Structures 29 (2021) 1699–1708

Contents lists available at ScienceDirect

Structures
journal homepage: www.elsevier.com/locate/structures

Efficient direct displacement-based seismic design approach for structures

with viscoelastic dampers
Yi Xiao, Ying Zhou *, Zhiqian Huang
State Key Laboratory of Disaster Reduction in Civil Engineering, Tongji University, Shanghai, China

A R T I C L E I N F O A B S T R A C T

Keywords: Viscoelastic damper (VED) is one of most effective structural control techniques, because it provides both
Viscoelastic damper additional stiffness and damping to the structures. Current seismic standards implement force-based methods for
Direct displacement-based design the design of VEDs, which generally require extensive iterative calculations. It also causes inconsistent risk
Design approach
degree and economic efficiency because it is displacements, rather than forces, that are directly related to
Minimum iteration
Nonlinear history analysis
structural damage. In the past decades, researchers have developed a direct displacement-based design (DDBD)
Design example methodology. The DDBD approach begins with selecting a target displacement and estimating of the equivalent
damping ratio, and then calculate the seismic demand of the structure using a substitutive equivalent elastic
system. However, because of the strong nonlinearity of the VED devices, it is difficult to appropriately proportion
the key parameters (i.e., damping, period) of the VED controlled structure system, which makes the DDBD
method ineffective. To solve this problem, this paper proposed an efficient DDBD approach that is in accordance
with the Chinese seismic design principles. The relationship between the equivalent damping and structure
period of the VED-structure system is established, so the proposed approach requires minimum iterative cal­
culations and the efficiency improves significantly. A step-by-step description of the approach is presented, in
which the two key VED parameters, namely the strain amplitude and its additional damping ratio contribution to
the structure, are determined by the target displacements. The approach is further examined and verified by a 24
story building. Nonlinear history analysis results confirm that, by using the proposed DDBD procedure, the
designed target story drift is achieved with no iterative calculations and the structural responses of the VED-
structure system satisfy the code requirements.

1. Introduction the system has a strong nonlinearity and the displacement is not
considered in the first design stage, it may not satisfy the displacement
A viscoelastic damper (VED) consists of viscoelastic material layers limits. However, only after nonlinear history analyses have been carried
assembled between steel plates [1–4]. Activated by the relative motion out can this been sort out and iteration designs are needed afterwards,
between the two plates, the viscoelastic material experiences a hyster­ which is extremely inefficient. Secondly, the force-based method designs
etic shearing deformation and generates both supplemental elastic the structures’ displacement under earthquakes to be constraint by the
stiffness and viscous damping, which are effective in reducing the performance limit, so various buildings designed using the same
deformation and other seismic responses of the building [5–8]. approach usually result in different displacement levels, leading to
Currently, the design approach for structures with VEDs in seismic inconsistent risk degree and economic efficiency [14,15].
standards adopts a force-based method, in which the strength verifica­ To address these challenges, researches have proposed to use a
tion is first performed and then followed by a displacement examination displacement based seismic design approach, because displacement is
[9–13]. For example, the USA seismic design standard ATC-40 [10] more directly related with damage than forces [14]. The Japan Society
adopts a capacity-spectrum based approach, which uses the character­ of Seismic Isolation (JSSI) manual adopts a passive control design
istic intersection point of a pushover capacity curve and a demand curve methodology based on performance curves, which indicate plots of ac­
to design the dampers, followed by a nonlinear history analysis exami­ celeration (force) reduction ratio, Ra, against displacement reduction
nation. This design approach faces several challenges. Firstly, because ratio, Rd [16]. Therefore, the JSSI approach aims to control both force

* Corresponding author.
E-mail address: [email protected] (Y. Zhou).

https://doi.org/10.1016/j.istruc.2020.12.067
Received 15 September 2020; Received in revised form 2 December 2020; Accepted 24 December 2020
Available online 13 January 2021
2352-0124/© 2021 Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.
Y. Xiao et al. Structures 29 (2021) 1699–1708

and drift responses. Based on the JSSI framework, several design ap­ • An example and verification of the approach using a 24 story frame
proaches were proposed [17–19]. However, the simplified Ra - Rd rela­ building.
tionship is developed based on Japanese design earthquakes and may be
incongruous with design principles in other countries [20]. In recent 2. Relationship between the equivalent damping and structure
decades, the direct displacement-based design (DDBD) method has period of VED-structure systems
caught vast attentions in the field of seismic design [14,21–24]. The
DDBD approach begins with selecting a target displacement, which the 2.1. Relationship among properties for VED devices
structure is designed to match under specific earthquake intensity. The
key methodology of the DDBD approach is to construct a substitutive Fig. 1 shows the configuration and typical force–displacement hys­
equivalent elastic system with equivalent structural period Teq and teresis of the VEDs. According to the Kelvin model and Fig. 1(b), the
equivalent damping ξeq that includes the effect of nonlinear plasticity, so effective stiffness Kd, the damping coefficient Cd, and the equivalent
the designed displacement can be targeted using the displacement damping ratio ξd of the VED device are [20,35]:
spectrum.
Fd0
Unfortunately, it is difficult to appropriately proportion Teq and ξeq Kd = (1)
δd0
for a structural equipped with VED devices, because it requires a
comprehensive understanding of the VED-structure system’s properties. Wd
As a result, the DDBD approach proposed in the past on design of VED- Cd = (2)
πωδ2d0
structure systems generally assume a high ξeq in the first place, and
examine if the assumption is appropriate after nonlinear history ana­ Wd Wd
lyses, which again requires inefficient iteration [25]. Other studies on ξd = = (3)
4πWe 2πFd0 δd0
applying DDBD to passive control structures also emphasizes issues
relating to effects of support flexibility, soft-story mechanism, frame- where ω is the loading frequency. Substitute Eqs. (2) and (3) to (1),
damped brace interactions, torsion and high mode effects [26–34]. gives:
In China, the seismic design of buildings adopts a two-stage design Cd π
Kd = (4)
principle [9]. The first stage designs the main structure to be elastic ξd T
under minor earthquakes (63% exceedance probability in 50 years),
where T is the loading period that equals to 2π/ω. The ξd varies with
which corresponds to the frequently occurring earthquake (FOE) in
the viscoelastic property and strain amplitude, and can also be expressed
FEMA 450 [11]. The second stage checks the system’s displacement
as the product of the damping ratio at reference strain ξd,ref and strain-
under major earthquakes (2–3% exceedance probability in 50 years),
modification coefficient μξ; similarly, Kd can be expressed as the product
which corresponds to the maximum considered earthquake (MCE) in
of the stiffness at reference strain Kd, ref and strain-modification coeffi­
FEMA 450 [11]. Therefore, the main design procedure is conducted
cient μK:
under minor earthquakes, for which the main structure is elastic. This
unique design criterion of the Chinese seismic design standards makes it ξd = μξ ξd,ref
(5)
less difficult to establish the relationship between Teq and ξeq for Kd = μK Kd,ref
implementing the DDBD approach.
For a specific type of VED developed by the second author [36,37], if
This paper presents an efficient DDBD approach for the design of
the reference strain is 100%, the regression formula for μξ and μK are:
structures with VEDs. The proposed approach is efficient because it
minimizes the requirement of iterations by appropriately proportion the μξ = 0.0375γ3d − 0.2587γ2d + 0.4256γd + 0.7903
key properties of the substitutive equivalent elastic system (i.e. Teq and 1.0575γd
(6)
μK = 0.4659 + 1.5168e−
ξeq) in the first place. The paper presents the following:
where γd is the strain amplitude of the VED and equals to δd0 divided
• Establishment of the relationship between Teq and ξeq for VED- by the viscoelastic material layer thickness h. Coefficients μξ and μK
structure systems; remain unchanged with various damper sizes and are both equal to one
• A step-by-step description of the proposed DDBD approach; and under the reference strain (100%). The comparisons between the curve
fitting results based on Eq. (6) and test results in [36] are displayed in

Fig. 1. VED.

1700
Y. Xiao et al. Structures 29 (2021) 1699–1708

Fig. 2. Comparisons between curve fitting and test results.

Fig. 2, which indicates that they agree well. mode. Therefore, the shear deformation of the j-th VED (in the i-th story)
is Δn0ϕri,msin(2πt/Tm), where ϕri,m = ϕi,m-ϕi-1,m. The energy dissipated by
all the VEDs for a cycle of variation WDE is calculated by the area circled
2.2. Relationship between the equivalent damping and structural period of by their hysteresis (ellipse-shaped):
VED-structure systems ( )
2π2 ∑ ∑ ni
WDE = Cd,j Δ2n0 ϕ2ri,m (8)
Fig. 3 shows a typical arrangement of the VEDs to structures, where Tm i j = 1
the VEDs are installed using stiff vertical connections to adjacent stories.
As a result, the shear deformation of the VEDs on i-th story, δd,i, equals to where Cd,j is the damping coefficient of the j-th VED, ni is the number
the i-th story displacement of the structure. Under a harmonic vibration of VED in the i-th story. The maximum strain energy of the VED-
corresponds to the m-th mode, the displacement of the i-th floor is: structural system WSE equals to its maximum kinetic energy:
( ) 2π2 ∑
Δi = Δn0 ϕi,m sin
2πt
(7) WSE = mi Δ2n0 ϕ2i,m (9)
Tm Tm2 i

where Δi is the displacement of the i-th floor, Δn0 is the maximum where mi is the i-th floor mass. The additional damping ratio
displacement of the roof. ϕi,m is the component of the i-th floor in the m- generated by the VEDs to the structural system of the m-th mode ξadd,m is
th mode vector. Tm is the structural period corresponding to the m-th then determined as:
( )
( ) ∑ ∑ni C ϕ2
WDE Tm i j = 1 d,j ri,m
ξadd,m = = ∑ 2
(10)
4πWSE 4π i mi ϕi,m

The structure-VED system’s performance is dominated by its first

mode with a fundamental period of T1 , the additional damping ratio
′

from the VEDs for the first mode is:

( )
( ′ ) ∑ ∑ni C ϕ2
T1 i j = 1 d,j ri,1
ξadd = ∑ 2
(11)
4π i mi ϕi,1

The Chinese seismic code [9] designs the structure under minor
earthquake, under which the main structure is required to be elastic, so
the structure-VED system’s equivalent damping ratio ξeq is:
ξeq = ξ0 + ξadd (12)

where ξ0 is the elastic viscous damping ratio of the main structure,

which equals to 0.05 and 0.02 for reinforced concrete structure and steel
structure, respectively.
Typically, the VEDs are arranged along the buildings with similar
stiffness ratio (damper’s effective stiffness over the lateral structural
stiffness) for each story [16], so the first mode shape of the structures
without VEDs is assumed to be the same with the VED-structure system
(ϕi,1 and ϕri,1). Because the main structure remains elastic and no plas­
ticity is generated, the equivalent period of the VED-structure system
(Teq) equals to its fundamental structural period. For the structure
Fig. 3. Typical arrangement of VEDs.

1701
Y. Xiao et al. Structures 29 (2021) 1699–1708

without VEDs and the VED-structure system, their fundamental period, under major earthquake achieve the ultimate strain of the VED γ du,
indicated as T1 and T1 respectively, are:
′
which can be measured by test. γd is determined as:
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅ γ du
√ ∑m ϕ2 γd = (19)
√ i i,1
μmax
√ i
T1 = 2π√∑ (13)
Ki ϕ2ri,1 where μmax = max(μi), and μi is the i-th story displacement ratio of
i
responses under major earthquakes to minor earthquakes. Assume that
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅
√ ∑
mi ϕ2i,1 the μi of the VED-structure system equals to the undamped structures,
√
√ gives:
(14)
′ i
Teq = T1 = 2π√∑ ∑ni
√
(Ki + Kd,j )ϕ2ri,1 θ0i,major
i j μi = (20)
θ0i
where Ki is the i-th story lateral stiffness of the main structure and Kd,j
where θ0i,major is the average (for all ground motions) peak inter-
is the stiffness of the j-th VED in the i-th story. According to Eqs. (4) and
story drift response of the undamped structure in the i-th story under
(5):
major earthquakes obtained by time-history analyses.
Cd,j π Eq. (19) and Eq. (20) assume that the drift response along the height
Kd,j = (15)
μξ,j ξd,ref Teq is rather uniform, so most of the VEDs implemented can achieve rela­
tively large strain amplitude (smaller than their capacity). However, if
Note that ξd,ref is the property of viscoelastic material and is consis­ several soft-story mechanisms are observed, μmax can be moderately
tent in all VEDs in the building (known value). μξ,j is a function of strain reduced. In such cases, the final design of the VEDs in the soft-story
amplitude γd (Eq. (6)), which will be targeted to a same value for all should be enhanced.
VEDs to achieve their full potential. Step 4: Establish the equivalent single-degree-of-freedom (SDOF)
Substitute Eqs. (15) and (13) to (14): system of the main structure. Because the main structure remains elastic
( ∑ ∑n i 2
) under the design earthquake intensity (minor earthquakes), its equiva­
1 1 1 Teq i ( j Cd,j )ϕri,1
= 2+ 2 ∑ 2
(16) lent period and damping ratio in the SDOF system equal to its funda­
2
Teq T1 Teq μξ ξd,ref 4π i mi ϕi,1
mental structural period T1 and elastic viscous damping ratio ξ0,
Substitute Eq. (11) to Eq. (16): respectively. The equivalent mass Meq is calculated by using the target
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ deformation shape along the building height:
ξadd ( )2
Teq = T1 1 − (17) ∑N ∑N
μξ ξd,ref Meq = mi Δd,i / mi Δ2d,i
(21)
i=1 i=1
Once the strain amplitude γd is determined from seismic performance
∑
i
objectives, μξ will be calculated from Eq. (6), then Eq. (17) shows that Δd,i = (Hk − Hk− 1 )(Rd,k θ0k )
the equivalent period of the VED-structure system Teq is determined by k=1
ξadd. Eq. (12) shows that the equivalent damping ratio ξeq is also
where Δd,i is the design displacement of the i-th floor in relative to
determined by ξadd. Therefore, Eq. (12) and Eq. (17) establish the rela­
ground. Hk-Hk-1 and Rd,kθ0k are the story height and target story drift of
tionship between Teq and ξeq, which provides the basis of the proposed
the k-th story, respectively.
iterative-free DDBD procedure.
Step 5: Determine the equivalent SDOF system characteristics of the
VED-structure system. The equivalent period Teq and damping ratio ξeq
3. Proposed direct displacement-based design approach
of the VED-structure are determined by Eqs. (17) and (12), respectively.
Because γd has been determined in step 3, the only unknown variation is
The proposed design method aims to provide detailed guidelines to
ξadd. The displacement response of the undamped structure and VED-
achieve the target performance in the design of structures under minor
structure system can be estimated using a simple spectrum analyses,
earthquake, at which the main structure remains elastic. This method­
determined as Sd(T1, ξ0) and Sd(Teq,ξeq), respectively. The Chinese
ology is adopted in the Chinese seismic design criterion. Once the
seismic code [9] has provided the acceleration spectrum for various T
number and parameters of the VED devices are determined, structural
and ξ, i.e. Sa(T, ξ). The displacement spectrum value Sd(T1, ξ0) and
analyses should be conducted to examine if the code requirement is
Sd(Teq,ξeq) are then determined as:
achieved.
The step-by-step procedure is: Sd (T1 , ξ0 ) = Sa (T1 , ξ0 )gT12 /(4π2 )
Step 1: Obtain the peak inter-story drift response of the undamped 2 (22)
Sd (Teq , ξeq ) = Sa (Teq , ξeq )gTeq /(4π2 )
structure (structure without VED) in the i-th story θ0i (average value for
all ground motions) by time-history analyses of the undamped structure The displacement reduction estimated by Sd(T1, ξ0) and Sd(Teq,ξeq)
under minor earthquakes. should equal to the target Rd determined in step 2:
Step 2: Determine the target displacement reducing factor Rd under
Sd (Teq , ξeq )
minor earthquakes, which equals to the ratio of the VED-structure sys­ Rd = (23)
Sd (T1 , ξ0 )
tem response to the undamped structure response.
∑N By using Eq. (23), ξadd is solved, so as Teq and ξeq.
i=1
Rd,i Step 6: Calculate the seismic base shear force Vb, the seismic force at
Rd =
N (18) the i-th floor Fi, and the seismic shear demand at the i-th story Vi:
Rd,i = min(Θi /θ0i , 1) Vb = Meq Sa (Teq , ξeq )
∑
N
where Rd,i is the target displacement reduction factor of the i-th Fi = Vb mi Hi / mi Hi
story, N is the story number, and Θi is the target story drift, which can be i=1 (24)
selected by engineers or refer to codes. ∑
N

Step 3: Determine the strain amplitude of the VEDs under minor Vi = Fj

earthquake γ d. To do that, let the largest strain amplitude of the VEDs
j=i

1702
Y. Xiao et al. Structures 29 (2021) 1699–1708

Step 7: Design the VEDs. Assuming that the lateral resisting behavior
of stories with Θi > θ0i will not be negatively affected by the enhance­
ment in other stories, one can expect that these stories can spontane­
ously satisfy the design requirement. Therefore, the VEDs will only be
installed in stories where the Θi < θ0i, on conditions that they are ar­
ranged continuously in the vertical direction. Note adjustments may be
needed if the above assumption in this step is not true for a specific
building. In the i-th story, the additional damping ratio ξadd,i = WDE,i/
(4π WSE,i), where WDE,i equals to the energy dissipated by VEDs Wd,i,
because the main structure is elastic. According to Eq. (3) and Fig. 2,
Wd,i = 2πVd,iΔri0ξd,i, where Vd,i and Δri0 are the total shear force of VEDs
and the inter-story displacement in the i-th story, respectively. ξd,i is the
equivalent damping ratio of all the VEDs in the i-th story. The maximum
strain energy of the VED-structural system WSE,i = 1/2ViΔri0. Substitute
the formula of Wd,i and WSE,i to that of ξadd,i, gives:
ξadd,i
Vd,i = Vi (25)
ξd,i

Because the VEDs will be arranged continuously in the vertical di­

rection, the ξadd,i/ξd,i is assumed to equal to ξadd/ξd, in which ξadd has
been determined in step 5 and ξd is determined by γ d (step 3), Eqs. (5)
and (6). Note that this assumption leads to proportional stiffness and Fig. 4. Example building elevation view.
capacity introduced by VEDs to the same level in stories where VEDs are
installed. The total viscoelastic material layer area Ai in the i-th story is: Table 1. The seismic intensity region is 8, site class is 2 and design
seismic group is 2, which are seismic hazard parameters described in [9]
Vdi
Ai = (26) and define the target spectrum. The design and analysis were conducted
fd
in the short span direction (Y direction).
where fd is the damping force per unit area under strain γd at the The frame was modeled in OpenSEES using a nonlinear bending-
maximum local design temperature: shear-type Multi-degree-of-freedom model [17,38,39]. In other words,
γd a lumped mass was assumed for each story and the mechanical prop­
fd = fd,ref μK (27) erties between adjacent stories were modeled using a bending spring
100%
and a shear spring. The parameters used in the model were obtained
where fd,ref is the damping force per unit area under reference strain from a static pushover analysis and are shown in Table 1. Specifically,
(100% to be consistent with μK) at the maximum local design temper­ the bending spring was assumed as elastic (elastic material, twoNode­
ature. μK is determined by Eq. (5). Then, for a certain size of VEDs, the Link element) and the shear spring model was modelled as trilinear for
number of dampers in each story can be determined. 1–2 story or and bilinear for 3–24 story (Hysteretic material, twoNo­
The viscoelastic material layer thickness h can be determined by the deLink element) depending on story materials. The trilinear shear spring
strain amplitude γ d and the target structural inter-story deformation: is defined by five key properties (K1, K2/K1, K3/K1, Qc and Qy) whereas
Θi (Hi − Hi− 1 ) the bilinear spring is defined by three properties (K1, K3/K1 and Qy).
h= (28) Here K1 is the story elastic shear stiffness, K2 is the story shear stiffness
γd
after concrete cracking, K3 is the story shear stiffness after steel yielding,
where Hi - Hi-1 is the story height of the i-th story. Qc is the story shear capacity when concrete cracks, and Qy is the story
At this stage, the material, size, and number of the VEDs in each story shear capacity when the steel yields. The MDOF model agrees very well
are determined. The designed VED-structure system is expected to in terms of dynamic properties with a 3D model, which demonstrates the
match the target drift Θi under minor earthquakes. Then, the nonlinear MDOF model’s feasibility of nonlinear history analyses [38,39]. The
history analyses can be carried out to examine if the designed system elastic modal damping ratio was assumed to be 2%.
satisfy the code requirement. The three natural periods obtained from the modal analysis using
OpenSEES software are 3.133 s, 1.049 s and 0.623 s, which are close to
4. Design example using the proposed method those determined by SNAP software in literature [38]. Five natural
earthquake ground motion records were selected from the PEER data­
4.1. Example building information and drift performance base [44] and two artificial earthquake ground motion were generated
for the evaluation of seismic performance according to [9]. The ground
A 24-story steel frame building with a total height of 84.75 m was motions were scaled to target peak ground accelerations (PGA) of 70
selected as the example building [38,39]. The building locates in Beijing cm/s2 and 400 cm/s2, respectively, to commensurate with the PGA
and was designed by Chinese and Japanese joint design team in 1984. requirement for minor and major earthquakes of intensity 8. Fig. 6
Figs. 4 and 5 show the elevation and plan views of the 24-story building, shows the comparison of these ground motions’ acceleration spectrum
respectively. Though the main structural system is dominated by steel scaled to the minor earthquake level with the target code-specified
materials, the designer chose to use the concrete encased steel columns spectrum at the site, and it demonstrates that they agree well with the
(with a peripheral dimension of 850 mm × 850 mm) and beams (with a target code spectrum at T1, T2 and T3, as requested by the standard.
peripheral dimension of 500 mm × 950 mm) in the 1st and 2nd stories. Further, the mean spectrum is very close to the code spectrum in almost
In upper stories (3–24 story), steel box sections (ranges from 450 × 450 the entire range of periods considered (T < 6 s), indicates that the some
× 45 to 450 × 450 × 25 profile [40], unit: mm) were used for columns shifts of structural periods (caused by adding VEDs or structural plas­
and H-shaped steel sections (ranges from 650 × 300 × 12 × 28 to 650 × ticity development under major earthquakes) will not affect the feasi­
250 × 12 × 22 profile [41], unit: mm) were used for beams. The bility of the selected ground motion for seismic performance evaluation.
structural concrete grade is C30 [42] and the steel grade is Q345 [43]. The seismic performances of the structure system were determined
Story mass and height are summaries in the 3rd and 2nd column of

1703
Y. Xiao et al. Structures 29 (2021) 1699–1708

Fig. 5. Example building plan view.

Table 1
Main parameters of the frame model.
Story Mass (t) Height (m) Bending spring stiffness (kN⋅m/rad) Shear model Shear spring parameters

K1 (kN/mm) K2/K1 K3/K1 Qc (kN) Qy (kN)

24 216 84.75 1.508E + 08 Bilinear 137.9 – 0.324 – 1794

23 1067 80.55 2.298E + 09 Bilinear 403.1 – 0.460 – 5037
22 1078 76.25 5.123E + 09 Bilinear 660.1 – 0.454 – 7418
21 1073 72.95 5.654E + 09 Bilinear 697.7 – 0.454 – 9370
20 1073 69.65 5.969E + 09 Bilinear 723.1 – 0.447 – 11,215
19 1080 66.35 7.449E + 09 Bilinear 768.1 – 0.426 – 12,922
18 1083 63.05 7.537E + 09 Bilinear 788.0 – 0.428 – 14,507
17 1085 59.75 7.555E + 09 Bilinear 802.5 – 0.398 – 16,091
16 1085 56.45 7.556E + 09 Bilinear 805.1 – 0.304 – 17,499
15 1087 53.15 7.537E + 09 Bilinear 802.1 – 0.202 – 18,977
14 1086 49.85 7.506E + 09 Bilinear 801.3 – 0.131 – 20,202
13 1087 46.55 7.519E + 09 Bilinear 811.7 – 0.091 – 21,173
12 1088 43.25 7.469E + 09 Bilinear 816.5 – 0.065 – 22,125
11 1088 39.95 7.420E + 09 Bilinear 831.0 – 0.048 – 23,053
10 1091 36.65 7.351E + 09 Bilinear 845.7 – 0.038 – 23,892
9 1091 33.35 7.283E + 09 Bilinear 851.3 – 0.032 – 24,632
8 1091 30.05 7.216E + 09 Bilinear 858.8 – 0.029 – 25,345
7 1094 26.75 7.819E + 09 Bilinear 884.3 – 0.027 – 25,972
6 1097 23.45 7.736E + 09 Bilinear 893.9 – 0.028 – 26,419
5 1098 20.15 7.952E + 09 Bilinear 919.5 – 0.031 – 26,722
4 1098 16.85 7.879E + 09 Bilinear 1001.7 – 0.038 – 26,629
3 1098 13.55 7.773E + 09 Bilinear 1574.9 – 0.062 – 23,692
2 1669 10.25 1.503E + 10 Trilinear 2518.0 0.284 0.044 5238 28,847
1 1787 5.25 1.540E + 10 Trilinear 4766.5 0.330 0.060 5912 29,519

using the mean value of those obtained from the seven ground motions.
The peak inter-story drift response of the undamped structure was ob­
tained by nonlinear history analyses for each of the three ground mo­
tions. Fig. 7(a) and (b) show the drift responses under minor and major
earthquakes, respectively. As is shown, the undamped structure satisfies
the code limitation under minor earthquakes (1/550 for story 1 ~ 2
reinforced concrete frame and 1/250 for story 3 ~ 24 steel frame), but
dose not meet the code limitation under major earthquakes (1/50) [9].
To further increase the structural seismic performance and investi­
gate the feasibility of the proposed DDBD approach, a stricter target drift
was set under minor earthquakes as 1/1000 for story 1 ~ 2 and 1/500
for story 3 ~ 24. Note that the improved seismic fortification objective
will not affect the approach’s feasibility in applying to the Chinese
seismic design methodology, as long as the main frame is still elastic
under minor earthquakes, which is certainly true with a stricter target
drift. The improved performance will be achieved by adding VEDs. The
Fig. 6. Acceleration spectra of selected ground motions and code- drift limitation under major earthquakes is still 1/50.
specific spectrum.

1704
Y. Xiao et al. Structures 29 (2021) 1699–1708

Fig. 7. Peak story drift response of the undamped structure.

4.2. Design process

The seismic design method proposed in Section 3 was used for the
VED design of the example building (Figs. 5 and 6). The VEDs developed
by [36,37] was used. The ultimate strain γdu of this type of VED is 400%.
The damper’s equivalent damping ratio under the reference strain
(100%) is ξd,ref = 31%. The damping force per unit under the reference
strain (100%) at the maximum local design temperature (36 ◦ C for
Beijing) is fd,ref = 1.19 × 10-3 kN/mm2. The step-by-step procedure is as
follows:
Step 1: The peak inter-story drift response of the undamped structure
under minor earthquakes was obtained by nonlinear history analyses for
each of the seven ground motions, as is show in Fig. 7(a).
Step 2: The target story drift Θi were set to 1/1000 for story 1 ~ 2 and
1/500 for story 3 ~ 24 under minor earthquakes, as discussed in Section
4.1. Then the target displacement reducing factor Rd was calculated by
Eq. (18) as 0.844.
Step 3: The peak inter-story drift response of the undamped structure
under major earthquakes was obtained by nonlinear history analyses
and shown in Fig. 7(b), with a large deformation concentrated in story 3.
According to Eq. (20) and its related discussion, the μmax was determined Fig. 8. The relationship between Rd and ξadd for the example building.
to be 10. The strain amplitude of the VEDs under minor earthquakes
γ d was 40% by Eq. (19). Substitute γd to Eq. (6), gives μξ = 0.92 and μK = Section 3. The final number of VEDs is shown in the last column of
1.46. Table 2. No VED was added in story 1 and story 21 ~ 24 because their
Step 4: For the equivalent SDOF system of the main structure, the peak drift responses already satisfied the target requirement (Fig. 7).
equivalent period, damping ratio, and mass equal to 3.133 s, 2% and
18771 t, respectively.
Step 5: Fig. 8 shows the relationship between Rd and ξadd for the 4.3. Performance examine
example building, calculated by Eqs. (12), (17), and (22) and (23). As is
shown, Rd decreases monotonically as ξadd increases. By targeting Rd to The VEDs were arranged in a typical configuration, as shown in
0.844 Step 2), gives ξadd equals to 2.6%. Substitute ξadd to Eqs. (17) and Fig. 3. Thus, the VEDs can be modeled by additional twoNodeLink ele­
(12), gives Teq = 2.987 s and ξeq = 4.6%. ments between adjacent stories, once their mechanical properties have
Step 6: The seismic base shear force Vb was 6418 kN by Eq. (24). The been appropriately simulated. Depending on the specific type of VEDs,
seismic force Fi and seismic shear demand Vi at each story are summa­ various mechanical models can be used, such of Maxwell, Kelvin-Voigt,
rized in Table 2. and fractional derivative models [45]. In this design example, a VED
Step 7: The total shear force of VEDs Vd,i and viscoelastic material with hybrid non-linearity was adopted [36,37] (consistent with Eq. (6)).
layer area Ai at each story are summarized in Table 2. The damping force The hybrid non-linearity is defined by the fact that the applying VED
per unit area under strain γd at 36 ◦ C is fd = 6.92 × 10–4 kN/mm2 by Eq. contains both softening and hardening behaviors. It is found that that
(27). The layer thickness is 16 mm by Eq. (28) (the typical story height is this type of more complex non-linear properties leads to higher energy
3.3 m). The viscoelastic material layer area of each damper was selected dissipation and deformation capacities comparing with conventional
as 400 × 400 mm2. After a preliminary analysis, it is found the VED linear-type VEDs. Based on experiments and mechanical analyses, Zhou
strain amplitude at story 3 exceeds its capacity (400%), so one more VED et al. [46] and Gong [20] proposed a mechanical model of the hybrid
was added to story 3, which is consistent with the discussion in Step 3 of non-linearity VED and its implementation in OpenSEES, where the VED
material is constructed by three pre-defined material objects acting in

1705
Y. Xiao et al. Structures 29 (2021) 1699–1708

Table 2 requirements. Yet, it is found that the proposed approach, which as­
The VED design. sumes proportional stiffness of the VEDs to structures along the height,
Story Fi (kN) Vi (kN) Vd,i (kN) Ai (mm2) Number of VEDs cannot produce a uniform distribution in the vertical direction. Future
research in VED stiffness arrangement should be investigated to maxi­
24 106 106 10 13,978 /
23 499 606 55 79,605 / mize VEDs’ potential in structural vibration control.
22 478 1083 99 142,369 /
21 455 1538 140 202,139 / 4.4. Discussions
20 434 1972 179 259,204 2
19 416 2388 217 313,921 2
18 397 2785 253 366,060 2
(1) Computational requirement
17 377 3162 288 415,562 2
16 356 3518 320 462,330 3 The proposed approach aims to nearly eliminate iterations and
15 336 3853 351 506,445 3 design VED-structure systems efficiently. However, in step 1 and step 3,
14 315 4168 379 547,783 3
nonlinear history analyses are used to obtain the undamped structure
13 294 4462 406 586,420 3
12 273 4735 431 622,351 4 story drift response under minor (θ0i) and major earthquakes (θ0i,major),
11 253 4988 454 655,540 4 respectively. For complex buildings, this might be computational
10 232 5220 475 686,072 4 demanding. In such cases, during the preliminary design stage, elastic
9 211 5431 494 713,855 4
response spectrum analysis can be used to determine θ0i, and θ0i,major can
8 190 5622 512 738,888 4
7 170 5792 527 761,234 4
be estimated by the drift obtained by elastic spectrum analysis multi­
6 149 5941 541 780,877 5 plied by a deflection amplification factor, as per seismic codes such as
5 129 6070 552 797,771 5 GB 50011–2017 [9] and ASCE 7–16 [48]. In step 7, nonlinear history
4 107 6177 562 811,898 5 analyses are used to obtain the story drift response of the VED-structure
3 86 6264 570 823,258 6
system under minor and major earthquakes. A simplified approach
2 99 6363 579 836,321 5
1 55 6418 584 843,485 / similar to those for the undamped structure can be adopted. Further, the
analyses of the VED-structure are merely for performance checking, and
may be not necessary in the preliminary design stage.
parallel, as shown in Fig. 9. The first pre-defined material is the viscous
dashpot model, and it is simulated using a ViscousDamper material with (2) Application for complex buildings
its stiffness parameter set to be very large. The second pre-defined ma­
terial is used to model the devices’ stiffness with the hardening effect According to [14], the torsion, P-Delta and high mode effect, which
considered. It is simulated by an Elastic MultiLinear material. The third may be nonnegligible in complex buildings, can be considered in the
pre-defined material is established to consider Mullins effect [47], and it DDBD framework by altering the target displacement profile along the
depicts softening effect with the increases of strain amplitude. In height. Their effects usually lead to enhancement in upper stories.
OpenSEES, it is simulated by use of several MinMax material controlled Meanwhile, the target displacement itself is very sensitive to structural
by stepped-up strain amplitude combined in parallel. Detailed assem­ type. For building frames, it is recommended to use a first-mode shear-
bling and material parameters in accordance with the designed VED dominated profile, in which the largest drift takes place in the first story
(400 mm × 400 mm × 16 mm) can be found in [20]. and the drift decreases with the increase of story (absolute) height. For
After adding the VED elements, nonlinear history analyses were brace frames, however, a linear profile may be assumed regardless of the
carried out for the designed VED-structure system using the same seven number of stories [14,49]. Unfortunately, so far there is no rigorous
ground motions as the undamped structure (Fig. 7). Fig. 10(a) and (b) studies of the displacement profile for a VED-structure. Therefore, in this
show the peak drift responses under minor and major earthquakes of the paper we use a displacement profile described by Eq. (21) and distribute
VED-structure system, respectively. Fig. 11 directly compares the mean base shear force to each story by Eq. (24), which enhances the design of
peak drift responses of both the undamped structure and VED-structure upper stories comparing with the building frame profile, so it might
system under minor and major earthquakes. include some effects of torsion, P-Delta and high modes. Once a rigorous
The results confirm that, by using the proposed DDBD procedure, the displacement profile is verified for VED-structures, it should replace Eq.
target story drift is achieved under minor earthquakes, and the struc­ (21) and those modifications to include torsion, P-Delta and high mode
tural responses of the VED-structure system satisfy the code effects, as described in [14], should be adopted. Further, placing VEDs in
the perimeter of buildings will help reduce torsional effects.

(3) Including VEDs’ support flexibility

The current proposed approach assumes that the stiffness of the

VEDs’ support to structures is infinite. In a typical VED arrangement
configuration, as shown in Fig. 3, this may be appropriate. However,
various studies have found that the effect of the dampers’ support is not
negligible, especially for those with a brace-type support [26–32]. The
support flexibility reduces the lateral capacity introduced by VEDs by
reducing their shear deformation in relative to story displacement. To
account for support flexibility, one can refer to [26] by either deter­
mining the minimum support stiffness or reducing the maximum
achievable damping in design.

5. Summaries and conclusions

An efficient direct displacement-based design (DDBD) approach was

proposed for structures with viscoelastic dampers (VED). The approach
Fig. 9. Assembling of OpenSEES material for modeling VEDs. is consistent with the Chinese seismic design principle. The relationship

1706
Y. Xiao et al. Structures 29 (2021) 1699–1708

Fig. 10. Peak story drift response of the VED-structure.

Fig. 11. Comparison of Mean peak drift response between undamped and VED-structure system.

between the equivalent damping and structure period of the VED- investigate the approach’s feasibility. Results show that, by using the
structure system was established, which helps to proportion key pa­ proposed DDBD procedure, the target story drift is achieved under minor
rameters in the DDBD procedure and nearly eliminate iterations. The earthquakes, and the structural responses of the VED-structure system
step-by-step procedure was introduced, in which the two key VED pa­ satisfy the code requirements.
rameters, namely the strain amplitude γd and its additional damping
ratio contribution to the structure ξadd, are determined by the target Declaration of Competing Interest
displacements. Some hypotheses are assumed in this study, including 1)
the VED-structure system is assumed to have the same first mode shape The authors declare that they have no known competing financial
with the undamped structure; 2) the VED-structure system is assumed to interests or personal relationships that could have appeared to influence
have the same displacement ratio of responses under major earthquakes the work reported in this paper.
to minor earthquakes with the undamped structure; 3) the lateral
resisting behavior of one story will not be negatively affected by en­
Acknowledgements
hancements in other stories; and 4) the VED force is proportional to the
story shear force to the same level at all stories where VEDs are installed.
The authors gratefully acknowledge the financial support from The
Note that these hypotheses indicate that the proposed approach may
National Nature Science Foundation of China (Grant No. 51878502).
face some difficulties when applying to irregular buildings, such as those
with irregular plan and vertical capacity discontinuity, but for regular
References
buildings, it is found to be appropriate and efficient.
The proposed approach was examined and verified by a 24 story [1] Soong TT, Dargush GF. Passive energy dissipation systems in structural
example building. Nonlinear history analyses were carried out to engineering. Chishester UK: Wiley; 1997.

1707
Y. Xiao et al. Structures 29 (2021) 1699–1708

[2] Christopoulos C, Filiatrault A. Principles of supplemental damping and seismic [26] Losanno D, Londono JM, Zinno S, Serino G. Effective damping and frequencies of
isolation. Pavia, Italy: IUSS Press; 2006. viscous damper braced structures considering the supports flexibility. Comput
[3] Takewaki I. Building Control with Passive Dampers: Optimal performance-based Struct 2018;207:121–31. https://doi.org/10.1016/j.compstruc.2017.07.022.
design for earthquakes. Chichester UK: Wiley; 2009. [27] Losanno D, Spizzuoco M, Serino G. An optimal design procedure for a simple frame
[4] Saidi I, Gad EF, Wilson JL, Haritos N. Development of passive viscoelastic damper equipped with elastic-deformable dissipative braces. Eng Struct 2015;101:677–97.
to attenuate excessive floor vibrations. Eng Struct 2011;33(12):3317–28. https:// https://doi.org/10.1016/j.engstruct.2015.07.055.
doi.org/10.1016/j.engstruct.2011.05.017. [28] Losanno D, Spizzuoco M, Serino G. Design and retrofit of multistory frames with
[5] Lu L-Y, Lin G-L, Shih M-H. An experimental study on a generalized Maxwell model elastic-deformable viscous damping braces. J Earthquake Eng 2019;23(9):
for nonlinear viscoelastic dampers used in seismic isolation. Eng Struct 2012;34: 1441–64. https://doi.org/10.1080/13632469.2017.1387193.
111–23. https://doi.org/10.1016/j.engstruct.2011.09.012. [29] Nuzzo I, Losanno D, Caterino N. Seismic design and retrofit of frame structures
[6] Chae Y, Ricles JM, Sause R. Modeling of a large-scale magneto-rheological damper with hysteretic dampers: a simplified displacement-based procedure. Bull
foe seismic hazard mitigation Pert I: Passive mode. Earthquake Eng Struct Dyn Earthquake Eng 2019;17(5):2787–819. https://doi.org/10.1007/s10518-019-
2013;42:669–85. 00558-8.
[7] Yamamoto M, Sone T. Damping systems that are effective over a wide range of [30] Fujita K, Moustafa A, Takewaki I. Optimal placement of viscoelastic dampers and
displacement amplitudes using metallic yielding component and viscoelastic supporting members under variable critical excitations. Earthquakes Struct 2010;1
damper in series. Earthquake Eng Struct Dyn 2014;43(14):2097–114. https://doi. (1):43–67. https://doi.org/10.12989/eas.2010.1.1.043.
org/10.1002/eqe.2438. [31] Losanno D, Serino G. Parametric non-linear analysis of reinforced concrete frames
[8] Chen C, Ricles JM, Karavasilis TL, Chae Y, Sause R. Evaluation of a real-time hybrid with supplemental damping systems. Int J Struct Constr Eng 2017;11(4):434–41.
simulation system for performance evaluation of structures with rate dependent [32] G. Lomiento N. Bonessio F. Braga Design criteria for added dampers and supporting
devices subjected to seismic loading. Eng Struct 2012;35:71–82. https://doi.org/ braces SIAPS 1 1 55 73 10.2140/siaps.2010.1.55.
10.1016/j.engstruct.2011.10.006. [33] Sutcu F, Takeuchi T, Matsui R. Seismic retrofit design method for RC buildings
[9] Ministry of Housing and Urban-Rural Development, China PR. GB 50011-2010, using buckling-restrained braces and steel frames. J Constr Steel Res 2014;101:
Code for seismic design of buildings, China Architecture & Building Press, Beijing, 304–13. https://doi.org/10.1016/j.jcsr.2014.05.023.
2016. [34] Ferraioli M, Lavino A. A displacement-based design method for seismic retrofit of
[10] ATC-40, Seismic evaluation and retrofit of concrete buildings, Applied Technology RC buildings using dissipative braces. Math Problems Eng 2018;2018:1–28.
Council, Redwood City, 1996. https://doi.org/10.1155/2018/5364564.
[11] FEMA 450, NEHRP recommended provisions for seismic regulations for new [35] Chopra AK. Dynamics of structures: theory and applications to earthquake
buildings and other structures Federal Emergency Management Agency 2004 engineering. 4th edition. Upper Saddle River, NJ, USA: Prentice hall; 2012.
Washington D.C.. [36] Gong S, Zhou Y. Experimental study and numerical simulation on a new type of
[12] Lee K-S, Fan C-P, Sause R, Ricles J. Simplified design procedure for frame buildings viscoelastic damper with strong nonlinear characteristics: Tests and Simulation on
with viscoelastic or elastomeric structural dampers. Earthquake Eng Struct Dyn a new Type of VED with Strong Nonlinearities. Struct Control Health Monit 2017;
2005;34(10):1271–84. https://doi.org/10.1002/eqe.479. 24(4):e1897. https://doi.org/10.1002/stc.1897.
[13] Zhou Y, Lu X, Weng D, Zhang R. A practical design method for reinforced concrete [37] Gong S, Zhou Y, Ge P. Seismic analysis for tall and irregular temple buildings: A
structures with viscous dampers. Eng Struct 2012;39:187–98. https://doi.org/ case study of strong nonlinear viscoelastic dampers: A tall and irregular temple
10.1016/j.engstruct.2012.02.014. with strong nonlinear VE dampers. Struct Design Tall Spec Build 2017;26(7):
[14] Priestley MJN, Calvi GM, Kowalsky MJ. Displacement-based seismic design of e1352. https://doi.org/10.1002/tal.1352.
structures. Pavia, Italy: IUSS Press; 2007. [38] D. Feng H. Xia W. Liu Design of a 24-story damped steel frame based on Chinse and
[15] Kim J, Choi H. Displacement-based design of supplemental dampers for seismic Japanese building codes 2015 San Diego, USA.
retrofit of a framed structure. J Struct Eng 2006;132(6):873–83. https://doi.org/ [39] Feng D, Xia H, Liu W, Cui H. Vibration reduction analysis of a high-rise steel frame
10.1061/(ASCE)0733-9445(2006)132:6(873). structure, 10th Japan-China building Structure Technology Symposium. China:
[16] JSSI Manual 2nd Edition. Design and construction manual for passively controlled Nanjing; 2013 (in Chinese).
buildings, Japan Society of Seismic Isolation, Tokyo, Japan, 2007. [40] Standardization Administration, China PR. GB/T 6728-2017, Cold forming hollow
[17] Pu W, Kasai K. Design method for RC Building structure controlled by elasto-plastic sectional steel for general structure, Beijing, 2017.
dampers using performance curve. 15th World Conference on Earthquake [41] Standardization Administration, China PR. GB/T 11263-2017, Hot rolled and cut T
Engineering. 2012. section steel, Beijing, 2017.
[18] Kasai K, Ogawa R, Pu W, Kiyokawa T. Passive control design method for elasto- [42] Ministry of Housing and Urban-Rural Development, China PR. GB 50010-2010,
plastic frame added with visco-elastic dampers. Nihon Kenchiku Gakkai Kozokei Code for design of concrete structures, China Architecture and Building Press,
Ronbunshu 2010;75(655):1625–33. https://doi.org/10.3130/aijs.75.1625. Beijing, 2010.
[19] Kasai K, Pu W. Passive control design method for mdof slip-hysteretic structure [43] Ministry of Housing and Urban-Rural Development, China PR. GB 50017-2017,
added with visco-elastic damper. Nihon Kenchiku Gakkai Kozokei Ronbunshu Standard for design of steel structures, China Architecture and Building Press,
2010;75(650):781–90. https://doi.org/10.3130/aijs.75.781. Beijing, 2017.
[20] S. Gong Research on seismic analysis and design methods of structures added with [44] PEER Pacific Earthquake Engineering Research Center, Strong Ground Motion
viscoelastic dampers with strong nonlinearity Ph.D dissertation 2018 Tongji Database 2013.
University Shanghai, China (in Chinese). [45] Lewandowski R, Chorążyczewski B. Identification of the parameters of the
[21] Calvi GM, Sullivan TJ. Development of a model code for direct displacement based Kelvin–Voigt and the Maxwell fractional models, used to modeling of viscoelastic
seismic design. Atti di Linea IV: Convegno Finale del progetto RELUIS; 2009. dampers. Comput Struct 2010;88(1-2):1–17. https://doi.org/10.1016/j.
p. 1–3. compstruc.2009.09.001.
[22] Giannakouras P, Zeris C. Seismic performance of irregular RC frames designed [46] Zhou Y, Chen P, Gong S. Mechanical model of a hybrid non-linear viscoelastic
according to the DDBD approach. Eng Struct 2019;182:427–45. https://doi.org/ material damping device with its verifications. Front Mater 2019;6:33.
10.1016/j.engstruct.2018.12.058. [47] L. Mullins Softening of Rubber by Deformation 42 1 1969 339 362 10.5254/
[23] O’Reilly GJ, Sullivan TJ. Direct displacement-based seismic design of eccentrically 1.3539210.
braced steel frames. J Earthquake Eng 2016;20(2):243–78. https://doi.org/ [48] ASCE 7–16. Minimum design loads for buildings and other structures. Am Soc Civil
10.1080/13632469.2015.1061465. Eng, Reston, Va 2016.
[24] Noruzvand M, Mohebbi M, Shakeri K. Modified direct displacement-based design [49] Malekpour S, Dashti F. Application of the direct displacement based design
approach for structures equipped with fluid viscous damper. Struct Control Health methodology for different types of RC structural systems. Int J Concr Struct Mater
Monit 2020;27(1). https://doi.org/10.1002/stc.2465. 2013;7(2):135–53. https://doi.org/10.1007/s40069-013-0043-2.
[25] Mazza F, Vulcano A. Nonlinear seismic analysis to evaluate the effectiveness of
damped braces designed for retrofitting RC framed structures. Int J Mech 2013;7
(3):251–61.

1708