You are on page 1of 22

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/260814173

Vulnerability assessment of container cranes under stochastic wind loading

Article  in  Structure and Infrastructure Engineering · June 2013


DOI: 10.1080/15732479.2013.834943

CITATIONS READS

16 1,297

2 authors, including:

Sourav Gur
Indian Institute of Technology (IIT) Patna
41 PUBLICATIONS   527 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

seismic sustainability of critical facilities View project

TiNi-based materials: Understanding and applications View project

All content following this page was uploaded by Sourav Gur on 02 April 2017.

The user has requested enhancement of the downloaded file.


Structure and Infrastructure Engineering
Maintenance, Management, Life-Cycle Design and Performance

ISSN: 1573-2479 (Print) 1744-8980 (Online) Journal homepage: http://www.tandfonline.com/loi/nsie20

Vulnerability assessment of container cranes


under stochastic wind loading

Sourav Gur & Samit Ray-Chaudhuri

To cite this article: Sourav Gur & Samit Ray-Chaudhuri (2014) Vulnerability assessment of
container cranes under stochastic wind loading, Structure and Infrastructure Engineering,
10:12, 1511-1530, DOI: 10.1080/15732479.2013.834943

To link to this article: http://dx.doi.org/10.1080/15732479.2013.834943

Published online: 30 Oct 2013.

Submit your article to this journal

Article views: 132

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=nsie20

Download by: [University of Arizona] Date: 02 February 2017, At: 08:33


Structure and Infrastructure Engineering, 2014
Vol. 10, No. 12, 1511–1530, http://dx.doi.org/10.1080/15732479.2013.834943

Vulnerability assessment of container cranes under stochastic wind loading


Sourav Gur1 and Samit Ray-Chaudhuri*
Department of Civil Engineering, Indian Institute of Technology Kanpur, Kanpur 208016, UP, India
(Received 29 January 2013; final version received 27 April 2013; accepted 1 June 2013; published online 30 October 2013)

Container cranes, an essential component in any seaport facility system, are highly susceptible to failure during natural
disasters such as severe windstorms. Damage or collapse of these structures can lead to significant amount of economic loss
in terms of both repair or replacement and downtime. To study how various parameters such as boom position, yaw angle,
gradient height and gradient wind speed affect the failure vulnerability of a crane, in this work, a representative container
crane has been modelled in a commercial software considering both material and geometric nonlinearities. Stochastic wind
fields are simulated as a stationary process by means of the spectral representation method using power spectrum in
conjunction with along- and across-wind coherence functions. Wind fields are also simulated as a non-stationary stochastic
process utilising the modulation functions from the available literature to consider the situations of thunderstorm
downbursts. Nonlinear time history analyses are then performed and a few responses are compared for stationary and non-
stationary wind field cases. It is observed that for the given simulation parameters, the responses considering stationary wind
fields are larger than those of the non-stationary wind fields. Failure vulnerability of the crane is then assessed in terms of
fragility curves. The results of this study show that the failure vulnerability may not be maximum for the set of parameter
values for which the aerodynamic forces are maximum. This study provides a better understanding of crane vulnerability
that may be utilised to achieve a better crane design.
Keywords: container crane; stochastic wind field; non-stationary wind field; nonlinear analysis; vulnerability assessment;
fragility curves

1. Introduction dockside cranes under wind-induced loading is not


Seaports are essential nodes of national and international uncommon. For examples, in 1996, the Zhanjiang port
transport systems of a country and often serve as regional of Guangdong was severely damaged by a typhoon, where
economic centres. In addition, when inland transportation 16 cranes were destroyed, and in 2003, at Pusan port of
systems become out of operation due to natural or South Korea, 8 cranes designed according to a modern
manmade disasters, ports play a major role by serving as code (BS 2573) were severely damaged due to a typhoon
local transportation lifeline systems for emergency and resulted in $31 million loss. There are several crane
transportation and rescue operations. Because of these failure incidents at the US east and gulf coasts (Zhang,
reasons, it is essential to minimise damage of ports under Sun, & Peng, 2009).
natural disasters such as severe windstorms. In a port Recent damage surveys show that the damage of
facility, container cranes are one of the most important container cranes due to severe wind storms is localised in
classes of structures that facilitate the movement of cargos most cases. Many cranes are found to fail during
between a ship and storage yards. Damage or collapse of windstorms due to failure at their bases and anchoring
such structures can hinder smooth port operation, resulting systems such as tie-down and stowage-pin even at a wind
in significant economic loss in terms of repair or speed much less than the design wind speed of non-
replacement, downtime and enduring loss of business operating conditions. It is also observed that usually one
due to permanent traffic rerouting. damage incident initiates cascading failure events invol-
Although significant developments in terms of design ving many nearby cranes and resulting in severe damage
and construction of modern-day huge cranes have been or even collapse of multiple cranes (Jordan, 1995;
achieved, recent crane failure incidents indicate that these McCarthy, Jordan, Lee, & Werner, 2007; McCarthy,
cranes are highly susceptible to natural hazards. Failure of Soderberg, & Dix, 2009; McCarthy & Vazifdar, 2004;
cranes due to seismic loading has gained significant Morris & Hoite, 1997). Furthermore, it was concluded that
attention in recent years, while the literature on wind- container cranes are highly vulnerable to sudden onset
induced failure is scant. However, failure of modern-day high-speed wind. In addition, to reduce failure of container

*Corresponding author. Email: samitrc@iitk.ac.in


q 2013 Taylor & Francis
1512 S. Gur and S. Ray-Chaudhuri

cranes during wind storm, different types of retrofitting 908. It is also found that the values of Cd and C l are more
schemes were suggested and even a modified design for a boom-up position than those of a boom-down
philosophy was proposed for the design of these cranes position.
(see McCarthy et al., 2007; McCarthy et al., 2009; To simplify the design procedure, most of the modern
McCarthy & Vazifdar, 2004). codes such as British Standard, BS 2573-1 (1983),
Several researchers such as Eden, Iny, and Butler American Society of Civil Engineers, ASCE 7-05 (2006)
(1981), Han and Han (2010), Huang, Wang, and Gu and European Standard, Eurocode 1-4 (2004) recommend
(2006), Lee and Kang (2007, 2008), Lee, Kim, Han, and performing static analysis using a representative model of
Han (2007), Lee, Shim, Han, Han, and Lee (2007), and the container crane that is to be designed. For this purpose,
Scarabino, Di Leo, Delnero, and Bacchi (2005) conducted different code suggestions are to use either the mean wind
numerical and experimental studies to determine values of field or the gust factor method to compute the wind
different governing parameters for the design of container loading. However, for a tall and unsymmetric structure
crane due to wind loading. Eden et al. (1981) studied the such as a container crane, these methods may not provide
effect of oncoming wind field on a container crane accurate estimation of responses. Recently, a few
responses by developing numerical models of cranes. researchers have carried out dynamic time history analyses
Results show that a 3-s gust factor gives the maximum to determine the influence of different wind field
value of design wind speed. It is also found that the out-of- parameters and structural configurations on response of
service gust wind speed is more conservative than in- container cranes (see Chen & Kareem, 2001; Kim,
service gust wind speed. Huang et al. (2006) carried out Shinozuka, & Ghang, 2004; Sun, Xu, & Ko, 1999).
both experimental and numerical studies to investigate the Although significant research efforts have been
effect of wind load on container cranes. It was observed directed to study the effect of wind loading on container
that the influence of Reynolds number (Re) at smooth flow cranes, as per authors’ knowledge, no systematic study has
is very less and the most unfavourable wind loading angle been performed so far to investigate the effect of different
is found to be between 158 and 308. In another numerical parameters on the failure vulnerability of a container
study, Zrnic, Oguamanam, and Bosnjak (2006) investi- crane. The main purpose of this study is thus to investigate
gated the dynamic response of boom girder of a mega the influence of structural configurations (i.e. boom
quayside container crane and observed that the deflection positions) and wind field properties (i.e. gradient height,
of the end point of boom girder increases with an increase gradient wind speed and yaw angle) on the failure
in wind velocity. Han and Han (2010) performed wind vulnerability of a representative container crane by
tunnel tests (1:200 scale container crane model) and simulating spatially varying stochastic wind fields with a
numerically analysed a crane to propose an alarm system set of simulation parameters.
using the uplift force as reference data. Scarabino et al.
(2005) conducted an experimental study on a 1:20 scale
port crane boom girder section and reported that the values 2. Stochastic wind field simulation
of drag coefficient (C d ) are more in a smooth flow Wind loads calculated according to different code
condition than those of a turbulent flow condition and the provisions are static in nature, and thus, a static analysis
influence of Reynolds number (Re) on drag coefficient is is usually carried out for design of a structure under wind
negligible. Lee, Kim, et al. (2007) and Lee, Shim, et al. load. However, a very high level of temporal wind speed
(2007) studied the effect of machinery house and boom fluctuation and spatial variation in the wind field can occur
positions on the structural stability of a container crane by during wind storms. Such effects can only be taken into
performing experiments on a 1:200 scale model of a account by performing a dynamic analysis. To incorporate
container crane. The results show that the effect of position the dynamic effect of wind loading, several design codes
of machinery house on the drag and lift coefficients are recommended an equivalent static wind load analysis by
almost negligible and the maximum value of uplifting means of the gust factor method. But, in the gust factor
force at any support occurs either for yaw angles between method, only the magnitude of mean wind load is
208 and 408 or between 1308 and 1508. increased by a constant factor, which may not be able to
Recently, two wind tunnel studies (Lee & Kang, 2007, capture effectively the influence of spatially varying wind
2008) were performed on a 1:150 scale container crane field on the response of structures such as container cranes.
model located in two types of atmospheric boundary layers Thus, to investigate the effect of spatially varying wind
(open coast and open terrain). Effect of Re, yaw angle and field on failure of container crane, wind fields are
boom position of container crane (boom-down and boom- simulated in this study.
up) on aerodynamic coefficients were studied. From this Wind fields can be simulated as stationary or non-
study, it was observed that the drag (C d ) and lift (C l ) stationary processes. For simulation of wind fields as a
coefficients become maximum at angles 408 and 1408, stationary process, wind speed at any point and any instant
respectively, and both become minimum at an angle of of time Vðx; y; z; tÞ is decomposed into two parts: (i) mean
Structure and Infrastructure Engineering 1513

wind speed V m ðzÞ, which is independent of time but varies (Blevins, 1990; Simiu & Scanlan, 1986):
along the height only; and (ii) fluctuating wind speed (also  a
called as mean-deducted) vðx; y; z; tÞ, which varies both in z
V m ðzÞ ¼ V g ; ð3Þ
space and time (see Caracoglia & Jones, 2003; Paola, Zg
1998). Thus, Vðx; y; z; tÞ can be expressed as follows:
where V m ðzÞ is the mean wind speed at a height z, V g is the
Vðx; y; z; tÞ ¼ V m ðzÞ þ vðx; y; z; tÞ: ð1Þ gradient wind velocity at gradient height Z g (the height at
which wind velocity becomes constant) and a is a power
The fluctuating part of the wind fields is simulated here law exponent that depends on terrain conditions. In this
by means of the spectral representation method (see study, a is taken as 1/7 considering an open terrain
Deodatis, 1996; Gurley & Kareem, 1998; Han & Luo, condition. For a systematic study, Z g is assumed to vary
2006; Karmakar, Ray-Chaudhuri, & Shinozuka, 2012; from 7.5 to 20.0 m with an increment of 2.5 m and,
Paola, 1998; Popescu, Deodatis, & Prevost, 1998; for each case of Z g , V g is assumed to vary from 5.0 to
Shinozuka, 1971; Shinozuka & Deodatis, 1991, 1996; 75.0 m/s. It may be noted that the chosen ranges of Z g and
Shinozuka & Jan, 1972; Yamazaki & Shinozuka, 1988). In V g cover a wide range of wind conditions.
this method, a one-dimensional multivariate (1D-mV)
fluctuating wind field surrounding the structure is
simulated. A brief description of simulation procedure 2.1.2 Yaw angle (u)
for stationary stochastic wind fields along with the Yaw angle or the wind incident angle is the angle made by
important parameters is provided in the following the direction of wind flow with respect to the axis of a
subsections. structure. Generally, it is observed that with the change in
Recent studies have highlighted the necessity of yaw angle, the response of a structure changes. In this
considering non-stationarity of wind fields especially study, the yaw angle is measured with respect to the x-axis
during the thunderstorm downbursts (see Chen & of the structure (as shown in Figure 1) and is assumed to
Letchford, 2004, 2007; Holmes & Oliver, 2000). In such vary from 08 to 1808 at an interval of 308.
cases, the mean wind speed is usually modelled as a
deterministic time-varying function whereas the fluctuat-
ing part is modelled as a evolutionary random process. 2.1.3 Shear velocity of wind field (u* )
Thus, during such events, the total wind fields can be
Shear velocity, also called the friction velocity, is the
expressed as follows:
velocity by which shear stress is expressed in the unit of
velocity. It is a useful parameter in wind engineering to
Vðx; y; z; tÞ ¼ V m ðz; tÞ þ v~ ðx; y; z; tÞ ð2aÞ
compare, for example, true velocity of a flow in a stream
with a velocity that relates shear stress between any two
with layers of flow. A shear velocity is used to describe shear-
related motion in a moving wind field. It can be expressed
V m ðz; tÞ ¼ V m ðzÞf ðtÞ; ð2bÞ as follows (see Ambrosini, Riera, & Danesi, 2002; Hansen
& Krenk, 1999; Lumley & Panofsky, 1964; Zhang, Li, &
where f ðtÞ is the time-modulating function of mean wind Peng, 2008):
speed and v~ ðx; y; z; tÞ is the non-stationary fluctuating part,
which can be simulated using evolutionary power spectral kV m ðzÞ
density function (see Chen, 2008; Liang, Ray-Chaudhuri, u* ¼ ; ð4Þ
lnðz=z0 Þ
& Shinozuka, 2007).
where u* is the shear velocity in m/s; k is the Von Karman’s
constant, usually taken as 0.4, V m ðzÞ is the mean wind
2.1 Simulation parameters speed at height z in m/s, and z0 is the roughness height,
Following parameters are considered to simulate the typically taken as 0.01 for smooth surface.
stationary stochastic wind field:

2.1.4 Power spectrum


2.1.1 Mean wind speed Kaimal, Wyngaard, Izumi, and Coté (1972) proposed a
It is considered that the wind is flowing horizontally with power spectral density function for the horizontal wind
the wind velocity being zero at the ground level and velocity fluctuations, which is suitable for both low and
varying only with the height according to the well-known high frequency spectrum area. This spectrum can be
power law, which can be expressed in the following form expressed as (see Ambrosini et al., 2002; Hansen & Krenk,
1514 S. Gur and S. Ray-Chaudhuri

Figure 1. Model of a container crane at boom-down position showing various nodes.

1999; Harris, 1990; Kaimal et al., 1972; Zhang et al., is ½V m ðzj Þ þ V m ðzk Þ=2 in m/s; Cx , Cy and C z are constants
2008) that may be taken as 6.0, 10.0 and 16.0, respectively.

1 1 z 1
Su ðv; zÞ ¼ 200 u2*   ;
2 2p V m ðzÞ 1 þ 50ðjvjz=2pV m ðzÞÞ 5=3 2.2 Stationary wind field simulation
ð5Þ For the simulation of mean wind speed, at first, the original
coordinate system ðx; y; zÞ is transformed to the wind flow
where Su ðv; zÞ is the Kaimal and Simiu power spectrum, z direction ðx1 ; y1 ; z1 Þ coordinate system. As it is assumed
is the height of the point in m where the wind velocity is that the wind flow is horizontal, to obtain the wind speed
required, v is the frequency in rad/s, V m ðzÞ is in m/s and u* along the direction of the structure, yaw angle (u, in
is in m/s. In this study, this spectrum has been used to degrees) is used for coordinate transformation (Figure 2).
simulate the wind field. This transformation of structural coordinate system into
the wind flow direction can be given as

2.1.5 Along- and across-wind coherence functions 8 9 2 38 9


>
> x1 >
> cosðuÞ cosð90 2 uÞ cosð90Þ >>x>>
If j and k are two points along and across the direction of >
< = 6> < >
7> =
wind flow, then the along- and across-wind coherence 6
y1 ¼ 6 cosð90 þ uÞ cosðuÞ 7
cosð90Þ 7 y :
>
> >
> 4 5>
> >
between these two points are expressed as (see Ambrosini : z1 >
> ; > >>
cosð90Þ cosð90Þ cosð0Þ : z ;
et al., 2002; Han & Luo, 2006; Hansen & Krenk, 1999;
Harris, 1990; Morfiadakis, Glinou, & Koulovari, 1996; ð7Þ
Zhang et al., 2008)
 
jvjC x dx
hjk ðv; dxÞ ¼ exp 2 ; ð6aÞ
2pV mjk ðzÞ
2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi3
jvj ðC y dyÞ2 þ ðC z dzÞ2
gjk ðv; dy; dzÞ ¼ exp42 5; ð6bÞ
2pV mjk ðzÞ

where hjk ðv; dxÞ and gjk ðv; dy; dzÞ are the along- and across-
wind coherence functions, respectively; dx, dy and dz are Figure 2. Coordinate transformation from structural system to
jxj 2 xk j, jyj 2 yk j and jzj 2 zk j, respectively (in m); V mjk ðzÞ wind flow direction.
Structure and Infrastructure Engineering 1515

Now, all the nodal coordinates of the structure are H jk ðvÞ ¼ H *jk ð2vÞ
transformed in the direction of wind flow and the absolute ð11cÞ
value of difference in coordinates between any two nodes j j ¼ 1; 2; . . . ; m; k ¼ 1; 2; . . . ; m 2 1; j . k;
and k is calculated. Now, considering a value of gradient
height Z g and gradient wind velocity V g, the mean wind
speed V m ðzÞ at any node with height z of the structure is  !
calculated by using Equation (3). 21 Im H jk ðvÞ
ujk ðvÞ ¼ tan   : ð11dÞ
The fluctuating part of the wind speed is considered as Re H jk ðvÞ
an 1D-mV, zero mean, stationary Gaussian stochastic field
with components vo1 ðtÞ, vo2 ðtÞ, vo3 ðtÞ, . . . , vom ðtÞ. It may be Finally, using the spectral representation method based
noted that a few recent studies have considered the wind on the fast Fourier transform (FFT) technique (Brigham,
field as a non-Gaussian process. However, in this study for 1988), the fluctuating part of the wind speed can be
the simplicity of simulation, it is assumed that the simulated as
fluctuating (mean deducted) part of the wind field may be (  )
considered as a Gaussian process. Thus, for this fluctuating ðiÞ
Xj MX 21
2p
part, the spectral density matrix S o ðvÞ can be expressed as vj ðpDtÞ ¼ Re Bjql exp ilp ; ð12Þ
q¼1 l¼0
M
(Popescu et al., 1998)
2 3 j ¼ 1; 2; . . . ; m p ¼ 0; 1; . . . ; M 2 1
o
S11 ðvÞ o
S12 ðvÞ ··· o
S1m ðvÞ
6 o 7 l ¼ 0; 1; . . . ; N 2 1;
6 S21 ðvÞ o
S22 ðvÞ ··· o
S2m ðvÞ 7
6 7
S ðvÞ ¼ 6
o
6 .. .. .. .. 7 7; ð8Þ where
6 . . . . 7 pffiffiffiffiffiffiffi
4 5
o
Sm1 ðvÞ o
Sm2 ðvÞ · · · o
Smm ðvÞ Bjql ¼ 2jH jq ðlDvÞj Dvexp½2iujq ðlDvÞexp½ifðiÞ
ql ; ð13aÞ

vu
Dv ¼ ; ð13bÞ
where N
2p
Sjjo ðvÞ ¼ Sj ðvÞ j ¼ 1; 2; . . . ; m; ð9aÞ Dt ¼ ; M ¼ 2N; ð13cÞ
MDv
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Sjko ðvÞ ¼ Sj ðvÞSk ðvÞ hjk gjk fql is the random phase angle, which is uniformly
ð9bÞ distributed over 0 to 2p. vu is the upper cut-off frequency
j; k ¼ 1; 2; . . . ; m; j – k assumed based on previous knowledge or using engineer-
ing judgements.
with Sj ðvÞ and Sk ðvÞ being the Kaimal and Simiu power
spectral density functions at jth and kth points,
respectively. 2.3 Non-stationary wind field simulation
Now, using the Cholesky’s method, the power spectral Spatially varying non-stationary wind fields can be
density matrix S o ðvÞ is decomposed into a lower triangular simulated by slightly modifying the procedure for the
matrix HðvÞ as follows (Popescu et al., 1998): simulation of stationary fields by utilising the evolutionary
power spectral density functions that define the non-
S o ðvÞ ¼ HðvÞH *T ðvÞ; ð10Þ stationary nature of the fields (Chen, 2008; Liang et al.,
2007; Spanos & Kougioumtzoglou, 2012). Thus, for the
where simulation of non-stationary fluctuating parts of wind
2 3 fields, Equations (9a) and (9b) can be modified as follows:
H 11 ðvÞ 0 0 ··· 0
6 7 o 2
6 H 21 ðvÞ H 22 ðvÞ 0 ··· 0 7 Sjj ðv; tÞ ¼ jAj ðt; vÞj Sj ðvÞ j ¼ 1; 2; . . . ; m; ð14aÞ
6 7
6 7 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
6 7
HðvÞ ¼ 6 H 31 ðvÞ H 32 ðvÞ H 33 ðvÞ ··· 0 o
7; Sjk ðv; tÞ ¼ jAj ðt; vÞkAk ðt; vÞj Sj ðvÞSk ðvÞ hjk gjk
6 .. .. .. .. .. 7
6 7 j; k ¼ 1; 2; . . . ; m; j – k;
6 . . . . . 7
4 5
H m1 ðvÞ H m2 ðvÞ H m3 ðvÞ · · · H mm ðvÞ ð14bÞ

ð11aÞ where Aj ðt; vÞ is generally a complex-valued deterministic


modulating function of both t and v for j ¼ 1; 2; . . . ; m.
Thus, once Aj ðt; vÞ is known, the fluctuating non-
H jj ðvÞ ¼ H jj ð2vÞ j ¼ 1; 2; . . . ; m; ð11bÞ stationary part of the wind field can be generated using
1516 S. Gur and S. Ray-Chaudhuri

the spectral representation method based on FFT (Liang noticeably due to the incorporation of the time-varying
et al., 2007). mean speed modulation factor. It can also be observed
During thunderstorm downbursts, the time-indepen- from this figure that the amplitude of the fluctuating part
dent mean wind speed is usually modelled by multiplying increases with the increase in the mean wind speed, as has
the mean speed with a time-varying modulating function been reported in the previous literature.
f ðtÞ. Therefore, to simulate a non-stationary wind field,
Aj ðt; vÞ and f ðtÞ are required in addition to those required
for the simulation of stationary wind field. Usually, 2.5 Wind force computation
mathematical models derived from measured wind fields Only the drag and lift forces at each node are considered in
are used to approximate Aj ðt; vÞ and f ðtÞ. this study. The drag force F D and the lift force F L at any
node due to wind field are calculated as (Simiu & Scanlan,
1986)
2.4 Simulation results
At first, simulation results are presented considering the 1  2
wind field as a stationary process. To simulate the F D ðx; y; z; tÞ ¼ r Vðx; y; z; tÞ C D A; ð15aÞ
2
fluctuating part of the wind speed, the values of different
parameters are chosen as: N ¼ 2048, M ¼ 4096 and 1  2
F L ðx; y; z; tÞ ¼ r Vðx; y; z; tÞ CL A; ð15bÞ
vu ¼ 4p. Realisations of wind speed time histories are 2
simulated for 1023.75 s with an interval of 0.25 s for all
nodes of the structural model. Figure 3 shows a realisation where r is the density of air taken as 1.225 kg/m3; C D and
of simulated wind speed data for Z g ¼ 20.0 m and CL are the drag and lift coefficients, respectively; and A
V g ¼ 75.0 m/s and for nodes 8, 34, 36, 54, 61 and 66 is the effective area. The effective area of any node is
(see Figure 1). It may be noted that nodes 34, 36, 54 and 61 given as
are located at the same height from the ground with nodes
A ¼ jAx cosðuÞj þ jAy cosðuÞj; ð16Þ
54 and 61 being in the same line along the direction of
wind flow. However, the time histories are different at all
where Ax is the tributary area perpendicular to x-axis and
these nodes. This clearly shows the spatially varying
Ay is the tributary area perpendicular to y-axis.
nature of the wind field. Also, by comparing the wind
speed time histories of nodes 8, 54 and 66, one can notice
the variation of wind speed along with height. Note that the
3. Numerical modelling and analysis
mean velocities at nodes 54 and 66 are same as the gradient
height that is taken as 30 m, which is lower than the height 3.1 Numerical modelling
of these two nodes. However, the mean velocity at node 8 In this study, a container crane of tower height 72.60 m and
is lower than that of the nodes 54 and 66 due to the power a total boom length of 131.0 m (back reach length 32.0 m,
law variation. rail span 29.0 m and outreach length 70.0 m) has been
To check the accuracy of the simulated field, the cross considered. The lifting capacity of the crane is 65.0 tones.
power spectral density function of wind speed between any From the aforementioned properties, the representative
two nodes is calculated from the simulated wind speed data. crane is classified as Super-Post-Panamax crane. The
Then considering 20 realisations, their average values are crane is modelled (67 nodes, connected by 109 elements)
also determined for different node pairs. Figure 4 shows in a widely used commercial software known as SAP 2000
the comparison of generated power spectral density (Computer and Structures Inc., 2009). The legs of the
functions (a sample and average computed from 20 models are assumed to be fixed at their bases. The reason
realisations) with the assumed one (i.e. Equations (9a) and behind this assumption is the fact that a crane is usually
(9b)). Only significant data are plotted in the graphs, i.e. anchored at their legs during a wind storm. Isotropic
data within the range of 0.1 –1.0 Hz are shown. It can be tendon-type steel [unit weight (g): 76; 972:86 N=m3 ,
observed from this figure that, in mean sense, the power modulus of elasticity (E): 1:999 £ 1011 N=m2 , Poisson’s
spectral density functions generated from the wind speed ratio ðy Þ: 0.30, shear modulus (G): 7:690 £ 1010 N=m2 and
data are in very good agreement with the assumed one. minimum yield stress ðF y Þ: 1:493 £ 109 N=m2 ] with 3%
For the simulation of non-stationary wind fields, the kinematic hardening is considered for material modelling.
modulating functions for the mean and fluctuating parts as Note that the numerical model of the crane used in this
mentioned earlier are adopted from Chen and Letchford study represents the design model of existing cranes with
(2004). Figure 5 shows a realisation of non-stationary similar size. To study the effect of boom positions, two
wind field (Figure 5(a) for node 8 and Figure 5(b) for node different boom positions (boom-down and boom-up) are
34). It can be observed from this figure that unlike the considered. For the boom-down position, the entire boom
stationary fields, the mean wind speed fluctuates girder is kept horizontal, whereas for the boom-up position
Structure and Infrastructure Engineering 1517

Figure 3. Wind speed time-histories for stationary wind field at nodes (a) 8, (b) 34, (c) 36, (d) 54, (e) 61 and (f) 66 (see Figure 1).
1518 S. Gur and S. Ray-Chaudhuri

Figure 4. Comparison of assumed and generated power spectral density functions between nodes: (a) 34 and 34, (b) 34 and 43, (c) 54
and 61 and (d) 61 and 66.

Figure 5. Wind speed time histories for non-stationary wind field at nodes (a) 8 and (b) 34 (see Figure 1).
Structure and Infrastructure Engineering 1519

Figure 6. Representative container crane model at two different boom positions: (a) boom-down and (b) boom-up.

only the outreach portion of boom girder is kept vertical. mode shape. Also, the second mode shape for the boom-
Figure 6 shows the representative crane models at two down position corresponds to the lateral motion of the
different boom positions. backreach while, for the boom-up position, the second
mode represents the lateral motion of the outreach portion
of the boom girder. In general, the lower modes
3.2 Eigenvalue and static analyses correspond to the vibration of boom.
An analysis is performed to estimate the natural For a container crane, various sources of static loads
frequencies and mode shapes of the models. The Ritz are present in addition to the self-weight of the structure.
vector method is used to extract the first few modes of the These loads are due to the machine house, trolley and
model for both boom positions. Table 1 provides the first hoisting, staircase and dummy mass. These loads are
six natural periods of the models. It can be observed from applied at each nodes of the crane model and an analysis is
these results that the periods in all modes of boom-up performed to obtain the state of stress and deformation of
model are higher than the corresponding boom-down the model. It is well known that, when large stresses are
position. This implies that the boom-up position is more present within a structure or the deformation/rotation are
flexible than the boom-down one. It is worth mentioning large, the equilibrium equations differ significantly from
that the fundamental period of these models as obtained the undeformed configuration to the deformed geometry of
here is comparable with those reported in the published the structure. In addition, due to geometric nonlinearity, a
literature (Soderberg, Hsieh, & Dix, 2009). tensile force in a member tends to resist the rotation of
Figures 7 and 8 show the first two mode shapes at elements and stiffen the structure, whereas a compressive
boom-down and boom-up positions, respectively. From force in a member tends to enhance the rotation of
these figures, it can be observed that the first mode shape elements (see Chen & Kareem, 2001; Kim et al., 2004;
corresponds to predominantly the lateral vibration of the Sun et al., 1999). Thus, for the static analysis, geometric
outreach portion of the boom girder at boom-down nonlinearities (i.e. P 2 D effect and large deformation) are
position. However, at boom-up position, the lateral considered. It has been observed during the static analysis
vibration of the entire structure corresponds to the first that the consideration of geometric nonlinearity requires a
moderate amount of iteration highlighting the presence of
geometric nonlinear behaviour.
Table 1. First six natural periods for different boom positions.

Boom-down Boom-up
Mode number position period (s) position period (s) 3.3 Wind load analysis
1 3.14162 3.66817 Wind load analysis is performed after the static load
2 2.38287 2.58208 analysis and holding the states of deformation and stress in
3 1.30613 2.11273 all members. Wind load time histories are applied at each
4 1.06751 1.50134 node in both horizontal and vertical directions to simulate
5 0.80840 0.99998
6 0.71979 0.73545 the drag and lift forces. For computation of wind forces,
the values of ðC D Þ and ðCL Þ are taken from Lee and Kang
1520 S. Gur and S. Ray-Chaudhuri

Figure 7. Mode shapes of the first two modes at boom-down condition: (a) first mode and (b) second mode.

(2007) and provided in Table 2. Note that these values Note that the analysis was restricted to buffeting forces
were obtained from wind tunnel testing of similar crane alone and the aerodynamic forces due to the self-excitation
structures and were defined with respect to total drag and phenomenon could not be considered due to software
lift forces of the crane model, which were obtained from limitations.
the base balance measurements. Due to unavailability of Different force and displacement responses are
these force coefficients for each member, these values recorded from the time history analysis. At first, results
have been used for each member. In fact, many articles are presented considering wind fields as stationary random
have followed a similar approach (Ambrosini et al., 2002; process. Figure 9 shows a sample of different response
Boonyapinyo, Yamada, & Miyata, 1994; Bourdeix, time histories when the analysis is performed for a yaw
Hemon, & Santi, 1998; Harikrishna, Shanmugasundaram, angle of 908 and considering Z g ¼ 20.0 m and
Gomathinayagam, & Lakshmanan, 1999; Huang et al., V g ¼ 75.0 m/s. The response time histories are displace-
2006; Karmakar et al., 2012; Scanlan, 1993). ment along y-axis of the top node (node 66), displacement
For each boom position and yaw angle, sample of node 7 along y-axis and moment rotation curve of
realisations of nodal wind force time histories are obtained bottom plastic hinge of member 7 about x- and y-axis.
by using the data generated from wind field simulations. Wind speed time history for the top node is also shown in
For each boom position, 2% Rayleigh damping is this figure. It may be observed from Figure 9(b),(c) that
considered for the first two modes. A nonlinear time displacements fluctuate about non-zero mean values
history analysis is then carried out using the direct without the change of sign. The moments at the base,
integration scheme. For nonlinear analysis, Newmark’s however, show a change in sign implying high-magnitude
average acceleration (g ¼ 0.5 and b ¼ 0.25) method with stress reversal (Figure 9(d),(e)) about both x- and y-axis.
a time interval of 0.25 s is used to estimate the response. This is because of the kinematic material hardening
considered for modelling these elements. Once the
member crosses the plastic moment limit and incurs in
the nonlinear range, a large permanent deformation
occurs. Reversals of rotation then produce negative
moments in this member.
Figure 10 shows how the responses differ when non-
stationary fields are considered instead of stationary fields.
Two response parameters, top drift (displacement of node
66) and base moment along with top node wind speed is
plotted in this figure considering stationary and non-
stationary wind fields. A horizontal line is shown in
Figure 10(a) to denote the mean wind speed. Response
time histories of the crane are obtained for the boom-down
position at a yaw angle of 608. One can observe from
Figure 10(b) and (c) that the displacement and moment
Figure 8. Mode shapes of the first two modes at boom-up responses match well for both stationary and non-
condition: (a) first mode and (b) second mode. stationary cases near the zone where the non-stationary
Structure and Infrastructure Engineering 1521

Table 2. Drag and lift coefficients used for different boom positions (Lee & Kang, 2007).

Yaw angle (u, in degrees) Boom-down drag coeff. Boom-down lift coeff. Boom-up drag coeff. Boom-up lift coeff.
0 0.611 2 0.111 0.889 2 0.167
30 0.944 2 0.194 1.139 2 0.167
60 1.111 2 0.167 1.153 2 0.194
90 0.833 2 0.167 0.853 2 0.167
120 1.083 2 0.131 1.193 2 0.111
150 1.000 2 0.167 1.167 2 0.028
180 0.722 2 0.111 0.889 2 0.028

Figure 9. (a) Wind speed at node 66 along y-axis (see Figure 1), (b) displacement of node 66 along y-axis, (c) displacement of node 8
along y-axis, (d) moment– rotation curve of bottom plastic hinge of member 7 about x-axis and (e) moment– rotation curve of member 7
about y-axis.
1522 S. Gur and S. Ray-Chaudhuri

Figure 10. Comparison of stationary and non-stationary wind field cases: (a) wind speed at node 66, (b) displacement of node 66 and
(c) base moment.

responses attain the peak value. This is because the periods mean and mean plus standard deviation cases. To get an
of the first few modes of vibration of the structure, which idea of dispersion of responses, see Figure 12. One can
dominate the responses, are very small compared with the observe from this figure that the coefficient of variation
dominant period of the mean wind speed modulating (COV) for non-stationary wind field case is always lower
function. It can be noted that the peak displacement and than that of the stationary wind field case for both boom-
base moment responses reduce by 22% and 27%, down and boom-up positions. It can also be observed that
respectively, due to the consideration of non-stationary the COV increases with gradient wind speed following
wind fields. almost a linear trend. In addition, the COV of boom-up
Figure 11 represents the variation of peak response position is found to be higher than that of the boom-down
parameters (i.e. normalised base moment, top drift (node position for a given gradient wind speed. These results
66) and rotational ductility of member 7) with gradient imply that a higher COV is associated with a higher
wind speed considering stationary and non-stationary wind response with the COV value being , 13% up to a gradient
fields. Results are obtained for boom-down position of the wind speed of 65 m/s.
crane at a yaw angle of 608 and presented in terms of mean
and mean plus standard deviation responses considering
20 realisations. Results indicate that each of these 4. Vulnerability assessment
responses considering stationary wind fields is always To determine the effect of different parameters on the
higher than that of the non-stationary wind fields for both failure probability of the container crane, a fragility
Structure and Infrastructure Engineering 1523

Figure 11. Mean and mean plus standard deviation responses in boom-down position considering stationary and non-stationary wind
fields: (a) base moment, (b) top drift (node 66) and (c) support rotational ductility of member 7.

analysis is performed by defining the failure of a crane in bridges, buildings and other structures (such as highway
terms of damage states and then generating fragility networks) have also been assessed (Jeong & Elnashai,
curves. A fragility curve provides the probability of 2011; Kappos & Panagopoulos, 2010; Sung & Su, 2011;
exceeding a particular damage state (which is usually Torbol & Shinozuka, 2012; Zhong, Gardoni, & Rosowsky,
defined in terms of an engineering damage parameter such 2012; Zhou, Banerjee, & Shinozuka, 2010). Studies have
as drift level and force level) when an event with a also been carried out on seismic fragility of container cranes
specified intensity has occurred. Generally, fragility and other port structures (Ray-Chaudhuri, Karmakar, Na, &
curves are expressed by two-parameter (median and log- Shinozuka, 2009).
standard deviation) log-normal distribution functions. In For generation of seismic fragility curves, peak ground
the past, fragility curves have been developed for regular acceleration of ground motion is typically used as the
structures such as buildings and bridges subjected to intensity measure. In this study, the gradient wind velocity
seismically induced loads (e.g. Pan, Agrawal, & Ghosn, ðV g Þ is considered as the intensity measure for estimating
2007; Pan, Agrawal, Ghosn, & Alampalii, 2010; the failure probability of a container crane under wind
Shinozuka, Feng, Lee, & Naganuma, 2000). Recently, loading. A brief description of the procedure to generate
probabilistic and time-dependent fragilities of concrete fragility curves is given in the following section.
1524 S. Gur and S. Ray-Chaudhuri

Figure 12. COV for peak base moment considering stationary and non-stationary wind fields: (a) boom-down and (b) boom-up
positions.

4.1 Damage state of these factors from an operational point of view. However,
Wind storms generally sustain for a very long duration in the absence of such limit states and for simplicity, in
imposing long-duration fluctuating wind load. This this study, the vulnerability assessment is carried out
fluctuating wind load may produce significant amount of considering the limiting value of base moments only.
stress along with stress reversal (see Figure 9) at different
elements of the structure resulting in low-cycle fatigue
4.2 Monte Carlo simulation
failure. In fact, past experiences have demonstrated that
the sections connected to the supports are prone to failure A Monte Carlo simulation is carried out to obtain the
due to large moment and axial force acting on them. To response statistics. To study the convergence of the
define a failure of container crane, in this study, the response parameter (in this case, normalised base moment),
required damage state is considered as the formation of a a convergence test is performed where mean and standard
plastic hinge at any of the legs connected to the supports. deviation values of the base moments are calculated. The
This assumption implicitly considers that the safety mean and standard deviation of the normalised base
mechanisms are designed well and the failure of the moments for different sample sizes at boom-down and
structure is not a result of failure of a safety link. boom-up positions are shown in Figure 13. It can be
Now, let us consider that the plastic moment-carrying observed from this figure that the response of the crane
capacity of any leg connected to a support as M P . Then, the converges to a particular value for a sample size of about 70
ratio R of the absolute value of the developed moment (M) (shown as a dotted vertical line) or higher for both boom
to the plastic moment (M P ) of the section can be used to positions. So, for further studies, for each set of gradient
define failure of the container crane. In the mathematical height and gradient wind speed, 70 different wind fields are
form, this can be expressed as simulated and used to perform nonlinear dynamic analyses
for obtaining response statistics.
jMj
MP ¼ FY Z P and R ¼ ; ð17Þ
MP
4.3 Estimation of fragility parameters
where Z P is the plastic section modulus and F Y is the yield The median and log-standard deviation of fragility curves
stress. Thus, the failure criterion is defined as follows: the for the considered damage state are estimated with the aid
section (and thus the structure) is assumed to have failed, if of the maximum likelihood method as proposed by
the value of this ratio (R) becomes greater than or equal to Shinozuka et al. (2000). To avoid crossing of fragility
unity for any loading case. Note that apart from the base curves for increasing intensity, a common log-standard
moments, there may be other factors such as permanent deviation is used as suggested by Shinozuka et al. (2000).
deformation of the structure and damage in other At first, for a given V g , Z g and u, the mean of maximum
components that may render the crane to be non- base moment response is estimated using a sample size of
operational. Damage states may thus be required in terms 70. Furthermore, these mean values are used to obtain
Structure and Infrastructure Engineering 1525

Figure 13. Mean and standard deviation of the normalised base moment versus sample size for (a) boom-down and (b) boom-up
positions.

damage states. Finally, the fragility parameters are deviation ðsV g Þ indicates the dispersion of the failure
estimated. gradient wind speeds.

5. Results and discussion 5.1 Effect of gradient height


The fragility parameters (i.e. the median and log-standard Figure 14(a),(b) provides the median value of failure
deviation of failure gradient wind speed) are estimated for gradient height for different yaw angles and boom
different parameters such as yaw angle, gradient height positions. From these figures, it can be observed that,
and boom positions. The median value of fragility curve irrespective of yaw angle, for both boom positions (boom-
ðmV g Þ gives the gradient wind speed value for which the down and boom-up), the median value of the failure
failure probability is 50%, whereas the log-standard gradient wind speed increases almost linearly with the

Figure 14. Median value of failure gradient wind speed with respect to gradient height for (a) boom-down and (b) boom-up positions.
1526 S. Gur and S. Ray-Chaudhuri

Figure 15. Ratio of log-standard deviation to median failure gradient wind speed with respect to gradient height for (a) boom-down and
(b) boom-up positions.

increase in gradient height. Such trend has been observed log-standard deviation (as mentioned in Section 3) of the
because, at a lower value of gradient height, the mean wind failure gradient wind speed for boom-down and boom-up
speed profile becomes very stiff, i.e. the mean value of positions considering damage corresponding to all six
wind speed reaches very high value at a comparatively gradient heights. It can be observed from Table 3 that for all
lower height causing higher wind load on the container yaw angles, the log-standard deviation values are slightly
crane. It can also be observed that the yaw angle of 608 is larger in the boom-down position than the boom-up
most critical for both boom positions. Furthermore, the position. Overall, the log-standard deviation value remains
boom-up position is more vulnerable than that of the below 20 m/s. Moreover, it may be noted that for both boom
boom-down position. In fact, for u ¼ 608, mV g is about positions, the log-standard deviation is maximum at a yaw
27 m/s at boom-up position for Z g ¼ 10 m, whereas at the angle of 608 and minimum at a yaw angle of 08. Thus, in
same value of Z g , mV g is about 50 m/s for the boom-down general, it can be concluded that the dispersion of fragility
position. curves will be larger for lower gradient height than for
Figure 15(a),(b) provides the ratio of log-standard higher gradient height as the median failure wind speed is
deviation to the median of failure gradient wind speed (i.e. larger for higher gradient height.
normalised log-standard deviation) for both boom
positions. It can be observed that the value of this ratio
is high for low-gradient wind speed. This is because, with 5.2 Effect of yaw angle
the increase in gradient height, the median value of failure Figure 16(a),(b) shows the variation of failure gradient
gradient wind speed increases while the log-standard wind speed with respect to the yaw angle for boom-down
deviation of the failure gradient wind speed remains and boom-up positions, respectively. From these figures, it
constant (see Table 3). Table 3 provides the common is observed that for a particular gradient height, at boom-
down position, the failure gradient wind speed value
Table 3. Log-standard deviation of failure gradient wind speed increases with the yaw angle in the sequence 608, 1208,
for boom-down and boom-up positions. 908, 308, 1508, 1808 and 08. However, at boom-up position,
Log-standard Log-standard the failure gradient wind speed value increases with the
Yaw angle deviation boom-down deviation boom-up yaw angle in the sequence 608, 1208, 08, 1808, 908, 308 and
u (degree) position (m/s) position (m/s) 1508. Note that with the change in yaw angle, the spatially
varying wind field changes imposing varying load at
0 11.560 10.491
30 15.422 11.097 different nodes and exciting different modes of vibration
60 19.932 17.652 of the crane. However, it is also observed that at both
90 16.539 12.503 boom positions, the yaw angle of 608 is most vulnerable.
120 17.895 16.229 Studying the log-standard deviation of the gradient wind
150 14.309 10.610 speed (not shown here), a similar trend as in gradient
180 12.799 11.693
height is also observed for different yaw angles, i.e. a high
Structure and Infrastructure Engineering 1527

Figure 16. Median value of failure gradient wind speed with respect to yaw angle for (a) boom-down and (b) boom-up positions.

normalised log-standard deviation is observed for a low more vulnerable than the boom-down position. For a 10-m
median value of V g . gradient height, it can be observed that at boom-up
position, the median value of failure gradient wind speed is
about 46% less than that of the boom-down position at a
5.3 Effect of boom position yaw angle of 608 and 25% less at a yaw angle of 1208.
Figure 17(a),(b) provides the comparison of median failure
gradient wind speeds between the boom-down and boom-
up positions for different gradient heights and yaw angles. 5.4 Comparison with previous studies
From Figure 17(a), it can be observed that for a particular In most of the previous studies (experimental and
value of yaw angle, the median value of failure gradient numerical), the emphasis was placed on the aspect of
wind speed is more at boom-down position than that of the determination of the drag and lift coefficients or the force
boom-up position. This implies that the failure of the crane coefficients for a proper design of container cranes. This
started earlier at boom-up position than boom-down study focuses on estimating the failure vulnerability
position. In other words, the boom-up position is much of container cranes under stochastic wind loading.

Figure 17. Influence of boom positions on the median value of failure gradient wind speed with respect to (a) gradient height and
(b) yaw angle.
1528 S. Gur and S. Ray-Chaudhuri

The results of this study show some important aspects, for a better design of container cranes, a failure
which were not observed previously. For example, vulnerability assessment should also be conducted. Other
previous studies (namely Lee, Kim, et al., 2007; Lee, specific findings of this study are as follows:
Shim, et al., 2007) have shown that at boom-up position,
. A comparison of responses by considering station-
the force coefficients become maximum at a yaw angle of
ary and non-stationary wind fields shows that for the
208 or 1708 and minimum at a yaw angle of 908. However,
assumed simulation parameters, the responses
this study shows that the failure vulnerability of container
considering stationary fields are larger than those
crane at boom-up position becomes maximum at a yaw
of the non-stationary fields.
angle of 608 and minimum at a yaw angle of 1508. Also,
. From the nonlinear dynamic analyses, it is found
Lee, Kim, et al. (2007) and Lee, Shim, et al. (2007) Lee
that although the displacement responses of the
and Kang (2007) have experimentally shown that at boom-
crane under a fluctuating wind field are one-sided, a
down position, the drag and lift coefficients become
stress reversal in flexure for some members can
maximum at yaw angles of 408 and 1408 and minimum at a
happen as a result of nonlinear deformations. It may
yaw angle of 908. However, the present study shows that
be noted that such members may be prone to failure.
the failure vulnerability becomes maximum at a yaw angle
. From the fragility analysis, it is found that the
of 608 and minimum at a yaw angle of 08 for the boom-
median value of failure gradient wind speed
down position. Thus, it is clear that the most unfavourable
increases almost linearly with an increase in
or favourable yaw angles for the force, drag and lift
gradient height.
coefficients are not same with those of the yaw angles
. It is observed that the failure vulnerability is
when failure vulnerability is considered. Note that, in
maximum at a yaw angle of 608 for both boom
previous studies, the drag and lift coefficients are
positions.
evaluated experimentally using scaled model and without
. For a 10-m gradient height and 608 yaw angle, it is
giving much attention to evaluate the behaviour of
also observed that the median failure gradient wind
members. However, in this study, the vulnerability of a
speed is about 46% less for a boom-up position
particular set of members is evaluated considering
when compared with a boom-down position.
simulated spatially varying stochastic wind fields.
The results of this study are limited to the parameter Note that like other parametric studies, the results of
space considered herein, i.e. (i) crane model, (ii) this study are limited to the parameter space considered
simulation parameters for wind field simulation, (iii) herein, and therefore, generalisation of these results will
assumed damage states and (iv) fragility model. It is require careful considerations.
imperative that the results will be different if a different set
of parameters are considered than those used in this study.
Therefore, the fragility results cannot be generalised for Acknowledgements
any crane, wind conditions or type of damage. The authors gratefully acknowledge Dr Ung Jin Na, Director, ITS
& Road Environment Division, Ministry of Land, Transport and
Maritime Affairs, Gwacheon 427-712, Korea for providing
details of the crane model used in this study. The work described
6. Conclusions in this paper is a part of master degree thesis of the first author
In this paper, a systematic study is conducted to estimate and was supported by the Ministry of Human Resource
the failure vulnerability of a container crane under Development (MHRD), Government of India.
spatially varying stochastic wind fields. Emphasis is
placed on understanding the effect of boom positions and
Note
wind field properties on the failure vulnerability of the
1. Email: sourav.gur.1987@gmail.com
crane. For this purpose, numerical models of a
representative (super-post-panamax) container crane in
boom-down and boom-up positions are developed using a
References
popular commercial software considering both material
and geometric nonlinearities. Stochastic wind fields are Ambrosini, R.D., Riera, J.D., & Danesi, R.F. (2002). Analysis
of structure subjected to random wind loading by simulation
simulated by means of the spectral representation method. in the frequency domain. Probabilistic Engineering Mech-
Finally, using the mean responses, vulnerability of the anics, 17, 233 –239.
crane is estimated by means of the fragility analysis American Society of Civil Engineers (ASCE 7-05) (2006).
wherein the failure due to flexural behaviour of base Minimum design loads for buildings and other structures.
members is considered as a damage state. It is observed New York, NY: ASCE.
Blevins, D.R. (1990). Flow induced vibration (2nd ed.). New
from this study that the failure vulnerability may not be York, NY: Van Nostrand Reinhold Company.
maximum for the set of parameter values for which the Boonyapinyo, V., Yamada, H., & Miyata, T. (1994). Wind-
aerodynamic forces are maximum. It is also concluded that induced nonlinear lateral – torsional buckling of cable-stayed
Structure and Infrastructure Engineering 1529

bridge. Journal of Structural Engineering, ASCE, 120, crane. In The fourth international symposium on compu-
486– 506. tational wind engineering (CWE2006), Yokohama, Japan.
Bourdeix, M.T., Hemon, P., & Santi, F. (1998). Wind induced Jeong, S.H., & Elnashai, A.S. (2011). Probabilistic fragility
vibrations of chimneys using an improved quasi-steady assessment method of structural intervention schemes.
theory for galloping. Journal of Wind Engineering and Journal of Structure and Infrastructure Engineering:
Industrial Aerodynamics, 74 – 76, 785– 794. Maintenance, Management, Life-Cycle Design and Perform-
Brigham, E.O. (1988). The fast Fourier transform and its ance, 8, 928– 938.
applications. Englewood Cliffs, NJ: Prentice-Hall. Jordan, M. A. (1995). Dockside container cranes: Reprinted from
British Standard (BS 2573-1) (1983). Rules for the design of PORTS ’95. In Proceedings, sponsored by the committee on
cranes Part 1: Specifications for classification, stress ports and harbors of the waterway, port, costal, and ocean
calculations and design criteria for structures (4th revision).: engineering division/ASCE, Tampa, FL, USA.
BSI. Kaimal, J.C., Wyngaard, J.C., Izumi, Y., & Coté, O.R. (1972).
Caracoglia, L., & Jones, N.P. (2003). Time domain vs. frequency Spectral characteristics of surface layer turbulence. Quar-
domain characterization of aeroelastic forces for bridge deck terly Journal of Royal Meteorological Society, 98, 563– 589.
sections. Journal of Wind Engineering and Industrial Kappos, A.J., & Panagopoulos, G. (2010). Fragility curves for
Aerodynamics, 91, 371– 402. reinforced concrete buildings in Greece. Journal of Structure
Chen, X. (2008). Analysis of alongwind tall building response to and Infrastructure Engineering: Maintenance, Management,
transient nonstationary winds. Journal of Structural Engin- Life-Cycle Design and Performance, 6, 39 – 53.
eering, ASCE, 134, 782– 791. Karmakar, D., Ray-Chaudhuri, S., & Shinozuka, M. (2012).
Chen, X., & Kareem, A. (2001). Nonlinear response analysis of Conditional simulation of non-Gaussian wind velocity
long-span bridges under turbulent winds. Journal of Wind profiles: Application to buffeting response of Vincent
Engineering and Industrial Aerodynamics, 89, 1335–1350. Thomas suspension bridge. Probabilistic Engineering
Chen, L., & Letchford, C.W. (2004). A deterministic – stochastic Mechanics, 29, 167– 175.
hybrid model of downbursts and its impact on a cantilevered Kim, H.K., Shinozuka, M., & Ghang, S.P. (2004). Geometrically
structure. Engineering Structures, 26, 619–629. nonlinear buffeting response of a cable-stayed bridge.
Chen, L., & Letchford, C.W. (2007). Numerical simulation of Journal of Engineering Mechanics, ASCE, 130, 848– 857.
extreme winds from thunderstorm downbursts. Journal of Lee, J.S., & Kang, H.J. (2008). Experimental study of wind load
Wind Engineering and Industrial Aerodynamics, 95, on a container crane located in a uniform flow and
atmospheric boundary layers. Journal of Wind Engineering
977– 990.
and Industrial Aerodynamics, 30, 1913– 1921.
Computer and Structures Inc. (2009). SAP2000 V. 14. Berkeley,
Lee, J.S., & Kang, H.J. (2007). Wind load on a container crane
CA, Computer and Structures Inc. Retrieved from http://
located in atmospheric boundary layers. Journal of Wind
www.csiberkeley.com
Engineering and Industrial Aerodynamics, 96, 193– 208.
Deodatis, G. (1996). Simulation of ergodic multivariate
Lee, S.W., Kim, Y., Han, D.S., & Han, G.J. (2007). Wind tunnel
stochastic process. Journal of Engineering Mechanics,
study on the structural stability of a container crane
ASCE, 122, 778– 787.
according to the boom shape. In The proceedings of the
Eden, J.F., Iny, A., & Butler, A.J. (1981). Cranes in storm winds. fourth WSEAS international conference on fluid mechanics,
Engineering Structures, 3, 175– 180. Gold Coast, Queensland, Australia.
European Standard (Eurocode 1-4) (2004). Actions on structures Lee, S.W., Shim, J.J., Han, D.S., Han, G.J., & Lee, K.S. (2007).
general actions part 1 –4: Wind actions. An experimental analysis of the effect of wind load on the
Gurley, K., & Kareem, A. (1998). A conditional simulation of stability of a container crane. Journal of Mechanical Science
non-normal velocity – pressure fields. Journal of Wind and Technology, 21, 448– 454.
Engineering and Industrial Aerodynamics, 77 – 78, 39 – 51. Liang, J., Ray-Chaudhuri, S., & Shinozuka, M. (2007).
Han, D.S., & Han, G.J. (2010). Reference data on uplift forces Simulation of non-stationary stochastic processes by spectral
for an alarm system to prevent an accident of a container representation. Journal of Engineering Mechanics, 133,
crane due to windblast. Journal of Mechanical Engineering 616– 627.
Science, 225(Part C), 1373– 1380. Lumley, J.L., & Panofsky, H.A. (1964). The structure of
Han, D.J., & Luo, J.J. (2006). Simulation of stochastic fluctuating atmospheric turbulence. New York, NY: Wiley.
wind field using the wave superposition method with random McCarthy, P., & Vazifdar, F. (2004). Securing cranes for storm
frequencies. In EPMESC X on computational methods in wind: Uncertainties and recommendations. In ASCE ports
engineering and science. Sanya, Hainan, China. conference, Houston, TX, USA.
Hansen, S.O., & Krenk, S. (1999). Dynamic along-wind response McCarthy, P., Jordan, M., Lee, K., & Werner, S. (2007).
of simple structures. Journal of Wind Engineering and Increasing hurricane winds dockside crane retrofit rec-
Industrial Aerodynamics, 82, 147– 171. ommendations. In ASCE ports conference, San Diego, CA,
Harikrishna, P., Shanmugasundaram, J., Gomathinayagam, S., & USA.
Lakshmanan, N. (1999). Analytical and experimental studies McCarthy, P., Soderberg, E., & Dix, A. (2009). Wind damage to
on the gust response of a 52 m tall steel lattice tower under dockside cranes: Recent failures and recommendations. In
wind loading. Computers and Structures, 70, 149– 160. TCLEE 2009 conference, Oakland, CA, USA.
Harris, R.I. (1990). Some further thoughts on the spectrum of Morfiadakis, E.E., Glinou, G.L., & Koulovari, M.J. (1996). The
gustiness in strong winds. Journal of Wind Engineering and suitability of the von Karman spectrum for the structure of
Industrial Aerodynamics, 33, 461– 477. turbulence in a complex terrain wind farm. Journal of Wind
Holmes, J.D., & Oliver, S.E. (2000). An empirical model of a Engineering and Industrial Aerodynamics, 62, 237– 257.
downburst. Engineering Structures, 22, 1167– 1172. Morris, C. A., & Hoite, S. (1997). The future of quayside
Huang, P., Wang, J.Y., & Gu, M. (2006). Wind tunnel test and container crane. In The post conference workshop, China
numerical simulation of mean wind loads on a container Ports, China.
1530 S. Gur and S. Ray-Chaudhuri

Pan, Y., Agrawal, A.K., & Ghosn, M. (2007). Seismic fragility of Soderberg, E., Hsieh, J., & Dix, A. (2009). Seismic guidelines for
continuous steel highway bridges in New York State. Journal container cranes. In TCLEE 2009 conference, Oakland, CA,
of Bridge Engineering, ASCE, 12, 689– 699. USA.
Pan, Y., Agrawal, A.K., Ghosn, M., & Alampalii, S. (2010). Spanos, P.D., & Kougioumtzoglou, I.A. (2012). Harmonic
Seismic fragility of multispan simply supported steel wavelets based statistical linearization for response evol-
highway bridge in New York State. II: Fragility analysis, utionary power spectrum determination. Probabilistic
fragility curves, and fragility surfaces. Journal of Bridge Engineering Mechanics, 27, 57 – 68.
Engineering, ASCE, 15, 462– 472. Sung, Y.C., & Su, C.K. (2011). Time-dependent seismic fragility
Paola, M.D. (1998). Digital simulation of wind field velocity. curves on optimal retrofitting of neutralised reinforced
Journal of Wind Engineering and Industrial Aerodynamics, concrete bridges. Journal of Structure and Infrastructure
74 – 76, 91 – 109.
Engineering: Maintenance, Management, Life-Cycle Design
Popescu, R., Deodatis, G., & Prevost, J.H. (1998). Simulation of
and Performance, 7, 797– 805.
homogeneous non-Gaussian stochastic vector fields. Prob-
abilistic Engineering Mechanics, 13(1), 1 – 13. Sun, D.K., Xu, Y.L., & Ko, J.M. (1999). Fully coupled buffeting
Ray-Chaudhuri, S., Karmakar, D., Na, U.J., & Shinozuka, M. analysis of long span cable supported bridges: Formulation.
(2009). Seismic performance evaluation of container cranes. Journal of Sound Vibration, 228, 569– 588.
In ATC and SEI 2009 conference on improving the seismic Torbol, M., & Shinozuka, M. (2012). The directionality effect in
performance of existing buildings and other structure, San the seismic risk assessment of highway network. Journal of
Francisco, CA. Structure and Infrastructure Engineering: Maintenance,
Scanlan, H.R. (1993). Bridge buffeting by skew winds in erection Management, Life-Cycle Design and Performance, 1, 1– 14.
stages. Journal of Engineering Mechanics, ASCE, 119, Yamazaki, F., & Shinozuka, M. (1988). Digital generation of
251– 269. non-Gaussian stochastic fields. Journal of Engineering
Scarabino, A., Di Leo, M.J., Delnero, J.S., & Bacchi, F. (2005). Mechanics, ASCE, 114, 1183– 1197.
Drag coefficients and strouhal numbers of a port crane boom Zhang, L.L., Li, J., & Peng, Y. (2008). Dynamic response and
girder section. Journal of Wind Engineering and Industrial reliability analysis of tall buildings subject to wind loading.
Aerodynamics, 93, 451– 460. Journal of Wind Engineering and Industrial Aerodynamics,
Shinozuka, M. (1971). Simulation of multivariate and multi- 96, 25 – 40.
dimensional random process. Journal of Acoustical Society Zhang, C.H., Sun, W., & Peng, X.Q. (2009). Numerical
of America, 49, 357– 368. simulation and analysis of interference effects on the port
Shinozuka, M., & Deodatis, G. (1991). Simulation of stochastic container group. In The Seventh Asia-Pacific Conference on
processes by spectral representation. Applied Mechanics Wind Engineering, Taipei, Taiwan.
Reviews, ASME Book No. AMR094 44, 191– 204. Zhong, J., Gardoni, P., & Rosowsky, D. (2012). Seismic fragility
Shinozuka, M., & Deodatis, G. (1996). Simulation of multi-
estimates for corroding reinforced concrete bridges. Journal
dimensional Gaussian stochastic field by spectral represen-
of Structure and Infrastructure Engineering: Maintenance,
tation. Applied Mechanics Reviews, ASME Book No.
AMR183 49, 29 – 53. Management, Life-Cycle Design and Performance, 8, 55 – 69.
Shinozuka, M., & Jan, C.M. (1972). Digital simulation of random Zhou, Y., Banerjee, S., & Shinozuka, M. (2010). Socio-economic
processes and its applications. Journal of Sound Vibration, effect of seismic retrofit of bridges for highway transpor-
25, 111– 128. tation networks: A pilot study. Journal of Structure and
Shinozuka, M., Feng, M.Q., Lee, J., & Naganuma, T. (2000). Infrastructure Engineering: Maintenance, Management,
Statistical analysis of fragility curves. Journal of Engineer- Life-Cycle Design and Performance, 6, 145– 157.
ing Mechanics, ASCE, 126, 1224 –1231. Zrnic, N., Oguamanam, D., & Bosnjak, S. (2006). Dynamics and
Simiu, E., & Scanlan, H.R. (1986). Wind effects on structure (2nd modeling of mega quayside container cranes. FEM
ed.). New York, NY: John Wiley and Sons. Transactions, 34, 193– 198.

View publication stats

You might also like