Engineering Structures 191 (2019) 575–582

Contents lists available at ScienceDirect

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

Experimental studies on reinforced concrete interior beam-column joints T

strengthened by unsymmetrical chamfers
⁎
Eddie Siu-shu Lama, , Bo Lib, Zhi-hang Xuea, Kwok-tung Leungc, Jeffery Yuet-kee Lamd
a
The Hong Kong Polytechnic University, Hong Kong
b
University of Nottingham Ningbo China, China
c
Architectural Services Department, HKSAR Government, Hong Kong
d
Nano and Advanced Materials Institute Limited, Hong Kong

A B S T R A C T

Beam-column joints (“BCJ”) are critical members in low-rise to mid-rise buildings. Failure of BCJ may lead to collapse of building. To strengthen non-seismically
designed BCJ, a comprehensive study has been commissioned to develop a strengthening strategy for BCJ using unsymmetrical chamfers. Specifically, chamfers are
proposed to be installed at beam-column corners and under the soffit only to alleviate the undesirable joint-shear failure. Tests were carried out on four 2/3-scale
BCJ, comprising one control specimen and three strengthened specimens including 300 mm chamfers with and without U-bars and 150 mm chamfers with U-bars.
The specimens were subjected to moderate level of axial load and loaded to failure under quasi-static cyclic loading. It has been shown that unsymmetrical chamfers
are effective to protect a non-seismically designed BCJ against failure at joint core. Mode of failure is shifted from joint-shear in the control specimen to column-
flexure in the strengthened specimens. To enhance the performance of BCJ, size of chamfer is more crucial in comparison with providing reinforcements inside the
chamfers.

1. Introduction shear failure [21,22]. Li et al. [12] has demonstrated that chamfers can
be provided without reinforcements and that external BCJ with one
Beam-column joints (“BCJ”) are crucial members in moment-re- chamfer (i.e. unsymmetrical chamfer) exhibited compatible perfor-
sisting structures taking up more forces than the adjacent structural mance to that of two chamfers. In this study, the method of strength-
members. It has been repeatedly reported in post-earthquake re- ening BCJ by installing chamfers unsymmetrically at soffit of beams is
connaissance that structures failed under earthquake due to insufficient further developed. As shown in Fig. 1, the strengthening strategy has
joint shear reinforcements [1,2]. In areas with no seismic provisions the advantages of minimum disruption to building layout and ease of
[3,4], BCJ are designed to gravity loads and wind load only. Thus, joint construction.
shear reinforcements are not required. Joint aspect ratio may also In this study, four BCJ specimens, including one control specimen
dominate the performance of BCJ [5]. Examples can be found in low- and three specimens strengthened using chamfer, were constructed and
rise buildings, like hospitals, facilities for public transportation, police tested to failure under cyclic horizontal displacement. All specimens
stations, fire stations, etc. in Hong Kong designed to the pre-2004 were subjected to constant axial load N applied to the column with N/
concrete code. To prepare for possible moderate seismic action, this Agfcu = 0.25, where Ag is gross area of column and fcu is cube strength
class of BCJ has to be strengthened. In this study, a method of of concrete. Performance was assessed by modes of failure, energy
strengthening of BCJ is proposed with due consideration of minimum dissipation, stiffness degradation and strains in chamfers.
impact to use of space and ease of construction.
Strengthening of reinforced concrete BCJ has always been a topic 2. Experiment program
attracting a lot of research interest. Examples include, concrete jack-
eting [6], shotcrete jacketing [7], FRC jacketing [8], steel jacketing 2.1. Method of strengthening
[9,10], ferrocement jacketing [4,11,12], FRP wrapping [13,14], com-
bined use of the above [15], etc. Alternatives include metallic haunches As shown in Fig. 2, chamfers were installed at the soffit of beams.
[16], passive energy dissipation devices [17], shape memory alloys Chamfer size Lc was based on the least dimension of beam depth and
[18], pre-stressed steel angles [19] and prestressed steel L-profiles [20]. column width. In the case of chamfers with reinforcements, R6 U-bars
Recently, chamfer expansion was successfully used to suppress joint at 50 mm spacing were installed. Three strengthening schemes are

⁎
Corresponding author.
E-mail address: [email protected] (E.S.-s. Lam).

https://doi.org/10.1016/j.engstruct.2019.03.099
Received 5 March 2018; Received in revised form 21 March 2019; Accepted 26 March 2019
Available online 03 May 2019
0141-0296/ © 2019 Elsevier Ltd. All rights reserved.
E.S.-s. Lam, et al. Engineering Structures 191 (2019) 575–582

Table 1
Flexural and compressive strength of mortar.
Specimen Compressive strength (Mpa) Flexural strength (MPa)

IJ-C150R 44.8 6.8

IJ-C300R 42.5 6.3
IJ-C300WR 54.0 7.5

of each specimen was estimated by testing 100 mm cubes at the date of

testing. The results are 53.2 MPa, 65.9 MPa, 57.3 MPa and 67.1 MPa for
the respective specimens IJ-NC, IJ-C150R, IJ-C300R and IJ-C300WR.

2.3. Test setup and loading sequence

Fig. 4 shows the test setup of a multi-purpose testing system [11].

Beams were connected to the strong floor through two hinges one at
each end to restrain the vertical displacement. Bottom column was
Fig. 1. Illustration of chamfers installed in the structure. hinged to the strong floor. Studies by Yang et al. [23] have indicated
that the above loading arrangement (loading at upper column) pro-
vided more realistic representation than loading at the beam ends. Axial
considered and are identified as C150R, C300R and C300WR. Here, load and horizontal displacement/force at upper column were provided
C150 and C300 represent 150 mm and 300 mm chamfer size Lc, re- by two actuators and monitored by two load cells and one LVDT. Strain
spectively. Chamfers with and without reinforcements are represented gauges were installed on reinforcements close to BCJ and on the sur-
by R and WR respectively. face. A pair of LVDTs was installed diagonally on BCJ to monitor the
To strength a BCJ, concrete cover of BCJ and a length equal to shear deformation. Four pairs of LVDTs were installed on the upper
chamfer size was removed. After preparing the exposed surface by an column to estimate the curvature.
impact hammer, R6 U-bars were installed. After formwork was erected, All specimens were subjected to moderate level of axial load and
mortar was applied and compacted by hammering. Formwork was loaded to failure under quasi-static cyclic loading according to the
dismantled 24 h after casting and chamfers were air-cured. loading protocol as shown in Fig. 5. Firstly, axial load of 0.25 fcuAg was
Mortar with water-cement ratio of 0.45 was used to casting the applied vertically at the upper column and kept constant throughout
chamfers. Mix proportion is 1:0.005:2.5:0.45:0.016 for cement:silica the test. Here, fcu is compressive strength of concrete and Ag is gross
fume:sand:water:superplasticizer. Compressive strength and flexural cross-sectional area of the column. Horizontal load was subsequently
strength of mortar were estimated by testing 100 × 100 mm cubes and applied to reach 75% of moment capacity of the beam. The corre-
40 × 40 × 160 mm prisms at 28 days. The results are shown in Table 1. sponding horizontal displacement was defined as Δ0.75. Yield dis-
placement is Δy = Δ0.75/0.75 based on linear extrapolation. After-
wards, the horizontal displacement was applied at increments of 0.5Δy
2.2. The specimens
at pre-peak stage and Δy at post-peak stage, respectively. Cyclic loading
was repeated twice at each displacement increment. The test termi-
Dimension and reinforcement details of the specimens are given in
nated when horizontal load reduced to 85% of the maximum. Post-peak
Fig. 3. The specimens represent an interior BCJ of a sub-frame with
displacement at 85% peak value is regarded as the ultimate deforma-
points of contra-flexure at the supports. Cross sections of columns and
tion.
beams are 300 mm × 300 mm and 275 mm × 400 mm
(breadth × depth) respectively. Transverse reinforcement is not pro-
vided in the joint zone. Concrete cover is 25 mm to stirrups. Measured 3. Experimental result
strength of reinforcements are listed in Table 2.
All specimens were air-cured after demoulding. Ready-mixed con- In what follows, drift ratio is the ratio of horizontal displacement at
crete at 150 mm slump was used. Compressive strength of concrete fcu the tip of upper column to distance between the supports of columns or

Fig. 2. The strengthening schemes: (a) Specimen IJ-300R; (b) Specimen IJ-150R; (c) Specimen IJ-300WR.

576
E.S.-s. Lam, et al. Engineering Structures 191 (2019) 575–582

Fig. 3. Specimen IJ-NC (unit: mm).

Table 2
Measured material properties of reinforcement.
Description Reinforcement Yield strength Ultimate strength
(MPa) (MPa)

High-yield deformed T25 557.1 662.6

bars T20 586.0 684.6
Mild steel bars R8 387.9 462.9
R6 461.0 528.2

2600 mm as shown in Fig. 4. Fig. 6 presents the crack pattern of beam-

column joints specimens at the peak load.

3.1. General observations and failure modes

Specimen IJ-NC Flexural cracks were first observed in the beams

and several vertical cracks then appeared in the joint. At a drift ratio of
0.82%, diagonal cracks appeared in the joint. With progressive in-
creased in the drift ratio, more flexural cracks emerged on the beams
and the existing cracks propagated. Shear cracks intersected at centre of
the joint. Only three cracks were observed on the lower column
throughout the loading history due to the effect of axial load. At peak
horizontal force, diagonal cracks fully developed in the joint and the
specimen failed by joint shear. At the post-peak stage, repeated opening
and closing of diagonal cracks on the joint further accelerated the
spalling of joint cover.
Specimen IJ-C150R The first flexural crack was found at the
bottom of beam. When the drift ratio was increased to 0.39%, cracks
Fig. 4. Test setup.
developed at the corners of chamfers and propagated into the joint.
When approaching the peak horizontal force, there was spalling of
mortar at upper column. When the drift ratio reached 2.17%, com- When the drift ratio reached 0.72%, diagonal cracks intersected with
pressive failure of upper column and likely failure at corners of chamfer each other in the joint. As the drift ratio increased, cracks at the corners
were observed. of chamfers propagated and intersected with that in the joint. When the
Specimen IJ-C300R Flexural cracks were first observed at the drift ratio reached 1.7%, longitudinal cracks appeared at the upper
beams and at the corners of chamfers. With increasing drift ratio, column and this was followed by spalling of mortar. The specimen
flexural cracks increased and propagated towards the neutral axis. exhibited obvious failure of column and some deterioration of

577
E.S.-s. Lam, et al. Engineering Structures 191 (2019) 575–582

peak horizontal force, spalling of mortar was observed at the upper

column and cracks extended into the joint. The specimen failed by a
combination of column failure, flexural failure of beams and dete-
rioration of chamfers.

3.2. Hysteretic behavior

Recognising the difference in concrete strength between the speci-

mens, adjustment is made to the horizontal force acting on specimen X
by the following equation.

Equivalent horizontal force = horizontal force of specimen X

× fcu, NC / fcu, X (1)

here, fcu,X and fcu,NC are compressive strength of concrete of specimen X

and specimen IJ-NC, respectively.
Fig. 7 shows the horizontal force against lateral displacement at the
upper columns for each specimen. Specimens IJ-C300R and IJ-C300RW
achieved highest peak horizontal force and experienced slower degen-
Fig. 5. Loading protocol.
eration of stiffness at post-peak stage. This resulted in improving the
loading capacity of BCJ, enhancing the stiffness and reducing the peak
chamfers. deformation.
Specimen IJ-C300WR Compared to specimen IJ-300R, cracks ex- Specimen IJ-C300WR performed better than specimen IJ-C150R in
tended more rapidly in the chamfers due to the absence of U-bars in the reaching higher peak horizontal force even though there was no U-bars
chamfers and relatively more cracks were observed. Cracks originated in the chamfers of the former. This suggests that chamfer size Lc is a
from the corners of chamfers propagated along the bottom of beams. At dominant factor affecting the performance of a strengthened BCJ. As

Fig. 6. BCJ of specimens at peak horizontal load: (a) Specimen IJ-NC; (b) Specimen IJ-C150R; (c) Specimen IJ-C300R; (d) Specimen IJ-C300WR.

578
E.S.-s. Lam, et al. Engineering Structures 191 (2019) 575–582

Fig. 7. Horizontal load against horizontal displacement (a) Specimen IJ-NC; (b) Specimen IJ-C150R; (c) Specimen IJ-C300R; (d) Specimen IJ-C300WR.

control specimen suffered progressive loss of stiffness whereas

strengthened specimens were able to maintain the stiffness. Table 3
shows peak values of equivalent horizontal force achieved by the spe-
cimens. Loading capacities increase with increasing LC/LBC ratio. Here,
LBC is the width of column. Specimen IJ-C150R performed better than
control specimen IJ-NC whereas highest loading capacities were
achieved by specimens IJ-C300WR and IJ-C300R.

3.3. Displacement ductility

Displacement ductility of each specimen is summarized in Table 4

based on envelops of hysteretic curves as shown in Fig. 8. Here, dis-
placements are mean values obtained from positive and negative en-
velops. Ductility µ is Δu/Δy. Δu is the ultimate displacement at 85%
loading capacity in the descending branch and Δy is the yield dis-
placement estimating by energy balancing method and secant method
based on general yielding [11].
Fig. 8. Envelopes of the hysteretic loops. Here, the value of displacements for each specimen are obtained
from both positive and negative. The ductility of test specimens is
Table 3 summarized in Table 4 based on the envelops of hysteretic curve as
Peak values of equivalent horizontal force (kN). shown in Fig. 8. Ductility µ is defined as the ration of ultimate dis-
Specimen IJ-NC IJ-C150R IJ-C300R IJ-C300WR
placement Δu to yielding displacement Δy, in which ultimate displace-
ment Δu is the displacement when loading capacity drop to 85% and Δy
LC/ LBC ratio – 0.5 1.0 1.0 is the yield displacement estimating by energy balancing method and
Pull → 99.3 127.2 134.6 141.9 secant method based on general yielding [11]. Energy balance method
Push ← −100.4 −101.5 −139.4 −129.6
Average 99.9 114.3 137.0 135.8
is based on adjusting the secant line to make area A1 equal to A2 as
Improvement – 14.4% 37.1% 35.9% shown in Fig. 9(a). The secant line starts from point O passing point I
and intersects at peak strength point A. The displacement at point A is
estimated as the yielding displacement. The general yielding method, as
compared with chamfer size Lc, chamfers with or without reinforce- shown in Fig. 9(b), is a tangent line starting from Point O and inter-
ment is of secondary importance. secting at peak strength Point H. Then vertical line passing Point H
Fig. 8 shows envelopes of the hysteretic loops of specimens against insects the load-displacement at Point I. The secant line starts from
equivalent horizontal force. At the initial stage, the envelopes are close Point O passing Point I and insects at peak load Point A. The dis-
to each other. When the equivalent horizontal force exceeds 70 kN, placement at Point A is estimated as the yielding displacement.

579
E.S.-s. Lam, et al. Engineering Structures 191 (2019) 575–582

Table 4
Summary of displacement ductility.
Specimen Balance of energy General yielding

Ultimate displacement Δu (mm) Yielding displacement Δy (mm) Ductility µ Yielding displacement Δy (mm) Ductility µ

Push← Pull→

IJ-NC −67.13 73.65 24.54 3.0 15.82 4.66

IJ-C150R −64.87 77.44 24.25 3.19 16.47 3.94
IJ-C300R −64.67 75.27 25.41 2.96 17.05 3.79
IJ-C300WR −87.16 70.63 27.49 3.17 25.15 3.47

Fig. 9. Definitions of yield displacement using (a) energy balance method, and (b) general yielding method.

Fig. 10. Cumulative energy dissipation for each specimen.

Fig. 11. Stiffness degradation versus horizontal displacement.
3.4. Energy dissipation
influencing the performance of a strengthening scheme using chamfers.
Energy dissipation at each cycle is calculated from each load-dis-
placement loop. Cumulative energy dissipation is computed by sum-
mating the energy dissipation in previous cycles. Fig. 10 shows cumu- 3.5. Stiffness degradation
lative energy dissipation of each specimen against horizontal
displacement. Energy dissipation of BCJ is improved by the proposed Stiffness is defined as the slope of a line connecting the maximum
strengthening method with chamfers. At the same energy dissipation, point and minimum point at each hysteresis loop. Fig. 11 shows var-
horizontal displacement of specimens IJ-C150R, IJ-C300R, and IJ- iation of stiffness against maximum horizontal displacement at each
C300WR are smaller than that of control specimen. As demonstrated by hysteresis loop. Ignoring the difference in concrete strength, strength-
comparing specimens IJ-C300WR and IJ-C150R (at about the same ened specimens are stiffer than the control specimen and all specimens
concrete strength), energy dissipation increases with increasing size of show degradation in stiffness with increasing horizontal displacement.
chamfer. This affirms that the length of chamfer is the major aspect In spite of the difference in chamfer size, stiffnesses of strengthened
specimens are similar and merges at the advanced stage of loading.

580
E.S.-s. Lam, et al. Engineering Structures 191 (2019) 575–582

compression chord. Further, similar compressive strains were obtained

from specimens IJ-C300R and IJ-C300WR indicating that the influence
of U-bars to chamfers is nominal.
Fig. 13 compares strains in U-bars installed in the chamfers of
specimens IJ-C300R and IJ-C150R. With progressive increase in hor-
izontal force under cyclic action, U-bars are subjected to high tension
and low compression indicating the effectiveness of U-bars in trans-
ferring tensile force between beams and columns. Generally, tensile
stress in U-bars increases with reducing chamfer size.
Fig. 14 plots the strain on the surfaces of chamfers against hor-
izontal displacements for specimens IJ-C300W and IJ-C300WR, both
with chamfer size equals to 300 mm. The chamfers were subjected to
high compressive strains up to −0.0015 and low tensile strain. Due to
the low tensile strain experienced by the chamfers under cyclic action,
specimen IJ-C300WR without U-bars could still provide performance
compatible to that of specimen IJ-C300W with the installation of U-
Fig. 12. Strain of mortar in chamfer.
bars.

4. Conclusions

Experimental studies on strengthening beam-column joints using

unsymmetrical chamfers were conducted. One control specimen and
three specimens strengthened with different chamfer size and with or
without U-bars were prepared and tested to failure under reversed
cyclic loads. The following conclusions can be drawn.

(1) Unsymmetrical chamfers installed at the soffit of beams are effec-

tive to prevent joint shear failure and to enhance the loading ca-
pacity of beam-column joints.
(2) Loading capacity of a beam-column joint is increased by over 30%
after installing 300 mm chamfers at the soffit of beams regardless of
with or without U-bars.
(3) In terms of strengthening, size of chamfer is more crucial in com-
parison with installing U-bars in chamfers. This is demonstrated by
comparing the performance of the specimens.
Fig. 13. Strain of reinforcement in chamfers of specimens IJ-C300R and IJ- (4) As the chamfers are subjected to high compressive strains and low
C150R. tensile strains, the use of U-bars in chamfer is of secondary im-
portance.
3.6. Strain at surfaces of chamfers
Acknowledgement
Strain gauges were installed on the surfaces of chamfers to measure
the strain of mortar parallel to edge of chamfers. As shown in Fig. 12, The authors wish to express their gratitude and sincere appreciation
compressive strain of chamfers increased rapidly with the horizontal to the financial support from the Architectural Services Department of
displacement and was up to −0.002 (specimen IJ-C150R) or −0.0008 the Hong Kong Special Administration Region (HKSAR), Hong Kong via
(specimens IJ-C300R and IJ-C300WR) when the tests terminated. Innovation and Technology Support Programme (ITP/043/16NP). In
Tensile strain of chamfers was less than 0.00025 in all specimens. This addition, the authors thank the technicians of Structural Engineering
demonstrates that the main contribution of chamfers is to provide the Research Laboratory of the Department of Civil and Environmental
Engineering, The Hong Kong Polytechnic University for their support to

Fig. 14. Comparison of strains on surface of chamfers (a) Specimen IJ-C300R; (b) Specimen IJ-C300WR.

581
E.S.-s. Lam, et al. Engineering Structures 191 (2019) 575–582

the experimental studies. 2013;56:897–909.

[12] Li B, Lam ES-S, Cheng Y-K, Wu B, Wang Y-Y. Strengthening of non-seismically
designed beam-column joints by ferrocement jackets with chamfers. Earthquakes
References Struct 2015;8:1017–38.
[13] Ghobarah A, Said A. Shear strengthening of beam-column joints. Eng Struct
[1] Doǧangün A. Performance of reinforced concrete buildings during the May 1, 2003 2002;24:881–8.
Bingöl earthquake in Turkey. Eng Struct 2004;26:841–56. [14] Li B, Kai Q. Seismic behavior of reinforced concrete interior beam-wide column
[2] Zhao B, Taucer F, Rossetto T. Field investigation on the performance of building joints repaired using FRP. J Compos Constr 2010;15:327–38.
structures during the 12 May 2008 Wenchuan earthquake in China. Eng Struct [15] Sasmal S, Ramanjaneyulu K, Novák B, Srinivas V, Kumar KS, Korkowski C, et al.
2009;31:1707–23. Seismic retrofitting of nonductile beam-column sub-assemblage using FRP wrap-
[3] Lam S, Xu Y, Chau K, Wong Y, Ko J. Prog Earthquake Resistant Des Build Hong Kong ping and steel plate jacketing. Constr Build Mater 2011;25:175–82.
2002. [16] Pampanin S, Christopoulos C, Chen TH. Development and validation of a metallic
[4] Li B, Lam ES-S, Wu B, Wang Y-Y. Seismic behavior of reinforced concrete exterior haunch seismic retrofit solution for existing under-designed RC frame buildings.
beam-column joints strengthened by ferrocement composites. Earthquakes Struct Earthquake Eng Struct Dyn 2006;35:1739–66.
2015;9:233–56. [17] Chung TS-K, Lam ES-S, Wu B, Xu Y-L. Experimental and numerical verification of
[5] Wong HF, Kuang JS. Effects of beam—column depth ratio on joint seismic beha- hydraulic displacement amplification damping system. Struct Eng Mech
viour. Proc Inst Civil Eng – Struct Build 2008;161(2):91–101. https://doi.org/10. 2009;33:1–14.
1680/stbu.2008.161.2.91. [18] Dolce M, Cardone D, Marnetto R. Implementation and testing of passive control
[6] Karayannis C, Chalioris C, Sideris K. Effectiveness of RC beam-column connection devices based on shape memory alloys. Earthquake Eng Struct Dyn
repair using epoxy resin injections. J Earthquake Eng 1998;2:217–40. 2000;29:945–68.
[7] Tsonos A-DG. Performance enhancement of R/C building columns and beam–- [19] Shafaei J, Hosseini A, Marefat MS. Seismic retrofit of external RC beam–column
column joints through shotcrete jacketing. Eng Struct 2010;32:726–40. joints by joint enlargement using prestressed steel angles. Eng Struct
[8] Beschi C, Riva P, Meda A. Corner beam-column joints seismic retrofitting with high 2014;81:265–88.
performance fiber-reinforced concrete jacketing. World Conference on Earthquake [20] Santarsiero G, Masi A. Seismic performance of RC beam–column joints retrofitted
Engineering (15WCEE): IAEE (International Association for Earthquake with steel dissipation jackets. Eng Struct 2015;85:95–106.
Engineering). 2012. [21] Chaimahawan P, Pimanmas A. Seismic retrofit of substandard beam-column joint
[9] Ghobarah A, Aziz TS, Biddah A. Rehabilitation of reinforced concrete frame con- by planar joint expansion. Mater Struct 2009;42:443.
nections using corrugated steel jacketing. Struct J 1997;94:282–94. [22] Chaimahawan P, Pimanmas A. Behaviour and analysis of beam–column joint
[10] Ruiz-Pinilla J, Pallarés F, Gimenez E, Calderon P. Experimental tests on retrofitted strengthened by cast-in-situ expansion. Mag Concr Res 2010;62:347–63.
RC beam-column joints underdesigned to seismic loads general approach. Eng [23] Yang H, Zhao W, Zhu Z, Fu J. Seismic behavior comparison of reinforced concrete
Struct 2014;59:702–14. interior beam-column joints based on different loading methods. Eng Struct
[11] Li B, Lam ES-S, Wu B, Wang Y-Y. Experimental investigation on reinforced concrete 2018;166:31–45.
interior beam–column joints rehabilitated by ferrocement jackets. Eng Struct

582