You are on page 1of 19

Dimensional Estimation of Residual-Drift Demands for

Bilinear Bridges under Near-Fault Ground Motions


Zhan Shu1 and Jian Zhang, A.M.ASCE2
Downloaded from ascelibrary.org by Universidad De Las Americas Puebla on 05/04/21. Copyright ASCE. For personal use only; all rights reserved.

Abstract: Residual drift of bridge columns is an important performance measure for structural damage and postearthquake repair decisions.
Yet there are very limited models available for estimating the residual drifts accurately. This paper presents an innovative method to realisti-
cally predict the residual drifts (including their trend and upper bound) of bilinear bridge deck-column systems directly from their inelastic me-
chanical properties and ground motion characteristics. The proposed estimation originated from the rigorous dimensional analysis of
nonlinear time-history responses of various bilinear bridge deck-column systems under near-fault ground motions. Under this framework, the
peak inelastic drifts and residual drifts were presented in dimensionless forms and showed remarkable order and correlation with structure-to-
pulse frequency and the dimensionless nonlinearity index that accounts for the preyielding strength, ground motion amplitude, and softening
or hardening postyielding behavior. Regressive equations for maximum displacement and residual-drift demands were proposed and vali-
dated. The corresponding error of this approach was shown to be lower than those of the existing codified methods. DOI: 10.1061/(ASCE)
BE.1943-5592.0001298. © 2018 American Society of Civil Engineers.
Author keywords: Residual drift; Peak drift; Seismic response estimation; Bilinear bridges; Near-fault ground motions.

Introduction To predict the PD demand, two methods are most commonly


used, the displacement coefficient method by FEMA documents
Bridges are critical components in transportation systems and are (FEMA 1997a, b, 2000, 2005) and the capacity spectrum method
particularly susceptible to damages and interruptions throughout (ATC 1996, 2011), which combine the nonlinear static analysis pro-
earthquake events due to their distributed nature (Tatano and cedures with results from linear dynamic analysis (Makris and
Tsuchiya 2008). The permanent residual drifts (RDs) particularly Psychogios 2006). More studies have been presented to modify and
impact their structural integrity and functionality after major earth- improve these methods (Vidic et al. 1994; Erduran and Kunnath
quakes. Over the decades, studies have proved the RD as an impor- 2010; Chopra and Goel 2000). Zhang et al. (2011) further elabo-
tant consideration in judging a structure’s postearthquake safety rated the estimation of the peak displacement for bridge columns by
and the economic feasibility of repair (Macrae et al. 1994; Ruiz- introducing a dimensionless nonlinearity index that better captured
García and Miranda 2005). For some extreme cases, RDs may be the nonlinear shear–flexure interaction behavior.
too large hence seriously jeopardize structural stability during The earlier studies calculating residual demands were conducted
earthquake aftershocks and make repairs uneconomical, indicat- by Riddell and Newmark (1979) and Mahin and Bertero (1981),
ing that the repair is on par with a complete structural replacement who identified some of the key behavioral aspects associated with
(Bazzurro et al. 2004; Mackie et al. 2011; Ruiz-García and RDs. In current practice, a commonly used method to estimate the
Aguilar 2015). residual demands is to assume that the RDs are equal to a certain
For the displacement-based or more advanced performance- percentage of the PDs, depending on factors such as the hysteretic
based design and evaluation methods that consider the robustness model (MacRae and Kawashima 1997). Such an approach has been
of bridge systems, it is desirable to explicitly evaluate the likely used to estimate the RDs for building structures (Christopoulos
magnitude of RDs. Shaking table tests (Kawashima et al. 1992; et al. 2003; Pampanin et al. 2003; Ruiz-García and Miranda 2005,
Macrae et al. 1994) revealed that the magnitude of RDs was influ-
2006a, b, c, 2010). In 2012, the ATC-58 documentation (FEMA
enced by factors such as the material type, structural configuration,
2012) was finalized, in which a set of equations was proposed to
and peak drifts (PDs). Consequently, various works have been done
relate residual story drift to peak transient story drift for generalized
to estimate the PD and RD demands, trying to avoid tedious labora-
building systems. Bojórquez and Ruiz-García (2013) presented the
tory tests or time-consuming numerical analyses (Zhang et al. 2011;
RD demands for steel moment-resisting frames against narrow-
Ruiz-García and Chora 2015).
band earthquake ground motions in Mexico. Their work also related
the RDs with the PDs, suggesting that limit residual demands be
limited to 0.5 or 1% and limit peak interstory drift demands be lim-
1
Associate Professor, Dept. of Structural Engineering, Tongji Univ., ited to within 2 or 3%. Erochko et al. (2011) performed a series of
Shanghai 200092, China. numerical tests for steel buildings, establishing an upper bound of
2
Associate Professor, Dept. of Civil and Environmental Engineering, RDs related to peak story displacements and yield displacements.
Univ. of California, Los Angeles, CA 90095 (corresponding author). For bridges, as the deck-column systems can usually be simpli-
ORCID: https://orcid.org/0000-0003-1214-5808. Email: zhangj@ucla.edu
fied to a Single-Degree-of-Freedom (SDOF) system, most existing
Note. This manuscript was submitted on May 22, 2017; approved on
May 8, 2018; published online on August 31, 2018. Discussion period studies have focused on grasping the nonlinearity features and seis-
open until January 31, 2019; separate discussions must be submitted for mic damage–preventing techniques. For example, the earthquake-
individual papers. This paper is part of the Journal of Bridge resistant design code in Japan provides a RD estimation for the
Engineering, © ASCE, ISSN 1084-0702. bridge-column system as a function of the PD value, the yield drift,

© ASCE 04018087-1 J. Bridge Eng.

J. Bridge Eng., 2018, 23(11): 04018087


and the postyield stiffness ratio (JSCE 2000). Tirasit and u þ 2 j s v s u_ þ v s uy~f s ðu; u_ Þ ¼ ap gðv p tÞ
€ (1)
Kawashima (2007) proposed an empirical hysteretic model based
on the observed column RDs and damages. Brown and Saiidi where uy = yield displacement; v s = (ke/ms)1/2 = circular frequency
(2011) investigated a bridge pier against the near-fault ground computed using the initial stiffness (ke) and structural mass (ms);
motions where RDs were evaluated. Kawashima et al. (2011) dem- and the function f~s ðu; u_ Þ = nonlinear resisting force-displacement
onstrated the bridge-pier responses and the RD values against im- relationship that depends on the yield displacement (uy) and post-
pulsive ground motions. Motaref et al. (2014) evaluated the RDs yield stiffness hardening ratio (« = kp/ke). Given a specific wave-
considering connection details and various seismic protective form [g(t)] of input ground acceleration, the peak displacement and
designs, such as the application of rubber pads, the engineered ce- the residual displacement of a bilinear SDOF system [characterized
mentitious components, and the carbon-fiber-reinforced polymer by the postyield stiffness ratio (« ), the natural frequency (v s), and
jacketing of column segments. the yield displacement (uy), or equivalently, the characteristic
The objective of this study was to offer a new way to realistically
Downloaded from ascelibrary.org by Universidad De Las Americas Puebla on 05/04/21. Copyright ASCE. For personal use only; all rights reserved.

strength (Qy)] become functions of six variables expressed in Eqs.


predict the trend and the upper bound of RDs directly from the char- (2a) and (2b)
acteristics of a bridge and ground motion, where a mathematically
sound approach based on physics was adopted, namely, the dimen- umax ¼ f ðv s ; j s ; uy ; ɛ; ap ; v p Þ (2a)
sional analysis. The methodology was first used by Makris and
Black (2004a, b, c) on elastic fixed-base structures and was later
extended to inelastic systems (Makris and Psychogios 2006), inelas- uend ¼ f ðv s ; j s ; uy ; ɛ; ap ; v p Þ (2b)
tic soil-structure systems (Zhang and Tang 2009), bridge columns
(Zhang et al. 2011), nonlinear damping systems (Zhang and Xi The variables appearing in Eqs. (2a) and (2b) involve only
2012), and adaptive stiffness and damping devices (Shu et al. two reference dimensions, length and time. According to
2017). Guided by such a concept, the estimation of PDs and RDs Buckingham’s P-theorem (Barenblatt 1996), the number of in-
for SDOF bilinear systems representing generalized bridge systems dependent dimensionless P-parameters is determined as fol-
is proposed and validated in this paper. The proposed equations lows: (6 variables) – (2 reference dimensions) = 4 P-parameters.
relate the inelastic drift demands with the structure-to-pulse fre- By normalizing the response to the characteristic length scale of
quency ratio, normalized yield displacement, postyield stiffness, ground motion, the normalized PD (PPD  umaxv 2p /ap) and the
and normalized strength. In comparison to existing codified predic- normalized RD (PRD  uendv 2p /ap) become functions of the nor-
tions, the method is efficient in reducing the scatter in the engineer- malized frequency (Pv  v s/v p); the structural damping
ing demands and can directly estimate the drift demands without ratio (P j  j s); the normalized structural yield displacement
resorting to the linear time-history responses and spectrum. (Puy  uyv 2p /ap); or equivalently, the normalized strength
(PQ  Qy/msap) and the postyield stiffness ratio (P«  « ), which
is formulated as Eqs. (3a) and (3b)
Dimensional Analysis and Simulation Scheme
PPD  umax v 2p =ap ¼ f PD ðPv ; P j ; Puy ; Pɛ Þ (3a)
This study estimated the inelastic drift demands (i.e., PD and RD
demands) based on a set of dimensionless parameters to describe
the normalized structural responses of a SDOF bilinear system. PRD  uend v 2p =ap ¼ f RD ðPv ; P j ; Puy ; Pɛ Þ (3b)
Seventy-five near-fault ground motions were selected due to their
distinct impulsive features and their devastating effects on From Eqs. (3a) and (3b), the PD and RD demands are normal-
structures. ized with pulse acceleration amplitude (ap) and frequency (v p). For
velocity pulse-type ground motions that demonstrate distinguish-
Dimensional Analysis Framework able pulses in their velocity time histories, the inelastic drift
demands could be normalized with pulse velocity amplitude (vp)
The comprehensive nonlinear seismic responses of bridges are eval- and frequency (v p), shown in Eqs. (4a) and (4b)
uated under the framework of dimensional analysis, which provides
an effective way of interpreting the otherwise largely scattered PPD  umax v p =vp (4a)
inelastic structural responses from time-history analysis using
recorded ground motions. By normalizing the inelastic drift
demands with respect to the energetic length scale of ground PRD  uend v p =vp (4b)
motions, similar responses (i.e., independent of the intensity of
ground motions) can be obtained. In this section, the dimensional To better estimate the inelastic drift demands, appropriate
analysis of a bilinear SDOF bridge model is introduced to provide dimensionless quantities need to be derived and then evaluated in
the logical and mathematic background behind the proposed terms of their influence on the dimensionless drift demands. In addi-
equations. tional to the previously defined P terms, Zhang et al. (2011) intro-
The dimensional analysis framework is based on the existence duced a dimensionless measure PNL to capture the complex hyste-
of a distinct time scale and length scale that characterize the most retic nonlinear behavior of a bilinear oscillating system with the
energetic component of the ground shaking. Such time and length understanding of the direct impact of the energy dissipation mecha-
scales emerge naturally from the distinguishable pulses and pulse- nism, which is expressed by Eq. (5)
like near-fault ground motions [e.g., the acceleration amplitude (ag)    
and duration of the ground motion (Tp)], which dominate a wide ms ag 1 ms ag 1
PNL  ¼ (5)
class of strong earthquake records. For a bilinear SDOF system sub- Qy 1 þ kp =ke Qy 1 þ ɛ
ject to a pulse-type ground motion that can be generally character-
ized by its pulse amplitude (ap) and frequency (v p, or equivalently, where ag = input peak ground acceleration. The first term [(ms ag)/
Tp = 2p /v p), its equation of motion can be expressed by Eq. (1) Qy] in Eq. (5) is similar to the inverse of the normalized strength

© ASCE 04018087-2 J. Bridge Eng.

J. Bridge Eng., 2018, 23(11): 04018087


(PQ) identified previously in the dimensional analysis except the for this study (including 25 acceleration pulse motions and 50 ve-
pulse amplitude (ap) is replaced by the peak ground acceleration locity pulse motions). The pulse-type motions were identified based
(ag). This term indicates that smaller strength relative to the struc- on improved methods to derive the pulse representations using dif-
tural mass or larger input ground motion will result in larger nonli- ferent pulse models (Tang and Zhang 2011). Detailed characteris-
nearity in the structure. The second term [1/(1þ« )] represents ei- tics are provided in Tables 1 and 2 for these selected ground
ther reduced or increased energy dissipation capability due to motions.
softening (kp < 0) or hardening (kp > 0) behavior with reference to Out of the 75 earthquake records, 20 acceleration pulse motions
the elastoplastic system. Therefore, the parameter PNL essentially and 40 velocity pulse motions were used to compute the nonlinear
combines the effect of PQ and postyielding stiffness ratio (P« ). responses and to generate the estimation formula for PD and RD. In
Subsequently, Eqs. (3a) and (3b) can be further simplified to addition, the last five acceleration pulse–type motions and 10 veloc-
ity pulse-type motions (shown as Earthquakes 21–25 in Table 1 and
 PD ðPv ; P j ; PNL Þ
PPD  umax v 2p =ap ¼ f
Downloaded from ascelibrary.org by Universidad De Las Americas Puebla on 05/04/21. Copyright ASCE. For personal use only; all rights reserved.

(6a) Earthquakes 46–50 and 71–75 in Table 2) were reserved as the


ground motion inputs to validate the estimated results.
 RD ðPv ; P j ; PNL Þ
PRD  uend v 2p =ap ¼ f (6b)
Sample Bridge Model
The proposed estimations were provided based on simple SDOF
bilinear systems. The mechanical properties of a SDOF bilinear sys-
Pulse-Type Ground Motions
tem could be sampled from structural pushover analyses (Chopra
Near-fault ground motions often show impulsive characteristics and Goel 2002; Chopra and Chintanapakdee 2004). In addition to
exhibiting distinguishable pulses in their velocity and displacement the basic nonlinear features of a bilinear oscillating system, some
time histories, also occasionally in acceleration time histories, sig- existing studies incorporated other features, such as different struc-
nificantly imposing higher demands on structures (Bertero et al. tural types, site conditions [ATC-40 (ATC 1996) and ATC-58
1978; Hall et al. 1995; Malhotra 1999). The relatively long period (FEMA 2012)], and shear–flexure interactions (Zhang et al. 2011).
pulses in near-fault ground motions are closely related to the seis- Nevertheless, to keep the estimation simple and applicable, this
mological aspects of earthquakes, such as source mechanism, fault study did not include these features in the inelastic drift demand
slip, and rupture directivity (Somerville 2003). estimations.
The pulses are directly related to the rise time and slip velocity As an example, a simplified bridge deck-column model was
of faulting and can be formally extracted with various established selected, representing the typical highway bridges that exist in the
methods (Mavroeidis and Papageorgiou 2003; Baker 2007; He and current bridge inventory in the United States. The primary curve
Agrawal 2008). The inelastic drifts are found to correlate well with of the selected bridge model was close to a bilinear SDOF system
the structure-to-pulse period ratio (Ts/Tp) (Cuesta and Aschheim (Zhang et al. 2011). The bridge model had a total deck mass of
2001; Alavi and Krawinkler 2004; Mavroeidis et al. 2004; Gillie 926.4 t, the initial elastic stiffness (ke) equaled 65 MN/m, and the
et al. 2010). A total of 75 pulse-like ground motions were selected natural frequency was 8.38 rad/s, which corresponded to a

Table 1. Acceleration pulse-type motions

Number Earthquake Year Magnitude (Mw) Station name Distance to fault (km) PGA (g) ap (g) v p (rad/s)
1 COYOTELK/CYC 1979 5.7 Coyote Lake Dam (SW abutment) 6.13 0.28 0.26 12.08
2 LIVERMOR/B-LFA 1980 5.4 Livermore—Fagundas Ranch 14.88 0.22 0.18 15.71
3 COALINGA/A-ATC 1983 5.1 Anticline Ridge Free-Field 12.47 0.86 0.23 22.44
4 COALINGA/A-ATP 1983 5.1 Anticline Ridge Pad 12.47 0.56 0.32 24.17
5 COALINGA/E-CHP 1983 4.9 Coalinga-14th and Elm (Old CHP) 12.18 0.20 0.19 34.91
6 COALINGA/F-CHP 1983 5.2 Coalinga-14th and Elm (Old CHP) 12.74 0.73 0.73 17.95
7 MORGAN/G06 1984 6.2 Gilroy Array #6 9.86 0.32 0.24 29.92
8 PALMSPR/HCP 1986 6.1 Hurkey Creek Park 29.83 0.23 0.20 22.44
9 LOMAP/LEX 1989 6.9 Los Gatos—Lexington Dam 5.02 0.52 0.57 5.32
10 WHITTIER/B-ALH 1987 5.3 Alhambra—Fremont School 14.02 0.21 0.18 18.48
11 WHITTIER/B-OBR 1987 5.3 LA—Obregon Park 15.19 0.34 0.23 24.17
12 CAPEMEND/PET 1992 7 Petrolia 8.18 0.63 0.60 7.95
13 NORTH392/GLB 1994 5.3 Sun Valley—Sunland 16.49 0.35 0.27 20.94
14 NORTHR/PAC 1994 6.7 Pacoima Dam (downstream) 7.01 0.50 0.53 12.82
15 NORTHR/SCS 1994 6.7 Sylmar—Converter Station 5.35 0.80 0.63 5.42
16 NORTHR/SYL 1994 6.7 Sylmar—Olive View Med FF 5.3 0.73 0.61 16.53
17 CHICHI03/CHY080 1999 6.2 CHY 080 22.37 0.47 0.39 5.61
18 SMADRE/altde 1991 5.6 Altadena—Eaton Canyon 13.17 0.36 0.31 16.11
19 SMADRE/opark 1991 5.6 LA—Obregon Park 27.4 0.21 0.18 20.27
20 SMADRE/4734A 1991 5.6 Pasadena—USGS/NSMP Office 17.13 0.33 0.24 16.98
21 WHITTIER/A-NOR 1987 6 Norwalk—Imp Hwy, SGrnd 20.42 0.24 0.20 8.61
22 NORTHR/LOS 1994 6.7 Canyon Country—W Lost Canyon 12.44 0.47 0.38 9.97
23 KOBE/KJM 1995 6.9 KJMA 0.96 0.85 0.68 6.90
24 NORTH392/STC 1994 5.3 Northridge—17645 Saticoy Station 13.87 0.22 0.18 20.94
25 CHICHI06/TCU080 1999 6.3 TCU 080 10.2 0.532 0.46 44.88

© ASCE 04018087-3 J. Bridge Eng.

J. Bridge Eng., 2018, 23(11): 04018087


Table 2. Velocity pulse-type motions

Number Earthquake Year Magnitude (Mw) Station name Distance to fault (km) PGA (g) vp (m/s) v p (rad/s)
26 SFERN/PUL 1971 6.6 Pacoima Dam (upper-left abutment) 1.81 1.43 1.02 4.59
27 IMPVALL/H-AEP 1979 6.5 Aeropuerto Mexicali 0.34 0.36 0.31 2.80
28 IMPVALL/H-AGR 1979 6.5 Agrarias 0.65 0.31 0.43 2.99
29 IMPVALL/H-EMO 1979 6.5 EC Meloland Overpass FF 0.07 0.38 0.86 2.09
30 IMPVALL/H-E03 1979 6.5 El Centro Array #3 12.85 0.27 0.27 1.23
31 IMPVALL/H-E04 1979 6.5 El Centro Array #4 7.05 0.47 0.70 1.43
32 IMPVALL/H-E05 1979 6.5 El Centro Array #5 3.95 0.53 0.84 1.70
33 IMPVALL/H-E06 1979 6.5 El Centro Array #6 1.35 0.44 1.02 1.65
34 IMPVALL/H-E07 1979 6.5 El Centro Array #7 0.56 0.46 0.81 1.73
Downloaded from ascelibrary.org by Universidad De Las Americas Puebla on 05/04/21. Copyright ASCE. For personal use only; all rights reserved.

35 IMPVALL/H-HVP 1979 6.5 Holtville Post Office 7.65 0.26 0.42 1.36
36 MAMMOTH/L-LUL 1980 5.9 Long Valley Dam (Upr L Abut) 19.83 0.99 0.28 6.16
37 COALINGA/H-Z14 1983 6.4 Parkfield—Fault Zone14 29.98 0.26 0.47 5.61
38 COALINGA/D-TSM 1983 5.8 Transmitter Hill 9.52 1.03 0.62 8.27
39 COALINGA/F-CHP 1983 5.2 Coalinga-14th and Elm (Old CHP) 12.74 0.73 0.40 17.95
40 MORGAN/G06 1984 6.2 Gilroy Array #6 9.86 0.32 0.39 5.76
41 PALMSPR/NPS 1986 6.1 North Palm Springs 4.04 0.67 0.55 4.95
42 SANSALV/GIC 1986 5.8 Geotech Investig Center 6.3 0.85 0.71 8.38
43 SANSALV/NGI 1986 5.8 National Geografical Inst 6.99 0.61 0.35 7.85
44 WHITTIER/A-DWN 1987 6 Downey—Co MaintBldg 20.82 0.23 0.26 7.85
45 WHITTIER/A-OR2 1987 6 LB—Orange Ave. 24.54 0.26 0.27 7.76
46 SUPERST/B-PTS 1987 6.5 Parachute Test Site 0.95 0.42 1.08 2.99
47 LOMAP/G02 1989 6.9 Gilroy Array #2 11.07 0.41 0.40 4.22
48 LOMAP/LEX 1989 6.9 Los Gatos—Lexington Dam 5.02 0.52 1.05 5.32
49 ERZIKAN/ERZ 1992 6.7 Erzincan 4.38 0.49 0.79 2.63
50 CAPEMEND/PET 1992 7 Petrolia 8.18 0.63 0.48 2.44
51 LANDERS/YER_225 1992 7.3 Yermo Fire Station 23.62 0.22 0.40 0.90
52 NORTHR/LOS_032 1994 6.7 Canyon Country—W Lost Cany 12.44 0.47 0.37 2.95
53 NORTHR/KAT_032 1994 6.7 Simi Valley—Katherine Rd. 13.42 1.07 0.43 6.83
54 NORTHR/SCS_032 1994 6.7 Sylmar-Converter Station 5.35 0.59 0.84 2.20
55 NORTHR/SCE_032 1994 6.7 Sylmer-Converter Sta East 5.19 0.84 0.75 2.07
56 KOBE/KJM_140 1995 6.9 KJMA 0.96 0.85 0.97 6.90
57 KOBE/TAZ_140 1995 6.9 Takarazuka 0.27 0.65 0.63 3.81
58 KOBE/TAK_140 1995 6.9 Takatori 1.47 0.68 1.08 3.05
59 KOCAELI/GBZ_184 1999 7.5 Gebze 10.92 0.24 0.41 1.28
60 CHICHI/CHY006_292 1999 7.6 CHY 006 9.77 0.31 0.58 2.63
61 CHICHI/CHY035_292 1999 7.6 CHY 035 12.65 0.26 0.33 4.19
62 CHICHI/CHY101_289 1999 7.6 CHY 101 9.96 0.45 0.61 1.09
63 CHICHI/TCU029_306 1999 7.6 TCU 029 28.05 0.22 0.61 1.14
64 CHICHI/TCU036_277 1999 7.6 TCU 036 19.84 0.13 0.61 1.14
65 CHICHI/TCU040_277 1999 7.6 TCU 040 22.08 0.15 0.42 1.14
66 CHICHI/TCU065_272 1999 7.6 TCU 065 0.59 0.82 0.99 1.30
67 CHICHI/TCU075_271 1999 7.6 TCU 075 0.91 0.33 0.75 1.26
68 CHICHI/TCU103_277 1999 7.6 TCU 103 6.1 0.13 0.58 0.81
69 CHICHI/TCU128_306 1999 7.6 TCU 128 13.15 0.19 0.77 1.00
70 CHICHI/TCU136_278 1999 7.6 TCU 136 8.29 0.17 0.39 0.69
71 CHICHI/TCU141_275 1999 7.6 TCU 141 24.21 0.10 0.51 1.30
72 STELIAS/059v2_160 1979 7.5 Icy Bay 26.46 0.16 0.37 2.90
73 YOUNTVL/2016a_061 2000 5 Napa Fire Station #3 14.15 0.60 0.40 9.24
74 CHICHI03/CHY024_270 1999 6.2 CHY 024 19.65 0.19 0.32 2.28
75 CHICHI03/CHY080_270 1999 6.2 CHY 080 22.37 0.06 0.69 5.61

structural period of approximately 0.75 s. The yield displacement excitations are presented in Figs. 1(a and c), showing permanent
(uy) of the sample bridge was 0.013 m, the postyield stiffness ratio RDs occurring toward the same directions as the PDs after both
was 5%, and the structural damping was assumed at 5% for the ground motions. The hysteretic behavior of the bilinear system
bridge model. is illustrated by Figs. 1(b and d), where the circles show the posi-
To demonstrate the PDs and the RDs, the sample bridge tions where the motion of the system stopped. The Kobe record
model was first excited with two pulse-type ground motions pushed the system to a PD of 3.66%. Furthermore, the motion of
(Number 23 from the 1995 Kobe earthquake and Number 26 the SDOF system rested with a permanent drift of 0.45%. In
from the 1971 San Fernando earthquake) selected from Tables 1 addition, the PD and RD for the San Fernando record were 5.41
and 2. The nonlinear time-history responses of the two ground and 0.76%.

© ASCE 04018087-4 J. Bridge Eng.

J. Bridge Eng., 2018, 23(11): 04018087


Drift time history
2

Drift (%)
0

-2
Record #23 Kobe KJM 1995
-4
0 10 20 30 40 50
Time (sec)
Total acceleration time history
Acceleration (g)

0.2
Downloaded from ascelibrary.org by Universidad De Las Americas Puebla on 05/04/21. Copyright ASCE. For personal use only; all rights reserved.

-0.2
0 10 20 30 40 50
Time (sec) Drift (%)
(a) (b)

Drift time history


4
2
Drift (%)

0
-2
-4
-6 Record #26 Pacoima dam San Fernando, 1971
-8
0 10 20 30 40
Time (sec)
Total acceleration time history
Acceleration (g)

0.2

-0.2
0 10 20 30 40
Time (sec)
(c) (d)

Fig. 1. Nonlinear responses from two presented motions (system properties: Qy/W = 8.77%, « = 5%, j = 5%): (a) PD and acceleration time history
against EQ #23 KOBE/KJM in Table 1; (b) system acceleration-drift behavior against EQ #23 KOBE/KJM; (c) PD and acceleration time history
against EQ #26 Pacoima Dam in Table 2; and (d) system acceleration-drift behavior against EQ #26 Pacoima Dam.

Prediction of Peak Transient Drifts and RDs Table 3. Nonlinear parameters for the six bilinear configurations

Purpose System « (%) Qy/W (%)


In addition to the prototype bridge-column model, nine more SDOF
bridge models with the same elastic stiffness (ke) but different post- To predict SYS1 5 8.8
yield stiffnesses (kp) and yield strengths (Qy) are presented and SYS2 0 8.8
were simulated to represent a wider range of bridge structures. The SYS3 −5 8.8
first six bilinear systems were used to generate the nonlinear drift- SYS4 8 8.8
estimating formula, whose configurations were as follows: (1) the SYS5 5 17.6
sample column model with intermediate postyield stiffness of kp = SYS6 5 35.2
5%ke and yield strength Qy (SYS1), (2) system with zero postyield To validate SYS7 0 17.6
stiffness of kp = 0%ke and yield strength Qy (SYS2), (3) system with SYS8 2 8.8
SYS9 −2 8.8
negative postyield stiffness of kp = –5%ke and yield strength Qy
SYS10 8 17.6
(SYS3), (4) system with larger postyield stiffness of kp = 8%ke and
yield strength Qy (SYS4), (5) system with intermediate postyield
stiffness of kp = 5%ke and yield strength 2Qy (SYS5), and (6) system for all the prediction and validation cases. Therefore, the presented
with intermediate postyield stiffness of kp = 5%ke and yield strength formula in the study cannot be applied to all generalized bilinear
4Qy (SYS6). In addition, four more configurations were selected to systems before further validations are made.
validate the estimated results. The detailed features for each system
are summarized in Table 3. Finally, because the study was limited
Estimation of Peak Transient Drifts
to the bridge systems, the bilinear systems were all modified from
the prototype bridge-column system presented earlier. Specifically, The estimation of the PD demand was presented first before the esti-
the elastic stiffnesses and the damping ratios remained unchanged mation of the RD demands. Zhang et al. (2011) proposed an

© ASCE 04018087-5 J. Bridge Eng.

J. Bridge Eng., 2018, 23(11): 04018087


equation to estimate inelastic displacement of bridge columns under estimation was relatively small, the damping ratio was also assumed
the dimensional analysis framework. The estimation assumed a to be fixed at 5%, similar to the work by Zhang et al. (2011).
fixed damping ratio at 5%. Then, the dimensional PD was written as Therefore, Eq. (9) closely follows Eqs. (6a) and (6b) with constant
Eq. (7) P j and indicates that the normalized PDs were related to the struc-
ture-to-pulse frequency (Pv ) and nonlinear index (PNL ) mainly.
umax v 2p The constant a4 in Eq. (9) is the upper bound for the first term (|Pv –
PPD ¼ ¼ 1:35jPv  1:0j1:25 SMF1  2:4SMF1 (7)
ap a1|a2 ) in the cases when the term (Pv – a1) approaches zero.
This study refined the aforementioned coefficients by minimizing
where the spectral modification factor (SMF1) is a function of the the error between the prediction and the numerically computed
nonlinearity index and aspect ratio given by Eq. (8) results. For a simplified bilinear system, the estimation formula
for the PD ratio under the dimensional framework was expressed
Downloaded from ascelibrary.org by Universidad De Las Americas Puebla on 05/04/21. Copyright ASCE. For personal use only; all rights reserved.

H by Eq. (10)
SMF1  ðPNL ÞCNL where CNL ¼ 0:8a2 and a ¼ 0:4 (8)
D
PPD ¼ 1:5Pv 1 P0:15
NL < 2:5PNL
0:15
(10)
where H:D = aspect ratio of the bridge columns. When the aspect ra-
tio was less than 3.0, column behaviors were dominated by shear– Fig. 2 shows the normalized PD demands (PPD) of the six bilin-
flexure interaction. For taller columns (i.e., higher aspect ratios), the ear systems against the structure-to-pulse frequency ratio (Pv ). The
behaviors were dominated mainly by flexure. Eqs. (7) and (8) physi- pluses correspond to the simulated values, and the circles are the
cally reflect the fact that a larger nonlinearity index (PNL) will predicted results from Eq. (10). The prediction errors are defined in
result in larger normalized peak displacement (PPD), and its effect subsequent sections. For the six bridge systems, the errors between
is more significant for columns with smaller aspect ratios. the simulated and predicted PD values are also shown in Fig. 2 for
Motivated by Eqs. (7) and (8), this study recommends the following each case. The errors were approximately 45% for all cases. Further
format to estimate the PDs for a bilinear system: evaluation of the presented formula is presented in the validation
part of the subsequent section.
PPD ¼ jPv  a1 ja2 PNL a3 < a4 PNL a3 (9)
Estimation of RDs
where a1, a2, and a3 = constants to be determined from regression
analysis aiming to minimize errors compared with numerical analy- For a SDOF bridge model, the RDs were sensitive to a set of param-
ses. Because the influence of the damping ratio on the drift eters, including the nonlinear system configurations and the input

Fig. 2. PPD–Pv relations for the six bilinear systems: (a) SYS1; (b) SYS2; (c) SYS3; (d) SYS4; (e) SYS5; and (f) SYS6.

© ASCE 04018087-6 J. Bridge Eng.

J. Bridge Eng., 2018, 23(11): 04018087


ground motion characteristics. For a bilinear system, RDs were a Fig. 3 plots the normalized residual PRD of two different bilinear
function of the lateral strength of the structure (Qy) relative to the SDOF systems versus Pv . The pluses represent the calculated PRD
earthquake demand and the inelastic postyield stiffness (« ). For the from nonlinear analyses, and the circles show the estimation by Eq.
nonlinear behaviors other than the bilinear behavior, the cyclic (12). It was noticed that larger RDs occurred when the postyield
unloading configuration of the system also mattered. For exam- stiffness ratios were smaller (i.e., SYS2 and SYS3). The SYS3 with
ple, the self-centering devices could significantly reduce the RD a negative postyield stiffness ratio was the highly nonlinear system,
demands due to their recentering capabilities during system which corresponded to larger RDs with more scattered distribution.
unloading. Then, comparing Fig. 3(a) with Figs. 3(e and f), one could notice
In addition, features of earthquake inputs, which include the that when Qy increased, the RD distribution became less scattered,
identification of pulses in the ground motions, the duration of and the errors for the prediction were consequently reduced.
ground shaking, and the magnitude of PDs, will also impact the Reviewing all six cases, it can be concluded that the proposed RD
Downloaded from ascelibrary.org by Universidad De Las Americas Puebla on 05/04/21. Copyright ASCE. For personal use only; all rights reserved.

RDs. Specifically, the duration of ground shaking for a pulse motion prediction was reasonable compared with the simulated data.
added some randomness to the RD. The system would stop at a cer- Furthermore, a detailed study was performed to evaluate the
tain point somewhere randomly located between zero and its maxi- influences of different types of pulse waveforms on RDs. There
mum plastic deformation (determined by the PD minus uy for a are a few pulse models available in the literature. For example,
bilinear system). Makris and Chang (2000) classified the pulse excitations into
The RDs were not only sensitive to Pv , but were also highly three distinctive types: Type A (one sine acceleration pulse),
sensitive to the ductility demand (i.e., PD/uy) and postyield stiffness Type B (one cosine acceleration pulse), and Type Cn (n main
ratio (« ). A set of regressive models are proposed here to predict cycles in the displacement time history). In addition, the
RD demands. The regression procedure adopts the Weibull function Mavroeidis and Papageorgiou (2003) pulse model was essen-
to identify the relation between PRD and Pv , as the Weibull func- tially an amplitude-modulated sinusoidal function where five
tion comprehends different variations for nonlinear curve fitting. parameters (namely A, fp, , g , and t0) control the amplitude,
The result of the mathematical approach finally suggested a slightly frequency, phase, oscillatory character, and envelope, respec-
better linear correlation between log(PRD) and log(Pv ). However, tively. Fig. 4(a) shows the pulses defined by Mavroeidis and
the fit between log(PRD) and log(Pv ) was still rather scattered, Papageorgiou (2003) corresponding to different values of the os-
indicating that the RD values were not merely correlated with Pv . cillatory variable g = 1.01, 2, 3, and 4, respectively, with a ve-
Therefore, the study systematically chose different formats of func- locity amplitude of A = 0.5 m/s and frequency of fp = 2 Hz. The
tions with values of « , PNL, PRD, and Pv that had physical mean- corresponding PD and RD spectra are also provided in Fig. 4,
ings as well as produced a minimal overall error. Based on the anal- where the solid lines represent the calculated values and the
ysis, the RD could first be estimated in a format close to the PD dash-dotted lines represent the estimations provided earlier.
formula, which is expressed by Eq. (11) Fig. 4 shows that the proposed estimations generally captured
the trend of both PDs and RDs. The peak values of PDs shifted
PRD ¼ c1 PcNL
2
 Pcv3 (11) toward larger frequencies when the oscillation cycles increased
(e.g., g = 3 and 4 cases). In addition, the peak values of RDs
where c1, c2, and c3 = constants to be determined. However, such occurred around the peak values of PDs. As cycles of the motion
expression could not emphasize the postyield stiffness ratio, which increased, the RDs generally diminished. In the validation sec-
had a large impact on RD demands. In addition, by setting c1, c2, tion, it is further shown through the spectra of two real earth-
and c3 constants, the influence of the nonlinearity was ignored, quake motions that the difference in waveforms disappeared in
which was undesirable. To minimize the overall error from the 60 the real earthquake motions, and the proposed predictions of
earthquake cases, a finalized formula estimating the RD demands PDs and RDs were reliable for distinctive waveforms.
was presented by Eq. (12)
8 Correlations between PD and RD
>
> P ðɛ < 5%Þ
< PD
PRD ¼ ð0:9  9ɛÞPNL  Pv NL
0:15 P 0:1
2:5 ð5%  ɛ  10%Þ Many existing studies incorporated the PD into the RD-estimation
>
> formula for more accurate results. For example, the Japan Society
:
0 ðɛ > 10%Þ of Civil Engineers (JSCE) code defines an estimation of residual
(12) displacement after system yields with Eq. (13) (JSCE 2000)
  uend ¼ CR ð m R  1Þð1  ɛÞuy ¼ CR ðumax  uy Þð1  ɛÞ (13)
and PRD  min PPD ; 0:8PNL 0:15
.
Eq. (12) first suggested that the RDs were the same as the PDs
for the systems with postyield stiffness ratios smaller than –5%. where m R = response ductility factor of a bilinear system (i.e., col-
Then, the RD-estimation formula was provided for the more gener- umns) ( m R = umax/uy); « = postyield stiffness ratio; and CR = RD
alized cases when the postyield stiffness ratios were between –5 and modification factor decided by the postyield stiffness ratio (« ).
10%. Moreover, for those special nonlinear systems when the post- Then, CR = 0.5 is recommended for RC bridge piers based on the re-
yield stiffness ratios were larger than 10%, the proposed formula sidual displacement response spectrum (Kawashima et al. 1998).
gave a zero RD estimation. It was also noticed that when « JSCE is a pioneering codified approach to estimating RD
approached 10%, estimation from the main RD-estimation formula demands with a focus on the bridge designs. However, two prob-
also provided a result close to zero. Comparing the main RD- lems exist for the estimation. First, the JSCE code offers limited PD
estimation formula with Eq. (11), c1 and c3 were no longer con- estimation, which is needed to predict the RD demands. Second,
stants, and the updated coefficients involved the postyield stiffness setting CR at a constant value turns out to be inaccurate. In fact,
ratio and the nonlinearity term. It was noted that PRD increased when the system yields (i.e., umax > uy), CR can be back-calculated
with PNL, which means that larger nonlinearity led to larger esti- from Eq. (14) with the RDs and PDs calculated from nonlinear
mated RD values. In addition, PRD decreased when « increased. analyses

© ASCE 04018087-7 J. Bridge Eng.

J. Bridge Eng., 2018, 23(11): 04018087


Downloaded from ascelibrary.org by Universidad De Las Americas Puebla on 05/04/21. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. PRD–Pv relations for the six bilinear systems: (a) SYS1; (b) SYS2; (c) SYS3; (d) SYS4; (e) SYS5; and (f) SYS6.

uend PRD with the work by Erochko et al. (2011), who established a similar
CR ¼ ¼ (14)
ð
ðumax  uy Þ 1  ɛ Þ ðPPD  Puy Þð1  ɛÞ RD upper bound expressed by (umax – uy) for nonlinear building
structures yet not under the dimensional analysis framework. For
The proposed PRD formula from Eq. (12) represents signifi- the rest of the cases, the distributions for CR were highly scattered,
cant improvements over the JSCE RD-estimating models. First, which indicated some more dynamic PRD–PPD correlations rather
the proposed estimation does not simply depend on a coefficient than a fixed CR value.
that could be multiplied by PD. Instead, RDs are directly eval- Subsequently, the study defined a more robust relationship
uated by the normalized drift demands without running nonlin- (namely, CR|P) between PRD and PPD. Similar to the RD- and PD-
ear time-history analyses. Second, the prediction does not rely estimation formula, a logarithmic interdependency between CR|P
on a provided PD estimation. Third, the proposed formula is in and Pv is expressed by Eq. (15)
an ordered form under the framework of dimensional analysis. CRjP ¼ t 1 Ptv2 (15)
The estimation is more reliable and meaningful than the existing
approaches because it not only incorporates the features of dif-
ferent bilinear systems but also considers the variations from the where t 1 and t 2 = coefficients to be determined. Similar to the RD-
ground motion characteristics. estimation formula, the study recommended that the CR|P formula
Fig. 5 shows the correlations among PRD, CR, and (PPD – Puy) also involve the nonlinearity configurations of the bilinear systems.
(1 – « ) for the six bilinear systems. The circles show the back- Moreover, this study established the PRD–PPD correlation under
calculated CR values from Eq. (14) based on the nonlinear analyses. the dimensional analysis framework, and the CR|P formula was
The JSCE-defined coefficient CR consequently became the tangent defined by dividing PRD by PPD. For a system with a postyield
for each dot. Two typical CR values (i.e., 0.5 and 1.0) are marked as stiffness between –5 and 10%, the updated expression is presented
the slopes in Fig. 5. The CR = 0.5 line is the recommended value for in Eq. (16)
RC bridge piers. It could then be noticed that CR values were greater
uend PRD ð0:9  9ɛÞP0:15 PNL 2:50:1

than 0.5 after many earthquakes. In addition, CR could approach 1, NL  Pv


CRjP ¼ ¼ ¼
indicating that the maximum RD values were represented by the umax PPD 1:5Pv 1 P0:15
NL
plastic displacement term [(umax – uy)(1 – « )]. It could be further
¼ ð0:6  6ɛÞPv PNL 1:5  1
0:1
noticed that the majority of the cases approached the 1:1 line for (16)
SYS3 with a negative postyield stiffness ratio. Such an observation
indicated that the upper bound of the PRD was defined by (PPD – Compared with Eq. (15), the coefficients t 1 = 0.6 – 6« and t 2 =
Puy)(1 – « ). Such an upper bound of RDs agrees to some extent PNL
0:1
–1.5 were associated with the bilinear system configurations.

© ASCE 04018087-8 J. Bridge Eng.

J. Bridge Eng., 2018, 23(11): 04018087


Downloaded from ascelibrary.org by Universidad De Las Americas Puebla on 05/04/21. Copyright ASCE. For personal use only; all rights reserved.

Fig. 4. Predictedand calculatednormalized PDsand RDsofabilinear bridge(ɛ ¼ 0:05, j ¼ 0:05,and Q=W ¼ 0:088) when subject to different pulsesdefined
by Mavroeidis and Papageorgiou (2003): (a) Acceleration waveform; (b) Dimensionless peak displacement; and (c) Dimensionless residual displacement.

Fig. 5. Correlations between PRD and (PPD – Puy)(1 – « ): (a) SYS1; (b) SYS2; (c) SYS3; (d) SYS4; (e) SYS5; and (f) SYS6.

© ASCE 04018087-9 J. Bridge Eng.

J. Bridge Eng., 2018, 23(11): 04018087


Downloaded from ascelibrary.org by Universidad De Las Americas Puebla on 05/04/21. Copyright ASCE. For personal use only; all rights reserved.

Fig. 6. Correlations between CR|P and Pv : (a) SYS1; (b) SYS2; (c) SYS3; (d) SYS4; (e) SYS5; and (f) SYS6.

Instead of a constant value, this proposed CR|P formula estab- Table 4. CQ/W coefficient and f « factor
lished a dynamic relation between PD and RD that also captured
Q/W (%) CQ/W « (%) f« « (%) f«
the system nonlinearity and the ground motion characteristics.
Fig. 6 shows the RD–PD correlations for the six bilinear sys- 5 1.2 ≤–4 1.000 6 0.443
tems. The triangles in Fig. 6 were computed by dividing the RDs 10 1.2 −3 0.980 7 0.415
by PDs obtained from the nonlinear analyses, and the circles are 15 1.3 −2 0.960 8 0.388
the estimated CR|P values according to the proposed formula. It 20 1.3 −1 0.940 9 0.360
was observed that the triangles were rather scattered, indicating 25 1.4 0 0.920 10 0.333
that the RD–PD correlation was complicated. Especially for 30 1.1 1 0.580 11 0.305
SYS2 and SYS3 when the postyield stiffness ratios were small, 35 0.95 2 0.553 12 0.278
the RDs may have varied from 0 to near 100% of the PDs. In 40 0.95 3 0.525 13 0.250
addition, the estimated formula could capture different bilinear 45 0.95 4 0.498 14 0.223
features from the system, and the CR|P values generally grasped ≥50 0.95 5 0.470 ≥15 0.200
the trends of the RD-to-PD correlations.
values of the CQ/W coefficient and the f « factor are provided in
Maximum Possible RD Table 4. Notice that the discrete data are provided in the table,
Due to the fact that randomness was involved in the RD distribu- which means that the other values in between need to be
tions, the study further explored the possible upper-bound index interpolated.
(CR|Pmax) of the RDs in addition to the PRD–PPD correlation pre- In addition, the data presented in Table 4 are also illustrated in
sented earlier. For example, from Fig. 6, it was noticed that the Figs. 7 and 8. For example, the CR|Pmax values for the six bilinear
CR|Pmax was approximately 0.55 for SYS1 with a 5% postyield stiff- systems were 1.2  0.47 = 0.56, 1.2  0.92 = 1.1, 1.2  1.0 = 1.2,
ness ratio and 8.8% Qy/W. For the typical bridge systems that could 1.2  0.388 = 0.47, 1.3  0.47 = 0.61, and 0.95  0.47 = 0.45. As
be simplified to a bilinear SDOF system, the study recommended an the CR|Pmax values were always smaller than 1.0, the updated RD
upper bound of RD-to-PD ratio (i.e., CR|Pmax) expressed by Eq. (17) upper-bound indices for the six systems were 0.56, 1.0, 1.0, 0.47,
0.61, and 0.45, which generally matched the obtained results from
CRjPmax ¼ CQ=W  f ɛ  1 (17) Fig. 6.

where CQ/W = coefficient considering the influences from the char-


Application of Suggested Equations
acteristic strength; and f « = factor considering the influences from
the postyield stiffness ratio. The study sampled the data from 10 The application of the suggested equations was simple. For exam-
Q/W values from 5 to 50% and 20 « values from –4 to 15%. The ple, for SYS1 excited by an impulsive excitation with ag = 0.4g and

© ASCE 04018087-10 J. Bridge Eng.

J. Bridge Eng., 2018, 23(11): 04018087


Downloaded from ascelibrary.org by Universidad De Las Americas Puebla on 05/04/21. Copyright ASCE. For personal use only; all rights reserved.

Fig. 7. CQ/W coefficient of CR|Pmax.

Fig. 8. f « factor of CR|Pmax.

v p = 15 rad/s, the PNL term could be computed as PNL = (ms ag)/ First, the PD and RD spectra were provided for two of the selected
[Qy (1 þ « )] = 4.11. Then, Pv = v s/v p = 0.56. The PPD = min[1.5  ground motions. Second, the inelastic drifts were estimated from
(0.56)−1, 2.5]  4.110.15 = 3.09, which was directly computed two existing codified methods as well as the proposed method.
from Eq. (10). Then, umax could be further computed based on the Then, the errors of these methods were defined and calculated to
impulsive features of an individual ground motion record, which verify the prediction accuracies. All results are presented in
was umax = PPD/v 2p  ap = 3.09/225  3.92 m = 0.0538 m. Notice Tables 5–8.
that the yield point of the sample bridge was 0.013 m.
Similarly, PRD = min[(0.9 – 9  0.05)  (0.56)−1.35, 0.8]  Spectra Validations with Two Earthquakes
4.11°0.1 5 = 0.99, which was computed from Eq. (12). Consequently,
uend = PRD/v 2p  ap = 0.99/225  3.92 m = 0.0172 m, which was Two earthquakes were selected for this section: acceleration pulse-
calculated based on the impulsive features of the ground motion re- type Earthquake 21 (i.e., Whittier/A-NOR in Table 1) and velocity
cord. As the estimation of the residual displacement might be less pulse–type Earthquake 71 (i.e., Chichi/TCU141_275 from Table 2).
accurate, the more meaningful upper bound of the RD was provided Fig. 9 shows the results for Earthquake 21, with the acceleration
as 0.56  0.0538 m = 0.301 m. time history of the record and its pulse representation [Fig.
9(a)], the normalized PDs [Fig. 9(b)], and the normalized RDs
[Fig. 9(c)] obtained from simulations using earthquake record
Validations of the Proposed Estimation (solid line, EQ), the pulse representation (dotted line, MP), and
the predictions proposed in this study (dash-dotted line). It was
This section describes testing of the accuracy of the proposed PD demonstrated that the predictions using the proposed formulas
and RD demand models provided in the previous sections. The last captured the general trend and upper bound for both PDs and
four bilinear configurations listed in Table 3 and 15 near-fault RDs. Although there was a noticeable difference between the
pulse-type ground motions (i.e., the last five records from Table 1 simulation results using the real earthquake motion and its pulse
and the last 10 records from Table 2) were used for the validations. representation at some frequencies, they generally agreed with

© ASCE 04018087-11 J. Bridge Eng.

J. Bridge Eng., 2018, 23(11): 04018087


Table 5. Validation data from SYS7 against the 15 ground motions

Computed ATC prediction JSCE prediction Proposed prediction


EQ number PD (%) RD (%) PD (%) Error (%) RD (%) Error (%) RD (%) Error (%) PD (%) Error (%) RD (%) Error (%)
21 0.92 0.34 1.02 10.87 0.15 20.13 0.26 8.92 0.70 23.54 0.43 9.50
22 1.70 0.64 2.15 26.81 0.49 9.01 0.65 0.40 1.28 24.82 0.82 10.63
23 7.28 4.18 5.49 24.50 6.07 25.95 3.44 10.18 3.60 50.57 2.03 29.57
24 0.49 0.09 0.64 30.89 0.03 12.38 0.04 8.84 0.17 64.90 0.12 7.59
25 0.99 0.57 1.92 94.31 0.17 40.34 0.29 28.53 0.11 89.01 0.08 50.15
46 5.59 2.19 2.14 61.65 4.38 39.17 2.59 7.21 4.10 26.76 2.09 1.75
47 4.38 1.22 2.46 43.89 3.17 44.62 1.99 17.66 2.75 37.15 0.88 7.62
Downloaded from ascelibrary.org by Universidad De Las Americas Puebla on 05/04/21. Copyright ASCE. For personal use only; all rights reserved.

48 5.45 3.36 2.07 62.06 4.24 16.01 2.52 15.46 3.62 33.49 1.43 35.58
49 2.61 2.22 1.57 39.84 1.40 31.44 1.10 42.73 1.35 48.41 0.61 61.68
50 3.67 0.72 2.99 18.58 2.46 47.25 1.63 24.78 1.69 53.83 0.65 2.05
71 7.28 4.18 5.49 24.50 6.07 25.95 3.44 10.18 3.60 50.58 2.02 29.58
72 1.87 0.74 2.37 26.47 0.66 4.21 0.73 0.27 2.25 20.17 1.01 14.74
73 2.33 0.87 1.83 21.48 1.12 10.62 0.96 3.94 1.84 21.04 0.66 8.99
74 1.83 0.28 0.71 60.91 0.62 18.49 0.71 23.72 2.41 31.94 0.63 19.51
75 0.80 0.30 0.63 20.82 0.12 22.33 0.20 12.44 1.08 35.06 0.38 9.94
Average 37.84 24.53 14.35 40.75 19.93
Note: SYS7: Qy/W = 17.6%; Dy = 0.4%; « = 0; and CR|Pmax = 1.

Table 6. Validation data from SYS8 against the 15 ground motions

Computed ATC prediction JSCE prediction Proposed prediction


EQ number PD (%) RD (%) PD (%) Error (%) RD (%) Error (%) RD (%) Error (%) PD (%) Error (%) RD (%) Error (%)
21 0.64 0.21 1.03 60.75 0.13 19.49 0.21 0.09 0.78 21.67 0.38 38.83
22 2.11 0.05 2.18 3.39 1.49 102.85 0.93 62.91 1.41 33.06 0.72 47.48
23 4.54 0.34 5.55 22.39 3.92 118.77 2.12 59.08 3.98 12.32 1.82 49.05
24 0.50 0.29 0.65 29.84 0.09 60.01 0.14 43.24 0.19 61.69 0.14 44.92
25 1.44 0.60 1.94 35.12 0.82 22.63 0.60 0.03 0.12 91.65 0.09 54.16
46 5.87 1.80 2.17 62.97 5.25 88.48 2.77 24.90 4.53 22.77 1.95 3.75
47 5.86 0.17 2.49 57.47 5.24 130.43 2.77 66.88 3.05 48.01 0.88 18.27
48 5.99 0.59 2.09 65.04 5.37 120.22 2.83 56.38 4.01 33.05 1.36 19.43
49 2.69 1.55 1.59 41.03 2.07 29.37 1.22 18.51 1.49 44.66 0.57 54.93
50 4.86 0.03 3.03 37.70 4.24 130.69 2.28 69.86 1.87 61.48 0.63 18.75
71 4.54 0.34 5.55 22.39 3.92 118.77 2.12 59.08 3.98 12.34 1.82 49.04
72 1.82 0.87 2.40 31.70 1.20 27.48 0.79 6.60 2.49 36.58 0.96 7.01
73 2.84 0.43 1.85 34.70 2.22 95.09 1.29 45.71 2.04 28.22 0.64 11.15
74 4.39 0.16 0.72 83.63 3.77 123.99 2.05 64.88 2.67 39.26 0.65 16.70
75 1.04 0.19 0.64 39.07 0.42 33.73 0.41 31.58 1.19 14.35 0.36 24.05
Average 41.81 81.47 40.65 37.41 30.50
Note: SYS8: Qy/W = 8.8%; Dy = 0.2%; « = 2%; and CR|Pmax = 0.66.

each other at most frequency ranges, which indicated that the physics involved with the nonlinear structural responses and the fi-
high frequency contents overriding the pulses had a limited delity of the proposed methods (see the error analysis in the next
impact on PDs and RDs. section). They can be used as an efficient assessment and predesign
Similarly, Fig. 10 shows the results for Earthquake 71, which tool.
was a velocity pulse-type motion. The velocity time histories were
compared between the earthquake motion and its pulse representa- Comparison of the Proposed Model with
tion in Fig. 10(a). It was shown that the proposed PD and RD esti-
Existing Methods
mations again captured well the general trend and upper bound of
the simulated responses from the real earthquake record and its The JSCE prediction and the ATC-58 prediction were selected as a
pulse representation. The differences in responses between the real reference to the presented formula. The two codified predictions
earthquake motion and its pulse representation were moderate. The and the proposed estimation model were comparable because that
results from these two earthquake motions gave a glimpse of the they were all developed using the nondegrading bilinear SDOF
complex yet promising nature of the proposed simplified methods models subject to ground motions.
to predict PDs and RDs for nonlinear systems. Although the inten- The ATC-58 estimation of the PD demand was designed for the
tion was not to replace the nonlinear time-history analyses with building system. Consequently, a SDOF one-story building was
these simplified predictions, the study demonstrated the underlying designed with similar nonlinear features to estimate the PD based

© ASCE 04018087-12 J. Bridge Eng.

J. Bridge Eng., 2018, 23(11): 04018087


Table 7. Validation data from SYS9 against the 15 ground motions

Computed ATC prediction JSCE prediction Proposed prediction


EQ number PD (%) RD (%) PD (%) Error (%) RD (%) Error (%) RD (%) Error (%) PD (%) Error (%) RD (%) Error (%)
21 0.63 0.28 1.03 62.10 0.13 23.40 0.22 8.95 0.78 23.36 0.57 45.73
22 3.78 3.65 2.18 42.14 3.18 12.38 1.82 48.33 1.42 62.35 1.08 68.01
23 11.02 3.20 5.56 49.59 10.42 65.53 5.52 21.01 4.00 63.67 2.75 4.12
24 0.54 0.35 0.65 19.64 0.10 45.60 0.17 32.30 0.19 64.50 0.14 39.04
25 2.18 2.03 1.94 11.07 1.59 19.93 1.01 46.37 0.12 94.47 0.09 88.77
46 10.72 4.72 2.17 79.71 10.13 50.42 5.37 6.02 4.56 57.47 2.95 16.49
47 9.02 4.12 2.49 72.35 8.43 47.70 4.50 4.16 3.06 66.02 1.34 30.88
Downloaded from ascelibrary.org by Universidad De Las Americas Puebla on 05/04/21. Copyright ASCE. For personal use only; all rights reserved.

48 11.56 5.94 2.09 81.88 10.97 43.52 5.80 1.22 4.03 65.13 2.07 33.45
49 10.19 2.65 1.59 84.42 9.60 68.12 5.10 23.96 1.50 85.30 0.86 17.63
50 11.36 3.12 3.03 73.32 10.77 67.34 5.69 22.68 1.88 83.42 0.96 18.99
71 8.79 3.10 5.56 36.78 8.19 57.91 4.38 14.52 4.00 54.45 2.75 4.03
72 3.13 2.99 2.40 23.17 2.53 14.54 1.49 47.77 2.50 19.92 1.45 49.20
73 4.71 4.60 1.85 60.66 4.12 10.32 2.30 48.86 2.05 56.53 0.97 77.17
74 11.67 3.24 0.72 93.84 11.08 67.12 5.85 22.34 2.68 77.00 0.98 19.36
75 1.86 1.70 0.64 65.82 1.27 23.07 0.85 45.57 1.20 35.50 0.54 61.97
Average 57.10 41.13 26.27 60.61 38.32
Note: SYS9: Qy/W = 8.8%; Dy = 0.2%; « = –2%; and CR|Pmax = 1.00.

Table 8. Validation data from SYS10 against the 15 ground motions

Computed ATC prediction JSCE prediction Proposed prediction


EQ number PD (%) RD (%) PD (%) Error (%) RD (%) Error (%) RD (%) Error (%) PD (%) Error (%) RD (%) Error (%)
21 0.90 0.17 1.02 13.74 0.14 6.65 0.21 9.67 0.70 22.41 0.08 18.39
22 1.67 0.05 2.15 28.70 0.37 37.88 0.57 61.48 1.26 24.48 0.16 13.08
23 3.72 0.20 5.48 47.57 2.40 118.13 1.51 70.12 3.56 4.32 0.40 10.50
24 0.49 0.04 0.64 32.20 0.01 11.62 0.02 8.55 0.17 64.94 0.05 2.35
25 0.91 0.27 1.91 110.15 0.14 28.93 0.22 12.34 0.11 88.23 0.08 43.09
46 4.51 0.46 2.14 52.53 3.20 121.22 1.87 62.62 4.05 10.27 0.41 2.20
47 3.38 0.88 2.45 27.36 2.06 69.77 1.35 27.75 2.72 19.44 0.17 42.13
48 4.98 0.15 2.06 58.52 3.66 140.99 2.09 77.80 3.58 28.04 0.28 5.14
49 1.96 1.04 1.57 20.02 0.64 40.24 0.70 34.47 1.33 32.13 0.12 93.67
50 3.02 0.26 2.98 1.33 1.70 95.73 1.19 61.54 1.67 44.60 0.13 8.70
71 3.72 0.20 5.48 47.57 2.40 118.13 1.51 70.12 3.56 4.33 0.40 10.50
72 1.90 0.36 2.36 24.09 0.59 23.81 0.67 32.87 2.22 16.71 0.20 17.01
73 2.14 0.07 1.83 14.50 0.82 70.43 0.78 66.73 1.82 14.91 0.13 5.75
74 1.23 0.31 0.71 41.76 0.24 12.68 0.36 7.86 2.38 94.41 0.12 31.05
75 0.68 0.11 0.63 7.19 0.07 12.27 0.11 0.90 1.07 56.56 0.07 11.92
Average 35.15 60.57 40.32 35.05 21.03
Note: SYS10: Qy/W = 17.6%; Dy = 0.4%; « = 8%; and CR|Pmax = 0.50.

on the equations provided in the study. The details of the ATC-58 (18), no residual shall occur when the system stays within its elastic
PD-estimation formula were not provided in the study. Moreover, the range. The second phase and the third phase were calibrated such
ATC-58 estimation of the RD demand is highlighted in this section that, for ductility ratios near 2, 4, and 6, the ratios of RD to PD were
because it could also be applied to the SDOF systems. Then, the three 0.15, 0.225, and 0.5, respectively. Ultimately, when the collapse
drift demand estimation approaches were compared with the simu- point was reached, the RD approached the PD. Such drift-to-inten-
lated results to show their accuracies. The ATC-58 prediction of resid- sity relation is illustrated in Fig. 11.
ual displacement is a piecewise estimation, which categorizes the re- However, the ATC-58 formula comes with shortcomings.
sidual displacement into three expressions according to different First, the ATC equations only predict RDs for a narrow range of
ductility ratios, which are shown in Eq. (18) structural systems considering limited nonlinearity, which was
8 shown to significantly influence the RDs. For example, the ATC-
>
> 0 umax  uy
< 58 equation contains no postyield stiffness (« ), and its influence
uend ¼ 0:3ðumax  uy Þ uy < umax < 4uy (18) on the RD estimation was not reflected except insofar as the post-
>
>
: yield stiffness affected the PD ratio. The equations were pre-
ðumax  3uy Þ umax  4uy
sented in this simple format owing to the lack of physical data to
The advantage of the ATC-58 approach is the dynamic cali- validate modeling of residual deformations and due to the com-
brated coefficient that links the RD with the PD. According to Eq. plexity of obtaining significant improvements in the RD estimates.

© ASCE 04018087-13 J. Bridge Eng.

J. Bridge Eng., 2018, 23(11): 04018087


Downloaded from ascelibrary.org by Universidad De Las Americas Puebla on 05/04/21. Copyright ASCE. For personal use only; all rights reserved.

Fig. 9. Predicted and calculated normalized PD and RD spectra of a bilinear bridge (ɛ ¼ 0:05, j ¼ 0:05, and Q=W ¼ 0:088) when subject to the
motion recorded at the Norwalk station during the 1987 Whittier earthquake: (a) Earthquake (EQ) motion and its M&P pulse representation (MP); (b)
Normalized peak displacements simulated using EQ motion and MP pulse, compared with estimated values; and (c) Normalized residual displace-
ments simulated using EQ motion and MP pulse, compared with estimated values.

Fig. 10. Predicted and calculated normalized PD and RD spectra of a bilinear bridge (ɛ ¼ 0:05, j ¼ 0:05, and Q=W ¼ 0:088) when subject to the
motion recorded at the TCU 141 station during the 1999 Chichi earthquake: (a) Earthquake (EQ) motion and its M&P pulse representation (MP); (b)
Normalized peak displacements simulated using EQ motion and MP pulse, compared with estimated values; and (c) Normalized residual displace-
ments simulated using EQ motion and MP pulse, compared with estimated values.

© ASCE 04018087-14 J. Bridge Eng.

J. Bridge Eng., 2018, 23(11): 04018087


In addition, similar to the JSCE formula, the ATC-58 estimation
µ=inf could not be functional without the estimated PD. ATC-58 has its
Ground Motion Intensity Residual own PD-estimation model for building structures, and its accuracy
was validated in the subsequent validation section. Results show that
the ATC-58 estimations of PDs slightly underestimated the PD
Transient demands. These deficiencies were eliminated by the proposed
method, which fully captured the bilinear behavior under the dimen-
sional analysis framework.
µ=4 Fig. 12 illustrates the normalized PDs (PPD) for the four bilinear
systems. Because PD was not estimated by JSCE, the pluses, dotted
lines, and squares represent the results from nonlinear analyses, the
Downloaded from ascelibrary.org by Universidad De Las Americas Puebla on 05/04/21. Copyright ASCE. For personal use only; all rights reserved.

proposed PD estimation, and the ATC-58 estimation, respectively.


µ=1 Fig. 13 shows the normalized RDs (PRD) for the four bilinear sys-
tems. The pluses, dotted lines, squares, and triangles represent the
results from nonlinear analyses, the proposed estimation, the ATC-
58 estimation, and the JSCE estimation, respectively. The results
Peak Inter-story Drift Ratio from the nonlinear analyses showed that the bilinear systems with
smaller postyield stiffness ratios (e.g., SYS9) triggered more nonli-
Fig. 11. Idealized model to estimate residual story drift from
nearity, increasing the RDs and their dispersions. For SYS10, as the
peak transient drift as a function of ground motion intensity
postyield stiffness ratio increased to 8%, the RDs and their disper-
(ATC-58).
sions became smaller. Comparing the proposed equation with the

Fig. 12. Validation of the PD estimations/through comparisons of the numerical simulations with the ATC-58 (FEMA 2012) estimations: (a) SYS7;
(b) SYS8; (c) SYS9; and (d) SYS10.

© ASCE 04018087-15 J. Bridge Eng.

J. Bridge Eng., 2018, 23(11): 04018087


Downloaded from ascelibrary.org by Universidad De Las Americas Puebla on 05/04/21. Copyright ASCE. For personal use only; all rights reserved.

Fig. 13. Validation of the RD estimations/through comparisons of the numerical simulations with the two codified methods of ATC-58 (FEMA
2012) and JSCE (2000): (a) SYS7; (b) SYS8; (c) SYS9; and (d) SYS10.

two codified approaches, it could be seen that the proposed formula preferred in this study. There were cases in which large PDs
and the JSCE formula provided better RD estimations, whereas the occurred while the structures finally rested at their original loca-
ATC-58 approach had larger errors. The detailed error percentages tions. In other words, for such cases, the RDi might be extremely
are listed in the subsequent section. In addition, when structure fre- small and negligible, creating huge errors following such error defi-
quency was smaller or closer to the pulse frequency (i.e., 0 < Pv ≤ nition. Consequently, a modified error equation was given by Eq.
1), the dispersion of the simulated data became relatively large, (20) to produce a more reasonable error estimation
N  
such that the errors increased in the neighborhood of the normalized
frequency, Pv = 1. 1X  RDie  RDi 
~e ¼ (20)
N i¼1  RDimax 

Error Estimation for the Proposed Model


The updated formula calculated the error of a RD prediction nor-
The accuracy of the proposed method was further validated by an malized by the maximum possible RD instead of the RD, where
error analysis. Conventionally, the overall normalized error is com- RDimax is the estimated maximum RD from the product of PDi and
puted by Eq. (19) CR|Pmax [defined by Eq. (17)]. The updated formula is applied to
N   evaluate the RD estimation errors in the study.
1X  RDe  RD 
i i
e¼ (19) Similar to the aforementioned studies, estimations of RDs for
N i¼1  RDi 
all four bilinear systems were performed. The results for the four
configurations of the bilinear systems are summarized in Tables
where N = number of cases computed; RDi = simulated RD for the 5–8. The footnote in each of Tables 5–8 shows the bilinear struc-
ith ground motion; and RDie = estimated result. However, the errors tural properties. The first column is the earthquake number. The
of the RD estimation calculated from such an equation were not second and third columns are the numerically simulated PDs and

© ASCE 04018087-16 J. Bridge Eng.

J. Bridge Eng., 2018, 23(11): 04018087


RDs for the bilinear systems. The fourth to seventh columns are Finally, it was recognized that the proposed RD estimations
the ATC estimations of PDs and RDs and their corresponding hinged largely on the dependence of residual displacement on the
errors. The eighth column is the JSCE estimation of RDs, and peak displacement (i.e., larger peak displacement likely resulted in
the errors are provided in the ninth column. The last four col- bigger residual displacement). Although the predictions of PDs and
umns are the proposed estimations of PDs and RDs and their cor- RDs in this study were easier and more accurate than the existing
responding errors. approaches, the method still could not fully account for the record-
In terms of the PD estimation, the ATC-58 formula based on the specific waveforms, the ground motion duration, more complex
elastic spectra was acceptable yet slightly underestimated over a hysteretic behavior, and the complete dynamic nature of the
wide range of periods, which also matched the conclusion made by responses. It is not intended to replace the nonlinear time-history
Chopra and Goel (2000). The ATC-58 estimations had maximum analyses. Instead, the proposed method can be used as an efficient
errors of 47% compared with the results from numerical analyses. assessment and predesign tool.
The proposed PD estimation was better than the ATC estimation
Downloaded from ascelibrary.org by Universidad De Las Americas Puebla on 05/04/21. Copyright ASCE. For personal use only; all rights reserved.

with a slight reduction of overall estimation errors.


For the RD estimation, the JSCE formula was more accurate for Acknowledgments
the cases with 0 or negative postyield stiffness ratios (i.e., SYS7 and
SYS9). The proposed RD formula outperformed the JSCE formula This research was partially funded by the National Science
for SYS8 and SYS10. The overall errors of the proposed RD estima- Foundation under Grant CMMI-0830391, Joy Pauschke, program
tion formula were between 20 and 30% except for SYS9, with a neg- manager. The idea of this paper originated during a study about
ative postyield stiffness ratio. However, the ATC-58 estimations self-centering devices. The authors also thank other scholars for
produced huge errors (e.g., 81% for SYS8), which made the estima- their help and intellectual exchange, in particular, Yuchuan Tang
tion meaningless for real applications. Therefore, it could be con- from Southeast University (China) for providing a collection of
cluded that the JSCE formula and the proposed formula provided near-fault ground motions and their pulse representations.
better RD estimations. However, the JSCE formula required PD as
an input. For this study, the PD values input into the JSCE formula
References
were the exact results obtained from the nonlinear time-history anal-
yses. In fact, the exact PD values were usually not available for the Alavi, B., and H. Krawinkler. 2004. “Behavior of moment-resisting frame
RD estimations, and using the estimated PD values as inputs might structures subjected to near-fault ground motions.” Earthquake Eng.
have created larger errors for the JSCE estimation. Comparatively, Struct. Dyn. 33 (6): 687–706. https://doi.org/10.1002/eqe.369.
the proposed PDs and RDs were estimated directly from structural ATC (Applied Technology Council). 1996. Seismic evaluation and retrofit
and ground motion characteristics without running nonlinear analy- of concrete building. ATC-40. Redwood City, CA: ATC.
ses. Finally, the estimated RD trends were reliable, which well fol- Baker, J. W. 2007. “Quantitative classification of near-fault ground motions
lowed the calculated RD demands. Of the four bilinear cases, the using wavelet analysis.” Bull. Seismol. Soc. Am. 97 (5): 1486–1501.
estimated RDs will increase if PNL increases and/or « decreases. https://doi.org/10.1785/0120060255.
Barenblatt, G. I. 1996. Scaling, self-similarity, and intermediate asymp-
totics. Cambridge, UK: Cambridge University Press.
Conclusions Bazzurro, P., C. A. Cornell, C. A. Menun, and M. Motahari. 2004.
“Guidelines for seismic assessment of damaged buildings.” In Proc.,
13th World Conf. on Earthquake Engineering, Paper No. 1708.
Estimating the residual deformation of bridge structures has
Vancouver, BC, Canada.
always been a challenging task. This paper aimed to develop a Bertero, V. V., S. A. Mahin, and R. A. Herrera. 1978. “Aseismic design
reliable demand model under the framework of dimensional anal- implications of near-fault San Fernando earthquake records.”
ysis for estimating the RD trend and the maximum RD of general- Earthquake Eng. Struct. Dyn. 6 (1): 31–42. https://doi.org/10.1002/eqe
ized nondegrading SDOF bridge models. Two regressive equa- .4290060105.
tions were proposed to directly estimate the PD and RD demands Bojórquez, E., and J. Ruiz-García. 2013. “Residual drift demands in
imposed by earthquake motions. The RD–PD correlation CR|P moment-resisting steel frames subjected to narrow-band earthquake
was also explored, and the maximum possible RD values were ground motions.” Earthquake Eng. Struct. Dyn. 42 (11): 1583–1598.
provided for a wide range of bilinear bridge-deck systems. https://doi.org/10.1002/eqe.2288.
Additionally, two existing estimations of RD demands [i.e., the Brown, A., and M. Saiidi. 2011. “Investigation of effect of near-fault
motions on substandard bridge structures.” Earthquake Eng. Eng. Vib.
ATC-58 approach and the JSCE (2000) approach] were also stud-
10 (1): 1–11. https://doi.org/10.1007/s11803-011-0042-8.
ied and compared with the proposed methods in this study. Chopra, A. K., and C. Chintanapakdee. 2004. “Inelastic deformation ratios
The simulation results showed that the normalized PD and RD for design and evaluation of structures: Single-degree-of-freedom bilin-
demands were strongly correlated to the postyield stiffness ratio, ear systems.” J. Struct. Eng. 130 (9): 1309–1319. https://doi.org/10
the structure-to-pulse frequency ratio, and the dimensionless nonli- .1061/(ASCE)0733-9445(2004)130:9(1309).
nearity index. Therefore, the estimations of PDs and RDs were pro- Chopra, A. K., and R. K. Goel. 2000. “Evaluation of NSP to estimate seis-
posed to be computed directly from structural and ground motion mic deformation: SDF systems.” J. Struct. Eng. 126 (4): 482–490.
characteristics, including the realistic features for the nonlinear sys- https://doi.org/10.1061/(ASCE)0733-9445(2000)126:4(482).
tems. Compared with the simulated results, the proposed model pro- Chopra, A. K., and R. K. Goel. 2002. “A modal pushover analysis procedure
vided RD estimations with acceptable errors for typical bridge sys- for estimating seismic demands for buildings.” Earthquake Eng. Struct.
tems with moderate hardening, and they were more accurate than Dyn. 31 (3): 561–582. https://doi.org/10.1002/eqe.144.
Christopoulos, C., S. Pampanin, and M. J. N. Priestley. 2003. “Performance-
the existing methods (e.g., ATC-58 and JSCE). However, larger
based seismic response of frame structures including residual deforma-
errors were experienced for bilinear systems with 0 or negative tions. Part I: Single-degree of freedom systems.” J. Earthquake Eng. 7
postyield stiffness ratios. Furthermore, for a given bilinear configu- (1): 97–118. https://doi.org/10.1080/13632460309350443.
ration, the errors became larger for the cases in which the structure Cuesta, I., and M. A. Aschheim. 2001. “Inelastic response spectra using
frequencies were smaller or near the neighborhood of the pulse conventional and pulse R-factors.” J. Struct. Eng. 127 (9): 1013–1020.
frequencies. https://doi.org/10.1061/(ASCE)0733-9445(2001)127:9(1013).

© ASCE 04018087-17 J. Bridge Eng.

J. Bridge Eng., 2018, 23(11): 04018087


Erduran, E., and S. K. Kunnath. 2010. “Enhanced displacement coefficient Eng. Mech. 130 (9): 1032–1044. https://doi.org/10.1061/(ASCE)
method for degrading multi-degree-of-freedom systems.” Earthquake 0733-9399(2004)130:9(1032).
Spectra 26 (2): 311–326. https://doi.org/10.1193/1.3381157. Makris, N., and S. Chang. 2000. “Effect of viscous, visco-plastic and fric-
Erochko, J., C. Christopoulos, R. Tremblay, and H. Choi. 2011. “Residual tion damping on the response of seismic isolated structures.”
drift response of SMRFs and BRB frames in steel buildings designed Earthquake Eng. Struct. Dyn. 29 (1): 85–107. https://doi.org/10.1002
according to ASCE 7-05.” J. Struct. Eng. 137 (5): 589–599. https://doi /(SICI)1096-9845(200001)29:1<85::AID-EQE902>3.0.CO;2-N.
.org/10.1061/(ASCE)ST.1943-541X.0000296. Makris, N., and T. Psychogios. 2006. “Dimensional response analysis
FEMA (Federal Emergency Management Agency). 1997a. NEHRP com- of yielding structures with first-mode dominated response.”
mentary on the guidelines for the seismic rehabilitation of buildings. Earthquake Eng. Struct. Dyn. 35 (10): 1203–1224. https://doi.org
Rep. No. FEMA-274. Washington, DC: FEMA. /10.1002/eqe.578.
FEMA (Federal Emergency Management Agency). 1997b. NEHRP guide- Malhotra, P. K. 1999. “Response of buildings to near-field pulse-like
lines for the seismic rehabilitation of buildings. Rep. No. FEMA-273. ground motions.” Earthquake Eng. Struct. Dyn. 28 (11): 1309–1326.
Washington, DC: FEMA.
Downloaded from ascelibrary.org by Universidad De Las Americas Puebla on 05/04/21. Copyright ASCE. For personal use only; all rights reserved.

https://doi.org/10.1002/(SICI)1096-9845(199911)28:11<1309::AID
FEMA (Federal Emergency Management Agency). 2000. Pre-standard -EQE868>3.0.CO;2-U.
and commentary for the seismic rehabilitation of buildings. Rep. No. Mavroeidis, G. P., G. Dong, and A. S. Papageorgiou. 2004. “Near-fault
FEMA-356. Washington, DC: FEMA. ground motions, and the response of elastic and inelastic single-degree-
FEMA (Federal Emergency Management Agency). 2005. Improvement of of-freedom (SDOF) systems.” Earthquake Eng. Struct. Dyn. 33 (9):
nonlinear static seismic analysis procedures. Rep. No. FEMA 440. 1023–1049. https://doi.org/10.1002/eqe.391.
Washington, DC: FEMA. Mavroeidis, G. P., and A. S. Papageorgiou. 2003. “A mathematical repre-
FEMA (Federal Emergency Management Agency). 2012. Seismic perform- sentation of near-fault ground motions.” Bull. Seismol. Soc. Am. 93 (3):
ance assessment of buildings. FEMA P-58. Washington, DC: FEMA. 1099–1131. https://doi.org/10.1785/0120020100.
Gillie, J. L., A. Rodriguez-Marek, and C. McDaniel. 2010. “Strength reduc- Motaref, S., M. Saiidi, and D. Sanders. 2014. “Shake table studies of
tion factors for near-fault forward directivity ground motions.” Eng. energy-dissipating segmental bridge columns.” J. Bridge Eng. 19 (2):
Struct. 32 (1): 273–285. https://doi.org/10.1016/j.engstruct.2009.09 186–199. https://doi.org/10.1061/(ASCE)BE.1943-5592.0000518.
.014. Pampanin, S., C. Christopoulos, and M. J. N. Priestley. 2003.
Hall, J. F., T. H. Heaton, M. W. Halling, and D. J. Wald. 1995. “Near-source “Performance-based seismic response of frame structures includ-
ground motion and its effects on flexible buildings.” Earthquake ing residual deformations. Part II: Multi-degree of freedom sys-
Spectra 11 (4): 569–605. https://doi.org/10.1193/1.1585828. tems.” J. Earthquake Eng. 7 (1): 119–147. https://doi.org/10.1080
He, W. L., and A. K. Agrawal. 2008. “Analytical model of ground motion
/13632460309350444.
pulses for the design and assessment of seismic protective systems.” J.
Riddell, R., and N. M. Newmark. 1979. Statistic analysis of the response of
Struct. Eng. 134 (7): 1177–1188. https://doi.org/10.1061/(ASCE)0733
nonlinear systems subjected to earthquakes. Structural Rep. Series 468.
-9445(2008)134:7(1177).
Champaign, IL: Civil Engineering, Univ. of Illinois.
JSCE (Japan Society of Civil Engineers). 2000. Earthquake resistant design
Ruiz-García, J., and J. D. Aguilar. 2015. “Aftershock seismic assess-
codes in Japan. Tokyo: Earthquake Engineering Committee.
ment taking into account postmainshock residual drifts.” Earthquake
Kawashima, K., G. A. MacRae, K. Hasegawa, T. Ikeuchi, and K. Oshima.
Eng. Struct. Dyn. 44 (9): 1391–1407. https://doi.org/10.1002/eqe
1992. “Ductility of steel bridge piers from dynamic loading tests.” In
.2523.
Stability and ductility of steel structures under cyclic loading, edited by
Ruiz-García, J., and C. Chora. 2015. “Evaluation of approximate methods
Y. Fukumoto and G. Lee. Boca Raton, FL: CRC Press.
to estimate residual drift demands in steel framed buildings.”
Kawashima, K., G. A. MacRae, J. Hoshikuma, and K. Nagaya. 1998.
Earthquake Eng. Struct. Dyn. 44 (15): 2837–2854. https://doi.org/10
“Residual displacement response spectrum.” J. Struct. Eng. 124 (5): 523–
530. https://doi.org/10.1061/(ASCE)0733-9445(1998)124:5(523). .1002/eqe.2611.
Kawashima, K., R. Zafra, T. Sasaki, K. Kajiwara, and M. Nakayama. 2011. Ruiz-García, J., and E. Miranda. 2005. Performance-based assessment of
“Effect of polypropylene fiber reinforced cement composite and steel existing structures accounting for residual displacements.” Technical
fiber reinforced concrete for enhancing the seismic performance of Rep. No. 153. Stanford, CA: John A. Blume Earthquake Engineering
bridge columns.” J. Earthquake Eng. 15 (8): 1194–1211. https://doi.org Center, Stanford Univ.
/10.1080/13632469.2011.569051. Ruiz-García, J., and E. Miranda. 2006a. “Direct estimation of residual dis-
Mackie, K. R., J. Wong, and B. Stojadinovic. 2011. “Bridge damage and placement from displacement spectral ordinates.” In Proc., 8NCEE,
loss scenarios calibrated by schematic design and cost estimation of Paper No. 1101. San Francisco, CA.
repairs.” Earthquake Spectra 27 (4): 1127–1145. https://doi.org/10 Ruiz-García, J., and E. Miranda. 2006b. “Evaluation of residual drift
.1193/1.3651362. demands in regular multi-storey frames for performance-based seismic
MacRae, G. A., C. Hodge, M. J. N. Priestley, and F. Seible. 1994. Route assessment.” Earthquake Eng. Struct. Dyn. 35 (13): 1609–1629. https://
5/405 separation. Shake table tests of as-built and retrofitted configura- doi.org/10.1002/eqe.593.
tion. SSRP Rep. No. 94/16, p. 246. San Diego: Structural Systems Ruiz-García, J., and E. Miranda. 2006c. “Residual displacement ratios for
Research Project, Dept. of Applied Mechanics and Engineering assessment of existing structures.” Earthquake Eng. Struct. Dyn. 35 (3):
Sciences, Univ. of California. 315–336. https://doi.org/10.1002/eqe.523.
MacRae, G. A., and K. Kawashima. 1997. “Post-earthquake residual dis- Ruiz-García, J., and E. Miranda. 2010. “Probabilistic estimation of residual
placements of bilinear oscillators.” Earthquake Eng. Struct. Dyn. 26 (7): drift demands for seismic assessment of multi-story framed buildings.”
701–719. https://doi.org/10.1002/(SICI)1096-9845(199707)26:7<701:: Eng. Struct. 32 (1): 11–20. https://doi.org/10.1016/j.engstruct.2009.08
AID-EQE671>3.3.CO;2-9. .010.
Mahin, S. A., and V. V. Bertero. 1981. “An evaluation of inelastic seismic Shu, Z., J. Zhang, and S. Nagarajaiah. 2017. “Dimensional analysis of
design spectra.” J. Struct. Div. 107 (9): 1777–1795. inelastic structures with negative stiffness and supplemental damping
Makris, N., and C. J. Black. 2004a. “Dimensional analysis of bilinear oscil- devices.” J. Struct. Eng. 143 (3): 04016184. https://doi.org/10.1061
lators under pulse-type excitations.” J. Eng. Mech. 130 (9): 1019–1031. /(ASCE)ST.1943-541X.0001658.
https://doi.org/10.1061/(ASCE)0733-9399(2004)130:9(1019). Somerville, P. G. 2003. “Magnitude scaling of the near fault rupture direc-
Makris, N., and C. J. Black. 2004b. “Dimensional analysis of rigid- tivity pulse.” Phys. Earth Planet. Inter. 137 (1–4): 201–212. https://doi
plastic and elasto-plastic structures under pulse-type excitations.” J. .org/10.1016/S0031-9201(03)00015-3.
Eng. Mech. 130 (9): 1006–1018. https://doi.org/10.1061/(ASCE) Tang, Y., and J. Zhang. 2011. “Response spectrum-oriented pulse iden-
0733-9399(2004)130:9(1006). tification and magnitude scaling of forward directivity pulses in
Makris, N., and C. J. Black. 2004c. “Evaluation of peak ground velocity near-fault ground motions.” Soil Dyn. Earthquake Eng. 29 (10):
as a ‘good’ intensity measure for near-source ground motions.” J. 1330–1346.

© ASCE 04018087-18 J. Bridge Eng.

J. Bridge Eng., 2018, 23(11): 04018087


Tatano, H., and S. Tsuchiya. 2008. “A framework for economic loss estima- Zhang, J., and Y. Tang. 2009. “Dimensional analysis of structures with
tion due to seismic transportation network disruption: A spatial comput- translating and rocking foundations under near-fault ground motions.”
able general equilibrium approach.” Nat. Hazards 44 (2): 253–256. Soil Dyn. Earthquake Eng. 29 (10): 1330–1346. https://doi.org/10.1016
https://doi.org/10.1007/s11069-007-9151-0. /j.soildyn.2009.04.002.
Tirasit, P., and K. Kawashima. 2007. “Seismic performance of square rein- Zhang, J., and W. Xi. 2012. “Optimal nonlinear damping for inelastic
forced concrete columns under combined cyclic flexural and torsional structures using dimensional analysis.” In Proc., Structures
loadings.” J. Earthquake Eng. 11 (3): 425–452. https://doi.org/10.1080 Congress, 20th Analysis and Computation Specialty Conf. Reston,
/13632460601031813. VA: ASCE.
Vidic, T., P. Fajfar, and M. Fischinger. 1994. “Consistent inelastic design Zhang, J., S. Xu, and Y. Tang. 2011. “Inelastic displacement demand of
spectra: Strength and displacement.” Earthquake Eng. Struct. Dyn. 23 bridge columns considering shear-flexure interaction.” Earthquake Eng.
(5): 507–521. https://doi.org/10.1002/eqe.4290230504. Struct. Dyn. 40 (7): 731–748. https://doi.org/10.1002/eqe.1056.
Downloaded from ascelibrary.org by Universidad De Las Americas Puebla on 05/04/21. Copyright ASCE. For personal use only; all rights reserved.

© ASCE 04018087-19 J. Bridge Eng.

J. Bridge Eng., 2018, 23(11): 04018087

You might also like