You are on page 1of 13

Simplified Formula for Earthquake-Induced Hydrodynamic

Pressure on Round-Ended and Rectangular


Cylinders Surrounded by Water
Piguang Wang 1; Mi Zhao 2; and Xiuli Du 3
Downloaded from ascelibrary.org by University of Central Florida on 12/17/20. Copyright ASCE. For personal use only; all rights reserved.

Abstract: An accurate and efficient numerical model is developed to calculate the earthquake-induced hydrodynamic pressure on uniform
vertical cylinders with an arbitrary cross section surrounded by water. According to the boundary conditions and using the variables
separation method, the three-dimensional Laplace equation governing the incompressible water is transformed into a two-dimensional (2D)
Helmholtz equation. As a key element, a circular boundary surrounding the structures is introduced so that the computational domain is
partitioned into unbounded and bounded domains. The unbounded domain is simulated by an exact artificial boundary condition, which is
derived by using the separation variable method. The impedance matrix of the entire domain is obtained by the finite-element method. The
hydrodynamic forces on rectangular and round-ended cylinders are calculated, which can be modeled as the product of an added mass
of water and the acceleration of the cylinder. However, these complicated expressions of the hydrodynamic forces are not suitable for
engineering application. Therefore, simplified formulas for the added mass of the round-ended and rectangular cylinders are obtained by
the curve-fitting method. The results indicate that the precision of the present added mass formulas is enough for engineering applications.
DOI: 10.1061/(ASCE)EM.1943-7889.0001567. © 2018 American Society of Civil Engineers.
Author keywords: Hydrodynamic pressure; Finite element; Artificial boundary condition; Arbitrary cross section; Added mass; Curve
fitting.

Introduction the response of the structure instead of the water. The research on
seismic hydrodynamic pressure began with the gravity dam
With the development of the economy and transportation, more and (Westergaard 1933) and the cylindrical tank (Jacobsen 1949).
more deep-water bridges have been constructed around the world Liaw and Chopra (1974) initially investigated the significance of
in recent years, especially in China. These deep-water bridges are hydrodynamic pressure on the dynamic response of cantilever cir-
always under threat of earthquakes because most of them are lo- cular cylinders. They discovered that water compressibility was
cated in zones with high seismic intensity. Some research has in- negligible for slender cylinders but important for squat cylinders
dicated that bridges on land usually suffer severe damage during vibrating at high frequency, and the surface waves were significant
earthquakes (Han et al. 2009; Wilson et al. 2014; Luke and Suren only at very low frequencies. If the water is incompressible, the
2016). The experiment conducted by Wei et al. (2013) also indi- seismic hydrodynamic pressure is a kind of inertia force that is
cated that fluid–structure interaction had significant effects on the equal to a product of the constant mass of water and the acceler-
dynamic response of bridge pile foundations submerged in water. ation of the cylinder. Since then, several researchers have studied
Therefore, seismic design and analysis is necessary for bridges on the hydrodynamic pressure on circular cylinders. Williams (1986)
land and in deep water. Earthquake analysis of deep-water bridges used the boundary integral method to investigate the dynamic re-
or bride cylinders requires special consideration because of the sponse of circular cylinders subjected to high-frequency horizontal
water–structure interaction. The water–structure should be con- ground excitation, where the cylinder can extend from the sea bed
sidered because it may affect the dynamic properties and dynamic to, or beyond, the still water level. Tanaka and Hudspeth (1988)
response of the water–structure interaction. To save the computa- proposed an eigenfunction solution for the dynamic response of
tional costs, the water is modeled as the earthquake-induced hydro- circular cylinders surrounded by compressible water subjected to
dynamic pressure on the structure instead of as a real medium earthquake excitation. The solution is limited to a vertical circular
domain. This water model is reasonable due to the concern about cylinder whose length is the same as the fluid depth. Han and Xu
(1996) developed a theoretical model of an added mass repre-
1
Lecturer, College of Architecture and Civil Engineering, Beijing Univ. sentation for a flexible circular cylinder vibrating in water and
of Technology, Beijing 100124, China. Email: wangpiguang1985@126 presented a simple formula for evaluating the natural frequencies
.com using the added mass representation. The simple formula is very
2
Full Professor, College of Architecture and Civil Engineering, Beijing useful to designers working in the area of the dynamics of sub-
Univ. of Technology, Beijing 100124, China (corresponding author). merged flexible structures. Chen (1997) presented a finite-difference
Email: zhaomi@bjut.edu.cn scheme to solve nonlinear hydrodynamic pressure acting on a cir-
3
Full Professor, College of Architecture and Civil Engineering, Beijing
cular cylinder. In the analysis, not only were the three-dimensional
Univ. of Technology, Beijing 100124, China. Email: duxiuli11@163.com
Note. This manuscript was submitted on December 9, 2017; approved equations of motion used, but the simultaneous action of three
on August 20, 2018; published online on December 4, 2018. Discussion components of ground acceleration was also included. Li and Yang
period open until May 4, 2019; separate discussions must be submitted (2013) presented an improved method of hydrodynamic pressure
for individual papers. This paper is part of the Journal of Engineering calculation for circular hollow cylinders and the complicated hy-
Mechanics, © ASCE, ISSN 0733-9399. drodynamic pressure expressions were simplified by curve fitting.

© ASCE 04018137-1 J. Eng. Mech.

J. Eng. Mech., 2019, 145(2): 04018137


Du et al. (2014) proposed a simplified formula of hydrodynamic bottom to above the surface along the z-axis. The cylinder whose
pressure on rigid circular cylinders considering water compressibil- axis line coincides with the z-axis is treated as a one-dimensional
ity in the time domain based on the analytical solution. Jiang et al. structure and the height of the cylinder is Hs . The mass of the
(2017) developed a simplified formula for the hydrodynamic force superstructure is simplified as a lumped mass. The water depth is
on a circular cylinder by elaborate data fitting, where the main h, the foundation is assumed to be rigid, and the earthquake exci-
parameters were radius of the cylinder and water height. In addition, tation is assumed to propagate along the x-direction. The system is
the earthquake-induced hydrodynamic pressure on axisymmetric initially at rest. The numerical solution for hydrodynamic pressure
cylinders has also been investigated by Liaw and Chopra (1975), caused by surroundings is proposed in this section.
Sun and Nogami (1991), Park et al. (1991), Avilés and Li (2001),
and Wei et al. (2015). However, these simplified formulas (SIM)
were only used to compute the hydrodynamic forces on circular Mathematical Formulation
cylinders. The hydrodynamic forces on cylinders with other cross- Assuming water to be incompressible and inviscid, the small-
sectional shapes and the corresponding simplified formulas need to amplitude irrotational motion is governed by the Laplacian equa-
Downloaded from ascelibrary.org by University of Central Florida on 12/17/20. Copyright ASCE. For personal use only; all rights reserved.

be investigated in detail. tion in the Cartesian coordinate system, that is


Circular cylinders are also commonly used in deep-water
bridges However, there have been few reports on the study of ∂2p ∂2p ∂2p
þ þ ¼0 ð1Þ
earthquake-induced hydrodynamic pressure on round-ended and ∂x2 ∂y2 ∂z2
rectangular cylinders. Based on the fundamental frequency reduc-
tion rate of the bridge cylinders with and arbitrary cross section, in which pðx; y; z; tÞ = hydrodynamic pressure. Correspondingly,
Yang and Li (2013a) presented a simple added mass method for the scalar hydrodynamic pressure p should satisfy next boundary
water–structure interaction analysis of deep-water bridges and ob- conditions
tained the expressions of added mass of circular and square cylin- ∂p
ders. Moreover, the Morison equation was used to estimate the ¼0 z¼0 ð2Þ
∂z
hydrodynamic force on circular cylinders (Song et al. 2013;
Yang and Li 2013b), which ignored the effect of the structure p¼0 z¼h ð3Þ
on the flow of water. Overall, the hydrodynamic pressure on cir-
cular cylinders have been studied extensively, but the research ∂p
on round-ended and rectangular cylinders is still far from the ¼ −ρü cosðn; xÞ; r ¼ aðθÞ ð4Þ
∂n
aim of convenient application.
In this study, we developed an accurate and efficient numeri- where n = vector normal to the surface of the cylinder described by
cal method for calculating the earthquake-induced hydrodynamic aðθÞ; ü = relative acceleration of the cylinder; and ρ = density of
pressure on uniform vertical cylinders with an arbitrary cross water. The boundary condition in Eq. (4) means that the outward
section surrounded by water, where the cylinder is treated as a normal acceleration of the water particle in contact with the cylin-
one-dimensional structure. Based on the proposed model, the hy- der is equal to that of the structural surface.
drodynamic pressures on rectangular and round-ended cylinders Applying separation of the variables to the hydrodynamic pres-
are obtained and the numerical solutions are fitted as the form sure, the depth dependence of the problem can be factored out as
of a Bessel function. Further, the simplified formulas for the added
mass of the round-ended and rectangular cylinders are obtained by X

pðx; y; z; tÞ ¼ pj ðx; y; tÞZj ðzÞ ð5Þ
the curve-fitting method, and are accurate and concise enough for j¼1
engineering applications.
Substituting Eq. (5) into Eq. (1) yields an ordinary differential
equation and a two-dimensional Helmholtz equation as follows:
Numerical Solution for the Hydrodynamic Pressure
∂ 2 Zj
þ λ2j Zj ¼ 0 ð6Þ
A uniform hollow cylinder with an arbitrarily shaped cross section ∂z2
surrounded by water is shown in Fig. 1, which extends from the sea
∂ 2 pj ∂ 2 pj
þ − λ2j pj ¼ 0 ð7Þ
∂x2 ∂y2
superstructure
where λj = unknown constant. Considering the boundary condi-
water tions Eqs. (2) and (3), the solution of ZðzÞ can be expressed as
Zj ðzÞ ¼ cos λj z ð8Þ
y
where λj ¼ ð2j − 1Þπ=2h.
r Accordingly, inserting Eq. (5) into Eq. (4) and using the ortho-
Hs h gonality of Zj ðzÞ, the boundary condition at the interface of water
z x and cylinder can be rewritten as
∂pj
¼ −ρüj cosðn; xÞ; r ¼ aðθÞ ð9Þ
cross section ∂n
o x
where üj ¼ ð2=hÞ∫ h0 ü cos λj zdz.
Moreover, the Sommerfeld radiation condition expressed in
Fig. 1. Definition of the problem.
Eq. (10) should be taken into consideration, that is

© ASCE 04018137-2 J. Eng. Mech.

J. Eng. Mech., 2019, 145(2): 04018137


y ZZ   ZZ Z
r ∂ p~ j ∂pj ∂ p~ j ∂pj ∂pj
þ dV þ λ2j p~ j pj dV ¼ p~ j dL
∂x ∂x ∂y ∂y ∂n
ð14Þ

2 o x Applying the standard Galerkin finite-element discretization


(Zienkiewicz et al. 2013) to the fluid domain V, which is discre-
C tized into linear quadrilateral fluid elements, the weak form of the
weighted residual formulation of Helmholtz equation, including the
boundary condition Eq. (9), becomes
Fig. 2. Analysis model. ðK þ λ2j MÞpj ¼ Fj ð15Þ
and
ZZ
Downloaded from ascelibrary.org by University of Central Florida on 12/17/20. Copyright ASCE. For personal use only; all rights reserved.

  M¼ NT ðx; yÞNðx; yÞdV ð16aÞ


pffiffiffi ∂pj
lim r − iλj pj ¼ 0 ð10Þ
r→∞ ∂r
ZZ  T 
∂N ðx; yÞ ∂Nðx; yÞ ∂NT ðx; yÞ ∂Nðx; yÞ
Thus, the problem is reduced to solve a two-dimensional Helm- K¼ þ dV ð16bÞ
∂x ∂x ∂y ∂y
holtz equation. If the cylinder is not circular or elliptical, explicit
mathematical solutions of the Helmholtz equation cannot be ob- Z
tained in general, and it becomes necessary to employ a discretiza- Fj ¼ − NT ðx; yÞρüj cosðn; xÞdL ð16cÞ
tion procedure to evaluate the hydrodynamic pressure pj in Eq. (7).
As shown in Fig. 2, to model the fluid domain efficiently, the where pj = vector containing the remaining unknown nodal hydro-
whole fluid domain is divided into two subdomains: the bounded dynamic pressure pj ; N = global shape function vector; and the
domain Ω1 surrounding the cylinder with the outer cylindrical superscript T = matrix transpose.
boundary surface at finite distance Ra (ΓC ) from the origin, and
unbounded domain Ω2 outside of Ω1 . Here, the bounded domain
is modeled by the finite-element method described in “Finite- Artificial Boundary Condition for the Unbounded
Element Model for the Bounded Domain” and the unbounded Domain
domain is represented by the exact artificial boundary condition The Helmholtz equation Eq. (7) in the polar coordinate system can
developed in “Artificial Boundary Condition for the Unbounded be rewritten as
Domain”.
∂ 2 pj 1 ∂pj 1 ∂ 2 pj
þ þ − λ2j pj ¼ 0 ð17Þ
∂r2 r ∂r r2 ∂θ2
Finite-Element Model for the Bounded Domain
The weighted residual concept provides an equivalent integral Applying the method of variable separation, pj can be ex-
formulation of the Helmholtz Eq. (7). The concept defines the pressed as
bounded domain V as a pressure filed, for which the integral equa- pj ¼ RðrÞΘðθÞ ð18Þ
tion can be written as
ZZ  2  where ΘðθÞ and RðrÞ satisfy the following equations:
∂ pj ∂ 2 pj 2p
p~ j þ − λj j dV ¼ 0 ð11Þ
∂x2 ∂y2 Θ 0 0 þ n2 Θ ¼ 0 ð19Þ

in which p~ j is any weighting function satisfied for the d2 R dR


requirements. r20 2
þ r0 − ðr20 þ n2 ÞR ¼ 0 ð20Þ
dr0 dr 0
The weighted residual formulation Eq. (11) may be reformu-
lated as where r0 ¼ rλj . The solution to Eq. (19) is
ZZ   ZZ
∂ p~ j ∂pj ∂ p~ j ∂pj ΘðθÞ ¼ B1 cos nθ þ B2 sin nθn ¼ 0; 1; 2; : : : ð21Þ
þ dV þ λ2 p~ j pj dV
∂x ∂x ∂y ∂y
ZZ      The solution to Eq. (20), satisfying the radiation condition at
∂ ∂pj ∂ ∂pj
− p~ j þ p~ j dV ¼ 0 ð12Þ infinity Eq. (10), is
∂x ∂x ∂y ∂y
RðrÞ ¼ D1 K n ðλj rÞ ð22Þ
According to the divergence theorem (Spiegel et al. 2009), the
last integral term in Eq. (12) can be rewritten as where K n ð·Þ is the modified Bessel function of the second kind of
ZZ      Z order n; and B1 , B2 , and D1 = unknown constants.
∂ ∂pj ∂ ∂pj ∂pj Thus, pj can be written as
p~ j þ p~ j dV ¼ p~ j dL ð13Þ
∂x ∂x ∂y ∂y ∂n X

pj ¼ ai K n ðλj rÞT i ðθÞ ð23Þ
where L = closed boundary that encloses the fluid domain V; and i¼1
n = unit normal vector with positive orientation away from the fluid
domain V. fT i ðθÞg ¼ f1; sin θ; cos θ; : : : ; sin iθ; cos iθ; : : : g i ¼ 1; 2; 3; : : :
Substitution of Eq. (13) into Eq. (12) yields the weak form of the
weighted residual formulation of the Helmholtz equation ð24Þ

© ASCE 04018137-3 J. Eng. Mech.

J. Eng. Mech., 2019, 145(2): 04018137


where n ¼ ½i=2; and ½· denotes rounding toward zero direction. Verification
By utilizing the orthogonality of the modes T i ðθÞ, Eq. (23) can be
rewritten as To test and verify the effectiveness of the proposed method, the
problem of a circular cylinder interaction with water is taken as
X

pj ¼ Pj;i T i ðθÞ ð25Þ a test case. The resultant circumferential hydrodynamic force acting
i¼1 on the unit length of the circular cylinder can be formulated as
Z (Du et al. 2014)
ε 2π
Pj;i ¼ ai K n ðλj rÞ ¼ i pj T i ðθÞdθ ð26Þ X

2π 0
F¼ Fj cos λj z ð34aÞ
in which εi ¼ 1 in the case of i ¼ 1, otherwise εi ¼ 2. j¼1
Differentiating Eq. (25) with respect to the r coordinate is
K 1 ðλj aÞ
∂pj X ∞
∂Pj;i Fj ¼ ρπa2 üj ð34bÞ
Downloaded from ascelibrary.org by University of Central Florida on 12/17/20. Copyright ASCE. For personal use only; all rights reserved.

r ¼ T ðθÞ ð27Þ λj aK 10 ðλj aÞ


∂r i¼1
∂r i

∂Pj;i where a = radius of the circular cylinder.


¼ ai λj rK n0 ðλj rÞ ð28Þ Fig. 3 shows the comparison of Fj on a circular cylinder
∂r
between the proposed method and the analytical solution with
Substituting Eq. (26) into Eq. (28), the results lead to a modal üj ¼ 1. Fig. 4 shows the comparison of F on an elliptical cylinder
dynamic-stiffness relationship on the artificial boundary r ¼ Ra, between the proposed method and the analytical solution proposed
which can be expressed as by Wang et al. (2018), where a and b denote semimajor and semi-
minor axes. It can be seen that the proposed method has a good
∂Pj;i
¼ −Si ðλj Ra ÞPj;i ð29Þ agreement with analytical solution.
∂r
K n0 ðλj Ra Þ
Si ðλj Ra Þ ¼ −λj Ra ð30Þ 1.2
K n ðλj Ra Þ
Proposed method
Eqs. (26), (27), (29), and (30) are the exact artificial boundary 1.0 Analytical solution
conditions for the wave radiation problem in this study.
0.8
a)
2

Coupled Equation
0.6
Applying the standard Galerkin finite-element discretization to the
Fj / (

unbounded domain Ω2 , an impedance matrix is obtained. Then, the


0.4
artificial boundary is discretized into line elements and is modeled
by the finite-element method. The impedance matrix of the fluid
extending to infinity calculated by the finite-element method is then 0.2
added to the fluid satisfying the equilibrium and continuity condi-
tions. The resulting equation can be described as 0.0
     0 2 4 6 8 10
SI SIB pj;I FI ja

¼ ð31Þ
SBI SB þ SB pj;B FB
Fig. 3. Hydrodynamic force Fj on a circular cylinder caused by
where S ¼ ðK þ λ2j MÞ is the impedance matrix of the fluid; the surrounding water.
subscripts I and B = nonartificial boundary and artificial boundary
nodes of the finite-element mesh representing the fluid; and S∞ B
is the impedance matrix of the unbounded domain, which can be
formulated as 1.0 1.0
" #
X∞
ε
S∞
B ¼ W Ti TTi Si i
W ð32Þ 0.8 0.8
i¼1

where W is the shape function matrix; and Tl = cylinder vector of 0.6 0.6
b2
b2

the values of the ith mode on artificial boundary nodes.


F/
F/

The hydrodynamic pressure pj caused by outer water can be 0.4 0.4


obtained by solving finite-element Eq. (31). With the hydrody- a=15 b=10 h=40 a=30 b=10 h=40
namic pressure applied on the cylinder, the hydrodynamic force
0.2 0.2
acting on the unit length of the cylinder along the x-axes and y-axes Proposed Proposed
are then calculated by Analytical Analytical
Z 2π 0.0 0.0
Fjx ¼ − pj cosðn; xÞaðθÞdθ ð33aÞ 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
0 z/h z/h
Z 2π Fig. 4. Hydrodynamic forces F on elliptical cylinders caused by
Fjy ¼ − pj sinðn; xÞaðθÞdθ ð33bÞ surrounding water with ü ¼ 1.
0

© ASCE 04018137-4 J. Eng. Mech.

J. Eng. Mech., 2019, 145(2): 04018137


2 1.8

unbounded domain
1.6

Acceleration (m/s2)
1 Included
Neglected

a2uj)
1.4
0

Fj / (
1.2
-1
1.0

-2
0 10 20 30 40 0.8
t (s) 2 4 6 8
Downloaded from ascelibrary.org by University of Central Florida on 12/17/20. Copyright ASCE. For personal use only; all rights reserved.

Ra / a
Fig. 5. Input acceleration time history of ground motion.
Fig. 7. Hydrodynamic force Fj on a square cylinder against Ra with
2a ¼ 4 m and λj ¼ 0.2.
0.6
Proposed model
Total hydrodynamic force (MN)

0.4 Abaqus
m ¼ ρs A, where ρs is the mass density of the cylinder and A is the
0.2 cross-sectional area. The mass of the superstructure is Ms ¼ δmH s ,
where δ is the ratio of the mass of the superstructure to the mass of
0.0 the cylinder. It was assumed that the cylinder can only distort in the
-0.2 first N mode shapes of vibration in vacuum (Liaw and Chopra
1974). Therefore, the elastic displacement of the cylinder under
-0.4 earthquake action can be written as
-0.6 X
N
us ¼ ϕi ðzÞqi ðtÞ ð35Þ
-0.8 i¼1
0 10 20 30 40
t (s) where ϕi ðzÞ = mode shapes in vacuum; and qi ðtÞ = generalized
coordinates. In this case, the solution for ϕi ðzÞ is found in the fol-
Fig. 6. Total hydrodynamic force on a rigid square cylinder. lowing form (Clough and Penzien 2003):

ϕi ðzÞ ¼ sin χi z − sinh χi z


The hydrodynamic force on a vertical cylinder with an arbi- sinh χi H s þ sin χi H s
trarily shaped cross section surrounded by water also can be ob- þ ðcosh χi z − cos χi zÞ ð36Þ
cosh χi H s þ cos χi H s
tained by commercial programs, e.g., ADINA, ANSYS, Abaqus,
and others. However, the fluid in these commercial programs must where χi can be obtained by the following equation
be modeled by a three-dimensional finite-element model (FEM).
In addition, the proposed model is the substructure method, which χδðcos χH s sinh χHs − sin χH s cosh χH s þ 1
considers the structure–water system as composed of two substruc-
þ cos χH s cosh χHs ¼ 0 ð37Þ
tures, namely, the structure and the body of water. This technique
avoids direct analysis of the large water–structure interaction sys-
tem. Therefore, the computational cost of the proposed model Expression for Seismic Response
is much less than the three-dimensional finite-element model. A
square cylinder that has a width of 2a ¼ 4 m and a height of Hs ¼ Employing the superposition method of mode of vibration and
20 m was adopted to verify the proposed model. The water depth is using the generalized coordinate to represent the amplitude of
h ¼ 20 m. The acceleration of the ground motion is shown in Fig. 5 the deformation, the equation of motion for response of the cylinder
and the peak acceleration is 0.2 g in this study. Fig. 6 shows the to a horizontal ground motion is expressed as
comparison of the total hydrodynamic force between the proposed
M i q̈i þ Ci q̇i þ K i qi ¼ −M 0i üg þ fi ð38Þ
method and the numerical model in Abaqus. The computation time
in Abaqus is 810.1 s and it is only 13.4 s in the proposed model.
The generalized properties denoted in this equation are defined
Fig. 7 shows the hydrodynamic force Fj on a square cylinder
by the following expressions, where the natural vibration frequen-
with 2a ¼ 4 m and λj ¼ 0.2 against the radius of the fluid domain
cies and the damping ratio are denoted by ωi and ξ i :
Ra in the case of the unbounded domain considered and neglected.
It can be seen that the precision of the proposed model is still high Z H
s
when Ra is close to the size of the structure. Mi ¼ mϕi ðzÞ2 dz ð39aÞ
0

Z Hs
Seismic Response of Elastic Cylinder Surrounded
M 0i ¼ mϕi ðzÞdz ð39bÞ
by Water 0
The Euler-Bernoulli beam model was used for the elastic cylinder
in this study. The mass per unit length of the cylinder is given by K i ¼ ω2i M i ð39cÞ

© ASCE 04018137-5 J. Eng. Mech.

J. Eng. Mech., 2019, 145(2): 04018137


horizontal
Ci ¼ 2ξ i ωi M i ð39dÞ

χ4i EI

Vertical
ω2i ¼ ð39eÞ 2a2
m 2a1
Z h
fi ¼ ðFg þ Fs Þϕi ðzÞdz ð39fÞ
0 2b2
2b1
where E = elastic modulus; I = area moment of inertia; and Fg and
Fs = hydrodynamic force on the cylinder caused by rigid motion
(a) (b)
and elastic motion. Hydrodynamic forces Fg and Fs can be ex-
pressed as Fig. 8. Cross section: (a) round ended; and (b) rectangular.
Downloaded from ascelibrary.org by University of Central Florida on 12/17/20. Copyright ASCE. For personal use only; all rights reserved.

X

ð−1Þjþ1 S j
Fg ¼ 2ρüg cos λj z ð40aÞ
j¼1
λj h
16
Z

Top deflection / Forced amplitude


X

Sj h Experimental (Goto and Toki, 1965)
Fs ¼ 2ρ üs cos ðλj zÞ dz cos λj z ð40bÞ Proposed
j¼1
h 0 12
Damping constant 0.086
Z 2π
h=1m 2a=76mm
Sj ¼ − Pj ða; θÞa cos αdθ ð40cÞ 8
0

where ug = displacement of the ground motion; Sj is called the


dynamic-stiffness coefficient; and Pj ða; θÞ is computed by the 4
method proposed in “Numerical Solution for the Hydrodynamic
Pressure” with ρuj ¼ 1 and λj ¼ ð2j − 1Þπ=2h. Substituting
Eq. (40) into Eq. (39f), the results are expressed as 0
0.0 0.5 1.0 1.5 2.0
f i ¼ −üg M gi − q̈i ðtÞM si ð41aÞ Load frequency / Resonance frequency in water

X
∞ Z Fig. 9. Comparison between proposed method and experimental
ð−1Þjþ1 Sj h
M gi ¼ −2ρ ϕi ðzÞ cos λj zdz ð41bÞ values of amplitude at the top of a circular cylinder.
j¼1
λj h 0

X
∞ Z 2
Sj h
M si ¼ −2ρ ϕi ðzÞ cos λj zdz ð41cÞ When the ground motion has two components, dynamic re-
j¼1
h 0
sponses of the cylinder can be solved by following steps: obtaining
the acceleration of the ground motion along the horizontal and ver-
The complex frequency function for generalized modal dis- tical directions, respectively; computing the dynamic responses of
placement obtained from Eq. (38) is expressed as the cylinder along the horizontal and vertical directions, respec-
ðM0i þ Mgi Þω2 U g tively; and obtaining the resultant responses of the cylinder along
Qi ðωÞ ¼ − ð42Þ the direction of the ground motion.
M i ½1 − 2iζ i ðωi =ωÞ-ðωi =ω2 Þ þ M si Fig. 9 shows the comparison of the amplitude at the top of a
circular cylinder excited by the harmonic motion between the pro-
where Qi ðωÞ ¼ ∫ ∞ −∞ qi ðtÞe
−iωt
dt; U g ¼ ∫ ∞
−∞ ug e
−iωt
dt; and ω =
posed method and the experimental values (Goto and Toki 1965).
frequency of the seismic loading. Then, generalized modal dis-
It can be seen that the proposed method agrees well with experi-
placement qi ðtÞ can be obtained by
mental values.
Z
1 ∞
qi ðtÞ ¼ Q ðωÞeiωt dω ð43Þ
2π −∞ i Expression of Dynamic-Stiffness Coefficient
The expression of the dynamic-stiffness coefficient for a cir- In the case of ground motion along the horizontal direction,
cular cylinder can be obtained from the analytical solution given the dynamic-stiffness coefficient for round-ended cylinder Sj is
by Du et al. (2014). In “Expression of Dynamic-Stiffness Coeffi- fitted as
cient,” expressions of the dynamic-stiffness coefficient for round- −β 1 K 1 ðα1 r0 Þ
ended and rectangular cylinders are obtained by the curve-fitting Sj ¼ ρπa21 ð44Þ
method. In this study, the numerical solutions in Eq. (40c) were α1 r0 K 10 ðα1 r0 Þ
fitted as a mathematical formula by curve fitting tool in software where correction factors α1 and β 1 are obtained by curve fitting.
MATLAB. Fig. 8 shows the schematic diagram of the round-ended Correction factors α1 and β 1 at different ratios b1 =a1 are shown in
and rectangular cylinders. The cross-sectional properties of the Table 1. In the case of ground motion along the vertical direction,
round-ended cylinder are I x ¼ πa21 =4 þ ð4=3Þa31 ðb1 − a1 Þ, I y ¼ the dynamic-stiffness coefficient for round-ended cylinder Sj is
ð4=3Þa1 ðb1 − a1 Þ3 þ 2πa21 ½a21 =8 þ ðb1 − a1 Þ2 =2 þ 4a1 ðb1 − a1 Þ= fitted as
ð3πÞ, and A ¼ πa21 þ 4a1 ðb1 − a1 Þ. The cross-sectional properties
of the rectangular cylinder are I x ¼ ð4=3Þa32 b2 , I y ¼ ð4=3Þa2 b32 , −β 2 K 1 ðα2 r0 Þ
Sj ¼ ρπb21 ð45Þ
and A ¼ 4a2 b2 . α2 r0 K 10 ðα2 r0 Þ

© ASCE 04018137-6 J. Eng. Mech.

J. Eng. Mech., 2019, 145(2): 04018137


Table 1. Correction factors α1 , β 1 , α2 , and β 2 Table 2. Correction factors α3 and β 3
b1 =a1 α1 β1 α2 β2 a2 =b2 α3 β3
1 1 1 1 1 5 0.915 0.970
1.2 1.088 1.070 0.978 1.030 4 0.941 0.994
1.5 1.181 1.142 0.955 1.063 3 0.978 1.027
1.8 1.249 1.193 0.941 1.088 2 1.041 1.082
2 1.290 1.221 0.935 1.099 1 1.168 1.188
2.5 1.361 1.272 0.922 1.119 1/2 1.376 1.309
3 1.415 1.308 0.911 1.130 1/3 1.472 1.375
4 1.500 1.357 0.894 1.138 1/4 1.554 1.423
5 1.530 1.379 0.886 1.143 1/5 1.616 1.456
Downloaded from ascelibrary.org by University of Central Florida on 12/17/20. Copyright ASCE. For personal use only; all rights reserved.

where correction factors α2 and β 2 are obtained by curve fitting. 2.0


a2/b2=5.0 numerical
Correction factors α2 and β 2 at different ratios b1 =a1 are shown in
Table 1. Figs. 10(a and b) show the comparative results for the cyl- a2/b2=5.0 Eq. (46)
inder with b1 =a1 ¼ 1.5, 2, and 4 when ground motion propagates 1.5
a2/b2=1.0 numerical
along the horizontal and vertical directions, respectively. The com-
parison shows that the results of the present fitting formulas are in a2/b2=1.0 Eq. (46)
1.0

Sj
good agreement with the numerical results. a2/b2=0.2 numerical
In the case of ground motion along the horizontal direction, the
dynamic-stiffness coefficient for rectangular cylinder Sj is fitted as a2/b2=0.2 Eq. (46)
0.5
−β 3 K 1 ðα3 r0 Þ
Sj ¼ 4ρa22 ð46Þ
α3 r0 K 10 ðα3 r0 Þ
0.0
0 1 2 3 4
where correction factors α3 and β 3 at different ratios a2 =b2 are
r0
shown in Table 2. Fig. 11 shows the comparative results for the
cylinder with a2 =b2 ¼ 0.2, 1, and 5. The comparison shows that Fig. 11. Comparative results of dynamic-stiffness coefficient for the
the results of the present fitting formulas Eq. (46) are in good agree- rectangular cylinder.
ment with the numerical results.
According to the data in Tables 1 and 2, correction factors α1 ,
β 1 , α2 , β 2 , α3 , and β 3 can be further fitted as
Fig. 12 shows the comparison of Eq. (47) and fitting data in
α1 ¼ −0.92ðb1 =a1 Þ−0.547 þ 1.92 ð47aÞ Table 1. Fig. 13 shows the comparison of Eq. (48) and fitting data
in Table 2. It can be seen that the there is a good agreement between
β 1 ¼ −0.544ðb1 =a1 Þ−0.756 þ 1.544 ð47bÞ the fitting formulas and the fitting data.

α2 ¼ 0.162ðb1 =a1 Þ−0.735 þ 0.837 ð47cÞ


Added Mass of the Cylinder
β 2 ¼ −0.17ðb1 =a1 Þ−1.302 þ 1.167 ð47dÞ The hydrodynamic force acting on a rigid cylinder caused by
the surrounding water can be conveniently modeled as the product
α3 ¼ 0.76 × 10−1.932a2 =b2 þ 1.1 × 10−0.0378a2 =b2 ð48aÞ of an added mass of water and the acceleration of the structure.
According to Eq. (40), the added mass per unit of the rigid cylinder
β 3 ¼ 0.454 × 10−1.66a2 =b2 þ 1.132 × 10−0.0317a2 =b2 ð48bÞ can be expressed as

1.4 1.4
b1/a1=1.5 numerical b1/a1=1.5 numerical
1.2 1.2 b1/a1=1.5 Eq. (45)
b1/a1=1.5 Eq. (44)

1.0 b1/a1=2.0 numerical 1.0 b1/a1=2.0 numerical

b1/a1=2.0 Eq. (44) b1/a1=2.0 Eq. (45)


0.8 0.8
Sj
Sj

b1/a1=4.0 numerical b1/a1=4.0 numerical


0.6 b1/a1=4.0 Eq. (44) 0.6 b1/a1=4.0 Eq. (45)

0.4 0.4

0.2 0.2
0 1 2 3 4 0 1 2 3 4
(a) r0 (b) r0

Fig. 10. Comparative results of dynamic-stiffness coefficient for the round-ended cylinder: (a) horizontal; and (b) vertical directions.

© ASCE 04018137-7 J. Eng. Mech.

J. Eng. Mech., 2019, 145(2): 04018137


1.8 1.4
Data in Tab. 1 2 Data in Tab. 1
1

Eq. (47a) 2 Eq. (47c)


1
1.6
Data in Tab. 1 2 Data in Tab. 1
1 1.2

2
Eq. (47b) 2 Eq. (47d)
1
1 and

2 and
1.4

1.0
1.2

1.0 0.8
1 2 3 4 5 1 2 3 4 5
Downloaded from ascelibrary.org by University of Central Florida on 12/17/20. Copyright ASCE. For personal use only; all rights reserved.

(a) b1 / a1 (b) b1 / a1

Fig. 12. Comparison of correction factors from fitting formula Eq. (47) and Table 1: (a) α1 and β 1 ; and (b) α2 and β 2 .

1.8 1.0
3 Data in Tab. 2
l=0.2 mg
1.6 3 Eq. (48a) 0.8
l=0.2 m1
3 Data in Tab. 2 l=0.5 mg
1.4 0.6
3

Eq. (48b) l=0.5 m1


3

z/h
and

1.2 0.4
3

1.0 0.2

0.8 0.0
0 1 2 3 4 5 0.0 0.5 1.0 1.5 2.0 2.5 3.0
a2 / b2 Added mass / 4a 22

Fig. 13. Comparison of correction factors from fitting formula Eq. (48) Fig. 14. Comparison of added mass distributions for the square
and Table 2. cylinders.

X

ð−1Þjþ1 Sj cos λj z
mg ðzÞ ¼ 2ρ ð49aÞ 20
j¼1
λj h
15

The added mass per unit of the flexible cylinder for the ith mode 10
of vibration can be expressed as
(U-U0 )/U0 %

R 5
X∞
Sj cos λj z 0h ϕi ðzÞ cos λj zdz
mi ðzÞ ¼ 2ρ ð49bÞ 0
j¼1
ϕi ðzÞh l=0.2 mi
-5 l=0.2 mg
Fig. 14 shows the added mass computed from Eq. (49) for the l=0.5 mi
-10
rigid motion and first mode of a square cylinder, where the param-
l=0.5 mg
eters are E ¼ 30,000 MPa, ρs ¼ 2,500 kg=m3 , h ¼ H s ¼ 80 m, -15
and a2 ¼ b2 . It can be seen that the added mass caused by flexible 0 5 10 15 20
motion of the cylinder is different from the added mass caused by Frequency (Hz)
rigid motion. The concept of an added mass to represent the water–
structure interaction has been applied in many different situations, Fig. 15. Evaluation of the approximate added mass representation.
such as the earthquake effects on dams (Westergaard 1933). For
convenience of calculation and engineering application, the added
mass obtained by assuming the structure is rigid is used to represent
the influence of water in contact with a structure. Consequently, the displacement in Eq. (42) are presented in Fig. 15. It can be seen
the added mass defined by Eq. (49a) is not an exact representation that the errors between exact curves and approximate curves are not
of the hydrodynamic effects. The added mass is taken given by high. The same conclusion was also given by Liaw and Chopra
Eq. (49b) to compute the exact increase in the seismic response (1974). Therefore, the effects of surrounding water on dynamics
of the square cylinder, and by Eq. (49a) to obtain an approximate of cylinders can be approximately represented by an added mass
increase. The effects of the surrounding water on the amplitude of obtained by assuming the cylinder is rigid.

© ASCE 04018137-8 J. Eng. Mech.

J. Eng. Mech., 2019, 145(2): 04018137


1.0 1.0
l=2 1=1.5 FEM l=2 1=1.5 FEM
l=2 1=1.5 SIM l=2 1=1.5 SIM
0.8 0.8
l=0.5 1=5 FEM l=0.5 1=5 FEM
l=0.5 1=5 SIM l=0.5 1=5 SIM
0.6 0.6

z/h
z/h 0.4 0.4

0.2 0.2

0.0 0.0
0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0
Downloaded from ascelibrary.org by University of Central Florida on 12/17/20. Copyright ASCE. For personal use only; all rights reserved.

(a) cm (b) cm

Fig. 16. Comparison of simplified formulas for the added mass distribution of a round-ended cylinder: (a) horizontal; and (b) vertical directions.

Simplified Formula for the Added Mass of the Rigid 1.0


l=2 2=0.25 FEM
Cylinder
l=2 2=0.25 SIM
The expression of the added mass per unit of the cylinder is given 0.8
l=0.5 2=4 FEM
by Eq. (49) and the total added mass of the cylinder can be ex-
pressed as M ¼ ∫ h0 mg ðzÞdz. In fact, it is very troublesome to cal- l=0.5 2=4 SIM
0.6
culate the added mass per unit of the cylinder and the total added

z/h
mass because of the complexity of Eq. (49a). In order to provide 0.4
convenience for engineering applications, it is necessary to simplify
the complicated expression. In this section, the complicated expres-
sion of the added mass is simplified by curve fitting. 0.2
Three dimensionless parameters, namely, width-to-depth ratio
(l ¼ 2a0 =h), length-to-width ratio (δ 1 ¼ b1 =a1 ) for a round-ended 0.0
cylinder, and width-to-length ratio (δ 2 ¼ a2 =b2 ) for a rectangular 0.0 0.3 0.6 0.9 1.2
cm
cylinder, are introduced. The proposed simplified formulas in this
study are in the range of 0.2 ≤ l ≤ 2, 1 ≤ δ 1 ≤ 5, and 0.2 ≤ δ 2 ≤ 5. Fig. 17. Comparison of simplified formulas for the added mass
distribution of rectangular cylinder.
Added Mass per Unit of the Cylinder
According to the study, it is found that Eq. (49a) can be well
simplified as d2 ¼ q14 lq15 þ q16 ð53Þ

mg ðzÞ q14 ¼ 0.589δ 1−1.146 þ 0.587 ð54aÞ


¼ d1 ½1 − z=hed2 ðz=h−1Þ  ð50Þ
m0
q15 ¼ 0.02δ −1.524 − 1.256 ð54bÞ
in which m0 ¼ ρπa20 for a round-ended cylinder and m0 ¼ 4ρa20 for 1

a rectangular cylinder; 2a0 = width of the cylinder perpendicular to


the direction of ground motion; and d1 and d2 can be obtained by q16 ¼ −0.0218δ −2.015
1 þ 2.203 ð54cÞ
curve fitting. Then, d1 and d2 at different l and δ 1 or δ 2 can be
In the case of ground motion propagating along the vertical
further fitted as simple mathematical formulas by curve fitting.
direction, the coefficient d1 for a round-ended cylinder is simpli-
Figs. 16 and 17 present the comparison of the added mass per unit
fied as
of the cylinder obtained by the SIM and the presented FEM for a
round-ended cylinder and rectangular cylinder, respectively. It can d1 ¼ q21 l2 − 0.365l þ q22 ð55Þ
be seen that the simplified formulas agree well with the numerical
solutions.
q21 ¼ 0.01δ 1−2.263 þ 0.0357 ð56aÞ
In the case of ground motion propagating along horizontal
direction, the coefficient d1 for a round-ended cylinder is simpli-
fied as q22 ¼ −0.183δ −1.393
1 þ 1.231 ð56bÞ

d1 ¼ q11 l2 þ q12 l þ q13 ð51Þ and the coefficient d2 is simplified as

d2 ¼ q43 l−1.23 þ 2.17 ð57Þ


q11 ¼ −0.232δ 1−0.485 þ 0.278 ð52aÞ

q12 ¼ 0.729δ 1−0.6 − 1.096 ð52bÞ q43 ¼ −0.349δ −0.459


1 þ 1.558 ð58Þ

q13 ¼ −0.571δ 1−0.824 þ 1.631 ð52cÞ In the case of ground motion propagating along the hori-
zontal direction, the coefficient d1 for a rectangular cylinder is sim-
and the coefficient d2 is simplified as plified as

© ASCE 04018137-9 J. Eng. Mech.

J. Eng. Mech., 2019, 145(2): 04018137


d1 ¼ q31 lq32 þ q33 ð59Þ p22 ¼ 0.13δ 1−0.585 þ 0.662 ð66bÞ

q31 ¼ −0.35δ −1.092


2 − 0.198 ð60aÞ p23 ¼ −0.047δ −1.422
1 þ 0.277 ð66cÞ

q32 ¼ 0.766 × 100.0471δ2 − 0.782 × 10−1.401δ2 ð60bÞ For a rectangular cylinder, the total added mass in the case of
ground motion propagating along the horizontal direction can be
simplified as
q33 ¼ 0.412δ−1.012
2 þ 0.952 ð60cÞ
M
and the coefficient d2 is simplified as ¼ p31 l−p32 þ p33 ð67Þ
m0 h
d2 ¼ q34 lq35 þ q36 ð61Þ (
( −11.122 × 10−9.191δ2 − 1.438 × 10−0.544δ2 δ2 ≤ 1
−8.153δ2 0.0865δ 2 p31 ¼
1.859 × 10 þ 0.938 × 10 δ2 ≤ 1 −0.779 × 10 −0.743δ 2
− 0.475 × 10 −0.0151δ 2
δ2 > 1
Downloaded from ascelibrary.org by University of Central Florida on 12/17/20. Copyright ASCE. For personal use only; all rights reserved.

q34 ¼ ð62aÞ
−0.0157δ 22 þ 0.194δ 2 þ 0.846 δ2 > 1 ð68aÞ
( (
0.296δ 2−0.548 − 1.521 δ2 ≤ 1 −0.469δ 22 þ 0.888δ 2 − 0.044 δ2 ≤ 1
q35 ¼ ð62bÞ p32 ¼ ð68bÞ
−1.26 δ2 > 1 −0.384 × 10−0.4δ2 þ 0.644 × 10−0.017δ2 δ2 > 1
( (
−0.0282δ−1.647
2 þ 2.121 δ2 ≤ 1 11.05 × 10−9.117δ2 þ 2.053 × 10−0.326δ2 δ2 ≤ 1
q36 ¼ ð62cÞ p33 ¼ ð68cÞ
−0.0095δ 2 −0.711δ2
0.004δ 22 − 0.0533δ 2 þ 2.143 δ2 > 1 1.09 × 10 þ 0.826 × 10 δ2 > 1

The error between simplified formulas (M SIM ) and numerical


Total Added Mass of the Cylinder solutions (M FEM ) is defined as
For a round-ended cylinder, the total added mass in the case of jM FEM − M SIM j
error ¼ × 100% ð69Þ
ground motion propagating along the horizontal direction can be M FEM
fitted as
Figs. 18 and 19 show the error of the simplified formulas for
M round-ended and rectangular cylinders, respectively. It can be seen
¼ p11 × 10−p12 l þ p13 ð63Þ
m0 h that the error of the present simplified formulas is less than 5%,
which satisfies the engineering precision demand.
p11 ¼ −0.459δ 1−0.749 þ 1.224 ð64aÞ

p12 ¼ −0.496δ 1−0.696 þ 1.292 ð64bÞ Validation


Seismic responses of an elastic cylinder were used to verify the
p13 ¼ −0.0592δ 1−1.221 þ 0.293 ð64cÞ proposed simplified formulas for the added mass of the round-
ended and rectangular cylinders surrounded by water. The exact
and the total added mass in the case of ground motion propagating added mass was obtained by the FEM in “Finite-Element Model
along the vertical direction can be fitted as for the Bounded Domain.” Cylinder parameters including density,
M Young’s modulus, damping ratio, height, and water depth were
¼ p21 × 10−q22 l þ q23 ð65Þ 2,500 kg=m3 , 30,000 MPa, 0.05, 80 m, and 70 m, respectively. The
m0 h
mass of the superstructure was equal to two times the cylinder’s
mass. The acceleration of the ground motion is shown in Fig. 5
p21 ¼ −0.128δ 1−1.566 þ 0.885 ð66aÞ and the peak acceleration was 0.2 g in this study.

1.2 0.6
1=1.5 1=1.5
1.0 1=2.0 1=2.0

0.8 1=3.0 0.4 1=3.0

1=5.0 1=5.0
error

error

0.6

0.4 0.2

0.2

0.0 0.0
0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0
(a) 2a1/ h (b) 2b1/ h

Fig. 18. Error of the simplified formulas for the added mass distributions of a round-ended cylinder: (a) horizontal; and (b) vertical directions.

© ASCE 04018137-10 J. Eng. Mech.

J. Eng. Mech., 2019, 145(2): 04018137


8 1.0 1.0
2=0.2
a2=5m a2=20m
b2=20m b2=10m
2=0.5 0.8 0.8
6
2=2

2=5.0
0.6 0.6
error

z / Hs

z / Hs
0.4 0.4
2
In water In water
0.2 0.2
In vacuum In vacuum
0
0.0 0.5 1.0 1.5 2.0
Downloaded from ascelibrary.org by University of Central Florida on 12/17/20. Copyright ASCE. For personal use only; all rights reserved.

0.0 0.0
2a3/ h 0 50 100 150 0.0 0.1 0.2 0.3 0.4 0.5
(a) displacement (mm) (b) displacement (mm)
Fig. 19. Error of the simplified formulas of the added mass distribu-
tions of rectangular cylinder. Fig. 22. Peak displacement along the height of the rectangular cylinder
in the case of ground motion along the horizontal direction:
(a) a2 ¼ 5 m, b2 ¼ 20 m; and (b) a1 ¼ 20 m, b1 ¼ 10 m.

1.0 1.0
a1=10m a1=10m 0.4 0.2
b1=5m Exact Exact
b1=8m
0.8 0.8 Simplified
Simplified
0.2 0.1

displacement (m)
displacement (m)
0.6 0.6
z / Hs

z / Hs

0.4 0.4 0.0 0.0

In water In water
0.2 0.2 In vacuum -0.2 -0.1
In vacuum
a1=10m a1=10m
0.0 0.0 b1=5m b1=8m
0.0 0.1 0.2 0.3 0.00 0.05 0.10 0.15 0.20 -0.4 -0.2
(a) displacement (m) (b) displacement (m) 0 10 20 30 40 0 10 20 30 40
(a) t (s) (b) t (s)
Fig. 20. Peak displacement along the height of the round-ended
cylinder in the case of ground motion along the horizontal direction: Fig. 23. Relative displacement time histories on top of the round-ended
(a) a1 ¼ 10 m, b1 ¼ 5 m; and (b) a1 ¼ 10 m, b1 ¼ 8 m. cylinder in the case of ground motion along the horizontal direction:
(a) a1 ¼ 10 m, b1 ¼ 5 m; and (b) a1 ¼ 10 m, b1 ¼ 8 m.

In the case of ground motion propagating along the horizontal


1.0 1.0 direction, two cases were conducted with a1 ¼ 10 m, b1 ¼ 5 m,
a1=10m a1=10m and a1 ¼ 10 m, b1 ¼ 8 m. In the case of ground motion propagat-
b1=10m b1=30m ing along the vertical direction, two cases were conducted with
0.8 0.8
a1 ¼ 10 m, b1 ¼ 10 m, and a1 ¼ 10 m, b1 ¼ 30 m. In the case of
ground motion propagating along the horizontal direction, two
0.6 0.6 cases were conducted with a2 ¼ 5 m, b2 ¼ 20 m, and a2 ¼ 20 m,
b2 ¼ 10 m. Figs. 20–22 show the maximum displacement along
z / Hs
z / Hs

the height in the case of surrounding water being considered


0.4 0.4 and neglected. The results indicate that the hydrodynamic forces
can significantly increase the seismic responses of the cylinder.
0.2 In water 0.2 In water Figs. 23–25 present the time histories of the displacement on top
In vacuum In vacuum of the cylinder with the added mass computed by the exact method
and simplified method. It can be seen that the there is a good agree-
0.0 0.0 ment between the exact method and proposed simplified formulas.
0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6
(a) displacement (mm) (b) displacement (mm)

Fig. 21. Peak displacement along the height of the round-ended Conclusions
cylinder in the case of ground motion along the vertical direction:
A numerical method was presented in this study to evaluate the hy-
(a) a1 ¼ 10 m, b1 ¼ 10 m; and (b) a1 ¼ 10 m, b1 ¼ 30 m.
drodynamic pressure on a uniform vertical cylinder with an arbitrary

© ASCE 04018137-11 J. Eng. Mech.

J. Eng. Mech., 2019, 145(2): 04018137


0.8 0.8
Exact Exact References
Simplified Simplified
Avilés, J., and X. Li. 2001. “Hydrodynamic pressures on axisymmetric
0.4 0.4 offshore structures considering seabed flexibility.” Comput. Struct.
displacement (m)

displacement (m)
79 (29–30): 2595–2606. https://doi.org/10.1016/S0045-7949(01)
00125-0.
0.0 0.0 Chen, B. F. 1997. “3D nonlinear hydrodynamic analysis of vertical cylinder
during earthquake. I: Rigid motion.” J. Eng. Mech. 123 (5): 458–465.
https://doi.org/10.1061/(ASCE)0733-9399(1997)123:5(458).
-0.4 -0.4 Clough, R. W., and J. Penzien. 2003. Dynamics of structures. 3rd ed.
a1=10m a1=10m Berkeley, CA: Computer & Structures, Inc.
b1=30m Du, X., P. Wang, and M. Zhao. 2014. “Simplified formula of hydrodynamic
b1=10m
-0.8 -0.8 pressure on circular bridge piers in the time domain.” Ocean Eng.
0 10 20 30 40 0 10 20 30 40 85: 44–53. https://doi.org/10.1016/j.oceaneng.2014.04.031.
Downloaded from ascelibrary.org by University of Central Florida on 12/17/20. Copyright ASCE. For personal use only; all rights reserved.

(a) t (s) (b) t (s) Goto, H., and K. Toki. 1965. “Vibrational characteristics and a seismic
design of submerged bridge piers.” In Proc., 3rd World Conf. in Earth-
Fig. 24. Relative displacement time histories on top of the round-ended quake Engineering, 107–122. Wellington, New Zealand: New Zealand
cylinder in the case of ground motion along the vertical direction: National Committee on Earthquake Engineering.
(a) a1 ¼ 10 m, b1 ¼ 10 m; and (b) a1 ¼ 10 m, b1 ¼ 30 m. Han, Q., X. Du, J. Liu, Z. Li, L. Li, and J. Zhao. 2009. “Seismic damage of
highway bridges during the 2008 Wenchuan earthquake.” Earthquake
Eng. Eng. Vib. 8 (2): 263–273. https://doi.org/10.1007/s11803-009
-8162-0.
Han, R. P. S., and H. Xu. 1996. “A simple and accurate added
0.15 0.6 mass model for hydrodynamic fluid-structure interaction analysis.”
Exact Exact J. Franklin Inst. 333 (6): 929–945. https://doi.org/10.1016/0016
0.10 Simplified 0.4 Simplified -0032(96)00043-9.
Jacobsen, L. S. 1949. “Impulsive hydrodynamics of fluid inside a cylindri-
displacement (m)
displacement (m)

0.05 0.2 cal tank and of fluid surrounding a cylindrical pier.” Bull. Seismol. Soc.
Am. 39 (3): 189–204.
0.00 0.0 Jiang, H., B. Wang, X. Bai, C. Zeng, and H. Zhang. 2017. “Simplified ex-
pression of hydrodynamic pressure on deep water cylindrical bridge
-0.05 -0.2 piers during earthquakes.” Bridge Eng. 22 (6): 04017014. https://doi
.org/10.1061/(ASCE)BE.1943-5592.0001032.
-0.10 a2=5m -0.4 a 2=20m Li, Q., and W. L. Yang. 2013. “An improved method of hydrodynamic pres-
b2=20m b 2=10m sure calculation for circular hollow piers in deep water under earth-
-0.15 -0.6 quake.” Ocean Eng. 72: 241–256. https://doi.org/10.1016/j.oceaneng
0 10 20 30 40 0 10 20 30 40 .2013.07.001.
(a) t (s) (b) t (s) Liaw, C. Y., and A. K. Chopra. 1974. “Dynamics of towers surrounded by
water.” Earthquake Eng. Struct. Dyn. 3 (1): 33–49. https://doi.org/10
Fig. 25. Relative displacement time histories on top of the rectangular .1002/eqe.4290030104.
cylinder in the case of ground motion along the horizontal direction: Liaw, C. Y., and A. K. Chopra. 1975. “Earthquake analysis of axisymmetric
(a) a2 ¼ 5 m, b2 ¼ 20 m; and (b) a1 ¼ 20 m, b1 ¼ 10 m. towers partially submerged in water.” Earthquake Eng. Struct. Dyn.
3 (3): 233–248. https://doi.org/10.1002/eqe.4290030303.
Luke, C., and C. Suren. 2016. “Seismic fragility performance of skewed
and curved bridges in low-to-moderate seismic region.” Earthquake
Struct. 10 (4): 789–810. https://doi.org/10.12989/eas.2016.10.4.789.
cross section, which was validated by a circular cylinder. Based on
Park, W. S., C. B. Yun, and C. K. Pyun. 1991. “Infinite elements for evalu-
the present method, the hydrodynamic forces on round-ended and
ation of hydrodynamic forces on offshore structures.” Comput. Struct.
rectangular cylinders caused by surrounded water were obtained. 40 (4): 837–847. https://doi.org/10.1016/0045-7949(91)90313-B.
The numerical solutions of the hydrodynamic forces that were mod- Song, B., F. Zheng, and Y. Li. 2013. “Study on a simplified calculation
eled as the product of an added mass of water and the acceleration of method for hydrodynamic pressure to slender structures under earth-
the cylinder were expressed as the form of Bessel functions by the quakes.” J. Earthquake Eng. 17 (5): 720–735. https://doi.org/10
curve-fitting method. However, these added mass expressions are so .1080/13632469.2013.771592.
complicated that their values are difficult to calculate. For conven- Spiegel, M. R., S. Lipschutz, and D. Spellman. 2009. Schaum’s outlines
ience of engineering application, the simplified formulas for the vector analysis. 2nd ed. New York: McGraw-Hill.
added mass of round-ended and rectangular cylinders were obtained Sun, K., and T. Nogami. 1991. “Earthquake induced hydrodynamic pres-
by the curve-fitting method. Numerical examples indicate that the sure on axisymmetric offshore structures.” Earthquake Eng. Struct.
proposed simplified formulas agree well with the exact method and Dyn. 20 (5): 429–440. https://doi.org/10.1002/eqe.4290200504.
their accuracy is enough for engineering applications. Tanaka, Y., and R. T. Hudspeth. 1988. “Restoring forces on vertical circular
cylinders forced by earthquakes.” Earthquake Eng. Struct. Dyn. 16 (1):
99–119. https://doi.org/10.1002/eqe.4290160108.
Wang, P., M. Zhao, and X. Du. 2018. “Analytical solution and simplified
Acknowledgments formula for earthquake induced hydrodynamic pressure on elliptical
hollow cylinders in water.” Ocean Eng. 148: 149–160. https://doi.org
This work is supported by the National Natural Science Foundation /10.1016/j.oceaneng.2017.11.019.
of China (51708010, 51678015, and 51421005). The support is Wei, K., N. Bouaanani, and W. C. Yuan. 2015. “Simplified methods for
gratefully acknowledged. The results and conclusions presented efficient seismic design and analysis of water-surrounded composite
are of the authors and do not necessarily reflect the view of the axisymmetric structures.” Ocean Eng. 104: 617–638. https://doi.org/10
sponsors. .1016/j.oceaneng.2015.05.001.

© ASCE 04018137-12 J. Eng. Mech.

J. Eng. Mech., 2019, 145(2): 04018137


Wei, K., W. C. Yuan, and N. Bouaanani. 2013. “Experimental and Eng. Struct. 70: 158–167. https://doi.org/10.1016/j.engstruct.2014.03
numerical assessment of the three dimensional modal dynamic response .039.
of bridge pile foundations submerged in water.” J. Bridge Eng. 18 (10): Yang, W., and Q. Li. 2013a. “A new added mass method for fluid-structure
1032–1041. https://doi.org/10.1061/(ASCE)BE.1943-5592.0000442. interaction analysis of deep-water bridge.” KSCE J. Civ. Eng. 17 (6):
Westergaard, H. M. 1933. “Water pressures on dams during earthquakes.” 1413–1424. https://doi.org/10.1007/s12205-013-0134-2.
Trans. ASCE 98 (2): 418–433. Yang, W., and Q. Li. 2013b. “The expanded Morison equation con-
Williams, A. N. 1986. “Earthquake response of submerged circular sidering inner and outer water hydrodynamic pressure of hollow
cylinder.” Ocean Eng. 13 (6): 569–585. https://doi.org/10.1016/0029 piers.” Ocean Eng. 69: 79–87. https://doi.org/10.1016/j.oceaneng.2013
-8018(86)90040-5. .05.008.
Wilson, T., H. Mahmoud, and S. Chen. 2014. “Seismic performance of Zienkiewicz, O. C., R. L. Taylor, and J. Z. Zhu. 2013. The finite element
skewed and curved reinforced concrete bridges in mountainous states.” method: Its basis and fundamentals. 7th ed. London: Elsevier.
Downloaded from ascelibrary.org by University of Central Florida on 12/17/20. Copyright ASCE. For personal use only; all rights reserved.

© ASCE 04018137-13 J. Eng. Mech.

J. Eng. Mech., 2019, 145(2): 04018137

You might also like