Engineering Structures 166 (2018) 31–45

Contents lists available at ScienceDirect

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

Seismic behavior comparison of reinforced concrete interior beam-column T

joints based on different loading methods
⁎
Hong Yanga, , Wentong Zhaoa, Zhenzhen Zhua, Jianping Fub
a
School of Civil Engineering, Chongqing Univ., Chongqing, China
b
Key Laboratory of New Technology for Construction of Cities in Mountain Area, Chongqing Univ., Chongqing, China

A R T I C LE I N FO A B S T R A C T

Keywords: Due to the difference in loading point, experimental researches on seismic behavior of beam-column joint can be
Beam-column joint generally divided into two types, namely loading at beam ends (BL method) and loading at column top end (CL
Loading method method). This paper presents an experimental investigation aimed at comparing the cyclic behavior of interior
Joint shear strength beam-column joints with different loading methods. Four groups of full-scale interior beam-column sub-
Longitudinal bar-slip
assemblages were tested under reversed cyclic loading. Two identical specimens were tested under two different
Joint shear deformation
loading methods in each group. On the basis of the experimental results, the paper discusses the distinctions of
cyclic behaviors under different loading methods in terms of crack pattern, joint shear strength, joint shear
transfer mechanism, joint shear deformation, longitudinal bar-slip, energy dissipation, and stiffness degradation.
In addition, a numerical simulation of these four groups of specimens is shown and discussed. The experimental
results and numerical analyses confirm that the loading method can have certain influence on the cyclic behavior
of the beam-column joints. The joint shear strength, joint shear deformation and longitudinal beam bar slip of
beam-ends loading specimens (BL specimens) are considerably larger than that of column-end loading specimens
(CL specimens), although the loading protocol used for CL method is more severe than BL method in terms of
number of loading cycles. In consequence, the influence of loading method should be considered when the
seismic behavior of the beam-column joint is evaluated according to a great number of existing beam-column
joint tests.

1. Introduction
Post-earthquake inspections indicate that the severe damage or
failure of beam-column joints can result in excessive drift or global
collapse of the structure [1–3]. Under cyclic loading, high shear stress is
concentrated in the beam-column joint region. Significant bond slip and
strength degradation occur when the yielding of longitudinal bars of the
beam penetrates into joint region and diagonal cracking develops. Bond
slip failure or shear failure of joint are considered undesirable as they
lead to strength degradation and stiffness degradation of the RC frame
[4]. Thus, joint shear deformation and bar slip deformation are con-
sidered as crucial parts in experimental research.
It is essential that beam-column joint can effectively transmit the
necessary shear forces through joint region. The joint shear force can be
transferred across the joint by a combination of two mechanisms,
namely strut mechanism and truss mechanism [4–7]. The truss counts
on the forces transferred by bond of longitudinal bars, the horizontal
transverse steel and the vertical column reinforcement to maintain the
truss action, so that a large amount of transverse reinforcement is
required in the joint region to sustain the joint strength. The diagonal
compression strut primarily depends on the compression stresses acting
on the concrete of beam and column critical sections and the concrete
strength of strut. In this case, the requirement of joint transverse re-
inforcement reduces to just an amount required to confine the concrete
properly. Different crack patterns can be observed under different
transfer mechanisms [5–12].
A number of research teams in different parts of the world have
studied the behavior of beam-column joints in order to improve its
response to seismic loads. In the experimental programs, cyclic loads of
the experiments can be introduced in two methods: loading on the
column-end (CL method) or loading on the beam-ends (BL method), as
shown in Fig. 1. Some research teams applied the cyclic loads by CL
method [6,9,10,12–14]. The deflected shape of the subassemblage
under CL method is similar to the deflected shape of a subassemblage in
planar frame under earthquake motion, as shown in Fig. 1(a). The
cyclic loading applied at the column end was generally displacement-
controlled with the increasing of story drift ratio [12–14]. However, it
is slightly difficult to control the boundary condition in this method.

⁎
Corresponding author at: School of Civil Engineering, Chongqing Univ., Chongqing 400044, China.
E-mail addresses: yangh@cqu.edu.cn (H. Yang), wentong21@163.com (W. Zhao), clocknine@sina.cn (Z. Zhu), fjp@cqu.edu.cn (J. Fu).

https://doi.org/10.1016/j.engstruct.2018.03.022
Received 17 October 2017; Received in revised form 10 March 2018; Accepted 11 March 2018
Available online 28 March 2018
0141-0296/ © 2018 Elsevier Ltd. All rights reserved.
H. Yang et al.
Fig. 1. Loading methods of beam-column joint.
Thus, the other research teams applied cyclic loads by BL method
[8,11,15–20]. The cyclic loading applied at the beam ends was gen-
erally load-controlled at first and displacement-controlled subsequently
with the increasing of displacement ductility factor [8,11,15–18]. This
method is more complicated due to the requirement of two coordinated
actuators (Fig. 1(b)), but the boundary condition is easier to control. A
significant number of experimental studies have been performed on
beam-column subassemblages by different loading methods, but it was
evaluated together without dividing the loading methods, which may
lead to the misjudgment of the seismic behavior [21–23].
The main objective of this research is to compare the difference in
the seismic behavior between the identical specimens with two loading
methods, and to quantitatively evaluate their difference. Four groups of
full-scale RC interior beam-column subassemblages were designed,
constructed and tested to compare the seismic behavior of the joint.
There were two identical specimens (same geometry and design details)
in each group, one loaded by CL method and the other one loaded by BL
method to allow a performance comparison to be established. The
conventional drift ratio based loading protocol and the conventional
displacement ductility based loading protocol were used for CL and BL
specimens respectively, in order to keep consistent with the loading
protocols of these two kinds of loading methods previously used by
other researchers [11,13–17]. The investigation was focused on com-
paring the difference of crack pattern, joint shear strength, joint shear
transfer mechanism, joint shear deformation, longitudinal bar-slip,
energy dissipation, and stiffness degradation, with the tested data
processed using the formulations extensively used by other research
groups. Then, the four groups of specimens were simulated satisfacto-
rily using non-linear finite element analysis, and the difference in the
simulated behavior between two loading methods was compared.
There are a large number of existing beam-column subassemblages
experimental results that were tested using different loading methods.
Therefore, they should not be analyzed together before the effects of
loading method on the seismic behavior of beam-column sub-
assemblages are investigated thoroughly. The target of this study is to
compare the difference of the test results under two loading methods
using their conventional loading protocol, but not to study how to
change the loading protocol of this two loading methods to obtain the
identical test results. Based on the research results of this paper, it is
clarified that loading methods affect the seismic behavior of beam-
column joint to different degrees, and the difference was quantitatively
32
Engineering Structures 166 (2018) 31–45
evaluated. This study contributes to combine a large number of existing
test results of BL specimens and CL specimens to make a comprehensive
evaluation of the beam-column joint seismic behavior considering the
effects of loading methods.
2. Test program
2.1. Specimen and design parameter
A total of eight specimens including four CL specimens (named CL-
n, n = 1, 2, 3, 4) and four BL specimens (named BL-n, n = 1, 2, 3, 4)
were divided into four groups (Group-n included CL-n and BL-n). Two
specimens in each group were designed with the same dimensional size
and reinforcement details, but loads were applied in different ways. The
specimens were designed based on the strong column-weak beam re-
quirement and detailed requirements of Chinese Code [24]. The joint
transverse reinforcements were close to the lower limit value calculated
according to the Chinese Code [24]. Fig. 2 shows the geometry, di-
mensions, and reinforcement details of each group of specimens.
2.2. Material specifications
The concrete compressive strength of all the specimens at the time
of testing is presented in Table 1, and the yield strength of the re-
inforcements is presented in Table 2. The specimens of different group
were prepared separately, and two specimens in the same group were
tested in different days, thus, there were differences in the strength of
reinforcements and concrete.
2.3. Test setup, instrumentation and loading sequence
The loading apparatuses of two loading methods are shown in
Fig. 3(a) and (b). A constant axial load was applied to the column of
both CL and BL specimens through an actuator at the column top, which
was maintained constant during the tests. The axial load ratio of Group
I, II, III and IV are 0.25, 0.3, 0.25 and 0.25, respectively. For CL spe-
cimens, the bottom of the column was hinge support, and both ends of
beams were roller supports. Loads applied at the top end of the column
was measured with a load cell attached in series with an actuator that
applied reversed cyclic lateral load, as shown in Fig. 3(a). The full
specimen under CL method in laboratory is shown in Fig. 3(c). The test
H. Yang et al. Engineering Structures 166 (2018) 31–45

Fig. 2. Geometry, dimensions, and reinforcement detailing of test specimens.

Table 1 consistent with other researchers who have used BL method

Mechanical properties of concrete on day of testing (average test results). [11,15,17,18]. The load-controlled loading was applied in the first two
Group I Group II Group III Group IV
cycles and displacement-controlled loading was applied in the sub-
sequent cycles. The first cycle was to get cracking load of the beam and
CL-1 BL-1 CL-2 BL-2 CL-3 BL-3 CL-4 BL-4 to check the instruments. The second cycle was to get four yield dis-
placement Δy (including top and bottom longitudinal beam bar yield
Compressive strength 35.5 28.8 38.2 37.5 34.3 29.8 30.2 32.8
displacement at positive and negative direction). The displacement-
fcu150 (N/mm2)
controlled loading started from the third loading cycle with the in-
creasing of displacement ductility factor. The displacement history of
Table 2 the BL specimens is shown in Fig. 4(b). The displacement ductility
Material properties of reinforcement bars obtained from material testing. factor μ is defined as the ratio of peak displacement Δ to the yield
displacement Δy .
Group No. Beam and column longitudinal Joint transverse reinforcement
reinforcement
To measure the strains of the reinforcements, strain gauges with 3-
mm gauge length were installed on longitudinal beam and column re-
Diameter, mm Yield strength, Diameter, mm Yield strength, inforcements as well as the transverse reinforcements in the joint re-
MPa MPa gion. In addition, the joint panel and the longitudinal beam bars at
Group I 16 528 12 407
beam plastic-hinge zone were installed with dial indicators to measure
Group II 20 540 12 368 the joint shear deformations and longitudinal bar-slips. Four dial in-
Group III 16 444 12 464 dicators were installed on bars welded in the beam longitudinal re-
Group IV 16 553 12 362 inforcement to measure the longitudinal bar-slips, and two dial in-
dicators were installed in the diagonal direction on the joint panel to
measure the joint shear deformations. The external instrumentation
was conducted under the displacement-controlled test method with the
plan is shown in Fig. 3(e).
increasing of story drift ratio. The loading sequence of CL specimens is
commonly adopted by other researchers [12–14]. The displacement
history of the CL specimen is shown in Fig. 4(a). For BL specimens, the 3. Test results and discussion
top and bottom of the column were hinge supports. Loads applied at
east and west beam ends were measured with two load cells attached in 3.1. Hysteretic load-displacement relationship
series with the two actuators that applied the reversed cyclic loads, as
shown in Fig. 3(b). The full specimen under BL method in laboratory is The hysteretic responses of the applied load-displacement for the
shown in Fig. 3(d). The generated loads at the beam ends were mea- four groups of specimen are illustrated in Fig. 5. The crack pattern of
sured using two load cells. The test was conducted under the load- the specimens at the same condition are illustrated in Fig. 6, corre-
controlled and displacement-controlled test method, in order to keep sponding to a 15% strength reduction of the peak resistance. All spe-
cimens displayed significant damage within the joint region with

33
H. Yang et al.
Fig. 3. Test setup and instrumentation.
concrete cover spalling.
For the CL specimens, the diagonal cracks in the joint core surfaces
were first observed at a drift ratio (DR) of 0.75% and developed
stronger and wider with the loading cycles repeated. The first yield of
the bottom beam reinforcement was developed at a DR close to 1.35%.
As the lateral drift increased, the lateral load resistance of the specimen
increased until reaching a peak load at an average DR of 2%.
Subsequently, the joint core began to crush and spall off.
Simultaneously, the CL specimens experienced strength and stiffness
degradation as a result of the bond degradation of longitudinal beam
reinforcements, widening of the diagonal shear cracks and the con-
sequent shear strength degradation within the joint core. It is noted that
some of the top beam reinforcement of the CL specimens yielded near
34
Engineering Structures 166 (2018) 31–45
the end of loading whiles some did not yield.
For the BL specimens, the diagonal cracks in the joint region pro-
pagated rapidly. The top beam reinforcement at the west side of the
joint and the bottom beam reinforcement at the east side of the joint
yielded simultaneously at a displacement ductility factor μ = 1. The top
beam reinforcement at the east side of the joint and the bottom beam
reinforcement at the west side of the joint yielded simultaneously at a
displacement ductility factor μ = −1. Diagonal cracks densely and
uniformly distributed in the whole joint core at this stage. Concrete
crush and concrete cover spall off within the joint region initiated
during the increase of μ from 1 to 2 and reached a peak load at μ = 2.
Pinching behavior observed in the hysteresis response of BL specimens
was more severe than that of CL specimens. As presented in Fig. 6, the
H. Yang et al.
Fig. 4. Loading history.
average crack interval of the beam flexural cracks of BL specimens was
smaller than that of their corresponding CL specimens.
As shown in Fig. 1, the beam ends of CL specimens were roller
supports so that the vertical deformations of beam ends were sig-
nificantly smaller than that of BL specimens. When loaded by CL
method, the horizontal displacements at column top were symmetric in
positive and negative direction. In this condition, the upward and
downward beam bending deflections were similar. Therefore, under CL
method, the bottom beam bars in asymmetric reinforced beam (top
beam bars are more than bottom beam bars) always yielded much
earlier than top beam bars. When loaded by BL method, the external
force was directly applied at beam ends that can enforce the top and
bottom beam bar yielded simultaneously. In this condition, because of
the asymmetrical reinforcement in beam, the generated loads at beam
ends in downward direction were significantly larger than that in up-
ward direction, as shown in Fig. 5. This distinction between CL and BL
method led to larger compression stress of diagonal strut in the joint
region of BL specimens.
It is noted that, when μ = 2 (see in Fig. 5), the west beam shear
force was significantly larger than east beam shear force in the BL
specimens, however the vertical deformations at east and west beam
ends were similar (except for the east side of BL-3). It can be deduced
that, if the loading protocol of BL specimens is consistent with that of
CL specimens, the top and bottom beam bars may not yield simulta-
neously, but the yield displacements of the two directions should just
have little difference. In consequence, if both of the CL and BL speci-
mens were loaded in the drift ratio based displacement-controlled
loading protocol, the discipline of test results and the comparison re-
sults should be similar to that in this research.
3.2. Joint shear strength and shear transfer mechanisms
As previously mentioned, the cyclic displacements of CL and BL
specimens were loaded using their conventional loading protocols re-
spectively. The longitudinal beam bars and the concrete in joint region
were subjected to different loading effects when loaded in different
loading method, which will lead to different degrees of inelastic de-
formations. Because of the difference in loading point, it is hard to
compare the test results of CL and BL specimens with respect to dis-
placement of column top end for CL specimens and displacement of
beam end for BL specimens.
For the sake of comparison, two specimens in each group are
evaluated at the identical representative damage degrees. Four feature
points representing four representative damage degrees of the joint are
selected: (1) the initial cracking of joint core, (2) the first yield of the
longitudinal beam reinforcement, (3) the peak strength, and (4) 15%
strength degradation of the peak strength. The evaluation results of the
35
Engineering Structures 166 (2018) 31–45
horizontal joint shear force of the four groups of joints are presented in
Table 3. The horizontal joint shear force can be calculated as follow [7],
MW + ME
Vjh = −Vc
jhb (1)
where MW and ME = moments at beam-joint interface on the west and
east side of the joint, respectively; jhb = distance between the tension
and compression resultants in the beam section at the beam-joint in-
terface; Vc = column shear force. MW and ME are calculated according
to the beam shear force VW and VE , as shown in Fig. 1. For CL specimens,
VW and VE are measured directly with the two tension-compression
sensors installed on the supports of the beams. For BL specimens, VW
and VE are measured with two load cells attached in series with the two
actuators that applied the reversed cyclic loads. Because of the varia-
tion of loading apparatuses between CL and BL method, CL specimens
were under the influence of P−Δ effect while BL specimens did not.
Therefore, in the calculation, the P−Δ effect is included in horizontal
joint shear force of CL specimens.
As presented in Table 3, the joint shear strength of BL specimens
were relatively larger than that of their corresponding CL specimens.
The horizontal joint shear force at the peak strength of BL specimens
were 8.0%, 9.0%, 9.4% and 9.6% larger than their corresponding CL
specimens in Group I, II, III and IV respectively. As previously men-
tioned, the top and bottom beam bar of the BL specimens yielded si-
multaneously at the early stage of loading, the bottom beam bar of the
CL specimens yielded at a DR of 1.35%, however some of the top beam
bar of the CL specimens yielded at the end of loading whiles some did
not yield. In other words, the tensile stresses of bottom beam bars in
each group were similar, while the tensile stresses of top beam bar of BL
specimens were larger than that of CL specimens, which is related to the
difference in loading method and testing apparatuses. The tensile
stresses of longitudinal beam bars contribute to the joint shear force.
Thus, the discrepancy of the joint shear force is primarily attributed to
the variation of loading protocols and loading apparatuses, which led to
the discrepancy of the tensile stresses between the top beam bar of CL
and BL specimens.
The crack patterns of specimens in the same loading method were
common to one another. Under the condition shown in Fig. 6, the
number of joint diagonal cracks of CL specimens was less than that of
BL specimens, and the concrete spalling of CL specimens was only
concentrated in the joint center compared to BL specimens. The dif-
ferent crack patterns of these two loading method can be attributed to
the contribution of different shear transfer mechanisms within joint.
Evidently, the CL and BL specimens are mainly governed by the diag-
onal strut and the truss mechanism respectively. This phenomenon,
which is related to the stress state of beam longitudinal reinforcement,
the distribution of bond stress and the degradation characteristics, is
H. Yang et al. Engineering Structures 166 (2018) 31–45

Fig. 5. Load-displacement hysteretic response of tested beam-column joints.

explained as below. beam reinforcement; h c = width of the joint; τc and τb = average bond
The equilibrium condition of top longitudinal beam bars in CL and stress of CL and BL specimens. At the first yield, the top and bottom
BL specimens at first yielding point mentioned previously are illustrated beam bar of the BL specimens yielded simultaneously, however, in CL
in Fig. 7. specimens, only the bottom beam bar yielded due to more steel bars
′ ′
For CL specimens were reinforced at the top. Therefore, f sc and f sb are similar, while fy is
′ significantly larger than f sc . Correspondingly, the average bond stress in
f sc Astop + τc h c πdb = f sc Astop (2a) BL specimens (τb ) are larger than that in CL specimens (τc ). The larger
distinction of top beam bar stress between CL and BL specimens re-
For BL specimens
inforcement leads to the larger distinction between τb and τc . This
′
f sb Astop + τb hc πdb = fy Astop phenomenon is primarily attributed to the variation of loading proto-
(2b)
cols and loading apparatuses between CL and BL method. The BL
′ ′ method enforces the top and bottom beam bar yielded simultaneously
where f sc , f sc , f sb and fy = stresses of the top reinforcements;
Astop = sectional area of top beam reinforcements; db = diameter of top while the CL method does not. In addition, the relatively more loading

36
H. Yang et al. Engineering Structures 166 (2018) 31–45

Fig. 6. Crack observation of specimens.

Table 3 is resisted primarily by a truss mechanism, while the joint shear in CL

Evaluation of horizontal joint shear force. specimens is resisted primarily by a diagonal compression strut.
Observation Loading method Horizontal joint shear force If a truss mechanism prevails, larger bond stress will be formed in
the beam bars at the early stage of loading period, and after the yielding
Group I Group II Group III Group IV of beam bars the joint will experience relatively severe bond degrada-
tion and yield penetration. In addition, the joint core will become more
First crack CL 565 652 641 628
BL 640 590 684 461
flexible when the joint shear is transferred by a relatively flexible truss
First yield CL 1057 1150 925 840 mechanism than by a relatively stiff strut mechanism [9]. In this case,
BL 1067 1184 991 858 larger bar slip and joint shear deformation of BL specimens will be
Peak CL 1120 1162 945 947 observed in comparison with an identical CL specimen governed by
BL 1190 1267 1034 1038
strut mechanism.
15% Degradation CL 883 1082 731 780
BL 868 1137 805 856

3.3. Damping and stiffness properties

cycles in CL specimens also expedite the bond degradation.
The equivalent viscous damping of all specimens was calculated
Accordingly, the distinction between CL and BL method mentioned
using the [Eqs. (3)–(5)]. In addition, the degree of stiffness degradation
above led to larger compression stress of diagonal strut and larger bond
of all specimens was also calculated using the [Eqs. (6), (7)]. The cal-
stress of top beam bar in the joint region of BL specimens.
culation results of equivalent viscous damping and stiffness degradation
The strain distribution of longitudinal beam bars at the latter three
are compared with respect to the feature point mentioned previously, as
feature points mentioned previously in positive loading direction
shown in Fig. 9. It is noted that there was no integrated cycle at the first
(Fig. 3) of Group I and II are evaluated. The evaluation results of top
crack, thus only the latter three feature points mentioned previously
longitudinal bean bar of Group I and bottom longitudinal beam bar of
were evaluated.
Group II are presented in Fig. 8. Steeper slope of strain distribution is
attributed to larger bond stress. More distinct change in the slope of ED
ξeq =
strain distribution is attributed to more significant bond degradation, 4πEsto (3)
and larger bond stress causes more contribution of truss mechanism.
According to Fig. 8, it can be stated that at first yield (point 2) and peak (δi + 1−δi )(Fi + 1 + Fi )
strength (point 3) the bond stress along beam bars of BL specimens were ED = ∑ 2 (4)
significantly higher compared with those of CL specimens. Whereas at
the 15% degradation (point 3) the bond stress along beam bars of BL F0 Δ0
Esto =
and CL specimens tended to be similar accompanied with bond de- 2 (5)
gradation. The comparison of the crack patterns and bond stress of
beam bars among the two group of specimens indicates that, before the (F+ve−F−ve)
kp =
significant strength degradation occurs, the joint shear in BL specimens (Δ+ve−Δ−ve) (6)

Fig. 7. Equilibrium condition of top longitudinal beam bar.

37
H. Yang et al.
Fig. 8. Comparison of strain distribution in the longitudinal reinforcement in beam.
(F+y−F−y )
ky =
(Δ+y −Δ−y ) (7)
where ξeq = equivalent viscous damping; ED = dissipated energy of
each cycle; Esto = equivalent elastic strain energy associated in each
cycle; δi = displacement at each step; Fi = load at each step; F0 = peak
load of each cycle; Δ0 = peak displacement of each cycle; k p = peak-to-
peak stiffness; F+ve and F−ve = maximum and minimum forces of the
intended cycle; Δ+ve and Δ−ve = maximum and minimum displacements
of the intended cycle; k y = peak-to-peak stiffness of the yielded cycle;
F+y and F−y = maximum and minimum forces of the yielded cycle; and
Δ+y and Δ−y = maximum and minimum displacements of the yielded
cycle.
According to Fig. 9(a) at the first yield BL specimens show higher
damping properties than CL specimens, except for Group III. At the
peak strength, the equivalent viscous damping of BL specimens are
56.2%, 61.5%, 12.6% and 4.1% higher than their corresponding CL
specimens in Group I, II, III and IV respectively. This discrepancy is
primarily attributed to the variation of loading protocols and loading
apparatuses. The top and bottom beam bar of the BL specimens yielded
simultaneously, which made the beam shear force of the BL specimens
in downward direction larger than that in CL specimens. While at the
Fig. 9. Equivalent viscous damping and stiffness degradation.
38
Engineering Structures 166 (2018) 31–45
15% degradation, the equivalent viscous damping of BL specimens are
1.6%, 17.1%, 22.6% and 31.7% lower than their corresponding CL
specimens in Group I, II, III and IV respectively.
In addition, as shown in Fig. 9(b), the degree of stiffness degrada-
tion of CL specimens are generally lower than BL specimens. At the 15%
degradation, the stiffness degradation ratio (kp/ky) of BL specimens are
43.3%,43.4%, 53.1% and 18.6% lower than CL specimens in Group I, II,
III and IV respectively. This phenomenon is primarily attributed to the
more severe material damage in BL specimens.
3.4. Joint shear deformation
Measurement of diagonal displacements within the joint region by
dial indicator I and II (Fig. 3(c)) enabled calculation of joint shear de-
formation as shown in Fig. 10. The overall joint shear strain can be
conventionally evaluated from measurements as follow [8,11],
hb2 + hc2
γ = γb + γc = (ΔI−ΔII)
2hb hc (8)
where γb and γc = joint shear strain at beam side and column side; hb
and h c = height and width of the joint core; ΔI and ΔII = deformations
measured by dial indicator I and II (‘‘+” for lengthening and ‘‘−‘‘for
H. Yang et al.
Fig. 10. Beam–column joint area deformation.
shortening). The calculation of the horizontal shear force is mentioned
previously. The joint shear stress is expressed as follow,
Vjh
τ=
bj h c (9)
where bj is effective joint width defined by ACI 318 M-14 [25].
The joint shear strain against joint shear stress of Group I and IV is
illustrated in Fig. 11. The joint shear stress was directly proportional to
the load applied at loading point. The measured joint shear deformation
increased after the maximum load recorded at 2.2% drift and μ = 2 for
CL-1/4 and BL-1/4, respectively, followed by significant degradation on
strength and stiffness. Fig. 12 shows a comparison of the joint shear
deformation envelopes for each group, with indication of the same
condition corresponding to a stress reduction of 15% with respect to the
peak stress. Joint shear stresses were slightly greater in BL specimens
than in their corresponding CL specimens. These results, in accordance
with Table.3, are primarily attributed to the variation of loading pro-
tocols and loading apparatuses. Joint shear strains were significantly
greater in BL specimens than that in CL specimens, and the average
joint shear strains measured at the 15% degradation of the positive and
negative loadings are 0.025–0.03 rad for CL specimens,
0.045–0.075 rad for BL specimens.
Although the loading protocol used for CL method is more severe
than BL method in terms of number of cycles, the joint shear de-
formation of CL specimens was significantly smaller than that of BL
specimens. Accordingly, the different cyclic behavior between CL and
BL specimens is primarily attributed to the variation of loading
Fig. 11. Joint shear stress-joint shear deformation hysteretic response of Group I and IV.
39
Engineering Structures 166 (2018) 31–45
protocols and loading apparatuses. The BL method enforces the top and
bottom beam bar yielding simultaneously, which lead to larger com-
pression stress of diagonal strut and larger bond stress of beam bar in
the joint region of BL specimens. It can be deduced that, if the number
of loading cycles used in CL method is consistent with that adopted in
BL method for the specimens in Group I, II, III and IV of this paper, the
joint shear deformations of CL specimens should decrease, and the
discrepancy of joint shear deformations between CL and BL specimens
should be larger than that in Figs. 11 and 12.
The joint shear strain of the four feature points mentioned pre-
viously of each group are compared in Fig. 13. As shown in this figure,
shear strains at the initial cracking point were similar between two
loading methods. Comparing the yield, peak, and 15% degradation
point, the difference of joint shear strain between the specimens in
different loading methods increased with the increase of loading dis-
placement. At the 15% degradation point, the average shear strain of
positive and negative direction of CL specimens were 68.3%, 49.8%,
36.5% and 64.6% lower than their corresponding BL specimens in
Group I, II, III and IV respectively. The relatively small distinction of
Group III is partially assigned to the smaller relative beam width to
column width ratio.
Concrete strut mechanism is beneficial to the control of joint dis-
tortion. Under truss mechanism, a large amount of transverse steel is
required in the joint area. On the contrary, under strut mechanism, the
requirement of joint reinforcement reduces substantially. The joint
shear transfer of CL and BL specimens, with identical reinforcement
details, are governed by strut mechanism and truss mechanism re-
spectively, and the joint core of CL specimens is therefore better
H. Yang et al.
(a) Group I
(c) Group III
Fig. 12. Comparison of envelope of joint shear deformation.
confined. In addition, the more diagonal cracks in the BL specimens (see
in Fig. 6) makes the joint core more flexible. Consequently, BL speci-
mens exert larger joint shear deformation.
3.5. Reinforcement slippage
Fig. 14 shows a comparison of the longitudinal beam bar slip en-
velopes for each group at east beam side, with indication of 15% de-
gradation condition (‘‘+” for the pull out direction and ‘‘−” for the
push in direction). It is noted that the stresses of tensile beam bars re-
lated to the beam shear force lead to different bending moment of CL
and BL method. According to Fig. 14, it indicates that the bar slips of BL
specimens considerably exceeded their corresponding CL specimens in
the pull out direction, although the bar slips of CL and BL specimens in
the push-in direction were generally closer to each other.
The slippage of longitudinal reinforcement at four feature points
mentioned previously are presented in Fig. 15. It can be seen that the
bar slips in the pull out direction at the yield, peak, and 15% de-
gradation point of the BL specimens were generally larger than that of
CL specimens, while in the push in direction CL and BL were similar. At
the 15% degradation point, the average bar slips in the pull out di-
rection of CL specimens were 53.0%, 48.4%, 49.5% and 32.7% lower
than their corresponding BL specimens in Group I, II, III and IV re-
spectively. For the pull out direction, the discrepancy of bar slips be-
tween the specimens in different loading methods increased with the
increase of loading displacement.
It is well known that the cyclic loading affects the bond degradation
of reinforcement [26–28], the bond slip increased with the increase of
the number of cycles. Although the loading protocol used for CL method
40
Engineering Structures 166 (2018) 31–45
(b) Group II
(d) Group IV
is more severe than BL method in terms of number of cycles, the
longitudinal bar slip of CL specimens was considerably smaller than
that of BL specimens. This fact effectively demonstrates that the larger
tensile force and bond stress is primarily attributed to the loading
protocol and loading apparatuses, and the BL method enforces the top
and bottom beam bar yielding simultaneously. Therefore, according to
the test results of longitudinal bar slip, the relatively larger tensile force
and bond stress of beam reinforcement results in relatively severe in-
elastic deformations in BL specimens.
It confirms to the experimental observation that the beam plastic
zone was distributed with denser flexural cracks (Fig. 6) under BL
method as a result of the larger tensile force of beam bar. The larger bar
slips in the pull out direction for BL specimens are primarily attributed
to the relatively larger tensile force. Within the width of the column, the
bond degradation of the longitudinal beam bars in BL specimens are
more serious than in CL specimens, suggesting a more severe bond slip
(see Fig. 8). Simultaneously, the yield penetration of the beam long-
itudinal reinforcement in BL specimens leads to relative small bar slip in
the push in direction.
4. Numerical simulation
The numerical modeling was conducted on the OpenSees platform.
The model adopted to simulate the response of the beam-column joint
subassemblages are presented in Fig. 16. The beam and column were
represented by Nonlinear Beam Column Element (NBCE) with three and
four integration points (IP) respectively, and the cross-sections of which
were discretized as Fiber Section (FS). The member sizes, cross-sectional
dimensions and steel reinforcement detailing were determined
H. Yang et al.
Fig. 13. Comparison of joint shear deformation-feature points curve.
according to those in Fig. 2. Concrete01 Material was used to model
confined and unconfined concrete behavior in beam and column, and
Steel02 Material was used to model the uniaxial cyclic behavior of
longitudinal reinforcement. The joint region was simulated by Beam
Column Joint Element (BCJE) proposed by Lowes and Altoontash [29], in
which the PINCHING4 Material with main parameters calculated by
bottle-shaped diagonal strut model and BARSLIP Material were used in
shear-panel and bar-slip springs, respectively [23,25,30].
The simulation model mentioned above was used to simulate the
four groups of specimens in CL and BL specimens. Fig. 17 shows the
comparison between the numerical and experimental load-displace-
ment diagrams of beam-column joint specimens. The numerical model
was able to predict the experimental response of tested specimens.
However, the tangent stiffness and the stiffness of both the unloading
and reloading branches were relatively overestimated, and the pinching
effect was less severe than that in experimental response. The dis-
crepancy between the numerical and experimental results is primarily
attributed to the simplification of the BCJE model, including the sim-
plification of shear-panel and bar-slip spring. The joint core of BCJE
model is simulated by a simplified rectangle shear-panel with the
constitutive relationship described by PINCHING4 Material. In this case,
the reloading and unloading stiffness in each cycle are linear, which is
different from the experimental result. The compressive strength of
diagonal strut exists a certain degree of discreteness, therefore, the ef-
fective compression strength exists partial error when using diagonal
strut model to determine the parameters of shear-panel. In addition,
according to analysis of source routine of the BARSLIP Material, the slip
corresponding to yield strength of reinforcement is relatively small,
41
Engineering Structures 166 (2018) 31–45
which lead to a relatively large initial stiffness.
To compare the local joint inelastic deformations of each test group,
the simulated envelopes of the bottom longitudinal beam bar slips in
pull out direction and the joint shear deformations are presented in
Fig. 18. The simulated deformations in the condition corresponding to
15% degradation of the test is also shown for comparison. Considering
the numerical stability of the global solution algorithm, the envelope
curve of BARSLIP Material conducted by Mitra and Lowes [30] did not
exhibit softening. As a result, decline of longitudinal bar slip occurred
in the last few cycles, which is plotted in Fig. 18(b) by short dash line.
As shown in Fig. 18, the deformation disciplines of the simulation re-
sults were consistent with the test result that both of the joint shear
strains and the bar slips in BL specimens were considerably greater than
in CL specimens. At the peak strength, the average simulated joint shear
stress of positive and negative direction of BL specimens were 5.4%,
7.3%, 8.9% and 5.6% larger than their corresponding CL specimens in
Group I, II, III and IV respectively. At the 15% strength degradation, the
average simulated joint shear strain of positive and negative direction
of CL specimens were 56.6%, 53.6%, 35.0% and 55.7% lower than their
corresponding BL specimens in Group I, II, III and IV respectively. At
the 15% strength degradation, the average simulated bar slip at pull out
direction of CL specimens were 48.6%, 54.3%, 26.6% and 49.9% lower
than their corresponding BL specimens in Group I, II, III and IV re-
spectively.
5. Conclusions
Four groups of interior beam-column joints experiments were
H. Yang et al.
Fig. 14. Comparison of envelope of slippage of beam bar.
conducted to investigate the effect of the loading methods on the joint
seismic behavior. On the basis of the tested results, the following can be
inferred:
1. The damage of the four CL specimens under cyclic loading was
concentrated in the center of the joint core with sparse and wider
diagonal cracks distributed in the joint region, while the damage of
four BL specimens occurred in the entire joint region with dense and
narrow diagonal cracks. During the loading period, the shear force
transfer capacity of longitudinal beam reinforcements in the BL
specimens was greater than that of CL specimens. The crack patterns
observed from the four groups of specimens and the strain dis-
tribution of the longitudinal beam reinforcements indicate that the
contribution of joint shear transfer mechanisms are significantly
affected by loading methods. The comparison between CL and BL
specimens was implemented by comparing the test results at iden-
tical feature points.
2. The yielding of top beam longitudinal reinforcement in the BL
specimens occurred significantly earlier than that in CL specimens,
causing the discrepancy of the tensile stresses of top beam bars in CL
and BL specimens, which in turn led to the variation of the max-
imum joint shear stress between CL and BL specimens. The max-
imum joint shear stress of BL specimens were 8.0–9.6% larger than
that of CL specimens. The joint shear strain response of interior
joints was not significantly affected by the variation of the max-
imum joint shear stress, but was significantly affected by the var-
iation of loading method. Based on the test results, the 15% de-
gradation of the joint shear strength was achieved when the joint
42
Engineering Structures 166 (2018) 31–45
shear strain reached a value between 0.025 and 0.03 rad for CL
specimens and a value between 0.045 and 0.075 rad for BL speci-
mens. At the 15% degradation point, the shear strain of CL speci-
mens were 36.5–68.3% lower than that of BL specimens. This phe-
nomenon is explained by the fact that the diagonal strut mechanism
is dominant in CL specimens and the truss mechanism is dominant
in BL specimens.
3. The longitudinal beam bar slip response was considerably affected
by the variation of loading method. The experimental results show
that the bar slips at the pull out direction of BL specimens were
significantly larger than that of CL specimens, while the bar slips at
the push in direction for CL and BL specimens were similar. At the
15% degradation point, the bar slips in the pull out direction of CL
specimens were 32.7–53.0% lower than that in BL specimens. The
specimens in BL method were subjected to more severe bond de-
gradation and yield penetration after beam yielding.
4. The bar slip and joint shear deformation characteristics of numerical
analysis results were consistent with the experimental results.
Experimental results and numerical analysis show that the influence
of loading method on the seismic behavior of beam-column joint
should be taken into account.
5. The discrepancy of the inelastic deformations of CL and BL speci-
mens is primarily attributed to the loading protocol and loading
apparatuses. The BL method enforces the top and bottom beam bar
yielding simultaneously, which leads to larger compression stress of
diagonal strut and larger bond stress of top beam bar in the joint
region of BL specimens.
6. Although the loading protocol used for CL method is more severe
H. Yang et al. Engineering Structures 166 (2018) 31–45

Fig. 15. Comparison of longitudinal beam bar slippage-feature points curve.

than BL method in terms of number of cycles, the joint shear de-
formation and longitudinal bar slip of CL specimens were con-
siderably smaller than that of BL specimens. In consequence, the
accumulated damage caused by number of loading cycles is a sec-
ondary factor in this research.
7. The seismic behavior of two identical specimens loading in CL and
BL method were considerably different. Thus, the influence of
loading method should be considered or the specimens should be
divided into two groups based on loading methods, when the
seismic behavior of the beam-column joint is evaluated according to
a great number of prior beam-column joint tests.
8. The deflected shape of the subassemblage under CL method is more
realistic, but BL method is easier to perform in laboratory.
Therefore, under the premise of ensuring the boundary condition in
laboratory, CL method is recommended to investigate the seismic
behavior of beam-column subassemblage.

Fig. 16. Simulated model of beam-column joint subassemblages.

43
H. Yang et al. Engineering Structures 166 (2018) 31–45

Fig. 17. Comparison of analytical results with experimental test results.

44
H. Yang et al.
Fig. 18. Envelope curves of inelastic deformations of beam-column joint specimens.
Acknowledgements
The authors would like to express their appreciation for the fi-
nancial support provided by the National Natural Science Foundation of
China under Grant Nos. 51178487 and 51478063.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in the
online version, at http://dx.doi.org/10.1016/j.engstruct.2018.03.022.
References
[1] Sezen H, Whittaker AS, Elwood KJ, Mosalam KM. Performance of reinforced concrete
buildings during the August 17, 1999 Kocaeli, Turkey earthquake, and seismic design and
construction practise in Turkey. Eng Struct 2003;25(1):103–14.
[2] Zhao B, Taucer F, Rossetto T. Field investigation on the performance of building struc-
tures during the 12 May 2008 Wenchuan earthquake in China. Eng Struct
2009;31(8):1707–23.
[3] Ghobarah A, Saatcioglu M, Nistor I. The impact of the 26 December 2004 earthquake and
Tsunami on structures and infrastructure. Eng Struct 2006;28(2):312–26.
[4] Park R, Paulay T. Reinforced concrete structures. New York: John Wiley; 1975.
[5] Kitayama K, Otani S, Aoyama H. Earthquake resistant design criteria for reinforced
concrete interior beam–column joints. In: Pacific conference on earthquake engineering,
Wairakei, New Zealand; 1987. p. 315–326.
[6] Leon RT. Shear strength and hysteretic behavior of interior beam-column joints. ACI
Struct J 1990;87(1):3–11.
[7] Paulay T, Priestley MJN. Seismic design of reinforced concrete and masonry buildings.
New York: John Wiley & Sons; 1992.
[8] Beckingsale CW. Post elastic behavior of reinforced concrete beam–column joints. [Ph.D.
thesis]. Christchurch, New Zealand: Dept. Civil Eng., Univ. Canterbury; 1980.
[9] Park R, Ruitong D. A comparison of the behavior of reinforced concrete beam-column
joints designed for ductility and limited ductility. NZSEE Bull 1988;21(4):255–78.
[10] Hakuto S, Park R, Tanaka H. Seismic load tests on interior and exterior beamcolumn
joints with substandard reinforcing details. ACI Struct J 2000;97(1):11–25.
[11] Chang B, Hutchinson T, Wang X, Englekirk R. Seismic performance of beam-column
45
Engineering Structures 166 (2018) 31–45
subassemblies with high strength steel reinforcement. ACI Struct J 2014;111(6):1329–38.
[12] Alaee P, Li B. High-strength concrete exterior beam-column joints with high-yield
strength steel reinforcements. Eng Struct 2017;145(7):305–21.
[13] Ashtiani MS, Dhakal RP, Scott A. Seismic performance of high-strength self-compacting
concrete in reinforced concrete beam-column joints. J Struct Eng 2014;140(5):472–82.
[14] Li B, Leong CL. Experimental and numerical investigations of the seismic behavior of
high-strength concrete beam-column joints with column axial load. J Struct Eng 2014.
http://dx.doi.org/10.1061/(ASCE)ST.1943-541X.0001191.
[15] Meinheit DF, Jirsa JO. The shear strength of reinforced concrete beam–column joints. In:
Tech. Rep. CESRL Report no. 77-1, Univ. Texas at Austin; 1977.
[16] Yoshioka K, Sekine M. Experimental study of prefabricated beam-column sub-
assemblages. ACI SP-123; 1991. p. 465–92.
[17] Liu A, Park R. Seismic load tests on two concrete interior beam-column joints reinforced
by plain round bars designed to pre-1970s codes. NZSEE Bull 1998;31(3):164–76.
[18] Li B, Yiming W, Pan T-C. Seismic behaviour of non-seismically detailed interior beam-
wide column joints; Part I: experimental results. ACI Struct J 2002;99(6):791–802.
[19] Del Vecchio C, Di Ludovico M, Balsamo A, Prota A, Manfredi G, Dolce M. Experimental
investigation of exterior RC beam–column joints retrofitted with FRP systems. ASCE J
Compos Constr 2014;18(4):1–13.
[20] Ruiz-Pinilla JG, Pallarés FJ, Gimenez E, Calderón PA. Experimental tests on retrofitted RC
beam-column joints under designed to seismic loads. Eng Struct 2014;59:702–14.
[21] Park R. A summary of results of simulated seismic load tests on reinforced concrete beam-
column joints, beams and columns with substandard reinforcing details. J Earthq Eng
2002;6(2):147–74.
[22] Kim J, Lafave JM. Key influence parameters for the joint shear behaviour of reinforced
concrete (RC) beam–column connections. Eng Struct 2008;29(10):2523–39.
[23] Mitra N. An analytical study of reinforced concrete beam–column joint behavior under
seismic loading. [Ph.D. thesis]. Seattle, WA: Dept. Civil Eng., Univ. Washington; 2007.
[24] GB 50010-2002, Code for design of concrete structures. Construction Ministry of China.
[25] ACI Committee 318. Building code requirements for structural concrete and commentary.
Farmington Hills, MI: American Concrete Institute; 2014.
[26] Verderame GM, Ricci P, Carlo GD, Manfredi G. Cyclic bond behaviour of plain bars. Part I:
experimental investigation. Constr Build Mater 2009;23:3499–511.
[27] Ismail MAF, Jirsa JO. Bond deterioration in reinforced concrete subject to low cycle loads.
ACI Struct J 1972;69(6):334–43.
[28] Li J, Gao X, Zhang P. Experimental investigation on the bond of reinforcing bars in high
performance concrete under cyclic loading. Mater Struct 2007;40(10):1027–44.
[29] Lowes L, Altoontash A. Modeling reinforced-concrete beam-column joints subjected to
cyclic loading. J Struct Eng 2003;129(12):1686–97.
[30] Mitra N, Lowes LN. Evaluation, calibration and verification of a reinforced concrete
beam-column joint model. J Struct Eng 2007;133(1):105–20.