You are on page 1of 12

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/223109492

Influence of stiffness degradation on strength


demands of structures built on soft soil site

ARTICLE in ENGINEERING STRUCTURES · OCTOBER 2002


Impact Factor: 1.84 · DOI: 10.1016/S0141-0296(02)00052-4

CITATIONS READS

34 251

2 AUTHORS:

Eduardo Miranda Jorge Ruiz-Garcia


Stanford University Universidad Michoacana de San Nicolás de …
111 PUBLICATIONS 2,329 CITATIONS 45 PUBLICATIONS 662 CITATIONS

SEE PROFILE SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Jorge Ruiz-Garcia
letting you access and read them immediately. Retrieved on: 04 February 2016
Engineering Structures 24 (2002) 1271–1281
www.elsevier.com/locate/engstruct

Influence of stiffness degradation on strength demands of structures


built on soft soil sites
Eduardo Miranda a,∗, Jorge Ruiz-Garcia b
a
Department of Civil and Environ. Engineering, Stanford University, Stanford, CA 94305-4020, USA
b
Department of Civil and Environ. Engineering, Stanford University, Stanford, CA 94305-4020, USA

Received 1 August 2001; received in revised form 19 March 2002; accepted 1 April 2002

Abstract

The effect of stiffness degradation on the lateral strength demands of inelastic single-degree-of-freedom systems subjected to
soft soil records is investigated. The modified-Clough model is used to represent structures that exhibit significant stiffness degra-
dation when subjected to reverse cyclic loading and the elastic-perfectly-plastic model is used to represent non-degrading structures.
The study is based on two sets of ground motions. The first set includes 100 accelerograms recorded in various recent earthquakes
on soft soil sites of Mexico City. The second set includes 16 ground motions recorded on bay mud sites in the San Francisco Bay
Area during the 1989 Loma Prieta earthquake. A special emphasis is placed on the effect of stiffness degradation on strength
reduction factors that permit the estimation of inelastic strength demands from elastic strength demands. Mean ratios of lateral
strength demands of stiffness degrading systems to lateral strength demands of non-degrading systems for both sets of ground
motions are presented. It is concluded that structures with stiffness degradation, and with periods of vibration shorter than the
predominant period of vibration of the ground motion, can experience lateral strength demands larger than those of non-degrading
structures in the same period range. Whereas for structures with periods of vibration equal or longer than the predominant period
of the ground motion, the lateral strength demands are typically smaller for stiffness-degrading structures than those of non-degrading
structures.  2002 Published by Elsevier Science Ltd.

Keywords: Lateral strength demand; Stiffness degradation; Soft soil sites; Strength reduction factor; Ductility demand

1. Introduction records obtained on firm soils has been the topic of sev-
eral investigations. Clough and Johnston [1,2] conducted
Present seismic design codes allow structures to one of the first studies of this type which investigated
undergo inelastic deformations in the event of strong the effects of stiffness degradation on single-degree-of-
earthquake ground motions. In current design procedures freedom (SDOF) systems subjected to four earthquake
inelastic behavior is accounted for through the use of ground motions. This study concluded that ductility
strength reductions factors that allow structures to be demands in the degrading systems were not significantly
designed for lateral forces that are smaller than those different from those observed in ordinary elasto-plastic
required by the structure to remain elastic during severe structures, except for structures having a period of
earthquakes. Reinforced concrete structures and other vibration less than 0.5 s, and recommended further
types of structures, even when they have very good investigations on the earthquake resistance of degrading
detailing, when subjected to reversed cyclic loading stiffness structures in the short period range.
exhibit a gradual loss of lateral stiffness usually referred Other early studies reached similar conclusions [3–8].
to as stiffness degradation. The effect of stiffness degra- In particular, the study by Riddell and Newmark [6] con-
dation on the seismic response of structures subjected to cluded that ordinates of the average spectra did not vary
significantly when various nonlinear models were used,
and that the use of the elastoplastic idealization pro-

Corresponding author. Tel.: +1-650-723-4450; fax: +1-650-725- vided, in almost every case, a conservative estimate of
6014. the average response to a number of earthquake motions.
E-mail address: miranda@ce.stanford.edu (E. Miranda). With the exception of the study by Al-Sulaimani, [8]

0141-0296/02/$ - see front matter  2002 Published by Elsevier Science Ltd.


PII: S 0 1 4 1 - 0 2 9 6 ( 0 2 ) 0 0 0 5 2 - 4
1272 E. Miranda, J. Ruiz-Garcia / Engineering Structures 24 (2002) 1271–1281

these early studies did not provide a quantitative measure Michoacan earthquake and only considered one acceler-
of the effect of stiffness degradation on the seismic ogram, namely the EW component of the ground motion
response of SDOF systems. recorded at the SCT site in Mexico City. Even though
Some recent studies have provided quantitative meas- both studies considered similar hysteretic models and the
ures of the effect of stiffness degradation on the seismic same earthquake ground motion record, the conclusions
response of SDOF systems. For example, Krawinkler from these studies differ considerably. The study by Tar-
and his co-workers [9,10] have presented average ratios quis [18] concluded that for earthquakes with narrow
of strength reduction factors of stiffness degrading sys- band elastic spectra, as those typically recorded on soft
tems to elastoplastic systems when subjected to a set of soil deposits, strength requirements for a given allowable
15 earthquake ground motions recorded on firm soils in ductility are strongly dependent on the force deformation
California. These studies show that strength reduction characteristics of the system. On the other hand, the
factors of stiffness-degrading structures are smaller than study by Meli and Ávila [19] concluded that strength
those of non-degrading structures for periods of demands for stiffness degrading systems show little dif-
vibration smaller than about 0.4 s. For longer periods, ference from those of the elastic perfectly-plastic sys-
the strength reductions factors of stiffness-degrading tems.
structures are, on average, up to 20% larger than those The objective of this study is to present the results of
of non-degrading structures, which means that in this an investigation whose main goal was to provide more
spectral region lateral strength demands of non-degrad- information on the effects of stiffness degradation in
ing structures are larger than those with stiffness degra- structures built on very soft soils when subjected to
dation. In the short period range, the larger the ductility earthquake ground motions. In particular this study tried
ratio the larger the difference between the strength to: (1) use a relatively large number of ground motions
reduction factors of stiffness-degrading and non-degrad- in order to identify the most important trends; (2) assess
ing structures. However, no clear trend with change in the spectral regions (as a function of the predominant
the level of ductility ratio could be observed in the period of the ground motion) where stiffness degradation
medium- and long-period range. Other recent studies of leads to larger lateral strength demands and hence cannot
the effect of stiffness degradation on the response of be neglected; (3) quantify the differences in lateral
SDOF systems include the studies by Lee at al. [11] and strength demands in stiffness-degrading systems with
that of Borzi and Elnashai [12]. The effect of stiffness respect to those in non-degrading systems; and (4) evalu-
degradation on the response of multi-degree-of-freedom ate the effects of stiffness degradation as a function of
structures has been the topic of some recent investi- the level of inelastic deformation and as a function of
gations [13,14]. The study by Gupta and Krawinkler [13] the ratio of the period of vibration to the predominant
found that MDOF structures with stiffness degradation period of vibration of the ground motion.
have a global (roof) drift similar to that observed of
SDOF systems with period equal to that of the funda-
mental mode of the MDOF structure. The study by 2. Load-deformation hysteretic models
Foutch and Shi [14] concluded that the hysteresis type
has only a minimum effect on ductility demands, and Most studies that have investigated the inelastic
that the effects of hysteresis type in the MDOF systems response of structures when subjected to earthquake
they investigated were similar to those of SDOF systems ground motions have used the elastic-perfectly plastic
reported by previous investigators. (EPP) model. This model, which is shown in Fig. 1a, is
All of the previously mentioned studies have only used in this study to represent structures with no stiffness
considered ground motions recorded either on rock or degradation. The EPP load-deformation hysteretic model
on firm soil. Recent studies [10,15–17] have concluded is characterized by having a constant loading stiffness
that seismic demands on structures built on very soft soil up to yielding which occurs at a yielding strength, Fy,
can be significantly different from those of structures on which is assumed to be equal in both loading directions.
rock or on firm sites. In particular, these studies have During yielding, the system has no lateral stiffness and
concluded that seismic demands of structures built on the unloading stiffness is equal to the loading stiffness.
soft soil are strongly dependent on the ratio of the period Only two parameters are needed to characterize this
of vibration of the structure to the predominant period model, the initial stiffness Ke and the yielding
of the ground motion. However, these studies did not strength, Fy.
consider the effect of stiffness degradation in combi- Many hysteretic models have been proposed to rep-
nation with ground motions recorded in soft soils. resent the load-deformation characteristics of reinforced
There have been only two studies that have investi- concrete structures when subjected to reversed cyclic
gated the effects of stiffness degradation considering loading. One of the first models to include the effect of
very soft soil records [18,19]. Both of these studies were stiffness degradation was the one proposed by Clough
conducted as a result of the September 19th, 1985 and Johnston [2], which is shown in Fig. 1b. This model
E. Miranda, J. Ruiz-Garcia / Engineering Structures 24 (2002) 1271–1281 1273

Fig. 1. Load-deformation hysteretic models used in this study: (a) Elastic-perfectly plastic; (b) Clough; (c) Modified-Clough.

has an elasto-plastic-perfectly-plastic envelope, however stresses, slippage of steel bars or other phenomena.
it differs from the EPP model in that, after the initial Thus, results from this study should not be used for
yielding, further loading branches are directed towards structures where, besides stiffness degradation, signifi-
the furthest unloading point in the direction of loading, cant pinching and/or strength degradation can occur.
thus with a lateral stiffness smaller than the initial stiff-
ness. The degree of degradation depends on the
maximum displacement in prior inelastic excursions. 3. Strength reduction factors
Thus, the larger the maximum inelastic displacement, the
smaller the loading stiffness. Similarly to the EPP model, The strength-reduction factor (i.e. reduction in
the unloading stiffness in the Clough model is always strength demand due to nonlinear hysteretic behavior)
kept equal to the initial elastic stiffness. Rµ is defined as [17] the elastic strength demand divided
Based on experimental results of a number of by the inelastic strength demand
medium-sized reinforced concrete members subjected to
reversed cyclic loading, Takeda, Sozen and Nielsen [20] Fy(m ⫽ 1)
Rm ⫽ (1)
proposed a hysteretic model to reproduce analytically the Fy(m ⫽ mi)
behavior of reinforced concrete structures. A trilineal
envelope characterizes this model, now commonly used where Fy (m=1) is the lateral yielding strength required
and known as the Takeda model. It also differs from to maintain the system elastic and Fy (m=mi) is the lateral
the Clough model by considering a degrading unloading yielding strength required to maintain the displacement
stiffness whose amplitude depends on the maximum dis- ductility demand, m, less or equal to a predetermined
placement demand in the direction of loading. Many target ductility, mi. Eq. (1) can also be rewritten as
other models have been proposed for stiffness degrading Cy(m ⫽ 1)
structures [3–8,21]. A good review of some of these earl- Rm ⫽ (2)
Cy(m ⫽ mi)
ier hysteretic models is provided in Otani, [7].
In order to represent structures with stiffness degra- where Cy (m=1) is the seismic coefficient (lateral yield-
dation the modified-Clough model is used in this study. ing strength divided by the weight of the structure)
This model is based on the Clough model, however, it required to avoid yielding; and Cy (m=mi) is the minimum
incorporates a modification that was pointed out by seismic coefficient required to control the displacement
Mahin and Bertero [22] and later on implemented in a ductility demand, m, to a target ductility demand, mi. For
computer program by Mahin and Lin [23]. As shown in earthquake-resistant design, Rµ corresponds to the
Fig. 1b, in the original Clough model, after unloading maximum reduction in lateral strength that can be used
from point A and reloading from point B, loading would in order to limit the displacement ductility demand to a
take place towards point C, which is not realistic. In the predetermined ductility mi in a structure that will have a
modified-Clough model reloading from point B is lateral strength equal to the design strength. For each
towards the unloading point A, and after reaching this earthquake record, the inelastic strength demands were
point further loading is directed towards point C. computed for a family of 50 SDOF systems undergoing
Several studies have concluded that the modified- six levels of inelastic deformation (µ equal to 1.5, 2–6).
Clough model is capable of reproducing the behavior of The inelastic strength demand Fy (m=mi) was computed
properly designed reinforced concrete structures where by iteration on the lateral yielding strength of the system
shear failure is avoided and the behavior is primarily for a specific period of vibration and specific values of
flexural. However, it is not capable of capturing neither m until the ductility displacement demand was, within a
pinching nor strength degradation caused by high shear tolerance of 1%, equal to the specified ductility ratio.
1274 E. Miranda, J. Ruiz-Garcia / Engineering Structures 24 (2002) 1271–1281

Thus, for a given ground acceleration time history, a Rµ The soils corresponding to this zone are in general hardly
spectrum can be constructed by plotting the strength compressible and exhibit a relatively high shearing
reduction factors (computed with Eq. (1)) of a family strength. The transition zone lies between the hill and
of SDOF systems (with different periods of vibrations) lake bed zone and is characterized by the variability of
undergoing different levels of inelastic deformation mi the soil profile. Sequences of hard soils, sands, silty
when subjected to the same ground motion. Of particular sands and soft clays are typical soil stratigraphies of this
relevance to this study is to investigate whether the area. Damage surveys [24] showed that most of the
strength reduction factors of stiffness-degrading systems structures that collapsed or suffered major damage dur-
are smaller, equal or larger than those of elastoplastic ing the 1985 Michoacan earthquake were located within
systems. the zone bounded by the dotted line in Fig. 2. Table 1
lists the first set of records considered in this study. For
each record earthquake magnitude and peak ground
4. Earthquake ground motions used in this study acceleration are given.
The second set includes 16 ground motions recorded
A total of 116 earthquake acceleration time histories on bay mud sites in the San Francisco Bay Area during
were considered in this investigation. The first set of the 1989 Loma Prieta earthquake. San Francisco Bay is
earthquake ground motions consists of 100 acceleration located in a basin about 15 km wide bounded by the
time histories recorded in the soft soil zone of Mexico active San Andreas and Hayward fault zones [25]. This
City in five recent earthquakes. Fig. 2 shows a map of region is characterized by a wide variety of geologic
the current seismic microzonation of Mexico City. Three deposits, from rock sites in the hill area, to estuarine
zones can be observed, the soft zone (“lake bed” zone), mud and clay deposits in the flatlands along the margins
the transition zone and the firm zone (hill zone). Mexico of the bay. The bay mud area is comprised of unconsoli-
City is located partly on an old lake bed that was formed dated, water-saturated, dark plastic clay and silty clay
by the Texcoco, the Chalco and the Xochimilco lakes. with well-sorted silt and sand dunes in some areas. It
Relatively thick deposits of lacustrine clay form the soft may contain more than 50% of water content and low
zone. In the lake bed zone the depth of these soft clay shear-wave velocities in the range of 67 to 116 m/s.
deposits varies from 10 m (33 ft) to 60 m (197 ft). These Table 2 summarizes the second set of records used in
clay deposits are very deformable and are characterized this study.
by very high water contents that reach more than 400%, Five percent damped absolute acceleration response
shear wave velocities as low as 40 m/s, and high plas- spectrum corresponding to three ground motions
ticity indexes. The firm soil (hill zone) represented by a recorded in the lake bed zone of Mexico City during the
dark gray in the figure is mainly formed by volcanic tuff. April 25, 1989 earthquake are shown in Fig. 3. It can
be seen that earthquake ground motions from this region
of Mexico City are characterized by having narrow band
spectra with peak spectral acceleration ordinates that are
typically between five and six times the peak ground
acceleration. These spectral amplifications are approxi-
mately two times larger than those that, on average,
occur for the same level of damping in spectra of ground
motions recorded on rock or firm sites. Furthermore, the
frequency content of these ground motions is charac-
terized by a clearly defined predominant frequency and
narrower spectra.
Miranda [15,17] showed that estimation of the pre-
dominant frequency of the ground motion was very
important in order to adequately assess seismic demands
on both linear elastic and nonlinear structures built on
soft soils. A good example of this observation is shown
in Fig. 4a, where strength reduction factors computed
with the elastoplastic model and required to control the
displacement ductility ratio to four are shown for the
same three ground motions used to produce the results
shown on Fig. 3. It can be seen that strength reduction
factors of ground motions recorded on soft soil are
Fig. 2. Location of Ground Motion Accelerographic Stations in Mex- characterized by very distinct peaks, where the strength
ico City where records used in this study were obtained. reduction factors are significantly larger than the dis-
E. Miranda, J. Ruiz-Garcia / Engineering Structures 24 (2002) 1271–1281 1275

Table 1
Earthquake ground motions recorded in Mexico City used in this study

Date Magnitude [Ms] Station name Station No. Comp. 1 PGA [cm/s2] Comp. 2 PGA [cm/s2]

09/19/85 8.1 SCT SC EW 167.9 NS 97.9

04/25/89 6.9 Alameda 01 EW 37.4 NS 45.5


04/25/89 6.9 C.U. Juarez 03 EW 37.4 NS 40.2
04/25/89 6.9 Xochipilli 06 EW 43.3 NS 43.3
04/25/89 6.9 Villa Gomez 09 EW 38.6 NS 38.6
04/25/89 6.9 P.C.C. Superficie 25 EW 42.5 NS 28.9
04/25/89 6.9 Villa del Mar 29 EW 46.5 NS 49.4
04/25/89 6.9 Jamaica 43 EW 31.2 NS 35.2
04/25/89 6.9 U. Colonia IMSS 44 EW 39.6 NS 52.3
04/25/89 6.9 Balderas 45 EW 51.4 NS 42.6
04/25/89 6.9 Rodolfo Menendez 48 EW 47.7 NS 27.7
04/25/89 6.9 San Simon 53 EW 30.5 NS 39.7
04/25/89 6.9 Tlatelolco 55 EW 32.8 NS 44.9
04/25/89 6.9 Liverpool 58 EW 40.0 NS 40.6
04/25/89 6.9 Candelaria 59 EW 45.2 NS 28.6
04/25/89 6.9 Roma RO EW 54.7 NS 45.4

10/24/93 6.6 Xochipilli 06 EW 9.9 NS 8.3


10/24/93 6.6 Villa del Mar 29 EW 11.4 NS 13.6
10/24/93 6.6 Jamaica 43 EW 8.4 NS 12.1
10/24/93 6.6 U.Colonia IMSS 44 EW 15.0 NS 12.2
10/24/93 6.6 Buenos Aires 49 EW 17.1 NS 14.4
10/24/93 6.6 Tlatelolco 55 EW 9.7 NS 8.3
10/24/93 6.6 Roma-B RO-B EW 8.5 NS 6.5
10/24/93 6.6 Roma-C RO-C EW 7.9 NS 10.5
10/24/93 6.6 SCT SC EW 10.5 NS 10.9

12/10/94 6.3 Xochipilli 06 EW 15.3 NS 16.4


12/10/94 6.3 Tlatelolco 08 EW 14.4 NS 14.8
12/10/94 6.3 Jamaica 43 EW 10.6 NS 12.1
12/10/94 6.3 Balderas 45 EW 13.7 NS 11.3
12/10/94 6.3 Buenos Aires 49 EW 15.7 NS 16.4
12/10/94 6.3 Tlatelolco 55 EW 12.8 NS 9.8
12/10/94 6.3 Cordova 56 EW 17.4 NS 17.2
12/10/94 6.3 Candelaria 59 EW 14.1 NS 14.1
12/10/94 6.3 Garibaldi 62 EW 15.1 NS 13.9
12/10/94 6.3 Roma RO EW 12.0 NS 14.2
12/10/94 6.3 Roma-A RO-A EW 16.5 NS 19.4
12/10/94 6.3 Roma-B RO-B EW 13.7 NS 10.3
12/10/94 6.3 SCT SC EW 15.0 NS 11.0

09/14/95 7.1 Alameda 01 EW 37.4 NS 45.5


09/14/95 7.1 C.U. Juarez 03 EW 25.9 NS 24.9
09/14/95 7.1 CUPJ 04 EW 26.8 NS 24.5
09/14/95 7.1 Tlatelolco 08 EW 28.5 NS 26.5
09/14/95 7.1 P. .Elias Calles 10 EW 30.0 NS 29.7
09/14/95 7.1 Jamaica 43 EW 24.3 NS 27.7
09/14/95 7.1 Tlatelolco 55 EW 19.4 NS 29.7
09/14/95 7.1 Cordova 56 EW 45.2 NS 44.1
09/14/95 7.1 Garibaldi 62 EW 25.8 NS 30.1
09/14/95 7.1 Roma RO EW 37.4 NS 28.6
09/14/95 7.1 Roma-B RO-B EW 25.0 NS 23.6
09/14/95 7.1 Roma-C RO-C EW 28.9 NS 31.1

placement ductility ratio. These peaks coincide with the be seen that, despite containing information of all three
predominant period of vibration of the ground motion. ground motions, this average does not represent an
Shown in the same figure is the average of the strength adequate characterization of the strength reduction fac-
reduction factors of these three ground motions. It can tors of any of the ground motions. To overcome this
1276 E. Miranda, J. Ruiz-Garcia / Engineering Structures 24 (2002) 1271–1281

Table 2
Earthquake ground motions recorded in the San Francisco Bay area used in this study

Date Magnitude [Ms] Station Name Station No. Comp. 1 PGA [cm/s2] Comp. 2 PGA [cm/s2]

10/17/89 7.1 Foster City (Redwood Shores) FC 90 277.6 360 63


10/17/89 7.1 Treasure Island (Fire Station) TC 0 112.0 90 97.9
10/17/89 7.1 Oakland (2-Story O.B.) OAK-2 290 238.1 200 187.3
10/17/89 7.1 Oakland (Outer Harbor Wharf) OAK-W 35 281.4 305 265.5
10/17/89 7.1 S.F. International Airport SFO 90 325.7 0 230.8
10/17/89 7.1 Emeryville EMRY 260 255.0 350 255.0
10/17/89 7.1 Larkspur Ferry Terminal LFT 270 134.7 360 94.6
10/17/89 7.1 Redwood City RC 43 270.0 233 222.0

Fig. 3. Acceleration Response Spectra (ζ=5%) for Three Ground


Motions Recorded in the Lake Bed Zone of Mexico City.

problem Miranda [15] suggested normalization of the


period of vibration, T, of the SDOF systems by the pre-
dominant period of vibration of the ground motion, Tg.
Strength reduction factors for these three ground motions
but with normalized periods, T/Tg, are shown in Fig. 4b.
Also shown in the figure is the average of the three nor-
malized spectra. It can be seen that this period normaliz-
ation leads to a good characterization of the inelastic
strength demands of structures on soft soils. It shows
that, for systems with periods of vibration that are two
or more times the predominant period, the strength
reduction factor is on average, approximately equal to
the displacement ductility ratio. For systems with periods
of vibration smaller than about two-thirds of the pre-
dominant period of the ground motion, the strength Fig. 4. Strength-reduction factor spectra (µ=4) computed for three
reduction factor is smaller than the displacement duc- ground motions recorded in the lake bed zone of Mexico City: (a) as
a function of T; (b) as a function of T/Tg.
tility ratio, while the opposite is true for systems with
periods of vibration between about two thirds and two
times the predominant period of the ground motion. For In this study the predominant period of vibration was
systems with periods of vibration that are approximately computed using the approach suggested by Miranda [15–
equal to the predominant period of vibration the 17] in which the predominant period of the ground
reduction in lateral strength demand due to inelastic motion is defined as the period at which the maximum
behavior can be very large, which indicates that, for ordinate of a 5% damped relative velocity spectrum
these structures, inelastic behavior is particularly occurs. Examples of the computation of the predominant
efficient in reducing lateral strength demands. period of the ground motion for a record obtained in the
E. Miranda, J. Ruiz-Garcia / Engineering Structures 24 (2002) 1271–1281 1277

San Francisco Bay Area and in Mexico City are shown ductility ratio. Fig. 6a shows mean constant-ductility
in Figs. 5a and 5b, respectively. strength reduction factors, Rµ, as a function of nor-
malized periods, T/Tg, computed for non-degrading
(RµEP) systems using the Mexico City set of ground
5. Results of statistical study motions. It can be seen that, in general, mean strength-
reduction factors are smaller than the target ductility for
5.1. Mean strength-reduction factors systems with T/Tg ratio less than 0.65. For systems with
T/Tg ratio between 0.65 and 2.0 the strength-reductions
A total of 69,600 strength-reduction factors were com- factors are greater, and in some T/Tg regions significantly
puted as part of this investigation (corresponding to 116 greater, than the target displacement ductility ratio
earthquake ground motions, 50 periods of vibration, six whereas for systems with T/Tg ratio greater than 2.0 the
levels of inelastic deformation, and two types of hyster- strength-reductions factors are approximately equal to
etic behavior). Mean strength-reduction factors corre- the target ductility for non-degrading systems and
sponding to each hysteretic behavior (i.e., modified- slightly greater than the target ductility for degrading
Clough model and elastic-perfectly plastic model) were systems. Fig. 6b shows mean constant-ductility strength
then computed for each period and each displacement reduction factors, Rµ, for stiffness-degrading (RµSD). It
can be seen that both types of hysteretic behavior
strength reduction factors show a very similar trend,
which primarily depends on the ratio of the period of

Fig. 5. Estimation of the predominant period of the ground motion: Fig. 6. Mean strength-reduction factors compared for the Mexico
(a) for Foster City (recording station FC, 10/17/89, comp 360); (b) for City ground motions set: (a) elastoplastic perfectly plastic systems; and
Villa del Mar (recording station 09, date 25/4/89, comp 90). (b) stiffness degrading systems.
1278 E. Miranda, J. Ruiz-Garcia / Engineering Structures 24 (2002) 1271–1281

vibration to the predominant period of the ground the Mexico City ground motion set. However, for both
motion, T/Tg, and the target displacement ductility ratio hysteretic behaviors considered here, the Mexico City
m and stiffness-degrading (RµSD). However it can be seen ground motion set produces larger strength reduction
that the maximum mean strength reduction factor, which factors for systems with periods of vibration between
occurs for T/Tg=1, is greater for stiffness degrading 0.65 and 2 times the predominant period of vibration,
structures than for non-degrading systems, suggesting while mean strength reduction factors of the San Franci-
that for structures with periods of vibration close to the sco Bay Area ground motion set are larger for T/Tg
predominant period of the ground motion, lateral smaller than 0.65 and for T/Tg greater than 2.
strength demands are larger in non-degrading systems
than in stiffness degrading systems. 5.2. Quantification of the effects of stiffness
Figs. 7a and 7b show similar plots of mean strength degradation
reduction factors for systems with elastoplastic perfectly
plastic hysteretic behavior and for the modified Clough In order to study further the effect of stiffness degra-
stiffness degrading system, respectively, when subjected dation on the lateral strength demands, and particularly
to set of ground motions recorded in the San Francisco to quantify the effect of stiffness degradation on lateral
Bay Area. It can be seen that, despite the fact that the strength demands, degrading to non-degrading inelastic
Mexico City deposits are characterized by significantly strength-demand ratios rm, were computed. This para-
lower shear wave velocities, mean strength-reduction meter defined as the ratio of the lateral strength demand
factors exhibit a similar trend than those computed using in the stiffness degrading system, FySD, to the lateral
strength demand in the non-degrading system
(elastoplastic perfectly plastic) FyEP.
FySD(m ⫽ mi)
rm ⫽ (3)
FyEP(m ⫽ mi)
This ratio is a measure of how larger or smaller are
the lateral strength demands in systems with stiffness
degradation compared to those in non-degrading sys-
tems. Multiplying and dividing (3) by the lateral strength
required to maintain the system elastic and substituting
(1) we get
Fy(m ⫽ 1)
FyEP(m ⫽ mi) RµEP
rm ⫽ ⫽ (4)
Fy(m ⫽ 1) RµSD
FySD(m ⫽ mi)
Thus, this ratio also represents the ratio of the strength
reduction factor in elastoplastic systems with respect to
that of systems with stiffness degradation. This ratio was
computed for each record, each T/Tg ratio and each level
of ductility mi for a total of 34,800 degrading to non-
degrading inelastic strength-demand ratios.
Fig. 8 shows mean degrading to non-degrading inelas-
tic strength-demand ratios computed for the Mexico City
ground motion set. Spectral regions and ductility ratios,
where this ratio is larger than one, correspond to situ-
ations in which lateral strength demands of stiffness
degrading systems are, on average, larger than those of
non-degrading systems. Similarly, values in which this
mean ratio is smaller than one correspond to situations
in which the lateral strength demands of non-degrading
systems are, on average, larger than those of stiffness
degrading systems. It can be seen that values T/Tg divid-
Fig. 7. Mean strength-reduction factors computed for the San Franci- ing regions where this ratio is, on average, larger than
sco Bay area ground motion set: (a) elastoplastic perfectly plastic sys- one, from those in which is on average smaller than one
tems; and (b) stiffness degrading systems. is not constant and varies with the level of inelastic
E. Miranda, J. Ruiz-Garcia / Engineering Structures 24 (2002) 1271–1281 1279

strength demands, computed for inelastic SDOF systems


with elastoplastic behavior, to estimate the inelastic
strength demands in structures with stiffness degradati-
on.
Mean degrading to non-degrading inelastic strength-
demand ratios computed for the San Francisco Bay Area
ground motion set are shown in Fig. 9. It can be seen,
that in general the trends for this second set of ground
motions is similar to that observed for the Mexico City
ground motion set, although in this case the curves are
not as smooth because this second ground motion set
has a much smaller number of ground motions. In parti-
cular the limiting T/Tg ratios dividing the spectral
regions, where it is unconservative to neglect the effects
of stiffness degradation, from spectral regions, where it
is conservative to neglect the effects of stiffness degra-
Fig. 8. Mean inelastic strength-demand ratio computed for the Mex-
dation, are approximately the same for both ground
ico City ground motion set.
motions sets. However, the rm ordinates for T/Tg greater
than one are larger than those obtained for the Mexico
deformation. These limiting values of T/Tg that divide City ground motions set.
the region where it is unconservative to neglect the For both sets of ground motions strength demands of
effects of stiffness degradation (i.e., stiffness degradation stiffness degrading systems are smaller than those of ela-
produces larger lateral strength demands) from spectral stoplastic systems for T/Tg ⬎1. This may appear to be
regions where it is conservative to neglect the effects of counter intuitive as one may perhaps expect higher
stiffness degradation (regions in which stiffness degra- strength demands for stiffness degrading systems,
dation leads to smaller lateral strength demands than regardless of their period. However, a similar trend has
those of non-degrading systems) increase, in general, also been reported for systems with periods longer than
with a decreasing level of inelastic behavior. As seen in 0.6 s when subjected to ground motions recorded on firm
this figure, these limiting T/Tg ratios range from 0.60 for soil [2,6,10].
m=6 to about 0.9 for m=1.5. It can also be seen that for Fig. 10 shows the coefficient of variation of the
m=1.5 there is another region where rm is also smaller degrading to non-degrading inelastic strength-demand
than one for T/Tg between 0.35 and 0.45. It can be seen ratios rm corresponding to the ground motions recorded
that, for periods of vibration that are small compared to on Mexico City. It can be seen that, in general, disper-
the predominant period of the ground motion, the lateral sion increases as the level of inelastic deformation
strength demands in systems that exhibit stiffness degra- increases for a T/Tg ratio greater than 0.5. Furthermore,
dation can be on average up to 22% larger than those of for a given displacement ductility ratio the coefficient of
elastoplastic systems. Meanwhile for systems with for variation is not constant and it depends on the T/Tg ratio.
T/Tg⬎1, rm decreases as the displacement ductility level
increases. For structures with periods from one to two
times the predominant period of the ground motion the
lateral strength demands of stiffness degrading systems
range from 10% smaller (for m=1.5) to 35% smaller (for
m=6) than those of the elastoplastic systems. Thus, struc-
tures in these period regions may have benefited from the
effects of stiffness degradation during the 1985 Mexican
earthquake, leading to smaller levels of damage than
those that would be inferred from the use of elastoplastic
models. For example, for a structure with an initial per-
iod of vibration equal to the predominant period of the
ground motion (i.e., T/Tg=1) the lateral strength demand
needed to keep the displacement ductility demand below
four for non-degrading systems is approximately 1.30
times greater than that of stiffness-degrading systems.
Degrading to non-degrading inelastic strength-demand
ratios can be useful in earthquake-resistant design Fig. 9. Mean inelastic strength-demand ratio computed for the San
because they allow the use of previous results of lateral Francisco Bay area ground motion set.
1280 E. Miranda, J. Ruiz-Garcia / Engineering Structures 24 (2002) 1271–1281

6. Summary and conclusions

The primary purpose of this investigation was to


assess the influence of stiffness degradation on strength
demands of structures built on very soft soil sites. For
this purpose, a comprehensive statistical study has been
presented of strength reduction factors computed for
inelastic SDOF systems with two types of hysteretic
behavior undergoing different levels of inelastic defor-
mation when subjected to a relatively large number of
earthquake ground motions. The modified-Clough model
was chosen to represent the structures that exhibit stiff-
ness-degrading behavior under reversed cycling loading,
while the elastic-perfectly plastic model was used to
simulate the structures with non-degrading character-
istics. Two sets of ground motions were used. A first
Fig. 10. Coefficient of variation of the inelastic strength-demand ratio set includes 100 ground motions recorded on soft soils
computed for the Mexico City ground motion set. deposits in Mexico City during five different earthquakes
while the second ground motion set comprises 16 ground
motions recorded in bay mud deposits of the San Franci-
sco Bay Area during the 1989 Loma Prieta earthquake.
The largest dispersions of the degrading to non-degrad- The following conclusions can be drawn from the
ing inelastic strength-demand ratios occur for when the results of this study. The strength-reduction factors that
period of vibration is approximately equal to the pre- allow the evaluation of inelastic strength demands from
dominant period of vibration of the ground motion. Fig. elastic strength demands vary significantly with the ratio
11 shows degrading to non-degrading inelastic strength- of the period of vibration, T, to the predominant period
demand ratios rm corresponding to mean plus one stan- of the ground motion, Tg, for both stiffness degrading
dard deviation. It can be seen that in this case for T/Tg and non-degrading systems. With the exception of sys-
ratios larger than one, the inelastic strength demands of tems with periods of vibration that are two or more times
stiffness degrading systems are very similar to those of the predominant period of the ground motion, strength
non-degrading systems. However, for systems with per- reduction factors computed with ground motions
iods of vibration smaller than the predominant period recorded on soft soils are significantly different than
of the ground motion, the inelastic strength demand of those computed from ground motions recorded on rock
degrading systems can be up to 40% greater than the or firm soil sites. Hence strength reduction factors
lateral strength demand for non-degrading systems. derived from firm site records should not be used for the
design or the evaluation of structures on soft soils. For
systems with periods of vibration smaller than about
two-thirds of the predominant period of the ground
motion, the strength reduction factor is smaller than the
displacement ductility ratio, while the opposite is true
for systems with periods of vibration between two thirds
and two times the predominant period of the ground
motion.
In order to quantify the effects of stiffness degradation
of the seismic response of structures on soft soils,
degrading to non-degrading lateral strength demand
ratios, rm were computed. For structures with periods of
vibration shorter than the predominant period of the
ground motion, the lateral strength demands of stiffness
degrading systems can be on average up to 20% larger
than those of non-degrading systems. This means that in
this spectral region structures to be built on soft soils
that may exhibit stiffness degradation when subjected to
cyclic reversals, in some cases need to be designed for
Fig. 11. Mean plus one standard deviation of the inelastic strength- higher lateral forces than non-degrading structures, in
demand ratio computed for the Mexico City ground motion set. order to control the ductility demands to the same level
E. Miranda, J. Ruiz-Garcia / Engineering Structures 24 (2002) 1271–1281 1281

in both types of structure. For structures whose period nonlinear systems subjected to earthquakes, Report No. 468,
of vibration is close to or larger than the predominant Dept. of Civil Engineering, University of Illinois at Urbana-
Champaign, IL, 1979.
period of vibration of the ground motion, lateral strength [7] Otani S. Hysteretic models of reinforced concrete for earthquake
demands for stiffness degrading systems are, on average, response analysis. Journal of the Faculty of Engineering, The
smaller than those of non-degrading system. Thus, in this University of Tokyo, Vol. XXXVI, No. 2, 1981; 125–159.
spectral region stiffness degradation is, on average, ben- [8] Al-Sulaimani GI. Inelastic response of degrading and nondegrad-
eficial to the structure by reducing lateral strength ing systems under seismic actions. Ph.D. Thesis, University of
Texas at Austin, TX, 1982.
demands. [9] Nassar A, Krawinkler H. Seismic demands for SDOF and MDOF
Limiting values of T/Tg that divide spectral regions systems. Rep. No. 95, John A. Blume Earthquake Engineering
where lateral strength demands of stiffness-degrading Centre, Stanford University, 1991.
structures are, on average, larger than those of non- [10] Rahnama M, Krawinkler H. Effects of soft soil and hysteresis
degrading structures from regions where lateral strength model on seismic demands. Rep. No. 108, John A. Blume Earth-
quake Engineering Centre, Stanford University, CA, 1993.
demands of stiffness-degrading structures are, on aver- [11] Lee LH, Hang SW, Oh YH. Determination of ductility factor
age, smaller than those of non-degrading structures considering different hysteretic models. Earthquake Eng and
depend on the level of inelastic deformation. In general, Struct Dyn 1999;21:957–77.
these limiting T/Tg decrease with increasing level of [12] Borzi B, Elnashai AS. Refined force reduction factors for seismic
inelastic behavior. design. Engineering Structures 2000;22:1244–60.
[13] Gupta A, Krawinkler H. Effect of stiffness degradation on defor-
mation demands for SDOF and MDOF structures. In: Proc 6th
Natl. Conf on Earthquake Engrg, Seattle, Washington, 1998.
[14] Foutch DA, Shi S. Effects of hysteresis type on the seismic
Acknowledgements
response of buildings. In: Proc 6th Natl Conf on Earthquake
Engrg, Seattle, Washington, 1998.
The second author would like to express his gratitude [15] Miranda E. Seismic evaluation and upgrading of existing build-
to the Consejo Nacional de Ciencia y Tecnologia ings. Ph.D. Thesis, University of California, Berkeley, CA, 1991.
[16] Miranda E. Evaluation of site-dependent inelastic seismic design
(CONACYT) in Mexico for the financial support pro- spectra. J of Structural Eng, ASCE 1993;119(5):1319–38.
vided through the loan-scholarship No. 124834 to pursue [17] Miranda E. Site-Dependent strength reduction factors. J of Struc-
his doctoral studies under the supervision of the first tural Eng, ASCE 1993;119(12):3505–19.
author. The authors also acknowledge the comments and [18] Tarquis F. Structural response and design spectra for the 1985
suggestions to improve this manuscript offered by two Mexico City earthquake. Report No. GD 89-1, Dept. of Civil
Engineering., Univ of Texas at Austin, TX, 1988.
anonymous reviewers. [19] Meli R, Ávila JA. The Mexico earthquake of September 19,
1985—Analysis of building response. Earthquake Spectra, EERI
1988;5:1–17.
[20] Takeda T, Sozen MA, Nielsen NM. Reinforced concrete response
References to simulated earthquakes. J of the Structural Div, ASCE
1970;96(12):2257–573.
[1] Clough RW. Effect of stiffness degradation on earthquake duc- [21] Kunnath SK, Reinhorn AM, Park YJ. Analytical modeling of
tility requirements. Report No. 66-16, Dept. of Civil Engineering, inelastic response of R/C structures. J of Structural Eng, ASCE
University of California, Berkeley, CA, 1966. 1990;116(4):996–1027.
[2] Clough RW, Johnston SB. Effect of stiffness degradation on [22] Mahin SA, Bertero VV. Nonlinear seismic response of a coupled
earthquake ductility requirements. In: Proc of the Japan Earth- wall system. J of the Structural Div, ASCE 1976;102(9):1759–
quake Engineering Symposium, Tokyo, Japan, 1966; 227–232. 980.
[3] Nielsen NN, Imbeault FA. Validity of various systems. In: Proc [23] Mahin SA, Lin J. Construction of inelastic response spectra for
of the Third Japan Earthquake Engineering Symposium, Tokyo, single-degree-of-freedom systems. Earthquake Engineering
Japan, 1970; 707–714. Research Center, Report UCB/EERC-83/17, University of Cali-
[4] Anagnostopoulus SA. Non-linear dynamic response and ductility fornia, Berkeley, CA, 1983.
requirements of buildings subjected to earthquakes. Report No. [24] Rosenblueth E, Meli R. The 1985 Mexico earthquake: causes and
349, Dept of Civil Engineering, Massachusetts Institute of Tech- effects in Mexico City. Concrete International,6, 8
nology, MA, 1972. 1985;198:23–34.
[5] Iwan WD. The Response of simple stiffness degrading structures. [25] Borcherdt RD. Effects of site conditions on strong ground shak-
In: Proc of the Sixth World Conf on Earthquake Engrg, New ing in the San Francisco bay region during the 1989 Loma Prieta
Delhi, India, 1977; 1094–1099. earthquake. In: Proc of the XI World Conf on Earthquake Engrg,
[6] Riddell R, Newmark NM. Statistical analysis of the response of Acapulco, Mexico, 1996; Paper No. 2028.

You might also like