You are on page 1of 8

構造工学論文集 Vol.

57B (2011 年 3 月) 日本建築学会

PEAK STORY DRIFT DISTRIBUTION IN PASSIVELY DAMPED MULTISTORY BUILDINGS


A simplified estimation method

Fides UMEMURA*, Hisashi UMEMURA** and Toshimi KABEYASAWA***

This paper describes a modified nonlinear static analysis method to estimate seismic response in buildings with passive damping
systems. Recently developed static methods for passively damped buildings provide a wide scope of application, in that some can consider
inelastic responses and some may be applied for various types of passive dampers. The proposed procedure not only considers these but
also includes the estimation of responses for arbitrarily located dampers, i.e. non-proportional story-wise passive damper distributions.
This is achieved by explicitly modeling passive dampers in the structural model used for static pushover analysis. Truss members with
linearly elastic stiffness are substituted in place of the damper’s force-deformation model using equal energy considerations. Validation
of the method is conducted by comparing estimated peak story drift distributions to those from dynamic analyses of reinforced concrete
buildings with various types of passive dampers, which showed promising results. In general, the results imply that there are different
numbers of significant modes which need to be considered in the estimation procedure depending on how unproportional the story-wise
distribution of the passive dampers are, and on the type of passive damper used. For non-proportional damper distributions, significant
number of modes was is inversely related to the relative amount of passive damper in the story, i.e. the more passive dampers in the
story, the less the number of significant modes, which shows different higher mode contribution in stories even for the same building.
For most cases of inelastic behavior of passively damped buildings, nonlinear time-history response analysis is required. However, such
responses can be very sensitive to the characteristics of individual ground motion used as seismic input and would therefore need a wide
set of records to cover all phase variation responses. Moreover, the number of dynamic analysis needed would greatly increase during
the design stage. This estimation method, which utilizes pushover and modal analysis for considering higher modes, can facilitate the
analysis and design of passively damped structures, particularly for designing optimal location of passive dampers.

Keywords: Nonlinear static analysis, Story drifts, Multistory buildings, Passive dampers

1. INTRODUCTION suitable or possible to distribute in such a manner, especially for buildings


with space or design constraints. During design, consideration of uncon-
Passive dampers are now being used for many types of building struc-
ventional damper configurations in order to optimize damper effect, i.e.
tures, from low-rise to high-rise buildings and for steel, reinforced con-
maximize seismic response reduction, will require a great number of dy-
crete, as well as wood construction. Apartment buildings in Japan are
namic time-history response analyses.
now typically constructed with some form of passive damping or base
In this paper, a modified nonlinear static analysis method for estimat-
isolation system or a combination of both. However, such buildings with
ing story drift response in passively damped multistory structures is pre-
passive dampers relied heavily on extensive testing and experimentation
sented. The proposed estimation method can make allowances for non-
for their earthquake-resistant design. A lot of previous researches on this
proportional passive damper distributions. The proposed procedure, un-
subject are also experimental in nature, that is, often limited to a particu-
like similar estimation methods, models the passive dampers explicitly
lar type of damper or to the properties of the test structure used. Hence,
and includes these models in the multiple-degree-of-freedom (MDOF)
despite the widespread use of passive dampers for the seismic design of
nonlinear frame-damper system used for pushover analysis.
new buildings, there is still no comprehensive analysis and design method
or guideline in place in Japan.
2. ESTIMATION PROCEDURE FOR PEAK STORY DRIFT RE-
The nonlinear static analysis procedure may be used as an alternative SPONSE
to dynamic analysis to estimate the response of buildings with a wide The proposed procedure for evaluating the peak story drift response
variety of passive dampers while considering nonlinear behavior of the of passively damped buildings is based on the nonlinear static analysis
1)2)3)
multistory main structure . For buildings with velocity-dependent approach, which is better known as the pushover analysis, and uses the
dampers, however, passive dampers have to be distributed such that pro- concept of the capacity spectrum method. Focus is given to the peak
portional damping may be assumed for the entire structure. Hence, a story drift of the building in response to the earthquake since damage to
common practice is to distribute dampers such that the damper stiffness structural and non-structural building elements is known to be closely re-
is in proportion to the building frame stiffness. However, it is not always
* Ph. D.
** Associate Prof., Dept. of Architecture and Design, Faculty of Engineering, Nagoya Institute of Technology, Ph. D.
*** Prof., Dept. of Architecture, Faculty of Engineering, University of Tokyo, Dr. Eng
r r r
lated to the story drifts, i.e. lateral displacements between two adjoining M1 M1 D1
T 1 = 2π = 2π P = 2π (4)
story levels. The building may be represented by either a two- or three- K1 Fi /D1 A1

dimensional mathematical model that explicitly includes the model/s of where mi and xi denote the story mass and story lateral displace-
the passive damping system. The estimation procedure are described in ment for the MDOF system, and M1 , D1 , A1 and T 1 denote the mass,
detailed steps hereafter, with a flowchart summary shown in Fig. 1. displacement, acceleration and period for the SDOF system respec-
INPUT: tively. Express the SDOF capacity curve in the spectral acceleration-
(a) Damper Amount   (b) Building Frame   (c) Input EGM  displacement (i.e., A1 -D1 ) format.
 and Location   Model   
   
(4) Superimpose the S pa -S d curve of the EGM for an assumed hEGM
STEPS:
? value over the SDOF capacity curve and get the intersection point.:
- (1) Truss Member Substitution
- for Passive Damper Models Using the capacity spectrum method, and the earthquake demand,

?
get the intersection point between the structural capacity curve and
(2) Nonlinear Pushover Analysis the seismic earthquake demand expressed in the spectral pseudo-
? acceleration-displacement (S pa -S d ) format for an initially assumed
(3) SDOF Conversion
viscous damping ratio hEGM .
? (5) Obtain the MDOF system response for the pushover load step cor-
(4) Spectral Capacity-Demand 
Intersection Point: Assume hEGM 
responding to the intersection point. Compute heq .:
(5b) Iteration until heq ≈ hEGM : (a)Using the initial SDOF system intersection point in Step 4, the es-
Compare and replace hEGM with heq
timated peak story deformations are taken as the deformation values
? 6
(5a) MDOF Story Response Estimates: at the corresponding pushover step of the MDOF system. Calculate
Get Peak Story Drifts and Calculate heq
the peak story drifts from the difference in deformation at adjoining
?
(6) Higher Mode Response stories.
From the MDOF response, compute the equivalent damping ratio heq
Fig. 1 Flowchart of the estimation procedure for the SDOF system as the sum of the effective damping ratios con-
tributed by the passive damping system hd , the hysteretic behavior
of the main structure hh , as well as the inherent viscous damping in
2.1. Procedural Steps
the MDOF system (assumed herein as equal to 0.05).
(1) Substitute truss members for all passive dampers.: Given the
building frame model and the type, amount and location of pas- heq = hd + hh + 0.05 (5)
sive dampers to be added, convert each passive damper’s force-
Denoting the deformation vector at the corresponding pushover step
deformation model into an elastic truss member model, i.e. into an
at the intersection point as {x}, hd may be calculated by modify-
equivalent linear stiffness Kt , and substitute it in place of the damper.
2)4)5)
ing the equation of the modal strain energy method for static
The substitute truss member is approximated based on the concept
3)
analysis ,
of equal energy considerations for the damper at its story location. P P
Ft j ∆x j
j j Kt j ∆x j
2

The method used for approximating Kt for various types of passive hd = P = P (6)
2π i Fi xi 2π i Fi xi
damper is discussed later.
for stories j with passive dampers added. Ft j are the axial forces in
(2) Perform nonlinear pushover analysis.: Push the frame-damper
the truss members which may be expressed in terms of the equivalent
model setup from Step (1) incrementally to obtain the load-
truss stiffness of the passive damper/s at story j, Kt j . Fi and xi are the
deformation relationships and capacities for each story. The initial
total pushover story force and story displacement for all the stories i
load pattern used may be based on the structure’s fundamental mode
respectively.
shape or on those prescribed by design specifications.
For a yielding building frame, the hysteretic energy dissipated by the
(3) Convert story capacities into a SDOF capacity curve.: Story capac-
dampers at the yielding k stories may be expressed by hh ,
ity curves of the MDOF model obtained in Step (2) are converted
P
into a capacity curve of the equivalent single-degree-of-freedom 2 k (Qy x − Qmax xy )
hh = P for k stories with µ > 1 (7)
π i Fi xi
(SDOF) system responding in the first mode. If the structure is
pushed with a load Fi at story i, values of the SDOF system proper- where Qy and xy are the story yield strength and story displacement
ties are calculated at each step using the equations below respectively and Qmax = Qy [1 + ρ(µ − 1)] is the maximum shear for
P bilinear systems with post-yield stiffness coefficient ρ and ductility
( mi xi )2
M1 = P (1)
mi xi 2 factor µ = xmax /xy = x/xy .
P
mi xi 2 (b)Finding the right intersection point involves iteration of Steps (4)
D1 = P (2)
mx
P i i and (5). Compare the calculated equivalent damping ratio of the
Fi
A1 = (3) structure heq to the value of the damping ratio used for the seismic
M1
Viscous (VD) Viscoelastic (VE) Hysteretic (HD)
(a) Force-displacement relationships for each damper type (b) Force-displacement relationship of
substitute truss member

(c) Parameters for each damper type


Fig. 2 Truss substitution for each type of passive damper

demand hEGM . Iterate until the two values are approximately the
2.2. Derivation of Substitute Truss Member Stiffness Kt for Passive
same. Dampers
Other response parameters, such as velocity and acceleration, may In the conventional static approach, the effect of viscous damping, or
also be obtained for the MDOF system from the state of the structure the velocity-dependent characteristics of the added dampers, cannot be
at the final pushover step used. taken into account in the model since the analysis is static. Instead, seis-
(6) Perform eigenvalue analysis at the pushover load step where heq ≈ mic demand is reduced based on the equivalent damping ratio which in-
hEGM to get higher mode shapes. Repeat pushover analysis using cludes the viscous damping due to the passive dampers. However, this
higher mode shape as lateral load coefficients.: The load pattern simplification assumes that the damper is distributed such that the en-
used to push the structure is determined based on the structural state tire frame-damper system essentially retains a proportional or classical
at peak response in the fundamental mode; that is, perform eigen- damping, which severely limits the possible damper placements over the
value analysis at the peak response load step and use the resulting height of the structure. Thus, it also limits the possible response reduction
mode shape {φn } vector at mode n multiplied by the story masses {m} capabilities of the dampers.
for the incremental lateral load pattern. Since the equivalent modal In order to overcome this limitation, models for the viscous damping
damping ratio heqn may directly be determined from the equations and energy dissipation effects of passive dampers are explicitly added to
below using the modal frequency ωn (or T n ) obtained from eigen- the structural model used for static pushover analysis. Equivalent lin-
value analysis, iteration is not necessary. ear stiffnesses, approximating the passive damper response reduction ef-
P fect, are substituted in place of the force-displacement model of the pas-
Kt j ∆x j 2
j
hdn = P (8) sive dampers using equal energy considerations (Procedure Step (1)). In
2πωn i mi xi 2
2

heqn = hdn + hhn (9) this manner, the proposed procedure expands on the applicability of the
static analysis method to include non-proportional passive damper distri-
To obtain the peak story displacements, the structure is pushed until butions over the stories; that is, it allows for a more arbitrary placement
xn , taken as point (T n , heqn ) of the seismic demand response spec- of dampers to multistory structures.
trum, and defined as the SDOF response displacement associated In this paper, the elastic truss member stiffness Kt is approximated by
with mode n. equating the energy dissipation per cycle of the damper to the energy
Peak story drifts from each higher mode are combined by adding dissipated by an equivalent truss member in the pushover analysis. For
the signed story drifts for both positive and negative signs (based the force-displacement loops of the three types of passive damper models
on the static correction method 6) ), assuming positive story drifts for considered in Fig. 2(a), energy dissipation per cycle represented by the
the fundamental mode response, and choosing the one yielding the shaded areas are each equated to the energy under the curve of the truss
largest drift values in absolute terms. in Fig. 2(b). Equations for Kt can then be derived. These are presented
  in Table 1 for each of the damper types considered where ω is the natu-

X
 N 


{∆x} =  ±∆xn  (10)


n=1

 ral circular frequency of the elastic structure. K sec is the secant stiffness
max
(= Fdy [1 + ρ(µ − 1)]/xdy µ), derived from the geometric relationship of
This particular modal response combination method was chosen be- the hysteretic damper properties of shear yield strength Fdy , yield dis-
cause it recognizes the relative signs of the lateral load and its corre- placement xdy and post-yield stiffness coefficient ρ, and is calculated by
sponding response, thus avoiding overestimations to be found in the approximating the ductility factor µ from the calculated peak base shear
more common SRSS method.
Table 1 Truss stiffness Kt used Table 2 Member cross-section properties
Damper Equation used to Concrete Web reinforcement
type approximate Kt Story Section fc 0 Reinforcement fy Spacing fwy
VD Kt = 2πCd ω (mm) (MPa) (MPa) (mm) (MPa)
p
VE Kt = π Kd 2 + (2Cd ω)2 COLUMNS Inner Outer
3-4 600x600 24 16φ25 12φ25 350 2φ13@100 300
HD Kt = 8Ksec
1-2 600x600 24 16φ25 16φ25 350 2φ13@100 300

4-story
BEAMS 1st row 2nd row
of the seismic demand response spectrum. 2-4 400x700 24 4φ25 - 350 2φ13@200 300
The truss substitution Step (1), which makes use of a simple and com- 1 400x750 24 4φ25 1φ25 350 2φ13@150 300
COLUMNS Inner Outer
mon concept in seismic response analysis, namely the equivalent lin-
9-12 850x850 24 16φ29 16φ29 400 2φ13@100 350
earization method, and yet has a potentially wide application. It may be 5-8 850x850 30 16φ32 16φ32 400 2φ13@100 350
applied to any passive damper type as long as the damper can adequately 3-4 850x850 36 16φ32 24φ35 400 3φ13@100 350
be represented by a linear model. Hence, the use of the simple truss sub- 2 850x850 36 20φ35 24φ35 400 3φ13@100 350
1 850x850 36 20φ35 28φ35 400 3φ13@100 350

12-story
stitution offers a simple approximation of the damper energy dissipation
BEAMS 1st row 2nd row
contribution compared to the more complex nonlinear MDOF model used 11-12 500x800 24 4φ25 - 350 3φ13@150 350
for the building structure. Note however, that because of this simplicity, 9-10 500x800 24 4φ25 1φ29 400 4φ13@150 350
7-8 500x800 24 4φ29 2φ29 400 4φ13@150 350
several assumptions were made regarding the passive damper response,
5-6 500x800 30 4φ32 2φ29 400 4φ13@150 350
such as (1) steady-state response, (2) equal energy dissipation for each 4 500x800 36 4φ32 2φ29 400 4φ13@150 350
damper regardless of story location, (3) essentially constant modal prop- 1-3 500x800 36 4φ32 3φ29 400 4φ13@150 350
erties for VD and VE, and peak base shear for HD.

3. ANALYTICAL MODELS AND GROUND MOTIONS USED


3.1. Building Models
A 4-story and a 12-story fixed-based regular reinforced concrete (RC)
building, designed to conform to the Building Standard Law of Japan7) ,
were used to obtain frame models for analysis. Common structural plan
and elevations for the two buildings are given in Fig. 3, while beam and
column member cross-section properties are tabulated in Table 2. The
buildings are modeled as nonlinear MDOF structural systems with hys-
teretic behavior after yielding. An inner frame is converted in a lumped-
mass shear-type model using static pushover analysis. Specifically, story
load-deformation curves are obtained through pushover analysis, which
then converted into bilinear elastoplastic backbone curves at each story
by considering equal energy under the curves until the limit displacement.
Story force-deformation curves and conversion to a shear-type system
story bilinear for the 4-story model are illustrated in Step (2) Nonlinear
Pushover Analysis of Fig. 5. The yield base shear coefficients obtained
for the 4- and 12-story structures are 0.59 and 0.26 respectively, while Fig. 3 Building frame plan and elevations

eigenvalue analyses gave the fundamental periods as 0.53sec and 0.89sec


respectively.

3.2. Earthquake Ground Motions (EGM)


Eleven seismic acceleration records from actual strong-motion earth-
quakes, as well as two artificial accelerogram provided by the Building
Center of Japan for seismic analysis purposes, are scaled to correspond
to a peak ground velocity (PGV) of 50cm/s and used as input motions in
the dynamic analyses. Record sources, with their original peak ground
acceleration (PGA) and PGV, are listed in Table 3 while their pseudo-
acceleration spectra are shown in Fig. 4. Time-history analysis of the Fig. 4 Pseudo-acceleration spectra and the mean for all EGM used
buildings under the set of EGM provided a fairly large scatter of story
drift responses while generally producing story yielding at the earthquake 3.3. Passive Damper Models

level used. Three simple types of passive dampers, namely the viscous damper
(VD), viscoelastic damper (VE) and hysteretic damper (HD), were stud-
Table 3 List of seismic records used (scaled to PGV= 50cm/s) ing the effective damping ratio hd based on the concepts of the modal

PGA PGV strain energy method. This method approximates hd with the assumption
ID Earthquake
(g) (cm/s) that the frequencies and mode shapes of the passively damped structure
ELC El Centro NS, Imperial Valley, 1940 0.35 33.5 is identical to the original building structure with the effect of shear stiff-
SCT SCT N90W, Mexico City, 1985 0.17 60.4
ness from passive dampers added. Thus, conventional eigenvalue analy-
SYL Sylmar Hospital, Northridge, 1994 0.84 129.00
sis could be used to determine the modal properties of the structure.
KBE KobeJMA NS, HyogokenNanbu, 1995 0.84 45.60
Using modal shapes and frequencies of the frame-damper model, the
HCH Hachinohe EW, Tokachi-oki, 1968 0.19 27.50
damping ratio for each mode n, hdn , for VD and VE may be expressed as
TFT Taft EW, 1952 0.18 17.75
PAC Pacoima Dam, San Fernando, 1971 1.08 57.60 P P
j C d j ∆φn j Ksec j ∆φn j 2
2
j
KSR Kushiro-oki JMA EW, 1993 0.65 41.45 hd = P P = (11)
2ωn i mi φni 2ωn 2 i mi φni 2
2

TOH Tohoku Univ NS, Miyagiken-oki, 1978 0.26 37.20 ∆φn j = φn j − φn( j−1) ∆φn1 for the first story (12)
IZT Izmit, Kocaeli, 1999 0.22 29.80
SKR Sakayra, Kocaeli, 1999 0.38 79.50 where j extends over the stories provided with dampers and i extends over
BCJ1 Artificial, provided by the BCJ 0.21 28.70 all the story masses. For a target hd , the damping coefficient Cd parameter
BCJ2 Artificial, provided by the BCJ 0.36 62.10 for VD may be obtained directly using Eq. 11, while Cd and storage
stiffness Kd parameters for VE requires iteration for varying values of
ied. VD represents a velocity-dependent damper type, VE represents Ksec and eigenvalue analysis. Storage stiffness Kd = Cd ω is assumed
dampers possessing both viscous damping and stiffness, and HD repre- herein.
sents a displacement-dependent damper. Their force-displacement loops For the case of HD, damper amount is expressed by the shear yield
and parameters are shown in Fig. 2(a) and 2(c) respectively. In model- strength Fdy . Parameter Kd may be obtained for a target hd by calculating
ing the frame-damper system, several assumptions were used to simplify Ksec from Eq. 11 and deriving from the bilinear hysteretic curve shape
and focus the investigation. Other factors that, in reality, affect the seis- of HD, assuming a post-yield stiffness ratio of 0.01, limit ductility ratio
mic performance of passive damping devices, such as temperature and of 5 and setting the damper yield displacement to 1/800rad to ensure
loading frequency were disregarded. Also, bracing and other types of yielding at low displacement levels. Using Kd , Fdy may be calculated
connecting members are neglected in the study. Dampers are added par- for an assumed peak base shear. Here, peak base shear is estimated by
P
allel to the frame to act against seismic forces at each story level and are Fb = S pa W, using the structure’s total weight for W and mean peak
assumed to have negligible mass. Moreover, the bending moment forces pseudo-acceleration at its fundamental period for S pa as given in the plot
are also disregarded by the assumption of a shear-type building model. in Fig. 4. This base shear is reduced by a factor Fh to account for the
Damper Model Quantification hysteretic damping effect as per Japanese code,
A commonly used approach in analyzing the energy dissipated by 1
Fh = (13)
passive damping devices is by expressing it in a corresponding viscous 1 + 10hd
damping ratio. Hence in this study, the “damper amount”, i.e. damper Therefore, the required HD shear yield strength at the first story is
parameters, of passive dampers added to a structure are quantified by us-
Fb Fh 1
Fdy1 = · (14)
1 + ρ(µd − 1) 1 + 1/k

Step (1) Truss Member Substitution Step (2) Nonlinear Pushover Analysis Step (3) SDOF conversion

Step (4) Capacity-Demand Diagram Step (5) MDOF Response Estimate Step (6) Higher Mode Response (heq = 0.02)

Fig. 5 Example of estimation procedure: 4-story with VD model under ELC ground motion
the error in estimates are due more to analytical approximations of the
conventional nonlinear static analysis rather than the approximations in-
troduced by the truss substitution step in the procedure. An explanation
may be found from the fact that only one mode was used in the esti-
mation since better results were obtained for estimations considering 3
modes. Overall, relatively good estimates were obtained from the es-
Fig. 6 Story-wise passive damper configurations UNIF and CENT timation procedure, based on the comparison of mean peak story drifts
obtained under all EGM of the estimation procedure and dynamic analy-

based on the frame-damper geometric relationships and where relative sis.

damper stiffness ratio is k = Kd /K f (= 1) and post-yield stiffness ratio is The effect of considering the contribution of higher modes is studied

ρ = 1/(1 + k). Required yield strengths for the upper stories Fdyi may next. In Fig. 8, partial results using 1 mode and 3 modes in the estimation

be obtained using the same equation by assuming a proportional stiffness procedure are presented for the 12-story model with VE added (hd = 2%)

distribution, i.e. by replacing the base shear Fb with shear at each story under the SCT and HCH ground motions. Using 3 modes considerably

Fi = Fb (Qyi /Qy1 ) where Qyi is the yield strength of the original building improves the estimates for the HCH record. On the other hand, plots

frame at each story i. like that shown for the SCT record show that in general, for EGMs with

A graphical example of the estimation process for one combination of relatively good results using only 1 mode, adding higher mode responses

building-damper model and EGM input is shown in Fig. 5. causes an overestimation of the upper story responses.
Mean results under all EGM records are presented in Fig. 9 for the
4. VALIDATION OF THE ESTIMATION PROCEDURE 12-story model with various passive damper types using 1 mode and 3
Validation of the estimation procedure was conducted in two parts, modes. Fig. 9 show that generally good estimates were obtained with
considering (1) a proportional passive damper distribution (UNIF) and VE using only 1 mode and with VD using 3 modes, but considerable un-
(2) a non-proportional distribution (CENT). For the UNIF configuration, derestimation occurred for HD even with 3 modes considered. Consider-
the story-wise distribution of dampers to the MDOF frame system was ing higher modes is commonly done to resolve underestimation in upper
conducted such that proportional viscous damping may be assumed in story response. The difference in the number of significant modes needed
the structure. This means that for VD, dampers are distributed to all sto- for VD and VE may be attributed to these damper’s difference in ef-
ries with a constant Cd , while for VE and HD, dampers are distributed fect on the dynamic response. Maximum dynamic response of the frame
such that the relative story stiffness distribution is the same as the orig- with VE is essentially due to the first mode, which is why adding more
inal frame (k is constant). For the CENT configuration, the same total response modes gives an overestimation. Similar plots for the 4-story
“amount”of damper used in the UNIF configuration is distributed. How- model show better overall results for all dampers. In general, considera-
ever, larger amounts are provided to the central or middle-third stories as tion of 3 modes gives better estimation although some overestimation is
graphically shown in Fig. 6. Specifically, dampers were reduced by half seen at upper stories when 1 mode is sufficient for estimating response.
on the lower- and upper-third stories, which is then redistributed equally For HD, the observed underestimation for many EGM cases may be
to the central stories. attributed to the consideration of the entire cyclic energy in the truss sub-
Peak story drifts based on the dynamic time-history response analyses stitution step, which overestimates the damping ability of the HD depend-
were considered as the “correct” response values with which the esti- ing on the EGM. Hence, the simple truss substitution applied for HD in
mated story drift values based on the proposed procedure were compared. this study needs to be modified to account for cyclic variations brought
about by different EGM properties.
4.1. Frames with Proportional Damper Configuration UNIF
Partial results of the analyses are shown in Fig. 7 for the 4-story model 4.2. Frames with Non-Proportional Damper Configuration CENT
with only one mode considered in the estimation procedure. In the fig- Investigations conducted on frames with the non-proportional CENT
ure, the lines show dynamic analysis results (DYN) while the squares damper configuration yielded similar observations and estimation results
show estimated values (EST). Dotted lines and hollow squares are the re- as the previous section for UNIF configuration. A significant difference
sults obtained for the frame alone (no dampers added), while solid lines between UNIF and CENT results is in the need to include more modes
and filled squares are the results for the frame with dampers added (with in the estimation procedure for CENT configuration, even for the 4-story
damper amount hd = 5%). model.
Fig. 7 shows good estimates of peak story drifts of the frame with In Fig. 10, mean peak story drift results for the 12-story model with
dampers under BCJ2 and SCT ground motions. On the other hand, re- VE for the set of EGM used are presented for comparison of estimates
sponse estimates were not good in some cases obtained under the ELC using 1 mode and 2 modes. Although reponses were computed for up to
and HCH ground motions. However, the discrepancy in estimates found 3 modes, consideration of the third mode in the mean response of the 12
for these ground motions are observed both for the frame alone (no story model is not so different to that of 2 modes and is thus omitted in
dampers added) and for the frame with dampers added, suggesting that this plot. It can be seen that the estimates were considerably better for
the combination of 2 modes in the lower- and upper-third stories, where
there are small amounts of dampers. On the other hand, estimates for the 5. CONCLUSION

central stories, where there are a large amount of dampers, are generally A modified nonlinear static analysis method for estimating peak story
better using only 1 mode. These results suggest that the number of sig- drifts in buildings with passive damping systems under seismic excita-
nificant modes in the response is closely related to the relative amount tions was proposed. The procedure is applicable to the analysis of ar-
of damper in the story, i.e. the less the relative amount of damper in the bitrarily located dampers to multistory structures, i.e. non-proportional
story, the more contribution can be observed from higher modes, and vice passive damper distributions over the stories. This is achieved by sub-
versa. stituting truss members with linearly elastic stiffness for the passive
For response estimation purposes, considering variable number of sig- damper locations in the structural model used for static pushover step.
nificant modes per story depending on the relative amount of damper Case studies of 4-story and 12-story regular RC buildings modeled as
added may considerably increase the accuracy of the method. However, MDOF shear-type frames with various types, amounts and story loca-
further studies in this direction, as well as refinement in the calculation of tions of passive dampers were conducted in order to study the effect of
the equivalent linear model used in the truss substitution step, are needed. passive dampers on the peak story drift distribution of the seismic re-
sponse. Three types of passive dampers, namely the viscous, viscoelastic
and hysteretic dampers, were considered.
Validation of the method showed promising results although higher

BCJ2

SCT

ELC

HCH
(a) VD (b) VE (c) HD
Fig. 7 Comparison of peak story drifts from the estimation procedure and from time history response analysis: 4-story
model with various damper types under different EGM using 1 mode
HCH
Fig. 10 Mean peak story drifts from the estimation pro-
cedure and from time history response analysis: 12-story
model with VE in CENT configuration

sive damper in the story. Further studies into the use of the procedure
would include the establishment of the appropriate number of modes to
be used and the conversion of the passive damper model into an equiva-
lent linear model.
SCT
(a) 1 mode (b) 3 modes References
1) Federal Emergency Management Agency (FEMA). “NEHRP guide-
Fig. 8 Comparison of peak story drifts from the estimation lines and commentary for the seismic rehabilitation of buildings.”Re-
procedure and from time history response analysis: 12-story ports No. FEMA 273 and 274. Washington, DC: 1997.
2) Constantionou MC, Soong TT, Dargush GF. “Passive energy dissipa-
model with VE under different EGM using 1 and 3 modes tion systems for structural design and retrofit.” MCEER Monograph
No. 1. Multidisciplinary Center for Earthquake Engineering Research,
accuracy may still be obtained with other considerations in the truss 1998.
3) Kasai K, Fu Y, Watanabe A. “Passive control systems for seismic
modelling. For buildings with HD, the procedure underestimated for damage mitigation.” Journal of Structural Engineering 1998; 124(5):
501-512.
many EGM cases. This may be attributed to the consideration of the 4) Chang KC, Soong TT, Lai ML, Nielsen EJ. “Viscoelastic dampers
entire cyclic energy in the truss substitution step, which overestimates as energy dissipation devices for seismic applications.” Earthquake
Spectra. 1993; 9(3): 371-387.
the damping ability of the HD depending on the EGM. Hence, the sim- 5) Lobo RF, Bracci JM, Shen KL, Reinhorn AM, Soong TT. “Inelas-
tic response of R/C structures with viscoelastic braces.”Earthquake
ple equations used in this study for the simple truss substitution in the Spectra.1993; 9(3): 419-446.
procedure still needs to be refined, particularly for the HD equation used 6) Chopra AK. “Dynamics of structures: theory and applications to
earthquake engineering.” Prentice-Hall Inc., 1995.
in this study, to account for cyclic variations brought about by different 7) Architectural Institute of Japan.“Design guidelines for earthquake re-
sistant reinforced concrete buildings based on inelastic displacement
EGM properties.
concept.” In Japanese. 1999.
For non-proportional damper distributions, the number of significant
modes was also found to be closely related to the relative amount of pas-

1 mode

3 modes
(a) VD (b) VE (c) HD
Fig. 9 Mean peak story drift comparison for the estimation procedure and time history response analysis: 12-story model
with various damper types using 1 and 3 modes

You might also like