on the resonance effect by dynamic soil–structure interaction: a revelation study
TRANSCRIPT
ORI GIN AL PA PER
On the resonance effect by dynamic soil–structureinteraction: a revelation study
Hamza Gullu • Murat Pala
Received: 27 May 2013 / Accepted: 7 January 2014� Springer Science+Business Media Dordrecht 2014
Abstract The present study makes an attempt to investigate the soil–structure resonance
effects on a structure based on dynamic soil–structure interaction (SSI) methodology by
direct method configuration using 2D finite element method (FEM). The investigation has
been focused on the numerical application for the four soil–structure models particularly
adjusted to be in resonance. These models have been established by single homogenous
soil layers with alternating thicknesses of 0, 25, 50, 75 m and shear wave velocities of 300,
600, 900 m/s-a midrise reinforced concrete structure with a six-story and a three-bay that
rests on the ground surface with the corresponding width of 1,400 m. The substructure has
been modeled by plane strain. A common strong ground motion record, 1940 El Centro
Earthquake, has been used as the dynamic excitation of time history analysis, and the
amplitudes, shear forces and moments affecting on the structure have been computed under
resonance. The applicability and accuracy of the FEM modeling to the fundamental period
of soils have been confirmed by the site response analysis of SHAKE. The results indicate
that the resonance effect on the structure becomes prominent by soil amplification with the
increased soil layer thickness. Even though the soil layer has good engineering charac-
teristics, the ground story of the structure under resonance is found to suffer from the larger
soil layer thicknesses. The rate of increment in shear forces is more pronounced on
midstory of the structure, which may contribute to the explanation of the heavily damage
on the midrise buildings subjected to earthquake. Presumably, the estimated moment ratios
could represent the factor of safeties that are excessively high due to the resonance con-
dition. The findings obtained in this study clearly demonstrate the importance of the
resonance effect of SSI on the structure and can be beneficial for gaining an insight into
code provisions against resonance.
H. Gullu (&)Department of Civil Engineering, University of Gaziantep, 27310 Gaziantep, Turkeye-mail: [email protected]
M. PalaDepartment of Civil Engineering, Adıyaman University, 02040 Adıyaman, Turkey
123
Nat HazardsDOI 10.1007/s11069-014-1039-1
Keywords Resonance � Soil–structure interaction � Finite element method �Site response analysis
1 Introduction
Resonance effect is an important subject in earthquake engineering practice. It is the result
of making the frequency of super structure to the frequency of supporting soil closer
(Takewaki 1988). This fact has been experienced in the several past earthquakes where
tuning of the natural period of a building structure with that of a surface ground caused
significant response amplifications on the buildings and resulted significant damage (The
earthquakes of 1970 Gediz, 1985 Mexico City, 1998 Adana-Ceyhan, etc.). Soil–structure
interaction (SSI) is a major topic that deals with the resonance phenomenon in detail. It
refers to the relationship between the characteristics of both the structure and the soil
stratum and is usually represented by modifying the dynamic properties of the structure.
This interaction causes energy dissipation and changes the natural modes of vibration of
the structure such as natural frequencies and the corresponding mode shapes (Wolf 1985;
Veletsos and Prasad 1989; Wenk et al. 1998).
It is increasingly desired to take into account the soil effects on the design of structures
particularly those located in active seismic zones. In recent years, numerous researchers
have performed studies on the effects of SSI on the dynamic seismic response of buildings
(Wolf and Song 2002; Aviles and Perez-Rocha 2003; Khalil et al. 2007). The analysis and
design process for dynamic loading generally assumes structures to be fixed at their bases.
However, supporting soil medium actually allows a movement to some extent due to
flexibility. This may reduce the overall stiffness of the structural system and may increase
the natural periods of the system. Considerable change in spectral acceleration with natural
period can be observed from the response spectrum curve. Such change in natural period
may considerably alter the seismic response of any structure (Stewart et al. 1999; Dutta and
Roy 2002; Dutta et al. 2004). Aviles and Perez-Rocha (1997) noted that SSI may be very
important for medium- and long-period structures when the predominant site periods are
large. Despite this dynamic SSI effects should be taken into account for stiff and/or heavy
structures supported on a relatively soft soil. These are generally small and may be
neglected for soft and/or light structures founded on stiff soils. In general, the natural
frequency of a soil–structure system due to SSI effects is obtained lower than the frequency
of the structure itself. In addition, radiation damping increases the total damping of a soil–
structure system comparing with the damping of the structure itself (Kramer 1996). Based
on the latest study of Khalil et al. (2007), it is apparently inferred that SSI effects directly
alter the resonance characteristics of the soil–structure system. Although many works have
been attempted about the SSI effects on soil and structures, there is limited number of work
that particularly involved the resonance effects from resonance models (Dutta et al. 2004).
Moreover, there is a lack of resonance study for models in resonance that systematically
examine the effects of soil layer thickness on the dynamic response of plane frame
structures under strong ground motion.
The aim of this work is to gain some insights into the reasons for earthquake damage to
engineered buildings due to the resonance effect on the basis of dynamic SSI methodology.
Investigation was carried out using some hypothetical SSI models which were adjusted so
the soil and structure were in resonance. The soil layer thickness in these models was
Nat Hazards
123
varied; however, a constant structure (a midrise building) resting on soil surface was used
throughout the study. The reason for choosing the midrise building is that majority of them
in previous severe earthquakes did not demonstrate good performance (Celebi 2000;
Sancio et al. 2002; Ulusay et al. 2004). Direct method configuration was used for the SSI
analysis. The SSI model was constructed by 2D finite element method with rectangular
meshes employing a common method of SAP2000. Site response analysis of soil layers to
the interaction from FEM modeling was carried out by the method of SHAKE. This study
is believed to contribute to engineers in practice when designing structures against
resonance.
2 A review on resonance effects from past earthquakes
The fundamental periods of structures can be crudely estimated from a rule of thumb
method in which the fundamental period of N-story building is approximately N/10 s
(Kramer 1996). They may range from about 0.05 s for a well-anchored piece of equipment,
0.1 s for a one story simple bent or frame, 0.5 s for a low structure up to about four stories
and between 1 and 2 s for a tall building from 10 to 20 stories (Arnold and Reitherman
1982). The fundamental periods of soils usually have values varying from 0.1 s (rock, stiff
or dense soils) to 1 s (soft or loose soils) (ICBO 1994). If the two fundamental periods are
matched each other, there is a high probability for the building will approach a state of
partial resonance (quasi-resonance). Experiences from historical earthquakes reflect that
long-period seismic waves from large-magnitude earthquake events can be amplified by
some four- to sixfolds due to resonance with flexible soil layers (Lam et al. 2001; Chandler
et al. 2002). The amplified motion may be subjected to further resonance with flexible tall
buildings where torsional inertia generated by dynamic coupling can create significant
horizontal rotation and result in a significant increase in the drift demand on individual
lateral load resisting elements. This torsional coupling effect of resonance is particularly
apparent in structures that respond elastically to an earthquake prior to initiation of damage
(Balendra et al. 2005).
The 1970 Turkish Gediz Earthquake demolished the paint workshop building of the
Tofas-Fiat automobile factory in Bursa, located 135 km away from the epicenter, while no
other building in Bursa was damaged. The main reason for the demolished structure was
found that the predominant periods of the structure and underlying soil were approximately
equal around a value of 1.2 s (Tezcan et al. 2002). Gullu (2001) obtained GIS-based
microzonation maps with respect to soil amplification and fundamental periods for the
town of Dinar comparing with the damages due to 1995 Dinar Earthquake. From the
fundamental period versus damage models, it is concluded that there is a likely resonance
effect which resulted in medium-heavy damage or collapse of the 3- to 4-story masonry
and reinforced concrete buildings which are overlaid by the soils with the predominant
periods of 0.3–0.5 s. In addition, 5- to 6-story reinforced concrete buildings over the sites
with the fundamental periods of 0.5–0.7 s may be damaged also by the resonance of soil–
structure. Similar results supporting the resonance effect in the damage during 1995 Dinar
Earthquake can be seen in Ansal et al. (2001). An investigation of 1998 Adana-Ceyhan
Earthquake indicates that one of the main causes of collapses or severe damage on the
midrise buildings (7–10 stories) in Ceyhan is double resonance. Because the spectrum of
recorded strong ground motion in Ceyhan has a predominant period frequency about
1.5 Hz (0.67 s), and predominant frequency of site conditions depicted by alluvial media is
found similar to the ones of structures and strong motion (Celebi 2000). A damage survey
Nat Hazards
123
conducted in Ceyhan found a strong correlation between the frequencies of site and
damaged buildings (Wenk et al. 1998). A recent study supports the previous ones such that
the resonance effect from the soil–structure or the ground motion soil–structure has been
important in the damage observed in Ceyhan (Yalcınkaya and Alptekin 2005).
1999 Kocaeli and Duzce Earthquakes attacked a densely populated area, Marmara
Region, in Turkey and severely shaked the region resulting devastating damages to
building and structures. Investigations on the localization of observed settlements around
buildings, the relative infrequency of observations of liquefaction in open fields and the
higher rate of severe ground failure for taller buildings indicate that ground strains asso-
ciated with soil–structure interaction may have contributed to the triggering and severity of
ground failure during the 1999 Kocaeli Earthquake (Sancio et al. 2002). Some previous
studies demonstrate a possible resonance effect on the damage of Fatih Mosque (Istanbul)
during the 1999 Kocaeli Earthquake. The predominant period of the structure (2.4 Hz NS–
2.5 Hz EW) was estimated as falling into the predominant period of seismic loading
(Beyen 2007). A seismic amplification study was conducted for the town of Avcılar
(Istanbul), located at about 120 km west from the epicenter of the 1999 Kocaeli Earth-
quake, using the NS component of this earthquake recorded at Izmit Meteorological
Station. It is found that the buildings at Avcılar, with the natural periods of vibration close
to anyone of 0.70, 1 and 1.60 s, are expected to experience relatively heavier damage due
to resonance effects as well as soil amplification (Tezcan et al. 2002). The resonance effect
can also be considered one of the contributing factors to the damage of some buildings in
Cay-Eber Earthquake of February 3, 2002. Fundamental periods of response spectra of the
strong ground motion recorded at Afyon station are found to be very close to the natural
period of tall structures which were collapsed or heavily damaged (Ulusay et al. 2004)
The 1985 Mexico City earthquake is one of the instructive earthquakes where the
resonance is well defined in many damaged buildings. The greatest damage occurred in the
Lake Zone underlain by soft soil (38–50 m depth) where the characteristics of site periods
were estimated from 1.9 to 2.8 s. Buildings less than five stories and modern buildings
greater than 30 stories were exposed to slight damage within this area. However, most of
the buildings in range from 5 to 20 stories, those fundamental periods were nearly equal to
or somewhat less than the characteristics site period, either were collapsed or badly
damaged. The possible reason for the damage was the double resonance (Kramer 1996).
It is shown from the past events that tuning of the natural period of a building structure
with that of a surface soil causes significant amplifications that result in the increasing of
inertial forces acting on the structure with a considerable damage. So, it is very important
to check the interactions between the vibration periods (or frequencies) of structures and
the supporting soil in order to determine how close they are to resonance.
3 Methodology of SSI analysis
SSI can significantly change the free-field ground motion at the foundation level and the
dynamic properties of the structure, while local site conditions may produce large
amplifications and important spatial variations in seismic ground motion (Veletsos and
Prasad 1989). In a dynamic SSI analysis, a bounded structure (which may be linear or
nonlinear), consisting of the actual structure and an adjacent irregular soil if present, will
interact with the unbounded (infinite or semi-infinite) soil which is assumed to be linear
elastic (Fig. 1). In SSI problems, the ability to predict the coupled response of the soil and
structure has great significance. Hence, SSI-related problems need a combination of soil
Nat Hazards
123
and structure models. Although structure models have a good basis in the literature, soil
models include complicated analysis due to their unbounded nature. In fact, the major
difficulty in modeling the soil region can be attributed to the characteristics of wave
propagation through the soil medium. Because soil has very complex characteristics due to
the heterogeneous, anisotropic and nonlinear natures in force versus displacement.
Moreover, the presence of fluctuations in water table further adds to its complexity (Dutta
and Roy 2002). Therefore, one of the most important objectives in a SSI study is to build
up reliable and easily implemented models (Wolf and Song 2002). The various rigorous
numerical methods developed for the analysis of dynamic SSI can be classified into two
main groups: the direct method and the substructure method. The direct method is applied
in this paper due to its simple implementation of SSI approach in a single by Eq. 1:
½M�f€ug þ ½K��fug ¼ �½M�f€uffðtÞg ð1Þ
where f€uffðtÞg is the specified free-field accelerations at the boundary nodal points, [M] is
the mass matrix, [K*] is the complex stiffness matrix and {u}is the vector of unknown
nodal point displacements (Kramer 1996).
In order to estimate SSI by the direct method, a finite bounded soil zone adjacent to the
structure (near-field) and the structure itself may be modeled using the finite element
method and the seismic free-field motion is applied at a fictitious interface enclosing the
soil–structure system. The effect of the surrounding unbounded soil (far-field) is approx-
imately analyzed by imposing transmitting boundaries along the fictitious (near-field/far-
field) interface (Zhang et al. 1999; Halabian and El Naggar 2002; Wegner et al. 2005). A
lot of transmitting boundaries have been developed to satisfy the radiation condition such
as a viscous boundary (Lysmer and Kuhlemeyer 1969), a superposition boundary (Smith
1974) and others (Liao and Wong 1984). The soil with the superstructure is modeled up to
the artificial boundary (Fig. 2), and the response of the soil and structure is determined
simultaneously by analyzing the idealized soil–structure system in a single step (Jaya and
Meher Prasad 2002). The artificial boundary (Fig. 2) is obtained by modifying the viscous
boundary (Lysmer and Kuhlemeyer 1969) to all degrees of freedom of the boundary nodes
for elastic wave propagation in semi-infinite medium. It is assumed that the wave energy
arrives at the boundary with equal probability from all directions. In order to get a proper
accuracy and reduce the effects of reflected waves by the transmitting boundary, it is
necessary to consider a large amount of soil around the structure when the direct method is
employed. In this way, the radiation condition is taken into account for an unbounded
medium. Thus, modeling of a significant part of the soil is an essential point in the direct
method. The distance between the artificial soil boundary and the building is usually
several times of the width of the structure. Applying a finite element mesh, the total
number of nodes in the unbounded soil medium will dominate to those of the soil–structure
system. Therefore, the direct method is usually used to study two-dimensional models
(Wegner et al. 2005). The mass matrix [Mb] of the bounded medium with degrees of
freedom on the boundary can be obtained as in structural dynamics:
½SbðxÞ� ¼ ½Kb� � x2½Mb� ð2Þ
where [Kb] and [Sb(x)] are the static-stiffness and dynamic-stiffness matrices of the
bounded medium, respectively (Halabian and El Naggar 2002).
FEM is usually applied for SSI problems because of its accuracy and convenient
standard algorithms in the public domain (Wegner et al. 2005). Even if care must be taken
about the possibilities of inaccuracy arising out of numerical limitations while interpreting
Nat Hazards
123
the results, it is the most powerful and versatile tool for solving SSI problems. The main
advantage of the FEM is that it can easily be employed for incorporating the effect of
material nonlinearity, nonhomogeneity and anisotropy of the supporting soil medium as
well as the geometry (Dutta and Roy 2002; Pitilakis et al. 2007). Even though the method
is able to treat soil domains of arbitrary layer geometry and to accommodate material
nonlinearity, anisotropy and inhomogeneity, it does not satisfy the radiation-toward-infi-
nity condition at the boundaries, a phenomenon inherent in SSI. Special boundary con-
ditions, as mentioned in the direct method, have to be developed to simulate the unbounded
nature of the soil medium. The FEM can be applied efficiently to geometries requiring
transmitting boundaries (Pitilakis et al. 2007). There are some finite element codes,
FLUSH (Lysmer et al. 1975), SASSI (Ostadan et al. 2000) and SAP2000, which are
designed specifically to perform SSI analyses incorporating an equivalent linear visco-
elastic model for the soil behavior. In this paper, SAP2000 was employed for the dynamic
Fig. 1 Problem definition of dynamic SSI (Wolf and Song 2002)
Fig. 2 Direct method configuration (Jaya and Meher Prasad 2002)
Nat Hazards
123
response of SSI by the FEM. The application of SAP2000 is extremely advantageous
because this numerical method significantly reduces the computational effort and makes
the solution of problem simpler.
4 Numerical application
In this study, 2D soil–structure models were adopted to account the resonance effect on the
response of reinforced concrete (RC) structure subjected to earthquake. The main reason
for the usage of 2D modeling is that 3D modeling deals with very complex motion
equations that produce a computationally intensive task and make the solution of SSI
problems more difficult even if it appears to be more reliable (Zhang et al. 1999). The
resonance effect from dynamic SSI was investigated concerning the four model cases
(Fig. 3) which were adjusted so the soil and the structure were in resonance. This allows us
to find out the resonance effects in the structure. Alternating soil layer thicknesses (SLT)
with 0 m (case 1), 25 m (case 2), 50 m (case 3) and 75 m (case 4) were assigned over the
bedrock. However, dynamic properties of the soil layers in shear wave velocity (Vs) were
determined by trial-and-error method where it was aimed to get shear wave values pro-
ducing the fundamental periods of soils close to the ones of the structure in order to have
the cases concerned in the resonance. Estimation of Vs is described in the paragraphs later.
In all cases, the structure was considered to rest on the ground surface. Here, it should be
noted that the bedrock in the SSI methodology of this study is considered to have relatively
high rigidity characteristics that define its shear wave velocity to become much greater than
the one of the soil layers.
A RC structure that is 18 m in height (H) with a 6-story (i.e., midrise building) and 9 m
in width (B) with a 3-bay was used for the models given in the cases by Fig. 3. The model
cases might give some insights for the reasons of the damage on those buildings. The story
heights and bay intervals of 3 m were taken to be constant in the frame. The structure has a
flexible shallow foundation that consists of a RC mat footing with 1 m depth. The sections
of structural elements were rectangular, and their dimensions were kept constant for all
stories. The cross sections of columns and beams were taken to be 400 9 400 mm and
200 9 600 mm, respectively. They were constant through the story height. The RC
structure was considered to be homogenous, isotropic and linear elastic with Young’s
modulus (modulus of elasticity) E = 28 GPa, Poisson’s ratio m = 0.20 and mass density
q = 24 kN/m3. The damping ratio of the structure was taken to be 8 %. Since plane strain
elements were employed in the ground, mass and stiffness of the structure were defined as
the quantities per unit length with respect to the out-of-plane direction.
For analyze purpose of the soil layers, isotropic, homogeneous and linear stress–strain
conditions were assumed. The damping ratio of 8 % was also used for the soil considering
the contribution of radiation and material damping for a mat footing feasible range of
footing size, by following the guideline prescribed in the literature (Gazetas 1991). Pois-
son’s ratio equal to 0.3 was selected, and the mass density of the soil layers taken into
account was 18 kN/m3. Some of the characteristics of soils and structure are summarized
in Table 1.
Following the suggestions of surveyed methodologies, while solving the SSI problem,
the side nodes of the discretized finite elements in the boundary are considered to be
connected with dashpots allowing only the horizontal movements and the bottom nodes are
considered to be pin support. The middle node at the base of the foundation is kept fixed in
order to prevent rigid body translation in the horizontal direction. The established SSI
Nat Hazards
123
model for this study was given in Fig. 4. In this SSI model, FEM was employed to
formulate the mass and stiffness matrices for the structural frames. Consistent mass matrix
was used to develop the formulation as accurate as possible. The soil layer under the
structure was modeled with maximum 7872 solid elements. The viscous boundary was
obtained by using link elements. Each element has four nodes with three degrees of
freedom in each node. The Lysmer and Waas transmitting boundary that is intended to
absorb the body surface waves on the lateral infinite boundary was used in the model
(Lysmer and Waas 1972). The Lysmer–Wass transmitting boundaries that consist of dash
pots were specified at the vertical edges of the finite element mesh to model radiation
damping and the artificial boundaries of the soil. The maximum dimension of the mesh of
solid elements was chosen from the literature (Lysmer et al. 1975). In order to increase the
accuracy of results as noticed by Wegner et al. (2005), a soil medium of 1,400 m in length
around the structure was taken into account. As mentioned above, SLT over the bedrock
was changed from zero to 75 m as the four cases of SSI model.
In the dynamic analysis of soil–structure models, one of the most common strong
ground motion records for dynamics research, the 1940 El Centro Earthquake, with a time
step of 0.02 s was used. The NS component of this earthquake ground motion that has
maximum amplitude of 0.319 g was selected as the excitation (Fig. 5).
By taking the characteristics of the structure, soil layers and FEM mesh employed above
into consideration, a trial-and-error method via FEM was performed to determine Vs of
SLT for the study cases (Fig. 3) such that the soil layers and structure are to be resonance.
Fig. 3 Cases in resonance considered in the study. The height and width of structures are 18 and 9 m whichare constant in all cases. The estimated shear wave velocities are 300 m/s (SLT = 25 m), 600 m/s(SLT = 50 m) and 900 m/s (SLT = 75 m). SLT soil layer thickness
Table 1 Some characteristics ofsoil and structure
Property Soil Structure
Poisson’s ratio 0.30 0.20
Unit weight (kN/m3) 18 24
Damping ratio (%) 8 8
Elasticity modulus (GPa) – 28
Nat Hazards
123
Vs of soil layers were estimated as 300, 600 and 900 m/s, respectively, for 25, 50 and 75 m.
Consequently, four resonance cases have been defined to examine the resonance effect on
the structure. In the case 1, the structure in the SSI model was solved for fixed base
condition (or cantilever); thus, it does not need to calculate Vs of soil. Following the Vs
estimations, the fundamental periods of the soil layers that represent to these estimated Vs
values have been also computed by FEM and obtained as 0.334 s (SLT = 25 m), 0.335 s
(SLT = 50 m) and 0.331 s (SLT = 75 m) for the soil layer thicknesses. Moreover, fun-
damental period of the employed structure that has H/B = 2 was computed for all soil
layers using FEM and estimated to be 0.333 s. The computation results of the study cases
are presented in Table 2 for VS of soil layers, in Table 3 for the fundamental periods of soil
and structure to be in resonance.
The applicability and accuracy of the numerical analysis of FEM for estimating the
fundamental periods of the soil layers have been verified with the computations of fun-
damental periods from the most common evaluation formula given by Eq. 3 and from the
site response analysis performed by SHAKE code (Schnabel et al. 1972) in both. Con-
sidering the shear waves traveling vertically upward through a single soil layer of thickness
Fig. 4 Finite element mesh and boundary conditions for the soil and structure in SAP 2000
Nat Hazards
123
SLT above bedrock, the fundamental period of horizontal vibration of the ground is given
by
T ¼ 4SLT
ð2n� 1ÞVs
ð3Þ
where n is an integer, 1, 2, 3,…, representing the various modes of vibration (Tezcan et al.
2002). The fundamental period of soil or characteristic site period depends on the thickness
and shear wave velocity of soil and provides a very useful indication of the period of
vibration at which the most significant amplification can be expected (Kramer 1996). For
Fig. 5 El Centro earthquake, May 1940, record used in the study (NS component, PGA = 0.32 g)
Table 2 Computed Vs of soillayers for the study cases, whichare to be in resonance, usingFEM from SSI model (the struc-ture rests on the ground surface)
Study case Depth of bedrock or soillayer thickness (SLT) (m)
Estimated Vs (m/s)
Case 1 0 (fixed or cantilever) –
Case 2 25 300
Case 3 50 600
Case 4 75 900
Table 3 Computed fundamentalperiods of soils and structure,which are to be in resonance,using FEM from SSI model
Study case Fundamental period (s) by FEM
Soil Structure (H/B = 2,H = 18 m, B = 9 m)
Case 1 (Fixed orcantilever)
(Fixed or cantilever)
Case 2 0.334 0.333
Case 3 0.335 0.333
Case 4 0.331 0.333
Nat Hazards
123
the first mode of vibration using Eq. 3, the fundamental periods of all layers (SLT 25, 50
and 75 m) were evaluated as 0.333 s.
SHAKE performs one-dimensional shear wave propagation analysis and computes the
response in a homogeneous and viscoelastic soil layers of infinite horizontal extent sub-
jected to vertically traveling shear waves. The nonlinear soil behavior is accounted by the
use of equivalent linear soil properties using an iterative procedure to obtain the values for
modulus and damping compatible with the strains developed in each layer. In this work,
the variation in dynamic shear modulus and damping ratio of the soil layers with respect to
the shear strain was established from the past studies (Seed and Idriss 1970; Vucetic and
Dobry 1991). The effective strain ratio was taken to be equal to 0.65. Using the employed
strong ground motion record (Fig. 5) as the input ground surface motion (i.e., object
motion), the wave propagation analysis by SHAKE was conducted for the soil profiles of
SLT = 25, 50 and 75 m, and a hypothetical record at the bedrock was obtained first. Next,
this hypothetical record as the bedrock input motion (i.e., object motion) was transferred to
the ground surface, and the free surface motion was computed on the ground surface as a
site response to the structure. Fast Fourier Transform algorithm was utilized to analyze the
earthquake record. The hypothetical bedrock motions for all soil layers were obtained as
same with each other (Fig. 6). As for the hypothetical ground surface motion, it was
calculated to be exactly equal to the real motion given in Fig. 5 for all soil layers.
Accordingly, the spectral amplitudes and amplifications for the soil layers due to site
response analysis were also estimated as same with each other on the ground surface. The
computed absolute spectral acceleration, spectral velocity and spectral displacement for the
bedrock and free ground surface are shown in Fig. 7. As seen from Fig. 7, that the surface
ground motion has greater amplitudes than the bedrock motion due to local site effects.
These amplitudes are more prominent up to the vibrations of 1-s period. This may be
attributed to the rigidity characteristics of soil. The amplifications seem more prominent at
the accelerations as compared to ones of velocity and displacement likely due to earth-
quake nature of response. Subsequent to the analysis by the SHAKE, the fundamental
periods of soil layers were determined using spectral amplitude ratios of the absolute
accelerations based on the study of Borcherd (1970), in which the spectral amplitude ratio
is calculated by the ratio of Fourier amplitude spectrum of a soil site to the one of a nearby
hard-rock site (reference site). Since this method is mostly offered when the distance
between the two sites is much smaller than their epicenter distances, it could be accurately
applied for this study. The estimated spectral amplification using spectral ratios of the
absolute acceleration for the soil layers is presented in Fig. 8. It is clearly appeared from
-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
0 2 4 6 8 10
Time (sec)
Bed
rock
Acc
eler
atio
n (g
)
Fig. 6 The estimatedhypothetical bedrock motion bySHAKE for the soil layers of 25,50 and 75 m (it was obtained assame with each other for all thelayers)
Nat Hazards
123
the Fig. 8 that the fundamental periods of the soil layers are nearly equal to 0.3 s. This
period is very close to the fundamental period of the midrise buildings and may be
beneficial for explanation of the reasons to the damage on the midrise buildings in the
earthquakes.
The evaluated fundamental periods from both the Eq. 3 and the SHAKE (site response
analysis) demonstrate that they are nearly equal to the ones obtained from the FEM. This
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Period (sec)
Abs
olut
e A
ccel
erat
ion
(g)
BedrockSurface
0
10
20
30
40
50
60
70
80
0.01 0.1 1 10
0.01 0.1 1 10
Period (sec)
Rel
ativ
e V
eloc
ity (
cm/s
ec)
0
5
10
15
20
25
1010.1
Period (sec)
Rel
ativ
e D
ispl
acem
ent (
cm)
Fig. 7 Spectral amplitudes forthe soil layers of 25, 50 and 75 m
Nat Hazards
123
relatively confirms the results of FEM in the adopted SSI model. It is realized from the
fundamental periods that SSI models for all study cases are in resonance, and the resonance
effect on the structures could be studied based on the parameters such as displacement,
velocity, acceleration, shear force and moment capacity of the structure. Consequently, the
particular role of resonance effect on the structural damage from the soil–structure inter-
action could be attempted and properly discussed.
5 Results and discussions
Time history analysis of dynamic soil–structure models for the study cases in resonance
was performed, and influence of the resonance on the RC structure for the alternating SLT
was investigated. For the case of SLT = 0 m (case 1), the RC structure was assumed to be
fixed base (cantilever). The resonance effects on the RC structure were discussed in
accordance with the outcomes of displacement, velocity, acceleration, shear force and
bending moment.
The soil effect on the structure in the resonance cases can be obviously seen by the time
history results of acceleration and displacement (Fig. 9). It is shown from the figure that
the influence of resonance increases with the increased SLT. This clearly emphasizes the
role of SLT inherently associated with the shear wave velocities of soils. From the analysis
of SSI, a careful study of the time history results considering the height of story versus the
amplitude (i.e., displacement, velocity and acceleration) and the amplitude ratio (i.e., the
ratios of displacement, velocity and acceleration of SLT to the fixed base) is given in
Figs. 10, 11, 12. The results reveal the observations of remarkable more increasing in the
lateral displacement of SSI models in resonance case (Fig. 10). As mentioned above, the
resonance effect changes with respect to the soil layer thickness. The increase in SLT of
25, 50 and 75 m is relatively greater than the fixed base structure (SLT = 0 m). However,
increment rate appears nearly to be similar for the higher SLT (from 50 to 75 m) where the
shear wave velocities are varied from 600 to 900 m/s or may be to higher values. This can
be clearly seen by the results from Fig. 9, which is attributed to the stiffness of the soil. The
ratios of story lateral displacement of SSI models to the fixed base model on the ground
surface (or ground story) are 5.5, 4.7 and 2.7, respectively, for the soil layer thicknesses of
75, 50 and 25 m. These ratios are slightly increasing toward to higher story (Fig. 10).
Similar observations can be obtained in the amplitudes of velocity and acceleration such
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.01 0.1 1 10
Period (sec)
Spe
ctra
l Am
plifi
catio
n
Fig. 8 Spectral amplification forthe soil layers of 25, 50 and 75 m
Nat Hazards
123
that the velocity and acceleration of story are increased with respect to SLT as the structure
height is raised (Figs. 11, 12). However, the increment ratio with respect to the fixed base
is more pronounced on the ground floor. This is an expected result that may be attributed to
the effect of shaking which often occurred on the ground surface intensively. It is clear that
one of the contributing factors to the resonance effect on the RC structure in this study is
the site amplification which was approximately computed as 2.8 from the site response
analysis of SHAKE (Fig. 8). Its size seems to be cable of creating a larger resonance effect.
Fig. 9 The time history versus amplitude at the top of structure with respect to SLT
Nat Hazards
123
This amplification value has similar amplification and fundamental period with the
response spectra of strong ground motion of 2002 Cay-Eber Earthquake (Ulusay et al.
2004) where collapses and heavy damages occurred in some of the midrise buildings
(Yesilcay apartment blocks) due to the resonance phenomenon as previously mentioned. It
is evident that local site conditions especially the soil layer thickness and shear wave
velocity of soil deposit play a significant role on the seismic response of buildings under
the resonance conditions. As a consequence of Figs. 9, 10, 11, 12, it is demonstrated that
the RC structure produces similar responses under the resonance cases of higher SLTs (i.e.,
[50 m) or higher shear wave velocities of soil (i.e., [600 m/s).
Distribution of shear force and bending moment of the SSI models for the fixed base and
three different SLT cases was investigated by selecting outside column at left side. The
shear forces of selected column for all the SSI models and the ratios of story shear forces of
models to the fixed base model are given in Figs. 13, 14, respectively. It can be seen from
Fig. 13 that the shear force on the ground floor of the RC building proportionally increases
with the increased SLT under the resonance. As the height of building increases, the shear
force on the RC structure decreases as expected. It is interesting to note that the top story of
the building produces similar shear forces no matter how SLT is varied (or shear wave
velocity of soil is varied). It is observed from the ground floor that the larger SLTs (i.e.,
[50 m) or larger Vs values (i.e., [600 m/s) produce similar effects of the shear force
probably due to the similar rigidity characteristics of soil. The findings from Fig. 13 clearly
indicate that even though the SLT owns good rigidity characteristics (i.e., Vs [ 600 m/s), it
may be capable of producing considerable shear forces by the larger soil layer thicknesses
(i.e., [50 m) on the ground story under the resonance conditions, for the RC building
investigated in this study.
As for the examination of the results from Fig. 14, that it can obviously be seen that the
rate of shear force increments is more pronounced on the midstory as compared with the
Fig. 10 a The storydisplacements of SSI modelswith respect to SLT. b The ratiosof story lateral displacement ofSSI models to the fixed basemodel (cantilever) with respect toheight of story
Nat Hazards
123
ones of remaining storys. This may be explained by means of the energy absorption
processing throughout the structural system of the RC building considered in this inves-
tigation (Lysmer and Kuhlemeyer 1969; Zhang et al. 1999). Some of the energy is
absorbed by the loop of shear forces, but this is probably less than the one from the loop of
bending moments on the structure. The shear failure developed in the structural system is
brittle and has no ductility. As a result of the shear failure, the strength of structural
elements in the buildings rapidly decreased. This may provide some contribution to the
explanation of the heavily damage on the midrise buildings under the resonance at some
past earthquakes (1998 Adana-Ceyhan Earthquake, 1999 Kocaeli, etc.). As mentioned
before, Celebi (2002) resulted that the collapses and heavy damage on the midrise
buildings in 1998 Adana-Ceyhan Earthquake were possibly occurred due to double reso-
nance. However, a possible change in the rate of shear forces in the midstorys might trigger
progression of the damage and increase the severity of damage or collapses as well as the
effects from double resonance. The findings from Fig. 14 reveal that since the moment
failure in the structure mostly becomes ductile, it is very important that the energy
absorption throughout the structural system is more convenient to be carried out by the
loop of moments.
The story bending moments of the RC structure are presented in Fig. 15. It is realized
from the figure that the moments in the cantilever case are significantly lower than the ones
from all the other cases of SLT. It is clear that the moments in the storys decrease with the
increased height of story. Moreover, the moments of higher SLTs ([50 m) or higher shear
wave velocities ([600 m/s) in the storys appear to be very close to each other. From the
results of Fig. 15, it can be said that the ground floor of the RC structure under the
resonance can be considerably suffered from the larger soil layer thicknesses (i.e.,[50 m),
Fig. 11 a The story velocities ofSSI models with respect to SLT.b The ratios of story velocities ofSSI models to fixed base modelwith respect to SLT
Nat Hazards
123
similar to the finding from Fig. 13, even though the soil layer has good engineering
characteristics (i.e., Vs [ 600 m/s).
The ratios of story lateral bending moments of SSI models to the fixed base model are
shown in Fig. 16. When focusing on the midstorys, it can be observed that the normalized
moment ratios do not produce a significant increment rate for the remaining storys below.
Fig. 12 a The storyaccelerations of SSI models withrespect to SLT. b The ratios ofstory accelerations of SSI modelsto fixed base model with respectto SLT
Fig. 13 The story shear forcesof SSI models with respect toSLT
Nat Hazards
123
This can be interpreted as robust evidence for the role of the shear forces of midstorys on the
heavily damage in the resonance cases, as mentioned before (Fig. 14). The moment ratios
(Fig. 16) are approximately yielded as 3, 4.5 and 5.5 for the SLTs of 25, 50 and 75 m,
respectively. Presumably, these ratios can be proposed for representing the factor of safeties
for the design of RC structure under the resonance. However, the factor of safety from the
moment ratios appears to be excessively high, which obviously indicates to higher bending
moments and large response amplification of the structure as well as the soil amplification. As
a result of Fig. 16, it can be concluded that the buildings designed for the resonance condi-
tions require high factor of safeties that could be proportionally involved with the normalized
moment ratios obtained in this study. But, in reality, it would not be economical approach to
design the buildings with the high factor of safety. Hence, increasing the rigidity of buildings,
using some instruments and techniques to absorb the energy and base isolation systems can be
used to decrease the resonance effect on the structures.
As well as the notifications in the past studies (Dutta and Roy 2002; Dutta et al. 2004), it
can be emphasized that the most conventional attempt in practice during the design process of
buildings is usually to ignore the SSI effect and to assume that the bedrock is to be fixed base.
However, using the SSI methodology in this study obviously shows that the responses on the
structure under the resonance are considerably influenced by the local site conditions (SLT,
Vs). Therefore, the buildings subjected to a possible resonance should be carefully designed
by taking the SSI effect into account. It should be noted that from a methodological point of
view, other investigations should be accomplished. The results presented here correspond to
only the excitation of 1940 El Centro Earthquake. Thus, a further work is needed particularly
to assess the earthquake dependence of resonance using different strong ground motion
records. The cases studied in the resonance situation can be surveyed in the aspect of multiple
layered soils rather than the single ones. This investigation is a numerical-based study,
however, can be extended to a laboratory work which may improve the applied methodology
as well as it may produce guidelines for the effect of SSI in a future course of study. Even
though all these tasks may enhance our insight on the resonance, the complicated nature of
resonance always has to be regarded in design. It is believed that the findings in this study are
valuable and will help for code provision purposes as well as attract attention to engineers in
practice to be careful when designing structures against resonance.
Fig. 14 The ratios of story shearforces of SSI models to fixedbase model with respect to SLT
Nat Hazards
123
6 Conclusion
In this preliminary study, the effects of resonance on the response of structure under the
seismic loading in the elastic range of vibration have been investigated by performing 2D
dynamic SSI analysis. The 4 soil–structure models (single homogenous soil layers with a
RC structure resting on the ground surface) adjusted to be in resonance have been used in
the investigation (Fig. 3). Direct method has been applied for the SSI analysis by using
Fig. 15 The story bendingmoments of SSI models withrespect to SLT
Fig. 16 The ratios of storybending moments of SSI modelsto fixed base model with respectto SLT
Nat Hazards
123
FEM for modeling the entire soil–structure system (Fig. 4). The applicability and accuracy
of the FEM for the fundamental periods of soil layers were confirmed by the most common
evaluation formula (Eq. 3) and the site response analysis by SHAKE (Figs. 6, 7, 8).
The investigation results that (Figs. 9, 10, 11 12, 13, 14, 15, 16) the resonance effect
(i.e., the amplitudes, shear force and moment) on the RC structure increases with the
increased SLT. For higher SLTs, this effect seems close to each other. The responses under
resonance are strongly influenced by soil amplification (Figs. 9, 10, 11, 12). Even though
the soil layer has good engineering characteristics (i.e., Vs [ 600 m/s), the ground floor of
the RC structure under the resonance can be considerably damaged from the larger soil
layer thicknesses (i.e.,[50 m) (Figs. 13, 14, 15, 16). The rate of shear force increments are
more pronounced on the midstorys as compared with the ones of remaining storys
(Fig. 14). This may contribute at the explanation of the heavy damage on the midrise
buildings under the resonance at some past earthquakes. The moment ratios (Fig. 16) are
likely supposed to be factor of safeties which are calculated relatively high due to the
resonance effect on the RC structure, as expected. The overall evaluation of this investi-
gation reveals that the resonance effects estimated from the SSI analysis fairly produce
greater responses on the RC structure. The practical relevance of the findings obtained in
this study can be considered to be high. They can be beneficial for gaining an insight into
code provisions as well.
Acknowledgments This study is supported by The Scientific Research Project Unit of University ofGaziantep. Dr K.Hazirbaba is gratefully acknowledged by the corresponding author of this paper forproviding post-doctorate fellowship at his research project (Grant No. G3238-33650) at University of AlaskaFairbanks. The authors are grateful to the anonymous reviewers for carefully reviewing the manuscript andproviding valuable comments.
References
Ansal A, Iyisan R, Gullu H (2001) Microtremor measurements for the microzonation of Dinar. Pure applGeophys 158(12):2525–2541
Arnold C, Reitherman R (1982) Building configuration and seismic design. Wiley, New YorkAviles J, Perez-Rocha LE (1997) Site effects and soil-structure interaction in the Valley of Mexico. Soil Dyn
Earthq Eng 17:29–39Aviles J, Perez-Rocha LE (2003) Soil–structure interaction in yielding systems. Earthq Eng Struct Dyn
32:1749–1771Balendra T, Lam NTK, Perry MJ, Lumantarna E, Wilson JL (2005) Simplified displacement demand
prediction of tall asymmetric buildings subjected to long-distance earthquakes. Eng Struct 27:335–348Beyen K (2007) Structural identification for post-earthquake safety analysis of the Fatih mosque after the 17
August 1999 Kocaeli earthquake. Eng Struct 30(8):2165–2184Borcherdt RD (1970) Effects of local geology on ground motion near San-Francisco Bay. Bull Seismol Soc
Am 60:29–61Celebi M (2000) Revelations from a single strong-motion record retrieved during the 27 June 1998 Adana
(Turkey) earthquake. Soil Dyn Earthq Eng 20:283–288Chandler AM, Lam NTK, Sheikh N (2002) Response spectrum predictions for potential near-field and far-
field earthquakes affecting Hong Kong: soil sites. Soil Dyn Earthq Eng 22:419–440Dutta SC, Roy R (2002) A critical review on idealization and modeling for interaction among soil–
foundation–structure system. Comput Struct 80:1579–1594Dutta SC, Bhattacharya K, Roy R (2004) Response of low-rise buildings under seismic ground excitation
incorporating soil–structure interaction. Soil Dyn Earthq Eng 24:893–914Gazetas G (1991) Formulas and charts for impedances of surface and embedded foundations. J Geotech Eng
ASCE 117(9):1363–1381Gullu H (2001) Microzonation of Dinar with respect to soil amplification by using geographic information
systems. Ph.D. Thesis, Istanbul Technical University, Institute of Science and Technology, p 289
Nat Hazards
123
Halabian AM, El Naggar MH (2002) Effect of non-linear soil–structure interaction on seismic response oftall slender structures. Soil Dyn Earthq Eng 22:639–658
ICBO (1994) Uniform Building Code. 1991 International Conference of Building Officials, Whitter,California
Jaya KP, Meher Prasad A (2002) Embedded foundation in layered soil under dynamic excitations. Soil DynEarthq Eng 22:485–498
Khalil L, Sadek M, Shahrour I (2007) Influence of the soil–structure interaction on the fundamental periodof buildings. Short Communication. Earthq Eng Struct Dyn 36:2445–2453
Kramer S (1996) Geotechnical earthquake engineering. Prentice Hall, New Jersey, p 653Lam NTK, Wilson JL, Chandler AM (2001) Seismic displacement response spectrum estimated from the
frame analogy soil amplification model. J Eng Struct 23:1437–1452Liao ZP, Wong HL (1984) A transmitting boundary for the numerical simulation of elastic wave propa-
gation problems. J Comput Phys 3:174–183Lysmer J, Kuhlemeyer RL (1969) Finite dynamic model for infinite media. J Eng Mech Div ASCE
95:859–877Lysmer J, Waas G (1972) Shear wave in plane infinite structure. J Eng Mech Div ASCE 98(EM1):85–105Lysmer J, Udaka T, Tsai CF, Seed HB (1975) FLUSH: a computer program for approximate 3-D analysis of
soil–structure interaction problems. Report EERC-75-30, Earthquake Engineering Research Center,University of California, Berkeley, CA, USA
Ostadan F, Chen CC, Lysmer J (2000) SASSI2000—a system for analysis of soil–structure interaction.University of California, Berkeley, CA, USA
Pitilakis D, Dietz M, Muir Wood D, Clouteau D, Modaressi A (2007) Numerical simulation of dynamicsoil–structure interaction in shaking table testing. Soil Dyn Earthq Eng 28(6):453–467
Sancio RB, Braya JD, Stewart JP, Youd TL, Durgunoglu HT, Onalp A, Seed RB, Christensen C, BaturayMB, Karadayılar T (2002) Correlation between ground failure and soil conditions in Adapazari,Turkey. Soil Dyn Earthq Eng 22:1093–1102
Schnabel PB, Lysmer J, Seed HB (1972) SHAKE: A Computer Program for Earthquake Response Analysisof Horizontally Layered Sites. Report No. UCB/EERC-72/12. Earthquake Engineering ResearchCenter, University of California, Berkeley. December
Seed HB, Idriss IM (1970) Soil moduli and damping factors for dynamic response analyses. ReportNo:EERC-70-10. University of California, Berkeley, California
Smith WD (1974) A nonreflecting plane boundary for wave propagation problems. J Comput Phys15:492–503
Stewart JP, Fenres GL, Seed RB (1999) Seismic soil–structure interaction in buildings I: analytical method.J Geotech Geoenviron Eng 125(1):26–37
Takewaki I (1988) Remarkable response amplification of building frames due to resonance with the surfaceground. Soil Dyn Earthq Eng 17:211–218
Tezcan S, Kaya E, Bal E, Ozdemir Z (2002) Seismic amplification at Avcılar, Istanbul. Eng Struct24:661–667
Ulusay R, Aydan O, Erken A, Tuncay E, Kumsar H, Kaya Z (2004) An overview of geotechnical aspects ofthe Cay-Eber (Turkey) earthquake. Eng Geol 73:51–70
Veletsos AS, Prasad A (1989) Seismic interaction of structures and soils: stochastic approach. J Struct EngASCE 115:935–956
Vucetic M, Dobry R (1991) Effect of soil plasticity on cyclic response. J Geotech Eng ASCE 117(1):89–107Wegner JL, Yao MM, Zhang X (2005) Dynamic wave–soil–structure interaction analysis in the time
domain. Comput Struct 83:2206–2214Wenk T, Lacave C, Peter K (1998) The Adana-Ceyhan earthquake of June 27, 1998. Reconnaissance Report
of the Swiss Society for Earthquake Engineering and Structural Dynamics, Zurich, SwitzerlandWolf JP (1985) Dynamic soil-structure interaction. Prentice-Hall, Englewood CliffsWolf JP, Song CH (2002) Some cornerstones of dynamic soil–structure interaction. Eng Struct 24:13–28Yalcınkaya E, Alptekin O (2005) Site effect and its relationship to the intensity and damage observed in the
June 27, 1998 Adana-Ceyhan Earthquake. Pure Appl Geophys 162:913–930Zhang X, Wegner JL, Haddow JB (1999) Three dimensional soil–structure–wave interaction analysis in
time domain. Earthq Eng Struct Dyn 36:1501–1524
Nat Hazards
123