , and ln is the wave number which may be related to the natural
circular frequency ln of order (l , n) by

ln4 =

mR 4 2
ln
D

(6)

The wave number ln and the amplitude ratio of ln to ln can be determined from the requirement
that the homogeneous equations representing the stress-free conditions along the roof edge have the
non-trivial solutions
c11
c
21

c12 ln 0
=
c22 ln 0

(7)

where
c11 = ln2 J n ( ln ) + ln J n ( ln ) n 2 J n ( ln )
- 95 -

c12 = ln2 I n ( ln ) + ln I n ( ln ) n 2 I n ( ln )

c21 = ln3 J n( ln ) + ln2 J n ( ln ) ln 1 + ( 2 ) n 2 J n ( ln ) + ( 3 ) n 2 J n ( ln )


c22 = ln3 I n( ln ) + ln2 I n ( ln ) ln 1 + ( 2 ) n 2 I n ( ln ) + ( 3 ) n 2 I n ( ln )

(8)

denoting Poissons ratio.


Velocity Potential and Pressure

The solution for the potential that satisfies the Laplace equation (1a), the wall condition (1b) and
the bottom condition (1c) is given by

= Ain (t ) cosh in
n = 0 i =1

where in denotes

z r
J n in cos n
R R

(9)

the i -th positive root of J n ( in ) = 0 .

On substitution of Eqs. (4) and (9) into Eq. (1d), the kinematic condition at the liquid surface can be
expressed as

Ain (t )
i =1

in2
H r
cosh in J n in = ln (t ) Z ln
g
R R l =0

(10)

where in denotes the natural circular frequency of free surface mode of order (i , n) given by
g
H
in tanh in
R
R

in =

On multiplying

( r R ) J n ( in r R )

(11)
on the both sides of Eq. (10), integration with respect to u = r R

over 0 u 1 and making use of the orthogonality relation of the Bessel function, Eq. (A1), the unknown
coefficient Ain

is determined in the form

Ain (t ) =

2
2
in n 2

1
1
g
H J ( ) in2
cosh in n in
R

a
l =0

(t )

iln ln

(12)

where
ailn =

1
in2
uZ ln J n ( in u )du

J n ( in ) 0

(13)

After substitution of Eq. (5) into Eq. (13) and making use of Eqs. (A2) - (A4) in the appendix, ain can be
obtained in explicit forms
ai 00 = 0, ai 01 = 1,

ailn = in2 ln 2 ln 2 J n ( ln ) + 2 ln 2 I n ( ln ) (l 1)
in + ln
in ln

(14)

On substitution of Eq. (12) into Eq. (9), the solution for the potential can be obtained in the form

- 96 -

2
g cosh ( in z R ) J n ( in r R )
ailnln (t ) cos n
2
2
J n ( in )
i =1 n in cosh ( in H R )
l =0

=
n =0

(15)

2
in

The hydrodynamic pressure on the floating roof can then be evaluated by substitution of Eqs. (4) and (15)
into Eq. (2)

g J n ( in r R )
2

g
Z lnln (t ) cos n
p = 2
a

t
(
)
+

iln
ln
2
2
l =0
n=0

i =1 in n in J n ( in ) l = 0

(16)

Equations of Motion

The final step to the solution is to determine the unknown modal displacements ln (t ) involved in
Eqs. (15) and (16). This is achieved by solving the equation of motion of the floating roof, as described
below.
The wind pressure pw on the floating roof can be expanded conveniently into Fourier series in

pw (r , , t ) = pn (r , t ) cos n

(17)

n=0

where the wind pressure distribution is assumed to be symmetric with respect to the axis = 0 , which is
parallel to the wind direction.
On substitution of Eqs. (4), (16) and (17) into Eq. (3), the equation of motion of the floating roof can
be written as

m
l =0

2
g J n ( in r R )
ailnln (t ) g Z lnln (t )

2
2
J n ( in ) l = 0
i =1 n in
l =0

ln

(t ) + ln2 ln (t ) Z ln = pn

2
in

(n = 0, 1, , ) (18)

( r R ) Zln on the both sides of Eq. (18), integration with respect to

On multiplying

u = r R over 0 u 1

and making use of an orthogonal relation of the free vibration modes yield

(
k =0

lk

kn + lkn ) kn (t ) + ln2 +
ln ln (t ) = ln
m

(19)

where lk denotes the Kronecker delta defined by

lk = 0 (l k ) ,

lk = 1 (l = k )

and
1

ln = uZ ln2 du

(20)

lkn =
ln =

1
2
g
1
aikn uZ ln J n ( in u ) du

2
2
2
0
m i =1 in n in J n ( in ) k = 0

1 1
uZ ln pn du
m 0

(21)

(22)

- 97 -

After substitution of Eqs. (5) and (13) into Eqs. (20) and (21), and making use of the integral formulas for
the Bessel functions (Abramowitz and Stegun [10]), ln and lkn can be obtained in explicit forms
01 = 1 4
2

1
ln = ln2 { J n ( ln )} J n 1 ( ln ) J n +1 ( n ) + ln ln J n ( ln ) I n +1 ( ln ) + J n +1 ( ln ) I n ( ln )

2
ln
+

lk n =

2
1 2
ln {I n ( ln )} I n 1 ( ln ) I n +1 ( ln ) (l 1)

m
i =1

2
in

2
2
in

g
ailn aikn
in2

(23)

(24)

ln can be evaluated by numerical integration of Eq. (22).


In the equation of motion (19), lkn and g m denote the added mass due to liquid and the
hydrostatic restoring force coefficient due to change of buoyancy, respectively, normalized by the mass
density m of the floating roof. Note that in the present formulation use has been made of the free
vibration modes in air, and consequently, the equation of motion for each mode is not uncoupled with each
other. The free vibration characteristics of the floating roof in liquid can be evaluated by solving the
homogeneous equation of motion (19) without the forcing term.
In the foregoing formulation no damping effects have be taken into account to derive the equation of
motion (19). By adding a proper damping term (e.g. Rayleigh damping or stiffness-proportional damping),
it is now possible to consider the damping effects such as material damping, drag fluid forces and friction
forces along the side wall. In the numerical examples discussed below the stiffness-proportional damping is
assumed using the damping ratio 0.01 for the fundamental mode in liquid with the number of
circumferential waves n =1.
Final Solutions

Once the modal displacements ln (t ) have been obtained by solving the equation of motion (19),
these can be substituted into Eqs. (4), (15) and (16) to evaluate the displacement w of the floating roof,
the velocity potential and the hydrodynamic pressure p on the floating roof, respectively.
According to Kirchhoffs theory of elastic plates, the relations between the displacement and the
curvatures and twist of the floating roof are given by

r =

2 w
1 w 1 2 w
1 2 w
1 w
,
=
+
,
=
2

2
2
2
r
r r r
r r r

Substitution of Eqs. (4) and (5) into Eq. (25) yields


2

r
r
ln

ln J n ln + ln I n ln cos n

R
R
n = 0 l =1


r n2
r

= ln (t ) ln ln J n ln 2 J n ln
R
n = 0 l =1
Rr R r

r n2
r

+ ln ln I n ln 2 I n ln cos n
R
Rr R r

r = ln (t )

- 98 -

(25)

ln
r 1
r
J n ln 2 J n ln
R
Rr R r

r = ln (t ) ln n
n = 0 l =1

r 1
r

+ ln n ln I n ln 2 I n ln sin n
R
Rr R r

(26)

The bending and twisting moments in the floating roof can then be evaluated from
M r = D( r + ), M = D( + r ) ,

M r = D(1 ) r

(27)

WIND TUNNEL TEST

Test Conditions and Apparatus

Wind tunnel test has been carried out in a boundary layer wind tunnel at Kajima Technical Research
Institute to measure the wind pressure distributing over the floating roof in an open-topped cylindrical tank.
The test conditions and the particulars of the test apparatus are shown in Table 1, and the arrangement of
measuring points in Fig. 2 and Table 2. The test apparatus is an open-topped cylinder of 250 mm in
diameter, 125mm high and 6mm thick. The floating roof of 236mm in diameter with a pontoon ring of
15mm wide and 2mm thick was pasted to the side wall, adjusting the liquid depth varying from 2 to 110mm
(Cases A - E). The tests have been performed twice for each liquid depth. As the displacement of the roof is
completely restraint, the aerodynamic wind-structure interaction effect is not included in the measured
pressure. For more details of wind tunnel test, see Uematsu et al. [9].
Table 1 Test Conditions and Apparatus
Geometric scale of wind tunnel flow
1/300
Mean wind velocity at the top of tank 10 m/s
Exponent in power distribution law
0.15
Outer diameter of tank
250 mm
Inside diameter of tank
238 mm
Diameter of floating roof
236 mm
Height of tank
125 mm
Liquid depth (Case A)
2 mm
Liquid depth (Case B)
27.85 mm
Liquid depth (Case C)
64.5 mm
Liquid depth (Case D)
101.15 mm
Liquid depth (Case E)
110 mm
Sampling period
0.001 s
Duration time
32.768 s

Measurement of Wind Pressure

Wind pressures were measured at 97 points on the roof surface by means of pressure sensors, as shown
in Fig. 2. The pressure measured at each point was normalized by the velocity pressure measured at the top
of model tank to obtain the wind pressure coefficient. The design wind pressure is then evaluated by
multiplying the measured wind pressure coefficient by the design velocity pressure at the top of tank.
The law of similarity between the model and real scales can be satisfied by
- 99 -

Fig. 2 Arrangement of Measuring Points on Floating Roof


Table 2 Radial Coordinates of Measuring Points
r/R f *
Points
r (mm)
1-24
109
0.916
25-48
83.25
0.7
49-72
54.5
0.5
73-96
35.75
0.3
97
0
0
* R f : Radius of floating roof

t
L V'
=
t ' L' V

(28)

where t , L and V denote the time, characteristic length and mean wind velocity at the top of tank in
real scale, and t ' , L '

and V ' denote the corresponding quantities in model scale. Equation (28) is

used later to evaluate the sampling period in real scale.


WIND RESPONSE ANALYSIS
Tank Model

The principal parameters of the tank model analyzed here are presented in Table 3. This is a model of
typical oil-storage tank of 40,000m3 capacity of 50m in diameter and 25m high. The liquid depth is varied
parametrically from 0.4 to 22m (Cases A - E).

The floating roof is considered to consist of the upper and

lower decks connected by the rims. It is assumed for simplicity that the contribution of the rims to the
bending rigidity can be ignored. The effective cross-sectional coefficient is evaluated by taking into
consideration the effect of local buckling at the compressive side.
The wind parameters assumed in the analysis are shown in Table 4, where the basic wind velocity, the
mean wind velocity at the top of tank and the design velocity pressure are evaluated by assuming the design

- 100 -

Table 3 Principal Parameters of Tank


Diameter of tank
50 m
Height of tank
25 m
Liquid depth (Case A)
0.4 m
Liquid depth (Case B)
5.57 m
Liquid depth (Case C)
12.9 m
Liquid depth (Case D)
20.23 m
Liquid depth (Case E)
22 m
Mass density of liquid
850 kg/m3
Total height of roof section
700 mm
Thickness of upper and lower decks
4.5 mm
Effective cross-sectional coefficient of 1050 mm2
roof section (per unit width)
Bending rigidity of roof
2.464105 kN-m
Mass density of roof
100 kg/m2
Poisson's ratio
0.3
Damping ratio
0.01
* These values are somewhat different from those adopted
in the wave tunnel test shown in Table 1, but the effect
of these changes may be considered to be slight.
Table 4 Wind Parameters
Geometric scale of wind tunnel flow
1/200*
Basic wind velocity
38 m/s
Exponent in power distribution law
0.10*
Design return period
100 years
Mean wind velocity at the top of tank 51.3 m/s
Design velocity pressure
1607 Pa
Sampling period
0.039 s
Duration time
1278 s

return period of 100 years. These evaluations are based on the recommendations for wind loads on
buildings published by Architectural Institute of Japan [11].
Free-vibration Characteristics

Prior to wind response analysis, free-vibration analysis of Eq. (19) has been performed to obtain the
natural periods of the floating roof in liquid. The modal numbers were truncated at 6 for the free surface
mode (subscript i in Eq. (15)), 5 for the radial elastic mode (subscript l in Eq. (15)) and 12 for the
circumferential mode (subscript n in Eq. (15)), after confirming the convergence. A part of the results are
presented in Table 5 for the case of H =12.9m (Case C).
Table 5 Natural Periods in Liquid (H=12.9m) (s)
n =0
n =1
n =2
n =3
n =4
n =5
1.949 8.581 3.365 1.396 0.722 0.432
l =1
0.688 0.766 0.401 0.246 0.165 0.118
l =2
0.362 0.212 0.139 0.099 0.074 0.057
l =3
0.133 0.092 0.067 0.051 0.041 0.033
l =4
0.065 0.049 0.038 0.031 0.025 0.021
l =5

- 101 -

Wind Pressure Distribution

Figures 3 and 4 illustrate an example of time-history and corresponding power spectrum density of
wind pressure at the center of roof in real scale. It can be observed that the negative (upward) wind
pressure acts with the mean pressure around 1kPa and the maximum negative pressure around 2.3kPa. It
can be noted that the wind pressure has sufficient energy over a range of natural periods of the floating roof
presented in Table 5 to excite significant sloshing response.
Figure 5 displays the contours of the mean and maximum wind pressure distributing over the roof
surface. These are the averages of two test results for each liquid depth. Negative wind pressure is observed
almost allover the roof surface. The magnitude of the wind pressure is not so affected by the liquid depth,
taking the mean pressure around 1.0-1.2kPa and the maximum negative pressure around 2.5kPa. The effect
of the liquid depth is found to be significant on the distribution of the wind pressure. For the liquid depth
shallower than the half-filled case the maximum pressure arises around the center of the roof, while it
moves to the windward direction as the liquid depth approaches to the fully-filled case.

Fig. 3 Time History of Wind Pressure at


the Center of Roof (H=12.9m)

Fig. 4 Power Spectrum Density of Wind


Pressure at the Center of Roof (H=12.9m)
Wind-induced Dynamic Response

Time-history analysis has been performed to predict the sloshing response of the floating roof under
wind loads. Newmarks method (the average acceleration method) was adopted with the time interval
t =0.039s for the duration time Td =1278s. The stiffness-proportional damping was assumed using the

damping ratio 0.01 for the fundamental mode in liquid with the number of circumferential waves n =1.
The wind pressure distributing over the roof surface was evaluated by multiplying the measured
pressure coefficients by the design velocity pressure given in Table 4, and then expanded into finite Fourier
series around the circumference. The Fourier series was truncated at n =12, after confirming that the
- 102 -

H=0.4m

H=5.57m

H=12.9m

H=20.23m

H=22m

Fig. 5 Mean and Maximum Wind Pressure

- 103 -

Fig. 6 Time Histories of Response at the


Center of Roof (H=12.9m)

Fig. 7 Time Histories of Response at the


Windward Edge (H=12.9m)

original wind pressure can be reproduced by Fourier composition. In view of symmetry with respect to the
wind direction =0, only the cosine coefficients were adopted in the analysis. The integration of Eq. (22)
has been evaluated numerically by dividing the integral region 0 u 1 into 10 segments to each of
which Gaussian 2 point formula was applied.
Figures 6 - 9 show the time-histories and corresponding Fourier amplitude spectra of roof
displacement, hydrodynamic liquid pressure and bending stresses at the center and the windward edge
( r =11.9m, =0) of the roof. The bending stresses were evaluated by dividing the bending moments by
the effective cross-sectional coefficient shown in Table 3. It can be observed that the contribution of the
fundamental mode with the number of circumferential waves n =1 at the period 8.581s is dominant to the
roof displacement, while the contribution of the fundamental modes with n =0 at the period 3.365s, n =2
at the period 1.949s and n =3 at the period 1.396s are significant to the liquid pressure and the bending
stresses.
Figure 10 displays the contours of the mean and maximum amplitudes of roof displacement
distributing over the roof surface. These are the averages from two test results for each liquid depth. The
positive (upward) roof displacement is observed almost allover the roof surface. The roof displacement
takes the maximum value at the windward edge, decreases as moving to the central portion of the roof and
again increases as approaching to the leeward edge. The roof displacement tends to increase as the liquid
- 104 -

Fig. 8 Fourier Amplitude Spectra of Response


at the Center of Roof (H=12.9m)

Fig. 9 Fourier Amplitude Spectra of Response


at the Windward Edge (H=12.9m)

depth increases.
Figure 11 displays the contours of the mean and maximum amplitudes of Misess stress distributing
over the roof surface. These are the averages from two test results for each liquid depth. The significant
bending stress is observed almost allover the roof surface. For the liquid depth shallower than the half-filled
case the maximum bending stress arises around the center of the roof, while it moves to the windward
direction as the liquid depth approaches to the fully-filled case. The bending stress tends to increase as the
liquid depth increases, amounting to around 120-130Mpa for the fully-filled case, which is about half the
yielding stress of the steel.

- 105 -

H=0.4m

H=5.57m

H=12.9m

H=20.23m

H=22m

Fig. 10 Mean and Maximum Roof Displacement

- 106 -

H=0.4m

H=5.57m

H=12.9m

H=20.23m

H=22m

Fig. 11 Mean and Maximum Misess Stress

- 107 -

Comparison with Earthquake Response

Earthquake response analysis was also performed for the purpose of comparison. As an input ground
motion acting in the direction =0, the EW component of the 2003 Tokachi-oki earthquake recorded at
Tomakomai K-NET Station (HKD 1290309260450EW) was used with t =0.01s and Td =300s.
The comparison has been made in Fig.12 between the maximum amplitudes of roof displacement and
Misess stress due to wind and earthquake excitations. The roof displacement due to wind is found to be
much smaller than that due to earthquake, whereas the significant bending stresses amounting to half the
yielding stress of the steel can be observed due to wind, which exceed the stresses due to earthquake. This
is because the displacement due to earthquake is dominated by the rigid-body mode which produces no
bending stress, while the displacement due to wind loads includes the elastic modes which give rise to large
bending stresses. Furthermore, while the maximum bending stresses due to earthquake arise around
r / R =0.4, the distribution of stresses due to wind loads is spreading almost allover the roof surface. The
reason for this is because, while only the natural modes with the number of circumferential waves n =1
contribute to the earthquake response, the other modes, particularly with the number of circumferential
waves n =0, n =2 and n =3, contribute to the wind-induced response, as observed in Figs. 8 and 9. This
result seems to suggest the importance of considering the wind effects in the design of floating roofs.

Fig. 12 Comparison between Wind and Earthquake Responses (H=22m)


CONCLUSIONS

Wind tunnel test in a turbulent boundary layer has been carried out to measure the wind pressure
distributing over the floating roof in an open-topped cylindrical liquid storage tank. The measured data was
then utilized to predict the wind-induced dynamic response of the floating roof, which was idealized as an
isotropic elastic plate of uniform stiffness and mass. An important finding in the present study is that,
although the roof displacement due to wind loads is much smaller than that due to earthquake, the
significant bending stresses arise due to wind loads, which exceed the stresses due to earthquake. This
- 108 -

result seems to suggest the importance of considering the wind effects in the design of floating roofs.
ACKNOWLEDGMENTS

This work has been carried out as a part of research project in the Advanced Research Center for
Seismic Experiments and Computations, Meijo University. The strong motion data was provided by the
K-NET of National Institute for Earth Science and Disaster Prevention, Japan.
REFERENCES

[1] T. Matsui, Y. Uematsu, K. Kondo, T. Wakasa, T. Nagaya: Wind Effects on Sloshing of a Floating Roof
in a Cylindrical Liquid Storage Tank, Proceedings of the ASME Pressure Vessels and Piping Division
Conference, Chicago, Illinois, U.S.A, PVP2008-61688, July 2008, to be published
[2] Nakagawa, K.: On the Vibration of an Elevated Water Tank II, Technical Reports of Osaka University,
Vol. 5 , No. 170, pp.317-336, 1955
[3] Yamamoto, Y.: The Liquid Sloshing and the Impulsive Pressures of Oil Storage Tanks due to
Earthquakes, Journal of High Pressure Institute of Japan, Vol. 3 , No. 1, pp.2-8, 1965, in Japanese
[4] Sakai, F., Nishimura, M., and Ogawa H.: Sloshing Behavior of Floating-Roof Oil Storage Tanks,
Computers and Structures, Vol. 19 , No. 1-2, pp.183-192, 1984
[5] Nishiguchi, H., Ito, M., Honobe, H., and Kanoh, T.: Sloshing Action Analysis and Safety Evaluation of
the Oil Tank by the Long Period Earthquake Motion, The Thermal and Nuclear Power, Vol. 56, No.
581, pp.89-94, 2005, in Japanese
[6] Yamauchi, Y., Kamei, A., Zama, S., and Uchida, Y.: Seismic Design of Floating Roof of Oil Storage
Tanks Under Liquid Sloshing, Proceedings of the ASME Pressure Vessels and Piping Division
Conference, Vancouver, Canada, PVP2006-ICPVT11-93280, 2006
[7] Matsui, T.: Sloshing in a Cylindrical Liquid Storage Tank With a Floating Roof Under Seismic
Excitation, Journal of Pressure Vessel Technology, Transactions of the ASME, Vol. 129, No. 4,
pp.557-566, 2007
[8] Matsui, T.: Sloshing in a Cylindrical Liquid Storage Tank With a Single-deck Type Floating Roof
Under Seismic Excitation, Proceedings of the ASME Pressure Vessels and Piping Division Conference,
San Antonio, Texas, PVP2007-26249, 2007
[9] Uematsu, Y., Koo, C., and Kondo, K.: Wind Loads on Open-topped Storage Tanks, Proceedings of
BBAA 6th International Colloquium on Bluff Bodies Aerodynamics and Applications, Milano, Italy,
2008
[10] Abramowitz, M., and Stegun, I. A.: Handbook of Mathematical Functions, 9th ed., Dover, New York,
1972
[11] Architectural Institute of Japan: Recommendations for Loads on Buildings, 4th ed., Architectural
Institute of Japan, Tokyo, 2004
APPENDIX

Let the i -th positive root of J n ( in ) = 0 be denoted by in , the following orthogonality relation

- 109 -

exists (Abramowitz and Stegun [10])


1 n2
1
uJ n ( in u ) J n jn u du = 2 in2
0

2
{ J n ( in )}

i= j

(A1)

i j

The integral formulas of the Bessel functions (Abramowitz and Stegun [10]) read
1

n +1

J n ( in u ) du =

J n +1 ( in ) =

in

J n ( in )

in2

in

J n ( in )

uJ ( u )J ( u ) du
1

ln

in

1
J ( ) J ( ) in J n ( ln ) J n 1 ( in )
= 2
in ln2 ln n 1 ln n in

uI ( u )J ( u ) du
1

ln

in

1
I ( ) J ( ) in I n ( ln ) J n 1 ( in )
= 2
in + ln2 ln n 1 ln n in

Noting that
J n ( in ) = J n 1 ( in ) nJ n ( in ) in = 0

i.e., J n 1 ( in ) = nJ n ( in ) in

the following integral formulas can be obtained


1

n +1

J n ( in u ) du =

in

J n +1 ( in ) =

in2

J n ( in )

(A2)

uJ ( u )J ( u ) du
0

ln

in

1
ln J n 1 ( ln ) J n ( in ) in J n ( ln ) J n 1 ( in )
= 2
in ln2

1
ln J n 1 ( ln ) nJ n ( ln ) J n ( in ) = 2 ln 2 J n ( ln ) J n ( in )
in2 ln2
in ln

(A3)

uI ( u )J ( u ) du
0

ln

in

1
ln I n 1 ( ln ) J n ( in ) in I n ( ln ) J n 1 ( in )
= 2
in + ln2

1
ln I n 1 ( ln ) nI n ( ln ) J n ( in ) = 2 ln 2 I n ( ln ) J n ( in )
in2 + ln2
in + ln

- 110 -

(A4)

9.

1 17

1)

1
2

2)

3)

0.4

4)

5)

2003

2 18

- 111 -

FEM

1)

2)

3)

2003

FEM

3 19

780mm 100,000kl 1/100

1)

2)

0.2%
0.3% 1.1%
0.1% 0.5% 1.0%

3)

4)

1 0
1 2 0
2

- 112 -

1983
2005

2023 B

- 113 -

WG WG

FS

18
19~23

K-NET

- 114 -

You might also like