Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Evaluation of Damping Ratios for the Seismic

Analysis of Tall Buildings


Cristian Cruz 1 and Eduardo Miranda, Ph.D. 2

Abstract: This paper examines damping ratios in tall buildings in California inferred from motions recorded in instrumented buildings
Downloaded from ascelibrary.org by "California,Univ Of Irvine" on 04/30/18. Copyright ASCE. For personal use only; all rights reserved.

during earthquake motions. A total of 41 measured seismic responses from 14 buildings with 20 or more stories are analyzed using a para-
metric least-squares system identification technique in the time domain. The reliability of the inferred damping ratios is first examined based
on the sensitivity of the structural response to variations in modal damping ratios. Using only inferred damping ratios deemed reliable, it is
found that the damping ratios of the fundamental mode decrease with increasing building height, reaching values lower than 2.5%—the
damping value that is commonly recommended for seismic design in the United States for buildings taller than 150 m. The variation of
damping of higher modes with frequency is then examined by studying the results of those buildings where damping ratios of at least the first
three modes are reliably inferred. It is shown that the variation of damping with frequency follows an approximately linear trend. No evidence
was found to suggest that a mass-proportional model could be appropriate. The amplitude dependency of damping is then analyzed for
buildings having four or more recorded earthquakes, finding no statistical correlation between the peak roof drift ratio and the identified
fundamental damping ratio. DOI: 10.1061/(ASCE)ST.1943-541X.0001628. © 2016 American Society of Civil Engineers.

Introduction no new strong-motion data was added. The work of Satake et al.
(2003) collected data from various sources in Japan to assemble
Estimation of the response of tall buildings to seismic loading a new database of damping values coming from 205 buildings.
requires knowledge of the damping in the building. Unlike mass The majority of the data, however, corresponds to low-amplitude
and stiffness matrices, which can be determined from the geometry motions: only 18 of the 205 buildings were subjected to earthquakes,
of the structure, the only reliable way to assess the damping in a of which only 4 were higher than 150 m. Fritz et al. (2009) collected
building is through measurements of the structural response when 1,572 damping ratio measurements from full-scale tests on 972 dif-
subjected to dynamic loads. Several studies have gathered data on ferent buildings. The database was compiled from smaller datasets
damping ratios from the recorded response of buildings. Using and other public sources, and had a fair share of strong motion data:
simplified methods, Blume (1970) estimated 38 values of damping 260 data points came from 93 buildings subjected to earthquakes,
ratio for the fundamental mode of 5 buildings, all in the range of but only 10 of those buildings were analyzed using parametric
20–30 stories, subjected to small tremors caused by underground system identification schemes. Furthermore, the vast majority of
nuclear detonations. In a later publication, Blume studied 230 damp- high-quality damping data corresponds to low-amplitude sources.
ing ratios from 74 buildings subjected to low-amplitude motions There is a general agreement that damping is expected to increase
(Scholl 1975). The resulting histograms for damping from these two with amplitude at small deformation levels (Tamura and Yoshida
publications represent the first attempt to describe uncertainty in 2008; Spence and Kareem 2013), and therefore, it is not clear to what
damping. Haviland (1976) analyzed a set of 244 damping values extent damping data from low-amplitude motions is applicable to
gathered from 139 buildings and constructed probability distribu- seismic analysis. Jeary (1986) showed that damping ratio increased
tions for the damping ratio of the fundamental mode. The damping with increasing amplitude in 17 buildings subjected to forced vibra-
estimates were collected through a literature survey of 39 references, tions at very low amplitudes (the peak roof displacement in these
where the vast majority of the data came from low-amplitude sour- tests was less than 3 mm), proposing a piecewise linear model for
ces such as nuclear, forced, micro-tremors, wind, and ambient vi- amplitude dependency. Li et al. (2000) confirmed this trend in a
brations. Davenport and Hill-Carroll (1986) developed a database series of observations of a 120-m-tall building subjected to ampli-
of 165 buildings subjected to full-scale damping tests, including all tudes up to a peak roof drift ratio of about 0.01%. Tamura et al.
the data from Haviland (1976). Unfortunately, the database did not (2000) corrected the piecewise linear model to take into account char-
add any new data on damping inferred from high-amplitude motion. acteristics observed in the Japanese damping database. An increase in
Lagomarsino (1993) extended this database to a total of 185 build- damping with amplitude has also been observed when comparing
ings, adding data points collected from other investigations. Again, values inferred from ambient vibrations with those obtained from
seismic motions. Celebi et al. (1993) and Celebi (1996) studied
1
Ph.D. Candidate, Dept. of Civil and Environmental Engineering, the damping ratios of five buildings, first under ambient vibrations
Stanford Univ., Stanford, CA 94305 (corresponding author). E-mail: and then subjected to the 1989 Loma Prieta earthquake, observing in
ccruzd@stanford.edu all cases that the damping ratios inferred from strong-motion data for
2
Associate Professor, Dept. of Civil and Environmental Engineering, the first mode were significantly larger than those obtained from am-
Stanford Univ., Stanford, CA 94305. E-mail: emiranda@stanford.edu
bient-vibration data. All these investigations lead to the conclusion
Note. This manuscript was submitted on November 12, 2015; approved
on June 18, 2016; published online on July 25, 2016. Discussion period that damping recommendations for earthquake-resistant design
open until December 25, 2016; separate discussions must be submitted should come from the analysis of strong-motion data.
for individual papers. This paper is part of the Journal of Structural En- There are few studies on damping values inferred from the
gineering, © ASCE, ISSN 0733-9445. seismic response of buildings. Using simplified spectral methods,

© ASCE 04016144-1 J. Struct. Eng.

J. Struct. Eng., 2017, 143(1): 04016144


Tanaka et al. (1969) analyzed the records of 17 buildings subjected subjected to multiple ground motions, providing an opportunity to
to the 1968 Saitama (Japan) earthquake, whereas Hart and make numerous observations of the damping ratios in these build-
Vasudevan (1975) studied 12 buildings subjected to the 1971 ings when subjected to earthquake ground motions. This work pro-
San Fernando earthquake. These two studies provide the only vides new data, analyzing the damping ratios inferred from motions
strong-motion data points in the database by Haviland (1976) and recorded in 14 buildings with 20 or more stories when subjected
its later extensions. It is important to note that these early studies, as to earthquake ground motions. The main objective is to determine
well as some low-amplitude investigations, used spectral methods damping ratios that should be employed in the seismic analysis of
such as the half-power bandwidth method to infer damping ratios. tall buildings when using linear elastic modal analysis and a fixed-
It has been well documented that this method is not reliable and base model, in order to best reproduce the measured responses.
produces biased results, overestimating damping due to several is- These assumptions are typically made in seismic design, where
sues including sampling resolution, zero padding, smoothing, mo- soil–structure interaction is rarely taken into account. Damping
dal interference, length of the signal, and stationarity problems ratios were inferred from recorded motions using a parametric sys-
(Stagner and Hart 1971; Hill-Carroll 1985; Jeary 1986). Therefore, tem identification technique based on the minimization of the error
Downloaded from ascelibrary.org by "California,Univ Of Irvine" on 04/30/18. Copyright ASCE. For personal use only; all rights reserved.

attention should be paid to damping ratios of buildings subjected to between the recorded and predicted responses in the time domain.
earthquake motions and inferred employing modern system iden- This technique is further investigated to establish confidence inter-
tification techniques. Beck and Jennings (1980) and McVerry vals for the damping estimates, as well as to assess the reliability
(1979) developed a system identification technique based on the of inferred damping ratios. Using only damping ratios deemed
least-squares minimization between the measured seismic response reliable, the relationship between the damping ratio of the funda-
of a building and that predicted by a parametric model. The former mental mode and the building height is evaluated and compared to
formulated the problem in the time domain and applied it to a published recommendations. In the same spirit, the relationship be-
42-story building in Los Angeles, and the latter did it in the fre- tween the damping ratio of higher modes and their corresponding
quency domain, analyzing 10 buildings subjected to the San modal frequencies is studied and compared to the commonly used
Fernando earthquake of 1971, of which three buildings had 20 sto- Rayleigh damping. Finally, the relationship between damping and
ries or more. Li and Mau (1991) and Mau and Aruna (1994) used a amplitude of structural response is examined.
similar formulation in the time domain to analyze three buildings
of 6, 13, and 15 stories. Li and Mau (1997) analyzed 21 buildings
subjected to the 1987 Whittier and 1989 Loma Prieta earthquakes, Buildings Analyzed
with only three buildings having 20 stories or more. Safak (1991)
A total of 41 seismic responses, coming from 14 buildings with
presented a discrete-time filter approach to the identification of
20 stories or more, were analyzed. Table 1 shows a summary of
structural parameters under earthquake conditions that was applied
the 14 buildings, including their location, number of stories, height,
to five buildings after the 1989 Loma Prieta earthquake (Safak and
lateral-force resisting system, material, and number of earthquakes
Celebi 1991, 1992; Celebi and Safak 1991, 1992; Celebi et al.
analyzed. Table 2 shows information about the earthquakes consid-
1993; Celebi 1996), only two of which exceeded 20 stories in
ered. All recorded data was obtained from the Center for Engineer-
height. Goel and Chopra (1997) identified the damping ratio of
ing Strong Motion Data (USGS, CGS, and ANSS 2014).
22 buildings under the Northridge earthquake of 1994; only three
of the buildings had 20 stories or more. Some of these studies were
summarized and form the basis for damping recommendations in System Identification
the ATC-72-1 document (PEER/ATC 2010) for analysis and design
of tall buildings. Reinoso and Miranda (2005) analyzed six tall System identification is a tool to obtain information about a system
buildings located in California and found damping ratios ranging from its measured dynamic response. System identification tech-
from 4 to 6% for concrete buildings and 0.5–8% for steel buildings. niques can be categorized in two types: nonparametric and para-
More recently, Bernal et al. (2015) analyzed the first mode damping metric. In a nonparametric technique, the system is treated like
ratio of 68 buildings subjected to several earthquakes and provided a “black box” and the objective is to identify a functional relation-
some equations to estimate damping. Only five buildings exceeded ship between the inputs and outputs of the system, typically
20 stories in height. In spite of the advances in modern system iden- through its transfer function. These models have several disadvan-
tification techniques, there is still a large variability and hence un- tages such as the lack of prior information about the structure, the
certainty in estimating damping ratios in tall buildings. The same fact that the inferred function inevitably requires a parametric
building can have different results when analyzed under different model for physical interpretation, and that they are not capable
earthquakes, and different identification methods may yield differ- of predicting the response and therefore are not suitable for the seis-
ent results for the same earthquake (Nakajima et al. 2012). Little mic design of a structure (Beck 1978). Briefly, a parametric system
attention has been devoted to the reliability of the inferred damping identification procedure consists of finding the set of parameters
ratios, which has led to damping databases that sometimes include that minimizes the difference between the response recorded by
different values for the same building analyzed by different inves- sensors installed in a building and the one predicted by a numerical
tigators. The lack of reliability testing has also led some of the pre- model of the building. Different identification techniques mainly
viously mentioned studies to report excessively large damping differ in the numerical model employed, in the objective function
ratios for higher modes, or modal period ratios that make no physi- used to measure the difference between the computed and mea-
cal sense. These issues, together with the scarcity of data for tall sured response, and in the optimization algorithm employed to find
buildings, highlight the need for new, reliable damping estimates the parameters that minimize the objective function. If a modal re-
for high-rise buildings. sponse history analysis is used to compute the seismic response of
The Center for Engineering Strong Motion Data (CESMD) is a the model of the structure, the problem involves the estimation of
cooperative center that integrates the earthquake strong-motion the modal periods, critical damping ratios, mode shapes, and modal
data collected by the U.S. Geological Survey (USGS) and the participation factors of the model that will best reproduce the re-
California Geological Survey (CGS). Their database of monitored corded structural response. This process can be conducted in either
structures includes several tall buildings, some of which have been the time domain (e.g., Beck and Jennings 1980) or in the frequency

© ASCE 04016144-2 J. Struct. Eng.

J. Struct. Eng., 2017, 143(1): 04016144


Table 1. List of Buildings Analyzed
Number Identifier Identifier number Location Stories Height (m) Structural system Material Number of earthquakes
1 LA-20 24464 North Hollywood 20 51.5 MF RC 4
2 SD-21 3300 San Diego 21 92.2 MF S 4
3 EM-30 1662 Emeryville 30 95.1 MF RC 1
4 OA-22 58312 Oakland 22 100.0 MF S 2
5 LA-32 24288 Los Angeles 32 102.7 MF S 8
6 SA-25 67900 Sacramento 25 103.6 BF + MF S 1
7 SF-42b 58437 San Francisco 42 129.2 SW S 1
8 SF-47 58532 San Francisco 47 171.9 MF S 1
9 SF-42a 1443 San Francisco 42 183.3 MF S 1
10 SF-62a 58389 San Francisco 62 188.3 SW + OR RC 1
11 SF-58 58411 San Francisco 58 191.4 SW + MF + OR RC 2
Downloaded from ascelibrary.org by "California,Univ Of Irvine" on 04/30/18. Copyright ASCE. For personal use only; all rights reserved.

12 SF-48b 1239 San Francisco 48 206.1 MF S 1


13 LA-54 24629 Los Angeles 54 218.1 MF S 7
14 LA-52 24602 Los Angeles 52 218.2 BF S 7
Note: BF = braced frames; MF = moment frames; OR = outriggers; RC = reinforced concrete; S = steel; SW = shear walls.
a
This building has a tuned liquid-mass damper acting in its transverse direction (this direction was not analyzed).
b
The height of this pyramidal-shaped building was considered from the ground to the setbacks above the 48th floor in the N–S direction.

Table 2. Earthquakes Considered in This Investigation Modal Minimization in the Time Domain
Number Earthquake Year MW Depth (km) In a linear, modal superposition model, the response of each mode
1 Landers 1992 7.3 1.1 is governed by the following equation of motion:
2 Calexico (1) 2010 7.2 32.3
3 Loma Prieta 1989 7.1 18.0 D̈n ðtÞ þ 2ωn ξ n Ḋn ðtÞ þ ω2n Dn ðtÞ ¼ −üg ðtÞ ð1Þ
4 Hector Mine 1999 7.1 6.0
5 Northridge 1994 6.7 19.0 where D̈n , Ḋn , and Dn correspond to the relative acceleration,
6 Big Bear 1992 6.5 1.0 velocity, and displacement of a single degree of freedom system
7 South Napa 2014 6.0 11.3 with unit mass, natural circular frequency ωn, and critical damping
8 Whittier 1987 5.9 9.5 ratio ξ n ; üg = ground acceleration, and the subindex n denotes the
9 Ocotillo 2010 5.7 6.9 mode number. The modal displacements unj at the jth degree of
10 Canyon dam 2013 5.7 11.0 freedom of the structure can then be computed as
11 Sierra Madre 1991 5.6 12.0
12 Alum Rock 2007 5.4 9.2 unj ðtÞ ¼ Γn ϕnj Dn ðtÞ ð2Þ
13 Chino Hills 2008 5.4 13.6
14 Borrego Springs 2010 5.4 14.0 where Γn = modal participation factor; and ϕnj = mode shape for
15 La Habra 2014 5.1 4.8
mode n evaluated at the jth degree of freedom. Multiplying Eq. (1)
16 Calexico (2) 2010 4.9 5.4
17 Inglewood 2009 4.7 15.0
by Γn ϕnj yields
18 Whittier Narrows 2010 4.4 18.0
ünj ðtÞ þ 2ωn ξ n u̇nj ðtÞ þ ω2n unj ðtÞ ¼ −Γn ϕnj üg ðtÞ ð3Þ
19 Encino 2014 4.4 9.9
20 Berkeley (2) 2011 4.0 9.8
21 Berkeley (1) 2011 3.8 9.6 where ünj , u̇nj , and unj correspond to the acceleration, velocity, and
22 LA Airport 2012 3.7 10.8 displacement of the jth degree of freedom for mode n relative to the
ground, respectively.
If only N modes are considered to have a significant influence in
domain (e.g., McVerry 1979). An important advantage of these the seismic response of the building, then the response of the struc-
methods is that the inferred parameters correspond to those that ture can be calculated as
need to be used by practicing structural engineers to reproduce X
N
the recorded response using conventional modal analyses. Another û¨ tj ðtÞ ≈ üg ðtÞ þ Γn ϕnj D̈n ðtÞ ð4Þ
family of methods for the identification of linear structures is the n¼1
use of discrete-time filters (Safak 1991). In this case, the identifi-
cation problem consists of obtaining the optimal parameters of the where û¨ tj = total (absolute) acceleration at the jth degree of
filter that will best reproduce the output signals. These parameters freedom.
of the filter can later be related to the dynamic properties of the The objective function considered in this study is defined as the
¨
difference squared between the predicted relative acceleration ûðtÞ
structure. These methods are widely used in other engineering
fields, so there are several software packages available to perform and the one measured by the sensors in the building üðtÞ normal-
the identification (e.g., Ljung 1999, 2003). Another commonly ized by the measured relative acceleration summed over all times
used family of methods uses a subspace approach (Van Overschee and sensors
and De Moor 1994; Di Ruscio 1996). The advantages of these
methods can be found in Viberg (1995) and Favoreel et al.
X
N sen X
τ
½üj ðiΔtÞ − û¨ j ðiΔtÞ2
JðθÞ ¼ Pτ 2
ð5Þ
(2000). There are numerous other system identification procedures, j¼1 i¼1 k¼1 ½üj ðkΔtÞ
whose details can be found in several literature review articles
(e.g., Ghanem and Shinozuka 1995; Beck 1996; Safak 2001; where N sen = number of sensors above ground level Δt and τ = the
Gevers 2006; Papagiannopoulos and Beskos 2012). time step and the number of points in the signal, respectively. This

© ASCE 04016144-3 J. Struct. Eng.

J. Struct. Eng., 2017, 143(1): 04016144


objective function provides an overall measure of fit between the
computed and measured relative accelerations. The purpose of the
normalization of the squared error at each sensing location in
Eq. (5) is to provide an equal weight to each sensor location. Please
notice that if this normalization is not done, the system identifica-
tion would converge toward parameters that minimize the differ-
ence between computed and measured relative accelerations in
sensor locations experiencing larger accelerations which tend to in-
crease with increasing height and therefore, in general, would tend
to produce better fits for sensors located higher in the building.
The structural parameters are found as those that minimize the
objective function
min JðθÞ ð6Þ
Downloaded from ascelibrary.org by "California,Univ Of Irvine" on 04/30/18. Copyright ASCE. For personal use only; all rights reserved.

where
08 9 8 9 2 31
> ω1 > > ξ1 > Γ1 ϕ11 ··· ΓN ϕN1
>
< >
= >
< > = 6
B . . .. .. .. 7C
θ¼B 6
@> .. >; > .. >; 4 . . . 7C
5A ð7Þ
>
: > >
; : ; >
ωN ξN Γ1 ϕ1N sen ··· ΓN ϕNN sen
Fig. 1. Continuum model (reprinted from Miranda and Akkar 2006,
When designing buildings, the most commonly used type of © ASCE)
analysis is linear elastic response spectrum modal analysis. Further-
more, the base of the structure is typically assumed as fixed.
Eqs. (2)–(4) provide the time history response of fixed-base struc- Mode Inclusion Criteria
tures, so identified modal properties by minimization of the objec-
tive function [Eq. (5)] provide the estimation of modal periods and It is well known that only a small number of modes contribute sig-
damping ratios that best reproduce the measured response in com- nificantly to the seismic response of buildings. Therefore, there is a
bination with the type of analysis in which these properties will be need for cut-off criteria to eliminate parameters of modes that have
used. In particular, the damping ratios identified in this study do not a negligible contribution to the response parameter being consid-
correspond to those of the superstructure alone or what is some- ered at instrumented locations in the structure. This is important
times referred to as the inherent structural damping, but rather they because, if a mode does not contribute significantly to the response
represent the damping ratio of the combined soil-foundation- of the structure, the identified damping ratio may not be reliable.
structure system which include radiation damping due to rocking This study uses cut-off criteria based on the Arias modal contribu-
of the soil–structure interaction. These dynamic properties are often tion factor defined in Eq. (8). This parameter measures the relative
referred to as the “effective” dynamic properties of the equivalent, contribution of the Nth mode with respect to the total calculated
fixed-base model of the building (Veletsos 1977). Therefore, the response with N modes of vibration. In this investigation, only
identified damping ratios correspond to those that will best repro- those modes with an Arias modal contribution higher than 5% were
duce the measured response of the building of a fixed-base linear considered as relevant for the response of the structure.
elastic model when using as input a recorded motion at the base of Pτ PN sen 2
the structure. k¼1 j¼1 ½ünj ðkΔtÞ
MCn ¼ PN Pτ PN sen 2
ð8Þ
Because the optimization problem is highly nonlinear, it is very i¼1 k¼1 j¼1 ½üij ðkΔtÞ
important to establish a good starting point for the minimization
algorithm in order to avoid convergence into local minima. To ob- There are some cases where the ground motion is not strong enough
tain a set of initial parameters, a simplified model of the building in certain range of frequencies to significantly excite the first, and
was used. The model considers the building as a continuous system sometimes also the second, mode of the structure. This may occur,
consisting of a cantilever flexural beam laterally attached to a shear for example, in the case of low-magnitude earthquakes that may
beam such that both undergo the same lateral deformation along the produce significant response of higher modes but negligible re-
height (Fig. 1), and it is governed by a small number of parameters: sponse of the first mode. In this case, the contribution of the first
the fundamental period T 1 , the modal damping ratios ξ n , and the mode may be very small and the damping ratio inferred for the first
lateral stiffness ratio α, which measures the relative contribution of mode may not be reliable. The inclusion criteria of inferred damp-
the flexural rigidity and shear rigidity in the flexural and shear ing ratios for the first two modes are based on the amplification Rn
beams, respectively. Based on these parameters, mode shapes, of the 1% damped floor acceleration spectral ordinates averaged in
period ratios, and modal participation factors are easily computed. the vicinity of the modal period in question for the motion recorded
This model has previously been used to obtain approximate floor at the roof with respect to that of the motion recorded at the base of
acceleration and drift ratio demands, and to identify the dynamic the building as follows:
properties of buildings (Miranda and Taghavi 2005; Reinoso and
Miranda 2005; Miranda and Akkar 2006). The mathematical S̄Roof
a ðT n Þ
Rn ¼ ð9Þ
description of the model, as well as recommended values for α S̄Ground
a ðT n Þ
depending on the structural system of the building, can be found
in Miranda and Reyes (2002), whereas closed form equations of where S̄a ðT n Þ corresponds to the average of the 1% damped floor
modes shapes, period ratios, and modal participation factors can acceleration spectral ordinates within the period range defined by
be found in Miranda and Taghavi (2005) and Miranda and Akkar ½0.9T n ; 1.1T n , plus 0.01g. The purpose of the shift in the spectral
(2006). ordinates is to avoid numerical issues when values are close to zero.

© ASCE 04016144-4 J. Struct. Eng.

J. Struct. Eng., 2017, 143(1): 04016144


Sa [g] Sa [g]
1994 Northridge 2008 Chino Hills
2.5 1.0
Roof Roof
2.0 Ground 0.8 Ground

1.5 0.6

1.0 0.4

0.5 0.2

0.0 0.0
0 2 4 6 8 0 2 4 6 8
Downloaded from ascelibrary.org by "California,Univ Of Irvine" on 04/30/18. Copyright ASCE. For personal use only; all rights reserved.

(a) Period [s] (b) Period [s]

Fig. 2. Roof and ground acceleration response spectrum (ξ ¼ 1%) for the E–W direction of building LA-52: (a) 1994 Northridge earthquake;
(b) 2008 Chino Hills earthquake

The cut-off criteria employed were R1 > 2.2 and R2 > 1.6. An than 1% in the objective function. It can be seen that the damping
example of the application of these criteria is illustrated in Fig. 2 interval for the 1992 Landers M W 7.3 earthquake is significantly
for building LA-52. Fig. 2(a) shows ground and roof lightly narrower than the one computed for 1994 Northridge MW 6.7 earth-
damped (1% damped) response spectra for the 1994 M6.7 quake, indicating that there was a larger sensitivity of the acceler-
Northridge earthquake where it can be seen that there is significant ation response to changes in the damping ratio in the Landers
amplification of spectral ordinates of the motion recorded at the earthquake than in the Northridge earthquake. Although a damping
roof relative to spectral ordinates of the motion recorded at the ratio was adequately identified for both events the damping ratio
ground for periods in the vicinity of the first mode period (approx- inferred from the Landers earthquake, an event with larger magni-
imately 6 s); with a computed amplification factor of R1 ¼ 6.97, tude than the Northridge earthquake produces a stronger response
indicating that the first mode is sufficiently excited during this to the first mode and therefore yields a more reliable estimate of the
earthquake. The vertical lines in these figures indicate the period damping ratio of the fundamental period of the building. Similarly,
range used to calculate the first mode amplification factor R1. the damping ratio inferred from the 2014 La Habra M W 5.1 earth-
Meanwhile, Fig. 2(b) shows ground and roof 1% damped response quake, although similar in value to that inferred from the Landers
spectra for the 2008 M5.4 Chino Hills earthquake where almost no earthquake, is less reliable because the objective function is not as
amplification of spectral ordinates occur for periods in the vicinity sensible to small changes in the damping ratio as it was during the
of the first mode in which an amplification factor of R1 ¼ 1.60 is Landers earthquake.
computed. Therefore, the damping ratio inferred for the fundamen-
tal mode under this particular earthquake is not considered in fur-
ther analyses. Application to Tall Buildings

Identified Damping Ratios for the Fundamental Mode


Reliability of the Damping Estimates
The damping ratios for the fundamental mode were first studied and
The reliability of the damping estimates was quantified by the sen- analyzed. Fig. 4 shows the identified damping ratios for the first
sitivity of the objective function J, defined in Eq. (5), to small mode, for both directions of each building, and its variation with
changes in the modal damping ratio. The identified parameters the building height. When more than one damping ratio was avail-
minimize J, so any increase or reduction in the damping ratio with able for a building, the one with the narrowest reliability interval
respect to the identified damping ratio will cause an increment in was plotted in this figure. It can be seen that, as previously observed
the value of the objective function. However, modes which are not by other investigators (Suda et al. 1996; Smith and Willford 2007;
significantly contributing to the acceleration response changes in Bernal et al. 2015) the damping ratio decreases with increasing
the damping ratio may cause negligible changes in the objective building height. Unlike the first two studies, which primarily in-
function, indicating that although the identified damping ratios cluded damping ratios inferred from buildings subjected to wind
minimize the objective function, they are less reliable and they and other low-amplitude vibrations, the study by Bernal et al.
may not necessarily correspond to those of the structure. In this (2015) is based entirely on instrumented buildings subjected to
study the reliability of the identified damping ratio was quantified seismic loading. This study significantly expands the study by
by identifying the interval of damping ratios that produce changes Bernal for tall buildings by including nearly three times more build-
in the objective function J that are within 1% of its minimum value. ings with more than 20 stories.
The width of this interval of damping ratios is a measure of the With the exception of one point (one building in one direction),
reliability of the inferred damping values, allowing assessment all the damping ratios in Fig 4 obtained for buildings taller than
of the reliability of the modal damping ratios obtained for the same 150 m have values lower than 2.5%—the damping value that is
building when subjected to different earthquakes. Fig. 3 shows the commonly recommended for seismic design of tall buildings in
variation of the objective function with changes in the damping ra- the United States (LATBSDC 2006; TBI 2010). A power regression
tio of the first mode of the LA-52 building when subjected to three was fitted to the data and also plotted in Fig. 4; the resulting co-
different earthquakes. In this figure, the small circles indicate the efficients can be seen in Table 3. Fig. 4 also compares the data to the
minimum of the objective function. The vertical lines show the in- damping recommendations provided by four different investiga-
tervals containing the damping values that cause a variation of less tions: Satake et al. (2003), Fritz et al. (2009), PEER/ATC-72-1

© ASCE 04016144-5 J. Struct. Eng.

J. Struct. Eng., 2017, 143(1): 04016144


1992 Landers 1994 Northridge 2014 La Habra
20

18

16

14

12

10

4
Downloaded from ascelibrary.org by "California,Univ Of Irvine" on 04/30/18. Copyright ASCE. For personal use only; all rights reserved.

0
0 2 4 6 8 10 0 2 4 6 8 10 0 2 4 6 8 10
1
[%] 1
[%] 1
[%]

Fig. 3. Variation of the objective function, J, with changes in the damping ratio of the first mode for building LA-52, during three different
earthquakes

Steel Buildings Concrete Buildings


8% 8%
Fritz et al. 2009 Bernal et al. 2015
7% Bernal et al. 2015 7% ATC-72-1 2010
ATC-72-1 2010 Fritz et al. 2009
6% 6%
Satake et al. 2003 Satake et al. 2003
5% This investigation 5% This investigation

4% 4%

3% 3%

2% 2%
1% 1%

0% 0%
0 50 100 150 200 250 300 0 50 100 150 200 250 300
(a) Building height (m) (b) Building height (m)

Fig. 4. Identified damping ratios for the fundamental mode versus building height, compared to some proposed recommendations: (a) steel buildings;
(b) reinforced concrete buildings

Table 3. Coefficients for Power Regression by Fritz et al. (2009) slightly overestimates the data, and may be
Material A B R2 Function form thought of as an upper bound. For concrete buildings [Fig. 4(b)]
the data is very scarce. However, it is worrisome that, with the ex-
Steel 0.90 −0.76 0.45 ξ1 ¼ A · HB
ception of Satake et al. (2003), all recommendations overestimate
Concrete 15.60 −1.31 0.81
the data for buildings higher than 150 m. The 2% constant value
recommended in the performance-based PEER guidelines also ap-
pears to be high for buildings taller than 180 m. Values of damping
(2010), and Bernal et al. (2015). For recommendations that use the ratios lower than 2% from this investigation are consistent to
number of stories as regressors, an average height of 3.96 m (13 ft) those reported by Smith and Willford (2007) or by Spence and
per story was considered to transform the number of stories into Kareem (2013).
building height. Caution should be taken when comparing these
regressions because they have been calculated from damping ratios
Damping of Higher Modes
that have been collected from different loading conditions, using
different system identification methods and assumptions. Never- To study the damping ratio of higher modes, the three buildings
theless, they all show a similar descending trend with building with the highest number of recorded events were selected: LA-32,
height. For steel buildings [Fig. 4(a)], it can be seen that the curve LA-52, and LA-54. SD-21 and LA-20 also have records available
by Bernal agrees with the data over all the height range. The equa- from several earthquakes, but their response is governed primarily
tion proposed by Satake et al. (2003) underestimates the damping by only the first two translational modes of vibration in each direc-
ratios inferred in this study. This is probably because most of the tion, so they were not considered in this part of the study. Fig. 5
data used by Satake and colleagues is based on very small vibration shows the relation between the identified damping ratios ξ n in each
amplitudes. For buildings over 150 m, the equation recommended of the three buildings in each of the earthquakes when plotted

© ASCE 04016144-6 J. Struct. Eng.

J. Struct. Eng., 2017, 143(1): 04016144


n LA-32 - Long. n LA-52 - Long. n LA-54 - Long.
8% 4% 4%

6% 3% 3%

4% 2% 2%

2% 1% 1%
R² = 0.56 R² = 0.76 R² = 0.48
0% 0% 0%
0.0 1.0 2.0 3.0 4.0 0.0 1.0 2.0 3.0 0.0 0.5 1.0 1.5 2.0
Frequency (Hz) Frequency (Hz) Frequency (Hz)
Downloaded from ascelibrary.org by "California,Univ Of Irvine" on 04/30/18. Copyright ASCE. For personal use only; all rights reserved.

n LA-32 - Trans. n LA-52 - Trans. n LA-54 - Trans.


8% 4% 4%

6% 3% 3%

4% 2% 2%

2% 1% 1%
R² = 0.70 R² = 0.44 R² = 0.64
0% 0% 0%
0.0 1.0 2.0 3.0 0.0 1.0 2.0 3.0 0.0 0.5 1.0 1.5 2.0
Frequency (Hz) Frequency (Hz) Frequency (Hz)

Fig. 5. Identified damping ratios versus modal frequencies, in the longitudinal and transverse directions, for three tall buildings in Los Angeles

as a function of their corresponding identified modal frequency f. Damping versus Amplitude


The plots include only the data that passed the reliability tests
Several previous investigations have studied the effect of amplitude
previously described. It can be seen that, in all cases, damping in- on damping ratios. Most of these studies have considered relatively
creases with increasing frequency. This is expected due to inertial small levels of lateral deformation under wind loading. A notable
soil–structure interaction effects in which radiation damping in- exception is the study by Bernal et al. (2015) that evaluated the
creases with increasing frequency for the rocking degree of free- correlation between damping ratios in the first mode and amplitude
dom. Also shown in the figures is a linear model fitted to the in buildings subjected to earthquake loading. Different studies have
data, indicating that damping ratios of higher modes approximately used different parameters as measures of amplitude. Usually, the
increase linearly with increasing frequency. However, the linear amplitude metric is based on some function of the measured accel-
trend does not entirely correspond to a stiffness-proportional damp- eration, like the peak value (e.g., Tamura and Suganuma 1996), the
ing because this type of damping model would force the linear RMS (e.g., Tamura et al. 1994), or the spectral ordinate of the first
trend to start from the origin. It can be seen that the intercept of mode (e.g., Bernal et al. 2015). Although some of these metrics
the regressions are, in all cases, different from zero. This is coherent may be appropriate for low-rise buildings whose response is often
with the findings of Satake et al. (2003) for the first three modes, dominated by the fundamental mode of vibration, high-rise build-
who concluded that higher mode damping ratios can be thought to ings are strongly influenced by higher modes, and experiencing
be an intermediate between constant damping for all modes and high levels of acceleration may not necessarily imply large levels
stiffness-proportional damping. It should be noted that in none of lateral deformation. Furthermore, buildings experiencing similar
of the three buildings there is a hyperbolic trend of decreasing levels of acceleration but having very different fundamental periods
damping ratios with increasing frequency, indicating that using a of vibration may have experienced very different levels of lateral
mass-proportional damping model is clearly inadequate. The most deformation demands. For this reason, this study evaluates possible
common approach to model the damping in structures is to consider changes in damping ratio as a function of the peak roof drift ratio
Rayleigh damping in which damping ratios are specified as a linear (PRDR), defined as the peak displacement at roof level relative to
combination of the stiffness- and mass-proportional models. the ground normalized by roof height. This is a commonly used
Typically, a damping ratio is specified to be the same in the first metric in earthquake engineering to describe the overall, or global,
and third mode or in the first and fourth modes. It was found that measure of lateral deformation experienced by a building during an
such a model did not provide a correct trend when selecting the earthquake. Observations made by other researchers for buildings
same damping ratio to the first and third modes of vibration, when under low-amplitude motions show that damping increases with
selecting the same damping ratio to the first and fourth modes of increasing PRDR, when PRDR is less than 0.01% (Jeary 1986;
vibration, nor when finding the best fit of the Rayleigh damping Li et al. 2000; Tamura et al. 2000). However, after reaching a cer-
model. Coefficients of determination were consistently much lower tain amplitude level, damping saturates and may even start decreas-
than when using a simple linear trend, indicating that the use of ing (Fukuwa et al. 1996; Tamura and Suganuma 1996; Tamura
Rayleigh is not appropriate for tall buildings and, as shown in Fig. 5, 2012; Spence and Kareem 2013; Bernal et al. 2015). Most of these
a linear variation provides a much better model of damping ratios in previous observations have been made under motions in which roof
tall buildings. drift ratios were smaller than 0.01%. A significant increase in

© ASCE 04016144-7 J. Struct. Eng.

J. Struct. Eng., 2017, 143(1): 04016144


1 LA-54 (H = 218 m) 1 LA-52 (H = 218 m) 1 LA-32 (H = 103 m)
5% 5% 5%
Longitudinal Longitudinal Longitudinal
4% Transverse 4% Transverse 4% Transverse
3% 3% 3%

2% 2% 2%

1% 1% 1%

0% 0% 0%
0.00% 0.10% 0.20% 0.30% 0.00% 0.05% 0.10% 0.15% 0.20% 0.00% 0.05% 0.10% 0.15%
Peak roof drift ratio Peak roof drift ratio Peak roof drift ratio

1 SD-21 (H = 92 m) 1 LA-20 (H = 52 m)
7% 12%
Longitudinal Longitudinal
6%
Downloaded from ascelibrary.org by "California,Univ Of Irvine" on 04/30/18. Copyright ASCE. For personal use only; all rights reserved.

10%
Transverse Transverse
5%
8%
4%
6%
3%
4%
2%
1% 2%
0% 0%
0.00% 0.05% 0.10% 0.15% 0.00% 0.10% 0.20% 0.30% 0.40%
Peak roof drift ratio Peak roof drift ratio

Fig. 6. Identified damping ratios for the fundamental mode as a function of the peak roof drift ratio

damping has also been observed when comparing values inferred they provide information of the best damping ratios to use in con-
from ambient vibrations with those from strong-motion data nection with the most typical model used in standard engineering
(Celebi et al. 1993; Celebi 1996). However, there is very little in- practice. The reliability of the inferred damping ratios was assessed
formation regarding the variation of damping with increasing am- based on quantifying the sensitivity of the objective function to
plitude under earthquake ground motions, and buildings that have changes in the damping ratio and the amplification of lightly
several recorded events provide an excellent opportunity to study damped roof acceleration spectral ordinates with respect to those
this trend. The variation of inferred damping ratios of the funda- computed with the ground motion. Using only reliable damping
mental mode with changes in PRDR is examined in Fig. 6 for estimates, the variation of the first-mode damping with height
all buildings with four or more earthquakes recorded in the build- was examined and compared to four different existing damping rec-
ing. The plots show the results obtained for all damping ratios, in- ommendations. Results of this investigation concur with previous
ferred from the various ground motions recorded at each building, investigations that have indicated that damping ratios decrease with
that pass the reliability restrictions previously described. The only increasing building height and that damping ratios in concrete
case that shows an increase of damping with amplitude was the structures are higher than those in steel structures. Furthermore,
longitudinal direction of LA-54; in all other cases no clear increase regardless of the material, it was found that all buildings higher
in damping ratio is observed with increases in level of response. To than 150 m had damping ratios lower than the usual 2.5% used
study the correlation between damping and amplitude, the damping in seismic design.
ratios of each building were normalized by the average of the Damping ratios associated with higher modes of vibration were
damping ratios inferred for each building in earthquakes causing examined for three tall buildings that recorded several earthquakes.
a PRDR of less than 0.03% in the same building. When the lon- It was found that damping ratios of higher modes increase with the
gitudinal direction of LA-54 was removed from the data set, no modal frequency, and it was shown that a linear variation is a better
statistical correlation was found between damping and amplitude. fit than the commonly used mass-proportional, stiffness-propor-
The best example of this lack of correlation is building LA-52, tional, or Rayleigh damping models. Finally, it was found that
where the same damping ratio is inferred from earthquakes causing in most cases the damping ratio remains approximately constant
PRDRs of 0.02 and 0.19%, meaning that an increase of 850% even with tenfold increases in peak roof drift ratios. In one case
in the amplitude of the motion does not cause a significant change the damping ratios decreased with increasing amplitude of defor-
in the identified damping ratio. This suggests that the trends mation. These observations on saturation and possible reductions in
observed under low-amplitude motions in which damping ratios damping ratios for earthquake loading are consistent with previous
saturate and do not increase with increasing amplitude may extend observations from wind-induced vibrations and free-vibration tests.
to seismic-induced motion.

Acknowledgments
Conclusions
The authors would like to acknowledge CONICYT—Becas Chile,
The damping ratios of 14 buildings with 20 stories or more, sub- and the Blume Earthquake Engineering Center at Stanford Univer-
jected to earthquake motions, were inferred using a parametric sity for the financial aid for to the first author for conducting doc-
modal system identification technique. Identified damping ratios toral studies at Stanford under the supervision of the second author.
correspond to those that will best reproduce the measured response Ground and structural motions used in this investigation were ob-
of the building of a fixed-base linear elastic model when using as tained from the California Strong Motion Instrumentation Program
input at recorded motion at the base of the structure. In other words, of the California Geological Survey and from the United States

© ASCE 04016144-8 J. Struct. Eng.

J. Struct. Eng., 2017, 143(1): 04016144


Geological Survey. Efforts to install, operate, and maintain seismic Celebi, M., Phan, L. T., and Marshall, R. D. (1993). “Dynamic character-
instrumentation in buildings as well as to process and disseminate istics of five tall buildings during strong and low-amplitude motions.”
earthquake records by these organizations are gratefully acknowl- Struct. Des. Tall Build., 2(1), 1–15.
edged. The authors also wish to thank two anonymous reviewers Celebi, M., and Safak, E. (1991). “Seismic response of Transamerica build-
ing. I: Data and preliminary analysis.” J. Struct. Eng., 10.1061/(ASCE)
for their careful review of our manuscript as well as for their sug-
0733-9445(1991)117:8(2389), 2389–2404.
gestions to improve it. Celebi, M., and Safak, E. (1992). “Seismic response of Pacific Park Plaza. I:
Data and preliminary analysis.” J. Struct. Eng., 10.1061/(ASCE)0733-
9445(1992)118:6(1547), 1547–1565.
Notation Davenport, A. G., and Hill-Carroll, P. (1986). “Damping in tall buildings:
Its variability and treatment in design.” Proc., American Society of Civil
The following symbols are used in this paper: Engineers (ASCE) Spring Convention, ASCE, New York, 42–57.
Dn ; Ḋn ; D̈n = relative acceleration, velocity, and displacement of Di Ruscio, D. (1996). “Combined deterministic and stochastic system iden-
the unit mass SDOF system associated to the nth tification and realization: DSR—A subspace approach based on obser-
Downloaded from ascelibrary.org by "California,Univ Of Irvine" on 04/30/18. Copyright ASCE. For personal use only; all rights reserved.

mode of vibration; vations.” Model Ident. Control, 17(3), 193–230.


g = acceleration of gravity; Favoreel, W., De Moor, B., and Van Overschee, P. (2000). “Subspace
J = objective function; state space system identification for industrial processes.” J. Process
N = number of modes considered to have a significant Control, 10(2), 149–155.
influence in the seismic response of the building; Fritz, W. P., Jones, N. P., and Igusa, T. (2009). “Predictive models for the
N sen = number of sensors above ground level; median and variability of building period and damping.” J. Struct. Eng.,
10.1061/(ASCE)0733-9445(2009)135:5(576), 576–586.
Rn = spectral amplification factor of the nth mode of
Fukuwa, N., Nishizaka, R., Yagi, S., Tanaka, K., and Tamura, Y. (1996).
vibration; “Field measurement of damping and natural frequency of an actual
S̄a ðT n Þ = average of the 1% damped floor acceleration steel-framed building over a wide range of amplitudes.” J. Wind.
spectral ordinates evaluated within the period range Eng. Ind. Aerodyn., 59(2), 325–347.
defined by ½0.9T n ; 1.1T n , plus 0.01g; Gevers, M. (2006). “A personal view of the development of system iden-
T n = period of the nth mode of vibration; tification: A 30-year journey through an exciting field.” IEEE Control
üg = ground acceleration; Syst., 26(6), 93–105.
unj , u̇nj , ünj = relative acceleration, velocity, and displacement of Ghanem, R., and Shinozuka, M. (1995). “Structural-system identification.
the nth mode of vibration evaluated at the jth I: Theory.” J. Eng. Mech., 10.1061/(ASCE)0733-9399(1995)121:
degree of freedom; 2(255), 255–264.
üj = measured relative acceleration at the jth degree of Goel, R. K., and Chopra, A. K. (1997). “Vibration properties of buildings
determined from recorded earthquake motions.” Rep. No. UCB/EERC-
freedom;
97/14, Earthquake Engineering Research Center, Univ. of California,
û¨ j = predicted relative acceleration at the jth degree of Berkeley, CA.
freedom; Hart, G. C., and Vasudevan, R. (1975). “Earthquake design of buildings:
û¨ tj = predicted absolute acceleration at the jth degree of Damping.” J. Struct. Div., 101(1), 11–30.
freedom; Haviland, R. (1976). “A study of the uncertainties in the fundamental trans-
α = lateral stiffness coefficient; lational periods and damping values for real buildings.” Research Rep.
Γn = modal participation factor of the nth mode of No. 5, Massachusetts Institute of Technology, Cambridge, MA.
vibration; Hill-Carroll, P. E. B. (1985). “The prediction of mean structural damping
Δt = time step of the record; values and their coefficients of variation.” M. Eng. thesis, Univ. of
ξ n = critical damping ratio of the nth mode of vibration; Western Ontario, London, ON, Canada.
τ = number of points in the signal; Jeary, A. P. (1986). “Damping in tall buildings—A mechanism and a pre-
dictor.” Earthquake Eng. Struct. Dyn., 14(5), 733–750.
ϕnj = mode shape for the nth mode of vibration evaluated
Lagomarsino, S. (1993). “Forecast models for damping and vibration peri-
at the jth degree of freedom; and
ods of buildings.” J. Wind. Eng. Ind. Aerodyn., 48(2), 221–239.
ωn = natural circular frequency of the nth mode of LATBSDC (Los Angeles Tall Buildings Structural Design Council).
vibration. (2006). “An alternative procedure for seismic analysis and design of
tall buildings located in the Los Angeles region.” Los Angeles, CA.
Li, Q. S., Liu, D. K., Fang, J. Q., Jeary, A. P., and Wong, C. K. (2000).
References “Damping in buildings: Its neural network model and AR model.” Eng.
Struct., 22(9), 1216–1223.
Beck, J. L. (1978). “Determining models of structures from earthquake Li, Y., and Mau, S. T. (1991). “A case study of MIMO system identification
records.” Rep. No. EERL 78-01, Earthquake Engineering Research applied to building seismic records.” Earthquake Eng. Struct. Dyn.,
Laboratory, California Institute of Technology, Pasadena, CA. 20(11), 1045–1064.
Beck, J. L. (1996). “System identification methods applied to measured Li, Y., and Mau, S. T. (1997). “Learning from recorded earthquake motion
seismic response.” Proc., 11th World Conf. on Earthquake Engineer- of buildings.” J. Struct. Eng., 10.1061/(ASCE)0733-9445(1997)123:
ing, Elsevier, New York. 1(62), 62–69.
Beck, J. L., and Jennings, P. C. (1980). “Structural identification using Ljung, L. (1999). System identification: Theory for the user, Prentice Hall,
linear models and earthquake records.” Earthquake Eng. Struct. Saddle River, NJ.
Dyn., 8(2), 145–160. Ljung, L. (2003). System identification toolbox user’s guide, MathWorks,
Bernal, D., Döhler, M., Kojidi, S. M., Kwan, K., and Liu, Y. (2015). Natick, MA.
“First mode damping ratios for buildings.” Earthquake Spectra, 31(1), Mau, S. T., and Aruna, V. (1994). “Story-drift, shear, and OTM estimation
367–381. from building seismic records.” J. Struct. Eng., 10.1061/(ASCE)0733-
Blume, J. A. (1970). “The motion and damping of buildings relative to 9445(1994)120:11(3366), 3366–3385.
seismic response spectra.” Bull. Seismol. Soc. Am., 60(1), 231–259. McVerry, G. H. (1979). “Frequency domain identification of structural
Celebi, M. (1996). “Comparison of damping in buildings under low- models from earthquake records.” Rep. No. EERL 79-02, Earthquake
amplitude and strong motions.” J. Wind. Eng. Ind. Aerodyn., 59(2), Engineering Research Laboratory, California Institute of Technology,
309–323. Pasadena, CA.

© ASCE 04016144-9 J. Struct. Eng.

J. Struct. Eng., 2017, 143(1): 04016144


Miranda, E., and Akkar, S. D. (2006). “Generalized interstory drift spec- Smith, R. J., and Willford, M. R. (2007). “The damped outrigger concept
trum.” J. Struct. Eng., 10.1061/(ASCE)0733-9445(2006)132:6(840), for tall buildings.” Struct. Des. Tall Build., 16(4), 501–517.
840–852. Spence, S. M., and Kareem, A. (2013). “Tall buildings and damping: A
Miranda, E., and Reyes, C. J. (2002). “Approximate lateral drift demands concept-based data-driven model.” J. Struct. Eng., 10.1061/(ASCE)
in multistory buildings with nonuniform stiffness.” J. Struct. Eng., ST.1943-541X.0000890, 04014005.
10.1061/(ASCE)0733-9445(2002)128:7(840), 840–849. Stagner, J. R., and Hart, G. C. (1971). “Damping estimation and digital
Miranda, E., and Taghavi, S. (2005). “Approximate floor acceleration de- filtering applied to structural motion studies.” Rep. No. UCLA-ENG-
mands in multistory buildings. I: Formulation.” J. Struct. Eng., 10.1061/ 7181, Univ. of California, Los Angeles.
(ASCE)0733-9445(2005)131:2(203), 203–211. Suda, K., Satake, N., Ono, J., and Sasaki, A. (1996). “Damping properties
Nakajima, K., et al. (2012). “Seismic response analysis of a semi-active-
of buildings in Japan.” J. Wind. Eng. Ind. Aerodyn., 59(2), 383–392.
controlled base-isolated building during the 2011 Great East Japan
Tamura, Y. (2012). “Amplitude dependency of damping in buildings and
earthquake.” Proc., Int. Symp. on Engineering Lessons Learned from
critical tip drift ratio.” Int. J. High-Rise Build., 1(1), 1–13.
the 2011 Great East Japan Earthquake, Japan Association for
Earthquake Engineering, Tokyo, 1025–1036. Tamura, Y., Suda, K., and Sasaki, A. (2000). “Damping in buildings for
wind resistant design.” Proc., Int. Symp. on Wind and Structures for
Downloaded from ascelibrary.org by "California,Univ Of Irvine" on 04/30/18. Copyright ASCE. For personal use only; all rights reserved.

Papagiannopoulos, G., and Beskos, D. (2012). “Damping identification for


building structures subjected to earthquakes: A review.” J. Serb. Soc. the 21st Century, Techno-Press, Daejon, Korea, 115–129.
Comp. Mech., 6(1), 129–147. Tamura, Y., and Suganuma, S. Y. (1996). “Evaluation of amplitude-
PEER/ATC. (2010). “Modeling and acceptance criteria for seismic design dependent damping and natural frequency of buildings during strong
and analysis of tall buildings.” PEER/ATC-72-1, Applied Technology winds.” J. Wind. Eng. Ind. Aerodyn., 59(2), 115–130.
Council in cooperation with the Pacific Earthquake Engineering Tamura, Y., Yamada, K., and Yohota, H. (1994). “Estimation of structural
Research Center, Redwood City, CA. damping of buildings” Proc., ASCE Structural Congress and IASS Int.
Reinoso, E., and Miranda, E. (2005). “Estimation of floor acceleration de- Symp., ASCE, New York, 1012–1017.
mands in high-rise buildings during earthquakes.” Struct. Des. Tall Tamura, Y., and Yoshida, A. (2008). “Amplitude dependency of damping in
Build., 14(2), 107–130. buildings.” 2008 Structures Congress: 18th Analysis and Computation
Safak, E. (1991). “Identification of linear structures using discrete-time fil- Specialty Conf., Structural Engineering Institute, Reston, VA.
ters.” J. Struct. Eng., 10.1061/(ASCE)0733-9445(1991)117:10(3064), Tanaka, T., Yoshizawa, S., Osawa, Y., and Morishita, T. (1969). “Period and
3064–3085. damping of vibration in actual buildings during earthquakes.” Bull.
Safak, E. (2001). “Analysis of earthquake records from structures: An over- Earthquake Res. Inst., 47, 1073–1092.
view.” Strong motion instrumentation for civil engineering structures, TBI Guidelines Working Group. (2010). “Guidelines for performance-
Springer, Netherlands, 91–107. based seismic design of tall buildings.” PEER Rep. No. 2010/05, Pacific
Safak, E., and Celebi, M. (1991). “Seismic response of Transamerica build-
Earthquake Engineering Research Center, Berkeley, CA.
ing. II: System identification.” J. Struct. Eng., 10.1061/(ASCE)0733-
USGS, CGS, and ANSS. (2014). “Center for engineering strong motion
9445(1991)117:8(2405), 2405–2425.
data.” 〈http://strongmotioncenter.org〉 (Oct. 18, 2014).
Safak, E., and Celebi, M. (1992). “Recorded seismic response of Pacific
Park Plaza. II: System identification.” J. Struct. Eng., 10.1061/ Van Overschee, P., and De Moor, B. (1994). “N4SID: Subspace algorithms
(ASCE)0733-9445(1992)118:6(1566), 1566–1589. for the identification of combined deterministic-stochastic systems.”
Satake, N., Suda, K. I., Arakawa, T., Sasaki, A., and Tamura, Y. (2003). Automatica, 30(1), 75–93.
“Damping evaluation using full-scale data of buildings in Japan.” J. Veletsos, A. S. (1977). “Dynamics of structure-foundation systems.” Struc-
Struct. Eng., 10.1061/(ASCE)0733-9445(2003)129:4(470), 470–477. tural and geotechnical mechanics, Prentice-Hall, Englewood Cliffs, NJ,
Scholl, R. E. (1975). “Effects prediction guidelines for structures subjected 333–361.
to ground motion.” Rep. No. JAB-99-115, URS/Blume Engineers, Viberg, M. (1995). “Subspace-based methods for the identification of linear
San Francisco, CA. time-invariant systems.” Automatica, 31(12), 1835–1851.

© ASCE 04016144-10 J. Struct. Eng.

J. Struct. Eng., 2017, 143(1): 04016144

You might also like