You are on page 1of 16

An Evaluation of the Force Reduction Factor in the Force-Based Seismic Design

Gakuho Watanabe and Kazuhiko Kawashima
Tokyo Institute of Technology, O-Okayama, Meguro, Tokyo, Japan, 152-8552




ABSTRACT
This paper presents an analysis of the force
reduction factors used in the force-based seismic
design of structures. The force reduction factors
are evaluated based on 70 free-field ground
motions. Scattering of the force reduction factors
depending on ground motions and the effect of
damping rations assumed in linear and nonlinear
responses are clarified. A new formulation of the
force reduction factors is presented.

KEY WORDS
Seismic design, Forced-based design, Response
modification factor, Force modification factor,
Seismic response


1. INTRODUCTION

In the force based seismic design, it is usual to
estimate the demand from a linear response of a
structure by dividing it by the force reduction
factor. The force reduction factor or response
modification factor, which is often called q-factor
or R-factor, has an important role in the estimation
of design force of a structure. An early study by
Newmark and Hall (1973) revealed the fact that
the equal displacement assumption and the equal
energy assumption provide a good estimation of
the force reduction factors at long and short
periods, respectively. This affected an important
effect to seismic design criteria worldwide.
Various researches such as Nassar and Krawinkler
(1991) and Miranda and Bertero (1994) have been
conducted on the force reduction factors. In
particular, Miranda and Bertero provided a
detailed review on the force reduction factors.
However, in spite of the importance in seismic
design, less attention has been paid to large
scattering of the force reduction factors depending
on ground motions. Since the scattering is so large,
only the mean values of the force reduction factor
is not sufficient to evaluate a force reduction
factor for design. Assumption of damping ratio for
evaluating the linear and nonlinear responses is
another important point. Although it has been
general to assume the same damping ratio for the
linear and nonlinear responses, it depends on how
the force reduction factors are used.
This paper present an analysis on the force
reduction factors based on 70 free-field ground
motions. The scattering of the force reduction
factors depending on ground motions and the
effect of assumption of damping ratios in linear
and nonlinear responses are clarified.


2. DEFINITION OF FORCE REDUCTION
FACTOR

If one idealizes a structure in terms of a
single-degree-of-freedom (SDOF) oscillator with
an elastic perfect plastic bilinear hysteretic
behavior as shown in Fig. 1, the force reduction
factor
µ
R may be defined as
) , , (
) , (
) , , , (
NL T
NL
Y
EL
EL
R
NL EL T
T F
T F
T R
ξ µ
ξ
ξ ξ µ
µ
(1)
in which T : natural period,
EL
R
F and
NL
Y
F :
maximum restoring force in an oscillator with a
linear and a bilinear hysteresis, respectively,
T
µ :
target ductility factor, and
EL
ξ and
NL
ξ :
damping ratio assumed in the evaluation of linear
and bilinear responses, respectively. The natural
period T may be evaluated based on the cracked
stiffness of columns. Representing
y
u the yield
displacement where the stiffness changes from the
cracked stiffness to the post-yield stiffness, a
target ductility factor
T
µ may be defined based
on the yielding displacement
y
u as
y
T
T
u
u
max
µ (2)
in which
T
u
max
is a target maximum
displacement of an oscillator. The post-yield
stiffness is assumed to be 0 in the present study.
Since the damping controls structural response,
it has to be clarified carefully. A structure under a
strong excitation generally exhibits strong
hysteretic behavior, and this results in an energy
dissipation in a structure. For example, the
flexural inelastic deformation of columns
contributes to energy dissipation in a bridge.
Hence, the evaluation of damping ratio depends
on the idealization of such an energy dissipation.
If one idealizes the energy dissipation in nonlinear
structural components by incorporating nonlinear
elements that represent the hysteretic behavior, the
energy dissipation in the nonlinear structural
components is automatically included in the
analysis. On the other hand, if one idealizes the
nonlinear structural components by elastic linear
elements, the energy dissipation in the nonlinear
structural components has to be included in the
analysis by other means. The equivalent viscous
damping ratio
h
ξ is generally used for such a
purpose as
W
W
h


π
ξ
4
1
(3)
in which W ∆ and W represent an energy
dissipation in a hysteretic excursion and the
elastic energy, respectively. For example, in an
oscillator with an elastic perfect-plastic bilinear
hysteresis, the equivalent damping ratio
h
ξ is
µ
µ
π
ξ
1 2 −

h
(4)
Fig. 2 shows the equivalent damping ratio by
Eq. (4). It is generally very large such as 0.4 at the
target ductility factor of 3-5.
In addition to such hysteretic energy
dissipation, there must be some other sources of
energy dissipation (for example, Kawashima,
Unjoh, Tsunomoto 1993). The radiation of energy
from a foundation to surround ground contributes
to energy dissipation. Structural damping such as
friction at connections may be important in many
structures (for example, Kawashima and Unjoh
1989). Viscous damping due to friction with air is
generally predominant in a structure with a long
natural period. It is general to idealize those
sources of energy dissipation in terms of the
equivalent viscous damping.
If one considers a structure in which the
flexural hysteretic energy dissipation is
predominant with other sources of energy
dissipation being a secondary importance, the
total damping ratio
eq
ξ of a SDOF oscillator
δ δ
NL
δ
EL
δ
F
F
EL
F
NL
O

Fig. 1 Definition of Force reduction Factor
may be provided as
oth h eq
ξ ξ ξ + (5)
in which
h
ξ is the damping ratio that accounts
the hysteretic energy dissipation by Eq. (3), and
oth
ξ is the damping ratio that accounts the energy
dissipation other than the hysteretic energy
dissipation.
In the evaluation of the force reduction factor
µ
R based on Eq. (1), how damping ratios are
assumed in the evaluation of the linear and the
nonlinear responses is important. If one assumes
the damping ratios as
eq EL
ξ ξ and
oth NL
ξ ξ (6)
the energy dissipation is essentially the same
between the linear and the nonlinear responses.
Hence, the force reduction factor by Eq. (1)
represents the difference of restoring force
between the linear and nonlinear responses. Thus,
Eq. (1) reflects the effect of nonlinear response of
an oscillator.
On the other hand, if one assumes the damping
ratios as
eq NL EL
ξ ξ ξ (7)
or,
oth NL EL
ξ ξ ξ (8)
the force reduction factor by Eq. (1) includes the
effect of different energy dissipation between the
linear and nonlinear responses, in addition to the
effect of nonlinear response. By assuming Eq. (7)
in the evaluation of nonlinear response, the
hysteretic energy dissipation in the nonlinear
structural components is counted by the
equivalent viscous damping in addition to the
inelastic excursion in the nonlinear elements. As a
consequence, the hysteretic energy dissipation in
the nonlinear structural components is counted
twice in the evaluation of nonlinear response. On
the other hand, if one assumes Eq. (8), the
hysteretic energy dissipation is not taken into
account in the evaluation of linear response.
It should be noted here that which is
appropriate among Eqs. (6), (7) and (8) depends
on how the force reduction factor is used. Based
on the original definition inherent to the force
reduction factor, it seems that Eq. (6) is the most
appropriate. Eq. (8) generally provides
conservative estimation for the force reduction
factors. If 0 ≈
oth
ξ , the difference of the force
reduction factors among Eqs. (6), (7) and (8) is
limited. It should be noted that
h
ξ ,
oth
ξ and
eq
ξ
depend on the type of a structure, mode shape,
hysteresis and the target ductility factor.
Although the equivalent damping ratio
h
ξ is
very high as shown in Fig. 2, it is not general to
assume such a high damping ratio in seismic
design of a bridge structure. It is because a bridge
structure is generally more complex than a SDOF
oscillator, and this makes the relative contribution
of the hysteretic energy dissipation of columns
less significant. Since it is general practice in a
standard bridge structure to assume about 0.05 for
the damping ratio including hysteretic energy
dissipation of columns,
eq
ξ is assumed to be 0.05
in the present study based on Eq. (6). Hence, it is
assumed here that
h
ξ and
oth
ξ is 0.03 and 0.02,
respectively.
Using those damping ratios, the force
reduction factors are evaluated in this study based
on Eq. (6). However, an analysis assuming Eqs.
(7) and (8) is also conducted for comparison with
the previous studies.


0
0.1
0.2
0.3
0.4
0.5
0.6
0 2 4 6 8
E
q
u
i
v
a
l
e
n
t

D
a
m
p
i
n
g

R
a
t
i
o

ξ
Ductility Factor
h

Fig.2 Equivalent Damping Ratio
h
ξ by Eq. (4)
3. REVIEW OF THE PAST
INVESTIGATIONS

An early study for the force reduction factor was
conducted by Newmark and Hall (Newmark and
Hall 1973). They used 10 ground motions
recorded in the 1940 Imperial Valley Earthquake.
They assumed
EL
ξ =
NL
ξ =0.05, and proposed a
force reduction factor as
( )
¦
¦
¦
¦
¦
¦
¦
¦
¦
¦
¦
¦
¦
≤ ≤ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅
≤ ≤

⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅

≤ ≤ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ −
≤ ≤ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅
(
(
,
\
,
,
(
j

≤ ≤ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅


) (
) ( /
) 4 / ( 1 2
) 4 / 10 / (

4
1 2
) 10 / 0 ( 1
2 1
1 1 1
1 1
1 1
1 2 / 1 log 513 . 2
1
1
T T T
T T T T T
T T T
T T T
T
T
T T
R
µ
µ
µ
µ
µ
µ

(9)
where,
A
V
T
ea
ev
φ
φ
π 2
1


1 2
1 1


µ
µ
T T (10)
V
D
T
ev
ed
φ
φ
π 2
1

in which, A, V and D represent peak ground
acceleration, velocity and displacement,
respectively, and
ea
φ ,
ev
φ and
ed
φ represent
the amplification for acceleration, velocity and
displacement, respectively.
Nassar and Krawinkler proposed a force
reduction factor, assuming
EL
ξ =
NL
ξ =0.05,
based on an analysis for 15 ground motions as
(Nassar and Krawinkler 1991)
¦ ¦
c
c R
/ 1
1 ) 1 ( + − µ
µ
(11)
where
T
b
T
T
T c +
+

α
α
α
1
) , ( (12)
in which α represents a ratio of the post-yield
stiffness to the initial elastic stiffness, and a and
b are coefficients depending on α . Nassar and
Krawinkler precisely analyzed the effect of
stiffness deterioration, and provided the
coefficients a and b depending on α .
Miranda and Bertero proposed a force
reduction factor, assuming
EL
ξ =
NL
ξ =0.05,
based on an analysis for 124 ground motions as
(Miranda and Bertero 1994)
1 1
) , (
1
> +
Φ

g
T T
R
µ
µ
(13)
where,
¦
¦
¦
¦
¦
¦
¦
¦
¦
¦
¦
¦
¦
¦
¦
¦
¦
⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅
¦
¦
¦
¦
¦
¦
¦
¦
¦
¦
(
(
,
\
,
,
(
j
− − − +
⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅
¦
¦
¦
¦
¦
¦
¦
¦
¦
¦
(
,
\
,
(
j
− − −

+
⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅
¦
¦
¦
¦
¦
¦
¦
¦
¦
¦
(
,
\
,
(
j
− − −

+
Φ
(soft)

4
1
ln 3 exp
4
3
3
1
(alluvium)
5
1
ln 2 exp
5
2
12
1
1
(rock)

5
3
ln
2
3
exp
2
1
10
1
1
2
2
2
g
g g
T
T
T
T
T
T
T
T T T
T
T T T
µ
µ
(14)
in which
g
T represents a most predominant
period.
It has been known that the equal energy
assumption provides a good estimation for the
force reduction factor at short periods while the
equal displacement assumption at long periods.
The force reduction factor provided by the equal
energy and the equal displacement assumptions
are given as
1 2 − µ
µ
R (equal energy) (15)
µ
µ
R (equal displacement) (16)
Application of Eqs. (15) and (16) and a
comparison of the present study to the previous
models will be described later.
4. FORCE REDUCTION FACTOR FOR
BILINEAR OSCILLATORS

Force reduction factors were evaluated for target
ductility factor
T
µ of 2, 4, 6 and 8 assuming an
elastic perfect-plastic bilinear hysteresis. Damping
ratio in the linear and nonlinear analyses is
assumed as
EL
ξ =0.05 and
NL
ξ =0.02 based on
Eq. (6). Seventy free field ground accelerations by
64 shallow earthquakes with depth less that 60 km
were used for analysis. They are classified into
three soil conditions depending on the
fundamental natural period of subsurface ground
g
T ; stiff ( 2 . 0 <
g
T s), moderate ( 6 . 0 2 . 0 < ≤
g
T s)
and soft ( ≥
g
T 0.6 s) (Japan Road Association
2002). Number of records in the stiff, moderate
and soft categories is 16, 39 and 15, respectively.
Distribution of peak ground accelerations on the
earthquake magnitudes and epicentral distances is
shown in Fig. 3. The peak accelerations are in the
range of 0.1-8m/s
2
, and the epicentral distances
are in the range of 10-500 km.
Fig. 4 shows the force reduction factors for the
70 ground motions. Only the results for
T
µ =4
and 6 are presented here since the results for other
target ductility factors show the similar
characteristics. It is seen in Fig. 4 that scattering
of the force reduction factors depending on
ground motions is significant. For example at
natural period of 1 second, the force reduction
factors varies from 1.9 to 10.3 depending on
ground motions for
T
µ =4 at the moderate soil
sites. It is apparent that such a large scattering of
the force reduction factors result in a large change
of sizing of a structure in seismic design.
Obviously smaller force reduction factors should
be assumed in design to provide conservative
design. It is observed in Fig. 4 that the
dependence of force reduction factors on the soil
condition is less significant. This will be discussed
later.
Since the scattering of the force reduction
factors depending on ground motions is so large
that the means +/- one standard deviations of the
force reduction factors were obtained for each
target ductility factor, natural period and soil
condition. Fig. 5 shows the mean values and the
mean values +/- one standard deviations of the
force reduction factors presented in Fig. 4. The
force reduction factors predicted by Eqs. (15) and
(16) based on the equal displacement and the
equal energy assumptions are also presented here
for comparison. The mean values of force
reduction factors increase as the natural periods
increase, and then they approach to
T
µ at long
period. It has been pointed out in the previous
researches that the Eq. (15) provides a good
estimation to the force reduction factor. However,
it is noted that Eq. (15) provides a good estimation
to the mean values, but it considerably
underestimates the force reduction factors
corresponding to the mean values minus one
standard deviations. On the other hand, Eq. (16)
0.1
1
10
10 100 1000 P
e
a
k

A
c
c
e
l
e
r
a
t
i
o
n

(
m
/
s
e
c
2
)
Epicentral Distance (km)
Stiff (Type-I)
Moderate (Type-II)
Soft (Type-IIII)

(a) Soil Condition
0.1
1
10
10 100 1000 P
e
a
k

A
c
c
e
l
e
r
a
t
i
o
n

(
m
/
s
e
c
2
)
Epicentral Distance (km)
6.5 - 7.0
7.0 - 7.5
7.5 - 8.0

(b) Earthquake Magnitude
Fig. 3 Classification of Ground Accelerations in
Terms of Soil Conditions and Earthquake
Magnitudes
provides better estimation to the mean values
minus one standard deviations. Taking account of
the force reduction factors having considerable
scattering depending on ground motions, it seems
reasonable to consider a certain redundancy in the
estimation of the force reduction factor in design.

0
4
8
12
16
0 1 2 3 4
Natural Period (s)
R
µ
f
a
c
t
o
r

0
4
8
12
16
20
24
0 1 2 3 4
Natural Period (s)
R
µ
f
a
c
t
o
r
(a) Stiff (Type-I) (a) Stiff (Type-I)

0
4
8
12
16
0 1 2 3 4
Natural Period (s)
R
µ
f
a
c
t
o
r

0
4
8
12
16
20
24
0 1 2 3 4
Natural Period (s)
R
µ
f
a
c
t
o
r
(b) Moderate (Type-II) (b) Moderate (Type-II)

0
4
8
12
16
0 1 2 3 4
Natural Period (s)
R
µ
f
a
c
t
o
r

0
4
8
12
16
20
24
0 1 2 3 4
Natural Period (s)
R
µ
f
a
c
t
o
r
(c) Soft (Type-III) (c) Soft (Type-III)
(1)
T
µ =4 (2)
T
µ =6
Fig. 4 Force Reduction Factors
Based on such a consideration, it is more
conservative to assume Eq. (16) instead of Eq.
(15) for a design purpose.
Fig. 6 shows the dependence of the standard
deviations of force reduction factors ) (
µ
σ R on
the natural periods T and the soil condition.
Similar to the mean values, the standard
deviations ) (
µ
σ R increase as the natural periods
increase, and decrease after taking peak values at
natural period of 1-2 second. Fig. 7 shows the
dependence of the standard deviations ) (
µ
σ R on
the target ductility factors
T
µ . The standard
deviations ) (
µ
σ R increase as the target ductility
factors increase. The relation may be
approximated by a least square fit as
¦
¦
¦
¦
¦
⋅ + −
⋅ + −
⋅ + −

(soft) 409 . 0 354 . 0
(moderate) 0.378 0.292
(stiff) 379 . 0 328 . 0
) (
T
T
T
R
µ
µ
µ
σ
µ
(17)
As the soil condition dependence of ) (
µ
σ R
is less significant as shown in Fig. 7, Eq. (17) may
be approximated as
T
R µ σ
µ
⋅ + − ≈ 4 . 0 3 . 0 ) ( (18)

Average + 1σ
Average - 1σ
Average
R
µ
= 2µ-1
R
µ
= µ


0
4
8
12
0 1 2 3 4
Natural Period (s)
R
µ
f
a
c
t
o
r

0
4
8
12
0 1 2 3 4
Natural Period (s)
R
µ
f
a
c
t
o
r

(a) Stiff (Type-I) (a) Stiff (Type-I)

0
4
8
12
0 1 2 3 4
Natural Period (s)
R
µ
f
a
c
t
o
r

0
4
8
12
0 1 2 3 4
Natural Period (s)
R
µ
f
a
c
t
o
r

(b) Moderate (Type-II) (b) Moderate (Type-II)

0
4
8
12
0 1 2 3 4
Natural Period (s)
R
µ
f
a
c
t
o
r

0
4
8
12
0 1 2 3 4
Natural Period (s)
R
µ
f
a
c
t
o
r

(c) Soft (Type-III) (c) Soft (Type-III)
(1)
T
µ =4 (2)
T
µ =6
Fig. 5 Mean and Mean +/- One Standard Deviation of the Force Reduction Factors
for 70 Ground Motions
0 1 2 3 4 5
R
µ


f
a
c
t
o
r
Natural Period (s)
µ
1 a
(µ−1)
1/b
P
c
Q
0

Fig. 8 Idealization of Force Reduction Factors
5. FORMULATION OF FORCE
REDUCTION FACTORS

To idealize the mean values of the force reduction
factors in Fig. 5, they are represented as
1 ) ( ) 1 ( + Ψ ⋅ − T R µ
µ
(19)

where,
1 ) (
) (
+

⋅ Ψ
− ⋅ a T b
e
a T
c T (20)
in which a , b and c are parameters to be
determined.
Since µ
µ
R at a T in Eq. (20), the
parameter a represents the period where
µ
R is
equal to µ (Point P) as shown in Fig. 8. Because
the gradient of
µ
R is
) (
) ( 1
) 1 (
a T b
e
a T b
c
dT
dR

− −
⋅ − µ
µ
(21)
it is ) 1 ( − ⋅ µ c at Point P. Consequently, the
parameter c represents the gradient at Point c
divided by 1 − µ . Representing Q as the point
where
µ
R takes the peak value, b / 1 represents
the period between Points P and Q.
Based on the definition, the following
condition has to be satisfied in
µ
R
0
lim
0


µ
R
T
(22)
Hence, the coefficient c can be eliminated as
ab
ae c / 1 (23)
Substitution of Eq. (23) makes Eq. (20) as
1 ) ( +

Ψ
bT
ae
a T
T (24)
It is noted that Eq. (19) automatically satisfies
the following condition
µ
µ

∞ →
R
T
lim (25)
0
2
4
6
8
0 1 2 3 4
Stiff (Type-I)
Moderate (Type-II)
Soft (Type-III)
Natural Period (s)
σ

(

R
µ

)

(a)
T
µ =4
0
2
4
6
8
0 1 2 3 4
Natural Period (s)
σ

(

R
µ

)

(b)
T
µ =6
Fig. 6 Natural Period Dependence of Standard
Deviations of the Force Reduction
Factors
0
2
4
6
8
10
0 2 4 6 8 10
Stiff (Type-I)
Moderate (Type-II)
Soft (Type-III)
Target Ductility
µ
T
σ

(
R
µ
)

Fig. 7 Target Ductility Factor Dependence of
the Standard Deviations of Force
reduction Factors
It is a feature of the above formulation that the
equation is simpler and the physical meaning of
the parameters a and b is clearer than the
previous studies.
The mean values of force reduction factors in
Fig. 5 were fitted by Eq. (19) using a nonlinear
least square method (Press et al 1996). Table 1
represents the a and b as well as the
regression coefficients. Although the regression
coefficient is not high enough for some
combinations such as
T
µ =2 and stiff sites, it may
be accepted in other conditions. As shown later,
the fitting is not necessarily poor for a
combination of
T
µ =2 and stiff sites.
Fig. 9 shows parameters a , b / 1 and b a / 1 + .
Parameter a is in the range of 1.0-1.4 second at
stiff and moderate sites, and 1.5-2.4 second at soft
sites. They are less sensitive to the target ductility
factor
T
µ between 2 and 8. As described before,
a represents the period where µ
µ
R , which
implies that the equal displacement assumption by
Eq. (16) provides the best estimation at period a .
Consequently, the accuracy of equal displacement
assumption is high at 1.0-1.4 second at stiff and
moderate sites, and 1.5-2.4 second at soft site.
As shown in Fig. 8, b a / 1 + represents the
natural period where
µ
R takes the peak value. It
is 1.5-2 second at stiff and moderate sites, and
0
1
2
3
4
0 2 4 6 8 10
a

(
s
)
Target Ductility µ
Τ
Stiff (Type-I)
Moderate (Type-II)
Soft (Type-III)

(a) Parameter a
0
1
2
0 2 4 6 8 10
1
/
b

(
s
)
Target Ductility µ
Τ
(b) Parameter b 1
0
1
2
3
4
0 2 4 6 8 10
a
+
1
/
b

(
s
)
Target Ductility µ
Τ

(c) Parameter b a 1 +
Fig. 9 Parameters a and a+1/b in Eqs. (20)
Table 1 Parameters a and b and Regression
Coefficients R (
NL
ξ =0.02 and
EL
ξ =
0.05)
Soil Conditions
T
η a, b
and R Type-I Type-II Type-III
a 1.29 1.12 2.35
b 2.77 2.18 1.69 2
R 0.379 0.701 0.851
a 1.24 0.989 1.52
b 2.39 1.62 1.05 4
R 0.673 0.842 0.886
a 1.34 1.03 1.85
b 2.15 1.24 0.821 6
R 0.717 0.869 0.878
a 1.36 1.20 1.74
b 1.67 1.11 0.611 8
R 0.776 0.899 0.895
0
0.2
0.4
0.6
0.8
0 2 4 6 8 10
P
e
r
i
o
d

(
s
)
Target Ductility µ
Τ
Stiff (Type-I)
Moderate (Type-II)
Soft (Type-III)

Fig. 10 Natural Periods where Force Reduction
Factors Take Values Predicted by the
Equal Energy Assumption (Eq. (15))
2.5-3.5 second at soft site. It slightly increases as
target ductility
T
µ increases.
The natural periods where
µ
R take the values
predicted by Eq. (15) based on the equal energy
assumption are obtained as shown in Fig. 10.
They are in the range of 0.2-0.36 second, 0.26-0.4
second and 0.4-0.6 second at stiff, moderate and
soft sites, respectively. They are much shorter than
the natural periods where the equal displacement
assumption provides the best approximation.
Fig. 11 compares the mean force reduction
factors presented in Fig. 5 to the values predicted
by Eq. (19). Although some discrepancies are
observed at larger target ductility factors, Eq. (19)
µ
T
= 2
µ
T
= 4
µ
T
= 6
µ
T
= 8
Empirical Model Mean

0
4
8
12
16
0 1 2 3 4
Natural Period (s)
R
µ
f
a
c
t
o
r

(a) Stiff (Type-I)
0
4
8
12
16
0 1 2 3 4
Natural Period (s)
R
µ
f
a
c
t
o
r

(b) Moderate (Type-II)
0
4
8
12
16
0 1 2 3 4
Natural Period (s)
R
µ
f
a
c
t
o
r

(c) Soft (Type-III)
Fig. 11 Application of Eq. (19) to the Mean
Force Reduction Factors Presented in
Fig. 5
0
4
8
12
0 1 2 3 4
Natural Period (s)
R
µ

f
a
c
t
o
r
Stiff (Type-I)
Moderate (Type-II)
Soft (Type-III)

(a)
T
µ =2
0
4
8
12
0 1 2 3 4
Natural Period (s)
R
µ

f
a
c
t
o
r

(b)
T
µ =4
0
4
8
12
0 1 2 3 4
Natural Period (s)
R
µ

f
a
c
t
o
r

(c)
T
µ =6
0
4
8
12
0 1 2 3 4
Natural Period (s)
R
µ

f
a
c
t
o
r

(d)
T
µ =8
Fig. 12 Effect of Soil Condition on the Force
Reduction Factors Predicted by Eq.
(19)
provides a good estimation for the mean force
reduction factors.
Fig. 12 shows the effect of soil condition on the
mean force reduction factors estimated by Eq.
(19). The effect of soil condition is less significant
on the force reduction factors, in particular at
small target ductility factors.
As shown in Fig. 4, scattering of the force
reduction factors around the mean values is
extensive. Hence, the force reduction factors
corresponding to the mean values m substituted
by a standard deviation ) (
µ
σ R are evaluated as
shown in Fig. 13. The mean and the standard
deviation of force reduction factors were
evaluated by by Eq. (19) and Eq. (18),
respectively, in this estimation. They are of course
close to the force reduction factors of the mean
minus one standard deviation directly computed
from the 70 ground motions (refer to Fig. 5). The
force reduction factors predicted by Eq. (15)
based on the equal energy assumption are
presented here for comparison. From Fig. 13, it is
seen that at
T
µ =4, the equal energy assumption
provides a good estimation at natural periods
longer than 0.5 second at stiff and moderate sites
and 1.2 second at soft sites, while it provides
underestimation at natural periods shorter than
those values. On the other hand, at
T
µ =8, the
equal energy assumption provides a good
estimation at 0.6 second at stiff and moderate sites
and 1 second at soft sites. It underestimates and
overestimates the force reduction factors
0
4
8
12
0 1 2 3 4
Natural Period (s)
R
µ
=3.87
R
µ
=2.65
R
µ

f
a
c
t
o
r
µ
Τ
= 8
µ
Τ
= 4

(a) Stiff (Type-I)
0
4
8
12
0 1 2 3 4
Natural Period (s)
R
µ
=3.87
R
µ
=2.65
R
µ

f
a
c
t
o
r
µ
Τ
= 8
µ
Τ
= 4

(b) Moderate (Type-II)
0
4
8
12
0 1 2 3 4
Natural Period (s)
R
µ
=3.87
R
µ
=2.65
R
µ

f
a
c
t
o
r
µ
Τ
= 8
µ
Τ
= 4

(c) Soft (Type-III)
Fig.13 Force Reduction Factors Corresponding to
Means minus One Standard Deviations
Table 2 Parameters a and b (
NL
ξ =
EL
ξ =
0.02)
Soil Conditions
T
µ

a and
b Type-I Type-II Type-III
a 0.152 0.225 0.361 2
b 0.289 1.60 1.12
a 0.289 0.348 0.600 4
b 2.46 1.28 0.902
a 0.397 0.432 0.800 6
b 1.81 1.14 0.768
a 0.507 0.513 0.916 8
b 1.14 1.04 0.632
Table 3 Parameters a and b (
NL
ξ =
EL
ξ =
0.05)
Soil Conditions
T
µ

a and
b Type-I Type-II Type-III
a 0.226 0.344 0.521 2
b 4.14 1.94 1.34
a 0.778 0.572 0.976 4
b 3.50 1.35 0.994
a 0.981 0.725 1.23 6
b 2.93 1.15 0.757
a 1.23 0.807 1.28 8
b 2.57 0.983 0.569
corresponding to the mean minus one standard
deviation at natural periods shorter and longer,
respectively, than the above natural periods.


6. EFFECT OF DAMPING RATIOS

In the preceding analysis, the force reduction
factors were evaluated based on Eq. (1) assuming
EL
ξ =0.05 and
NL
ξ =0.02. However in the past
researches, damping ratios were usually assumed
as
EL
ξ =
NL
ξ =0.05. Consequently, the same
analysis presented in the preceding chapters was
conducted by assuming
EL
ξ =
NL
ξ =0.05 based
on Eq. (7) using the same ground motion data set.
For comparison, an analysis was also conducted
assuming
EL
ξ =
NL
ξ =0.02 based on Eq. (8).
Tables 2 and 3 show the parameters a and b
determined for a combination of
EL
ξ =
NL
ξ =0.02
and
EL
ξ =
NL
ξ =0.05, respectively. Fig. 14
compares a and b a / 1 + thus determined. Also
presented in Fig. 14 are a and b a / 1 + used in
the preceding chapter (
EL
ξ =0.05 and
NL
ξ =0.02,
refer to Fig. 9). It is seen in Fig. 14 that both a
and b a / 1 + at the same target ductility factors
are the shortest for a combination of
EL
ξ =
NL
ξ =0.02 and the longest for a
combination of
EL
ξ =0.05 and
NL
ξ =0.02.
Parameters a and b a / 1 + for a combination of
EL
ξ =
NL
ξ =0.05 are between the two cases.
ξ
NL
= ξ
EL
= 0.02
ξ
NL
= 0.02, ξ
EL
= 0.05
ξ
NL
= ξ
EL
= 0.5

0
1
2
3
0 2 4 6 8 10
µ
Τ
Target Ductility
a

(
s
)
0
1
2
3
0 2 4 6 8 10
µ
Τ
Target Ductility
a
+
1
/
b

(
s
)

(a) Stiff (Type-I) (a) Stiff (Type-I)
0
1
2
3
0 2 4 6 8 10
µ
Τ
Target Ductility
a

(
s
)
0
1
2
3
0 2 4 6 8 10
µ
Τ
Target Ductility
a
+
1
/
b

(
s
)

(b) Moderate (Type-II) (b) Moderate (Type-II)
0
1
2
3
0 2 4 6 8 10
µ
Τ
Target Ductility
a

(
s
)
0
1
2
3
0 2 4 6 8 10
µ
Τ
Target Ductility
a
+
1
/
b

(
s
)

(c) Soft (Type-III) (c) Soft (Type-III)
(1) Parameter a (2) Parameter a+1/b
Fig. 14 Dependence of Parameters a and a+1/b on the Assumption of Damping Ratios
Fig. 15 compares the mean values of the force
reduction factors based on the three assumptions
of damping ratios. Original force reduction factors
computed from the 70 ground motions are also
presented here for comparison. A systematic
difference of the force reduction factors is
observed reflecting the dependence of a and
b a / 1 + on the damping rations. The combination
of
EL
ξ =
NL
ξ =0.02 provides the largest
estimation for the force reduction factors, while
the combination of
EL
ξ =0.05 and
NL
ξ =0.02
provides the smallest estimation. The combination
of
EL
ξ =
NL
ξ =0.05 provides the estimation
between the two cases.


ξ
NL
= ξ
EL
= 0.02
ξ
NL
= 0.02, ξ
EL
= 0.05
ξ
NL
= ξ
EL
= 0.5
Present Study
Means


0
5
10
15
20
0 1 2 3 4
R
µ

f
a
c
t
o
r
Natural Period (s)

0
5
10
15
20
0 1 2 3 4
R
µ

f
a
c
t
o
r
Natural Period (s)

(a) Stiff (Type-I) (a) Stiff (Type-I)

0
5
10
15
20
0 1 2 3 4
R
µ

f
a
c
t
o
r
Natural Period (s)

0
5
10
15
20
0 1 2 3 4
R
µ

f
a
c
t
o
r
Natural Period (s)

(b) Moderate (Type-II) (b) Moderate (Type-II)

0
5
10
15
20
0 1 2 3 4
R
µ

f
a
c
t
o
r
Natural Period (s)

0
5
10
15
20
0 1 2 3 4
R
µ

f
a
c
t
o
r
Natural Period (s)

(c) Soft (Type-III) (c) Soft (Type-III)
(1)
T
µ =4 (2)
T
µ =8
Fig. 15 Dependence of Force Reduction Factors on the Assumption of Damping Ratios
7. COMPARISON WITH THE PREVIOUS
STUDIES

Fig. 16 shows a comparison of the force reduction
factor in the present study by Eq. (19) to Nassar
and Krawinkler by Eq. (11) and Miranda and
Bertero by Eq. (13). Since it is assumed in Eqs.
(11) and (13) that
EL
ξ =
NL
ξ =0.05, the same
damping ratios are assumed in the present study
for comparison. The original mean values of the
force reduction factors computed from the 70
ground motions are also presented here for
comparison. It is noted that definition of soil
conditions is not the same among three researches.
Hence they are classified into stiff, moderate and
soft. In the Miranda and Bertero formulation,
g
T
Present Study
Means Miranda et al.
Nassar et al.


0
4
8
12
16
0 1 2 3 4
Natural Period (s)
R
µ

f
a
c
t
o
r

0
4
8
12
16
0 1 2 3 4
Natural Period (s)
R
µ

f
a
c
t
o
r

(a) Stiff (Type-I) (a) Stiff (Type-I)

0
4
8
12
16
0 1 2 3 4
Natural Period (s)
R
µ

f
a
c
t
o
r

0
4
8
12
16
0 1 2 3 4
Natural Period (s)
R
µ

f
a
c
t
o
r

(b) Moderate (Type-II) (b) Moderate (Type-II)

0
4
8
12
16
0 1 2 3 4
Natural Period (s)
R
µ

f
a
c
t
o
r

0
4
8
12
16
0 1 2 3 4
Natural Period (s)
R
µ

f
a
c
t
o
r

(c) Soft (Type-III) (c) Soft (Type-III)
(1)
T
µ =4 (2)
T
µ =8
Fig. 16 Comparison with Previous Studies

was assumed 1.5 second at soft (alluvial) site in
Eq. (14).
From Fig. 16, it is seen that the present study
provides a quite similar result to the formulations
by Miranda & Mertero and Nassar & Krawinkler
if the same damping ratios are assumed in the
evaluation of linear and nonlinear responses.


8. CONCLUSIONS

An analysis was conducted for the force reduction
factor based on response of SDOF oscillator using
70 free-field ground motions. Based on the
analysis presented herein, the following
conclusions may be deduced:

1) A new formulation as shown in Eqs. (19) and
(24) was developed. The formulation is simpler
than the past formulations. Parameters a and
a +1/ b express the natural period where
µ
R is
equal to µ and
µ
R takes a peak value,
respectively.
2) Difference of the damping ratios assumed in
the evaluation of linear and nonlinear responses
(
EL
ξ and
NL
ξ ) provides a systematic difference
in the force reduction factors. The combination of
EL
ξ =
NL
ξ =0.02 provides the largest estimation
for the force reduction factors, while the
combination of
EL
ξ =0.05 and
NL
ξ =0.02
provides the smallest estimation. The combination
of
EL
ξ =
NL
ξ =0.05 provides the estimation
between the two cases. Hence, the damping ratios
have to be carefully assumed keeping how the
force reduction factors are used in mind.
3) Scattering of the force reduction factors
depending on ground motions is significant.
Although it has been pointed out that the equal
displacement assumption by Eq. (15) provides a
good estimation to the force reduction factors, it
provides a good estimation only to the mean
values; however, it considerably underestimates
the mean minus one standard deviation. On the
other hand, the equal energy assumption by Eq.
(16) provides a better estimation to the force
reduction factors corresponding to the mean
minus one standard deviation, although it provides
too conservative estimation to the mean values.
Taking account of the considerable scattering of
the force reduction factors depending on ground
motions, it is conservative to assume the equal
energy assumption instead of the equal
displacement assumption.
4) The response modification factors in the
present study by Eqs. (19) and (24) provides quite
close force reduction factors proposed by Nassar
and Krawinkler, and Miranda and Betero, if the
damping ratios are assumed as
EL
ξ =
NL
ξ =0.05.


REFERENCES

Japan Road association (2002), Part V Seismic
design, Design specifications of highway
bridges,” Maruzen, Tokyo, Japan.
Kawashima, K, MacRae, G. A., Hoshikuma, J.
and Nagaya, K. (1998). “Residual displacement
response spectra,” Journal of Structural
Engineering, 124(5), 513-530, ASCE
Kawashima, K. and Unjoh, S. (1989). “Damping
characteristics of cable-stayed bridges
associated with energy dissipation at movable
support,” Structural Engineering and
Earthquake Engineering, Proc. Japan Society of
Civil Engineers, 404/I-11, 123-130.
Kawashima, K., Unjoh, S. and Tsunomoto, M.
(1993). “Estimation of damping ratio of
cable-stayed bridges for seismic design,”
Journal of Structural Engineering, 119(4),
1015-1031, ASCE.
Miranda, E. and Bertero, V. (1994). “Evaluation
of strength reduction factors for earthquake
resistant design,” Earthquake Spectra, 10(2),
357-379.
Nassar, A. A. and Krawinkler, H. (1991).
“Seismic demands for SDOF and MDOF
systems,” Report No. 95, The John A. Blume
Earthquake Engineering Center, Stanford
University, California, USA.
Newmark, N. M. and Hall, W. J. (1973). “Seismic
design criteria for nuclear reactor facilities,”
Report No. 46, Building Practices for Disaster
Mitigation, National Bureau of Standards, U.S.
Department of Commerce, 209-236.
Press, W.H., Teukolsky, S.A., Vetterling, W.T. and
Flannery, B.P. (1996). “Numerical recipes in
Fortran 77,” Second Edition, The Art of
Scientific Computing, Cambridge University
Press, 678-683.
Priestley, M. J. N., Seible, F. and Calvi, G. M.
(1996). “Seismic design and retrofit of bridges,”
John Wiley & Sons, New York, USA.
Takeda, T., Sozen, M. A. and Nielsen, N. N.
(1970). “Reinforced concrete response to
simulated earthquake,” Journal of Structural
Engineering, 96(12), 2557-2573, ASCE

Representing u y the yield displacement where the stiffness changes from the cracked stiffness to the post-yield stiffness. and this results in an energy dissipation in a structure. It is generally very large such as 0. Unjoh. The natural period T may be evaluated based on the cracked stiffness of columns. Kawashima. If one considers a structure in which the flexural hysteretic energy dissipation is predominant with other sources of energy dissipation being a secondary importance. 2 shows the equivalent damping ratio by Eq. there must be some other sources of energy dissipation (for example.4 at the target ductility factor of 3-5. it has to be clarified carefully. On the other hand. respectively. a target ductility factor µT may be defined based on the yielding displacement u y as u µ T = max T (2) uy in which ∆W and W represent an energy dissipation in a hysteretic excursion and the elastic energy. µ T . Hence. The radiation of energy from a foundation to surround ground contributes to energy dissipation. Since the damping controls structural response. the energy dissipation in the nonlinear structural components has to be included in the analysis by other means. µT : target ductility factor. Structural damping such as friction at connections may be important in many structures (for example. It is general to idealize those sources of energy dissipation in terms of the equivalent viscous damping. Tsunomoto 1993). Kawashima and Unjoh 1989). respectively. In addition to such hysteretic energy dissipation. The equivalent viscous damping ratio ξ h is generally used for such a purpose as (1) ξh = FYNL (T . Viscous damping due to friction with air is generally predominant in a structure with a long natural period. the evaluation of damping ratio depends on the idealization of such an energy dissipation. 1 Definition of Force reduction Factor factor Rµ may be defined as R µ (T . ξ EL ) If one idealizes the energy dissipation in nonlinear structural components by incorporating nonlinear elements that represent the hysteretic behavior. ξ NL ) = EL FR (T . the flexural inelastic deformation of columns contributes to energy dissipation in a bridge. if one idealizes the nonlinear structural components by elastic linear elements. For example. FR and FYNL : maximum restoring force in an oscillator with a linear and a bilinear hysteresis. ξ EL . ξ NL ) 1 ∆W ⋅ 4π W (3) EL in which T : natural period. and ξ EL and ξ NL : damping ratio assumed in the evaluation of linear and bilinear responses. µ T . respectively. the equivalent damping ratio ξ h is 2 µ −1 (4) ξh = π µ Fig. . The post-yield stiffness is assumed to be 0 in the present study.F FEL FNL O δ δEL δNL δ Fig. in an oscillator with an elastic perfect-plastic bilinear hysteresis. A structure under a strong excitation generally exhibits strong hysteretic behavior. (4). the energy dissipation in the nonlinear structural components is automatically included in the analysis. the total damping ratio ξ eq of a SDOF oscillator in which u max T is a target maximum displacement of an oscillator. For example.

In the evaluation of the force reduction factor Rµ based on Eq.0. As a consequence. mode shape. Based on the original definition inherent to the force reduction factor. Using those damping ratios. (6).05 for the damping ratio including hysteretic energy dissipation of columns. (8) generally provides conservative estimation for the force reduction factors. (3).3 0. (6). it seems that Eq. (8). how damping ratios are assumed in the evaluation of the linear and the nonlinear responses is important. the hysteretic energy dissipation in the nonlinear structural components is counted twice in the evaluation of nonlinear response. It should be noted here that which is appropriate among Eqs. Although the equivalent damping ratio ξ h is very high as shown in Fig. the force reduction factor by Eq.02. if one assumes the damping ratios as ξ EL = ξ NL = ξ eq (7) or. ξ eq is assumed to be 0. On the other hand.05 in the present study based on Eq. if one assumes Eq. it is not general to assume such a high damping ratio in seismic design of a bridge structure. Hence. (8) ξ EL = ξ NL = ξ oth the force reduction factor by Eq. the hysteretic energy dissipation in the nonlinear structural components is counted by the equivalent viscous damping in addition to the inelastic excursion in the nonlinear elements. in addition to the effect of nonlinear response. (7) and (8) is also conducted for comparison with the previous studies. respectively. (7) and (8) is limited. However.1 00 2 4 6 Ductility Factor 8 Fig. (6). hysteresis and the target ductility factor. It should be noted that ξ h . On the other hand. If ξ oth ≈ 0 . (7) in the evaluation of nonlinear response. (1) includes the effect of different energy dissipation between the linear and nonlinear responses. Thus. If one assumes the damping ratios as ξ EL = ξ eq and ξ NL = ξ oth (6) the energy dissipation is essentially the same between the linear and the nonlinear responses.03 and 0.5 0. (6).6 Equivalent Damping Ratio ξ h 0. it is assumed here that ξ h and ξ oth is 0. the hysteretic energy dissipation is not taken into account in the evaluation of linear response. It is because a bridge structure is generally more complex than a SDOF oscillator. (1) reflects the effect of nonlinear response of an oscillator. the force reduction factors are evaluated in this study based on Eq. 2. Eq. (1) represents the difference of restoring force between the linear and nonlinear responses. ξ oth and ξ eq depend on the type of a structure. and this makes the relative contribution of the hysteretic energy dissipation of columns less significant.2 0. an analysis assuming Eqs. (7) and (8) depends on how the force reduction factor is used. the difference of the force reduction factors among Eqs. and ξ oth is the damping ratio that accounts the energy dissipation other than the hysteretic energy dissipation. (6) is the most appropriate. Since it is general practice in a standard bridge structure to assume about 0.4 0. Eq. . (4) may be provided as ξ eq = ξ h + ξ oth (5) in which ξ h is the damping ratio that accounts the hysteretic energy dissipation by Eq. By assuming Eq. (1).2 Equivalent Damping Ratio ξ h by Eq. Hence.

and proposed a force reduction factor as 1 ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ (0 ≤ T ≤ T1 / 10)  2.05.513 log ( / 2 µ −1 ) 1   T   2 µ − 1  4T     1  ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ (T1 / 10 ≤ T ≤ T1 / 4) Rµ =   ′  2 µ − 1 ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ (T1 / 4 ≤ T ≤ T1 )  ′ Tµ / T1 ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ (T1 ≤ T ≤ T1 ) µ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ (T ≤ T ≤ T ) 1 2  (9) where. stiffness to the initial elastic stiffness. velocity and displacement. It has been known that the equal energy assumption provides a good estimation for the force reduction factor at short periods while the equal displacement assumption at long periods. Nassar and Krawinkler proposed a force reduction factor. and φea . (15) and (16) and a comparison of the present study to the previous models will be described later. A . based on an analysis for 15 ground motions as (Nassar and Krawinkler 1991) Rµ = {c( µ − 1) + 1}1 / c (11) where 1+ T in which α represents a ratio of the post-yield c(T .  2  3  1 1 3    − 1+ exp−  ln T −    5   2  10T − µT 2T    ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ (rock)  2    1 2 1    exp− 2 ln T −   −  1+ Φ= 12T − µT 5T 5       ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ (alluvium)   2   T g 3T g 1    T     1 + 3T − 4T exp− 3 ln T − 4  g         ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ (soft)  (14) in which Tg represents a most predominant period. α ) = Tα α + b T (12) Application of Eqs.05. velocity and displacement. based on an analysis for 124 ground motions as (Miranda and Bertero 1994) µ −1 +1 > 1 Rµ = (13) Φ (T . and a and b are coefficients depending on α . respectively. Nassar and Krawinkler precisely analyzed the effect of stiffness deterioration. They used 10 ground motions recorded in the 1940 Imperial Valley Earthquake. They assumed ξ EL = ξ NL =0. V and D represent peak ground acceleration. The force reduction factor provided by the equal energy and the equal displacement assumptions are given as Rµ = 2 µ − 1 (equal energy) Rµ = µ (equal displacement) (15) (16) φ V T1 = 2π ev φ ea A T1′ = T1 µ 2µ − 1 (10) φ D T1 = 2π ed φ evV in which. respectively. . REVIEW OF INVESTIGATIONS THE PAST An early study for the force reduction factor was conducted by Newmark and Hall (Newmark and Hall 1973). Miranda and Bertero proposed a force reduction factor. assuming ξ EL = ξ NL =0. assuming ξ EL = ξ NL =0. Tg ) where. φev and φed represent the amplification for acceleration.05. and provided the coefficients a and b depending on α .3.

They are classified into three soil conditions depending on the fundamental natural period of subsurface ground Tg .02 based on Eq.1 10 Stiff (Type-I) Moderate (Type-II) Soft (Type-IIII) 100 1000 Epicentral Distance (km) (a) Soil Condition Peak Acceleration (m/sec2) 10 1 0. This will be discussed later. Obviously smaller force reduction factors should be assumed in design to provide conservative design. It is observed in Fig. and then they approach to µT at long period.5 .2 s). natural period and soil condition. stiff ( Tg < 0. Distribution of peak ground accelerations on the earthquake magnitudes and epicentral distances is shown in Fig. (15) provides a good estimation to the force reduction factor.0 100 1000 Epicentral Distance (km) (b) Earthquake Magnitude Fig. and the epicentral distances are in the range of 10-500 km. the force reduction factors varies from 1. 3. (6). Only the results for µT =4 and 6 are presented here since the results for other target ductility factors show the similar characteristics.7. It is seen in Fig. moderate and soft categories is 16.one standard deviations of the force reduction factors were obtained for each target ductility factor. On the other hand. (15) provides a good estimation to the mean values. 4 that scattering of the force reduction factors depending on ground motions is significant. The peak accelerations are in the range of 0. respectively. The force reduction factors predicted by Eqs.3 depending on ground motions for µ T =4 at the moderate soil sites.9 to 10. Fig.6 s) (Japan Road Association 2002).2 ≤ Tg < 0. 3 Classification of Ground Accelerations in Terms of Soil Conditions and Earthquake Magnitudes factors depending on ground motions is so large that the means +/. It has been pointed out in the previous researches that the Eq. 4. 6 and 8 assuming an elastic perfect-plastic bilinear hysteresis. moderate ( 0.4.5 .0 . 4. Damping ratio in the linear and nonlinear analyses is assumed as ξ EL =0. However. FORCE REDUCTION FACTOR FOR BILINEAR OSCILLATORS Force reduction factors were evaluated for target ductility factor µ T of 2. 4 shows the force reduction factors for the 70 ground motions.7. 4 that the dependence of force reduction factors on the soil condition is less significant. Fig. Eq. For example at natural period of 1 second.0 7. Since the scattering of the force reduction Peak Acceleration (m/sec2) 10 1 0.1-8m/s2. It is apparent that such a large scattering of the force reduction factors result in a large change of sizing of a structure in seismic design. Seventy free field ground accelerations by 64 shallow earthquakes with depth less that 60 km were used for analysis. but it considerably underestimates the force reduction factors corresponding to the mean values minus one standard deviations. 5 shows the mean values and the mean values +/. The mean values of force reduction factors increase as the natural periods increase.8.5 7.05 and ξ NL =0.one standard deviations of the force reduction factors presented in Fig. (16) . Number of records in the stiff. (15) and (16) based on the equal displacement and the equal energy assumptions are also presented here for comparison.1 10 6. it is noted that Eq. 39 and 15.6 s) and soft ( Tg ≥ 0.

4 Force Reduction Factors (c) Soft (Type-III) (1) µ T =4 (2) µ T =6 provides better estimation to the mean values minus one standard deviations. . Taking account of the force reduction factors having considerable scattering depending on ground motions. it seems reasonable to consider a certain redundancy in the estimation of the force reduction factor in design.24 20 16 12 Rµ factor 8 4 0 0 1 2 3 4 Rµ factor 16 12 8 4 0 0 1 2 3 4 Natural Period (s) Natural Period (s) (a) Stiff (Type-I) 24 20 (a) Stiff (Type-I) 16 Rµ factor 16 12 8 4 12 Rµ factor 8 4 0 0 1 2 3 4 0 0 1 Natural Period (s) Natural Period (s) 2 3 4 (b) Moderate (Type-II) 24 20 (b) Moderate (Type-II) 16 Rµ factor 16 12 8 4 12 Rµ factor 8 4 0 0 1 (c) Soft 2 3 4 0 0 1 Natural Period (s) Natural Period (s) 2 3 4 (Type-III) Fig.

and decrease after taking peak values at natural period of 1-2 second.1σ 12 Rµ factor Rµ factor 8 4 0 0 1 2 3 Natural Period (s) 4 12 8 4 00 Rµ = 2µ-1 Rµ = µ 1 2 3 Natural Period (s) 4 (a) Stiff (Type-I) 12 Rµ factor Rµ factor 8 4 00 1 2 3 Natural Period (s) 4 12 8 4 00 (a) Stiff (Type-I) 1 2 3 Natural Period (s) 4 (b) Moderate (Type-II) 12 Rµ factor 8 4 0 0 1 2 3 Natural Period (s) 4 Rµ factor 12 8 4 0 0 (b) Moderate (Type-II) 1 2 3 Natural Period (s) 4 (c) Soft (Type-III) (1) µ T =4 (c) Soft (Type-III) (2) µ T =6 Fig. 7.Average + 1σ Average Average .One Standard Deviation of the Force Reduction Factors for 70 Ground Motions Based on such a consideration.328 + 0. 6 shows the dependence of the standard deviations of force reduction factors σ ( Rµ ) on the natural periods T and the soil condition. Similar to the mean values.379 ⋅ µ T  σ ( R µ ) = − 0. Fig. The relation may be approximated by a least square fit as (stiff)  − 0. it is more conservative to assume Eq.378 ⋅ µ T (moderate) (17)  − 0.3 + 0.292 + 0. the standard deviations σ ( Rµ ) increase as the natural periods increase. Fig. Eq. 7 shows the dependence of the standard deviations σ ( Rµ ) on the target ductility factors µT . 5 Mean and Mean +/. (16) instead of Eq. (15) for a design purpose.4 ⋅ µT (18) .409 ⋅ µ (soft) T  As the soil condition dependence of σ ( Rµ ) is less significant as shown in Fig. (17) may be approximated as σ ( Rµ ) ≈ −0. The standard deviations σ ( Rµ ) increase as the target ductility factors increase.354 + 0.

(20). the coefficient c can be eliminated as c = 1 / ae ab (23) Substitution of Eq. 6 Natural Period Dependence of Standard Deviations of the Force Reduction Factors 10 8 σ (Rµ) 6 4 2 0 0 2 4 6 8 Target Ductility µT 10 Stiff (Type-I) Moderate (Type-II) Soft (Type-III) in which a . 8 Idealization of Force Reduction Factors where. 1 / b represents the period between Points P and Q. FORMULATION OF REDUCTION FACTORS FORCE To idealize the mean values of the force reduction factors in Fig. the following condition has to be satisfied in Rµ T →0 Fig. 7 Target Ductility Factor Dependence of the Standard Deviations of Force reduction Factors lim Rµ = 0 (22) Hence. Representing Q as the point where Rµ takes the peak value. 5. b and c are parameters to be determined. Based on the definition. (20) as Ψ (T ) = T −a ae bT +1 (24) 5. 8. Ψ (T ) = c ⋅ T −a e b⋅(T −a ) +1 (20) 8 6 σ ( Rµ ) 4 2 0 0 1 2 3 Natural Period (s) (b) 4 µ T =6 Fig. Because the gradient of Rµ is dRµ 1 − b(T − a ) = c( µ − 1) ⋅ (21) dT e b (T −a ) it is c ⋅ ( µ − 1) at Point P. (19) automatically satisfies the following condition lim R µ = µ T →∞ (25) . the parameter c represents the gradient at Point c divided by µ − 1 . Since Rµ = µ at T = a in Eq. they are represented as (19) Rµ = ( µ − 1) ⋅ Ψ (T ) + 1 It is noted that Eq.8 Rµ factor 6 σ ( Rµ ) 4 2 0 0 1 Stiff (Type-I) Moderate (Type-II) Soft (Type-III) µ c (µ−1) P Q 1 a 0 1 1/b 2 3 4 Natural Period (s) 5 2 3 Natural Period (s) (a) 4 0 µ T =4 Fig. Consequently. the parameter a represents the period where Rµ is equal to µ (Point P) as shown in Fig. (23) makes Eq.

611 R 0.379 0.989 1.2 0 0 2 4 6 8 Target Ductility µΤ 10 Stiff (Type-I) Moderate (Type-II) Soft (Type-III) Fig.05) 4 3 a (s) 2 1 0 0 2 4 Stiff (Type-I) Moderate (Type-II) Soft (Type-III) ηT 2 4 6 8 a.701 0.0-1.Table 1 Parameters a and b and Regression Coefficients R ( ξ NL =0. 8.4 second at stiff and Target Ductility µΤ (c) Parameter a + 1 b Fig. As described before.869 0. Consequently.717 0.12 2.886 a 1.4 0.0-1.776 0.74 b 1. a + 1 / b represents the natural period where Rµ takes the peak value.03 1. (15)) moderate sites. it may be accepted in other conditions.24 0.24 0.5-2.02 and ξ EL = 0. 9 Parameters a and a+1/b in Eqs.895 (a) Parameter a 2 1/b (s) 1 0 0 2 4 6 Target Ductility µΤ 6 8 10 Target Ductility µΤ (b) Parameter 1 b 8 10 4 3 a+1/b (s) 2 1 0 0 2 4 6 8 10 It is a feature of the above formulation that the equation is simpler and the physical meaning of the parameters a and b is clearer than the previous studies. The mean values of force reduction factors in Fig.05 R 0. b Soil Conditions and R Type-I Type-II Type-III a 1.35 b 2.4 second at stiff and moderate sites. and 1.4 second at soft sites.5-2 second at stiff and moderate sites.77 2. (19) using a nonlinear least square method (Press et al 1996).52 b 2.4 second at soft site.6 0. which implies that the equal displacement assumption by Eq.29 1.39 1. 1 / b and a + 1 / b .18 1.899 0. the accuracy of equal displacement assumption is high at 1. They are less sensitive to the target ductility factor µT between 2 and 8.34 1.11 0.85 b 2. Table 1 represents the a and b as well as the regression coefficients. a represents the period where Rµ = µ . As shown in Fig. Although the regression coefficient is not high enough for some combinations such as µT =2 and stiff sites.673 0.5-2.8 Period (s) 0.36 1.15 1. 5 were fitted by Eq.62 1. (16) provides the best estimation at period a . 9 shows parameters a . (20) 0.842 0. It is 1.67 1. the fitting is not necessarily poor for a combination of µT =2 and stiff sites. and 1.851 a 1.69 R 0. Fig.878 a 1. 10 Natural Periods where Force Reduction Factors Take Values Predicted by the Equal Energy Assumption (Eq.821 R 0.20 1. As shown later. Parameter a is in the range of 1. and .

respectively.2-0. 10.6 second at stiff.µT = 2 µT = 4 µT = 6 µT = 8 16 12 Rµ factor 8 Empirical Model Mean Rµ factor 12 8 4 0 0 1 2 3 4 Stiff (Type-I) Moderate (Type-II) Soft (Type-III) Natural Period (s) (a) µ T =2 12 Rµ factor 8 4 0 0 1 2 3 4 4 0 0 1 2 3 4 Natural Period (s) (a) Stiff (Type-I) 16 12 Rµ factor Natural Period (s) (b) µ T =4 12 Rµ factor 1 2 3 4 Natural Period (s) 8 4 0 0 8 4 0 0 1 2 3 4 (b) Moderate (Type-II) 16 12 Rµ factor 8 4 0 0 1 2 3 4 Rµ factor 12 8 4 0 Natural Period (s) Natural Period (s) (c) µ T =6 0 1 Natural Period (s) 2 3 4 (c) Soft (Type-III) Fig.4-0. It slightly increases as target ductility µT increases. 11 compares the mean force reduction factors presented in Fig. (15) based on the equal energy assumption are obtained as shown in Fig.5-3. (19) . (19) to the Mean Force Reduction Factors Presented in Fig. They are much shorter than the natural periods where the equal displacement assumption provides the best approximation. (19). They are in the range of 0. 5 2. Although some discrepancies are observed at larger target ductility factors. Fig. 11 Application of Eq. moderate and (d) µ T =8 Fig.26-0. Eq.5 second at soft site. 12 Effect of Soil Condition on the Force Reduction Factors Predicted by Eq. 0. The natural periods where R µ take the values predicted by Eq. 5 to the values predicted by Eq. (19) soft sites.4 second and 0.36 second.

507 1. The force reduction factors predicted by Eq. The mean and the standard deviation of force reduction factors were evaluated by by Eq.289 2. Fig.65 1 2 3 Natural Period (s) 4 Rµ =3.14 1.28 0. (19). On the other hand. As shown in Fig. the equal energy assumption provides a good estimation at natural periods longer than 0. while it provides underestimation at natural periods shorter than those values.569 close to the force reduction factors of the mean minus one standard deviation directly computed from the 70 ground motions (refer to Fig. 5). in particular at small target ductility factors. at µT =8.725 1.6 second at stiff and moderate sites and 1 second at soft sites.87 µΤ = 8 µT 2 4 6 8 (c) Soft (Type-III) Fig.778 0.397 1.13 Force Reduction Factors Corresponding to Means minus One Standard Deviations provides a good estimation for the mean force reduction factors.23 0. 4. in this estimation.35 0. the equal energy assumption provides a good estimation at 0.14 0. (19) and Eq.981 0.93 1.60 1.976 3. 12 shows the effect of soil condition on the mean force reduction factors estimated by Eq.344 0.768 0.65 1 2 3 Natural Period (s) 4 Rµ =3.632 Table 3 Parameters a and b ( ξ NL = ξ EL = 0. From Fig. 13. respectively.65 1 2 3 Natural Period (s) 4 Rµ =3.2 second at soft sites.02) µT a and 2 4 6 b a b a b a b a b Type-I 0.94 1.23 2.521 4. the force reduction factors corresponding to the mean values m substituted by a standard deviation σ ( Rµ ) are evaluated as shown in Fig.994 0.14 (a) Stiff (Type-I) 12 Rµ factor 8 4 0 0 µΤ = 4 Rµ =2.348 0.04 0.87 µΤ = 8 Table 2 Parameters a and b ( ξ NL = ξ EL = 0.757 1.800 1. The effect of soil condition is less significant on the force reduction factors.05) (b) Moderate (Type-II) 12 Rµ factor 8 4 0 0 µΤ = 4 Rµ =2.807 1.81 0.57 0.600 1.361 1. scattering of the force reduction factors around the mean values is extensive.34 0. (18).50 1.46 0.12 0.152 0.902 0.5 second at stiff and moderate sites and 1.12 Rµ factor 8 4 0 0 µΤ = 4 Rµ =2. it is seen that at µT =4. It underestimates and overestimates the force reduction factors . (15) based on the equal energy assumption are presented here for comparison.28 2. They are of course a and b a b a b a b a b Soil Conditions Type-I Type-II Type-III 0.572 0.15 0.513 0.225 0. Hence.983 0. 13.289 0.226 0.916 1.87 µΤ = 8 8 Soil Conditions Type-II Type-III 0.432 0.

ξNL = ξEL = 0.02. Consequently. ξEL = 0. EFFECT OF DAMPING RATIOS In the preceding analysis.02 and the longest for a combination of ξ EL =0.5 3 a (s) 2 1 0 0 2 4 6 8 Target Ductility µΤ 10 a+1/b (s) 3 2 1 0 0 2 4 6 8 Target Ductility µΤ 10 (a) Stiff (Type-I) (a) Stiff (Type-I) 3 a (s) 2 1 0 0 2 4 6 8 Target Ductility µΤ 10 a+1/b (s) 3 2 1 0 0 2 4 6 8 Target Ductility µΤ 10 (b) Moderate (Type-II) 3 a (s) 2 1 0 0 2 4 6 8 Target Ductility µΤ 10 a+1/b (s) 3 2 1 0 0 (b) Moderate (Type-II) 2 (c) Soft (Type-III) (1) Parameter a 4 6 8 Target Ductility µΤ 10 (c) Soft (Type-III) (2) Parameter a+1/b Fig. respectively.02 ξNL = 0.05 and ξ NL =0. 14 compares a and a + 1 / b thus determined. respectively. .02.02. Fig. refer to Fig.02. the same analysis presented in the preceding chapters was conducted by assuming ξ EL = ξ NL =0. However in the past researches.05. (1) assuming ξ EL =0.05.02 and ξ EL = ξ NL =0. than the above natural periods. the force reduction factors were evaluated based on Eq. 9). Tables 2 and 3 show the parameters a and b determined for a combination of ξ EL = ξ NL =0. Parameters a and a + 1 / b for a combination of ξ EL = ξ NL =0. an analysis was also conducted assuming ξ EL = ξ NL =0.05 and ξ NL =0. (8).05 are between the two cases.05 based on Eq. 6.02 based on Eq. 14 are a and a + 1 / b used in the preceding chapter ( ξ EL =0. Also presented in Fig.05 ξNL = ξEL = 0. 14 Dependence of Parameters a and a+1/b on the Assumption of Damping Ratios corresponding to the mean minus one standard deviation at natural periods shorter and longer. It is seen in Fig. 14 that both a and a + 1 / b at the same target ductility factors are the shortest for a combination of ξ EL = ξ NL =0. For comparison.05 and ξ NL =0. (7) using the same ground motion data set. damping ratios were usually assumed as ξ EL = ξ NL =0.

02 provides the largest estimation for the force reduction factors.02 ξNL = 0.05 and ξ NL =0. The combination of ξ EL = ξ NL =0.05 ξNL = ξEL = 0. A systematic difference of the force reduction factors is observed reflecting the dependence of a and a + 1 / b on the damping rations. The combination of ξ EL = ξ NL =0. Original force reduction factors computed from the 70 ground motions are also presented here for comparison.Present Study ξNL = ξEL = 0.02 provides the smallest estimation. 15 Dependence of Force Reduction Factors on the Assumption of Damping Ratios Fig.02.5 20 Rµ factor Rµ factor 15 10 5 0 0 1 2 3 Natural Period (s) 4 20 15 10 5 0 0 Means 1 2 3 Natural Period (s) 4 (a) Stiff (Type-I) 20 Rµ factor Rµ factor 15 10 5 0 0 1 2 3 Natural Period (s) 4 20 15 10 5 0 0 (a) Stiff (Type-I) 1 2 3 Natural Period (s) 4 (b) Moderate (Type-II) 20 Rµ factor Rµ factor 15 10 5 0 0 1 2 3 Natural Period (s) 4 20 15 10 5 0 0 (b) Moderate (Type-II) 1 2 3 Natural Period (s) 4 (c) Soft (Type-III) (1) µ T =4 (c) Soft (Type-III) (2) µ T =8 Fig. while the combination of ξ EL =0. . ξEL = 0. 15 compares the mean values of the force reduction factors based on the three assumptions of damping ratios.05 provides the estimation between the two cases.

16 shows a comparison of the force reduction factor in the present study by Eq. (13). The original mean values of the force reduction factors computed from the 70 ground motions are also presented here for comparison. (19) to Nassar and Krawinkler by Eq.Present Study Miranda et al. COMPARISON WITH THE PREVIOUS STUDIES Fig. 16 Comparison with Previous Studies 7. moderate and soft. In the Miranda and Bertero formulation. (11) and (13) that ξ EL = ξ NL =0. (11) and Miranda and Bertero by Eq. Hence they are classified into stiff. Since it is assumed in Eqs. Tg . 16 12 Nassar et al. Means 16 12 Rµ factor Rµ factor 8 4 0 0 1 2 3 Natural Period (s) 4 8 4 0 0 1 2 3 Natural Period (s) 4 (a) Stiff (Type-I) 16 12 16 12 (a) Stiff (Type-I) Rµ factor 8 4 0 0 1 2 3 Natural Period (s) 4 Rµ factor 8 4 0 0 1 2 3 Natural Period (s) 4 (b) Moderate (Type-II) 16 12 16 12 (b) Moderate (Type-II) Rµ factor 8 4 0 0 1 2 3 Natural Period (s) 4 Rµ factor 8 4 0 0 1 2 3 Natural Period (s) 4 (c) Soft (Type-III) (1) µ T =4 (c) Soft (Type-III) (2) µ T =8 Fig.05. It is noted that definition of soil conditions is not the same among three researches. the same damping ratios are assumed in the present study for comparison.

while the combination of ξ EL =0. 404/I-11. S. the damping ratios have to be carefully assumed keeping how the force reduction factors are used in mind. K. Hoshikuma. (19) and (24) provides quite close force reduction factors proposed by Nassar and Krawinkler.02 provides the smallest estimation.” Journal of Structural Engineering. Kawashima.. REFERENCES Japan Road association (2002).05 provides the estimation between the two cases. Tokyo. G. Kawashima.” Journal of Structural Engineering. Parameters a and a +1/ b express the natural period where R µ is equal to µ and Rµ takes a peak value. 119(4). Hence. V. The combination of ξ EL = ξ NL =0. 1015-1031. 10(2). (19) and (24) was developed. and Bertero. “Estimation of damping ratio of cable-stayed bridges for seismic design.05 and ξ NL =0. ASCE. 124(5). and Unjoh. if the damping ratios are assumed as ξ EL = ξ NL =0. Japan. Miranda. Design specifications of highway bridges. the equal energy assumption by Eq. K. 16. ASCE Kawashima.was assumed 1. 3) Scattering of the force reduction factors depending on ground motions is significant. (1994). . Taking account of the considerable scattering of the force reduction factors depending on ground motions. Based on the analysis presented herein. Although it has been pointed out that the equal displacement assumption by Eq. From Fig. Unjoh. and Nagaya. the following conclusions may be deduced: 1) A new formulation as shown in Eqs. “Evaluation of strength reduction factors for earthquake resistant design.02 provides the largest estimation for the force reduction factors. A. (15) provides a good estimation to the force reduction factors.” Structural Engineering and Earthquake Engineering.” Maruzen. it considerably underestimates the mean minus one standard deviation. 8. although it provides too conservative estimation to the mean values. 357-379. CONCLUSIONS An analysis was conducted for the force reduction factor based on response of SDOF oscillator using 70 free-field ground motions. S. (16) provides a better estimation to the force reduction factors corresponding to the mean minus one standard deviation. (1989). Proc. “Residual displacement response spectra.05. 2) Difference of the damping ratios assumed in the evaluation of linear and nonlinear responses ( ξ EL and ξ NL ) provides a systematic difference in the force reduction factors. (1993). The formulation is simpler than the past formulations. and Tsunomoto. (14). (1998). The combination of ξ EL = ξ NL =0.5 second at soft (alluvial) site in Eq. K. K. 123-130. however. J. 4) The response modification factors in the present study by Eqs. 513-530. MacRae. E..” Earthquake Spectra. M. it provides a good estimation only to the mean values. “Damping characteristics of cable-stayed bridges associated with energy dissipation at movable support. Japan Society of Civil Engineers. and Miranda and Betero. Part V Seismic design. it is seen that the present study provides a quite similar result to the formulations by Miranda & Mertero and Nassar & Krawinkler if the same damping ratios are assumed in the evaluation of linear and nonlinear responses. On the other hand. it is conservative to assume the equal energy assumption instead of the equal displacement assumption. respectively.

T. W. 678-683.. Building Practices for Disaster Mitigation. 46. Blume Earthquake Engineering Center. New York. (1991).” Journal of Structural Engineering. The John A.H. A.” John Wiley & Sons. ASCE . Newmark. (1996). S. M. W. A. 209-236. N. USA. “Seismic demands for SDOF and MDOF systems. M. and Flannery. N.” Report No. (1970). Press. J. U. W.” Second Edition. and Hall. 95. 96(12). J. A. B. Takeda. California. Seible. and Nielsen. and Krawinkler. Cambridge University Press. (1973). F. G. Vetterling. H.. N.Nassar.. “Seismic design and retrofit of bridges. Stanford University. Department of Commerce.” Report No. “Reinforced concrete response to simulated earthquake. “Numerical recipes in Fortran 77. Sozen. Teukolsky.S. (1996). T. M.. “Seismic design criteria for nuclear reactor facilities.A.P. The Art of Scientific Computing. 2557-2573. Priestley. M. USA. N. National Bureau of Standards. and Calvi.