Sei sulla pagina 1di 23

Engineering Structures 88 (2015) 51–73

Contents lists available at ScienceDirect

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

A numerical model for simulation of RC beam-column connections


Morteza Omidi ⇑, Farhad Behnamfar
Dept. of Civil Engineering, Isfahan University of Technology, Isfahan 8415683111, Iran

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

Article history: Beam to column connections in reinforced concrete (RC) frames are among the elements having essential
Received 20 February 2014 effects in determining the performance and behavior of structure under different loads. Role of these con-
Revised 6 January 2015 nections against the lateral loads especially strong earthquakes is such important that experience of past
Accepted 18 January 2015
earthquakes confirm their effect in defining value and extent of structural damages. Therefore, to appro-
priately address the seismic performance of new or existing RC frames, engineers need models able to
predict the connection behavior with an acceptable accuracy and enough simplicity. In this research, a
Keywords:
numerical model for simulating the elastic and inelastic behavior of RC beam-column connections is pre-
RC beam-column connections
Finite Element Model (FEM)
sented. This model consists of a rigid offset element and beam and column elements with concentrated
Numerical model plasticity. The rigid offset element is calibrated to give a good estimation of the initial stiffness based on
Simulation the shear demand ratio of the connection. Each of the beam and column elements with concentrated
Elastic and inelastic behavior plasticity includes two rotational springs in series. One spring represents the nonlinear behavior of beam
and column and the other contains the nonlinear behavior of the connection. Each one of the rotational
springs possesses its own moment-rotation response curve. Comparison of the simulated response with
that of the existing experimental results confirms that the suggested model follows the seismic response
of beam-column connections with a very good accuracy.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction the inelastic behavior of connections and the bending response of


frame members were modeled by two inelastic rotational springs
Past earthquake observations and subsequent laboratory works located at the two end points of an elastic element.
suggest that connections have a key role in reinforced concrete Otani [2] used a bilinear idealization of an envelope curve to
(RC) structures in determining the ductile/nonductile and strength compute the characteristic points of the curve. It was assumed that
degradation behavior of moment frames. Failure of connections bond stresses are constant along the development length of the
can result in a global structural collapse. Practical connection mod- reinforcing bars and that the reinforcing embedment length is long
els are still a prevailing need for engineers to anticipate the enough to develop sufficient steel forces. The fixed end rotation
response of connections and to determine their effects on the gen- was found to be proportional to the square of the moment acting
eral structural behavior. In the last four decades and more, numer- at the beam-column interface. Takeda et al. rule [3] was used as
ous studies have been undertaken on RC connections and several the associated hysteretic rule. The pinching effect due to bond
numerical models have been proposed for simulation of behavior stress-slip and shear sliding were included by the ones suggested
and failure mechanisms of connections of RC moment frames. In by Banon et al. [4]. Several other models have been proposed in
a general view, the mentioned studies can be divided in two cate- the literature; however, they do not generally consider the joint
gories, rotational hinge models, and multi-spring models. The rota- shear behavior, but only the beam and column moment-rotation
tional hinge models generally consist of a bilinear or trilinear capacities (Fillipou et al. [5], Fillipou and Issa [6]).
pushover curve, or a set of formulas for hysteretic behavior of con- Alath and Kunnath [7] presented a zero-length rotational spring
nection under cyclic loads. Such models have been proposed based for simulating the shear deformations of connection. They deter-
on analytical formulations and experimental observations. The first mined and calibrated the envelope of shear stress–shear strain
attempt within this category was reported by Giberson [1], where relation of the above spring based on experimental results. Pamp-
anin et al. [8] determined variation of the principal tensile stresses
⇑ Corresponding author.
in the center of connection against joint’s shear distortion and sug-
gested a moment-rotation relation for the connection springs. A
E-mail addresses: omidi_morteza@yahoo.com (M. Omidi), farhad@cc.iut.ac.ir
(F. Behnamfar). similar model for prediction of shear behavior of exterior joints

http://dx.doi.org/10.1016/j.engstruct.2015.01.025
0141-0296/Ó 2015 Elsevier Ltd. All rights reserved.
52 M. Omidi, F. Behnamfar / Engineering Structures 88 (2015) 51–73

under monotonic loading was proposed by Sharma et al. [9]. In the incorporated it into a macro-model-based finite element analysis
models proposed by Pampanin et al. [8] and Sharma et al. [9], bond in which beams were modeled with fiber elements. The joint model
slip is only indirectly considered and no explicit consideration of consisted of a series of springs to characterize bond-slip behavior
bond slip is provided. under large tension.
The basic idea in the multi-spring models is simulating behavior With reviewing the literature as above, it is observed that var-
of the beam-column joints with isolation of the shares of different ious models have been proposed in the past to predict the inelastic
factors including bonding behavior of the longitudinal bars and response of RC beam-column connections. A number of the sug-
concrete rupture at the beam-column interface. Such models are gested models are relatively simplistic and engineers have to take
based on fundamentals of mechanics and are experimentally cali- important simplified assumptions or possess necessary experi-
brated. In this respect, Biddah and Ghobarah [10] presented a mental data for calibration of the models, e.g., Giberson [1], Alath
model consisting of discrete rotational springs for modeling of and Kunnath [7] and Pampanin et al. [8]. In other models, the con-
shear deformations and bond slip of connections. The shear nection response has been simulated using continuum-type ele-
stress–shear strain relation was assumed to be ideally consisting ments, e.g., Biddah and Ghobarah [10,25], Elmorsi et al. [12],
of a tri-linear curve based on the softening truss theory [11] and Ghobarah and Youssef [13], Altoontash [15], Shin and LaFave
the deformations due to bond slip were simulated using a bilinear [17], Anderson et al. [21] and Fleury et al. [26]. In the latter model,
model referring to previous analytical and experimental data. while there is a lesser need to simplifications and experimental
Elmorsi et al. [12] proposed a model in which beams and columns data, most of the models are neither efficient enough nor suffi-
were introduced using elastic elements and were connected to ciently robust to predict seismic response of larger systems, like
panel zones utilizing nonlinear transitional elements. A more com- multistory buildings. Also, most of the latter models are not easily
plex model introduced by Ghobarah and Youssef [13], benefitted incorporated in the conventional design softwares. Moreover, a
from four hinged rigid elements surrounding the joint. They were large part of the suggested models are suitable only for exterior
connected diagonally by two shear springs to account for shear joints or are not appropriate for analysis of joints with no or low
deformations of joint. Lowes and Altoontash [14] and then Altoon- transverse reinforcement.
tash [15] presented refined and simplified models in the same line Regarding the mentioned limitations of the existing models of
using the modified compression field theory (MCFT) introduced in RC connections, in this research having illuminated by the model
1986 by Vecchio and Collins [16] to define the envelope of the of Birely et al. [27,28], a practical model is presented for elastic
shear stress–shear strain relation of the panel zone. Joints with and inelastic behavior of interior RC connections. Consistency with
no transverse reinforcement, typical in gravity load designed the common commercial design softwares, high speed in modeling,
(GLD) RC frames, were excluded from this study. Shin and LaFave computational efficiency, and having an acceptable accuracy for
[17] used a method similar to Lowes and Altoontashto to deter- both modern and older types of RC joints, are the major advantages
mine the share of the shear deformation of connections in the of the proposed model with respect to the previous works. In the
general behavior of frames. More than 50 experimental shear next sections, the database used, characteristics of the suggested
stress–shear strain curves were used by them for calibration pur- model, and comparison of the responses predicted to those
poses. Jeremic and Bao [18] simplified the model of Lowes and observed in other works are described.
Altoontash for simulating the response of the beam-column con-
nection. They used both ordinary linear and zero-length elements,
and incorporated rotation at the end interfaces of members due to 2. The reference database
bending distortion. Tajiri et al. [19] introduced a new macro ele-
ment for modeling of RC connections for analysis of elasto-plastic 2.1. Description
frames. Their macro element consisted of rigid segments repre-
senting concrete sections that remained plane after deformation To calibrate the proposed model, a database consisting of 23
and of axial springs representative of concrete, bars, slip and shear experimental subassemblages and 167 finite elements (FE) analyt-
deformations. Bao et al. [20] proposed a model simulating seismic ical samples, is populated. The basic experimental (BE) samples
response of interior joints and progressive collapse of RC struc- have been selected such that they cover a wide range of numerical
tures, based on the model by Lowes and Altoontash [14]. They uti- values of prime factors affecting the joint response, including
lized a simplified bilinear relation for the spring simulating the demand-capacity ratios of joint shear and column axial force, com-
reinforcement slip and a symmetric multi-linear moment-rotation pressive strength of concrete, column-to-beam depth ratio, col-
relation, owing its characteristics to the MCFT, for shear springs of umn-to-beam flexural strength ratio, detail of transverse
joint. reinforcement of joint, and longitudinal rebars arrangement of col-
Anderson et al. [21] proposed a cyclic shear stress–strain model umn and beam. Both modern and older details are considered. The
for joints without transverse reinforcement calibrated using mea- BE samples include: the A-M-Z4 sample of Cheok et al. [29], 20
sured data from tests on joints without transverse reinforcement samples consisting of B01–B10, C01, C03, D01–D08 out of B, C,
subjected to a range of displacement histories and joint shear stress and D series of samples tested by Shiohara and Kusahara [30,31],
demands. Kim and LaFave [22] constructed an extensive database of and AL1 and AL2 samples of Li et al. [32].
RC beam-column joint test results including 341 experimental sub- The above subassemblages exclude: slabs or transverse beams,
assemblies having at least a minimum amount of joint transverse eccentric beams, smooth reinforcing bars, column splice failure
reinforcement. In their work, influence of different parameters on modes, and rebar anchoring failure. For simulation of seismic
joint shear stress and/or strain behavior at identified key points loads, all experimental samples have been tested under reversed
was assessed and the strength values were compared with ACI rec- cyclic lateral loading. The analytical part of the database has been
ommendations. Wang et al. [23] proposed a theoretical model for produced in this research, using the nonlinear finite element mod-
the shear strength of RC beam–column joints under seismic loading. eling of extra subassemblages. The analytical samples, being 167 in
The reinforced concrete in the joint panel was idealized as a homog- number, are offshoots of B01, D07, and A-M-Z4 BE samples by
enous material and contribution of the shear reinforcement (includ- varying the joint’s prime affecting factors as described above.
ing both the horizontal and vertical shear reinforcements) was taken Characteristics of the 23 experimental samples selected from
into account using the nominal tensile strength concept of concrete. literature and 167 analytical samples produced in this research
Yu and Tan [24] proposed a component-based joint model and are given in Appendix A.
M. Omidi, F. Behnamfar / Engineering Structures 88 (2015) 51–73 53

2.2. The design parameters The above instructions, give the shear strength of the connec-
tion ðV n Þ as a function of compressive strength of concrete, effec-
The main parameters used to identify the samples of the data- tive horizontal cross section of joint, connection type, and the
base are: (1) shear demand-capacity ratio (SDR or v j;max =v n , volumetric ratio of the confining transverse reinforcement in the
described in Section 2.3) varied between 0.5 and 2.27, (2) column connection. The maximum shear force in the joint ðV j Þ can be cal-
axial force ratio, P=Ag f c , being 0.0–0.5, where P is the applied col- culated using equilibrium of forces acting on the connection
umn axial load, Ag is the gross area of the column, and f c is the con- (Fig. 1) just before failure, as Eq. (1):
crete compressive strength, (3) detail of the connection’s
V j ¼ T pr;A þ T pr;B  V Col ð1Þ
transverse reinforcement, being from completely consistent to
inconsistent (cases I–IV, Section 2.4), (4) column-to-beam flexural where V j is the joint shear force, T pr;A and T pr;B are, respectively, the
P P
strength ratio, M nc = Mnb , that is varied between 0.72 and 2.24, flexural tension force in the beam on A and B sides of the joint, and
where Mn is the nominal bending strength of column (index c) and V Col is the column shear force. The maximum joint shear force
beam (index b) and the sum of bending capacities is taken at the ðV j;max Þ can be calculated using equilibrium of forces acting on the
connection’s center, (5) column-to-beam depth ratio: 0.5–2.0, (6) connection (Eq. (1) and Fig. 1) just before failure.
Beam’s tensile reinforcement ratio: 0.89–3.98%, (6) longitudinal
reinforcing bar distance ratio, which is the distance between cen- 2.4. Detail of the transverse reinforcement
troids of tensile and compressive reinforcements divided by the
height of the cross section: 0.50–0.80. In this study, four different cases including case I (consistent)
Also, the concrete compressive strength of the specimens lies in and cases II–IV (inconsistent) are taken for the transverse rein-
29–32.4 (MPa). Distribution of the above design values between forcement in the connection, as of Table 1.
the database samples is given in Table A5 of Appendix A. In Table 1, s is spacing of the transverse reinforcement in the
plastic hinge zone well as the connection, and hc is the column’s
2.3. The shear force demand ratio (SDR) of the connection depth. According to ASCE41-06 and FEMA356, the spacing of con-
sistent transverse reinforcement of a connection shall be no more
SDR is the ratio of the maximum joint shear force to nominal than hc =3, otherwise the reinforcement is inconsistent.
joint shear strength ðV j;max =V n Þ and calculated using the require-
ments of FEMA356 and ASCE/SEI 41-06 [33,34]. 2.5. Behavioral mechanisms of the connections
Table 1
Different cases for the joint’s transverse reinforcement. According to the observations reported in the literature, the
Cases I II III IV
mechanisms of behavior of the interior RC connections can be cat-
egorized as follows:
Spacing of ties ðsÞ s 6 h3c hc
< s 6 h2c hc
< s 6 hc s > hc
3 2
Rein. condition Ca N.C.b N.C.b N.C.b
(1) Mechanism A: yielding of longitudinal bars of beam at the
a
Conforming. connection resulting in the formation of beam plastic hinge
b
Non-conforming. at the same location.
(2) Mechanism B: yielding of longitudinal bars of column at the
connection resulting in the formation of column plastic
hinge at the same location.
(3) Mechanism C: yielding of transverse reinforcement of the
connection, forming of diagonal tensile cracks, and shear
failure of the joint.
(4) Mechanism D: yielding of transverse reinforcement of beam
and column at the connection, and shear failure of them.
(5) Mechanism E: bond failure of beam bars passing through the
joint.

In this study, bond mechanism is ignored. It has no important


Fig. 1. Joint shear free body diagram.
consequences on the results of this study as described in Section

Fig. 2. The stress–strain diagrams for: (a) concrete, (b) longitudinal bars, (c) transverse bars; of the connection A-M-Z4 [29].
54 M. Omidi, F. Behnamfar / Engineering Structures 88 (2015) 51–73

3.1. As observed in Table A5 in Appendix A, sometimes a combina- 3.2. Material properties


tion of the above mechanisms is responsible for failure of the con-
nection. It should be noted that the FE samples of the database 3.2.1. Concrete
were first developed for the BE samples to evaluate the accuracy The FE model needs the following concrete properties:
of the FE analysis, through comparison of the key factors
qffiffiffiffi
between analysis and test. The factors compared included: load– 0
 The elasticity modulus ðEc Þ, being equal to 4700 f c (MPa), in
displacement curve of the connections, failure moment, pattern 0
of cracking, and failure mode. The comparison showed that an which f c is the concrete compressive strength.
acceptable conformance existed between the two groups of results  The ultimate axial tensile strength, or the rupture modulus, f r ,
qffiffiffiffi
such that the FE results can be used as a basis for connection model 0
where f r ¼ 0:6 f c (MPa).
development along with the BE samples. To see how the FE analy- 0
 Ultimate axial compression strength ðf c Þ.
sis results of this work follow those of the BE’s, the A-M-Z4 sample
 The Poisson’s ratio, m, assumed to be equal to 0.2.
is selected for comparison, as presented in the next section.
 The axial stress–strain relation in compression. In this research
the Hognestad equation is used for the same propose, which is
written as follows [36]:
3. Nonlinear FE analysis of A-M-Z4 [29]
8   2 
Geometrical characteristics and section details of the A-M-Z4 >
< f 00c 2eec  eec ec 6 e0
0 0
connection are shown in Fig. A1. The specimen was pinned at col- fc ¼ h i ð2Þ
>
: f 00 1  0:15ð ec e0 Þ
umn base and roller supported at the beam ends and the column c ecu e0 e0 6 ec 6 ecu
top as shown in Fig. A2.
00
where f c is the maximum compressive stress in concrete calculated
from Eq. (3):
3.1. The FE modeling
00 0
f c ¼ ks f c ð3Þ
The FE model of the connection is developed using ANSYS [35]. 0
Plain concrete is modeled with Solid65. This is an 8-noded element ks is a constant being equal to 1, 0.97, 0.95, 0.93, and 0.92 for f c ’s of
having three translational degrees of freedom (Dof’s) at each node 15, 20, 25, 30, and P35 MPa, respectively. In Eq. (2), the strain e0 is
with the ability to simulate cracking and crushing in concrete. The determined using Eq. (4):
Link8 element is used for modeling of reinforcement. This is a two-
node truss-type element having three translational Dof’s at each
node. Steel plate modeling is carried out utilizing Solid45 element 300
being an 8-noded element with three transitional DOF’s at each
250
node. Ideally, the bond strength between the concrete and steel
reinforcement should be considered. However, in this study, per-
Load (kN)

200
fect bond between the materials was assumed. To provide the per-
fect bond, the link element for the steel reinforcing was connected 150 A
to the corresponding nodes of adjacent concrete solid elements, so
B
that the two materials shared the same nodes. While it is possible 100
that in maximum tension areas a complete bond does not exist C
between concrete and reinforcement bars, use of the above method 50
D
in an FEM modeling proves to be satisfying in this research and will
be shown to result in a model well simulating the studied experi- 0
mental samples in initial stiffness and ductility. 0 10 20 30 40 50 60
Similar to the actual test, an axial force equal to 10% of the gross Displacement (mm)
compression capacity of column (0.1Ag f c Þ is applied to the upper
Fig. 4. The load–displacement curve of A-M-Z4. The points A, B, C, and D correspond
node of column as a constant load. Despite the cyclic test, the lat- to initiation of cracks in concrete, yielding in tensile reinforcements of beam,
eral load is applied to the analytical model monotonically to trace yielding in compressive reinforcements of beam, and the peak loads of the
the envelope behavior. connection, respectively.

1 ANSYS EXPERIMENTAL
ELEMENTS AUG 2 2010
09:08:30

300

250
Lateral load (kN)

200

150

100
Y

Z X
50

0
0 10 20 30 40 50 60
Displacement (mm)

Fig. 3. The FE model and load–displacement curve of the A-M-Z4 sample [29].
M. Omidi, F. Behnamfar / Engineering Structures 88 (2015) 51–73 55

 00 
f ments are taken to be 500 and 510 MPa, respectively. The Poisson’s
e0 ¼ 1:8 c ð4Þ
Ec ratio of steel is assumed to be 0.3.

The concrete compressive strength in test was reported to be 3.3. Comparison of the FE and experimental results
30.7 MPa. Using this value, the concrete stress–strain curve is illus-
trated as Fig. 2a. The corresponding values of the elasticity and rup- According to the available test results, in this section the load–
ture modulus are 26.04 GPa and 3.88 MPa, respectively. displacement curves, failure moment and cracking patterns are
compared as follows.

3.2.2. Steel reinforcement 3.3.1. The load–displacement curve


Based on the reference experimental studies, the stress–strain The analytical and experimental load–displacement curves in
relation of the longitudinal and transverse rebars are introduced the first quadrant are shown in Fig. 3. The analysis seems to have
as multi-linear curves shown in Fig. 2b and c. According to the test, been successful in following the envelope of the experimental
the yield strength of the longitudinal and transverse reinforce- results. The peak loads obtained for the samples are 279 and

700
700
600
600
500 500
Stress (MPa)

Stress (MPa)
400 400 A
A
300 300 B
B
200 200
C C
100 100
D D
0 0
0.00 0.01 0.02 0.03 0.04 0.05 0.000 0.002 0.004 0.006 0.008 0.010
Strain Strain

(a) (b)
35 600

30 500
25
Stress (MPa)

Stress (MPa)

400
20 A
300 A
15 B
200 B
10
C C
5 100
D D
0 0
0 0.005 0.01 0.015 0.02 0 0.0005 0.001 0.0015 0.002 0.0025 0.003
Strain Strain
(c) (d)
600

500
Stress (MPa)

400
A
300
B
200
C
100
D
0
0 0.0005 0.001 0.0015 0.002 0.0025 0.003
Strain
(e)
Fig. 5. The stress–strain curves of the critical elements for: (a) the tensile longitudinal rebars of beam, (b) the compressive longitudinal rebars of beam, (c) concrete in
compression, (d) transverse rebars of the connection, and, (e) transverse rebars of beam. The points A, B, C, and D correspond to initiation of cracks in concrete, yielding in
tensile reinforcements of beam, yielding in compressive reinforcements of beam, and the peak loads of the connection, respectively.
56 M. Omidi, F. Behnamfar / Engineering Structures 88 (2015) 51–73

282 kN with the associated displacements of 49 and 51 mm, from Table 2


the analysis and the test, respectively. The offset length coefficient proposed by Birely et al. [27,28].

b All ACI compliant ACI non-compliant


Proposed optimal 0.4 0.6 0.4
3.3.2. The failure pattern of the connection
In this section the failure pattern of the AM-Z-4 sample is inves-
tigated using an FEM model and the analysis results are compared tensile and compressive steel elements of the longitudinal rebars,
with those of the experiments. critical compressive concrete element, and critical elements of
Fig. 4 illustrates the analytical load–displacement curve of the the transverse reinforcements of the panel zone and beam, are
A-M-Z4 assemblage. Also, the stress–strain curve of the critical shown in Fig. 5. The critical elements of the beam are located at

Fig. 6. Strain distribution just before failure: (a) in the connection A-M-Z4, (b) in the concrete elements of beam at the column face.

Fig. 7. Stress distribution before collapse in: (a) concrete elements of A-M-Z4, (b) longitudinal reinforcement, (c) cracking of connection just before failure, (d) the test
assemblage view at the failure [29].
M. Omidi, F. Behnamfar / Engineering Structures 88 (2015) 51–73 57

Table 3
Error associated with the rigid offset model of Birely et al. [27,28].

Model All specimens ACI compliant ACI non-compliant


Avg. Std. dev. Avg. Std. dev. Avg. Std. dev.
FEMA356a 0.53 0.07 0.50 0.08 0.54 0.07
ASCE/SEI 41-06b 0.04 0.14 0.04 0.17 0.06 0.13
Proposed optimalb 0.02 0.15 0.01 0.17 0.02 0.14
Fully rigid joint ðb ¼ 1Þb 0.24 0.12 0.17 0.14 0.25 0.12
Centerline ðb ¼ 0Þb 0.19 0.17 0.28 0.21 0.16 0.16
a
Beam and column effective stiffnesses per FEMA356.
b
Beam and column effective stiffnesses per ASCE/SEI Standard 41-06.

failure, the longitudinal bars and the compressive concrete of beam


are well in the nonlinear region while the transverse reinforce-
ments of beam and connection still behave elastically.
For concrete (Fig. 5c), the stress remains almost constant
between a strain value of 0.005–0.015, whereas the input Hognes-
tad stress–strain curve suggests that the ultimate strain should be
taken as 0.0038. This happens because the FE technique used is
based on the smeared crack and not the discrete crack approach.
As shown in Fig. 6, checking of the tensile and compressive stress
contours of the concrete elements just before failure shows that
only a negligible part of the concrete volume, corresponding to
the concrete cover of the compressive rebars of beam at the con-
nection, posses strains excessive to 0.0038. Therefore, this should
have only a marginal effect on the credibility of results.
In Fig. 7, distribution of stress in concrete and reinforcement
elements are shown at the verge of failure. It is observed that
the longitudinal rebars of beam have yielded at the connection
and the concrete elements of tensile and compressive zones adja-
Fig. 8. Nonlinear model of RC joints suggested by Birely et al. [27,28]. cent to the connection have attained their maximum capacity
values. The above facts show that a plastic hinge forms in the
beam at the connection. The elements of the ties of the connec-
tion have also yielded while the longitudinal rebars of column
the side of the connection. The mentioned diagrams actually dis- remain in the elastic region, except in the connection. Based on
play the stress–strain variation of these elements during the anal- the experimental results, formation of the plastic hinge in beam
ysis process up to application of the peak lateral load. was responsible for failure. Moreover, shear cracks appeared in
The common benchmark points of the curves are labeled with A, the connection. It is interesting to note that similarly, in the
B, C, and D. These points correspond to initiation of cracks in con- nonlinear FEM analysis first bending cracks appear in the
crete, yielding in tensile reinforcements of beam, yielding in com- tensile region of beam and then shear cracks develop in the con-
pressive reinforcements of beam, and the peak loads of the nection. With increase of load, cracks extend to the mid span of
connection, respectively. Fig. 5 clearly shows that at connection beam.

Fig. 9. Dual-hinge components: (a) beam spring, and (b) joint spring [27,28].
58 M. Omidi, F. Behnamfar / Engineering Structures 88 (2015) 51–73

Fig. 7c shows the crack pattern of the connection in the last step the recommendations of FEMA356 and ASCE41-06 using the 26-
before failure. specimen dataset. The error was computed as the difference
Fig. 7d certifies the FE analysis results as it shows formation of a between the experimental and simulated values, normalized by
plastic hinge in the beams at the connection and almost no damage the experimental values.
in the columns out of the connection zone. For developing the nonlinear model, Birely et al. utilized a beam
element with concentrated plasticity. In their model the connec-
tion zone was assumed to be completely rigid ðb ¼ 1Þ and the plas-
4. A review on the Birely’s discrete connection model
tic hinge of beam at the connection composed of two nonlinear
hinges in series. One of the hinges accounted for bending response
Since the model proposed in this research is an extension and
of beam and the second one represented the nonlinear behavior of
modification of that suggested by Birely et al. [27,28] for discrete
joint. The moment-rotation characteristics of the beam bending
modeling of RC connections, a view on basic components of Birely’s
hinge was determined by moment–curvature analysis of beam.
model is in perspective.
That of the joint hinge was proposed to follow a bilinear path with
Birely et al. proposed a linear and a nonlinear model for simu-
the stiffness of each part being a function of the joint’s geometry
lating lateral behavior of interior RC connections using a database
and the shear strength of concrete. Fig. 8 illustrates the nonlinear
of 45 specimens. In the linear model, a modified offset length was
connection model of Birely et al., and Fig. 9 shows the moment–
proposed for better estimation of the initial stiffness of connection
rotation curves of each nonlinear spring.
and displacements corresponding to yielding of beam. The offset
For calculating the upper limits of rotations of beam and joint
length, or the length of a short rigid beam inserted between the
hinges, Birely et al. divided the dataset connections to three cate-
span beam and the connection’s central node, was determined as
gories of ductile joints meaning connections with displacement
a coefficient ðbÞ of connection’s dimension. It was observed that
ductilities ðlD Þ not less than 4, intermediate or limited ductility
in some cases the models with no rigid offset ðb ¼ 0Þ, were stiffer
joints with 1 6 lD < 4, and brittle joints with lD < 1, and pre-
than the experimental assemblage. To decrease the associated
inaccuracy, the latter specimens were removed from the database sented the evaluation results accordingly. It was shown that the
when calibrating the linear model, resulting in a reduced database proposed nonlinear model resulted in the least average errors for
of 26 samples. The coefficient b was presented as of Table 2 for the load and displacement corresponding to the point with the
three categories of connections, including all joints in the dataset, maximum strength; being 1% and 4%, respectively. For ductile
ACI compliant joints, and ACI non-compliant joints in the dataset. and intermediate joints, while the average errors in estimation of
Table 3 shows the average and standard deviation of error in the maximum strengths were acceptably small (8% and 12%,
estimation of the initial stiffness using coefficient b proposed by respectively), those for the associated maximum displacement
Birely et al. [27,28] and those of the models developed based on were relatively large (61% and 20%, respectively).

Fig. 10. The end rigid zones for modeling of beam-column connection. (a) Fully flexible connection; (b) fully rigid connection; (c) connection with a rigid offset length equal
to a factor b of beam and column depths.
M. Omidi, F. Behnamfar / Engineering Structures 88 (2015) 51–73 59

In this research, motivated by the Birely et al.’s model, a more under service loads on member’s stiffness. The connection zone is
practical model is presented for simulation of elastic and inelastic usually assumed to be rigid and this is simulated with defining
lateral behavior of the interior RC joints. Use of a larger database rigid offsets at the beam and column end parts at the connection.
with a wider range of design parameters, using a more efficient cal- In this research, the rigid offset length recommended based on
ibration procedure, accounting for nonlinear behavior of column, the results of the reference database, is calibrated and evaluated.
ability to simulate elastic as well as post-yield behavior of connec- First, the recommendations of ASCE/SEI 41-06 and FEMA356 are
tion in a unified analytical model, considerable increase of ability assessed. Regarding Fig. 10b, FEMA356 recommends to assume the
and accuracy of model in simulation of elastic and inelastic behav- total connection as a rigid zone and defines an effective bending
ior of interior RC connections, are among the advantages of the stiffness for the frame element as a function of the axial force to
proposed model with regard to that of Birely et al. be equal to ð0:5  0:7ÞEIg (E = modulus of elasticity of con-
crete,Ig = the gross moment of inertia of member) [33]. In contrast,
5. The proposed model ASCE/SEI 41-06 introduces a smaller stiffness, equal to
ð0:3  0:7ÞEIg depending on the axial load. This reference deter-
The model suggested in this research, consists of a rigid offset mines the rigid offset length based on the ratio of bending capacity
length for the linear model and of the concentrated plasticity beam of column to beam to be: (a) equal to full depth of the column
P P
and column elements for the nonlinear model. The linear model is when M = M < 0:8, (b) equal to full depth of the beam
P nc P nb
meant to estimate the initial stiffness of the connection and the when M nc = Mnb > 1:2, and (c) average of beam and column
P P
nonlinear model is to simulate the connection’s stiffness due to depths when 0:8 < M nc = Mnb < 1:2.
yielding of beam and column’s reinforcement and strength reduc- In this model, according to Fig. 10c, the offset length was deter-
tion of connection after it was damaged due to shear failure. mined as a factor b times the joint’s dimension considering the
Evaluation of past experimental data shows that various design effective stiffness coefficients of ASCE41-06 for beams and col-
parameters affect the seismic behavior of RC joints. In this research umns. The b factor varies with the SDR of connection. For this pur-
the shear demand ratio of connection ðSDR ¼ v j;max =v n Þ, where v j is pose, first for each of the samples of the database (including 23
the maximum joint shear force and v n is the nominal joint shear experimental and 167 analytical samples introduced in Section
strength, is considered to be a prime factor and will be utilized 2.1.), the initial stiffnesses of the linear model,ðksim Þi , and the
to categorize the calibration coefficients of the proposed linear experimental or FE samples ðkExp=FEM Þi are calculated as the secant
and nonlinear models of connection. On this basis, the connections stiffnesses for a point on the associated load–displacement curves
existing in the database are divided into three groups of those with corresponding to a displacement of 0.75Dy , where Dy is the yield
SDR 6 1.2, connections with SDR P 1.5, and the connections in displacement. Then Eq. (5) is used to calculate the error, or better:
between. The parameters required for each part of the numerical relative difference, of the initial stiffness estimated by the pro-
model will be calibrated on the same basis. Characteristics of both posed model, ei , as:
linear and nonlinear models are described comprehensively in the ðkExp=FEM Þi  ðksim Þi
following sections. ei ¼ ð5Þ
ðkExp=FEM Þi

5.1. The linear model

The rigid offset model that is simply able to be implemented in


professional softwares, is an effective tool for simulation of the
connection’s flexibility in the linear elastic analyses of the RC
frames. In linear modeling of a frame, the shear, bending, and tor-
sional stiffnesses of the frame elements are determined using
member dimensions and material characteristics. For RC elements
it is customary to use the effective stiffnesses calculated by reduc-
ing the member’s gross stiffness to simulate the effect of cracking

Table 4
The offset rigid length factor ðbÞ calibrated using the reference database.

b v j;max =v n
6 1:2 > 1:2& < 1:5 P 1:5
Proposed optimal 0.38 0.43 0.50
Fig. 11. Details of the numerical model.

Table 5
Evaluation of the initial stiffness error of the offset rigid length models based on the reference database.

Model v j;max =v n
6 1:2 > 1:2& < 1:5 P 1:5
Avg. Std. dev. Avg. Std. dev. Avg. Std. dev.
FEMA 356a 0.22 0.070 0.20 0.080 0.19 0.090
ASCE/SEI 41-06b 0.028 0.16 0.01 0.15 0.033 0.14
Proposed optimalb 0.014 0.054 0.015 0.071 0.019 0.116
Fully rigid joint ðb ¼ 1Þb 0.14 0.088 0.10 0.096 0.080 0.087
Centerline ðb ¼ 0Þb 0.13 0.17 0.16 0.18 0.17 0.21
a,b
The effective stiffnesses of beam and column are calculated based on FEMA356 and ASCE/SEI 41-06, respectively.
60 M. Omidi, F. Behnamfar / Engineering Structures 88 (2015) 51–73

The coefficient b of the rigid offset length is calculated as the value 5.2. The nonlinear model
minimizing sum of the squares of the errors in initial stiffness. The
calculated values of b for each range of SDR values are shown in Although the suggested rigid offset model can acceptably pre-
Table 4. dict the initial stiffness of the connection, this model is not able
Assessment of the experimental and FE analysis results showed to follow the stiffness reduction due to beam and column yielding
that in the studied connections the transverse reinforcement con- and strength reduction due to damage of the connection in shear.
dition (consistent or non-consistent according to Table 1) does not In this section, to simplify the nonlinear modeling of the response
have a considerable effect on elastic behavior and initial stiffness of of RC connections using commercial software packages, a model is
the connections. Therefore, it is not considered in the developed presented in which the moment-rotation response of the concen-
linear model. Also, the same results show that the initial stiffness trated plastic hinges of the beam and column elements are modi-
of connection increases with the increase of SDR, a fact in very fied such that it simulates the flexibility and strength reduction
good agreement with Table 4. due to the damage of connection.
In Table 5, the average and standard deviation of error of the The suggested model is presented in Fig. 11, including beams
initial stiffness estimated by the reference database based on the and columns with concentrated plastic hinges. The effective stiff-
suggested b factors and the calculated errors on the basis of mod- nesses of the beams and columns out of the plastic hinges are cal-
eling recommendations of FEMA356 and ASCE/SEI 41-06 are pre- culated based on ASCE/SEI 41-06. The rigid offset length is
sented. In addition, the error is mentioned for two cases, determined as a function of the shear demand ratio of the connec-
including the fully flexible (Fig. 10)a and the fully rigid (Fig. 10b) tion using the values suggested in Table 4. In this model each of the
connections applying the effective stiffness recommendations of beam and column elements include two rotational springs placed
ASCE/SEI 41-06. in series at the plastic hinge location. One spring represents the
The numerical values of Table 5 show that applying the model- nonlinear behavior of the beam or column and the other one sim-
ing recommendations of FEMA356 results in a considerable ulates that of the connection. Each spring possesses its own
increase of the initial stiffness of connections with any shear moment-rotation, to be discussed in the following.
demand. On the other hand, the modeling requirements of ASCE/
SEI 41-06 make an underestimation of the same quantity when 5.2.1. The moment-rotation response of the nonlinear springs of the
the shear demand ratio is less than 1.5 and overestimates it for lar- beam and column
ger ratios, but is considerably more accurate than FEMA356 in both The nonlinear moment-rotation curves of beam and column are
cases. Also, it is observed that for a rigid connection, the recom- respectively determined with analysis of moment–curvature of
mendations of ASCE/SEI 41-06 do better than FEMA356 in estimat-
ing the initial stiffness.
Table 5 represents the fact that the rigid offset length suggested
in this study, somewhat enhance the accuracy of the initial stiff-
ness compared to the suggested model of ASCE/SEI 41-06. For
the connections with shear demand ratios equal to or less than
1.2 and equal to and larger than 1.5, the average error decreases
respectively from 2.8% and 3.3% of ASCE/SEI 41-06 to 1.4% and
1.9% of the suggested model.
According to Table 3, in the linear model of Birely et al., the
average and standard deviation for the ACI-consistent samples
are 0.01 and 0.17, and for the ACI-non consistent samples are
0.02 and 0.14, respectively. With the much more extensive data-
base used in this research (23 experimental and 167 FE numerical
samples compared with 26 experimental samples of Birely et al.),
the error of the proposed linear model ranges between 0.054 to Fig. 13. Components of shear deformation of connection. (a) The beam-column
0.116 in comparison to 0.14 to 0.17 of the Birely et al.’s model. panel zone and directions of tension and compression struts in which under lateral
forces, (b) the panel zone before deformation, (c) the panel zone after deformation
Therefore the proposed model can estimate the initial stiffness of
and its average shear strain.
the connection with higher accuracy.

Fig. 12. The moment-rotation response of the beam and column’s nonlinear Fig. 14. The moment-rotation response of the panel zone’s nonlinear spring.
springs.
M. Omidi, F. Behnamfar / Engineering Structures 88 (2015) 51–73 61

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
beam and column sections. First the beam and column’s moment– 2
a2 þ b
curvature responses are calculated and then with the use of flexi- c¼ ðDd1 þ Dd2 Þ ð6Þ
2ab
ble length of beam, (LBeam;flexible ¼ ðLb  bhc Þ=2Þ, and column,
where c is the average joint’s shear strain. Other parameters are
(LColumn;flexible ¼ ðLc  bhb Þ=2Þ, the above responses are transformed
shown in Fig. 13.
to moment-rotation curves. In the post-yield range, the moment–
Then using the equilibrium and deformation consistency equa-
curvature response is converted to the moment-rotation curve by
tions, the nonlinear moment-rotation response of the joint’s repre-
multiplying curvature by the length of the plastic hinge. Length
sentative spring is determined as follows [27,38]:
of the plastic hinge is taken to be equal to half of the beam depth
according to Corley [37]. The general model for the moment- Mj ¼ X v :V j ð7Þ
rotation behavior of the members is introduced as a tri-linear rela- hj ¼ X c :c ð8Þ
tion as of Fig. 12 including the cracking, yielding, and failure points.
where V j and c are the shear strength and the shear strain of the
panel zone, respectively, and the X c and X v factors are determined
5.2.2. The moment-rotation response of the nonlinear spring of the
as follows using the geometry of the frame:
panel zone
   
For determining the nonlinear moment-rotation response of the hc jd hc
panel zone, first the shear force-shear strain curves of the connec- Xv ¼ 1  jd 2 1   ð9Þ
Lb Lc Lb
tions of the dataset are calculated using the FE analysis. Such hb
curves are depicted utilizing the force–displacement relation of Xc ¼ 1  ð10Þ
Lc ð1  hc =Lb Þ
the assemblage. For the latter purpose, equilibrium of forces acting
on the free body diagram of connection (Fig. 1) is considered and in which hb and hc are beam and column’s depths, respectively, and
having the shear force of column at each load step, the connection’s jd is the moment arm of the beam at critical section. These param-
shear force, Vj, is calculated using Eq. (1). The shear strain of con- eters are shown in Fig. 11.
nection due to diagonal tensile and compressive deformations of After computing the moment-rotation response of the nonlin-
the panel zone (Fig. 13) is calculated using Eq. (6): ear springs of the panel zones available in the reference database,

Vj/Vn≤1.2 1.2<Vj/Vn<1.5
0.6 0.6

0.5 0.5

0.4 0.4
I I
P/Agf'c
P/Agf'c

0.3 0.3
II II
0.2 0.2
III III
0.1 IV 0.1 IV

0.0 0.0
0.000 0.005 0.010 0.015 0.020 0.025 0.030 0 0.005 0.01 0.015 0.02 0.025 0.03
Average value of maximum joint shear Average value of maximum joint
strain (rad) shear strain (rad)
(a) (b)

Vj/Vn≥1.5
0.6

0.5

0.4
P/Agf'c

0.3 I
II
0.2
III
0.1 IV

0.0
0.000 0.005 0.010 0.015 0.020 0.025 0.030
Average value of maximum joint shear
strain (rad)
(c)
Fig. 15. Variation of the average values of maximum shear strain capacity of connection under different axial load ratios and conditions of ties for connections with:
(a) v j =v n 6 1:2, (b) 1:2 < v j =v n < 1:5, (c) v j =v n P 1:5. Cases I–IV are defined in Table 1.
62 M. Omidi, F. Behnamfar / Engineering Structures 88 (2015) 51–73

Table 6 Simulated Experimental


Suggested values for the maximum joint shear strain ðcmax Þ.
300
P
Ag f c
Trans. reinf. v j;max =v n
6 1:2 > 1:2& < 1:5 P 1:5 250

Lateral load (kN)


a
6 0:1 I: C 0.025 0.022 0.014
P 0:4 s 6 h3c 0.020 0.017 0.011 200
6 0:1 II: N:C:b 0.022 0.017 0.013
P 0:4 hc
< s 6 h2c 0.016 0.014 0.010 150
3
6 0:1 III: N:C:b 0.020 0.015 0.010
P 0:4 hc
< s 6 hc 0.013 0.009 0.007 100
2
6 0:1 IV: N.C.b 0.016 0.008 0.007
P 0:4 s > hc 0.010 0.006 0.004 50
a
A-M-Z-4
Conforming.
b
Non-conforming. 0
0 10 20 30 40 50 60
Displacement (mm)

Table 7 Fig. 16. The experimental and simulated responses for the A-M-Z-4 sample.
Suggested values for the factor relating the moment of the panel zone to the beam
moments at cracking, yielding, and failure pointsðab;i Þ. where ab;i and ac;i are the calibrated modification factors for the
ab;i v j;max =v n moment of the panel zone’s nonlinear spring at the beam and
column.
6 1:2 > 1:2& < 1:5 P 1:5
Rotation of this spring at the failure point is determined based
ab;cracking 1.237 0.981 0.686 on Eq. (12) as a function of shear strain associated with the maxi-
ab;yield 1.045 0.998 0.723
mum shear strength of the panel zone ðcmax Þ and system geometry.
ab;ultimate 0.922 0.885 0.619
Then the rotation corresponding to the cracking and yield points
are introduced as factors ki of the rotation at the failure point
(Eq. (13)), as follows:
Table 8 hu ¼ X c :cmax ð12Þ
Suggested values for the factor relating the moment of the panel zone to the column
moments at cracking, yielding, and failure pointsðac;i Þ. hj;i ¼ ki :hu ði ¼ cracking and yield pointsÞ ð13Þ
ac;i v j;max =v n The panel zone spring acts only up to the peak load. To make a bet-
6 1:2 > 1:2& < 1:5 P 1:5 ter estimation of displacement ductility of the connection, the max-
imum rotations are limited to certain values for each of the
ac;cracking 0.983 0.749 0.726
ac;yield 0.928 0.789 0.785 nonlinear springs of the beam, column, and panel zone. These limit
ac;ultimate 0.839 0.685 0.668 rotations are calibrated as factors k of the rotation corresponding to
the failure point, as in Eq. (14):
hfail ¼ k:hultimate ð14Þ

Table 9 5.2.3. The proposed values for maximum shear strain of the panel zone
Suggested values for the factors relating ultimate to cracking and yield rotations (ki Þ.
ðcmax Þ
ki v j;max =v n Investigation and evaluation of the shear force-shear strain
6 1:2 > 1:2 & < 1:5 P 1:5 curves of the connections of the dataset shows that:
kcracking 0.0314 0.0433 0.0471
kyield 0.3622 0.3841 0.5957
(1) For any conditions of transverse rebars of connection (con-
sistent or non-consistent), increase of the axial load ratio
or the shear demand ratio results in reduction of the maxi-
this response is idealized again as a tri-linear curve as in Fig. 14.
mum shear strain capacity (Fig. 15).
The moment-rotation response of this spring is calibrated such
(2) A non-consistent transverse reinforcement (s > hc =3Þ has a
that in which the moment of the panel zone at the beam and col-
detrimental effect on the maximum shear strain capacity
umn elements are calculated at cracking, yielding, and failure
and considerably decreases its value (Fig. 15). With the
points as factors ai of the corresponding beam and column
change of the ties condition from I to IV, the failure of con-
moments at the same points, i.e.:
nections having an SDR > 1 changes from a ductile (often
M jb;i ¼ ab;i :M b;i and M jc;i ¼ ac;i :M c;i yield of beam in bending) to a brittle mechanism due to
ði ¼ cracking; yield and ultimate pointsÞ ð11Þ shear failure of the joint or sometimes the beam (Table A5).

Fig. 15 shows the pattern of variation of the average values of


Table 10 maximum shear strain capacity under different axial load ratios
Suggested values for the factor relating ultimate and failure rotations ðkÞ. and various conditions of transverse reinforcement for certain val-
k v j;max =v n ues of the shear demand ratio.
The above behavior mechanisms and the results presented in
6 1:2 1:2 > & < 1:5 P 1:5
Fig. 15 are considered in determining the maximum shear strain
kbeam 3.50 2.54 2.21 capacity of connection. In addition, design parameters affecting
kcolumn 3.23 2.75 2.08
behavior of the connections such as the axial load ratio ðP=Ag f c Þ,
kjoint 2.81 2.72 2.57
the shear demand ratio (v j;max =v n Þ, and condition of the transverse
M. Omidi, F. Behnamfar / Engineering Structures 88 (2015) 51–73 63

Simulated Experimental Simulated Experimental


100
70 90

60 80

Lateral load (kN)


Lateral load (kN)
70
50
60
40
50
30 40
30
20
20
10
B01 10 B02
0 0
0 10 20 30 40 0 10 20 30 40
Displacement (mm) Displacement (mm)

Simulated Experimental Simulated Experimental

140 70

120 60
Lateral load (kN)

Lateral load (kN)


100 50

80 40

60 30

40 20

20 10
B03 B04
0 0
0 10 20 30 40 0 10 20 30 40
Displacement (mm) Displacement (mm)

Simulated Experimental Simulated Experimental


100 100
90 90
80 80
Lateral load (kN)

Lateral load (kN)

70 70
60 60
50 50
40 40
30 30
20 20
10 B05 10 B06
0 0
0 10 20 30 40 0 10 20 30 40
Displacement (mm) Displacement (mm)

Fig. 17. The experimental and simulated responses for the B series samples.

rebars, are accounted for. Table 6 displays the values suggested for ing to Tables 7–9. The values have been optimized for estimation of
the maximum shear strain capacity of connection ðcmax Þ corre- stiffness of the plastic hinge (a combination of nonlinear springs of
sponding to the point of initiation of shear strength degradation. the beam, column, and joint in series) at cracking, yielding, and
These values are in fact means of the maximum shear strains for failure in shear. This is implemented to ensure that the model
the connections existing in the studied dataset. For intermediate can accurately predict the pre-yield (cracking) and post-yield
values, a linear interpolation can be used. behavior (yielding of beam and column and strength degradation
due to damage of connection) of the response.
6. Calibration of the numerical model As seen in Tables 7 and 8, increase of shear demand ratio of the
connection reduces the associated moment of the connection at
The coefficients ab;i ; ac;i , and ki , mentioned in Section 5.2.2, are different stages compared with the values corresponding to smal-
calculated using the dataset for different values of v j;max =v n accord- ler shear demands, as expected. In the same line, according to
64 M. Omidi, F. Behnamfar / Engineering Structures 88 (2015) 51–73

Simulated Experimental Simulated Experimental

70 70

60 60

Lateral load (kN)


50 50
Lateral load (kN)

40 40

30 30

20 20

10 10
B07 B08
0 0
0 10 20 30 40 0 10 20 30 40
Displacement (mm) Displacement (mm)

Simulated Experimental Simulated Experimental

140 140

120 120
Lateral load (kN)

Lateral load (kN)


100 100

80 80

60 60

40 40

20 20
B09 B10
0 0
0 10 20 30 40 0 10 20 30 40
Displacement (mm) Displacement (mm)

Fig. 17 (continued)

Simulated Experimental Simulated Experimental


100 100
90 90
80 80
Lateral load (kN)

70 70
Lateral load (kN)

60 60

50 50

40 40

30 30

20 20

10 10
C01 C03
0 0
0 10 20 30 40 50 0 10 20 30 40 50

Displacement (mm) Displacement (mm)

Fig. 18. The experimental and simulated responses for the C series samples.

Table 9, lambda values increases for larger shear demands meaning connections. Based on the experimental and FE analysis results,
that the rotation ductility of the connection considerably decreases with increase of the SDR the failure mechanism of connection
for large shear demands as is again expected. changes from a ductile condition due to the yield of beam and
The k factor is also presented in Table 10 for each of the non- sometimes column in bending to a brittle condition because of
linear springs of beam, column, and panel zone according to SDR the shear failure of connection or occasionally beam, resulting
of connection. The purpose of this factor is better representation in an overall reduction of ductility (Table A5). As observed in
of the displacement ductility of connection with minimizing the Table 10, increase of SDR results in decrease of the k factor for
average error in estimation of the displacement corresponding all of the nonlinear hinges, in agreement with the above
to the starting point of strength degradation of the dataset reasoning.
M. Omidi, F. Behnamfar / Engineering Structures 88 (2015) 51–73 65

Simulated Experimental
60 Simulated Experimental
60

50
50
Lateral load (kN)

Lateral load (kN)


40 40

30 30

20 20

10 10
D01 D02
0 0
0 10 20 30 40 50 0 10 20 30 40 50
Displacement (mm) Dispalcement (mm)

Simulated Experimental Simulated Experimental

60 60

50 50
Lateral load (kN)

Lateral load (kN)


40 40

30 30

20 20

10 10
D03 D04
0 0
0 10 20 30 40 50 0 10 20 30 40 50
Displacement (mm) Displacement (mm)

Simulated Experimental Simulated Experimental


80
90
70
80
60 70
Lateral load (kN)

Lateral load (kN)

50 60
50
40
40
30
30
20
20
10 10
D05 D06
0 0
0 10 20 30 40 50 0 10 20 30 40 50
Displacement (mm) Displacement (mm)

Fig. 19. The experimental and simulated responses for the D series samples.

7. Implementation of the proposed numerical model and (2) Assigning the boundary conditions.
comparison with the test results (3) Introducing the effective stiffness coefficients of member
sections regarding the axial load ratio according to, e.g.,
7.1. Implementation ASCE41-06.
(4) Assigning the rigid offset length to the end portions
The different steps of introducing the proposed model to a sam- of the beam and column members, based on
ple analysis software can be summarized as follows: Table 4.
(5) Introducing the nonlinear hinges of beam, column and
(1) Introduction of the dimensions of connection, beam and joint, around the connection, consisting at each connection
column sections, and the material properties of concrete and side of a rotational spring for the connecting member in
steel. series with another spring for the joint (Fig. 11). The
66 M. Omidi, F. Behnamfar / Engineering Structures 88 (2015) 51–73

Simulated Experimental Simulated Experimental


120
90
80 100

70
Lateral load (kN)

80

Lateral load (kN)


60
50 60
40
30 40

20
20
10
D07 D08
0 0
0 10 20 30 40 50 0 10 20 30 40 50
Displacement (mm) Displacement (mm)

Fig. 19 (continued)

Simulated Experimental Simulated Experimetal


80 100

70 90
80
60
Lateral load (kN)

70
Lateral load (kN)

50
60
40 50

30 40
30
20
20
10 10
AL1 AL2
0 0
0 20 40 60 80 100 0 20 40 60 80 100
Displacement (mm) Displacement (mm)

Fig. 20. The experimental and simulated responses for the AL1 and AL2 samples.

Table 11
The difference percentage of the numerical model relative to the experimental results in estimating the displacement and force for the point of maximum strength.

Key points Difference v j;max =v n


6 1:2 > 1:2 & < 1:5 P 1:5
Peak disp. Avg. 1.65 2.76 4.73
Std. dev. 8.65 3.16 8.90
Peak load Avg. 0.82 0.84 1.63
Std. dev. 3.65 2.88 3.54

moment-rotation properties of these springs are defined as (1) In A-M-Z-4 connection it is observed that the beam springs
of Sections 5.2.1 and 5.2.2. rotations have reached their final values ðhfail Þ, while springs
of column and joint have remained in the elastic region.
7.2. Comparison with the test results Therefore, the numerical model recognizes the beam plastic
hinges as being responsible for failure of joint.
To evaluate the proposed numerical connection model, the (2) In all of the B-series connections, both beam and joint
dataset introduced in Section 2 consisting of 23 experimental sam- springs rotations entered the nonlinear region. A similar
ples is utilized. For this purpose, the pushover analysis of the condition exists for column springs too, except B04 and
experimental samples is carried out using the proposed connection B06. With the latter exception, in all of the B-series connec-
model in SAP2000 [39]. Figs. 16–20 show results of the pushover tions the joint spring rotations have reached their ultimate
analysis along with the cyclic behavior of the dataset samples in value (hfail Þ. In B04 and B06 samples the beam springs have
the first quadrant. reached to their final rotation capacities.
For each group of the experimental samples, results of the anal- (3) In C series connections all of the springs (beam, column, and
ysis with the proposed numerical model can be summarized as joint) enter the nonlinear region. In C01 and C03 the joint and
follows: column spring rotations have reached to their final capacities.
M. Omidi, F. Behnamfar / Engineering Structures 88 (2015) 51–73 67

Table A1
The experimental samples B01, D07 [30], and A-M-Z4 [29] and the corresponding finite element models.

The basis sample B01 SH- SH-B2 SH- SH- D07 SH- SH- SH- A-M- CH1 CH2 CH3
B1 B3 B4 D1 D2 D3 Z-4
Concrete compressive 29 32.4 30.7
strength (MPa)

Beam depth  width (mm) 240  240 240  170 203  406
Longitudinal bars Top 4D13 4U14 4U16 4U18 4U20 7D13 4U10 6U10 4U12 9#3 9U12 9U14 9U16
Bottom 4D13 4U14 4U16 4U18 4U20 7D13 4 6 6 5#3 7 7 7
+2#4

Column Longitudinal bar distance ratio 0.80 0.72 0.73


Tensile reinforcement ratio (%) 0.98 1.19 1.55 1.96 2.42 2.54 0.89 1.34 1.94 0.94 1.35 1.85 2.41
Depth  width (mm) 240  240 240  340 254  457
Longitudinal bars 4D13 4U14 4U16 4U18 4U20 4D13 4U10 4U10 4U12 8#3 8U12 8U14 8U16
2D13 2U10 2U10 2U12 4#4 4U12 4U14 4U16
4U5 4U5 4U5 4U5
Longitudinal bar distance ratio 0.8 0.86 0.70
0.65 0.70
0.30
Tensile reinforcement ratio (%) 0.98 1.19 1.55 1.96 2.42 1.04 0.58 0.58 0.83 1.17 1.24 1.66 2.15
Joint Joint hoops D6-2sets D6-2sets U5-12sets
Column-to-beam depth ratio 1.00 2.00 1.12
v j;max =v n 1.00 1.12 1.31 1.46 1.62 1.22 0.57 0.75 1.02 0.95 1.25 1.49 1.64

Table A2
Characteristics of the sample series B [30].

The basis sample B02 B03 B04 B05 B06 B07 B08 B09 B10
Concrete compressive strength (MPa) 29
Beam depth  width (mm) 240  240
Longitudinal bars 5D13 5D16 4D13 5D13 5D13 4D13 4D13 5D16 5D16
SD345 SD390 SD345 SD390
Longitudinal bar distance ratio 0.8 0.65 0.5 0.65 0.5
Tensile reinforcement ratio (%) 1.22 1.92 0.98 1.22 1.22 1.07 1.18 2.09 2.30
Column depth  width (mm) 240  240
Longitudinal bars 5D13 5D16 6D13 5D13 5D13 4D13 4D13 5D16 5D16
2D13 5D13
SD345 SD390 SD345 SD390
Longitudinal bar distance ratio 0.8 0.8 0.8 0.65 0.5 0.65 0.5
0.5 0.5
Tensile reinforcement ratio (%) 1.22 1.92 1.47 1.80 2.67 1.07 1.18 2.09 2.30
Joint Joint hoops D6(SD295) 2Sets
Column-to-beam flexural strength ratio 1 1.48 1.35 1.78 1

Table A3
Characteristics of the sample series C and D [30].

The basis sample C01 C03 D01 D02 D03 D04 D05 D06 D08
Concrete compressive strength (MPa) 31 32.4
Beam Depth  width (mm) 240  240 240  120 240  170
Longitudinal bars 3D13 5D13 7D13 7D16
+2D13
SD345
Longitudinal bar distance ratio 0.8 0.72
0.5
Tensile reinforcement ratio (%) 1.31 2.62 1.81 2.54 3.98
Column Depth  width (mm) 240  240 240  340
Longitudinal bars 5D13 2D13 3D13 5D13 2D13 3D13 4D13 3D16
SD345
Longitudinal bar distance ratio 0.8 0.86
Tensile reinforcement ratio (%) 1.22 0.33 0.50 0.84 0.33 0.50 0.67 0.79
Joint Joint hoops D6(SD295) 2Sets
Column-to-beam flexural strength ratio 1.03 1.10 0.99 1.42 2.23 0.72 1.03 1.33 0.98

(4) In the D series connections, all of the column springs umn spring rotations have reached their ultimate
enter the nonlinear region and except of D07, the joint capacities.
spring remain in the elastic region. In D01, D04, D05, (5) In AL1 and AL2 that are connections having weak seismic
and D08, the beam springs remained elastic while in details (including no ties in the connection), the joint springs
others they behaved nonlinearly. In D03 and D07 the enter the nonlinear region. In AL1 in which the column-
beam spring rotations and in the other samples the col- to-beam flexural strength ratio is less than one, the column
68 M. Omidi, F. Behnamfar / Engineering Structures 88 (2015) 51–73

Table A4
Characteristics of concrete and bars in the AL1 and AL2 samples [32].

Sample Concrete compressive strength Yield stress of the longitudinal bars of beam and column Yield stress of the transverse bars of beam and column
(MPa) (MPa) (MPa)
AL1 30.3 473 252
AL2 32.1

Table A5
Characteristics and design parameters of the database samples.

Joint specimens Joint description P=Ag f c v j;max =v n Transverse reinforcement Failure mechanism
Test or FEM Model
A-M-Z-4 Reference (Table A1) 0.1 0.95 I A,C A,C
A-M-Z-4-aI Similar to A-M-Z-4 0 0.93
A-M-Z-4-bI 0.2 0.96
A-M-Z-4-cI 0.3 0.97
A-M-Z-4-dI 0.4 0.98
CH1 Reference (Table A1) 0.1 1.25 I A,C A,C
CH1-aI Similar to CH1 0 1.24
CH1-bI 0.4 1.26
CH2 Reference (Table A1) 0.1 1.49 I A,C A,C
CH2-aI Similar to CH2 0 1.46
CH2-bI 0.4 1.52
CH3 Reference (Table A1) 0.1 1.62 I A,C A,C
CH3-aI Similar to CH3 0 1.56
CH3-bI 0.4 1.64
A-M-Z-4-aII Similar to A-M-Z-4 0 0.93 II A,C A,C
A-M-Z-4-II 0.1 0.94
A-M-Z-4-dII 0.4 0.95
CH1- aII Similar to CH1 0 1.18 II A,C A,C
CH1-II 0.1 1.19
CH1-bII 0.4 1.26
CH2- aII Similar to CH2 0 1.32 II A,C A,C
CH2-II 0.1 1.39
CH2-bII 0.4 1.49
CH3- aII Similar to CH3 0 1.48 II A,C A,C
CH3-II 0.1 1.59
CH3-bII 0.4 1.60
A-M-Z-4-aIII Similar to A-M-Z-4 0 0.82 III A,C A,C
A-M-Z-4-III 0.1 0.93
A-M-Z-4-dIII 0.4 0.96
CH1- aIII Similar to CH1 0 1.09 III A,C A,C
CH1-III 0.1 1.22
CH1-bIII 0.4 1.25
CH2- aIII Similar to CH2 0 1.2 III A,C A,C
CH2-III 0.1 1.31
CH2-bIII 0.4 1.33 A,C,D A,C
CH3- aIII Similar to CH3 0 1.17 III A,C A,C
CH3-III 0.1 1.33 A,C,D A,C
CH3-bIII 0.4 1.42
A-M-Z-4-a IV Similar to A-M-Z-4 0 0.72 IV A A
A-M-Z-4-IV 0.1 0.80
A-M-Z-4-dIV 0.4 0.82
CH1-aIV Similar to CH1 0 0.88 IV A A
CH1-IV 0.1 0.91
CH1-bIV 0.4 0.93 A,D A,C
CH2-aIV Similar to CH2 0 0.90 IV A A
CH2-IV 0.1 1.01 A,D A,C
CH2-bIV 0.4 1.05
CH3-aIV Similar to CH3 0 0.99 IV D C
CH3-IV 0.1 1.01
CH3-bIV 0.4 1.17
B01 Reference (Table A1) 0 1.00 I A,B,C A,B,C
B01-aI Similar to B01 0.1 1.13 A,C A,C
B01-bI 0.2 1.17
B01-cI 0.3 1.23
B01-dI 0.4 1.25
B01-eI 0.5 1.27
SH-B1 Reference (Table A1) 0 1.12 I A,B,C A,B,C
SH-B1-aI Similar to SH-B1 0.1 1.29
SH-B1-bI 0.4 1.37 A,C A,C
SH-B1-cI 0.5 1.39 A,B,C A,B,C
M. Omidi, F. Behnamfar / Engineering Structures 88 (2015) 51–73 69

Table A5 (continued)

Joint specimens Joint description P=Ag f c v j;max =v n Transverse reinforcement Failure mechanism
Test or FEM Model
SH-B2 Reference (Table A1) 0 1.31 I A,B,C A,B,C
SH-B2-aI Similar to SH-B2 0.1 1.49
SH-B2-bI 0.4 1.56 A,C A,C
SH-B2-cI 0.5 1.58 A,B,C A,B,C
SH-B3 Reference (Table A1) 0 1.46 I A,B,C A,B,C
SH-B3-aI Similar to SH-B3 0.1 1.69
SH-B3-bI 0.4 1.76 A,C A,C
SH-B3-cI 0.5 1.77 A,B,C A,B,C
SH-B4 Reference (Table A1) 0 1.62 I A,B,C A,B,C
SH-B4-aI Similar to SH-B4 0.1 1.86
SH-B4-bI 0.4 1.88 A,C A,C
SH-B4-cI 0.5 1.96 A,B,C A,B,C
B01-II Similar to B01 0 1.00 II A,B,C A,B,C
B01-aII 0.1 1.09 A,C A,C
B01-dII 0.4 1.17
B01-eII 0.5 1.19 A,B,C A,B,C
SH-B1-II Similar to SH-B1 0 1.17 II A,B,C A,B,C
SH-B1-aII 0.1 1.21 A,C A,C
SH-B1-bII 0.4 1.35
SH-B1-cII 0.5 1.37 A,B,C A,B,C
SH-B2-II Similar to SH-B2 0 1.39 II A,B,C A,B,C
SH-B2-aII 0.1 1.46
SH-B2-bII 0.4 1.55
SH-B2-cII 0.5 1.58
SH-B3-II Similar to SH-B3 0 1.49 II A,B,C A,B,C
SH-B3-aII 0.1 1.66
SH-B3-bII 0.4 1.75 A,C A,C
SH-B3-cII 0.5 1.79 A,B,C A,B,C
SH-B4-II Similar to SH-B4 0 1.63 II A,B,C A,B,C
SH-B4-aII 0.1 1.84
SH-B4-bII 0.4 1.85 A,C A,C
SH-B4-cII 0.5 1.89 A,B,C A,B,C
B01-III Similar to B01 0 1.08 III A,B,C A,B,C
B01-aIII 0.1 1.15 A, C A, C
B01-dIII 0.4 1.18
B01-eIII 0.5 1.23 A,B,C A,B,C
SH-B1-III Similar to SH-B1 0 1.16 III A,B,C A,B,C
SH-B1-aIII 0.1 1.20 A,C A,C
SH-B1-bIII 0.4 1.26
SH-B1-cIII 0.5 1.32 A,B,C A,B,C
SH-B2-III Similar to SH-B2 0 1.38 III A,B,C A,B,C
SH-B2-aIII 0.1 1.44
SH-B2-bIII 0.4 1.57 A,C A,C
SH-B2-cIII 0.5 1.63 A,B,C A,B,C
SH-B3-III Similar to SH-B3 0 1.55 III A,B,C A,B,C
SH-B3-aIII 0.1 1.59
SH-B3-bIII 0.4 1.72 A,C A,C
SH-B3-cIII 0.5 1.77 A,B,C A,B,C
SH-B4-III Similar to SH-B4 0 1.75 III A,B,C A,B,C
SH-B4-aIII 0.1 1.78
SH-B4-bIII 0.4 1.81 A,C A,C
SH-B4-cIII 0.5 1.84 A,B,C,D A,B,C
B01-IV Similar to B01 0 1.01 IV A,B,C A,B,C
B01-aIV 0.1 1.06 A A
B01-dIV 0.4 1.11
B01-eIV 0.5 1.15
SH-B1-IV Similar to SH-B1 0 1.09 IV A,B,C A,B,C
SH-B1-aIV 0.1 1.16 A A
SH-B1-bIV 0.4 1.21 A,D A
SH-B1-cIV 0.5 1.29 A,D A,C
SH-B2-IV Similar to SH-B2 0 1.22 IV A,B,C A,B,C
SH-B2-aIV 0.1 1.25 A,D A
SH-B2-bIV 0.4 1.28 A
SH-B2-cIV 0.5 1.36 A,C
SH-B3-IV Similar to SH-B3 0 1.29 IV A,B,C A,B,C
SH-B3-aIV 0.1 1.31 D A,C
SH-B3-bIV 0.4 1.33 C
SH-B3-cIV 0.5 1.37
SH-B4-IV Similar to SH-B4 0 1.21 IV A,B A,B
SH-B4-aIV 0.1 1.25 D A,C
SH-B4-bIV 0.4 1.35 C
SH-B4-cIV 0.5 1.39
D07 Reference (Table A1) 0 1.22 I A,B,C A,B,C

(continued on next page)


70 M. Omidi, F. Behnamfar / Engineering Structures 88 (2015) 51–73

Table A5 (continued)

Joint specimens Joint description P=Ag f c v j;max =v n Transverse reinforcement Failure mechanism
Test or FEM Model
D07-aI Similar to D07 0.1 1.27 A,C A,C
D07-bI 0.2 1.28
D07-cI 0.3 1.30
D07-dI 0.4 1.33
SH-D1 Reference (Table A1) 0 0.56 I A A
SH-D1-aI Similar to SH-D1 0.1 0.57
SH-D1-bI 0.4 0.60
SH-D2 Reference (Table A1) 0 0.75 I A A
SH-D2-aI Similar to SH-D2 0.1 0.76
SH-D2-bI 0.4 0.79
SH-D3 Reference (Table A1) 0 1.02 I A A
SH-D3-aI Similar to SH-D3 0.1 1.03
SH-D3-bI 0.4 1.07
D07-III Similar to D07 0 1.24 III A,B,C A,B,C
D07-aIII 0.1 1.27 A,C A,C
D07-dIII 0.4 1.29
SH-D1-III Similar to SH-D1 0 0.50 III A A
SH-D1-aIII 0.1 0.52
SH-D1-bIII 0.4 0.54
SH-D2-III Similar to SH-D2 0 0.72 III A,B,C A,B,C
SH-D2-aIII 0.1 0.74 A A
SH-D2-bIII 0.4 0.76
SH-D3-III Similar to SH-D3 0 1.00 III A,B,C A,B,C
SH-D3-aIII 0.1 1.04 A A
SH-D3-bIII 0.4 1.08
D07-IV Similar to D07 0 1.02 IV A A
D07-aIV 0.1 1.09 A,D A
D07-dIV 0.4 1.17 A,C,D A,C
SH-D1-IV Similar to SH-D1 0 0.52 IV A A
SH-D1-aIV 0.1 0.54
SH-D1-bIV 0.4 0.56
SH-D2-IV Similar to SH-D2 0 0.69 IV A A
SH-D2-aIV 0.1 0.72
SH-D2-bIV 0.4 0.74
SH-D3-IV Similar to SH-D3 0 0.95 IV A A
SH-D3-aIV 0.1 0.97
SH-D3-bIV 0.4 0.99 A,D A,C
B02 Reference (Table A2) 0 1.21 I A,B,C A,B,C
B03 0 2.19 I A,B,C A,B,C
B04 0 0.97 I A,B,C A,C
B05 0 1.21 I A,B,C A,B,C
B06 0 1.21 I A,C A,C
B07 0 0.99 I A,B,C A,B,C
B08 0 1 I A,B,C A,B,C
B09 0 2.23 I A,B,C A,B,C
B10 0 2.27 I A,B,C A,B,C
C01 Reference (Table A3) 0 1.16 I A,B,C A,B,C
C03 0 1.56 I A,B,C A,B,C
D01 0 0.89 I A,B A,B
D02 0 0.89 I A,B A,B
D03 0 0.89 I A,B A,B
D04 0 1.25 I A,B B
D05 0 1.25 I A,B B
D06 0 1.25 I A,B A,B
D08 0 2.23 I B B
AL1 Reference (Table A4) 0 1.55 IV B,C B,C
AL2 0 1.28 IV A,C A,C

springs behave nonlinearly while the beam springs in predicting the connection response under lateral load both in
remained elastic. In AL2 connection having a column-to- elastic and plastic regions with an acceptable accuracy. This con-
beam flexural strength ratio larger than unity, the column clusion is true both for modern and traditional (seismically weak,
springs remain elastic despite the beam springs. In both AL1 and AL2) details of the interior RC connections.
samples the joint spring rotations have reached their
ultimate values. 7.3. Evaluating the presented numerical model
(6) In all of the cases, the simulated curves drop down abruptly
after reaching to the ultimate rotation capacities of nonlin- The push-over curves of the numerical model are compared to
ear springs of beam, column, or joint, per case. the envelope of the experimental hysteretic curves. The existing
differences in estimation of the displacement and force corre-
Comparing the simulated and experimental results (Figs. )16– sponding to the point of maximum strength are calculated, and
20 and reviewing the failure mechanisms reported in Table A5, it their averages (Avg.) and standard deviations (Std. dev.) are men-
can be said that the numerical model has been quite successful tioned in Table 11.
M. Omidi, F. Behnamfar / Engineering Structures 88 (2015) 51–73 71

The differences reported in Table 11 being small show that the acceptable accuracy in simulating the elastic and inelastic behavior
model suggested in this research is able to simulate the response of of the connection.
connections with good accuracy. The range of the mean of differ- Although the model is illuminated from that of Birely et al.
ence in estimating the displacement at point of maximum error [27,28], there are important differences completely distinguishing
is from 1.65% for the connections with a shear demand ratio not the two. In addition to difference in the constitutional relations and
larger than 1.2 to 4.73% for the same ratio being not smaller than method of calibrating the model, some other basic dissimilarities
1.5. For estimation of the maximum strength, the average differ- exist between the models, including:
ence varies from 0.82% for connections with a shear demand ratio
equal to or smaller than 1.2 to 1.63% for the shear demand ratio not 1. The model of Birely does not possess the ability to simulate
smaller than 1.5. stiffness reduction of the connection due to yielding in column,
According to Birely et al., the average error of their nonlinear i.e., the nonlinear bending behavior of column is not included.
model in estimating the maximum lateral load capacity of connec- This is while the same factor has an important role in changing
tions with different ductility levels is 1–12% with a standard devi- behavior of the connection when the column is weaker than the
ation of 8–10%, and in estimating the associated displacement is beam in bending. In the proposed numerical model the nonlin-
4–61% with a standard deviation of 53–90%. ear bending behavior of column is included with the concen-
Comparing the above values with those presented in Table 11 trated plasticity.
shows that the nonlinear model presented in this research pos- 2. Response of the nonlinear spring of the panel zone in the model
sesses much smaller errors in estimation of the response values of Birely is bilinear in which stiffness of each branch is a func-
at the same point and can better predict the nonlinear behavior tion of the connection geometry and the elastic modulus of con-
of the said connections. crete. In the proposed model, a tri-linear response curve is used
for the same spring adding the initial cracking of the panel zone.
This in turn has resulted in including effects of parameters such
as the axial load ratio of column, the shear demand ratio and
8. Comparison of the presented numerical model with other the transverse bars of connection.
existing models 3. The Birely model consists of two distinct linear and nonlinear
models each one utilizing a different offset length, such that
As mentioned, so far many experimental studies have been the linear model is only able to simulate the elastic behavior
implemented on the connections and several numerical models and the nonlinear model simulating only the post-yielding
have been proposed to simulate the behavior and failure mecha- behavior. In the presented model, the linear/nonlinear path
nisms of them. Most of the suggested models are too complicated can follow the elastic and post-yielding behaviours in the
to be widespreadly used for common applications. In comparison, framework of a unique numerical model.
the prime characteristics of the numerical model presented in this 4. Regarding accuracy, the proposed model is more accurate in
research are: (1) consistency and ability to be used in common estimating either the initial stiffness (elastic region) and the
commercial structural software packages; (2) high speed of devel- force and displacement at the point of maximum strength
oping the connection model; (3) computational efficiency; (4) (inelastic region) of connection.

9. Conclusions

In this research, a linear-nonlinear numerical model for simu-


lating the elastic and inelastic behavior of RC connections able to
be simply implemented in commercial software packages was pro-
posed. A test database consisting of 23 experimental subassem-
blages along with an analytical database of 167 finite elements
models, developed in this study, was utilized. The essential param-
eters affecting the joint response, including demand-capacity
ratios of joint shear and column axial forces, compressive strength
of concrete, column-to-beam depth ratio, column-to-beam flexural
strength ratio, detail of transverse reinforcement of joint, and lon-
gitudinal rebars arrangement of column and beam, were taken into
account.
The mechanisms of behavior included were those of the interior
Fig. A1. The A-M-Z4 subassemblage (Dimensions in mm) [29]. RC connections consisting of yielding of longitudinal bars of beam
(or column) at the connection (beam or column plastic hinge),

Fig. A2. The connection A-M-Z4 [29], (a) beam section, (b) column section.
72 M. Omidi, F. Behnamfar / Engineering Structures 88 (2015) 51–73

Fig. A3. The boundary conditions and details of the experimental samples of series B, C, and D of Shiohara and Kusahara [30,31].

with the connections used in this study. The shear demand ratio of
connection was utilized to categorize the calibration coefficients of
the proposed linear and nonlinear models of connection.
In the linear part, to appropriately estimate the initial stiffness
of connection, rigid offset lengths were provided in the beam and
column at the connection point. The force–displacement curves
calculated by the proposed rigid offset lengths were compared
with those by the recommended offset lengths of FEMA356 and
ASCE/SEI 41-06, and the experimental results. Errors of the connec-
tion’s initial stiffness in average were 20.3% for FEMA356, 5.4% for
ASCE/SEI 41-06, and 1.6% for the proposed model, showing the very
good accuracy of this model in estimating the initial stiffness.
In the nonlinear model, use was made of the beam and column
elements with concentrated plasticity. Each of these elements
included two rotational springs placed in series. A spring is repre-
Fig. A4. Details of the experimental samples AL1 and AL2 of Li et al. [32]. sentative of the nonlinear behavior of beam and column and the
other of the nonlinear behavior of the connection. The moment-
rotation response of the nonlinear spring was considered to be
yielding of transverse reinforcement of the connection resulting in tri-linear considering the initial cracking of the panel zone. This
forming of diagonal tensile cracks and shear failure of the joint, in turn resulted in including effects of parameters such as the axial
and, yielding of transverse reinforcement of beam or column at load ratio of column, the shear demand ratio and the transverse
the connection resulting in beam or column shear failure. The bond bars of connection. Certain limits were considered for rotations
mechanism is ignored in this study. It was shown that the latter of the beam, column, and panel zone nonlinear springs to include
assumption had no important consequences on the results analysis limitation of the ductility capacity of the connection. Contrary to
M. Omidi, F. Behnamfar / Engineering Structures 88 (2015) 51–73 73

Fig. A5. Dimensions and details of beam and column in the AL1 and AL2 samples [32].

some other existing models, in the proposed model, the linear/non- [16] Vecchio FJ, Collins MP. The modified compression field theory for reinforced
concrete elements subjected to shear. J Am Concrete Inst 1986;83:219–31.
linear path can follow the elastic and post-yielding behaviours in
[17] Shin M, LaFave JM. Modeling of cyclic joint shear deformation contribution in
the framework of a unique numerical model. RC beam-column connection to overall frame behavior. Struct Eng Mech
The results of evaluating the numerical model illustrated that 2004;18(5):645–69.
the average of errors in estimating the displacement and force at [18] Jeremic B, Bao Y. Modeling of reinforced concrete beam-column joint. ECI 248
term project report, Civil and Enviromental Engineering Department,
the maximum capacity were 3.1% and 1.1% respectively, represent- University of California, Davis; 2005.
ing the very good ability of the model in simulating the post- [19] Tajiri S, Shiohara H, Kusuhara F. A new macroelement of reinforced concrete
yielding behavior of the connections. beam-column joint for elasto-plastic plane frame analysis. In: Eighth national
conference of earthquake engineering, San Francisco, California; 2006. No. 674.
[20] Bao Y, Kunnath SK, Tawi SE, Lew HS. Macromodel-based simulation of
progressive collapse: RC frame structures. J Struct Eng 2008;134(7):1079–91.
Appendix A [21] Anderson M, Lehman D, Stanton J. A cyclic shear stress–strain model for joints
without transverse reinforcement. Eng Struct 2007;30(4):941–54.
A.1. The database of this study [22] Kim J, LaFave JM. Key influence parameters for the joint shear behavior of
reinforced concrete (RC) beam-column connections. Eng Struct
2007;29:2523–39.
Characteristics of the samples selected or produced for calibra- [23] Wang GL, Dai JG, Teng JG. Shear strength model for RC beam–column joints
tion in this study are given in Tables A1–A5 and Figs. A1–A5. under seismic loading. Eng Struct 2012;40:350–60.
[24] Yu J, Tan KH. Experimental and numerical investigation on progressive
collapse resistance of reinforced concrete beam column sub-assemblages.
References Eng Struct 2013;55:90–106.
[25] Ghobarah A, Biddah A. Dynamic analysis of reinforced concrete frames
including joint shear deformation. Eng Struct 1999;21(11):971–87.
[1] Giberson MF. Two nonlinear beams with definitions of ductility. J Struct Div, [26] Fleury F, Reynouard JM, Merebet O. Multi component model of reinforced
ASCE 1969;95:137–57. concrete joints for cyclic loading. J Eng Mech 2000;126(8):804–11.
[2] Otani S. Inelastic analysis of RC frame structures. ASCE J Struct Div [27] Birely A, Lowes L, Lehman DE. A model for the practical nonlinear analysis of
1974;100(7):1433–49. reinforced-concrete frames including joint flexibility. Eng Struct
[3] Takeda T, Sozen MA, Nielsen N. Reinforced concrete response to simulated 2012;34:455–65.
earthquakes. ASCE J Struct Eng Div 1970;96(12):2257–73. [28] Birely A, Lowes L, Lehman DE. Practical linear and nonlinear models of
[4] Banon H, Biggs JM, Irvine HM. Seismic damage in reinforced concrete frames. reinforced concrete beam-column joints in existing structures. In: 9th US
ASCE J Struct Div 1981;107(9):1713–29. national and 10th Canadian conference on earthquake engineering, Toronto
[5] Fillipou FC, Popov EP, Bertero VV. Effects of bond deterioration on hysteretic (Ontario, Canada); 2010. No. 694.
behaviour of reinforced concrete joints. Berkeley (CA): EERC, Univ. of [29] Cheok G, Lew HS. Performance of 1/3 scale model precast concrete beam-
California; 1983. Report no UCB/EERC 83/19. column connections subjected to cyclic inelastic loads. Report no. 2. NISTIR
[6] Fillipou FC, Issa A. Nonlinear analysis of reinforced concrete frames under 4433, NIST, Gaitherburg, MD; June, 1991.
cyclic load reversals. Berkeley (CA): EERC, Univ. of California; 1988. Report no. [30] Shiohara H, Kusuhara, F. An overlooked failure mechanism of reinforced
UCB/EERC-88/12. concrete beam-column joints. In: 7th CUEE and 5th ICEE joint conference,
[7] Alath S, Kunnath SK. Modeling inelastic shear deformation in RC beam-column Tokyo Institute of Technology, Tokyo (Japan); 2010.
joints. In: Tenth conference on engineering mechanics. University of Colorado, [31] Shiohara H. Reinforced concrete beam-column joint: failure mechanism
Boulder; 1995. p. 822–5. overlooked. J Struct Construct, AIJ 2008;73(631):1641–8. in Japanese.
[8] Pampanin S, Magenes G, Carr A. Modeling of shear hinge mechanism in poorly [32] Li B, Tran CT, Pan TS. Experimental and numerical investigations on the seismic
detailed RC beam column joints. In: Concrete structures in seismic regions: fib behavior of lightly reinforced concrete beam-column joints. J Struct Eng
2003 symposium, May 6–8, Athens, Greece. Athens: Technical Chamber of 2009;135(9):1007–18.
Greece; 2003. Paper no. 171. [33] FEMA 356. Prestandard and commentary for the seismic rehabilitation of
[9] Sharma A, Eligehausen R, Reddy GR. A new model to simulate joint shear buildings. Prepared by the American Society of Civil Engineers for the Federal
behavior of poorly detailed beam–column connections in RC structures under Emergency Management Agency, Washington, DC (FEMA Publication No.
seismic loads, Part I: Exterior joints. Eng Struct 2011;33(3):1034–51. 356); 2000.
[10] Biddah A, Ghobarah A. Modeling of shear deformation and bond slip in [34] ASCE/SEI 41. Seismic rehabilitation of existing buildings. American Society of
reinforced concrete joints. J Struct Eng Mech 1999;7:413–32. Civil Engineers, Reston, Virginia; 2007.
[11] Hsu TTC. Unified theory of reinforced concrete. Boca Raton (FL): CRC Press Inc.; [35] ANSYS Inc, Release 10.0 Documentation for ANSYS, United States; 2005.
1993. [36] Desayi P, Kirshan S. Equation for the stress–strain curve of concrete. J Am
[12] Elmorsi M, Kianoush MR, Tso WK. Modeling bond-slip deformations in Concrete Inst 1964;61:345–50.
reinforced concrete beam-column joints. Can J Civil Eng 2000;27(3):490–505. [37] Corley GW. Rotational capacity of reinforced concrete beams. J Struct Div,
[13] Youseef M, Ghobarah A. Modeling of RC beam-column joints and structural ASCE 1966;92(ST5):121–46.
walls. J Earthquake Eng 2001;5:93–111. [38] Omidi M. Study of non-linear static behavior and equivalent non-linear model
[14] Lowes LN, Altoontash A. Modeling reinforced-concrete beam-column joints of reinforced concrete beam-column joints. MSc thesis. Department of Civil
subjected to cyclic loading. J Struct Eng 2003;129(12):1686–97. Engineering, Isfahan University of Technology, Iran; 2011.
[15] Altoontash A. Simulation and damage models for performance assessment of [39] Computer and Structures Inc, SAP2000 Analysis References, Berkeley,
reinforced concrete beam-column joints. Stanford, CA, University, PhD Thesis; California; 2008.
2004.