Sunteți pe pagina 1din 20

Soil Dynamics and Earthquake Engineering 92 (2017) 706725

Contents lists available at ScienceDirect

Soil Dynamics and Earthquake Engineering


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

Eect of dierent aspects of multiple earthquakes on the nonlinear behavior MARK


of RC structures

F. Hosseinpour , A.E. Abdelnaby
Department of Civil Engineering, The University of Memphis, Memphis, TN 38152, USA

A R T I C L E I N F O A BS T RAC T

Keywords: Observations from eld investigations have shown that a structure may be subjected to more than one
Degrading models earthquake in a relatively short period of time. Since after every single earthquake, a structure faces a stiness
Multiple earthquakes and strength degradation, it may not be able to withstand subsequent shaking especially when seismic retrot is
Vertical earthquake component not an option due to the short time intervals between the successive earthquakes. Therefore, it is necessary to
Earthquake direction
consider eects of repeated shaking on the behavior of structures prone to multiple earthquakes. Although
Aftershock polarity
Damage
many studies have been performed so far, there are some signicant decits which need more attention. Some
Irregularity of the important limitations in most of the recent studies include: (1) the use of ground motion data from
uncorrelated events occurring at dierent locations and times; (2) utilization of simple models that do not
contain appropriate damage features; and (3) no proper consideration to the eects of earthquake direction,
aftershock polarity, structure irregularity, and the vertical component of the earthquake. The objective of this
study is to overcome the aforementioned limitations through using a robust nite element model that simulates
the degrading behavior of reinforced concrete structures subjected to as-recorded seismic sequences. For this
purpose, the nonlinear response of two eight-story reinforced concrete buildings (both regular and irregular in
height) is evaluated. The buildings are subjected to a suite of ground motion records obtained from the recent
20102011 Christchurch earthquake sequence. Dynamic response history analyses are conducted to investigate
the eects of: 1) damage from previous ground motions; 2) earthquake direction; 3) aftershock polarity; and 4)
the vertical component of the earthquake. The results presented in this study indicate that earthquake direction
(in the irregular building), structure irregularity, and the vertical earthquake component can have a
considerable eect on the response of structures subjected to multiple earthquakes. Findings also show that
aftershock polarity can signicantly change the response of the irregular structure.

1. Introduction Christchurch (New Zealand, 20102011), Chile (2010), Nepal


(2015), and the most recent one, Kumamoto (Japan, 2016). The latter
Multiple earthquakes occur at many regions around the world sequences occurred at the time of preparation of the present paper and
where complex fault systems exist. These fault systems usually do not are still ongoing. Two foreshocks of magnitude 6.2 and 6 followed by a
relieve all accumulated strains at once when the rst rupture takes magnitude 7 main shock between 14 and 15 April 2016 (UTC) caused
place. Therefore high stresses form at dierent locations causing severe damage and injuries. Furthermore many casualties have been
sequential ruptures until the fault system is completely stabilized. reported for Kumamoto earthquakes so far. Fig. 1 shows the earth-
The sequential ruptures along the fault segment(s) lead to multiple quakes with magnitudes more than 4 in Kyushu (the region that
earthquakes which are often hard to distinguish them as fore-, main- Kumamoto city is located) between 14 and 17 April 2016 (UTC) based
and after-shocks, or a sequence of earthquakes from proximate fault on the information in IRIS [2]. As it can be seen in Fig. 1, there are 51
segments [1]. Since usually the time spans between seismic sequences earthquakes (with Mw >4) in about 84 h. It should be noted that some
are short and there is not enough time to retrot the structure against earthquake magnitudes were based on body-wave magnitudes (mb)
subsequent ground motions, the structures suer from stiness and and the others were based on moment magnitude (Mw). Therefore,
strength degrading as a result of accumulated damage. Scordilis relationship [3] was employed to convert mb to Mw as
Damage of structures due to successive shaking has been reported follows:
in recent earthquake sequences including Tohoku (Japan, 2011),


Corresponding author.
E-mail address: fhssnpur@memphis.edu (F. Hosseinpour).

http://dx.doi.org/10.1016/j.soildyn.2016.11.006
Received 1 May 2016; Received in revised form 26 August 2016; Accepted 1 November 2016
Available online 24 November 2016
0267-7261/ 2016 Elsevier Ltd. All rights reserved.
F. Hosseinpour, A.E. Abdelnaby Soil Dynamics and Earthquake Engineering 92 (2017) 706725

Fig. 1. Fore-, main- and aftershocks of Kumamoto-Japan earthquake in Kyushu region between 14 and 17 April 2016 (UTC).

Fig. 2. Map of central Canterbury showing the Greendale fault surface trace (black line), the epicenter of the dareld earthquake (blue star), and the vertical PGAs at selected strong-
motion stations. (For interpretation of the references to color in this gure legend, the reader is referred to the web version of this article.) [42].

Mw=0. 85( 0. 04)mb +1. 03((0. 23), 3. 5mb6. 2 (1) Aschheim and Black [4] employed degrading systems for the rst
time to study the eect of damage from previous earthquake on the
Almost none of the existing design codes include a reliable method response of SDOF structures. They demonstrated that the eect of
to consider the eect of seismic sequences (multiple earthquakes) on previous earthquake on the peak displacement of the SDOF system is
the behavior of structures. The most widely used approach for seismic not signicant and displacement responses of the damaged and
design of structures is based on a single (strongest) earthquake. In undamaged SDOF structures are similar after the structure experiences
recent years, researchers investigated the eect of seismic sequences on the peak displacement. Amadio et al. [5] investigated nonlinear
dierent aspects of inelastic behavior of structures [1,435]. behavior of SDOF systems under multiple earthquakes. They employed

707
F. Hosseinpour, A.E. Abdelnaby Soil Dynamics and Earthquake Engineering 92 (2017) 706725

Fig. 3. Main shock and aftershocks of Christchurch earthquake 2010 (source: GNS Science [43]).

dierent hysteretic models and showed that multiple earthquakes can completely changed the response of the structures. Findings also
cause signicant damage accumulation. They also indicated that the demonstrated that the nonlinear behavior of RC structures is signi-
most vulnerability is related to the elastic-perfect plastic systems and cantly aected by multiple earthquakes. Raghunandan et al. [10] also
hence they can be considered as the controlling system. In this study, a studied aftershock collapse vulnerability of RC frame structures. They
moment resisting steel frame was also analyzed under seismic indicated that collapse capacity of the structure under aftershock will
sequences. Findings showed a more reduction in behavior factor for be reduced signicantly if the damage level resulting from the main
the MDOF structure than the equivalent SDOF system is observed. shock is signicant. They also employed dierent physical damage
Hatzigeorgiou and Liolios [6] studied nonlinear behavior of RC frames measures to see their eectiveness in predicting the aftershock collapse
under repeated strong ground motions. They employed 45 seismic capacity of the damaged buildings. They divided the damage indicators
sequences to investigate the inelastic behavior of eight RC frames into two groups: (1) Observable; and (2) Non-observable. Based on a
including both regular and vertically irregular structures. The results of considered ranking, it was concluded that the drift of the damaged
this study showed that the response and design of RC frames are building is the best damage measure. Faisal et al. [11] investigated the
signicantly aected by seismic sequences. They also proposed an eect of multiple earthquakes on ductility demands of three-dimen-
empirical expression to estimate the ductility demands under multiple sional inelastic concrete frames with dierent number of stories (3, 6,
earthquakes using a simple combination of ductility demands under 12, and 18 stories) and observed a signicant increase in the ductility
single earthquakes. Ruiz-Garcia and Negrete-Manriquez [7] evaluated demands when considering multiple earthquakes. They also proposed
the drift demands in the steel frames with dierent number of stories empirical relationships to predict maximum story ductility demands of
under as-recorded seismic sequences (both far-eld and near eld). the considered frames under seismic sequences. Zhang et al. [12] also
The results showed that the frequency contents of the main shock and evaluated the damage of concrete gravity dams under as-recorded
aftershock are signicantly dierent. They indicated that as-recorded multiple earthquakes and concluded that accumulated damage and
aftershocks do not have an important eect on the increase of peak and design of concrete gravity dams are highly aected by as-recorded
residual drift demands and that articial sequences can cause an seismic sequences. They also proposed that considering the isolated
overestimation in median peak and residual drift demands. design earthquake underestimates the dam damage and hence the
Abdelnaby [1] and Abdelnaby and Elnashai [8,9] studied the eects traditional codes need to be modied and multiple earthquakes should
of multiple earthquakes on degrading RC structures. They used ber be considered in these codes. Hatzivassiliou and Hatzigeorgiou [13]
based models to investigate the nonlinear behavior of concrete studied the eects of seismic sequences on three-dimensional rein-
structures under both articial and as-recorded seismic sequences. In forced concrete buildings. They observed that the seismic demands are
this study dierent degrading features for steel and concrete were increased by multiple earthquakes. They also investigated the eect of
considered. Based on the results of this study, degrading models siting conguration on the ductility demand and indicated that the

708
F. Hosseinpour, A.E. Abdelnaby Soil Dynamics and Earthquake Engineering 92 (2017) 706725

Table 1
Stations.

Station Latitude Longitude Na

Ashburton District Council (ADCS) 43 54 15S 171 44 51E 2


Arthurs Pass Police Station (APPS) 42 57 02S 171 34 03E 2
Ashley School (ASHS) 43 16 34S 172 35 45E 3
Christchurch Canterbury Aero Club (CACS) 43 29 05S 172 31 48E 9
Christchurch Botanic Gardens (CBGS) 43 31 52S 172 37 11E 10
Christchurch Cathedral College (CCCC) 43 32 23S 172 38 50E 3
Christchurch Hospital (CHHC) 43 32 15S 172 37 38E 12
Christchurch Cashmere High School (CMHS) 43 34 02S 172 37 27E 7
Canterbury Ring Laser (CRLZ) 43 34 35S 172 37 23E 3
Cust School (CSTC) 43 18 50S 172 22 52E 2
Opus Building, Ground Floor (D06C) 43 32 31S 172 36 51E 2
5th Floor Christchurch Police Stn (D08C) 43 32 03S 172 37 57E 2
Basement of Christchurch Police Station (D09C) 43 32 02S 172 37 57E 3
Sign of the Kiwi (D13C) 43 36 30S 172 38 41E 3
Kennedy Bush Reserve (D14C) 43 38 03S 172 37 28E 3
Summit Road (D15C) 43 35 11S 172 43 32E 3
Dareld High School (DFHS) 43 29 29S 172 06 07E 5
Dunsandel School (DSLC) 43 40 09S 172 11 52E 2
Greendale (GDLC) 43 35 16S 172 05 19E 3
Godley Drive (GODS) 43 34 48S 172 46 13E 4
Halswell School (HALS) 43 35 33S 172 34 10E 5
Hillmorton High School (HHSS) 43 33 33S 172 35 34E 2
Hororata School (HORC) 43 32 29S 171 57 35E 2
Hulverstone Drive Pumping Station (HPSC) 43 30 12S 172 42 07E 6
Heathcote Valley Primary School (HVSC) 43 34 52S 172 42 33E 22
Kaiapoi North School (KPOC) 43 22 41S 172 39 49E 8
Lyttleton Port Cashin Quay (LCQC) 43 36 37S 172 43 46E 2
Lincoln Crop and Food Research (LINC) 43 37 29S 172 28 04E 8
Lyttelton Port Company (LPCC) 43 36 34S 172 43 29E 7
Lyttleton Port Oil Wharf (LPOC) 43 36 36S 172 42 53E 5
McQueen's Valley (MQZ) 43 42 28S 172 39 13E 2
New Brighton Library (NBLC) 43 30 31S 172 43 52E 8
Christchurch North New Brighton School 43 29 50S 172 43 04E 3
(NNBS)
Oxford (OXZ) 43 19 39S 172 02 17E 2
Panorama Road (PARS) 43 34 10S 172 45 02E 3
Christchurch Papanui High School (PPHS) 43 29 40S 172 36 24E 5
Pages Road Pumping Station (PRPC) 43 31 39S 172 40 58E 8
Riccarton High School (RHSC) 43 32 16S 172 33 51E 8
Rakaia School (RKAC) 43 45 11S 172 01 23E 2
Rolleston School (ROLC) 43 35 40S 172 22 51E 2
Southbridge School (SBRC) 43 48 37S 172 15 08E 3
Shirley Library (SHLC) 43 30 25S 172 39 48E 5
Selwyn Lake Road (SLRC) 43 40 36S 172 19 03E 3
Styx Mill Transfer Station (SMTC) 43 28 09S 172 36 49E 6
Springeld Fire Station (SPFS) 43 20 23S 171 55 44E 2
Swannanoa School (SWNC) 43 22 16S 172 29 43E 2
Templeton School (TPLC) 43 33 06S 172 28 19E 2
Waikari (WAKC) 42 57 53S 172 42 18E 2

a
Number of records with PGA >=0.05g in both horizontal and vertical directions.

level of increase in ductility demand is strongly aected by the siting researchers employed ber based models [1,8,9,3638]. Fiber based
conguration. models or distributed plasticity models consider the nonlinearity not
Although some studies have been performed so far, there are still only in the connections, but also along the whole length of the
some limitations. Most of the employed models include either system members. These models divide the cross sections into dierent
level based or component level based models. System level based materials and bers. For each material, dierent models considering
models are the models considering the structure as single degree of degrading eects can be employed and then the models are assembled
freedom. Although these models are simple, they have some decits. In to show the behavior of the cross section and member. Using ber
system level based models, the localized behavior, force redistribution based models, the local degradation in concrete and steel reinforce-
and the eect of higher modes are neglected. Component level based ments in a reinforced concrete structure can be easily modeled.
models or lumped plasticity models have also been extensively used to However these models are computationally expensive, especially in
show the behavior of the MDOF structures under seismic sequences. case of 3-D structures, but they are more reliable than lumped
Although these models are not computationally expensive, they do not plasticity models.
account for localized degradation in strength and stiness. In these Most of the present studies have also used repeated or uncorrelated
models, the length of the plastic hinges is considered as zero. Instead of random earthquakes (articial sequences) to investigate the behavior of
beam-column connections, the rest of the structure assumes to behave the structures under multiple earthquakes. Since the characteristics of
elastically. To overcome the limitations of the previous models, some the main shock and aftershock, such as magnitude, intensity, frequency

709
F. Hosseinpour, A.E. Abdelnaby Soil Dynamics and Earthquake Engineering 92 (2017) 706725

Fig. 4. Horizontal and vertical seismic sequences in RHSC and HVSC stations.

content, and duration are dierent and characteristics of the ground buckling, yielding, and Bauschinger eect (reduction in yielding stress
motions are dierent in dierent site conditions as well, using articial due to the changing of the loading direction) in steel reinforcements are
sequences leads to unreliable results. The eects of aftershock polarity, considered. The nonlinear response of the structures under 48 real
earthquake direction, structure irregularity, and the vertical compo- seismic sequences from Christchurch-New Zealand earthquakes 2010
nent of the earthquake on the response of the structures under multiple 2011 are studied and the eect of irregularity, damage from previous
earthquakes are also other limitations that have not been studied events, aftershock polarity, vertical component of the earthquake, and
suciently so far. earthquake direction on the structural demands will be investigated.
This study aims to overcome the aforementioned limitations and
evaluate the behavior of the concrete structures under as-recorded 2. Input ground motions
seismic sequences. For this purpose, two eight-story reinforced con-
crete buildings, including one regular and one irregular (in height) As it was mentioned in previous section, since the characteristics of
buildings, are modeled using ber based modeling in Zeus-NL soft- the main shock and aftershock are dierent, using repeated sequences
ware, developed at university of Illinois at Urbana-Champaign [39]. In as multiple earthquakes may not be reliable. Furthermore, because of
modeling the structures, dierent degrading features like stiness and the dierent site conditions, using random ground motions from
strength degradation and pinching eect (changing in stiness from uncorrelated earthquakes and putting them in sequences to investigate
tension to compression and vice versa) in concrete as well as the local the behavior of the structures cannot be reliable as well. For this

710
F. Hosseinpour, A.E. Abdelnaby Soil Dynamics and Earthquake Engineering 92 (2017) 706725

Fig. 4. (continued)

purpose, this study employs as-recorded sequences from Christchurch- Fig. 4). The response spectra for individual ground motions and
New Zeeland 20102011 earthquakes. sequences were also drawn (see Fig. 5). To check the acceleration
A magnitude 7.1 earthquake struck the Canterbury Region of New consistency of the records with the region of the designed buildings and
Zealand at 4:35 am on 4 September 2010 [40]. The earthquake's the soil type, horizontal response spectra for soil type B as well as
epicenter was located some 25 miles (40 km) west of Christchurch vertical response spectra based on EC8 [45] were obtained (see Fig. 5).
near the town of Dareld, and the focus was located about 6 miles As it can be seen in Fig. 5, the general shapes and trends of the
(10 km) beneath the surface. It was caused by right-lateral movement response spectra are similar for the employed seismic sequences and
along a previously unknown regional strike-slip fault in the western EC8 [45].
section of the Canterbury Plains. The fault appeared about 5056 miles
(8090 km) southeast of the boundary between the Australian and 3. Model development
Pacic tectonic plates [41]. This earthquake caused a high vertical
acceleration in many stations as shown in Fig. 2. Fig. 2 also shows the Two planar concrete structures are employed to represent both
main shock's epicenter and fault surface trace. The earthquake was regular and irregular (in height) RC buildings for this study. These
followed by many large magnitude aftershocks (see Fig. 3). The buildings were designed by Hatzigeorgiou and Liolios [6] and where
recorded PGAs for aftershocks in some stations were much higher they modeled the buildings using the lumped plasticity modeling
than that of the main shock. approach (see Fig. 6). They consist of typical beamcolumn RC frame
Based on the proximity to the fault region (source to site distance) buildings without shear walls, located in a high-seismicity region of
and the peak ground accelerations (more than 0.05g), 48 stations with Europe considering both gravity and seismic loads [6].
dierent number of ground motions in sequences were selected. The The dead loads (excluding self-weight) and live loads are equal to
records were downloaded from GeoNet processed strong-motion data 20 and 10 kN/m, respectively, and are directly applied on the beams.
[44]. Table 1 shows the stations and the number of ground motions in All oors are assumed to be rigid in plan to account for the diaphragm
sequences with PGA more than 0.05g in both horizontal and vertical action of concrete slabs. Material properties are assumed to be 20 MPa
directions. For each station, the records were put in sequences (see for the concrete compressive strength (concrete grade C20) and

711
F. Hosseinpour, A.E. Abdelnaby Soil Dynamics and Earthquake Engineering 92 (2017) 706725

Fig. 5. Horizontal and vertical response spectra in RHSC and HVSC stations.

500 MPa for the yield strength of both longitudinal and transverse scalar model is used to simulate the stiness degradation and recovery.
reinforcements (steel grade S500s) [6]. Stiness recovery scheme is introduced to simulate crack opening and
In order to accurately capture the degrading behavior of the RC closure. In addition, the evolution of yield surface is controlled by the
buildings on the concrete and steel material level, a distributed strength function of eective strength that accounts for strength
plasticity model is developed using ber-based nite element analysis deterioration due to existence of cracks parallel to the loading direction
tool, Zeus-NL [39]. [8]. To simulate the behavior of steel reinforcements, modied
For this purpose the cross section of the elements are divided into Menegotto-Pinto steel model [47,48] is used to consider the
dierent segments based on the material types. Models and relation- Bauschinger eect, inelastic buckling of reinforcements after crushing
ships considering degrading in steel and concrete are employed and of concrete cover, and reinforcement fracture. Fig. 7 shows cyclic
assembled to demonstrate the behavior of the cross section and stress-strain behavior of steel reinforcements and concrete. Table 2
member. also shows natural periods of the rst three modes.
To consider degradation in concrete, the plastic-damage model More details on modeling can be found in Abdelnaby [1] and
proposed by Lee and Fenves [46] is employed. The concept of fracture- Abdelnaby and Elnashai [8,9].
energy-based multiple-hardening is used to represent tensile and For dynamic analyses, Newmark integration algorithm [49] is
compressive damage independently. A thermodynamically consistent employed. Newmark integration is a well-known implicit method that

712
F. Hosseinpour, A.E. Abdelnaby Soil Dynamics and Earthquake Engineering 92 (2017) 706725

Fig. 5. (continued)

nt nr
has been used in many structural dynamic problems and investigations di i
on seismic behavior of the structures so far. In this method, two max [ , ]tolerance convergence
dref ref
parameters, and , are employed to get structural responses in each i =1 i =1 (4)
step based on initial conditions as follows:
where di , i , dref , ref ,nt , and nr are iterative displacement i, iterative
ui +1 = ui + (t)[(1 )ui + (ui +1)] (2) rotation i, reference displacement, reference rotation, number of
translational freedoms, and number of rotational freedoms respec-
1
ui +1 = ui + (t)ui + (t)2 [( )ui +ui +1] tively.
2 (3)
Dierent time-steps from 0.001 to 0.02 s are also chosen and both
where ui , ui+1, ui ,ui+1,ui , and ui+1 are displacements, velocities and Newton-Raphson method (number of initial reformations=number of
accelerations in steps i and i+1 respectively and t shows time-steps. In iterations) and modied Newton-Raphson method (number of initial
this study, optimal values of 0.25 and 0.5 are considered for and reformations=0) are employed for iterative strategy. Further detail on
respectively. the employed algorithm, convergence criteria, time stepping, and
The convergence criterion is based on the maximum iterative iterative strategy can be found in [50].
increment of displacements (see Eq. (4))

713
F. Hosseinpour, A.E. Abdelnaby Soil Dynamics and Earthquake Engineering 92 (2017) 706725

Fig. 6. Concrete buildings designed and modeled by Hatzigeorgiou and Liolios [6].

Fig. 7. (a) Cyclic stressstrain behavior of steel under tension and compression [48] (b) uniaxial cyclic behavior of concrete model [46].

4. Seismic behavior of damaged and undamaged structures sidered and the eect of the damage on the story drifts and number of
plastic hinges for both regular and irregular buildings was evaluated.
In order to examine the eect of the damage from previous events, For this purpose, the damaged and undamaged buildings were
multiple earthquakes with dierent number of sequences were con- subjected to aftershocks. The damaged buildings had already experi-

714
F. Hosseinpour, A.E. Abdelnaby Soil Dynamics and Earthquake Engineering 92 (2017) 706725

Table 2 Fig. 10 indicates that the total drifts for both damaged and undamaged
Natural periods of regular and irregular buildings. regular building are almost the same and for irregular building there is
a slight dierence between the demands. As the PGA of the main shock
Models T1 (s) T2 (s) T3 (s)
increased, the dierence between the responses also increased (see
Regular Irregular Regular Irregular Regular Irregular Fig. 11). To see the eect of the vertical irregularity on the structural
responses, the average increases of the total residual drifts in both
Hatzigeorgiou 1.233 0.9673 0.4718 0.4469 0.2651 0.2746
regular and irregular structures were calculated for similar stations as
and liolios[6]
Present study 1.2282 0.9754 0.4593 0.4417 0.2604 0.2747 follows:

Fig. 8. Total residual drifts in damaged and undamaged buildings (a) regular building (b) irregular building.

enced a level of degradation from the main shock while the undamaged
Averageincreasesoftotalresidualdrifts
buildings had not experienced any previous events. The ndings
generally showed an increase in structural demands (see Figs. 8 and Residual drift (damaged )-Residual drift (undamaged )
=
9). However in some cases that the main shock PGA was not high Residual drift (undamaged ) (5)
(almost less than 0.15 g), there was not a signicant dierence between
the demands of the damaged and undamaged structures (see Fig. 10). Considering damage from the previous event, the average increase

715
F. Hosseinpour, A.E. Abdelnaby Soil Dynamics and Earthquake Engineering 92 (2017) 706725

Fig. 9. Number of plastic hinges in damaged and undamaged buildings (a) regular building (b) irregular building.

of total residual drifts under subsequent events for the irregular excitations.
structure was almost twice the regular structure (78.3 and 40.1 for The moment-curvature diagram for some plastic hinges in beams
the irregular and regular structures respectively). Fig. 12 indicates the and columns of both regular and irregular buildings are also provided
plastic hinge locations in the damaged and undamaged regular and in Fig. 13.
irregular buildings. As it can be seen, considering the damage from the The higher demands in the irregular structure indicate that this
main shock causes an increase in number of plastic hinges and this structure is more sensitive to the damage from the previous event. This
increase is more for the irregular building. Fig. 12 shows the location of can be as a result of force distribution under seismic sequences which is
plastic hinges developed in the damaged and undamaged regular and more symmetric for the regular building.
irregular buildings due to main shock and aftershocks. As it can be
seen, most of the plastic hinges formed in the irregular building are in 5. Eect of earthquake direction
columns while for the regular building, they are mostly formed in the
beams. This indicates a better performance for the regular building. In To see the performance of the irregular structure in dierent
addition, a soft-story behavior in the irregular building occurred for the directions, dierent sequences were employed and applied in both
undamaged case causing the development of plastic hinges at the rst- east-west and west-east directions. Because of the symmetry, the
story columns only. In contrary, the regular building developed multi- regular structure was not evaluated for the eect of earthquake
ple plastic hinges along the building stories and hence providing more direction. Results indicated that in most cases changing the earthquake
energy dissipation and force redistribution in response to aftershock direction signicantly aects the residual drifts. In some stations the

716
F. Hosseinpour, A.E. Abdelnaby Soil Dynamics and Earthquake Engineering 92 (2017) 706725

Fig. 10. Total drift demands PGA: main shock (MSH)=0.1 g, aftershock 1 (ASH1)=0.1 g, aftershock 2 (ASH2)=0.07 g (a) regular building (b) irregular building.

dierences between the total residual drifts are more than 10 cm for also aected by earthquake direction in most cases. Therefore it is
two directions (see Fig. 14). Results of this study also indicated that necessary to include the eect of earthquake direction when consider-
considering just a main shock to evaluate the structural response in ing the nonlinear behavior of irregular buildings under seismic
dierent directions may cause an underestimation in the dierence sequences.
between structural demands in dierent directions. As it can be seen in
Fig. 15, for the rst 160 s (main shock), the total drift demands in two 6. Eect of aftershock polarity
directions are not signicantly dierent but the dierence increases
under subsequent events. Changing in total drift demands with earth- The other important part of this study includes investigating the
quake direction shows dierent performances in dierent directions. structural behavior based on the aftershock direction (aftershock
This is due to the distinctive variations and degradations of the polarity). Efraimiadou et al. [14] investigated the eect of aftershock
structural properties such as stiness, strength, frequency and damp- polarity analytically and studied structural pounding between adjacent
ing in each direction. Fig. 16 shows that the number of plastic hinges is buildings under strong earthquakes. They proposed that the inelastic

717
F. Hosseinpour, A.E. Abdelnaby Soil Dynamics and Earthquake Engineering 92 (2017) 706725

Fig. 11. Total drift demands (a) regular building- PGA: MSH=0.47 g, ASH=0.16 g (b) irregular building PGA: MSH=0.4 g, ASH=0.07 g.

seismic behavior of the structures is strongly aected by aftershock drift demands with changing the aftershock direction. The results were
polarity and the sign of the ground motions should be considered in compared for the cases when the main shock and aftershocks were in
generating articial sequences. Raghunandan et al. [10] also men- the same direction and when they were in dierent directions. Findings
tioned that when the residual drift resulting from a main shock is high, indicated that aftershock direction has a slight eect on the response of
the direction of the aftershock with respect to the main shock is the regular building and drift demands are not signicantly aected by
signicant and used randomized main shock polarity to study after- aftershock polarity (see Figs. 17(a) and 18(a)). Unlike the regular
shock collapse vulnerability of RC frame structures. Ruiz-Garca and building, total residual drift demands in the irregular building were
Aguilar [15] employed lumped plasticity modeling approach and highly aected by aftershock direction for many sequences and
investigated the eect of aftershock polarity on the structural responses increased when the main shock and aftershocks were in the same
of a four-story steel frame. They concluded that the aftershock collapse direction (see Fig. 17(b)). The dierence between the responses was
capacity and aftershock capacity against demolition are signicantly more for the sequences including aftershocks with high PGAs. As it can
aected by aftershock polarity. be seen in Fig. 18(b) for the rst 500 s, there is no signicant dierence
To investigate the eect of aftershock polarity in this study, between the total drift demands but after that when an aftershock with
dierent sequences were employed to see the variations of the total a high PGA (0.52 g) is applied to the structure, the responses

718
F. Hosseinpour, A.E. Abdelnaby Soil Dynamics and Earthquake Engineering 92 (2017) 706725

Fig. 12. Plastic hinge locations PGA: MSH=0.429 g, ASH1=0.094 g, ASH2=0.157 g, ASH3=0.112 g, ASH4=0.125 g, ASH5=0.059 g, ASH6=0.122 g, ASH7=0.067 g (a) regular
building (b) irregular building.

Fig. 13. Momentcurvature behavior of plastic hinges in one end of the beams and columns.

719
F. Hosseinpour, A.E. Abdelnaby Soil Dynamics and Earthquake Engineering 92 (2017) 706725

Fig. 14. Dierence between total residual drifts in two direction.

Fig. 15. Total drift demands for dierent directions.

signicantly change. The higher variations of total drifts and total drifts for the irregular structure was almost twice the regular structure.
residual drifts in the irregular building can be due to the high In other words main shocks had lower impacts on the regular building
sensitivity to the earthquake direction which was shown in the previous and because of the symmetry, changing the direction of subsequent
section. Besides that, it was shown in Section 4 that considering the events causes smaller eects on the structural responses in regular
damage from the previous event, the average increase of total residual structure compared to the irregular one.

720
F. Hosseinpour, A.E. Abdelnaby Soil Dynamics and Earthquake Engineering 92 (2017) 706725

Fig. 16. Dierence between number of plastic hinges in two directions.

Fig. 17. Total residual drifts based on the aftershocks direction with respect to the main shock direction (a) regular building (b) irregular building.

721
F. Hosseinpour, A.E. Abdelnaby Soil Dynamics and Earthquake Engineering 92 (2017) 706725

Fig. 18. Total drift demandsPGA: MSH=0.29 g, ASH1=0.05 g, ASH2=0.08 g, ASH3=0.52 g (a) regular building (b) irregular building.

7. Eect of vertical component of the earthquake in Fig. 19, considering both horizontal and vertical earthquake
components causes a very slight change (less 5 mm in most stations)
Most of the previous studies in the eld of earthquake engineering in total residual drifts.
have neglected the eects of vertical ground motion and are usually Since the vertical component causes an increase in axial force in
guided by horizontal motion. The main reason for this practice is found columns, the number of plastic hinges in columns was also calculated.
in circumstances that engineering structures are intended primarily for As it can be seen in Fig. 20, considering vertical component of the
vertical load transfer and thereby it is implied that they have sucient earthquake causes an increase in the number of the plastic hinges in
resistance to dynamic forces caused by vertical motion [51]. columns. Therefore, it can be concluded that the eect of vertical
To see the eect of the vertical earthquake component in this study, component of the earthquake on the columns responses shouldn't be
multiple earthquakes in both vertical and horizontal directions were neglected in the design of the reinforced concrete structures under
employed and applied to the regular and irregular buildings. The multiple earthquakes.
results showed that considering the vertical component of the earth- It should be noted that because of facing a very high level of
quake does not change the total residual drifts signicantly. Fig. 19 inelasticity and thus structural collapse under some sequences, those
shows the dierence between total residual drifts when considering sequences were either not considered or some of their aftershocks were
both horizontal and vertical earthquake components. As it can be seen removed.

722
F. Hosseinpour, A.E. Abdelnaby Soil Dynamics and Earthquake Engineering 92 (2017) 706725

Fig. 19. Dierence between total residual drifts when considering earthquake vertical direction (a) regular building (b) irregular building.

8. Conclusion and number of plastic hinges and may cause high total residual
drifts in the irregular structure.
The present study investigated the nonlinear behavior of concrete Considering just a main shock to investigate the structural demands
structures under real multiple earthquakes. For this purpose two in dierent directions may cause unreliable results.
regular and irregular (in height) RC buildings were modeled (consider- Aftershock polarity can signicantly change the total drift demands in
ing degrading features of both concrete and steel reinforcements) using the irregular structure especially when sequences contain aftershocks
Zeus-NL software. In this study the eect of irregularity, damage from with high PGAs, but has a slight eect in the regular structure.
previous events, earthquake direction, aftershock polarity and vertical The vertical component of the earthquake may have a very slight
component of the earthquake were considered. Based on the results, eect on total drift demands. However it increases the number of
the following conclusions may be drawn: plastic hinges in columns and it should be considered in design of
the structures under multiple earthquakes.
The eect of the damage from the previous event is more important
in the irregular structure and the average increase of total residual Because the results of this study are based on certain types of
drifts under subsequent ground motions is almost twice the regular structures, and since the frequency content and duration of a ground
structure. motion are also other important parameters as well as the PGA, further
Changing the earthquake direction aects the total drift demands research is required to have more reliable ndings.

723
F. Hosseinpour, A.E. Abdelnaby Soil Dynamics and Earthquake Engineering 92 (2017) 706725

Fig. 20. Number of plastic hinges in columns (a) regular building (b) irregular building.

References [6] Hatzigeorgiou GD, Liolios AA. Nonlinear behaviour of RC frames under repeated
strong ground motions. Soil Dyn Earthq Eng 2010;30:101025.
[7] Ruiz-Garca J, Negrete-Manriquez JC. Evaluation of drift demands in existing steel
[1] Abdelnaby AE. Multiple earthquake eects on degrading reinforced concrete frames under as-recorded far-eld and near-fault mainshockaftershock seismic
structures [Doctoral dissertation]. Urbana-Champaign, IL: Univ. of Illinois; 2012. sequences. Eng Struct 2011;33(2):62134.
[2] IRIS, IEB export: 56 earthquakes as a sortable table, http://ds.iris.edu/ieb/ [8] Abdelnaby A, Elnashai A. Performance of degrading reinforced concrete frame
evtable.phtml?caller=IEB & st=2016-04-14 & et=2016-04-18 & nma=4xma=10 & systems under Tohoku and Christchurch earthquake sequences. J Earthq Eng
ob=time-desc & li=1000 & xla=56.000 & nla=22.000 & xlo=159.000 & nlo=127.000 2014;18(7):100936.
& sbl=1 & pbl=1 & caller=self & name=Japan%20Region & zm=4 & mt=ter & [9] Abdelnaby A, Elnashai A. Numerical modeling and analysis of RC frames subjected
rgn=Japan%20Region & title=IEB%20export%3A%2056%20earthquakes%20as to multiple earthquakes. Earthq Struct 2015;9(5):95781.
%20a%20sortable%20table. & stitle=from%202016-04-14%20to%202016-04- [10] Raghunandan M, Liel AB, Luco N. Aftershock collapse vulnerability assessment of
18%2C%20with%20magnitudes%20from%204%20to%2010%2C%20all%20depths reinforced concrete frame structures. Earthq Eng Struct Dyn 2015;44(3):41939.
%2C%20with%20priority%20for%20most%20recent%2C%20and%20limited%20to [11] Faisal A, Majid TA, Hatzigeorgiou GD. Investigation of story ductility demands of
%201000; 2016 [Last access: 08.22.16]. inelastic concrete frames subjected to repeated earthquakes. Soil Dyn Earthq Eng
[3] Scordilis EM. Empirical global relations converting M S and m b to moment 2013;44:4253.
magnitude. J Seismol 2006;10(2):22536. [12] Zhang S, Wang G, Sa W. Damage evaluation of concrete gravity dams under
[4] Aschheim M, Black E. Eects of prior earthquake damage on response of simple mainshockaftershock seismic sequences. Soil Dyn Earthq Eng 2013;50:1627.
stiness-degrading structures. Earthq Spectra 1999;15(1):124. [13] Hatzivassiliou M, Hatzigeorgiou GD. Seismic sequence eects on three-dimensional
[5] Amadio C, Fragiacomo M, Rajgelj S. The eects of repeated earthquake ground reinforced concrete buildings. Soil Dyn Earthq Eng 2015;72:7788.
motions on the non-linear response of SDOF systems. Earthq Eng Struct Dyn [14] Efraimiadou S, Hatzigeorgiou GD, Beskos DE. Structural pounding between
2003;32:291308. adjacent buildings subjected to strong ground motions. Part I: the eect of dierent

724
F. Hosseinpour, A.E. Abdelnaby Soil Dynamics and Earthquake Engineering 92 (2017) 706725

structures arrangement. Earthq Eng Struct Dyn 2013;42(10):150928. mainshockaftershock sequence-type ground motions. Soil Dynam Earthq Eng
[15] Ruiz-Garca J, Aguilar JD. Aftershock seismic assessment taking into account 2013;45:112.
postmainshock residual drifts. Earthq Eng Struct Dyn 2015;44(9):1391407. [35] Zhai CH, Wen WP, Li S, Chen Z, Xie LL. The damage investigation of inelastic
[16] Di Sarno L. Eects of multiple earthquakes on inelastic structural response. Eng SDOF structure under the mainshockaftershock sequence-type ground motions.
Struct 2013;56(11):67381. Soil Dyn Earthq Eng 2014;59:3041.
[17] Dong Y, Frangopol DM. Risk and resilience assessment of bridges under mainshock [36] Vasilopoulos AA, Beskos DE. Seismic design of space steel frames using advanced
and aftershocks incorporating uncertainties. Eng Struct 2015;83:198208. methods of analysis. Soil Dyn Earthq Eng 2009;29(1):194218.
[18] Fakharifar M, Chen G, Sneed L, Dalvand A. Seismic performance of postmainshock [37] Sextos AG, Katsanos EI, Manolis GD. EC8-based earthquake record selection
FRP/steel repaired RC bridge columns subjected to aftershocks. Compos Part B procedure evaluation: validation study based on observed damage of an irregular
2015;72:18398. R/C building. Soil Dyn Earthq Eng 2011;31(4):58397.
[19] Fragiacomo M, Amadio C, Macorini L. Seismic response of steel frames under [38] Li Z, Hatzigeorgiou GD. Seismic damage analysis of RC structures using ber beam-
repeated earthquake ground motions. Eng Struct 2004;26(13):202135. column elements. Soil Dyn Earthq Eng 2012;32(1):10310.
[20] Goda K. Nonlinear response potential of mainshockaftershock sequences from [39] Elnashai AS, Papanikolaou V, Lee D. Zeus NL a system for inelastic analysis of
Japanese earthquakes. Bull Seismol Soc Am 2012;102(5):213956. structures. Mid-America Earthquake Center, University of Illinois at Urbana-
[21] Han R, Li Y, van de Lindt J. Seismic risk of base isolated non-ductile reinforced Champaign; 2002, [Program Release September 2002].
concrete buildings considering uncertainties and mainshockaftershock sequences. [40] Potter SH, Becker JS, Johnston DM, Rossiter KP. An overview of the impacts of the
Struct Saf 2014;50:3956. 20102011 Canterbury earthquakes. Intl J Disaster Risk Reduct 2015;14:614.
[22] Han R, Li Y, Lindt J. Impact of aftershocks and uncertainties on the seismic [41] Christchurch earthquakes of 201011. Encyclopdia Britannica, http://www.
evaluation of non-ductile reinforced concrete frame buildings. Eng Struct britannica.com/event/Christchurch-earthquakes-of-2010-2011; 2016 [Last
2015;100:14963. access: 04.28.16].
[23] Hatzigeorgiou GD, Beskos DE. Inelastic displacement ratios for SDOF structures [42] Gledhill K, Ristau J, Reyners M, Fry B, Holden C. The Dareld (Canterbury, New
subjected to repeated earthquakes. Eng Struct 2009;31:274455. Zealand) Mw 7.1 earthquake of September 2010: a preliminary seismological
[24] Li Q, Ellingwood BR. Performance evaluation and damage assessment of steel report. Seismol Res Lett 2011;82(3):37886.
frame buildings under main shockaftershock earthquake sequences. Earthq Eng [43] GNS Science, http://www.gns.cri.nz/Home/Our-Science/Natural-Hazards/
Struct Dyn 2007;36(3):40527. Recent-Events/Canterbury-quake/Recent-aftershock-map; 2014 [Last access: 04.
[25] Li Y, Song R, van de Lindt JW. Collapse fragility of steel structures subjected to 28.16].
earthquake mainshockaftershock sequences. J Struct Eng 2014. [44] GeoNet processed strong-motion data. ftp.geonet.org.nz/strong/processed/
[26] Loulelis D, Hatzigeorgiou GD, Beskos DE. Moment resisting steel frames under Proceedings. [Last access: 4.28.16].
repeated earthquakes. Earthq Struct 2012;3(34):23148. [45] EN 1998-1 Eurocode 8: design of structures for earthquake resistance; Part 1:
[27] Mahin SA. Eects of duration and aftershocks on inelastic design earthquakes. In: general rules, seismic actions and rules for buildings 2005; European Committee
Proceedings of the seventh world conference on earthquake engineering. vol. 5; for Standardization, Brussels.
1980. p. 67780. [46] Lee J, Fenves G. Plastic-damage model for cyclic loading of concrete structures. J
[28] Moustafa A, Takewaki I. Response of nonlinear single-degree-of-freedom struc- Eng Mech 1998;124(8):892900.
tures to random acceleration sequences. Eng Struct 2011;33(4):12518. [47] Menegotto M, Pinto PE. Method of analysis for cyclically loaded reinforced concrete
[29] Riddell R, Garcia JE, Garces E. Inelastic deformation response of SDOF systems plane frames including changes in geometry and nonelastic behavior of elements
subjected to earthquakes. Earthq Eng Struct Dynam 2002;31:51538. under combined normal force and bending. In: Proceedings of the IABSE
[30] Ruiz-Garca J. Mainshockaftershock ground motion features and their inuence symposium on resistance and ultimate deformability of structures acted on by
in buildings seismic response. J Earthquake Eng 2012;16(5):71937. welldened repeated loads, Final report. Lisbon; 1973.
[31] Song R, Li Y, van de Lindt JW. Impact of earthquake ground motion characteristics [48] Gomes A, Appleton J. Nonlinear cyclic stressstrain relationship of reinforcing bars
on collapse risk of post-mainshock buildings considering aftershocks. Eng Struct including buckling. Eng Struct 1997;19:8226.
2014;81:34961. [49] Newmark NM. A method of computation for structural dynamics. J Eng Mech
[32] Yaghmaei-Sabegh S, Ruiz-Garca J. Nonlinear response analysis of SDOF systems ASCE 1959;85:6794.
subjected to doublet earthquake ground motions: a case study on 2012 Varzaghan [50] Elnashai AS, Papanikolaou V, Lee DH. ZEUS-NL user manual. University of Illinois
Ahar events. Eng Struct 2016;110:28192. at Urbana-Champaign, Mid-America Earthquake Center; 2002.
[33] Zafar A, Andrawes B. Seismic behavior of SMAFRP reinforced concrete frames [51] Varevac D, Dragani H, Gazi G. Inuence of the vertical component of earthquake
under sequential seismic hazard. Eng Struct 2015;98:16373. on large span RC beams. Teh Vjesn 2010;17(3):35766.
[34] Zhai CH, Wena W-P, Chen ZH, Lia SH, Xie L-L. Damage spectra for the

725