You are on page 1of 16

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/325868153

Application of an Innovative SMA Ring Spring System for Self-Centering Steel


Frames Subject to Seismic Conditions

Article  in  Journal of Structural Engineering · August 2018


DOI: 10.1061/(ASCE)ST.1943-541X.0002127

CITATIONS READS

0 181

7 authors, including:

Wei Wang
Tongji University
92 PUBLICATIONS   527 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Seismic Resilience in Steel and Composite Structures View project

Progressive collapse behavior of steel and composite structures View project

All content following this page was uploaded by Wei Wang on 23 October 2018.

The user has requested enhancement of the downloaded file.


Application of an Innovative SMA Ring Spring System for
Self-Centering Steel Frames Subject to Seismic Conditions
Cheng Fang, M.ASCE 1; Wei Wang, M.ASCE 2; James Ricles, P.E., M.ASCE 3; Xiao Yang 4;
Qiuming Zhong 5; Richard Sause, P.E., M.ASCE 6; and Yiyi Chen 7

Abstract: This paper presents an innovative shape memory alloy (SMA) ring spring system used for high-performance steel beam-to-
Downloaded from ascelibrary.org by Tongji University on 10/22/18. Copyright ASCE. For personal use only; all rights reserved.

column connections. The ring spring system, which includes a series of SMA outer rings and high-strength alloy inner rings stacked in
alternation with mating tapered faces, is used in conjunction with high-strength bolts in typical extended end-plate connections to achieve
favorable self-centering and energy dissipation response while ensuring ease of construction. An experimental investigation on three proof-of-
concept connection specimens with varying SMA ring thicknesses and bolt preload levels is conducted first, where the key properties,
including strength, stiffness, ductility, self-centering capability, and energy dissipation, are discussed in detail. Following the experimental
program, a three-dimensional (3D) numerical study is carried out that enables a more in-depth interpretation of the test results and supports
the design of full-scale self-centering connections for a prototype steel moment frame. A system-level time-history analysis is then performed
on the frame with various layouts of the SMA connections, and it is shown that, compared with a conventional steel frame, the structure with
the SMA connections exhibits comparable maximum interstory drifts but significantly reduced residual drifts. The results also reveal that a
reasonable balance between structural performance and cost may be achieved by an appropriate arrangement of the SMA connections within
the structure. DOI: 10.1061/(ASCE)ST.1943-541X.0002127. © 2018 American Society of Civil Engineers.
Author keywords: Shape memory alloy (SMA); Ring spring system; Self-centering connection; Cyclic performance; Numerical study;
Time-history analysis.

Introduction the development of low-damage and high-performance alternatives


to conventional structural systems, and one of the promising
The extensive damage to steel beam-to-column connections after concepts is to introduce self-centering capabilities into a structure
the 1994 Northridge and 1995 Kobe earthquakes resulted in signifi- (Ricles et al. 2001; Garlock et al. 2005; Lin et al. 2013a, b). The
cant repair efforts and related economic losses (Miller 1998; goal of self-centering structural design is to minimize postearth-
Nakashima et al. 1998). After the 2011 Christchurch earthquake, quake structural damage and to eliminate permanent (residual)
hundreds of buildings in the affected zone were either completely interstory drift. One way of offering such capabilities in a building
or partially demolished, and it has been estimated that the total cost frame is to utilize a unique class of smart metals, i.e., shape
of the rebuild will be as much as NZ$40 billion (Wood et al. 2016). memory alloys (SMAs).
Lessons drawn from these major earthquakes provide impetus for SMAs have attracted great attention in various fields, including
1
aerospace, automotive, medicine, and robotics (Jani et al. 2014). As
Assistant Professor, State Key Laboratory of Disaster Reduction in a result of a solid-to-solid phase transformation, SMAs can exhibit
Civil Engineering, Dept. of Structural Engineering, Tongji Univ., Shanghai
two important macroscopic phenomena, namely, the shape memory
200092, China.
2
Professor, State Key Laboratory of Disaster Reduction in Civil effect (SME) and superelastic effect (SE) (Lagoudas 2008). When
Engineering and Dept. of Structural Engineering, Tongji Univ., loaded at a relatively low temperature, i.e., that below the marten-
Shanghai 200092, China (corresponding author). Email: weiwang@tongji sitic finish temperature (M f ) of the material, SMA is transformed
.edu.cn from the twinned martensitic phase to the detwinned martensitic
3
Professor, Advanced Technology for Large Structural Systems Engi- phase. This transformation leads to residual strain upon unloading,
neering Research Center, Dept. of Civil and Environmental Engineering, but heating the deformed SMA can promote strain recovery. This
Lehigh Univ., Bethlehem, PA 18015. phenomenon is known as SME. When loaded at a relatively high
4
Graduate Student, Dept. of Structural Engineering, Tongji Univ.,
temperature, i.e., above the austenitic finish temperature (Af ), SMA
Shanghai 200092, China.
5
Graduate Student, Dept. of Structural Engineering, Tongji Univ., exhibits SE, allowing it to undergo 8–10% recoverable strain
Shanghai 200092, China. immediately upon unloading. The potential for utilizing SMAs
6
Professor, Advanced Technology for Large Structural Systems Engi- in civil engineering applications was not recognized until the ini-
neering Research Center, Dept. of Civil and Environmental Engineering, tiation of two pioneering research projects, i.e., the Manside and
Lehigh Univ., Bethlehem, PA 18015. Istech projects, in the 1990s (Chang and Araki 2016). These re-
7
Professor, State Key Laboratory of Disaster Reduction in Civil Engi- search programs led to successful implementation of SMA devices
neering and Dept. of Structural Engineering, Tongji Univ., Shanghai in retrofitting several historic masonry structures. Since then, sig-
200092, China.
nificant research efforts have led to the development of SMA-based
Note. This manuscript was submitted on August 30, 2017; approved on
March 1, 2018; published online on June 4, 2018. Discussion period open braces (Yang et al. 2010; Massah and Dorvar 2014; Qiu and
until November 4, 2018; separate discussions must be submitted for indi- Zhu 2017a, b), base isolators (Bhuiyan and Alam 2013; Gur
vidual papers. This paper is part of the Journal of Structural Engineering, and Mishra 2013), dampers (Jennings and van de Lindt 2014;
© ASCE, ISSN 0733-9445. Silwal et al. 2015), displacement restrainers (Guo et al. 2012),

© ASCE 04018114-1 J. Struct. Eng.

J. Struct. Eng., 2018, 144(8): 04018114


concrete structures (Tazarv and Saiid Saiidi 2015; Billah and Alam from the conventional extended end-plate connection type. The
2016; Chen and Andrawes 2017), and steel connections (Fang et al. SMA ring spring system is the key part that constitutes the pro-
2014; Moradi and Alam 2014). Among the two properties of SMA, posed connection. As shown in Fig. 1(a), each ring spring system,
SE (which is also the focus of the current study) seems to be more typically used in conjunction with a high-strength bolt, includes a
appealing to the seismic research community due to the inherent series of SMA outer rings and high-strength alloy (HSA) inner
benefits of hysteretic damping, self-centering, and stable cyclic rings stacked in alternation with mating tapered faces. When the
repeatability. connection rotates, the bolt rows further away from the center
In particular, providing beam-to-column connections with a SE of rotation of the connection tend to be pulled, leading to expansion
property is considered to be one feasible approach to promote self- of the SMA outer rings as the HSA inner rings are compressed.
centering behavior. The main concept is to utilize SMA compo- This wedging action can generate a large force from the axial com-
nents to meet the ductility and energy dissipation requirements ponent of the contact force between the outer and inner rings, a
at the connection zone, while maintaining the adjacent members mechanism that significantly contributes to the moment resistance
(e.g., beams and columns) within the elastic range. SMA bolts of the entire connection. It should be noted that the HSA inner rings
are the most widely considered candidate for such connections. have to be sufficiently stiff and strong such that the main deforma-
Downloaded from ascelibrary.org by Tongji University on 10/22/18. Copyright ASCE. For personal use only; all rights reserved.

Abolmaali et al. (2006) found that T-stubs, which are typical sub- tion occurs in the SMA outer rings. The shear resistance of the con-
assemblies forming the entire steel connection [Eurocode 3 (CEN nection is provided by the high-strength bolts as well as the friction
2005)], have satisfactory self-centering properties when equipped between the end plate and the column face.
with SMA bolts. Speicher et al. (2011) further demonstrated the The superelastic characteristic of SMA causes the ring spring
benefit of using SMA bolts for shear tab connections through a system to exhibit a strong deformation recovery tendency during
series of physical tests. Ma et al. (2007) studied an end-plate unloading, resulting in deformation recovery of the entire connec-
SMA connection type through a numerical investigation, and the tion. When cyclic load is applied to the connection, energy is dis-
feasibility of these connections was later verified experimentally sipated by the hysteretic damping characteristic of the SMA
by Fang et al. (2014, 2017). It was shown that these connections material, as well as the friction between the outer and inner rings,
have excellent self-centering capability with moderate energy where the friction depends on the treatment of the contact surfaces.
dissipation. With initial confidence gained by these studies, more The required number of SMA outer rings depends on the required
diverse SMA-based steel connection types were proposed and ex- deformation of the connection, e.g., based on the maximum ex-
amined, including H beam-to-tubular column steel connections pected drift for the considered earthquakes. For typical moment-
(Wang et al. 2015b), combined SMA bolt and steel angle connec- resisting connections, only a small number of SMA outer rings
tions (Wang et al. 2015a, 2017), and SMA bolt-washer hybrid con- are required due to their pronounced deformability. This enables
nections (Yam et al. 2015). Apart from studies of individual a compact design of the connection, and hence spatial interference
connection performance, the overall structural behavior was also can be minimal. When the deformation capacity of the ring spring
investigated (DesRoches et al. 2010; Sultana and Youssef 2016).
system is reached, the gaps between the inner rings are closed,
However, the current SMA-based self-centering connections
and further deformation occurs in the adjacent steel members. This
may also have limitations. A number of studies have warned that
“locking” mechanism can effectively prevent damage to the SMA
the SMA bolts are susceptible to fracture in the threaded area
outer rings when the connection deformation demand is unexpect-
(Tamai et al. 2004; Abolmaali et al. 2006; Fang et al. 2015b),
edly large.
and the resulting loss of connection load carrying capacity may
The behavior of individual SMA outer rings can be understood
greatly increase the risk of building collapse under strong earth-
by conducting uniaxial cyclic compressive tests on a simple ring
quakes. Decreasing the stress demand over the threaded area by
spring system. Utilizing the test setup shown in Fig. 1(b), the typ-
increasing the net threaded-to-shank area ratio can improve the bolt
ical hysteretic response of the SMA outer rings used in this study is
ductility, but the associated material waste from the machining pro-
obtained. The SMA outer ring was fully loaded with 30 repeated
cess is considerable, and therefore the resulting connection design
loading cycles, and it shows satisfactory self-centering capability,
may be uneconomical. In addition, a certain level of prestrain needs
full flag-shaped hysteretic loops, and good repeatability. The geo-
to be applied to the SMA bolts, but this procedure can be difficult
for slender bolts due to unwanted twisting effects during preload- metric details of the SMA outer ring are given in Fig. 1(a).
ing. Moreover, a more efficient energy dissipation mechanism may
be desirable for SMA connections. In order to address the issues Test Specimens
identified above, a novel type of self-centering connection, which
employs a SMA ring spring system, is proposed in this paper. The A total of three proof-of-concept beam-to-column connection spec-
study commences with an experimental investigation on three imens were designed and tested. The test parameters are (1) the
proof-of-concept connection specimens, and a numerical study thickness of the SMA outer rings and (2) the bolt preload level.
is subsequently carried out to calibrate the test results and to facili- Two groups of SMA outer rings, with two ring thicknesses (T)
tate the design of full-scale self-centering connections for a typical of 3 and 5 mm, were produced for this test program. These rings
steel moment frame. Finally, a system-level time-history analysis is were machined with a permissible geometric tolerance of 0.05 mm
performed that reveals some key engineering demand characteris- from commercial Ti-50.8at.%Ni (i.e., 50.8% atomic percentage of
tics of the structure under a suite of seismic excitations. nickel with the balance from titanium) SMA raw tubes, and the
rings were then heat treated at 400°C for 30 min, followed by water
quenching. The dimension of the SMA outer rings, as detailed in
Experimental Program Fig. 1(a), was determined based on a preliminary numerical study
conducted by the authors and coworkers (Fang et al. 2015a) such
that an effective utilization of the SMA material can be achieved.
Basic Design Concept The inner rings were made from 38CrMoAl alloy and were de-
In order to achieve the required connection performance while en- signed to fit the tapered angle of the SMA outer ring. Grease was
suring ease of construction, the proposed connection is modified applied between the outer and inner rings.

© ASCE 04018114-2 J. Struct. Eng.

J. Struct. Eng., 2018, 144(8): 04018114


Downloaded from ascelibrary.org by Tongji University on 10/22/18. Copyright ASCE. For personal use only; all rights reserved.

(a)

50
T = 3mm

40

First cycle
Load (kN)

30

20
Last cycle

10

0
0 1 2 3 4 5 6
Deformation (mm)

(b)

Fig. 1. Connections with SMA ring springs: (a) geometric configurations and dimensions of test specimens; and (b) typical behavior of individual
SMA outer ring.

Each connection specimen consists of a Q345 (nominal ring is 5 mm, this arrangement results in a total deformation capac-
f y ¼ 345 MPa) H180 × 120 × 12 × 16 built-up steel beam, a ity of 15 mm before bolt preload. Each internal bolt, which is ex-
Q345 H300 × 250 × 16 × 20 built-up steel column, a series of pected to have lower deformation demand, is equipped with a ring
beam and column stiffeners, and a connection with eight high- spring system containing two SMA outer rings, leading to a 10-mm
strength bolts, i.e., four external ones and four internal ones. For deformation capacity before preload. To study the initial stiffness of
each external bolt, the associated ring spring system includes three the connection, two levels of bolt preload were considered, namely,
SMA outer rings. As the deformation capacity of one SMA outer P1.5 and P2.0, corresponding to precompression deformations of

© ASCE 04018114-3 J. Struct. Eng.

J. Struct. Eng., 2018, 144(8): 04018114


Table 1. Details of test specimens
Specimen Δp per SMA Number of SMA rings Number of SMA rings Total Δp per Total Δp per Thickness of SMA
code ring (mm) per external bolt per internal bolt external bolt (mm) internal bolt (mm) ring (mm)
P1.5T3 1.5 3 2 4.5 3.0 3.0
P2.0T3 2.0 3 2 6.0 4.0 3.0
P1.5T5 1.5 3 2 4.5 3.0 5.0
Note: Δp = value of precompression deformation via bolt preload.

1.5 and 2.0 mm for each SMA outer ring, respectively. After de- ring spring system was subjected to compressive deformation.
ducting the precompression deformations, the ring spring systems According to visual observations, the ring spring system in the
are expected to provide a connection rotation capacity correspond- specimens was almost fully compressed at 4% drift. No damage
ing to at least 4% drift. For easy reference, each specimen is des- to the ring spring system was observed at the end of the test,
Downloaded from ascelibrary.org by Tongji University on 10/22/18. Copyright ASCE. For personal use only; all rights reserved.

ignated with a specimen code, starting with the precompression for and the high-strength bolts remained tightened when the specimens
each SMA outer ring, and ending with the thickness of the SMA returned to their initial centered position. These observations imply
outer ring (e.g., P1.5T3), as detailed in Table 1. It is worth men- that the connection may be well prepared for aftershocks or future
tioning that the current test specimens were not designed based on earthquakes. In addition, no shear slippage was observed between
any prototype building. They are proof-of-concept connections the end plate and column face during the entire loading history.
used to verify the feasibility of the proposed connection concept. When the tests were completed, the ring spring systems were re-
moved from each specimen for a more detailed inspection. Minor
Test Setup, Instrumentation, and Loading Protocol wearing scratches were observed over the greased contact surfaces
of both the inner and outer rings. The scratches may be attributed to
The test setup is schematically shown in Fig. 2(a). The steel column an imperfect uniaxial compression state of the ring spring system,
was oriented horizontally and fixed to the strong floor. The column since an inclined gap exists between the end plate and column face
was slightly oversized to remain elastic for repeatable use. The steel during the rotation of the connection. The wear of the greased con-
beam was oriented vertically and attached to the steel column using tact surfaces may gradually increase the friction force, but it is be-
the proposed connection. The load was applied near the free end lieved that the minor scratches should have limited influence on the
of the beam at a distance of 1.35 m from the column face. The out- overall connection behavior. It is noted that if wearing scratches
of-plane movement of the steel beam was prevented by a lateral are severe, which is likely to occur when grease is not applied,
brace located in the region between the load and column face. the self-centering capability of the ring spring system would be
The typical instrumentation layout is shown in Fig. 2(b). A series compromised, and the resulting material imperfection may also
of strain gauges and rosettes were placed on the key structural com- cause premature fracture of the SMA outer ring. In addition, severe
ponents, including the steel beam, steel column, end plate, and stiff- wearing scratches should be avoided from a reusability point of
eners. The beam tip displacement, column end displacements, local view.
concentrated connection rotation, and shear slippage between the The moment-drift curves of the specimens are shown in Fig. 4.
end plate and the column face were measured via a suite of linear The moment is calculated through multiplying the beam tip load by
variable differential transformers (LVDTs). the lever arm Lr . Flag-shaped hysteretic responses are observed
The loading procedure for the specimens involved two steps. with negligible residual deformation. The moment-drift curves first
Bolt preload was first applied by tightening the high-strength bolts,
follow a linear ascending curve, and the initial yield moment ranges
and this step was aided by placing temporary LVDTs at the bolt nut
between 12 and 20 kN-m, depending on the size of the SMA outer
to ensure an appropriate level of precompression applied to the ring
rings or the preload level. In particular, because of an increase in
spring system. Subsequently, a cyclic load was applied onto the
bolt preload, the yield moment of specimen P2.0T3 is increased
connection by a 200-kN hydraulic-servo actuator, where the load-
compared with that of specimen P1.5T3. Under the same bolt pre-
ing sequence is given in Table 2. The drift value is calculated by
load level, the yield moment seems to be proportional to the thick-
θ ¼ Δ=Lr , where Δ is the displacement of the loading point, and
ness of the SMA outer ring. After the linear ascending branch, a
Lr is the lever arm, i.e., the distance between the loading point and
nonlinear moment-drift response occurs with wide hysteretic loops.
the column flange face (1.35 m). The ambient temperature during
With a combination of forward transformation of the SMA and fric-
the tests ranged between 22 and 28°C. In addition, tension coupon
tion action, pronounced hardening is exhibited after attaining the
tests, conforming to ISO 6892-1 (ISO 2009), were conducted to
yield moment. At 4% drift, the peak moment of the specimens
obtain the material properties of the steel members, and the results
ranges between approximately 40 and 60 kN-m, which largely
are given in Table 3. The fracture elongation percentage was ob-
depends on the size of the SMA outer rings. The hysteretic re-
tained based on a gauge length of 100 mm.
sponse is generally stable, although a decrease in the reverse pla-
teau is observed with an increase in drift, which could be caused by
Test Results and Discussions transformation induced fatigue (TIF), a phenomenon commonly
observed in SMA (Lagoudas 2008).
The behavior of the specimens is also reflected by the strain
Test Observations gauge readings (readings from the rosettes are presented using
In general, the three test specimens exhibited anticipated deforma- von Mises equivalent strains). As shown in Fig. 5, the strains of
tion modes of behavior. The main inelastic deformation occurred in the beam flanges near the connection are lower than the yield strain,
the connection, while the beam, column, and end plate remained indicating that the beam remained elastic throughout the entire test-
elastic, suggesting that minimal postearthquake repair would be re- ing procedure. Similar results are observed in the other parts of the
quired. As shown in Fig. 3, the deformation of the connection fea- beam and the steel column. However, Fig. 5 shows that the beam
tured a gap opening between the end plate and column face as the stiffeners experienced yielding at larger drifts. This is due to the

© ASCE 04018114-4 J. Struct. Eng.

J. Struct. Eng., 2018, 144(8): 04018114


Downloaded from ascelibrary.org by Tongji University on 10/22/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 2. Test arrangement: (a) test setup; and (b) instrumentations.

large thrust force applied onto the stiffeners when the associated Further Discussion on Test Results
beam flange is the center of compression during the rotation of Stiffness and strength are two fundamental characteristics for steel
the connection. Nevertheless, the local yielding of the stiffeners beam-to-column connections. In Eurocode 3 (CEN 2005), connec-
seems to have a limited influence on the overall connection behav- tion stiffness is classified through comparisons with the rigidity of
ior. Generally speaking, the connection behavior satisfies the initial the adjacent steel beam EI b =Lb , where E is the Young’s modulus
design intention that the SMA ring spring system provides the and I b and Lb are the second moment of area and span of the beam,
deformation capacity, while the main structural members remain respectively. A connection can be classified as rigid provided that
almost undamaged during earthquakes. its initial stiffness K ini is greater than 25EI b =Lb and 8EI b =Lb for

© ASCE 04018114-5 J. Struct. Eng.

J. Struct. Eng., 2018, 144(8): 04018114


Table 2. Loading protocols for test specimens nominally pinned, partial-strength, or full-strength connection.
Sequence Drift level (%) Number of cycles per drift level The criteria for nominally pinned and full-strength connections
are Mmax =M b;pl < 0.25 and M max =Mb;pl > 1.0, respectively, and
1 0.375 3
a partial-strength connection is that with Mmax =M b;pl in between.
2 0.5 3
3 0.75 3
Based on these limits, it can be seen that the test specimens are all
4 1.0 4 partial-strength connections. It should be noted that, in practice, the
5 1.5 2 classes of the connections are adjustable by ensuring an appropriate
6 2.0 2 match between the connection stiffness/strength and beam size.
7 3.0 2 In addition, no degradation of moment resistance is observed by
8 4.0 2 the end of the tests. This indicates that the specimens exhibit a duc-
tility capacity of at least 4% drift under the cyclic loading, and
therefore, they can be potentially employed for ductility class high
(DCH) structures and special moment frames (SMFs), as required
Table 3. Typical steel material properties from coupon tests by Eurocode 8 (CEN 2004) and AISC (2010), respectively. Apart
Downloaded from ascelibrary.org by Tongji University on 10/22/18. Copyright ASCE. For personal use only; all rights reserved.

Yield strength Ultimate strength Fracture from the favorable ductility performance, the specimens also show
Sample plate f y (MPa) fu (MPa) elongation (%) a satisfactory self-centering response with no induced residual de-
Beam web and 364 502 26.9 formation after cyclic loading to 4% drift. The self-centering ten-
stiffeners dency of the specimens can be further evaluated by the restoring
Beam flange 392 514 25.9 moment M rst , as illustrated in Fig. 6(a). A positive value of M rst
End plate 405 513 27.0 indicates no residual deformation, whereas a negative value means
that residual deformation is developed; in addition, the higher
M rst is, the greater the self-centering tendency that the structure ex-
hibits (but possibly with reduced energy dissipation). As shown in
moment-resisting frames and nonsway braced frames, respectively.
Fig. 6(a), the restoring moments of the three specimens follow a
A connection is considered as nominally pinned when K ini ≤
similar trend, i.e., a decrease with an increase in drifts. However,
0.5EI b =Lb . For the cases where 0.5EI b =Lb ≤ K ini ≤ 25EI b =Lb
the values of M rst are always kept above zero, indicating a reliable
(or 8EI b =Lb ), the connection can be defined as semirigid. The ini-
self-centering response. Moreover, it is found that increasing the
tial stiffness K ini of the current specimens is summarized in Table 4,
precompression level of the ring spring system leads to increased
in which the values of K ini are obtained based on the concentrated M rst at early stages of loading, but due to more rapid TIF-induced
connection rotation response rather than the overall drift response. degradation under a higher strain level, the reverse transformation
The concentrated connection rotation is defined as the rotation con- plateau decreases more quickly, and as a result the influence of in-
tributed by the connection only, i.e., that excluding the beam de- creasing precompression on M rst becomes less at larger drifts.
formation. Assuming a beam length of 3.6 m, i.e., span-to-depth The flag-shaped self-centering response is accompanied by
ratio = 20, the test specimen connections fall into the categories energy dissipation. A common index used to reflect the energy dis-
of semirigid and rigid connections for moment-resisting frames sipation capacity of a structure/member is equivalent viscous
and nonsway braced frames, respectively. The results also show damping (EVD) ξ eq , which is expressed by
that K ini could be increased by increasing either the bolt preload
level or the size of the SMA outer rings. WD
ξ eq ¼ ð1Þ
A similar approach for the classification of strength is given in 2πW E
Eurocode 3 (CEN 2005). By comparing the connection maximum
moment resistance (M max ) against the plastic moment resistance of where W D = energy loss per cycle, i.e., the area enveloped by each
the connected beam (Mb;pl ), a connection can be classified as a complete moment-drift hysteretic cycle; and W E = energy absorbed

Fig. 3. Typical deformation and failure modes of specimens.

© ASCE 04018114-6 J. Struct. Eng.

J. Struct. Eng., 2018, 144(8): 04018114


50 connections equipped with normal SMA bolts (Fang et al. 2014,
P1.5T3
40 2017), where the typical EVDs at small drifts are much lower.
30
It is believed that the improved EVD for the current specimens
is attributed to the friction effect over the tapered face.
20
Moment (kNm)

10

0
Numerical Study of SMA Connections
-10
3D Modeling Strategy and Validation
-20
Following the experimental study, a 3D numerical study was per-
-30
formed to serve two main purposes: (1) to enable a more in-depth
-40 Test result interpretation of the test results and (2) to support the design of full-
-50
FE prediction scale self-centering connections for a typical SMF. The overall
Downloaded from ascelibrary.org by Tongji University on 10/22/18. Copyright ASCE. For personal use only; all rights reserved.

-0.05 -0.04 -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05 dynamic response of the SMF subjected to seismic excitations will
Drift (rad) be discussed later in the next section.
The specimens were simulated using the nonlinear finite-
50
P2.0T3 element (FE) analysis program ABAQUS (2012). Except for the
40 SMA outer rings, which were modeled by 3D eight-node linear
30
brick elements (C3D8 elements), the other structural components
were discretized by reduced integration C3D8R elements with
20 hourglass control. For the greased tapered faces, a friction coeffi-
Moment (kNm)

10 cient of 0.15 was provisionally assumed, and no penetration was


allowed in the normal contacting direction. For the remaining con-
0
tact pairs, e.g., that between the end plate and the column face, a
-10 general friction coefficient of 0.3 was considered. It should be
-20 noted that these coefficients were taken based on typical values
[Eurocode 3 (CEN 2005)], which were not fully justified. A more
-30
detailed numerical investigation on the influences of friction on the
-40 Test result behavior of ring spring systems is available from Fang et al.
FE prediction (2015a), which is beyond the focus of this study. The welds at the
-50
-0.05 -0.04 -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05 beam, column, and stiffeners were all simulated via tie interactions.
Drift (rad) Taking advantage of symmetry, only one half-side of the specimens
was modeled for computational efficiency, as shown in Fig. 7.
80
P1.5T5 For the material properties, idealized multilinear stress-strain re-
60 sponse with the von Mises yield criterion was considered for the
steel components and the HSA inner rings. An integrated Auricchio
40 model (Auricchio 2001) was used to capture the superelastic behav-
ior of the SMA outer rings. The key uniaxial material property for
Moment (kNm)

20 SMA is illustrated in Fig. 7. It should be noted that it is very dif-


ficult to conduct material tests of coupons directly extracted from
0
the SMA outer rings; therefore, the material properties used in this
-20 study were based on the relevant test data on SMA bars, which
were provided by the same supplier, with the same composition
-40 (commercial Ti-50.8at.%Ni), and had experienced the same heat
treatment procedure. The uncertainty of the material properties
-60
Test result might be a source of modeling discrepancy. Another issue associ-
FE prediction ated with the Auricchio model is its incapability of simulating the
-80
-0.05 -0.04 -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05 possible TIF phenomenon of the SMA material. This was ad-
Drift (rad) dressed by a simple yet effective approach, namely, the SMA-steel
hybrid modeling strategy. In each SMA outer ring, a thin layer of
Fig. 4. Moment-drift responses of specimens. steel cylindrical sheet was embedded, which was expected to ex-
pand synchronously with the surrounding SMA. This steel part has
the same Young’s modulus and yield strength as that of the SMA,
but exhibits plasticity, which mimics the TIF behavior that induces
in a linear system undergoing the same positive and negative maxi- residual deformation. The thickness of the steel cylindrical sheet is
mum deformations and forces as those of the considered nonlinear adjustable to ensure reasonable agreement between the FE predic-
system. As shown in Fig. 6(b), the EVD first increases with an in- tion and test result. In this study, the thickness of the steel part is
crease in drifts and is then stabilized between 12 and 16% when the taken as 12% of the overall thickness (T) of the SMA outer ring
drift exceeds 1%. It is also found that even at the first cycle, a rea- such that the ring spring behavior can be well simulated.
sonable amount of energy has been dissipated, leading to EVDs of The testing procedures used for the test specimens were consid-
up to 10 at 0.375% drift. The result implies that these connections ered in the numerical study. Bolt preload was first applied to enable
could start contributing to energy dissipation at small structural the required precompression of the ring spring system, and sub-
deformations. This behavior is quite different from that of the sequently a displacement controlled load was applied at the end

© ASCE 04018114-7 J. Struct. Eng.

J. Struct. Eng., 2018, 144(8): 04018114


2000 5000
P1.5T5:
1500 Beam stiffeners
Yield microstrain = 1784 4000 T7

Equivalent Microstrain
1000 T8

3000
Microstrain

500

0
Yield microstrain = 1705
2000
-500

-1000 1000

P1.5T5: Steel beam


-1500 Yield microstrain = 1784
S1 0
S2
-2000
Downloaded from ascelibrary.org by Tongji University on 10/22/18. Copyright ASCE. For personal use only; all rights reserved.

-0.05 -0.04 -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05 -0.05 -0.04 -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05
Drift (rad) Drift (rad)

Fig. 5. Typical strain gauge readings at key locations.

Table 4. Stiffness and strength of test specimens


Stiffness Strength
Specimen K ini K ini =ðEI=LÞ a
Classification b
M max M max =M b;pl Classification
P1.5T3 22,341 13.5 Semirigid 39.8 0.28 Partial strength
P2.0T3 29,034 17.5 Semirigid 45.6 0.32 Partial strength
P1.5T5 27,483 15.6 Semirigid 58.3 0.40 Partial strength
Note: The unit for K ini is kNm=rad, and the unit for Mmax is kNm.
a
L=Db ¼ 20 is assumed.
b
Based on the criterion for moment resisting frames.

10 20
Equivalent viscous damping (%)
Restoring moment M rst (kNm)

8 16

6 12

4 8

2 4
P1.5T3 P1.5T3
P2.0T3 P2.0T3
P1.5T5 P1.5T5
0 0
0.00 0.01 0.02 0.03 0.04 0.05 0.00 0.01 0.02 0.03 0.04 0.05
(a) Drift (rad) (b) Drift (rad)

Fig. 6. Discussion on test results: (a) self-centering ability; and (b) energy dissipation ability.

of the beam. The typical deformation modes of the model at two Reasonable agreement is also found between the FE-predicted
representative stages, namely, immediately after bolt precompres- and experimental moment-drift responses, as shown in Fig. 4. The
sion and at 4% drift, are shown in Fig. 7. The predicted deformation initial stiffness and moment resistance of the specimens are well
modes are consistent with the test observations, and it is reaffirmed predicted, and the general hysteretic response and self-centering
that the main contribution to the connection rotation is from com- properties are sufficiently captured. In particular, the decrease of
pression of the ring spring system. Consistent with the test results, the reverse transformation plateau with increasing drift (caused
the von Mises stress contour suggests that at 4% drift, the beam and by TIF) is reasonably reflected by the steel cylindrical sheet.
column remain elastic, but the beam stiffeners experience minor Because of certain limitations of the modeling technique, however,
yielding. the predicted loading and unloading curves do not exactly follow

© ASCE 04018114-8 J. Struct. Eng.

J. Struct. Eng., 2018, 144(8): 04018114


Downloaded from ascelibrary.org by Tongji University on 10/22/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 7. Representative FE model and typical deformation mode (Specimen P1.5T5).

the test results. For instance, the predicted peak moment at 4% drift All the connections of the conventional MRF are rigid/full-strength
tends to be slightly lower than the test result. It is believed that this connections. Following the strong column–weak beam rule, first
discrepancy may be caused by variations of the friction effect over yielding is expected to occur over the beam section near the
the tapered faces at different compression states of the ring spring beam-to-column connection region. The panel zones are reinforced
system, and it is deduced that a more pronounced friction force, by doubler and continuity plates to ensure adequate yield resis-
leading to increased moment resistance, may exist over the tapered tance. The sections selected for the beams and columns at different
faces when the ring spring system approaches the fully compressed floor levels are noted in Fig. 8.
condition. The variation of friction effect may also explain the The conventional prototype steel frame building (hereafter,
slight discrepancy of the unloading response, especially the frame-STL) is modified in this study to enable the use of the pro-
descending branch when the model is initially unloaded (i.e., at posed SMA self-centering connections. Two types of modified
reversal of friction). This variation is difficult to reflect in the frame buildings are considered, namely, frame-SMA1 and frame-
FE model, but its influence on the overall connection behavior SMA2. The two modified frame buildings have the same framing
is acceptably small. arrangement and member sizes as frame-STL, but differ in connec-
tion arrangement. For frame-SMA1, all the conventional rigid con-
nections are replaced by SMA connections, where this strategy
Prototype Building aims to minimize the damage to the main frame and to maximize
Following the validation study, self-centering connections were de- the self-centering performance of the structure. Perhaps as a more
signed for a prototype building. The conventional (original) proto- economical solution, frame-SMA2 adopts partially distributed
type steel frame is a five-story, 4 × 3 bay office building (located at SMA connections, as illustrated in Fig. 8. This arrangement sheds
ground type B soil) designed by Dimopoulos et al. (2013) accord- light on the potential for obtaining desirable system-level behavior
ing to Eurocode 8 (CEN 2004). In the longitudinal direction, the with a well-balanced cost of construction through a proper alloca-
structure includes four identical moment-resisting frames (MRFs), tion of the SMA connections.
and along the transverse direction, the lateral load is resisted by
braced frames. The yield strengths for the steel beams and steel
SMA Connection Design
columns are 300 and 350 MPa, respectively. The typical plan
and elevation views of the building are shown in Fig. 8, and the As currently there is no codified recommendation available for de-
analysis performed in this study is based on a typical 2D MRF signing self-centering connections, the following rules are prelimi-
in the longitudinal direction. Being slightly different from the narily considered: (1) the initial stiffness of the SMA connections
frame considered in Dimopoulos et al. (2013), the roof beam in prior to decompression is comparable to the rigid connection
this study is considered to be the same as that for the lower floor. criterion such that the conventional and modified buildings could

© ASCE 04018114-9 J. Struct. Eng.

J. Struct. Eng., 2018, 144(8): 04018114


Downloaded from ascelibrary.org by Tongji University on 10/22/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 8. Conventional and modified prototype buildings and typical SMA connection responses.

have similar initial elastic stiffness as well as fundamental period modeling assumption, the typical moment-concentrated rotation
of vibration; (2) the yield moment of the SMA connection is taken response of the designed connections is shown in Fig. 8.
as approximately 80% of the yield moment resistance (M e ) of the
steel beam in order to ensure that the inelastic deformation first
occurs in the connection while the steel beam is still elastic; (3) the System-Level Analysis
ring spring system is designed to provide a concentrated connection
rotation capacity of at least 0.04 radians; and (4) at 0.03-radian System-Level Modeling
concentrated rotation, the corresponding moment resistance of the
connection approaches the plastic moment resistance (Mp ) of the To facilitate nonlinear dynamic analysis of the prototype frame
buildings, a simplified model for the SMA connections was devel-
steel beam, which means that a certain level of beam yielding is
oped using the nonlinear dynamic analysis program OpenSees
expected at large structural deformations.
(Mazzoni et al. 2006), as schematically illustrated in Fig. 9(b).
Based on these basic design rules, the self-centering connections
The beam and column members were simulated via nonlinear
with SMA ring spring systems were designed to match the section
force-based beam-column elements with fiber sections, such that
properties of the IPE400 and IPE450 beams used for the prototype
the development and distribution of plasticity along the length
building. As shown in Fig. 9(a), a connection typically consists of
and depth of the members are included. Each fiber was modeled
six bolt rows to provide an adequate level of moment resistance,
using Steel01 material with a postyielding stiffness ratio of 0.01.
where each bolt includes one ring spring system. Taking the
For the SMA connection, rigid elements were used to represent
SMA connection for IPE400 beams for instance, the external diam- the end plate and the corresponding column flange, while the gap
eter (De ) and thickness (T) of the SMA outer rings are 65 and opening response was simulated by applying appropriate spring
6 mm, respectively, which are larger than those used for the test sets in between. For each side of the connection, six rows of
specimens. Each SMA outer ring provides a deformation capacity zero-length spring element sets, representing the associated SMA
of 11 mm, and the precompression for each outer ring is 3 mm. ring spring systems, were arranged along the height of the beam-to-
The newly designed connections were simulated with ABAQUS column interface according to their positions. The zero-length
(2012), where the same material property and modeling strategy spring elements for each row of SMA ring springs include
as those considered in the validation study were adopted. However, two parallel elements with different hysteresis models, namely,
caution should be exercised when applying the same SMA material a flag-shaped hysteresis model and steel hysteresis model. A
parameters to SMA outer rings with different sizes, since these SelfCentering material was used for the flag-shaped hysteresis
rings may behave differently; i.e., there may be a size effect. of the SMA ring springs. A Hardening material was used to supple-
Although it has been found that SMA bars with different diameters ment the flag-shaped hysteresis model, so that the aforementioned
may have similar stress-strain responses after appropriate heat treat- TIF phenomenon of the SMA material can be included. In order
ment (Wang et al. 2016), no test data are available for the SMA to reasonably match the ABAQUS results, the steel hysteresis
rings. This uncertainty in stress-strain response is not considered model was assumed to contribute to 20% of the total strength.
in the current study, but should be kept in mind when interpreting In addition, two zero-length contact spring elements were added
the analysis results. Nevertheless, the ABAQUS predictions (with to the top and bottom of the interface to allow simulation of the gap
the modeling approach validated) for the connections should be opening mechanism. These springs have a linear elastic response in
adequate for demonstrating the fundamental system-level seismic compression, but deform freely in tension (i.e., without force). The
behavior of steel frames with SMA connections. Employing this value of the compression stiffness was taken as 20 times the axial

© ASCE 04018114-10 J. Struct. Eng.

J. Struct. Eng., 2018, 144(8): 04018114


Downloaded from ascelibrary.org by Tongji University on 10/22/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 9. Full-scale SMA connections: (a) detailed design; and (b) OpenSees model.

© ASCE 04018114-11 J. Struct. Eng.

J. Struct. Eng., 2018, 144(8): 04018114


stiffness of the beam, a value that provides sufficient accuracy and The maximum interstory drift (MID) and residual interstory
ensures good numerical convergence. A spring element with a large drift (RID) responses are considered as two major engineering
elastic stiffness was added at the midheight of the model to provide demand parameters for assessing the overall dynamic response
the shear resistance of the connection. The panel zone of the col- of the prototype frame buildings subjected to the 20 ground mo-
umn includes two vertical and two horizontal rigid elements, where tions. The median and 84th percentile values of the two quantities
the nodes at the intersection point are connected by a zero-length over the heights of the frames are shown in Figs. 10(b and c). It is
rotational spring element to reflect the bilinear elastoplastic shear observed that, although the SMA frames typically have less energy
force-deformation panel zone behavior. dissipation than frame-STL, all three frames exhibit similar MIDs,
with median values of approximately 1.5 and 2.5% under the DBE
and MCE, respectively. The comparable median MID values are
Ground Motions attributed to the nonaccumulation of deformation for the SMA
Each of the three prototype buildings was subjected to a suite of 20 frames with flag-shaped hysteretic responses, whereas the inelastic
earthquake ground motions recorded at locations with site class B deformations of frame-STL, which are not as easily recovered, ac-
conditions, with an average shear wave velocity between 360 and cumulate during the seismic response. In addition, it is found that,
Downloaded from ascelibrary.org by Tongji University on 10/22/18. Copyright ASCE. For personal use only; all rights reserved.

800 m=s [Eurocode 8 (CEN 2004)]. These earthquake records were during the numerical simulation using the 20 ground motions, the
taken from 9 events in different countries and regions. The event largest displacement that the bolt row springs experience is approx-
magnitudes Mw of the ground motions range from 6.53 to 7.62. The imately 15 mm (corresponding to a maximum hoop strain level of
ground motions were scaled to the design basis earthquake (DBE) around 4.5% according to the 3D ABAQUS model), which is
level using the scaling procedure recommended by Somerville smaller than the maximum deformation capacity of 24 mm for
(1997), and they were further scaled by 1.5 to represent the maxi- IPE 400. This indicates that the SMA ring springs are not fully
mum considered earthquake (MCE) level. The details of the scaled compressed when the MID is achieved.
ground motions are given in Table 5. For the RID response, frame-SMA1 and frame-SMA2 have es-
sentially no residual drift under the DBE, whereas frame-STL has
median and 84th percentile RIDs of 0.07 and 0.24%, respectively.
Seismic Response Results and Discussions Under the MCE, the median and 84th percentile RID values for
Prior to the dynamic time-history analysis, a static push-pull analy- frame-STL increase to 0.21 and 0.56%, respectively. In addition,
sis, employing an inverted triangularly distributed lateral force, was the RID values for frame-STL have a much larger dispersion than
performed to understand several key system-level hysteretic char- the MID values (although individual ground motion responses are
acteristics of the prototype frames, as shown in Fig. 10(a), where V not shown in the figure). On the other hand, both SMA frames have
is the base shear and W is the seismic weight. As anticipated, small RID values under the MCE, with the median values generally
frame-STL exhibits a full hysteretic loop response with a large smaller than 0.05%, and the 84th percentile values below 0.15%.
static residual drift upon unloading. Frame-SMA1 and frame- It is of interest to note from Fig. 10(c) that frame-SMA1 and frame-
SMA2 tend to have less static residual drift. The static lateral re- SMA2 have comparable residual drift levels, although the number
sistance of the two SMA frames is slightly less than that of the of the SMA connections used for the latter is half of that for the
conventional frame, because the yield moments of the SMA con- former. This phenomenon can be explained by the principle of
nections are designed to be less than the yield moments of the steel probabilistic self-centering (Eatherton and Hajjar 2011), where
beams, as mentioned before. The three frames have a similar initial the dynamic residual drift of any system is often less than the maxi-
stiffness, and as a result the natural periods of the corresponding mum possible static value achieved after load is slowly decreased
prototype buildings are close, ranging between 1.070 and 1.094 s. from the peak drift. In other words, as long as the static reversal

Table 5. Information on selected ground motions


Scale factor
Earthquake name Station Component Magnitude (Mw ) Distance (km) DBE MCE
Imperial Valley 1979 Cerro Prieto H-CPE237 6.53 15.19 2.05 3.08
Loma Prieta 1989 Hollister-S&P HSP000 6.93 27.67 0.72 1.08
Loma Prieta 1989 Woodside WDS000 6.93 33.87 3.49 5.24
Loma Prieta 1989 WAHO WAH090 6.93 17.47 1.20 1.80
Manjil 1990 Abbar ABBAR-T 7.37 12.56 0.70 1.05
Cape Mendocino 1992 Fortuna-Fortuna Blvd FOR000 7.01 15.97 2.47 3.71
Cape Mendocino 1992 Rio Del Overpass-FF RI0360 7.01 14.33 1.25 1.88
Landers 1992 Desert-Hot Springs LD-DSP000 7.30 21.78 2.37 3.56
Northridge 1994 LA-W 15th St W15090 6.69 25.6 2.86 4.29
Northridge 1994 Moorpark-Fire Sta MRP180 6.69 16.92 1.94 2.91
Northridge 1994 N Hollywood-Cw CWC270 6.69 7.89 1.33 2.00
Northridge 1994 Santa Susana Ground 5108-360 6.69 1.69 1.95 2.93
Northridge 1994 LA-Brentwood VA 0638-285 6.69 12.92 2.12 3.18
Northridge 1994 LA-Wadsworth VA 5082-235 6.69 14.55 1.54 2.31
Kobe 1995 Nishi-Akashi NIS090 6.90 7.08 1.19 1.79
Kobe 1995 Abeno ABN090 6.90 24.85 2.49 3.74
Chichi 1999 TCU105 TCU105-E 7.62 17.18 2.39 3.59
Chichi 1999 CHY029 CHY029-N 7.62 10.97 1.32 1.98
Chichi 1999 CHY041 CHY041-N 7.62 19.83 1.40 2.10
Hector 1999 Hector HEC090 7.13 10.35 1.04 1.56

© ASCE 04018114-12 J. Struct. Eng.

J. Struct. Eng., 2018, 144(8): 04018114


0.4 0.4 0.4

0.2 0.2 0.2

V/W

V/W
V/W

0.0 0.0 0.0

-0.2 -0.2 -0.2

-0.4 -0.4 -0.4


Frame-STL Frame-SMA1 Frame-SMA2
-0.04 -0.02 0.00 0.02 0.04 -0.04 -0.02 0.00 0.02 0.04 -0.04 -0.02 0.00 0.02 0.04
Roof drift (rad) Roof drift (rad) Roof drift (rad)

(a)
Downloaded from ascelibrary.org by Tongji University on 10/22/18. Copyright ASCE. For personal use only; all rights reserved.

5 DBE level 5 MCE level

4 4
STL (med.)
STL (med.)
STL (84p.)
STL (84p.) SMA1 (med.)
Story

Story
SMA1 (med.)
3 3 SMA1 (84p.)
SMA1 (84p.) SMA2 (med.)
SMA2 (med.)
SMA2 (84p.)
SMA2 (84p.)
2 2

med. = median med. = median


th th
1 84p. = 84 percentile 1 84p. = 84 percentile
0.00 0.01 0.02 0.03 0.00 0.01 0.02 0.03 0.04
Maximum inter-story drift (rad) Maximum inter-story drift (rad)

(b)

5 5 MCE level
DBE level

4 4
Story

STL (med.)
Story

STL (med.)
3 3
STL (84p.) STL (84p.)
SMA1 (med.) SMA1 (med.)
SMA1 (84p.) SMA1 (84p.)
2 SMA2 (med.) 2 DS1 DS2 SMA2 (med.)
SMA2 (84p.) SMA2 (84p.)
med. = median med. = median
th th
1 DS1 84p. = 84 percentile 1 84p. = 84 percentile
0.000 0.001 0.002 0.003 0.004 0.005 0.000 0.002 0.004 0.006 0.008 0.010
Residual inter-story drift (rad) Residual inter-story drift (rad)

(c)

Fig. 10. System-level analysis results: (a) static push-pull response; (b) maximum interstory drift response; and (c) maximum residual interstory drift
response.

force necessary for pushing back the deformed structure is rea- than 0.2%, such that “no structural realignment is necessary for
sonably decreased (i.e., as in frame-SMA2), the self-centering structural stability, but the building may require adjustment and
response can be significantly improved during dynamic loading. repairs to nonstructural and mechanical components.” A more re-
The finding suggests that the SMA connections can be selectively laxed second class, DS2, requires RID to be less than 0.5%, such
used in critical regions to achieve satisfactory self-centering re- that realignment of structural frame and related structural repairs
sponses with reduced cost. are economically feasible, and degradation in structural stability
From a codified assessment point of view, a set of RID limits are is limited. The analysis results show that frame-SMA1 and
stipulated in FEMA P-58 (FEMA 2012). A total of four classes frame-SMA2 satisfy class DS1 under almost all the MCE-level
are defined, where the first class, DS1, requires the RID to be less ground motions. For frame-STL, however, 5 out of 20 DBE-level

© ASCE 04018114-13 J. Struct. Eng.

J. Struct. Eng., 2018, 144(8): 04018114


ground motions result in RID values that exceed 0.2%. For the Acknowledgments
MCE, 35% of the ground motions result in RID values greater than
0.5%, and thus frame-STL fails to satisfy the DS2 limit for these The financial support from the Natural Science Foundation of
ground motions. The results clearly demonstrate the positive role of China (NSFC) with Grant Nos. 51778456 and 51778459 is grate-
the SMA connection in reducing RID for steel frames. fully acknowledged. The study was also supported by the
Collaborative Research Project under International Joint Research
Laboratory of Earthquake Engineering at Tongji University and by
Summary and Conclusions the Fundamental Research Funds for the Central Universities.

This paper presents a new type of self-centering connection that is


equipped with an innovative SMA ring spring system. An exper- References
imental investigation on three proof-of-concept connection speci-
ABAQUS. 2012. Analysis user’s manual: Version 6.12. Waltham, MA:
mens was conducted first, where the ring thickness and bolt preload Dassault Systèmes.
level were the main testing parameters. The experimental investi- Abolmaali, A., J. Treadway, P. Aswath, F. K. Lu, and E. McCarthy. 2006.
Downloaded from ascelibrary.org by Tongji University on 10/22/18. Copyright ASCE. For personal use only; all rights reserved.

gation was followed by an extended numerical study providing a “Hysteresis behavior of t-stub connections with superelastic shape
basis for the design of full-scale self-centering connections in a typ- memory fasteners.” J. Construct. Steel Res. 62 (8): 831–838. https://doi
ical steel moment frame. A system-level time-history analysis was .org/1016/j.jcsr.2005.11.017.
then performed, which demonstrated the seismic performance of AISC. 2010. Seismic provisions for structural steel buildings. Chicago, IL:
prototype structures with different arrangements of the proposed AISC.
Auricchio, F. 2001. “A robust integration-algorithm for a finite strain shape-
SMA connection. The main findings and conclusions are summa-
memory-alloy superelastic model.” Int. J. Plast. 17 (7): 971–990.
rized as follows: https://doi.org/1016/S0749-6419(00)00050-4.
• The test specimens showed desirable flag-shaped hysteretic re- Bhuiyan, A. R., and M. S. Alam. 2013. “Seismic performance assessment
sponse with negligible residual deformation. Inelastic deforma- of highway bridges equipped with superelastic shape memory alloy-
tion of the test specimens occurred in the SMA ring spring based laminated rubber isolation bearing.” Eng. Struct. 49: 396–407.
systems of the connections, while the beam, column, and end https://doi.org/1016/j.engstruct.2012.11.022.
plate remained elastic, suggesting that minimal postearthquake Billah, A. H. M. M., and M. Alam. 2016. “Performance-based seismic de-
repair would be required. sign of shape memory alloy-reinforced concrete bridge piers. I: Devel-
• The test specimen connections are classified as semirigid con- opment of performance-based damage states.” J. Struct. Eng. 142 (12):
nections in terms of stiffness, although they could be considered 04016140. https://doi.org/1061/(ASCE)ST.1943-541X.0001458.
CEN (European Committee for Standardization). 2004. Design of struc-
to be rigid connections in nonsway braced frames; from a
tures for earthquake resistance: I: General rules, seismic actions
strength point of view, the specimen connections are classified and rules for buildings. Eurocode 8. Brussels, Belgium: CEN.
as partial-strength connections. CEN (European Committee for Standardization). 2005. Design of steel
• The test specimens exhibited a deformation capacity of at least structures: Part 1-8: Design of joints. Eurocode 3. Brussels, Belgium:
4% drift, with no degradation of moment resistance under cyclic CEN.
loading, and therefore they can be potentially employed for duc- Chang, W. S., and Y. Araki. 2016. “Use of shape-memory alloys in con-
tility class high (DCH) structures and special moment frames struction: A critical review.” Proc. ICE Civ. Eng. 169 (2): 87–95. https://
(SMFs), as specified by Eurocode 8 (CEN 2004) and AISC doi.org/10.1680/jcien.15.00010.
(2010), respectively. Chen, Q., and B. Andrawes. 2017. “Cyclic stress-strain behavior of
• The test specimens showed a satisfactory self-centering re- concrete confined with NiTiNb-shape memory alloy spirals.” J. Struct.
Eng. 143 (5): 04017008. https://doi.org/1061/(ASCE)ST.1943-541X
sponse, with no residual deformation after load cycles up to
.0001728.
4% drift, and reasonably good energy dissipation capacity with DesRoches, R., B. Taftali, and B. R. Ellingwood. 2010. “Seismic perfor-
an equivalent viscous damping of up to 16%. mance assessment of steel frames with shape memory alloy connec-
• A detailed 3D numerical investigation was performed to enable tions. I: Analysis and seismic demands.” J. Earthquake Eng. 14 (4):
a more in-depth interpretation of the test results and to facilitate 471–486. https://doi.org/1080/13632460903301088.
the design of self-centering SMA connections prototype build- Dimopoulos, A. I., T. L. Karavasilis, G. Vasdravellis, and B. Uy. 2013.
ings. Compared with a conventional steel frame building, the “Seismic design, modelling and assessment of self-centering steel
frame buildings with SMA connections had comparable maxi- frames using post-tensioned connections with web hourglass shape
mum interstory drifts but significantly decreased residual drifts pins.” Bull. Earthquake Eng. 11 (5): 1797–1816. https://doi.org/10
under seismic loading. .1007/s10518-013-9437-4.
Eatherton, M., and J. F. Hajjar. 2011. “Residual drifts of self-centering sys-
• Employing some SMA connections in a steel frame, along with
tems including effects of ambient building resistance.” Earthquake
conventional connections, can lead to satisfactory self-centering Spectra 27 (3): 719–744. https://doi.org/1193/1.3605318.
capability for the frame. This finding suggests the possibility of Fang, C., W. Wang, C. He, and Y. Y. Chen. 2017. “Self-centring behaviour
optimizing the use of the SMA connections to achieve desirable of steel and steel-concrete composite connections equipped with NiTi
system-level behavior balanced against the cost of construction. SMA bolts.” Eng. Struct. 150: 390–408. https://doi.org/1016/j.engstruct
It is worth mentioning that the full-scale SMA ring spring sys- .2017.07.067.
tems in the system-level building models were simulated based on Fang, C., M. C. H. Yam, A. C. C. Lam, and L. K. Xie. 2014. “Cyclic per-
the SMA material parameters obtained from small-scale SMA ring formance of extended end-plate connections equipped with shape
spring specimens. There could be a size effect as SMA rings with memory alloy bolts.” J. Constr. Steel Res. 94: 122–136. https://doi
different sizes may behave differently, and this is worth future in- .org/1016/j.jcsr.2013.11.008.
Fang, C., M. C. H. Yam, A. C. C. Lam, and Y. Y. Zhang. 2015a. “Feasibility
vestigations. In addition, varying environmental temperatures may study of shape memory alloy ring spring systems for self-centring seis-
have certain influence on the SMA behavior, and in particular, the mic resisting devices.” Smart Mater. Struct. 24 (7): 075024. https://doi
transformation temperature must be lower than the working temper- .org/1088/0964-1726/24/7/075024.
ature to ensure superelasticity. The influence of temperature was Fang, C., M. C. H. Yam, H. W. Ma, and K. F. Chung. 2015b. “Tests on
not examined in detail in this study. superelastic Ni-Ti SMA bars under cyclic tension and direct-shear:

© ASCE 04018114-14 J. Struct. Eng.

J. Struct. Eng., 2018, 144(8): 04018114


Towards practical recentring connections.” Mater. Struct. 48 (4): Qiu, C., and S. Zhu. 2017a. “Performance-based seismic design of
1013–1030. https://doi.org/1617/s11527-013-0212-4. self-centering steel frames with SMA-based braces.” Eng. Struct.
FEMA. 2012. Seismic performance assessment of buildings. Vol. 1: 130: 67–82. https://doi.org/1016/j.engstruct.2016.09.051.
Methodology. FEMA P-58-1. Washington, DC: FEMA. Qiu, C., and S. Zhu. 2017b. “Shake table test and numerical study of self-
Garlock, M. M., J. M. Ricles, and R. Sause. 2005. “Experimental studies centering steel frame with SMA braces.” Earthquake. Eng. Struct. Dyn.
of full-scale posttensioned steel connections.” J. Struct. Eng. 131 (3): 46 (1): 117–137. https://doi.org/10.1002/eqe.2777.
438–448. https://doi.org/1061/(ASCE)0733-9445(2005)131:3(438). Ricles, J. M., R. Sause, M. Garlock, and C. Zhao. 2001. “Post-tensioned
Guo, A., Q. Zhao, and H. Li. 2012. “Experimental study of a highway seismic resistant connections for steel frames.” J. Struct. Eng. 127 (2):
bridge with shape memory alloy restrainers focusing on the mitigation 113–121. https://doi.org/1061/(ASCE)0733-9445(2001)127:2(113).
of unseating and pounding.” Earthquake Eng. Eng. Vib. 11 (2): Silwal, B., R. J. Michael, and O. E. Ozbulut. 2015. “A superelastic viscous
195–204. https://doi.org/1007/s11803-012-0110-8. damper for enhanced seismic performance of steel moment frames.”
Gur, S., and S. K. Mishra. 2013. “Multi-objective stochastic-structural- Eng. Struct. 105: 152–164. https://doi.org/1016/j.engstruct.2015.10
optimization of shape-memory-alloy assisted pure-friction bearing .005.
for isolating building against random earthquakes.” Soil Dyn. Earth- Somerville, P. 1997. Development of ground motion time histories for
quake Eng. 54: 1–16. https://doi.org/1016/j.soildyn.2013.07.013. phase 2 of the FEMA/SAC steel project. Rep. No. SAC/DB-97/04.
ISO. 2009. Metallic materials: Tensile testing: Part 1: Method of test at Sacramento, CA: SAC Joint Venture.
Downloaded from ascelibrary.org by Tongji University on 10/22/18. Copyright ASCE. For personal use only; all rights reserved.

room temperature. ISO 6892-1. Geneva, Switzerland: ISO. Speicher, M. S., R. DesRoches, and R. T. Leon. 2011. “Experimental re-
Jani, J. M., M. Leary, A. Subic, and M. A. Gibson. 2014. “A review of sults of a NiTi shape memory alloy (SMA)-based recentering beam-
shape memory alloy research, applications and opportunities.” Mater. column connection.” Eng. Struct. 33 (9): 2448–2457. https://doi.org
Des. 56: 1078–1113. https://doi.org/1016/j.matdes.2013.11.084. /1016/j.engstruct.2011.04.018.
Jennings, E., and J. W. van de Lindt. 2014. “Numerical retrofit study of Sultana, P., and M. A. Youssef. 2016. “Seismic performance of steel mo-
light-frame wood buildings using shape memory alloy devices as seis- ment resisting frames utilizing superelastic shape memory alloys.”
J. Construct. Steel Res. 125: 239–251. https://doi.org/1016/j.jcsr
mic response modification devices.” J. Struct. Eng. 140 (7): 04014041.
.2016.06.019.
https://doi.org/1061/(ASCE)ST.1943-541X.0000953.
Tamai, H., Y. Kitagawa, and T. Fukuta. 2004. “Application of SMA rods to
Lagoudas, D. C. 2008. Shape memory alloys: Modeling and engineering
exposed-type column bases in smart structural systems.” In Proc., 13th
applications. New York, NY: Springer.
World Conf. on Earthquake Engineering. Ottawa, Ontario, Canada:
Lin, Y. C., R. Sause, and J. M. Ricles. 2013a. “Seismic performance of a
Canadian Association for Earthquake Engineering.
large-scale steel self-centering moment resisting frame: MCE hybrid
Tazarv, M., and M. Saiid Saiidi. 2015. “Reinforcing NiTi superelastic SMA
simulations and quasi-static pushover tests.” J. Struct. Eng. 139 (7):
for concrete structures.” J. Struct. Eng. 141 (8): 04014197. https://doi
1227–1236. https://doi.org/1061/(ASCE)ST.1943-541X.0000661.
.org/1061/(ASCE)ST.1943-541X.0001176.
Lin, Y. C., R. Sause, and J. M. Ricles. 2013b. “Seismic performance of a Wang, W., T. M. Chan, and H. L. Shao. 2015a. “Seismic performance
steel self-centering moment resisting frame: Hybrid simulations under of beam-column joints with SMA tendons strengthened by steel
design basis earthquake.” J. Struct. Eng. 139 (11): 1823–1832. https:// angles.” J. Constr. Steel Res. 109: 61–71. https://doi.org/1016/j.jcsr
doi.org/1061/(ASCE)ST.1943-541X.0000745. .2015.02.011.
Ma, H., T. Wilkinson, and C. Cho. 2007. “Feasibility study on a self- Wang, W., T. M. Chan, H. L. Shao, and Y. Y. Chen. 2015b. “Cyclic behav-
centering beam-to-column connection by using the superelastic behav- ior of connections equipped with NiTi shape memory alloy and steel
ior of SMAs.” Smart Mater. Struct. 16 (5): 1555–1563. https://doi.org tendons between H-shaped beam to CHS column.” Eng. Struct. 88:
/1088/0964-1726/16/5/008. 37–50. https://doi.org/1016/j.engstruct.2015.01.028.
Massah, S. R., and H. Dorvar. 2014. “Design and analysis of eccentrically Wang, W., C. Fang, and J. Liu. 2016. “Large size superelastic SMA bars:
braced steel frames with vertical links using shape memory alloys.” Heat treatment strategy, mechanical property and seismic application.”
Smart Mater. Struct. 23 (11): 115015. https://doi.org/1088/0964 Smart Mater. Struct. 25 (7): 075001. https://doi.org/1088/0964-1726/25
-1726/23/11/115015. /7/075001.
Mazzoni, S., F. McKenna, M. Scott, and G. Fenves. 2006. Open system for Wang, W., C. Fang, and J. Liu. 2017. “Self-centering beam-to-column
earthquake engineering simulation (OpenSees): User command lan- connections with combined superelastic SMA bolts and steel angles.”
guage manual. Berkeley, CA: Pacific Earthquake Engineering Research J. Struct. Eng. 143 (2): 04016175. https://doi.org/1061/(ASCE)ST.1943
Center, Univ. of California. -541X.0001675.
Miller, D. K. 1998. “Lessons learned from the Northridge earthquake.” Wood, A., I. Noy, and M. Parker. 2016. “The Canterbury rebuild five years
Eng. Struct. 20 (4–6): 249–260. https://doi.org/1016/S0141-0296(97) on from the Christchurch earthquake.” Reserve Bank N. Z. Bull. 79 (3):
00031-X. 1–16.
Moradi, S., and M. S. Alam. 2014. “Feasibility study of utilizing supere- Yam, M. C. H., C. Fang, A. C. C. Lam, and Y. Y. Zhang. 2015. “Numerical
lastic shape memory alloy plates in steel beam-column connections for study and practical design of beam-to-column connections with shape
improved seismic performance.” J. Intell. Mater. Syst. Struct. 26 (4): memory alloys.” J. Constr. Steel Res. 104: 177–192. https://doi.org
463–475. https://doi.org/1177/1045389X14529032. /1016/j.jcsr.2014.10.017.
Nakashima, M., K. Inoue, and M. Tada. 1998. “Classification of damage to Yang, C. S. W., R. DesRoches, and R. T. Leon. 2010. “Design and analysis
steel buildings observed in the 1995 Hyogoken-Nanbu earthquake.” of braced frames with shape memory alloy and energy-absorbing hybrid
Eng. Struct. 20 (4–6): 271–281. https://doi.org/1016/S0141-0296(97) devices.” Eng. Struct. 32 (2): 498–507. https://doi.org/1016/j.engstruct
00019-9. .2009.10.011.

© ASCE 04018114-15 J. Struct. Eng.

View publication stats J. Struct. Eng., 2018, 144(8): 04018114

You might also like