Australian Earthquake Engineering Society 2016 Conference, Nov 25-27, Melbourne, Vic

BEHAVIOUR OF RC SPANDREL BEAM IN EXTERIOR WIDE BEAM-

COLUMN CONNECTIONS

H. Behnam(1), J.S. Kuang(2), K. Abdouka(3)

(1)
PhD Candidate, Hong Kong University of Science and Technology, Hong Kong. hbehnam@connect.ust.hk
(2)
Professor, Hong Kong University of Science and Technology, Hong Kong. cejkuang@ust.hk
(3)
Senior Lecturer, Swinburne University of Technology, Melbourne, Australia. kabdouka@swin.edu.au

Abstract

According to architectural and structural needs, heavily loaded and long span moment frames are often built
with the beams wider than the framing column. Wide beam-column connections in this structural typology
play a significant role in determining the overall seismic behaviour of the structures. This paper presents the
results of a set of experiments performed on two full-scale exterior wide beam-column connections. The
specimens have the same dimensions and reinforcement detailing, except for the reinforcement detail in
spandrel beam. The control specimen had both longitudinal and transverse reinforcement within the spandrel
beam while the other specimen had no reinforcement in the spandrel beam. According to the results of the
test, the failure mode in the control specimen was ductile with beam flexural hinging followed by joint and
spandrel beam torsional failure, while it changed to the brittle torsional failure of spandrel beam in the other
specimen. The specimen with no reinforcement in its spandrel beam exhibited brittle torsional failure with an
average reduction of 37% in the wide beam flexural strength capacity compared to the control specimen. In
addition to the experimental study, nonlinear three-dimensional finite element analysis was conducted to
model the behaviour of tested specimens using monotonic loading analysis and to investigate the load
transfer mechanism in wide beam-column connection. The results from both experimental and numerical
investigation indicated that the level of joint shear stresses and the level of spandrel beam torsional stresses
should be controlled to achieve an acceptable and adequate seismic performance.

Keywords: wide beam-column connection; full-scale test; spandrel beam; finite element analysis; joint
shear stresses; torsional stresses.

Introduction
In the areas of moderate to low seismicity, such as Australia, Hong Kong, and the majority of European
countries, reinforced concrete moment resisting frames with wide beam-column connections have been used
extensively. This structural system offers a medium storey height as well as lateral stiffness compared to
reinforced concrete conventional frames and flat-slab structures. In comparison with conventional frames, it
offers lower floor-to-floor height, less congested joint region, simplified formwork, and fast construction at a
Australian Earthquake Engineering Society 2016 Conference, Nov 25-27, Melbourne, Vic

reduced cost. Compared to flat-slab structures, it affords longer spans and more freedom on the column grid
arrangements (LaFave and Wight 1999; Benavent-Climent 2007; Li and Kulkarni 2009; Elsouri and Harajli
2015; Davey et al. 2013). Because of these advantages, there is an increasing trend in adopting this structural
system even in the regions of high seismicity.

A literature review on the wide beam-column connections behaviour can reveal two important features. First,
exterior wide beam-column connections are not only susceptible to joint shear failure but also, they are
extremely vulnerable to torsional failure of the spandrel beam. Second, the torsional behaviour of transverse
beams strongly influences the overall seismic behaviour of wide beam-column connections. (LaFave and
Wight 2001; Benavent-Climent 2009 and 2010; Goldsworthy, and Abdouka 2012; Siah et al., 2003; Stehle et
al., 2001).

Despite these facts, the new wide beam frame buildings are designed and constructed around the world, with
no particular consideration given to the torsional behaviour of the spandrel beams. Furthermore, the majority
of existing wide beam frame buildings were designed and built when the relevant code of practice had no
seismic design provisions. The severe damage and collapse of these buildings in the recent earthquakes are
the results of Inadequate reinforcement detailing. Most of the existing wide beam-column frame building
share the following shortcomings: (1) absence of transverse reinforcement in the joint region; (2) insufficient
bond length of the beam bars; (3) lack of torsional reinforcement in the spandrel beam to sustain torsion; and
(4) substantially wide beam which exceeds the limits prescribed by the current seismic codes (Benavent-
Climent 2007; Fardis 2009; Lopez-Almansa et al. 2013; Dominguez et al. 2016). Therefore, analysis, design,
and detailing of the exterior wide beam-column connections are critical issues that need to be carefully
addressed.

The existing test database on exterior wide beam-column connection is very limited, and it cannot address all
behavioural aspects of the wide beam-column connections. Therefore, modelling is necessary. Nonlinear
three-dimensional finite element model (FEM) is capable of accurately modelling the mechanical properties,
crack formation and propagation, deflections, and possible failure mechanisms in reinforced concrete
members. However, considering the popularity of FEM, relatively few studies have been reported to date of
applying FEM to wide beam-column connections (Li and Kulkarni 2009 and 2010). The most challenging
aspect is how to accurately modelling the concrete in tension and compression and reinforcing bars in the
model. In the past, different theories and formulations for constitutive modelling of concrete have been
developed and implemented in FEM computer softwares. Among the constitutive models for simulating the
behaviour of concrete, the concrete damaged plasticity (CDP) model has proved to provide the most stable
regime. The CDP has the ability to capture the post-peak behaviour with reliable accuracy when compared to
the experimental results (Jankowiak and Lodygowski 2005; Birtel and Mark 2006; Mohamed 2014;
Genikomsou and Polak 2015; Wosatko et al. 2015;).

The objective of this study is to explore the influence of the spandrel beam on the performance of the
exterior wide beam-column connection. Two full-scale exterior wide beam-column connections were tested
to determine the overall seismic performance of the connections and to evaluate the important aspects of
their responses quantitatively. The tests are an extension of our previous research on the seismic performance
of wide beam-column connections, which was part of a program for earthquake resistance structures.
Furthermore, nonlinear three-dimensional FEM was conducted to study the load transfer mechanism within
the connection region. The CDP, which is offered in Abaqus (Abaqus 2013) was adopted for the
representation of concrete. The numerical results are compared to the test results regarding deflections,
strength and crack patterns.
Australian Earthquake Engineering Society 2016 Conference, Nov 25-27, Melbourne, Vic

Experimental program
The experimental program aims at investigating the influence of the spandrel beam on the seismic behaviour
of wide beam-column connections. Figs. 1 and 2 show the geometry and reinforcing details of the tested
specimens, designated as S4, and S5. The specimens representing a portion of the framing system are
obtained by terminating the beam at its mid-span and the columns at their mid-heights, where inflection
points are likely to occur under the lateral loads. Due to the laboratory limitation, the beam span in the frame
was only considered to be 3 meters. Testing beam-column specimens with the shorter beam is a common
practice in this field (Elsouri and Harajli 2015; Kotsovou and Mouzakis). Specimens have an identical beam
length of 1500 mm with a cross-section of 750 mm × 300 mm and the same column height of 3100 mm with
a cross-section of 300 mm × 360 mm. The columns were reinforced with bars of 16 mm diameter (10D16),
amounting to a 1.85% column reinforcement ratio. Specimen S4 contained a shallow depth spandrel beam
with a cross-section of 360 mm × 300 mm. The spandrel beam in this specimen consisted of 8D16
longitudinal bars and having D10 closed stirrup at 70 mm distance (Fig. 1). In the specimen S5, there was no
reinforcement in the spandrel beam. The widths of the beam in both specimens were less than the limit set by
the current ACI Code (ACI 318-14:2014, bb< bc+1.5hc) but exceeds the limit that established by New
Zealand code (NZS3101:2006, bb< bc+0.5hc). The column stirrups were continued through the joint. In both
specimens, three layers of closed stirrups were located between beam top and bottom reinforcement in beam-
column joint. Each layer consisted of four legs of a steel bar with a diameter of 10 mm. Column cross-
section and reinforcement detail are shown in Figs. 1 and 2.

Fig. 1. Dimensions and reinforcement details of specimen S4 (unit: mm).

Australian Earthquake Engineering Society 2016 Conference, Nov 25-27, Melbourne, Vic

Fig. 2. Dimensions and reinforcement details of specimen S5 (unit: mm).

Expected capacities
The concrete compressive strength on the test day was 34.7 MPa and 34.2 MPa, in specimens S4 and S5,
respectively. The average yield stress of the longitudinal reinforcement (D 16 mm) and stirrup (D 10 mm)
were 558 MPa and 511 MPa. The main design parameters of the specimens such as the column (Mn,c) and
beam (Mn,b) nominal flexural capacities, column-to-beam flexural strength ratio (Mr), estimated lateral load
(Vb,e), factored torsion applied to spandrel beam (Tu), nominal torsional moment strength (Tn), joint shear
demand (Vj,e), and joint shear capacity (Vn) were estimated based on the measured properties of materials and
ACI 318-14 criteria. The specimen capacities were calculated using the actual material strength and they are
referred to as expected strengths, as summarised in Table 1.

Table 1 shows the minimum column-to-beam flexural strength ratio (Mr) was 1.53, which satisfies the
“strong column-weak beam” philosophy. The factored torsional moment on the side face of the column, Tu,
was estimated by multiplying the overstrength factor of 1.25 by the applied torsional moment. The ratios of
Tu/Tn in specimens S4, and S5 were 1.56, and 6.63, respectively. ACI 352R-02 requires these ratios to be
lower than one. Therefore, torsional failure of the spandrel beam was expected in both specimens. In
specimen S5, there was no reinforcement in the spandrel beam and the torsional capacity was taken as the
concrete cracking torque.
Australian Earthquake Engineering Society 2016 Conference, Nov 25-27, Melbourne, Vic

Table 1. Main design parameters and expected capacities.

Specimen Mn,c Mn,b Mra Vb,eb Tu Tu/Tnc Vj,ed γ ee γACI
kNm kNm kN kNm kN
S4 187 243.8 1.53 184.7 114 1.56 1292 2.03 1.25
S5 187 243.3 1.54 184.3 114 6.63f 1293 2.05 1
a
M r   M n,c  M n,b  1.2
b
Vb,e  M n,b 1.32
cT
n  2 Ao At f yt cot  s , where Ao= 0.85Aoh=0.85x1y1, θ =45°, and s is the spacing of stirrup in the spandrel beam (s =
70mm).
dV
j ,e  1.25 As f y  Vcol , where As is the beam flexural reinforcement area and Vcol is the column shear force.
e
e  V j ,e Vn  V j ,e  f c bc hc 
f
Tn  Tcr  0.33 2
f c Acp Pcp

Table 1 also shows that the expected normalised joint shear stress ratio (γe) was larger than the ACI
requirement. The γe should be smaller than 1.25, and 1.0 for specimens S4, and S5, respectively. The ACI
code only considers the column cross-sectional area (bc×hc) to resist joint shear stresses. Since the joint shear
force was greater than the joint shear capacity, joint shear failure was expected in all specimens.

Test setup, loading sequence, instrumentation, and measurements

The schematic test setup is shown in Fig. 3. The test setup was arranged on the test floor, and for the
convenience of applying loading and testing, each beam-column assembly was rotated 90° from the true
orientation (column positioned horizontally and beam positioned vertically). The beam tip was linked to an
actuator with a swivel connector to apply the lateral load. Initially, an axial load of 480 kN (12.8% of the
column axial capacity) was applied to the column through a hydraulic jack. The axial force was applied in a
force controlled mode and was maintained constant through the test. Then, the lateral cyclic displacement
(Δ) was applied to the top of the beam in two opposite directions. The lateral load history consisted of
several sets of three cycles with different horizontal displacements amplitudes, as shown in Fig 4. The drift
ratio (%) is expressed as the ratio of the applied displacement to the beam length plus half of the column
depth. The same loading history and test setup were used for both specimens. The similar test setup was
adopted in previous research to study the behaviour of both gravity designed specimens and moment
resistance connections (Hung and Jen 2007, Wong and Kuang 2008, among them). Effect of gravity induced
moment on the beam is important but by using aforementioned test setup gravity moment is not being
incorporated in the test. Since the ultimate capacity of the connection and comparative behaviour of two
specimens was the main objective of this paper, ignoring the gravity induced moment on the beam is
acceptable.

Specimens were instrumented extensively by fixing strain gauges at critical locations identified on the
reinforcement bars to record the magnitude of the reinforcement strains that were developed at different
loading stages. The locations of the strain gauges on the longitudinal and transverse steel bars in the beam
and column are shown in Figs. 2 to 5, as red colour spots. Linear variable displacement transducers (LVDTs)
were used to measure the displacement at various locations on the specimen as shown in Fig. 3.
Australian Earthquake Engineering Society 2016 Conference, Nov 25-27, Melbourne, Vic

Fig. 3. Schematic view of the test setup (unit: mm).

75 5
60 4
45 3
Displacement (mm)

30 2

Drift ratio %
15 1
0 0
-15 -1
-30 -2
-45 -3
-60 -4
-75 -5
0 6 12 18 24 30 36 42 48 54 60
Step Number

Fig. 4. Cyclic displacement history.

Experimental results and observations

The final failure patterns of the specimens were obtained from repeated reverse cyclic loading, as shown in
Fig 5. In the specimen S4, first flexural cracks were formed across the full width of the beam at a 0.2% drift
ratio. In the drift ratio of 1.5%, flexural cracks continued to develop into full width of the beam near the
beam-column interface. At a drift ratio between 2% and 3%, the diagonal shear cracks on the joint tended to
widen, indicating the start of the joint shear failure. Torsional cracks were appeared in the spandrel beams
starting at a drift ratio of 0.75%. These inclined torsional cracks developed around the spandrel beam and
Australian Earthquake Engineering Society 2016 Conference, Nov 25-27, Melbourne, Vic

extended to the intersection region of the column sides and beam at a drift ratio of 3%. The joint shear and
torsional failure of spandrel beam after beam yielding were evident due to the wide-opened diagonal shear
cracks, and the visible expansion of concrete from the joint and the spandrel beam region. In specimen S5,
the first hairline flexural crack was developed in the joint interface section at a drift ratio of 0.2%. At a 1%
drift ratio, the torsional cracks developed rapidly, wrapping around the spandrel beams diagonally. At 2%
drift ratio, the spandrel beam was completely separated from the column face and a complete loss in the
torsion capacity of the spandrel beams occurred, which triggered a considerable reduction in the lateral load
capacity. In the subsequent cycles, the separated spandrel beam started to slide along the shear planes near
the column faces. The crack pattern at the end of the test was characterised by severe torsion cracking in the
spandrel beam near the column face and minor flexural cracks at the beam.

(a) Specimen S4 (b) Specimen S5

Fig 5. crack pattern at 5% drift ratio (end of test)

200 200
160 S4 160 S5
120 120
80 80
Lateral Load (kN)
Lateral Load (kN)

40 40
0 0
-40 -40
-80 -80
-120 -120
-160 -160
Beam capacity limit Beam capacity limit
-200 -200
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
Drift % Drift %

(a) Specimen S4 (b) Specimen S5

Fig. 6. Lateral load-drift hysteretic response.
Australian Earthquake Engineering Society 2016 Conference, Nov 25-27, Melbourne, Vic

Fig 6 illustrates the lateral load versus the drift ratio of the specimens in the form of hysteresis curves.
Specimen S4 reached its theoretical capacity at a drift ratio of 3% in both loading directions. Then, its
strength degraded until the end of the test. The hysteresis loops of this specimen showed considerable
pinching, which was primarily attributed to the formation of diagonal shear and torsion cracking in the joint
and spandrel beam. Specimen S5 experienced significant strength loss compared to the specimen S4 due to
the torsional failure of the spandrel beam. This specimen did not reach its theoretical capacity, while it
attained the maximum capacities of -114.5 kN and 120.8 kN which are approximately 62% and 65% of the
theoretical strength. The strength of this specimen decreased significantly from the following loading cycles
until the end of the test. As shown in Fig. 6(b), there is a certain level of lateral force transfer that appears to
be maintained up to the 5% drift ratio. The residual load-carrying capacities in this specimen were -69.4 kN
and 78.4 kN, which are 38% and 42% of the theoretical capacities. This load level is consistent with the
moment transferred directly from the central part of the beam to the column.

The test results showed that the behaviour of the spandrel beam strongly affects the overall load-
displacement relationship, crack pattern, joint shear capacity and failure mode of the specimens. The
maximum torsion applied to the spandrel beam was determined by measuring the strain in the beam
longitudinal reinforcement, which was 84, and 39.7 kNm in the specimens S4, and S5, respectively. The
ratio of the maximum applied torsion to the spandrel beam torsional capacity was 1.15, and 2.30,
respectively. In the specimen S5, the maximum torsion was 57% larger than the spandrel beam cracking
torque capacity, and the specimen failed in torsion. In the specimen S4, the actual torsion acting on the
spandrel beam was almost 15% larger than the spandrel beam capacity. However, it exhibited a very severe
torsion cracking and concrete crushing in the spandrel beam. The crushing of the concrete occurred because
basically, the cross section of the spandrel beam was not enough to resist applied torsion. According to the
ACI 352R-02, in the exterior wide beam-column connections, the spandrel beam should be designed for full
equilibrium torsion from the beam and slab bars. However, before determining the required transverse and
longitudinal reinforcement for the torsion the adequacy of the cross-section for torsion must be checked.
Cross-sectional dimensions are limited to help reduce the unsightly cracking and to prevent crushing of the
inclined concrete compression struts due to the shear and torsion.

Finite element simulations

To further enhance the understanding of the complicated behaviour of the wide beam-column connections, a
nonlinear finite element analysis was carried out. In the numerical study, the geometrical properties of the
model were kept similar to the test specimens. Eight-noded hexahedral (brick) elements (C3D8R) were used
for concrete with reduced integration to avoid the shear locking effect. Two-noded linear truss elements
(T3D2) were used to model reinforcements. The embedded method was adopted to simulate the bond
between the concrete and the reinforcement which was assumed perfect. After carrying out mesh sensitivity
study, the mesh size of 30 mm was adopted in all models. Analogous to the test condition, a loading was
defined in two steps. First, the column axial load was applied up to the required level. In the second phase, a
lateral displacement was applied in the beam end. Static analysis in Abaqus/Standard with viscosity
regularisation was performed.
Australian Earthquake Engineering Society 2016 Conference, Nov 25-27, Melbourne, Vic

Material models
Concrete Damaged Plasticity (CDP) model (Lubliner et al. 1989; Lee and Fenves 1998) is used as the
constitutive model for concrete material. The concrete material parameters used as input variables in the
CPD model are the density, modulus of elasticity Ec, Poisson’s ratio v, compressive and tensile strengths of
the concrete and damage parameter in compression and tension. The concrete behaviour in tension is
characterised by a stress-crack displacement response, as shown in Fig. 7, where ft is the maximum tensile
strength and Gf denotes the fracture energy of concrete that represents the area under the tensile stress-crack
displacement curve. Adopting stress-crack width displacement based on the fracture energy can prevent
mesh-sensitivity and allow for numerical convergence. The fracture energy of concrete Gf (N/m) for ordinary
normal weight concrete can be obtained from CEB-FIP Model Code 2010.

3.2

f't 2.8

Gf = area under the

Tensile Stress

2.4

stress-crack opening
relation
σt

1.6

1.2

0.8

0.2f't 0.4

Crack width, δ
0
0 0.0 5 0.1 0.1 5 0.2

w1=Gf/f't wu=5Gf/f't

Fig. 7. Uniaxial tensile stress-crack width relationship for concrete.

The concrete stress-strain behaviour under compressive loading was modelled in three phases (Fig. 8).
Equations for the assumed compressive stress-strain diagram were suggested by (Birtel 2006) and are given
in Eqs. (1a, b, and c) by slight modifications.

 c(1)  Ec c ,  c  0.4 fc Ec (1a)

2
c  c 
c  
0  0 
 c (2)  f , 0.4 f c Ec   c  0.0035
 c
1  c  2  c (1b)
0
1
 2   f c 0  c2 
 c (3)     c   , 0.0035   c   u (1c)
 2 fc 2 0 
Australian Earthquake Engineering Society 2016 Conference, Nov 25-27, Melbourne, Vic

Eq. (1a) represents the linear-elastic branch in which εc is a variable changing from zero to 0.4fc/Ec, and Ec is
the initial modulus of elasticity. The linear branch ends at the stress level of 0.4fc. Eq. (1b) describes the
ascending branch up to the strain level of 0.0035. The corresponding strain level at the peak stress is defined
as ε0 = 2fc/Ec; ηc is the material constant. The stress and strain compatibility at the strain level of εc = 0.4fc/Ec,
for Eq. (1a) and (1b) gives the value of ηc. Eq. (1c) shows the third and descending branch; λ is the constant
crushing energy as a material property. Using the stress and strain compatibility at the strain level of εc =
0.0035, for Eq. (1b) and (1c) enables the value of λ to be determined. Damage parameters were introduced
in the CPD model in tension and compression according to Figs. 9 and 10, respectively (Genikomsou and
Polak (2015)). The uniaxial stress-strain relation of the reinforcement is modelled as elastic with Young’s
modulus (Es) and Poisson’s ratio (v) having typical values of 200,000 MPa and 0.3, respectively. Table 2
summarises the main concrete parameters used in the model.

40
1 2 3

fc 35
Compressive stress, σc

30

25

20

15

0.4fc
10

0.2fc 5

0.0035
0
0 0.0 05 0.0 1 0.0 15 0.0 2 0.0 25 0.0 3

ε0=2fc/Ec Strain εc

Fig. 8. Uniaxial compressive stress-strain relationship for concrete.

1 1
0.95
Compressive damage parameter

0.95
0.8 0.8
Tensile Damage parameter

0.6 corresponding to 0.6 corresponding to

σt = 0.2ft in σc = 0.2fc in
dc

descending branch
dt

0.4 0.4 descending branch

0.2 0.2

0 0
0 w1 0.05 0.1 0.15 0 0.01 0.02 0.03 0.04
wu
Tensile displacement: mm ε0 Compressive strain εu

Fig. 9. Tensile damage parameter-strain Fig. 10. Compressive damage parameter-strain

relationship for concrete. relationship for concrete (simplified in linear form).
Australian Earthquake Engineering Society 2016 Conference, Nov 25-27, Melbourne, Vic

Table 2. Properties of concrete.

Specimen fc Eca ε0b ftc Gf Poisson’s Density
MPa MPa MPa N/mm Ratio (v) tonne/mm3
S4 34.7 27700 0.00251 2.8 0.083 0.2 2.4E-009
S5 34.2 27500 0.00248 2.8 0.083 0.2 2.4E-009
Ec  4700 f c  MPa  , Module of elasticity
a

b
 0  2 f c Ec ,
c
measured using four-point loading tests on test day.

In addition to defining the material parameters, the following data were also provided.

 σb0/σc0 is the ratio of biaxial compressive to uniaxial compressive yield stress, set to 1.16;
 Kc is the coefficient determining the shape of the deviatoric cross-section, configured toa value of 0.667;
 ψ is the dilatation angle, physically interpreted as the internal friction angle of concrete, taken as 40;
 ε is the potential flow eccentricity, assumed with a default value of 0.1; and
 μ is the viscosity parameter used for the viscoplastic regularisation of the concrete constitutive equations
in the Abaqus/Standard analyses, taken as 0.00001.

Comparison of FEA predictions with experimental results

Fig. 11 presents the analysis results together with the experimental hysteretic loops in terms of load-
displacement. As illustrated in this figure, the lateral force-displacement curve predicted by the FEA follows
most of the experimental curve closely. Both results show similar post-peak softening behaviour.

200 200

150 S4 150 S5
100 100
Lateral Load (kN)
Lateral Load (kN)

50 50

0 0

-50 -50
Cyclic loading Cyclic loading
-100 test -100 test
Monotonic Monotonic
-150 loading FEA -150 loading FEA

-200 -200
-90 -75 -60 -45 -30 -15 0 15 30 45 60 75 90 -90 -75 -60 -45 -30 -15 0 15 30 45 60 75 90
Displacement mm Displacement mm

(a) Specimen S4 (b) Specimen S5

Fig. 11. Load-displacement from experiment and numerical analysis.
Australian Earthquake Engineering Society 2016 Conference, Nov 25-27, Melbourne, Vic

Fig. 12 shows the cracking pattern on the tension side of the beam at 5% drift, from the FEA. The obtained
FEA crack patterns were similar to the cracks that observed during the test and shown in Fig 5. The
simulated response of the test specimens was in good agreement with the results found from the experiment.
The modelled responses verified the ability of the selected material parameters and constitutive models to
capture the behaviour of the connections up to the end of the tests. The ratio of the predicted maximum
lateral force in the beam to that of the test results was within an error of ± 9%. The presented analyses
indicate that the proposed model can simulate the nonlinear behaviour of RC wide beam-column
connections, and it can be used for parametric studies on various aspects influencing the behaviour of wide
beam-column connections.

(a) Specimen S4 (b) Specimen S5

Fig. 12. Cracking pattern on tension surface at 5% drift

In addition to the crack pattern, the stress distribution in tension reinforcement was compared. Beam bar
yielding for the beam top reinforcement anchored inside the column core occurred in the 1.67%, and 1.5%
drift cycles in the specimens S4, and S5, respectively. Beam bars anchored outside the column core of the
specimen S5 never yielded. The entire beam bar yielded at a 2.5% drift ratio in specimens S4. In the
specimens S4 joint shear failure and spandrel beam, torsional failure occurred at a drift ratio of 3%.
Specimen S5 failed in a brittle torsional mode. The FEA showed a same yielding pattern as the
reinforcement.

Figs. 13 and 14 represents the concrete compressive stress trajectories for tested specimens. Fig 13a and 14a
clearly show that the primary load transfer mechanism through and around the joint is the diagonal strut
mechanism. Fig 13b shows the stress distribution of the portion of wide beam outside the column width is
non-uniform. Part of the loading in this region is transferred to the beam-column joint directly through the
diagonal paths. The remaining portion of loading passes towards the transverse beams first and then transfers
to the joint core through torsion. The concrete compressive struts cannot develop in outside portion of the
beam in specimen S5, because of the lack of the spandrel beam reinforcement.
Australian Earthquake Engineering Society 2016 Conference, Nov 25-27, Melbourne, Vic

Diagonal strut
mechanism

Compressive struts
Directly to column
core

Compressive struts
Going through
spandrel beam

(a) Side view (b) Top view

Figure 13. Load transfer paths in concrete in specimen S4

Diagonal strut
mechanism

Compressive struts
Directly to column
core

(a) Side view (b) Top view

Figure 14. Load transfer paths in concrete in specimen S4

Australian Earthquake Engineering Society 2016 Conference, Nov 25-27, Melbourne, Vic

From the observed performance of the specimens and also from the results of the numerical modelling, it can
be concluded that the spandrel beam has a vital effect on the seismic behaviour of the exterior wide beam-
column connections. Providing both longitudinal and transverse reinforcements in spandrel beam is
necessary to resist torsion and to develop adequate concrete struts. According to the results of the test and
numerical analysis, it is suggested that spandrel beam should be designed for full equilibrium torsion.
However, before determining the required transverse and longitudinal reinforcement for the torsion the
adequacy of the cross-section for torsion must be checked. Cross-sectional dimensions are limited to help
reduce the unsightly cracking and to prevent crushing of the inclined concrete compression struts due to the
shear and torsion

Conclusions

An experimental and numerical investigation of two full-scale exterior reinforced concrete wide beam-
column connections was carried out to study the influence of the spandrel beam on the overall behaviour of
wide beam-column connections. The spandrel beam in specimen S4 included both longitudinal and
transverse reinforcements, but there was no reinforcement within the spandrel beam of the specimen S5. The
following conclusions can be drawn.

1- Beam bar yielding for the beam top reinforcement anchored inside the column core occurred in the
1.67%, and 1.5% drift cycles in the specimens S4, and S5, respectively. Beam bars anchored outside
the column core of the specimen S5 never yielded due to the torsional failure of spandrel beam. The
entire beam bar yielded at a 2.5% drift ratio in specimens S4.
2- The failure mode of the specimens S4 was beam flexural hinging followed by joint shear and
spandrel beam torsional failure. However, specimen S5 without reinforcement in its spandrel beam
failed in a brittle torsional mode with an average strength loss of 37%.
3- The ratio of the predicted maximum lateral force in the beam to that of the test results was within an
error of ± 9%. The presented analyses indicate that the proposed model can simulate the nonlinear
behaviour of wide beam-column connections.
4- The numerical analysis clearly showed that the primary load transfer mechanism through and around
the joint is the diagonal strut mechanism. Therefore, the joint region should be confined by stirrups
(similar to the conventional joints).
5- The presence of the spandrel beam’s reinforcement altered the orientations of concrete diagonal
struts in the spandrel beams.
6- According to the results of the test and numerical analysis, providing both longitudinal and
transverse reinforcement within the spandrel beam is necessary to achieve adequate seismic
performance. Therefore, it is suggested to design the spandrel beam for full equilibrium torsion.
However, before determining the required transverse and longitudinal reinforcement for the torsion
the adequacy of the cross-section for torsion must be checked.

Acknowledgments
The support of the Hong Kong Research Grants Council (HKRGC) under Grant Number 16209115 is
gratefully acknowledged.
Australian Earthquake Engineering Society 2016 Conference, Nov 25-27, Melbourne, Vic

References

ABAQUS. Analysis user’s manual 6.10. (2013), Dassault Systems Simulia Corp. SIMULIA, Providence, IR.
American Concrete Institute (ACI) Committee. (2014). “Building code requirements for structural concrete
(ACI 318-14) and commentary.” ACI 318-14, Farmington Hills, MI.
American Concrete Institute (ACI)-ASCE Committee. (2002). “Recommendations for design of beam-
column joints in monolithic reinforced concrete structures.” ACI 352R-02, Farmington Hills, MI.
Benavent-Climent, A. (2007). “Seismic behaviour of RC wide beam-column connections under dynamic
loading.” J. Earthquake Eng., 11(4), 493-511.
Benavent-Climent, A., Cahis, X., Zahran, R. (2009). “Exterior wide beam-column connections in existing
RC frames subjected to lateral earthquake loads.” Eng. Struct., 31(7), 1414-1424.
Benavent-Climent, A., Cahis, X., Vico, J. (2010). “Interior wide beam-column connections in existing RC
frames subjected to lateral earthquake loading.” Bulletin of Earthquake Engineering, 8(2), 401-420.
Birtel V, Mark P. (2006). “Parameterised finite element modelling of RC beam shear failure,”:95-108.
CEB-FIP Model. (1990). “CEB-FIB Model Code 1990: Design code.” Bulletin d’Information 195, London:
Thomas Telford.
Davey, M., Abdouka, K., Al-Mahaidi, R. (2013). “Post-tensioned band beams as moment resisting frames
under earthquake loading: a state-of-the-art review.” Aust. J. of Struct. Eng., 14(3), 193-205.
Dominguez, D., Lopez-Almansa, F., Benavent-Climent, A. (2016). “Would RC wide-beam buildings in
Spain have survived Lorca earthquake (11-05-2011)?” Eng. Struct., 108, 134-154.
Elsouri, A., and Harajli, M. (2015). “Interior RC wide beam-narrow column joints: potential for improving
seismic resistance.” Eng. Struct., 99, 42-55.
Fardis, M. N. (2009). Seismic Design, Assessment and Retrofitting of Concrete Buildings: Based on EN-
Eurocode 8, Springer, London.
Genikomsou AS, Polak MA. (2015). “Finite element analysis of punching shear of concrete slabs using
damaged plasticity model in ABAQUS.” Eng Struct, 98:38-48.
Goldsworthy, H. M., and Abdouka, K. (2012). “Displacement-based assessment of non-ductile exterior wide
band beam-column connections.” J. Earthquake Eng., 16(1), 61-82.
Hung-Jen, L., & Jen-Wen, K. (2007). “Eccentric reinforced concrete beam-column connections subjected to
cyclic loading in principal directions.” ACI structural journal, 104(4), 459.
Jankowiak T, Lodygowski T. (2005). “Identification of parameters of concrete damage plasticity constitutive
model.” Foundations of civil and environmental engineering., 6:53-69.
Kotsovou, G., and Mouzakis, H. (2012). “Exterior RC beam–column joints: new design approach.” Eng.
Structures, 41, 307-319.
LaFave, J. M., and Wight, J. K. (2001). “Reinforced concrete wide-beam construction vs. conventional
construction: resistance to lateral earthquake loads.” Earthquake Spectra, 17(3), 479-505.
LaFave, J. M., and Wight, J. K. (1999). “Reinforced concrete exterior wide beam-column-slab connections
subjected to lateral earthquake loading.” ACI Struct. J., 96(4), 577-585
Australian Earthquake Engineering Society 2016 Conference, Nov 25-27, Melbourne, Vic

Lee J. and Fenves G.L. (1998). “Plastic-damage model for cyclic loading of concrete structures”. Journal of
Engineering Mechanics, 124:8, 892–900.
Li, B., and Kulkarni, S. A. (2009). “Seismic behaviour of reinforced concrete exterior wide beam-column
joints.” J. Struct. Eng., 136(1), 26-36.
Lopez-Almansa, F., Dominguez, D., Benavent-Climent, A. (2013). “Vulnerability analysis of RC buildings
with wide beams located in moderate seismicity regions.” Eng. Struct., 46, 687-702.
Lubliner J., Oliver J., Oller S. and Onate E. (1989). “A plastic-damage model for concrete”. International
Journal of Solids and Structures, 25:3, 229-326.
Mohamed AR, Shoukry MS, Saeed JM., (2014). “Prediction of the behavior of reinforced concrete deep
beams with web openings using the finite element method.” Alexandria Eng Journal, 53:329-39.
Siah, W., Stehle, J., Mendis, P., Goldsworthy, H. (2003). “Interior wide beam connections subjected to
lateral earthquake loading.” Eng. Struct., 25(3), 281-291.
Stehle, J. S., Goldsworthy, H., Mendis, P. (2001). “Reinforced concrete interior wide-band beam-column
connections subjected to lateral earthquake loading.” ACI Struct. J., 98(3), 270-279.
Wong, H. F., and Kuang J. S. (2008). “Effects of beam-column depth ratio on joint seismic
behaviour.” Proceedings of the Institution of Civil Engineers-Structures and Buildings 161(2), 91-101.
Wosatko A, Pamin J, Polak MA., (2015), “Application of damage-plasticity models in finite element
analysis of punching shear.” Comput Struct, 151:73-85.