Earthquake Response of 10-Story Story-Drift-Controlled RC Frames With Hysteretic Dampers
Earthquake Response of 10-Story Story-Drift-Controlled RC Frames With Hysteretic Dampers
Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct
Fig. 1. Schemes of buildings with hysteretic dampers: (a) BRB, and (b) bending plates.
Fig. 2. Assumed interaction between the main frame and the damper system: (a) schematic combination, and (b) restoring force characteristics.
With regard to the control of deformation, previous researchers two common arrangements of hysteretic dampers in directly
have also investigated the applicability of displacement-based connected type frames [19]. Fig. 1(a) shows one case when BRBs
design approaches to framed structures with hysteretic dampers are installed, connecting the beam–column joint of the lower story
[8,13,17,18]. For instance, Teran-Gilmore and Virto-Cambray [17] with the beam of the upper story at the midspan; the connection is
introduced a displacement-based approach for the preliminary given by bolted or welded steel plates and should remain elastic
design of low-rise buildings with BRBs. They showed that the and rigid. Fig. 1(b) shows one arrangement of bending plates,
proposed methodology yields an adequate level of seismic design connected to the lower story by elastic braces and to the beams
when the areas of braces are determined as a function of the value of the upper story at the midspan. Some common devices used in
of the fundamental period required by an entire system to control this type of frame are buckling-restrained braces [7,20], steel shear
the level of damage in the main structure. Kim and Seo [8] proposed panels and steel bending plates [21–23].
a design methodology for low-rise steel frames with BRBs aimed
From Fig. 1, the restoring force of the entire system (main
at producing story drifts that are the same in all stories within a
frame + damper system) is assumed to be the combination
target displacement. In the same context, Kawano and Hirata [18]
of two springs connected in parallel so that the main frame
recently proposed a method to optimize damper strengths so that
and the damper system share the same deformation. The spring
the story-drift responses of steel frames are evenly distributed
throughout the stories. corresponding to the damper system is inelastic and the spring for
Thus, after identifying possible new directions on the control of the main frame can be either elastic or inelastic. Fig. 2(a) shows the
deformation of a R/C main structure with hysteretic dampers, an scheme of this assumption.
effort to introduce a deformation-controlling scheme is presented The idealized skeleton curves are illustrated in Fig. 2(b). For R/C
in this study through the ‘constant yield story-drift ratio’ scheme. structures, the skeleton curve for the main frame should include
The ‘constant yield story-drift ratio’ scheme is introduced as an the effect of stiffness reduction due to section cracking, yield of
alternative to control the deformations of hysteretic dampers reinforcement and cyclic loading. In Fig. 2(b), QS , QFy and QDy are
and those of the overall structure, by determining the yield the yield shear strength of the entire system, main frame and
deformation of dampers from the structural characteristics of the damper system, respectively. 1Fc , 1Fy , 1Dy , 1max , µF , µD are the
R/C main structure. cracking story drift, the yield story drift of the main frame, the yield
The objective of this paper is to examine the influence of story drift of the damper system, the maximum story drift, the
the ‘constant yield story-drift ratio’ scheme on the seismic ductility of the main frame and the ductility of the damper system,
performance of R/C building structures with hysteretic dampers respectively. α and ρ define the shear at the cracking point QFc
through a series of parametric studies. This scheme is used and the equivalent stiffness Keq for the main frame, respectively.
to investigate the reduction in story drift demand and seismic From the common arrangements of hysteretic dampers shown in
damage after installing damper to an R/C main frame. This paper Fig. 1, this study considers two R/C main frames: one with BRBs
concludes with a brief discussion on the definition of mechanical and another with generic hysteretic damper models.
properties of hysteretic dampers in accordance with the proposed
scheme.
2.2. Description of analyzed buildings
2. Methodology and structural modeling
Two ten-story R/C main frames are investigated under similar
2.1. R/C buildings with hysteretic dampers design parameters. The symmetric plan consists of 3 by 4 bays each
of 7 m with a typical height of 3.5 m, as shown in Fig. 3. Hysteretic
Hysteretic dampers (the damper system) are installed in the dampers are placed at each frame and at each story in the direction
frames of an R/C main structure (the main frame). Fig. 1 illustrates under study.
J.A. Oviedo A. et al. / Engineering Structures 32 (2010) 1735–1746 1737
Fig. 3. Analyzed buildings: (a) typical plan, (b) elevation of the R/C main frame with BRBs, and (c) elevation of the R/C main frame with generic hysteretic damper models.
The gravitational loads (dead and live) per unit area are
assumed to be the same for all stories, with a typical floor load of
10.02 kN/m2 . Prior to the installation of the hysteretic dampers,
the structural designs of both main frames were established based
on the Building Standard Law of Japan (BSLJ) [24] for a story-drift
angle limit of 1/200. Both R/C main frames were intentionally
kept unchanged to observe the influence of the yield deformation
Fig. 4. Element model for passive damper [28].
and the yield strength of dampers on the structural response.
Structural designs represent the strong-column and weak-beam
collapse mechanism, typical for ductile moment-resisting frame coefficients were applied in accordance with FEMA-274 [27] to
buildings. determine the equivalent stiffness of the main frame, Keq (Fig. 2(b)).
The BSLJ stipulates a vertical distribution of earthquake- For Building B, numerical analyses were carried out by the
equivalent forces according to the seismic lateral strength computer program STERA-3D [28], used for the frame analysis
coefficient at the i-th story, Ci , and the seismic lateral strength of R/C buildings. The hysteresis rule for columns and beams is
distribution factor at the i-th story, Ai , defined by the degrading trilinear Takeda model. For all non-linear time-
history analyses, the inherent viscous damping ratio is 0.02, the
1 2T integration time step is 0.005 s and the post-elastic stiffness ratio
Ai = 1 + √ − αi (1)
αi 1 + 3T is 0.01.
Ci = Ai C0 (2)
2.3. Hysteretic damper model
T = 0.02H (3)
Qi = Ci αi W (4) In Building A, the BRBs are modeled by inelastic truss elements
that can undergo large compressive strains without buckling. In
where αi is the normalized weight above the i-th story. The
Building B, the model for hysteretic dampers is more generic, as
standard base shear coefficient C0 is 0.2, T is the natural period,
shown in Fig. 4. From a general standpoint, hysteretic dampers
and H is the building height. The lateral shear strength at the i-th
story, Qi , is determined by Eq. (4). W is the building weight. link the story shear Q and the story drift ∆ of the story at which
The first frame model (hereafter Building A) corresponds to they are installed. Therefore, the damper model consists of a
the application of BRBs to a two-dimensional R/C frame model as line element with a non-linear shear spring whose deformation
shown in Fig. 3(b). One pair of BRBs was installed at the center bay is linked to the nodal displacements of the frame to which the
of each story of frame C , according to Fig. 3(a). Frame C is selected damper is connected. Details of this damper model can be found
to represent the response of the whole building. The gravitational elsewhere [28]. The hysteresis model for both inelastic elements is
load at each floor is assumed to be proportional to the tributary of bilinear type with kinematic hardening. Isotropic hardening was
loading area of frame C . The total weight of this two-dimensional not considered.
model is 14,720 kN.
The second frame model (hereafter Building B) corresponds to 2.4. Parametric study
the application of generic hysteretic damper models to a three-
dimensional R/C frame model as shown in Fig. 3(c). One hysteretic The damper strength ratio β (ratio of the yield strength of the
damper was installed at the center bay of each story for each damper system to the yield strength of the entire system, hereafter
frame in the direction under evaluation, for a total of five hysteretic the strength ratio) and the yield story drift ratio ν (ratio of the
dampers per story. The total weight of this three-dimensional yield story drift of the damper system to the yield story drift of the
model is 58,898 kN. Table 1 summarizes the structural properties R/C main frame, hereafter the drift ratio) are the main parameters.
of both R/C main frames. A series of non-linear time-history analyses corresponds to the
For Building A, numerical analyses were carried out by the combination of (1) two building models (Buildings A and B), (2)
computer program Drain-2DX [25]. The bilinear model and the eight strength ratios (β = 0, 0.2 to 0.8), (3) four drift ratios (ν =
degrading bilinear Takeda model [26] were used for the hysteresis 0.4, 0.6, 0.8 and 1.0), and (4) different input ground motions. The
rules of columns and beams, respectively. Stiffness reduction factor ν will be discussed in detail in the following section.
1738 J.A. Oviedo A. et al. / Engineering Structures 32 (2010) 1735–1746
Table 1
Structural properties of the analyzed R/C main frames.
Element Section geometry (m) Concrete strength (MPa) Reinforcement ratio (%)
Building A Building B
Fig. 5. Assumed schemes for the restoring force at the i-th story: (a) constant yield story-drift ratio, ν , (b) constant stiffness ratio, k, and (c) variation of k for different ν .
Referring to Fig. 2(b), the strengths of the damper system and start to dissipate energy. Control over the yield story drift and
the main frame are related by uniform distribution of the stiffness ratio k along the height are
automatically given by employing ν .
QS = QFy + QDy (5)
The yield story drift 1iDy and the horizontal stiffness j KDi at the
QDy = β QS (6) i-th story for the damper system are determined by
QFy = (1 − β)QS . (7)
1iDy = ν 1iFy (11)
The stiffness i
Keq
and the yield story drift 1iFy
at the i-th story of
each R/C main frame were determined by pushover analysis. The
i
j KD
i
= j QDy /1iDy . (12)
vertical distribution of lateral load used for the pushover analyses The idealized restoring force at the i-th story is depicted in Fig. 5(a).
was proportional to the first mode shape. The strength of the main It can be observed that when the strength of the damper system
frame is intentionally kept unchanged while the strength of the i
increases from j QDy i
to j+1 QDy , as a result of a larger β , the stiffness
damper system changes. Thus, the following equations result for of the damper system increases to meet j+1 QDy i
. The value of ν is
a certain strength ratio j β (the symbol ‘j’ is introduced hereafter to constant for all stories and strength ratios, and the stiffness ratio
show how the parameters change when β changes):
is constant for all stories for j β . On the other hand, unlike the
QFy ‘constant yield story-drift ratio’ scheme, when a constant stiffness
j QS = (8) ratio is assumed for all strength ratio cases (e.g., [5]) as shown
(1 − j β)
in Fig. 5(b), the stiffness of the damper system is the same for all
j QDy = j βj QS (9) values of β but different at each story, and there is no direct control
over the yield story drift of the damper system. From Figs. 2(b) and
j KD
jk = (10) 5, the relationship between k and ν is
Keq
β
where j QS , j QDy and j k are the yield shear strength of the entire k= . (13)
system, the yield shear strength of the damper system and the (1 − β) ν
stiffness ratio for j β , respectively. QFy is the yield shear strength The behavior of Eq. (13) is shown in Fig. 5(c). This plot suggests that
of the main frame system for all values of β . At each value of β , the it is neither practical nor economical to control the strength and
total base shear resisted by the damper system QDy was distributed story drift in full (β → 1.0 and ν → 0.1) because the required
along the height like the horizontal force distribution used for the stiffness for the damper system would become too large. Although
design of the main frame (Eq. (2)) to determine the shear resisted the behavior in Fig. 5(c) does not change, the applicable range for
i
by the damper system at the i-th story, QDy . the drift ratio depends on the stiffness of the main frame and can
be narrowed or widened.
2.5. ‘Constant yield story-drift ratio’ scheme Here, it should be emphasized that the proposed scheme
does not aim to produce story drifts that are constant in all
The drift ratio ν is introduced to define the yield deformation of stories, as investigated in [8,18]. The proposed scheme is used to
hysteretic dampers and to examine its implications on the seismic investigate and search for trends of the ratio of the maximum
performance. The value of the drift ratio ν is defined from 0 to 1.0, floor displacement D and story drift ∆ to the maximum floor
and it is intended to be constant for all stories and strength ratios. displacement Do and story drift 1o of the R/C main frame without
ν = 1.0 means that both the damper system and the main frame dampers. Fig. 6 depicts the pushover curves of Building A and B. The
yield at the same story-drift level, and it can be understood as vertical axis denotes the story shear normalized by the building
the lowest protection to the main frame since the damper system weight W . The horizontal axis denotes the story drift normalized
will have the smallest stiffness and require larger story drifts to by the story height h. From Fig. 6(b) and (c), it can be observed
J.A. Oviedo A. et al. / Engineering Structures 32 (2010) 1735–1746 1739
Fig. 6. Pushover curves: (a) R/C main frame, (b) Building A: β = 0.2 and ν = 0.4, and (c) Building A: β = 0.2 and ν = 1.0.
that the yield story drift of the damper system, 1Dy , is different
at each story and agrees very well with the assumed values from
Eq. (11), as desired. From these two figures, it can be also noted
that the yield story drift of dampers at the upper stories tends to be
larger than the assumed values due to the effect of overall bending
deformation.
The mechanical properties of the hysteretic dampers were Fig. 7. Deformed configuration of the BRB at the i-th story.
defined according to the methodology presented. For Building A,
the BRBs are modeled by a single inelastic truss element. Fig. 7 Table 2
i Input ground motions.
shows the deformed configuration at the i-th story. QDy is the story
shear to be resisted by both BRBs at the i-th story. Earthquake source Input motion PGA (m/s2 ) PGV (m/s) Td (s)
In Fig. 7, L is the bay length, h is the story height and Lb is the El Centro NS (1940) ElCentro 3.44 0.35 54
total length of the BRB. δby
i
and 1iDy are the yield axial deformation BCJ-L2 (synthesized) BCJ-L2 3.55 0.50 96
and the yield story drift of an individual BRB at the i-th story. JMA-Kobe NS (1995) Kobe 8.18 0.91 60
El Centro NS (1940) ElCentro50 5.05 0.51 54
From Fig. 7, the area of an individual BRB at the i-th story can be
i El Centro NS (1940) ElCentro75 7.57 0.76 54
expressed in terms of the yield shear strength QDy and the yield El Centro NS (1940) ElCentro100 9.87 0.98 54
story drift 1iDy as BCJ-L2 (synthesized) BCJ75 4.85 0.75 96
BCJ-L2 (synthesized) BCJ100 6.45 1.00 96
JMA-Kobe NS (1995) Kobe75 6.75 0.75 60
δby
i
JMA-Kobe NS (1995) Kobe100 9.00 1.00 60
cos θ ≈ (14)
1iDy Td:duration.
i
1 QDy Lb Three different source acceleration records were selected: two
Aiby = (15) well-known earthquake ground motions, El Centro NS (1940) and
2 1iDy E cos2 θ
JMA-Kobe NS (1995), and a synthesized record widely used in
where Aiby represents the required cross-sectional area for a BRB Japan, BCJ-L2. The records were scaled to meet different levels of
i seismic intensity (Peak Ground Velocity (PGV) = 0.50, 0.75 and
to remain elastic under the force QDy with a yield story drift of
1.00 m/s).
1iDy . E is the elastic modulus. Thus, the geometry of a BRB at the
i-th story for each value of ν and β is determined from Eqs. (9), 2.8. Evaluation of seismic performance
(11) and (15). It is worth mentioning that because both Aiby and
δby
i
are determined, this framework and the BRB model used in the It is well known that the damage to a structure depends on
analyses enforced the use of different yield stresses for the BRBs at the amount of hysteretic energy dissipated and on its cumulative
each floor (further discussion is presented at the end of this paper). effect. In this study, the seismic damage in the main frame is mainly
Although a required yield stress could be rather difficult to adjust assessed by means of the cumulative plastic strain energy ratio ēiF
in practical design, this situation might be convenient as it would defined as
grant more flexibility for the sizing of BRBs. Moreover, if a yield i
EHF
stress is fixed as in common design practice, there is no control ēiF = (16)
i
QFy 1iFy
over the yield story drift of BRBs.
For Building B, the yield shear strength QDy i
and the yield story where EHFi i
, QFy , 1iFy and ēiF are the hysteretic energy dissipated, the
drift 1iDy of the damper system at the i-th story for each value yield shear strength, the yield story drift and the cumulative plastic
strain ratio of the main frame at the i-th story, respectively.
of ν and β are determined from Eqs. (9) and (11). QDy
i
is equally
distributed to each damper of the i-th story.
3. Results of earthquake response analysis and discussion
Table 2 summarizes the input motions used for the non-linear Fig. 8 illustrates the variation of the input energy ratio EI /EI0 .
time-history analyses. The vertical axis denotes the ratio of the input energy EI to the
1740 J.A. Oviedo A. et al. / Engineering Structures 32 (2010) 1735–1746
input energy without hysteretic dampers EI0 . It is clearly noticed dissipated by the main frame is seen for all values of ν and β .
that there is a higher variability in the EI /EI0 ratio for β > 0.5, and This reduction is more significant for low values of ν with low
there is not a remarkable influence of the drift ratio for β ≤ 0.5. values of β . On the other hand, for Building B, some increase in
It can also be observed that the influence of the input motion the hysteretic energy dissipated by the main frame was observed.
characteristics is more significant than the influence of the PGV This situation in Building B is because the efficiency of the damper
level. In the case of severe input motions such as Kobe, the system in dissipating the hysteretic energy decreases for large
input energy increases remarkably with increasing strength ratios. values of β . The dampers, especially at the upper stories, affected
This phenomenon can be attributed to the concentration of large by the overall bending deformation of the building along with large
impulses in the acceleration record accompanied with large plastic yield strengths, do not work beyond the elastic range. As a result,
deformation. For values of β lower than 0.4, at which the natural the main frame takes all the inelastic work and sustains even more
period does not change considerably, the input energy is practically seismic damage than in the case without dampers.
the same as that of the building without BRBs, and is scarcely In general, a remarkable reduction in the participation of the
affected by the drift ratio ν . A similar situation was reported by main frame with decreasing drift ratios was noted. This trend is
Inoue and Kuwahara [5]: the input energy characterized by an attributed to the large amount of hysteretic energy dissipated by
equivalent velocity was almost independent of the restoring force the damper system for low drift ratios. The damper system starts
characteristics of the structural system. dissipating the hysteretic energy before the main frame yields, and
therefore it sustains more inelastic cycles.
3.2. Cumulative plastic strain energy ratio Regarding the strength ratio β , a relatively flat region within
the middle third of the β axis can be seen. This region ranges
Fig. 9 shows the cumulative plastic strain energy ratio of the from β of 0.2 to 0.5 in most cases; however, its upper limit
main frame ēF normalized by that of the main frame without tends to increase as the PGV increases. This relatively flat region
dampers ēF 0 . For Building A, a reduction of the hysteretic energy suggests that strength ratios within this range would reduce the
J.A. Oviedo A. et al. / Engineering Structures 32 (2010) 1735–1746 1741
Fig. 10. Distribution of the normalized maximum story-drift ductility along the height: Building A.
participation of the main frame and increase the energy dissipated evaluated throughout the structural design process as it indicates
by the damper system. In terms of damage protection, it appears an extent of structural damage. Thus, Figs. 10 and 11 show a plot
that the protection to the main frame is maximized within this of the vertical distribution of the maximum story-drift ductility of
range of β , and that the larger the drift ratio, the less effective the the main frame µF normalized by that of the main frame without
protection to the main frame is. hysteretic dampers µF 0 , for various input motions. Hereafter,
One important point to note is that low values of ν tend to results for values of β ≤ 0.5 are presented as this range provided
widen the range of β at which the protection of the main frame the largest protection to the main frame, as discussed in the
is maximized and kept almost invariant. This ‘uniform’ range previous section.
has a relevant meaning in engineering practice because, if the Although no clear tendency for the amount of reduction or in-
structural design were defined within this range, the structural crease of the story-drift ductility demand to the main frame can
performance would be less affected in the case of modifications be observed from Figs. 10 and 11, the following trends and as-
to the design caused by uncertainties such as construction and pects are identified. (1) The µF /µF 0 ratio becomes uniform along
on-site installation practices and/or material strength reliability. the height and is lower than unity for most input motions. In par-
Consequently, an optimum strength ratio should be defined ticular, reductions of the story-drift ductility in the main frame
within this ‘uniform’ range. Inoue and Kuwahara [5] proposed (µF /µF 0 < 1.0) are seen for low values of ν and β . It should be
an expression to estimate the optimum value for steel moment noted that these reductions can be as large as 50%. However, some
frames. increase in the story-drift ductility (µF /µF 0 > 1.0) is also pre-
sented for some input ground motions after installing the dampers.
3.3. Vertical distribution of the story drift ratio (2) For increasing values of β , the µF /µF 0 ratio tends to become
smaller at the upper stories than those at the lower stories in Build-
The story-drift ductility (ratio of the maximum story drift to ing A, while Building B experiences exactly the opposite behav-
the yield story drift) is a structural response parameter commonly ior. Here, as discussed in the previous section, the deformations of
1742 J.A. Oviedo A. et al. / Engineering Structures 32 (2010) 1735–1746
Fig. 11. Distribution of the normalized maximum story-drift ductility along the height: Building B.
dampers at the upper stories in Building B decrease, the dampers Building B, the change of fundamental period is small and most
remain mainly elastic and the main frame goes into a larger inelas- analysis cases presented reductions in the story-drift ductility of
tic response. (3) The µF /µF 0 ratio was found to be more dependent the main frame. However, analysis cases corresponding to ElCentro
on the input ground motion characteristics than on the value of ν . and ElCentro50 input motions responded elastically and presented
(4) An increasing value of ν tends to reduce the protection to the larger µF /µF 0 ratios caused by a high variability of the input energy
main frame regardless of the value of β . with increasing values of β (Fig. 12(a)).
The fundamental period for each analysis case changes accord- It should be noted that the maximum value of the story-
ing to the value of ν and β . In general, the fundamental period drift ductility is independent of the amount of hysteretic energy
shortens with increasing values of β and decreasing values of dissipated, number of inelastic cycles and amplitude of response
ν . However, this change is more noticeable when β changes, as below this maximum level, which deteriorate the mechanical
shown in Fig. 12(c). A decrease or increase in the story-drift duc- properties of a structure and lead to fatigue damage. Therefore,
tility demand to the main frame after installing dampers can be the vertical distribution of the normalized cumulative plastic strain
comprehensibly understood by examining the fluctuation of the energy is presented in the following section.
input energy within the range of fundamental periods of all anal-
ysis cases. Akiyama [29] showed that the input energy spectrum 3.4. Vertical distribution of the cumulative plastic strain energy ratio
smoothens with the extent of plastification and that an elastic en-
ergy spectrum with a damping ratio of 0.1 may describe an en- The results are emphasized for two source motions hereafter:
velope of the input energy of an inelastic system (especially for El Centro NS (1940), widely used for research studies; and BCJ-L2,
the longer period range). Thus, Fig. 12 shows the elastic energy re- commonly used in Japan to check and ensure safety in the building
sponse spectra (expressed by an equivalent velocity Ve ) of the three design against Level-2 earthquakes [30]. In the previous section,
source records for damping ratios of h = 0.02 and h = 0.1 and the cumulative plastic strain energy ratio ēF of the whole main
the variation of the fundamental period T of analyzed buildings. It frame was discussed; however, a reduction of the total ēF does not
can be clearly seen that the energy response spectra shape softens necessarily bring along the fact that the damage is also reduced in
for a larger damping ratio (i.e., large extent of plastification). all stories. Figs. 13 and 14 show the distribution along the height
Since all analysis cases responded inelastically in Building A, it of the ratio ēF /ēF 0 . The vertical axis represents the story number.
can be seen in Fig. 12(b) that a period decrease means a larger input Two common behaviors can be observed: (1) the lower the drift
energy, especially for Kobe and BCJ-L2 acceleration records. This ratio, the larger the reductions of the ēF /ēF 0 ratio in all stories,
effect can be observed in Fig. 10 when the µF /µF 0 ratio increases and (2) there are neither relevant concentrations at any story nor
with increasing values of β . Here, the µF /µF 0 ratio decreases any increase of the ēF /ēF 0 ratio for ν ≤ 0.8. Another important
with decreasing values of ν as the input energy does not change consequence of low values of ν is that the distribution of the ēF /ēF 0
significantly and the stiffness of the entire system increases. In ratio along the height tends to remain constant. This suggests an
J.A. Oviedo A. et al. / Engineering Structures 32 (2010) 1735–1746 1743
Fig. 12. Energy response spectra and variation of fundamental period T : (a) energy response spectra for h = 0.02, (b) energy response spectra for h = 0.1, and (c) variation
of fundamental period with respect to ν and β .
Fig. 13. Distribution of the normalized cumulative plastic strain energy ratio along the height: Building A.
almost uniform control over the protection to the main frame. For 3.5. Maximum floor displacement
large values of ν and β , a concentration of seismic damage in the
middle stories with ēF /ēF 0 ratios even larger than 1.0 was observed. Fig. 15 shows the distribution along the height of the
For Building B in Fig. 14, the ēF /ēF 0 ratio becomes larger at normalized floor displacement for a few input motions. The
the upper stories when β increases. As mentioned earlier, this vertical axis represents the floor number and the horizontal axis
indicates that the damper system does not dissipate enough depicts the maximum floor displacement, D, normalized by the
hysteretic energy at the upper stories. It should be pointed out that maximum floor displacement without hysteretic dampers, D0 . The
the hysteretic dampers remained elastic at most upper stories for effect of the additional stiffness introduced by the dampers is
β ≥ 0.6. evident in reducing the floor displacements. As the input energy
A clear difference is observed by comparing the trends of the was found to be almost invariant for low strength ratios, it was
vertical distributions of the µF /µF 0 ratio and the ēF /ēF 0 ratio expected that the floor displacements for β ≤ 0.5 would present
(Figs. 10, 11, 13 and 14). The drift ratio ν has a larger influence significant reduction. This situation was clearly observed in the
on the vertical distribution of the ēF /ēF 0 ratio than on the µF /µF 0 floor displacement response. For almost all values of ν and β ,
ratio. Here, although the choice of a lower value of ν might not the floor displacements were lower than those of the structure
bring relevant changes in the reduction of the story-drift ductility without dampers. The reduction of floor displacements certainly
of the main frame, the cumulative damage could be significantly suggests less story inelastic work and less seismic damage in the
reduced. main frame.
1744 J.A. Oviedo A. et al. / Engineering Structures 32 (2010) 1735–1746
Fig. 14. Distribution of the normalized cumulative plastic strain energy ratio along the height: Building B.
In Building A, the equivalent horizontal stiffness and the Eq. (20) suggests that, to achieve a certain 1Dy , four paths are
cross-sectional area of BRBs were determined by Eqs. (12) applicable: (1) adjust the plastic and elastic lengths, or (2) adjust
and (15), respectively. This definition stands for an equivalent the stresses, or (3) adjust both lengths and stresses, or (4) adjust
single element with constant cross-sectional area and equivalent the inclination angle θ . Nowadays, due to some limitations in the
stiffness. However, a BRB is commonly assumed to comprise two available steel material, the adjustment of the elastic or plastic
elastic parts and a plastic part, as shown in Fig. 16. The idealized length is more common. Based on Fig. 3(b), Table 3 shows some
BRB is modeled by three springs connected in series. Lb is the total examples of yield story drifts calculated from Eq. (20) for Lb =
length of the BRB, Lbp is the plastic length, Lbe is elastic length, 4.95 m and θ = 45°. It can be seen that by having larger cross-
and kbe and kbp are the axial stiffness of the elastic and plastic sectional areas for the elastic part and by increasing the length of
part, respectively. kbeq is the equivalent axial stiffness. The inelastic the plastic part, larger yield story drifts can be obtained. The largest
behavior is intended to occur within the plastic length, and the yield story drift for a certain yield stress is given theoretically when
elastic length corresponds to elastic and rigid connections. both the plastic and elastic parts share the same stress under the
The total axial deformation δby is given by yield axial force Pby . However, this might not ensure yield only in
the plastic part. Smaller yield story drifts can also be obtained by
δby = δbp + 2δbe (17) using lower yield stresses for the steel.
For instance, in case of bending plates (generic hysteretic
Pby Lbe Lbe
δbe = = σbe (18) damper model, Building B), a wider range of yield story drifts could
Abe E E be obtained compared with the case of BRBs. A yield story drift
Pby Lbp Lbp 1Dy could be obtained by adjusting the geometry, the moment of
δbp = = σby (19) inertia of the section, or the height of the device according to the
Abp E E
way the device is installed into the main structure and the manner
where δbp and δbe are the axial deformation of the plastic and it dissipates the hysteretic energy.
elastic parts, respectively. Pby is the yield axial force of the plastic
part, E is the elastic modulus for the steel, and σbe and σby are the
4. Conclusions
stresses under the yield force Pby in the elastic and plastic parts,
respectively. σbe is always lower than σby .
The earthquake response of ten-story story-drift-controlled R/C
Thus, referring to Fig. 16 and Eqs. (14) and (17)–(19), the yield
building structures with hysteretic dampers has been presented
story drift of a BRB is given by
through a parametric study with a series of non-linear time-
2σbe Lbe + σby Lbp history analyses. The drift ratio ν and the strength ratio β are the
1Dy = . (20) main parameters for the definition of the mechanical properties of
E cos θ
J.A. Oviedo A. et al. / Engineering Structures 32 (2010) 1735–1746 1745
Fig. 15. Distribution of the normalized maximum floor displacement along the height: Buildings A and B.