Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

Soil Dynamics and Earthquake Engineering 164 (2023) 107583

Contents lists available at ScienceDirect

Soil Dynamics and Earthquake Engineering


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

The role of soil structure interaction (SSI) on seismic response of tall


buildings with variable embedded depths by experimental and
numerical approaches
Mohammed El Hoseny a, b, *, Jianxun Ma a, **, Walid Dawoud b, Davide Forcellini c
a
School of Human Settlements and Civil Engineering, Xi’an Jiaotong University, Xi’an 710054, Shaanxi Province, China
b
Faculty of Engineering at Shoubra, Benha University, Cairo 11629, Egypt
c
Faculty of Civil Engineering, University of San Marino, Via Consiglio dei Sessanta, 99 Serravalle, 47899, San Marino

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

Keywords: The soil-structure interaction is an integral part of the seismic evaluation, especially for structures with variable
Shaking table test embedment depths. The purpose of this work is to ensure the adequacy of the small scaling factor in the dynamic
Soil-structure interaction analysis and study the seismic behavior of tall building with considering variable embedment depths and soil
Embedded depth
structure interaction effects. Therefore, scaled model tests of a 15-story steel structure were performed using
Seismic analysis
Tall building
shaking table tests and numerical simulations. Three scaled earthquake records, namely: Chi-Chi (1999),
Plaxis 3D model Northridge (1994), and Kobe (1995), were selected. The numerical scaled coupled models were implemented
with Plaxis 3D software and validated with the experimental observations. Afterward, the experimental and
numerical results of scaled models were verified with prototype models to ensure the adequacy of the small
scaling coefficient. The results generated from small-scale models derived from experimental and numerical
simulations were in good agreement. In addition, these results achieved good accuracy with full-scale field
conditions in the seismic analysis. Therefore, a small geometric scaling coefficient of 1:50 achieved good ac­
curacy for buildings with embedded parts to represent the dynamic behavior of structures. Within the frame
results, it is noted that the SSI has an essential role in amplifying the lateral deflection of structures compared
with a fixed base and thus the embedded depths significantly affect the lateral seismic response of the considered
structures.

1. Introduction several consequences, such as foundation embedment effects and wave


inclination [3]. Many researchers [4–10] compared the SSI effects and
The seismic assessment of structural dynamic response is commonly the fixed conditions demonstrating that the flexible base with consid­
based on the assumption that the structure is fixed at the base, but in ering SSI may be suitable in the seismic analysis of structural elements
some cases, the investigation of Soil-Structure Interaction (SSI) becomes due to the increased degree of freedom (i.e., rotations and settlements of
necessary to represent the complex soil structure system realistically. In the foundation) and the modified dynamic characteristics compared
this regard, provisions are provided in Chapter 8 of ASCE/SEI 41–17 [1], with a fixed base [6]. In this regard, Forcellini [11] performed 3D nu­
which can be used to deem the SSI effects in the case of simplified nu­ merical simulations on several buildings to account for the effects of SSI
merical models. Two types of interactions are recognized for considering by developing analytical fragility curves. Furthermore, the same author
SSI: inertial and kinematic. The inertial interaction is due to the vibra­ proposed a theoretical model [12] that considers the effects of SSI on
tions of the structure and foundation, causing displacements of the simplified buildings.
foundation relative to the free field motion [2]. The kinematic interac­ Other researchers [13] studied experimental, numerical, and
tion occurs from the restriction of relatively rigid foundations to analytical modeling to observe the dynamic soil-foundation interaction.
conform to the ground motion, modifying the foundation motion with Two physical models involving a shallow foundation resting on a sand

* Corresponding author. School of Human Settlements and Civil Engineering, Xi’an Jiaotong University, Xi’an 710054, Shaanxi Province, China.
** Corresponding author.
E-mail addresses: mohammed.elhoseny@stu.xjtu.edu.cn (M. El Hoseny), majx@mail.xjtu.edu.cn (J. Ma).

https://doi.org/10.1016/j.soildyn.2022.107583
Received 31 July 2022; Received in revised form 3 October 2022; Accepted 8 October 2022
Available online 18 October 2022
0267-7261/© 2022 Elsevier Ltd. All rights reserved.
M. El Hoseny et al. Soil Dynamics and Earthquake Engineering 164 (2023) 107583

deposit were tested on a shaking table to analyze the soil–foundation assess the couple effects of the behavior of SSI and embedment. In
interaction. The numerical, analytical, and experimental results were particular, the aim is to propose the appropriate small-scale coefficient
validated and verified with each other and achieved good results. In of coupled systems with embedded structural elements to represent the
addition, the seismic behavior of the shallow foundation on the sandy full-scale field conditions. The full-scale field conditions comprise 15-
soil has been noted and observed. Ref. [14] simulated a 20-story story concrete moment resisting frames with variable embedded
building with three foundation systems supported on two soil profiles. depths supported on silty clay soil. To perform these models in a labo­
3D nonlinear dynamic analyses were performed in the time domain by ratory with a small capacity of shaking table, a series of small-scale
using FLAC3D software. It was demonstrated that the SSI effects lead to shaking table tests of 15-story steel frame structures with different
an overall increase in damping and deformability of the embedded depths were placed directly on the shaking table as a fixed
soil-foundation-structure system compared to the fixed-base structure. base and on silty clay soil to simulate a flexible foundation soil. Three
In addition, Parkash [7] evaluated the SSI effects of the embedded different embedded depths are taken into account: 1) without embedded
structure by a simplified method for Multi Degree Of Freedom (MDOF). depth (No basement-S15), 2) with one basement story-S15+1b, and 3)
The author estimated the effects of increasing the embedment depth on with two basement stories S15+2b to highlight the role of embedment
the lateral displacements at the top of the superstructure. Moreover, length on seismic response of the structure. The shaking table tests were
with the increment of the embedment foundation level, the foundations’ accomplished at the American University in Cairo (AUC) laboratories,
natural frequency increases [8]. Cairo, Egypt. The size of the uniaxial shaking table is 1.30 m × 1.70 m,
In the study of Turan et al. [15], a single degree of freedom with with maximum displacement ±75 mm and maximum loading of 5 tons.
different embedded depths was studied under different frequency con­ Furthermore, the testing frequency range of the shaking table is 0.1–20
tents. Other studies reported in the literature the effects of embedded Hz. Three different earthquake displacement records: Chi-Chi (1999),
massive foundations on the filtering action exerted by embedded foun­ Northridge (1994), and Kobe (1995), were applied. The experimental
dations. In particular [16], demonstrated that the foundation mass may measurements were validated with proposed numerical scaled models
have strong beneficial effects on filtering. Stewart et al. [9] observed and then verified with prototype models to evaluate the small-scale
that the static stiffness of the larger embedded foundation depth in­ factor in the dynamic analyses and the performance of the superstruc­
creases the damping due to the effects of contact areas of the embedded ture under seismic loadings.
elements. One of the main effects of the SSI consists of rising energy
dissipation through radiation damping and hysteretic soil damping that 2. Full description of the coupled prototype and scaled models
affects overall system damping. The trapping and scattering of the waves
radiated from the base of buildings into the soft upper layer is significant 2.1. Characteristics of the coupled prototype models
on the free field soft soil, especially in duration due to the resonance case
[17]. In addition, one of the main consequences of SSI under earthquake The configurations of concrete moment resisting frame models
loadings concerns the modification of the foundation input motion considered in the numerical analyses are presented in Fig. 1. The three
compared with the free-field ground. Such an effect was described with prototype buildings consist of three-dimensional fifteen-story concrete
analytical correlation expressions by Sotiriadis et al. [18]. moment-resisting frames with a 8 m × 8 m square plan with variable
In order to study the SSI and the effects of the embedded structure embedded depths (0, 3 m, and 6 m). Each prototype has double bays in
elements in the seismic analysis of buildings, many experimental both directions, and the bay span is 4 m. The total height from the top of
shaking table tests were performed. The full-scale laboratory SSI test the raft foundation to the roof floor level is 45 m, with a typical floor
achieves more accurate structural behavior, but it is limited due to the height of 3 m. All materials used in the prototypes are reinforced con­
high costs and required time. Ref. [19] applied full-scale soil-structure crete with the same physical and mechanical properties (weight per unit
interaction tests for a simple frame with cross-bracings laid on a shallow volume, γ = 25kN/m3, Poisson’s ratio, ⱱ = 0.2, the modulus of elas­
foundation in Greece to discuss the dynamic foundation impedance ticity, E = 25742.96 MPa, and the average compressive strength, ƒc =
function. Others [20] performed a full-scale soil-foundation system by 30 MPa). The seismic-resistant system is composed of ordinary moment-
experimental and numerical approaches under earthquake loads to resisting frames. Three Dimensional-Finite Element (3D FE) models of
check a reasonable accuracy between the two approaches. In addition, structures were built using SAP2000 software [28] to determine the
the motion of the foundation and the forces transmitted to the soil were preliminary designs of structural elements, in which the columns and
observed. Ref. [21] carried out a large-scale factor of 1:6 on the shaking beams are coupled through the floor slabs. The columns and beams el­
table for tall buildings supported on pile group foundations on soft soil. ements are represented as frame elements, while the slab element is
Others [22] performed numerical simulation by OpenSees with a represented as shell elements. The nonlinear time histories were applied
shaking table test with a scaling coefficient of 1:4 for saturated soft soil in this study with step-by-step dynamic analyses. The equation of
under vertical propagating waves. An option to reduce costs and time
consists of conducting smaller-scale experiments coupled with numeri­
cal simulations [19,20]. This approach was adopted by Ref. [23], which
applied a scale factor of 1:30 for a fifteen-story concrete moment
resisting building frames founded on soft soil. The observed results from
the experimental observations were in agreement with the numerical
simulations.
Moreover, Goktepe et al. [24] performed experimental tests on a
small capacity shaking table with a scaling of 1:45 for six-story concrete
moment resisting frames rested on silty sand soil. Their results revealed
that the experimental measurements achieved acceptable accuracy with
numerical simulations. Although several numerical and experimental
studies had been achieved on the soil-structure interaction phenomena
by various researchers over the past years [7,8,22,25,26,27], few studies
applied the small-scale tests to monitor the dynamic structural behavior
considering SSI and variable embedment lengths under seismic loadings.
In this background, the principal novelty of this research consists of
performing a parametric study with several values of the embedment to Fig. 1. Configurations of the prototype SSI models.

2
M. El Hoseny et al. Soil Dynamics and Earthquake Engineering 164 (2023) 107583

motion is solved in the nonlinear analysis internally at each output time acceleration, and a is acceleration. This is known as the “Cauchy con­
step and at each load function time step. Furthermore, a maximum dition”, which can also be defined in terms of shear wave velocity as
sub-step size smaller than the output time step is determined to mini­ follows:
mize the amount of non-linear iteration. The geometric nonlinearity and
(Vs)p √̅̅̅
P-delta effects are considered in the nonlinear analysis. Also, the cracked = λ (1)
(Vs)m
sections for the reinforced concrete sections are considered according to
ACI318 [29]. As a result, the slab thickness for all floors is 160 mm, and Another comprehensive study [38] was performed to set scaling re­
all beam dimensions are 300 mm (in-width) x 600 mm (in-depth), while lations for the SSI system model under dynamic loading and defined the
the reinforced concrete columns taper every five floors, where the first entire problem in terms of density, geometric, and strain scaling factors.
five floors have columns of C1 with dimensions 1100 mm × 1100 mm, This method relates the density (λp) and geometric (λ) scaling factors
the second five floors consist of columns C2 with dimensions 1000 mm and then derives the strain scaling factor (λε) from shear wave velocity
× 1000 mm, and the last five floors have columns C3 with dimensions tests on both prototype and scaling model soil, as in Eq. (2):
900 mm × 900 mm. The prototype models with basement stories have
λ
retaining walls with a thickness of 250 mm. By simulating the 3D nu­ λε = ( )2 (2)
merical analyses of three prototype models with a fixed base, the first
(Vs)p
(Vs)m
natural frequencies and total masses are shown in Table 1. It is observed
that the first natural frequency of the superstructure with a fixed base Further, many researchers [23,35,39–41] stated that the strain
increases with raising the embedment depth of structural elements, in scaling factor (λε) should be kept equal to one, and the Cauchy condition
which the amplification factor of frequencies between S15+2b and S15 (Eq. (1)) should be satisfied to achieve an adequate model for dynamic
models is 1.15. SSI simulation in the experimental tests. Consequently, the similitude
In the case of a flexible base, the soil structure prototype models are rules of different parameters in terms of the geometric scaling factor (λ)
supported on silty clay with a shear wave velocity of 220 m/s, friction between the prototype and scaled model are shown in Table 2 [24,42].
angle of 31.8◦ , Poisson’s ratio of 0.3, and a unit weight of 17.8 kN/m3. Based on the similitude rule formulas and the allowable limits of the
To reduce the undesirable boundary effects and avoid any reflection of shaking table specifications, the appropriate geometric scaling factor (λ)
the propagating waves in numerical analysis, the ratio between the soil is adopted as 1:50 for the experimental shaking table tests. Table 3 il­
plane dimension to the structure plane dimension is greater than 5 [30, lustrates the size and weight of the test models after applying similitude
31]. In addition, the ratio between width to length of the soil container is laws. Therefore, the scaled superstructures are fifteen-story steel frame
2:3, considered the most appropriate ratio in the SSI system problem structures without embedded depth and with embedded depths (0.06 m
[32–34]. Therefore, the soil block size over bedrock to be contributed in and 0.12 m). The length and width of superstructures are to be 0.16 m.
the numerical analysis is assumed to be 70 m and 50 m in length and Also, the total height is determined to be 0.9 m. By considering the
width, respectively. The total depth of the soil block over bedrock is similitude rule, the geotechnical model dimensions composed of soil
assumed to be 40 m. block and laminar soil container are scaled to be 1.40 m in length, 1.00
m in width, and 0.80 m in depth.
To conclude the natural frequencies for scaled models, the prototype
2.2. Characteristics of the coupled scaled models frequencies (as seen in Table 1) should be scaled by a factor of λ− 0.5
according to the similitude rule. In addition, the masses of scaled models
2.2.1. Scaled models are calculated based on the density of the model (ρm) and should be
The scaling model of the considered SSI problem includes geometric, equal to the density of the prototype (ρp). Therefore, Eq. (3) calculates
kinematic, and dynamic similitudes [35]. To ensure geometric similar­ the density of the prototype, in the case of no basement (S15), as an
ity, the scaled model and prototype should have homologous physical example, where (Mp) and (Vp) are the mass and volume of the prototype
dimensions, while satisfying kinematic similitude requires that the structure, respectively.
scaled model and prototype result in homologous particles at homolo­ /
gous points at a homologous time. The dynamic similitude describes a ρp =
Mp
=
1761000
= 611.46 kg m3 (3)
condition where homologous parts of the scaled model and prototype Vp 8 × 8 × 45
experience homologous net forces. This means that after applying the Consequently, the mass (mm) of the scaled S15 without a basement
geometric scaling of physical dimensions, both the kinematic and dy­ can be estimated in Eq. (4) as:
namic similitude rules should be satisfied by utilizing other parameters.
Thus, the main objective of scale modeling is to achieve both the kine­ mm = ρm × Vm = 611.46 × (0.16 × 0.16 × 0.9) = 14.08 Kg (4)
matic and dynamic similitude rules with respect to the prototype with
adequate accuracy. where, Vm is the scaled structural model volume. Based on the above
The adopted methodology [36] mentioned that a scaled model to the calculation, the required masses and natural frequencies of the scaled
prototype with a similar density is desired, fixing another component of structural models for three cases are summarized in Table 4.
the scaling relations. Reference [37] concluded that in 1-g scale Based on the known dimensions, first natural frequencies of three
modeling, the dimensionless product a/g (Froude’s number) must be scaled models, and masses of three scaled models, 3D numerical models
kept equal to unity, implying that the ratio of the specific stiffness (E/ρ) were achieved by the SAP2000 software package to conclude the sizes of
of the scaled model to prototype is equal to the geometric scaling factor physical scaled models’ elements. Both floor slabs and retaining walls
λ, where E is the modulus of elasticity, ρ is density, g is gravitational are modeled as two-Dimensional (2D) shell elements, and a one-
Dimensional (1D) frame element is for columns, as shown in Fig. 2.
The mild steel is adopted with grade 240/350 in the design. In addition,
Table 1
First natural frequencies and total masses of the three prototype cases with fixed
base. Table 2
S15 S15+1b S15+2b Similitude rule for shaking table tests.

Natural Frequency (Hz) 0.723 0.768 0.831 Mass density 1 Acceleration 1 Length λ
Total Mass (tones) 1761 1822 1883 Force λ3 Shear wave velocity λ0.5 Stress λ
Stiffness λ2 Time λ0.5 Strain 1
S15 without a basement story, S15+1b with one basement story, and S15+2b with
Modulus λ Frequency λ − 0.5 EI λ5
two basement stories.

3
M. El Hoseny et al. Soil Dynamics and Earthquake Engineering 164 (2023) 107583

Table 3
The geometric dimensions of the prototype and scaled coupled system according to the similitude rule.
Geometric scaling factor (λ = Structure width Structure length Structure height Soil length Soil width Soil depth Soil volume Soil mass (kg)
1/50) (m) (m) (m) (m) (m) (m) (m3)

1:1 8 8 45 70 50 40 140,000 249,200,000

1:50 0.16 0.16 0.9 1.4 1.0 0.8 1.12 1993.6

According to the similitude rule, the minimum dimensions of the


Table 4
laminar box were determined as 1.4 m in length, 1.0 m in width, and 0.8
Required first natural frequencies and total masses of three scaled models.
m in depth. By allowing 80 mm on each side for construction, the final
S15 S15+1b S15+2b dimensions of the laminar box were estimated to be 1.48 m in length and
Natural Frequency (Hz) 5.11 5.43 5.87 1.08 m in width. As mentioned in many manuscripts [24,32,34,42], the
Total Mass (kg) 14.08 14.57 15.06 rubber and aluminum layers were employed for laminar shear box
materials. To conclude the sizes of rubber and aluminum layers used in
the laminar box, a 3D numerical model was built by SAP2000 software
the modulus of elasticity (E) of steel material is 2.1E+08 kN/m2, the
under the condition that the natural frequencies of the laminar shear box
Poisson ratio is 0.3, and the weight per unit volume is 7.69 t/m3. After
and soil block are identical to reduce the interferences between box and
checking the safety of all elements, it was found that the steel floor plate
soil. Therefore, the natural frequency of the laminar shear box can be
is 160 mm in length, 160 mm in width, and 4.5 mm in thickness, while
estimated by Eq. (5) as follows:
the dimensions of the four vertical columns are determined to be 6.5 mm
in length, 1.5 mm in width, and 900 mm in total height from the rigid Vs
ƒ= Hz (5)
base. In addition, four and eight vertical plates represent the Retaining 4H
Walls (R.W.) for S15+1b and S15+2b models, respectively. Each plate has
160 mm in length, 60 mm in height, and 1.5 mm in thickness. The where Vs is the shear wave velocity of soil in the scaled SSI model with a
connections between Floors, columns, retaining walls, and foundation value of 31.13 m/s, and (H) is the soil depth of 0.8 m. Therefore, the
are provided using welding. The target first natural frequencies and natural frequency of the laminar shear box is calculated to be 9.72 Hz.
masses of the three scaled models based on sizes of elements are sum­ Based on the natural frequency and dimensions of the laminar shear box,
marized in Table 5 and matched with the required characteristics in it consists of ten aluminum hollow box sections and nine layers of solid
Table 4. It can be observed that the difference between required and rectangular rubber, as shown in Fig. 4. The aluminum box section and
target frequencies is less than 0.6% for the three scaled models; at the rubber layers were represented as 1D frame and 2D shell elements in the
same time, the deviation of total masses is less than 2.2%. Therefore, the
three steel models were manufactured in the workshop, as displayed in Table 5
Fig. 3. Target first natural frequencies and masses of three scaled models without base
plate from numerical analysis.
2.2.2. Laminar shear box S15 S15+1b S15+2b
According to various researchers [31,34,43,44], the laminar shear
Natural Frequency (Hz) 5.08 5.44 5.85
box is the most efficient and appropriate type to represent the infinity Total Mass (kg) 13.84 14.29 14.74
soil in actual conditions; thus, it was applied in the experimental models.

Fig. 2. 3D numerical scaled structural models.

4
M. El Hoseny et al. Soil Dynamics and Earthquake Engineering 164 (2023) 107583

frequency of the box. As a result, the first natural frequency of the empty
laminar box was observed at 9.50 Hz, which matched with required and
target values.

2.2.3. Soil properties


The soil used in the laboratory was brought from a real site as a case
study to realize the actual dynamic behavior of the building. Undis­
turbed soil samples were taken from the field conditions to the soil
laboratory to investigate the properties of this soil. The soil is charac­
terized as silty clay with a natural water content of 21%, and the average
unit weight is 17.8 kN/m3. The consolidated-undrained direct shear test
was conducted on representative soil samples based on ASTM D6528
[47]. The effective stresses were applied to specimens with different
values according to chosen specimens’ depths, where the maximum
effective stress was 14.26 kPa at a depth of 0.8 m. The peak internal
friction angle was found to be 31.8◦ with a cohesion of 60 kPa. The shear
modulus was found to be 1758 kPa, assuming Poisson’s ratio of 0.3; the
elastic modulus is estimated to be 4571 kPa. By knowing the shear
modulus and density of selected soil, the shear wave velocity was
concluded. The dilatancy angle was estimated as 1.8◦ as recommended
calculation in a reference [48]. A summary of the soil properties is given
in Table 6.
After a few days from transmitting of soil, the soil properties were
shifted. Therefore, it was required to convert these soil layers in the
laboratory as in field conditions with the same properties to simulate the
actual case. The soil was remolded to simulate the actual field conditions
Fig. 3. Scaled structure models used in shaking table tests. in the laminar box, which the laminar box was divided into 16 layers
with a thickness of 50 mm for each layer (as Fig. 6). The soil was
SAP model, respectively. By trial and cycles to achieve the required
frequency of the laminar box, it was made from hollow aluminum box
frames with dimensions of 80 mm (width) x 40 mm (height) x 2 mm
(thickness) and connected by welding as a rigid connection. On the other
hand, the flexible rubber layers with dimensions 40 mm (width) x 50
mm (height) were separated between aluminum box frames to reduce
the undesirable boundary effects and allow the box to deform in a shear
beam manner. Consequently, the natural target frequency from the
numerical analysis is calculated to be 9.43 Hz which matches with the
required value. For constructing the laminar shear box, the rubber layers
and aluminum sections were connected by the high-strength material of
kemapoxy165. In addition, a 36 mm thick wooden base plate was pro­
vided and bolted to the shaking table by 14M20 to prevent the soil from
resting on the shaking table directly. The interior container sides and the
base were covered and sealed with two layers of plastic sheeting to
maintain the water content percentage of the soil. In addition, a 20 mm
thick polystyrene foam sheet was provided inside the box to reduce the
reflecting waves in the lateral boundaries [45,46]. The components of
this box are illustrated in Fig. 5. Before placing the soil layers in the
laminar box, a sine sweep test was applied to check the fundamental Fig. 5. Components of the laminar shear box.

Fig. 4. Laminar shear box of soil.

5
M. El Hoseny et al. Soil Dynamics and Earthquake Engineering 164 (2023) 107583

Table 6 dynamic actuator (Avg. Dyn. Act.) displacements of the earthquakes that
Mechanical properties of underlying soil for scaled SSI problem from laboratory were in agreement with input scaled earthquakes. Furthermore, Table 7
tests. presents the characteristics of the three earthquakes.
Parameter Symbol Magnitude Unit

Average unit weight γ 17.8 kN/m3 3. Experimental observations and numerical simulations
Young’s modulus E 4571 kN/m2
Shear modulus G 1758 kN/m2 3.1. Shaking table tests and finite element analyses of the three scaled
Poisson’s ratio ν 0.3 – models in the case of no SSI (fixed base)
Compression wave velocity Vp 58.23 m/s
Shear wave velocity Vs 31.13 m/s
Cohesion c 60 kN/m2 The results of scaled models with a fixed base on the shaking table
Friction angle ø 31.8 (◦ ) are illustrated in this section. The masses of the scaled structural models,
Dilatancy angle ψ 1.8 (◦ ) including the base plate, were measured as 16.23 kg, 16.72 kg, and
17.21 kg for S15, S15+1b, and S15+2b models, respectively, that matched
with the numerical studies. Each scaled model was secured sequentially
remolded by mixing 4 L of water with a soil weight of 52 kg and com­
on a shaking table by four bolts with a diameter of 20 mm to achieve a
pacted using a free hammer to reach the required height to achieve the
fixed base. The instrumentations including LVDTs and accelerometers
same properties of the actual soil. During the filling of the laminar soil
were installed to measure the displacements and accelerations, respec­
box, many specimens were prepared and tested to check its character­
tively. Four LVDTs were installed at the shaking table, Level 7 and Level
istic using a direct shear test. The maximum difference between undis­
15, to monitor the behavior of the scaled structural models. In addition,
turbed and remolded soil properties is 10% occurred in internal friction
two accelerometers were set up at the shaking table and the top of scaled
angle. Consequently, the properties of soil used in the laminar soil box
structural models to study the behavior of structures and check the
are considered in agreement with the field samples, are shown in
recorded displacements in which the acceleration, velocity, and
Table 6.
displacement are related to each other within the time domain. Fig. 10
To execute the basement stories, the soil in the laminar box was
shows the arrangements of LVDTs and accelerometers for the S15 model
hand-excavated to depths of 60 mm and 120 mm successively for S15+1b
as an example in a fixed base case.
and S15+2b models, respectively. After placing the scaled models in the
A sine sweep test was applied on three scaled structure models to
specified place in the soil, it was essential to ensure no gaps around the
determine the fundamental frequencies in the case of a fixed base. The
sidewalls. Backfilling and compaction with the same soil were carried
sine sweep test involves a logarithmic frequency sweep holding with
out to achieve complete contact at the soil-structure interface.
constant displacement at the level of the shaking table. The frequency of
the shaking table was increased from 1 Hz to 10 Hz with an increment
2.3. Adopted displacement earthquake records 0.5 Hz. The first resonance between the shaking table and the structural
model presented the fundamental frequency of the structural model. The
The shaking table can be operated using an electric actuator test was repeated three times to ensure fundamental frequency with
controlled by hydraulic with an electric servo valve and a controller adequate accuracy. Fig. 11 (a) and 11(b) display displacement time
which simulates dynamic displacement time histories. Therefore, three histories of the sine sweep test with different frequencies at the roof floor
earthquake records, including Chi-Chi (1999), Northridge (1994), and level for S15 and S15+2b models, respectively. By examining these figures,
Kobe (1995), have been adopted as displacement time histories. The it is shown that the peak displacement amplitudes for S15 and S15+2b
earthquakes have been applied at the rigid base of the shaking table on scaled models arise at frequencies around 5 Hz and 6 Hz, respectively.
both fixed and flexible bases. According to the similitude laws in Table 2, The maximum response of lateral displacement clearly indicates the
the displacement values were reduced by a scale factor of 0.02. In resonance frequency. It is concluded that the resonance value corre­
addition, the time values were scaled by a factor of 0.14142. Therefore, sponds with the fundamental frequency of the structural model. The
the original and scaled displacement time histories of earthquakes are maximum lateral displacement of the S15 scaled model was recorded as
shown in Figs. 7–9. Also, the dynamic actuator has a Linear Variable 18.433 mm at a resonance frequency of 5 Hz. However, in the S15+2b
Differential Transformer (LVDT) that measures the output displacement model, the maximum displacement at the resonance frequency of 6 Hz
under seismic loads. Therefore, Fig. 7(b)-9(b) show the output average was reduced to 16.892 mm. By applying the same concept for the S15+1b
model, the first resonance for the S15+1b model was determined as 5.5 Hz
with an amplitude displacement of 17.541 mm. Consequently, the
maximum deviation of natural frequencies between numerical simula­
tions (as seen in Table 5) and experimental records is less than 2.5% for
three scaled models, which is in very good agreement.
After the verification of the adequacy of the structural models’
characteristics, the three scaled displacement earthquakes were per­
formed on the shaking table level to the scaled structural models with a
fixed base. The three scaled models’ results were determined and pre­
sented in terms of the absolute lateral displacements, which means the
displacement at floor level from LVDT measurement includes the
translational component of the shaking table and the distortion
component of the superstructure. Subsequently, Table 8 shows the
maximum absolute lateral displacements measurements from Experi­
mental (Exp.) tests of three scaled models at levels 15 and 7 under three
scaled earthquakes in the fixed base case (without SSI).
Based on previous numerical analyses for three scaled models with a
fixed base (in section 2.2.1), nonlinear time histories dynamic analysis
was performed under three scaled earthquakes. Consequently, the
maximum absolute lateral displacements derived from Numerical
Fig. 6. Layers of soil. (Num.) analyses at levels 15 and 7 are also displayed in Table 8 under

6
M. El Hoseny et al. Soil Dynamics and Earthquake Engineering 164 (2023) 107583

Fig. 7. Chi-Chi earthquake (1999) (a) original record (b) Scaled record.

Fig. 8. Northridge earthquake (1994) (a) original record (b) Scaled record.

Fig. 9. Kobe earthquake (1995) (a) original records (b) Scaled records.

Table 7
Characteristics of the actual earthquakes [49].
Earthquake Country Date PGA (g) Mw (R) Duration (s) Station

Chi-Chi Taiwan Sep.1999 0.36 7.6 120 CHY006


Northridge USA Jan.1994 0.57 6.7 40 CDMG STATION 24278
Kobe Japan Jan.1995 0.34 6.9 40 KAKOGAWA

three different scaled earthquakes. The error percentage between experimental and numerical results. It is observed that numerical and
experimental and numerical results can be calculated as follows: experimental scaled models achieve sufficient accuracy, in which the
maximum percentage error is less than 9%.
|experimental − numerical results|
X100 = % (6)
numerical results
Therefore, Table 8 also presents the error percentage between

7
M. El Hoseny et al. Soil Dynamics and Earthquake Engineering 164 (2023) 107583

wave at the bottom of the soil.


The numerical model of this case was built up in Plaxis software [48]
by considering a linear elastic perfectly plastic material model under
Mohr-Coulomb (MC) failure criterion for soil model, as already pro­
posed in many studies [50,51] to simulate nonlinear soil behavior under
seismic forces. In addition, the soil properties were considered in
Table 6. The mesh soil was built up by considering the previous con­
tributions [52–54] in order to define the number of the elements based
on the seismic signal wavelength and the maximum frequency above
which the spectral content of the input can be considered negligible.
Consequently, the top movement of the soil was added from the
numerical analysis in Fig. 12. A validation procedure was applied to
ensure that the results at the top of the soil from the numerical simu­
lations were in good agreement with the experimental measurements.
For the soil structure coupled system in the laboratory, the structural
models were placed on the designed location in the middle of the soil.

Fig. 10. Final arrangements of the LVDTs and accelerometers in a fixed


base case.

3.2. Shaking table tests and finite element analyses of the three scaled
models in the case of SSI effects (flexible base)

The laminar soil container was secured and fixed on the shaking
table. As demonstrated in sections 2.2.2 and 2.2.3, the laminar box was
filled with specific layers of soil in one day. After finishing the surface,
the top of the soil was covered and sealed with a plastic sheet to
maintain the water content in the soil mix. On the second day before
placing the scaled structures, a Sine curve was applied at the base of the
shaking table level to verify the properties of the soil derived from
laboratory tests. Consequently, the lateral displacement at the middle
point of the top soil block was measured by LVDT. Therefore, Fig. 12 Fig. 12. Experimental and numerical lateral displacements at the top level of
displays the resulted sine wave at the top soil block with an input sine the middle soil surface with shaking table movement as Sine wave.

Fig. 11. Lateral displacement time histories at the roof floor level subjected to sine sweep test under different frequencies for structural models (a) S15 model (b)
S15+2b model.

Table 8
The experimental and numerical maximum absolute lateral displacement records (mm) of three scaled structural models at levels 15 and 7 under three scaled seismic
loadings without considering SSI (fixed base).
Chi-Chi Northridge Kobe

Exp. Num. Error % Exp. Num. Error % Exp. Num. Error %

S15 Level15 5.92 6.07 2.5 3.06 3.31 7.5 2.36 2.51 6.0
Level 7 5.37 5.49 2.2 2.41 2.43 0.8 1.98 2.08 4.8
S15+1b Level15 5.80 6.01 3.5 2.76 2.96 6.7 2.35 2.50 6.0
Level 7 5.12 5.26 2.6 2.16 2.15 0.5 1.86 2.04 8.8
S15+2b Level15 5.00 5.05 1.0 2.49 2.34 6.4 2.17 2.12 2.3
Level 7 4.72 4.81 1.9 2.12 1.99 6.5 1.78 1.80 1.1

8
M. El Hoseny et al. Soil Dynamics and Earthquake Engineering 164 (2023) 107583

The instrumentations of structures were similar to those with fixed base permanent slips occur. The properties of interface elements are related
cases. As a result, Fig. 13 presents the final setup of the soil-structure to adjacent soil properties by using the strength reduction factor (Rinter).
interaction system on the shaking table for three coupled scaled This factor relates to the interface strength (adhesion and wall friction)
models. After applying three scaled displacement earthquakes on the SSI to the soil strength (cohesion and friction angle). The strength reduction
system at the rigid shaking table base level, Table 9 shows the maximum factor applied in the model was set up to be 0.67 according to Ref. [48].
absolute lateral displacement measurements for three scaled models at Therefore, the cohesion and friction angle of interface elements are
levels 15 and 7 in the case of a flexible base. In this case, the absolute reduced by Rinter compared with soil properties.
displacements include the translational component of the shaking table, The analysis of the soil and structure was performed in a single step
the rocking component caused by foundation rotation, and the distor­ which is called a direct method, following [56–58] that applied such an
tion component of the superstructure. approach for multi-story buildings. The three scaled earthquakes had
According to mechanical properties and dimensions of scaled been applied at the bedrock base to simulate the dynamic response of
models, 3D numerical scaled models with a flexible base were built as in superstructures. Nonlinear time-history dynamic analysis was carried
Fig. 14. The dimensions of the soil elements were increased from the out in this study to determine the actual behavior dynamic response of
structure to the lateral boundaries that were realistically modeled with structure models.
damping dashpots in order to absorb the spurious wave reflections in­ The numerical simulations of the three scaled structural models in
side the soil under applied dynamic loads. Consequently, two dashpots terms of maximum absolute lateral displacements at levels 15 and 7 with
were implemented along the longitudinal and transversal directions, as flexible bases are presented in Table 9. By comparing the numerical
shown in Fig. 14. The dashpots provide normal and shear stresses that predictions with experimental observations, it is concluded that the
absorb the waves at the lateral boundaries by equivalent viscous experimental and numerical results are in good agreement with suffi­
dampers based on the pressure and shear wave velocities, respectively. cient accuracy, with the maximum error percentage not exceeding 13%.
These artificial boundaries may absorb waves, whether non-harmonic or
harmonic waves, as mentioned in Ref. [55]. In addition, the base mesh
3.3. Numerical analyses of the real models without SSI effects (fixed
was represented as fixed (with no) deformations by allowing the direct
base)
waves to propagate along the upward direction [48]. The interface be­
tween soil block and structural elements (foundations, retaining walls)
This section evaluates the real buildings under three original earth­
was modeled in order to simulate the frictional zone with the soil with
quakes with a fixed base. To ensure the small-scaled coefficient achieved
an elastic-plastic model used to describe the SSI effects. In particular, the
the adopted accuracy, the real model results are compared with the scale
elastic behavior occurs when small displacement develops within the
models results. By referring to the characteristics of three prototype
interface, while the plastic interface behavior is considered as soon as
models in section 2.1, the numerical simulations were performed using

Fig. 13. Final setup of the flexible models on the shaking table.

9
M. El Hoseny et al. Soil Dynamics and Earthquake Engineering 164 (2023) 107583

Table 9
The experimental and numerical maximum absolute lateral displacement records (mm) of three scaled structural models at levels 15 and 7 under three scaled seismic
loadings considering SSI (flexible base).
Chi-Chi Northridge Kobe

Exp. Num. Error % Exp. Num. Error % Exp. Num. Error %

S15 Level15 7.66 8.77 12.6 3.93 3.72 5.6 3.56 3.80 6.3
Level 7 6.70 7.08 5.4 2.93 2.73 7.3 2.62 2.77 5.4
S15+1b Level15 7.08 7.71 8.2 3.73 3.65 2.2 3.16 3.25 2.7
Level 7 6.26 6.53 4.1 2.67 2.55 4.7 2.41 2.65 9.0
S15+2b Level15 6.20 6.53 5.0 3.43 3.27 4.9 2.85 2.64 7.9
Level 7 5.27 5.75 8.3 2.35 2.26 4.0 2.30 2.11 9.0

that the calculated scale coefficients have sufficient accuracy, which are
ranged from 62 to 46. Therefore, the similitude coefficient (λ = 1:50)
can represent the seismic behavior of real models with reasonable ac­
curacy in the case of a fixed base.

3.4. Numerical analyses of the real models considering SSI effects


(flexible base)

In order to assess the seismic behavior of real buildings under seismic


scenarios, three earthquake records were applied (Chi-Chi, Northridge,
and Kobe) by performing the nonlinear time histories dynamic analysis
method. This method consists of applying the selected input motions at
the base of the soil model to consider the SSI effects. The numerical
simulations were performed with the PLAXIS 3D finite element program,
in which the prototype models’ characteristics were mentioned in sec­
tion 2.1. In addition, the boundary conditions were applied as in the
scaled numerical models with a flexible base to reduce reflected waves
back into the soil medium. After referring to the similitude rules in
Table 2 and examining the subsoil properties in the scaled models in
Table 6, the subsoil properties were concluded in real cases.
Fig. 14. 3D Numerical scaled models with a flexible base. The maximum absolute lateral displacements considering SSI effects
are presented in Table 11 for three prototype models at levels 15 and 7.
Furthermore, the maximum absolute lateral displacements derived from
the SAP software based on the finite element method. Three real
experimental observations and numerical simulations of scaled models
displacement time histories of earthquakes were applied at the fixed
at levels 15 and 7 are shown in the same table. In addition, the geometric
base. The maximum absolute lateral displacements of three prototype
scaling coefficients between real and scaled results are calculated and
models at levels 15 and 7 with a fixed base condition are given
provided, with values ranging from 64 to 41. It is noticeable that the
comparatively under three earthquakes in Table 10. In addition, the
scaled models’ results (experimental and numerical) and the numerical
maximum absolute lateral displacements obtained from Experimental
simulations of the full-scale field conditions are achieved reasonable
(Exp.) observations and Numerical (Num.) simulations of scaled models
results. Therefore, it can be seen that the specified scaling factor (herein,
at levels 15 and 7 are shown. To verify and check the determined geo­
λ = 1:50) can generally represent the real coupled system under seismic
metric scaling coefficient (herein, λ = 1:50) achieved good results in
loads.
dynamic analyses, the maximum absolute lateral displacements from
real cases are divided by average maximum absolute lateral displace­
ments derived from experimental and numerical results. It is concluded

Table 10
Accuracy between real and scaled models of maximum absolute lateral displacements (m) at levels 15 and 7 under three different earthquakes in the case of fixed base.
Chi-Chi (1999) earthquake Northridge (1994) earthquake Kobe (1995) earthquake

Scaled model Real Geometric Scaled model Real Geometric Scaled model Real Geometric
average scale average scale average scale
Exp. Num. Num. Exp. Num. Num. Exp. Num. Num.
(adopted 1/λ = (adopted 1/λ = (adopted 1/λ =
(m) (m) (m) (m) (m) (m) (m) (m) (m)
50) 50) 50)

S15 Level 0.0059 0.0061 0.368 61 0.0031 0.0033 0.147 46 0.0024 0.0025 0.136 55
15
Level 0.0054 0.0055 0.316 58 0.0024 0.0024 0.115 48 0.0019 0.0020 0.107 55
7
S15+1b Level 0.0058 0.0060 0.347 59 0.0027 0.0029 0.144 51 0.0023 0.0025 0.130 54
15
Level 0.0051 0.0052 0.300 58 0.0021 0.0021 0.111 53 0.0018 0.0020 0.103 54
7
S15+2b Level 0.0050 0.0050 0.310 62 0.0025 0.0023 0.138 57 0.0022 0.0021 0.122 57
15
Level 0.0047 0.0048 0.285 60 0.0021 0.0020 0.107 52 0.0017 0.0018 0.102 58
7

10
M. El Hoseny et al. Soil Dynamics and Earthquake Engineering 164 (2023) 107583

Table 11
Accuracy between real and scaled models of maximum absolute lateral displacements (m) at levels 15 and 7 under three different earthquakes in the case of flexible
base.
Chi-Chi (1999) earthquake Northridge (1994) earthquake Kobe (1995) earthquake

Scaled model Real Average Scaled model Real Average Scaled model Real Average
Geometric scale Geometric scale Geometric scale
Exp. Num. Num. Exp. Num. Num. Exp. Num. Num.
(adopted 1/λ = (adopted 1/λ = (adopted 1/λ =
(m) (m) (m) (m) (m) (m) (m) (m) (m)
50) 50) 50)

S15 Level 0.0077 0.0087 0.469 57 0.0039 0.0037 0.160 42 0.0035 0.0038 0.150 41
15
Level 0.0067 0.0071 0.360 52 0.0029 0.0027 0.126 45 0.0026 0.0027 0.110 41
7
S15+1b Level 0.0071 0.0077 0.450 61 0.0037 0.0036 0.150 41 0.0031 0.0032 0.140 44
15
Level 0.0063 0.0065 0.349 54 0.0026 0.0025 0.124 48 0.0024 0.0026 0.108 43
7
S15+2b Level 0.0062 0.0065 0.406 64 0.0034 0.0033 0.148 44 0.0028 0.0026 0.140 52
15
Level 0.0053 0.0057 0.331 60 0.0023 0.0022 0.120 53 0.0023 0.0021 0.107 48
7

4. Discussion 4.1. Dynamic behavior of scaled models considering SSI effects compared
with a fixed base
As mentioned above, the numerical predictions from nonlinear finite
element analyses for both real and scaled coupled systems are in good By comparing the maximum lateral displacement derived from the
agreement with experimental observation for the three performed SSI system with fixed base cases in Table 12, the SSI considerations have
earthquakes and both fixed and flexible bases. Therefore, it can be amplified compared with the fixed base conditions under different
obviously seen that the small scaling factor (herein, λ = 1:50) can seismic loads, agreed with many researchers [4,24,42]. The amplifica­
represent the seismic behavior of the real coupled system for practical tion values of the lateral displacements in case SSI compared with fixed
considerations. In addition, the obtained numerical approach can be base are generated by the rocking component and the distortion
used to study other parametric. component of the superstructure that is different from the fixed base
For studying the SSI and embedded depths effects on the seismic case. It may be observed that the SSI effects have amplified up to 1.7
response of the superstructure, the relative lateral displacement of the times on the lateral response displacements compared with those of
superstructure is presented. The components of the absolute lateral fixed base for three scaled models under the Chi-Chi earthquake, while
displacements of the superstructure include the translational component the maximum amplifications reach up to around 1.9 times under
of the earthquake, the rocking component caused by foundation rotation Northridge and Kobe earthquakes for three scaled models. Therefore,
(in a flexible base only), and the distortion component of the super­ considering a fixed base in the seismic analysis is an underestimation of
structure. By examining the results of the laboratory measurements, the the lateral response of structures, and the conventional seismic analysis
translational component of the earthquake is common in fixed and is not adequate by neglecting the SSI effects.
flexible base cases. Therefore, the displacement time history of the
shaking table has been subtracted from the displacement time history of
4.2. Dynamic behavior of scaled models with variable embedded depths
the measured story to determine the lateral displacement of a certain
considering a fixed base and SSI effects
level relative to the shaking table level. Table 12 shows the maximum
lateral displacements of two measured stories for three scaled models
In this section, the effects of the embedded depths on the seismic
with fixed and flexible bases relative to shaking table movements under
behavior of superstructures are presented. According to Table 12, the
three earthquake records. The concluded results have two discussion
S15+2b model has the lowest values of the maximum lateral displacement
items: the first is the SSI effects compared with a fixed base, while the
compared with other models under different earthquake records,
second item is the effects of variable embedded depths with consider­
whether in fixed (No SSI) or flexible (SSI) bases. To study the embedded
ation of fixed and flexible bases. Finally, the relative lateral displace­
depth effects in Table 12, the maximum lateral displacements of scaled
ment of real system models with the height of the superstructure is
models with one basement (S15+1b) and two basements (S15+2b) are
discussed to extend parametric study in future work.
divided by the results produced without basement case (S15) under
different earthquakes to produce a reduction ratio. As a result, in the
S15+1b model fixed case, the maximum reduction ratios are 0.81, 0.93,
and 0.94 under Chi-Chi, Northridge, and Kobe earthquakes, respec­
Table 12 tively, but in the case S15+2b model, the maximum reduction ratios are
The experimental maximum lateral displacement records (mm) relative to 0.73, 0.74, and 0.83 under Chi-Chi, Northridge, and Kobe earthquakes,
shaking table movement at levels 15 and 7 for three scaled models under three respectively. By applying the same calculations in the case of a flexible
scaled earthquakes. base, the reduction ratios reach up to 0.84, 0.98, and 0.93 under Chi-Chi,
S15 S15+1b S15+2b Northridge, and Kobe earthquakes, respectively. In the case of two
basements stories (S15+2b), these ratios are 0.60, 0.82, and 0.65,
NO SSI NO SSI NO SSI
SSI SSI SSI respectively. It may be noticed that the structure with lower embedded
depth has slightly affected the lateral response of structures, whether in
Chi-Chi (1999) Level15 2.62 4.04 2.12 3.42 1.94 3.26
Level 7 2.03 3.30 1.89 3.01 1.48 1.98 the case of fixed or flexible bases, in which the maximum reduction
Northridge Level15 2.20 3.78 2.10 3.73 1.78 3.10 percentage is 19%. However, in the structure with two basement stories,
(1994) Level 7 1.72 2.85 1.60 2.84 1.27 2.40 the lateral displacement is significantly reduced under different earth­
Kobe (1995) Level15 1.87 3.44 1.76 3.22 1.63 2.25 quakes compared with the structure without a basement, especially
Level 7 1.52 2.46 1.45 2.45 1.26 1.69
considering SSI effects. Therefore, the maximum reduction percentage

11
M. El Hoseny et al. Soil Dynamics and Earthquake Engineering 164 (2023) 107583

reaches up to 40%. Based on the aforementioned, the neglecting of exhibits the absolute lateral displacement time history at the roof floor
embedded structures in the seismic analysis produces overestimation in level that is subtracted from the absolute lateral displacement time
calculating the lateral response of structures. Consequently, the cost of history at the foundation level to create the relative lateral displacement
building is affected. time history at the roof floor level. It is shown that the maximum relative
displacement at the roof floor level is +0.274 m with a positive sign,
despite the maximum absolute displacements at the roof and foundation
4.3. Relative lateral displacements of the real coupled systems
levels of − 0.47 m and − 0.276 m, respectively, with a negative sign (out
of phase). By comparing Fig. 15(a) and (b), It is observed that the
The relative lateral displacement in this section is concluded by
amplification ratio of maximum absolute lateral displacement at the roof
subtracting the movement at the foundation level from the movement of
floor level is 1.277 (0.470/0.368). However, in the case of maximum
a certain level to realize the SSI effects. The three components (trans­
relative lateral displacement, the amplification ratio increases up to 2
lational, rocking, distortion) are in or out of synchronizing with each
(0.274/0.137). These behaviors are because: (a) the three displacement
other; they may be in-phase or out-of-phase, mutually offset, or over­
components may be in or out of sync with each other. In addition, in-
lapped. Therefore, the amplification ratios between the flexible and
phase or out-of-phase, mutually offset or overlapped. (b) The absolute
fixed bases in the absolute lateral displacement case may differ from the
lateral displacement time history at the roof floor level in the flexible
relative lateral displacement case. To observe this procedure, Fig. 15
base case is highly shifted compared with earthquake motion and
shows the absolute and relative displacement time histories for the S15
foundation motion due to the flexible base of the superstructure.
prototype model at roof floor level with fixed and flexible bases under
By following the identical sequences for other cases, the maximum
the Chi-Chi earthquake (as an example). According to Fig. 15(a), in the
relative lateral movements along the superstructure height are pre­
fixed base case, the absolute displacement time history at the roof floor
sented to study the seismic behavior of real superstructures considering
level is subtracted from displacement time history at the foundation base
SSI and embedded depths effects. The maximum relative lateral dis­
(earthquake motion) to produce relative lateral displacement at a roof
placements of the three prototypes with fixed and flexible bases under
floor level. It is noted that the maximum relative lateral displacement is
three earthquakes are displayed in Fig. 16(a)-16(c). It becomes apparent
− 0.137 m, located distant from the maximum absolute displacement of
that the maximum relative lateral displacements of buildings along the
− 0.368 m. On the other hand, in the case of the flexible base, Fig. 15(b)
height considering SSI are amplified compared with a fixed base which
agreed with numerical and experimental scaled models’ results. The
maximum amplification ratios for the three prototype flexible models at
roof floor level reach up to 2, 1.34, and 1.60 compared with the fixed
bases under Chi-Chi, Northridge, and Kobe earthquakes, respectively.
Therefore, the SSI effects have amplified the lateral displacement
compared with a fixed base. On the other hand, the consideration of
embedded depth effects has reduced structures’ lateral displacements,
especially considering SSI effects, in which the maximum reduction ratio
at the roof floor level in the S15+2b flexible base is around 0.7 compared
with the S15 model under different earthquakes. Therefore, it is rec­
ommended to consider the soil-structure interaction and embedded
depth of structural elements in the seismic analysis to maintain the
safety and cost of the superstructure.
To give confidence in the seismic behavior of superstructures with
embedded depths, Table 13 summarizes the maximum foundation ro­
tations of full-scale field conditions models under three different
earthquakes to predict the rocking component of the superstructures. As
expected, the maximum rotation angle generally decreases with
increasing embedded depth. The rocking component can be calculated
based on the relationship mentioned by Trifunace et al. [2], in which the
relative lateral displacement from a flexible base comprises the distor­
tion and rocking components due to SSI effects [59]. According to
Table 13 and Fig. 16, in the S15 model structure, on average, 23.6% of
the maximum lateral deflections at a roof level were due to the rocking
component, while 76.4% took place as a result of the distortion
component. In the S15+1b model structure, on average, 20.4% of the
maximum lateral deflections at a roof level resulted from the rocking
component, while 79.6% occurred due to the distortion component.
These values in the S15+2b model structure were 19% and 81%,
respectively. For example, under the Chi-Chi earthquake, the maximum
lateral deflection at the top of the fixed base 15-story without a base­
ment was recorded to be 0.14 m due to the distortion component only,
while the maximum lateral deflection at the top of the same model with
a flexible base was 0.27 m, with 0.064 m due to the rocking component,
and 0.206 m, a result of the distortion component. It is observed that the
distortion component derived from the flexible base is larger than that of
the fixed base case with an amplification ratio of 1.47, which is accepted
by other research [21].
Fig. 15. Absolute and relative displacement time histories for S15 model at the
roof floor level under Chi-Chi earthquake with different bases: (a) fixed base (b)
flexible base.

12
M. El Hoseny et al. Soil Dynamics and Earthquake Engineering 164 (2023) 107583

Fig. 16. Maximum relative lateral displacements (m) of 15-story prototypes without basement, with one basement, and with two basements in the case of fixed and
flexible cases under three earthquakes: (a) Chi-Chi (1999) (b) Northridge (1994) (c) Kobe (1995).

field condition to investigate other parametric.


Table 13
The obtained results showed that the amplified lateral displacements
Numerical maximum rotation of foundation base under three earthquakes.
occurred for the case with flexible base (SSI effect) under different
Chi-Chi (1999) Northridge (1994) Kobe (1995) earthquakes, and this circumstance may compromise structural safety.
S15 0.070 0.065 0.050 This amplification referred to rocking and distortion components that
◦ ◦ ◦

S15+1b 0.056 0.053 0.039 were different from in fixed base cases. In addition, the amplification
◦ ◦ ◦

S15+2b 0.046 0.043 0.033


◦ ◦ ◦

ratios in terms of absolute and relative lateral displacements were


different in both fixed and flexible bases. These behaviors were because
5. Conclusions the measured displacement time histories may be in or out of sync, in-
phase or out-of-phase, and mutually offset or overlapped with input
The contribution of this paper is to propose a small scaling coefficient motions.
(1:50) able to represent a real 15-story concrete moment resisting The parametric study on the embedment showed a reduction of the
building frames with variable embedded depths (no basement, one lateral displacements, especially in the case of the flexible base, in which
basement, and two basement floors) in the dynamic analyses. The lateral the maximum reduction ratio at the roof floor level in the S15+2b flexible
response of the structures for scaled models was considered by per­ base was around 0.7 compared with the S15 model under different
forming a series of shaking table tests. For each case, two types of earthquakes. Therefore, it is demonstrated that the mutual effects of SSI
foundation conditions (fixed and flexible bases) were studied. The de­ and embedment have significant effects on the dynamic behavior of tall
tails of physical models, laminar container, and the soil mix were buildings and thus both these issues need to be considered in the
explained to achieve a more accurate dynamic soil-structure interaction. assessment of structural dynamic response.
The scaled models were investigated both experimentally and numeri­ The present study provided insightful results on the role of SSI on the
cally under three different earthquake scenarios and then evaluated the seismic response of several configurations with variable embedded
real models numerically. To evaluate the small scaling coefficient in the depths by experimental and numerical approaches. However, several
dynamic analyses, the maximum absolute lateral displacements of the issues require further research. The principal limitations of the paper
prototype models were divided into those investigated experimentally consist of (1) considering three input motions selected as reference
and numerically of scaled models in the case of the fixed and flexible earthquakes. This seismic scenario will be extended in future work.
bases. It can be concluded that the small scaling factor (herein, λ = 1:50) Then, (2) the assumption of a rigid bedrock, since the aim of the paper
can capture the dynamic response of the full-scale field conditions with was to discuss the role of the embedded depths by comparing between
variable embedded depths having sufficient accuracy under seismic relative results. Other bedrock conditions will be performed in future
loadings. Therefore, the numerical model can represent the full-scale analyses. Another limitation is (3) that the seismic response may be

13
M. El Hoseny et al. Soil Dynamics and Earthquake Engineering 164 (2023) 107583

affected by the irregularity of the building. In this regard, future [13] Massimino MR, Maugeri M. Physical modelling of shaking table tests on dynamic
soil–foundation interaction and numerical and analytical simulation. Soil Dynam
buildings will be considered and tested.
Earthq Eng 2013;49:1–18. https://doi.org/10.1016/J.SOILDYN.2013.01.023.
[14] Scarfone R, Morigi M, Conti R. Assessment of dynamic soil-structure interaction
Author Statement effects for tall buildings: a 3D numerical approach. Soil Dynam Earthq Eng 2020;
128:105864. https://doi.org/10.1016/J.SOILDYN.2019.105864.
[15] Turan A, Hinchberger SD, El Naggar MH. Seismic soil–structure interaction in
Jianxun Ma: Methodology, resources, funding acquisition; buildings on stiff clay with embedded basement stories. Can Geotech J 2013;50:
Mohammed El Hoseny, Davide Forcellini and Walid Dawoud: 858–73. https://doi.org/10.1139/cgj-2011-0083.
[16] Conti R, Morigi M, Rovithis E, Theodoulidis N, Karakostas C. Filtering action of
conceptualization, writing—review and editing, Mohammed El embedded massive foundations: new analytical expressions and evidence from 2
Hoseny, Walid Dawoud, and Jianxun Ma: formal analysis, investi­ instrumented buildings. Earthq Eng Struct Dynam 2018;47:1229–49. https://doi.
gation, data curation, writing—original draft preparation, visualization. org/10.1002/eqe.3014.
[17] Guéguen Philippe, Bard PY, Chávez-García FJ. Site-city seismic interaction in
All authors have read and agreed to the published version of the revised
Mexico City - like environments: an analytical study. Bull Seismol Soc Am 2002;92:
manuscript. 794–811. https://doi.org/10.1785/0120000306.
[18] Sotiriadis D, Klimis N, Margaris B, Sextos A. Analytical expressions relating free-
field and foundation ground motions in buildings with basement, considering soil-
Funding statement structure interaction. Eng Struct 2020;216:110757. https://doi.org/10.1016/J.
ENGSTRUCT.2020.110757.
The authors are grateful for the financial support of the National [19] Amendola C, de Silva F, Vratsikidis A, Pitilakis D, Anastasiadis A, Silvestri F.
Foundation impedance functions from full-scale soil-structure interaction tests. Soil
Natural Science Foundation of China (NSFC) under [grant number Dynam Earthq Eng 2021;141:106523. https://doi.org/10.1016/J.
51178390]. SOILDYN.2020.106523.
[20] Paolucci R, Shirato M, Yilmaz MT. Seismic behaviour of shallow foundations:
shaking table experiments vs numerical modelling. Earthq Eng Struct Dynam 2008;
37:577–95.
Declaration of competing interest [21] Yang J, Li P, Lu Z. Large-scale shaking table test on pile-soil-structure interaction
on soft soils. Struct Des Tall Special Build 2019;28:1–19. https://doi.org/10.1002/
The authors declare that they have no known competing financial tal.1679.
[22] Cheng X, Cui C, Sun Z, Xia J, Wang G. Shaking table test and numerical verification
interests or personal relationships that could have appeared to influence
for free ground seismic response of saturated soft soil. Math Probl Eng 2018;2018.
the work reported in this paper https://doi.org/10.1155/2018/3416315.
[23] Tabatabaiefar Sayed HR. Determining seismic response of mid-rise building frames
Data availability considering dynamic soil-structure interaction. 2012.
[24] Goktepe F, Celebi E, Jawad A, Omid AJ. Numerical and experimental study on
scaled soil-structure model for small shaking table tests. Soil Dynam Earthq Eng
Data will be made available on request. 2019;119:308–19. https://doi.org/10.1016/j.soildyn.2019.01.016.
[25] Yung-Yen Ko, Cheng-Hsing Chen. Soil–structure interaction effects observed in the
in situ forced vibration and pushover tests of school buildings in Taiwan and their
Acknowledgments modeling considering the foundation flexibility. Earthq Eng Struct Dynam 2010;39
(9):945–66. https://doi.org/10.1002/eqe.976.
[26] Lu X, Zhou Y, Lu W, Zhao B. Shaking table model test on shanghai world financial
The authors are grateful for the support of the construction labora­
center tower. Pac Conf Earthq Eng 2007;36:439–57. https://doi.org/10.1002/eqe.
tory at the American University in Cairo (AUC), Egypt especially, [27] Bagheri M, Jamkhaneh ME, Samali B. Effect of seismic soil–pile–structure
thankfulness to Dr. Omar El Kadi for supporting us in this laboratory. In interaction on mid- and high-rise steel buildings resting on a group of pile
foundations. Int J GeoMech 2018;18:04018103. https://doi.org/10.1061/(asce)
addition, the authors will also like to record their gratitude to my Chi­
gm.1943-5622.0001222.
nese friends in our office for supporting and the reviewers for their [28] SAP. Integrated finite element analysis and design of structures graphic user
proposed valuable and fruitful comments as well as for their powerful interface manual. Berkley, California, USA: Computers and Structures, Inc.; 2000.
reading and suggestions. [29] ACI Committee 318. Building code requirements for structural concrete. 2014.
[30] Chen B, Lu XL, Li PZ, Chen YQ. Modeling of dynamic soil-structure interaction by
ANSYS program. J Earthq Eng Eng Vib 2002;22:126–31.
References [31] Rayhani MH, El Naggar MH. Numerical modeling of seismic response of rigid
foundation on soft soil. Int J GeoMech 2008;8:336–46. https://doi.org/10.1061/
(asce)1532-3641(2008)8:6(336).
[1] ASCE. Seismic evaluation and retrofit of existing buildings. Virginia: Structural
[32] Chau KT, Shen CY, Guo X. Nonlinear seismic soil-pile-structure interactions:
Engineering Institute of American Society of Civil Engineers; 2017.
shaking table tests and FEM analyses. Soil Dynam Earthq Eng 2009;29:300–10.
[2] Trifunac MD, Stewart JP, Fenves GL, Seed RB. Seismic soil-structure interaction in
[33] Gazetas G. Vibrational characteristics of soil deposits with variable wave velocity.
buildings. I: analytical methods; seismic soil-structure interaction in buildings. II:
Int J Numer Anal Methods GeoMech 1982;6:1–20. https://doi.org/10.1002/
empirical findings. J Geotech Geoenviron Eng 2000;126:668–72. https://doi.org/
nag.1610060103.
10.1061/(asce)1090-0241. 2000)126:7(668.
[34] Pitilakis D, Dietz M, Wood DM, Clouteau D, Modaressi A. Numerical simulation of
[3] Stewart JP, Fenves GL, Seed RB. Seismic soil-structure interaction in buildings. I:
dynamic soil-structure interaction in shaking table testing. Soil Dynam Earthq Eng
analytical methods. J Geotech Geoenviron Eng 1999;125:26–37.
2008;28:453–67. https://doi.org/10.1016/j.soildyn.2007.07.011.
[4] Adam MA, Kamal OA, El-Hoseny M. Variation of seismic response of mid-rise RC
[35] Sulaeman A. The use of lightweight concrete piles for deep foundation on soft soils.
buildings due to soil structure interaction effects. Int J Civ Eng Technol 2016;7:
Universiti Tun Hussein Onn Malaysia; 2010.
220–40.
[36] Meymand Philip J. Shaking table scale model tests of nonlinear soil-pile-
[5] Adam M, Chouw N, Von Estorff O. Significance of soft local site in altering the non-
superstructure interaction in soft clay. Berkeley California: University of California;
uniform ground motions and the building response. Al-Azhar Univ Eng J 2004;8:
1994.
154–69.
[37] Moncarz PD. Theory and application of experimental model analysis in earthquake
[6] Veletsos AS, Meek JW. Dynamic behaviour of building-foundation systems. Earthq
engineering. 1981.
Eng Struct Dynam 1974;3:121–38.
[38] Iai S. Similitude for shaking table tests on soil-structure-fluid model in 1g
[7] Parkash S. Evaluation of soil-structure interaction effects of a massive embedded
gravitational field. Soils Found 1989;29:105–18.
structure during seismic excitation. In: Proc. 13th WCEE; 2004. vancouver, B.C.,
[39] Moss RES, Crosariol V, Kuo S. Shake Table Testing to quantify seismic soil-
Canada.
structure-interaction of underground structures. In: International conferences on
[8] Chowdhury Indrajit, Dasgupta SP. Dynamics of structure and foundation - a unified
recent advances in geotechnical earthquake engineering and soil dynamics. San
approach: 1. fundamentals. CRC Press; 2008.
Diego, California: Missouri University of Science and Technology; 2010.
[9] Stewart JP, Comartin C, Moehle JP. Implementation of soil-structure interaction
[40] Turan A, Hinchberger SD, El Naggar H. Design and commissioning of a laminar soil
models in performance based design procedures. 13th World Conference on
container for use on small shaking tables. Soil Dynam Earthq Eng 2009;29:404–14.
Earthquake Engineering. 2004. Vancouver, B.C., Canada.
https://doi.org/10.1016/j.soildyn.2008.04.003.
[10] El Hoseny M, Ma J, Josephine M. Effect of embedded basement stories on seismic
[41] Lee CJ, Wei YC, Kuo YC. Boundary effects of a laminar container in centrifuge
response of low-rise building frames considering SSI via small shaking table tests.
shaking table tests. Soil Dynam Earthq Eng 2012;34:37–51. https://doi.org/
Sustainability 2022;14:1275. https://doi.org/10.3390/su14031275.
10.1016/j.soildyn.2011.10.011.
[11] Forcellini D. Analytical fragility curves of shallow-founded structures subjected to
[42] Tabatabaiefar SHR, Fatahi B, Samali B. Numerical and experimental investigations
Soil-Structure Interaction (SSI) effects. Soil Dynam Earthq Eng 2021;141:106487.
on seismic response of building frames under influence of soil-structure interaction.
[12] Forcellini D. A novel framework to assess soil structure interaction (SSI) effects
Adv Struct Eng 2014;17:109–30. https://doi.org/10.1260/1369-4332.17.1.109.
with equivalent fixed-based models. Appl Sci 2021;11:10472.

14
M. El Hoseny et al. Soil Dynamics and Earthquake Engineering 164 (2023) 107583

[43] Cheung WM, Qin X, Chouw N, Larkin T, Orense R. Experimental and numerical Eng Geol Environ 2020;79:2949–69. https://doi.org/10.1007/s10064-020-01742-
study of soil response in a laminar box. In: Proceeding of 2013 NZSEE conference; w.
2013. [52] Forcellini D. Numerical simulations of liquefaction on an ordinary building during
[44] Tang L, Ling X, Xu P, Gao X, Wang D. Shake table test of soil-pile groups-bridge Italian (20 May 2012) earthquake. Bull Earthq Eng 2019;17:4797–823.
structure interaction in liquefiable ground. Earthq Eng Eng Vib 2010;9:39–50. [53] Forcellini D. Soil-structure interaction analyses of shallow-founded structures on a
https://doi.org/10.1007/s11803-009-8131-7. potential-liquefiable soil deposit. Soil Dynam Earthq Eng 2020;133:106108.
[45] Valsangkar AJ, Dawe JL, Mita KA. Earthquake engineering. In: Shake table studies [54] Mina D, Forcellini D. Soil–structure interaction assessment of the 23 November
of seismic response of single partially supported piles. University of Toronto Press; 1980 Irpinia-Basilicata earthquake. Geosciences 2020;10:152.
2019. p. 327–34. https://doi.org/10.3138/9781487583217-042. [55] Lysmer J, Kuhlemeyer RL. Finite dynamic model for infinite media. J Eng Mech Div
[46] Gohl WB, Finn WDL. Seismic response of single piles in shaking table studies. Proc 1969;95:859–77. https://doi.org/10.1061/jmcea3.0001144.
Fifth Can Conf Earthq Eng 1987:435–44. [56] Manafpour AR, Moradi V. Investigating conventional FE modelling for dynamic
[47] ASTM D6528. Astm D6528. Standard test method for consolidated undrained soil-structure interaction under horizontal and vertical ground motions. In: 15th
direct simple shear testing of cohesive soils. n.d. world conference on earthquake engineering, vol. 356. Lisbon: Lisbon Portugal;
[48] Brinkgreve RB. Plaxis 3D. Netherland: Plaxis Bv; 2012. 2012. p. 24–8.
[49] PEER PEER. Strong motion database. University of California at Berkeley; 2010. htt [57] Karabork T, Deneme IO, Bilgehan RP. A comparison of the effect of SSI on base
p://peer.berkeley.edu/peer_ground_motion_database. isolation systems and fixed-base structures for soft soil. Geomech Eng 2014;7:
[50] Goktepe F, Celebi E, Omid AJ. Numerical and experimental study on scaled soil- 87–103. https://doi.org/10.12989/gae.2014.7.1.087.
structure model for small shaking table tests. Soil Dynam Earthq Eng 2019;119: [58] Saez E, Lopez-Caballero F, Modaressi-Farahmand-Razavi A. Influence of 2d and 3d
308–19. https://doi.org/10.1016/j.soildyn.2019.01.016. soil modeling on dynamic nonlinear ssi response. 14th world conference on
[51] Goktepe F, Sahin M, Celebi E. Small shaking table testing and numerical analysis of earthquake engineering (14WCEE). 2008. p. 12–7. Beijing, China.
free-field site response and soil-structure oscillation under seismic loading. Bull [59] Kramer SL. Geotechnical earthquake engineering. Pearson Education India; 1996.

15

You might also like