Download as pdf or txt
Download as pdf or txt
You are on page 1of 24

EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS

Earthquake Engng Struct. Dyn. 2006; 35:1875–1898


Published online 12 September 2006 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/eqe.610

Torsional balance of plan-asymmetric structures with frictional


dampers: Experimental results

Ignacio J. Vial‡, § , Juan C. de la Llera∗, †, ¶ , José L. Almazán and Vı́ctor Ceballos∗∗


Department of Structural and Geotechnical Engineering, Universidad Católica de Chile, Santiago, Chile

SUMMARY
This investigation deals with the measured seismic response of a six-storey asymmetric structural model
with frictional dampers. Its main objective is to experimentally prove the concept of weak torsional balance
for mass- and stiffness-eccentric model configurations. The goal is to control the torsional response of
these asymmetric structures and to achieve, if possible, a weak form of torsional balance by placing
the so-called empirical centre of balance (ECB) of the structure at equal distance from the edges of
the building plan. The control of the dynamic response of asymmetric structures is investigated herein
by using steel–teflon frictional dampers. As expected from theory, experimental results show that the
mean-square and peak displacement demand at the flexible and stiff edges of the plan may be similar
in magnitude if the dampers are optimally placed. Frictional dampers have proven equally effective in
controlling lateral-torsional coupling of torsionally flexible as well as stiff structures. On the other hand,
it is shown that impulsive ground motions require larger frictional capacities to achieve weak torsional
balance. Copyright q 2006 John Wiley & Sons, Ltd.

Received 14 June 2005; Revised 23 February 2006; Accepted 6 June 2006

KEY WORDS: torsional balance; experimental results; plan-asymmetry; frictional dampers; empirical
centre of balance; response reduction factors

∗ Correspondence to: Juan C. de la Llera, Department of Structural and Geotechnical Engineering, Universidad
Católica de Chile, Casilla 306, Correo 22, Santiago, Chile.

E-mail: jcllera@ing.puc.cl

E-mail: ijvial@puc.cl
§ Master of Science.
¶ Professor.
 Assistant Professor.
∗∗ Graduate Student.

Contract/grant sponsor: Chilean National Fund for Science and Technology, Fondecyt; contract/grant number:
1020774

Copyright q 2006 John Wiley & Sons, Ltd.


1876 I. J. VIAL ET AL.

INTRODUCTION

Practice shows that plan-asymmetric structures show a larger rate of failure than their symmetric
counterparts when subjected to earthquakes. Since these structures do occur very often in practice
as a result of architectural considerations, it is relevant to look for procedures that enable the
designer to control their torsional response. Plan-asymmetry leads to an uneven lateral deformation
demand among resisting planes and recorded responses show that the inelastic deformation may
tend to concentrate in few resisting planes [1]. Thus, plan-asymmetry is considered by codes and
the engineering profession as an undesirable property of a structure. Moreover, current building
codes wisely limit the design of asymmetric structures by incorporating procedures to account for
the planwise irregular displacement distribution, leading to different capacities and detailing of
resisting planes across the building plan.
In a recent article, a procedure was proposed to optimally place frictional dampers (FD) in the
building plan in order to achieve a weak form of torsional balance [2]. This weak form leads to
equal mean-square values of the lateral displacements at the stiff and flexible edges of the building
plan. It does not imply minimum rotation of the building plan but rather a bound for the lateral
displacements at the stiff and flexible edges. As opposed to this definition, the strong form of
torsional balance looks for an optimal location in plan of the damper such that the building plan
rotation is minimized.
Few analytical and experimental investigations have focussed in lateral-torsional coupling of
multistorey structures with FD. The dynamic response of such single-storey asymmetric structures
has been considered recently [2]. A new concept, denoted as the empirical centre of balance (ECB)
was developed to define a performance objective for asymmetric structures. In theory, the principle
established says that if the ECB is located at equal distance from both edges of the building
plan, the mean-square value of the response at both edges will be the same, thus leading to the
so-called torsionally weak balanced design. In that investigation, the evaluation of the response of
asymmetric structures with FD is done only for weakly optimal locations of the FD. Performances
relative to other optimal damper location criteria such as the mirror rule [3] were also evaluated.
It can also be shown that once the structure is torsionally balanced in the weak sense, an increase
in the radius of gyration of the supplemental damping capacity, preserving the optimal damping
eccentricity, will tend to even the mean-square displacements at different resisting planes across
the building plan (strong balance).
Because of their stable frictional properties, diagonal elements with steel–teflon slotted bolted
connections (SBC) [4] were developed for the structure. The sliding capacities of the dampers
were easily modified by pre-tensioning calibrated flexible springs, thus enabling to test different
height- and planwise distributions of dampers. Results presented are by no means restricted to a
specific FD as long as the stick-and-slip phenomenon is minimized. Furthermore, it is important
to recall that the energy dissipation of FD is proportional to the maximum deformation and that
the device works well for a wide range of deformation amplitudes beyond its elastic limit. Finally,
dynamic properties for the steel–teflon sliding interphases were also characterized in terms of
sliding velocities and pressure [5].
Consequently, the objective of this research is to experimentally prove the concept of weak
torsional balance in mass and stiffness asymmetric multistorey structures with FD. Three different
mono-symmetric mass-asymmetric and three stiffness-asymmetric building configurations were
considered in the study, covering a wide range of system parameters. The linear-elastic experimental
models are intended as proof-of-concept models, which means that they do not necessarily represent

Copyright q 2006 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
DOI: 10.1002/eqe
BALANCE OF PLAN-ASYMMETRIC STRUCTURES WITH FRICTIONAL DAMPERS 1877

faithfully the detailed response of a specific real-life prototype. Moreover, results such as response
reduction factors are applicable in the strict sense, only to the set of model structures considered
herein. However, the authors are confident that the general trends shown by the experimental results
are more general since they also confirm the analytical results presented elsewhere [2].

EXPERIMENTAL SET-UP

The proof-of-concept structural model considered in this study is a six-storey frame building with
one identical bay of span length of 860 mm (Figures 1 and 2) in the X - and Y -direction, with
identical storey heights, h = 500 mm, and total building height H = 3000 mm. The model was built
in duralumin and an approximate geometric scale E L = 1/7. To maintain an acceleration
√ scale
√ EA
equal to 1 (E A = E L /E T2 = 1), time was scaled down in the model by a factor 7 (E T = E L ).
Hollowed beams and columns with square box-shapes of sides 30 and 40 mm, respectively, and
2.0 mm thickness, are used in the model. Prestressed steel cables of variable diameter ( = 1.5, 2,
and 3 mm) were used as braces and connected to the joint-brackets to increase lateral stiffness of
the structure. The floor diaphragm is a square steel plate of side 820 mm with a square central
orifice of side 410 mm, plate thickness t = 10 mm, and weight w = 386N each. These plates were
bolted in each corner to a horizontal flange connected to the L-shaped stiffeners (Figure 1). The
total approximate weight of the structural model is WT = 8135N .
The shaking table used is a 6 degree-of-freedom flight simulator which was adapted in 1998 by
this research group as a shaking table (Figure 1). The simulator was manufactured by MOOG and

N
2

ch#21
ch#20
820

410 Y
860

820

410

PL 820x820x10
ch#22
1

860 mm
A B
(b) X

A B
860

ch#4 ch#2
Helicoidal
2

spring.
1200
ch#6 ch#1
860 mm

Y
1

ch#5 ch#3
(± 20 mm)
1600

(a) (c) X

Figure 1. Experimental set-up with dynamic simulator and building model: (a) six-storey building model
with FD; (b) typical building plan; and (c) moving platform.

Copyright q 2006 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
DOI: 10.1002/eqe
1878 I. J. VIAL ET AL.

2800
40 820 40 Ch-26S
Level 6 (+3000)
30 x 30mm (typ) e=2mm Ch-24N Ch-23N

38W
c h# Ch-25N
32E eccentric weight
ch#
W=250N

Level 5 (+2500) Ch-22S


Ch-21N Ch-20N

37W
c h# .)
31E typ
c h# m(
.0 m
=2
le Ch-18N
cab Ch-19S
Level 4 (+2000)
Ch-17N Ch-16N

36W
c h#
30E
mm c h#
40 x 40mm (typ) e=2mm

= 1.5
me
1 6m duralumin
Ch-15S
3000 mm

Level 3 (+1500) A
Ch-14N Ch-13N

35W
c h#
29E A
c #
h
470

Level 2 (+1000) Ch-12S


Ch-11N Ch-10N
30

34W
c h#
28E
c h#

Level 1 (+500) Ch-9S


Ch-8N Ch-7N

cab 33W
le c h#
=v
ar 27E
(typ c h# 240
.)
25
82.5
100

860 mm 280 mm 860 mm


1 2 Detail A-A A B

Frames A,B. Frames 1,2

Figure 2. Location of displacement transducers and accelerometers in the six-storey-building, channels


27–38 are diagonal displacement transducers, and channels 1–26 are floor acceleration sensors.

is based on the original idea proposed by Steward [6]. Each of the 6 electromechanical actuators
control a single degree of freedom, which by a non-linear kinematic transformation enable to
control the three translational displacements of the platform and the three angles of rotation, pitch,
yaw, and roll. In this version, the platform and actuators are able to support a maximum load
of 15 kN, approximately. The limits of the dynamic platform for linear displacement, velocity,
and acceleration are ±25 cm, ±40 cm/s and 0.55g, respectively; on the other hand, the rotation
limits for displacement, velocity and acceleration are ±20◦ , ±35◦ /s and ±500◦ /s2 , respectively. A
conventional PID-controller is used for the axial displacement of each actuator, which is transferred
from an external computer in 20-byte frames to the platform computer in a RS-485 protocol at a
sampling rate of 60 Hz (sampling time h = 0.017 s). Digital conversion of data was performed by
a 64 channel AT-MIO-64E National Instruments board and data processing was done by using the
Data Acquisition Toolbox of Matlab [7].
The model was instrumented with 26 acceleration channels and 12 displacement channels. The
physical location of the sensors in the model is indicated in Figures 1 and 2; their layout enables us

Copyright q 2006 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
DOI: 10.1002/eqe
BALANCE OF PLAN-ASYMMETRIC STRUCTURES WITH FRICTIONAL DAMPERS 1879

to compute all floor accelerations in the two principal directions of the structure and the rotational
motion of each floor about a vertical axis; it also enable us to evaluate directly the storey drifts
and the displacements of each FD in the structure. Six accelerometers were placed on the moving
platform to measure the input motions (Figure 1(c)), 18 were place in horizontal triplets at each
floor to measure the lateral accelerations, and 2 were placed to measure vertical accelerations on
the 4th and 6th floors. The 12 displacement potentiometers were placed diagonally in pairs along
resisting planes A (west) and B (east). Moreover, accelerations are used to calculate storey shears
and torques as well as response modification factors due to lateral-torsional coupling.
The 6 asymmetric configurations considered in this investigation, 3 mass-asymmetric and
3 stiffness-asymmetric, are presented in Figure 3. In order to fictitiously achieve different torsional-
to-lateral frequency ratios in the structure, horizontal steel arms of length la = 2800 mm were fixed
to the structure at each floor level (Figure 2). Because of their section properties and as made
of steel, such arms are more than 13 times stiffer than any building storey and, hence, behave
approximately as rigid extensions of the diaphragm. Additional lumped masses, wm = 250N each,
were fixed on these arms at variable distances from the geometric centre (GC). This allowed
a displacement of the CM of the structure, leading to 3 different mass-eccentric systems. The 3
mass-eccentric configurations are denoted as M1 , M2 , and M3 , and are defined by locating one half
of the total additional mass at the left, centre, and right end of the arms, respectively. Naturally, the

A B A B
2

C =1. 5 mm C =1. 5 mm
C =1. 5 mm
1

M1 S1
A B A B
2

C =1. 5 mm C =3. 0 mm
C =1. 5 mm C =1. 5 mm
1

M2 S2
A B A B
2

C =1. 5 mm C=3. 0 mm
C =1. 5 mm
1

(a) M3 (b) S3

Figure 3. Asymmetric models considered: (a) mass-eccentric configurations; and


(b) stiffness-eccentric configurations.

Copyright q 2006 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
DOI: 10.1002/eqe
1880 I. J. VIAL ET AL.

Table I. Nominal building model properties.


Model Weight (N ) e y /L x Nominal periods Tx , Ty , T x
M1 8135 0 0.36; 0.25; 0.38 0.95
0.36; 0.26; 0.43 (K exp ) 0.84
Mass-eccentric M2 8135 0.28 0.36; 0.26; 0.27 1.33
0.36; 0.26; 0.30 (K exp ) 1.20
M3 8135 0.50 0.36; 0.26; 0.30 1.20
0.36; 0.26; 0.33 (K exp ) 1.09

S1 8135 −0.20 0.45; 0.26; 0.41 1.10


Stiffness-eccentric S2 8135 −0.16 0.29; 0.26; 0.35 0.83
S3 8135 −0.31 0.33; 0.26; 0.37 0.89

M3 configuration had also a large tendency to overturn due to the asymmetry in the location of the
masses. These mass-eccentric configurations lead to nominal mass eccentricities emy /L x ≈ 0, 0.28,
and 0.5, respectively (Table I). All these configurations are nominally symmetric in stiffness and
include two prestressed cable diagonals of diameter  = 2 mm in planes 1 and 2, and a single
prestressed diagonal of diameter  = 1.5 mm in planes A and B.
On the other hand, for stiffness-eccentric configurations, prestressed cables of variable diameters
were placed in resisting planes A and B as shown in Figure 3. The 3-stiffness-eccentric configura-
tions are denoted as S1 , S2 , and S3 , and are obtained by placing (i) a single  = 1.5 mm diameter
prestressed cable in resisting plane A, (ii) a  = 3.0 mm prestressed cable in plane A and one
 = 1.5 mm in plane B, and (iii) a single  = 3.0 mm prestressed cable in plane A, respectively.
These configurations lead to nominal stiffness eccentricities esy /L x ≈ −0.20, −0.16 and −0.31,
respectively (Table I). The S1 –S3 models are all nominally symmetric in mass and include two
prestressed cable diagonals of diameter  = 2 mm in planes 1 and 2 (Figure 2).
The first three uncoupled nominal and experimental vibration periods for these model structures
as well as the nominal torsional-to-lateral frequency ratios are presented in Table I. As shown in
this table, the masses and prestressed cables were selected in order to have torsionally flexible
systems ( = 0.83 and 0.89), a limit case ( = 0.95), and torsionally rigid structures ( = 1.20
and 1.33).
A closed-up of a typical steel–teflon damper used in the building is also presented in Figure 1.
The damper is composed by two teflon elements of 30 mm × 170 mm and a movable steel plate of
30 mm × 90 mm. Each damper has 2 fixed bolts and 2 slotted holes with a maximum displacement
of ±20 mm. Since the steel plate is in between the two teflon elements, each damper provides
two frictional steel–teflon interphases. Normal loads are applied by two calibrated flexible linear
springs in order to distribute the normal loads more uniformly. Each FD was placed in series with a
tubular diagonal element of diameter  = 16 mm and thickness t = 1.5 mm. A total of 12 dampers
were installed in resisting planes A and B.

RESPONSES CONSIDERED

The testing results and analysis presented herein are for 2 different earthquake records: Newhall
(Northridge, 1994; PGA = 0.59g) scaled to 20% of its peak ground acceleration, and El Centro

Copyright q 2006 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
DOI: 10.1002/eqe
BALANCE OF PLAN-ASYMMETRIC STRUCTURES WITH FRICTIONAL DAMPERS 1881

(Imperial Valley, 1940; PGA = 0.31g) scaled to 30% of its peak ground acceleration. The reason
for scaling down the records was to preserve the primary structure within the elastic range. Since
the models considered were mono-symmetric, only a single component of ground motion was
used in each test.
The location of the ECB is computed from the measured building responses at the GC of the
building plan from [2]

E[u y (l)u  (t)]


p∗ = − = − y  ( y / ) (1)
E[u  (t)2 ]

where E[·] represents the expected value of (·); u y (t) and u  (t) are the displacement and rotation
of the building plan measured at the GC;  y and  are the corresponding standard deviations
of the building response;  y  is the linear correlation coefficient between the displacement and
rotation; and p ∗ is the location of the ECB measured from the GC. Weak torsional balance occurs
if p ∗ = 0, or equivalently  y  = 0, and implies that the mean-square displacement of any resisting
plane in the plan, equidistant along the axis of symmetry to the GC, is the same. Recall also that
p ∗ physically represents the location of the vertex of the parabola defined by the mean-square
value of the lateral displacements of the building plan along its axis of symmetry.
Other response values are computed directly from recorded accelerations and displacements
by using simple equilibrium or kinematic equations. The responses are: (1) the maximum lateral
displacements at resisting planes A and B, u A and u B ; (2) the mean peak normalized displacements
(d) (d)
(reductions) at these edges, denoted as û A = u A /u A and û B = u B /u B , respectively, where the
super index (d) denotes the structure with FD; (3) the standard deviation of the lateral displacements
( j)
at both edges of the building plan,  A and  B ; and (4) the jth storey shear in the X -direction, Vx ,
and torque T ( j) . Moreover, to quantify the torsional effects in the structure, a global parameter
called the torsional index, , is defined. This parameter represents a ratio between the maximum
absolute difference between the lateral displacements at the flexible and stiff edges over the
displacements at the GC of the same structure, i.e.

max u f − u s 
 = 100 (%) (2)
max u 0 

Finally, a comparison is presented between the torsional amplifications obtained for the structural
model with and without FD. The purpose of these comparisons is to estimate the torsional response
reduction factors in mass and stiffness-asymmetric structures with FD. Although floor deformations
are readily available by double integration of the recorded floor accelerations, it was decided to
evaluate the superstructure deformations directly from the displacement transducers, and total
storey shears and torques from the measured floor total accelerations. Notice that storey shears
and torques are computed accurately since the building masses are known in this case. Therefore,
the following normalized storey shear and torque are defined for each storey:
(d) (d)
( j) (Vx ) j ( j) (Tx ) j
V̂x = and T̂x = , j = 1, . . . , 6 (3)
W WL x
(d) (d)
where the quantities (Vx ) j and (Tx ) j represent the peak storey shear and torque values for the
jth storey, respectively.

Copyright q 2006 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
DOI: 10.1002/eqe
1882 I. J. VIAL ET AL.

DYNAMIC SYSTEM PROPERTIES

Dynamic behaviour of FD
The calibration of the steel–teflon FD was performed using a universal testing machine (Instron
model 8872) and considering sinusoidal input signals. Since the damper has two sliding surfaces,
the coefficient of friction inferred from the experiments would be
Fs
= (4)
S
¯

where Fs is the measured sliding force, S is the total area of sliding (contact), ¯ = ( n dS)/S is
the average normal force on the contact surface, n = (4keq )/S is the normal stress, and  is the
individual spring deformation. The nominal stiffness of each spring is
d4G
kn = (5)
8n D 3
where n = 10, is the number of coils of the spring, D = 0.7 cm and d = 0.2 cm are the external
and internal diameters of the coil, and G = 8 × 105 kg/cm2 is the shear modulus of steel. Thus,
the nominal stiffness of each spring is kn = 46.65 kg/cm.
Shown in part (a) of Figure 4 is the experimental identification of the spring stiffness
(Equation (4)). The theoretical stiffness of 46.65 kg/cm compares well with the identified spring
stiffness of 52.5 kg/cm (11.1% error). To verify this stiffness value, let us consider the case of a FD
with lateral deformation ±0.25 cm, axial spring deformation 0.63 cm (9 turns of the screw), and
excitation frequency of 10 Hz. From part (b) of Figure 4, the maximum frictional force developed
was 23.5 kg. From part (d), the maximum value of the damping coefficient is max = 0.18, and the
estimated output sliding force is (Equation (3)) Fs = max N = 0.18 × (4 × 0.63 × 52.5) = 23.8 kg,
which is very close to the maximum observed dynamic force of 23.5 kg.
Shown in parts (b) through (d) of Figure 4 is the dynamic response of a typical FD. All 12 FD
were tested for different sinusoidal input displacements, frequency (ranging from 0.1 to 10 Hz)
and normal forces. Damper deformation amplitudes were ±0.5 and ±0.25 cm at low and high
frequencies, respectively (part (b) of Figure 4). It is apparent from the hysteretic curves that
the effect of stick-slip in the dampers was negligible. The almost rectangular hysteretic cycles are
extremely stable and their shape does not show any significant stick-slip effect. As the deformation
velocity decreases near the peak deformation, the frictional force also decreases gradually. The
dependency of the friction coefficient  with deformation velocity is shown in part (c) of Figure 4.
It has been previously shown [5], that the friction coefficient (v) can be related to deformation
velocity v by the equation
(v) = max − (max − min ) · e−a f v (6)
where min represents the value of the friction coefficient for low velocity, max is the asymptotic
value for high velocity, and a f is a coefficient that represents the transition between both values.
Furthermore, the hysteretic cycles in part (b) and the results of part (d) show how max varies
with the frequency input signal and the applied normal load for 3, 6, 9, and 12 turns of the screw.
Although max tends to decrease with larger normal loads, this effect is not large, at least for the
dampers considered. On the other hand, max is sensitive to the excitation frequency, and increases
from 6 to 18.8% from low to large frequencies. The experimentally identified values of min , max

Copyright q 2006 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
DOI: 10.1002/eqe
BALANCE OF PLAN-ASYMMETRIC STRUCTURES WITH FRICTIONAL DAMPERS 1883

45 40
Spring force = 33 kg. 9 turns, 10 Hz., 23.5 kg.
Total normal force = 4*33 = 132 kg. 3, 6, 9, 12 turns, 1hz
Normal spring force (kg)

F =µN = 0.18*132 = 23.8 kg.


s
20
30

Force (kg)
0
Kident= 52.5 kg/cm.

15

Experimental 3, 9 turns, 10hz


curve
0
0 3 6 9 12 0 0.2 0.4 0.6
(a) Number of turns (1 turn=0.07cm) (b) δ (cm)

30 30
µ = 16.6%, µ = 5.8%, a = 0.43s/cm
max min f
25 25
µmax for 9 turns = 18%
20 20
(%)
µ (%)

15 15
max
µ

10 10
3 turns
6 turns
5 5
9 turns
12 turns
0 0
0 5 10 15 20 0 2 4 6 8 10
(c) Velocity (cm/s) (d) Frequency (Hertz)

Figure 4. Identification of the frictional behaviour of dampers.

(asymptotic value) and a f (Equation (6)) for the 12 FDs considered, vary from 4.8 to 6.9%, 13.9
to 18.8%, and 0.25 to 0.43, respectively. The mean values of min , max and a f are 5.85%, 15.5%
and 0.36.

Experimental stiffness matrix


In order to estimate a good parametric model of the structure for this and future research, the stiff-
ness matrix K of the six-storey base model M1 was determined experimentally. The experimental
procedure used in the determination of K is theoretically trivial, but experimentally quite tedious.
It first estimates each column of the flexibility matrix J , by applying a single force in each of
the degrees of freedom at the nominal CM of the structure, and then, by carefully measuring the
displacements at the CM of each floor. Repeating this procedure for all degrees of freedom of the
system, the complete flexibility matrix is obtained. However, due to unforeseen inaccuracies in
the experimental procedure and sensors, the resulting matrix is not exactly symmetric. Therefore,
the best symmetric flexibility matrix can be defined as
Jsym = 12 (J + J T ) (7)

Copyright q 2006 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
DOI: 10.1002/eqe
1884 I. J. VIAL ET AL.

and the experimental stiffness matrix K is obtained by inversion of J , i.e.


−1
K exp = Jsgm (8)

By using K exp , the estimated vibration periods of the nominally symmetric structure are presented
in Table II. These vibration periods are compared with those obtained by using the theoretical
stiffness matrix for the model (Table I). The observed differences are less than 11.2% for the first
ten modes. Shown also in Table I are the estimated uncoupled torsional-to-lateral frequency ratio
x  for the mass-asymmetric systems, with K exp .

Damper design
Dampers were designed for the base configuration M1 , with a symmetrical plan distribution of their
capacities, the identified stiffness matrix described above, and the heightwise distribution proposed
by Uliarte [8]. Although not a main focus of this article, the design starts with an estimation of
the global stiffness and supplemental damping to be added to the structure to achieve a desired
performance. The performance objective in this case was to reduce the standard deviation of storey
drifts to a 45% of its original value. Notice that in this case, as opposed to a typical building design,
the building model and ground excitation have less uncertainty. The ground motion considered
in the design is the El Centro earthquake (El Centro, 1940) scaled down to 30% of its peak
acceleration. The earthquake response was estimated analytically, and the response reduction, i.e.
the ratio between the standard deviation of storey drifts with and without dampers was computed
(Table III).

Table II. Building periods from analytical and identified stiffness matrix.
Mode # 1 2 3 4 5 6 7 8 9 10

Direction  X Y  X Y  X  Y
Texp (s) 0.425 0.358 0.261 0.143 0.116 0.084 0.079 0.066 0.058 0.049
Tanal (s) 0.381 0.357 0.253 0.127 0.116 0.085 0.075 0.066 0.053 0.049
Error (%) 10.35 0.279 3.07 11.19 0 1.19 5.06 0 8.62 0

Table III. Analytical and experimental standard deviation of interstorey drift for
model M1 and the El Centro base motion.
Total damper Analytical, Experimental, Estimation
Storey capacity (N ) (d)
 / (d)
 / error (%)

1 43.2 0.445 0.558 20.21


2 109.9 0.429 0.495 13.38
3 96.1 0.433 0.504 14.33
4 66.7 0.438 0.515 14.95
5 39.2 0.437 0.580 24.62
6 53 0.433 0.598 27.68

Mean 0.436 0.54 19.19

Copyright q 2006 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
DOI: 10.1002/eqe
BALANCE OF PLAN-ASYMMETRIC STRUCTURES WITH FRICTIONAL DAMPERS 1885

Response reductions may be computed for different values of supplemental stiffness (parameter
) and damping (parameter ) and plotted in the so-called iso-performance curves presented else-
where [9]. In this case, the design of damper capacities were based on the response of the base
model M1 . The performance objective was to reduce in 55% the maximum standard deviation
of storey drift; this was achieved by placing a global supplemental damping of  = 15.2%. Since
ideal FD do not add equivalent stiffness to the structure, the change in vibration frequencies is
negligible. The resulting heightwise distribution of damper capacities is presented in Table III.
Of the total capacity 61% was allocated in the lower 3 stories, and the remaining 39% in the
top 3 stories. Capacities were distributed symmetrically on resisting planes A and B, providing
a total damper capacity of only 5% of the total weight of the structure. This small supplemental
capacity will enable to torsionally balance only some of the asymmetric models considered herein,
but it was intentionally kept constant to show the different dynamic behaviour of the models.
Furthermore, the required FD capacities were achieved by experimental calibration of the normal
forces on the dampers prior to their installation in the building. Damper calibration was performed
at the building nominal fundamental frequency of 2.78 Hz.
By using the identified stiffness matrix and nominal mass matrix for the structural system and
the true damper capacities, the experimental performance of the building model is compared with
the theoretical response of the structure in Table III. The unforeseen errors in the theoretical
prediction of interstorey drift vary from 13.4 to 27.7% and can be due to several factors, including
the unpredictable interaction between the building and shaking table and tracking errors in the
input ground motions. It is important to remark that since the damper design is based on the El
Centro ground motion, such design will not necessarily be optimal for other ground motions and
model configurations.

MASS-ASYMMETRIC SYSTEMS

A summary of the experimental El Centro (scaled to 30%) and Newhall (scaled to 20%) responses
are presented in Table IV. The responses considered are: (1) the peak and standard deviation
values of the lateral displacements at resisting planes A and B, u A and u B , (2) the building
plan rotational displacement L x u  , (3) the maximum of the torsional index  (%), and (4) the
correlation coefficient x between the lateral displacement and building plan rotation. For the sake
of brevity, the results presented in Table IV correspond to the 4th-storey response; other stories
follow similar trends.
Several observations can be stated from these results. By looking first to the base configuration
M1 with and without dampers it is concluded that FD reduce the peak displacement in resisting
plane A from 0.41 to 0.26 cm (37%) and from 1.21 to 0.86 cm (29%) for both earthquakes, respec-
tively. This reduction is somewhat less than the 0.55 originally specified for the design, but it is still
quite significant given the 5% FD capacity introduced in the structure. Please notice that for the
same M1 configuration and the structure without dampers subjected to EI Centro ground motion,
the displacement values of resisting planes A and B are significantly different. The asymmetry
shown by the benchmark model M1 , which may be attributed to several experimental factors, does
not cause an inconvenience since the purpose of the proof-of-concept model is just to compare
the response of the models with and without dampers. By including the dampers, the correlation
between the lateral and torsional motions decreases; indeed, this correlation coefficient decreases
from x = 0.81 to x = 0.70 with FD. Thus, dampers improve the response of the structure, but
their capacity and location do not allow to reach weak torsional balance.

Copyright q 2006 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
DOI: 10.1002/eqe
1886

Copyright q
Table IV. Peak displacement responses of the 4th-storey for: (a) El Centro (30%); and (b) Newhall (20%) ground motions.

(a) El Centro (b) Newhall


uA uB bu  A B  uA uB bu  A B 
(cm) (cm) (cm) (cm) (cm) (%) x (cm) (cm) (cm) (cm) (cm) (%) x
M1 w/o FD 0.4125 0.3145 0.0994 0.1492 0.0955 27.73 0.8137 1.2091 0.7499 0.4224 0.4419 0.2357 49.57 0.8247
FD, ed = 0 0.2556 0.2107 0.0575 0.0612 0.0446 25.52 0.7004 0.8608 0.6181 0.1690 0.2133 0.1429 23.66 0.8342
Mass- 0.2684 0.4704 0.1394 0.0792 0.1563 37.75 0.5995 1.1060 0.3306 0.1995 0.4036 39.90

2006 John Wiley & Sons, Ltd.


M2 w/o FD −0.9048 −0.9665
eccentric FD, ed = 0 0.1792 0.3188 0.0987 0.0358 0.0760 39.77 −0.8665 0.5728 0.9252 0.2918 0.1454 0.2697 39.89 −0.9437
FD, ed = 0.25L x 0.2091 0.2936 0.0798 0.0350 0.0604 32.20 −0.7843 0.5817 0.8393 0.2726 0.1460 0.2291 44.56 −0.9566
FD, ed = 0.50L x 0.2019 0.2789 0.1123 0.0381 0.0583 51.62 −0.5457 0.6232 0.6357 0.2736 0.1695 0.1972 45.39 −0.9340
M3 w/o FD 0.1699 0.5064 0.2211 0.0595 0.2494 70.45 −0.9606 0.3591 1.2038 0.5007 0.1544 0.6564 67.91 −0.9866
FD, ed = 0 0.1140 0.3948 0.1697 0.0226 0.0968 75.27 −0.9382 0.3072 0.9718 0.4273 0.0884 0.3464 67.95 −0.9730
FD, ed = 0.25L x 0.0946 0.2335 0.1193 0.0203 0.0551 93.30 −0.7874 0.3883 0.6040 0.3493 0.0955 0.1526 89.88 −0.9329
FD, ed = 0.50L x 0.1601 0.2203 0.1553 0.0357 0.0511 110.58 −0.3477 0.3712 0.5386 0.3497 0.1020 0.1174 93.72 −0.8618
I. J. VIAL ET AL.

S1 w/o FD 0.3570 0.6396 0.3071 0.1319 0.2739 72.82 −0.6964 0.8873 1.1277 0.6487 0.2728 0.5216 73.26 −0.6349
FD, ed = 0 0.1767 0.4095 0.1506 0.0399 0.0883 58.09 −0.7204 0.5053 1.0994 0.5565 0.1207 0.3349 75.98 −0.8452
FD, ed = 0.25L x 0.3299 0.2225 0.1279 0.0693 0.0515 48.43 0.3482 0.7497 0.7139 0.4378 0.1749 0.1932 61.34 −0.1023
FD, ed = 0.50L x 0.3704 0.1839 0.1523 0.1077 0.0443 57.66 0.7417 0.7781 0.7115 0.4456 0.2044 0.1839 61.25 0.1071
Stiffness- S2 w/o FD 0.1971 0.4122 0.1762 0.0426 0.1225 71.53 −0.8113 0.4506 1.0344 0.5222 0.0903 0.3073 97.93 −0.8546
eccentric FD, ed = 0 0.1364 0.2700 0.1139 0.0304 0.0472 60.68 −0.4606 0.3474 0.6861 0.2966 0.0568 0.1612 65.39 −0.8417
FD, ed = 0.25L x 0.1558 0.1256 0.0307 0.0268 0.0278 21.97 −0.0760 0.3814 0.5199 0.1600 0.0760 0.1157 38.01 −0.5588
FD, ed = 0.50L x 0.2011 0.1796 0.0816 0.0373 0.0332 54.67 0.1387 0.4624 0.7075 0.3828 0.0946 0.1571 83.74 −0.4765
S3 w/o FD 0.2182 0.3399 0.1775 0.0386 0.1492 101.11 −0.8749 0.4562 1.0237 0.6886 0.0930 0.3595 124.09 −0.8744
FD, ed = 0 0.1474 0.2842 0.1690 0.0313 0.0710 79.86 −0.6833 0.3091 1.0168 0.5336 0.0560 0.2963 87.71 −0.9421
FD, ed = 0.25L x 0.1713 0.2348 0.1167 0.0287 0.0474 81.80 −0.4662 0.4320 0.7137 0.5121 0.0810 0.1660 116.88 −0.6207
FD, ed = 0.50L x 0.2127 0.2393 0.1331 0.0378 0.0440 89.45 −0.1529 0.4873 0.6545 0.4953 0.1000 0.1485 111.64 −0.3826

DOI: 10.1002/eqe
Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
BALANCE OF PLAN-ASYMMETRIC STRUCTURES WITH FRICTIONAL DAMPERS 1887

Configuration M Configuration M Configuration M


1 2 3
1.2

0.6

(1) w/o FD
0
Edge displacement (cm)

σuA=0.442cm σ =0.2cm σ =0.154cm


uA
uA
σ =0.236cm σ =0.404cm σuB=0.657cm
uB uB

1.2

(2) with FD, e =0


Edge A

d
0.6
Edge B
0
σuA=0.214cm σuA=0.145cm σuA=0.0885cm
σuB=0.143cm σuB=0.27cm σuB=0.346cm
0 3 6 9 12 15

(3) with FD, ed =0.25 Lx


Time (s) 1.2

0.6

0
Edge displacement (cm)

σuA=0.146cm σuA=0.096cm
σuB=0.229cm σuB=0.153cm

x
1.2

(4) with FD, e =0.50 L


0.6

d
0
σ =0.169cm σ =0.102cm
uA uA
σuB=0.197cm; σuB=0.117cm
0 3 6 9 12 15 0 3 6 9 12 15
Time (s) Time (s)

Figure 5. Measured edge displacements at the 4th storey of mass-asymmetric structures


subjected to the Newhall ground motion (20%).

For the intermediate mass-eccentric system M2 , peak displacements of the model with FD at
plane A and êd = ed /L x = 0 are reduced relative to the case without dampers in 33 and 4.5%
for the two ground motions, respectively. It is clear that FD are less effective in reducing the
building response to impulsive-type ground motions. Two other locations of the dampers were
tested êd = 0.25 and 0.5. As the damper eccentricity increases, the correlation between lateral
displacement and plan rotation does reduce considerably in the case of El Centro, and, hence,
the standard deviation of the displacements of planes A and B become more similar, although
far from being optimally balanced. In the case of Newhall, the lowest correlation is achieved at
the damper location êd = 0.5, implying that for this specific damper design and ground motion,
the dampers are not able to control the torsional response of the building plan. This is consistent
with a previous theoretical investigation [2] which showed that for impulsive-type ground motions,
larger capacities should be used to torsionally balance the building response. Furthermore, for the
strong mass-eccentric system M3 and êd = 0, lateral displacements in resisting plane A are reduced
in 33 and 14.4% for the two ground motions. As before, the correlation between building plan
rotation and lateral displacement at the GC decreases as the damper eccentricity increases, reaching
a minimum of x = −0.348 for the El Centro ground motion and êd = 0.5. Such reduction in
motion correlation is much less in the case of the impulsive Newhall record. Nevertheless, the
important observation is that the trends follow exactly the expected theoretical behaviour, i.e. as
the correlation between plan rotation and translation reduces, the standard deviation of the edge

Copyright q 2006 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
DOI: 10.1002/eqe
1888 I. J. VIAL ET AL.

displacements become more similar in magnitude, leading to a torsionally balanced state of the
system in the weak sense.
Results presented in Table IV are complemented by those presented next in Figures 5–8. Shown
in Figure 5 are the 4th-storey edge displacements for the mass-eccentric building configurations,
the different damper locations, and the Newhall record. Included as a reference in these figures
are the values of the standard deviation of the displacements of resisting planes A (west) and
B (east). The traces show that in spite of the small damper capacity, the FD are able to mod-
ify considerably the response of the structure. Of course, damping reduces the duration of the
structural vibration, reduces the amplitude of these vibrations, and most important, modifies the
relative amplitudes of the displacements measured at both edges. The relationship between these
response traces and the results presented in Table IV are apparent. For instance, the minimum cor-
relation for M2 (although quite large) occurs for êd = 0.5, which is quite consistent with the closer
similitude between the amplitude of the traces of displacements in resisting planes A and B for
this damper eccentricity. These results will become clearer as we look into the stiffness-eccentric
configurations.

with FD,
w/oeFD
=0.25L with FD, e =0 w/o FD with FD, e =0 w/o FD with FD, e =0
d x d d d
6 6 6 6 6 6

5 5 5 5 5 5

4 4 4 4 4 4
Storey

3 3 3 3 3 3

2 2 2 2 2 2

1 1 1 1 1 1
0 0.3 0.6 0 0.3 0.6 0 0.3 0.6 0 0.3 0.6 0 0.5 1 0 0.3 0.6
σu (cm) σu (cm)
(a) with FD, e =0.25L with FD, e =0.5L with FD, e =0.25L with FD, e =0.5L
d x d x d x d x
6 6 6 6

5 5 5 5

4 4 4 4
Storey

3 3 3 3

2 2 2 2
E
B
W
A
1 1 1 1
0 0.3 0.6 0 0.3 0.6 0 0.3 0.6 0 0.3 0.6
(b) σ (cm) σ (cm) (c) σ (cm) σ (cm)
u u u u

Figure 6. Standard deviation of measured edge displacements of mass-asymmetric structures subjected to


the Newhall ground motion (20%): (a) M1 ; (b) M2 ; and (c) M3 .

Copyright q 2006 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
DOI: 10.1002/eqe
BALANCE OF PLAN-ASYMMETRIC STRUCTURES WITH FRICTIONAL DAMPERS 1889

Configuration M Configuration M Configuration M


1 2 3
0 0 0
10 10 10

10 10 10
E[u(p)(t,e )2]
d
y

10 10 10

10 10 10
0 70 140 0 70 140 0 70 140
(a)

Configuration S Configuration S Configuration S


1 2 3
0 0 0
10 10 10

10 10 10
E[uy (t,e ) ]
2
d
(p)

10 w/o FD 10 10
with FD, e =0 cm
d
with FD, e =0.25L
d x
with FD, e =0.5L
d x

10 10 10
0 70 140 0 70 140 0 70 140

(b) Plan location (cm)

Figure 7. Experimental estimation of empirical centre of balance (ECB) of the 4th


storey for the building configurations subjected to the Newhall ground motion (20%):
(a) mass-eccentric; and (b) stiffness-eccentric.

A more visual form of looking at the effect of FD in controlling the displacement response of
the mass-eccentric systems is presented in Figure 6. In this figure, vertical profiles of the standard
deviation of the displacements in planes A and B are presented. If the two lines are far apart, it
means that the displacement demand at both building edges differs considerably. For instance, in
the base system M1 , experimental results show that the discrepancy between the displacements
at both edges is partially corrected by the symmetric inclusion of FD. The same occurs for the
other mass-eccentric configurations M2 and M3 , which show the positive effect that FD have in
balancing the demand at both resisting planes. Although weak torsional balance is not achieved as
explained above, FD help in reducing the magnitude of building displacements and the difference
between displacements in planes A and B. Please recall that the dampers were designed for the El
Centro ground motion, and, hence, such design is not necessarily optimal for the Newhall record.
Although a more robust design could have been obtained by using a design spectra, the idea was
to test the robustness of the design procedure based on the iso-performance curve and a single
ground motion.

Copyright q 2006 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
DOI: 10.1002/eqe
1890 I. J. VIAL ET AL.

Configuration M 1 Configuration M 2 Configuration M 3


6 6 6

5 5 5

4 4 4
Storey

3 3 3

2 2 2

1 1 1
0 6 12 18 24 30 0 6 12 18 24 30 0 6 12 18 24 30
Shear (V/W) %

w/o FD
6 with FD, ed=0 6 6
with FD, ed=0.25Lx
with FD, ed=0.5Lx
5 5 5

4 4 4
Storey

3 3 3

2 2 2

1 1 1
0 3 6 9 12 15 0 3 6 9 12 15 0 3 6 9 12 15
(a) (b) Torque (T/WLx) % (c)

Figure 8. Maximum normalized storey shear and torque for mass-asymmetric configurations subjected to
the Newhall ground motion (20%): (a) M1 ; (b) M2 ; and (c) M3 .

Shown in Figure 7 is the mean-square value of displacement at different locations along the
building plan (assuming a rigid floor diaphragm). Traces of the mean-square displacements are
arbitrarily shown between ±140 cm (which corresponds to include the rigid horizontal arms);
in this case the building displacements occur in the range of ±43 cm. The response of the
M1 –M3 mass-eccentric configurations are presented for the case of the Newhall record. As pre-
sented elsewhere [2], the location of the minimum of these curves corresponds to the location of
the ECB and is optimal when it coincides with the GC of the plan ( p ∗ = 0). Therefore, from part
(a) of this figure, it is easy to see that for mass-eccentric systems and the FD capacity selected
of 0.05W , we do not achieve weak torsional balance. For instance, in the case of configuration
M3 , the damper location leading to the ECB closest to the optimal is êd = 0.5. It is apparent from
this analysis, that for an impulsive ground motion like Newhall, larger damper capacities would
be needed in order to control the torsional response of these mass-eccentric configurations. An
example of the weak torsional balance of the M2 mass-asymmetric will be presented later on in
the article.

Copyright q 2006 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
DOI: 10.1002/eqe
BALANCE OF PLAN-ASYMMETRIC STRUCTURES WITH FRICTIONAL DAMPERS 1891

Finally, shown in Figure 8 are the normalized storey shears and torques measured in the structure
for the Newhall record. Response reduction factors are quite notorious for the nominally symmetric
configuration M1 and less important for the M2 and M3 configurations. In all cases, storey shears
and torques are larger for the lower stories. Besides, larger response reductions are observed for
storey torques, say 39%, than for storey shears (13%). These reductions are quite significant,
especially since we have included a relatively small frictional capacity in the structure. Therefore,
for these models FD are capable of controlling not only the building plan displacements, but also
(in a less effective way) the storey shear and torque. It is also apparent that a planwise damper
distribution that works well for storey shears does not necessarily work well for storey torques.

STIFFNESS ASYMMETRIC SYSTEMS

In this section, analogous results as those presented previously in Table IV and Figures 5–8 are
obtained and interpreted from the measured response of stiffness-asymmetric configurations S1 –S3 .
In this particular case, the reader should look at the lower 3 blocks of Table IV and Figures 7,
and 9–11.
Again, several observations are stated from these results. Because in this experiment the flexible
edge coincides with the location of resisting plane B (east), responses will be arbitrarily compared

Configuration S Configuration S2 Configuration S3


1
1.2

(1) w/o FD
0.6

0
σuA=0.273cm σ =0.0903cm σ =0.093cm
uA uA
σ =0.522cm σuB=0.307cm σuB=0.359cm
uB

1.2

(2) with FD, ed=0


Edge A
0.6
Edge B
0
Edge displacement (cm)

σ =0.121cm σuA=0.0568cm σuA=0.0561cm


uA
σ =0.335cm σ =0.161cm σ =0.296cm
uB uB uB

(3) with FD, ed=0.25 Lx


1.2

0.6

0
σuA=0.175cm σuA=0.076cm σuA=0.081cm
σ =0.193cm σ =0.116cm σuB=0.166cm
uB uB
(4) with FD, ed=0.50 Lx

1.2

0.6

0
σ =0.204cm σuA=0.0946cm σuA=0.1cm
uA
σ =0.184cm σuB=0.157cm σ =0.148cm
uB uB
0 3 6 9 12 15 0 3 6 9 12 15 0 3 6 9 12 15
Time (s)

Figure 9. Measured edge displacements of the 4th of stiffness-asymmetric configurations


subjected to the scaled Newhall ground motion.

Copyright q 2006 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
DOI: 10.1002/eqe
1892 I. J. VIAL ET AL.

w/o FD with FD, ed=0 w/o FD with FD, ed=0 w/o FD with FD, ed=0
6 6 6 6 6 6

5 5 5 5 5 5

4 4 4 4 4 4
Storey

3 3 3 3 3 3

2 2 2 2 2 2

1 1 1 1 1 1
0 0.4 0.8 0 0.25 0.5 0 0.25 0.5 0 0.25 0.5 0 0.25 0.5 0 0.25 0.5

with FD, e =0.25L with FD, e =0.5L with FD, e =0.25L with FD, e =0.5L with FD, e =0.25L with FD, e =0.5L
d x d x d x d x d x d x
6 6 6 6 6 6

5 5 5 5 5 5

4 4 4 4 4 4
Storey

3 3 3 3 3 3

2 2 2 E 2 2 2
B
W
A
1 1 1 1 1 1
0 0.25 0.5 0 0.25 0.5 0 0.25 0.5 0 0.25 0.5 0 0.25 0.5 0 0.25 0.5
(a) σ (cm) σ (cm) (b) σ (cm) σ (cm) (c) σ (cm) σ (cm)
u u u u u u

Figure 10. Standard deviation of measured edge displacements of stiffness-asymmetric structures subjected
to the Newhall ground motion (20%): (a) S1 ; (b) S2 ; and (c) S3 .

only for that plane. For the El Centro (30%) ground motion, peak displacement reduction factors
computed between the structure with dampers and êd = 0, and the structure without dampers, are
36, 35, and 16% for the S1 –S3 configurations, respectively. Please notice that the response reduction
factors for resisting plane A tend to be larger than these values. This reduction is somewhat less
than the 0.55 specified originally for the design, but it is still quite significant given the small global
FD damper capacity introduced in the structure. As opposed to the previous case, for the S1 and S2
stiffness-asymmetric structures subjected to El Centro, the correlation coefficient x changes its
sign for the range of values considered for the normalized damper eccentricity êd . For instance, in
the S1 structure, the correlation coefficient crosses zero within the damper eccentricity range êd = 0
to êd = 0.25. Indeed, the correlation coefficient for êd = 0.25 is x = 0.35, which is low, leading
to similar values of the standard deviation of the building edge displacements. A similar situation
occurs with the S2 structure, which optimum will be within the range êd = 0.25 and êd = 0.5. In
this case, since the correlation coefficient for êd = 0.25 is x = − 0.08, i.e. essentially 0, the values
of the standard deviation of the edge displacements, say 0.027, are practically the same for both
edges. This result can be stated in a different form: for êd = 0.25, the ECB almost coincides with

Copyright q 2006 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
DOI: 10.1002/eqe
BALANCE OF PLAN-ASYMMETRIC STRUCTURES WITH FRICTIONAL DAMPERS 1893

Configuration S1 Configuration S2 Configuration S3


6 6 6

5 5 5

4 4 4
Storey

3 3 3

2 2 2

1 1 1
0 5 10 15 20 25 0 5 10 15 20 25 0 5 10 15 20 25
Shear (V/W) %

w/o FD
with FD, ed=0
6 with FD, ed=0.25Lx 6 6
with FD, ed=0.5Lx

5 5 5

4 4 4
Storey

3 3 3

2 2 2

1 1 1
0 3 6 9 12 15 0 3 6 9 12 15 0 3 6 9 12 15
(a) (b) Torque (T/WL x) % (c)

Figure 11. Maximum normalized storey shear and torque for stiffness-asymmetric configurations subjected
to the Newhall ground motion (20%): (a) S1 ; (b) S2 ; and (c) S3 .

the GC of the building plan and the mean-square value of the displacement at planes A and B are
the same. Therefore, this proves experimentally that the concept of the ECB is correct.
This result also carries over to the case of configuration S1 and the Newhall record. In this case,
the optimal damper eccentricity value is in between the range êd = 0.25 and êd = 0.5. However, for
configurations S2 and S3 , no zero crossing is observed for the correlation x and, hence, larger
damper capacities should have been needed to control the torsional response due to this impulsive
ground motion. The same observation carries over to the very asymmetric S3 configuration and El
Centro ground motion, which would require larger damper capacities to place the ECB at the GC
of the building plan.
Notice also for the above experimental results and configurations, for which weak torsional
balance is achieved, that the displacement reduction factors increase with the optimal location
of êd . Just as an example, consider the S2 configuration and El Centro ground motion, the dis-
placement reduction at resisting plane B increases from 35 to 70% by appropriately locating the

Copyright q 2006 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
DOI: 10.1002/eqe
1894 I. J. VIAL ET AL.

dampers in plan. The torsional index  also shows smaller values for the optimal location of the
dampers, reaching, for instance, 0.22 for the S2 configuration and El Centro. This implies that the
largest difference between the two edge displacements is 22% of the maximum deformation at
the CM.
Let us revisit Figure 7 in order to state graphically the same results that were stated in Table IV.
Since the figure presents only the results for the Newhall record, the only case where we
strictly achieve torsional balance with the given 0.05W FD capacity is S1 . Notice that the
two displacement parabolas corresponding to êd = 0.25 and êd = 0.5 have vertices that are close to
0 recall that the location of the vertex of the parabola gives the location of the ECB. As discussed
before, the optimum value of ed is somewhere in between and could be achieved by simply cal-
ibrating the damper frictional capacity. Of course, this is not practical since the designer does not
know, as in this case, the ground motion a priori. It is apparent from the response of structures
S2 and S3 to the Newhall record that no exact optimum is achieved in this case and that the best
option would be to place the dampers at êd = 0.5. Another interesting observation inferred from
Figure 7 is that the curvature of the parabolas shown is controlled by the planwise distribution
of the damper capacity. The goal is to distribute such capacity across the building plan as far as
possible from the ECB. In the limit (fictitious), if the capacity is placed very far from the ECB
(located at the GC), the structure will tend to behave as a nominally symmetric building plan.
Shown in Figure 9 are the 4th storey displacements in resisting planes A and B. As before,
two traces far apart from one another mean that the displacement demand at both building edges
differs considerably. The first observation is the large displacement reduction that can be achieved
by manipulating the planwise distribution of dampers. For instance, for structure S1 , which is
the only one that has an optimum damper location, the maximum vibration amplitudes (or the
corresponding standard deviations) become very similar for 0.25  êd  0.5. This is consistent with
the results presented earlier in Figure 7 for the ECB. Although no optimum damper location is
obtained for the S2 and S3 structures, the condition 0.25  êd  0.5 leads in both cases to good
displacement reduction factors and reasonable mean-square values of the response.
The standard deviation values of the heightwise profile of displacements of the stiffness-eccentric
systems considered are presented in Figure 10. Notice again the torsional balance caused by the FD
in the S1 structure; the standard deviation values at both edges are essentially identical, implying
weak torsional balance. The results are not as impressive for the S2 and S3 structures; however, the
reader may see that the ed = 0.25 and ed = 0.5 conditions are quite reasonable designs for these
structures, respectively. As a more general result, the response reductions observed in the building
height are very consistent and significant. These results imply that the idea of designing a coupled
structure to achieve weak torsional balance, may be an adequate and economic criterion. Please
recall that the dampers were designed in this case for the El Centro ground motion and not for
the Newhall record.
Finally, shown in Figure 11 are the normalized storey shears and torques measured in the 3
structural configurations for the Newhall record. In general, storey shear and torque reduction
factors are less significant than displacement reduction factors. In all cases storey shears and
torques tend to be larger in the lower stories. Besides, larger response reductions are observed for
storey torques, say between 33% and 50%, than for storey shears, say 25%. These reductions are
quite significant in spite of including supplemental frictional capacity in the structure. However,
the damper distribution in S1 leading to the smallest storey shear (êd = 0) is not the same as the
one that leads to weak torsional balance (0.25  êd  0.5). Indeed, the condition of weak torsional
balance leads to similar storey shears as the condition without dampers, but storey torques are

Copyright q 2006 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
DOI: 10.1002/eqe
BALANCE OF PLAN-ASYMMETRIC STRUCTURES WITH FRICTIONAL DAMPERS 1895

decreased considerably. It is important to note that in all cases and stories of this model, the FD
do not amplify the storey shear and torque of the original structure. Therefore, FD are capable
of controlling not only the building plan displacements, but also the storey shear and torque. It is
concluded again that a damper distribution that works well for storey shears does not necessarily
work well for storey torques.

ANALYTICAL EXAMPLES FOR MULTISTOREY BUILDING MODELS M2 AND S1

Since the damper design for the tested building model was done for the El Centro ground motion,
the reader may ask what would it be such design for the Newhall record. Would it be possible
to achieve weak torsional control of the mass- and stiffness-eccentric configurations that were
uncontrolled? The answer is affirmative and it will be shown by controlling the M2 mass-eccentric
model for the Newhall ground motion. A more general procedure to design multistorey structures
with FD will be presented in a sequel article. In general, this proposed design procedure will
consider 7 steps: (i) construct the iso-performance curves for the structure and ground motion
considered; (ii) select a design performance for the structure; (iii) identify the global dissipation
capacity required for the structure; (iv) estimate the inelastic demand on each damper; (v) select
the damper capacities; (vi) define their optimal location in plan; and (vii) validate the response by
inelastic analysis.
An iso-performance curve is a locus of equal response reduction factor in the space defined by
co-ordinates of equivalent supplemental stiffness and damping for a given ground motion. Thus,
several pairs of equivalent stiffness and damping may lead to a similar performance of the structure.
In this particular case, since ideal FD are purely dissipative, the desired design pairs will be located
along the ordinates of the iso-performance graph [9]. A section of the iso-performance graphs for
no increase in stiffness is presented in the first row of plots of Figure 12. Each point represents a
weakly balanced multistorey system with equivalent linear viscous damping ratios eq and subjected
to the Newhall and El Centro record ((a) and (b), respectively). The ordinates show the minimum
reduction factors for the standard deviation of edge drifts, normalized to the standard deviation
of storey drifts at the GC of the system without FDs. The heightwise distribution of equivalent
viscous damping in the structure was assumed to be proportional to the first mode interstorey drift.
The optimal planwise damper location (i.e. ECBi=1:6 = 0) was estimated numerically using the
Optimization Toolbox of Matlab [7]. Shown in parts (a) and (b), are the edge reductions considering
model M2 subjected to the Newhall and El Centro ground motions, respectively. Two damping
configurations are considered for both cases, a symmetrical configuration (ed = 0) and the optimal
configuration. These curves show that there exists a minimum damper capacity to achieve torsional
balance in the structure (hatched zone) that is much larger than the 0.05W used experimentally.
Besides, the optimal configuration of dampers leads to equal demand among resisting planes as
shown by the superposition of the traces for edges A and B.
On the other hand, in parts (c) and (d) of Figure 12, the true inelastic response of the 6th
storey with and without FDs is plotted for the S1 and M2 models, respectively, and subjected
to the Newhall ground motion. Part (c) validates the experimental results presented in Figure 7
(part (b)) for the S1 structure. For such case, torsional balance is achieved by considering the same
total damper capacity 0.05W that was used for the symmetric experimental model and for the El
Centro ground motion with ed = 0.5L x . Also, part (d) of this figure shows the response reduction
and torsional balance for the M2 model, which could not be balanced experimentally (Figure 7,

Copyright q 2006 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
DOI: 10.1002/eqe
1896 I. J. VIAL ET AL.

1
Edge A (e =0)
d
Edge B (e =0)
d
Drift Reduction (σd /σ0 )
CG

Edge A (Optimal Config.)


Edge B (Optimal Config.)
A,B

0.75

ed=0 e*
d
e*d e =0
d
0.5

0.25
0.1 0.2 0.3 0.4 0.1 0.2 0.3 0.4
(a) Supplemental Damping (ξ) (b) Supplemental Damping (ξ)

2
Edge A Edge A
Edge B Edge B
1

0
δ (cm)

System S w/o FD System M2 w/o FD


1
1

0
FD in optimum position FD in optimum position

0 3 6 9 12 15 0 3 6 9 12 15
(c) time (s) (d) time (s)

(e) (f)

Figure 12. Results from analytical building examples: (a, d, f ) M2 —Newhall;


(b) M2 —El Centro; and (c, e) S1 —Newhall.

Copyright q 2006 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
DOI: 10.1002/eqe
BALANCE OF PLAN-ASYMMETRIC STRUCTURES WITH FRICTIONAL DAMPERS 1897

Table V. Analytical results for the M2 building model subjected to Newhall (20%).
Without dampers Dampers in optimum location
Storey x  Edge A Edge B F y (kg) ed (cm) x  Edge A Edge B

1 −0.6537 0.764 1.422 60.68 28.921 −0.0235 0.509 0.520


2 −0.6551 0.776 1.361 18.73 25.544 0.0078 0.505 0.502
3 −0.6565 0.793 1.305 19.35 39.764 0.0051 0.480 0.485
4 −0.6573 0.780 1.335 91.37 25.691 0.0138 0.490 0.481
5 −0.6582 0.771 1.353 22.29 54.132 0.0230 0.465 0.457
6 −0.6591 0.840 1.203 21.65 45.062 0.0322 0.394 0.387

Mean −0.6566 0.787 1.330 0.0097 0.474 0.472


F y/W 28.77 %

part (a)). However, by using a total damper capacity of 29% of the weight of the structure and the
heightwise distribution presented in Table V, the system can be balanced in the weak sense.
Moreover, in parts (e) and (f) of Figure 12, the correlation coefficients (x  ) for systems S1
and M2 are presented. The correlation x  is reduced significantly by selecting the proper damper
capacities and location. For example, for the 6th storey of the Si model and ed = 0.5L x , x 
decreases from −0.836 for the system without dampers to −0.090 for the case with optimal
damper location. Finally, for the M2 model and optimal capacities and damper location, x 
decreases from −0.659 to 0.032.

CONCLUSIONS

In this investigation, a 6-storey asymmetric model was used to prove experimentally the concept
of weak torsional balance introduced earlier [2]. Experimental response modification factors were
computed for mass and stiffness asymmetric models with FD subjected to impulsive and non-
impulsive ground motions. All inelasticity present in the models is localized in the dampers.
Although results presented herein are limited to only 6 different asymmetric configurations, they
show trends that are intuitively appealing and coincide with results of analytical studies carried
on earlier for a much larger number of single and multistorey structures [2, 10]. Measured results
have shown (in 3 out of the 6 cases considered) that when the position of the ECB coincides
with the location of the GC, the mean-square values of the displacements at opposite edges of
an asymmetric structure will be identical; a condition that has been denoted previously as weak
torsional balance. It was apparent that the stiffness-eccentric systems could be torsionally balanced
with the frictional capacity selected of 0.05W ; however, much larger values would be required for
structures with mass asymmetry and impulsive ground motions. For instance, torsional balance
in the M2 model subjected to the Newhall record would be achieved for a frictional capacity 6
times larger. Since the optimal damper locations depend on the ground motion characteristics, their
selection in design should be based on averages obtained for several ground motions. In spite of
the small supplemental damper capacity introduced, frictional dampers allow peak displacement
reductions of elastic structures that range between 20–40% without increasing the storey shears and
torques—indeed, storey torques are reduced, in general, by introducing the dampers. Consequently,

Copyright q 2006 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
DOI: 10.1002/eqe
1898 I. J. VIAL ET AL.

the results of this experimental investigation are promissory in showing that frictional dampers are
able to control torsion in multistorey asymmetric structures by changing the location of the ECB
through the variation of the normal forces in the FD.

ACKNOWLEDGEMENTS
This investigation was funded by the Chilean National Fund for Science and Technology, Fondecyt
initially through Grant #1020774 and later through Grant #1050691. The authors also acknowledge the
two anonymous reviewers that contributed through their comments to improve this manuscript.

REFERENCES
1. de la Llera JC, Chopra AK, Almazán JL. Three-dimensional inelastic response of an RC during the Northridge
earthquake. Journal of Structural Engineering (ASCE) 2001; 127(5):482–489.
2. de la Llera JC, Almazán JL, Vial IJ. Torsional balance of plan-asymmetric structures with frictional dampers:
analytical results. Earthquake Engineering and Structural Dynamics 2005; 34:1089–1108.
3. Goel RK. Effects of supplemental viscous damping on earthquake response of asymmetric-plan systems.
Earthquake Engineering and Structural Dynamics 1997; 27:125–141.
4. Grigorian CE, Yang TS, Popov EP. Slotted bolted connection energy dissipators. Earthquake Spectra 1993;
9(3):491–504.
5. Constantinou M, Mokha A, Reinhorn A. Teflon bearings in base isolation, II: modeling. Journal of Structural
Engineering (ASCE) 1990; 116(2):455–474.
6. Steward D. Platform with six degrees of freedom. Proceedings of the Institution of Mechanical Engineers,
vol. 180, Pt. 1.15, London, England, 1965; 371–386.
7. Matlab. Reference Guide. The Mathworks, Inc: Natick, MA, 2004.
8. Uliarte RJ. Sistemas Pasivos de Disipación de Energı́a para la Reducción de Vibraciones en Estructuras: Análisis
y Diseño. School of Engineering, Pontificia Universidad Católica de Chile: Santiago, Chile, 1998.
9. Watt I. Performance based design of structures with energy dissipation devices. Master Thesis, School of
Engineering, Universidad Católica de Chile, Santiago, Chile, 2006.
10. de la Llera JC, Vial I, Almazán JL, Fighetti E. Balanced design of asymmetric buildings with energy dissipation
devices. Proceedings of the 8NCEE, San Francisco, CA, 2006.

Copyright q 2006 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2006; 35:1875–1898
DOI: 10.1002/eqe

You might also like