Download as pdf or txt
Download as pdf or txt
You are on page 1of 18

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/230001366

Cyclic‐demand spectrum

Article  in  Earthquake Engineering & Structural Dynamics · July 2002


DOI: 10.1002/eqe.171

CITATIONS READS
64 345

1 author:

Praveen K. Malhotra
StrongMotions Inc.
84 PUBLICATIONS   1,615 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Performance-based seismic design of a liquid nitrogen tank in California. View project

Ground Motion Analysis and Interpretation View project

All content following this page was uploaded by Praveen K. Malhotra on 14 October 2018.

The user has requested enhancement of the downloaded file.


EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS
Earthquake Engng Struct. Dyn. 2002; 31:1441–1457 (DOI: 10.1002/eqe.171)

Cyclic-demand spectrum

Praveen K. Malhotra∗; †
Factory Mutual Research; 1151 Boston-Providence Turnpike; P.O. Box 9102; Norwood; MA 02062; U.S.A.

SUMMARY
The seismic response spectrum denes the amplitude of the load, but it does not specify the number
of cycles for which the load must be resisted by the structure. The amplitude by itself is not su-
cient to evaluate the seismic resistance of a structure, because the structure’s strength, stiness and
energy-dissipation capacity reduce with an increase in the number of load cycles. This paper presents
a cyclic-demand spectrum, which, in conjunction with the amplitude spectrum, provides a more com-
plete denition of the seismic load, hence a way to consider the degradation in strength, stiness and
energy-dissipation capacity in a rational manner. Similarly to three amplitude parameters (peak ground
acceleration, peak ground velocity, and peak ground displacement), three cyclic-demand parameters are
introduced for sti, moderately sti, and exible systems. A design example is presented to illustrate
the use of the cyclic-demand spectrum. Copyright ? 2002 John Wiley & Sons, Ltd.

KEY WORDS: duration; strong motion; fatigue; strength degradation; performance-based design

INTRODUCTION

The response spectrum provides an important yet incomplete denition of the seismic load. It
denes the amplitude of the load (force and deformation), but it does not specify the number
of cycles for which the load must be resisted by the structure. The amplitude by itself is not
sucient to evaluate the seismic resistance of a structure, because the structure’s strength,
stiness and energy-dissipation capacity depend on the number of load cycles.
To address the cyclic demand, researchers have proposed numerous denitions of strong-
motion duration [1–9]. Duration, however, is an indirect measure of the cyclic demand, and
duration derived from the acceleration history is useful for only sti systems, which respond
to ground accelerations. Moderately sti to exible systems respond to ground velocities and
displacements. For such systems, duration, derived from the acceleration history, is of limited
value. A single parameter of duration cannot address the cyclic demand, even indirectly, for
all structural systems.

∗ Correspondence to: Praveen K. Malhotra, Factory Mutual Research, 1151 Boston-Providence Turnpike, P.O. Box
9102, Norwood, MA 02062, U.S.A.
† E-mail: praveen.malhotra@fmglobal.com

Received 19 June 2001


Revised 7 November 2001
Copyright ? 2002 John Wiley & Sons, Ltd. Accepted 7 November 2001
1442 P. K. MALHOTRA

PROBLEM DEFINITION AND OBJECTIVE

Figure 1 shows the processed acceleration, velocity and displacement histories of ground mo-
tions recorded in horizontal directions during the magnitude 6:7Mw 1994 Northridge, California
earthquake [10]. Figure 2 shows the acceleration–deformation response spectra of the motions
shown in Figure 1, for 10% of critical damping. The spectral acceleration is plotted versus
deformation and the natural period is indicated by radial tick-marks. The spectral acceleration
SA, spectral deformation SD, and period T are related to each other as follows [11]:
 2
T
SD = SA × (1)
2

Because of incompatibility among the processed acceleration, velocity and displacement histo-
ries, the spectra shown in Figure 2 were computed by the method described in Reference [12].
According to Figure 2, an ‘equivalent’ linear system [13; 14], with period T = 2s and 10%
of critical damping (derived from the secant stiness method [15; 16]), will undergo a defor-
mation of SD = 50 cm and experiences an acceleration of SA = 0:5 g in the North direction. In
other words, a structure, which has an ultimate base-shear capacity of 0:5W (50% its weight)
and exhibits 10% of critical damping, will withstand the ground motions shown in Figure 1,
if the structure can deform 50 cm in the North direction and 14 cm in the East direction (see
Figure 2). However, 50 cm is the maximum deformation in the North direction, which occurs

Figure 1. Processed acceleration, velocity and displacement histories of horizontal motions recorded at
Sylmar Hospital free-eld site during the magnitude 6:7Mw 1994 Northridge, California earthquake [10].

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1441–1457
CYCLIC-DEMAND SPECTRUM 1443

Figure 2. Acceleration–deformation response spectra of the motions shown in Figure 1,


for 10% of critical damping.

Figure 3. Deformation history of a system with 2 s period and 10% of critical damping, subjected to
the North direction motion shown in Figure 1(a).

once during the response of the structure (at t = 4:78 s, see Figure 3). The structure must
withstand not only the maximum deformation half-cycle (shown by dashed line), but sev-
eral other half-cycles of lesser deformation. Each half-cycle causes some damage to the
structure. The amplitude spectra, shown in Figure 2, provide no indication of the cyclic
demand.
The objective of this paper is to present a cyclic-demand spectrum, which, together with
the amplitude spectrum, provides a more complete denition of the seismic load.

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1441–1457
1444 P. K. MALHOTRA

NUMBER OF CYCLES

We will now determine a uniform-amplitude deformation history, consisting of a certain num-


ber of cycles, which causes the same damage to the structure as the non-uniform deformation
history shown in Figure 3.

Cumulative damage model

Con [17] and Manson [18], independently, proposed a simple model for low-cycle fatigue
failure. It states that the number of uniform-amplitude cycles to failure is
1
Nf = (2)
C · upc
where up is the plastic deformation; and c and C are structural damage parameters determined
from experiments. From tests done on steel samples, c was found to be approximately equal
to 2 [19; 20]. Assuming that each cycle contributes equally to the damage, the damage caused
by a single cycle of deformation amplitude up is
1
D(up ) = = C · upc (3)
Nf
The damage caused by the deformation history shown in Figure 3 can be obtained by adding
the damage caused by individual cycles, using Miner’s rule [21], i.e.

n
D=C · upic (4)
i=1

D = 1 implies complete damage (fatigue-failure). In this model, the load–sequence (relative


occurrence of large cycles with respect to small cycles) and the load–rate (frequency) eects
are ignored. Additional discussion on the model may be found in Reference [20].
Under severe shaking, the plastic deformation is a major contributor to the total deformation,
i.e. up ≈ u. (This assumption is later shown to give a conservative estimate of the number
of cycles). Because the deformation history shown in Figure 3 does not contain symmetric
cycles (with equal positive and negative amplitudes), it was decided to accumulate damage
from each half-cycle instead of each full-cycle. There are several other ways to accumulate
damage caused by individual cycles [20], but the method of half-cycles was used because of
its simplicity. The damage expression (Equation (4)) may then be re-written as
C 2n
D= · uic (5)
2 i=1
where, ui is the deformation amplitude of the ith half-cycle, and n is now the number of
half-cycles.
Dividing the cumulative damage D (Equation (5)) by the damage caused by a full-cycle of
largest amplitude umax , we obtain the equivalent number of cycles (of amplitude umax ), which
cause the same damage to the structure as the entire deformation history, i.e.
 c
1  2n ui
Ncy = · (6)
2 i=1 umax

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1441–1457
CYCLIC-DEMAND SPECTRUM 1445

Figure 4. Uniform amplitude deformation history that will cause the same damage as the non-uniform
deformation history shown in Figure 3, for damage exponent c = 2 (Equation (6)).

By applying Equation (6) to the deformation history shown in Figure 3, we obtain Ncy = 2,
for c = 2. In other words, the uniform-deformation history shown in Figure 4 causes the same
damage to the structure as the non-uniform deformation history shown in Figure 3. The time
scale in Figure 4 is arbitrary because the load–rate (frequency) eect is ignored.

Eect of non-linear force–deformation relationship on number of cycles

The deformation history shown in Figure 3 was obtained from an ‘equivalent’ linear analysis
using secant stiness—the stiness was assumed to remain constant throughout the response
history. In reality, the stiness reduces with increase in deformation. Therefore, in a true
non-linear analysis, the smaller amplitude cycles are associated with greater stiness, and
will become even smaller compared with the maximum amplitude cycle. This will result
in a fewer number of equivalent cycles (Equation (6)). Therefore, the ‘equivalent’ linear
assumption made in this study gives a conservative estimate of the number of cycles.

Eect of elastic deformation on number of cycles

As per the Con-Manson model (Equation (2)), the damage is caused only by plastic de-
formations. The deformation history shown in Figure 3 has some contribution from elastic
deformations. If the elastic component is removed, the smaller amplitude cycles will become
even smaller compared to the maximum amplitude cycle. Several smaller amplitude cycles
will be completely eliminated, as they do not have any plastic component. This too will
reduce the number of equivalent cycles (Equation (6)). Therefore, the inclusion of elastic
deformations in Equation (6) gives a conservative estimate of the number of cycles.

FACTORS AFFECTING NUMBER OF CYCLES

As per Equation (6), the number of equivalent cycles depends on the damage exponent c and
the shape of the deformation history. It does not depend on the amplitude of the deformation

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1441–1457
1446 P. K. MALHOTRA

Figure 5. Cyclic-demand spectra for ground motion shown in Figure 1(a), for three values of damping
and a xed value of the damage exponent c = 2.

history. The shape of the deformation history depends on the ‘eective’ period and damping
of the system, and the shapes of the acceleration, velocity and displacement histories.

Eect of period and damping

Figure 5 shows the number of cycles versus period for three values of damping (& = 5, 10 and
20% of critical) and a xed value of the damage exponent c = 2. From here on, we will refer
to these plots as the cyclic-demand spectra. Note that the number of cycles Ncy approaches
∼ 2 for very sti (short-period) systems and ∼ 1 for very exible (long-period) systems,
irrespective of the damping ratio. For systems that are neither very sti nor very exible, the
number of cycles reduces with increase in the damping ratio.
For very sti systems, the deformation history is directly proportional to the ground ac-
celeration history [11]. Therefore, the number of cycles for such systems is obtained directly
from the acceleration history, irrespective of the damping ratio. For very exible systems,
the deformation history is the same as the ground displacement history [11]. The number of
cycles for such systems is, therefore, obtained directly from the ground displacement history,
irrespective of the damping ratio. As in the case of the amplitude spectrum, the eect of
damping is most pronounced for systems that are neither very sti nor very exible.

Eect of damage exponent

Figure 6 shows the cyclic-demand spectra for 10% of critical damping and three values of
the damage exponent c = 1:5, 2 and 3. According to Equation (6), the higher the value of
c, the greater the damage caused by large-amplitude cycles compared to the small-amplitude
cycles. Therefore, the number of cycles reduces with increase in the value of c.

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1441–1457
CYCLIC-DEMAND SPECTRUM 1447

Figure 6. Cyclic-demand spectra for the ground motion shown in Figure 1(a) for 10% of critical damping
and three values of damage exponent c.

Figure 7. Cyclic-demand spectra for the motions shown in Figure 1(a) (North) and Figure 1(b) (East)
for 10% of critical damping and damage exponent c = 2.

Eect of shape of ground motion histories

Figure 7 shows a comparison between the cyclic-demand spectra of ground motions shown
in Figure 1(a) (North direction) and Figure 1(b) (East direction). Except for very short-
periods, the cyclic demand in the East direction is more than that in the North direction.
This is because the velocity and displacement histories in the North direction have a distinct
pulse, which is signicantly higher than the remaining history. As a result, although the load

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1441–1457
1448 P. K. MALHOTRA

amplitudes are higher in the North direction (Figure 2), the cyclic demand is lower in that
direction. Pulse-like ground motions, therefore, impose a large amplitude demand [22], but
for a fewer number of cycles.

CYCLIC-DEMAND PARAMETERS

We will now summarize the information provided by the cyclic-demand spectrum by a few
key parameters. By recognizing that structural systems fall into three broad categories—sti,
moderately sti, and exible—and the response of these systems is controlled by the acceler-
ation, velocity and displacement histories, respectively, we can characterize the cyclic demand
of a ground motion history by these parameters:
 
1  2n xgi 2
NA = · (7)
2 i=1 PGA
 
1 2n ẋgi 2
NV = · (8)
2 i=1 PGV

1 2n  x 2
gi
ND = · (9)
2 i=1 PGD
where, xgi , ẋgi and xgi are the amplitudes of ith peaks in the acceleration, velocity and dis-
placement histories; PGA, PGV and PGD are the peak values of ground acceleration, velocity
and displacement. Note that the NA , NV and ND values depend on the shape rather than the
amplitude of the acceleration, velocity and displacement histories.
For very sti systems, the number of cycles in the cyclic-demand spectrum approaches
NA , whereas for very exible systems the number of cycles approaches ND (Figure 5). For
moderately-sti systems, the number of cycles is proportional to NV . The parameter NA is
therefore useful for sti systems, the parameter NV for moderately sti systems, and the
parameter ND for exible systems. The ground motions with similar NA , NV and ND values
are expected to have similar cyclic-demand spectra. Table I lists the NA , NV and ND parameters
for 36 rock motions and Table II lists these parameters for 35 soil motions. These are arranged
in ascending order of the parameter NV .
In nearly all cases, NA ¿NV ¿ND , because of progressively reduced contribution of high-
frequencies from the acceleration to velocity to displacement history. The eect of earthquake
magnitude, distance and soil types on the NA , NV and ND values is outside the scope of this
study. Ground motions with a dominant pulse in the velocity history have a smaller NV value.

SMOOTH CYCLIC-DEMAND SPECTRUM

The site-specic values of NA , NV and ND parameters could be obtained from their respective
‘attenuation’ relationships (yet to be developed). It is not yet certain if NA , NV and ND will
reduce or increase with increase in the distance from the seismic source. The objective here
is to construct a cyclic-demand spectrum from known values of NA , NV and ND parameters

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1441–1457
CYCLIC-DEMAND SPECTRUM 1449

Table I. Cyclic-demand parameters for various strong-motion records on rock sites.


Ground motion record∗ NA NV ND

Name and direction Earthquake†


Cape Mendocino 0◦ Petrolia 1.43 0.63 0.84
Yerba Buena Island 90◦ Loma Prieta 2.67 0.83 1.15
Pacoima Dam Downstream 175◦ Northridge 2.03 0.84 1.82
Gilroy #1 Gavilan College 0◦ Loma Prieta 3.58 1.16 1.10
Petrolia 90◦ Petrolia 3.25 1.21 0.78
Pacoima Dam Downstream 265◦ Northridge 1.74 1.45 1.52
Kobe University 90◦ Kobe 3.11 1.54 1.84
Kobe University 0◦ Kobe 3.14 1.61 1.46
Pacoima Kagel Canyon 0◦ Northridge 2.77 1.61 3.12
Cape Mendocino 90◦ Petrolia 1.88 1.75 1.51
Agnew State Hospital 0◦ Loma Prieta 5.98 1.83 1.92
Mt. Wilson Seismic Station 0◦ Sierra Madre 3.96 1.88 1.32
LA Temple & Hope 180◦ Northridge 7.46 1.89 2.28
Castaic Old Ridge Route 90◦ Northridge 3.23 2.00 2.11
Gilroy #1 Gavilan College 90◦ Loma Prieta 3.83 2.00 1.09
Yerba Buena Island 0◦ Loma Prieta 7.51 2.19 3.20
Big Bear Lake 0◦ Big Bear 5.71 2.37 1.21
Petrolia 0◦ Petrolia 4.32 2.70 1.18
Castaic Old Ridge Route 0◦ Northridge 3.23 2.72 1.08
San Gabriel 180◦ Whittier 3.38 2.85 1.69
San Gabriel 270◦ Whittier 5.64 2.95 2.63
Cantua Creek School 270◦ Coalinga 4.22 2.99 2.62
LA Temple & Hope 90◦ Northridge 11.5 3.25 3.62
Cantua Creek School 0 Coalinga 4.00 3.61 1.79
Joshua Tree 90◦ Landers 8.46 3.61 1.93
Mt. Wilson Seismic Station 90◦ Sierra Madre 6.96 3.67 1.37
Big Bear Lake 270◦ Big Bear 9.37 3.87 1.01
LA Baldwin Hills 0◦ Northridge 10.7 3.90 3.22
Pacoima Kagel Canyon 90◦ Northridge 4.77 4.18 2.06
Malibu Pt. Dume 0◦ Northridge 8.30 4.22 3.42
Agnew State Hospital 90◦ Loma Prieta 4.96 4.37 1.59
Malibu Pt. Dume 90◦ Northridge 6.22 5.42 3.76
Joshua Tree 0◦ Landers 5.81 5.84 2.92
LA City Terrace 180◦ Northridge 6.77 5.85 3.43
LA City Terrace 90◦ Northridge 8.08 5.90 3.57
LA Baldwin Hills 90◦ Northridge 5.85 5.97 2.91
∗ Records may be downloaded from these web-sites: COSMOS Virtual Data Center (http://db.cosmos-eq.org/),
PEER Strong Motion Database (http://peer.berkeley.edu/smcat/), UCSB Strong Motion Database (http:
//smdb.crustal.ucsb.edu/), or CDMG/CSMIP (http://www.consrv. ca.gov/dmg/).
† 1989 Loma Prieta, CA 7M ; 1992 Petrolia, CA 7M ; 1992 Landers, CA 7:3M ; 1992 Big Bear, CA 6:5M ;
w w w w
1994 Northridge, CA 6:7Mw ; 1995 Kobe, Japan, 6:9Mw ; 1983 Coalinga, CA 6:5Mw ; 1987 Whittier, CA 6:1Mw ;
1991 Sierra Madre, CA 5:8Mw ; 1986 Palm Springs, CA 6:2Mw .

in the same manner as the amplitude spectrum is constructed from known values of PGA,
PGV and PGD.
The fact that for very sti systems Ncy = NA and for very exible systems Ncy = ND , leads to
the belief that the relationship between the number of cycles and the NA , NV and ND parameters

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1441–1457
1450 P. K. MALHOTRA

Table II. Cyclic-demand parameters for various strong-motion records on soil sites.
Ground motion record∗ NA NV ND

Name and direction Earthquake†


Sylmar County Hospital 0◦ Northridge 2.12 0.76 0.96
Rinaldi LADWP Receiving Station 228◦ Northridge 3.00 0.83 1.15
El Centro #6 230◦ Imperial Valley 3.74 0.97 0.91
El Centro #8 230◦ Imperial Valley 4.45 1.12 0.79
Treasure Island 90◦ Loma Prieta 2.18 1.14 1.32
Jensen Filter Plant 22◦ Northridge 2.44 1.30 0.88
El Centro #6 140◦ Imperial Valley 5.14 1.34 1.06
Sylmar County Hospital 90◦ Northridge 2.18 1.37 1.91
Gilroy #3 0◦ Loma Prieta 4.95 1.46 0.94
Newhall LA County Fire Station 0◦ Northridge 4.42 1.53 1.03
Newhall LA County Fire Station 90◦ Northridge 4.78 1.55 1.11
El Centro #8 140◦ Imperial Valley 2.10 1.67 0.89
Yermo Fire Station 270◦ Landers 4.35 1.69 0.97
Los Angeles University Hospital 5◦ Northridge 2.74 1.78 2.87
Kobe Takatori 90◦ Kobe 5.99 1.79 1.78
El Centro #1 140◦ Imperial Valley 10.2 1.85 1.94
Rinaldi LADWP Receiving Station 318◦ Northridge 7.49 1.86 1.70
Hollister South Street and Pine 0◦ Loma Prieta 3.24 1.91 1.41
Whitewater Trout Farm 180◦ Palm Springs 4.05 1.92 0.94
Kobe Takatori 0◦ Kobe 4.53 2.01 1.89
Arleta Nordho Ave. Fire Station 90◦ Northridge 5.18 2.03 2.33
Gilroy #2 0◦ Loma Prieta 3.53 2.03 1.48
Jensen Filter Plant 292◦ Northridge 2.71 2.03 1.74
Gilroy #2 90◦ Loma Prieta 3.79 2.11 1.13
Yermo Fire Station 0◦ Landers 11.6 2.12 1.19
Treasure Island 0◦ Loma Prieta 2.93 2.15 2.79
Gilroy #3 90◦ Loma Prieta 3.79 2.22 1.77
Tarzana Cedar Hill Nursery 90◦ Northridge 4.63 2.25 0.77
Whitewater Trout Farm 270◦ Palm Springs 3.79 2.29 0.90
El Centro #1 230◦ Imperial Valley 10.3 2.89 1.63
El Centro 0◦ El Centro 6.59 3.52 3.00
Hollister South Street and Pine 90◦ Loma Prieta 6.09 3.80 2.26
Arleta Nordho Ave. Fire Station 0◦ Northridge 5.49 3.81 2.45
Tarzana Cedar Hill Nursery 0◦ Northridge 8.76 4.17 1.37
Los Angeles University Hospital 95◦ Northridge 9.29 5.08 2.82
∗ Records may be downloaded from these web-sites: COSMOS Virtual Data Center (http://db.cosmos-eq.org/),
PEER Strong Motion Database (http://peer.berkeley.edu/smcat/), UCSB Strong Motion Database (http:
//smdb.crustal.ucsb.edu/), or CDMG/CSMIP (http://www.consrv.ca.gov/dmg/).
† 1989 Loma Prieta, CA 7M ; 1992 Petrolia, CA 7M ; 1992 Landers, CA 7:3M ; 1992 Big Bear, CA 6:5M ;
w w w w
1994 Northridge, CA 6:7Mw ; 1995 Kobe, Japan, 6:9Mw ; 1983 Coalinga, CA 6:5Mw ; 1987 Whittier, CA 6:1Mw ;
1991 Sierra Madre, CA 5:8Mw ; 1986 Palm Springs, CA 6:2Mw ; 1940 El Centro, CA 6:9Mw .

depends primarily on the stiness of the system. A measure of stiness of a system is the
ratio between its natural period and the ‘central’ period of the ground motion:

PGD
Tg = 2 · (10)
PGA

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1441–1457
CYCLIC-DEMAND SPECTRUM 1451

Figure 8. Cyclic-demand spectra of 71 (36 rock + 35 soil) ground motions for 10% of critical
damping and c = 2. The average spectrum is shown by thick dashed line.

Figure 9. Smooth cyclic-demand spectrum for 10% of critical damping. TV and TD are determined
from Equations (11) and (12).

A system with period T ¡Tg may be considered sti and that with period T ¿Tg may be
considered exible [12]. The lower the T=Tg ratio, the stier the system. The cyclic-demand
spectra of various ground motions were normalized along the horizontal axis with respect to
period Tg to align the sti and exible regions of various spectra. Figure 8 shows the plots of
normalized cyclic-demand spectra of 71 ground motions (36 rock +35 soil) for 10% of critical
damping and c = 2. The average of 71 cyclic-demand spectra is shown by thick dashed line
in this gure.
The average spectrum is redrawn in Figure 9 by a dashed line. Also shown in this gure
are the average of 71 NA and ND values, and the control-periods TV and TD computed from

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1441–1457
1452 P. K. MALHOTRA

Figure 10. Ratios between the actual and the smooth cyclic-demand spectra of 71 (36 rock + 35 soil)
motions for 10% of critical damping and c = 2. The average of all ratios is shown by thick white line.

the average of 71 PGA, PGV and PGD values:


PGV
TV = 2 · (11)
PGA
PGD
TD = 2 · (12)
PGV
A smooth spectrum consisting of straight-line segments is obtained by least-square tting.
From known values of NA , NV and ND , the smooth spectrum may be constructed by using
the control periods (0:035TV , 0:12TV , 0:3TV , TV , TD and 6:2TD ) and the amplication factors
(1.2, 1.5 and 1.6) applied to NA , NV and ND values.
To conrm that the smooth spectral shape (Figure 9) is a reasonable representation of the
actual spectra, the ratios between the actual and the smooth spectra were computed for 71
ground motions. Shown in Figure 10, these ratios were found to range between 0.37 and 3.1.
The average of these ratios, shown by a thick white line, is close to unity throughout the
period range. Therefore, the smooth spectrum represents the actual spectrum in an ‘average
sense’, although at certain periods it may signicantly deviate from the actual spectrum.
The smooth spectra were also obtained for other values of damping. Table III lists the
amplication factors for 2, 5, 10 and 20% of critical damping. The eect of damping on the
control periods was found to be small. The spectral shape was simplied to conservatively
represent all values of damping. The simplied shape is shown in Figure 11.

EXAMPLE

Let us consider the design of a hypothetical single-storey building with a roof weighing 1 MN.
The site-specic ground motion parameters that have 10% chance of being exceeded in 50
years (475-year return period) are listed in Table IV. The performance objective for the 475-
year return period event is to limit the horizontal deformation of the support columns to 10 cm.

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1441–1457
CYCLIC-DEMAND SPECTRUM 1453

Table III. Amplication factors for dierent values of damping used in constructing smooth cyclic-
demand spectrum from known values of NA , NV and ND (Figure 11).
Damping A V D
(% of critical)
2 2.1 2.7 3.4
5 1.5 1.8 2.1
10 1.2 1.5 1.6
20 1.1 1.3 1.4

Figure 11. Construction of smooth cyclic-demand spectrum from known values of NA , NV


and ND . Amplication factors A , V and D are listed in Table III. TV and TD are given by
Equations (11) and (12).

Table IV. Site-specic amplitude and cyclic-demand parameters that have 10% chance of being exceeded
in 50 years (475-year return period).
Amplitude parameters Cyclic-demand parameters
PGA PGV PGD NA NV ND
0:4 g 40 cm= s 15 cm 4 2.5 2

Using the PGA, PGV and PGD values and the Newmark–Hall amplication factors A , V
and D [23], we construct the smooth amplitude spectra shown in Figure 12(a) for 5, 10 and
20% of critical damping. Now, using NA , NV and ND values and the amplication factors A ,
V and D (Table III), we construct the smooth cyclic-demand spectra shown in Figure 12(b)
for 5, 10 and 20% of critical damping.
The force–deformation relationship for support columns for the rst four cycles of ±10 cm
horizontal deformation is shown in Figure 13. Note a signicant reduction in the size of
the hysteresis loop from 1st to 4th cycle. The ‘equivalent’ damping values, evaluated from
the size of the hysteresis loop, are listed in the 2nd column of Table V. In Figure 13,

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1441–1457
1454 P. K. MALHOTRA

Figure 12. Site-specic design spectra for 5, 10 and 20% of critical damping: (a) Amplitude spectra;
and (b) cyclic-demand spectra.

Figure 13. Cyclic force–deformation relationship for support columns for rst 4 cycles
of ±10 cm horizontal deformation.

note also a reduction in the maximum horizontal force from 1st to 4th cycle. The force
amplitude (smaller of the positive and negative peaks) for dierent cycles are listed in the
3rd column of Table V.

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1441–1457
CYCLIC-DEMAND SPECTRUM 1455

Table V. Hysteretic damping and force amplitude for the support columns at ±10 cm horizontal
deformation in various cycles (see Figure 13).
Cycle Hysteretic damping Force amplitude
(% of critical) (kN)
1st 16 38
2nd 12 36
3rd 10 34
4th 8 34

We need to determine the number of columns required to support the roof of this building
so that the building meets the desired performance objective.
As a rst approximation, we assume that the ‘eective’ damping of the system is 10% of
critical. For 10 cm horizontal deformation (performance objective), the spectral acceleration
from the 10% damped spectrum is 0.3g (Figure 12(a)). Therefore, the building must possess
an ultimate base-shear capacity of 0:3W = 300 kN for yet unknown number of cycles. From
Figure 12(a), the ‘eective’ period of the building will be 1:15 s. Now, from Figure 12(b),
the number of cycles for 1:15 s period and 10% of critical damping is Ncy = 3:5. Therefore,
to meet the performance objective, this building must have an ultimate base-shear capacity of
300 kN and exhibit 10% damping for 3.5 cycles of deformation ±10 cm. In the 3.5th cycle,
the damping is 9% of critical and the horizontal strength of each column is 34 kN (Table V).
Although, the damping in the 3.5th cycle is less than the assumed value of 10%, the average
damping during the rst three-and-half cycles is (16 + 12 + 10 + 8=2)=3:5 = 12% of critical.
Therefore, no further iteration is necessary. The required number of columns is 300=34 = 9.
In the above analysis, no consideration has been given to smaller-than-design seismic events,
which may signicantly increase the number of load cycles applied to the structure. Also, in
arriving at the number of cycles, it was assumed that the damage exponent is c = 2 for the
column material. The analysis is not a substitute for non-linear response history analysis, but
is considered adequate for design of simple systems or preliminary design of complex systems.
The site-specic amplitude and cyclic-demand parameters listed in Table IV can be used to
select strong-motion records for nonlinear response history analysis.

CONCLUSIONS

1. The cyclic-demand spectrum proposed in this paper complements the amplitude spectrum
to provide a more complete denition of the seismic load. To withstand a ground shaking
with desired performance objectives, a structure must resist the load (specied by the
amplitude spectrum) for a certain number of cycles (specied by the cyclic-demand
spectrum).
2. The cyclic-demand reduces with increase in system damping. The eect of damping is
most pronounced for systems which are neither too sti nor too exible.
3. The cyclic-demand parameters (NA , NV and ND ) proposed in this paper are analogous to
the amplitude parameters (PGA, PGV and PGD). NA (similar to PGA) is useful for sti
systems, NV (similar to PGV) is useful for moderately sti systems, and ND (similar to
PGD) is useful for exible systems.

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1441–1457
1456 P. K. MALHOTRA

4. NA , NV and ND provide a more direct measure of the cyclic demand than any denition
of strong-motion duration.
5. NA , NV and ND are not aected by the threshold acceleration. Bracketed duration, for
example, can signicantly change if the threshold is raised from 0.03 to 0.05g.
6. NA , NV and ND are not aected by the lengths of ‘silent’ gaps in the ground motion
histories, caused by time dierence between main and sub-events.
7. Site-specic cyclic-demand parameters can be used with site-specic amplitude parame-
ters to select strong-motion records for non-linear response history analysis.
8. Pulse-like ground motions may subject structures to large load amplitudes but generally
the associated number of cycles is low. Both the amplitude and the number of cycles
should be considered in assessing the damage potential of ground motions.

APPENDIX A

Nomenclature

A , V , D Cyclic-demand amplication factors (Figure 11)


c Damage exponent (Equation (2))
C Coecient in low-cycle fatigue model (Equation (2))
D Measure of fatigue damage (D = 1 implies complete damage)
n Number of half cycles in acceleration, velocity, displacement
or deformation history
NA Cyclic-demand parameter for sti systems (Equation (7))
Ncy Equivalent number of uniform amplitude cycles (Equation (6))
ND Cyclic-demand parameter for exible systems (Equation (9))
Nf Number of cycles to failure (Equation (2))
NV Cyclic-demand parameter for moderately-sti systems (Equation (8))
PGA Peak ground acceleration
PGD Peak ground displacement
PGV Peak ground velocity
SA Spectral acceleration
SD Spectral deformation
SV Spectral velocity
T Natural period
TD Control period in smooth spectrum (Equation (12))
Tg ‘Central’ period of ground motion (Equation (10))
TV Control period in smooth spectrum (Equation (11))
u Total deformation (elastic + plastic)
ui Deformation amplitude of ith half-cycle
umax Maximum deformation (Figure 3)
up Plastic deformation
W Weight of the structure
xg Ground acceleration
ẋg Ground velocity
xg Ground displacement

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1441–1457
CYCLIC-DEMAND SPECTRUM 1457

ACKNOWLEDGEMENT

Two anonymous reviewers provided valuable comments, which improved the paper.

REFERENCES

1. Housner GW. Intensity of ground shaking near the causative fault. Proceedings of the 3rd World Conference
on Earthquake Engineering, vol. 1, Auckland, New Zealand, 1965; 81– 94.
2. Housner GW. Measures of severity of ground shaking. Proceedings of the U.S. National Conference on
Earthquake Engineering, Ann Arbor, Michigan, 1975; 25 – 33.
3. Ambraseys NN, Sarma SK. Response of earth dams to strong earthquakes. Geotechnique, 1967; 17(2):181–283.
4. Husid LR. Charactersticas de terremotos. Analisis general. Revista del IDIEM vol. 8, Santiago de Chile, 1969;
21– 42.
5. Bolt BA. Duration of strong ground motions. Proceedings of the 5th World Conference on Earthquake
Engineering, vol. 1, Rome, 1973; 1304 –1313.
6. Trifunac MD, Brady AG. A study of the duration of strong earthquake ground motion. Bulletin of Seismological
Society of America, 1975; 65(3):581– 626.
7. Kawashima K, Aizawa K. Bracketed and normalized durations of earthquake ground acceleration. Earthquake
Engineering and Structural Dynamics, 1989; 18:1041–1051.
8. Kramer SL. Geotechnical earthquake engineering, Prentice-Hall, Upper Saddle River, NJ, 1996.
9. Bommer JJ, Martinez-Pereira A. The eective duration of earthquake strong motion. Journal of Earthquake
Engineering, 1999; 3(2):127–172.
10. California Strong Motion Instrumentation Program. California Division of Mines and Geology (CDMG) web-
site: http://www.consrv.ca.gov/dmg/.
11. Chopra AK. Dynamics of structures: theory and applications to earthquake engineering. Prentice-Hall,
Englewood Clis, NJ, 1995.
12. Malhotra PK. Response spectrum of incompatible acceleration, velocity and displacement histories. Earthquake
Engineering and Structural Dynamics, 2001; 30(2):279 –286.
13. Shibata A, Sozen MA. Substitute structure method for seismic design in R=C. Journal of the Structural Division,
ASCE, 1976; 102(ST1):1–18.
14. Priestly MJN, Siebel F, Calvi GM. Seismic Design and Retrot of Bridges. Wiley: New York, NY, 1996.
15. Jennings PC. Equivalent viscous damping for yielding structures. Journal of Engineering Mechanics Division,
ASCE, 1968; 94(EM1):103 –116.
16. Iwan WD, Gates NC. The eective period and damping of a class of hysteretic structures. Earthquake
Engineering and Structural Dynamics, 1979; 7(3):199 –212.
17. Con LF, Jr., A study of the eect of cyclic thermal stresses in ductile metals. Transactions of ASME, 1954;
76:931– 950.
18. Manson SS. Behavior of materials under conditions of thermal stress. NACA TN 2933, 1954.
19. Boyer HE. Atlas of Fatigue Curves. American Society of Metals: OH, 1999.
20. Krawinkler H. Zohrei M. Lashkari-Irvani B. Coe NG. Hadidi-Tamjed H. Recommendation for experimental
studies on the seismic behavior of steel components and materials. Report No. NSF=CEE-83220, Stanford
University, Stanford, CA, September, 1983.
21. Miner MA. Cumulative damage in fatigue. Journal of Applied Mechanics, 12:1945.
22. Malhotra PK. Response of buildings to near-eld pulse-like ground motions. Earthquake Engineering and
Structural Dynamics, 1999; 28(11):1309 –1326.
23. Newmark NM, Hall WJ. Earthquake Spectra and Design, Earthquake Engineering Research Institute, Oakland,
CA, 1982; 35 – 36.

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1441–1457

View publication stats

You might also like