Anda di halaman 1dari 20

Engineering Structures 81 (2014) 412–431

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Effects of joint flexibility on lateral response of reinforced concrete


frames
Jalil Shafaei a,b, Mohammad Sajjad Zareian c,⇑, Abdollah Hosseini a, Mohammd Sadegh Marefat a
a
School of Civil Engineering, College of Engineering, University of Tehran, P.O. Box 11365-4563, Tehran, Iran
b
Department of Civil Engineering, Shahrood University of Technology, P.O. Box 36199-95161, Shahrood, Iran
c
Department of Civil Engineering, Ferdowsi University of Mashhad, P.O. Box 91775-1111, Mashhad, Iran

a r t i c l e i n f o a b s t r a c t

Article history: It is known that shear deformation of beam-column joints have a significant contribution to lateral
Received 13 September 2013 response of reinforced concrete structures. Most of the available analysis programs assume rigid joints
Revised 28 September 2014 regardless of reinforcement details in the joint region. Slip of the beam longitudinal reinforcement in
Accepted 30 September 2014
the joint and shear deformation of joint panel significantly decrease rigidity of the joint, especially in
Available online 25 October 2014
non-seismically detailed beam-column joints. In the current study, these effects were modelled in the
OpenSees software framework and a modified joint element for analysis of multi-storey frames was used.
Keywords:
The model was verified by experimental results and the verified model was used to analyze five and ten
Reinforced concrete frames
Beam-column joint
storey frames with various joint details using nonlinear static and incremental dynamic analyses. The
Joint flexibility results of frames with flexible joints were compared with those of frames with rigid joints assumption.
Seismic detailing The analytical results confirmed that the modified joint element successfully predicted experimental cyc-
Nonlinear static analysis lic behaviour of beam-column joint specimens and also was found that existing RC structures with defi-
IDA analysis cient beam-column joints are very vulnerable when subjected to severe earthquakes and the rigid joint
assumption was not appropriate when assessing the behaviour of existing non-seismically detailed struc-
tures. The results also showed that the effect of joint deformation on lateral response of RC frames
depends on the peak acceleration values of the ground excitation.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction analysis of RC structures. Thus, it is necessary that inelastic behav-


iour of the joint panel is explicitly simulated in predicting the
Post-earthquake inspections of damaged reinforced concrete response of RC structures when subjected to earthquake loading
(RC) Buildings [1–5] as well as laboratory tests of non-seismic RC [7].
beam-column joints [6–8] were demonstrated that non-seismi- It has been common in seismic analysis of RC moment-resisting
cally detailed beam-column joints suffered serious damage and structures that beam-column joints are modelled with rigid joint
significantly affect the global structural performance and cause assumption [7,23] as shown in Fig. 1a. This assumption gives rise
to structural collapse. In RC moment-resisting frames, the overall to an overestimation of the structure stiffness and underestimation
earthquake resistance is significantly affected by the behaviour of of the storey drifts and hence to an improper assessment of the
beam-column joints. According to previous experimental [8–12] period and seismic performance of the existing structures.
and analytical [13–22] research studies, the nonlinear behaviour However, it has been demonstrated that, in ductile RC moment-
of RC beam-column joints is significantly affected by bond-slip of resisting structures designed based on current seismic codes, the
the beam longitudinal reinforcements and diagonal shear cracking joint regions are actually not rigid when subjected to significant
in the joint region that cause to degradation of strength, unloading/ seismic loading, resulting in diagonal cracking and significant
reloading stiffness, and energy dissipation of beam-column joint shear deformation in the joint region. As a consequence of this
sub-assemblages, which should be addressed in the nonlinear and possible bond slip of the beam longitudinal reinforcement in
the joint, total joint deformations can be considerable. Paulay
and Priestley [7] stated that ‘‘typically, 20% of the inter-storey
⇑ Corresponding author.
deflection due to earthquake forces may originate from joint
E-mail addresses: jshafaie@ut.ac.ir, js.shafaei@gmail.com (J. Shafaei), zareian@
stu.um.ac.ir, sajjad.zareian@gmail.com (M.S. Zareian), hosseiniaby@ut.ac.ir (A. Hosseini),
deformations. For this reason it is strongly recommended that no
mmarefat@ut.ac.ir (M.S. Marefat). allowance for rigid end regions be made in the lateral force

http://dx.doi.org/10.1016/j.engstruct.2014.09.046
0141-0296/Ó 2014 Elsevier Ltd. All rights reserved.
J. Shafaei et al. / Engineering Structures 81 (2014) 412–431 413

Nomenclature

0
fc standard cylinder compressive strength (MPa) dMx maximum displacement attained during seismic
fct concrete tensile strength (MPa) loading
fy yield strength of steel reinforcement (MPa) b model parameter
Ec concrete modulus of elasticity (MPa) Qy calculated yield strength
D1 and D2 changes in length of the diagonal gauges in the joint dE incremental dissipated hysteretic energy
region, P⁄ = P/(bhfcm) axial load index
D undistorted length of the diagonal gauge b, h width and height of the cross-section respectively
/ inclination angle of the diagonal gauge to the hor- x, x0 ratio of tension and compression reinforcement
izontal line respectively
h and b vertical and horizontal distances between the end L⁄ shear span index
points of the diagonal gauge qs = (b0h0/bs) transverse reinforcement ratio
Ts tension force b0, h0 width and height of the confined concrete core to
Vcol column shear force the centreline of stirrups
Vjh horizontal joint shear force s spacing of stirrups
vjh horizontal joint shear stress fyt yield strength of transverse reinforcement
Aj effective joint area qd diagonal reinforcement ratio
DI Park and Ang’s damage index a confinement effectiveness factor

analysis of ductile frames’’. Therefore, the rigid joint assumption due to beam flexure and joint deformation (see Fig. 2a). The non-
may result in misinterpretation of the global performance charac- linear behaviour of the rotational spring was determined assuming
teristics of structures and could consequently lead to miscalcula- the point of contra-flexure at the centre of the member. Otani [28]
tion of strength and ductility demands on constituent structural proposed a two components beam model. Which is composed of
members and this situation is very critical in structures with two parallel deformable elements, one linearly elastic and the
non-seismic reinforcement detailing [24]. other nonlinear, two nonlinear springs at the ends of these ele-
ASCE/SEI 41 [25] and FEMA-356 [26] suggest to model the ments, and two rigid links outside the springs (see Fig. 2b). The
beam-column joint as a rigid panel. In this model the joint panel two rigid links represent the beam-column joint region, while
is treated as a completely rigid element having dimensions equal the two nonlinear springs model the fixed-end rotations due to
to those of the joint (see Fig. 1a) although ASCE/SEI 41 [25] and slippage of the reinforcement inside the joint region. Bond stress
FEMA-356 [26] both ignore joint shear deformations, but do was assumed to be constant along the development length and
account for the effects of reinforcement slippage in the joint region the compressive reinforcement was assumed not to slip. The rota-
to determine a reduced effective flexural stiffness. ASCE41-06 and tion due to slip was evaluated as the elongation of the tensile rein-
FEMA-356 potentially overestimate the stiffness of RC moment forcement along the development length divided by the distance in
frames by recommending that beam-column joints ‘‘shall be repre- between the tensile and compressive reinforcement bars.
sented as a stiff or rigid zone’’. ASCE/SEI 41 [25] and FEMA-356 [26] El-Metwally and Chen [29] and Alath and Kunnath [30] decou-
suggests a backbone curve as shown in Fig. 1b, for modelling the pled the nonlinear behaviour of the beam, column, and joint panel,
joint shear deformation in nonlinear analysis of RC structures. by introducing zero-length rotational spring elements between the
Several analytical models have been proposed for modelling the joint panel and adjacent members. El-Metwally and Chen [29]
seismic response of RC beam-column joints by considering the developed a model in which a zero length rotational spring was
nonlinear behaviour in the joint region in order to address the placed between beams and columns to characterize the inelastic
effects of bond-slip and shear deformation in structural analysis behaviour within the joint (see Fig. 2c). A macro element to
as shown in Fig. 1c. Giberson [27] proposed a beam element with account for shear deformation in the joint regions of RC structures
two nonlinear rotational springs attached to two ends of a per- was proposed by Alath and Kunnath [30], where the relative rota-
fectly elastic element in order to account for nonlinear action tional deformation between the beam and the column is modelled
Column

Center line
Q
Qy
b
a

1.0 C
B
Beam
E D c
Rigid joint panel zone A
or

(a) Rigid beam-column joint (b) Recommended joint strength-shear deformation


envelope curve for joint modelling
Fig. 1. Beam-column joint modelling based on the recommendations of ASCE41/FEMA-356.
414 J. Shafaei et al. / Engineering Structures 81 (2014) 412–431

Inelastic rotational spring Rigid line


Inelastic rotational
element
spring

Elastic line element


Elastic beam Inelastic line element
(a) Giberson [27] (b) Otani [28]

Zero length Inelastic Inelastic rotational


rotational spring spring

Beam column
line element
Rigid element

(c) El-Metwally and Chen [29] (d) Alath and Kunnath [30]

2
For joint shear
5 1

3 1

For bond-slip 4 1

(e) Biddah and Ghobarah [31] (f) Youssef and Ghobarah [33]

Bar-slip
spring
Beam end

Shear panel Interface shear Shear panel Column end


zone spring

(g) Lowes and Altoontash [34] (h) Altoontash [24]


Fig. 2. Existing models for considering nonlinear behaviour in the joint region, beam and column.

by a zero-length rotational spring element. This formulation were demonstrated through comparisons against experimental
assumes that the columns and beams are connected by rigid links results of twelve RC exterior beam-column joints.
which are capable of independent rotations (see Fig. 2d). Biddah Youssef and Ghobarah [33] and Lowes and Altoontash [34] used
and Ghobarah [31] proposed an element which accounts for shear continuum-type panel elements combined with spring elements to
deformation and bond-slip of reinforcing bars in the joint region. represent nonlinear behaviour of the joint region. Youssef and
This element was the modified version of the model was proposed Ghobarah [33] proposed a model in which the joint was repre-
by Alath and Kunnath [30] and the joint region is idealized by two sented by four rigid members that enclose the joint, with pin con-
springs in series, one representing shear deformation and the other nection between the between these rigid elements and with shear
representing bond-slip to include the effect of the bond slip on the springs connecting the diagonal. The impact of bar-slip within the
seismic behaviour of the RC beam-column joints (see Fig. 2e). Favv- joint and concrete crushing at the joint perimeter was represented
ata et al. [32] proposed a behavioural model that can be success- using three concrete and three steel springs at each face of the con-
fully implemented in a well-established nonlinear dynamic nection region between the beams and columns and the joint panel
analysis program. The validity and accuracy of the proposed model (see Fig. 2f). Lowes and Altoontash [34] proposed an element for
J. Shafaei et al. / Engineering Structures 81 (2014) 412–431 415

Fig. 3. Dimensions and reinforcement details of specimens J1, J2 and J3 (dimensions in millimetres).

modelling the inelastic response of interior RC beam-column joints inspected by Marefat et al. [36]. In addition to flexural deforma-
in two-dimensional structural analyses that was composed of one tions, slip of longitudinal reinforcements and shear deformations
shear panel, four interface-shear springs, and eight bar-slip springs was considered in the analytical model. These effects were mod-
(see Fig. 2g). Various inelastic behaviour mechanisms such as the elled in the OpenSees software framework [37] and verified by
shear failure of the joint panel, the loss of shear load transfer due experimental result of a test program was conducted at the struc-
to cracking at the beam-column interfaces, and failure of bond tural laboratory of the University of Tehran. The verified beam-col-
on the longitudinal reinforcement were also taken into account. umn joint model was used to analyze two multi-storey buildings
The material model of the shear stress–strain behaviour of the joint including five and ten storey frames, and then the response of
panel was defined based on the modified compression field theory frames with seismic and non-seismic detailing joint elements were
[35] (MCFT). The cyclic response parameters were calibrated based compared with the response of frames with rigid joint assumption.
on experimental data. Altoontash [24] simplified the model
proposed by Lowes and Altoontash [34] by introducing a model 2. Description of experimental program
consisting of four zero-length rotational springs located at beam-
and column-joint interfaces, which simulate the member-end rota- For verification of selected beam-column joint element, the
tions due to bond-slip behaviour, while the panel zone component experimental program was conducted at the structural laboratory
with a rotational spring remains to simulate the shear deformation of the University of Tehran. Three half scale external RC beam-col-
of the joint (see Fig. 2h). umn joint with seismic and non-seismic reinforcement detail and
From the review of literature it was found that modelling of identical dimensions were tested by applying lateral cyclic loading
beam-column joints has recently received considerable attention of increasing amplitudes. Tested specimens were comprised of one
and different modelling methods were proposed by several unit having seismic reinforcement detailing and two units having
researchers. The existing models include all the components con- non-seismic reinforcement detailing which are the representatives
tributing to the joint deformations, such as bond-slip of beam rein- of the reinforced concrete construction practice in Iran before the
forcements and shear deformation of the joint region. Thus, in 1970s. Specimen J1 represented a seismic beam-column joint
order to accurately describe the behaviour of beam-column joints, designed to satisfy ACI Committee 318M-11 [38] requirements,
the properties and load–deformation response of the components whereas specimens J2 represented a non-seismic beam column
and parameters should be experimentally verified. joint detailing with a shear strength deficiency in the term of
In this paper the beam-column joint model was proposed by absence of transverse steel hoops in the joint panel zone and J3
Lowes and Altoontash [34], was used for investigating influence is similar to J2 with insufficient beam bottom bare development
of nonlinear behaviour of the joint region on lateral response of length in the joint panel zone.
typical RC moment resisting buildings that were designed and con-
structed in Iran in the late 1970s. These building generally have 2.1. Dimensions and reinforcement details of specimens
non-seismic reinforcement detailing in the joint region and are
vulnerable when subjected to earthquake loading. Typical dimen- External beam-column joint sub-assemblages were considered
sions and reinforcement details of the typical buildings were to be isolated from an existing five-storey residential RC
selected by collecting structural data and reinforcing details of building and having an inter-storey height of 3.5 m and a beam
more than 70 existing RC buildings in Tehran, Iran that were effective span of 5.0 m. The design of the beam-column joint
416 J. Shafaei et al. / Engineering Structures 81 (2014) 412–431

sub-assemblages was based on the typical reinforcement details


and average dimensions of existing buildings that were con-
structed in Iran before the 1970s. The points of contra-flexure were
assumed at the mid-height of columns and the mid-span of beams.
Dimensions and reinforcement details of all specimens were iden-
tical except for the reinforcement details in the joint region as
shown in Fig. 3. The longitudinal and transverse reinforcement of
all specimens for the beam and column satisfied ACI Committee
318M-11 [38] seismic requirements (see Fig. 3). Specimen J1 had
sufficient shear reinforcement in the joint, while specimen J2
was designed as a shear-deficient specimen provided with no
transverse shear reinforcement in the joint region and specimen
J3 was designed as a shear-deficiency and anchorage critical spec-
imen with the bottom longitudinal reinforcement in the beam
anchored 75 mm from the column face and with no transverse
reinforcement installed in the joint region. The dimensions and
reinforcement details of all specimens are shown in Fig. 3.

Fig. 4. Overview of test setup and instrumentation (photograph of test specimen J2,
2.2. Material properties
image by Jalil Shafaei).

All test specimens were constructed using normal weight and


ready mixed concrete with targeted 28-day concrete cylinder com- ity and a ±200 mm stroke in a displacement controlled mode to
pressive strength of 23 MPa and concrete specified slump of identify the cyclic behaviour of the beam-column joint. For each
80 mm, with maximum aggregate sizes of 19 mm. Each entire specimen a total of 10 electrical resistance strain gauges were
specimen was cast as a single element lying horizontally on the attached to the longitudinal and transverse reinforcement at criti-
laboratory floor and was wet-cured for 7 days. Mechanical proper- cal locations to record the magnitude of reinforcement strains that
ties of concrete on the day of testing and the steel reinforcement developed at different loading stages as shown in Fig. 3, and also 13
for 8 mm and 14 mm diameters are shown in Tables 1 and 2, linear variable displacement transducers (LVDTs) were used to
respectively. record the beam rotation, column rotation and joint distortion. A
schematic drawing of the test setup is shown in Fig. 4.
2.3. Test setup, instrumentation and load history The loading procedure for all specimens was based on accep-
tance criteria specified in ACI Committee 374.1-05 [39] utilizing
The test setup with the specimen supports and other key com- displacement (drift) control to preliminary target drifts. Three fully
ponents is illustrated in Fig. 4. For all specimens a constant axial reversed cycles were applied at each drift ratio and then a half-
load of 220 kN (equal to 0.16 Agf’c) was applied by a horizontal domain drift ratio cycle was adopted. An initial drift ratio of
1000 kN capacity hydraulic jack to the column in a force controlled 0.20% was selected, being effectively within the linear elastic
mode and was maintained constant throughout the test. After response range of the specimen and subsequent drift ratios were
application of the 220 kN axial load to the column, quasi static lat- set to not less than one and one-quarter times the previous drift
eral cyclic loading of increasing amplitudes was applied at the top ratios, and not more than one and one-half times the aforemen-
of the beam by a hydraulic actuator with a ±100 kN loading capac- tioned drift ratio. This loading sequence was intended to ensure
that displacements were increased gradually in steps that were
neither too large nor too small. The selected drift was intended
Table 1
to cause displacements that simulated high levels of inelastic
Mechanical properties of concrete at day of testing.
deformations such as would be experienced by the frame during
Specimen Specimen Cylinder Prism E-modulus a severe earthquake. Fig. 5 shows the lateral cyclic loading protocol
description compressive tensile (MPa)
according to ACI Committee 374.1-05 [39].
strength strength
(MPa) (MPa)
Specimen J1 Confined joint 23.0 3.0 22,540
according to ACI
318-08 code 8
Specimen J2 Unconfined joint 23.3 3.0 22,687 6.0%
with no transverse
6
4.5%
reinforcement 3.5%
4 2.75%
Specimen J3 Unconfined joint 24.7 3.1 23,500 2.2%
and inadequate 1.75%
2 1.0%
1.4%
Drift (%)

0.75%
anchorage length 0.25%
0.5%
of beam bottom 0
bars 0.2% 0.35%
-2

-4
Table 2
-6
Material properties of reinforcement bars obtained from material testing.

Bar diameter Yield strength Ultimate Yield Ultimate -8


(mm) (MPa) strength (MPa) strain (%) strain (%) 0 150 300 450 600 750 900

8 350 410 0.18 18


Step
14 460 680 0.20 13
Fig. 5. Lateral cyclic loading protocol.
J. Shafaei et al. / Engineering Structures 81 (2014) 412–431 417

Drift= ± 1.0 % Drift= ± 3.5% Drift= ± 6.0%


(a) Specimen J1

Drift= ± 1.0 % Drift= ± 3.5% Drift= ± 4.5%


(b) Specimen J2

Drift= ± 1.0 % Drift= ± 3.5% Drift= ± 4.5%


(c) Specimen J3
Fig. 6. Damage progression and crack observation of specimens (images by Jalil Shafaei).

2.4. Test observations of specimens and continuous stiffness and strength degradation with increasing
displacement with respect to the seismic specimen (J1), which is
Crack observation and the progress of damage during the test in primarily attributed to the joint shear failure of specimen J2 (see
successive steps of all specimens are illustrated in Fig. 6. As shown Fig. 7b) and joint shear failure and bond failure of beam bottom
in Fig. 6a specimen J1 was controlled by the flexural mechanism of reinforcement of specimen J3 (see Fig. 7c). Out-cycle and in-cycle
the beam at the column face, whereas specimens J2 and J3 failed by decline of strength and stiffness were appeared after yielding of
concrete crushing and slip of beam bottom longitudinal bars in the the specimens J2 and J3 in a gradual manner as shown in Fig. 7b
joint region and shown in Fig. 6b and c respectively. The first flex- and c. This phenomenon may be assigned to the shear and bond
ural cracks of all specimens appeared at the bottom of the beam at failure in the joint region. It is worth noting that in the hysteretic
0.25% drift, and for specimens J1 and J2 these cracks gradually curve of specimen J1 as shown in Fig. 7a, the first reversal of cycles
propagated to a height of 750 mm, whereas for specimen J3 these is approximately the same with the second and third ones. The
cracks propagated to a height of 300 mm for the pull direction, and hysteresis loops of specimen J1 are larger and remained stable
to 650 mm for the push direction. Yield of longitudinal and trans- without apparent strength deterioration and stiffness degradation
verse reinforcement was measured by strain gauges, with first with respect to specimens J2 and J3. The hysteretic responses of the
yield of the beam longitudinal reinforcement occurring at a drift specimens show that the ultimate load and deformation capacity
ratio of 1.1% for specimen J1, while yield of the beam longitudinal of specimen J1 were higher than specimens J2 and J3 for both
reinforcement was not recorded for specimens J2 and J3. the push and pull directions.

2.5. Hysteretic load–displacement curves 2.6. Hysteretic joint shear stress–deformation response

The hysteretic load–displacement responses of specimens are Measurement of diagonal displacements within the joint region
presented in the form of displacement versus corresponding hori- enabled calculation of joint shear distortion. The joint shear defor-
zontal applied load to top of the beam as shown in Fig. 7. Fig. 7a mation represents the damage condition of the joint with increas-
illustrates the response of specimen J1, specimen designed accord- ing drift ratio. As shown in Fig. 4, a set of two LVDTs crossing each
ing to ACI Committee 318M-11 [38] requirement, exhibited ductile other were mounted diagonally on the joint panel with respect to
load–displacement hysteretic response with no notable pinching the beam’s axis to measure the joint shear deformations. Using two
or strength drops by the end of testing. The hysteretic loops of LVDTs on side of the joint region in each specimen, angular
the non-seismic specimens (J2 and J3) show considerable pinching changes were obtained at each drift ratio of loading protocol. The
418 J. Shafaei et al. / Engineering Structures 81 (2014) 412–431

60 60
50 (a) Specimen J1 Flextural Capacity of 50 (b) Specimen J2
Flextural Capacity of

Force (kN)
40 the beam 40
32kN 32kN the beam
Force (kN)
30 30
20 20
10 Joint detail 10 Joint detail
0 0
-10 -10
-20 -20
-30 -30
-40 -40
Flextural Capacity of 41.5kN Flextural Capacity of 41.5kN
-50 the beam Drift ratio (%) -50 the beam Drift ratio (%)
-60 -60
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6

60
50 (c) Specimen J3
Flextural Capacity of
40
Force (kN)

32kN the beam


30
20
10 Joint detail
0
-10
-20
-30
-40
Flextural Capacity of 41.5kN
-50 the beam Drift ratio (%)
-60
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6

Fig. 7. Load–displacement hysteretic response of tested beam-column joints.

D1 þ D2
cav e ¼ ð1Þ
D sin 2/

where D1 and D2 are the changes in length of the diagonal gauges in


the joint region (see Fig. 8a), D is the undistorted length of the diag-
onal gauge, and / is the inclination angle of the diagonal gauge to
the horizontal line, where tan / = h/b, with h and b respectively
being the vertical and horizontal distances between the end points
of the diagonal gauge (see Fig. 8a). Eq. (1) gives an idea of the aver-
age shear deformation in the joint region even after development of
diagonal cracks in the instrumented region. As shown in Fig. 8b, the
joint horizontal shear force and joint shear stress can be calculated
according to Eqs. (2) and (3) respectively.
(a) Joint shear deformation (b) Joint shear force
V jh ¼ T s  V col ð2Þ
Fig. 8. Parameter for calculation of joint shear deformation and joint shear force.
V
v jh ¼ jh ð3Þ
Aj
shear deformation was calculated from measured contraction and
extension of the diagonal gauge lines measured by LVDTs. The where Ts is the tension force in the beam longitudinal reinforce-
average joint shear deformation angle, cavg, can be calculated using ment, Vcol is the column shear force as shown in Fig. 8b and Aj is
Eq. (1). effective joint area.

6 6 6
Joint shear stress (MPa)

Joint shear stress (MPa)

Joint shear stress (MPa)

(a) Specimen J1 (b) Specimen J2 (c) Specimen J3


4 4 4

2 2 2
Push Push Push
0 0 0
Pull Pull Pull
-2 -2 -2

-4 -4 -4

-6 -6 -6
-0.01 -0.008 -0.006 -0.004 -0.002 0 0.002 0.004 -0.01 -0.005 0 0.005 0.01 0.015 -0.04 -0.03 -0.02 -0.01 0 0.01
Joint shear deformation (rad) Joint shear deformation (rad) Joint shear deformation (rad)

Fig. 9. Joint shear stress versus joint shear deformation relationships of tested beam-column joints.
J. Shafaei et al. / Engineering Structures 81 (2014) 412–431 419

To gain an insight into the behaviour of beam-column joint with diagonal cracking in the joint zone began at a drift of 1.75%
sum-assemblages, the joint shear stress-deformation response (see Fig. 6a). For specimen C2, spalling of cover concrete in the joint
was compared among three specimens in Fig. 9. As shown in region occurred at a drift ratio of between 1.4% and 2.0%, and from
Fig. 9, seismically detailed specimen J1 showed smaller joint shear a drift of 2.0% the cracking was mainly concentrated in the joint
deformation in comparison to non-seismically detailed specimens region and formed a typical X-shaped pattern with X-shape cracks
J2 and J3. For specimens J1, joint concrete damage was not signif- in the joint zone commenced at a drift ratio of 0.75%, indicating
icant by the end of testing, but little concrete spalling was observed imminent joint failure (see Fig. 6b). For specimen C3, vertical
cracks in the joint panel formed at a drift ratio of 0.5% due to ten-
sile bond-slip of the beam bottom reinforcement, and diagonal
2 shear cracks in the joint zone developed at a drift ratio of 0.75%
for the reverse loading direction (see Fig. 6c).
Specimen J1
1.6
Value of damage index

Specimen J2
2.7. Park and Ang’s damage index of tested specimens
Specimen J3
1.2
Several quantitative dimensionless measures of the deteriora-
tion of RC components based on displacement ductility ratio and
0.8 hysteretic dissipated energy have been reported in the literature
[40]. The damage index model by Park and Ang [41] has been
0.4
widely used because of its simplicity and the fact that it has been
calibrated using various experimental tests. In the current study, in
order to investigate the seismic performance of non-seismically
0 and seismically detailed beam-column joints the damage index
0 1 2 3 4
of the non-seismic joints were examined and compared with the
Drift (%)
seismic joints. Damage achieved by each specimen was evaluated
Fig. 10. Comparisons of the Park and Ang’s damage indices for all tested specimens. using Park and Ang’s damage model [41] bas applied to RC
beam-column joints by Karayannis et al. [40] and Karayannis and

(a) Components of the beam-column (b) Internal and external displacements


joint model and rotations
Fig. 11. Beam-column joint element proposed by Lowes and Altoontash.

Fig. 12. Hysteretic one-dimensional load–deformation model.


420 J. Shafaei et al. / Engineering Structures 81 (2014) 412–431

Sirkelis [42]. This model is expressed as a linear combination of springs simulate stiffness and strength degradation due to bond-
ultimate displacement and dissipated energy by the following slip damage, and interface-shear springs simulate reduced capacity
expression: for shear transfer at the joint perimeter due to crack opening. A rel-
Z atively stiff elastic load–deformation response is assumed for the
dMx b
DI ¼ þ dE ð4Þ interface-shear components. The deformation of the component
du Q y :du
is based upon the displacement at the four internal DOF in the
where dMx is the maximum deflection attained during seismic load- shear panel along with the combination of displacements at the
ing; du is the ultimate deflection capacity under monotonic loading; twelve exterior DOF.
b is a model parameter that depends on the value of shear force,
axial force, and amount of longitudinal and confinement reinforce- 3.2. Envelope curve
ments; Qy is the calculated yield strength; and dE is the incremental
dissipated hysteretic energy. Values of dMx, Qy and dE are obtained To facilitate implementation of the joint model, the load–defor-
from the experimental results of the joint specimens, whereas the mation response of each components of the joint element was
value of du that comprises contributions from deformations of defined using a general one-dimensional hysteretic model devel-
beams, columns and joint, was estimated based on the empirical oped by Lowes and Altoontash along with modifications were used
formula for the calculation of ultimate drift ratio according to CEN in this paper. This one dimensional model was defined by a multi-
Eurocode 8 [43]. linear response envelope, a tri-linear unload–reload path, and
three damage rules that control the evolution of the response path.
1 The four states that define a hysteretic uniaxial material model
du ¼ 0:016
cel (loading, reloading, and unloading states) are shown in Fig. 12.
 0:225 Load–deformation paths for the loading states (states 1 and 2)
maxð0:01; x0 Þ f
aq yt
ðL Þ 25 s f cm ð1:25Þ100qd
0:35
 0:3P f cm ð5Þ are defined by the backbone envelopes. The backbones can be
maxð0:01; xÞ
modified during the analysis to simulate the cyclic deterioration.
where cel is equal to 1.0 in this study (applies to the secondary seis- Whenever there is a deformation reversal, the unloading/reloading
mic elements); P⁄ = P/(bhfcm) is the axial load index; b and h are states (states 3 and 4) are redefined based on specific rules for each
width and height of the cross-section respectively; x and x0 the material model and the unloading/reloading path endpoints.
ratio of tension and compression reinforcements respectively; fcm Calibration of the model requires 16 parameters to define the
is the concrete compressive strength; L⁄ is the shear span index response envelope (material states 1 and 2), 6 parameters to define
and is equal to M/Vh; qs = (b0h0/bs) is the transverse reinforcement the two unload–reload paths (material states 3 and 4) and 12
ratio, b0 and h0 are the width and height of the confined concrete parameters to define the hysteric damage rules.
core to the centerline of the stirrup, respectively; s is the spacing In this paper the cyclic response parameters were calibrated
of stirrups; fyt is the yield strength of transverse reinforcement; qd with experimental result of External RC beam-column sub-assem-
is the diagonal reinforcement ratio; and a is the confinement effec- blages. Joint transverse reinforcement and bond-slip of the longitu-
tiveness factor derived from the following equation. dinal reinforcements were assumed to contribute to shear panel
   P 2! stiffness and strength. The multi-linear response envelope for all
s s bi tested specimens is shown in Fig. 13. As shown Fig. 13, due to
a¼ 1 1 1 ð6Þ
2b0 2h0 6b0 h0 installation of an unequal longitudinal reinforcement at top and
bottom of the beam section, the envelope curves of beam-column
where bi denotes the centerline spacing of the longitudinal rein- joints were not symmetrical.
forcements (index by i) laterally restrained by a stirrup corner or The unload–reload path was defined by three parameters [37]:
a cross-tie along the perimeter of the cross-section. In the cases
where the calculated ultimate displacement du had a higher value 1. rDisp: It’s the ratio of the deformation which reloading occurs
than the experimentally observed value the latter was used. The to the maximum historic deformation demand.
ultimate displacement from tests was defined as the displacement 2. rForce: It’s the ratio of the force which reloading begins to force
at a predefined 20% drop below the maximum strength [7]. The corresponding to the maximum historic deformation demand.
model parameter b is taken to be 0.25 for the tested specimens as 3. uForce: It’s the ratio of strength developed upon unloading from
recommended by Altoontash [24]. The calculated values of damage negative load to the maximum strength developed under
indices based on the above-described model are presented and monotonic loading.
compared in Fig. 10 for seismic and non-seismic specimens. As
shown in Fig. 10, seismic specimen show quite lower damage indi-
ces at each drift ratio as compared with the non-seismic specimens
during the initial loading.
1
Joint shear stress/√f'c (MPa)

0.8
3. Description of the joint model
0.6
0.4
3.1. Main consideration
0.2
0
Fig. 11 shows the detailed idealization of the beam-column
-0.2
joint element developed by Lowes and Altoontash [34]. This joint Joint J1
element has four exterior nodes, each with three degrees of free- -0.4
Joint J2
dom (DOF), thus the joint is compatible with traditional 2D -0.6
Joint J3
beam-column elements as shown in Fig. 11b. The joint model com- -0.8
prises of eight zero-length bar-slip springs, four interface shear -1
-0.06 -0.04 -0.02 0 0.02 0.04 0.06
springs and a panel that deforms only in shear. The shear-panel
Joint rotation (rad)
component with a zero-length rotational spring simulates strength
and stiffness loss due to shear failure of the joint region, bar-slip Fig. 13. Envelope curves of beam-column joint specimens.
J. Shafaei et al. / Engineering Structures 81 (2014) 412–431 421

Table 3 and deterioration in the strength developed in the vicinity of the


Calibrated parameters of unload/reload for beam-column joint specimens. maximum and minimum deformation demands (reloading
Joint type rDisp rForce uForce strength deterioration) [34]. Fig. 14 shows the impact of these
Joint J1 0.25 0.40 0.01 three different damage modes on the hysteretic response.
Joint J2 0.5 0.25 0.01 The calibrated pinching parameters for all tested specimens (J1,
Joint J3 0.5 0.10 0.01 J2 and J3) are presented in Table 4.

4. Analytical verification of joint model


4. The calibrated parameters of unload-reload path for seismically
The joint element calibration procedures were evaluated
and non-seismically detailed beam-column joints (J1, J2 and J3)
through comparison of simulated and observed response using
are presented in Table 3.
experimental results of the tested specimen with various joint
reinforcement detailing. In this process, appropriate materials
3.3. Cyclic response and elements were selected from software library and their param-
eters were analytically determined for the selected beam-column
Stiffness degradation, strength deterioration and pinching joint element. Specimens were subjected to simulated earthquake
behaviour often occur due to anchorage-zone behaviour, inade- loading by forcing the beam tip through a cyclic displacement his-
quate confinement and crack opening and closing in RC elements. tory and a constant column axial load was applied using load con-
This behaviour was represented using the three damage rules that trol. The sub-assemblage with actual test boundary conditions was
define evolution of the response envelope and unload–reload paths modelled in OpenSees software framework. Fig. 15 shows beam-
as a function of load–deformation history. Hysteretic damage was column joint sub-assemblage in test setup and simulated model
simulated through degradation in unloading stiffness (unloading for beam-column joint unit with considering joint deformations
stiffness degradation), deterioration in strength achieved at previ- using selected joint element and beam-column joint unit with rigid
ously unachieved deformation demands (strength deterioration), joint assumption.

(a) Unloadind stiffness degradation (b) Reloadind stiffness degradation (c) Strength degradation
Fig. 14. Hysteretic damage rules of Pinching4 material.

Table 4
Calibrated pinching parameters for tested specimens.

Joint type Unloading stiffness degradation Reloading strength deterioration


gK1 gK2 gK3 gK4 Limit gD1 gD2 gD3 gD4 Limit
Joint J1 0.7 0.0 0.09 0.0 0.9 0.15 0.0 0.22 0.0 0.95
Joint J2 0.7 0.0 0.15 0.0 0.9 0.15 0.0 0.25 0.0 0.95
Joint J3 0.7 0.0 0.15 0.0 0.9 0.25 0.0 0.25 0.0 0.90

(a) Beam-column joint sub-assemblage in (b) Simulated model with (c) Simulated model without
experimental test setup (images by Jalil Shafaei). considering joint flexibility. considering joint flexibility.
Fig. 15. Experimental and simulated model of beam-column joint sub-assemblages.
422 J. Shafaei et al. / Engineering Structures 81 (2014) 412–431

45 45
Experimental Experimental
30 30
Analytical Rigid Joint

15 15
Force (kN)

Force (kN)
0 0

-15 -15

-30 -30

-45 -45

-60 -60
-6 -4 -2 0 2 4 6 -6 -4 -2 0 2 4 6
Drift (%) Drift (%)
(a) Simulation of specimen J1 with considering joint (b) Simulation of specimen J1 without considering joint
flexibility flexibility
45 45
Experimental Experimental
30 30
Analytical Rigid Joint

15 15
Force (kN)

Force (kN)

0 0

-15 -15

-30 -30

-45 -45

-60 -60
-6 -4 -2 0 2 4 6 -6 -4 -2 0 2 4 6
Drift (%) Drift (%)
(c) Simulation of specimen J2 with considering joint (d) Simulation of specimen J2 without considering joint
flexibility flexibility
45 45
Experimental Experimental
30 30
Analytical Rigid Joint

15 15
Force (kN)

Force (kN)

0 0

-15 -15

-30 -30

-45 -45

-60 -60
-6 -4 -2 0 2 4 6 -6 -4 -2 0 2 4 6
Drift (%) Drift (%)
(e) Simulation of specimen J3 with considering joint (f) Simulation of specimen J3 without considering joint
flexibility flexibility
Fig. 16. Comparison of analytical results with experimental test results.

4.1. Overview of the OpenSees software framework nonlinear beam-column element formulations and global solution
algorithms. OpenSees has available several models for describing
OpenSees is an object-oriented, open source framework devel- the behaviour of materials (steel and concrete) and the global cyc-
oped to simulate the response of structural systems subjected to lic behaviour of RC structural components. The member stiffness
earthquakes by researchers at the Pacific Earthquake Engineering and forces are obtained by numerically integrating the stiffness
Research (PEER) Center [37]. OpenSees was chosen for dynamic and forces of sections along the member length. The section defor-
and static analysis of selected frame because of the relative ease mation is used to obtain the strain in each fibre, based on the plane
with the new joint element formulation and because it includes section assumption. The fibre stress and stiffness are updated
J. Shafaei et al. / Engineering Structures 81 (2014) 412–431 423

5000 4000

4000 3000

Microstrain
Microstrain
3000 2000

2000 1000

1000 0

0 Experimental -1000 Experimental


Analytical Analytical
-1000 -2000
-3 -2 -1 0 1 2 3 -4 -2 0 2 4
Drift (%) Drift (%)
(a) Simulation results of specimen J1
4000

3000
Microstrain

2000

1000

0
Experimental
Analytical
-1000
-4 -2 0 2 4
Drift (%)
(b) Simulation results of specimen J2
1000 3000

500
2000
Microstrain

0
Microstrain

-500 1000

-1000
0
-1500 Experimental Experimental
Analytical Analytical
-2000 -1000
-4 -2 0 2 4 -4 -2 0 2 4
Drift (%) Drift (%)
(c) Simulation results of specimenJ3
Fig. 17. Comparison of simulation results of strain gauges with experimental results.

according to the corresponding material models, followed by behaviour of longitudinal steel reinforcement. The model uses a
upgrading of the section force resultant and the corresponding bilinear backbone, and the Menegotto–Pinto model for loading
stiffness [44]. and unloading rules [37]. Monotonic steel material properties were
identified based on the reported materials tests as shown in
4.2. Material and element simulation models Table 2.
The column and beam section were discretized as fibre section
Uniaxial stress–strain concrete and steel materials were used to using 20 mm meshes. The confined core and the unconfined cover
model the experimental specimens at the section level. Concrete03 concrete were defined using Concrete03 material described before.
material was used to model confined and unconfined concrete Steel02 was used to model longitudinal steel reinforcement in the
behaviour. This material model defines the uniaxial response of section using the as-built locations. Beam and columns were mod-
concrete in compression and tension. The model features linear elled on the member level using an inelastic fibre element (nonlin-
tension softening to represent inelastic tension action. The uncon- earBeamColumn element) along the element length, with five
fined concrete parameters were used for cover concrete that were integration points within the field span. Joint was modelled with
experimentally obtained from the reported materials tests at the beamColumnJoint element proposed by Lowes and Altoontash
test date (see Table 1) for tested specimens. Concrete strain at [34]. Pinching4 have been used for modelling shear-panel and
maximum strength was assumed as 0.0035 [26]. The maximum bar-slip components of beam column joint element [37].
tensile stress was taken from splitting test results as shown in Fig. 16a, c and e show the simulation and experimental results
Table 1. The modification for the core concrete was implemented of load–displacement response of tested beam-column joints with
by using confined concrete model proposed by Mander et al. considering joint flexibility for specimens J1, J2 and J3 respectively.
[44]. Steel02 material model was used to model the uniaxial cyclic As exhibited in Fig. 16a, c and e, the simulation analytical model
424 J. Shafaei et al. / Engineering Structures 81 (2014) 412–431

(a) Typical floor plan of the structures

(b) 5-storey frame (c) 10-storey frame


Fig. 18. Detail and dimension of five and ten storey frames (dimensions in millimetres).

region (J1), the second group having no transverse reinforcement


in the joint region (J2) and the third group having both no trans-
Table 5
Construction detailing and modelling assumptions for frames. verse reinforcement in the joint region and inadequate anchorage
length of beam bottom bars (J3). The dead and live design loads
Frame description Modelling Construction
were assumed to be 6.5 kN/m2 and 2.0 kN/m2 respectively. The
assumption detailing
effective seismic mass was considered to be the dead load plus
Frame with rigid panel zone Rigid panel zone Equal for all types
20% of the live load [45].
Frame with rigid centerline of construction detailing
Frame with flexible joint J1 Flexible panel Seismic joint In the current study three types of joint assumption were con-
Frame with flexible joint J2 zone Unconfined joint sidered for modelling of frames, including rigid joint assumption
Frame with flexible joint J3 Unconfined joint and without considering joint dimension (according to Paulay and
inadequate anchorage Priestley [7] recommendations), rigid joint assumption with con-
length
sidering joint dimension (according to ASCE/SEI 41 [25] and
FEMA-356 [26] recommendations) and flexible joint assumption
with considering nonlinear behaviour of joint based on the rein-
was able to successfully predict the experimental response of forcement detailing of the joint. Properties of all models are shown
tested specimens. Moreover as shown in Fig. 16b, d and f, there in Table 5.
is a considerable difference between experimental results and sim- Frame simulation with Beam-column joint element and rigid
ulation response of beam-column joint with rigid joint assumption joint element are shown in Fig. 19. The frames were modelled
showing that rigid joint assumption did not predict lateral using the following uniaxial materials in OpenSees software
response of frames precisely especially for non-seismically detailed framework:
beam-column joint with non-seismic detailing.
Fig. 17a–c shows the comparison of analytical and experimental 1. Cover concrete of beam and column: Concrete03.
results of strain gauges on the beam longitudinal reinforcement for 2. Confined core concrete of beam and column: Concrete03.
seismically and non-seismically detailed specimens. As shown in 3. Reinforcing bars of beam and column: Steel02.
Fig. 17a–c, the analytical model was able to capture the general
response of strain history of longitudinal bars with reasonable Moreover, nonlinearBeamColumn was used for modelling the
accuracy. beam and column elements and five integration points were
assigned in each element. For modelling of the beam-column joint
5. Properties of five and ten storey existing frames element, the calibrated experimental backbone relationship was
assigned to the exterior, interior and roof joints. The beam-column
Five and ten storey RC moment frames was selected for the joint element was able to model shear deformation of the panel
assessment of joint behaviour on lateral response of RC frames. zone, as well as slip of the beam longitudinal reinforcement in
Dimension and properties of frames are shown in Fig. 18. Frame the joint regions.
components including beams and columns were designed and
detailed base on ACI Committee 318M-11 [38] requirements. The 6. Numerical simulation results
main difference of frames was in the joint reinforcement detailing.
The joint reinforcement details were divided into three groups, Effect of the different reinforcement detailing in the joint region
with the first group having adequate reinforcement in the joint and various joint modelling assumption on lateral response of
J. Shafaei et al. / Engineering Structures 81 (2014) 412–431 425

Fig. 19. Simulation model for (a) 10-storey frame with joint element, (b) 10-storey frame with rigid joint, (c) 5-storey frame with joint element, and (d) 5-storey frame with
rigid joint.

moment resisting frame was investigated through performing non- 6.1. Nonlinear static analysis
linear static and incremental dynamic analyses. The modal proper-
ties of 5 and 10 storey frames are presented in Table 6. As shown in In the nonlinear static pushover procedure, monotonically
Table 6, considering joint deformations in modelling of frames increasing forces are applied to an analytical model of the structure
increased the fundamental period of frames with considering joint until the displacement of the control node exceeds the target dis-
deformations with respect to those of frame with rigid joints. The placement. In the current study all selected frames were subjected
first period of frame with centerline joint modelling assumption to vertical and lateral loads based on the nonlinear static procedure
and joint detailing similar to specimens J1, J2 and J3 increased by given by the ATC40 [46]. Nonlinear static analyses of the frame
17%, 55%, 62% and 70% for five storey frame and increased by were carried out using OpenSees software framework. For this pur-
20%, 97%, 110% and 123% for ten storey frame with respect to frame pose, a constant gravity load, equal to the total dead load plus 20%
with rigid joint element respectively. of the live load based on the ASCE 7 [45], was applied to each

Table 6
Modal properties of the studied frames.

Frame name Period (s) Increasing 1st period relative to frame with rigid joint (%)
1st 2nd 3rd
5 Storey Frame with rigid panel zone 0.69 0.26 0.14 0
Frame with rigid centerline 0.81 0.31 0.16 17
Frame with flexible joint J1 1.07 0.36 0.17 55
Frame with flexible joint J2 1.12 0.37 0.18 62
Frame with flexible joint J3 1.17 0.38 0.18 70
10 Storey Frame with rigid panel zone 1.01 0.38 0.23 0
Frame with rigid centerline 1.15 0.44 0.26 20
Frame with flexible joint J1 1.68 0.61 0.34 97
Frame with flexible joint J2 1.77 0.64 0.35 110
Frame with flexible joint J3 1.86 0.67 0.36 123

200 250

160 200
Base shear (kN)

Base shear (kN)

120 150

80 100 Rigid Joint


Rigid Joint
Rigid CL Joint Rigid CL Joint
Joint J1 Joint J1
40 50
Joint J2 Joint J2
Joint J3 Joint J3
0 0
0 200 400 600 0 200 400 600 800 1000
Roof displacement (mm) Roof displacement (mm)
(a) (b)

Fig. 20. Base shear-roof displacement response for (a) 5-storey frame and (b) 10-storey frame.
426 J. Shafaei et al. / Engineering Structures 81 (2014) 412–431

frame. For lateral seismic loads, the analysis was performed by model because considering the joint element increased the flexibil-
assuming a vertical distribution proportional to the storey shear ity of the frame. Therefore, fundamental period of the building was
distribution calculated by combining modal responses from a also affected by the inclusion of the joint flexibility, as shown
response spectrum analysis of the building, including sufficient Table 6.
modes to capture at least 90% of the total building mass [26]. As
a result, an inverted triangular distribution over the height was 6.2. Incremental dynamic analysis (IDA)
used as the lateral load pattern. P–D effect was also considered
in the nonlinear static pushover analysis. The nonlinear static analysis results identified that the initial
Fig. 20 shows the pushover curves for both the positive and stiffness of the simulated frame was different between the model
negative loading direction in terms of the base shear versus the with joint element and that with rigid joints. The change of the ini-
roof displacement for five and ten storey frames. For each simu- tial stiffness resulted in change of the fundamental period of the
lated frame, the lateral responses were compared between the frame. This observation highlighted the need to perform dynamic
model with considering various joint modelling assumptions and analysis of the frame. Incremental dynamic analysis (IDA) is a
the model with rigid joints. From the pushover curves presented method, in which through consecutive nonlinear dynamics analy-
in Fig. 20, it was found that the initial stiffness of simulated frames sis subjected to several scaled ground motions, the ability to pre-
reduced as joint element were included in the computational dict the seismic capacity and demands is gained. According to

1 1

Peak record acceleration (g)


Peak record acceleration (g)

0.8 0.8

0.6 0.6

Rigid Joint Rigid Joint


0.4 0.4
Rigid CL Joint Rigid CL Joint
Joint J1 Joint J1
0.2 Joint J2 0.2 Joint J2
Joint J3 Joint J3
0 0
0 2 4 6 8 10 12 14 0 2 4 6 8
Maximum inter storey drift (%) Maximum inter storey drift (%)
(a) (b)
Fig. 21. IDA curves for all simulated frames with different joint detailing for (a) 5-storey frame and (b) 10-storey frame.

(a) 1000 (b) 800 0.25g


0.25g
0.50g 0.50g
Roof displacement (mm)

800 0.75g 0.75g


600 1.00g
Base shear (kN)

1.00g

600
400
400

200
200

0 0
Rigid Rigid CL Joint J1 Joint J2 Joint J3 Rigid Rigid CL Joint J1 Joint J2 Joint J3

(c) 1000 (d) 1000 0.25g


0.25g 0.50g
0.50g 0.75g
Roof displacement (mm)

800 800
0.75g 1.00g
Base shear (kN)

1.00g
600 600

400 400

200 200

0 0
Rigid Rigid CL Joint J1 Joint J2 Joint J3 Rigid Rigid CL Joint J1 Joint J2 Joint J3

Fig. 22. Maximum roof displacement and base shear for all simulated frames with different joint detailing. (a and b) Roof displacement and base shear for 5-storey frame (c
and d) Roof displacement and base shear for 10-storey frame.
J. Shafaei et al. / Engineering Structures 81 (2014) 412–431 427

recommendations of FEMA P695 [47] entitled ‘‘Quantification of model when subjected to Northridge earthquake record scaled to
Structure Seismic Performance Factors’’ for IDA analysis, several PGA of 0.5 g. The existing frames with flexible joint assumption
earthquake records required for incremental dynamic analysis. suffered larger maximum roof displacement than that of the
All analytical models were built in OpenSees software framework, frames with rigid joint assumption. This is mainly due to the
and ground motions with different scale factors were imposed to increased deformations in the joints caused by bond slip and joint
each structural frame. Name and specification of the input ground shear distortion. As shown in Fig. 23, joint detailing and its proper
motion records used for this dynamic analysis are available in modelling effectively increase the roof displacement. It can be con-
FEMA P695 [47]. The seismic response of frames was evaluated cluded that dynamic responses of simulated frames were strongly
by means of IDA analysis response. To draw IDA curves, each affected by the modelling of the joint flexibility and thus consider-
record was scaled from 0.1Sa (T1, 0.05) to 1.0Sa (T1, 0.05), results ing the joint flexibility is essential for simulating the earthquake
of OpenSees processing were recorded. Although each earthquake response of existing RC buildings constructed prior to the 1970s
record had an IDA curve, for assessing the performance of struc- having non-seismically detailed beam-column joints.
tures average of all earthquake records was used. Quantities of
maximum storey drift for five and ten storey frames with different 6.2.3. Inter storey drift
joint assumption are shown in Fig. 21. It is shown that the peak Inter storey drift is an important engineering parameter and is
drift was greater in the model with considering joint element espe- an indicative of structural and nonstructural performance. The
cially for non-seismically detailed frame. average maximum drifts for various types of joints detailing and
In this section results of IDA analysis for scaled ground motions various joint modelling assumption according to four scale factors
according to FEMA P695, in terms of maximum base shear, maxi- are shown in Fig. 24. As shown in Fig. 24, the maximum inter sto-
mum roof displacement, roof displacement time history, inter sto- rey drifts of frames with flexible joint modelling significantly
rey drift, maximum plastic rotation of beams and joints, and increased compared to frames with rigid joint assumption. How-
hysteresis curve of joints and beams for simulated frames are ever, because of bar-slip in the beam bottom longitudinal rein-
presented. forcement in frame with joint detailing similar to specimens J3,
the inter storey drift significantly increased in comparison with
6.2.1. Maximum base shear and roof displacement other types of frames. It can be concluded that increasing the inter
The Maximum base shear and roof displacement for five and ten storey drift resulted in structural and non-structural damages.
storey frames with five types of joints based on four scale factors
are shown in Fig. 22. The results showed that considering joint 6.2.4. Maximum plastic rotation of beams and joints
modelling in frames with non-seismically detailed joints had a sig- In frames with seismic joints detailing, forming plastic hinges in
nificant effect with respect to frames with seismically detailed beams were the essential part of the plastic deformation and
joints. For example in 5-storey frame with joint detailing similar energy dissipation. Whereas in frame with non-seismic joint
to specimens J1, J2 and J3, maximum roof displacement increased detailing, the joints performed such as weak link between beam
10%, 20% and 50% with respect to frame with rigid joint assumption and column, so absorbing of energy in beam plastic hinges signif-
respectively. In terms of base shear the results show that consider- icantly reduce. It can be concluded that maximum rotation of
ing joint modelling reduced base shear with respect to frame with joints in frame with non-seismic joint detailing is greater than that
rigid joint assumption. of joints in frame with seismic joint detailing, as shown in Fig. 25.
Moreover, plastic rotation of beams in frame with seismic joint
6.2.2. Roof displacement time history detailing is more than in non-seismic ones. Also joint rotations in
As examples of the results, Fig. 23 shows the comparison of the frames with lack of beam bottom bar anchorage length in the joint
roof displacement time histories for the simulated frames with region were increased about 90% with respect to frames with
joint detailing similar to specimens J1, J2 and J3 and with rigid joint unconfined joint detailing.
Roof displacement (mm)

400
Roof displacement (mm)

400

200 200

0 0
-200 -200
Frame with Rigid Joint
-400 -400 Frame with Rigid Joint
Frame with Rigid CL Joint Frame with Joint J1
-600 -600
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Time (Sec) Time (Sec)
Roof displacement (mm)

400
Roof displacement (mm)

400

200 200

0 0

-200 -200

-400 Frame with Rigid Joint -400 Frame with Rigid Joint
Frame with Joint J2 Frame with Joint J3
-600 -600
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Time (Sec) Time (Sec)

Fig. 23. Roof displacement time history for 5 storey frame subjected to the Northridge earthquake scaled to PGA of 0.5 g.
428 J. Shafaei et al. / Engineering Structures 81 (2014) 412–431

5 5 5 5
4 4 4 4
3 3 3

Storey
Storey

Storey
Storey
Rigid Rigid Rigid Rigid
Rigid CL Rigid CL Rigid CL Rigid CL
2 2 2 2
Joint J1 Joint J1 Joint J1 Joint J1
1 Joint J2 1 Joint J2 1 Joint J2 1 Joint J2
Joint J3 Joint J3 Joint J3 Joint J3
0 0 0 0
0 1 2 3 0 2 4 6 0 2 4 6 8 10 0 5 10 15
Drift (%) Drift (%) Drift (%) Drift (%)
SF=0.25g SF=0.5g SF=0.75g SF=1.0g
(a)

10 10 10 10
8 8 8 8

Storey
6
Storey

6 6
Storey

Storey
Rigid Rigid Rigid Rigid
Rigid CL Rigid CL 4 Rigid CL Rigid CL
4 4 4
Joint J1 Joint J1 Joint J1 Joint J1
2 Joint J2 2 Joint J2 2 Joint J2 2 Joint J2
Joint J3 Joint J3 Joint J3 Joint J3
0 0 0 0
0 0.5 1 1.5 0 1 2 3 4 0 2 4 6 8 0 2 4 6 8
Drift (%) Drift (%) Drift (%) Drift (%)
SF=0.25g SF=0.5g SF=0.75g SF=1.0g
(b)
Fig. 24. Average inter-storey drift of all simulated frames for different scale factor for (a) 5-storey frame and (b) 10-storey frame.

1
Peak acceleration (g)

1 Peak acceleration (g) 1


Peak acceleration (g)

0.8 0.8 0.8


0.6 0.6 0.6
0.4 0.4 0.4
Beam Rotation
0.2 Beam Rotation Beam Rotation
Joint Rotation 0.2 0.2
Joint Rotation Joint Rotation
0
0 0
0 0.02 0.04 0.06 0 0.02 0.04 0.06 0 0.04 0.08 0.12
Rotation (Rad) Rotation (Rad) Rotation (Rad)
(a) (b) (c)
1
1 1
Peak acceleration (g)

Peak acceleration (g)

Peak acceleration (g)

0.8
0.8 0.8
0.6 0.6 0.6
0.4 0.4 0.4
Beam Rotation
0.2 Beam Rotation Beam Rotation
Joint Rotation 0.2 0.2
Joint Rotation Joint Rotation
0 0 0
0 0.01 0.02 0.03 0.04 0 0.01 0.02 0.03 0.04 0 0.02 0.04 0.06 0.08
Rotation (Rad) Rotation (Rad) Rotation (Rad)
(d) (e) (f)
Fig. 25. Comparison of maximum plastic rotation of beams and joints for (a) 5-storey frame with joint J1, (b) 5-storey frame with joint J2, (c) 5-storey frame with joint J3, (d)
10-storey frame with joint J1, (e) 10-storey frame with joint J2, and (f) 10-storey frame with joint J3.

Beams plastic rotation in frames with rigid and centerline joint in structures. As shown in Fig. 27, in frame with seismic joint detail-
modelling assumption, for five and ten storey frames are shown in ing, J1, plastic hinge can be developed without any sign of stiffness
Fig. 26. As shown in Fig. 26, beam plastic rotation was approxi- and strength degradation. In the frame with non-seismic joint
mately equal in these two types of modelling. detailing, J2, hysteresis curve showed high rate of stiffness and
strength degradation, and hence dissipation of energy in beam plas-
6.2.5. Beams and joints hysteresis response tic hinge is limited. In the frame with non-seismic joint detailing, J3,
The shape of hysteresis curve of primary element such as beams hysteresis curve showed brittle joint shear failure accompanied by
is an importance factor to identify the ability of energy dissipation slippage of the beam bottom longitudinal reinforcement. As an
J. Shafaei et al. / Engineering Structures 81 (2014) 412–431 429

Peak record acceleration (g)


1

Peak record acceleration (g)


0.8 0.8

0.6 0.6

0.4 0.4
Rigid Joint Rigid Joint
0.2 0.2
Rigid CL Joint Rigid CL Joint
0 0
0 0.02 0.04 0.06 0.08 0 0.01 0.02 0.03 0.04
Rotation (Rad) Rotation (Rad)
(a) (b)
Fig. 26. Comparison of maximum plastic rotation of beams for (a) 5-storey frame and (b) 10-storey frame.

250 250 250

Moment (kN.m)
150 150
Moment (kN.m)

150
Moment (kN.m)

50 50 50

-50 -50 -50

-150 -150 -150


Frame with Joint J1 Frame with Joint J2 Frame with Joint J3
-250 -250 -250
-0.03 -0.02 -0.01 0 0.01 0.02 0.03 -0.03 -0.02 -0.01 0 0.01 0.02 0.03 -0.03 -0.02 -0.01 0 0.01 0.02 0.03
Rotation (Rad) Rotation (Rad) Rotation (Rad)
(a) Beams
150 150 150
Frame with Joint J1 Frame with Joint J2 Frame with Joint J3
Moment (kN.m)

100
Moment (kN.m)

100 Moment (kN.m) 100


50 50 50
0 0 0
-50 -50 -50
-100 -100 -100
-150 -150 -150
-0.08 -0.06 -0.04 -0.02 0 0.02 -0.08 -0.06 -0.04 -0.02 0 0.02 -0.08 -0.06 -0.04 -0.02 0 0.02
Rotation (Rad) Rotation (Rad) Rotation (Rad)
(b) Joints
Fig. 27. Example of moment–rotation curve for beams and joints subjected to the 1994 Northridge earthquake record.

example, moment–rotation curve of several beams and joints sub- the experimental observations and analytical results, the following
jected to the 1994 Northridge earthquake record are shown in conclusions are drawn:
Fig. 27.
1. The general behaviour mode for the seismically detailed speci-
7. Conclusion men (J1) was flexural yielding in the beam at the column face
whereas for the non-seismically detailed specimens (J2 and
Three half-scale external RC beam-column joints were tested at j3) joint shear failure occurred generally before the beam sec-
the structural laboratory of the University of Tehran by simulta- tion reached its ultimate flexural strength.
neously applying a constant axial load and lateral cyclic loading 2. The average ductility for specimens J2 and J3 decreased by 27%
of increasing amplitudes. Tested specimens were comprised of and 54% with respect to the specimen J1 and the average peak
one unit having seismic reinforcement detailing and two units hav- load decreased by 14% and 42% in the pull direction and by
ing non-seismic reinforcement detailing, as is typical of pre-1970 11% and 13% in the push direction for specimens J2 and J3 in
Iranian construction practice before the introduction of seismic comparison to specimen J1.
code provisions. Based on the experimental tests, analytical verifi- 3. Pinching of the force–displacement hysteretic curve for the
cation of beam-column joint element proposed by Lowes and non-seismically detailed specimens (J2 and J3) was significantly
Altoontash was performed. According to verification study, the cal- increased and the energy dissipation capacity was decreased by
ibrated joint elements were incorporated in the analytical model 55% and 67% with respect to the seismically detailed specimen
and nonlinear static and incremental dynamic analyses were per- (J1).
formed on the five and ten storey frames with the seismically 4. A reliable 2D analytical model of the joint under cyclic loading
and non-seismically detailed joint using various modelling was developed using OpenSees software framework. The mod-
assumption to investigate how the structural responses are chan- ified beam-column joint model was calibrated by test results
ged by including joint flexibility in the frame model. Based on and good agreement was found between the experimental
430 J. Shafaei et al. / Engineering Structures 81 (2014) 412–431

and numerical hysteretic behaviour of the tested specimens. [3] Ghobarah A, Saatcioglu M, Nistor I. The impact of the 26 December 2004
earthquake and Tsunami on structures and infrastructure. Eng Struct
The model was able to capture the modes of failure, peak load
2006;28:312–26.
and initial stiffness of the tested specimens. The levels of strain [4] Zhao B, Taucer F, Rossetto T. Field investigation on the performance of building
in the longitudinal reinforcement were close for the tested structures during the 12 May 2008 Wenchuan earthquake in China. Eng Struct
specimens and the corresponding models. 2009;31:1707–23.
[5] Kam WE, Pampanin S, Elwood KJ. Seismic performance of reinforced concrete
5. Lateral structural responses of frames were strongly affected by buildings in the 22 February Christchurch (lyttelton) earthquake. New Zeal Soc
whether joint flexibility was included in modelling or not. By Earthq Eng Bull 2001;44.
modelling joint flexibility, the lateral load carrying capacity of [6] Paulay T, Park R, Priestley MJN. Reinforced-concrete beam-column joints under
seismic actions. J Am Concr I 1978;75:585–93.
the structure reduced, and also the fundamental period of the [7] Paulay T, Priestley MJN. Seismic design of reinforced concrete and masonry
structure was increased. The changes of the responses were buildings. New York: John Wiley & Sons; 1992.
more considerable for the structures whose joints had non-seis- [8] Park R. A summary of results of simulated seismic load tests on reinforced
concrete beam-column joints, beams and columns with substandard
mic detailing. reinforcing details. J Earthq Eng 2002;6:147–74.
6. The first period of frame with centerline joint modelling [9] Paulay T. Equilibrium criteria for reinforced-concrete beam-column joints. ACI
assumption and joint detailing similar to specimens J1, J2 and Struct J 1989;86:635–43.
[10] Leon R, Jirsa JO. Bidirectional loading of R.C. beam-column joints. Earthq
J3 increased by 17%, 55%, 62% and 70% for five storey frame Spectra 1986;2:537–64.
and increased by 20%, 97%, 110% and 123% for ten storey frame [11] French CW, Moehle JP. Effect of floor slab on behavior of slab-beam-column
with respect to frame with rigid joint element respectively. connections. ACI Struct J, Special Publ Des Beam-Column Joints Seismic Resist
1991;SP-123:225–58.
7. Modelling the joint region as a rigid element significantly over-
[12] Beres A, Pessiki SP, White RN, Gergely P. Implications of experiments on the
estimated the stiffness of moment resisting frames and cause to seismic behavior of gravity load designed RC beam-to-column connections.
miscalculation of structural responses especially for frames Earthq Spectra 1996;12:185–98.
with non-seismic joint detailing due to the significant joint [13] Mullas MG, Filippou FC. Analytical procedures in the study of seismic response
of reinforced concrete frames. Eng Struct 1990;12:37–48.
shear distortions and bar pull out in the joint regions. [14] Celik OC, Ellingwood BR. Modeling beam-column joints in fragility assessment
8. The current study highlighted that joint flexibility is essential of gravity load designed reinforced concrete frames. J Earthq Eng
for simulating existing RC structures constructed in Iran before 2008;12:357–81.
[15] Melo J, Fernandes C, Varum H, Rodrigues H, Costa A, Arêde A. Numerical
the 1970s with non-seismic joint detailing and conventional modelling of the cyclic behaviour of RC elements built with plain reinforcing
analyses (rigid joint assumption) may not reflect the realistic bars. Eng Struct 2011;33:273–86.
responses of those types of RC structures under earthquake [16] Sharma A, Eligehausen R, Reddy GR. A new model to simulate joint shear
behavior of poorly detailed beam-column connections in RC structures under
loading. seismic loads, part I: exterior joints. Eng Struct 2011;33:1034–51.
[17] Sagbas G, Vecchio FJ, Christopoulos C. Computational modeling of the seismic
It should be notated that there is a large number of RC buildings performance of beam-column subassemblies. J Earthq Eng 2011;15:640–63.
[18] Birely AC, Lowes LN, Lehman DE. A model for the practical nonlinear analysis
in Iran that constructed before development of seismic codes, and of reinforced-concrete frames including joint flexibility. Eng Struct
hence these building are vulnerable to earthquake loading due to 2012;34:455–65.
non-seismic detailing in the joint region. So it is very essential to [19] Mergos PE, Kappos AJ. A gradual spread inelasticity model for R/C beam-
columns, accounting for flexure, shear and anchorage slip. Eng Struct
retrofit these building as soon as possible with practical and
2012;44:94–106.
effective retrofit techniques. The first step for retrofitting of these [20] Fernandes C, Varum H, Costa A. Importance of the bond-slip mechanism in the
buildings is detailed seismic performance assessment to evaluate numerical simulation of the cyclic response of RC elements with plain
the behaviour of building close to reality. For this reason there is reinforcing bars. Eng Struct 2013;56:396–406.
[21] Unal M, Burak B. Development and analytical verification of an inelastic
high demand to introduce a simple and practical analytical method reinforced concrete joint model. Eng Struct 2013;52:284–94.
for using in the commercial analytical software. The next step of [22] Alva GMS, El Debs ALHdC. Moment–rotation relationship of RC beam-column
the current research program is proposing a simple analytical connections: experimental tests and analytical model. Eng Struct
2013;56:1427–38.
method for modelling the various types of joint detailing in the [23] Ghobarah A, Biddah A. Dynamic analysis of reinforced concrete frames
commercial software such as ETABS and still is in progress. including joint shear deformation. Eng Struct 1999;21:971–87.
[24] Altoontash A. Simulation and damage models for performance assessment of
reinforced concrete beam-column joints. Stanford, California: Stanford
Acknowledgment University; 2004.
[25] ASCE/SEI 41. Seismic rehabilitation of existing buildings. Reston, Virginia:
American Society of Civil Engineers; 2007.
The authors are grateful to all the sponsors of this research
[26] FEMA-356. Prestandard and commentary for the seismic rehabilitation of
program for their support. The authors acknowledge the techni- buildings. Washington, D. C: Federal Emergency Management Agency; 2000.
cians of the University of Tehran structural laboratory for their [27] Giberson MF. Two nonlinear beams with definition of ductility. J Struct Div
(ASCE) 1969;95:137–57.
role in facilitating the experimental works. The employees of
[28] Otani S. Inelastic analysis of RC frame structures. J Struct Div-ASCE
High Performance Computing Center (HPCC) of the University of 1974;100:1433–49.
Tehran are also thanked for their assistance in performing the non- [29] El-Metwally SE, Chen WF. Moment rotation modeling of reinforced-concrete
linear static and incremental dynamic analyses. The first author beamcolumn connections. ACI Struct J 1988;85:384–94.
[30] Alath S, Kunnath SK. Modeling inelastic shear deformations in RC beam-
would also like to acknowledge Professor Jason Ingham from the column joints. In: Engineering mechanics proceedings of 10th conference.
University of Auckland for his helpful comments. The authors also University of Colorado at Boulder, Boulder, Colorado: ASCE; 1995. p. 822–5.
want to express their gratitude to all the Engineering Structures [31] Biddah A, Ghobarah A. Modelling of shear deformation and bond slip in
reinforced concrete joints. Struct Eng Mech 1999;7:413–32.
Journal reviewers for providing constructive comments on the [32] Favvata MJ, Izzuddin BA, Karayannis CG. Modelling exterior beam-column
original manuscript. joints for seismic analysis of RC frame structures. Earthq Eng Struct D
2008;37:1527–48.
[33] Youssef M, Ghobarah A. Modelling of RC beam-column joints and structural
References walls. J Earthq Eng 2001;5:93–111.
[34] Lowes L, Altoontash A. Modeling reinforced-concrete beam-column joints
[1] Sezen H, Whittaker AS, Elwood KJ, Mosalam KM. Performance of reinforced subjected to cyclic loading. J Struct Eng 2003;129:1686–97.
concrete buildings during the August 17, 1999 Kocaeli, Turkey earthquake, and [35] Vecchio FJ, Collins MP. The modified compression-field theory for reinforced-
seismic design and construction practise in Turkey. Eng Struct concrete elements subjected to shear. J Am Concr I 1986;83:219–31.
2003;25:103–14. [36] Marefat MS, Khanmohammadi M, Bahrani MK, Goli A. Experimental
[2] Dogangun A. Performance of reinforced concrete buildings during the May 1, assessment of reinforced concrete columns with deficient seismic details
2003 Bingol earthquake in Turkey. Eng Struct 2004;26:841–56. under cyclic load. Adv Struct Eng 2006;9:337–47.
J. Shafaei et al. / Engineering Structures 81 (2014) 412–431 431

[37] Mazzoni S, McKenna F, Scott MH, Fenves GL. Opensees command language [42] Karayannis CG, Sirkelis GM. Strengthening and rehabilitation of RC beam-
manual. Berkeley (USA): Pacific Earthquake Engineering Research Center, column joints using carbon-FRP jacketing and epoxy resin injection. Earthq
University of California. <http://opensees.berkeley.edu/>; 2007. Eng Struct D 2008;37:769–90.
[38] ACI Committee 318M-11. Building code requirements for structural concrete [43] CEN Eurocode 8. Design of structures for earthquake resistance-part 3:
(ACI 318m-11) and commentary. Farmington Hills, Michigan, USA: American Assessment and retrofitting of buildings. Brussels; 2005.
Concrete Institute; 2011. [44] Mander JB, Priestley MJN, Park R. Theoretical stress–strain model for confined
[39] ACI Committee 374.1-05. Acceptance criteria for moment frames based on concrete. J Struct Eng-ASCE 1988;114:1804–26.
structural testing and commentary (ACI 374.1-05). Farmington Hills, [45] ASCE 7. Minimum design loads for buildings and other structures. Reston,
Michigan, USA: American Concrete Institute; 2005. Virginia: American Society of Civil Engineers; 2005.
[40] Karayannis CG, Chalioris CE, Sirkelis GM. Local retrofit of exterior RC beam- [46] ATC40. Seismic evaluation and retrofit of concrete buildings. Redwood City,
column joints using thin RC jackets – an experimental study. Earthq Eng Struct California: Applied Technology Council; 1996.
D 2008;37:727–46. [47] FEMA P695. Quantification of building seismic performance factors.
[41] Park YJ, Ang AHS. Mechanistic seismic damage model for reinforced-concrete. J Washington, D.C: Federal Emergency Management Agency, prepared by the
Struct Eng-ASCE 1985;111:722–39. American Society of Civil Engineers (ASCE); 2009.

Anda mungkin juga menyukai