You are on page 1of 19

Smart Materials and Structures

PAPER
Recent citations
Multi-level SMA/lead rubber bearing isolation - Development and experimental validation
system for seismic protection of bridges of anchorage systems for shape memory
alloy cables
Fei Shi et al

To cite this article: Sasa Cao et al 2020 Smart Mater. Struct. 29 055045 - Analysis and design of a novel and
compact X-structured vibration isolation
mount (X-Mount) with wider quasi-zero-
stiffness range
Jing Bian and Xingjian Jing
View the article online for updates and enhancements.

This content was downloaded from IP address 202.3.77.210 on 04/04/2021 at 14:30


Smart Materials and Structures

Smart Mater. Struct. 29 (2020) 055045 (18pp) https://doi.org/10.1088/1361-665X/ab802b

Multi-level SMA/lead rubber bearing


isolation system for seismic protection
of bridges
Sasa Cao1,2, Osman E Ozbulut2, Suiwen Wu3, Zhuo Sun1 and Jiangdong Deng1
1
Department of Civil Engineering, Guangzhou University, Guangzhou 510006, People’s Republic of
China
2
Department of Engineering Systems and Environment, University of Virginia, Charlottesville, VA
22904, United States of America
3
Department of Civil and Environmental Engineering, MS258, University of Nevada, Reno, Reno, NV
89557, United States of America

E-mail: ozbulut@virginia.edu

Received 17 February 2020


Accepted for publication 16 March 2020
Published 20 April 2020

Abstract
In performance-based seismic design, bridges are expected to satisfy specific performance
objectives under several levels of seismic hazard. In this paper, a multi-level SMA/lead rubber
bearing (ML-SLRB) isolation system was proposed to ensure both isolation efficiency and
capability to limit excessive bearing displacements under different levels of earthquake
excitations. The ML-SLRBs also offer advantages such as the ability to provide re-centering
forces and good fatigue and corrosion-resistant. The ML-SLRB isolation system consists of
three groups of SMA cables, each is designed to be activated at a certain seismic hazard level,
and a conventional lead rubber bearing. First, the design and working mechanism of this new
isolation system were described in detail. Then, a design procedure was proposed for seismic
isolation of bridge structures with ML-SLRBs. Next, the hysteretic response of ML-SLRBs was
simulated in a general-purpose structural engineering software. A four-span continuous
box-girder bridge was designed and modeled with different isolation systems including
ML-SLRBs. Nonlinear dynamic analyses of the isolated bridges were conducted under both
far-fault and near-fault earthquakes. Results show that compared to isolations systems that do
not adapt their stiffness according to increasing seismic demand, e.g. the isolators with a
bilinear force–displacement response, the proposed isolation system exhibits high isolation
efficiency at small or moderate earthquakes, while effectively limits the bridge displacements to
avoid pounding and girder unseating under extreme earthquakes.
Keywords: isolated bridges, shape memory alloy, multi-level hazard, displacement control,
rubber bearings, isolation efficiency

(Some figures may appear in colour only in the online journal)

1. Introduction by the isolation systems limits the displacement response of


the superstructure. However, achieving satisfactory perform-
Seismic isolation has been considered as one of the effect- ance of isolated bridges under strong earthquakes still remains
ive earthquake-resistant design measures for girder bridges challenging [1]. The reasons for this challenge are three-fold:
over the past decades. Seismic isolation systems decouple the (1) the low force demand in the system is at the sacrifice
superstructure of the bridge from piers and thereby consider- of significant in-plane peak and residual displacements of
ably reduce the earthquake forces that are transferred to the girders, which could lead to pounding of the bridge to the
superstructure. In addition, supplemental damping provided abutment or even girder unseating [2–8]; (2) the isolation

1361-665X/20/055045+18$33.00 1 © 2020 IOP Publishing Ltd Printed in the UK


Smart Mater. Struct. 29 (2020) 055045 S Cao et al

bearing is susceptible to instability at large deformations; and or even unseating may occur if SMA elements with low stiff-
(3) the damaged bearings are required to be replaced. A num- ness are used, especially during strong earthquakes. However,
ber of isolated bridges have been significantly damaged or a smart seismic isolation system should be capable of adjust-
completely collapsed due to lack of sufficient displacement ing its stiffness during the course of motion [29]. To ensure
limiting capacity during the past major earthquakes (e.g. 1994 both isolation efficiency and capability to limit excessive bear-
Northridge, 1995 Kobe, 1999 Duzce, 1999 ChiChi and 2008 ing displacements, an isolation system with multiple mechan-
Wenchuan earthquakes, etc) [6, 9–11]. isms are needed. To this end, several isolation devices with
In order to control the displacement response and improve multi-level performance have been proposed [30–33]. How-
the re-centering capacity of the seismic isolation systems, ever, these devices adjust the energy dissipation capacity to the
shape memory alloys (SMA) that can be incorporated into an earthquake levels at the sacrifice of damages to specific com-
isolation system have gained attention more recently [12–17]. ponents. Therefore, significant residual deformations can be
SMAs can contribute to the characteristics of an isolation sys- observed in such systems due to lack of restoring forces and
tem in several ways. First, the hysteresis of SMAs is flag- they also require replacement after an event. To improve the
shaped with considerable energy dissipation capability. In re-centering ability, Calvi et al [34]. proposed a BowC isol-
addition, they have almost no residual displacement upon ation device that has a curved surface similar to the friction
unloading as a result of its superelastic effect. Furthermore, pendulum bearing but with different ring zones with increas-
SMAs have good fatigue and corrosion resistance [18]. Due ing friction coefficients from inside to outside. SMA ele-
to their appealing properties, a number of researchers have ments can also provide multi-level stiffness as well as the re-
investigated the performance of SMA-based isolators in high- centering force for conventional bearings if a judicious design
way bridges, in which the SMAs act as both damper and is achieved.
restrainer. In this paper, a new SMA isolation system that can enable
Several researchers have studied the use of SMA elements seismic protection of bridge structures under multi-level seis-
in friction-based isolation systems. Dolce et al [19] investig- mic hazards was proposed. The proposed system, named as
ated the seismic performance of bridges isolated with steel- multi-level SMA/lead rubber bearing (ML-SLRB) consists
PTFE sliding bearings and SMA-based auxiliary devices. of three groups of SMA cable and a lead rubber bearing.
Ozbulut and Hurlebaus [20] evaluated the performance of In the following, first, a detailed description and working
sliding-type isolation systems with an SMA device consider- mechanism of ML-SLRBs were discussed. Then, the design
ing environmental temperature changes. They [14] also stud- of bridges with ML-SLRB isolation systems was described.
ied the optimal design parameters of the superelastic fric- Next, a numerical model to simulate the response of ML-
tion base isolators for seismic design of bridge structures SLRBs was presented. A four-span continuous box girder
under near-field ground motions. Casciati et al [12, 21] pro- bridge was selected as a case-study structure and dynamic
posed an isolation system that combines a sliding system analyses of the bridge isolated with different isolation sys-
with multiple SMA bars for seismic protection of highway tems were conducted. The performance of the ML-SLRBs was
bridges. comparatively evaluated under both far-fault and near-fault
Some other researchers have explored the performance of earthquakes.
SMA-based rubber bearing isolators. Choi et al [22] proposed
an isolation bearing that consists of an elastomeric bearing
and SMA wires for highway bridges. Dezfuli and Alam pro-
2. Multi-level SMA/lead rubber bearing (ML-SLRB)
posed shape memory alloy wire-based natural rubber bearings
[13] and studied their performance for steel-girder highway
2.1. Description and design principles
bridges [23]. They also examined the combination of SMA
wires with lead rubber bearings and studied the seismic fragil- An ideal isolation system should have the ability to adjust its
ity of both steel girder bridges [24] and long-span cable-stayed design properties such as stiffness and damping with induced
bridges [25]. Ozbulut and Hurlebaus [26] conducted a sensitiv- earthquake excitation. This will ensure isolation efficiency at
ity analysis to evaluate the performance of SMA/rubber-based frequent to moderate earthquake levels while limiting excess-
(SRB) isolation system for seismic protection of bridges. They ive displacements at high seismic hazard levels. To achieve
also carried out a comparative performance assessment of these characteristics, a multi-level SMA/lead rubber bearing
SMA/rubber-based and SMA/friction-based isolation systems (ML-SLRB) that consists of a rubber bearing with a lead core,
for seismic protection of bridges [27]. Mishra et al [28] studied i.e. a lead-rubber bearing (LRB) and three groups of SMA
the optimal parameters for SMA-based rubber bridge bearings cables as shown in figure 1(a) was proposed in this study. In
through nonlinear random vibration analysis and developed the proposed isolation system, the rubber bearing provides low
closed-form expressions for estimating these optimal lateral stiffness while the lead core dissipates energy. Three
parameters. groups of SMA cables as shown in figure 1(b) are included
Note that in all of the above studies only a single level of in the design to alter the stiffness of the isolation bearing and
SMA element has been adopted together with either rubber- dissipate additional energy as the intensity of seismic event
based or friction-based bearings. Therefore, in these isolation increases. The SMA cables also provide a re-centering cap-
systems, if SMA elements with very high stiffness are used, the ability to the isolation system when the earthquake loading
isolation efficiency will be low, whereas large displacements ceases.

2
Smart Mater. Struct. 29 (2020) 055045 S Cao et al

Figure 1. (a) Multi-level SMA/rubber bearing and (b) three groups of SMA cables.

SMA cables selected as the shape memory alloy compon- third group of SMA cables are connected to the steel plates of
ent of the proposed bearing offer unique advantages. Wires the bearing by hinges.
are the most common form of SMA products in the market, Figure 3 shows a typical shear force–displacement relation-
however; the use of SMA wires is not realistic for most of ship of various isolation systems and the design spectrum in
the civil engineering applications where high axial stiffness acceleration–displacement (A–D) format. Two seismic haz-
and large force capacity are needed [13, 14, 35]. On the other ard levels corresponding to a probability of exceedance 10%
hand, large-diameter SMA bars are difficult to fabricate, have in 50 years (E1) and 5% in 100 years (E2) are considered
higher costs and are vulnerable to early fracture, especially if in Guidelines for Seismic Design of Highway Bridges [41]
they are machined. SMA cables, composed of a large num- and target response spectra at these hazard levels are shown
ber of small diameter wires, possess the advantages of SMA in figure 1. Here, a higher seismic hazard level (named as
wires such as ease fabrication, low-cost, and excellent supere- E3) that was assumed to be twice of E2 spectrum was also
lasticity and can produce large forces [36–40]. considered to represent extreme earthquakes. The pushover
The connection of the specific group of SMA cables to response of an isolated bridge with lead rubber bearing can
the cover plates of the proposed isolation system is shown in be idealized as a bilinear curve. According to the capacity
figure 2. The first group of SMA cables is connected to top and spectrum method, the performance point of the bridge with
bottom steel plates by hooks at the corners. The second and the LRBs can be denoted by u1 , u2 and u3 for frequent, moderate,

3
Smart Mater. Struct. 29 (2020) 055045 S Cao et al

Table 1. Performance objectives of ML-SLRB.

Performance Earthquake Return period


level level (year) Performance objectives

1 Frequent or 475 High isolation effi-


moderate ciency; immediate
earthquake use; no damage to the
(E1) system; no repair; no
closure
2 Design 2000 High isolation effi-
earthquake ciency; small residual
(E2) displacement; no clos-
ure; minor repair
3 Extreme Over 2000 No girders unseat-
earthquake ing; no collapse;
Figure 2. The specific connection of SMA cables.
(E3) repair/limited closure

Figure 3. Typical shear force–displacement relationships of


different isolation bearings.
Figure 4. The idealized response of ML-SLRB.

and large earthquakes respectively [42]. It can be seen that


the extensive displacements will occur at high seismic haz-
ards. When each group of SMA cables was included into
2.2. Working mechanism
the bearing system for the ML-SLRB, the pushover curve
was modified as shown in figure 3. The performance points In ML-SLRB, each group of SMA cables is designed to
for the ML-SLRB with all SMA groups are denoted by be activated at different seismic hazard levels as shown in
u1 ′′′ , u2 ′′′ and u3 ′′′ . It can be seen that the displacement figure 4. During a frequent earthquake, the first group of
response can successfully be controlled at each seismic hazard SMA cables is tensioned and experience phase transforma-
level. tions at a displacement level of u1 , while the second and the
Three performance objectives adopted here for the design third group are not activated as a result of available slacks.
of ML-SLRB include: (i) no damage in a low or moderate When the isolation displacement reaches u20 , the second group
level of earthquake; (ii) no severe damage in design earth- of SMA cables is also tensioned, while the third group of
quake; (iii) no girder unseating during an extreme earthquake. SMA cables is still in slack. The second group of SMA
With an increase in seismic demand, the stiffness of ML- cables experiences martensitic transformation at a displace-
SLRB increases step-by-step. In this way, the isolation effi- ment of u2 , which leads to another change in the overall
ciency was ensured during a small or moderate earthquake, stiffness of the bearing. During a very large earthquake, the
while the relative displacements between piers and girders are third group of SMA cables is also activated at a displace-
limited during extreme earthquakes. The specific performance ment level of u30 to limit the relative displacements between
objectives at various levels of earthquakes are summarized in piers and girders. They exhibit phase transformations at a
table 1. displacement of u3 .

4
Smart Mater. Struct. 29 (2020) 055045 S Cao et al

Figure 5. The tensile stress-strain relationship of SMA cable.

Table 2. Parameters of lead rubber bearing component of Table 3. The initial dimension of each group of SMA cables (unit:
ML-SLRB. mm).
Parameters Values Slacks Cable lengths

a (mm) 1770 u10 u20 u30 l1 l2 l3


h (mm) 222 0 50 50 7108 1833 228
K 0 (kN m−1 ) 62 300 100 100 1883 243
Keff (kN m−1 ) 13 200 150 268
Fy (kN) 817
α 0.154


The controlling design parameters include yield force F y , 2
l2 = (a + u20 ) + h2 (2)
yield displacement uy , initial stiffness K 0 and ratio of post-
yield to initial stiffness α for lead-rubber bearing compon- √
ent, yielding forces f yi , yield displacements ui , stiffness K i and l3 = u230 + h2 (3)
cable slack ui0 (i = 1 ~ 3)of each group of SMA cables.
A commercially available lead rubber bearing was selec- where l1 , l2 and l3 is the length of the first, second and third
ted as the first component of the ML-SLRB. Table 2 provides group of SMA cables, respectively. Remember that a is the
the geometric and structural properties of the selected bearing. side length of the bearing plate and u20 and u30 are the initial
The top and bottom steel plates for the bearing have a square slacks provided in the second and third SMA cable groups. For
geometry with a side length a, and the height of the rubber the selected dimensions of the lead rubber bearing, the possible
compound is denoted as h. The parameters K 0 , Keff , Fy and α slack lengths and the corresponding lengths of each group of
are the initial stiffness, effective stiffness, yielding force and SMA cables are provided in table 3.
ratio of post-yield to initial stiffness, respectively.
For the SMA cable component of the ML-SLRB, the stress-
2.3. Design procedure
strain behavior was taken as shown in figure 5. This mech-
anical response of SMA cable was obtained from uniaxial The design of ML-SLRB involves several steps as illustrated
tensile tests conducted at the University of Virginia. The tested in figure 6. The key parameters needed to be determined for
cable had a diameter of 8 mm and consisted of seven strands. ML-SLRB are cross-sectional area and slack displacements of
Each strand was composed of seven wires with a diameter each group of SMA cables. Here, all groups of SMA cables are
of 0.885 mm. The details of the experimental testing were assumed to have the same cross-sectional area to simply both
provided in another study [37]. design and fabrication procedures. The design procedure for
The length of each group of SMA cables shown in figure 1 seismic isolation of bridges with ML-SLRB is discussed step-
can be determined as follows: by-step below.
√ Step 1: Determine the design parameters of the lead-rubber
l1 = 2( a2 + h2 + a) (1) bearing component of ML-SLRB. Follow the guidelines and

5
Smart Mater. Struct. 29 (2020) 055045 S Cao et al

Figure 6. Design procedures of ML-SLRB.

Figure 7. Experimental and simulated stress-strain relationships of SMA cables.

provisions provided for the design of bridges with typical SMA cables. Moreover, u10 and u20 should be large enough
LRBs. to ensure that the second and third group of SMA cables will
Step 2: Select ground motion records that represent differ- not fracture before the isolation system attains its maximum
ent hazard levels. Develop a finite element model of a bridge displacement ulimit .
system isolated with LRBs. Conduct nonlinear dynamic ana- Step 4: Choose the diameter, which is correlated to
lyses on the bridge model under the selected accelerograms the cross-section area, for SMA cables. Conduct nonlinear
to compute the maximum displacements of isolation systems: dynamic analyses again on the updated bridge model with ML-
u1 , u2 and u3 (see figure 3) corresponding to the three seismic SLRB under the selected accelerograms to compute the max-
hazard levels, respectively. imum displacements of the isolation system and maximum
Step 3: Chose no slack, i.e. u10 = 0, to ensure the first moments at the bottom of piers at three seismic hazard levels.
group SMA cables will be activated immediately with the seis- Repeat this step for different ML-SLRB with a different cross-
mic loading. Considering the seat-width of piers or abutments, sectional area of SMA cable.
choose a maximum limiting displacement, ulimit , of isolation Step 5: Compare the responses for the isolation displace-
system according to seismic design codes. Choose u30 consid- ment and pier moments obtained for the bridge structure isol-
ering the values of u3 and ulimit . The slack in the third group of ated with ML-SLRB for the different cross-sectional area
SMA cable, u30 , should be 5–10 cm less than the smaller of u3 of SMA cable. Choose an optimum cross-sectional area,
and ulimit . Select an initial slack for the second group of SMA i.e. equivalent diameter, for SMA cables, with which the
cables, u20 , which could be a value between u1 and u2 . The ML-SLRB can limit the displacement of the isolation sys-
slack u20 should be smaller than u30 to ensure the second group tem below ulimit without a considerable increase in pier
of SMA cables will be activated earlier than the third group moments.

6
Smart Mater. Struct. 29 (2020) 055045 S Cao et al

Figure 8. Simplified stress-strain relationships of SMA cables.

3.1. Simulation of SMA cable groups

To model the SMA cables, a procedure proposed by Tazarv


and Saiidi was used to obtain the simplified flag shape model
[44]. In this procedure, the austenite modulus was taken as the
average slope between 15% and 70% of observed phase trans-
formation start stress level called as yield strength (f yo ), the
post-transformation stiffness was taken as the average slope
between 2.5% and 3.5% of strain on the upper plateau of 6%
strain loading cycle, and the martensite modulus was taken as
the slope of the curve between 8% and 9% strain. A ‘multilin-
ear elastic element’ and a ‘multilinear plastic element’ were
used to simulate the stress-strain relationship of a group of
SMA cables. Isotropic hysteretic rules were adopted in the
multilinear plastic element to simulate the energy dissipation
capacity of the SMA cables. The cyclic tensile response of
SMA cables obtained from experimental testing [37] was used
for selecting the model parameters and the model verifica-
Figure 9. Force–displacement response of ML-SLRB.
tion. Figure 7 compares the hysteretic stress-strain response of
SMA cables obtained from experimental testing and numerical
simulations. Some residual displacements can be observed in
Step 6: Check the response of the new bridge sys-
the numerical prediction of the SMA cable response. This is
tem. If the response satisfies the design principle shown
due to the use of ‘multilinear plastic element’ in the developed
in figure 4, the design procedure is completed. If not,
model, which always exhibit some plastic deformation. How-
u20 and u30 of SMA cables can be adjusted to fulfill the
ever, it can be seen that the residual displacements are very
multi-level performance target. They can be increased if
small at low strain amplitudes while it is less than 0.5% for
they are activated early and causes large forces, or they
the loading at 7.7%. In general, the skeleton curve and the area
can be decreased if they are not activated timely. Then,
encased by hysteretic loop of simulation match with experi-
repeat steps 3–5 until the multi-level performance targets are
mental results well.
obtained.

3.2. Simulation of ML-SLRB


3. Modeling of SMA cables and ML-SLRB
To model ML-SLRB in SAP 2000, three ‘multilinear elastic’,
In this section, a numerical model to represent the lateral three ‘multilinear plastic’ and a ‘plastic (Wen)’ elements are
force–displacement response ML-SLRB was developed using arranged in parallel. Each group of SMA cables was simu-
SAP 2000 software package [43]. lated by a multilinear elastic element and a multilinear plastic

7
Smart Mater. Struct. 29 (2020) 055045 S Cao et al

Figure 10. A typical 4-span continuous box-girder bridge (unit: m).

Figure 11. Section of the girder (unit: m).

element while the lead rubber bearing was modeled by a


‘plastic (Wen)’ element. The lateral force–displacement rela-
tionship of each cable was computed based on their deformed 1.45
shape. 6.25 6.25 6.25 2.5
As shown in figure 8, the simplified flag shape stress-strain
relationship of SMA cables was represented by multilinear
lines between different points i, where i = 1, 2, 3, 4 and 5.
The stress and strain at these points can be denoted by σi 1.5 3.75 1.5 3.95 1.5 3.75 1.5 1.5 3.75 1.5
and εi .
Then, the tensile force and overall length of each SMA a Elevation view b Side view
cable group at a point i are given by: 1.45
6.25 6.25
fji = σi Aj (4)

6.25
lji = lj (1 + εi ) (5)

where j denotes the jth group of SMA cables and Aj is the


c Plan view
cross-sectional area of the jth group of SMA cables.
Considering the configuration of three groups of SMA
cables of the proposed ML-SLRB, when each group of SMA Figure 12. A pier (unit: m).
cables is tensioned to the strain of points i in figure 8, the cor-
responding lateral displacements can be computed as: a + x1 h
f1ix = σi A1 √ f1iy = σi A1 √

( )2 2
(a + x1 ) + h2
2
(a + x1 ) + h2
1
x1i = (l1i − 2a) − h2 − a (6a) (7a)
2
a + x2 h
√ f2ix = σi A2 √ f2iy = σi A2 √
2 2
x2i = l22i − h2 − a (6b) (a + x2 ) + h2 (a + x2 ) + h2
(7b)
√ x3 h
f3ix = σi A3 √ f3iy = σi A3 √ (7c)
x3i = l23i − h2 (6c) x3 2 + h2 x3 + h2
2

where xji are the lateral displacements of the proposed ML- The horizontal force components fjix can be used to describe
SLRB when the jth SMA group of SMA cables attain strain the lateral force–displacement relationships of each group of
εi at an ith point in figure 8. In addition, the horizontal and SMA cables.
vertical components of forces of each group of SMA cables, Figure 9 illustrates the force–displacement response of ML-
fjix and fjiy can be computed by: SLRB predicted by the developed model under a quasi-static

8
Smart Mater. Struct. 29 (2020) 055045 S Cao et al

Table 4. The stiffness of the 6 DOF springs.

Kx (kN m−1 ) Ky (kN m−1 ) Kz (kN m−1 ) Rx (kN m Rad−1 ) Ry (kN m Rad−1 ) Rz (kN m Rad−1 )

4 070 000 4 070 000 15 630 000 74 940 000 74 940 000 32 610 000

Figure 13. Design earthquake spectra and the spectra of scaled ground motions: (a) far-fault earthquakes and (b) near-fault earthquakes.

Table 5. Ground motions used in nonlinear dynamic analyses. loading and unloading history. The response of the lead-rubber
Moment PGA PGV bearing component of the isolation system is also provided on
ID No. Vs_30 magnitude Fault type (g) (cm s−1 ) Pulse the same plot for comparison purposes. It can be seen that
with the addition of SMA cables to the lead-rubber bearing,
Far-fault the stiffness of the bearing experience a step-by-step increase
FF01 356 6.7 Thrust 0.52 66.7 for the increased loading levels. In particular, there is a consid-
FF02 309 6.7 Thrust 0.48 44.4 erable increase in the stiffness of the isolation system at large
FF03 326 7.1 Strike-slip 0.82 65.8
displacements, which can help to prevent girder unseating at
FF05 275 6.5 Strike-slip 0.35 36.1
high seismic hazards. In addition, the area under the hysteretic
FF08 256 6.9 Strike-slip 0.24 38.1
FF09 276 7.5 Strike-slip 0.36 58.8 response of the ML-SLRB, which indicates the energy dissipa-
FF10 523 7.5 Strike-slip 0.22 42.2 tion, is almost twice as large as that of the lead-rubber bearing.
FF11 354 7.3 Strike-slip 0.24 51.1
FF13 289 6.9 Strike-slip 0.53 38.0 4. Dynamic analysis of a bridge structure isolated
FF14 350 6.9 Strike-slip 0.56 45.4
with ML-SLRB
FF17 208 6.5 Strike-slip 0.45 36.0
FF19 259 7.6 Thrust 0.44 115.0
FF21 316 6.6 Thrust 0.21 27.6 4.1. Description of the bridge model
FF22 425 6.5 Thrust 0.35 30.5 In this section, a case study was performed in order to explore
Near-fault
the effectiveness of the proposed ML-SLRB in seismic protec-
NF04 349 6.5 Strike-slip 0.42 106.8 Yes
tion of bridges. A typical 4-span continuous box-girder bridge
NF05 371 6.9 Strike-slip 0.38 55.6 Yes
NF06 275 6.7 Strike-slip 0.49 95.5 Yes
in China, shown in figure 10, was selected for this purpose. The
NF09 282 6.7 Thrust 0.87 167.3 Yes span lengths are all equal to 30 m. The section of the girder and
NF10 441 6.7 Thrust 0.73 122.8 Yes structure of a bent are shown in figures 11 and 12, respectively.
NF12 306 7.6 Thrust 0.82 127.7 Yes The total mass of the deck is 11 510 t. Lead rubber isolation
NF14 276 7.1 Strike-slip 0.52 79.3 Yes bearings are installed on all bents. The fundamental period of
NF20 376 6.9 Strike-slip 0.64 55.9 No the original bridge was 1.21 s. A Rayleigh damping of 5% was
NF21 462 6.9 Strike-slip 0.51 45.5 No assumed for the first and second modes.
NF22 514 7.0 Thrust 1.43 119.5 No SAP 2000 was employed to model the bridge system with
NF23 380 6.7 Thrust 0.73 70.1 No the ML-SLRB isolation system. In this model, the bents and
NF24 281 6.7 Thrust 0.42 53.2 No girders were assumed elastic and simulated by elastic-beam
NF25 297 7.5 Strike-slip 0.31 73.0 No
elements. The ML-SLRB isolators were modeled as discussed
NF26 434 7.6 Thrust 0.56 91.8 No
earlier. The piles and soil were simulated by 6 DOF springs,

9
Smart Mater. Struct. 29 (2020) 055045 S Cao et al

Figure 14. (a) Displacement response history of ML-SLRB isolation system; force response history of (b) first group of SMA cables, (c)
second group of SMA cables, and (d) third group of SMA cables.

10
Smart Mater. Struct. 29 (2020) 055045 S Cao et al

the scaled far-fault and near-fault ground motions are shown


in figure 13.

4.3. Design of bridge structure with ML-SLRB

The procedure described in section 2.3 was followed to design


the selected bridge with an ML-SLRB isolation system. The
properties of the LRB component of the isolation system were
selected as shown in table 2. Considering the geometric prop-
erties of the selected bearing, the initial lengths of SMA cables
of ML-SLRB can be computed and are provided in table 3.
Then, the cross-sectional area of each SMA cable group and
slacks in each SMA group were determined through nonlinear
dynamic analyses.
First, nonlinear dynamic analyses of the bridge system with
LRB under the scaled near-fault accelerograms were conduc-
ted. The mean values for u1 , u2 and u3 are found to be 26 mm,
97 mm and 250 mm respectively. According to Guidelines for
Figure 15. Lateral force–displacement relationships of different Seismic Design of Highway Bridges [41], the maximum dis-
isolation systems.
placement of the isolation system, which is controlled by max-
imum shear deformation of LRB, should be below 200 mm.
As a result, the slacks u20 and u30 were chosen as 50 mm and
three translational and three rotational springs, assigned to the 150 mm respectively. Then, an initial diameter (10 mm) was
base of piers. The stiffness of the springs should be derived selected for the SMA cables and the numerical simulations are
by accounting for the properties of the soil and piles and their conducted. The diameter of the SMA cables was increased to
interaction. Table 4 provides the stiffness coefficients selec- meet the desired maximum displacement limit of 200 mm. The
ted in this study. In the table, Kx , Ky and Kz are the trans- final diameter of the SMA cables was determined as 25 mm.
lational stiffness in longitudinal, transverse and vertical dir- Figure 14 shows the total and component force response
ection, respectively; Rx , Ry and Rz are the rotational stiffness histories one of the ML-SLRBs at the middle pier of the bridge
around longitudinal, transverse and vertical direction, respect- subjected to ground motion FF01 scaled to the E3 level. It can
ively. The same 6 DOF springs were adopted for all piers for be seen that the first group of SMA cables were always ten-
simplicity. For convenience, the results from only the right sioned; however, the second and third groups of SMA cables
isolator and the right column of the middle pier were presen- were activated only when the later displacement was larger
ted in the following sections to discuss the maximum isolator than their respective slacks. It can be seen from figures 14(c)
displacement and column bottom moment of the bridge. and (d) that the second and third groups of SMA cables were
not tensioned at the beginning. The second group of SMA
cables was activated at about 1.9 s when the relative displace-
4.2. Selected ground motion records
ment exceeds 50 mm, and the third group of SMA cables
Three design earthquake spectra were chosen to represent E1, was activated at about 2.11 s when the displacement exceeds
E2 and E3 respectively., The equivalent velocities of the shear 150 mm. Throughout the time history, the forces of the second
wave of the soil for the selected side were 208–514 m s−1 . and the third group of SMA cables increased whenever the
Fourteen far-fault and 14 near-fault ground motion records isolator displacement exceeds 50 mm and 150 mm respect-
from FEMA P-695 [45] were used in nonlinear dynamic ana- ively. This observation confirms the working mechanism of
lyses. Previous studies have shown that pulse-like near-fault SMA cables.
earthquakes, which are usually identified by forward directiv-
ity effects, may result in larger seismic demands [46–48].
4.4. Comparative seismic response evaluation
Therefore, here, seven pulse-like near-fault and seven non-
pulse-like near-fault earhquakes are considered. The charac- To evaluate the performance of the proposed ML-SLRB com-
teristic properties of selected records such as peak ground paratively with conventional LRB and SMA-LRBs that con-
acceleration (PGA), peak ground velocity (PGV) and the sist of only a single group of SMA cables, additional nonlin-
average shear-wave velocity in the topmost 30 m of sedi- ear response history analyses of the bridge described above
ments (Vs_30) are provided in table 5. The accelerograms were conducted. In particular, four cases were considered for
were scaled to E1, E2 and E3 respectively using Seismo- the isolation of the selected bridge. Figure 15 shows the lateral
Match software [49]. In the scaling procedure of Seismo- force–displacement response of the four isolation systems. For
Match, the improved method proposed by Al Atik and Abra- Case 1, the isolation system consists of only the LRB com-
hamson was adopted [50]. The baseline correction and filtering ponent of the ML-SLRB designed above, i.e. this case repres-
were applied by whenever correction of the scaled accelero- ent a typical LRB isolation system. Case 2 and Case 3 are the
grams were needed [51]. The design spectra and the spectra of bridges isolated with SMA-LRB systems, where the isolation

11
Smart Mater. Struct. 29 (2020) 055045 S Cao et al

Figure 16. Hysteretic loops of the right isolator at the middle pier under far-fault earthquake FF01 scaled to E1, E2, and E3 hazard levels.

12
Smart Mater. Struct. 29 (2020) 055045 S Cao et al

Figure 17. Hysteretic loops of the right isolator at the middle pier under near-fault earthquake NF04 scaled to E1, E2, and E3 hazard levels.

13
Smart Mater. Struct. 29 (2020) 055045 S Cao et al

Figure 18. Comparison of the maximum displacement of the middle pier of the bridge under (a) far-fault earthquakes and (b) near-fault
earthquakes.

Figure 19. Maximum moment of the bottom section of the right column of the middle pier under (a) far-fault earthquakes and (b) near-fault
earthquakes.

system consists of an LRB and only first cable group of ML-


SLRB with a cable diameter of 25 mm for Case 2; and with
a larger cable diameter selected such that the equivalent stiff-
ness of the isolation system matches with that of ML-SLRB
at limit displacement of 200 mm for Case 3. Case 4 repres-
ents the bridge isolated with ML-SLRB with 25 mm diameter
SMA cables.
Nonlinear dynamic analyses were performed in the lon-
gitudinal direction of the bridge systems for Case 1 to
Case 4 using the far-fault and near-fault ground motions
provided in table 4 and scaled to three seismic hazard levels.
Figures 16 and 17 show the hysteretic loops for one of the isol-
ators at the middle pier for the bridge isolated with different
systems under a far-fault (FF01) and near-fault (NF04) ground
motion records, respectively. It can be seen compared to the
Figure 20. Connection of adjacent girders for selected case-study SMA-LRB with high stiffness (Case 3), the ML-SLRB sys-
bridge. tem (Case 4) exhibited good energy dissipation at all seismic

14
Smart Mater. Struct. 29 (2020) 055045 S Cao et al

Figure 21. Time-histories of the left bearing in the first pier under: (a) NF 04, (b) NF 05, (c) NF 06, (d) NF 09, (e) NF 10, (f) NF 12, and (g)
NF 14 ground motions.

hazard levels and exhibited a step-by-step increase in its stiff- between the girder and piers of the bridge under far-fault and
ness. This helped to ensure isolation efficiency at small or near-fault earthquakes for different isolation systems. As can
moderate earthquakes while preventing excessive displace- be seen from the figure, at relatively low seismic hazards, all
ments at high hazard level. On the other hand, when an SMA- isolation systems have similar responses but there are large
LRB with relatively lower stiffness was used (Case 2), the differences in response for the E3 hazard level. The largest
response of the isolation system was similar to that of the LRB isolation displacements were observed for the LRB system
system (Case 1). This suggests that the use of only one group of (Case 1) at all seismic hazard levels. The isolation displace-
SMA cables will only be efficient in limiting the isolation dis- ment responses for Case 3 and Case 4 were considerably smal-
placements at high seismic hazard levels if a very large cross- ler than that of bridge isolated by LRB (Case 1) during all
sectional area is selected for the SMA cables, i.e. (Case 3). levels of motions, especially under E3. In particular, the isol-
To enable quantitative comparison, the seismic responses ation displacements decreased 31.2% and 27.1%, respectively
of the isolated bridges under all ground motion records were for the Case 3 and Case 4 compared to Case 1 for the far-
computed and the mean results are illustrated in figures 18 fault earthquakes, and decreased 34.4% and 31.3% for the
and 19. Figure 18 shows the maximum relative displacements near-fault earthquakes. Also, the maximum displacements for

15
Smart Mater. Struct. 29 (2020) 055045 S Cao et al

Case 3 and Case 4 were successfully limited below the design presented for seven pulse-like near-fault earthquakes scaled to
value of 0.2 m. Some reductions in the isolation displacement E3 level. It can be seen that the SMA-LRB system with high
were also obtained for Case 2 but these reductions were not stiffness (Case 3) or ML-SLRB system (Case 4) successfully
substantial (below 12.1%). In addition, compared to far-fault control the bearing displacement. On the other hand, bearing
earthquakes, near-fault earthquakes create higher displace- displacements over 300 mm was observed for Case 1 and Case
ment demands. Figure 18 (b) also compares the maximum rel- 2 under NF04, NF05, NF06, NF09 and NF14.
ative displacements between the girder and piers of the bridge It should be also emphasized that the bridges designed with
under pulse-like and non-pulse-like near-fault earthquakes LRB isolation systems does not commonly suffer from resid-
scaled to E3 hazard level. It can be observed that the near- ual displacements as opposed to bridges designed with flat
fault earthquakes with pulse cause slightly higher displace- friction bearings. However, it can be seen in figure 21, the LRB
ment demands compared to no pulse near-fault earthquakes. system (Case 1) exhibits residual displacements of 29 mm and
Figure 19 shows the maximum moments of the bottom 12 mm under NF 04 and NF 09 respectively, which might
section of the right column of the middle pier for all bridge require some repair. On the other hand, SMA-LRB and ML-
systems under different seismic hazard levels. As seen in SLRB systems (Case 3 and Case 4) return to their original pos-
figure 19, the maximum moments increased significantly for ition for all ground motion cases.
the SMA-LRB system with high stiffness (Case 3). Com-
pared to the LRB system, there were 38%, 78.8%, and 35.2%
increases in the column moments at E1, E2, and E3 seismic 5. Conclusions
hazard levels for the far-fault earthquakes and 36.1%, 69.3%,
and 32.0% increases for the near-fault earthquakes, respect- In this paper, a new isolator system was proposed to achieve
ively. Although there was an increase in the column moment the protection of bridge structures under multi-level seismic
for the ML-SLRB system as well, these increases were relat- hazard by increasing re-centering and energy dissipation cap-
ively smaller with values of 9.8%, 31.0%, and 33.8% for far- abilities of the system. This system incorporates three groups
fault earthquakes and 9.6%, 26.7%, and 37.4% for near-fault of SMA cables with different configurations and a conven-
earthquakes. Note that especially for E2 design level seismic tional lead-rubber bearing isolator. With increasing earth-
hazard, the ML-SLRB isolation system reduced the column quake levels, each group of SMA cables was sequentially
moments by more than half compared to Case 3 while keep- activated to enable higher damping and force capacities. The
ing the isolation displacements quite similar to that case. In design principle of the proposed isolator and a numerical
addition, figure 19(b) illustrates the maximum moments of the model was developed through SAP2000 to represent its hys-
bottom section of the right column of the middle pier under teretic response. A case study was presented to show the effect-
pulse-like and non-pulse-like near-fault earthquakes. A some- iveness of the proposed ML-SLRBs in seismic response pro-
what higher moment demands were observed for the pulse- tection of a typical box-girder bridge subjected to far-fault
like near-fault earthquakes compared to no pulse near-fault and near-fault earthquakes scaled to different seismic hazard
earthquakes. levels. The study leads to the following conclusions:
In general, the unseating of bridges are in a relationship
with the minimum support length (MSL) of bridge girders (a) Compared to conventional lead rubber bearings, the ML-
shown in figure 20, which may be different from bridges to SLRBs can more effectively limit the isolation displace-
bridges. On the other hand, the pounding of bridges has a close ments at high seismic hazard levels, especially under near-
relationship with the dynamic characteristics of the adjacent fault ground motions.
spans. In particular, if two adjacent spans have very differ- (b) However, the maximum moments of the pier of the ML-
ent fundamental periods, the possibility of pounding increases. SLRB bridge system increase as compared with the LRB
However, both the unseating possibility of a bridge and pound- bridge system. In practical design, more attention should
ing are directly correlated to displacement demand on bear- be paid on the internal forces of piers and piles when the
ings. The larger the displacement of bearings is, the more ML-SLRB bridge system is adopted.
likely unseating and pounding problems will occur. There- (c) Compared to SMA-based rubber bearings, the ML-SLRBs
fore, the displacement of bearings is an important parameter can control both the isolation displacements and forces
for pounding and unseating problems. transferred to the foundation of the bridge for different
The configuration of the connection of adjacent girders is levels of earthquake excitations. On the other hand, when
shown in figure 20 for the selected case-study bridge. The an SMA-based rubber bearing that incorporates only the
MSL and clearance between adjacent girders are 1450 mm and first group SMA cables of ML-SLRBs with the same
100 mm, respectively. Since this is a continuous bridge and its cable cross-sectional area is used, there is no considerable
MSL is quite large, the unseating is not expected to be a prob- decrease in the displacement response. But if an SMA-
lem for the studied bridge. However, to provide insights for the based rubber bearing with larger diameter first group SMA
design of other types of bridges, the displacement response of cables of ML-SLRBs is used, then there are substantial
bearings in a transitional pier is discussed below. increases in base moments.
Figure 21 illustrates the time-histories of the left bearing (d) The proposed ML-SLRB can achieve multi-performance
in the first pier for four bridge design cases. Since pulse-like under various earthquake levels by gradually activating
near-fault earthquakes result in larger demands, the results are different groups of SMA cables. This helps to ensure high

16
Smart Mater. Struct. 29 (2020) 055045 S Cao et al

isolation efficiency under low or moderate earthquakes, [13] Dezfuli F H and Alam M S 2013 Shape memory alloy
while the bridge displacements are limited to avoid pound- wire-based smart natural rubber bearing Smart Mater.
ing and girder unseating under extreme earthquakes. Struct. 22 045013
[14] Ozbulut O E and Hurlebaus S 2011 Optimal design of
(e) Compared with no pulse near-fault earthquakes, the pulse- superelastic-friction base isolators for seismic protection of
like near-fault earthquakes resulted in higher displacement highway bridges against near-field earthquakes Earthq.
and force demands due to their velocity pulse. Eng. Struct. Dyn. 40 273–91
(f) The isolation efficiency of ML-SLRB will decrease with [15] Shinozuka M, Chaudhuri S R and Mishra S K 2015
the increase of earthquake intensity due to the gradual Shape-memory-alloy supplemented lead rubber bearing
(SMA-LRB) for seismic isolation Probab. Eng. Mech.
increase of stiffness caused by the activation of three 41 34–45
groups of SMA cables in a sequence. [16] Yang J 2012 Development of SMA Isolation Bearing and Its
Analysis of Damping Effect in Bridge (Liaoning: Liaoning
Technical University) (in Chinese)
Acknowledgments [17] Dezfuli F H and Alam M S 2014 Performance-based
assessment and design of FRP-based high damping rubber
The project was financially supported by the National Natural bearing incorporated with shape memory alloy wires Eng.
Science Foundation of China (Grants 51608136, 51278134), Struct. 61 166–83
[18] Dolce M and Cardone D 2001 Mechanical behaviour of shape
Scientific Research Fund of Institute of Engineering Mechan- memory alloys for seismic applications 2. Austenite NiTi
ics, China Earthquake Administration (Grant 2019D19) and wires subjected to tension. Int. J. Mech. Sci. 43 2657–77
Science and Technology Planning Project of Guangzhou City [19] Dolce M, Cardone D and Palermo G 2007 Seismic isolation of
(Grant 201804010291). bridges using isolation systems based on flat sliding
bearings Bull. Earthq. Eng. 5 491–509
[20] Ozbulut O E and Hurlebaus S 2010 Evaluation of the
ORCID iDs performance of a sliding-type base isolation system with a
NiTi shape memory alloy device considering temperature
Sasa Cao  https://orcid.org/0000-0003-1014-5855 effects Eng. Struct. 32 238–49
Osman E Ozbulut  https://orcid.org/0000-0003-3836-3416 [21] Casciati F and Faravelli L 2009 A passive control device with
SMA components: from the prototype to the model Struct.
Control Health Monit. 16 751–65
References [22] Choi E, Nam T H Oh J T et al 2006 An isolation bearing for
highway bridges using shape memory alloys Mater. Sci.
[1] Jangid R S and Kelly J M 2001 Base isolation for near-fault Eng. A 438 1081–4
motions Earthq. Eng. Struct. Dyn. 30 691–707 [23] Dezfuli F H and Alam M S 2016 Seismic vulnerability
[2] Deb S K and Deb S K 2004 Seismic base isolation—an assessment of a steel-girder highway bridge equipped with
overview Curr. Sci. 87 1426–30 different SMA wire-based smart elastomeric isolators Smart
[3] Li J, Peng T and Yan X 2008 Damage investigation of girder Mater. Struct. 25 075039
bridges under the Wenchuan earthquake and corresponding [24] Dezfuli F H and Alam M S 2017 Smart lead rubber bearings
seismic design recommendations Earthq. Eng. Eng. Vib. equipped with ferrous shape memory alloy wires for
7 337–44 seismically isolating highway bridges J. Earthq. Eng.
[4] Raheem S E A 2009 Pounding mitigation and unseating 22 1042–67
prevention at expansion joints of isolated multi-span [25] Li S, Dezfuli F H Wang J et al 2017 Longitudinal seismic
bridges Eng. Struct. 31 2345–56 response control of long-span cable-stayed bridges using
[5] Matsagar V A and Jangid R S 2003 Seismic response of shape memory alloy wire-based lead rubber bearings under
base-isolated structures during impact with adjacent near-fault records J. Intell. Mater. Syst. Struct. 29 703–28
structures Eng. Struct. 25 1311–23 [26] Ozbulut O E and Hurlebaus S 2010 Seismic assessment of
[6] Roussis P C, Constantinou M C Erdik M et al 2003 bridge structures isolated by a shape memory
Assessment of performance of seismic isolation system of alloy/rubber-based isolation system Smart Mater. Struct.
Bolu Viaduct J. Bridge Eng. 8 182–90 20 015003
[7] Karalar M, Padgett J E and Dicleli M 2012 Parametric analysis [27] Ozbulut O and Hurlebaus S 2011 Energy-balance assessment
of optimum isolator properties for bridges susceptible to of shape memory alloy-based seismic isolation devices
near-fault ground motions Eng. Struct. 40 276–87 Smart Struct. Syst. 8 399–412
[8] Shen J, Tsai M H Chang K C et al 2004 Performance of a [28] Mishra S K, Gur S Roy K et al 2015 Response of bridges
seismically isolated bridge under near-fault earthquake isolated by shape memory–alloy rubber bearing J. Bridge
ground motions J. Struct. Eng. 130 861–8 Eng. 21 1–15
[9] Housner G W and Thiel C C 1995 The continuing challenge: [29] Fenz D M and Constantinou M C 2008 Spherical sliding
report on the performance of state bridges in the Northridge isolation bearings with adaptive behavior: theory Earthq.
earthquake Earthq. Spectra 11 607–36 Eng. Struct. Dyn. 37 163–83
[10] Bruneau M, Wilson J C and Tremblay R 1996 Performance of [30] Xu L and Li J 2016 Design and experimental investigation of a
steel bridges during the 1995 Hyogo-ken Nanbu (Kobe, new type sliding retainer and its efficacy in seismic
Japan) earthquake—a North American perspective Can. J. fortification Eng. Mech. 33 111–18 (in Chinese)
Civ. Eng. 23 678–713 [31] Xu W 2011 Seismic Behavior Analysis of a New Structure
[11] Investigation Group of 921 1999 Chichi Earthquake. System Based on Multilevel Displacement Restricting and
Investigation Report of Damage of Bridges on 921 Chichi Energy Dissipation Device (Dalian: Dalian University of
Earthquake (Taiwan: NCEER) (in Chinese) Technology) (in Chinese)
[12] Casciati F, Faravelli L and Saleh R A 2009 An SMA passive [32] Peng T, Yu X Wang L et al 2012 Development and test of a
device proposed within the highway bridge benchmark multi-defense aseismic spherical bearing J. Tongji Univ.
Struct. Control Health Monit. 16 657–67 (Nat. Sci.) 40 992–5 (in Chinese)

17
Smart Mater. Struct. 29 (2020) 055045 S Cao et al

[33] Yuan W, Cao X and Rong Z 2010 Development and [42] Chopra A K and Goel R K 1999 Capacity-demand-diagram
experimental study on cable-sliding friction aseismic methods based on inelastic design spectrum Earthq. Spectra
bearing J. Harbin Eng. Univ. 31 1593–600 (in Chinese) 15 637–56
[34] Calvi P M, Moratti M and Calvi G M 2016 Seismic isolation [43] Sap C S I 2007 Integrated Software for Structural Analysis &
devices based on sliding between surfaces with variable Design (Berkeley, CA: Computer and Structures, Inc)
friction coefficient Earthq. Spectra 32 2291–315 [44] Tazarv M and Saiidi M S 2015 Reinforcing NiTi superelastic
[35] Dezfuli F H, Li S Alam M S et al 2017 Effect of constitutive SMA for concrete structures J. Struct. Eng. 141 04014197
models on the seismic response of an SMA-LRB isolated [45] Fema F E M A 2009 Quantification of Building Seismic
highway bridge 148 113–25 Performance Factors (Washington, DC: Applied
[36] Reedlunn B, Daly S and Shaw J 2013 Superelastic shape Technology Council)
memory alloy cables: I. Isothermal tension experiments Int. [46] Shahi S K and Baker J W 2014 An efficient algorithm to
J. Solids Struct. 50 3009–26 identify strong-velocity pulses in multicomponent ground
[37] Ozbulut O E, Daghash S and Sherif M M 2016 Shape memory motions Bull. Seismol. Soc. Am. 104 2456–66
alloy cables for structural applications J. Mater. Civ. Eng. [47] Mazza F and Mazza M 2016 Nonlinear seismic analysis of
28 1–10 irregular rc framed buildings base-isolated with friction
[38] Carboni B, Lacarbonara W and Auricchio F 2015 Hysteresis pendulum system under near-fault excitations Soil Dyn.
of multiconfiguration assemblies of nitinol and steel Earthq. Eng. 90 299–312
strands: experiments and phenomenological identification J. [48] Mazza F 2018 Seismic demand of base-isolated irregular
Eng. Mech. 141 04014135 structures subjected to pulse-type earthquakes Soil Dyn.
[39] Sherif M M and Ozbulut O E 2018 Tensile and superelastic Earthq. Eng. 108 111–29
fatigue characterization of NiTi shape memory cables Smart [49] Seismosoft 2013 Seismomatch V2. 1—a computer program
Mater. Struct. 27 015007 for spectrum matching of earthquake records
[40] Fang C, Zheng Y Chen J et al 2019 Superelastic NiTi SMA [50] Atik L A and Abrahamson N 2010 An improved method for
cables: thermal-mechanical behavior, hysteretic modelling nonstationary spectral matching Earthq. Spectra
and seismic application Eng. Struct. 183 533–49 26 601–17
[41] Ministry of Transport of the People’s Republic of China 2008 [51] Antoniou S, Pinho R and Bianchi F 2015 SeismoSignal V5.1:
Guidelines for seismic design of highway bridges JTG/T A Computer Program for Signal Processing of
B02-01-2008 Strong-Motion Data 5.1 edn (Pavia, Italy: Seismosoft Ltd)

18

You might also like