You are on page 1of 16

Soil Dynamics and Earthquake Engineering 142 (2021) 106554

Contents lists available at ScienceDirect

Soil Dynamics and Earthquake Engineering


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

Investigating the seismic isolation effect of the cushioned pile raft


foundation in soft clay through dynamic centrifuge tests
Fayun Liang a, *, Tongda Li a, Yu Qian b, Chen Wang a, Yajie Jia a
a
Department of Geotechnical Engineering, Tongji University, Shanghai, 200092, China
b
Department of Civil and Environmental Engineering, University of South Carolina, Columbia, SC, USA

A R T I C L E I N F O A B S T R A C T

Keywords: Cushioned pile-raft foundation (Cushioned-PR) is a new type of foundation for sea-crossing bridges in deep water
Centrifuge test (more than 50 m in depth). Cushioned-PR is able to tolerate a certain amount of slide that occurs between the raft
Cushioned pile raft foundation system and the piles under strong earthquakes. This characteristic could reduce the seismic motion transmitted to the
Connected pile raft foundation system
superstructure to maintain the integrity of the bridge. Thus, Cushioned-PR has become popular for bridges under
Seismic load
Ground motion
the threat of strong earthquakes. However, the dynamic responses of the Cushioned-PR under various earthquake
Soil-structure interaction intensities in soft clay have not been thoroughly understood. It is essential to better evaluate the isolation effect
of Cushioned-PR to improve the design. In this study, the non-linear behavior of Cushioned-PR and the isolation
effect of its interposed layer are evaluated using centrifuge tests. A series of dynamic centrifuge tests are per­
formed with a 3 × 3 pile group foundation embedded in soft clay under different earthquake motions. Both the
cushioned pile raft foundation system (Cushioned-PR) and the connected pile raft foundation system (Connected-
PR) are tested in the centrifuge tests. Acceleration and residual displacement of the super-structure, and bending
moment of columns and piles, are monitored during the experiments. The results show that the interposed layer
in Cushioned-PR has a significant impact on the bending moment and residual displacement of the pile-raft
system. Based on the observation from this study, the maximum bending moment of the column in the
Cushioned-PR is around 28%–54% of the values from the Connected-PR. On the other side, the maximum
horizontal displacement of Cushioned-PR is about 2–10 times larger than that of the Connected-PR. The results
show that Cushioned-PR can effectively reduce the seismic excitation transmitted to the super-structure if the
seismic intensity is high. However, the benefit of Cushioned-PR is marginal under low-intensity earthquakes.

1. Introduction [24]. The behavior of Connected-PR has been well documented.


Although Connected-PR is a good solution in many cases, it has very
With great demands of transportation infrastructure due to economic limited tolerance of deformation. If the deformation exceeds a certain
growth, many bridges are under construction or planning, especially limit, Connected-PR could lead to catastrophic failure, for example, the
those wide crossing rivers or seas [21]. Generally, the connected pile raft accident of the Kobe line of Hyogoken-Nambu Earthquake [31].
foundation (Connected-PR) is recognized as an economical and sus­ In the seismic zones, earthquake is the major natural hazard leading
tainable foundation form and is widely used due to its promising per­ to structural damage [2,4,12], where Connected-PR may not always be
formance [8,11,16,20,25,28,36,37,39]. This is because piles below the the best choice. A new foundation type has been proposed to overcome
raft foundation can reduce settlement, which not only solves the prob­ the deficiency of Connected-PR [7,35,45], whose piles are disconnected
lem of bearing deformation but also improves the stability and from the raft, namely the cushioned pile raft foundation system (Cush­
serviceability of a bridge [22]. Analytical methods have been proposed ioned-PR). For Cushioned-PR, a cushion, typically a layer of granular
to calculate the bearing capacity and deformation of Connected-PR [6, materials, is located between the piles and the raft. The cushion can also
28,40]. Further, centrifuge testing is performed to study the response of be referred as the interposed layer. When a bridge with Cushioned-PR
Connected-PR under pseudo-static and dynamic loading conditions encounter an earthquake, the cushion is able to tolerate a certain

* Corresponding author.
E-mail addresses: fyliang@tongji.edu.cn (F. Liang), 1810188@tongji.edu.cn (T. Li), yuqian@sc.edu (Y. Qian), cwang33@tongji.edu.cn (C. Wang),
happyvsmorehappy@126.com (Y. Jia).

https://doi.org/10.1016/j.soildyn.2020.106554
Received 5 July 2020; Received in revised form 15 December 2020; Accepted 15 December 2020
Available online 25 December 2020
0267-7261/© 2020 Elsevier Ltd. All rights reserved.
F. Liang et al. Soil Dynamics and Earthquake Engineering 142 (2021) 106554

amount of slide occurring between the raft and the piles under strong centrifugal acceleration, and the actual problem is simulated by using
earthquakes. This way, Cushioned-PR can effectively reduce the seismic the scale model [47,49]. In this study, centrifuge tests are performed on
motion transmitted to the super-structure to maintain the integrity of the centrifuge device (TLJ-150) in the Geotechnical Centrifuge Model­
the bridge [15,17,41]. ling Laboratory at Tongji University [29], as shown in Fig. 1. This device
Cushioned-PR is firstly applied in the Rion-Antirion bridge and has is able to reach a maximum centrifugal acceleration of 50 g, and a
quickly become popular [23,33,34,45]. There has been a considerable maximum vibration acceleration of 20 g. A customized laminar shear
amount of studies focusing on the performance of Cushioned-PR under box, shown in Fig. 1, has a rubber liner inside to absorb possible
static loadings. Previous researches suggest that the interposed layer reflecting waves, which can better represent the boundary conditions in
could improve the overall performance and influence the static and the field. The laminar shear box has the dimensions of 500 mm in length,
dynamic characteristics of foundations [3,14,19,27,38,48]. Also, nu­ 400 mm in width, and 550 mm in height, stacked by 22 aluminum rings.
merical simulations of the complex interactions between the compo­ Each ring has a thickness of 24 mm. Sliders are arranged between each
nents (raft, piles, and soil) of Cushioned-PR indicate the connection form two adjacent rings, and the relative displacement between every two
of Cushioned-PR benefits the structure stability under dynamic load [43, adjacent rings is up to 5 mm.
44,46]. Previous studies have investigated the bearing capacity and
settlement behavior of Cushioned-PR. Dynamic responses of
2.2. Soil sample preparation
Cushioned-PR are also studied by both experimental and numerical
approaches [1,17,32]. Unfortunately, the vibration isolation effect of
The soil used in this study is 100% Kaolinite type of clay material.
Cushioned-PR, especially Cushioned-PR in soft clay under dynamic
The following procedures are executed for the sample preparation:
loadings, has not been thoroughly understood [18].
This study aims to evaluate the vibration isolation effect of
1. The Kaolin clay powder is mixed with water in a 1:1 ratio by mass to
Cushioned-PR and its dynamic responses during seismic excitations in
prepare clay slurry.
soft clay. A series of centrifuge tests are performed to investigate the
2. Before pouring the clay slurry into the laminar shear box, one layer
dynamic response of both Cushioned-PR and Connected-PR in soft clay
of 2 cm-thick coarse sand is uniformly laid at the bottom of the
under different input motions. Parameters include acceleration, strain,
laminar shear box as a permeable layer for drainage. This layer will
and displacement are quantified and compared between the two types of
prevent excess moisture accumulates at the bottom of the clay layer
foundation. The influences of the different peak base acceleration (PBA)
in the subsequent steps to maintain the soft clay layer to be uniform.
and the earthquake waves are also analyzed through the Fourier trans­
3. The slurry is conditioned by applying a preloaded of 1 g for 24 h.
form and response-spectrum.
Then the soil sample is subjected to centrifuge consolidation under
the maximum centrifugal acceleration of 50 g, which typically
2. Laboratory experiment, equipment, and materials
required 24 h s or longer of continuous spinning. The soil sample
properties are checked by common soil property tests, including the
2.1. Centrifuge testing system
Proctor compaction test, Atterberg limit tests, consolidation test, and
triaxial shear test. Table 1 provides the soil properties. As shown in
For centrifuge tests, the gravity field is simulated by increasing the
Fig. 2, the measured soil strength form the T-bar readings increases

Fig. 1. Centrifuge facility.

2
F. Liang et al. Soil Dynamics and Earthquake Engineering 142 (2021) 106554

Table 1 respectively. The total mass of the super-structure together with the
Basic properties of the Kaolinite type clay used in this study. foundation is approximately 381,018 kg for the prototype, in which the
Property Average value/Range mass of the pile cap is approximately 198,768 kg. For Cushioned-PR, the
3 piles were not connected to the raft. The average contact pressure
Bulk unit weight kN (m ) 17.2
Water content: % 63.4 imposed by the super-structure to the foundation was about 46 kN/m2,
Liquid limit: % 73.5 based on the weight of the structure (381,018 kg) and the contact area
Plastic limit: % 43.5 (81 m2). The prototype dimensions of the pile are 0.52 m-diameter, 13.5
Compression index 0.558 m-long, nine piles in a 3 × 3 arrangement are embedded in soft clay.
Recompression index 0.12
Stiffness (MPa) 26
Dimensions of the laboratory model are also shown in Figs. 3 and 4. The
Effective frictional angle (◦ ) 25 natural period of a bridge is typically within the range of 0.4 s–2 s [15,
16]. The primary period of the super-structure in this study is approxi­
mately 0.5 s. Other relevant parameters of the physical model and the
similarity information are shown in Table 3 and Table 4.
Strain gauges are attached to piles to measure the strain variations
under excitations. For Connected-PR, the piles are rigidly connected to
the raft, both the free field movement of the soil and the inertia of the
super-structure have a great impact on the response of piles under vi­
bration. For Cushioned-PR, the strain variation of the pile during an
earthquake is more influenced by the free field movement of the soil.
Note the strain gauges are installed evenly along the pile shaft in the
form of the full Wheatstone bridge. The bending moment of the pile at
different depth during an earthquake can be calculated from the strain as
below:
EI
Mp = − εp (4)
Fig. 2. Undrained shear strength of kaolin clay measured using T-bar. y

approximately linearly with depth. The regression shows the where Mp is the bending moment of the pile, EI is rotational stiffness, εp
following relationship between shear strength Su and depth under is the maximum bending strain at the outer fiber of the pile cross-
the normal consolidation condition: section, y is the distance from the neutral axis to farthest fiber in the
cross-section. Similar approaches are applied in calculating the bending
Su = 1.81d (1) moment of pile, beam, or other structural components in the literature
According to Table 1, the effective unit weight of the soil sample is [13,30]. As shown in Fig. 3, the prototype scale of each centrifuge
7.2, then the effective vertical stress can be calculated by Equation (2) sample had the dimensions of 25 m in length, 20 m in width, and 20 m in
height. Fig. 4(a) shows the exact places where strain gauges were
(2) installed. Fig. 4(b) depicts the plan view of the raft. The nine piles are

σ v = 7.2d
placed in 3 × 3 array with the center-to-center distance of five times of
So the average ration of the undrained shear strength to the effective
the pile diameter. The minimum space between piles and the edge of the
vertical stress is 0.25, shown in Equation (3).
laminar shear box is 9 m on the prototype scale.
/ ′
Su σ v = 0.25 (3) For Connected-PR, super-structure, column, raft, and piles are rigidly
assembled together. After assembling, the model is gently pressed into
The same soil is used in all the tests in order to make a fair com­ the soft clay to the designated depth. Extra caution is excised during the
parison, avoiding possible variation due to the interactions between the pressing step to ensure the model maintains upright and avoids over
piles and different soil. disturbing the soil. The driven angle of the model is checked frequently
during the pressing stage. While for Cushioned-PR, only piles are gently
2.3. Experiment model preparation pressed into the soft clay to the designated depth. Then, extra soil above
the pile is removed. Gravels are placed above the piles and carefully
The foundation systems are designed to be single-degree-of-freedom tapped to form the interposed layer, in other words, the cushion. For
structures for this study. For all the tests, the centrifugal acceleration is Rion-Antirion Bridge, the grain size distribution of the gravel layer is
50 g consistently, and the similarity ratio of the foundation between the from 10 mm to 80 mm [34]. According to the similarity ratio, sieves with
model used in the experiment and the prototype structure is 1:50. an aperture size of 0.25 mm and 2 mm are used to obtain the proper
Table 2 shows the similitude ratio of the relevant physical quantities. granular materials. The gravels used to form the interposed layer ranges
Schematics of the sample, the dimensions of other components, and the from 0.25 mm to 1.5 mm in this study. The thickness of the interposed
sensors used in the laboratory test are shown in Fig. 3 and Fig. 4. The layer is 12 mm as shown in Fig. 3 and the interposed layer is compacted
dimensions of the super-structure and the raft are 4.3 m × 4.3 m × 4.3 m by a stone plate to achieve a target density of 1.75 g/cm3 [51]. The
and 9 m × 9 m × 0.9 m (length × width × height) for the prototype, density is double-checked before any test. Typically, the interposed
layer should be stiffer than the clay around the pile [17]. As shown in
Table 2 Table 1, the stiffness of the soil here is 26 MPa. The stiffness of the
Similitude relationships. interposed layer used in this test is 56 MPa, larger than that of the soil.
Physical quantity The ratio of similitude After preparing the interposed layer, the super-structure, column, and
raft are assembled together to put on top of the interposed layer.
Dynamic property time 1/50
frequency 50
acceleration 50
displace 1/50 2.4. Centrifuge test and input motions
force 1/(502)
strain 1 After the above steps, the laminar shear box is mounted on the
stress 1
shaking table. The centrifuge frame is then spun up to 50 g for about 1 h

3
F. Liang et al. Soil Dynamics and Earthquake Engineering 142 (2021) 106554

Fig. 3. Schematics of the centrifuge test specimen setup: (a) Cushioned-PR system model; (b) Connected-PR system model.

to completely dissipate the excess pore pressure. The inputted motions recorded. If the test results were similar to the measured data of the last
and the centrifuge tests are summarized in Table 5. test under white noise, the test conditions were stable. The tests were
White noise is a continuous sound caused by many frequencies of continuing. Fig. 5 shows the ratio of acceleration Fourier spectrum be­
equal intensity [26,50]. The power spectral density of white noise is tween the soil bottom and the shaking table under white noise excita­
consistent in the frequency domain. For its special characteristics, white tion. By Fourier transform the relationship between the input signal X(ω)
noise is often used to analyze the resonance frequency of soil samples and the output signal Y(ω) can be expressed as follow:
and structures in a dynamic centrifuge test. The PBA of white noise in ( )
− mω2 + icω + k Y(ω) = X(ω) (5)
this test is 0.05 g. To ensure the stability and reliability of the test data,
before any experiment with a seismic loading, white noise is used to Or can expressed as follow:
check the system. The acceleration, horizontal displacement, vertical
settlement, pore pressure, and the strain of the model are measured and Y(ω) = H(ω)X(ω) (6)

4
F. Liang et al. Soil Dynamics and Earthquake Engineering 142 (2021) 106554

Fig. 4. Pile and raft model.

Table 3 Table 4
The similitude relations of superstructure (S) and raft (R). The similitude relations of pile and column.
Physical quantity Model(S) Prototype Model(R) Prototype Physical quantity Model Prototype Model Prototype
(S) (R) (Pile) (pile) (Column) (Column)

Material steel concrete aluminium concrete Material aluminium concrete aluminium concrete
Modulus of elasticity 200 34.5 70 34.5 length L(m) 0.25 12.5 0.03 1.5
(GPa) Modulus of 200 34.5 70 34.5
Poisson’s ratio 0.25 0.2 0.33 0.2 elasticity (GPa)
Density ρ(kg/m3) 7800 2500 2700 2500 Poisson’s ratio 0.25 0.2 0.33 0.2
Size L(m) (0.086)3 (4.3)3 (1.8)2 × (9)2 × 0.9 Diameter D(m) 0.01 0.5 0.01 0.5
0.018 Thickness t(m) 0.001 0.021 0.002 0.028
Quality m (kg) 2.142075 198767.5 1.57464 182250 Modulus E (Pa) 7.00E+10 2.06E+11 7.00E+10 2.06E+11
Mass similarity ratio 1 45.27 1 48.73 Flexural rigidity 36.89 2.11E+08 29.91 2.39E+08
Lateral moment of 1.51E-03 612535.18 4.294043E- 1242489.38 EI(N⋅m2)
inertia J (kg⋅m2) 03 The similarity 1 48.93 1 53.17
The similarity ratio of 1 52.69 1 49.24 ratio of EI
J Compressive 2.42E+06 6.78E+09 3.52E+06 8.55E+09
rigidity EA(N)
The similarity 1 52.95 1 49.30
Take ω as a variable, H(ω) is called the transfer function. Based on ratio of EA
the transfer function, the resonant frequencies of the SODF can be Pile spacing s(m) 0.06 3 – –
The similarity 1 50 1 50
calculated. For other vibration systems, the resonant frequencies can be ratio of s
calculated through H(ω) like SODF.
In this test, vibrations in the range of 0.4–4 Hz appear to be accurate
enough. In this range, the ratio maintains a steady value of 1. This in­
Table 5
dicates that the acceleration of the soil base is consistent with the ac­
Test cases.
celeration of the shaking table. Thus, the shaking table acceleration can
be considered as the input acceleration of the soil base. The model is pre- Test Foundation Imputed motion
case form
excited several times until the data recorded by sensors remains stable.
After the trials, the system is restored to its initial state, excited again T-1 Cushioned- Earthquake White Acce100 Mexico- El-
PR motion noise City Centro
with the selected seismic wave. The steps are repeated for all tests. The
PBA (g) 0.05 0.1, 0.2, 0.1 0.1,
period between any two tests is kept to be at least 3 h to guarantee the 0.3 0.2,
excess pore water pressure is fully dissipated. 0.3
To analyze the soil behavior under earthquakes, the resonance fre­ T-2 Connected- Earthquake White Acce100 Mexico- El-
quency of the soil is calculated. Fig. 6 presents the transfer function of PR motion noise City Centro
PBA (g) 0.05 0.1, 0.2, 0.1 0.1,
the soil under white noise excitation. The accelerometer (A5) placed at 0.3 0.2,
the bottom of the soil layer is used to record the accelerations (as the 0.3
system input), the accelerometer (A4) placed at the surface of the soil

5
F. Liang et al. Soil Dynamics and Earthquake Engineering 142 (2021) 106554

Fig. 5. The ration of acceleration Fourier spectrum between soil bottom and
shaking table.

Fig. 6. The transfer function of soil bottom to soil surface under white
noise excitation.

layer is used to record the accelerations (as the system output), as shown
in Fig. 3. The transfer functions of the soil, which are used to compare
Cushioned-PR and Connected-PR are calculated. The resonant frequency
of the soil with Cushioned-PR is 1.05 Hz. While the resonant frequency
of the soil with Connected-PR is 1.07 Hz. The peak value of the accel­
eration transfer function for the soil sample is about 9.2. This shows that
the resonant frequencies of the soil for the two foundation setups are
very close to each other. Hence, the tests with different foundations are
comparable.
Fig. 7 gives the time-history of the seismic waves selected for this
study: the ACCe100 wave, the El-Centro wave, and the Mexico City
wave. Both the frequency and the peak base acceleration (PBA) of the
earthquake waves would influence the performance of the soil- Fig. 7. The earthquake acceleration time histories with a peak acceleration of
foundation-structure system. The El-Centro wave has a higher amount 0.1 g: (a) the Acce100 wave; (b) the El-Centro wave; (c) the Mexico city wave.
of energy in the early period, while the Mexico City wave has a higher
amount of energy in the late period. The ACCe100 wave is a synthetic
seismic wave that can well characterize the seismic engineering prop­
erties in the coastal areas of China. Fig. 8 showed the response spectra of
earthquake motions when the PBA was 0.1 g. The dominant period at
which the peak response occurs was 0.58 s for the El-Centro wave and
2.02 s for the Mexico City wave. For the ACCe100 wave, the normalized
response spectrum has a platform value of 0.3 g, and the corresponding
characteristic period is 1.0 s.
Before every official test, the setup is carefully calibrated and tuned
through a trial and error process to ensure the seismic waves applied is
close enough to the desired waves. The centrifuge tests performed in this
study consist of three variables, the connection form between piles and
the raft (rigid connected or disconnected), the types of ground motions,
and the ground motion intensities. Note all the piles are friction piles in
this study. The results from the experiments are discussed in the Fig. 8. The response spectrums of the ideal motions inputted to the control
following sections. system of the shake table (Damping ratio: 5%).

6
F. Liang et al. Soil Dynamics and Earthquake Engineering 142 (2021) 106554

3. Test results and discussions Table 6


The peak acceleration of superstructure under different motions.
In this section, the dynamic responses of Cushioned-PR and Foundation Input Peak positive acceleration of superstructure
Connected-PR are analyzed and compared, including acceleration, form motion
Small seismic Moderate Strong
bending moment, horizontal displacement, and vertical settlement. (0.1 g) seismic (0.2 g) seismic (0.3
g)

3.1. Dynamic responses of the super-structure Cushioned-PR ACCe100 0.41 g 0.54 g 0.63 g
El-Centro 0.28 g 0.37 g 0.55 g
Mexico 0.49 g
To evaluate the seismic resistance of the foundations, it is necessary City
to compare the dynamic responses of the super-structure. Fig. 9 shows Connected-PR ACCe100 0.33 g 0.76 g 0.89 g
the measured acceleration time history of the super-structure. The peak El-Centro 0.22 g 0.50 g 0.80 g
positive accelerations of the super-structure under different motions are Mexico 0.55 g
City
summarized in Table 6. Taking the dynamic responses of Connected-PR
and Cushioned-PR under the action of the El-Centro earthquake waves
as an example, the difference increases progressively with the increasing PBA was 0.3 g, the peak value of the positive acceleration was 0.55 g for
of PBA. When the PBA is 0.1 g, the peak value of the positive acceler­ Cushioned-PR, while 0.80 g for Connected-PR. The isolation effect is
ation was 0.28 g for Cushioned-PR, while 0.22 g for Connected-PR. negligible for Cushioned-PR when the PBA at a low level, such as 0.1 g.
When the PBA was 0.2 g, the peak value of the positive acceleration The peak value of the positive acceleration of Cushioned-PR is even
was 0.37 g for Cushioned-PR, while 0.50 g for Connected-PR. When the

Fig. 9. The acceleration time history of superstructure.

7
F. Liang et al. Soil Dynamics and Earthquake Engineering 142 (2021) 106554

higher than the value of Connected-PR when the PBA is 0.1 g. However, Table 7
when the PBA is 0.3 g, the peak acceleration of Cushioned-PR is about The frequency corresponding to the peak value of the Fourier spectrum of the
68%~71% of the value of Connected-PR. Hence, Connected-PR is superstructure.
preferred for places where the anticipated seismic intensity is low. While Foundation Input The frequency corresponding to the peak value (Hz)
the Cushioned-PR would be a better choice for places where the antic­ form motion
Small seismic Moderate Strong
ipated seismic intensity is high. (0.1 g) eismic (0.2 g) seismic (0.3
Fig. 10 shows the Fourier spectrum of the super-structure under the g)
ACCe100 waves. The frequency corresponding to the peak value of the Cushioned-PR ACCe100 0.64 0.64 0.51
Fourier spectrum is summarized in Table 7. The dynamic response of the 1.28 0.92 0.64
structure is not only related to the amplitude of the ground acceleration 1.46 0.96 0.91
but also the spectral characteristics of the ground motion. Through Connected-PR ACCe100 0.64 0.64 0.64
1.39 0.83 0.92
Fourier transform, the earthquake data can be transformed from the 1.46 0.92 0.96
time domain to the frequency domain. Therefore, the spectrum infor­
mation of the ground motion can be obtained. Where T represents the
time duration of acceleration. ω represents the response frequency of the
super-structure, t represents the time variable. The amplitude spectrum
FS(ω) can be expressed as below:

Fig. 10. The Fourier amplitude spectrums of superstructure.

8
F. Liang et al. Soil Dynamics and Earthquake Engineering 142 (2021) 106554

√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
√⎡ T ⎤2 ⎡ T ⎤2 of the structure remained the same in the case of Connected-PR, which is
√ ∫ ∫
√ about 1.62 Hz. As shown in Fig. 13(b), the frequency corresponding to
FS(ω) = √⎣ a(t)sinωtdt⎦ + ⎣ a(t)cosωtdt⎦ (7)
the natural vibration period of the structure is around 1.65 Hz, 1.55 Hz,
and 1.44 Hz, with the peak acceleration of 0.1 g, 0.2 g, and 0.3 g,
0 0

where a(t)sinωt and a(t)cosωt represents the imaginary part and the real respectively. Compared with Figs. 6 and 13 shows there are clear dif­
part of the finite Fourier transform F(ω), respectively. ferences between the resonant frequencies of the structures. Note the
Comparing the Fourier amplitude spectrum of Connected-PR and soil samples would not resonate with the models. Observed from the
Cushioned-PR under the motion of ACCe100 as an example, when the experiments, the frequency corresponding to the natural vibration
PBA of the inputted motion is 0.1 g, the peak Fourier amplitudes of two period of the structure decreases with the increases of PBA. When
foundations are almost the same. When the PBA is 0.2 g and 0.3 g, the bridges encounter a strong earthquake, there may be a considerable
peak Fourier amplitude of Connected-PR is significantly larger than that amount of slippage between the cushion and the foundation. This would
of Cushioned-PR. The spectrum characteristics of ground motions are change the stiffness of the structure. For Cushioned-PR, the frequency
the same for the two foundations. The seismic acceleration a (t) trans­ corresponding to the natural vibration period of the structure would
mitted from the foundation causes the difference. For Cushioned-PR, the change when the structure is subjected to a strong earthquake. The
interposed layer located between the raft and the piles reduces the changes in the structural stiffness would lean to the nonlinear behavior
seismic load transmitted to the super-structure, which facilitates motion of the structure.
isolation.
Fig. 11 shows the influence of the PBA on the measured peak super- 3.2. The bending moment of the column
structure acceleration (PSA) with Cushioned-PR and Connected-PR
under different waves. The PSA increases rapidly when using The inertial loading has a significant influence on the responses of a
Connected-PR, compared with Cushioned-PR. Fig. 11 suggests that both system under dynamic loadings. By analyzing the changes in the
the wave intensity and the foundation type would affect the PSA values. bending moment of the column, the isolation effect of Cushioned-PR and
The response spectrum reveals the relationship between the dynamic Connected-PR can be compared. The peak positive bending moment
characteristics of the structure and the ground motions. It is used to values of the column are summarized in Table 8. The values obtained
calculate the internal force and the deformation of the structure under from the two types of foundations under the action of the El-Centro wave
an earthquake, as shown in Fig. 12. When the ACCe100 wave is applied are analyzed. During a small earthquake (i.e. PBA less than 0.1 g), the
with a PBA of 0.1 g, the peak values of the spectral acceleration of the ratio of peak bending moment of Cushioned-PR to the value of
raft are almost the same for the two types of foundations. When the Connected-PR is 28%. During a moderate earthquake (i.e. PBA is 0.2 g),
ACCe100 wave and the El-Centro wave are applied with a peak accel­ the ratio of peak bending moment of Cushioned-PR to the value of
eration of 0.2 g and 0.3 g, respectively, the peak value of the spectral Connected-PR is 54%. While during a strong earthquake (i.e., PBA is 0.3
acceleration of the raft for Cushioned-PR is smaller than that of g), the ratio of peak bending moment of Cushioned-PR to the value of
Connected-PR. This demonstrates that the peak internal force and Connected-PR is 47%. The results confirmed that Cushioned-PR per­
deformation of Cushioned-PR under stronger earthquake motions are forms better than Connected-PR in reducing the seismic impact on the
generally smaller than that of Connected-PR, indicating a better seismic structures.
performance of Cushioned-PR. However, exceptions can also be
observed, such as shown in Fig. 12(d) and (e). 3.3. The vertical load of piles during the earthquake
Structures would anticipate certain damages when an earthquake
happens. This may change the dynamic characteristics of the structures. In the traditional pile foundation design approach, piles are typically
One of the key dynamic characteristics of a structure is the resonant assumed to bear all of the loads transferred from the structure. However,
frequency. The resonant frequency of the structure can be calculated by in-situ tests results indicate that there could be considerable amount of
the transfer function. Fig. 13 presents the transfer functions of the soil contact pressure between the raft and the soil, thus the loading from the
surface to the raft for both Cushioned-PR and Connected-PR under the superstructure is shared by the piles and the soil together. In this study,
El-Centro waves. The input signal is recorded by the accelerometer (A3) the axial force of piles is calculated with strain measurements from
placed at the soil surface, the output signal is recorded by the acceler­ strain gauges installed along the piles (see Fig. 4). Note the vertical
ometer (A1) placed on the raft, as shown in Fig. 3. The acceleration data conversion factor, kvp, is used to convert the vertical strain to load of the
is used to analyze the resonance frequency of the structures. As shown in pile, was calibrated before any test. With the calibrated conversion
Fig. 13 (a), the frequency corresponding to the natural vibration period factors, the axial force can be calculated as below:

Fig. 11. Influence of PBA on PSA for Cushioned-PR and Connected-PR.

9
F. Liang et al. Soil Dynamics and Earthquake Engineering 142 (2021) 106554

Fig. 12. The response spectrums of different earthquake waves (raft).

Lvp = kvp εv (8) the vertical load ratio of piles was only calculated before and after the
shaking. Before and after the shaking, av was zero. The vertical load
where Lvp is the vertical load shared by piles; kvp is the conversion factor sharing ratio of piles can be obtained by
in the vertical direction; εv is the vertical strain measured by the strain
Lvp Lvp
gauges. According to the information above, the axial force of piles can αap = = (10)
Lv Lvp + Lvs
be calculated before the shaking, during the shaking, and after the
shaking.
where Lvs is the vertical load supported by the soil.
Meanwhile, the total vertical load of the foundation model imposed
Table 9 summarizes the axial force of the piles before, during, and
by the structure can be calculated as below:
after the shaking, respectively. Table 10 shows the contact earth pres­
Lv = ms × g + ms × av (9) sure between the raft and the soil. For Connected-PR, before the earth­
quake motion was inputted, the raft was in contact with the soil. In this
where, Lv is the total vertical load of the foundation model; ms is the phase, the load of the structure was shared by the piles and the soil
weight of the structure; and av is the vertical acceleration of structure. together. According to Table 9, the vertical load sharing ratio of piles
The vertical acceleration of structure av was not measured during the was around 85%–96%. While after the shaking, the settlement of the soil
shaking due to equipment limitations. Thus, the total vertical load of the due to the consolidation was larger than the settlement of the pile toe.
foundation model Lv cannot be calculated during the shaking. Therefore, Thus, the raft was separated from the soil. In other words, the piles

10
F. Liang et al. Soil Dynamics and Earthquake Engineering 142 (2021) 106554

Fig. 13. The transfer function of soil surface to super-structure.

in Table 9, before the shaking, the vertical load sharing ratio of the piles
Table 8
was around 51%–68%. After the shaking, the soil and the interposed
Peak bending moment of column.
layer were densified, the vertical load sharing of the soil and the inter­
Foundation Input Peak positive bending moment of column (kN⋅m) posed layer increased. The vertical load sharing ratio of piles was around
form motion
Small seismic Moderate Strong 49%–65%, which has been slightly decreased. The order of piles in terms
(0.1 g) seismic (0.2 g) seismic (0.3 of the axial force magnitude was similar to that of Connected-PR. The
g)
axial force of piles did not change much before and after the earthquake
Cushioned-PR ACCe100 22.18 37.88 54.44 motions. Fig. 14 shows the center pile axial force of the two kinds of
El-Centro 7.56 16.61 26.69 foundation forms under the action of different PBAs. The axial force of
Mexico 24.36
the center pile for Cushioned-PR was nearly the same with the different
City
Connected-PR ACCe100 44.84 58.17 63.88 PBAs. In contrast, the axial force of the center pile for Connected-PR
El-Centro 26.47 30.42 56.40 increased clearly with the increasing of the PBA. According to the
Mexico 56.07 experimental results in this study, the vertical loading sharing ratio of
City
piles for Cushioned-PR was not as sensitive as the case of Connected-PR
to the earthquake magnitudes.

Table 9 3.4. The settlement of structure and raft during an earthquake


The axial force of piles.
Foundation PILE Before/ The intensity of inputted motion (El- The movement of a structure under earthquake motions can be
form During/ Centro) divided into vertical vibration, horizontal sliding/displacement, and
After the
Small Moderate Strong torsional vibration. It would be helpful to analyze the individual types of
shaking
seismic seismic seismic motion separately. The horizontal displacement for Cushioned-PR is
(0.1 g) (0.2 g) (0.3 g) anticipated to be high and needs to be controlled within the design limit.
Cushion-PR J1- Before 372.65 kN 353.07 kN 469.37 kN Fig. 15 shows the time history of the horizontal displacements for both
Center During 335.58 kN 306.51 kN 400.03 kN Cushioned-PR and Connected-PR under different motions. For
pile After 337.16 kN 382.11 kN 454.42 kN
Cushioned-PR, the horizontal constraint is provided by the fraction be­
J2-Side Before 338.32 kN 320.01 kN 425.34 kN
pile During 277.01 kN 259.32 kN 402.12 kN tween the raft and the interposed layer. For Connected-PR, piles are
After 305.66 kN 346.34 kN 410.82 kN rigidly connected to the raft. Nevertheless, the horizontal constraint for
J3- Before 89.86 kN 84.69 kN 111.42 kN Cushioned-PR is much less compared with the horizontal constrain a
Corner During 115.07 kN 109.84 kN 129.82 kN Connected-PR can provide, so the horizontal displacement for
pile After 81.22 kN 91.69 kN 104.19 kN
Connected- J1- Before 707.29 kN 1054.74 kN 1113.36
Connected-PR is relatively smaller.
PR Center kN When the ACCe100 wave is applied with a PBA of 0.1 g, 0.2 g, and
pile During 2072.09 2658.12 3576.56 0.3 g, the residual horizontal displacement ratios between Cushioned-PR
kN kN kN and Connected-PR are calculated as 240.48%, 260.87%, and 219.31%,
After 708.02 kN 862.77 kN 954.30 kN
respectively. When the El-Centro wave is applied, the residual hori­
J2-Side Before 550.39 kN 459.98 kN 514.75 kN
pile During 1141.28 2591.17 kN 4028.49 zontal displacement ratios between Cushioned-PR and Connected-PR
kN kN are calculated to be 963.64%, 787.5%, and 631.81%, respectively.
After 641.88 kN 631.65 kN 698.41 kN When Mexico City wave is adopted with PBA of 0.1 g, the residual
J3- Before 133.69 kN 90.24 kN 129.14 kN horizontal displacements of Cushioned-PR are larger than that of Con­
Corner During 371.36 kN 911.15 kN 1130.83
pile kN
nected-PR.
After 133.69 kN 105.19 kN 15.58 kN The capability to resist vertical settlement under dynamic loadings is
important to evaluate the seismic performance of a foundation. Fig. 16
shows the vertical settlement of the two kinds of foundations under
supported all the loads transferred from the structure after the shaking. different earthquake motions. For the Cushioned-PR and the Connected-
For Cushioned-PR, the piles were not connected to the raft. Thus, the PR, the vertical settlement increase with the increasing of PBA. Under
raft maintained its contact with the soil, which means the load of the the same earthquake loading, the vertical settlement for Cushioned-PR is
structure was always shared by the piles and the soil together. As shown larger than that of Connected-PR. The type of ground motions also

11
F. Liang et al. Soil Dynamics and Earthquake Engineering 142 (2021) 106554

Table 10
The contact earth pressure between the raft and the soil.
Foundation Before/After the Acce100 El-Centro Mexico City
form shaking
Small seismic Moderate seismic Strong seismic Small seismic Moderate seismic Strong seismic Small seismic
(0.1 g) (0.2 g) (0.3 g) (0.1 g) (0.2 g) (0.3 g) (0.1 g)

Cushioned-PR Before 46 kN/m2 46 kN/m2 46 kN/m2 46 kN/m2 46 kN/m2 46 kN/m2 46 kN/m2


After 46 kN/m2 46 kN/m2 46 kN/m2 46 kN/m2 46 kN/m2 46 kN/m2 46 kN/m2
Connected-PR Before 4.9 kN/m2 3.0 kN/m2 4.3 kN/m2 3.6 kN/m2 5.9 kN/m2 0.5 kN/m2 5.0 kN/m2
After N/A N/A N/A N/A N/A N/A N/A

together with large rotation angles during an earthquake. While for


Cushioned-PR, sliding takes place between the raft and the interposed
layer, in which the interposed layer could isolate the vibration. Thus, the
rotation angles of Cushioned-PR are smaller than those of Connected-PR
under the same earthquake motion. On the other hand, the interposed
layer would densify due to the interaction between the raft and the
interposed layer under a seismic motion. The density of the interposed
layer increases and contributes to more settlement. After the same
earthquake excitation, the settlement of Cushioned-PR is larger than
that of Connected-PR.

3.5. Dynamic response of the piles

Fig. 14. The axial force of the center pile. Note that the movement of the actual structure under earthquake
motions could be significantly different from the corresponding free-
influences the vertical settlement of the foundation systems, as shown in field motion, the pile-soil interaction may not be the only factor
Fig. 16 (a) and (b). The vertical settlements under ACCe100 waves are dictating the pile behavior. The connection form of the foundation
clearly larger than those under El-Centro waves. For Cushioned-PR, the system would also influence the complex impedance, the velocity
vertical settlement is not only due to the subsidence of the ground during admittance, and the reflected wave signal of velocity at the pile head [9,
earthquakes but also the deformation of the interposed layer. Increase 10], so it is necessary to study the vibration behavior of the pile. In this
the stiffness of the interposed layer may reduce the vertical settlement study, the maximum bending moment of piles can be calculated with the
under the earthquake motions. maximum transient strain based on the time history of strain gauge
Although the horizontal displacement of Cushioned-PR could be readings.
much higher than the value of Connected-PR, it is well below the Fig. 18 and Fig. 19 show the calculated bending moment profile
maximum allowable value under normal conditions. Taking Rion- along with the piles of the Cushioned-PR and the Connected-PR. As
Antirion Bridge as an example, the horizontal displacement is never shown in Fig. 18(b), when the applied excitation is the ACCe100 wave
reported to be a problem during service. While a large displacement in with a PBA of 0.2 g, the calculated bending moments are 90.10 kN m,
both horizontal directions is observed during a recent strong earthquake 145.69 kN m, 21.77 kN m and 2.68 kN m, respectively, along with the
[42]. Fortunately, the bridge is flexible enough to bear the large pile of Cushioned-PR. While the values are 167.49 kN m, 151.18 kN m,
displacement between the cushion and the raft, keeping the bridge 69.36 kN m, and 6 kN m along with the pile of Connected-PR. The
intact. The vertical and horizontal displacement will not influence the maximum bending moment to the pile head of the Cushioned-PR is
application of Cushioned-PR for normal service. Large slidings between about 53.79% of that for Connected-PR.
the raft and interposed layer under strong earthquakes are anticipated Observations of the maximum bending moment in this study are well
but that is acceptable considering the benefits of protecting the bridge aligned with the predictions from the numerical simulations conducted
structures during extreme events. by Tradigo et al. [43] and Baziar et al. [5]. The maximum bending
The total amount of foundation settlement includes the settlement moment appears at the upper-middle portion of the pile for
due to the densification of the soil layer and the settlement due to the Cushioned-PR. While for Connected-PR, the maximum bending moment
foundation rocking motions. For Cushioned-PR, the piles are not rigidly occurs close to the pile head.
connected to the raft, constraints in the horizontal and vertical direction When the intensity of excitation is small, the maximum bending
are weaker compared with Connected-PR, in which the piles are rigidly moment of the piles is close to each other for the two types of founda­
connected to the raft and the connection between the ground and the tions. While the differences between the maximum bending moment of
structure is relatively stronger. The correlation between the settlement two foundation types become significant when the intensity increases.
and the rotation for Cushioned-PR is different from Connected-PR. This phenomenon indicates that the connection form and the interposed
Table 11 shows the rotation angles of the different models. Fig. 17 layer affect the propagation of the inertial loading and the interaction
shows the settlement and rotation angle of the two different foundations between the soil and the piles. Furthermore, this phenomenon shows the
under the El-Centro wave with a PBA of 0.2 g. As shown in Fig. 17, the benefit of the isolation effect of Cushioned-PR.
maximum rotation angle of Connected-PR is generally larger than that of
Cushioned-PR. The main reason of this phenomenon is that the 4. Summary and conclusion
connection form may influence the dynamic response of the structure. In
most cases in this study, the rotation angle of the Connected-PR was A series of centrifuge tests are conducted to compare the seismic
generally larger than that of the Cushioned-PR under the action of responses of Cushioned-PR and Connected-PR to investigate the vibra­
earthquake motion. This phenomenon demonstrated that the dynamic tion isolation effect of Cushioned-PR. The synthetic ACCe100 wave, El-
response of the Connected-PR could be larger than that of the Centro wave, and Mexico City wave are employed to test the two types
Cushioned-PR. For Connected-PR, the seismic acceleration of the free of foundations, respectively. The peak response acceleration of the
field motion can easily transfer to the structure. Thus, settlement occurs super-structure, the natural period, and the natural frequency of the

12
F. Liang et al. Soil Dynamics and Earthquake Engineering 142 (2021) 106554

Fig. 15. Comparison of displacement-time history measured location in centrifuge tests.

13
F. Liang et al. Soil Dynamics and Earthquake Engineering 142 (2021) 106554

Fig. 16. The vertical settlement during earthquake: (a) under ACCe100; (b) under El-Centro.

vibration period of Cushioned-PR increases, while the value of


Table 11
Connected-PR shows very marginal difference. Cushioned-PR can
The peak positive rotation angle of the foundation forms.
effectively reduce the earthquake load imposed on the structure under a
Foundation Input The peak rotation angle of the foundation forms strong earthquake. Under strong earthquakes (i.e., PBA is 0.3 g), the
form motion (rad)
peak acceleration of the super-structure of Cushioned-PR is up to 40%
Small seismic Moderate Strong less than the value obtained for Connected-PR. The benefit would be less
(0.1 g) seismic (0.2 g) seismic (0.3
pronounced when the magnitude of the earthquake is smaller.
g)

Cushioned-PR ACCe100 0.058 0.087 0.146 (2) Cushioned-PR can effectively reduce the peak super-structure
El-Centro 0.035 0.043 0.075
acceleration and the bending moment of the column compared
Mexico 0.031
City with Connected-PR. The maximum bending moment of the col­
Connected-PR ACCe100 0.086 0.174 0.201 umn in the Cushioned-PR is around 28%–54% of the values from
El-Centro 0.031 0.072 0.106 the Connected-PR.
Mexico 0.035
(3) Because of the connection form, the horizontal constraint of
City
Cushioned-PR is smaller than the horizontal constraint of
Connected-PR. During the earthquake excitation, the maximum
system, the horizontal and vertical displacement of the system, and the horizontal displacement of Cushioned-PR is about 2–10 times
bending moment of piles, are all analyzed. Based on the results obtained larger than the value of Connected-PR. This higher horizontal
from this study, the following conclusions can be drawn: displacement of Cushioned-PR is generally acceptable during
normal service, however, it should be fully evaluated before field
(1) Structures would encounter certain damage when an earthquake applications.
happens. This will change the dynamic characteristics of the (4) The bending moment for the pile under earthquake motions is
structure; for example, the natural vibration period. Using caused by soil-structure interaction. For Cushioned-PR, the
transfer function and Fourier response spectrum to analyze the interposer layer and the connection form can reduce the inertial
change of natural vibration period and frequency of Cushioned- loading transmitted to the foundation when a relatively stronger
PR and Connected-PR under seismic load, the isolation effect of earthquake is applied to the system. The pile head restrain is
different foundations can be quantified. weaker for Cushioned-PR as compared with the conditions for
Connected-PR. The maximum peak bending moment of the pile
The test results show that with the increasing of PBA, the natural shifts from the top of the pile to the upper middle part of the piles,

Fig. 17. Vertical settlement versus rotation angle of foundation: (a) Cushioned-PR; (b) Connected-PR.

14
F. Liang et al. Soil Dynamics and Earthquake Engineering 142 (2021) 106554

Fig. 18. The bending moment of the center pile: (a) PBA = 0.1 g; (b) PBA = 0.2
g; (c) PBA = 0.3 g Fig. 19. The bending moment of the corner pile: (a) PBA = 0.1 g; (b) PBA =
0.2 g; (c) PBA = 0.3 g
along with a decreasing of the magnitude. This observation
agrees with the numerical simulation studies in the literature. the work reported in this paper.

Declaration of competing interest Acknowledgments

The authors declare that they have no known competing financial This work was supported by the National Natural Science Foundation
interests or personal relationships that could have appeared to influence of China (Grant No. 51878487). Financial support from these

15
F. Liang et al. Soil Dynamics and Earthquake Engineering 142 (2021) 106554

organizations is gratefully acknowledged. [26] Kuo HH. White noise theory. In: Kannan D, Lakshmikantham V, editors. In
Handbook of Stochastic Analysis and Applications. New York: Marcel Dekker;
2002. ch.
References [27] Liang F, Chen L, Shi X. Numerical analysis of composite piled raft with cushion
subjected to vertical load. Comput Geotech 2003;30(6):443–53. https://doi.org/
[1] Allmond JD, Kutter BL. Design considerations for rocking foundations on 10.1016/s0266-352x(03)00057-0.
unattached piles. J Geotech Geoenviron Eng 2014;140:4014058. https://doi.org/ [28] Loli M, Bransby M, Anastasopoulos I, et al. Interaction of caisson foundations with
10.1061/(ASCE)GT.1943-5606.0001162. a seismically rupturing normal fault: centrifuge testing versus numerical
[2] ASCE/SEI 41 Seismic evaluation and retrofit of existing buildings. Reston: simulation. Geotechnique 2012;62(1):29–43. https://doi.org/10.1680/geot.9.
American Society of Civil Engineers; 2014. P.153.
[3] Azizkandi AS, Rasouli H, Baziar MH. Load sharing and carrying mechanism of piles [29] Ma X, He Z, Zhu H, et al. Development of a new geotechnical centrifuge at tongji
in non-connected pile rafts using a numerical approach. Int J Civ Eng 2019;17(6): university in Shanghai. Proc. 6th IC Phys Modell Geotech 2006:151–6. https://doi.
793–808. https://doi.org/10.1007/s40999-018-0356-2. org/10.1201/noe0415415866.ch14.
[4] Banerjee S, Prasad GG. Seismic risk assessment of reinforced concrete bridges in [30] Mylonakis G. Simplified model for seismic pile bending at soil layer interfaces.
flood-prone regions. Struct Infrastruct Eng 2013;9(9):952–68. https://doi.org/ Soils Found 2001;41(4):47–58. https://doi.org/10.3208/sandf.41.4_47.
10.1080/15732479.2011.649292. [31] Nakamura S, Yoshida N, Iwatate T. Damage to Daikai subway station during the
[5] Baziar MH, Rafiee F, Azizkandi AS, et al. Effect of super-structure frequency on the 1995 Hyogoken-Nambu earthquake and its investigation. Japan Society of Civil
seismic behavior of pile-raft foundation using physical modeling. Soil Dynam Engineers, Committee of Earthquake Engineering; 1996. p. 287–95. https://doi.
Earthq Eng 2018;104:196–209. https://doi.org/10.1016/j.soildyn.2017.09.028. org/10.1201/noe0415804844.ch23.
[6] Bhaduri A, Choudhury D. Serviceability-based finite-element approach on [32] Park HJ, Ha JG, Kwon SY, et al. Investigation of the dynamic behaviour of a storage
analyzing combined pile–raft foundation. Int J GeoMech 2020;20(2):04019178. tank with different foundation types focusing on the soil-foundation-structure
https://doi:10.1061/(ASCE)GM.1943-5622.0001580. interactions using centrifuge model tests. Earthq Eng Struct Dynam 2017;46(14):
[7] Cao X, Wong I, Chang M. Behavior of model rafts resting on pile-reinforced sand. 2301–16. https://doi.org/10.1002/eqe.2905.
J Geotech Geoenviron Eng 2004;130(2):129–38. https://doi.org/10.1061/(asce) [33] Pecker A. Design and construction of the rion antirion bridge[C]//Geotechnical
1090-0241(2004)130:2(129). engineering for transportation projects. In: Proceedings of Geo-Trans. Los Angeles;
[8] Clancy P, Randolph MF. An approximate analysis procedure for piled raft 2004. p. 216–40. https://doi.org/10.1061/40744(154)7.
foundations. Int J Numer Anal Methods GeoMech 1993;17(12):849–69. https:// [34] Pecker A. Enhanced seismic design of shallow foundations: example of the Rion
doi.org/10.1016/0148-9062(94)91196-7. Antirion Bridge [C]. In: Proc. 4th Athenian Lecture on Geotechnical Engineering,
[9] Cui C, Meng K, Xu C, et al. Effect of radial homogeneity on low-strain integrity Athens; 2006.
detection of a pipe pile in a viscoelastic soil layer. Int J Distributed Sens Netw [35] Paolucci R, Figini R, Petrini L. Introducing dynamic nonlinear
2018;14(10):1–7. https://doi.org/10.1177/1550147718806459. soil–foundation–structure interaction effects in displacement-based seismic design.
[10] Cui C, Meng K, Wu Y, et al. Dynamic response of pipe pile embedded in layered Earthq Spectra 2019;29(2):475–96. https://doi.org/10.1193/1.4000135.
visco-elastic media with radial inhomogeneity under vertical excitation. Geomech [36] Horikoshi K, Randolph MF. Centrifuge modeling of piled raft foundation on clay.
Eng 2018;16(6):609–18. https://doi.org/10.12989/gae.2018.16.6.609. Geotechnique 1996;46(4):741–52. https://doi.org/10.1680/geot.1996.46.4.741.
[11] Cunha RP, Poulos HG, Small C. Investigation of design alternatives for a piled raft [37] Prakoso WA, Kulhawy FH. Contribution to piled raft foundation design. J Geotech
case history. J Geotech Geoenviron Eng 2001;127(8):635–41. https://doi.org/ Geoenviron Eng 2001;127(1):17–24. https://doi.org/10.1061/(ASCE)1090-0241
10.1061/(ASCE)1090-0241(2001)127:8(635). (2002)128:8(707).
[12] Deng L, Kutter BL, Kunnath SK. Seismic design of rocking shallow foundations: [38] Rasouli H, Fatahi B. A novel cushioned piled raft foundation to protect buildings
displacement based methodology. J Bridge Eng 2014;19:404043. https://doi.org/ subjected to naomal fault rupture. Comput Geotech 2019;106:228–48. https://doi.
10.1061/(asce)be.1943-5592.0000616. org/10.1016/j.compgeo.2018.11.002.
[13] Durante MG, Sarno LD, Sica S, et al. Experimental measurements of geotechnical [39] Reul O, Randolph MF. Design strategies for piled rafts subjected to nonuniform
system in 1-g tests[C]//Imeko Tc4 International Symposium & International vertical loading. J Geotech Geoenviron Eng 2004;130(1):1–13. https://doi.org/
Workshop on Adc Modelling & Testing Research on Electric & Electronic 10.1061/(ASCE)1090-0241(2004)130:1(1).
Measurement for the Economic Upturn. 2014. [40] Roy J, Kumar A, Choudhury D. Natural frequencies of piled raft foundation
[14] Fioravante V, Giretti D. Contact versus noncontact piled raft foundations. Can including super-structure effect. Soil Dynam Earthq Eng 2018;112:69–75. https://
Geotech J 2010;47(11):1271–87. https://doi.org/10.1139/T10-021. doi.org/10.1016/j.soildyn.2018.04.048.
[15] Gazetas G, Mylonakis G, Nikotaou A. Simple Methods for the Seismic Response of [41] Schexnayder C, Alarcon FA, Antillo ED, et al. Observations on bridge performance
Piles Applied to Soil-Pile-Bridge Interaction. Third International conference on during the Chilean Earthquake of 2010. J Construct Eng Manag 2014;140(4):
Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics April B4013001. https://doi.org/10.1061/(ASCE)CO.1943-7862.0000452.
2-7Ill; 1995. p. 1547–56. St. Louis. [42] Sokos E, Tselentis GA, Paraskevopoulos P, et al. Towards earthquake early warning
[16] Gazetas G, Zarzouras O, Drosos V, et al. Bridge-Pier Caisson foundations subjected for the Rion-Antirion bridge, Greece [J]. Bull Earthq Eng 2016;14(9):2531–42.
to normal and thrust faulting: physical experiments versus numerical analysis. https://doi.org/10.1007/s10518-016-9893-8.
Meccanica 2015;50(2):341–54. https://doi.org/10.1007/s11012-014-9997-7. [43] Tradigo F, Pisanò F, Di Prisco C, Mussi A. Non-linear soil–structure interaction in
[17] Ha JG, Ko KW, Jo SB, et al. Investigation of seismic performances of cushioned pile disconnected piled raft foundations. Comput Geotech 2015;63:121–34. https://doi.
raft foundations using dynamic centrifuge tests. Bull Earthq Eng 2019;17:2433–58. org/10.1016/j.compgeo.2014.08.014.
https://doi.org/10.1007/s10518-018-00530-y. [44] Tradigo F, Pisanò F, Di Prisco C. On the use of embedded pile elements for the
[18] Han XL, Li YK, Ji J, et al. Numerical simulation on the seismic absorption effect of numerical analysis of disconnected piled rafts. Comput Geotech 2016;72:89–99.
the cushion in rigid-pile composite foundation. Earthq Eng Eng Vib 2016;15(2): https://doi.org/10.1016/j.compgeo.2015.11.005.
369–78. https://doi.org/10.1007/s11803-016-0329-x. [45] Wong IH, Chang MF, Cao XD. Raft foundations with disconnected settlement-
[19] Han XL, Xiao CA, Li JX, et al. Spring Constitutive Model of rigid pile composite reducing piles. London: Thomas Telford; 2000. p. 469–86. https://doi.org/
foundation and application in design of raft foundation. J Cent S Univ 2013;20(4): 10.1680/daorf.27657.0017.
1079–84. https://doi.org/10.1007/s11771-013-1587-7. [46] Xu R, Fatahi B. Geosynthetic-reinforced cushioned piles with controlled rocking for
[20] Horikoshi K, Randolph MF. Centrifuge modeling of piled raft foundation on clay. seismic safeguarding. Geosynth Int 2018:1–78. https://doi.org/10.1680/
Geotechnique 1996;46(4):741–52. https://doi.org/10.1680/geot.1996.46.4.741. jgein.18.00018.
[21] Kawaguchi K, Tatsuoka F. Several geotechnical design and construction issues with [47] Yan K, He J, Cheng Q, et al. A centrifugal experimental investigation on the seismic
Akashi Strait Bridge. Soils Found 2010;50(6):829–45. https://doi.org/10.3208/ response of group-pile foundation in a slop with an inclined weak intercalated
sandf.50.829. layer. Soil Dynam Earthq Eng 2020;130:698–713. https://doi.org/10.1016/j.
[22] Katzenbach R, Leppla S, Choudhury D. Foundation systems for high-rise structures. soildyn.2019.105961.
USA and UK: CRC press, Taylor & Francis Group; 2016 (ISBN: 978-1-4987-4477-5), [48] Zheng J, Abusharar SW, Wang X. Three-dimensional nonlinear finite element
1-298. modeling of composite foundation formed by CFG–lime piles. Comput Geotech
[23] Ko KW, Park HJ, et al. Evaluation of dynamic bending moment of disconnected 2008;35(4):637–43. https://doi.org/10.1016/j.compgeo.2007.10.002.
piled raft via centrifuge test. Can Geotech J 2019;56(12):1917–28. https://doi.org/ [49] Zhou Y, Sun Z, Chen Y. Zhejiang University benchmark centrifuge test for LEAP-
10.1139/cgj-2018-0248. GWU-2015 and liquefaction responses of a sloping ground. Soil Dynam Earthq Eng
[24] Kumar A, Choudhury D, Katzenbach R. Effect of earthquake on combined pile–raft 2018;113:698–713. https://doi.org/10.1016/j.soildyn.2017.03.010.
foundation. Int J GeoMech 2016;16(5). 04016013, https://doi: 10.1061/(ASCE) [50] Zong Z, Xia Z, Liu H, Li Y, Huang X. Collapse failure of prestressed concrete
GM.1943-5622.0000637. continuous rigid-frame bridge under strong earthquake excitation: testing and
[25] Kumar A, Choudhury D. Development of new prediction model for capacity of simulation. J Bridge Eng 2016;21(9). https://doi.org/10.1061/(ASCE)BE.1943-
combined pile-raft foundations. Comput Geotech 2018;97:62–8. https://doi.org/ 5592.0000912. 04016047.
10.1016/j.compgeo.2017.12.008. [51] Zhang G, Zhang JM. Constitutive rules of cyclic behavior of interface between
structure and gravelly soil. Mech Mater 2008;41(2009):48–59. https://doi.org/
10.1016/j.mechmat.2008.08.003.

16

You might also like