Professional Documents
Culture Documents
Studies On Damping Behavior of Vertically Mixed Structures With Upper Steel and Lower Concrete Substructures
Studies On Damping Behavior of Vertically Mixed Structures With Upper Steel and Lower Concrete Substructures
net/publication/318363169
CITATIONS READS
12 640
3 authors, including:
Xiangdong He
University of Utah
5 PUBLICATIONS 50 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Dynamic load transfer rule of track-subgrade laminated system under high speed railway View project
All content following this page was uploaded by Xiangdong He on 24 May 2018.
RESEARCH ARTICLE
1
State Key Laboratory of Disaster Reduction
in Civil Engineering, Tongji University, Summary
Shanghai, China This paper presents 2 models to simulate the damping behavior of a 12‐story vertically mixed
2
Research Institute of Structural Engineering structure with upper steel and lower concrete substructures (S/RC structure). One is the modal
and Disaster Reduction, Tongji University, strain energy damping model based on modal damping ratio. The other is the assembled Rayleigh
Shanghai, China
damping model based on damping matrix that is obtained by combining Rayleigh damping matrix
Correspondence
of steel components with that of concrete ones. Then a 12‐story S/RC frame and a 12‐story RC
Ying Zhou, State Key Laboratory of Disaster
Reduction in Civil Engineering, Tongji frame are designed and used for shaking table test. Based on the test, the expressions for the
University, Shanghai, 200092, China. damping behavior of steel and concrete substructures of the S/RC frame are derived, and these
Email: yingzhou@tongji.edu.cn
expressions are utilized to form 2 different damping models of the S/RC structure separately. By
Funding information
comparing the damping behavior of the 2 models in analysis with what has been identified in the
National Key Research and Development
Program of China, Grant/Award Number: tests, the feasibility of the 2 models are assessed in both frequency domain and time domain.
2016YFC0701101 Theoretical analysis and experimental results show that the assembled Rayleigh damping model
is not applicable to conventional modal analysis because of its nonproportional characteristics.
However, the modal strain energy damping model based on equivalence of the dissipated modal
strain energy of structure, which embodies the essence of damping, can give better predictions
on damping behavior of the S/RC frame in both time and frequency domains. Finally, some sug-
gestions are put forward on the selection of damping parameters in practical seismic design for
vertically mixed structures.
KEY W ORDS
complex mode, damping behavior, modal strain energy, Rayleigh damping, vertically mixed
structure
1 | I N T RO D U CT I O N
High‐rise buildings have been increasingly put into use as urbanization is progressing. To enhance the ability of resisting winds and earthquakes for
tall building structures, measures of vibration control[1–7] and developing new kinds of structural system[8–11] are constantly taken. In recent
decades, various steel‐concrete mixed structural systems have been increasingly employed in practical engineering applications in order to satisfy
diversified demands. Vertically mixed structure, usually consisting of a lower concrete structure and an upper steel structure, has been gradually put
into use. The vertically varied lateral stiffness of this structural system can well meet deformation and load demands along the height and easily
make big rooms in upper stories of high‐rise building a reality. Therefore, this structural system has already been successfully constructed in several
major projects, such as Wuhan International Securities Building and Shanghai World Financial Center in China.[12]
Nomenclature: m, Mass; c, Viscous damping coefficient; k, Stiffness; F(t), External force; x€ , Acceleration; x_ , Velocity; x, Displacement; α, Mass‐proportional damping
coefficient; β, Stiffness‐proportional damping coefficient; ϕi, The ith mode column vector; ωi, Frequency of the ith mode; N, Total number of vibration mode; ξi,
Modal damping ratio of the ith mode; ξj’, Equivalent damping ratio of the jth mode; [K]i, Lateral stiffness of the ith element; Hrp(s), Acceleration transfer function;
Hd(s), Displacement frequency response function; Ha(s), Acceleration frequency response function; ξdf,RC, Identified modal damping ratio of 12‐story RC frame;
(ξdf)RC, Identified modal damping ratio of RC substructure in 12‐story S/RC frame; (ξdf)S, Identified modal damping ratio of steel substructure in 12‐story S/RC
frame; (ξdf)S/RC, Identified modal damping ratio of 12‐story S/RC frame; (ξt)S/RC, Modal damping ratio of 12‐story S/RC frame in test
Struct Design Tall Spec Build. 2017;26:e1392. wileyonlinelibrary.com/journal/tal Copyright © 2017 John Wiley & Sons, Ltd. 1 of 14
https://doi.org/10.1002/tal.1392
2 of 14 LU ET AL.
However, for seismic analysis, it is quite difficult to determine damping parameter for the whole structure. This can be attributed to the sig-
nificant difference on energy dissipation mechanisms between the upper and lower substructures that are made of different materials. Apparently,
it is difficult to assess the seismic behavior of vertical mixed structure. Villaverde[13] summarized that current methods for establishing damping
models of this kind of structural system can generally be classified into two categories: the decoupled method and the coupled method. The
decoupled method divides the whole system into two subsystems with each part analyzed and designed separately. This method avoids the
damping irregularity but ignores the interaction between substructures, leading to significant huge calculation errors. The coupled method treats
the structure as an entirety, thus avoiding the decoupling error. However, the damping matrix is nonproportional, and thus, the classic mode super-
position method is not applicable to seismic design any more.
Current codes for building design, such as Code for Seismic Design of Buildings in China (GB50011‐2010)[14] and IBC,[15] do not offer com-
prehensive provisions concerning the design of vertically mixed structural system. In practice, an expedient method has been frequently imple-
mented, which calculates the structural responses by presuming the structure is made of a homogenous material and then getting the envelope
of these results for seismic design. However, this method may not cover the responses of all members and will overestimate the responses of most
members, increasing the construction expenditure.[16] Besides, the accuracy of computation and analysis as well as the safety of the structure
designed by this method still remain a problem. Therefore, in order to precisely evaluate the seismic behavior of vertically mixed structures, further
investigations on damping properties of this structural system should be comprehensively launched.
In this paper, in order to establish an appropriate damping expression, two damping models, namely, the modal strain energy damping
model (MSEDM) based on modal damping ratio and the assembled Rayleigh damping model (ARDM) based on damping matrix, are chosen
for the vertically mixed structures. A 12‐story RC frame and a 12‐story S/RC frame are then designed and tested on the shaking table at
Tongji University to evaluate these models. Both damping models are validated by comparing their capabilities to predict the experimental
results in both frequency domain and time domain. Finally, suggestions on the selection of damping parameters in practical seismic design
are put forward.
2 | D A M P I N G M O D E L FO R V E R T I C A L L Y M I X E D S T R U C T U R E S ( S / R C )
To demonstrate energy dissipation in elastic deformation stage, damping term is considered in the dynamic motion equation. Theoretically,
any nonnegative functions can be employed to present the damping behavior of structure and extend the system from conservative one to
nonconservative one.[17] For mathematic convenience, velocity‐dependent viscous damping model has been widely used in practice, as is
shown in Equation 1
m⋅x€ þ c⋅x_ þ k⋅x ¼ FðtÞ; (1)
where m stands for mass, c for viscous damping coefficient, k for stiffness, F(t) for external force. Also, x€, x_ , andxmean acceleration, velocity,
and displacement, respectively.
Different from stiffness k and mass m, damping coefficient c cannot be directly calculated via physical quantity, such as dimension, density,
elastic modulus of structure, and so forth. The damping parameters of the model are commonly derived by equating the dissipated energy
measured in the tests with what is obtained by fictitious viscous damping. Then, the equivalent damping coefficient[18] can be obtained. In the
calculation of dynamic responses, two fundamental damping expressions are commonly used: modal damping coefficient and damping matrix, both
of which could be derived from the equivalent damping coefficient.[19]
" #
Cc1 0
C¼ Cc2 þ Cs1 ; (2)
0 Cs2
where Cc2 + Cs1 represents the common part of damping matrix for concrete substructure and steel substructure.
Regular Rayleigh damping is proportional, and thus, the corresponding dynamic equation of multi‐degree‐of‐freedom system could be
decoupled using orthogonal transformation. But the assembled Rayleigh damping, as shown in Equation 2, is substantially nonproportional, namely,
nonzero exists in the nondiagonal location of the assembled damping matrix. Thus, the formed damping matrices do not meet modal orthogonality,
and the dynamic equation for multi‐degree‐of‐freedom system could not be decoupled into a series of equations of single‐degree‐of‐freedom
LU ET AL. 3 of 14
system at each mode. To be exact, modal damping coefficients (ratios) actually do not exist, and there will be substantial huge error if the assembled
Rayleigh damping is adopted in modal transformation.
In addition, another advantage of adopting regular Rayleigh damping is its controllability for the damping behavior of the structure. Figure 1
exhibits the relationship between circular frequency and modal damping ratio derived from Rayleigh damping, where α and β are the mass‐propor-
tional and stiffness‐proportional damping coefficients, respectively. Both coefficients are conventionally evolved from specified damping ratios ξi
and ξj, and natural frequencies ωi and ωj correspond to two referenced vibration modes. It is usually suggested that ωi and ωj be taken as the first
modal frequency and an appropriate higher modal frequency, respectively. Once the damping ratios at two certain modes are determined, the cor-
responding damping curve could be subsequently determined, and the damping ratios in any other modes could then be specified. As illustrated in
the Figure 1, smaller damping ratios would be induced between the two specified frequencies, whereas larger damping ratios would occur beyond
the specified frequencies. Therefore, the damping behavior of the structure could be completely defined.[20] However, this important characteristic
no longer exists in the assembled Rayleigh damping model.
Matrices transformation on the assembled Rayleigh damping to obtain the modal damping matrix are shown in Equations 3 and 4:
2ωi ξ i ϕTi Mϕi ¼ ϕTi ðac Mc þ βc K c Þϕi þ ϕTi ðas Ms þ βs K s Þϕi ; (3)
" #
1 ðac −as Þ ðac −as ÞϕTi K c ϕi ϕTi Mc ϕi
ξi ¼ as þ ω βs þ
2
þ ⋅ T ; (4)
2ωi 2ωi 2ωi ϕTi Mc ϕi ϕi Mϕi
where the bold letter means matrix; the subscript s and c separately denote steel and concrete, respectively; ϕi is the ith mode column vector; ωi is
the frequency of the ith mode. ξi is the modal damping ratio of the ith mode. As shown in Equation 4, the expression ϕTi Mϕi is the structural modal
mass of the ith mode, which is independent of structural frequency. The expression, ϕTi K c ϕi =ϕTi Kϕi , ϕTi Mc ϕi =ϕTi Mϕi , transforms the stiffness and
mass matrices of concrete system using modal matrices of global structure. Because the modal matrices do not correspond to the matrices of con-
crete system to be transformed, the modal damping matrices obtained will probably not have a specific relationship with structural frequency.
It is known that regular Rayleigh damping model is adopted for each subsystem, and thus, the damping behavior for both components is under
control by predesignating the Rayleigh coefficients. However, in ARDM, the analytical process by assembling the damping submatrices of steel and
concrete frame is substantially a mathematical calculation, involving no distinct physical background. Consequently, the damping ratios derived
from the modal transformation on the formed assembled Rayleigh damping matrices do not have definite physical meaning and are out of control.
In other words, the pre‐existing advantage, in terms of specifying the damping behavior of structures prior to dynamic calculation in regular Ray-
leigh damping matrix, cannot be fully reflected by the ARDM.
the specific form of energy and thus is suitable for coupled system with different ways of damping such as soil–structure system where the soil
dissipate energy by viscous damping and the overall structure by hysteretic damping. Then the equivalent damping ratio of the coupled system
can be worked out through the energy‐weighted function.
The modal strain energy ratio, shown in Equation 5, is actually a special case of energy dissipation ratio. This expression specifies energy as
elastic strain energy, and the modal strain energy ratio is defined as dissipated elastic strain energy to store energy in a system. Thus, a rational
equivalent damping ratio based on energy conservation has been developed.
N
∑ ξ i fϕgTj ½K i fϕgj
0
ξ j ¼ i¼1N ; (5)
∑ fϕgTj ½K i fϕgj
i¼1
where ξj' is the equivalent damping ratio of jth mode; ξi is the damping ratio of the ith element; {ϕ}j is the displacement column vector of the jth mode;
[K]i is the lateral stiffness of ith element; N denotes the number of elements.
The modal strain energy damping method, with its simple form and clear physical meanings, has been widely used by many codes, such as
Guide Specifications for Seismic Isolation Design in USA[28] and Manual for Design of Highway Bridges in Japan,[29] to calculate modal damping
ratio of vibration isolation systems. Additionally, this method has been applied in an attempt to get the damping expression of vertically mixed
structures as well.[30,31]
3 | T ES T O V E R V I E W O F 1 2‐ S T O RY S/ R C F R A M E
A 12‐story S/RC frame was designed and constructed on the basis of a standard 12‐story RC frame[32] in Tongji University, shown in Figure 2. This
frame consists of a 7‐story RC frame at the bottom, a 4‐story Steel (S) frame at the upper, and a steel reinforced concrete story in the middle serv-
ing as a transition story. The section steel in the transition story extending from the steel frame is designed to be truncated at the half depth of the
top beams within the RC frame component. In this way, potential weak layer resulting from sudden truncation could then be largely avoided. Addi-
tionally, longitudinal reinforcement in columns extends to the top of transition story, where lateral moment caused by the upper steel frame under
the external excitation exists and may lead to potential damages.
The geometric dimensions and reinforcement of RC frame within 12‐story S/RC frame are identical with those of the standard 12‐story RC
frame. The cross sections of each column are 50 mm × 60 mm in RC frame, whereas I‐section is used in steel frame; consequently, lateral stiffness
in principal horizontal directions for both substructures is different. Additionally, these columns are oriented such that the weak axis of columns in
steel frame is in the direction of strong axis of column section in RC frame and vice versa. With this orientation, the system could be characterized
as a mixed one with flexible upper parts and rigid bottom parts. At the same time, the system has relatively uniform lateral stiffness along the
height. The loading cases of the test are displayed in Table 1.
The vertically mixed frame is designed, targeting at 7° seismic fortifications. Material parameters are directly acquired from tests, and the initial
damping ratio is assumed to be 0.02. Inter‐story drift ratios for S frame and RC frame of the 12‐story S/RC do not exceed 1/300 and 1/550, respec-
tively, which meets the limitation values regulated by Chinese code. According to the results of material tests, the adjusted acceleration similarity
ratio is 2.78.
The MSEDM method and the ARDM method are chosen to be further studied in this paper. Shaking table tests are then performed on the 12‐story
S/RC to evaluate the validity and applicability of these two methods. Before numerically modeling the damping properties of vertically mixed struc-
tures, damping ratios of RC and steel substructures of this S/RC frame should be determined firstly. Therefore in this section, based on the shaking
table test results, the damping expression of RC frame and steel frame are established through frequency domain analysis, which lays foundation
for the establishment of a damping expression for the whole vertically mixed structure.
(d)
(e)
FIGURE 2 Test configurations. (a) 12‐story RC frame, (b) 12‐ story S/RC frame, (c) elevation, (d) reinforcement of 12‐story standard frame, and (e)
reinforcement of 12‐story S/RC frame (unit: mm). RC = reinforced concrete; S/RC = vertically mixed structure with upper steel and lower concrete
substructures
6 of 14 LU ET AL.
Notes. EL stands for El Centro wave, KB for Kobe wave, PGA stands for peak ground acceleration, SH for Shanghai artificial wave, and WN for white noise,
S/RC for vertically mixed structure with upper steel and lower concrete substructures .
" #
N ϕri ϕpi ϕri ϕpi
Hrp ðsÞ ¼ ∑ s2 þ ; (6)
i¼1 ai ðs−λi Þ ai s−λi
ai ¼ 2jωdi Mi ; (7)
λi ¼ −ξ i ωi þ jωdi ; (8)
qffiffiffiffiffiffiffiffiffiffi
ωdi ¼ ωi 1−ξ 2 ; (9)
where p represents point excitation and r represents the identification point; ϕri and ϕpi, respectively, represent the corresponding coordinates at
the p and r point under the ith vibration mode of the structure.
If regarding the numerator and the denominator in Equation 6 as polynomials in variable s, the transfer function Hrp(s) can then be expressed in
form of a rational fraction as
2N
∑ xt stþ1
t¼1
H ðsÞ ¼
rp
2N
; (10)
∑ yt st−1 þ s2N
t¼1
n
E ¼ ∑ ½eðsk Þ2 : (12)
k¼1
xt and yt can be numerically determined when the objective function E reaches its minimum, thereby deriving the complete expression of the accel-
eration transfer function of Equation 10.
By assuming that the denominator of Equation 10 equals to 0, N groups of complex roots, namely, Si, the complex roots in the solutions of
dynamic equation with damping, are obtained. Based on these calculated results, the identified model damping ratios can finally be derived via com-
plex modal analysis method.
As frequency response function illustrated in the Figure 3, the fitted curve is in good consistent with sample points under the first white noise
excitation in X direction, showing high identification precision.
FIGURE 3 Fitted curve of identified damping of 12‐story RC frame. FRF = frequency response function; RC = reinforced concrete
expression. Consequently, as far as the author is concerned, it would be more appropriate to adopt stress for analysis on member section level,
while use displacement for analysis on structural member level.
To characterize the relationship between drift ratio and modal damping ratio of RC frame, displacement amplitude in the corresponding vibra-
tion modal should first be obtained. The analytical procedures are as the followings:
1. Obtain natural vibration frequency ω of the frame and corresponding acceleration amplification factor Ha by frequency response function of
white noise excitation.
2. By applying Laplace transformation to dynamic equations on displacement and acceleration, respectively, the relative equation between dis-
placement frequency response functions Hd(s) and acceleration frequency response functions Ha(s) are therefore derived.[33] As shown in
Equation 13, the displacement amplification factor Hd can be obtained based on the acceleration amplification factor Ha, that can be easily
obtained during the test;
s2
Ha ðsÞ ¼ s2 Hd ðsÞ ¼ : (13)
mðs2 þ 2ξωs þ ω2 Þ
1. Because the acceleration time domain signals in base region under each white noise are basically the same, the corresponding displacement
signals could be considered unchanged. Thus, it is reasonable to substitute displacement amplification factor for relative magnitude of dis-
placement under each excitation in numerical calculations.
Identified modal damping ratio ξdf,RC (influenced by frequency reduction), ξd,RC (free from frequency reduction), and displacement amplification
factor Hd,RC are obtained by the procedure discussed above, shown in Table 2. It is observed that after the loading cases of frequently occurred
earthquake accelerograms, the natural frequency slightly decreases whereas the damping ratio increases mildly, resulting from that the painting
layer of the RC frame has some cracks. The rate of reduction of structural frequency is assumed to be proportional to that of damping ratio. There-
fore, the last column of Table 2 shows damping ratio free from the influence of frequency reduction, which is the modified damping ratio of the 12‐
story standard frame and derived by linear fitting. However, both the displacement and acceleration responses of the whole frame are at a rela-
tively low level, and the maximum strain at the bottom sections of columns is less than 200 με according to the test data, therefore the frame could
still be considered in elastic phase during the experiments.
The displacement of the RC frame has been recorded in the test, hence the corresponding displacement drift ratio can be calculated and we
substitute the displacement amplification factor for it as is mentioned before. On the other hand, the modal damping ratio has also been identified
as mentioned in the above. Consequently, the nonlinear relationship between the damping ratio ξd,RC and displacement amplification factor of RC
frame can be fitted by the binomial regression model. Additionally, in order to get the lower bound of damping ratios, data of damping ratio for the
second mode are also applied in data fitting, as can be seen in Table 2. The sample points and regression curve are presented in Figure 4.
Mapping the relationship between displacement amplification factor and the damping ratio in standard 12‐story RC frame to the 12‐story S/
RC frame structure, the damping ratio of the RC frame (ξdf)RC within the 12‐story S/RC frame can be obtained, shown in Table 3.
Note. RC = reinforced concrete; S/RC = vertically mixed structure with upper steel and lower concrete substructures; WN = white noise.
For MSEDM,
As we have mentioned before, ξ ′j , the equivalent damping ratio, is derived from test by means of rational fraction polynomial method and com-
plex modal analysis and already known. ξ RC is obtained in Section 4.2 and is also known.
fϕgTj ½K S fϕgj and fϕgTj ½K RC fϕgj can be reached by Matlab program. Therefore, ξ S can then be arrived at.
For ARDM method, the case is similar.
2 3
Cc1 0
6 7
C¼4 Cc2 þ Cs1 5; (15)
0 Cs2
Cc ¼ αc Mc þ βc K c ; (18)
Cs ¼ αs Ms þ βs K s ; (19)
where Rayleigh damping was applied. Matrix C, Cc1, and Cc2 are Rayleigh damping matrices. Therefore, the damping matrix of S frame can then be
obtained by formulas presented above. Finally, damping ratio of the steel frame is able to be derived by analyzing the damping matrix.
Because there is no painting layer encasing S frame, it could be assumed that the frequency of S frame remains unchanged after these two
white noises excitations and the damping ratio of S frame only relates to the displacement responses. Therefore, the damping ratio of S frame
under other six vibration conditions could then be obtained by displacement extrapolation based on derived damping ratios in Table 4. The whole
results are shown in Table 5.
TABLE 5 Derived damping ratio of steel frame within 12‐story S/RC frame
1 0.49 3.24/3.42
2 0.64 3.62/3.95
3 0.25 2.62/2.56
4 0.17 2.43/2.31
5 0.26 2.65/2.60
6 0.25 2.62/2.56
7 0.09 2.23/2.03
8 0.08 2.2/1.98
Note. ARDM = assembled Rayleigh damping model; MSEDM = modal strain energy damping model; S/RC = vertically mixed structure with upper steel and
lower concrete substructures.
10 of 14 LU ET AL.
5 | T I M E D O M A I N A N A L Y S I S F O R S/ R C ST R U C T U R E
To compare the reliability of the two damping models, frequency analysis has already been carried out in Section 4 to test the predictions of
damping behaviors on the basis of each damping model. In this section, the predictions based on both damping models of vertically mixed struc-
tures will be compared with results from the shaking table test in time domain.
6 | D E T E R M I N A T I O N OF D A M P I N G P A R A M ET E R I N P R A C T I C A L D E S I G N
The modal damping ratios derived from the MSEDM method, and the assembled damping matrices formed by the ARDM method, have been pre-
viously studied and validated in both frequency domain and time domain. This section will present further analysis on the selection of damping
parameters in practical design of vertically mixed structures.
LU ET AL. 11 of 14
FIGURE 6 Comparison of displacements for both damping models. ARDM = assembled Rayleigh damping model; MSEDM = modal strain energy
damping model
1 0.286 0.278 0.551 0.545 0.323 0.303 0.415 0.407 0.272 0.257 0.466 0.462
3 0.404 0.402 0.563 0.563 0.500 0.500 0.508 0.507 0.293 0.272 0.428 0.416
5 0.468 0.455 0.713 0.700 0.565 0.546 0.664 0.648 0.339 0.324 0.463 0.450
6 0.526 0.509 0.534 0.527 0.488 0.477 0.506 0.508 0.391 0.379 0.421 0.405
7 0.392 0.386 0.541 0.530 0.446 0.443 0.549 0.555 0.276 0.271 0.422 0.420
8 0.337 0.334 0.444 0.447 0.380 0.375 0.458 0.446 0.250 0.247 0.435 0.424
9 0.329 0.287 0.385 0.379 0.412 0.396 0.433 0.404 0.274 0.244 0.386 0.371
10 0.269 0.258 0.387 0.378 0.438 0.425 0.491 0.475 0.240 0.229 0.340 0.331
12 0.418 0.392 0.416 0.406 0.483 0.460 0.446 0.449 0.316 0.297 0.324 0.319
3 0.202 0.197 0.168 0.161 0.278 0.270 0.143 0.137 0.326 0.314 0.070 0.063
6 0.246 0.232 0.177 0.166 0.281 0.271 0.081 0.077 0.314 0.301 0.078 0.071
7 0.255 0.241 0.196 0.185 0.258 0.250 0.104 0.098 0.315 0.302 0.086 0.078
8 0.150 0.144 0.249 0.233 0.287 0.278 0.131 0.128 0.054 0.053 0.104 0.097
9 0.213 0.202 0.290 0.271 0.311 0.300 0.143 0.139 0.292 0.277 0.126 0.118
12 0.359 0.343 0.288 0.272 0.281 0.273 0.165 0.157 0.304 0.288 0.149 0.138
Working stress no more than 1/2 yield stress Welded steel structure 2–3
RC structure with obvious cracks 3–5
Bolted and riveted steel structure 5–7
Prestressed concrete structure 2–3
Close to yield stress Welded steel structure 5–7
RC structure 7–10
Bolted and riveted steel structure 10–15
Prestressed concrete structure (prestress not completely lose) 5–7
Prestressed concrete structure (prestress essentially lose) 7–10
LU ET AL. 13 of 14
7 | C O N CL U S I O N S
In this study, two damping models for S/RC structure, namely, the MSEDM based on modal damping ratio and the ARDM based on damping matrix,
are investigated and validated for the sake of modeling damping behavior of the S/RC structure as accurately as possible. Then, a 12‐story S/RC
frame and a 12‐story RC frame are tested on shaking table. Based on the test results, the validity of both models are evaluated by frequency
domain and time domain analysis. Finally, some main conclusions are drawn as follows:
1. Damping matrices formed by the assembled Rayleigh damping characterize nonproportional damping, and therefore, the modal damping coef-
ficients do not theoretically exist here if assembled Rayleigh damping is adopted in modal transformation. Errors would be inevitably intro-
duced by forced decoupling approach. In addition, the energy dissipation capacity of vertically mixed structures (if represented by regular
Rayleigh damping matrices) is completely controlled by mathematical calculation, implying the absence of the advantage to specify the
damping behavior of studied structure prior to the dynamic calculation for ARDM method.
2. The MSEDM method is established according to the principle of energy equivalence during vibration, it not only exhibits the essence of
damping but also can be applied to determine the damping ratio of coupled structural systems.
3. The shaking table test of the 12‐story RC frame shows that the damping ratios derived by the MSEDM method not only nonlinearly relate to
displacement but also increase with the emergence of structure damages. Based on the shaking table tests performed on the 12‐story RC
frame and the 12‐story S/RC frame, numerical relation between damping ratio and drift ratio is developed.
4. Time domain and frequency domain analysis are carried out to evaluate the applicability of the MSEDM method and the ARDM method based
on the test results from the 12‐story RC frame and the 12‐story S/RC frame. It can be concluded that MSEDM method predicts precisely in
both time domain and frequency domain whereas the accuracy of ARDM is relatively low. Therefore, the MSEDM is preferentially advised to
be applied in seismic analysis of vertically mixed structures.
5. Some suggestions are put forward on the selection of damping parameters in practical seismic design. Based on theoretical analysis, it is con-
cluded that priority should be given to damping model based on modal strain energy approach when equivalent base shear method, modal
analysis method, and time history analysis method are taken.
ACKNOWLEDGEMEN TS
Financial support from the National Key Research and Development Program of China (2016YFC0701101) is highly appreciated. This work was
also supported by the Fundamental Research Funds for the Central Government Supported Universities.
RE FE R ENC E S
[1] Z. Lu, X. Y. Chen, D. C. Zhang, K. S. Dai, Earthquake Eng. Struct. Dynam. 2016, 11. https://doi.org/10.1002/eqe.2826
[2] Z. Lu, X. L. Lu, W. S. Lu, S. F. Masri, Earthquake Eng. Struct. Dynam. 2012, 41(5), 987.
[3] Z. Lu, D. C. Wang, S. F. Masri, X. L. Lu, Smart Struct. Syst. 2016, 18(1), 93.
[4] Z. Lu, X. L. Lu, H. J. Jiang, S. F. Masri, Eng. Comput. 2014, 31(4), 810.
[5] S. M. Gong, Y. Zhou, Struct. Control Health Monit. 2016. https://doi.org/10.1002/stc.1897
[6] Z. Lu, D. C. Wang, Y. Zhou, Struct. Design Tall Spec. Build. 2017, 2. https://doi.org/10.1002/tal.1359
[7] Z. Lu, S. F. Masri, X. L. Lu, Struct. Control Health Monit. 2011, 18(1), 79.
[8] Z. Lu, X. Y. Chen, X. L. Lu, Z. Yang, Earthquake Eng. Struct. Dynam. 2016, 45(9), 1537.
[9] Y. Zhou, C. Q. Zhang, X. L. Lu, Struct. Control Health Monit. 2016. https://doi.org/10.1002/stc.1864
[10] K. S. Dai, J. Z. Wang, R. F. Mao, Z. Lu, S. E. Chen, Struct. Design Tall Spec. Build. 2016. https://doi.org/10.1002/tal.1350
[11] Z. Lu, Y. L. Yang, X. L. Lu, C. Q. Liu, Appl. Sci. 2017, 7(2), 201. https://doi.org/10.3390/app7020201
[12] X. L. Lu, J. J. Zhu, Y. Zou, J. Earthquake Tsunami 2011, 3(4), 273.
[13] R. Villaverde, J. Struct. Eng. 1997, 123(8), 1011.
[14] GB50011‐2010. Code for seismic design of buildings in China; 2010 (in Chinese).
[15] Council I C. International building code: 2009 [M]. International code council, 2009.
[16] A. V. Papageorgiou, C. J. Gantes, Comput. Struct. 2010, 88(19–20), 1124.
[17] S. Adhikari, J. Sound Vib. 2006, 293(1–2), 156.
[18] W. Huang, J. Qian, Z. Zhou, Q. S. Fu, Soil Dynam. Earthquake Eng. 2015, 72, 119.
[19] R. W. Clough, J. Penzien, Dynamics of Structures [M], 2nd ed., Computers & Structures, Inc., Berkeley, California 2003.
[20] A. C. Finley, J. Struct. Eng. 2008, 134(4), 581.
[21] A. K. Chopra, Dynamics of Structures: Theory and Applications to Earthquake Engineering [M], Pearson Education, Inc., New Jersey 2001.
[22] C. J. Chang, B. Mohraz, Comput. Struct. 1990, 36(6), 1067.
14 of 14 LU ET AL.
[23] Y. K. Wang, Seismic Analysis of Coupled Structural Systems with Non‐proportional Damping [D], Polytechnic University at New York, New York 1999.
[24] J. S. Hwang, K. C. Chang, M. H. Tsai, Eng. Struct. 1997, 19(1), 55.
[25] W. Judith, J. Struct. Eng. 2009, 135(3), 282.
[26] A. Feriani, F. Perotti, Earthquake Eng. Struct. Dynam. 1996, 25(7), 689.
[27] J. M. Roesset, R. V. Whitman, R. Dobry, J. Struct. Div. 1973, 99(3), 399.
[28] AASHTO, Guide Specifications for Seismic Isolation Design [S], American Association of State Highway and Transportation Officials, Washington, D.C.
1991.
[29] JPWRI, Manual for Menshin Design of Highway Bridges[S], Public Work Research Institute, Tsukuba, Japan 1992.
[30] K. Kawashima, H. Nagashima, H. Iwasaki, J. Jpn. Soc. Struct. Eng. 1994, 40, 953.
[31] Y. T. Xue, C. J. Wei, R. F. Sun, T. H. Shi, X. Han, Earthquake Resist. Eng. Retrofitting 2008, 30(02), 91.
[32] X. L. Lu, P. Z. Li, Y. Q. Chen, Benchmark Test of a 12‐story Reinforced Concrete Frame Model on Shaking Table [R], State Key Laboratory of Disaster Reduc-
tion in Civil Engineering, Tongji University, Shanghai 2004 (in Chinese).
[33] G. Li, J. Li, Theory and Application of Dynamic Detection of Engineering Structures [M], Science press, Beijing 2002 (in Chinese).
[34] A. P. Jeary, J. Wind Eng. Ind. Aerod. 1996, 59(2–3), 103.
[35] Q. S. Li, J. Q. Fang, A. P. Jeary, et al., Earthquake Eng. Struct. Dynam. 2000, 29(12), 1845.
[36] Q. S. Li, D. K. Liu, J. Q. Fang, A. P. Jeary, C. K. Wong, Eng. Struct. 2000, 22(9), 1216.
[37] Q. S. Li, K. Yang, C. K. Wong, A. P. Jeary, J. Wind Eng. Ind. Aerod. 2003, 91(9), 1175.
[38] N. M. Newmark, W. J. Hall, Earthquake Spectra and Design [M], Earthquake Engineering Research Institute, Berkeley, California 1982.
[39] S. Adhikari, Comput. Struct. 2000, 77(6), 625.
[40] M. Lou, G. Chen, J. Eng. Mech. 2003, 129(8), 935.
[41] X. Zhou, D. Dong, Y. P. Su, Chin. Civil Eng. J. 2003, 36(5), 30.
Zheng Lu is an associate professor of College of Civil Engineering, Tongji University. His research interests are structural control and earth-
quake engineering.
Xiangdong He is a graduate student of Tongji University in China. His research focuses on seismic design of tall buildings.
Ying Zhou is a professor of structural engineering in Tongji University. Her research interests lie in earthquake resilient building design, ground
motion selection, and structural dynamic testing technology.
How to cite this article: Lu Z, He X, Zhou Y. Studies on damping behavior of vertically mixed structures with upper steel and lower con-
crete substructures. Struct Design Tall Spec Build. 2017;26:e1392. https://doi.org/10.1002/tal.1392