Soil Dynamics and Earthquake Engineering 77 (2015) 402–415

Contents lists available at ScienceDirect

Soil Dynamics and Earthquake Engineering

journal homepage: www.elsevier.com/locate/soildyn

Centrifuge modeling of offshore wind foundations under

earthquake loading
Hao Yu a,n, Xiangwu Zeng a,1, Frank H. Neff a,1, Bo Li b,2, Jijian Lian c,3
a
Department of Civil Engineering, Case Western Reserve University, Cleveland, OH 44106, United States
b
Department of Civil Engineering, Wenzhou University, Zhejiang, China
c
School of Civil Engineering, Tianjin University, Tianjin, China

art ic l e i nf o a b s t r a c t

Article history: The construction of large offshore wind turbines in seismic active regions has great demand on the
Received 12 December 2012 design of foundations. The occurrence of soil liquefaction under seismic motion will affect the stability of
Received in revised form the foundations and consequently the operation of the turbines. In this study, a group of earthquake
10 May 2015
centrifuge tests was performed on wind turbine models with gravity and monopile foundations,
Accepted 25 June 2015
respectively, to exam their seismic response. It was found that the seismic behavior of models was
quite different in the dry or saturated conditions. Each type of foundation exhibited distinct response to
Keywords: the earthquake loading, especially in the offshore environment. In the supplementary tests, several
Cementation remediation methods were evaluated in order to mitigate the relatively large lateral displacement of pile
Centrifuge test
foundation (by fixed-end pile and multi-pile foundation) and excessive settlement of gravity foundation
Densification
(by densification, stone column, and cementation techniques).
Foundation
Liquefaction & 2015 Elsevier Ltd. All rights reserved.
Liquefaction mitigation technique
Offshore wind turbine
Seismic response
Stone column

1. Introduction may cause excessive settlement and/or tilting of the structures. So

In recent years, offshore wind farms have become attractive
due to higher quality in wind resources at sea, requirement of no
land, and less impact on communities. However, the harsh off-
shore environment and loading conditions impose high demands
on the design of foundations. In particular, the fact that some
recently planned offshore wind farms will be built in seismic
active regions, such as the coastal regions of Southern Europe,
North America, and Eastern Asia, brings even greater challenges to
the design of foundations.
Earthquakes can cause significant damage to both foundations
and wind towers depending on the performance of the combined
system of superstructures, foundations, and surrounding soils. In
the offshore environment, soils can be softened by the increased
pore water pressure under earthquake loading. In the worst
scenario, earthquake-induced soil liquefaction can sharply reduce
the bearing capacity and lateral support of the foundation, which
n
Corresponding author. Tel.: þ 1 216 368 2950; fax: þ1 216 368 5229.
E-mail addresses: hxy88@case.edu (H. Yu), xxz16@case.edu (X. Zeng),
bxl102@case.edu (B. Li), jjlian@tju.edu.cn (J. Lian).
1
Tel.: þ1 216 368 2923; fax: þ1 216 368 5229.
2
Tel.: þ86 577 86697015; fax: þ 1 216 368 5229.
3
Tel.: þ86 22 27409525; fax: þ 86 22 27403841.
far, the understanding of the seismic resistance of offshore wind
foundations is very limited.
With little field data and experience, the seismic evaluation and
design of offshore wind tower foundations are commonly based
on the criteria developed for onshore wind turbines, oil platforms
[5,18] or general buildings [10,30,4]. However, these criteria
cannot completely address problems with wind turbines due to
the different loading environment and structural properties [47].
For offshore wind turbines, the effect of overturning moment
induced by heavy superstructure on the foundation is much more
severe than that of the vertical load. In addition, the natural
frequency of offshore wind turbine systems can be pretty close
to the dominant frequencies of an earthquake motion.
Various types of offshore foundations, i.e. gravity base, mono-
pile, tripods, suction buckets, and floating tension leg platforms
[39] have been proposed in recent years in order to provide
sufficient support for the wind tower. Byrne and Houlsby [11]
conducted a comprehensive review on the design of offshore
foundations and presented design calculations to investigate the
effects of foundation sizes and other critical parameters. Innova-
tive research using centrifuge modeling [46,27,29] has greatly
contributed to the design of structures in complex marine envir-
onment and the understanding of soil–structure interaction. In
recent years, a number of wind farms are located or planned to be

http://dx.doi.org/10.1016/j.soildyn.2015.06.014
0267-7261/& 2015 Elsevier Ltd. All rights reserved.
 H. Yu et al. / Soil Dynamics and Earthquake Engineering 77 (2015) 402–415
built in water less than 20 m deep. Gravity base and monopile
foundations, due to their simplicity and low-cost, as well as a great
deal of existing research and field experience, are the most
common and preferred foundations for offshore wind turbines
built in shallow water.
A gravity foundation depends mainly on its massive self-weight
induced bearing on the seabed to provide stability against the loads
transferred either from the superstructure or from the adjacent soil
and water. The effect of soil liquefaction on gravity structures was
studied by Lee and Focht [37], and Walker and Blair [61]. Dynamic
soil–structure interaction under seismic loading was investigated by
Ghosh and Madabhushi [28], Srbulov [55], and Dashti et al. [16]. The
pile supported structures tend to fail in the form of overturning or
tilting as the surrounding soil loses its lateral support during soil
liquefaction. Such failures have been reported by Yoshida and
Hamada [64], Tokimatsu et al. [57], and Adhikari and Bhattacharya
[2]. Both experimental and numerical analyses have been conducted
in recent years on the pile structures in order to study the complex
soil-pile-superstructure interaction during earthquakes [62,19,32,3].
However, the existing methods have not been applied to the case of
offshore wind turbine under earthquake loading. Although some
recent seismic investigations were conducted on full-scale turbine
models [36,65,48], the results still need to be calibrated and validated
by experimental data.
Over the past few decades, a number of techniques have been
developed to minimize liquefaction-induced hazards. For example,
soil densification has been widely used in the field. The effective-
ness of soil densification on the settlement during the liquefaction
was evaluated by Liu and Dobry [38], Mitchell et al. [45], and
Coelho et al. [15]. The technique of stone column, initially studied
by Seed and Booker [54], is currently accepted as one of the most
effective liquefaction countermeasures. The installation of stone
columns can effectively improve the stiffness of soil and reduce
the build-up of pore water pressure, and hence the associated
settlement by quick drainage during and immediately after the
earthquake [49]. The performance of stone column under seismic
loading can be found in theoretical analysis, model tests [50,44,1]
and case histories [43]. Cementation of soil is also considered as a
feasible technique to stabilize the foundation soil and therefore, to
reduce settlement induced by cyclic loading [20,25,31,13]. How-
ever, quantitative evaluation on how effective the improvement by
using this technique has not been well developed yet.
In this study, a group of earthquake centrifuge tests was
performed on two types of wind turbine models with a gravity
Fig. 1. Centrifuge models with (a) gravity base and (b) pile foundation (dimensions in prototype scale under 50 g).
403
base foundation or a monopile, respectively. The models took into
account the dimensions and masses of prototype wind turbines in
order to examine the loss of serviceability, i.e. settlement and
overturning, under earthquake loading. Tests in dry conditions
were also conducted to highlight the different structural responses
and amplified instability in saturated conditions. The experimental
results are useful in selecting and designing offshore wind founda-
tions in seismic areas. In addition, some supplementary tests were
conducted to evaluate the techniques that can mitigate seismic-
induced instability. It was found that the fixed-end pile founda-
tion, simulating a pile penetrating into bedrock, may provide a
design to minimize the lateral displacement under earthquake
loading. The multi-pile foundation would also effectively reduce
the lateral movement without amplifying the seismic response at
the tower head. At the same time, the techniques of densification,
stone columns, and cementation were found to be effective in
mitigating the excessive settlement of gravity base foundation in
the offshore environment.
2. Test program
2.1. Centrifuge models
As shown in Fig. 1, two simplified models (dimensions are shown
in prototype scale for tests at 50 g) were adopted according to the
structural features of offshore wind turbines in the field. Due to the
constrain of the size of the centrifuge models, the prototype structure
it represents would be a medium size wind turbine. The models with
each part labeled in Fig. 1 had the same superstructure but different
foundations. The superstructure consists of tower head and wind
tower. The tower head was simplified as a lumped mass at the top of
the tower rod with large slenderness ratio. Such structure was
expected to induce high overturning moment during an earthquake
compared with a typically short and broad building model. Details of
the model are described in Table 1.
The gravity foundation (Fig. 1a) was modeled by an aluminum
block which is much heavier than the superstructure. It was
embedded 1.5 m below the ground surface and the average
contact pressure of the foundation on the subsoil was about
70 kPa. The pile foundation (Fig. 1b) was fabricated from a solid
cylinder with the diameter and length of 0.9 m and 4.5 m (in
prototype scale), respectively. In the tests, the pile foundation was
fully driven into the soil. The end of the pile was standing on the
404 H. Yu et al. / Soil Dynamics and Earthquake Engineering 77 (2015) 402–415
Table 1
Wind turbine models in the centrifuge tests.
Dimension (m) Weight (ton)
Tower head 1.75  1.75  1.25 10.6
Wind tower L ¼ 13; D ¼ 0.5 6.25
Foundation Gravity base 3.75  3.75  2 75.9
Pile L ¼ 4.5; D ¼ 0.9 7.25
In prototype scale; L ¼ length; D¼ diameter.
base of the container simulating a pile resting on the bedrock in
the field.
The pile foundation with relatively large bending stiffness and
short length can be treated as a rigid body. Therefore, it is unlikely
to have a structural failure of the pile attributed to the inertial
force from the superstructure and kinematic force from the
ground movement [59]. In addition, the lateral displacement of
“rigid” pile at ground surface can be considered as a consequence
of the rotation of the pile [66], which might provide a straightfor-
ward way to check its stability under earthquake loading.
2.2. Testing facility and test procedures
The centrifuge tests were performed on the geotechnical centri-
fuge at Case Western Reserve University. The centrifuge arm has an
effective radius of 1.37 m. The payload capacity is 20 g-ton with a
maximum acceleration of 200 g for static tests and 100 g for dynamic
tests. The centrifuge is equipped with a hydraulic shaker designed by
TEAM Corporation. The details of the centrifuge were reported by
Figueroa et al. [23]. The earthquake generated by the shaker was
recorded by an accelerometer mounted on the shake table along the
direction of the seismic wave.
A rigid container was employed in the tests. The internal
dimensions of the container are 53.3 cm (length)  24.1 cm
(width)  17.7 cm (height). Comparing with other types of model
containers such as equivalent shear beam containers, laminar
boxes, or active boundaries containers, the rigid container might
not be the ideal choice for earthquake centrifuge tests. However,
the advantages of simple design, easy to use, and low cost make
the rigid container still widely used in centrifuge modelling. The
effects of the boundary as well as the size of rigid container on the
seismic responses in centrifuge tests have been well studied by
Elgamal et al. [21], Yang et al. [63], and Coelho et al. [14]. The use of
a rigid container in the earthquake centrifuge tests may exacerbate
the responses of both soil and structure. Since the generated
seismic wave can be reflected by the rigid sidewall and trans-
mitted back into the model, it may result in the intensified
vibration of the wind turbine and the delayed dissipation of pore
water pressure, therefore inducing additional structural settle-
ment and rotation. Yang et al. [63] have systematically compared
the seismic responses of earth dam by using both rigid and
laminar containers. They found that the laminar container models
experienced smaller acceleration and displacement than rigid
container models in loose sands; however for dense sands, the
two different containers resulted in nearly identical responses.
Takahashi et al. [56] conducted dynamic centrifuge model tests on
quay walls and drew the conclusion that the effect of a rigid
sidewall on the deformation behavior and acceleration character-
istics could be greatly mitigated as long as sufficient distance to
the sidewall was provided. However, the critical distances were
not well defined for different seismic waves and different size of
containers. Therefore, the effect of rigid sidewall on the seismic
behaviors of wind turbine should be further studied.
The soil layer was constructed with well graded Toyoura sand.
Toyoura sand is a uniform sand with angular particles with the
following physical properties: emax ¼ 0:98; emin ¼ 0:6; Gs ¼
2:65; and D50 ¼ 0:17 mm. During the model preparation, the
sand was uniformly poured from the height of 80 cm to keep the
relative density at about 68%. The thickness of the soil layer was
4.5 m. In saturated tests, the sand layer was saturated by de-aired
water under vacuum. The water table was maintained at 1.5 m
above the ground surface to simulate condition which would exist
at an offshore location. Instead of using viscous fluids such as
silicon oil, glycerin–water mixture, or Methycellulose, de-aired
water was used in this study as the pore fluid in order to observe
the entire process of development of excess pore pressure under
the seismic excitation which includes the buildup, steady state,
and dissipation. It would not affect the validity of the experiments
as long as the increase in permeability of the soil is properly taken
into account (in this case, increased by 50 times) in the data
interpretation and analysis.
The models were instrumented with accelerometers (ACC),
pore pressure transducers (PPT), and Linear Variable Differential
Transducers (LVDTs). The locations of these transducers were
presented in Fig. 1. The container was placed on the centrifuge
arm and the model was gradually spun up to a centrifugal
acceleration of 50g. The input motion generated by the hydraulic
shaker is a 1D synthetic earthquake (Fig. 2) at the base of the
model. The dominant frequency of input motion was about 1 Hz
which appeared to be smaller than the natural frequency of the
two turbine models. The time histories including accelerations in
the structures and soils, excess pore water pressures, settlements
of gravity model and ground surface in the free field, and the
lateral displacements of structures were recorded by the installed
sensors. Centrifuge scaling laws [53] were employed to evaluate
corresponding prototype behavior of the models.
The centrifuge tests reported in this paper were conducted for
general research purpose and not simulating one particular
structure in the field. One of the goals of our study is to compare
the seismic behaviors and resistances of a wind turbine with two
different foundations, especially in the event of soil liquefaction.
The wind turbine models with a gravity base foundation and a
monopile foundation have quite different natural frequencies. A
relatively simply earthquake with one dominant frequency was
employed as the input motion as the first step of the research. The
selected seismic wave was previously used in the VELACS project
which studied on the effects of earthquake-like loading on a
variety of models. It is a “baseline” type of input motion used by
quite a number of researchers. The recorded results in this study
can therefore be compared with those in various centrifuge tests
using the same input motion, as reported in the VELACS database.
0.8
0.4
Acceleration (g)
0.0
-0.4
-0.8
0 5 10 15 20 25
Time (sec)
Fig. 2. Typical input acceleration.
 H. Yu et al. / Soil Dynamics and Earthquake Engineering 77 (2015) 402–415 405

3. Experimental results 3.1. Seismic behavior of the soil

In this study, the models were tested in both dry and saturated In the test of Case 1 on gravity foundation models (Fig. 1a), the
conditions. The summary of the performed tests is listed in Table 2. soil behavior under seismic loading was measured both under the
The results are explained in the following sections. To check the structure (ACC4 and ACC5) and in the free field (ACC6–ACC8) in
repeatability of the experimental results, a few tests were carried out order to investigate the effect of surcharge on the dynamic soil
twice. Examples of the experimental results in two repeated tests are responses. In the dry condition, the soil accelerations reported in
shown in Fig. 3, indicating a good repeatability. the free field were quite similar to the input acceleration (Fig. 2).
According to the time histories recorded by ACC5 and ACC7
(Fig. 4a), which were placed at different locations but the same
Table 2 depth, there was almost no difference in the soil responses under
Summary of performed centrifuge modeling tests. the structure and in the free field. However, in the saturated tests,
the soil exhibited quite different behavior at the two locations.
Case Foundation type Test condition Note
Pore pressure transducers were employed to record excess pore
1 Gravity base Onshore/offshore
water pressure in the soil. With the increase of pore water
2 Monopile Onshore/offshore Pile rested on bedrock pressure ratio (ratio of increment of pore water pressure Δu to
the effective stress of soil σ 0 ) at PPT4 (Fig. 5a), the soil in the free
Tests with liquefaction-mitigation techniques
3 Monopile Onshore/offshore Pile anchored to bedrock field was softened, resulting in the stiffness reduction of soil
4 Gravity base Offshore Densification around ACC7 (Fig. 5b) during the time period of 2–7.5 s. When
5 Gravity base Offshore Stone columns pore water pressure began to dissipate after strong shaking, the
6 Gravity base Offshore Stone columns
soil gradually regained its stiffness. However, the generated pore
7 Gravity base Offshore Cementation
8 Gravity base Offshore Cementation
pressure ratio under the structure (PPT2, Fig. 5a) was much lower
than that in the free field, and correspondingly there was no

0.6 0.6

0.0
0.0

-0.6
ACC1 -0.6 ACC1
Acceleration (g)
Acceleration (g)

0.2 0.2

0.0 0.0

-0.2 ACC2 -0.2 ACC2

0.4 0.5

0.0
0.0

-0.4
ACC3 -0.5 ACC3
0 5 10 15 20 25 0 5 10 15 20 25
Time (sec) Time (sec)

1.0 PPT1 1.0 PPT1

0.5 0.5
0.0 0.0
1.0 PPT2 1.0 PPT2
0.5 0.5
0.0 0.0
1.0 PPT3 1.0 PPT3
0.5 0.5
0.0 0.0
1.0 PPT4 1.0 PPT4
0.5 0.5
0.0 0.0
0 5 10 15 20 25 0 5 10 15 20 25
Time (sec) Time (sec)
Displacement (cm)
Displacement (cm)

20 15
10 10
Lateral
Lateral

5
0 Max. Rotation: 1.5° Max. Rotation: 1.3°
0
Res. Rotation: 1.3° Res. Rotation: 1.1°
-10 -5
0 5 10 15 20 25 0 5 10 15 20 25
Time (sec) Time (sec)
Fig. 3. Test results of pile model in the saturated condition (a) in original test and (b) in repeated test.
406 H. Yu et al. / Soil Dynamics and Earthquake Engineering 77 (2015) 402–415

0.5 ACC5 PPT2

ACC7 1.0 PPT4

Pore Water Pressure Ratio

Acceleration (g)

0.5
0.0

0.0

PPT5
1.0
-0.5

Pore Water Pressure Ratio

0 5 10 15 20 25
Time (sec)
0.5

0.5 ACC4

0.0 0.0
0 5 10 15 20 25
-0.5
Acceleration (g)

0.5 Time (sec)

ACC6

0.0
0.5 ACC5
ACC7
-0.5
0.5 ACC7
Acceleration (g)

0.0
0.0
-0.5
0 5 10 15 20 25
Time (s)
Fig. 4. Accelerations recorded in the dry soil (a) in gravity model and (b) in -0.5
pile model. 0 5 10 15 20 25

0.5 ACC8
visible change in the soil behavior (ACC5, Fig. 5b). This was due to
the fact that the gravity structure applied higher initial confining
Acceleration (g)

pressure, and therefore increased the shear strength of the subsoil

and its resistance to soil liquefaction. According to the recording of
PPT5 in the free field (Fig. 5a), the pore pressure ratio reached to
1 during the strong shaking, which was shown by the significant
reduction of soil stiffness around ACC8 (Fig. 5b), indicating that the
top layer was completely liquefied in the free field.
In the test of Case 2 on pile models (Fig. 1b), ACC4–ACC7 were
embedded at different depths to measure the responses in the
free-field soil which were not influenced by either pile foundation
or the container boundaries. The recorded time histories of
accelerations (Fig. 4b) in the dry condition displayed almost the
same waveform and magnitudes as the input motion. In the
saturated tests, the changes in soil behavior with depth were
quite evident. From bottom to top, the pore pressure ratios
(Fig. 6a) gradually increased, while the soil stiffness decreased
(Fig. 6b). Such vibration attenuation was quite noticeable at the
top of the layer (ACC7), implying soil liquefaction.
3.2. Seismic behavior of the structure
As shown in Fig. 1a, ACC1 and ACC2 measured the accelerations
of gravity base in the horizontal and vertical directions, respec-
tively. ACC2 was installed some distance off the center of the
foundation, and thus the recorded time histories would capture
the rocking motion of the foundation. ACC3 measured the
responses of the tower head. Comparing the recorded responses
of the gravity model in the two different conditions (Fig. 7), the
horizontal accelerations in the foundation (ACC1) and tower head
0.0
-0.5
0 5 10 15 20 25
Time (sec)
Fig. 5. Recorded (a) pore water pressure ratios and (b) accelerations of soils in the
saturated test of gravity model.
(ACC3) were weakened in the saturated tests due to the increased
soil damping with cyclic loading. However, the rocking of the
gravity base (ACC2) was greatly amplified in the saturated condi-
tion, especially during the strong shaking.
The arrangement of accelerometers in the pile model was
exactly the same as in the gravity model. The pile was standing
on the base of the container without fixity. ACC2 was mounted
vertically on the platform on top of the pile foundation (Fig. 1b).
Comparing the recorded results in the two different conditions
(Fig. 8), the frequencies of structural responses in the saturated
tests were obviously lowered, but the amplitude in each part of
the model was slightly increased when liquefaction occurred.
When comparing the behaviors of different models, the seismic
response of the pile model was larger than the gravity model both
in the dry and saturated conditions. In terms of the lateral
displacements, both gravity model and pile model remained stable
 H. Yu et al. / Soil Dynamics and Earthquake Engineering 77 (2015) 402–415 407

1.0 PPT1 0.6

0.5
0.0
0.0
1.0 PPT2
Pore Water Pressure Ratio

-0.6 ACC1

Acceleration (g)
0.5
0.1
0.0
1.0 PPT3 0.0
0.5
-0.1
0.0 ACC2
1.0 PPT4 0.4
0.5
0.0
0.0
0 5 10 15 20 25 -0.4
ACC3
Time (sec)
0 5 10 15 20 25
Time (sec)
0.6

0.0
0.6
-0.6 ACC4
0.6 0.0
Acceleration (g)

0.0
-0.6 ACC1

Acceleration (g)
-0.6 ACC5
0.6 0.1

0.0 0.0

-0.6 ACC6 -0.1

0.6 ACC2
0.4
0.0

-0.6 ACC7 0.0

0 5 10 15 20 25
Time (sec) -0.4 ACC3
0 5 10 15 20 25
Fig. 6. Recorded (a) pore water pressure ratios and (b) accelerations of soils in the
Time (sec)
saturated test of pile model.
Fig. 7. Accelerations of gravity model in (a) dry tests and (b) saturated tests.

in the dry condition (Fig. 9a), while the dynamic vibrations were
significantly amplified in the saturated condition (Fig. 9b). How-
ever, the residual displacement for the gravity model was still
within a low range and the structure eventually rotated only 0.31
after shaking (Table 3), which indicated that the gravity base had a
greater resistance to the overturning moment under seismic
loading. On the other hand, the lateral displacement of pile model
accumulated during the liquefaction process and the residual tilt
was almost 1.31. Therefore, it can be concluded that the pile
foundation used in this study was more susceptible to the earth-
quake induced tilt, especially in the offshore environment.
3.3. Settlements of structure and ground surface in the gravity model
In the tests on a gravity model, LVDTs were used to measure the
settlements of the structure and ground surface in the free field.
According to the results shown in Fig. 10a, the settlements of both
structure and ground were negligible in the dry condition. The
time history of structural settlement was accompanied by many
fluctuations due to the rocking of the foundation, while the ground
settlement looked quite smooth.
In the saturated test (Fig. 10b), the dissipation of pore water
pressure after liquefaction greatly amplified the ground settlement
from 6 cm to 28 cm. Although the soil underlying the structure did
not liquefy, the structure still settled more than 30 cm, which was
even larger than the ground settlement. This amplification might
be caused by the soil–structure-interaction (SSI), which is dis-
cussed in the following section.
3.4. Interpretation of experimental results and discussion
Based on the experimental results presented in previous sec-
tions, it was noticed that the gravity foundation had a better
lateral resistance due to its massive weight and large contact area
with the soil, but a more significant structural settlement in the
offshore condition; the pile foundation resting on bedrock had
minimal settlement, but was susceptive to a larger lateral dis-
placement due to the lack of support by surrounding soils. These
results agreed well with the characteristics and performance of
offshore foundations in the field. It seemed hard to choose,
between the two distinct types of foundations, which one is better
suited for offshore wind turbine in seismic areas. The analysis of
the measured results in this section is to provide a more specific
guidance to estimate the seismic resistance of each type of
foundations.
If an onshore wind turbine is constructed in an area with
relatively dense sandy soil, the greater stiffness of the soil may
offer sufficient bearing to the wind turbine. The structural settle-
ment and tilt will be minimized under earthquake excitation. On
the other hand, the stiff soil can efficiently transmit the seismic
energy to the foundation, and consequently affect the safety of the
structure and the operation of the machine. It agrees well with the
facts that, in the dry tests, the soil responses at different depths
(Fig. 4) were quite similar to the input motion without much
attenuation. In addition, the recorded accelerations by ACC1 and
ACC3 of the gravity model were more significant in the dry
condition than in the saturated condition (Fig. 7).
408 H. Yu et al. / Soil Dynamics and Earthquake Engineering 77 (2015) 402–415

Lateral Displacement (cm)

0.6 4

Gravity
0.0 0

-0.6 ACC1 -4
4
Acceleration (g)

0.1

Pile
0.0
0
-0.1
ACC2
4

Fixed Pile
0.4

0.0 0

-0.4 -4
ACC3
0 5 10 15 20 25
0 5 10 15 20 25
Time (sec) Time (sec)

Lateral Displacement (cm)

0.6
4

Gravity
0.0 0

-0.6 -4
ACC1 20
Acceleration (g)

0.2
Pile 10
0.0 0

-0.2 ACC2 -10

4
Fixed Pile

0.4
0
0.0

-0.4 -4
ACC3
0 5 10 15 20 25
0 5 10 15 20 25
Time (sec)
Time (sec)
Fig. 9. Time histories of lateral displacements of each model (a) in the dry tests and
Fig. 8. Accelerations of pile model in (a) dry tests and (b) saturated tests.
(b) in the saturated tests.

In the offshore condition, the softened soil tended to dissipate Table 3

the seismic energy better. However, the development of pore Comparison of rotational angle of each model in the saturated tests.
water pressure and occurrence of soil liquefaction would induce
Maximum rotation (deg) Residual rotation (deg)
instability or even failure of the structure. For the gravity model,
the generated pore pressure ratio was much lower under the Gravity base model 0.5 0.3
structure than in the free field (Fig. 5a) due to the higher initial Pile model 1.5 1.3
confining pressure and static shear stress induced by the structure Fixed-end pile model 0.28 o0.1
[24,38,16]. The minor attenuation of ACC5 (Fig. 5b) also implied Multi-pile model 0.7 0.6

that no liquefaction occurred under the gravity base. The amplified

structural settlement essentially resulted from the effect of SSI.
The increased rocking of gravity base in the saturated condition
(Fig. 7b, ACC2), as well as the resulting shear stress by the relative
movement of foundation and underlying soil, were expected to
produce local failure of subsoil and local subsidence by extruding
the weakest soil away from underneath the foundation [17]. The
recorded fluctuations in the time history of structural settlement
shows the effect of SSI. This mechanism might also be amplified by
the vibration of the superstructure.
The tests on the pile foundation simulated the condition of a
monopile standing on bedrock without an anchor. In the dry
condition, the surrounding soil provided sufficient lateral support
during earthquakes. However, in the saturated condition, the
slender pile foundation tended to tilt by rotating in the softened
soil under the strong shaking. During liquefaction, the soil stiffness
and strength of the top layer were lost, contributing to the large
residual tilt angle after the earthquake. The peak accelerations and
root mean square values of the recorded signals during the strong
shaking (2–7.5 s) are compared in Table 4. This shows that the
acceleration at the pile tip (ACC1) during liquefaction has
exceeded the input motion, due to the interaction between soil
and pile. As reported in the studies by Kagawa [33], Tokimatsu and
Asaka [58] and Mao et al. [40], the responses of the pile tip,
including acceleration, displacement, shear stress, and moment,
tend to increase with the stiffness ratio of the pile and the
surrounding soil. Accordingly, the seismic behavior of the pile
foundation might be amplified when interacting with the softened
soil undergoing liquefaction. In addition, the natural frequency of
the system was significantly reduced during soil liquefaction,
which might increase the dynamic amplification factor of the
structure and enhance the effect of inertial force induced by
superstructure on the pile foundation [59].
4. Evaluation of remediation techniques
As the result of some unsatisfactory behaviors of each type of
foundation in the offshore condition, several mitigation techniques
 H. Yu et al. / Soil Dynamics and Earthquake Engineering 77 (2015) 402–415 409

the container in order to simulate the pile foundation anchored to

Structure the bedrock. In the dry condition (Fig. 11a), the structure behaved
0
Free-Field pretty similarly to the pile model without fixity (Fig. 8a). On the
other hand, in the saturated tests (Fig. 11b) the occurrence of
liquefaction amplified the difference in structural response
between pile model and fixed-end pile model. As demonstrated
Settlement (cm)

in Table 4, the fixed-end pile had effectively restrained the move-

ment of the foundation (ACC1 and ACC2). However, the accelera-
-3 tion in the tower head was increased, which might adversely
influence the normal operation and efficiency of the wind turbine.
As shown in Fig. 9, the lateral displacement of the fixed-end
pile model was very small in the dry tests. Although the dynamic
vibration was not obviously reduced in the saturated condition,
the residual tilt was less than 0.11 (Table 3) which can be
neglected. Therefore, the fixed-end pile foundation exhibited great
-6
improvement to rotational displacement under liquefaction. How-
0 5 10 15 20 25 ever, the high cost of deep piling should be taken into considera-
Time (sec) tion. There are also other engineering methods to improve lateral
stability of offshore wind turbines, such as a multi-leg foundation.
The selection of a suitable foundation should be based on the cost,
0 Structure efficiency, and local soil condition.
Free-Field
-5 4.2. Multi-pile model

-10 Since the single pile supported wind turbines tend to fail in the
Settlement (cm)

form of tilting or overturning as soil ceases to provide lateral support

-15 during soil liquefaction, quite a number of studies have been
performed on multi-pile foundations for offshore structures. McClel-
-20 land [41] presented the effects of pile size selection on the stability of
the deep foundation for offshore structures. Bea et al. [6] analyzed the
-25 structural reliability of a platform with multi-pile foundations under
extreme environment, i.e. hurricanes and earthquakes. Brandenberg
-30 et al. [8] performed earthquake centrifuge tests to investigate the

-35

0 5 10 15 20 25 0.6

Time (sec) 0.0

Fig. 10. Settlements of gravity model in the (a) dry test and (b) saturated test.
-0.6 ACC1
Acceleration (g)

0.1
Table 4
Comparison of structural responses between pile model and fixed-end pile model 0.0
in the saturated tests. -0.1 ACC2
Pile model Fixed-end pile model 0.4

0.0
(Max.) g (RMS) g (Max.) g (RMS) g
-0.4 ACC3
ACC1 0.916 0.350 0.659 0.284
0 5 10 15 20 25
ACC2 0.208 0.075 0.137 0.039
ACC3 0.360 0.155 0.500 0.268 Time (sec)
Input motion Max.¼ 0.649 g; RMS ¼0.265 g

0.6
Calculated from 2 s to 7.5 s. Max. ¼ maximum value; RMS¼ root mean square value.
0.0

were evaluated in order to improve the lateral resistance of the

Acceleration (g)

-0.6 ACC1
pile foundation and to reduce the settlement of the gravity
0.1
foundation under earthquake shaking.
0.0
-0.1
4.1. Fixed-end pile model ACC2
0.6

In the field condition, if the bedrock is at a reasonable depth 0.0

and of good quality, it is considered that the stiff rock can provide
the structure with sufficient bearing capacity and lateral resistance -0.6 ACC3

during seismic shaking. However, driving the pile into the bedrock 0 5 10 15 20 25

will inevitably increase the demand on installation facilities and Time (sec)
the difficulty of construction, as well as the cost of the project. In Fig. 11. Accelerations of fixed-end pile model (a) in the dry test and (b) in the
two supplementary tests in Case 3, the pile was fixed to the base of saturated test.
410 H. Yu et al. / Soil Dynamics and Earthquake Engineering 77 (2015) 402–415

responses of pile group foundations during soil liquefaction. Chen
et al. [12] have compared the structural properties of monopile and
tripod offshore wind turbine foundations and indicated that the
tripod foundation had greater stiffness, longer lifetime under cyclic
loading and even better stress-control capacity than the monopile
foundation [12].
Tripod foundations, as an alternative to traditional monopile
foundations, are one of the most favorable offshore multi-pile
foundations because of its light weight and extraordinary resis-
tance to the challenging marine environment. The tripod founda-
tion is typically installed in water depth more than 25 m, a deeper
water depth than monopile and gravity base foundations. The
reduction of soil stiffness. Among current techniques that can mitigate
the seismic-induced settlement in both onshore and offshore struc-
tures, the most common methods can be divided into two categories:
compaction/densification methods and soil reinforcement methods
by either replacing existing foundation soil or introducing additional
materials into the ground. Without any field case histories available,
1.0 PPT1
0.5

Pore Water Pressure Ratio

three legs of tripod are piled to the seabed, supporting the 0.0
transition piece and turbine. As shown in Fig. 12, the tripod 1.0 PPT3
foundation was modeled by using three aluminum rods of 0.5 m
diameter integrated by a triangular transition piece with each side 0.5
of 4 m and driven into ground soil by 3 m (in prototype scale). The
configurations of soil deposit, wind tower, and input motion were 0.0

exactly the same as those in the monopile model test. 1.0 PPT4
In the shallow depth of 0.5 m below the ground surface, the
accelerometers and pore pressure transducers were installed in 0.5
pairs right under the structure, outside the structure but adjacent
0.0
to the tripod leg and in the free field. The recorded time histories
of pore water pressure ratio are compared in Fig. 13. The lowest 0 5 10 15 20 25
pore water pressure was recorded right beneath the structure. The Time (sec)
underlying soil was reinforced by the three legs of the tripod Fig. 13. Recorded time histories of pore pressure ratio in tripod foundation
foundation and the confining pressure had been increased. The model test.
higher residual soil strength can also be observed from the
recorded acceleration in ACC1 (Fig. 14). The seismic response of 0.3 ACC1
the tower head was measured by ACC6. As shown in Fig. 15, the
maximum recorded acceleration during the earthquake was less 0.0
than 0.2 g, which was just half of that measured in monopile
foundation model. In addition, the wind turbine model with tripod -0.3
Acceleration (g)

foundation rotated only 0.61 after the earthquake (Table 3), which 0.3 ACC3
was 50% smaller than the model by using the monopile founda-
tion. The results clearly demonstrated that the multi-pile founda- 0.0
tion is an effective way of resisting lateral movement of wind
turbines under seismic loading. -0.3
0.3 ACC4

4.3. Settlement mitigation techniques for gravity foundations

0.0

It was observed in the previous experiments that earthquake

-0.3
shaking can cause significant settlement to an offshore wind turbine 0 5 10 15 20 25
with gravity foundations. This behavior can be largely attributed to Time (sec)
the build-up of pore water pressure during shaking and hence a
Fig. 14. Recorded time histories of acceleration in tripod foundation model test.

0.4 Tripod

0.2

0.0

-0.2
ACC6 (g)

-0.4
0.4 Monopile

0.2

0.0

-0.2

-0.4
0 5 10 15 20 25
Time (sec)

Fig. 12. Centrifuge model with tripod foundation (dimensions in prototype scale Fig. 15. Comparison of recorded acceleration in tower head between tripod
under 50 g). foundation model and monopile foundation model.
 H. Yu et al. / Soil Dynamics and Earthquake Engineering 77 (2015) 402–415 411

these techniques have not been widely used in offshore wind farms.
However, their effectiveness in reducing seismic-induced settlement D=1m
can be evaluated by performing centrifuge tests. For a specific project,
the selection of a suitable liquefaction mitigation technique should
take into account a number of factors which include feasibility,
geotechnical condition, environment, cost, and performance. In this
study, five supplementary tests were conducted by employing three
commonly used techniques, which include densification (Case 4);
stone columns (Cases 5 and 6); and cementation (Cases 7 and 8). The
results were used to compare with the data recorded in Case 1.

4.3.1. Densification of soil

Based on the observation in past earthquakes, densification is
considered as an effective technique to increase liquefaction 3.75m 1.9m 1.9m
resistance and to mitigate the induced settlements of shallow
foundations. The successful use of compaction as the treatment for
A2
loose and liquefiable marine soils in offshore projects include
cases such as Brest Naval Port in France [7], Pointe Noire in Gabon
[42], and Lagos Dry Dock in Nigeria [26]. In the test for Case 4, the
entire depth of soil deposit underneath the structure was com-
pacted to a dense state with a relative density of about 85%. To
keep the compacted soil undisturbed, only the ground settlement
and structural response were recorded in this case.
In comparison with the results in Case 1 which was without
compaction, the foundation settlement in Case 4 was significantly
reduced from 33 cm to less than 10 cm (Fig. 19). It was found by
Sadrekarimi and Ghalandarzadeh [49] that the densification tech-
nique was more efficient than stone columns for liquefaction
resistance and settlement reduction during an earthquake. How- 1.5m

ever, the compacted soil under the wind turbine maintained most P5 A1
A5
P3 0.5m

1.5m
of its initial stiffness and hence transmitted a large amount of P4 P2
seismic energy to the structure. Liu and Dobry [38] reported that P1 A3 A4

the foundation acceleration would increase with the compaction

depth. Therefore, it is important in offshore engineering to analyze
the effect of compacted density and compaction depth to control 8m 3m

the settlement while at the same time to minimize the seismic Fig. 16. Schematic view of stone columns configuration with two different tests:
impact to the upper structure. (a) Case 5; (b) Case 6 and (c) configuration of testing model and transducers.

4.3.2. Stone columns

It is well known that the installation of stone columns (gravel
drains) can increase foundation stiffness and speed up pore
pressure dissipation, and hence decreasing the seismic-induced
settlement and liquefaction potential. For example, stone columns
in a seawall in Patras, Greece served as drainage for excess pore
pressure that built up during construction and other dynamic
loading, and also provided the foundation with additional strength
under earthquake loading [60]. Unlike the dynamic compaction
method, construction of stone columns produces quite lower noise
and disturbance to the adjacent offshore structures, which might
be suitable for a site close to existing offshore structures. After
conducting careful site investigation and numerical analysis of
ground performance under cyclic loading, hundreds of 10 m deep
stone columns were installed at the particularly challenging
ground to improve the performance of gravity foundation of wind
turbines in the Hill of Fiddes wind farm, Aberdeenshire [34].
Centrifuge tests with two different configurations (Fig. 16) were
performed in this study to investigate how the stone columns
would influence the responses of soil and wind turbine structure
under earthquake loading.
The material representing the stone columns was poorly
graded fine gravels with the average particle size of 3 mm. In both
cases, the stone columns were installed up to the bottom of the
container by using a casing. Each column had the same diameter
of 1 m in prototype scale. In Case 5, the stone columns were
installed under the four corners of the foundation, while in Case 6,
eight columns were placed at both corners and edges. The
configuration of the model and transducers were shown in
Fig. 16c (the locations of transducers were different from those
in the previous tests). The testing procedures were exactly the
same as those mentioned before.
Fig. 17 shows the measured responses of the structure and soil
in the models with stone columns. The pore water pressures
generated under the structure (PPT1) and adjacent to the stone
columns (PPT2 and 3) were clearly lower than those in the model
without stone columns (Fig. 5a). However, there was no significant
change in pore water pressure generated in the free field (PPT4
and 5). The results also indicated that the drainage benefit
increased with the area of the stone columns. It can be seen in
Case 6 that the excess pore water pressure in the adjacent field
(Fig. 17b, PPT1, 2 and 3) dissipated earlier and faster than that in
Case 5. In addition, the stiffening effect of the stone columns on
the structural response and settlement was quite obvious. For Case
6 which had a larger replacement area, the installation of stone
columns greatly preserved the overall foundation stiffness. As a
result, the ground which had higher shear strength and bearing
capacity mitigated the SSI effect by resisting the penetration of the
gravity foundation during the strong shaking and therefore
reduced the settlement. As shown in Fig. 19, the measured
structural settlement in Case 6 was reduced by about 50%.
However, the seismic wave was able to transfer effectively to the
structure through the stiff composite ground. As a result, the
412 H. Yu et al. / Soil Dynamics and Earthquake Engineering 77 (2015) 402–415

0.6 1.0
ACC1 PPT1
0.0 0.5
-0.6 0.0
0.4 ACC2 1.0 PPT2

Pore Water Pressure Ratio

0.0 0.5
Acceleration (g)

-0.4 0.0
0.6 ACC3 1.0 PPT3
0.0 0.5
-0.6 0.0
0.6 ACC4 1.0 PPT4
0.0 0.5
-0.6 0.0
0.6 ACC5 1.0 PPT5
0.0 0.5
-0.6 0.0
0 5 10 15 20 25 0 5 10 15 20 25
Time (sec) Time (sec)

0.6 1.0 PPT1

ACC1
0.0 0.5

-0.6 0.0
0.4 ACC2 1.0 PPT2

Pore Water Pressure Ratio

0.5
Acceleration (g)

0.0
-0.4 0.0
0.6 ACC3 1.0 PPT3
0.0 0.5
-0.6 0.0
0.6 ACC4 1.0 PPT4
0.0 0.5
-0.6 0.0
0.6 ACC5 1.0 PPT5
0.0 0.5
-0.6 0.0
0 5 10 15 20 25 0 5 10 15 20 25
Time (sec) Time (sec)
Fig. 17. Testing results of gravity model with stone columns: (a) Case 5 and (b) Case 6.

recorded accelerations in the case with stone columns in the
gravity base (ACC1) and tower head (ACC2) were slightly higher
than the ones without using such technique.
4.3.3. Cementation of soil
The injection of cement grout for the purpose of improving
foundation stability has been used in the field for a few decades
now. Although the unit weight may be less than natural soils, the
cemented soil usually has higher strength, lower permeability, and
lower compressibility than the original soil. The Port and Harbor
Research Institute in Japan first employed cement grouting tech-
nique in large scale projects in soft marine soils. Mitchell et al.
[43–45] reviewed the application of cementation at Jackson Lake
Dam as a liquefaction mitigation method and found that the
improved foundation reduced the development of excess pore
water pressure, lateral spreading, and settlement of the dam. The
Hotel and Terminal building at Kobe Harbor survived during the
earthquake in Japan in 1995 because of the 15.8 m deep cementa-
tion improvement. Further studies indicated that no liquefaction
or lateral flow occurred in the foundation soils [9]. The measure-
ments of dynamic properties of cemented soil by tests including
cyclic triaxial, resonant column, and wave propagation methods,
as reported by Saxena et al. [51,52], Fernandez and Santamarina
[22], and Khan et al. [35]. The results demonstrated that artificial
cementation of hazardous sandy soil can significantly mitigate the
seismic-induced damage. In the supplementary tests, the founda-
tion soil in Cases 7 and 8 were prepared by mixing different
proportions of Toyoura sand and Portland cement. The mixed
contents were pluviated into a mold and compacted to a medium
dense state with a relative density of 68% (calculated based on the
properties of uncemented sand). Curing of the sample was
accomplished by submerging in de-aired water for two weeks.
The schematic view of model test is shown in Fig. 18. The wind
turbine model stood right on the cemented sand area. Only the
structural settlement and acceleration were measured during the
earthquake shaking.
In Case 7, the whole depth of the soil underlying the structure
was cemented with a cement ratio of 15% by weight. The cured
mixture was brittle but had good compressive and shear strength,
which generated the least settlement during the strong shaking
(Fig. 19). However, the cemented sand behaved like the dense soil
without cementation and therefore greatly magnified the struc-
tural response in the earthquake. The excessive acceleration in the
tower head ACC2 (Table 5) might have a negative impact on the
performance of the wind turbine.
Based on the research by Khan et al. [35] and Saxena et al. [52],
the modulus of cemented sand is mostly affected by the cement
ratio in addition of the water content and curing time. In Case 8,
the cement–sand mixing ratio was reduced to 5%, and the depth of
the cemented sample was half of that in Case 7 (Fig. 18). Such
attempt did not significantly influence the structural settlement,
which was still at a negligible level. The quantitative relationship
between cement content and compressibility of cemented soil
under dynamic load needs to be further studied in the laboratory.
As shown in Table 5, although the structural response was less
significant in Case 8 than in Case 7, it was still higher than those by
using other techniques.
 H. Yu et al. / Soil Dynamics and Earthquake Engineering 77 (2015) 402–415 413

5. Summaries and conclusions

A group of centrifuge tests was performed on offshore wind

turbine models with two types of foundations, i.e. gravity and
monopile foundations. Comparing the measured structural
responses under earthquake loading, each foundation has its
own advantages and limitations. Tests in dry condition were also
conducted in order to compare the results with those recorded in
saturated conditions. When an earthquake caused soil liquefac-
tion, the models with different foundations exhibited quite differ-
ent behaviors. Some remediation techniques were studied in the
supplementary tests in order to mitigate the seismic-induced
displacement. The test results can be summarized as

(1) Dense dry soil can provide sufficient strength against bearing
failure, but it can also amplify dynamic vibration imposed on
the turbine.
(2) A gravity foundation with large self-weight has good resis-
tance to overturning moments. However, the excessive settle-
ment in the saturated condition may influence the stability of
the wind structure.
(3) Monopile foundations without fixity seem to be quite suscep-
tible to earthquake loading, and can experience quite large
lateral tilt as the result of soil liquefaction. Although a fixed-
end pile foundation can effectively improve the lateral resis-
tance, the amplified response in the tower head may affect the
operation of the wind turbine.
(4) The multi-pile foundation, such as the tripod foundation, is
considered to be an effective alternative to monopile in
resisting the lateral displacement without amplifying the
seismic response at the tower head.
(5) The effects of soil–structure-interaction (SSI) play a significant
Fig. 18. (a) Plan view of cemented sand foundation and (b) schematic view of role in the seismic behavior of offshore wind turbines, as they
centrifuge models for Cases 7 and 8.
are likely to influence the structural settlement, foundation
response, and natural frequency of the system. However, the
observation and analysis of SSI are pretty difficult, especially
0
when there is soil liquefaction.
-5 (6) The results in the supplementary tests demonstrated the
effectiveness of stone columns, densification, and cementation
Settlement (cm)

-10 techniques in reducing seismic-induced settlement. However,

the amplified structural acceleration should be taken into
-15 Case 1
Case 4 consideration in design.
-20 Case 5 (7) The reported study provides useful information in selecting a
Case 6 suitable foundation for offshore wind turbines in seismic active
-25 Case 7 areas. However, for a project in the field, the foundation design
Case 8 should take into account various factors such as geological condi-
-30
tion of the seabed, load conditions, geotechnical techniques, avail-
-35 ability of construction materials, environmental impacts, and cost.
0 5 10 15 20 25
Time (sec)
Fig. 19. Settlements of gravity model in different cases.
Acknowledgment

Table 5 The authors would like to acknowledge the financial support

Comparison of structural responses in different cases (calculated from 2 s to 7.5 s).
from National High-Tech R&D Program of China (863 Program)
Case Techniques ACC1 ACC2 (Research grant 2012AA051702), the National Science Foundation
for Innovative Research Groups (Research grant 51021004), and
Max. (g) RMS (g) Max. (g) RMS (g) the International S&T Cooperation Program of China (ISTCP)
(Research grant 2012DFA70490).
1 N/A 0.306 0.172 0.220 0.098
4 Densification 0.391 0.191 0.244 0.111
5 Stone column 0.288 0.158 0.226 0.096 References
6 Stone column 0.384 0.217 0.196 0.109
7 Cementation 0.490 0.247 0.368 0.164
[1] Adalier K, Elgamal A, Meneses J, Baez JI. Stone column as liquefaction counter-
8 Cementation 0.388 0.184 0.301 0.156
measure in non-plastic silty soils. Soil Dyn Earthq Eng 2003;23:571–84.
414 H. Yu et al. / Soil Dynamics and Earthquake Engineering 77 (2015) 402–415
[2] Adhikari S, Bhattacharya S. Dynamic instability of pile-supported structures in
liquefiable soils during earthquakes. Shock Vib 2008;15:665–85.
[3] Ahmadi MM, Ehsani M. Dynamic analysis of piles in sand based on soil–pile
interaction. In: Proceedings of the 14th world conference on earthquake
engineering. Beijing, China; 2008.
[4] American Society of Civil Engineers. Minimum design loads for buildings and
other structures. ASCE 7-05; 2006.
[5] API. Recommended practice for planning, designing and construction fixed
offshore platforms – load and resistance factor design. American Petroleum
Institute; 1993.
[6] Bea RG, Jin Z, Valle C, Ramos R. Evaluation of reliability of platform pile
foundations. J Geotech Geoenviron Eng 1999;125:696–704.
[7] Boulard J. La forme de radoub prefabriquee no.10 du port militaire de Brest.
Travaux; 1974. p. 17–29.
[8] Brandenberg SJ, Boulanger RW, Kutter BL, Chang D. Static pushover analyses of
pile groups in liquefied and laterally spreading ground in centrifuge tests.
J Geotech Geoenviron Eng 2007;133:1055–66.
[9] Bruce DA. An introduction to the deep soil mixing methods as used in
geotechnical applications. Technical report no. FHWA-RD-99-138; 2000.
[10] Building Seismic Safety Council/National Institute of Building Sciences. NEHRP
recommended provisions for seismic regulations for new buildings and other
structures. FEMA 450; 2004.
[11] Byrne BW, Houlsby GT. Foundations for offshore wind turbines. Philos Trans R
Soc Lond Ser A, Math Phys Eng Sci 2003;361:2909–30.
[12] Chen D, Huang K, Bretel V, Hou L. Comparison of structural properties
between monopile and tripod offshore wind-turbine support structures. Adv
Mech Eng 2013 [Article ID 175684, 9 pp.].
[13] Clough GW, Iwabuchi J, Rad NS, Kuppusamy T. Influence of cementation on
liquefaction of sands. J Geotech Eng 1989:115.
[14] Coelho P, Haigh SK, Gopal Madabhushi SP. Boundary effects in dynamic
centrifuge modelling of liquefaction in sand deposits. In: Proceedings of the
16th ASCE engineering mechanics conference. Seattle: University of Washing-
ton; July 16–18 2003.
[15] Coelho, PALF, Haigh SK, Madabhushi SPG, Brien T. Centrifuge modeling of the
use of densification as a liquefaction resistance measure for bridge founda-
tions. In: Proceedings of the 13th world conference on earthquake engineer-
ing. Vancouver, B.C., Canada; 2004.
[16] Dashti S, Bray JD, Pestana JM, Riemer MR, Wilson D. Mechanisms of
seismically-induced settlement of buildings with shallow foundations on
liquefiable soil. J Geotech Geoenviron Eng 2010;136:151–64.
[17] Dashti S, Bray JD, Pestana JM, Riemer M, Wilson D. Centrifuge testing to
evaluate and mitigate liquefaction-induced building settlement mechanisms.
J Geotech Geoenviron Eng 2010;136:918–29.
[18] DIN EN ISO 19901-6. Petroleum and natural gas industries – specific require-
ments for offshore structures – Part 6: marine operations. English version
prEN ISO 19901-6:2006.
[19] Dobry R, Abdoun, T. Recent studies on seismic centrifuge modeling of
liqefaction and its effect on deep foundations. SOAP-3. In: Prakash S, editor.
Proceedings of the 4th international conference on recent advances in
geotechnical earthquake engineering and soil dynamics. San Deigo, CA; 2001.
[20] Dupas JM, Pecker A. Static and dynamic properties of cement. J Geotech Eng
1979;105:419–36.
[21] Elgamal A, Yang Z, Adalier K, Sharp MK. Effect of rigid container size on
dynamic earth dam response in centrifuge experiments. In: Proceedings of the
16th ASCE engineering mechanics conference. Seattle, WA: University of
Washington; July 16–18 2003.
[22] Fernandez A, Santamarina JC. The effect of cementation on the small strain
parameters of sands. Can Geotech J 2001;38:191–9.
[23] Figueroa JL, Saada AS, Dief H. In: Kimura T, Kusakabe O, Takemura J, editors.
Development of the geotechnical centrifuge at Case Western Reserve Uni-
versity Centrifuge 98, 1; 1998. p. 3–8.
[24] Finn WDL, Steedman RS, Yogendrakumar M. Seismic response of gravity
structures in a centrifuge. In: Proceedings of the 17th annual offshore
technology conference. Houston, Texas; 6–9May 1985.
[25] Frydman S, Shaal B, Baker R, Steinbach J, Horn H, Hendron D. Liquefaction
study of cemented sands. J Geotech Eng 1980;106:275–97.
[26] Gambin MP. Menard dynamic compaction, a new method for improving
foundation beds off-shore. In: Proceedings of the international symposium.
Brugge; 1982. p. 3.91–93.95.
[27] Gaudin C, Cluckey EC, Garnier J, Phillips R. New frontiers for centrifuge
modelling in offshore geotechnics. In: Proceedings of the 2nd international
symposium on frontiers in offshore geotechnics. Perth, Australia; 2010. p. 155–
88.
[28] Ghosh B, Madabhushi SPG. Centrifuge modelling of seismic soil structure
interaction effects. Nucl Eng Des 2007;237:887–96.
[29] Gourvenec S, Acosta-Martinez HE, Randolph MF. Centrifuge model testing of
skirted foundations for offshore oil and gas facilities. In: Proceedings of the
international conference on offshore site investigation and geotechnics.
London, UK; 2007.
[30] ICC. International Building Code 2006. International Code Council. IL, USA:
Country Club Hills; 2006.
[31] Iwabuchi J. Influence of cementation on liquefaction resistance of sands [Ph.D.
thesis]. VA: Virginia Polytechnic Institute and State University; 1986.
[32] Jakrapiyanun W. Physical modeling od dynamic soil-foundation-structure-
interaction using a laminar container [Ph.D. thesis]. San Diego: Department of
Structural Engineering, University of California; 2002.
[33] Kagawa T. Effects of liquefaction on lateral pile responses. Piles under
Dynamic Loads. . ASCE Geotechnical Special Publication; 1992. p. 207–23.
[34] Keller Foundations. Wind turbine capability. Keller Windfarm Capability
brochure. 〈http://issuu.com/kelleruk/docs/keller_wind_turbine_doc〉.
[35] Khan Z, Majid A, Cascante G, Hutchinson DJ, Pezeshkpour P. Characterization
of a cemented sand with the pulse-velocity method. Can Geotech J
2006;43:294–309.
[36] Lavassas I, Nikolaidis G, Zervas P, Efthimiou E, Doudoumis IN, Baniotopoulos
CC. Analysis and design of the prototype of a steel 1-MW wind turbine tower.
Eng Struct 2003;25:1097–106.
[37] Lee KL, Focht JA. Liquefaction potential at Ekofisk Tank in North Sea. J Geotech
Eng Division 1975;101(GT1):1–18.
[38] Liu L, Dobry R. Seismic response of shallow foundation on liquefiable sand. J
Geotech Geoenviron Eng 1997;123:557–67.
[39] Malhotra S. Design and construction considerations for offshore wind turbine
foundations in North America. Geotech Spec Publ 2010;2:1533–42.
[40] Mao CQ, Li YD, Yu H, Tao J. Numerical simulation of single pile dynamic
response and parameter analysis under horizontal seismic. J Hunan Inst Eng
2007;17:83–7.
[41] McClelland B. Design of deep penetration piles for ocean structures. J Geotech
Eng 1974;100:709–47.
[42] Menard L. La consolidation dynamique comme solution aux problemes de
fondation: pour la construction de quais, terminaux, reservoirs de stockage et
iles artificielles sur sols compressibles. In: Proceedings of the 7th international
Harbour conference. Antwerp; 1978.
[43] Mitchell JK, Wentz FJ. Performance of improved ground during the Loma
Prieta Earthquake University of California. Berkeley UCB/EERC Report 91/12;
1991.
[44] Mitchell JK, Baxter CDP, Munson TC. Performance of improved ground during
earthquakes. Soil improvement for earthquake hazard mitigation. ASCE
Geotech Spec Publ 1995;49:1–36.
[45] Mitchell JK, Cooke HG, Schaeffer J. Design considerations in ground improve-
ment for seismic risk mitigation. In: Proceedings of the specialty conference
on geotechnical earthquake engineering and soil dynamics III. ASCE Geotech-
nical Publication. Seattle, Washington: vol. 1; 1998. p. 580–613.
[46] Murff JD The geotechnical centrifuge in offshore engineering. In: Proceedings
of offshore technology conference. Houston; 1996. OTC 8265.
[47] Ntambakwa E, Rogers M. Wind turbine structures, dynamics, loads and
control. AWEA Seismic Forces for Wind Turbine Foundations final paper;
2009.
[48] Prowell I, Veletzos M, Elgamal A. Full scale testing for investigation of wind
turbine seismic response. In: Proceedings of the 7th world wind energy
conference. Kingston, Ontario, Canada; 2010.
[49] Sadrekarimi A, Ghalandarzadeh A. Evaluation of gravel drains and compacted
sand piles in mitigating liquefaction. Ground Improv 2005;9:91–104.
[50] Sasaki Y, Taniguchi E. Shaking table tests on gravel drains to prevent
liquefaction of sand deposits. Soils Found 1982;22:1–14.
[51] Saxena SK, Avramidis A, Reddy KR. Dynamic moduli and damping ratios for
cemented sands at low strains. Can Geotech J 1998;25:353–68.
[52] Saxena SK, Reddy KR, Avramidis A. Liquefaction resistance of artificially
cemented sand. J Geotech Eng Division 1998;114:1395–413.
[53] Schofield AN. Cambridge geotechnical centrifuge operations. Geotechnique
1980;30:227–68.
[54] Seed HB, Booker JR. Stabilization of potentially liquefiable sand deposits using
gravel drains. ASCE J Geotech Eng Division 1977;103:757–68.
[55] Srbulov M. Geotechnical earthquake engineering: simplified analyses with
case studies and examples. Geotech Geol Earthq Eng 2008;9:262.
[56] Takahashi H, Morikawa Y, Ichikawa E. Effects of rigid sidewall of specimen
container on seismic behavior. In: Proceedings of the 7th international
conference on physical modelling in geotechnics (ICPMG 2010). Zurich,
Switzerland; 28th June–1st July 2010. p. 177–82.
[57] Tokimatsu K, Mizuno H, Kakaurai M. Building damage associated with
geotechnical problems. Soil Found 1996 [Special Issue of Geotechnical Aspects
of the 1995 Hyogoken Nanbu Earthquake].
[58] Tokimatsu K, Asaka Y. Effect of liquefaction-induced ground displacements on
pile performance in the 1995 Hyogeken-Nambu earthquake. Soil Found
1998:163–77 [Special Issue].
[59] Tokimatsu K, Suzuki H, Sato M. Effects of inertial and kinematic interaction on
seismic behavior of the pile with embedded foundation. Soil Dyn Earthq Eng
2005;25:753–62.
[60] Vibroflotation Group. Stone columns: offshore stone columns. 〈http://www.
vibroflotation.com/Vibro/vibroflotation_fr.nsf/site/Stone-Columns.
Offshore-Stone-Columns〉.
[61] Walker S, Blair-Fish P. Piled vs gravity offshore structures for seismically active
arctic regions. In: Proceedings of the conference. Energy sources technology
conference and exhibition. Dallas: ASME; 1987.
[62] Wilson DW. Soil–pile–superstructure interaction in liquefying sand and soft
clay [Ph.D. thesis]. CA: Department of Civil & Environmental Engineering,
College of Engineering, University of California at Davis; 1998.
 H. Yu et al. / Soil Dynamics and Earthquake Engineering 77 (2015) 402–415 415

[63] Yang Z, Elgamal A, Adalier K, Sharp M. Container boundary effect on seismic
earth dam response in centrifuge model tests. In: Proceedings of the 11th
international conference on soil dynamics & earthquake engineering and 3rd
international conference on earthquake geotechnical engineering. Berkeley:
University of California; 2004.
[64] Yoshida N, Hamada H. Damage to foundation piles and deformation pattern of
ground due to liquefaction-induced permanent ground deformations. In:
Proceedings of the 3rd Japan–U.S. workshop on earthquake resistant design
of lifeline facilities and countermeasures for soil liquefaction, Technical Report
NCEER 91-0001. Buffalo, NY: NCEER; 1991. p. 147–61.
[65] Zhao X, Maisser P. Seismic response analysis of wind turbine towers including
soil-structure interaction. Proc Inst Mech Eng Part K: J Multi-body Dyn
2006;220:53–61.
[66] Zhang L. Nonlinear analysis of laterally loaded rigid piles in cohesionless soil.
Comput Geotech 2009;36:718–24.