Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

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

net/publication/282588614

Equivalent Viscous Damping Model for Short-Period Reinforced Concrete


Bridges

Article  in  Journal of Bridge Engineering · September 2015


DOI: 10.1061/(ASCE)BE.1943-5592.0000803

CITATIONS READS
11 956

3 authors, including:

Easa Khan Mervyn Kowalsky


University of Nebraska at Lincoln North Carolina State University
5 PUBLICATIONS   44 CITATIONS    142 PUBLICATIONS   5,788 CITATIONS   

SEE PROFILE SEE PROFILE

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

Seismic Load Path Effects in Reinforced Concrete Bridge Columns and Wall Piers View project

Durability of the Grouted Shear Stud Connection at low temperatures View project

All content following this page was uploaded by Mervyn Kowalsky on 11 December 2015.

The user has requested enhancement of the downloaded file.


Equivalent Viscous Damping Model for Short-Period
Reinforced Concrete Bridges
Easa Khan1; Mervyn J. Kowalsky2; and James M. Nau3

Abstract: This paper investigates the effect of spectral shape (intensity and width of the constant acceleration region) and postyield stiffness
Downloaded from ascelibrary.org by North Carolina State University on 12/11/15. Copyright ASCE. For personal use only; all rights reserved.

ratio on equivalent viscous damping for short-period RC bridge columns (effective period < 1 s). The modified Takeda degrading stiffness
hysteretic model, with parameters appropriate to bridge columns (often termed thin Takeda in the literature), is used for analysis. Insight
regarding the importance of these parameters is provided, and a new equivalent viscous damping model is proposed that includes the effect of
spectral shape and postyield stiffness ratio, as well as effective period and ductility. The proposed damping model is compared with two exist-
ing models. The results indicate that significant improvement is achieved in predicting the peak displacement using the proposed damping
model when compared with existing models. DOI: 10.1061/(ASCE)BE.1943-5592.0000803. © 2015 American Society of Civil Engineers.
Author keywords: Equivalent viscous damping; Direct displacement-based seismic design; Short-period single-degree-of-freedom (SDOF)
systems.

Background are not typically satisfied under seismic excitation, such as consid-
eration that systems are subjected to harmonic excitation of a pre-
Over the last several decades, it has been agreed that inertial forces scribed frequency (real ground motions exhibit varied frequency
induced by ground motion are poorly correlated with damage content). Details of the various assumptions and their impact on
(Priestley 1993; Priestley et al. 2007; Calvi et al. 2013). To better equivalent viscous damping are discussed in Dwairi et al. (2007).
control inelastic lateral displacements, displacement-based design Jennings (1968) and Hadjian (1982) found that if the equal energy
procedures have been developed. One such method is the direct dis- principle is used, different methods of treating the period shift, as
placement-based design (DDBD) approach, outlined in its current explained later, lead to different equivalent viscous damping ratios.
form by Priestley et al. (2007). The DDBD approach uses the concept Furthermore, for a given period shift, variations in the hysteretic
of equivalent linearization to predict the response of inelastic sys- model also yield variations in the equivalent damping ratio. Some
tems with equivalent linear properties of effective stiffness and of the studies that are considered important with regard to period
equivalent viscous damping. Several studies have been conducted to shift and hysteretic models are highlighted in the following text.
verify the performance of this method for a variety of structural sys- Rosenblueth and Herrera (1964) proposed equivalent stiffness
tems in which the effective periods are greater than 1 s. However, it is at maximum displacement as the basis for defining the period shift,
the authors’ hypothesis that the current models used to characterize in which the equivalent viscous damping is estimated by equating
equivalent viscous damping overestimate the peak inelastic displace- the dissipated energy per cycle of the nonlinear system to that of
ment for short-period structural systems (those with an effective pe- the equivalent linear system for harmonic excitation. They defined
riod less than 1 s). This overestimation is owing to the underestima- period shift as the ratio of effective to initial period of the oscilla-
tion of equivalent viscous damping, which is the focus of this study. tor, which is equal to the square root of ductility for the case of the
The concept of equivalent viscous damping was first proposed by elastic–perfectly plastic hysteretic model; for greater detail, see
Dwairi et al. (2007). Note that this definition is used in the current
Jacobsen (1930) to approximate the steady forced-vibration response
study, which is in line with DDBD technique. Gulkan and Sozen
of linear single-degree-of-freedom (SDOF) damped systems. In
(1974) obtained the equivalent damping ratio from physical
this method, the stiffness of the equivalent system is set equal to that
experiments by equating the input energy of the inelastic system to
of the real system, and the damping ratio is estimated by equating
the energy of the equivalent linear system. Although they used the
the energy dissipated per cycle of the real damping force to that of
same equivalent stiffness definition as Rosenblueth and Herrera
the equivalent damping force. Later, Jacobsen (1960) extended this
(1964), Kowalsky et al. (1995) used secant stiffness at maximum
idea to yielding SDOF systems on the basis of an equivalent damp-
deformation and the Takeda hysteretic model to derive an equation
ing ratio. Jacobson’s study was based on several assumptions that
for the equivalent damping ratio using an unloading stiffness factor
of 0.5 and postyield-to-initial-stiffness ratio of 0.05. Miranda and
1
Ph.D. Student, North Carolina State Univ., Raleigh, NC 27606 (cor- Ruiz-GarcÚa (2002) compared these three methods and found that
responding author). E-mail: ekhan2@ncsu.edu for a given ground motion and level of inelastic behavior, the larg-
2
Professor of Structural Engineering, North Carolina State Univ., est response is predicted by Gulkan and Sozen (1974), who used
Raleigh, NC 27606. E-mail: kowalsky@ncsu.edu the smallest value of equivalent damping, followed by Kowalsky
3
Professor of Structural Engineering, North Carolina State Univ.,
et al. (1995) and then by Rosenblueth and Herrera (1964), who
Raleigh, NC 27606. E-mail: nau@ncsu.edu
Note. This manuscript was submitted on April 30, 2014; approved on predicted the lowest response with the largest value of equivalent
April 22, 2015; published online on September 25, 2015. Discussion pe- viscous damping. The work of Iwan (1980) and a more recent
riod open until February 25, 2016; separate discussions must be submitted study by Kwan and Billington (2003) were based on statistical
for individual papers. This paper is part of the Journal of Bridge investigations in which their proposed empirical equations for the
Engineering, © ASCE, ISSN 1084-0702. equivalent period and equivalent damping ratio were obtained by

© ASCE 04015047-1 J. Bridge Eng.

J. Bridge Eng., 04015047


minimizing the RMS of the error in the maximum displacements the equivalent viscous damping of short-period (less than 1 s)
from the actual inelastic and equivalent linear systems. Iwan SDOF bridge systems. For this purpose, eight different design spec-
(1980) considered 12 ground motions with a period range from tra are considered. The eight shapes consist of five with varying
0:2 to 4 s, whereas Kwan and Billington (2003) considered six hys- widths of the constant acceleration region (while keeping the same
teretic models and 20 ground motions in the period range from intensities) and three with different intensities (while keeping the
0:1 to 1:5 s. It was reported by Dwairi et al. (2007) that these two same width of the constant acceleration region). In addition, instead
methods yield similar estimates of equivalent damping as those of using individual ground motions, multiple suites of seven ground
from Gulkan and Sozen (1974) for RC members. motions are used. Only the modified Takeda degrading stiffness
It is important to note that Rosenblueth and Herrera (1964), hysteretic model, shown in Fig. 1(a), is used with a postyield stiff-
Gulkan and Sozen (1974), and Kowalsky et al. (1995) used a period ness ratio r and with unloading stiffness factor a ¼ 0:5 and reload-
shift based on equivalent stiffness of the SDOF system, whereas ing stiffness factor b ¼ 0:0. This hysteretic model is used because
Iwan (1980) and Kwan and Billington (2003) obtained the period it has been shown to be the most appropriate for bridge columns
Downloaded from ascelibrary.org by North Carolina State University on 12/11/15. Copyright ASCE. For personal use only; all rights reserved.

shift via minimizing the RMS of the error in displacement. forming plastic hinges (Priestley et al. 2007).
Although it is common to all methods that the equivalent damping
ratio is a function of ductility with modification for different hyste-
Equivalent Viscous Damping
retic models, none of the studies aim to consider the effect of spectral
shape and postyield stiffness ratio on the response of short-period
Equivalent viscous damping equations are often expressed as the
structural systems. It was observed by Mahin and Bertero (1981)
sum of two components, as shown in Eq. (1)
that moderate hardening with a postyield stiffness ratio of 0.05
can significantly increase displacements in structural systems with jeq ¼ jel þ jhyst (1)
short-to-medium-period ranges, but that it has a marginal influence
on the displacement response for long periods. where jel ¼ damping in the elastic range; and jhyst ¼ damping due
Blandon and Priestley (2005) proposed equivalent viscous to hysteretic energy dissipation. Damping in the elastic range is
damping as a function of hysteretic model type, ductility, and effec- used to represent the damping that is not captured by the hysteretic
tive period (which corresponds to the effective stiffness at maxi- model adopted for the analysis, such as energy dissipation due to
mum displacement) for SDOF systems. Their study used six earth- nonlinearity in the elastic response, soil–structure interaction, and
quake records, six different hysteretic models, and periods between interaction with nonstructural elements. Traditionally, this term is
0 and 4 s in increments of 0.5 s. They observed that the variation of taken as 5% of the critical damping for RC structures; a lower value
individual ground motions was significant, and further study was is often used for steel structures.
suggested to minimize the scatter of individual ground motions The simplest approach for estimating the hysteretic damping is
from proposed average damping values. Dwairi et al. (2007) pro- proposed by Jacobsen (1960), as given by Eq. (2)
posed a new equivalent damping model considering 100 real earth-
2 A1
quake records (FEMA 2005), four different hysteretic rules, and jhyst ¼ (2)
periods from 0 to 5 s at intervals of 0.1 s. Their model is also a func- p A2
tion of hysteretic characteristics, ductility, and effective period and
was developed using average nonlinear time-history analysis where A1 ¼ area of the hysteretic loop; and A2 ¼ area of the rigid–
results. Note that none of the studies consider the effect of postyield perfectly plastic rectangle that encloses the loop. Fig. 1(b) shows
stiffness ratio or spectral shape on equivalent viscous damping, the graphical representation of the parameters.
both of which are considered in this study. Grant et al. (2005) found that the elastic and hysteretic components
of damping should not be added directly because hysteretic damping
is tied to the secant stiffness, whereas elastic damping is tied to either
Research Objective initial or tangent stiffness. They proposed factors that are applied to
the elastic damping before combination with hysteretic damping,
The objective of the research described in this paper is to investigate depending on hysteretic model type and choice of initial versus tan-
the effect of design spectrum shape and postyield stiffness ratio on gent stiffness damping. The model is expressed as Eq. (3)

Fig. 1. Modified Takeda degrading stiffness hysteretic model: (a) hysteretic model parameters; (b) hysteretic area for damping calculation

© ASCE 04015047-2 J. Bridge Eng.

J. Bridge Eng., 04015047


k ¼ mλ (3)

where k ¼ multiplier on the elastic damping; m ¼ displacement


ductility; and λ depends on hysteretic and elastic damping models.
For the thin Takeda model used for bridge columns, λ ¼ 0:378
for tangent stiffness-proportional damping and 0.340 for initial
stiffness-proportional damping.
Equivalent viscous damping models proposed by Priestley et al.
(2007), Blandon and Priestley (2005), and Dwairi et al. (2007) are
shown for the thin Takeda hysteretic model in Eqs. (4)–(6b),
respectively. Eq. (4) assumes 5% tangent stiffness-proportional
damping and includes the adjustments proposed by Grant et al.
Downloaded from ascelibrary.org by North Carolina State University on 12/11/15. Copyright ASCE. For personal use only; all rights reserved.

(2005), whereas Eqs. (5), (6a), and (6b) are more generic and allow
the selection of the value and type of elastic damping model inde-
pendent of the hysteretic damping Fig. 2. Comparison of seven compatible accelerograms spectra with
design spectrum
 
m 1
jeq ¼ 0:05 þ 0:444 for r ¼ 0 (4)
pm
 " #
0:95 1 1
jeq ¼ kjel þ 1  0:5  0:1rm 1 þ
1:3p m ðTeff þ 0:85Þ4
(5)
 
m 1
jeq ¼ kj el þ Cst for r ¼ 0 (6a)
mp

Cst ¼ 0:5 þ 0:4ð1  Teff Þ for Teff  1


Cst ¼ 0:5 for Teff > 1 (6b)

where Teff ¼ effective period; and r ¼ postyield stiffness ratio of


the SDOF system.
Fig. 3. Average design spectra of suite of seven compatible
Methodology for Development of Equivalent Viscous accelerograms
Damping Model

The methodology implemented in this study is similar to the


approach outlined by Blandon and Priestley (2005) and Dwairi Step 2: Determination of Design Displacement Demand
et al. (2007) and is outlined in the following text. The process The effective period and ductility level ( m ) are selected, and the
is applied to several parametric studies, as will be discussed equivalent viscous damping ( jeq ) is estimated. For the first iteration,
throughout the remainder of the paper. the equivalent viscous damping is estimated using the damping Eq.
(4) proposed by Priestley et al. (2007). The design displacement
Step 1: Generation of Average Compatible demand (Dd ) is obtained by entering the average design spectra
Design Spectrum with the effective period (Teff ) and finding the corresponding spec-
A design spectrum and a suite of seven accelerograms are selected. tral displacement according to the curve that represents the equiva-
These accelerograms are made spectrum compatible with the design lent viscous damping as shown in Fig. 3. Linear interpolation is
spectrum at 5% damping using the computer program WAVGEN used to find the damping value if it lies between two damping values
developed by Mukherjee and Gupta (2002), in which the recorded of the average design spectra.
accelerogram is modified such that the temporal variations in its fre-
quency content are retained in the synthesized accelerogram. Once
the seven spectrum-compatible records are obtained, the displace- Step 3: Determination of SDOF System Properties
ment response spectra for each are calculated using SPECTRA by The value of inertial mass is chosen, which is kept constant for all
Carr (2007) for 20 different values of damping ranging from 5 to analysis. The effective stiffness (Keff ) of the equivalent system is
65%. The average values of the design spectra for these seven com- determined from the mass (m) and effective period using Eq. (7).
patible accelerograms at the corresponding 20 values of damping The required strength (Vb ) that corresponds to the design displace-
are then obtained, which represent average compatible design spec- ment is determined using Eq. (8), and the yield displacement (Dy ) is
tra. For example, Fig. 2 shows the individual as well as the average obtained with Eq. (9)
design spectra of seven compatible accelerograms along with the
design spectrum for 5% damping. Some scatter is observed for indi-
vidual accelerograms; however, the average design spectra closely  2
2p
match the design spectrum. The average design spectra for four of Keff ¼ m (7)
Teff
the damping values are shown in Fig. 3.

© ASCE 04015047-3 J. Bridge Eng.

J. Bridge Eng., 04015047


Vb ¼ Keff Dd (8) The computed new design displacement and known effective period
(from Step 2) are used to determine the new value of equivalent
damping. Steps 3–5 are repeated until convergence is achieved. It
Dd
Dy ¼ (9) was found that the number of iterations for convergence varies from
m 2 to 15 and increases as the period of the SDOF system reduces and
the ductility increases.
Note that the methodology can be applied to other hysteretic
Step 4: Characterization of SDOF System for Nonlinear model types by using appropriate parameters required for that hys-
Time-History Analysis teretic model in the nonlinear time-history analysis.
The yield strength (Fy ) of the SDOF system is obtained as a function
of Vb , postyield stiffness ratio (r), and ductility using Eq. (10), Investigation of Effect of Design Spectrum Shape
whereas the cracked-section stiffness (Kcr ) is obtained according to on Equivalent Viscous Damping
Downloaded from ascelibrary.org by North Carolina State University on 12/11/15. Copyright ASCE. For personal use only; all rights reserved.

Eq. (11). Note that a single value of r is used in the first parametric
study, and a range of values is used in the second. The iterative pro- For short-period structures (Teff < 1 s), it is the authors’ hypothesis
cess is carried out for each value of r selected that equivalent damping is a function of the shape of the design
spectrum in addition to the parameters investigated by Blandon and
Vb Priestley (2005) and Dwairi et al. (2007), i.e., effective period, duc-
Fy ¼ (10)
1 þ rð m  1Þ tility, and hysteretic model. To evaluate this hypothesis, a paramet-
ric study was undertaken, which included five ductility levels (1.5,
Fy 2, 3, 4, and 6), effective periods ranging from 0 to 1 s with an incre-
Kcr ¼ (11) ment of 0.2 s, and the thin Takeda hysteretic model with a postyield
Dy
stiffness ratio (r) of 0.025. This postyield stiffness ratio is represen-
tative of RC bridge structures. Three different design spectra are
The nonlinear time-history analysis is performed using the suite used to determine the effect of spectral intensity and width of the
of seven compatible accelerograms that are also used in Step 1 for constant acceleration region for one suite of seven compatible
generation of the average design spectrum. The peak results of each accelerograms.
analysis are averaged (ignoring the sign of the response) to deter-
mine the average nonlinear time-history displacement (DNTHA ).
Effect of Spectral Intensity on Equivalent
The nonlinear time-history analysis is accomplished using the
Viscous Damping
computer program RUAUMOKO 3D (Carr 2009) using Newmark’s
average acceleration integration scheme with b ¼ 0:25. The same To evaluate the effect of spectral intensity on damping, three dif-
integration scheme is used in the program SPECTRA to generate ferent design acceleration spectra are generated according to
displacement response spectra from the accelerograms. AASHTO (2009). These spectra have peak ground accelerations
(PGAs) of 0.31, 0.53, and 0.64g, labeled as DS1, DS2, and DS3,
Step 5: Convergence Criteria respectively, in Fig. 4(a). Each spectrum has the same width of the
constant acceleration region (Tsa ) but different intensity levels.
The error between design displacement (Dd ) and the average non- The width of the constant acceleration region is equal to the differ-
linear time-history displacement (DNTHA ), determined in Steps 3 ence of the periods Tf and Ti , as shown in Fig. 4(a). The corre-
and 4, respectively, is calculated as an absolute difference between sponding displacement spectra are shown in Fig. 4(b). Analysis is
the two. This value is checked against the tolerance level, which is carried out using the methodology explained earlier for the param-
taken as 1.5% of the design displacement. If the error is less than the eters discussed in the previous paragraph for the three design spec-
tolerance level, then the value of the equivalent damping has con- tra shown in Fig. 4. Results are presented in the following text.
verged; otherwise, a new value of design displacement is chosen, In Fig. 5, the equivalent viscous damping needed to equate the
which is taken as halfway between the previous and current values. design and nonlinear time-history displacement is plotted against

Fig. 4. Three design spectra with different intensity levels: (a) acceleration response spectra; (b) displacement response spectra

© ASCE 04015047-4 J. Bridge Eng.

J. Bridge Eng., 04015047


Downloaded from ascelibrary.org by North Carolina State University on 12/11/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. Comparison of the equivalent viscous damping obtained for three design spectra with different intensities for the short-period range of the
SDOF system: (a) Teff ¼ 0:2 s; (b) Teff ¼ 0:4 s; (c) Teff ¼ 0:6 s; (d) Teff ¼ 0:8 s; (e) Teff ¼ 1:0 s

ductility for the three design spectra of Fig. 4 for five effective peri- widths of the constant acceleration region as shown in Fig. 6(a).
ods. The results indicate that the equivalent viscous damping is the The PGA of 0.53g was selected near the midpoint of the range from
same for all three spectral intensity values that correspond to the 0.31 to 0.64g. All five spectra have the same intensities in the con-
same ductility level and effective period. For example, in Fig. 5(a) stant spectral acceleration region labeled as Tsa1 , Tsa2 , Tsa3 , Tsa4 ,
at a ductility of 4, all three spectral shapes give damping values of and Tsa5 for spectral shapes S1–S5, respectively. The five ranges
approximately 31%. This trend is the same for all period ranges of have been selected to represent the width of the constant accelera-
Figs. 5(a–e). Thus, it can be concluded that equivalent viscous tion region that is representative of Site Classes A–E as categorized
damping is independent of spectral intensity level across all short- by AASHTO (2009). The corresponding displacement spectra are
period ranges of structural systems considered. Furthermore, equiv- shown in Fig. 6(b). Analysis is executed using the methodology pre-
alent viscous damping increases with increasing ductility and with viously described using the three design spectra (S2–S4) shown in
decreasing effective period. Fig. 6; S1 and S5 are included in a larger study described later in
this paper. Results are presented in the following text.
Fig. 7 shows the equivalent viscous damping needed to equate
Effect of Width of Constant Acceleration Region on
the design and nonlinear time-history displacements considering
Equivalent Viscous Damping
the ductility and effective period for three design spectra, with vary-
To evaluate the effect of the width of the constant acceleration ing widths of the constant acceleration region. The results indicate
region, five design acceleration spectra are generated according to that the equivalent viscous damping increases as the width of the
AASHTO (2009) considering a constant PGA of 0.53g with varying constant acceleration region increases for all effective periods. For

© ASCE 04015047-5 J. Bridge Eng.

J. Bridge Eng., 04015047


Downloaded from ascelibrary.org by North Carolina State University on 12/11/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 6. Five design spectra with different widths of the constant acceleration region: (a) acceleration response spectra; (b) displacement response
spectra

Fig. 7. Comparison of the equivalent viscous damping obtained for five design spectra with different widths of the constant acceleration region for
the short-period range of the SDOF system: (a) Teff ¼ 0:2 s; (b) Teff ¼ 0:4 s; (c) Teff ¼ 0:6 s; (d) Teff ¼ 0:8 s; (e) Teff ¼ 1:0 s

© ASCE 04015047-6 J. Bridge Eng.

J. Bridge Eng., 04015047


example, in Fig. 7(a), for a ductility of 4, the damping correspond- The iterative procedure discussed in the methodology section is
ing to S2, S3, and S4 is 24, 27, and 31%, respectively. This trend is carried out to determine the equivalent viscous damping that equa-
generally the same for all period ranges of Figs. 7(a–e). Thus, it can tes the design displacement with the average nonlinear time-history
be concluded that equivalent viscous damping is dependent on the displacement. To understand the effect of different ground motions,
width of the constant acceleration region of the design spectra with effective period, and postyield stiffness ratios, the results of the
equivalent damping increasing as the width of the constant accelera- parametric study are presented in Fig. 8 only for spectral shape S2,
tion region increases. where equivalent damping is plotted against ductility for different
values of effective periods. Each line in the plot represents the aver-
age value of damping obtained from 49 ground motions divided
Estimation of Equivalent Viscous Damping for a Large into seven suites, i.e., each line in Fig. 8 represents the average of
Parametric Study seven data points, and the vertical error bar represents one standard
deviation (SD) associated with these data points. The error bar is
Downloaded from ascelibrary.org by North Carolina State University on 12/11/15. Copyright ASCE. For personal use only; all rights reserved.

Following the results of the initial parametric study, only the width shown above the average line and not below for clarity. To further
of the constant acceleration region is kept as a spectral shape vari- elaborate, consider Fig. 8(a) with an effective period of 0.2 s and a
able in the larger parametric study described here. The details of ductility of 4. The average damping value of the seven data points is
each parameter considered are presented in the following list: approximately 31% with a SD of approximately 3%. It should be
• Five design spectrum shapes as shown in Fig. 6 are used to rep- noted that the SD reduces as the effective period increases for the
resent the different widths of the constant acceleration region. same ductility level, i.e., in Fig. 8(a) for a ductility of 6, the SD for
• A thin Takeda hysteretic model with unloading (a ) and reload- effective periods of 0.2 and 1 s is 4 and 2%, respectively. Also, for
ing ( b ) stiffness factors of 0.5 and 0.0, respectively, is used the same effective period, the SD increases with increases in ductil-
with different postyield stiffness ratios of 0.0, 0.025, 0.05, and ity. For example, in Fig. 8(a), for a ductility of 1.5 and 6, the SD is 2
0.075. and 4%, respectively, corresponding to an effective period of 0.2 s.
• Five ductility levels of 1.5, 2, 3, 4, and 6 are used. Furthermore, with increases in postyield stiffness ratio, the average
• The effective period is measured from 0 to 1 s, at a step size of damping value reduces significantly. For example, at a ductility of 6
0.2 s. and effective period of 0.2 s, the damping is 37 and 23% for r ¼ 0
• Forty-nine earthquake accelerograms were used, grouped into [Fig. 8(a)] and r ¼ 0:075 [Fig. 8(d)], respectively. Thus, it is clear
seven suites according to site class. The 49 ground motions that the effect of postyield stiffness ratio is significant and should be
cover Site Classes B–E, with one suite for Soil Class B and incorporated in the damping model for short-period SDOF systems.
two suites for each Soil Class C, D, and E. Each ground motion Similar trends are observed for spectral shapes S1, S3, S4, and S5.
was made spectrum-compatible with the five design spectral Markis and Kampas (2013) have also noted that the second slope
shapes of Fig. 6. All ground motions were recorded in seismic stiffness of the bilinear system may be used as the characteristic stiff-
events with moment magnitudes ranging from 6.0 to 7.5. ness for SDOF systems, illustrating the potential importance of r.

Fig. 8. Effect of ground motion, effective period, and postyield stiffness ratio: (a) r ¼ 0:0; (b) r ¼ 0:025; (c) r ¼ 0:5; (d) r ¼ 0:075

© ASCE 04015047-7 J. Bridge Eng.

J. Bridge Eng., 04015047


The data are also generated for design spectra S1, S3, S4, and S5 between S1 and S5 for a ductility of 4 and effective period of 0.2 s
of Fig. 6 for the same parameters as for S2. To see the effect of is 21% (50–29), whereas the SDs corresponding to this point for S1
design spectrum shape, the results are presented only for r ¼ 0 in and S5 are 3 and 6%, respectively. Note that the same trends are
Fig. 9. Note that the difference between Figs. 7 and 9 is that analysis observed for r values of 0.025, 0.05, and 0.075. Thus, to accurately
for Fig. 7 is performed for only one suite of ground motions contain- predict damping for short-period SDOF systems, the width of the
ing seven accelerograms, whereas in Fig. 9, it is performed for 49 constant acceleration region will be included in the proposed damp-
ground motions comprising seven suites. ing model.
Fig. 9 represents average damping corresponding to different
ductilities for the five design spectra shapes, where each figure cor-
responds to specific effective periods with r ¼ 0. It is clear from
Figs. 9(a–e) that as the width of the constant acceleration region of Proposed Equivalent Viscous Damping Model
the spectra increases, the equivalent viscous damping increases.
Downloaded from ascelibrary.org by North Carolina State University on 12/11/15. Copyright ASCE. For personal use only; all rights reserved.

For example, for an effective period of 0.2 s and ductility of 4, the The proposed damping model [Eqs. (12a)–(12f)], which applies for
average damping values for S1–S5 are 29, 31, 35, 42, and 50%, effective periods of less than 1 s, is obtained by applying the method
respectively. It is important to note that the effect of the width of of least squares to the analysis results. This model accounts for the
the constant acceleration region is significant because it affects width of the constant acceleration region and postyield stiffness ra-
equivalent damping more than the scatter that arises from individ- tio through parameters Tsa (see Fig. 6) and r, respectively, where as
ual ground motions. For example, the difference of damping the existing models (Blandon and Priestley 2005; Dwairi et al.

Fig. 9. Equivalent viscous damping grouped by spectral shapes S1, S2, S3, S4, and S5: (a) Teff ¼ 0:2 s, r ¼ 0:0; (b) Teff ¼ 0:4 s, r ¼ 0:0;
(c) Teff ¼ 0:6 s, r ¼ 0:0; (d) Teff ¼ 0:8 s, r ¼ 0:0; (e) Teff ¼ 1:0 s, r ¼ 0:0

© ASCE 04015047-8 J. Bridge Eng.

J. Bridge Eng., 04015047


2007) do not. Additionally, the displacement ductility ( m ) and existing models of Blandon and Priestley (2005) and Dwairi et al.
effective period (Teff ) are also included in the proposed model (2007), given by Eqs. (5) and (6a)–(6b), respectively. Because
Dwairi’s equation is developed only for r ¼ 0, both existing
A models along with the proposed model are analyzed for r ¼ 0.
jeq ¼ 0:05 þ ð m  1ÞB (12a)
mp All 49 ground motions are used for all short-period ranges of
structural systems and spectral shapes S2 and S4. Additionally,
A ¼ 0:14 þ 0:43ð1  Teff þ 1:68Tsa  2:5rÞ (12b) two spectral shapes (S0 and S6) with the same PGA as S1–S5
but different Tsa values (Tsa01 ¼ 0:18 s and Tsa6 ¼ 1:0 s, respec-
B ¼ 1:13 þ 0:26Tsa  3:22r (12c) tively) are considered. These values are outside the range used
for development of the model and hence are a good test of the
Tsa ¼ Tsa if 0:34  Tsa  0:65 (12d) applicability of the model across all possible Tsa values as defined
by the AASHTO guide specifications. Two ductility levels, 3 and
Downloaded from ascelibrary.org by North Carolina State University on 12/11/15. Copyright ASCE. For personal use only; all rights reserved.

6, which represent medium and high ductility levels, are consid-


Tsa ¼ 0:34  0:30ð0:34  Tsa Þ if Tsa < 0:34 (12e) ered. To conduct this analysis, a two-step process is performed as
follows.
Tsa ¼ 0:65 þ 0:15ðTsa  0:65Þ if Tsa > 0:65 (12f )
Step 1: Estimation of Displacement for Equivalent
The model in Eqs. (12a)–(12f) has been validated for values of Linear System (Dd )
Tsa between 0.27 and 0.8 s. This range of Tsa covers a large portion
The equivalent viscous damping for the models of Blandon and
of the possible values across the United States as defined in the
Priestley (2005) and Dwairi et al. (2007), as well as the proposed
AASHTO guide specifications (Table 1). This table represents re- model, is determined for a ductility of 3 and 6. Then, the effective
gional values of Tsa as a function of site classes. period and equivalent damping are entered into the average dis-
In Fig. 10, the label S1_Avg represents the average damping placement spectra to determine the displacement of the equivalent
obtained from the data, and S1_PM represents the corresponding linear system (Dd ) in the same way as shown in Fig. 3 of the meth-
damping estimated from the proposed damping model for spectral odology discussed previously.
shape S1. The same notation applies for S2–S5. It is clear from
Figs. 10(a–e) that the best fit is obtained for S2; however, a perfect
match was not possible for all spectral shapes covering the full Step 2: Determination of Average Nonlinear
range of r factors. Therefore, for spectral shapes S4 and S5, the Time-History Displacement (DNTHA )
damping model underestimates damping (S4_PM and S5_PM) for
Steps 3 and 4 of the methodology are carried out to determine
an effective period of 0.2 s, and for spectral shape S1, it slightly
(DNTHA ) as explained previously. The ratio of the displacement
overestimates damping (S1_PM) at effective periods of 0.4 and
for the equivalent linear system (Dd ) to the average nonlinear
0.6 s when compared with the data. This underestimation of damp-
time-history displacement (DNTHA ) is calculated for all three mod-
ing can cause overestimation of displacement demand, but such a els, and the results are presented in Fig. 11, in which each bar
variation at the lower period range (from 0 to 0.2 s) is preferable to represents the average displacement ratio for 49 ground motions.
larger scatter over a wider period range (from 0.2 to 1 s). Each error bar represents one SD from the average value. In Figs.
Furthermore, such a combination of low effective period and high 11(a–h), three ratios of displacement obtained from the three
ductility is rarely encountered in practice. It should also be noted models are shown for spectral shapes S0, S2, S4, and S6. A dis-
that the proposed model covers a broader range of postyield stiff- placement ratio of 1 means that the damping model accurately
ness ratios (from 0 to 0.075) and design spectral shapes than other estimates the damping, whereas a value less than 1 means overes-
models that are based only on one value of r with no consideration timation of damping, and greater than 1 means underestimation
of spectral shape. of damping. The proposed model is more comprehensive and
hence more accurate. The additional variables in the model do
not unduly affect computational effort.

Comparison of Existing and Proposed


Damping Models Conclusion

Analyses were conducted to determine the accuracy of the pro- The effect of different parameters on the equivalent damping ratio
posed damping model, represented by Eqs. (12a)–(12f), and the for short-period SDOF systems was investigated. It was found that
the equivalent damping ratio is a function of the width of the con-
Table 1. Ranges of Tsa Values for Several Regions of the United States stant acceleration region of the response spectrum and is signifi-
(Data from AASHTO 2009) cantly affected by the postyield stiffness ratio, in addition to the pre-
viously known parameters, including ductility, effective period, and
Tsa (s) range for site class ground-motion. In previous research, equivalent viscous damping
Region A and B C D E models were proposed on the basis of average results for a suite of
ground motions considering one value of the postyield stiffness ra-
West Coast 0.31–0.35 0.4–0.46 0.46–0.53 0.82–0.94
tio. However, it was found from this study that the postyield stiff-
Mountain West 0.25–0.27 0.34–0.38 0.39–0.43 0.43–0.68
ness has a significant effect on response, even more so than that
Central 0.20–0.24 0.31–0.31 0.35–0.36 0.57–0.63
obtained from ground-motion variation. Therefore, the effect of
Eastern 0.18–0.21 0.26–0.29 0.31–0.32 0.32–0.6
these parameters was proposed in a new design expression for
Alaska 0.24–0.37 0.33–0.48 0.37–0.56 0.39–0.99
equivalent viscous damping for short-period SDOF bridge systems.

© ASCE 04015047-9 J. Bridge Eng.

J. Bridge Eng., 04015047


Downloaded from ascelibrary.org by North Carolina State University on 12/11/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 10. Proposed damping model fitted with data: (a) Teff ¼ 0:2 s, r ¼ 0:0; (b) Teff ¼ 0:4 s, r ¼ 0:0; (c) Teff ¼ 0:6 s, r ¼ 0:0; (d) Teff ¼ 0:8 s,
r ¼ 0:0; (e) Teff ¼ 1:0 s, r ¼ 0:0

An extensive study was conducted using 49 ground motions large; and (2) a period of 0.4 s, when ductility is large, and the
and three different design spectral shapes for the thin Takeda hys- width of constant acceleration region is low. Furthermore, the SD
teretic model with various levels of ductility. Four levels of post- of equivalent damping is also reduced in comparison to the other
yield stiffness ratios were considered. Five oscillators with effec- two existing models. These improvements were achieved in the
tive periods from 0 to 1 s were considered with increments of proposed damping model because of the incorporation of depend-
0.2 s. New damping relations were obtained using an iterative pro- ency of design spectrum shape and postyield stiffness ratio, in
cedure that was a function of ductility, effective period, postyield addition to a tighter tolerance level for convergence. Considerable
stiffness ratio of the hysteretic model, and width of the constant scatter is observed in the damping ratios obtained for individual
spectral acceleration region of the design spectrum. The proposed ground motions, and the proposed damping model may be
damping model was also compared against two existing models improved further if additional ground-motion parameters are
for short-period SDOF systems. It was found that existing models incorporated. Of course, such modifications complicate the model
underestimate equivalent damping and overestimate the displace- and ultimately are difficult to include in design, where known
ment, whereas the proposed model was capable of estimating the ground-motion parameters are few.
equivalent damping more accurately. The largest errors for the It is important to note that the proposed damping model is
proposed model occur for (1) the short-period range of 0.2 s, when derived and verified using the AASHTO design spectral shape.
both the ductility and width of the constant acceleration region are Other spectral shapes could benefit from this model, although care

© ASCE 04015047-10 J. Bridge Eng.

J. Bridge Eng., 04015047


Downloaded from ascelibrary.org by North Carolina State University on 12/11/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 11. Comparison of the displacement ratio for three damping models: (a) m ¼ 3 and S0; (b) m ¼ 6 and S0; (c) m ¼ 3 and S2; (d) m ¼ 6 and
S2; (e) m ¼ 3 and S4; (f) m ¼ 6 and S4; (g) m ¼ 3 and S6; (h) m ¼ 6 and S6

should be used in doing so. It is the authors’ opinion that this model References
might give better results for other design spectra in comparison to
damping models that do not account for spectral shape. AASHTO. (2009). AASHTO guide specifications for LRFD seismic bridge
design, Washington, DC.
Blandon, C. A., and Priestley, M. J. N. (2005). “Equivalent viscous damping
equations for direct displacement based design.” J. Earthquake Eng.,
Acknowledgments
9(2), 257–278.
Calvi, G. M., Priestley, M. J. N., and Kowalsky, M. J. (2013).
The authors wish to thank North Carolina State University for its “Displacement-based seismic design of bridges.” Struct. Eng. Int.,
support during the conduct of this research. In addition, the 22(2), 112–121.
comments by the reviewers were most helpful in improving the Carr, A. J. (2007). RUAUMOKO user’s manual, Univ. of Canterbury,
quality of the manuscript. Christchurch, New Zealand.

© ASCE 04015047-11 J. Bridge Eng.

J. Bridge Eng., 04015047


Carr, A. J. (2009). RUAUMOKO 3D user’s manual, Univ. of Canterbury, Kowalsky, M. J., Priestley, M. J. N., and MacRae, G. A. (1995).
Christchurch, New Zealand. “Displacement-based design of RC bridge columns in seismic regions.”
Dwairi, H., Kowalsky, M. J., and Nau, J. M. (2007). “Equivalent damping Earthquake Eng. Struct. Dyn., 24(12), 1623–1643.
in support of direct displacement-based design.” J. Earthquake Eng., 11 Kwan, W. -P., and Billington, S. L. (2003). “Influence of hysteretic behavior
(4), 512–530. on equivalent period and damping of structural systems.” J. Struct. Eng.,
FEMA. (2005). “Improvement of nonlinear static seismic analysis proce- 10.1061/(ASCE)0733-9445(2003)129:5(576), 576–585.
dures.” FEMA 440, ATC-55 project, Redwood City, CA. Mahin, S., and Bertero, V. V. (1981). “Evaluation of inelastic seismic
Grant, D. N., Blandon, C. A., and Priestley, M. J. N. (2005). “Modeling design spectra.” J. Struct. Div., 107(9), 1777–1795.
inelastic response in direct displacement-based design.” Rep. ROSE Markis, N., and Kampas, G. (2013). “The engineering merit of the ‘effective
2005/03, IUSS Press, Pavia, Italy. period’ of bilinear isolation systems.” Earthquake Struct., 4(4), 397–428.
Gulkan, P., and Sozen, M. (1974). “Inelastic response of reinforced concrete Miranda, E., and Ruiz-GarcÚa J. (2002). “Evaluation of approximate meth-
structures to earthquakes motions.” ACI J., 71(12), 604–610. ods to estimate maximum inelastic displacement demands.” Earthquake
Hadjian, A. H. (1982). “A re-evaluation of equivalent linear models Eng. Struct. Dyn., 31(3), 539–560.
for simple yielding systems.” Earthquake Eng. Struct. Dyn., 10(6), Mukherjee, S., and Gupta, V. K. (2002). “Wavelet-based generation of spectrum-
Downloaded from ascelibrary.org by North Carolina State University on 12/11/15. Copyright ASCE. For personal use only; all rights reserved.

759–767. compatible time-histories.” Soil Dyn. Earthquake Eng., 22(9), 799–804.


Iwan, W. D. (1980). “Estimating inelastic response spectra from elastic Priestley, M. J. N. (1993). “Myths and fallacies in earthquake engineering
spectra.” Earthquake Eng. Struct. Dyn., 8(4), 375–388. conflicts between design and reality.” Proc., Tom Paulay Symp.: Recent
Jacobsen, L. S. (1930). “Steady forced vibrations as influenced by damp- developments in lateral force transfer in buildings, Univ. of California,
ing.” Trans. ASME, 52(15), 169–181. San Diego, 229–252.
Jacobsen, L. S. (1960). “Damping in composite structures.” Proc., 2nd Priestley, M. J. N., Calvi, G. M., and Kowalsky, M. J. (2007). Direct
World Conf. on Earthquake Engineering, 2, 1029–1044. displacement-based design of structures, IUSS Press, Pavia, Italy.
Jennings, P. C. (1968). “Equivalent damping for yielding structures.” Rosenblueth, E., and Herrera, I. (1964). “On a kind of hysteretic damping.”
J. Engrg. Mech. Div., 94(1), 103–116. J. Engrg. Mech. Div., 90(4), 37–48.

© ASCE 04015047-12 J. Bridge Eng.

View publication stats J. Bridge Eng., 04015047

You might also like