Structures 52 (2023) 205–219

Contents lists available at ScienceDirect

Structures
journal homepage: www.elsevier.com/locate/structures

Feasibility study of self-centring beam-to-column joints with posttensioned

SMA strands
Hui Qian a, Jiannan Ye a, Jundong Gao a, *, Liming Cai b, Jijian Zhou b, Haidong Zhang c
a
School of Civil Engineering, Zhengzhou University, Zhengzhou 450001, China
b
Henan Architectural Design and Research Inst. Co., Ltd., Zhengzhou 450099, China
c
CSCEC Zhongyuan Architectural Design Institute Co., Ltd, Zhengzhou 450004, China

A R T I C L E I N F O A B S T R A C T

Keywords: The hysteretic behaviour of self-centring beam-to-column joints with posttensioned shaped memory alloy (SMA)
Numerical study strands under cyclic loading was studied in this paper. The material test of the dog bone-shaped SMA coupon was
Hysteretic behaviour carried out and used to verify the applicability of the SMA constitutive in finite element (FE) models. The FE
Self-centring beam-to-column joint
models of self-centring joints were established by ABAQUS, and the accuracy of the FE models was verified by the
Posttensioned SMA strands
Parametric analysis
relevant test results available from the existing literature. With the verified FE modelling technique, the FE model
of a novel self-centring joint was generated, and the numerical results were compared with the hysteretic per­
formance of regular joints. Afterwards, the influences of the steel angle thickness, initial posttensioning force,
vertical distance of two SMA strands on one side of the beam, and SMA strand number on the hysteretic per­
formance of the novel joint were discussed in detail. The novel joint exhibited excellent seismic performance. The
initial posttensioning force, numbers and vertical distance of SMA strands have significant effects on the stiffness,
bearing capacity and self-centring performance of the proposed joints, but have little effects on the energy
dissipation. The bearing capacity, stiffness, energy dissipation as well as the residual deformation of the proposed
joints improve with increasing steel angle thickness.

1. Introduction revealed that it had a high bearing capacity and stiffness, with little
residual deformation. The seismic performance of post-tensioned (PT)
The steel frame structures have been widely used in earthquake re­ steel connections was tested by Garlock et al. [5,6]. The angle size, the
gions. As the main bearing component, the beam-to-column joint is an bolt gauge length, the initial posttensioning forces of steel strands, the
important part of the steel frame structure. It plays a role in transmitting number of steel strands, and the length of reinforcing plates were the
and distributing the internal force, coordinating the deformation be­ investigated parameters. In addition, Garlock et al. [7,8] described the
tween the components, and ensuring the integrity of the structure [1]. behaviour of PT frames and PT frame systems, as well as an approach to
Previous earthquake disasters around the world have shown that beam- seismic design for PT steel frame systems based on performance. And
to-column joints are the most vulnerable part of steel frame structures, nonlinear dynamic time history analysis was used to evaluate the effects
and the destruction or even collapse of the entire structure will happen of the connection strength, the panel zone strength, and an increase in
caused by damage to beam-to-column joints [2]. As a result, many the connection strength in the upper floors on the seismic response of PT
innovative concepts and methods that can improve the seismic perfor­ moment resisting frames (MRFs). Pan et al. [9] compared the mechan­
mance of beam-to-column joints have been closely paid attention to by ical properties of self-centring joints and regular joints with steel angle
scientists and engineers around the world. in detail by quasi-static test and numerical simulation analysis. The
Self-centring beam-to-column joints with top and seat angle have numerical analysis of PT connections with stiffened angles was per­
been the subject of intensive investigation by a large number of scholars. formed by Shiravand et al. [10], who also proposed the approach to
Ricles et al. [3,4] firstly proposed self-centring beam-to-column con­ predict the behaviour of PT connections with stiffened angles. Moradi
nections with top and seat angle energy dissipation and posttensioned et al. [11] established a finite element (FE) model of self-centring joints
steel strands. The seismic performance test of the joint specimen with steel angles by ANSYS. After verified with the existing tests, the

* Corresponding author.
E-mail address: [email protected] (J. Gao).

https://doi.org/10.1016/j.istruc.2023.03.177
Received 7 January 2023; Received in revised form 27 March 2023; Accepted 29 March 2023
Available online 6 April 2023
2352-0124/© 2023 Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.
H. Qian et al. Structures 52 (2023) 205–219

relative significance of 16 factors was assessed by using a statistical Table 1

design-of-test approach. Furthermore, Moradi et al. [12] used the Chemical compositions of the SMA coupon (wt.%).
response surface methodology to predict and optimize the lateral Ni Ti C N H O
response characteristics of self-centring beam-to-column joints with
55.86 44.1238 0.01 0.002 0.001 0.0032
steel angle. The optimal combination of maximizing the initial stiffness,
load capacity, and ultimate drift of the joints, or minimizing the amount
of steel in the beam section was obtained. Cai et al. [13] proposed a new 2. Material test on SMA coupon
self-centring post-tensioned precast beam-to-column connection
assembled by unbonded strands and steel angles, and studied the seismic 2.1. Test study
performance of different parameters including the initial posttensioning
force, beam depth, and type of steel angles. The seismic performance of A NiTi SMA coupon was tested to measure the material properties,
self-centring beam-column joints with steel angles under main after­ and the chemical composition is shown in Table 1. The SMA material
shocks was studied by Yu et al. [14]. The results showed that the damage will exhibit superelasticity when the temperature is higher than the
controllability is still maintained under the action of main aftershocks, austenite processing temperature Af [28], thus an appropriate heat
and the seismic performance of the joints is better than that of tradi­ treatment was carried out. The machined dog bone-shaped bar coupon
tional joints. with a diameter of 10 mm (shown in Fig. 1 (a)) was subjected to 400℃
Shape memory alloy (SMA) is an emerging smart material with self- for 40 min, followed by water quenched to enhance the superelastic
sensing, self-diagnosis, and self-healing functions. Because of the unique properties [29]. The coupon testing setup is illustrated in Fig. 1(b),
shape memory effect, superelasticity, and high damping of SMA, it has where an extensometer with a 50 mm gauge length was employed to
been widely concerned in the research and application of vibration measure the axial strain. The coupon was subjected to a cycle loading
control of civil engineering structures [15]. Ma et al. [16,17] proposed a protocol with a 1 % strain increment until 6 % strain, as shown in Fig. 1
new extended end-plate connection with SMA bolts and studied its (b). The strain exceeding 6 % would significantly compromise the hys­
feasibility by numerical models. After that, Fang et al. [18] conducted teretic responses of the SMA [30]. Thus, in practical applications of
eight full-scale tests on the extended end-plate connection with SMA pretensioned SMA strands in self-centring joints, it is necessary to ensure
bolts. The feasibility of a new class of partially restrained connections that the strain of SMA strands does not exceed 6 %. The SMA coupon
using SMA was evaluated by Ocel et al. [19]. After the quasi-static test, after material test is shown in Fig. 1 (b).
these connections exhibited a high level of energy dissipation, large The hysteresis curve of the tested NiTi SMA coupon is shown in Fig. 2
ductility capacity, and no strength degradation. A beam-column (a). Elastic modulus is 40GPa. The yield point of 450 MPa at 1.34 %
connection incorporating superelastic nickel-titanium (NiTi) SMA ten­ strain is the initiation of the SMA phase transformation. When subjected
dons was proposed by Speicher et al. [20]. This connection was to the maximum strain of 6 %, the SMA coupon exhibits excellent
compared to three other connections with steel tendons, martensitic superelasticity and the residual strain (εr) is 0.4 %, as tabulated in
NiTi tendons, and superelastic NiTi tendons paralleled with aluminium Table 2. The constitutive model was established according to the
tendons. Wu et al. [21] investigated the mechanical properties and en­ Ref. [31], as illustrated in Fig. 2 (b).
ergy dissipation capacities of the beam-to-column joints with martensite
NiTi SMA threaded rods by the quasi-static tests and the retests after the 2.2. Numerical modelling and validation
heating repair was conducted. And Wang et al. [22] proposed a novel
connection with superelastic SMA tendons and steel tendons. A nu­ A finite element (FE) model was established to simulate the SMA
merical simulation of end-plate beam-to-column connections with NiTi coupon by ABAQUS, as shown in Fig. 3. The model was meshed with the
SMAs was studied by Yam et al. [23], and an improved high strength first order incompatible mode solid element C3D8I. The parameters
(HS)-SMA hybrid solution was proposed for the issue of shear slippage. presented in Table 2 was inputted into ABAQUS. As shown in the Fig. 3,
Sultana et al. [24] investigated the seismic performance of steel moment two reference points, RP1 and RP2, were coupled to the ends of SMA
resisting frames (SMRFs) with SMA connections at certain locations coupon, respectively. A fixed constraint was implemented by point RP1,
based on maximum inter-storey drift (MID), maximum residual inter- and the loading displacements in accordance with the test were applied
storey drift (MRID), and damage scheme. The steel beam-to-column to RP2. Three FE models with mesh sizes of 1 mm, 2 mm and 3 mm were
connections with SMA bolts and plates through a numerical approach established, and stress–strain curves of FE models with three mesh sizes
and investigated the performance of eleven different connection models were compared, as shown in Fig. 4. The curve of the FE model with a
under cyclic loading was studied by Farmani et al. [25]. Chowdhury mesh size of 2 mm agrees well with the experimental test, while the
et al. [26] evaluated the effects of short-length regular post-tensioned stress of the FE model with a mesh size of 1 mm is slightly higher, and
strands and SMA hybrid strands in steel connections. Wang et al. [27] that of the FE model with a mesh size of 3 mm is slightly lower. The
investigated the earthquake resistant behaviour of a steel SC column elastic modulus E, yield stress fy and ultimate stress fu were calculated
equipped with NiTi SMA bolts and presented an analytical model with and compared with the experimental test in Table 3. The results show
the corresponding design procedure. that the FE models with mesh sizes of 1 mm and 2 mm are accurate
In this paper, a self-centring steel joint with superelastic SMA strands enough to compare with the test, and the result of the 3 mm model is
was proposed, and the seismic performance of the novel self-centring slightly lower than that of the test. Otherwise, the time required to
steel joints was studied. The material test of SMA coupon was carried calculate the FE model with a mesh size of 1 mm is 1741 s, the time
out, and the FE model of the SMA coupon was well verified. The con­ required to calculate the FE model with a mesh size of 2 mm is 231 s,
struction form of novel steel joints was given and the FE model of self- while the time required to calculate FE model with a mesh size of 3 mm
centring steel joints was established. The parametric analysis of the is 175 s. Therefore, the FE model with a mesh size of 2 mm was
self-centring steel joints was carried out, and the effects of SMA strand considered reasonable based on both calculation accuracy and effi­
posttensioning force, vertical distance of two SMA strands on one side of ciency. The mesh size of FE model was set as 2 mm.
beam, SMA strand number and steel angle thickness on the seismic The FE hysteresis curves of the tested SMA coupon are compared
performance of self-centring steel joints were considered. with the test ones, as shown in Fig. 4. The numerically predicted hys­
teresis curves agrees well with the test results. The difference observed
in the low-stress section can be attributed to the simplification of the
SMA constitutive model used in the FE simulation. However, this
simplification is acceptable in FE simulations and the model overall

206
H. Qian et al. Structures 52 (2023) 205–219

Fig. 1. Material test on a NiTi SMA coupon.

Fig. 2. Test curves and constitutive model of SMA coupon.

Table 2
Material properties of SMA.
Material d/mm EA/GPa σMs(fy)/MPa σMf/MPa σAs/MPa σAf/MPa εy/% εr/% εL
NiTi SMA 10 40 450 580 270 180 1.34 0.4 0.048

where the transformation stresses (σMs, σ Mf, σ As, and σ Af), Young’s moduli (EA) and maximum transformation strain (εL) are key isothermal parameters for superelastic
SMA in ABAQUS [18].

u u u

x y

Fig. 3. FE model of SMA coupons.

accurately reflects the stress–strain behaviour of SMAs. The constitutive

model and material parameters of the SMA coupon will be further used
Fig. 4. Mechanical properties of SMA and comparation of different mesh sizes.
in subsequent numerical studies in this paper.

207
H. Qian et al. Structures 52 (2023) 205–219

Table 3 consistent with those in the Ref. [9].

Comparison results of FE models with different mesh sizes. The FE models of carbon steel beam-to-column joints tested in
Mesh EFE/ fy,FE/ fu,FE/ Etest/ fy,test/fy, fu,test/fu, Ref. [9] were established by using the FE software ABAQUS. There were
size (MPa) (MPa) (MPa) EFE FE FE eight parts in the models, including beams, columns, steel angles, rein­
1 mm 368.2 457.1 587.6 1.09 0.98 0.99 forcing plates, shim plates, anchorage plates, bolts, and posttensioned
2 mm 359.7 447.5 575.8 1.11 1.01 1.01 steel strands. The posttensioned steel strands were meshed with the truss
3 mm 355.6 439.4 566.8 1.12 1.02 1.02 element T3D2 [32], while the remaining parts were meshed with the
first order incompatible mode solid element C3D8I, which could effec­
tively eliminate shear self-locking and decrease computational
3. Development and verification of FE models
complexity [33]. The overall mesh sizes of beams and columns were set
as 50 mm, and finer meshes were prescribed in the area around the panel
3.1. Development of FE models
zone. The mesh sizes for steel angles, reinforcing plates, and shim plates
were 20 mm.
Elaborate FE models were developed to conduct numerical simula­
Fig. 6 displays the FE models of JD1 and JD5 joints in Ref. [9]. Joint
tion on the existing test results of carbon steel beam-to-column joints in
JD1 is a self-centring joint with posttensioned strands, while joint JD5 is
Ref. [9]. The cross-section dimension of welded beams was I-500 × 300
a regular joint. The details of boundary conditions in FE models are also
× 28 × 14 mm (i.e., outer height H = 500 mm, flange width B = 180
shown in Fig. 6. Both the top and bottom ends of the column are hinged
mm, flange thickness tf = 28 mm, web thickness tw = 14 mm), and the
to simulate the point of contraflexure [34]. The loading amplitude is
welded column section size was I-500 × 500 × 20 × 20 mm. The geo­
applied using the displacement at the reference point RP3 coupling of
metric details of panel zone and other components’ layout are presented
in Fig. 5 (1). Moreover, the dimensions of bolt spacing in steel angles and
strand spacing in beam end are shown in Fig. 5 (2) and (3), and the Table 4
overall dimensions of joints are shown in Fig. 5. The dimensions of steel Parameters of carbon steel joint specimens [9].
angles as well as the initial posttensioning force of strands are shown in
Specimen Steel angle (b × d Initial posttensioning force of Joint type
Table 4. It is noted that both the self-centring joint and regular joint were × l) (mm) strands (kN)
considered in Ref. [9]. The 1 × 7 steel strand with the nominal diameter
JD1 200 × 16 × 300 60 Self-centring
of 15.2 mm and tensile strength of 1860 MPa was employed as post­ JD2 200 × 14 × 300 60 steel joint
tensioned strands. The bolt pretension force of M20 high-strength bolts JD3 200 × 16 × 300 120
in grade 8.8 was 125kN. The mechanical properties of all components, JD4 200 × 14 × 300 120
involving beams, columns, steel angles, reinforcing plates, shim plates JD5 200 × 16 × 300 — Regular steel
JD6 200 × 14 × 300 joint
and anchorage plates machined by Q345B hot rolled plates, were
—

Fig. 5. Geometric details of joints [9] (all dimensions in mm).

208
H. Qian et al. Structures 52 (2023) 205–219

u xu y u z
u xu y u z
x y z

x y z

y
y

x z
x z

Fig. 6. Joint FE models.

Fig. 7. Deformation mode comparison of two typical joints.

the beam top flange, and the loading protocol is consistent with the test shim plate, and the shim plate and beam [36]. The tangential property
in Ref. [9]. for the contact pairs was finite-sliding with the penalty method, and the
The pretension forces for the bolts and steel strands were applied default hard contact model was utilized for the behaviour normal to the
using the BOLT LOAD method [35], which is a more convenient and contact surfaces [37]. The tangential behaviour of the bolt shank and
effective method than the temperature method for defining pretension hole wall contact pair was considered to be frictionless, while the fric­
in steel strands. Eight different contact pairs were included in FE models, tion coefficient for other contact pairs was taken according to the test in
involving the bolt shank and hole wall, the bolt nut and steel angle, the Ref. [9]. Furthermore, the “tie” constraint was used to simulate the weld
steel angle and reinforcing plate, the steel angle and shim plate, the bolt connection between shim plates with column flange and reinforcing
head and column, the bolt head and beam, the reinforcing plate and plates with beam flange [38].

209
H. Qian et al. Structures 52 (2023) 205–219

F
F F
F F
F

F F F
F F
F

F F
F F

Fig. 8. Comparison of hysteresis curves of joints in Ref. [9].

The material properties in carbon steel joint FE models obtained and columns. The comparison results indicate the ability of the FE model
from Ref. [9] were converted into true stress versus logarithmic plastic to accurately predict the deformation modes of self-centring joints.
strain relationship before input into ABAQUS [39]. The FE models were Fig. 8 shows the comparison of FE hysteresis curves and test results.
run with general static analysis, accounting for geometric nonlinearity Obviously, the numerically predicted F-Δ curves agree well with the test
and material plasticity. The numerically obtained load–displacement (F- ones. The test curves of joint JD1 are significantly lower than the
Δ) curves and failure modes were compared with the test results in simulation ones, which is caused by bolt slipping during the loading
Ref. [9]. process [9]. In addition, the test curves of other joints are in close
agreement with the simulated ones. It can be hence found that the FE
models are able to provide accurate simulation of hysteresis curves for
3.2. Verification of FE models both self-centring and regular beam-to-column joints.
The test and FE skeleton curves are compared in Fig. 9. The FE
The comparisons of numerical demonstration modes with test ones in skeleton curves are in acceptable accordance with the test results. It is
Ref. [9] of two typical joints are shown in Fig. 7. It can be seen that the noted that the skeleton curves of joint JD1 and JD3 experienced strength
gap between steel angle and column as well as the deformation of steel degradation in the last cycle during cyclic loading test, which could be
angles are precisely captured. Meanwhile, the plastic deformation of explained by the fact that the initial defects and bolt slipping in the test
joints is also reflected by the stress distributions of steel angles, beams

210
H. Qian et al. Structures 52 (2023) 205–219

F
Fig. 9. Comparison of skeleton curves.

Table 5
Comparison results of carbon steel joints in Ref. [9].
Specimens K0,test/ FIGO,test/ Kθ,test/ Fu,test/ K0,FE/ FIGO,FE/ Kθ,FE/ Fu,FE/ K0,test FIGO,test Kθ,test Fu,test
(kN/mm) kN (kN/mm) kN (kN/mm) kN (kN/mm) kN K0,FE FIGO,FE Kθ,FE Fu,FE

JD1 1.82 14.47 0.29 63.86 1.85 22.99 0.33 77.55 0.99 — — —
JD2 1.71 21.20 0.30 69.32 1.80 21.31 0.30 69.79 0.95 0.99 0.97 0.99
JD3 1.17 39.41 0.30 87.78 1.32 39.08 0.31 91.00 0.89 1.01 0.95 0.96
JD4 1.11 36.92 0.28 84.62 1.26 40.60 0.28 85.04 0.88 0.91 1.02 0.99
JD5 0.11 — — 37.79 0.10 — — 41.98 1.02 — — 0.90
JD6 0.07 — — 30.82 0.08 — — 31.22 0.89 — — 0.99
Average — — — — — — — — 0.94 0.97 0.98 0.97
St. dev — — — — — — — — 0.06 0.05 0.04 0.04

Fig. 10. The proposed self-centring joint.

was not considered in the FE models. in Table 5, where the FE values are compared with the test ones.
The initial stiffness K0, imminent gap-opening resistance FIGO, stiff­ The mean values of K0,test/K0,FE, Kθ,test/Kθ,FE, FIGO,test/FIGO,FE and Fu,
ness after the gap opening Kθ and ultimate resistance Fu are used to test/Fu,FE are between the range of 0.94 and 0.98, and the standard de­
analyse the mechanical behaviour of self-centring beam-to-column viations are all less than 0.1. It is clear that the FE values for the initial
joints [40]. The initial stiffness, imminent gap-opening resistance and stiffness, imminent gap-opening resistance, ultimate resistance, and
stiffness after the gap opening are fitted linearly from the skeleton stiffness after the gap opening are a little higher than test ones. This
curves [41], and the ultimate resistance was determined by the maybe explained by the fact that the FE displacement was smaller than
maximum value of the skeleton curves [42]. The results are summarized the measured displacement due to the ignorance of the gap between

211
H. Qian et al. Structures 52 (2023) 205–219

Table 6
Geometric configurations of the proposed joints.
Components Material Section/mm Length/mm

Beam Q345B I-500 × 300 × 20 × 25 1200

Column Q345B I-500 × 500 × 20 × 25 2000
Steel angle Q235B L-200 × 200 × 25 300
Anchorage plate Q345B 500 × 20 500
SMA strands NiTi SMA Φ-32 mm 1700

devices in the simulation. Specially, the imminent gap-opening resis­

tance, stiffness after the gap opening and ultimate resistance of joint JD1
were not included in the average calculation, since the structural per­
formance was greatly affected by the bolt slippage in Ref. [9]. Therefore,
it can be concluded that the failure models, hysteresis curves, and
structural properties numerically obtained are in close agreement with
the test results, highlighting the accuracy and applicability of the
developed FE models.

4. Self-centring beam-to-column joints with posttensioned SMA

strands Fig. 11. Loading protocol.

4.1. Joint configurations of the regular joint (RJ) without posttensioned SMA strands is also
established for comparative analysis.
A novel self-centring joint (SJ) is proposed, as shown in Fig. 10 (a). The Chaboche model [43,44] of Q235B and Q345B materials were
The proposed self-centring joint is made up of carbon steel column, employed for modelling the hysteretic performance of the proposed
beam, stiffeners, steel angles, and superelastic SMA strands. The dif­ joint. The Chaboche model parameters of Q235B and Q345B presented
ference between the novel self-centring joint and the regular self- in Ref. [45] are illustrated in Table 7. In addition, the constitutive model
centring joint is that the SMA strands are used as the pretension provi­ of the SMA coupon shown in Fig. 2 (b) is adopted for SMA strands
sion component. In actual engineering structures, it is challenging to herein.
provide sufficient recovery pretension force for joints once the steel
strands yield, which may result in structural damage. The excellent
superelasticity of SMA enables it to maintain elasticity under significant 4.2. Results of FE numeration
displacement, which is particularly important for pretensioned steel
strands. The stress distribution results of the proposed joints at the loading
Beams and columns are made of Q345B welded H-section steel, while displacement of 30 mm are shown in Fig. 12, and the grey area in beams
the Q235B equilateral steel angle with a thickness (t) of 25 mm is and columns represents that the regions exceeds its stress at zero
adopted. The steel angles are welded to the beam and column flanges equivalent plastic strain (290 MPa). The plastic deformations of two
and the welding position is marked with red lines, as shown in Fig. 10 joints are mainly concentrated on the steel angles, which proves damage
(b). The constraint was strictly set according to the bolt position speci­ controllability of the joints. A mild plastic deformation on the beam of
fied in the Chinese standard GB 50017 to ensure the consistency of the specimen SJ is resulted from the large force of SMA strands, which can
yield strength between the welded angle steel and the bolted joint. For be observed in Fig. 12 (a). Moreover, the stress of SMA strands slightly
the feasibility study of pretensioned SMA strands in self-centring joints, exceeds its yield stress. In practical engineering, the initial post­
the utilization of welded connections in the ABAQUS model can avoid tensioning force should be controlled within a range to avoid the
complex contact conditions and significantly accelerate the convergence yielding of the beams, columns and SMA strands.
of the simulation. The load–displacement (F-Δ) curves of specimens RJ and SJ are
A total of 4 SMA strands with a diameter (Φ) of 32 mm are employed, shown in Fig. 13. The proposed self-centring joint specimen SJ demon­
which are symmetrically placed on both sides of the beam web. The strates excellent self-centring performance, with a typical “double flag”
vertical distance of two SMA strands on each side (h) is 280 mm, as hysteresis curve. In contrast, specimen RJ shows good energy dissipation
shown in the top right corner of Fig. 10 (a), and the initial posttensioning capacity but has a large residual deformation. It can also be seen that
force of SMA strands (P0) is 230kN. Geometric configurations of the specimen SJ has higher bearing capacity than that of specimen RJ.
beam, column, and steel angle are shown in Table 6. The steel angles are The skeleton curves, stiffness degradation curves, energy dissipation
employed for energy dissipation, whilst the posttensioned SMA strands curves and residual displacement curves of specimen RJ and SJ are
are used to offer the self-centring ability. The self-centring, high compared in Fig. 14. The skeleton curves of specimens RJ and SJ in
damping and high energy dissipation can be fully utilized by the pro­ Fig. 14 (a) are S-shaped with central symmetry, and the structural
posed joint. properties are summarized in Table 8. The initial stiffness (K0) of spec­
The proposed self-centring joint FE model is developed by ABAQUS imen SJ is about twice that of specimen RJ, and the ultimate resistance
software with the previous verified modelling technique, as shown in (Fu) of specimen SJ is about 1.8 times that of specimen RJ. An obvious
Fig. 10 (b). A tie constraint is used to simulate welding connections increment of the stiffness and bearing capacity is achieved by specimen
between steel angles and corresponding flanges. The element types and SJ compared with that of specimen RJ, which is due to the post­
mesh sizes adopted in the proposed self-centring joint FE model are the tensioning force provided by the SMA strands.
same as those of FE models stated in Section 3. The boundary conditions The stiffness of specimen SJ is higher than that of specimen RJ during
shown in Fig. 10 (b) are similar with those of the test joints in Ref. [9]. the loading, as shown in Fig. 14 (b). The stiffness degradation amplitude
The loading protocol is shown in Fig. 11, where the displacement of specimen SJ is larger than that of specimen RJ. Specifically, the
applied at the beam end increases with an interval of 5 mm, and the minimum stiffness of specimen SJ is 16.35kN•mm− 1, while that of
maximum loading displacement is 30 mm. Meanwhile, the joint model specimen RJ is 9.12 kN•mm− 1.
It can be found from Fig. 14 (c) that the energy dissipation curves of

212
H. Qian et al. Structures 52 (2023) 205–219

Table 7
Chaboche model parameters of Q235B and Q345B in Ref. [45].
Material σ |0 / Q∞/ biso Ckin,1 γ1 Ckin,2 γ2 Ckin,3 γ3 Ckin,4 γ4
MPa MPa /MPa /MPa /MPa /MPa

Q235B 228 88 9.1 × 106 5607 52 4321 49 2472 47 1630 35

Q345B 290 88 2.43 × 108 1000 29 1000 29 232 29 232 29

Fig. 12. Deformations and stress distribution at displacement of 30 mm.

concluded that the excellent self-centring capability but limited energy

consumption is provided by SMA strands. It is worth noting that the
residual rotation (Δr) of specimens SJ and RJ are 0.04 % and 1.1 %,
respectively. The self-centring joint is considered to have favourable
performance when the residual rotation is less than 0.20 % at a rotation
of 2.00 % [46].

5. Parametric analysis of proposed self-centring joint

5.1. The influence of initial posttensioning force of SMA strand

F

In order to discuss the influence of SMA strand initial posttensioning

forces on the structural behaviour of the proposed joints, four joint
models named SJP with different initial posttensioning forces were
developed. The parameters of each model are listed in Table 9.
The F-Δ curves of joint models with different SMA strand post­
tensioning forces are shown in Fig. 15. With the increase of the SMA
strand posttensioning force, the self-centring performance of the joint
becomes better and the bearing capacity gradually increases. A good
energy dissipation capacity is performed by the hysteresis curves of four
Fig. 13. F-Δ hysteresis curve of specimens RJ and SJ. joint models. The good self-centring performance is achieved on speci­
mens SJP1 and SJP2, while large residual deformation is observed in
specimens RJ and SJ are extremely close to each other, and the total specimens SJP3 and SJP4.
energy dissipation (Wt) of two specimens is very similar, as shown in Fig. 16 presents the skeleton curves, stiffness degradation curves,
Table 8. Nevertheless, the residual deformation of specimen SJ is much energy dissipation curves, and residual displacement curves of joint
smaller than that of specimen RJ, as shown in Fig. 14 (d). It can be models SJP. The bearing capacity and stiffness increase with the

213
H. Qian et al. Structures 52 (2023) 205–219

•
F

K
W

Fig. 14. Structural properties of specimens RJ and SJ.

Table 8
Structural properties of specimens RJ and SJ.
specimen K0/ FIGO/ Kθ/ F u/ Δ r/ Wt/
(kN•mm− 1) (kN) (kN•mm− 1) (kN) (mm) kJ

RJ 27.67 – – 271.5 17.62 26.25

SJ 57.74 213.7 5.55 488.76 0.76 26.51

Table 9
F

Model parameters of specimens SJP.

Specimen t/mm h/mm Φ/mm P0/kN P
SJP1 25 280 32 230
P
SJP2 25 280 32 190
P
SJP3 25 280 32 150 P
SJP4 25 280 32 110

increment of SMA strand posttensioning force, but the contribution to

energy dissipation is little. And within the elastic range of SMA strands,
the smaller residual deformation of the joint is caused by the larger SMA Fig. 15. F-Δ curve of specimens SJP.
strand posttensioning force.
The SMA strands force-lateral force (Pre-RF) curves of specimens SJP SJA, were designed to analyse the influence of steel angle thicknesses.
are shown in Fig. 17. The lateral force at the beam top is taken as the The only difference between each joint model is the steel angle thick­
horizontal axis, and the vertical axis represents the force of SMA strands. ness, as listed in Table 10.
The SMA strand forces start from the initial posttensioning force, The F-Δ curves of joint models SJA are shown in Fig. 18. The better
respectively, and return to the initial posttensioning forces after the end self-centring performance is noted in specimens SJA with thinner steel
of each cycle. It is noted that the SMA strand force is a little lower than angle. Specimens SJA1 and SJA2 experience large residual deformation,
the initial posttensioning force in the last loading cycle of specimen while specimens SJA3 and SJA4 can achieve self-centring. It can be
SJP1, which is owing to the yielding of SMA strands. Therefore, the concluded that the most important function of steel angles is energy
yielding of posttensioned SMA strands should be avoided in practical dissipation, whist the self-centring performance will be affected by the
engineering. steel angle thickness.
The skeleton curves, stiffness degradation curves, energy dissipation
5.2. The influence of steel angle thickness curves and residual displacement curves of joint models with different
steel angle thicknesses are shown in Fig. 19. The steel angle thickness
Four joint FE models with different thicknesses of steel angle, named has positive influences on the bearing capacity, stiffness and energy

214
H. Qian et al. Structures 52 (2023) 205–219

P
P
P
P

•
F

K
P
P
P
P

P P
P P
P P
P P
W

Fig. 16. Structural performance of specimens SJP.

P
P
P
P
Pre

t
t
t
t

RF
Fig. 17. Pre-RF curve of specimens SJP.
Fig. 18. F-Δ curve of specimens SJA.

Table 10
5.3. The influence of SMA strand vertical distance
Model parameters of specimens SJA.
The vertical distance changes of SMA strands on each side will lead to
Specimen t/mm h/mm Φ/mm P0/kN
different elongation deformation of SMA strands. Three joint models
SJA1 30 280 32 230 named SJW with different vertical distances of SMA strands were
SJA2 28 280 32 230
designed. The model parameters of SJW joints are shown in Table 11. It
SJA3 25 280 32 230
SJA4 22 280 32 230 should be noted that four SMA strands are symmetrically placed.
Fig. 20 shows the F-Δ curves of joint models with different vertical
distances of SMA strands. Hysteresis curves of these three joint models
dissipation of specimens SJA. However, with the growth of steel angle are close to each other, together with the similar bearing capacity and
thickness, the residual deformation also increases. It should be noted energy dissipation. It can be proved that little variation of hysteresis
that the beams and columns experience residual deformation, once the curves will be caused by different vertical distances of SMA strands on
plastic deformation exceeds the deformation that the SMA strand can each side.
recover. The skeleton curves, stiffness degradation curves, energy dissipation

215
H. Qian et al. Structures 52 (2023) 205–219

t
t
t
t

•
F

K
t
t
t
t

t t
t t
t t
t t
W

Fig. 19. Structural performance of specimens SJA.

residual deformation of joint models is influenced by the vertical dis­

Table 11
tance of SMA strands.
Model parameters of specimens SJW.
Specimen t/mm h/mm Φ/mm P0/kN 5.4. The influence of SMA strand number
SJW1 25 360 32 230
SJW2 25 320 32 230 The self-centring force provided by SMA strands are directly affected
SJW3 25 280 32 230
by their numbers n. Two joint models named SJC with SMA strand
numbers of 4 and 6 were established, as listed in Table 12.
Fig. 22 shows the F-Δ curves of joint models with different SMA
strand numbers. It can be seen that the bearing capacity and self-
centring performance in specimen SJC1 are greater than those in spec­
imen SJC2. This is due to greater self-centring force provided by more
SMA strands.
The skeleton curves, stiffness degradation curves, energy dissipation
curves and residual displacement curves of joint models SJC are shown
in Fig. 23. Slightly growths in terms of the bearing capacity, stiffness and
energy dissipation are observed with increasing number of SMA strands,
accompanied by lower residual deformation.
F

6. Conclusion

h The material tests of SMA coupons were carried out and used to
h verify the applicability of the SMA constitutive relation in FE models.
h The FE models of carbon steel self-centring and regular joints were
established using ABAQUS software, and the accuracy of which were
verified by the test in Ref. [9]. With the verified FE modelling technique,
the FE model of the novel self-centring joint with posttensioned SMA
strands was developed, and the numerically structural performance of
Fig. 20. F-Δ curve of specimens SJW. the proposed self-centring joint was compared with those of the regular
joint. The influences of steel angle thickness as well as initial post­
curves and residual displacement curves of specimens SJW are shown in tensioning force, vertical distance, and numbers of SMA strands on the
Fig. 21. Little effect of the vertical distance of SMA strands on the hysteretic performance of self-centring joints were analysed. The
bearing capacity, stiffness and energy dissipation of the proposed joint following conclusions can be made:
models is observed. Nevertheless, closer vertical distance of SMA strands
results in larger residual deformation. It can be concluded that only the

216
H. Qian et al. Structures 52 (2023) 205–219

h
h
h

•
F

K
h
h
h

h h
h h
h h
W

Fig. 21. Structural performance of specimens SJW.

verified FE modelling technique was further employed for the

Table 12
subsequent analysis of a novel joint.
Model parameters of specimens SJC.
(3) A novel self-centring beam-to-column joint with posttensioned
Specimen t/mm h/mm Φ/mm P0/kN n SMA strands was proposed and the FE models were established.
SJC1 25 280 32 150 6 Self-centring joint model SJ experienced higher stiffness and
SJC2 25 280 32 150 4 bearing capacity, smaller residual deformation, and similar en­
ergy dissipation, compared with its regular counterpart RJ.
(4) The increment of SMA strand initial posttensioning force and
numbers made noticed contribution to structural behaviour of the
proposed joint. The structural performance improved with
increasing steel angle thickness, except for the residual defor­
mation. The vertical distance of SMA strands on each side had
little effect on the structural performance.

The hysteretic performance of the novel joint is studied in detail by

FE analysis. It is proved that the self-centring beam-to-column joint with
posttensioned SMA strands exhibited expected performance, high­
lighting the feasibility of the proposed joint.
F

CRediT authorship contribution statement

Hui Qian: Conceptualization, Methodology, Writing – original draft,

Writing – review & editing, Supervision, Funding acquisition. Jiannan
Ye: Investigation, Data curation, Formal analysis, Writing – original
draft. Jundong Gao: Investigation, Conceptualization, Methodology,
Writing – review & editing. Liming Cai: Investigation, Writing – review
& editing. Jijian Zhou: Writing – review & editing. Haidong Zhang:
Fig. 22. F-Δ curve of specimens SJC. Writing – review & editing.

(1) The material properties of NiTi SMA were tested and further Declaration of Competing Interest
employed to verify the applicability of the constitutive model in
FE models. The verified constitutive model was adopted for SMA The authors declare that they have no known competing financial
strands in the subsequent analysis of a novel joint. interests or personal relationships that could have appeared to influence
(2) The accuracy of the developed FE models was verified by the the work reported in this paper.
relevant test results available from the existing literature. The

217
H. Qian et al. Structures 52 (2023) 205–219

•
F

K
W

Fig. 23. Structural performance of specimens SJC.

Acknowledgements: [15] Song G, Ma N, Li HN. Applications of shape memory alloys in civil structures. Eng
Struct 2006;28(9):1266–74.
[16] Ma HW, Wilkinson T, Cho CD. Feasibility study on a self-centring beam-to-column
The work presented in this paper was supported by the National connection by using the superelastic behavior of SMAs. Smart Mater Struct 2007;
Natural Science Foundation of China (No. 51978631), Program for 16(5):1555.
Innovative Research Team (in Science and Technology) in University of [17] Ma HW, Cho CD, Wilkinson T. A numerical study on bolted end-plate connection
using shape memory alloys. Mater Struct 2008;41(8):1419–26.
Henan Province (Grant no. 23IRTSTHN006), and Key Scientific [18] Fang C, Yam MCH, Lam ACC, et al. Cyclic performance of extended end-plate
Research Project in University of Henan Province (No. 23A560009). connections equipped with shape memory alloy bolts. J Constr Steel Res 2014;94:
122–36.
[19] Ocel J, DesRoches R, Leon RT, et al. Steel beam-column connections using shape
References memory alloys. J Struct Eng 2004;130(5):732–40.
[20] Speicher MS, DesRoches R, Leon RT. Experimental results of a NiTi shape memory
[1] Nakashima M, Roeder CW, Maruoka Y. Steel moment frames for earthquakes in alloy (SMA)-based recentring beam-column connection. Eng Struct 2011;33(9):
United States and Japan. J Struct Eng 2000;126(8):861–8. 2448–57.
[2] McCormick J, Aburano H, Ikenaga M, et al. Permissible residual deformation levels [21] Wu ZY, He XH, Zhang YC, et al. Experiment on hysteretic behavior of beam-to-
for building structures considering both safety and human elements. Proceedings of column connection using martensite NiTi shape memory alloy threaded rods in
the 14th world conference on earthquake engineering. Seismological Press Beijing, steel frame. Journal of Building Structures 2011;32(10):97–106. in Chinese.
2008: 12-17. [22] Wang W, Chan TM, Shao H, et al. Cyclic behavior of connections equipped with
[3] Ricles JM, Sause R, Garlock MM, et al. Posttensioned seismic-resistant connections NiTi shape memory alloy and steel tendons between H-shaped beam to CHS
for steel frames. J Struct Eng 2001;127(2):113–21. column. Eng Struct 2015;88:37–50.
[4] Ricles JM, Sause R, Peng SW, et al. Experimental evaluation of earthquake resistant [23] Yam MCH, Fang C, Lam ACC, et al. Numerical study and practical design of beam-
posttensioned steel connections. J Struct Eng 2002;128(7):850–9. to-column connections with shape memory alloys. J Constr Steel Res 2015;104:
[5] Garlock MM, Ricles JM, Sause R. Cyclic load tests and analysis of bolted top-and- 177–92.
seat angle connections. J Struct Eng 2003;129(12):1615–25. [24] Sultana P, Youssef MA. Seismic performance of steel moment resisting frames
[6] Garlock MM, Ricles JM, Sause R. Experimental studies of full-scale posttensioned utilizing superelastic shape memory alloys. J Constr Steel Res 2016;125:239–51.
steel connections. J Struct Eng 2005;131(3):438–48. [25] Farmani MA, Ghassemieh M. Shape memory alloy-based moment connections with
[7] Garlock MM, Sause R, Ricles JM. Behavior and design of posttensioned steel frame superior self-centring properties. Smart Mater Struct 2016;25(7):075028.
systems. J Struct Eng 2007;133(3):389–99. [26] Chowdhury MA, Rahmzadeh A, Moradi S, et al. Feasibility of using reduced length
[8] Garlock MM, Ricles JM, Sause R. Influence of design parameters on seismic superelastic shape memory alloy strands in post-tensioned steel beam–column
response of post-tensioned steel MRF systems. Eng Struct 2008;30(4):1037–47. connections. J Intell Mater Syst Struct 2019;30(2):283–307.
[9] Pan ZH. Research on steel frame system with self-resetting ability. Tsinghua [27] Wang B, Zhu S, Qiu CX, et al. High-performance self-centring steel columns with
University; 2010. in Chinese. shape memory alloy bolts: Design procedure and experimental evaluation. Eng
[10] Shiravand MR, Mahboubi S. Behavior of post-tensioned connections with stiffened Struct 2019;182:446–58.
angles under cyclic loading. J Constr Steel Res 2016;116:183–92. [28] DesRoches R, McCormick J, Delemont M. Cyclic properties of superelastic shape
[11] Moradi S, Alam MS. Lateral load–drift response and limit states of posttensioned memory alloy wires and bars. J Struct Eng 2004;130(1):38–46.
steel beam-column connections: parametric study. J Struct Eng 2017;143(7): [29] Qian H, Li ZA, Pei JZ, et al. Seismic performance of self-centring beam-column
04017044. joints reinforced with superelastic shape memory alloy bars and engineering
[12] Moradi S, Alam MS. Multi-criteria optimization of lateral load-drift response of cementitious composites materials. Compos Struct 2022;294:115782.
posttensioned steel beam-column connections. Eng Struct 2017;130:180–97. [30] Wang W, Fang C, Liu J. Large size superelastic SMA bars: Heat treatment strategy,
[13] Cai XN, Pan ZF, Zhu YZ, et al. Experimental and numerical investigations of self- mechanical property and seismic application. Smart Mater Struct 2016;25(7):
centring post-tensioned precast beam-to-column connections with steel top and 075001.
seat angles. Eng Struct 2021;226:111397. [31] Auricchio F, Taylor RL, Lubliner J. Shape-memory alloys: macromodelling and
[14] Yu HR, Li WB, et al. Seismic behavior of self-centring beam-to-column connections numerical simulations of the superelastic behavior. Comput Methods Appl Mech
with top-and-seat angles under mainshock-aftershock sequences. J Vibrat Shock Eng 1997;146(3–4):281–312.
2022;41(12):238–46. in Chinese.

218
H. Qian et al. Structures 52 (2023) 205–219

[32] Ahmadi O, Ricles JM, Sause R. Modeling and seismic collapse resistance study of a [40] Zhang A, Zhang Y, Li R, et al. Cyclic behavior of a prefabricated self-centring
steel SC-MRF. Soil Dyn Earthq Eng 2018;113:324–38. beam–column connection with a bolted web friction device. Eng Struct 2016;111:
[33] Yuan HX, Gao JD, Theofanous M, et al. Initial stiffness and plastic resistance of 185–98.
bolted stainless steel T-stubs in tension. J Constr Steel Res 2020;173:106239. [41] Ricles JM, Sause R, Wolski M, et al. Post-tensioned moment connections with a
[34] Yuan HX, Gao JD, Theofanous M. Numerical modelling and design of stainless steel bottom flange friction device for seismic resistant self-centring steel MRFs. 4th
double extended end-plate beam-to-column joints. Int J Steel Struct 2022;22(6): International conference on earthquake engineering. 2006.
1743–59. [42] Gao JD, Du XX, Yuan HX, et al. Hysteretic performance of stainless steel double
[35] Askariani SS, Garivani S, Hajirasouliha I, et al. Innovative self-centring systems extended end-plate beam-to-column joints subject to cyclic loading. Thin-Walled
using shape memory alloy bolts and energy dissipating devices. J Constr Steel Res Struct 2021;164:107787.
2022;190:107127. [43] Chaboche JL. Time-independent constitutive theories for cyclic plasticity. Int J
[36] Gao JD, Yuan HX, Du XX, et al. Structural behaviour of stainless steel double Plast 1986;2(2):149–88.
extended end-plate beam-to-column joints under monotonic loading. Thin-Walled [44] Chaboche JL. Constitutive equations for cyclic plasticity and cyclic viscoplasticity.
Struct 2020;151:106743. Int J Plast 1989;5(3):247–302.
[37] Francavilla AB, Latour M, Piluso V, et al. Simplified finite element analysis of [45] Liu XY, Bu YD, Wang YQ, et al. Ultra-low cyclic fatigue fracture of Q235B and
bolted T-stub connection components. Eng Struct 2015;100:656–64. Q345B steels and their butt welded joints. Int J Steel Struct 2022;22(2):430–49.
[38] Chen Y, Chen C, Chen C. Study on seismic performance of prefabricated self- [46] Federal Emergency Management Agency (FEMA). Seismic performance assessment
centring beam to column rotation friction energy dissipation connection. Eng of buildings: Volume 1-methodology, FEMA P-58-1. Washington, D.C: Federal
Struct 2021;241:112136. Emergency Management Agency; 2018.
[39] Gao JD, Yuan HX, Qian H, et al. Structural behaviour and design of stainless steel
end-plate beam-to-exterior column minor-axis joints. Thin-Walled Struct 2022;181:
110119.

219