You are on page 1of 20

EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS

Earthquake Engng Struct. Dyn. 2002; 31:16231642 (DOI: 10.1002/eqe.179)

Earthquake response of elastic SDF systems with non-linear


uid viscous dampers

Wen-Hsiung Lin and Anil K. Chopra;


Department of Civil and Environmental Engineering; 721 Davis Hall, University of California;
Berkeley; CA 94720-1710; U.S.A.

SUMMARY
The steady-forced and earthquake responses of SDF systems with a non-linear uid viscous damper
(FVD) are investigated. The energy dissipation capacity of the FVD is characterized by the supple-
mental damping ratio sd and its non-linearity by a parameter designated . It is found that the struc-
tural response is most eectively investigated in terms of sd and  because (1) these two parameters
are dimensionless and independent, and (2) the structural response varies linearly with the excitation
intensity. Damper non-linearity has essentially no inuence on the peak response of systems in the
velocity-sensitive spectral region, but dierences up to 14% were observed in the other spectral re-
gions. The structural deformation is reduced by up to 25% when sd = 5%; and by up to 60% when
sd = 30%. Non-linear FVDs are advantageous because they achieve essentially the same reduction in
system responses but with a signicantly reduced damper force. For practical applications, a procedure
is presented to estimate the design values of structural deformation and forces for a system with non-
linear FVD directly from the design spectrum. It is demonstrated that the earthquake-induced force in a
non-linear FVD can be estimated from the damper force in a corresponding system with linear FVD, its
peak deformation, and peak relative velocity; however, the relative velocity should not be approximated
by the pseudo-velocity as this approximation introduces a large error in the damper force. Finally, a
procedure is presented to determine the non-linear damper properties necessary to limit the structural
deformation to some design value or the structural capacity for a given design spectrum. Copyright ?
2002 John Wiley & Sons, Ltd.
KEY WORDS: non-linear uid viscous damper; earthquake response

INTRODUCTION

Supplemental damping devices act by dissipating seismic input energy, thereby reducing seis-
mic demands on the structure. Fluid viscous dampers (FVDs) [13] are a type of damping
device. Numerous experimental and analytical investigations have focused on linear FVDs

Correspondence to: Anil K. Chopra, Department of Civil and Environmental Engineering, 721 Davis Hall,
University of California, Berkeley, CA 94720-1710, U.S.A.
E-mail: chopra@ce.berkeley.edu

Received 18 June 2001


Revised 12 December 2001
Copyright ? 2002 John Wiley & Sons, Ltd. Accepted 18 December 2001
1624 W.-H. LIN AND A. K. CHOPRA

[46], because they can be modelled simply by a linear dashpot. While being eective in re-
ducing seismic demands on the structure, linear FVDs may develop excessive damper forces in
applications where large structural velocities can occur, e.g. in long period structures subjected
to intense ground shaking, especially in the near-fault region. Recently, both researchers [79]
and earthquake engineering professionals [10; 11] have begun to focus on FVDs exhibiting
non-linear forcevelocity relationship because of their ability to limit the peak damper force
at large structural velocities while still providing sucient supplemental damping.
Experimental results [1; 1214] have demonstrated that the forcevelocity relation for non-
linear FVDs can be analytically expressed as a fractional velocity power law involving two
damper parameters. Thus, the governing equation for an SDF system becomes non-linear,
which is dicult to parameterize and use in developing fundamental understanding of the ef-
fect of non-linear FVDs on system response. Consequently, a system with non-linear dampers
is usually replaced by an equivalent linear system, with its properties determined using dier-
ent methods: equating the energy dissipated in the two systems [7; 1517]; equating power
consumption in the two systems [8]; replacing the non-linear viscous damping by an array of
frequency- and amplitude-dependent linear viscous models [18]; and random vibration theory
[1922].
The dynamic response of SDF systems with non-linear viscous damping has been analysed
without using the equivalent linear system approximation [9]. Based on extensive numerical
examples, systems with similar damping levels were identied, and then their responses were
compared to evaluate the inuence of damper non-linearity on system response.
This investigation aims to demonstrate how non-linear FVDs aect the dynamic response of
SDF systems. Presented rst are the dynamic characteristics of a non-linear FVD, where we
demonstrate that its energy dissipation capacity can be characterized by supplemental damping
ratio sd , and its non-linearity by a parameter . Thus, non-linear FVDs characterized by the
same sd value but dierent  values dene energy-equivalent dampers. We then demonstrate
that the response of structures with non-linear FVDs are most eectively investigated in terms
of sd and  because (1) these two dimensionless parameters are independent, and (2) the
structural response varies linearly with the excitation intensity. Next, we study the inuence
of supplemental damping and damper non-linearity on steady-forced harmonic vibration and
earthquake of SDF systems. Based on the results obtained for the range of  considered, we
demonstrate the accuracy and identify the limitation of the equivalent linear viscous system
(same sd but  = 1). Finally, for a given earthquake design spectrum we consider two design
issues: (i) how to estimate structural deformation (and forces), damper force, and foundation
shear; and (ii) how to select a non-linear FVD that limits the structural deformation to a
design value.

NON-LINEAR FLUID VISCOUS DAMPER

The force (fD )-velocity (u)


relation for non-linear uid viscous dampers (FVDs) can be
analytically expressed as a fractional velocity power law:
|u|
fD = c sgn(u) (1)
where c is the experimentally determined damping coecient with units of force per velocity
raised to the  power;  is a real positive exponent with typical values in the range of

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:16231642
EARTHQUAKE RESPONSE OF ELASTIC SDF SYSTEMS 1625

0.351 for seismic applications [10; 11]; and sgn() is the signum function. Equation (1)
becomes fD = c1 u for  = 1, which represents a linear FVD and fD = c0 sgn(u) for  = 0,
which represents a pure friction damper. Thus,  characterizes the non-linearity of FVDs.
The energy dissipated by the damper of Equation (1) during a cycle of harmonic motion
u = u0 sin !t is
  2=!  2=!
ED = fD du = fD u dt = c |u|1+ dt (2)
0 0

Integrating Equation (2) results in [7]:

ED =  c ! u0+1 (3)

where the constant  is

22+ 2 (1 + =2)
 = (4)
(2 + )

and () is the gamma function. For a linear FVD, ( = 1),  = 1 and Equation (3) becomes

ED = c1 !u02 (5)

In the limit case of pure friction dampers, ( = 0),  = 4= and Equation (3) reduces to
ED = 4c0 u0 .
Non-linear and linear FVDs dissipate an equal amount of energy per cycle of harmonic
motion if the two results (Equations (3) and (5)) for ED are the same; this equality leads to

(!u0 )1
c = c1 (6)

Substituting Equation (6) into Equation (1) gives the damper force for energy-equivalent
FVDs:
fD (t) (!u0 )1
= |u|
sgn(u) (7)
fDo ( = 1)  u0
where fDo ( = 1) = c1 u0 is the peak force of the linear FVD. Equation (7) for harmonic
motion u = u0 sin !t is plotted in Figure 1 for three values of  : 1; 0:5, and 0. Although all
three hysteresis loops enclose the same area, their shape changes from an ellipse for a linear
viscous damper ( = 1) to a rectangle for a friction damper ( = 0). The peak value of the
damper force is thus
 1
fDo () 1 !u0
= (8)
fDo ( = 1)  u0

For dampers undergoing harmonic motion, the peak velocity is u0 = !u0 , and Equation (8)
reduces to fDo ()=fDo ( = 1) = 1= , which gives 0.9 for  = 0:5 and =4 for  = 0 (Figure 1).
Thus, for a xed displacement amplitude, peak force fDo () in the non-linear FVD (1) is

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:16231642
1626 W.-H. LIN AND A. K. CHOPRA

f D (t )
f Do ( = 1)

= 0.5 =1
1

=0

_1 0 1 u / uo

_1

Figure 1. Hysteresis loops for uid viscous dampers with three values of  undergoing harmonic motion.

2.5
=1
2 = 0.7
fDo () fDo (=1)

= 0.5
= 0.35
1.5

0.5

0
0 1 2 3 4
.
V / uo

Figure 2. Peak force in non-linear FVDs undergoing harmonic motion as a function


of the pseudo-to-real velocity ratio V= u0 .

less than peak force fDo ( = 1) in the corresponding linear FVD and fDo () becomes smaller
as  is decreased.
Because the system response is most sensitive to damping at ! = !n , energy equivalence
should be established at ! = !n for dampers undergoing non-harmonic motion, where !n is
the natural vibration frequency of the SDF system. Replacing ! by !n in Equation (8) gives
 1
fDo () 1 V
= (9)
fDo ( = 1)  u0
where V = !n u0 is the spectral pseudo-velocity for the SDF system. Plotted in Figure 2,
Equation (9) shows that peak force fDo () in the non-linear damper equals peak force
fDo ( = 1) in the corresponding linear damper at V= u0 = 1=(1) ( = 1:23; 1:24, and 1.25 for

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:16231642
EARTHQUAKE RESPONSE OF ELASTIC SDF SYSTEMS 1627

 = 0:7; 0:5, and 0.35, respectively); fDo ()fDo ( = 1) for V= u0 1=(1) ; and fDo ()
fDo ( = 1) for V= u0 1=(1) .

EQUIVALENT LINEAR VISCOUS DAMPING

We characterize the energy dissipation capacity of energy-equivalent non-linear FVDs by


supplemental damping ratio sd and their non-linearity by . For a linear single-degree-of-
freedom (SDF) system with mass m, stiness k, and a non-linear FVD dened by Equation
(1), the supplemental damping ratio sd due to the FVD is dened based on the concept of
equivalent linear viscous damping [23; Section3:9] as follows:
ED ED
sd = = (10)
4ESo 2ku02
where ESo is the elastic energy stored at the maximum displacement, u0 . Substituting Equa-
tion (3) evaluated at ! = !n into Equation (10) gives sd as a function of the displacement
amplitude, u0 :
 c  c
sd = (!n u0 ) = (!n u0 )1 (11)
2ku0 2m!n
Equation (11) reduces to the amplitude-independent damping ratio sd = c1 =2m!n for a linear
FVD ( = 1) and to sd = 2c0 =ku0 for a friction damper ( = 0). Equation (11) is reminiscent
of Jacobsens [15] dimensionless parameter, which is expressed in terms of the excitation
intensity instead of u0 and used to obtain an approximate solution for steady-forced vibration
of an SDF system with a non-linear FVD.

EQUATION OF MOTION AND SYSTEM PARAMETERS

Equation of motion

The equation governing the motion of the SDF system with mass m, elastic stiness k, linear
viscous damping coecient c, and a non-linear FVD subjected to ground acceleration
u g (t) is
|u| = mu g (t)
mu + cu + ku + c sgn(u) (12)
Given c and  = 1 values, Equation (12) is non-linear, therefore, the response u of the
system depends non-linearly on the excitation intensity. Thus, parameterizing this equation and
studying the eect of supplemental damping on system response become complicated because
of the non-linear term involving two parameters c and , wherein c is not a dimensionless
parameter [8; 9]. Therefore, we replace c by Equation (11) for energy-equivalent FVDs and
divide the resulting equation by m to obtain
2sd !n
u + 2!n u + !n2 u + |u| = u g (t)
(!n u0 )1 sgn(u) (13)


Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:16231642
1628 W.-H. LIN AND A. K. CHOPRA


where !n = k=m and  = c=2m!n are the natural vibration frequency and the damping ratio
of the system, respectively; and sd is the supplemental damping ratio due to the non-linear
FVD. Equation (13) governs the motion of SDF systems with energy-equivalent non-linear
FVDs, which are characterized by the same sd value but dierent  values. In particular,
when  = 1 and  = 0 in Equation (13), we obtain the governing equations for linear-viscous
and friction-supplemental damping, respectively.
Although Equation (13) is non-linear and involves the unknown displacement amplitude
u0 (for  = 1) in the supplemental damping term, it oers the following advantages over
Equation (12): (i) the response u of the system varies linearly with the excitation intensity,
i.e. scaling the u g (t) by doubling the peak ground acceleration u go will double u(t); (ii) eects
of non-linear FVDs on the system response can be investigated in terms of two independent,
dimensionless parameters, sd and ; and (iii) the accuracy of the corresponding linear viscous
system in estimating the response of the system with non-linear FVDs can be evaluated.
Although not readily apparent from Equation (13), the linear dependence of u on u go can be
proved. For this purpose, we rewrite Equation (13) in terms of the normalized displacement
u = u=(ust )0 , where (ust )0 = mu go =k = u go =!n2 is the peak static deformation:

2sd !n2 u01 u g (t)


u + 2!n u + !n2 u +  |u| = !n2
sgn(u) (14)
 u go

Equation (14) indicates that u is independent of the excitation intensity, demonstrating the
linear dependence of u on (ust )0 and hence, on u go .

System parameters

As indicated by Equation (13), the response of energy-equivalent SDF systems with non-linear
FVDs is controlled by four parameters: (i) damper non-linearity parameter , which controls
the shape of the damper force hysteresis loop (Figure 1); (ii) supplemental damping ratio
sd , which represents the energy dissipation capacity of the FVD independent of the  value;
(iii) natural vibration period of the system Tn = 2=!n ; and (iv) damping ratio, , which
represents the inherent energy dissipation capacity of the system.
Consider four dierent values of  : 1 (linear FVD), 0.7, 0.5, and 0.35, and three values
of supplemental damping ratio sd = 0:05; 0:15, and 0.30. These values of  and sd cover the
range of non-linear FVDs commonly used in the seismic protection of structures [10; 11]. The
inherent damping of the SDF system was xed at  = 5% and its natural vibration period Tn
was varied from 0.05 to 5 sec.

DYNAMIC RESPONSE

Steady-state response to harmonic excitation

The two key response quantities selected to study steady-forced vibration of SDF systems
with non-linear FVDs are (i) peak deformation response u0 ; and (ii) total force fTo = mu t0 ,
which is transmitted to the foundation and u t0 is the total acceleration of the structural mass.

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:16231642
EARTHQUAKE RESPONSE OF ELASTIC SDF SYSTEMS 1629

6 =1 6
=1
= 0.7 = 0.35
5 sd=5% 5
sd=5%
4 4

Rd
3 3
15% 15%
2 2

1 30% 1 30%

0 0
0 0.5 1 1.5 2 0 0.5 1 1.5 2
/n /n
(a)

6 =1 6 =1
= 0.7 = 0.35
5 sd=5% 5 =5%
sd

4 4
TR

3 15% 3 15%

2 2

1 30% 1 30%

0 0
0 0.5 1 1.5 2 0 0.5 1 1.5 2
/n /n
(b)

Figure 3. Comparison of (a) deformation response factors Rd (b) transmissibility (TR)


for systems ( = 5%) with non-linear (1) and linear ( = 1) FVDs for three supplemental
damping ratios, sd = 5; 15, and 30%.

Deformation response factor. Figure 3(a) shows the deformation response factor, Rd = u0 =
(ust )0 , which is the ratio of the deformation amplitude u0 to the maximum value of static
deformation (ust )0 , plotted against the frequency ratio !=!n for systems with non-linear FVDs
(1) together with the classical Rd curve of the corresponding linear viscous system ( = 1).
These results indicate that for smaller values of sd , the inuence of damper non-linearity on
Rd is very small over the entire range of excitation frequencies; this inuence increases for
larger values of sd and smaller  values, in part because damper non-linearity shifts the
resonant frequency closer to the natural frequency of the system.

Transmissibility. The transmissibility (TR) is the ratio of (1) the total force fT0 transmitted to
the foundation to the amplitude p0 of the applied harmonic force; or (2) the total acceleration
u t0 of the structural mass to the peak ground acceleration u go . Figure 3(b) plots TR against
frequency ratio !=!n for systems with non-linear FVDs (1) together with the classical
TR curve of the corresponding linear viscous system ( = 1). The inuence of damper non-
linearity on TR is larger than on Rd , especially for larger sd and smaller  values. Consistent
with earlier ndings [9], damper non-linearity increases TR, hence the transmitted force fT0
and structural acceleration u t0 . As mentioned earlier, although the transmitted force is larger,
the maximum damper force is smaller in a non-linear FVD.

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:16231642
1630 W.-H. LIN AND A. K. CHOPRA

Table I. Ground motions considered.


Excitation Earthquake Station PGA (g) PGV (cm=s) PGD (cm)
LMSR01 Loma Prieta Agnews State Hospital 0.172 25.9 12.6
LMSR02 Loma Prieta Capitola 0.443 29.2 5.5
LMSR03 Loma Prieta Gilroy Array #3 0.367 44.6 19.3
LMSR04 Loma Prieta Gilroy Array #4 0.212 37.8 10.1
LMSR05 Loma Prieta Gilroy Array #7 0.226 16.4 2.5
LMSR06 Loma Prieta Hollister City Hall 0.247 38.5 17.7
LMSR07 Loma Prieta Hollister Di Array 0.279 35.6 13.0
LMSR08 Loma Prieta Sunnyvale Colton Ave 0.207 37.3 19.1
LMSR09 Northridge Canoga ParkTopanga Canyon 0.420 60.7 20.3
LMSR10 Northridge LAFaring Rd 0.273 15.8 3.3
LMSR11 Northridge LAFletcher 0.240 26.2 3.6
LMSR12 Northridge Glendale Los Palmas 0.206 7.4 1.7
LMSR13 Northridge Hollywood Storage FF 0.231 18.2 4.8
LMSR14 Northridge LA CrescentaNew York 0.159 11.3 3.0
LMSR15 Northridge NorthridgeSaticoy 0.368 28.9 8.4
LMSR16 San Fernando LA Hollywood Stor Lot 0.174 14.8 6.3
LMSR17 Superstition Hills BRW 0.156 13.9 5.3
LMSR18 Superstition Hills El Centro Imp Co Center 0.358 46.3 17.6
LMSR19 Superstition Hills PLC 0.186 20.6 5.4
LMSR20 Superstition Hills Westmoreland Fire station 0.172 23.5 13.1

EARTHQUAKE RESPONSE

Earthquake excitation and spectral regions

A total of 20 ground motions were considered (Table I), including strong motions recorded
during the Loma Prieta (1989), Northridge (1994), San Fernando (1971), and Superstition
Hills (1987) earthquakes. This large-magnitude-small-distance (LMSR) set of motions comes
from Krawinkler [24].
The mean pseudo-velocity (V ) response spectrum for the ground motion ensemble is shown
in Figure 4. Three spectral regions are identied as follows: the acceleration-sensitive region
(Tn 0:6 sec), the velocity-sensitive region (0:66Tn 63 sec), and the displacement-sensitive
region (Tn 3 sec).

Response quantities

The response quantities of interest are: (i) peak deformation u0 , to which the peak lateral
force fSo = ku0 and internal forces in the structure are related; (ii) peak damper force fDo ;
(iii) peak acceleration u t0 of the mass; (iv) peak value of the force fTo transmitted to the
foundation; and (v) peak value of relative velocity u0 , which is necessary to compute the
exact force in a non-linear FVD and to verify the accuracy of its approximate value as given
by Equation (9).

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:16231642
EARTHQUAKE RESPONSE OF ELASTIC SDF SYSTEMS 1631

Spectral Regions
Acceleration Velocity Displacement
sensitive sensitive sensitive
100
= 5%
10%
20%

Pseudo Velocity (cm/s) 35%


10

0.5
0.02 0.1 1 10 50
Tn (sec)

Figure 4. Mean pseudo-velocity response spectrum for the LMSR-ensemble of 20 ground motions
with spectral regions identied.

Response history

Figure 5 shows the response of three systems with equal supplemental damping (sd = 15%),
but dierent values of  to a selected ground motion. Since the natural vibration period
of all three systems is the same, the time variation of their response is similar. Damper
non-linearity, characterized by the parameter , has very little inuence on the deformation,
velocity, and acceleration of the system, but aects the damper force signicantly, primarily
near the response peaks.

Inuence of damper non-linearity

Although the mean response spectra for deformation, relative velocity, and total acceleration
presented in Figures 6(a)(c), respectively, are aected very little by damper non-linearity, the
inuence increases at longer periods and for smaller values of , implying more non-linearity.
If the ratio of responses r for  = 0:35 and 1 are plotted for three response quantities, as
shown in Figure 7, clearly, then damper non-linearity has essentially no inuence on system
response in the velocity-sensitive spectral region and small inuence in the acceleration- and
displacement-sensitive regions. In particular, dierences up to 14 and 12% in deformation and
relative velocity, respectively, were observed in the acceleration-sensitive region and up to 17%
in total acceleration was observed in the limited portion of the displacement-sensitive region
considered (Figure 7). These observed maximum dierences, however, occurred because of the
small absolute value of the response (Figure 6). Thus, system response is only weakly aected
by damper non-linearity. This observation has the useful implication for design applications
that, for a given sd , the response of systems with non-linear FVDs can be estimated to a
sucient degree of accuracy by analysing the corresponding linear viscous system ( = 1).

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:16231642
1632 W.-H. LIN AND A. K. CHOPRA

0.6 1
=1
0.4 = 0.5
= 0.35 0.5
0.2

u / n( ust )o
st o
u/(u )

0 0

_0.2 .
_0.5
_0.4

_1
0 5 10 15 20 0 5 10 15 20
(a) (b)
0.3 10

0.2
Total Acceleration (g)

5
0.1

f / w (%)
0 0

D
_0.1
_5
_6.69
_0.2
_7.25
_10 _9.44
0 5 10 15 20 8 10 12 14
(c) t (sec) (d) t (sec)

Figure 5. Response history for (a) deformation, (b) velocity, (c) total acceleration, and (d) supplemental
damper force of an SDF system (Tn = 1sec and  = 5%) with sd = 15% subjected to the LMSR02 ground
motion, time range, while the scale is dierent in (d).

20
50 =0 0.6 =0 =1
=0 sd sd
Relative Velocity,u (cm/s)

sd =0.7
Total Acceleration, u (g)

5%
Deformation, D (cm)

15 5% .. 0.5 5% =0.5
t
o

40
. =0.35
o

15% 15%
15% 0.4
30 30%
10 30% 0.3
30% 20
0.2
5
10 0.1

0 0 0
0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5
(a) T (sec) (b) T (sec) (c) T (sec)
n n n

Figure 6. Mean response spectra for (a) deformation, (b) relative velocity, and (c) total acceleration
for SDF systems with  = 5% and supplemental damping sd = 0; 5; 15, and 30% due to
non-linear FVDs with dierent  values.

Inuence of supplemental damping

As expected, supplemental damping reduces structural response, with greater reduction achie-
ved by the addition of more damping (Figure 6); the reduction achieved with a given amount
of damping is dierent in the three spectral regions. As Tn 0, supplemental damping does not

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:16231642
EARTHQUAKE RESPONSE OF ELASTIC SDF SYSTEMS 1633

1.5

r (=0.35) / r (=1) 1 0.88

0.86 0.83

0.5 Deformation
Relative Velocity
Total Acceleration

0
0 1 2 3 4 5
T (sec)
n

Figure 7. Inuence of damper non-linearity on mean peak responses, r: deformation, relative velocity,
and total acceleration. Ratio of r( = 0:35) to r( = 1) is plotted for systems with sd = 30%.

aect response because the structure moves rigidly with the ground. And as Tn , supple-
mental damping again does not aect the response because the structural mass stays still while
the ground underneath moves. The response reduction is signicant over the range of periods
considered. Presented in Figure 8(a) are the ratios of the deformations of two systems, one
with supplemental damping and the other without. As little as 5% supplemental damping re-
duces the deformation response by 25, 21, and 18% averaged over the acceleration-, velocity-,
and displacement-sensitive spectral regions, respectively. The corresponding reductions are 45,
40, and 35% for moderate supplemental damping (sd = 15%) and 60, 55, and 46% for large
supplemental damping (sd = 30%). Consistent with earlier observations, the reduction in re-
sponses is essentially unaected by damper non-linearity in the velocity-sensitive region and
only weakly dependent in the acceleration- and displacement-sensitive regions.
The ratio of responses with and without supplemental damping, presented in Figure 8(b)
for deformation, relative velocity, and total acceleration indicates that supplemental damping
reduces all responses. However, supplemental damping is more eective in reducing struc-
tural deformation and, therefore, internal forces are compared to relative velocity or total
acceleration; the latter two responses are reduced to a similar degree.

Damper force

The response spectrum for damper force shown in Figure 9(a) permits two salient observa-
tions: (i) the damper force is larger for larger dampers, as indicated by their sd values; and
(ii) for a selected sd for supplemental damping, the damper force is smaller for non-linear
FVDs; the more non-linear the damper (i.e. smaller the  value), the smaller is the damper
force (Figure 9(b)). The reduction in damper force due to non-linearity of the damper in-
creases with the system period; for a system with Tn = 2 sec, the force in a non-linear damper
is 83, 73, and 66% of the force in a linear damper for  = 0:7, 0.5, and 0.35, respectively
(Figure 9(b)). The percentage reduction in the damper force as a result of damper non-
linearity is essentially independent of sd , the amount of supplemental damping (Figure 9(c)).
Non-linear FVDs are advantageous because they achieve essentially the same reduction in

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:16231642
1634 W.-H. LIN AND A. K. CHOPRA

1
= 5%
sd
0.8

D( ) / D( =0) 15%
sd 0.6

0.4 =1
sd

30%
= 0.7
0.2 = 0.5
= 0.35
0
0 1 2 3 4 5
Tn (sec)
(a)

0.8
r ( ) / r ( =0)
sd

0.6

0.4
sd

Deformation
Relative Velocity
0.2
Total Acceleration

0
0 1 2 3 4 5
Tn (sec)
(b)

Figure 8. (a) Ratios of mean peak deformations with and without supplemental damping; sd = 5; 15,
and 30%;  = 1; 0:7; 0:5, and 0.35. (b) Ratios of mean peak responsedeformation, relative velocity,
and total accelerationwith and without supplemental damping; sd = 30% and  = 0:35.

response (Figure 6) but with a signicantly reduced damper force (Figure 9(a)). The above
observations are valid for the range of system period considered, except for very short-period
systems (Tn 0:1 sec).

DESIGN ISSUES

Estimating structural deformation and forces

The design deformations and forces for linear systems are usually determined directly from
the earthquake design spectrum without any response history analysis (RHA). Can this also
be done for systems with a non-linear FVD with known properties c and ? As observed
earlier, the structural deformation (and hence internal forces) is more or less unaected by the
damper non-linearity parameter  and is essentially the same as for the corresponding linear

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:16231642
EARTHQUAKE RESPONSE OF ELASTIC SDF SYSTEMS 1635

_
15 1.5 50
=1
sd
= 30%
= 0.7
1
= 30% = 0.5 1 0
sd
= 0.35 0.7
0.5

0.5 = 0.35 50

Percent Reduction (%)


(=1)
10

Do
/ w (%)

0 100

() f
0 1 2 3 4 5
Tn (sec)

(b)
Do

15%

Do
_
f

50

f
1.5
= 0.35
5
1 0

5%
0.5 = 5% 50
sd
sd = 15%
sd = 30%
0
0 1 2 3 4 5 0 100
0 1 2 3 4 5
Tn (sec) T (sec)
n

(a) (c)

Figure 9. (a) Mean damper force spectra for FVDs with sd = 5; 15, and 30% and  = 1, 0.7, 0.5, and
0.35; (bc) ratios of mean damper force in non-linear and linear dampers.

system (Figure 6). Therefore, these design quantities can be determined directly from the
design spectrum for period Tn and total damping  + sd of the corresponding linear system.
However, because sd contains the unknown value of peak deformation, which in turn depends
on sd among other system parameters, it cannot be determined directly from Equation (11).
Thus, an iterative procedure is necessary:
1. Assume an initial value of sd (0)
= 0, determine D(0) = D(Tn ; ) from the design spectrum.
Set u0 =D(0) and compute sd (1)
from Equation (11).
2. Determine D = D(Tn ;  + sd ) from the design spectrum. Set u0 = D(1) and compute sd
(1) (1) (2)

from Equation (11).


3. For the ith iteration, determine D(i) = D(Tn ; +sd
(i)
) from the design spectrum. Set u0 = D(i)
and compute sd from Equation (11).
(i+1)

4. Repeat (3) until two successive values of sd are suciently close.
Example 1
Consider a one-storey frame with lumped weight w = 100 kips, natural vibration period Tn =
1 sec, damping ratio  = 5%, and a non-linear FVD with  = 0:5 and c = 3 kips=(in=sec)0:5 .
Determine peak deformation D and peak lateral force fSo resisted by the structure (exclud-
ing the damper force) for the elastic design spectrum [25] for ground motions with peak
acceleration u go = 0:25g, peak velocity ugo = 0:25 48 = 12 in=sec, and peak displacement
ugo = 0:25 36 = 9 in.

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:16231642
1636 W.-H. LIN AND A. K. CHOPRA

Table II. Iterative solutions of Example 1.


Iteration i sd
(i)
(%) D(i) (in) Vbo =w = A=g ei
0 0 4.396 0.449
1 19.5 2.361 0.241 1
2 26.7 2.034 0.208 0.267
3 28.7 1.954 0.200 0.072
4 29.3 1.932 0.197 0.020
5 29.5 1.926 0.197 0.006

Solution
In the velocity-sensitive region of the spectrum, which includes the Tn = 1 system under con-
sideration, the pseudo-velocity V = v ugo is constant [25], where V = 3:38 0:67 ln() for
84.1% non-exceedance probability with  dened in percent. Hence, the peak deformation
D = V=!n and the peak base shear Vbo = (A=g)w (excluding the damper force), where pseudo-
acceleration A = !n V . The results of the above iterative procedure are summarized in Table II;
the convergence criterion chosen is ei = |sd(i)
sd
(i1)
|=sd
(i)
60:01. Thus, the peak values of de-
formation and base shear for the system are u0 = 1:926 in and Vbo = 0:197w. Supplemental
damping has reduced the deformation from 4.936 to 1.926 in and base shear from 0:449w to
0:197w.

Estimating damper force

Equation (9) relates the peak forces fDo () and fDo ( = 1) in non-linear and linear dampers,
respectively, with both undergoing the same displacement and velocity amplitudes u0 and u0 .
Thus, the peak force fDo () in the non-linear FVD with known  can be expressed as
 1
c1 u0 V
fDo ()approx = (15)
 u0
where V = !n u0 and  is given by Equation (4). The non-linear damper force fDo () can
be computed from Equation (15) if u0 and u0 are known. They can be estimated as the
peak values of deformation and relative velocity of the corresponding linear system. This is
a reasonable approximation in light of the earlier results of Figure 6. These expectations are
conrmed by examining the error in the approximate value for the mean fDo () computed
from Equation (15) and is relative to its exact value presented in Figure 9(a). Figure 10(a)
indicates that Equation (15) is almost exact in the velocity-sensitive region of the spectrum,
overestimates fDo () (by at most 15%) in the acceleration-sensitive region, and underestimates
(by at most 7%) in the displacement-sensitive region. Moreover, the accuracy of Equation
(15) deteriorates slightly with the increase in sd .
The actual velocity u0 of the corresponding linear system required in Equation (15) and
to compute fDo ( = 1) = c1 u0 is not readily available, because the velocity spectrum is not
plotted routinely. If the velocity u0 is replaced by the pseudo-velocityan assumption that is
often made [26]Equation (15) reduces to
fDo ( = 1)
fDo ()approx = (16)


Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:16231642
EARTHQUAKE RESPONSE OF ELASTIC SDF SYSTEMS 1637

100 100
=1
sd = 5% =0.7 = 5%
50 50 sd
=0.5
=0.35
0 0
Relative Errors (%) in Mean Damper Force

_ _
50 50
0 1 2 3 4 5 0 1 2 3 4 5

100 100

sd = 15% = 15%
50 50 sd

0 0

_ _
50 50
0 1 2 3 4 5 0 1 2 3 4 5

100 100

= 30% = 30%
50 sd 50 sd

0 0

_ _
50 50
0 1 2 3 4 5 0 1 2 3 4 5
Tn (sec) T (sec)
n
(a) (b)

Figure 10. Relative errors in the mean damper force in a non-linear FVD estimated from the
response of a system with linear FVD using its (a) relative velocity u0 (Equation (15)) and
(b) pseudo-velocity (Equation (16)).

The resulting estimate of the damper force is not accurate, as can be seen by the large
errors in Figure 10(b), which increase with the systems period, damper non-linearity, and
supplemental damping ratio. The variation of error with period can be explained by examining
the V= u0 ratio, which appears in Equation (15) but is set equal to 1 in Equation (16). A plot
of this ratio determined from the mean velocity and pseudo-velocity spectra of the 20 ground
motions is shown in Figure 11 and indicates that V= u0 is greater than 1 for short-period
systems, approximately equal to 1 around Tn = 0:6 sec, and less than 1 for long-period systems
(Figure 11). These trends explain why Equation (16) underestimates the damper force fDo ()
for short-period systems, where V= u0 1 and overestimates fDo () for long-period systems,
where V= u0 1 (Figure 10(b)). Thus, velocity should not be approximated by the pseudo-
velocity, a result consistent with the earlier investigation [8].

Estimating foundation shear

The total shear force transmitted to the foundation includes the resisting forces in the structure,
fS , and in the supplemental damper, fD :
fT (t) = fS (t) + fD (t) (17)

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:16231642
1638 W.-H. LIN AND A. K. CHOPRA

7
= 10%
6 = 20%
= 35%
5

. 4
o
V/u
3

0
0.02 0.1 1 10 50
Tn (sec)

Figure 11. Ratio of mean pseudo-velocity to mean velocity for the LMSR-ensemble
of 20 ground motions.

Following tradition for structures without supplemental damping, the force associated with
structural damping is not included in Equation (17). The mean response spectrum for the
foundation shear in systems with linear FVD ( = 1) and non-linear FVD ( = 0:35) is pre-
sented in Figures 12(a) and 12(b), respectively.
How to calculate the peak value of the transmitted force, fTo , directly from the earthquake
design spectrum without any RHA? For a linear SDF system undergoing harmonic motion,
fTo is related to fSo and fDo , and the peak values of fS (t) and fD (t), respectively, by [Section
3.5 in Reference 23]

fTo = fSo2 + f2
Do (18)

The response of a linear, lightly damped SDF system to a wide-frequency-band random ex-
citation, e.g. many earthquake ground motions, is a narrow-band random process with the
centre frequency of the band equal to the natural vibration frequency !n of the system [27].
The system response to such earthquake excitations has the appearance of a harmonic of fre-
quency !n but with slowly varying random amplitude and random phase (e.g. Figure 6.4.1
in Reference [23]). However, this type of response function may be less representative of
systems with non-linear FVDs, because such systems are neither linear nor lightly damped.
This becomes apparent by examining Figure 5, especially for systems with smaller values of
. In spite of this observation, Figure 5 suggests that Equation (18) may be approximately
valid. To test this hypothesis, the mean values of fSo and fDo available from the earlier results
of RHA are substituted; fSo = (A=g)w, where A = !n2 D and D is given in Figure 6(a), and
fDo is given in Figure 9(a). Approximate results obtained in this manner are presented in
Figures 12(a) and (b), and the error in these results in Figures 12(c) and (d). As expected,
the errors are larger when the damper is non-linear; however, the error is only 1015% even
when the damper is highly non-linear ( = 0:35).
To determine the transmitted force directly from the earthquake design spectrum, fSo and
fDo are determined for the system with energy-equivalent linear FVD with total damping,

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:16231642
EARTHQUAKE RESPONSE OF ELASTIC SDF SYSTEMS 1639

0.4 0.4
Exact Exact
Eq. (18) Eq. (18)
0.3 Design Spectrum 0.3 Design Spectrum
fTo / w

0.2 0.2

0.1 0.1

0 0
0 1 2 3 4 5 0 1 2 3 4 5
Tn (sec) Tn (sec)
(a) (b)

40 40
Eq. (18) Eq. (18)
Design Spectrum Design Spectrum
(%)

20 20
To
Relative error in f

0 0

_ _
20 20

_ _
40 40
0 1 2 3 4 5 0 1 2 3 4 5
T (sec) Tn (sec)
n
(c) (d)

Figure 12. Mean response spectrum for foundation shear fTo in SDF systems with sd = 30% and
(a)  = 1 and (b)  = 0:35; relative error in fTo for (c)  = 1 and (d)  = 0:35.

 + sd :
fSo = mA(Tn ;  + sd ); fDo ( = 1) = 2msd !n u0 (19)
The force fDo () in the non-linear damper is then computed from Equation (15), with the
values of u0 and u0 determined from the deformation and relative velocity spectra, respectively.
Results obtained in this manner are also included in Figures 12(a) and 12(b) and the associated
errors in Figures 12(c) and 12(d). This approximate value of fTo obtained directly from the
standard pseudo-acceleration spectrum and the relative velocity spectrum is accurate enough
for most design applications.

Selecting a non-linear FVD

How to select the properties c and  of a non-linear FVD to satisfy a design requirement?
Suppose that the structural deformation cannot exceed some design limit for a new structure or
the deformation capacity of an existing structure. As shown earlier, the structural deformation

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:16231642
1640 W.-H. LIN AND A. K. CHOPRA

is essentially unaected by the damper non-linearity parameter  and it is essentially the same
as that for the corresponding linear system (Figure 6(a)). The total damping capacity that must
be provided in the system to limit the deformation of a linear system to a design value can
be determined directly from the design spectrum. Subtracting the inherent damping in the
structure from the total damping required gives sd , the necessary supplemental damping.
Many dierent non-linear FVDs can be chosen to provide the required sd . For a selected
value of , Equation (11) provides an equation for c .
2msd !n
c = (!n D)1 (20)

Figure 6 suggests that the selected damper dened by Equation (20) should satisfy the design
constraint reasonably well. The structural deformation should be (i) very close to the allow-
able value in the velocity-sensitive spectral region, (ii) less than the allowable value in the
acceleration-sensitive spectral region, but (iii) exceed only slightly the allowable value in the
displacement-sensitive spectral region.

Example 2
Consider the one-storey system and the elastic design spectrum dened in Example 1. Suppose
that the allowable lateral deformation of the frame is 1:93 in. Determine the properties c and
 of a non-linear FVD necessary to satisfy this design constant.
Solution
For Tn = 1 sec, the spectrum gives D = (Tn =2)V ugo , where V = 3:38 0:67 ln . From these
equations it can be deduced that the minimum damping required to limit D to 1:93 in is
34.5%. Subtracting the inherent damping  = 5% in the structure, supplemental damping should
provide at least  = 29:5%. Many dierent non-linear FVDs can be chosen to provide the
required sd . If we choose  = 0:5, Equation (4) gives  = 1:113. Substituting  , sd = 0:295,
m = 100=g kips sec2 =in, D = 1:93in, and !n = 2 into Equation (20) gives c = 3kips=(in sec)0:5 .
Similar calculations starting with  = 0:7 gives c = 1:91 kips=(in=sec)0:7 and  = 0:35 leads to
c = 4:21 kips=(in=sec)0:35 . All three of these non-linear dampers satisfy the design criterion.

CONCLUSIONS

This investigation of steady-forced and earthquake responses of elastic single-degree-of-


freedom (SDF) systems with non-linear uid viscous dampers (FVD) has led to the following
conclusions:

1. The dynamic characteristics of a non-linear FVD can be described by its energy dis-
sipation capacity, represented by supplemental damping ratio sd , and its non-linearity
by a parameter , which denes the hysteresis loop shape.
2. The most eective way to determine the response of structures with non-linear FVDs is
in terms of sd and  because (i) these two dimensionless parameters are independent,
and (ii) the structural response varies linearly with the excitation intensity.
3. For systems with same supplemental damping sd , the inuence of damper non-linearity
on harmonic response is very small over the entire range of excitation frequencies for

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:16231642
EARTHQUAKE RESPONSE OF ELASTIC SDF SYSTEMS 1641

smaller values of sd ; this inuence increases for larger values of sd and smaller 
values in part because damper non-linearity shifts the resonant frequency. Damper non-
linearity has more inuence on transmissibility than on structural deformation.
4. Damper non-linearity essentially has no inuence on the peak responsesdeformation
u0 , relative velocity u0 , and total acceleration u t0 of systems in the velocity-sensitive
spectral region. Dierences up to 14% in u0 and u0 were observed in the acceleration-
sensitive region and up to 17% in u t0 in the limited portion of the displacement-sensitive
region considered. These maximum dierences, however, occurred because of the small
absolute value of the response.
5. Supplemental damping reduces structural responses, with greater reduction achieved by
increasing the damping. The reduction achieved for a given sd is slightly dierent in
the three spectral regions, the largest being in the acceleration-sensitive region. The
deformation is reduced by up to 25% when sd = 5%; and up to 60% when sd = 30%.
6. Supplemental damping is more eective in reducing structural deformation, and hence
internal forces, compared to relative velocity or total acceleration; the latter two re-
sponses are reduced to a similar degree. These reductions are essentially unaected by
damper non-linearity in the velocity-sensitive region and are only weakly dependent in
the acceleration and displacement-sensitive regions.
7. Non-linear FVDs are advantageous because they achieve essentially the same reduction
in system responses but with a signicantly reduced damper force, e.g. for a system
with Tn = 2 sec, the force in a non-linear damper is 83, 73, and 66% of the force in
the corresponding linear damper for  = 0:7; 0:5, and 0.35, respectively.
8. The design values of structural deformation and forces for a system (period Tn and
inherent damping ) with non-linear FVDs can be estimated directly from the design
spectrum for the period Tn and total damping  + sd . However, sd must be determined
iteratively as it contains the unknown value of peak deformation.
9. The peak value of earthquake-induced force in a non-linear FVD can be estimated with
reasonable accuracy from the peak damper force in the corresponding linear system, its
peak deformation, and relative velocity. The relative velocity should not be replaced
by pseudo-velocity as this approximation introduces large error in the damper force.
10. The peak values of the structural force and damper force determined from the earth-
quake design spectrum can be combined according to Equation (18) to estimate the
peak value of the shear force transmitted to the foundation.
11. For a given design spectrum, a procedure was presented to determine the non-linear
damper properties c and  necessary to ensure that the structural deformation does not
exceed some design limit for a new structure or the deformation capacity of an existing
structure.

ACKNOWLEDGEMENTS

This research investigation was supported by the National Science Foundation under the grant CMS-
9626586. The authors have beneted greatly from discussions with Professor Nicos Makris, and are
grateful for his advice. We also wish to thank Professor Helmut Krawinkler for providing the LMSR
set of ground motions.

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:16231642
1642 W.-H. LIN AND A. K. CHOPRA

REFERENCES

1. Pekcan G, Mander JB, Chen SS. The seismic response of the 1 : 3 scale model R.C. structure with elastomeric
spring dampers. Earthquake Spectra 1995; 11(2):249 267.
2. Soong TT, Dargush GF. Passive Energy Dissipation Systems in Structural Engineering. Wiley: New York,
1997.
3. Makris N. Theoretical and experimental investigation of viscous dampers in applications of seismic and vibration
isolation. PhD dissertation, State University of New York, Bualo, Amherst, NY, 1992.
4. Constantinou MC, Symans MD. Experimental and analytical investigation of seismic response of structures with
supplemental uid viscous dampers. Report No. NCEER-92-0032, National Center for Earthquake Engineering
Research, Bualo, NY, 1992.
5. Tsopelas P, et al. Experimental and analytical study of systems consisting of sliding bearing, rubber restoring
force devices, and uid dampers. Report No. NCEER-94-0002, National Center for Earthquake Engineering
Research, Bualo, NY, 1994.
6. Reinhorn AM, Li C, Constantinou MC. Experimental and analytical investigation of seismic retrot of structures
with supplemental damping: Part 1Fluid viscous damping devices. Report No. NCEER-95-0001, National
Center for Earthquake Engineering Research, Bualo, NY, 1995.
7. Symans MD, Constantinou MC. Passive uid viscous damping systems for seismic energy dissipation. ISET
Journal of Earthquake Technology, Paper No. 382, 1998; 35(4):185 206.
8. Pekcan G, Mander JB, Chen SS. Fundamental considerations for the design of non-linear viscous dampers.
Earthquake Engineering and Structural Dynamics 1999; 28:1405 1425.
9. Terenzi G, Dynamics of SDOF systems with non-linear viscous damping. Journal of Engineering Mechanics
1999; 125(8):956 963.
10. Asher JW, Young RP, Ewing RD. Seismic isolation design of the San Bernardino county medical center
replacement project. Journal of Structure and Design of Tall Buildings 1996; 5:265 279.
11. Rodriguez S, Seim C, Ingham T. Earthquake protective systems for the seismic upgrade of the Golden Gate bri-
dge. Proceedings of the Third U.S.Japan Workshop on Protective Systems for Bridges, Berkeley, CA, 1994.
12. Terenzi G. Eetti dissipative nellisolamento sismico [Damping eects in the seismic isolation]. PhD
dissertation, University of Florence, Florence, Italy (in Italian), 1994.
13. Seleemah AA, Constantinou MC. Investigation of seismic response of buildings with linear and non-linear uid
viscous dampers. NCEER Report No. 97-0004, National Center for Earthquake Engineering Research, Bualo,
NY, 1997.
14. Chiarugi A, Quenette R, Terenzi G. Sperimentazione ed analisi del comportamento dinamico di dispositivi
siliconici [Experimental tests and analysis of the dynamic behavior of uid viscous devices]. Ingegneria Sismica,
Patron, Quarto Inferiore, Italy, 1995; 12(3):122 (in Italian).
15. Jacobsen LS. Steady forced vibration as inuenced by damping. Transactions of the American Society of
Mechanical Engineers 1930; 52:169 181.
16. Den Hartog JP. Forced vibrations with combined Coulomb and viscous friction. Transactions of the American
Society of Mechanical Engineers 1931; 53:107115.
17. Fabunmi JA. Extended damping models for vibration data analysis. Journal of Sound and Vibration 1985;
101(2):181192.
18. Rakheja S, Sankar S. Local equivalent constant representation of non-linear damping mechanisms. Engineering
& Computers 1986; 3:1117.
19. Caughey TK. Equivalent linearization techniques. Journal of the Acoustical Society of America 1963; 35:
1706 1711.
20. Booton RC. Non-linear control systems with random inputs. Transactions of the IRE Circuit Theory 1954;
CT-1:9 18.
21. Roberts JB. Response of non-linear mechanical systems to random excitations. Part 2: Equivalent linearization
and other methods. Shock Vibration Digest 1981; 13:15 29.
22. Roberts JB. Stationary response of oscillators with non-linear damping to random excitations. Journal of the
Sound and Vibration 1976; 50:145 156.
23. Chopra AK. Dynamics of Structures: Theory and Applications to Earthquake Engineering (2nd edn). Prentice-
Hall: Upper Saddle River, NJ, 2001.
24. Krawinkler H, et al. Seismic demands and their dependence in ground motion. Eleventh PEER Synthesis
Reports, Pacic Earthquake Engineering Research Center, University of California at Berkeley, Berkeley, CA,
2001.
25. Newmark NM, Hall WJ. Earthquake Spectra and Design. Earthquake Engineering Research Institute, Berkeley,
CA, 1982; 2937.
26. Soong TT, Constantinou MC. Passive and Active Structural Vibration Control in Civil Engineering. Springer:
New York, 1994.
27. Crandall SF, Mark WD. Random Vibration in Mechanical Systems. Academic Press: New York, 1963.

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:16231642

You might also like