By
................
\
\
\
...........
............
......,
.,......
_,..,.J'
Report No. CK 9203B
February 1992
Kajirna Corporation
Kajima Corporation
'
', l
t
This report summarizes each of the studies that have been conducted in California as
a part of the CUREeKajima Research Project #5, entitled .. Design Guidelines for
Ductility and Drift Limits." This research project has been supponed by a grant
provided by the Kajima Corporation
Universities for Research in Earthquake
and administered
Engineering).
by CUREe
(California
gratefully acknowledged.
The report consists of seven chapters. The first six chapters summarize the six different
srudies that have been conducted according to the agreed team research project plan.
~etail
REPORTS
1.
..
2.
Krawinlder, H., Nassar, A., and Rahnarna, M., "Evaluation of Damage Potential
of Recorded Ground Motions, .. June, 1991.
3.
Ground :Motions and its Implications for Design of Structures, July, 1991.
4.
Miranda, E., and Benero, V. V., "Evaluation of Seismic Performance of a TenStory RC Building," July, 1991.
5.
Anderson,
J.C., M"U"anda,
6.
Hart, G., and Ekwueme, C. G., Japanese Concrete Frame Building Response,"
July, 1991.
7.
Qi, X., and Moehle, J.P., "Displacement Design Approach for Reinforced
Concrete Structures Subjected to Earthquakes," January, 1991.
Chapter 7, after a brief review of the studies reported in the above seven reports,
(summarized in the first six chapters), presents guidelines for the development of a
reliable method for estimating the values of response reduction factor R and discusses
how these values could be used to improve present U.S. and Japanese code procedures
for earthquake resistant design.
This report summarizes only the work done by researchers of the C'UREe team. The
valuable contributions of the Kajima team to this joint research project are recognized
and gratefully aclalowledged.
'
TABLE OF CONTENTS
Chapter
1
Title
Review of StateofthePractice and
1.11.17
Design
2.1 2.19
Ground Motions
3.1 3.19
4.1 4.50
5.1 5.49
6.1 .. 6.22
by Jack P. Moehle
7
7.17.9
by
Vitelmo V. Bertero
James C. Anderson
Helmut Krawinkler
(
Eduardo Miranda
and
The CUREe and The Kajima Research Teams
/
ACKNOWLEDGEMENTS
This report summarizes the results obtained in the studies that have been conducted under
task 1 of a research project on topic 5 entitled "Design Guidelines For Ductility And Drift
Limits." This particular project is part of a program of research which is supported by a
grant given by Kajima Corporation and administered by CUREe, the California Universities
for Research in Earthquake Engineering. The financial support from Kajima is gratefully
acknowledged.
The main authors of this research paper wish to thank Mr. Nobumasa Tanaka, Group
Manager of the Kajima Research Team, and the members of this team for their valuable
comments. Thanks are extended to Hatem Goucha and Brad Young for their assistance
during the final preparation of this report.
TABLE OF CONTENTS
1. IN1'RODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1. 1 INTRODUCTORY REMARKS
.............................
1. 2 OBJECTIVES
............ ...............................
1. 3 SCOPE ................................................
1134
2121
22222223242424252626273232
3233343535363638394I45454545454546474852535454555656565656575757
4. 5
4.6
4. 7
4. 8
5759
616161626263
646565
6666666667686869697171
7375
8787
88888989899090919192
92Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
6. RESEARCH, DEVELOPMENT, AND EDUCATIONAL NEEDS .......... 976. 1 GENERAL REMARKS REGARDING THE NEED FOR IDEAL
SOLUTION ..................................
976. 2 RESEARCH AND DEVELOPMENT NEEDS TO IMPROVE THE
ESTABLISHMENT OF SLEDRS . . . . . . . . . . . . . . . . . . . . . . . . . . . 996. 2. 1 SLEDRS for Service Level EQs. . . . . . . . . . . . . . . . . . . . . . . 996. 2. 2 SLEDRS for Survivability .......................... 1006. 3 RESEARCH AND DEVELOPMENT NEEDS TO IMPROVE THE
ESTABLISHMENT OF SIDRS ............................ 1006. 3. 1 SIDRS for Strength, Cy ........................... . 1016. 3. 1 (a) Code Procedures to Determine SIDRS for Cy ... . 1026. 3. 2 Implications of Recent Research Results with Respect to
104Rationale for R Code Values ............
6. 3. 2 (a) The Case of Rock and Firm Soil Sites ........ . 1066. 3. 2 (b) Case of Very Soft Soil Sites ............... . 1066. 3. 2 (c) Recommendations for Improving Code SIDRS .. . 1086. 3. 2 (d) SIDRS for Lateral Displacement and IDI ...... . 109Table
111Figures ............................................. . 113o
131131132144
8. REFERENCES
146
144145
1. INTRODUCfiON
1.1 INTRODUCfORY
REMARKS
One of the most promising approaches for developing efficient methods for improving
earthquakeresistant design (EQRD) is predicting the response of structures to earthquake
(EQ) ground motions through the use of an energy balance equation. The energy balance
equation for a general oscillatory system subjected to base acceleration can be written as:
~ =
where
Es
+ Eo
(1. 1)
Es
Eo
(1. 2)
and
Eo
where
= E,
+~
(1. 3)
~ =
~ =
kinetic energy
(1. 4)
2
where F. is the restoring elastic force and vis the relative displacement of the reactive mass,
equal to F/K where K is the elastic stiffness.
From Eq. 1.4 it is clear that for any given linear elastic perfectly plastic system the maximum
~
that can be stored is reached when F. reaches the elastic limit or yielding force Fy of the
2Fy usually demands large amounts of material, and therefore an increase in the weight and
cost of the structure, it would seem logical to try to increase vy as much as possible by
decreasing K. However, because of the need to have a serviceable structure and to control
damage of nonstructural components in general, there are restrictions on the maximum
value that can be selected for vy (or for the minimum value forK). From this discussion it
should be clear that lateral displacement (lateral drift) plays an important role in the EQRD
of structures, and that there is a need to establish the maximum value that can properly be
selected. This is not an easy task.
From the above discussion it becomes clear that to achieve economical EQRD the E. must
be minimized to the value required to achieve a serviceable structure. The problem, then
is to find out what can be done to minimize E. .
equations results in the following expression for the energy stored in the structure.
Es
= E.  (E( + EJ
(1. 5)
Although, strictly speaking, for any given structure undergoing given ground motions the
input energy depends on the behavior (elastic or inelastic) of the structure, for the sake of
simplicity it may be assumed in the first approximation that
behavior.
E. is independent of structural
Then, considering that to achieve economical EQRD the energy stored in the
structure must be minimized, it is clear from Eq. 1.5 that it is necessary to dissipate a
significant part of the total input energy either by viscous damping, by inelastic
deformations, or by a combination of the two.
Although the advantages of controlling the seismic response of civil engineering structures
by increasing the equivalent viscous damping have long been recognized, the concept of
using plastic deformation of the structural material to dissipate part of the seismic input
energy does not appear in the technical literature of the United States until the 1950s. In
1956, Housner discussed the use of limit design for EQRD [1]. The use of the concept of
ductility and ductility ratio in EQRD of reinforced concrete (RC) structures was introduced
in the United States for the first time in 1961 with the publication of the Portland Cement
Association (PCA) Manual, "Design of Multistory Reinforced Concrete Buildings for
3Earthquake
Motions" [2].
significant
experimental and analytical research efforts have been devoted to the development of
reliable methods of EQRD based on a combination of strength and ductility. In 1969, the
advantages of using plastic methods for the EQRD of ductile momentresistant steel frames
was noted [3], and in 1975 and 1977 computer programs for the application of inelastic
design methods for the EQRD of ductile momentresistant RC frames based on the use of
design ductility were developed and proposed for use in structural design practice [4 and 5].
Despite these accomplishments, the practical application of inelastic EQRD in the United
States is not widely used. This also seems to be the case worldwide, except for countries
such as Mexico and New Zealand, where building codes have explicitly introduced the use
of ductility ratio, J.L, in the estimation of seismic design forces, and allow the use of limit
design methods. In New Zealand, the seismic code is based on combined limit design and
capacity design procedures which incorporate the concepts of strength and ductilit).
The slow progress in the use of limit design and capacity design procedures for EQRD is
not surprising. The definition of the term "ductility ratio" and its evaluation are precise only
for the case of ideal linear elasticperfectly plastic behavior, which, in reality, is the
exception rather than the rule. Furthermore, even though the advantages of providing a
structure with the largest ductility that is economically feasible are generally recognized, the
term "ductility" is used very loosely to express either the deformability of the structure or
the ductility ratio. Although the deformability, ductility and ductility ratio are interrelated
parameters, their values and significance in the actual behavior of structures can be quite
different. There is an urgent need to get a worldwide agreement regarding the proper use
of these technical terms and their proper evaluation and application to EQRD of structures.
1. 2 OBJECTIVES
The ultimate goals of this report are to review the stateofthepractice and the stateoftheart in the use of the concepts of ductility, ductility ratio, and drift for attaining efficient
EQRD, and then to identify the research, development and educational needs required to
improve the proper use of these concepts.
4
1. 3 SCOPE
To achieve these goals, we will discuss both the need for applying the concepts of ductility
and control of drift and their proper use. Initial emphasis is placed on the importance of
recognizing the differences and interrelationships between deformability, ductility, and
ductility ratio, and on the need to unify the ways in which the different types of ductility
ratios are estimated from the actual seismic response of structures. It is then shown that
to achieve high energy dissipation capacity and overall effective seismic performance, it is
necessary to utilize highly redundant combined structural systems with multiple lines of
structural defense.
inelastic behavior of these systems with the highest ductility ratio that is economically
feasible.
Different
definitions of drift indices are presented, and their relationship to strength and ductility are
discussed.
After a brief statement of EQRD problems, the stateofthepractice and the stateoftheart
in the use of the concepts of ductility and drift limits are reviewed. The review of the stateofthepractice focuses on how the concept of displacement ductility ratio (J..L,) and interstory
drift index (IDI) are used in present building seismic codes. Although emphasis is placed
on the Japanese and U.S. codes, the New Zealand, Mexican and European seismic codes
are also analyzed. Results obtained from the analyses and critical comparisons of these five
codes are summarized and discussed.
reviewed by assessing the implications of lessons learned from recent earthquakes and
research results. Comparisons of the stateoftheart and the stateofthepractice are used
to identify more fully research, development and educational needs. Short and long term
solutions are formulated for the proper use of the concepts of J..L, and IDI for EQRD of
buildings, placing emphasis on the need for an energy approach.
2. 1. 1 General Remarks It is well recognized and accepted that in EQRD all structural
members and their connections and supports, i.e.,all critical regions whose yielding strength
may be reached and exceeded by a severe earthquake, should be designed (sized and
detailed) with large ductility and stable hysteretic behavior so that the entire structure will
also be ductile and display stable hysteretic behavior. There are two main reasons for.this
ductility requirement:
internal forces, to develop its maximum potential strength, which is given by the combination
of the maximum strength of all components; and second, large structural ductility allows the
structure to move as a mechanism under its maximum potential strength, which will result
in the dissipation of large amounts of energy. (It should be noted that attempts should be
mad, to develop a mechanism of a complete nature as opposed to a partial or local
mechanism, as in the case of soft stories.) While these two reasons have been recognized
in the past, only the second has been emphasized because the large dissipation of energy
was used to justify the reduction of the design strength that would be required if only linear
elastic behavior were permitted. Although this reduction is justifiable in certain cases, the
authors have previously expressed their concerns [69] about excessively large reductions in
the required elastic strength or in the Linear Elastic Design Response Spectra (LEDRS)
through the indiscriminate use of large values for the structural ductility ratio. For clarity
and convenience in discussing the reasons for this concern, a glossary of the terms to be
used in the discussion is given below.
6Ductility:
deformation after its initial yield without any significant reduction in yield strength.
structure or element can undergo without a significant loss of initial yielding resistance to
the initial yield deformation.
The above definitions are illustrated in Fig. 2.1 for the case of a Ductile MomentResistant
Space Frame (DMRSF).
2. 1. 1 (a) Needs for Recognizing the Differences Between Deformability. Ductility and
Ductility Ratio. Although the ductility ratio depends on the ductility, which in turn depends
on the plastic deformability, so that the three terms are interrelated, there are essential
differences in their quantification that need to be recognized.
Defonnability vs. Ductility: While one structure can have significantly greater deformability
than another, its ductility (particularly its usable ductility) can be smaller. For example, this
can be the case of a very flexible RCDMRSF vs. a stiff but very ductile shear wall. It is
clear from analysis of Fig. 2.2 that if the DMRSF is too flexible, i.e. the
the maximum lateral deformation,
~f"''
.dfy
the DMRSF ductility that can be used could be smaller than the available and usable shear
wall ductility.
Ductility vs. Ductility Ratio: The difference between these two terms is clearly illustrated
in Fig. 2.2. While the shear wall usually has smaller ductility than a DMRSF, it can have
a significantly higher ductility ratio.
2. 1. 1 (b) Advantages of Providing Structural Components and Their Connections with the
Largest Ductility Economically Feasible.
component should be such that the structure has the opportunity to develop its maximum
7
potential strength according to the maximum strength of its components. The need for this
is illustrated in Fig. 2.3, where the strengths of a simple structure composed of a ductile
momentresisting frame and two coupled walls are depicted as the sum of the resistance
functions of each of their components. This figure illustrates that, in order for a structure
to develop its maximum potential strength RT as determined by the sum of the maximum
strength of each component (RT =
and J.i.F 2_1.0. To allow the structure to move as a mechanism under its maximum potential
strength, the ductility ratio of the walls, particularly wall
w.,
than that of the DMRF. This figure also illustrates the difference between ductility ratio
and deformability. While the ductile momentresisting frame has a larger deformability than
the walls, its ductility ratio can be smaller than that of the individual walls. The available
frame ductility ratio cannot be used effectively because the frame has significantly larger
deformability (flexibility) than the wall components do, resulting in a relatively earlier
failure of the wall components.
(3D)
interaction between DMRF and walls, it is possible that the maximum strength of the entire
structure will exceed the sum of its components if the strength of each is determined
considering it as acting independently.
(Fig. 2.4) of the behavior observed in the experiments conducted on the 7story RC
DMRSFwall structures of the U.S.Japan Cooperative Research Program. Results of these
experiments are discussed in detail in Ref. 6. The beneficial 3D interaction was identified
as a consequence of the effects of outriggering action of frames on the wall, as illustrated
in the isometric view of Fig. 2.5. The wall, rocking around the compressive edge during its
ductile axialflexural behavior, tends to lift up the surrounding girders of the DMRSF that
frame into the walls. The girders resist this movement and, in doing so, develop higher
shear forces and increase the axial compression in the wall, which results in an increase in
its axialflexural capacity. This outriggering action results in a significant enhancement of
the lateral strength of the whole structure.
82. 1. 1 (c) Quantification of the Ductility Ratio. Though the use of the concept of ductility
ratio for the EQRD of structures was introduced in U .~.earthquake engineering literature
in the early 1950s, though its application to RC structures was presented in 1961 in the PCA
Manual [2], and though tremendous experimental and analytical research efforts have since
been devoted to its evaluation and application, even today it continues to be an ambiguous
parameter.
researchers and practicing engineers, after recognizing the need to survey, analyze and
evaluate the main parameters
analytical and experimental research and in practice to describe the inelastic mechanical
characteristics of reinforced concrete materials, sections, regions, members, subassemblages,
structures and whole soilbuilding systems, made the following statement:
These
Furthermore,
it is highly questionable
whether
the
parameters
for
9
deformations, and the degradation in stiffness and strength that has been
observed under seismic conditions."
2. 1. 1 (d)
Concluding Remarks.
design, it is possible to use the concept of ductility and/or ductility ratio in a vague manner.
But when such philosophy has to be applied in the EQRD of real structures, the philosophy
has to be quantified, and therefore it is necessary to use unambiguous parameters which can
be evaluated numerically with reliability. Such parameters are usually the displacement
ductility ratio, JJ., , and/or the rotation ductility ratio, J.l.e Preliminary designs are usually
based on a selected maximum JJ. 6 , which is determined based on the maximum values of JJ.e
which can be developed or which can be accepted at the critical regions of the structural
members.
Assuming that the values of JJ. 6 can be selected and evaluated reliably, the problem that
remains is to use this ratio or parameter correctly in the design process of a structure. To
discuss the solution of this problem it is advisable to review briefly the stateofthepractice
and stateoftheart in EQRD of RC structures. This is done after the discussion for the
needs to control IDI.
design since a significant portion of the hazard to occupants and of the total
cost of repairing earthquake
10damage.
Nonstructural
damage
is more dependent
on the relative
For clarity and convenience, in discussing the importance of control in the earthquakeresistant design of any engineering system, and particularly in case of building structures, the
following glossary of terms to be used in the discussion is introduced.
illustrated in Fig. 2.6 [12] and Fig. 2.7 [11].
2. 2. 2 Drift Definitions
Drift: Relative lateral displacement between two points (two floors).
lnterstory Drift
= A ;  A ~~
11
(Fig. 2. 6)
AiAi1
h .
(Fig. 2. 6)
The glossary is
11
= RT (Fig. 2. 7)
2. 2. 3 The Need for Drift Design. As discussed in more detail in Ref. 12, the control of the
drift of a structural system under earthquake excitation is important for at least three
different reasons:
damage to nonstructural components; (2) to limit structural damage and avoid structural
instability (P  4) problems; and (3) to avoid human discomfort under frequent minor or
even occasional moderate earthquake shaking.
Story drifts and drift ductility factors may also be useful in providing information on the
distribution of structural damage. Unfortunately, conventionally computed story drifts may
not adequately reflect the potential structural or nonstructural damage to multistory
buildings. In some structures, a substantial portion of the horizontal displacements results
from axial deformations in the columns. Story drifts due to these deformations are not
usually a source of damage [Fig. 2.7 (a)].
A better index of both structural and nonstructural damage is the tangential story drift
index, RT . As schematically indicated in Fig. 2.7 (b), the intent of this index is to measure
the shearing distortion within a story. For the displacement components shown in Fig. 2. 7.
(c) and assuming that floor diaphragms are rigid in their own plane, the average tangential
drift index is equal to
RT
= H(uu )
:Jl
L (u6 +
+ 
tL
Cl
u_
 u4''
z
(2. 1)
in which L is the bay width and H is the story height. This first term on the righthand side
of Eq. 2.1 is the conventional story drift index, and the second is a correction applied for
each bay accounting for the slope of the floors above and below the story. It may not be
12appropriate to average the values of RT for a story when the pattern of axial column
deformations varies greatly across the structure (e.g.,frames with structural walls).
Although drift indices and, in particular, tangential drift indices, provide a good measure of
the distribution of structural deformations, it may be difficult to compute corresponding
yield drifts. One possible method for computing the yield drift is taking the drift present
at the appropriate location when the building, loaded with equivalent seismic lateral forces,
reaches its yield displacement; another is computing a story sheardrift relationship for a
subassemblage consisting of the story in question with appropriate boundary conditions.
The general philosophy of EQRD for structures (particularly for building structures) other
than essential facilities has been well established and proposes:
(3) to avoid collapse or serious damage in rare major earthquake ground shakings
(where structure could be damaged but should not collapse).
The above philosophy is in complete accord with the concept of comprehensive design.
However, current design methodologies fall short of realizing the objectives of this general
philosophy [13]. A typical hierarchy of earthquake limit states is shown in Table 2.1 [14].
As pointed out by Dowrick [14], even up to the present it has been common practice to
design normal structures or equipment to meet only the two criteria designated above as (1)
and (3). Indeed, design usually has only been carried out explicitly for criterion (3), on the
assumption that the other two criteria (particularly criterion (1)) would be satisfied
automatically.
13The growing concern over the costs of earthquake damages (direct, functional and indirect)
and the difficulty of repairing much postyield damage, points out the need for more
attention to be given to control of damage and repairability at the design stage. These
needs have been clearly emphasized by the 1989 Lorna Prieta earthquake.
The control of
damage will, of course, also help to improve human safety, which is the traditional
fundamental criterion. The main source of damage is deformation; thus to control damage,
it is necessary to control deformation and particularly to control interstory drift.
In summary, achievement of reliable and efficient EQRD requires satisfaction not only of
the criterion for strength and toughness but also the criteria for deformation
and
repairability. It should be noted that strength, toughness, deflection control and repairability
are interrelated and hard to define.
Assuming that the values of the IDI can be estimated reliably, the problem that remains is
how to use correctly this ratio or index in the design process of a structure. To discuss the
solution of this problem it is beneficial to review briefly the stateofthepractice and stateoftheart in earthquakeresistant
design of RC structures.
Tahle 2.1 IIIERARCIIY OF LIMIT STATE DESIGN CRITERIA FOR DIFFERENT LEVELS OF EARTHQUAKE HAZARD [14]
(A)
Serviceability
limit state
'Usual' design
cart hqua ke
Serviceability
eartiHJuake
(or OBE)
Response
condition
(I)
Undnmaged
Elnstic
Normal
st rue! ures
or
equipment
Critical
st rue! urcs
or
equipment
(2a)
No collapse
Postyield cycling
Limited deformation
(Repairable)
(2b)
Typical
return period (yr)
Response
condition
Survivability
limit stale
Survivability
earthquake
(or SSE)
(3)
510
or TA"'" T 11 15 (?)
VI
50100
5001000
(5)
(4)
Preyield
5001000
Precollapse
preyield
Typical
ret urn period (yr)
(C)
(D)
Ultimate
limit state
As (2a)
or
Precollapse
500010 000
17
R (RESISTANCE)
~F,
~w
~Wult
A (DEFORWA TION)
~ :~EAR
~ REAL 30
.EW.
BEHAVIOR
...a.:ALL
,..  
~.
..,
A (DEFORMATION)
1
~F.
Fig. 2.3
A(OEFORMATION)
18
R (Resistance)

w2 2 8
SHEAR WALL 1
P.w2 6 7
     . . , S H E A R WALL 2
p.F 1
p.F2~
1H#.,__!..l,~,._.....r
~w1 y 6w2 y
Fig. 2.4
D UCTJLE MOMENT
RESISTING FRAME
(DMRF)
f1 (Deformation)
19
r
II.
!"
r
rl
h.
!'
60
H
,DISPLACEMENTS
RESULTING FROM
'
CHORD
(A
ROTATION},~
f:Ltf
r
i 
A )
i1 T
rL
DRIFT PRODUCING
DAMAGE
~ . ; ._~,.~
HORIZONTAL ++
STORY DRIFT
(c) DISPLACEMENT
COt.PONENTS FOR
COMPUTING R
21
3. STATEOFTHEPRACTICE
AND STATEOFTHEART
OF EQRD OF RC STRUCfURES
3. 1
PROBLEMS
IN DESIGN
AND CONSTRUCTION
OF EQRESISTANT
STRUCfURES
The problem areas have been identified and discussed in detail by the authors in a series
of publications [79]. Only the three main problem areas that have been identified are
enumerated herein. The first problematical area in EQRD is in establishing the critical
earthquake
determining
(superstructure
and
visualization (for preliminary design) and prediction of the real supplies to the building at
the moment that an earthquake strikes.
The supplies and demands, in general,, involve the mechanical characteristics of stiffness,
strength, stability, and energy absorption and dissipation capacities.
Evaluation of the
demands and prediction of the supplies are not straightforward. The determination of the
demands, which is usually done by numerical analysis using mathematical models of the
entire soilfoundationbuilding system, depends on the interaction of this system as a whole
with the excitations that originate from changes in the system environment.
determination also depends on the intimate interrelation
betw~
The
itself. Specific problems encountered in the three problematical areas of the earthquakeresistant design of structures  critical earthquake input, demands on the building, and
supply capacities of the building are discussed in Refs. 79.
While a sound preliminary structural design and reliable analyses of this design are
necessary, they do not ensure an efficient earthquakeresistant
structure.
The seismic
response of a structure depends on the state of the entire soilfoundation and superstructure
system at the time that earthquake shaking occurs, which means that the response depends
not only on construction, but on maintenance as well. A design will be effective only if the
and maintenance in the seismic performance of structures has been recognized, insufficient
effort has been made to improve these practices through, for example, supervision and
inspection.
3. 2 STATEOFTHEPRACTICE
This review will focus only on the stateofthepractice of EQRD of buildings as reflected
by present building seismic codes, and will emphasize how the concepts of J. 6 and IDI are
used and/or how they could be used to improve the stateofthepractice according to
present knowledge.
Before discussing and comparing the seismic codes for the United
States, Japan, New Zealand, Mexico and Europe, it is convenient to make the following
general remarks regarding code estimation of demands and supplies.
3. 2. 1 Estimation of Demands in Present Seismic Codes. Although the review has focused
on U.S. seismic codes, the problems identified below are common to most codes in the
world.
estimation of demands, uncertainties that can be grouped into two categories: (1) specified
seismic forces; and (2) methods used to estimate response to these seismic forces.
3. 2. 1 (a) Strength For regular buildings, statically equivalent lateral seismic forces can be
derived as follows.
V = C.W =(C.., I R) W
c. is the
(3. 1)
c.p is the
to a Smoothed Linear Elastic Design Response Spectral (SLEDRS) for acceleration, S., (C.p
= C.R = S/g),
and R is the reduction factor. Although in most of the codes the values of
J. 6 ,
23
Structural response is usually estimated using linear elastic analyses of the effects induced
directly by the above statically equivalent lateral forces or by these forces multiplied by load
factors depending on whether the design will be performed using allowable (service or
working) stress or the strength method.
codes recommend or encourage the use of limit analysis and limit design methods.
3. 2. 1 (b) Stiffness and Drift. Most of the seismic codes address design for lateral stiffness
and for drift at service level. There are only a few codes that address estimations of the
change in stiffness during the response to major shaking and the maximum drift that can
occur.
There are also only a few codes that explicitly require that the contribution of
torsion should be considered in estimating the maximum lateral drift, and very few give any
guidelines regarding how to deal with the effect of multicomponent seismic excitations [15].
o.
evaluated as follows:
(3. 2)
where
cd
the deflection amplification factor (this factor is specified for different types
of structures)
oxe
The ATC 306 [16] limits the maximum IDI, and consequently
hazard group to which the building belongs.
o., according
to the seismic
limitations is
24
i
I
3. 2. 1 (c) P.1 Effects. Few codes give explicit requirements and/or recommendations
regarding how to estimate these effects. While ATC 306 [16] recommends that the P.1
effect does not need to be considered when the stability coefficient e is less than 0.10
(3. 3)
9
where
.1
= the
V,
h,.
P,
designstory drift
In accordance with this method the commentary to ATC 306 recommends that when
a is
greater than 0. 10, the design story drifts be multiplied by a factor 1I ( 1  a) > 1. NEHRP88
[17] requires multiplication by a factor 0.9 [11(1a) > 1.
The 1988 UBC has a very vague requirement regarding P.1 effects. This code requires
checking whether at working load level the ratio of secondary moment ( == Pw
primary moment ( == V w
c5 X&) to
dead and live load; V..., is the seismic shear at working load level; and h is the height of the
story). There is no doubt that there is a need for more rational code procedures to estimate
the demands regarding the stability effects at ultimate limit states.
3. 2. 2 (a) Strength. Most of the RC EQRD codes require that the supplied strength be
estimated
I
\
25
evaluated as a function of just the minimum specified strength of the materials, and then
reduced by a strength reduction factor. There are a few codes, like the New Zealand Code,
in which the design and detailing of the critical regions of the structure is based on the
probable supplied strength capacity to the members.
EQRD codes specify minimum size and reinforcement detailing according to the ductility
ratio that is expected to be developed, this is done in an implicit way. It can be concluded
that the stateofthepractice, as reflected by most of the present EQRD codes for RC
buildings, does not appear to have included the use of the concept of energy dissipation
capacity in a rational and reliable way through the use of the ductility ratios.
3. 2. 2 (b) Stiffness. Deformation and Stability Capacities. Most of the RC codes give only
empirical expressions to estimate the socalled "effective linear elastic stiffness." Some of
them, such as the 1988 UBC which is based on the ACI (31883) [19] (with some additional
provisions), include the effect of cracking in the estimation of the effective linear stiffness.
However, the UBC, as well as most of the present codes, does not specify how to evaluate
the changes in stiffness of the whole soilfoundation, superstructure and nonstructural
components system induced by increasing damage.
procedures that, based on the supplied local energy dissipation capacity of the structural
members (rotational ductility ratio and degradation with repeated cycles, i.e. local hysteretic
behavior), will lead to the estimation of the global deformational capacity of the structure
under not only monotonically increasing deformation but also under generalized (repeated
reversal) deformation.
The socalled TriService Technical Manual [20], which will be discussed in more detail
later, recommends a method called "Method 2; Capacity Spectrum Method" which requires
estimating the "ForceDisplacement Capacity Curve." Although the way that this curve is
used is questionable, there is no doubt that the requirement to estimate this curve is a step
forward toward solution of this issue.
26
J.J 6
and in IDI, some general remarks about their use are presented.
present practice
emphasizes the use of strength in the preliminary design of structures. More specifically,
in most of the present codes, the preliminary design is based on only base shear strength
with a requirement to check the drift by elastic analysis. The insistence on only using
strength as primary criterion is perhaps a consequence of the following two reasons: first,
the practice of trying to design for ultimate or safety limit state by reducing the actual
inelastic design to one at working stress where linear elastic analysis can be used; and
second, the assumption that there is a unique relation between strength and stiffness without
recognizing that (particularly in the case of reinforced concrete structures) it is possible to
change significantly the strength of a structure without changing its stiffness.
While
preliminary design based on base shear strength could be justified for design where
serviceability (elastic response) controls, it cannot be accepted for cases where the design
is controlled by the ultimate (safety) limit state where large plastic deformation is accepted:
at this limit state, base shear strength of a given designed structure is insensitive to variation
of deformation and, therefore, to damage. Once structures yield, the base shear strength
27
remains constant, while the deformation can take any value, from its yielding value up to
the maximum value, at which collapse (sudden significant drop in resistance) occurs. In
view of the above insensitivity of the base shear strength to damage in the inelastic (plastic)
range of response (behavior), there have been some proposals to base preliminary design
on only lateral stiffness, i.e. on only controlling the interstory drift. However, a practical
method for this type of design has yet to be developed.
The authors believe that the most rational approach is one which recognizes the importance
of strength and stiffness (control of deformation) and which also recognizes that these two
factors, while strongly interrelated in the case of elastic response, clearly have a weaker
relation to each other in the case of inelastic response.
deformation at ultimate limit states depends on the yielding strength provided to the
structure. To control inelastic deformation, it is necessary to provide the structure with a
minimum yielding strength. Therefore, to achieve an efficient preliminary EQRD, there is
a need to consider two requirements simultaneously the strength (based on the rational use
of
J.1. 6 )
3. 3. 1 (a)
Examples.
cases
= MCEQ, 0.40g
of
(vJ.
28Required:
(v)ocrvia:
0.005h(v) .....
0.02h
Solution
FIRST CASE: Rectangular Acceleration Pulse [Fig. 3.1. (a)]
Design for Serviceability: Linear elastic response. In this case, for td I T ;:: 0.5, the value
of the maximum Dynamic Load Factor (DLF) ..... is 2.0. See Fig. 3.2 [22].
(R,.j..,,_.;"" = 2
(O.lOg) m
(R,.joervioe = 0.20 W
T = 21t
lm ~ k = (21ti m
~k
(3. 4)
T2
= (v..rvioe)lh,
say (IDI).
= R_,.lk]
0.005
and then
29
> (voervia:)e>:>ep~ablc
If (v oervia:)c~ooman:~
(decrease) and a new iteration in the design procedure has to be started. Assuming that the
preliminary design for serviceability was satisfactory, then we can go to:
= 80 =
for a
R,/F 1
= R, = 0.20W,
J.L 6
= 0.20W/(0.40g)m = 0.5
This is a ductility ratio which not only cannot be
developed but which also results in a vmax = 80 vclastic = 80 0.005h = 0.4h, i.e. an (IDI)~ X
Increase in stiffness will lead to a decrease in the (v...J..rv1"'' but as the T will decrease the
td I Twill increase and then the v"'/v.lasti will increase. Thus the solution is to try to increase
the strength; but by how much? It depends on how much p, 6 can be developed economically
and on how large (IDI)max can be tolerated to control damage.
Assume that l.h < 6 and (IDI) < 0.02. From Fig. 3.3 it can be seen that to have a p, 6 =
vmax/vclastic = 6
+
R,/FI
= 1.05,
Check lateral stiffness. If we keep the same k as for the service design,
vmax = (vclast;.)max
= 0.063h. This
results in an IDI which is too high. Usually the maximum acceptable is 0.02. Thus to limit
the IDI it will be necessary either to decrease acceptable J. 6 or to increase stiffness or a
combination of these two solutions.
For example:
doubling k (which will reduce the period T to a value equal to Tl ./2 and
therefore
td
=1.4 td {f=l.8;
(3. 5)
TJ/2
and reducing J. 6 to 4 will result in a required yielding strength of
R,..
Vmax
Then
::;: 4 .6(0.00Sh)
(3. 6)
0.023h
impossible.
To demonstrate that the same type of design is required even by the less
demanding type of acceleration pulse, which can be considered to be the triangular pulse,
the above example is repeated for this kind of pulse and considering a
td I T = 1 rather than 1.25, and assuming T = I sec.
For td I T = 1 and (vJ..,"';"" = O.lOg, from the graph of (Fig. 3.2), the
= 1.47 in
=0.005.
F.trec~ive = 0.40W
Assuming R = R, = (R,.Jscl'\;a: from the graph (Fig. 3.4) it can be seen that for
td I T = 1, and
= 22
0.005 = 0.11,
improvement to the solution of the FIRST CASE, the supplied yielding resistance has to be
increased, either by a decrease of the acceptable J.La. and/or by an increase of the lateral
stiffness.
Assume that k is doubled; Tis decreased by T/ v'2 and consequently ~ I T = 1.41. With
this value of td I T and considering a J. 6 = 6, the required R,,.,/F.trective
0.28W and
vmax = (vewtJ0.28W
6 = (0.005/2) h
(0.28/0.15)
6 = 0.028
= 0. 7.
Then R_. =
323. 3. 2 (b) Concluding Remarks Although none of the above assumed ground motions are
realistic representations of usual EQ shaking, the examples illustrate very clearly the need
to consider simultaneously the strength (through the proper use of JJ. 6) as well as the
deformations (through the acceptable limits of IDI). Similar procedures can be applied in
the case of more realistic EQ ground motions. The only basic information that is needed
is the:
where SIDRS
3. 4
STATEOFTilEART
EQRD
According to the previous discussion, J1. 6 should be used throughout the whole design
procedure, and particularly in the final step, where the final designing and detailing of the
critical regions of the structures are done. However, as the importance of this last step
concerns other parts (phases) of the whole research project, only the use of JJ., in estimating
the demands will be discussed herein, specifically in: (1) establishing the design EQs, and;
(2) the preliminary design of the structure.
A.
REDUCTION
33of J.J., but also of the provided overstrength, OVS, and of the increase in damping, ~.
due to large deformations.
B.
DERIVATION
OF IDRS
THROUGH
STATISTICAL
STUDIES
OF THE
o [24].
This
method can be considered as a part of the overall energy approach to the EQRD [25,
26].
Method A, which is very simple, is already widely used and has been included in codes of
several countries. However, as the proponents of this method point out, the method is only
valid for very limited types of structures. The application of this method to the design of
most real buildings is questionable [2427].
applied to simple cases, its general application in practice will require extensive integrated
analytical
and
nonstructural
experimental
studies on real
component systems.
3D soilfoundationsuperstructure
and
3. 4. 2 Use of u. in Preliminary Design. For the purposes of this discussion, the different
ways of conducting the preliminary design of EQresistant structures as regards the use of
the ductility concept in the sizing of the structural members can be classified in the following
three groups:
(a) #1a is not used at all. The critical internal forces in the members are
obtained through linear elastic distribution (LED) of forces; (b) implicit and partial use of
#1a
moments that have been obtained through a LED of forces; (c) use of limit design
approach.
Different methods, varying from the one based on simple plastic theory, which
34assumes infinite ductility, to those based on a more general plastic theory which consider
"linear elastic serviceability conditions," as well as realistic limitation of IJ.e and
incorporating
J, 6 ,
compatibility and serviceability methods, with serviceability methods being the most
promising of the two. This group also covers methods that include the possible occurrence
of shakedown phenomena, which are, at present, being developed.
3. 4. 3 Concluding Remarks. In summary, the authors believe that the future of EQRD is
an energy approach in which the concept of ductility is used by combining the methods B
and c, i.e., Be, with proper consideration of the possibility of shakedown phenomena.
However, this is considered a longterm approach. In present practice, as will be discussed
later, most of the methods that are used can be classified as under the combination Aa.
Although methods that can be classified as combined Ab are being used and have been
investigated recently [2830], the results of these investigations indicate the need for further
studies regarding:
(1) the proper limits in the amount of redistribution; and (2) the
In view of the above remarks, and the fact that it is usually not practical to change radically
the stateofthepractice, the authors would like to formulate for the immediate or very near
future the following compromise solution: to conduct the preliminary design using improved
Ab or Ac (or even Aa) methods; but ideally this should be complemented with time history
nonlinear dynamic analyses of the response of the preliminary designed structure to the
predicted maximum credible EQ (MCEQ) ground shakings that can occur at the site of the
structure during its service life. If this is not possible, the least that should be conducted
is a static nonlinear analysis of the response of the structure under a monotonically
increasing lateral load. Before this compromise solution can be applied in practice, it is first
necessary to identify the improvements needed, and then to carry out the studies required
to achieve these improvements.
35
of the role of
responding in the inelastic (plastic) range under practically a constant strength, the degree
or level of damage depends upon the amount of the plastic deformation that the structure
undergoes. Thus, to control damage it is necessary to control deformations. The question
is how to achieve such control at the different levels of EQ shaking that can occur during
the life of the structure.
three different limit states, for a preliminary design it usually is convenient to try to satisfy
only the first (serviceability) and the last (safety) limit states.
3. 5. 1 Control of IDI at Serviceability Level. Because at service level the structure should
respond in its elastic range, then for a structure with a given damping and a total mass and
its distribution, the control of IDI depends on the stiffness. The question is what is or are
proper limits for IDI to control nonstructural damage as well as to protect other contents
of the building and perhaps avoid human discomfort. At present it is generally recognized
that the limits for IDI depend on the function (occupancy) of the building, type of
nonstructural , elements and how these nonstructural
structure. Depending on these factors, acceptable limits for IDI vary from 0.0006 to 0.006.
Although the estimation of the IDI at the service level is usually based on linear elastic
analyses, there are many uncertainties regarding the effective stiffness of the structural
members (particularly on RC structures due to cracking), the deformation of the foundation,
the effects of the nonstructural components, etc. Furthermore, analyses of the deformation
should be based on a realistic 3D model that considers properly the effects of torsion under
the multicomponents of ground motions. The estimation of torsional deformations offers
serious difficulties.
36The most practical procedure for including the criteria of IDI control in the preliminary
design is to use the SLEDRS (tripartite on logarithmic scale or even the standard SLEDRS
for displacement, Sd on normal scale). If the limit IDI is specified assuming uniform IDI
along the structure (or, according to type of structure, considering possible concentrated IDI
at top or bottom story by multiplying average IDI by a factor smaller than one), the
maximum acceptable displacement can be estimated and from the SLEDRS the maximum
acceptable period of the structure can be found. This preliminary estimation of the period
facilitates the selection of the PSa that has to be used for the design for strength. Once the
preliminary design for strength has been completed, a realistic 3D model of this preliminary
design can be used to estimate the maximum IDI that can occur under the multidirectional
EQ ground motion.
3. 5. 2 Control of IDI at the IDtimate or Safety Limit State. The acceptable maximum IDI
to control damage varies with the type of structure and the function of the structure.
Usually it varies from 0.01 to 0.03. The IDI demands for different types of EQs can be
estimated by evaluating the IDRS for different values of J.L, or Cy. The problem is how to
estimate the effective lateral stiffness or period T.
P~
be used for one type of analysis and reduced section properties for another?
design codes in the U.S. are not specific about this matter.
The seismic
properties used in lateral analysis in general, and in seismic analysis in particular, varies
widely.
371.
Although elastic material behavior is usually assumed for the sake of simplicity,
3. Not all RC members in a structure, and not all cross sections along a particular member,
are in the same state of behavior at the same time.
4. For many beams and other asymmetrically reinforced members, the stiffness properties
for positive bending and negative bending are different.
..
"'
5. Uncertainties regarding the actual contribution of the floor slab and of the deformations
of the joints.
6. The stiffness of reinforced concrete members and structures varies with the time, and
with the history of past exposure to wind forces and EQ ground motions.
7. The stiffness of RC members and structures varies with the level of deformations.
Analytical and experimental studies reported in Ref. 31 indicate that for motions which are
within the workingstress design limits of members, the measured fundamental periods of
concrete structures are generally slightly less than the periods computed using grossconcretesection properties.
motions up to the yield level, the stiffness of the building is usually somewhere between the
computed values based on the gross concretesection properties and the crackedsection
properties. Based on this observation, the same reference suggests that for force analysis,
the gross concretesection properties and the clearspan dimensions be used, and the effect
of nonseismic structural and nonstructural elements be considered. For drift calculations,
38either the lateral displacements determined using the above assumptions should be doubled,
or the centertocenter dimensions along with the average of the gross section and the
crackedsection properties, or onehalf of the gross section properties should be used.
Furthermore, the nonseismic structural and nonstructural elements should be neglected if
they do not create a potential torsional reaction.
Similar sets of assumptions have been proposed by research workers who have been
concerned about the choice of member stiffnesses to be used in the PA analysis of concrete
structures.
combinations of gravity and wind loads, MacGregor and Rage (32) recommend using 40%
of the gross section moment of inertia for beams and 80% of the gross section moment of
inertia for columns.
In Ref. 34 and 35 the effects of strength and stiffness and of the type of ground motion on
nonlinear displacement response of SDOF systems is investigated.
show that nonlinear displacement response is equal to the linear response spectral values
if the system has certain strength which is determined by dimensionless parameters for
strength, initial period, and type of ground motion.
Recent results obtained by some of the authors (36) have' shown that the nonlinear
displacements are very sensitive to the dynamic characteristics of the ground motions and
in some cases the displacement can be significantly higher than those computed from a
linear elastic response, particularly if high JJ. 6 are used in the derivation of the yielding
39strength (see Fig. 3.5 for the case of Corralitos and Hollister ground motions recorded
during the 1989 Lorna Prieta EQ). This observation agrees with the results obtained by
Kappos in Ref. 37. These observations also agree with results reported in Ref. 38, which,
based on results obtained, recommend the use of the following empirical formula for
estimating the deflection amplification factor Cd (defined as the ratio of absolute maximum
interstory displacement to the corresponding value from a linear time history analysis).
In
cd
(3. 7)
In Ref. 39, Hatamoto et al. propose a newly automated seismic design method for RC
frames which aims at a uniform energy dissipation throughout the building frame, so that
the resulting damage is uniformly distributed as much as possible over all elements.
difficult to find the structural principal axes, and the centers of structural mass, stiffness and
resistance may not be coincident.
may not coincide with the structural principal axes. Therefore, when applied excitations are
in the direction of structural reference axes, the structural response may not be maximum.
This has been clearly illustrated and discussed in Ref. 15. In this reference it is shown that
the seismic components and their input direction can significantly affect the response of a
torsionally sensitive structural system.
reference axes may remarkably underestimate the response, because the structural maximum
response is dependent on the seismic input direction and its magnitude.
._   m
h
\i gmax
vglllliX
____.
... t
l :__~~
    
'\
'
J..    . .
(c) TRIANGULAR
ACCELERATION PULSE
(b) RECJ'ANGULAit
ACCELERATION PULSE
(n) SIJOFS
t .. 7'\
I.6
2.0
I
1.6
I ,~i
'V /?
~ ~
1/
y
......J
1/v
:
0.4
v
I
0
0.05
VI
I
0.10
r.b_
0.5
'r~lr
1.0
0.2
10
r....
"" ~
0.4
1"
'/
"lli_
I
I
0
2.0
f
;J
0.6
'L
0.8
lrt
I
I
....J,..
1\
lrt
0.2
......,o
0.8
1.0

I. 2
v
b
1.2
"\
I
1/
I. 4
1.0
3.0
4.0
42
'
' / / / /Y/
7z;r/f
".:I
WI
I I I
I
0. I
0.1
I I I
1..0
10
40
43
I
~0
; .
I ;
I 'II I
/,
I I VI I I
I :
20
'
I I /1
. \/1 I I 1/
1
I
: 1/ ., 1 I 111
lA Ill I I/ l
I I II 17 :7
. 'I
I
I
r: .fll_ ~ L L _y_
I I
17'
.l /J I l I 1/
l.E Tl. _/..{ 1f +V4 tT _jI VI' / . . . v I
II
I
!
 10.8 _l_
.__H itfTitftV:~JrJtr+
~~~~/~~~~~~~~~~~~
1
8 ~~~~~+14~~~~1
I
1 I
I
I /
I /1 I 1 /
I
I
I
I
I I I I I
I
I I . I I I y I ;r I (A
I~ I
I Tl I I I I I
I
5 ~~~~~~~~+Y*V~+.Y_I~Y~T~~~~~~~~~~~~~41~1~14~~~
!/1 ~I II VI/ I~ I I ; II I I
I 09 ~
1 l!i v
v 1
lf~/l_j Y~/~y
/:
1.0
t '
1 1 1
/1
r
I
/;7
Y Y A
::
I / 1....(' I I I I I
/ ' / Y L/1 /1
I I I I II
~;~y:~'Yl\
v~r~
0.2~lLL
'~"':
~ 'i
1 "".1
I~ ~:
'Y'Ii;
1.20+
A I "!/I~
?1'""'1 _...;......_ :;( I
1.60~
,.!___
p' r ~tziQO:
~ u_
F
~ !.(
)',{
L1
t.J
(111
P.esislonce
function
0.1
vr ,
;m~~~~,t~~:~llf u.~ai~
0.8 //>
o.s
/n ~
f
Oisclacement
function
I
l
irianqulor culse
l
1.._~1~~~l..'l.'__,.!,1~~p_odl....,l.!... I'_:_'l7''..!...'...:..'~...J......J:......;'W'....!',..~1:i
0.1
0.2
0.5
!J'
0.8 1
10
20
'
44
CORRALITOS
DISP.(in]
90 DEG
~~~~
14
~=5%
j.J.=l
 J1. = 2
 J1. = 4
 J1. = 6
12
10

8
6
2
0
0.5
0.0
DISP.(inJ
14
12
10
2.5
2.0
1.5
1.0
PERIOD (sec)
HOLLISTER
J.l.=l
 J1. = 2
   J1. = 4
 J1. = 6
3. 0
DEG
90
~=5%/
I
,,
.,.
,.. .
'
'' /
"'
8
6
2
0
0. 0
0.5
1.5
2. 0
1.0
PERIOD (sec)
2.5
3. 0
454. STATEOFTilEPRACTICE:
REVIEW OF CODES
4.1 COMPARISON OF TilE BASIC SLEDRS: UNITED STATES' UBC AND JAPAN'S
BUilDING STANDARD LAW (BSL)
Comparison of the Basic SLEDRS adopted by UBC and BSL for the static lateral load
procedures, which are illustrated in Fig. 4.1, shows that:
4. 1. 1 Shape of the SLEDRS The two above codes assume that, for low values ofT, there
is a constant value for the amplified acceleration (or response pseudoacceleration)
which
neglects the decrease in the amplification of the acceleration to one as the period of the
structure tends to zero. While UBC adopts a sharp transition from the constant value of
the pseudoacceleration to the decreasing values, the BSL adopts a smooth decreasing curve
which seems a more rational transition.
4. 1. 2 SLEDRS Values Comparing the values for the three standard type of soils (soil
profile types) S1 , S2 , and S3 , it is clear that except for low values of the Fundamental
Period (T
Natural
0.6 sec.) and values ofT > 2 sees. ,the spectral values specified by BSL are
higher than those required by UBC. The differences in the case of soil profile type S3 (soft
soil) reach significant values. For example, in case ofT = 1.6 sec., the value of the BSL is
1.28/1.6
= 0.55.
Only the special soil profile S4 introduced by UBC compares well with the values of the S3
given by the BSL (see Fig. 4.1).
4. 2. 1 Japan's BSL. This is the only code of those analyzed herein that explicitly specifies
twolevel design earthquakes:
several times during the use of the buildings and should be resisted with almost no damage;
and severe earthquake ground motions, with a low probability of occurrence during the use
of the buildings and which should not cause collapse or harm human lives.
464. 2. 1 (a) ServiceLevel Design Earthquake. The stresses caused by moderate earthquake
ground motions shall not exceed the allowable stresses for temporary loads.
These
allowable stresses are: for concrete, twothirds of the specified concrete strength; for steel,
the yield stress. The lateral seismic shear coefficient Q; of the ith story above ground level
shall be determined in accordance with the following formula.
(4. 1)
where C; and
w; are
accordance with the formula in Eq. 4.3, and the weight of the reactive mass above the ith
story.
(Qs).
= (V s),
= (Cs).(WJ,
(4. 2)
(WJ. =The weight of the building, which shall be the sum of dead load and the applicable
(C.),
= Z.R,.A; Co = Z.R.(1)(0.2)
(4. 3)
where
Z =
the seismic hazard zoning coefficient, which for Zone A (the zone of highest seismic
risk in Japan) is
1.0.
R = the design spectral coefficient (see Fig. 4.1) which is similar to the C factor = C of
UBC.
A;
Co
the standard shear coefficient assumed to be 0.2 for moderate earthquake motions.
This defines the SLEDRS at service level.
47Based on above Eqs. 4.1 to 4.3, the base shear for the zone of highest seismic risk is:
(Qs).
(4. 4)
(C.). = 0, 2 R. defines the SLEDRS at the service limit state and is shown in Fig. 4.2.
analysis design procedures be used to ensure the minimum resistance against severe
earthquake motions.
For this level earthquake, the ultimate lateral shear strength of each story shall not be less
than the necessary ultimate lateral shear Qun =; V.., determined in accordance with the
following formula:
Vun
=Qun
(4. 5)
where: W R is the weight of the reactive mass under severe ground motions, and C" is given
by Eq. 4.3,arid for severe earthquake motions (i.e. at the ultimate or safety limit state) the
Cu
= {J.O){R..)u = (C.)u
(4. 6)
i.e., the direct spectral values given the SLEDRS shown in Fig. 4.1.
D.
behavior which depends on the available ductility. For RC with excellent ductility
0.3
D,
2.50.
48F" =
of each story (Fe varies from 1 to 1.5); and F.= is a Shape Factor that depends on
the variation of R. = the lateral stiffness, and varies from 1 to 1.5.
For the sake of simplicity, let us assume that Fe = 1. and Fce = 1., then
Vun = Q.,
= (C.)u WR = (0.3)(RJuWR
1 for RCSMRSF
(4. 7)
(4. 8)
The corresponding smoothed design spectra for the safety level of these two types of RC
structural systems are shown in Fig. 4.3.
Level. Because UBC does not specify design earthquake at service level, it is not possible
to compare the two codes.
IDtimate or Safety Limit State Level. The design seismic coefficient (C.)u at this level
specified by BSL can be obtained from Eqs. 4. 7 and 4.8 and is graphically shown in Fig. 4.3.
According to BSL this Smoothed Inelastic Design Response Spectra (SIDRS) should be used
with limit analysis to design structures. The term "limit analysis" is used herein to denote
an analysis in which is allowed redistribution of internal forces due to plastic actions.
For structures other than for essential facilities the UBC permits damage that will not
jeopardize human life. Neither UBC nor BSL defines explicitly what constitutes acceptable
damage.
UBC recognizes that the level of acceptable damage has to be different for
49occupancy.
Hazardous Structures (Categories I and II, respectively). These structures should survive
severe (extreme) earthquake ground motions without any significant nonstructural and
structural damage, i.e., without yielding; thus, their response should remain in the linear
elastic range. If this is so, the seismic forces for which they should be designed against are
those specified by the SLEDRS.
essential and hazardous facilities can be designed for seismic forces that are significantly
smaller than those required by the SLEDRS. Specifically for the case of RCSMRSF, the
UBC allows them to be designed for
i.e.
(0.104)(SLEDRS)(l.4) = (0.146)(SLEDRS)
= (SLEDRS)/7
(4. 9)
In other words, it is allowed to deamplify the SLEDRS by a factor of 117. (As it was noted
previously, the deamplification adopted by BSL is only 1/3.3.) Although the authors agreed
in principle with the following explanation given in the Commentary published by SEAOC
[40],
"In the judgment of SEAOC the details of design and construction often
dominate seismic performance.
50Japan's BSL, on the other hand, appears to accept the same level of damage independent
of the type of occupancy or function of the facility for which the structure is designed. The
reasons for this philosophy are not clear and require investigation.
UBC permits significant deamplification of the established SLEDRS through the use of the
socalled "system quality factor," R.,.
"
_1 (Z/(1.25)S) W
Rw
rm
_1 (2.75Z[) W
(4. 10)
Rw
or
(4. 11)
working stress level rather than to a first yielding strength level, to obtain the base shear at
the yielding level, it is necessary to multiply it by the load factor U, which for RC is 1.4.
Then
(4. 12)
It should be noted that this V" is used with the ACI (Ref. 31889) strength method which
is not a capacity (limit analysis) method. However if it is assumed that the ACI strength
method is used optimizing the design in such a way that a sufficient number of critical
regions reach their plastic moment to form (develop) a collapse mechanism, it is possible
to compare directly the design base shear of Eq. 4.12 with that used by BSL, i.e. Eqs. 4.7
and 4.8.
As for RCSMRSF and Shear Wall structural systems, the R,... values are 12 and 8
respectively, and therefore
51
V,
(4. 13)
12
and
4
V, = 1. (SLEDRS) W = 0.175 (SLEDRS) W for RCShear Walls
8
(4. 14)
The two equations are plotted in Fig. 4.3, which permits direct comparison of the UBC and
BSL design earthquakes for severe (extreme) earthquake ground motions. From analysis
of the curves shown in Fig. 4.3, it is clear that the seismic coefficients specified by the BSL
are significantly higher than those specified by UBC. For example, for Soil Profile S 1 and
a structure with a T = 1 sec. and having a RCSMRSF structural system, UBC allows to
design for a strength of Vu = 0.059 W, while the BSL requires Vu = 0.192 W, i.e. near three
times that required by UBC. Furthermore, the Japanese require that W include some live
load, and this is not generally required by UBC. Note that if the ACI strength method is
used without any plastic redistribution and without attempting to design all critical regions
for just their required strength, the resulting designed structure can have significant story
yielding overstrength.
It appears that either Japan's BSL is too conservative in its seismic coefficient or the UBC
is not conservative enough. Note that even the seismic coefficients specified by the BSL for
moderate (service) ground motion (Fig. 4.2), are larger than the seismic coefficients specified
by UBC for severe earthquake ground motion. It appears that UBC uses
R...
values which
are too large for reducing the SLEDRS, and therefore implies the use of an unrealistically
high ductility ratio in reducing the required strength.
It should be clearly noted that it is not possible to judge the effectiveness of the earthquake
resistant construction of any country just by looking at the code requirements of the seismic
forces to be used in the design. The quality and effectiveness of any earthquakeresistant
construction depends on the design philosophy (selection ofbuilding configuration, structure,
52layout, and structural system) among many other factors or aspects, including the following
main ones:
1.
Code requirements which include not only those regarding the seismic forces, but even
more importantly, the minimum requirements for the materials, sizing, and detailing
of the structurai and nonstructural components, as well as inspection during all the
phases of design and construction.
2.
3.
Maintenance (preservation).
The degree of damage during an earthquake depends on the deformation (stiffness) of the
structure.
However, as
previously pointed out, the resulting deformations are related to the yielding strength
provided to the structure.
If this
minimum strength is not available the inelastic deformations can reach unacceptable values.
for a linear
viscous damping mechanism, which is usually considered in setting up the structural equation
of motion and solving it to obtain through linear elastic analysis the LERS for SDOF
systems. This is clearly illustrated in Fig. 4.4, where, for a given specific recorded ground
53
compared with the 5% damped IRS obtained considering the effect of the dissipation of
energy induced by the inelastic (plastic) behavior of the SDOF when its yielding strength
(or its yielding seismic coefficient Cy) is reduced considering different values of the ductility
ratio J.L.
The above observation that "the reduction in structural response due to viscous damping
mechanism, i.e., due to an increase of damping ratio
~'is
induced by early yielding due to increase in acceptable ductility ratio p." is also clearly
illustrated by comparing the spectra of these reductions as recommended
by several
researchers, illustrated in the graphs of Fig. 4.5 [41]. From these graphs it is clear that the
reduction due to J.L decreases as the frequency increases. On the other hand, the reduction
due to increase in
increases as the frequency increases (see Figs. 4.4 and 4.5, aas well as
Table 4.1, where it has to be noted that the displacement region of the spectra corresponds
to the lowest frequency and the acceleration region to the highest frequency). Therefore
it appears that any attempt to reduce the SLEDRS due to inelastic behavior (i.e. due to J.L)
by selecting a new SLEDRS corresponding to an increased value of
is questionable. This
is the main reason for questioning the procedure recommended in what is usually called the
TriServices Manual approach [20, 42].
4. 3. 1 TriServices ManuaJ Approach. This manual recommends that essential buildings
(as well as all other buildings) be designed to resist two levels of earthquake ground
motions. This is a step forward in improving design approach. The first level of motion is
designated as EQ1 and the second and larger amplitude of motion is designated as EQ11.
The lateral forceresisting structural systems of these facilities will be designed to resist EQI
by elastic behavior (i.e. using SLEDRS).
Two
54
overstresses of individual elements, calculating the Inelastic Demand Ratios (IDR) which
are equal to the ratio of Demand Forces {M 0 , V0 , F 0 ) to the calculated yield or plastic
Capacities of the Elements <Me, Ve Fe). If the IDR exceed acceptable values, or some
other four conditions, the structure must be analyzed in accordance with Method 2. The
acceptable values for the IDR vary from 1 to 2.0 for Essential facilities; 1.25 to 2.5 for High
Risk facilities; and from 1.25 to 3.0 for all other facilities. These values seem reasonable,
however in the practical application of this method they are usually increased up to 10.
A stepbystep approach
questioned (particularly the determination of the change in the value of the fundamental
period or the effective period of vibration), there is no doubt that conceptually this
recommended procedure, which requires the estimation of the capacity of the structure to
resist lateral forces, is a step forward towards the improvement of earthquakeresistant
design methodology.
Two seriously questionable procedures in Method 2 are: first, the recommended procedure
to estimate the demand curve which is obtained by estimating a reduction (or deamplification)
in response amplification
comparison of the capacity spectrum with the demand curve. The reduction in response is
done by reducing the spectral values of the SLEDRS corresponding to EQII through the
use of a larger value of the viscous damping ratio
problem with this method is that conceptually the actual SIDRS cannot be obtained from
a SLEDRS by only increasing the value
of~
previously and illustrated in Fig. 4.4. Usually this method can be quite conservative,
particularly for structures having a certain period T (T close to the value of the predominant
55period [frequency] of the ground motion) and possessing a ductility displacement ratio equal
to or larger than 1.5. It is recommended
computed capacity response spectrum be compared with the demand spectra obtained by
estimating the IDRS corresponding to the Cy, as well as the
estimated lateral resistancedeformation
p.~
The direct comparison of the capacity spectrum (estimated according to the TriService
manual) with the earthquake demand curve can be questioned because the estimated change
in the effective period of vibration can only become effective in reducing the response of
the structure in the event that the ground motion time history is of such nature that it
induces on the structure a response that consists of repeated cycles of deformation reverSals
whose peak intensity (magnitude) increases with time from the first yielding to a reversal
. requiring the maximum available displacement ductility [see Fig. 4.8 (a)]. The comparison
will be completely incorrect in the case of impulsive types of EQ ground motion, which
would require the structure to develop its deformation in just one cycle of inelastic behavior,
as is illustrated in Fig. 4.8 (b).
4. 3. 2
Several
comparative designs of RC and steel buildings have been conducted in cooperative studies
among Japanese and U.S. designers [4449]. Because the Japanese BSL design procedure
is based on a specified story yielding strength, while the U.S. UBC design procedure is based
on the yielding strength of the member, it should be expected that the UBC would result
in design with higher overstrength. However, it is believed that this higher overstrength can
not overcome the significantly higher lateral strength required by the BSL. From the results
of these comparative designs it becomes clear that the current Japanese code requires
lateral strengths significantly higher (2 to 3 times) than the U.S. code. On the other hand,
in most of these studies it has been concluded that the drift requirements of the U.S. code
are more severe than the Japanese Code. This last conclusion needs further studies because
it appears that this conclusion has been derived by comparing only the drifts obtained under
56
different lateral load conditions and not under time history analysis of the response to the
same or similar ground motions.
4. 4
proportion members for design loading U. Then "Capacity" design is used to ensure that
during earthquake loading the preferred mechanism of energy dissipation is maintained.
That is, the action arising from "overstrength" under flexural or combined flexural axial and
shear stresses at the critical inelastic regions (usually called plastic hinges) are used as
design actions (forces) for other elements in the structure to avoid brittle failure elsewhere.
fr
~.
The "CM:I::!I1tu:ug1h" used in capacity design is calculated using steel strengths of 1.25
fr
for
Grade 275 (Grade 40) reinforcing steel, 1.4 fr for grade 380 (55 ksi) reinforcing steel and
~=1.
+ 1. 7 4
I. OOD + 1. 3 4 + E
0.90D + E
U = 1.40D
U =
U =
(4. 15)
57
4. 4. 5 (a) Earthquake Provisions. The NZS starts by stating a series of requirements that
are to be adopted to comply with the standard such as Symmetry and Ductility [which means
the ability of the building or member to undergo at least repeated and reversing inelastic
deflections beyond the point of first yield while maintaining a substantial proportion (;:::
85%) of its initial maximum load carrying capacity].
4. 4. 5 (b) Method of Analysis. According to the NZS, buildings shall be analyzed by the
equivalent static force method. However under certain special circumstances an additional
spectral modal analysis shall be conducted.
4.4.6 Equivalent Static Force Analysis: Total Horizontal Seismic Force or Shear: V, at the
(4. 16)
W,=D+C,L
(4. 17)
where: W, = total reduced gravity load above the level of imposed lateral ground restraint
and is given by the expression 4. 17, where C, varies from 0 to 2/3 depending upon the
intensity of the live load L; D = the dead load; and the seismic design coefficient
Cd = CSMR
(4. 18)
58C
basic seismic coefficient, to be determined from Fig. 4.9in accordance with the seismic
zone (A, B or C), the subsoil flexibility and the period T.
specified in the NZS, it appears that the spectral values for C can be interpreted to
represent the Design Earthquake (SIDRS) for the severe (extreme) earthquake ground
motions.
S 
structural type factor, intended to reflect the potential seismic performance of different
structural systems. The specified level of S (which varies form 0.8 to 5 for Reinforced
Concrete Structures
an~
ability of the structural type to dissipate energy in a number of load cycles (i.e. sho.uld
reflect the available ductility), and secondarily its degree of redundancy where
appropriate.
M or Mp =
factor related to the relative direct risk to life of a failure [corresponds to the
importance factor (occupancy type or function of structure) I of UBC]. However, the
value of R can be as high as 2 while the largest value of I is 1.25.
Assuming R = 1, i.e., buildings with normal occupancy or usage, the seismic design
coefficient corresponding to a RC frame responding elastically and located on rigid or
intermediate subsoils will be given by the spectra shown in Fig. 4.10, i.e.
59Ve
= Ce W, = (CSMR)
W,
= [C(5x0.8)
(1)] W,
= [4C]
W,
(4. 19)
For a similar structure but taking advantage of its maximum available ductility (actually
using its maximum capacity of energy dissipation) the design can be conducted using the
SIDRS Cu
V.,
= C.,W, = [Cx0.8x0.8]
W,
= [0.64C]
W,
(4. 20)
In other words the SIDRS C" can be considered to be obtained from the SLEDRS Ce by
reducing it by 4/0.64 = 1/0.16 = 6.25. Note that the NZS, similar to the UBC, reduces the
SLEDRS Ce by a constant value which is independent ofT. Furthermore, the NZS specifies
that for structures responding elastically, the seismic forces depend on the SLEDRS Ce, i.e.
V = 4C W, . However, for the design of structures containing highly hazardous content the
required V = (1.28C)W, , i.e., is obtained by reducing the SLEDRS Ce by 3.125. The
rationale for this reduction is not very clear and it appears to be incompatible with the fact
that these structures should remain practically elastic (i.e. respond in their elastic range).
Note that in the case of essential facilities (hospitals, etc.) the reduction is even higher
(3.91).
Similar incompatibility exists in the UBC. This code for hazardous and essential facilities
using RCSMRSF allows reducing the SLEDRS by a factor of 12/(1.25 1.4) = 6.86. The
rationale for this large reduction needs to be investigated. While it is true that any small
amount of yielding (ductility) can reduce significantly the ordinates
of the LERS
corresponding to structures with period T equal or close to the predominant period of the
ground shaking T, , in general it is difficult to justify a constant reduction of the order of
3.91 in the case of the NZS and particularly of 6.86 in the case of UBC.
60structures the UBC requires the use of the strength method which basically is not a capacity
method.
In general, for identical seismic forces, the application of limit design (based on
limit analysis) will result in a structure with smaller ultimate strength (or "overstrength")
than if it is designed just by the application of the ACI strength method. Only in the case
of "optimal strength design" (where every critical section is provided with just the demanded
strength) will this result in a structure with the same ultimate strength as that obtained from
applying limit design method.
redistribution of the design moments obtained from an elastic analysis to 70% of the
moment for the section obtained from an elastic envelope covering all appropriate
combinations of loads; i.e. moment reduction should not exceed 30%. The ACI strength
method limits the redistribution (increase or decrease) by 20%.
Neglecting the effect of the amount of redistribution in the design method as well as
differences in the material sizing and detailing requirements among the three codes and
comparing only the required seismic forces as illustrated in Fig. 4.11, it is clear that:
1.
To account for the effects of inelastic behavior the three codes reduce the seismic
forces required for structures responding elastically through a reduction factor that is
independent of the period of the structure.
2.
The BSL requires the use of the largest lateral seismic forces; for structures with
T < 2 sec. the NZS requires the smallest forces. For structures with T close to 1 sec.,
the BSL requires forces which are more than 3 times the forces required by UBC.
3.
Of all the codes reviewed herein, the largest reduction of the seismic forces that are
required to obtain the design forces for safety to keep the response of the structures
linear elastic (i.e. to the SLEDRS) is permitted by the UBC (12/1.4 = 8.57) This
reduction is attributed primarily to the effects of dissipation of energy (ductility). The
NZS allows a reduction of 5/0.8 = 6.25 and the BSL is the one that specifies the
smallest reduction (110.3 = 3.3). In chapter 6 a more detailed discussion will be
61
It should be noted that seismicity m the highest seismic region of New Zealand is
significantly lower than that in the corresponding highest seismic region in the United States
and Japan.
This fact is clearly reflected in the SLEDRS adopted by New Zealand when
compared with that of the United States and Japan (Fig. 4.11).
4. 6 COMPARISON OF REQUIRED
4. 6. 1 NZS Requirements.
forces specified at the safety level (multiplied by the factor K/SM, K being a factor by which
the values of C are scaled: K = 2.0 for the defined static forces, and 2.2 in the case of
dynamic modal forces) shall not exceed 0.0006 of the story height where nonstructural
elements are not properly separated nor 0.010 of the story height multiplied by the zone
factor which varies from 1. 0 for Zone A to 2/3 for Zone C. The NZS does not have a
specific requirement limiting the drift under service seismic forces. However there is a
general building deformation requirement limiting the interstory drift index (IDI) to 0.002.
Using this value, the total lateral load (normalized with respect to the weight of the reactive
mass) interstory drift index relationship is shown in Fig. 4.12. It can be seen that to reach
the maximum IDI of 0.010 the structure will have to develop a ductility ratio of 5 which is
the value that the NZS appears to have adopted as feasible to achieve. However, it should
be noted that from the values specified for structural type factor S, it appears that the NZS
considers a maximum displacement ductility ratio of 5/0.8 = 6.25.
4. 6. 1 (a)
Calculations of Deformations.
through a linear elastic analysis using the specified forces multiplied by K/SM (which for
62RCSMRSF is 3.125). If this is so, and considering that the maximum deflection at service
load is 0.002 h, it would appear that the designed structures can never reach the maximum
= 0.00625 h
< 0.01 h.
4. 6. 1 (b) Building Separation. The NZS specifies that parts of a building, or buildings on
the same site separated from each other shall have a minimum clear space from each other
either 1.5 times the sum of their computed deflections or of 0.004 h, whichever is the larger
and in any case not less than 25 mm (1.0 in.).
4. 6. 2 UBC Requirements.
Calculated story drift shall not exceed 0.04/R,., nor 0.005 times
> 65 feet.
calculated story drift shall not exceed 0.03/R,., nor 0.004 times the story height.
The
specified story drifts are those due to the specified seismic design forces using linear elastic
analysis. Thus they are at the working stress level. The UBC does not specify directly the
maximum acceptable drift under the extreme earthquake ground motions.
However,
according to the UBC requirement for deformation compatibility, it is assumed that the
maximum displacement should not exceed 3(R,/8) times the displacement resulting from
the required lateral forces.. Thus the maximum interstory drift for SMRSF structures of
building with height :::; 65 feet should not exceed:
m~ (O~~l
= O.O!Sh
(4. 21)
(i)
Rw (0.005h)
= 0.0225h
(4. 22)
Note that, according to the above requirement, if the structure is designed just to satisfy the
requirement that the IDI under the working lateral forces is equal to 0.04/R,., , then the
63maximum displacement ductility ratio that can be used is 0.015/[(0.04/R...) 1.4] = 3.21, which
is smaller than the specified R... modified
to yielding level R,
= R,)1.4 = 8.57.
For
buildings with a height exceeding 65 feet, the serviceability requirements are controlled by
0.03/R...
= 0.0025, thus
= 0.0035.
The maximum
acceptable deformation would be (0.03/R..,) x (3/8R...) = 0.0125. In this case the maximum
ductility that can be used is 0.0125/0.0035 = 3.57. Considering a building with T 2_1.64
sees. the minimum required strength would be: at working 0.03W; and at yielding 0.042W.
Then the idealized loaddeformation relationship is that illustrated in Fig. 4.12.
motions (i.e. V = 0.2R.W = 0,2 w for R. = 1.0) the story drift shall not exceed 0.005 of the
story height. This value can be increased to 0.00833 if the nonstructural members will have
no severe damage at this increased story drift limitation. Regarding the maximum interstory
drift under severe earthquake ground motions requiring a lateral shear strength of 0.30W
(see Figs. 4.3 and 4.11), to the best of the author's knowledge the BSL does not have any
direct or explicit specifications.
requirements
requirements to the severe motions level we have Fig. 4.12. It is clear from this figure that
if the maximum interstory drift index should not exceed the value of 1.5% (as recommended
by ATC and UBC), the maximum ductility implied in the BSL design would be 0.015/0.0075
= 2 for
1.2 for
cases where the nonstructural elements cannot be damaged. From these results it is very
clear that if Japanese designers try to satisfy only the minimum elastic stiffness required by
the BSL, they cannot take advantage of the maximum displacement ductility ratio that the
BSL appears to consider for reduction in SLEDRS for strength, Cr (i.e. 110.3 = 3.3).
It should be noted that from inspection of RC ductile moment frames constructed in Japan
it would appear that they are stiffer than required by the BSL. Furthermore, the above BSL
specifications apply only for buildings up to 60 meters (about 20 stories) in height. For
buildings higher than 60 meters (197 ft) special permission from the Ministry of
64
Construction following a detailed review of the dynamic behavior of the structure by the
Board of Technical Members is required. While it is true that the Japanese designers use
somewhat shallower girders than those used in the United States (to decrease story height),
the width of the floor girders is significantly wider than those used in the United States.
Furthermore, in Japan the span of the bays is usually smaller than in the United States.
4. 7 MEXICO:
DISTRICT
These standards recognize that acceptable damage is different for essential facilities (group
A) than for standard occupancy facilities (group B). However this differentiation is based
on adopting for group Aa seismic coefficient 50% higher than that for group B, or in other
words, it appears that it is not required to keep the essential facilities in the elastic range,
since this will require for group Aa seismic coefficient significantly higher than just 1.50
times the one adopted for group B. To keep the behavior of a RCSMRSF for group A in
its elastic range requires a design with a seismic coefficient of about 4 times (value of Q')
the values adopted for the inelastic design of the same type of structure for group B (Fig.
4.13). Even in the case that for an essential facility it would be possible to accept structural
damage equivalent to a displacement ductility 2 (e = 2), which would demand considerable
damage at the most critical region of the structure, the seismic coefficient should be not just
50% but 100% higher than that for group B.
It should be noted that the Mexican Federal District Standards are the only ones of those
considered herein which specify in their static method of analysis that when the T of the
building tends to zero, the required strength seismic coefficient
c. tends
4. 7. 1 Mexico Reguired Strength at illtimate (Safety) Limit State. The required maximum
strength design spectra for RCSMRSF located in Zones I and III are shown in Fig. 4.13.
It is obtained from the SLEDRS by reducing it by a reduction factor of Q, which, contrary
to other codes, is period dependent.
65
< 0.6 in the case of Zone II) the SLEDRS is divided by Q' = 1 + (T/T.)(Q,1), for any
value of T > T. = 0.2 sec. for Zone I and T > 0.6 sec; for Zone III the value of Q' = Q
=4
for the case of SMRSF. The reinforced concrete code requires that the design of
SMRSF be conducted using a strength method with a factored seismic load of 1.1 E. The
required strengths are shown in Fig. 4.13 for Zone III.
It should be noted again that from analysis of the IRS for different values of JJ. 6 , it becomes
clear that in general the required yielding seismic coefficient is not equal to the C required
by the SLEDRS divided by JJ. 6 The reduction varies with the T. This is clearly shown from
analysis of the values of the reduction for the Oakland Outer Harbor Channel 3 (Fig. 4.14);
while for value ofT > 0.5 sec. the reduction due to a JJ. 6 = 2 is higher than 2 (for T
=1
sec. the reduction for JJ. 6 = 2 is higher than 3). For T < 0.25 sec. and forT > 2 sees. the
reduction is smaller than 2. Furthermore, for values of JJ. 6 > 2 the reduction does not
increase in direct proportion to the values of JJ. 6 For example, for T = 2.0 sees. the
reductions for JJ. 6 = 3 through JJ. 6 = 6 are practically the same.
4. 7. 2 Mexico Requirements for lnterstory Drift. The 1987 Mexican regulations reqmre
that the maximum lateral IDI not exceed 0.006, except where elements that did not form
part of the structure were attached in such a manner that they would not be damaged by
structural deformations.
noted that these values include the inelastic deformation that was usually based on a Q
JJ. 6
= 4,
thus it appears that implicitly the Mexican code limits the linear elastic IDI to
0.0015 or 0.003, depending on whether the nonstructural elements can be damaged or not
due to the lateral deformations.
strengthinterstory drift index, according to the 1987 Mexican Code. From Fig, 4.12, it can
be seen that in the elastic range for either short or long T, the IDI requirements of the 1987
Mexican Code are the less severe except for the Japanese BSL. At the ultimate Limit State
of Safety and for short period structures, except for the New Zealand Standards, the limit
for the IDI is the most severe of all the regulations considered herein.
66The 1987 Code also requires that all types of construction shall be separated from the
neighboring pieces of land at a distance not less than 5 em (2 in) or the horizontal (lateral)
displacement
calculated
The horizontal
displacement should be computed with the reduced seismic forces (i.e. reduced by Q
= ~J. 6)
and then multiplied by the "factor of seismic behavior" Q = # 6 , and the result increased by
0.001, 0.003 and 0.006 of the floor level above the ground for Zones I, II and III,
respectively.
IN EUROPEAN SEISMIC
CODES
Recommendations
Technical Committee 1, Working Group 1.3, Seismic Design, 1st Ed., 1988
Seismic Design of Concrete Structures, CEB, 1987.
Both documents advocate a very similar approach, but with a few important differences.
These differences are identified here wherever appropriate.
this summary.
Recommended values for corner periods, plateau values, and shape of descending branch
vary between documents.
67
The CEB alludes to the use of different and site specific spectra, but the consensus in both
documents is that the ground acceleration A and elastic response spectrum shape are
adequate to describe the seismic demand for design purposes.
considered.
The ECCS recommends translation of the elastic response spectrum shape for damping
different from 5%, and also gives values for absolute and relative ground displacements to
be used in design.
4. 8. 2 (b) Design Spectrum. As illustrated in Fig. 4.16, a design spectrum shape is created
from the ground motion spectrum in a manner similar to ATC 306 [16], i.e.,the plateau is
extended toT = 0 and the liT line is changed to 1/(T)k, with a recommended value of 2/3
for k. The corner period
R(1)=Ro
Ro
R(1)=(I_)"
(4. 23)
To
R(1)~0.3Ro
This spectrum shape, R(T), multiplied by the ground acceleration A, is used as the basis for
deriving the seismic design forces. Again, the approach is similar to the Rfactor approach
in ATC 306, except that the Rfactors are called "behavior factors" and are denoted as q
(in ECCS) or K (in CEB). Thus the ordinates of the design spectrum C(T) are given as
68C(T) A
R(T)
K
(4. 24)
In equivalent static force analysis, the C(T) values are used as base shear coefficients and
are distributed over the height in a manner similar to U.S. codes. In modal analysis, the
C(T) values are used as spectral values for SRSS superposition
In the design process, strength design is used with rather complex combinations of different
load effects. The seismic load effects show up usually without load factors, i.e. ,again similar
to ATC. A significant difference from ATC is that portions of the live loads are considered
to be seismically effective.
4. 8. 2 (c) Behavior Factors. These factors differ significantly between steel and RIC, and
are quite different from the Rfactors used in the U.S.
For RC structures with ductility level III (highest ductility level), the CEB factors are
RC frame systems
K = 5
K = 4
For steel structures, the ECCS recommends the qfactors given in Table 4. 2. These qfactors contain the ratio
ex~/ cx 1
associated with mechanism motion to lateral length associated with first yielding in the
structure. Thus the ECCS accounts for "mudme overstrength" due to redistribution in the
inelastic range.
for damage control, but reference is also made to pounding. Drift criteria are set at either
the elastic design force level or the maximum expected displacement.
Throughout the
documents it is assumed that the maximum expected displacements can be obtained as the
elastic displacements computed under design loads times the behavior factor. The ECCS
69states that the spacing between adjacent structural units shall not be smaller than the sum
of the maximum expected displacements computed for the units.
Drift criteria for damage control differ significantly between ECCS and CEB, but are very
stringent in both cases.
under a
"scnicr:ability earthquake" (not defined) shall not exceed either 0.003h or 0.006h, depending
on the type of nonstructural elements attached to the structure. The CEB limits the elastic
interstory design deflections under specified design loads to (0.01/K)h.
is neglected
A~
4. 8. 3 Evaluation of Ductility Considerations. The global design is similar to, that proposed
in the ATC 306 document [ I6]. Elastic ground motion response spectra form the basis for
design. These spectra are modified for design, principally by increasing the spectral values
for T > T0
proportional to I I T 213 rather than I I T. No reasons are given for this modification, but
it probably is to account for multimode effects and to provide larger safety for long period
structures (ATC 306 arguments). Design force levels (for strength design) are obtained by
dividing this modified spectrum by behavior factors that may be as high as 8 and as low as
2.
70
The implicit
assumption (deduced from the criterion that inelastic displacements are equal to elastic
displacement times behavior factor) is that behavior factors and ductility ratios are identical.
In the CEB the behavior factors are independent of the real strength of the structure, thus,
no consideration is given to the increase in structure strength beyond the strength design
level. In the ECCS the behavior factors can be increased by the ratio of structure strength
to first yielding strength.
considerably smaller force level than RC structures (CEB), and significant differences in
ductility demand have to be expected.
In both the ECCS and the CEB the design is based on explicit strength considerations and
only implicit ductility considerations. As a consequence, the expected ductility demands will
be different for the same type of structures, dependent on structural period and many other
It is well established that the ratio of strength reduction factor (behavior
considerations.
factors) over ductility ratio is not a constant. This ratio is period dependent and is greatly
affected by site conditions, particularly for soft sites. For such sites the use of an elastic
response spectrum for design of structures that are expected to respond inelastically in a
severe earthquake is not appropriate unless site specific strength reduction factors are
considered.
At the element
distinguishing between "nondissipative" elements and "dissipative" elements. The former are
"brittle" elements which cannot be counted upon to dissipate energy through inelastic
deformations.
behavior factor q < 6, < 4, and < 2, respectively. Specific detailing requirements are given
for the three classes.
In the ECCS great emphasis is placed on "safety checks" whose purpose it is to force
inelastic deformations into dissipative zones (e.g.," ... dissipative zones must be part of the
beams and not of the columns ... ")and prevent overloads in nondissipative zones. The
"
71
guaranteed that the axial yielding of the tensile diagonals precedes the ultimate strength of
their connections as well as the collapse of beams and columns." Implementation of this
concept is very rigorous, much more so than in the 1988 UBC, and includes also
consideration of resistance factors (usually 111.2) and material overstrength factors.
The CEB does not contain similar overload safeguards [note the parallel to the 1988 UBC],
but is rather explicit in detailing requirements for individual elements.
codes, but their intended purpose is equally nontransparent. Specification of drift limitations
at the design force level, which is arrived at by dividing a severe earthquake spectrum by
judgmental behavior factors, cannot provide consistent damage control. It will be difficult
to relate _this drift to that anticipated in a "serviceability" earthquake for which damage
control is intended. The often quoted statement that maximum expected drift in a severe
earthquake is equal to elastic design drift times behavior factor should be looked upon with
great caution.
P~
effects.
4. 8. 5 Summary. The basic concepts of seismic design in both documents are rather similar
to those implemented in modern U.S. codes (ATC 306, UBC 1988). Significant differences
to U.S. codes are noted in specified numerical values ofjudgmental design parameters (e.g.,
behavior factors, drift limits) and in design details. Most of the differences point toward
more stringent design requirements for European practice. Both European codes, similarly
to present U.S. codes, are strength based rather than ductility based, and advocate a single
level design without a clear distinction between damage control (serviceability) and collapse
control (safety) requirements.
Percent of critical
damping
2.5
2.2
0.5
2.0
1
2
5
7
10
20
Acceleration
Velocity
1.8
i .4
1.2
1.1
1.0
4.0
3.6
3.2
2.8
6.4
5.8
.. 5.2
4.3
1.9
2.6
1.9
1.5
1.3
1.5
1.1
1.2
Structural
St=~ct......:::.es
F:=ame
System
\,
..
mainlv reactincr i:1 .... enclncr:
structures
sa u 1 a...L.
~ 8
bracings
Cantilever structures
Structures with reinforced concret.e
walls
Mixed structures
a u/ a 1
a u/ a 1
a u/ al
see recommendations
for reinforced
conc:ete
75
f
I
1.2
UBC
1.0
r' I
I ,
o.s : 1
lt
lt
0.6
IJ
0.4
II
I
l
0.2
0
0
2.0
1.0
3.0
T(sec)
0.4
0.2
0
0
2.0
1.0
T(sec)
3.0
0.6 
SHEAR WALL
_,JAPAN'S BSL
0.4 .........
SMRSF
0.3
SHEAR WALL
..
~/
I
0.18 .,.0.2
__
0.12~
/'
SMHSF

'
'.]
'..

...

0\
I
__._us unc
~
0
0
2.0
1.0
SOIL PROFILE
s.
3.0
(T (sec)
Fig. 4.3 COMPARISON OF JAPAN's BSL AND US's UBC DESIGN EQ's (SIDRS) AT SAFETY LEVEL
~.4
1.2
II
! It
1.0
I
I
'
I
I
I~
''
II
~=0%
~ =2%
~=5%
~ = 10%
~
20%
~=50%
=60%
0.8
FOR~=
5%
0. 6
0.4
0. 2
0. 0
0.0
0. 5
1.0
1.5
2.0
PERIOD (sec)
2.5
3. 0
J
78OUCTIUTY=S
OUCTIUTY=2
E:LC:HAOAloiSIIoiOHRAZ
~1J,!C:~
0.8
c::
z~
0.6
o.a ................
. . .
!lli2.Qlli :.~~
~F!f_!oi~A_R.!\..H.A.t. .L_
.
I t~ ~~~
.
....................
i ;:. ..~;..:;..:.:.~:. .:....
0.6
u
c::::
:J
c...
::::E
0.4
~
0.4
<
w
Cl
0.2
0.2
o~r~~~~.~~~~r~~~~+~~~~~
0.1
10
0~.~~~~~T~~~~~~
25
0.1
10
25
FREQUENCY, Hz
FREQUENCY, Hz
Comparison of deamplification fac:ors for 2% damping.
OUCTIUTY=S
OUCTIUTY=2
ELC:HAOAio.ISIIoiOHRAZ
~1._!C:~
RIOOELLN(WI.IARK
c:
0
,.....
u
. .
~
,.,,::
::
:
: ; : : ::
       
:;:
NE:WI.IARKHALL
. :;;.:
/,
E:LC:HAOAI.ISII.IOHRAZ
0.8
.................
!,!1..,!C:~
RIOOELLN(WI.IARK
; ::
:~.:
......
: ; :::
:::::
~ lgrrj~_R.!\..~.L~1T: j :1;
;:
~,
0.6
t?u
c::
:J
c..
::::E
<
w
Cl
0.2
: ~ .
0~.~~~~~~~T~~r.~.
10
25
0.1
FREQUENCY, Hz
I=REQUENCY, Hz
10
25
791.0
0.9
0.8
C"'
I'll
VI
....
Damping
0.7
0.6
Transition Zone
to 10% Damping
5%
I'll
L.
CIJ
o.s
CIJ
u
u
<
0.4
I'll
....
u
Demand
Pt.
L.
0.3
iO%
~Damping
llJ
c.
VI
0.2
Pt. A
Capacity~
0. I
o.s
I
1.0
I
1.5
"1
i.S
2.0
Period, T (sec.)
0.8
SPECTRAL
ACCELERATION
Sa(g)
0.6 :I
/
0.5
SLEDRS
= 5%
~\
IMPULSIVE
EQ
IDRS
Sl
= 1.5, ~ = 6%
rHARMONIC
EQ
0.4
0.2

0.1 _:'
0.5
1.0
T(secs)
1.5
2.0
2.5
80
i vg
.
GROUND MOTION
/""'\
rv u
\.._)
v C\U /'\ . .
1 RESISTANCE
HYSTERETIC BEHAVIOR
RESISTANCEDEFORMATION
RELATIONSHIP
(a) HARMONIC 1YPE OF EQ
'
..
vg
GROUND MOTION
RESISTANCE
RESISTANCEDEFORMATION
RELATIONSHIP
(b) IMPULSIVE 1YPE OF EQ
020
I
I
I
I
I
I
I
I
I
Cl7
Q1
CoefficientQ10
....:.;:"
Zone c.;
I
I
1
Zon~
'
oos
I
I
I
I
I
I
I
I
.... I .
i a;s . . ac
I
...
Zone A
015
I
I
I
I
I
I
02
r.....

.,..._

1
I
I
I
I
I
I
T
T
04
I
I
I
I
~"
....
......
r
..,..,
.......

"'"';
~.J.
..:::::
I
I
r......
I """t T
 oc:s
075
065
I
I
ous
os
. I
12 ..
to
0
in seconds
'!
I
I
I
!'..
T
06
Period
I
I
I
I
1' I
:::::j,
I4co
SLEDRS: Ce
0.8
SIDRS: Cu
ZONE A
0.6
FLEXIBLE SUBSOILS
0.4
i
0.2
'....

"
0.33}
0.3
..:::_          0.026 }
0.024
cu
II
....... ' 
I
0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
(T (sec)
Fig. 4.10 NEW ZEALAND NZS 4203, 1984: SLEDRS (CJ AND SIDRS (Cu)
82
Cy
1.2 ~~
usc
l.O
1988
JAPANSSL
 ..........
...
0. 8
0. 6
NZS4203
,
.
...
...
0.4
...
.. .
... .. ..
. ...
... ...
~~~~:
MEXICO DF1987
, ~ 
o. 2
'
.........

0. 0
o.s
0. 0
l.O
1.5
2.0
2.5
3.0
PERIOD (sec)
INELASTIC DESIGN SPECTRA (SIDRS)
0 4
0 3
JAPANBSL
......................... .
0.2
usc
0. l
.. .
1988
NZS4203
. ..
....
...... ..... .. .
...


0 0
0. 0
05
1.0
1.5
2.
2.5
PERIOD (sec)
Fig. 4.11  COMPARISON OF SLEDRS AND SIDRS
30
0.'
SOIL TYPE 1
SHORT PERIOD
JAPANBSL
0. J
....::,;;o~;,;;~ llrnll)
...
......
....
0.2
::
. ......./
o.1
1988 usc
/ ./1984 NZS
~~t~ r' ,,... ..
,~
1987 MEX OF
I
I
...
...
..
.....
:;f./
0. 0
0. 0 0 s
0. 0
0. 0 10
o. a 1s
0. 0 20
0.025
Cy
0.16
. 0. 14
SOIL TYPE 1
LONG PERIOD
0.12
/.:.::(,;c;n;Ci;~um.umt>
JAPANBSL
.:
0. 10
/ .................
0. 0 B
0. 0 6
0. 0 4
./1984 NZS
'~~:
.
. r...~.~,9=8~8~U~a=c
a.o2
i98"7""MExi5i="
.'
0. 0
0. 0
s
'INTERSTORY DRIFT INDEX
0. 0 0 s
0. 0 10
0. 0 l
0. 0 2 0
o. a 2s
1
84a Sa
V
ca g  sw
0:1 (1985
0.40 1,_.;......;.
_ _'"'
0.30
01 (1976)
3
2
T (Seconds)
0.0
Oakland Harbor
1.2
DEG
   J.L
 J.L
 J.L
 J.L
 J.L
 J.L
1.0
0.8
0.6
0.4
I,
',/
'
= 1
= 2
= 3
= 4
= 5
= 6
"
....... ',,.,
~,,~~''  ,.
0.2
35
,.
....... 
., ..
""=:: . , ,
~
,
'' , ..
':.._..... _.,
' .....
""...
'
"
'
0.0
0.0
0.5
l.5
2.0
Period (sec)
1.0
2.5
3.0
85
A
0
'
~.:ID
dOdd 0
=5
RCT)!
R <TJ .1.
Ro
2
03R

To
1l
A
j
.0
dO
co
ON DUCTILITY
DESIGN
5. 1 GENERAL REMARKS
Usually the practice in EQRD is based on official code procedures. Thus, the stateofthepractice on J.L 6 and IDIbased EQRD has been judged by analyzing and comparing the
present EQRD codes of the United States, Japan, New Zealand, Mexico and Europe (CEB).
In judging the results of the analyses conducted and particularly the observations made from
the comparisons of the codes, it is necessary to keep in mind the following facts:
1.
The studies presented herein analyzed and compared the code regulations regarding
seismic forces and stiffness requirements.
(drawings), which are related to the degree and quality of inspection]; and (c) the
maintenance
or preservation
and
nonstructural system.
2.
The seismic forces specified in the code reflect the seismicity of the country and this
seismicity varies not only from one country to another, but within countries it also
varies from one region to another.
earthquake ground motions in the regions of highest seismic risk in New Zealand is
smaller than the severity of earthquake ground motions in the zones of high seismic
risk in Japan or the United States.
different countries cannot be the same. From inspection of the SLEDRS shown in
88Fig. 5.1, it is clear that the seismic risks in Mexico City and in the worst zone of New
Zealand are significantly lower than those corresponding to the region of highest
seismic risk in Japan and the United States. From the point of view of comparing how
ductility is used to reduce the required linear elastic seismic forces, the ideal would
have been to normalize all the SLEDRS to a spectral acceleration of l.Og.
the United States and Japan have similar required SLEDRS which are somewhat smaller
(up to 20% for T = 3.0 sees.) than the NZS.
5. 2. 2 Sites with Soft Soils (Soil Type 3). From the analysis of the SLEDRS plotted in Fig.
5.1 (b), it can be seen that, except for buildings with a T < 0.6 sees., the Japanese BSL is
the most severe spectrum. For T < 0.6 sees. the UBC is the most stringent. If the special
soil Profile S4 introduced in the 1988 UBC is also considered, then the UBC becomes as
severe as the Japanese BSL and becomes even more severe forT > 2.0 sees. It should also
be noted that the Mexican D.F. SLEDRS it is as severe as the Japanese BSL forT > 3.2
sees. Then for tall buildings with T > 2.0 sees. and up to T
soil (soft clay, UBC type S4 or Zone III of Mexico City), the UBC is the most severe
SLEDRS. For these tall buildings the CEB SLEDRS is the least demanding.
89
For the case of reinforced concrete SMRSF, the resulting SIDRS are
plotted in Fig. 5.1 (c). Comparing these SIDRS with the SLEDRS [Fig. 5.1 (a)] from which
they were obtained, it becomes clear that the largest reductions are those of the UBC,
which is particularly true for buildings with T < 1.5 sees. For T > 1.5 sees., the minimum
specified seismic coefficient specified by UBC
(c/~ ~
uses a constant reduction of 12/1.4 = 8.6. The Japanese BSL uses the smallest reduction:
1.0/0.3 = 3.3. For tall buildings with T > 1.5 sees. and on firm soil the most severe SIDRS
is the Japanese BSL, which, up toT = 3.0 sees., is more than 33% higher than any one of
the other SIDRS.
5. 3. 2 Sites with Soft Soils (Soil Type 3) The SIDRS at the yield level for building on this
type of soil are plotted in Fig. 5.1 (b).
SLEDRS of Fig. 5.1 (d) allows evaluation of the deamplification that is involved going from
the SLEDRS to the SIDRS.
Japanese BSL and amounts to a constant reduction using a coefficient 3.33 = 1/0.3. The
largest reduction is that recommended by UBC, which is 12/1.4 = 8.6. From the SIDRS
shown in Fig. 5.1 (d), it is clear that for tall buildings with a T > 1. 7 sees. and up to a T
= 3.0 sees., the yielding strength required by BSL exceeds by more than 30%, 82%, and
121% those specified by the Mexican D.F.,CEB, and NZS codes, respectively. The yielding
strength required by UBC for tall buildings having T > 2.0 sees. is the lowest one of all the
codes considered herein. Even when the UBC Special Soil Profile S4 is considered, still the
corresponding SIDRS require significantly smaller yielding strength than that of the
Japanese BSL, Mexican D.F.,and CEB regulations.
905. 4 STATEOFTilEPRACTICE
Although all of the EQRD codes reviewed herein have regulations limiting the maximum
IDI for certain limit states, none of these codes have recommendations regarding how the
specified limitations should be directly introduced into the preliminary EQRD of a building
structure.
were introduced at the preliminary phase of EQRD, however, in present practice (to the
best of the authors' knowledge) the IDI limits specified by codes are checked by analysis of
the already finished preliminary design of the structure. As will be discussed later in more
detail, there is a need to develop practical code procedures that directly include the control
of damage in the initial phase of the preliminary EQRD.
imposing as one of the main design parameters first a minimum lateral stiffness or a
maximum acceptable IDI at service level and then a maximum acceptable IDI at the safety
(collapse) level. To have an idea of how the different codes attempt to control the damage,
it is convenient to plot the loaddeformation diagrams according to the requirements of
these codes. This has been done in Fig. 5.2. To be able to discuss more specifically the
problem of how the maximum acceptable IDI can affect the design of tall buildings, the
code expected loaddeformation, or more specifically, the base shear IDI diagrams for the
different codes considered (analyzed) herein have been plotted for two different groups of
T: short T (T < 0.3 sees.); and long T (T > 1.6 sees.).
5. 4. 1 (a) Short T: From the results plotted in Fig. 5.2 (a), it can be seen that there is a
very large variation in what is considered to be minimum adequate lateral stiffness (or
maximum acceptable IDI at service level). Although, except for the Japanese BSL, none
of the codes considered herein clearly defines what is the strength that is required by service
EQ ground motions, it is clear that the New Zealand Standards is the code requiring the
largest lateral stiffness, and, therefore the one that will result in better damage control
under their service EQs. This is specifically true in the case where
91nonstructural elements can be damaged (because of the way that they are connected to the
structure). In this case the NZS require the IDI be smaller than 0.0006 which is near 1/2,
1/4, and 1/6 of those specified by CEB, BSL, and UBC, respectively. In the case of Mexico
D.F., the code requirements for the minimum lateral stiffness become very severe for the
case of buildings located in Zone III (Soil profile S3 , i.e. soft soil [Fig. 5.2 (c)]). For this
case the lateral stiffness required is practically the same as that required by the NZS. Why
are there such large differences?
nonstructural elements used in these countries are different: while New Zealand is using
very stiff partitions and claddings, the United States uses very flexible nonstructural
elements. However the authors believe that at present the differences in the stiffnesses 'and
damageability of these nonstructural elements cannot, on their own, justify the difference
in the maximum acceptable IDI specified by the codes.
investigated further.
5. 4. 1 (b) Long T For building with T > 1.6 sees. and on finn soils (Soil Profile S), the
code expected strengthIDI diagrams are plotted in Fig. 5.2 (b). From this figure it can be
seen that the results regarding the minimum required lateral stiffness (or maximum
acceptable IDI limits) are similar to those discussed above for the case of Short T. In the
case of tall buildings located on soft soils [Fig. 5.2 (d)], the Mexican code becomes as severe
as the NZS regarding the lateral stiffness requirements.
From the above analyses it appears that while the Japanese BSL is the one requiring the
largest strength for service EQ ground motions, as far as lateral stiffness is concerned, these
laws are the most relaxed of all the codes considered, particularly for the case in which the
nonstructural components cannot be damaged.
EQs, it would appear that under service EQs, the IDI should not exceed 0.003 when
nonstructural components can undergo the same amount of deformation; however, this limit
needs further research.
92
5. 4. 2 Maximum Acceptable IDI at IDtimate Limit States (Collapse). (Fig. 5.2). While
some of the codes explicitly specify the maximum acceptable lateral displacement (in the
case of Mexico D.F.), there are others that limit the IDI in an implicit way (in the case of
UBC). The Japanese BSL does not specify any limit, however, from discussions with the
Kajima research team it appears that in practice the Japanese designers limit the maximum
acceptable IDI to 0.01. The Mexico D.F. explicitly specifies that the maximum IDI should
not exceed the values of 0.006 and 0.012, depending on whether the nonstructural
components can or cannot be damaged. The UBC implicitly specifies that the IDI shall not
exceed the values of 1.5% in the case of buildings with Short T (less than 65 ft. in height)
and 1.125% in the case of buildings greater in height.
different values for the maximum acceptable IDI, depending on Seismic Hazard Exposure
Groups (SHEG).
For essential facilities (SHEG III), the IDI limit is 0.01 which is
significantly more stringent than the acceptable limits for the other two SHEG which is
0.015. However, these ATC recommendations clarified that for the case of buildings in the
SHEG I, i.e. structures of ordinary importance, the requirement of 0.015 can be relaxed by
as much as 113, resulting in an IDI of 0.020,provided that their height is less than 3 stories
and that there are no brittletype finishes.
The authors would like to emphasize that the above limits of 1.0% and 1.5% are a
consensus judgment from experience collected from observations and analyses conducted
on what occurred in previous earthquakes. Compliance with these limits will insure not only
human safety but also damage control. From experience gained in previous earthquakes,
it can be concluded that structures designed and constructed in accordance with present
UBC will in general be capable of undergoing lateral deformations significantly higher than
those corresponding to an IDI limit of 0.015 without collapsing, however the damage will
be very high.
deformations in: the main building of the Olive View Hospital during the San Fernando
earthquake
of 1971; the Imperial County Building during the 1979 Imperial Valley
earthquake; and the 21 story building in the Pino Suarez complex during the 1985 Mexico
City earthquake (although one of the 21 story buildings collapsed, the other remained
93standing with IDI larger than 1.5% ). It should also be noted that while the authors agree
with the UBC specification limiting the IDI to 0.0125, this limit should be connected with
a minimum required yielding strength. Unfortunately, the minimum UBC required strength
is believed to be too low, thus design of tall buildings that attempt to provide the building
structure with just this minimum strength in case of severe EQ ground motions will undergo
IDI significantly larger than the maximum acceptable by the code.
regions of high seismic risk, it would appear that while the minimum requirements for
yielding strength imposed by the BSL are the more realistic, the initial stiffness seems low,
suggesting the need for these laws to limit the maximum IDI.
SOIL TYPE 1
SOIL TYPE 3
l.O
1. 0
1987 CES
a.a
0. I
0.1
p9~~l'!~~\
.
0.(
....
'':.:_..
..
0.'
....
......
1987 MEX OF
....... _._
.'
..;"'

o.o ~~
0. 0
0.5
~.\
!
;~.
I
,:,:.....
0.2
1954 NZS \
0. (
~
1987 CE3
,...
\.
1. 0
LO
l.
'\
...~'
.....
0.2

.....
.
0. 0
o.o
05
1.0
PERIOD (sec)
(c)
0. J
(d)
SOIL TYPE 1
JAPANSSL
.
SOIL TYPE :3
JAPANBSL
0 .l
...
..
1967
..
,:::: ~z:v~.~.~:
1987 MEX OF
:::: ~:~:: .
0.2
c:::
l.O
:. 0
a.'
..
0. 2
1.5
PERIOD (sec)
a.'
....._ .
~,.
.
1967 MEX OF
a.1
. _____
~==~=~:~==
.:.:.:.~

a.o ~~
o. a
a.s
1.0
l.S
LO
l.O
a.o
a.o
a.5
l. 0
PERIOD (sec)
1. 5
PERIOD (sec)
2. 0
2.5
l.a
SOil. TYPE 1
SIIOIH PEiliOD
J/\1' ANOSl
o. I
.......
.......
a. u
a.
II
a.
ll
SOil TYPE 1
LONG PEiliOD
JAPANOSI.
/:.;(,;o,;;u~;.;;~.i;"'"'J
...
....
....
a. 1 o
.....
.
....
....
/./
0. 3
........./
1907 CEO
0. 01
/:'
rr~r
; i ../ ......
i i /i . :'
0. a1
.
.
0
... !.99.?..~~~gf....
0. 03
o.a
.o
o.o
a.aos
o .a 10
! i /.' .....
NlS
;'ri ,.,.:'.' /.
I
~~L'?,O_I\_tllL
0. I
0.
o IS
0,010
o .ozs
; iJ //./
o. a 10
o. a IS
a.aos
'INTEnSTOI'lY DlliH INDEX
o.o
0,15
1900 UDC
rl.' ~,:
(d)
190~
~.
Cy
:rt.~.,'
a. al
1900 IJOC
0 ,OJO
'.OJ s
r,
0 .I
SOIL TYPE J
SIIORT PEI!IOD
a. 1 a
0. 15
o. I o
0. OS
SOIL TYPE J
LONG PERIOD
a. o
o.o
o.a
o.oos
0.010
o.o 15
o.ozs
a.oas
a.ola
a .o IS
a .on
INTI!IlSTOilY DRifT INDEX
a.aJS
\0
0\
I
GNEN:
1. Function of building.
2. Site of building.
REQUIRED:
SOLUTION:
with records
obtained at sites with similar site topography and local soil (geotechnical)
conditions located in tectonically similar regions and if necessary to generate
them
through
relationships
and applicable
982. Based on information collected above under item 1., to establish the
serviceability design EQ in the form of a SLEDRS
displacement (IDI).
3. Based on item 1., to establish the design EQ for the ultimate limit states
considering not only human safety but also control of damage in the forms of
SIDRS for strength Cy (in function of J.L 6) and for maximum acceptable IDI.
It should be noted that the value to be adopted for acceptable J.L 6 and IDI will
depend on the function of the building. For very important essential facilities,
the value of acceptable
J.L 6
the procedure
for the
preliminary EQRD of the structure is similar to that described in Ref. 5. Thus, what is
needed is to develop reliable information described above under items 1, 2 and 3. The
following is a brief discussion of the research and development needs to improve EQRD of
buildings.
6.
RESEARCH
AND
DEVELOPMENT
NEEDS
TO
IMPROVE
THE
ESTABLISHMENT OF SLEDRS
6. 2. 1 SLEDRS for Service Level EQs. As has been previously mentioned, except for the
Japanese BSL, none of the codes reviewed herein specify or recommend the use of design
EQ at service level, i.e. their design procedures are based on the use of only one design EQ
which is established at ultimate limit state. There is an urgent need to develop a reliable
preliminary EQRD procedure based on twolevel design, where the following two limit
states are considered:
during ground motion that is expected to occur several times during the life of the structure;
and then survivability and control of damage under a rare but possibly severe (extreme) EQ
99ground motion.
Although these needs were recognised several years ago [4, 5, 10], they
have recently been emphasized in the U.S. [51, 52]. To be able to develop reliable
procedures such as twolevel EQRD, it is necessary to conduct statistical and probabilistic
analyses of data available regarding what can be considered service EQ ground motions,
then to develop reliable SLEDRS that consider the LERS of all these ground motions.
Because reliable measured data on EQ ground motions are scarce, design spectra are
currently formulated using insufficient statistical information. Data from records of severe
ground motions of earthquakes that have occurred during the last seventeen years has
altered the previous statistical base so dramatically that drastic changes in the LEDRS and,
therefore, in the codespecified C, have been required. Examples of such ground motions
are: the 1968 Hachinohe earthquake, the 1971 San Fernando earthquake, the 1978 Sendai
earthquake, the 1979 Imperial Valley earthquake, the 1985 Chilean and Mexico City
earthquakes, the latter being perhaps the most dramatic, the 1986 San Salvador earthquake,
and the 1989 Lorna Prieta earthquake. Until 1971, the recorded NS component of the 1940
El Centro earthquake was considered as representative of extreme EQ ground motion. The
records obtained during the 1971 San Fernando Valley and the 1979 Imperial Valley
earthquakes demonstrated, however, that the damage potential of this El Centro component
was very low compared with that of some of the recorded San Fernando and Imperial Valley
motions.
6. 2. 2 SLEDRS for Survivability For this limit state, it is necessary to collect and/or
develop reliable data regarding the most severe EQ ground motions that can occur at the
site of the building during a certain established period. To achieve this it is necessary first
to answer the following question: "What is the right engineering parameter for defining the
severity of any given (recorded) EQ ground motion?" This question had been analyzed in
the studies reported in Refs. 25 and 26, which conclude that the best parameter to measure
the damage potential of an EQ ground motion seems to be its Energy Input,
expected hysteretic energy,
, or the
demonstrate very clearly that EQs such as the 1940 El Centro (which is usually used as a
100
MCEQ to check the safety of designed structures) have a damage potential to structures (as
measured by its
EJ
is illustrated in Fig. 6.1. Furthermore, as illustrated in Figs. 6.2 and 6.3, the SLEDRS
assumed by the 1985 SEAOC (which is the one used in the 1988 UBC) and ATC 306 are
significantly smaller than the LERS corresponding to the recorded ground motions during
the 1985 EQs in Mexico and Chile, and during other EQs such as the 1986 San Salvador
and the 1971 San Fernando EQs. Thus, if such ground motions were to occur in the United
States in the future, the values of the SLEDRS adopted by present United States code
requirements will significantly underestimate the response that could occur. It should be
noted that similar conclusions can be drawn if the LERS of these ground motions are
compared with the SLEDRS spectra adopted by the Japanese BSL as well as any of the
other codes considered herein.
6. 3. 1 SIDRS for Strength. C,. Ideal Solution: For any given site the ideal solution would
be to derive the SIDRS directly from statistical and probabilistic analyses of the IRS
corresponding to all recorded motions at the selected site or at similar sites located in
tectonically similar regions, and even of records derived through the use of theoretical
consideration.
The IRS for the selected ground motions should be computed taking into
account the different (hysteretic) models according to the expected behavior of the type of
structural system and materials to be used. Procedures and techniques for computing these
IRS have been discussed in detail in Refs. 24 and 27. There are a large number of such IRS
presently available; a few of these IRS illustrating the variation of the resistance coefficient
Cy with the ductility ratio
J.1.
1. The shape of the IRS (i.e. the variation of Cy with T) varies significantly depending on
the predominant frequency (or period
TJ of the recorded
101
depends on the site conditions (soil profile and topography) in which the record was
obtained [compare the IRS for the Pacoima (rock) record with the Hollister (firm
alluvium) and that of the Mexico (very soft clay)].
period
(The explanation
for this
deamplification has been given in Ref. 24). The longer the T, the larger seems to be
the deamplification.
3. The degree of reduction of the LERS (J. = 1) due to J. > 1 decrease for T #= T, and
tends to zero as the T of the structure tends to zero.
From comparison of the IRS presented in Fig. 6.4 with the SIDRS adopted by codes which
are given in Fig. 5.1, it can be observed that:
1.
For sites on firm or medium stiff soils there are already several recorded ground
motions whose IRS exceeds the SIDRS adopted by the codes reviewed herein, even
the one adopted by the Japanese BSL, which is by far the most conservative one.
These are the cases of IRS for the Chile, Pacoima, Derived Pacoima, Corralitos,
Capitola, and Hollister records. Note that this is true even in cases of J. = 6 which
is very difficult to achieve and/or to justify its use.
2.
For soft soil sites (soil profile 53 or 5 4), particularly with soft clays whose depth exceeds
40 feet, from the IRS corresponding to the recorded ground motions in Mexico City,
it appears that the SIDRS corresponding to the Cr adopted by the codes will be
exceeded, except for the case of the Japanese BSL for low and medium rise buildings
(up to 20 stories), even when a ductility J. = 6 could be developed and used. This
conclusion would be particularly true in the case of soft clay which can resist and
102therefore transfer peak ground acceleration up to 0.30gto the foundation, which seems
to be the value considered by the Japanese BSL for soft soils.
6. 3. 1 (a) Code Procedures to Determine SIDRS for Cy. As has been discussed earlier, the
SIDRS for Cy specified by codes are obtained by deamplifying the LEDRS through the use
of a reduction factor R which depends on !J. 6 Some of the authors have been analyzing the
rationale for the R code values since they were introduced in the codes. As discussed in
more detail in Refs. 8, 9, 21 and 36, it is very difficult to judge the rationale for the values
recommended
for the values of R that ATC 306 and the values of R,.. that SEAOC
(adopted by 1988 UBC) have recommended for reducing the SLEDRS to the recommended
SIDRS. These difficulties arise from the lack of discussion or even any indication of how
these values have been derived and what they are meant to physically represent. In Chapter
4 of the A TC 306 Commentary, it is stated that R "is an empirical response reduction factor
intended to account for both damping and the ductility inherent in the structural system at
displacements
displacement of the structural system." In evaluating this statement, it should be noted that
the SLEDRS selected by ATC is already based on 5% damped LEDRS.
Therefore, the
equivalent viscous damping expected in clean structures should not be significantly greater.
If the values of R and R,.. are assumed to depend only on !J. 6 , then the studies reported in
Refs. 8 and 9 clearly demonstrate that for any selected resistance
function~
damping ratio,
and ductility, the reduction factor varies with the period of the structure, decreasing as T
decreases.
This has been clearly confirmed by the results shown in Fig. 6.5 which, for
From the above discussion it appears that the recommendation of a constant value for R
(or R...), i.e. that the value be independent ofT for the structure, cannot be justified solely
on the basis of the ductility built up in a structure. The values recommended for R (or R,.)
appear too high, particularly for structures with a T < T, (the smaller is the TIT, ratio, the
103smaller is the reduction), if the designer attempts to design the structure with just the
strength required by the code, based on the use of the specified or recommended values of
R (orR,.,). Fortunately, as shown in previous publications [8,9,2l],the resulting code design
generally produces a significant overstrength, and the shorter the T of the structure is, the
larger this overstrength seems to be.
"The
response modification factor, R, and ... have been established considering that structures
generally have additional overstrength capacity, above that whereby the design loads cause
significant yield." The authors believe that this overstrength, OVS, together with builtin
toughness is a "blessing" because structures designed according to presently
specified design seismic forces (UBC or recommended ATC values) would be able to safely
withstand MCEQ shaking. The first "significant effective yielding" of properly designed
(sized), detailed, constructed and maintained highly redundant structures with relatively
short Tis not only considerably higher than that on which the code design is based, but such
structures also have a significant overstrength beyond their first effective yielding. The
resulting overstrength usually totals two to three times the minimum codespecified effective
yield strength.
"it is very difficult to rationalize (justify) quantitatively the values recommended by ATC for
R." If the value of R alone is used in the design of reinforced concrete framewall dual
systems, i.e. without any other requirements, the resulting design will not be reliable. The
use of a specific value for R should be tied to other requirements.
104concrete ductile momentresisting space frame members and structural walls. The authors
believe that this is not enough, and suggest that the preliminary design using an ATerecommended approach (or that of the 1988 UBC) be subjected to a limit analysis to obtain
an estimate of the actual maximum resistance of the structure as it will be constructed, and
that a value approximately three to five (depending on the structural type and fundamental
period T) times the minimum yielding strength required by ATC be ensured. Furthermore,
the design of walls (sizing and detailing) against shear (as well as against members of ductile
momentresisting space frames against shear) should be based on their maximum resistance.
= 5.5 and
= 5%)
(represented by Cy) significantly higher than that required by ATC (represented by C.) or
a strength capacity [defined in the figure as Overstrength Factor (OVF)Rcquircd
= required
Cy I ATC's C.] significantly higher than that required by the ATC 306 provisions.
Reference 25 shows that a structure with T
the minimum required resistance (C.) by the ATC 306 provisions is required to develop
ductility displacement ratios, J.L, well beyond the value usually considered as acceptable (J.L
= 5).
All the above results have been confirmed by the studies that are being conducted on the
recorded ground motions during the 1989 Lorna Prieta EQ [36]. The plot presented in Fig.
6. 7 can also be interpreted as giving an indication of the overstrength needed beyond the
first significant yielding if the design is conducted by using 1988 UBC. For example, if we
assume a RCSMRSF, the R corresponding to first significant yielding, Ry, rather than the
UBC specified R,., , would be R.)1.4
= 12/1.4 = 8.6.
= 8.6 on
the plots ofR .. of Fig. 6.5,as it is done in Fig. 6.8,the differences between this line and the
R,. (corresponding to the J.L considered) can be considered as representing the required
overstrength (OVS)Rc<juircd.
is assumed first that the spectral ordinates are representing the spectral acceleration in
terms of g; thus, they are equivalent to the seismic coefficient C.
Secondly, it is also
105assumed that the code specified (SLEDRS)cooE is equal to the actual LERS of the critical
EQ ground motion. The reasons why the plots of Fig. 6.8 have just given an approximate
idea and not the accurate representation of the required overstrength follow.
Rovs
= ~  R .. )IR .. = (OVS).pc<l,./(SIDRS)cooE
It should be noted
that strictly speaking the (LERS)AcruAL is not the same as the (SLEDR)yeooE . Thus when the
actual LERS of a critical ground motion, or the statistical evaluated (mean or mean plus
one standard deviation) strength demanded for all possible critical ground motions in the
form of a (SLEDRS)AcwAL , differs from the specified code, (SLEDRS)cooE , the use of the
plots of Fig. 6.8 to evaluate
Rovs
in the following examples. First, consider the case of a structure with a T tending to zero
seconds. According to the plots of Fig. 6.8 the
however, because as T tends to zero the required linear elastic strength, i.e., the
(LERS)AcruAL , tends to the value of the peak ground acceleration as specified by the code
in its (SLEDRS)cooE . In fact, as illustrated in Figs. 6.10 and 6.11, the
= 3.1,and not 8.6. Thus the
Rovs
(~)To
1.0/0.32
code specified (SLEDRS)cooE does not agree with the (LERS)AcruAL or (SLEDRS)AcruAL the
plots of Fig. 6. 8 cannot be used to evaluate required overstrength. Better representation of
required overstrength and of
Rovs are illustrated below for two cases: (a) the case of rock
or firm alluvium sites; and (b) the case of very soft soil sites.
6. 3. 2 (a) The Case of Rock and Firm Soil Sites The expected normalized spectra are
illustrated in Fig. 6.10. From the analysis of the graphs of Fig. 6.10 it appears that, in the
case of structures located on rock or firm alluvium having
J.1.
largest values for structures with T in the range of 0.1 to 0.5 sec. If the structure develops
a
J.1. ~
=0, 2 sees.
Considering an
effective peak ground acceleration of 0.40 g., i.e., Z = 0.40, the maximum values of OVS
are about
106
ovs
~.
= 6, the
OVS
3.6
0.600.13=0.47
0.4 0.13=0.27
= 2.5
= 2.1
Rov
the ROV
= 0.6.
6. 3. 2 (b) Case of Very Soft Soil Sites In the case of deep deposit (larger than 40 feet)
very soft, highly compressible clay, the OVS spectra are quite different from those for rock
and firm alluvium. Fig. 6.11 illustrates the OVS for EQ records on soft soil having a T,
(predominant period of soil) equal to 2 sees. [36].
From Fig. 6.11 it is clear that the normalized specified code (SLEDRS) for values of T
between 0. 8 sees. and 4 sees. underestimates the values obtained from recorded ground
motions. However, as UBC for Zone 4 specifies a Z
in the normalized
ordinates. However, it is clear that for periods around 2 sees. the underestimation is more
than two times. As a result, the EPA is required to be smaller than 0.20 g. As there have
already been. recorded ground motions in such types of soils with EPA significantly higher
than 0.20g. (actually, up to near 0.30g.),it is believed that the code specified spectra when
combined with the Z = 0.4 (which seems to be conservative for soil type S4) is not
conservative for structures with T near the T, of soft soil. From Fig. 6.11 it can be seen that
the required normalized OVS and the
Because the code specified (SLEDRS)cooE differs significantly from the (SLEDRS)AcruAL
.. 1.,, ,
of~
<MEAN
are quite different from the code value of 8.6. Also, as the actual
values of R .. plotted in Figs. 6.11 and 6.8 are obtained dividing the actual values of the
(LERS) by the (IRS) for different values of JJ., it is clear that computation of Rovs from Eq.
I
I
= (Ry
= 8.6 and
of R .. given by
There are already many recorded ground motions with (LERS)AcruAL for a certain range of
T which exceeds the (SLEDRS)cooE Examples of these cases are shown in Figs. 6.3,6.7and
6.11. Figures 6. 7 and 6.11 show not only that the spectral ordinates of the code (SLEDRS)
can be significantly exceeded, but that even if a
p. 6 =
accepted (which is doubtful because of the damage that will be involved with such high
displacement ductility ratio) there is a nee.d for significant overstrength beyond the strength
required by Code, particularly for structures with relatively short T, in order to have safe
design. From the results plotted in Fig. 6.5, 6.10, and 6.11, it is clear that the longer the
value of predominant period of the ground motions, T,, the larger the range of the period
of structures for which significant overstrength is required.
obtained on soft soils, and particularly those obtained in Mexico City, it would appear that
if very soft clays could resist and therefore induce to the building foundation EQ shaking
with peak acceleration equal to 30% g, the actual IRS will exceed those specified by present
codes, i.e. the (SIDRS)cooE , including the the BSL code for tall buildings.
Lessons learned from analysis of performance of buildings during recent destructive EQs
and results from recent research indiCate that lowrise buildings usually have large
overstrength with respect to that required by the U.S. codes the taller the building the
smaller the overstrength.
mediumrise buildings (particularly those located on sites with very soft soils) are the ones
that have to be suspected of becoming a serious threat to life and/or of incurring large
economic loss in the case of a major EQ.
6. 3. 2 (c)
Recommendations
requires improving first the reliability of the specified SLEDRS and then the reliability of
the specified values of R. Therefore, it is recommended:
appropriate instruments, networks and array to record strong motions in the free field
and at the foundation of structures.
processing and probabilistic methods of analyzing strong motion data and quantifying
seismic hazard.
2. To develop more reliable methods for estimating the values of R. This requires a
more precise definition of R. The definition illustrated in Fig. 6.12 is proposed as a
basis for improving the evaluation of R [21, 36].
regarding possible critical ground motions at any given building site, it will be possible
to obtain the SIDRS directly from the recorded motions, so that there will be no need
to specify the R" . Therefore for the proper use of the definition of Fig. 6.12 in
evaluating reliable values of R, what remains is an urgent need for calibration of the
real strength of structures that have been designed according to present code.
Besides the factors indicated in Fig. 6.12, there is another parameter which can affect R.
One of these extra parameters depends on the soilstructure interaction, which can increase
or decrease the severity of the recorded ground motions at the free field, which are usually
the records considered in the determination of the SLEDRS as well as the SIDRS. Another
parameter is the change in the period of vibration, T, with the degree of damage: as
J.L
increases, the values ofT will increase, and might affect the response.
It has to be noted that all of the above recommendations have been made to improve the
use of J.L 6 in defining reliable SIDRS for preliminary design. This improvement is necessary
but not sufficient.
The damage and failure of a building does not depend only on the
strength given to the structure and the intensity of the ground motions. It also depends on
the number of cycles of inelastic deformations that the structure is required to undergo
during its response, particularly the number of inelastic deformation reversals, which
depends on the duration of strong motion. Thus, it is necessary to define a way to include
the effects of duration of strong motions. These effects will be studied in Task #2 of the
overall research.
Recently conducted studies show that the selected acceptable J.L 6 should
109depend on the duration of the strong motions. The most promising method for including
the effects of duration of strong motions is through the use of an energy approach.
6. 3. 2 (d) SIDRS for Lateral Displacement and IDI. As has been previously discussed,
practical and simplified methods have been developed to estimate lateral drift [33  35].
These methods are based on the assumption that except for buildings with T 1 I T,
< 1, the
nonlinear lateral displacements are similar to the linear lateral displacements. From studies
conducted by the authors [36], it appears that the nonlinear displacements are very sensitive
to the dynamic characteristics of the ground motions and of the structure, and they can be
significantly different than those obtained based on linear behavior as it is illustrated in Figs.
3.5 and 6.13. Note from this last figure that in the case of ground motions with long T,, as
those occurring in very soft soils (the case of Mexico SCT record), the nonlinear
displacement can be significantly smaller (nearly 50% smaller) than the linear displacement
for structures with T 1 close to the value of the T 8 On the other hand, for values T < 2/3T,
== 1.4 sec., the nonlinear displacements are significantly higher (in some cases 3 times
higher). Thus, the authors believe that the ideal solution for finding reliable SIDRS for
lateral displacement is to estimate the IRS for displacement through time history nonlinear
analysis for all the possible types of critical ground motions that can occur at the selected
site and for different SDOF models and then derive a SIDRS through statistical and
probabilistic analyses of these IRS.
Based on the derived IRS, it is possible to estimate lower and upper bound IRS for the
interstory drift for MultiDegreeofFreedom Systems (MDOFS). A lower bound is obtained
by assuming constant IDI through the entire height of the structure, and an upper bound can
be estimated by assuming a soft story [25, 36]. Lower bounds of IDis IRS for different
recorded ground motions and considering
building site, once all the possible critical EQ ground motions have been selected and their
IRS for the IDI have been obtained, it is possible through statistical and probabilistic
analyses to derive a lower bound SIDRS for the IDI of MDOFS. This SIDRS can be used
for the preliminary design of a MDOFS.
110amplification factor by which the SIDRS' lower bound should be multiplied to take into
consideration the fact that in the response of MDOFS it is very difficult to achieve a
constant IDI through the whole height of the building. The taller the building (i.e. the
longer the T), the larger the amplification factor should be. Thus there is an urgent need
to conduct integrated analytical and experimental studies to find an empirical expression for
TABLE 6.1 VALUES OF NORMALIZED OVS AND OF Rovs COMPUTED FROM FIG. 6. 11
J.L=2
(sees)
0.9
J.L=4
1.55  0.32
1.23
(1.23/0.32 = 3.84)
1.10  0.32
(0. 78/0.32
= 0.78
= 2.43)
J.L=6


0.80  0.32
0.58
(0.58/0.32 = 1.81)
I
2.0
3.0
1.40  0.10
(1.22/1.18
= 1.22
' 0. 75  0.18
6. 77) . (0.57/0.18
=
0.80  0.14 = 0.66
(0.66/0.14 = 4.71)
= 0.57
= 3.16)
0.50  0.14 = 0.36
(0.36/0.14 = 2.57)
= 0.32
= 1.78)
0.45  0.14 = 0.31
(0.31/0.14 = 2.21)
0.50  0.18
(0.32/0.18
E1 /m (inch 2 /sec2 )
Chile
20000
Mexico
20000
u2

u4

15000
u2
uG
15000
10000
10000
5000
5000

u4

uaG
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.0
0.5
Period (.sec)
1.0
2.0
1.5
2.5
3.0
Period (.sec)
I
El Centro
......
......
San Salvador
VJ
I
20000
20000
15000
15000
uG
10000
10000
5000
5000
....
o.o
0.5
. 
1.0
L5
2.0
2.5
3.0
o.o
0.5
....~
...
1.0
1.5
.......
~.
2.0
Period (sec)
Period (.sec)
2.5
3.0
114
2.4
2.0
CHILE
MEXICO
( N 1 OE COMPONENT
AT LLOLLEO}
(EW COMPONENT
AT SCT)
1.6
0.4
0.0
2
T {Seconds)
Spa (g)
3.0
,,,,"
I I
; :__Pacoima
I
2. 5 
(5% Damping)
,,'
I
I
.,
II
II
II
2.0
1'
II
II
~~~
'I I
I I I
\'I
I
I I
I
I'
'v''
I,# '
I
San Salvador
1.5
I
,_..
,_..
Ul
1.0
0.8
0.5
0. 0
0.0
0.5
1.0
1.5
2.0
Period (second)
2.5
3.0
116
Deri.ved Paco:i.ma
Mi.yagi.
l.2
L2
ul
u2
l.O
o.a
I
I
0.4
0.2
u4
uS
,,'
o.e
u'
1
1
,' , ...\
0.,
....
,'
'.12
l.O
ul
..;?~~~~~:::,;.~==
o.o
o.c
0.2
o.o
o.o
0.!
l.O
l.!
PecJ.oc
c,.
2.:1
2.0
3.0
0.0
0.5
1.2
2.0
1.0
PerJ.oQ
(secl
~'
    p2
  ~J
,.2
,.s
,.
1.0
   p4
0.1
0.1
 ,_,
 p!t
o.c
3.0
p1
1.0
2.5
(secl
o.c
0.4
o.c
0.2
o.:z
o.o
o.a
o.o
0 .5
1.0
1.!5
J.O
2.!5
l.O
a.o
0.!5
Pec.l.od (sec I
Cy
1.0
2 .5
2.0
1.'
J.O
Period (SeC I
Cy
_,_,
l.:Z
1.0
 p2
pl
1.0
0.1
___ ,_,
0.1
1.::1
_,_.
Bid~.
__
,_,
,.2
Chan. 3
_,_.,_,
pl
 p5
o.c
Oalkand 7St.ory

_,_,
o.c
0.4
o.c
o.:z
O.l
o.o
o.o
o.a
0.!5
1. 0
1.!5
Pec.l.od (SeC I
:z.o
:z.,
J.O
o.o
0.!5
1.0
:.a
1.!5
Pedod <cl
(~=5%)
2.!5
l.O
117
Cy
Ch.ile
1.2
1.2
l.O
1.0
o.a
o.a
0.6
0.6
0.4
0.4
0.2
0.2
0.0
Mex.ico
ul
u2
ul
u4
...,
uS
0.0
o.o
0.5
l.O
1.5
Per.1ocl
Cy
2.0
2.5
3.0
0.0
0.5
(sec)
l.O
2.0
Per.1ocl
Cy
El Centro
1.2
San
2.5
3.0
2.5
3.0
(sec)
Salvador
1.2
ul
u2
uJ
u4
u5
l.O
l.O
... ,
o.a
o.a
0.6
0.4
0.2
0.0
0.0
0.0
0.5
l.O
1.5
Per.1ocl
2.0
2. 5.
3.0
0.0
0.5
(secJ
1.0
1.5
Per.1od
Cy
Paco.ima
1.2
1.2
l.O
1.0
o.a
o.a
0.6
0.,
0.4
0.4
0.2
0.2
2.0
(sec)
Taft
ul
u2
ul
u4
u5
... ,
0.0
0.0
0.0
0.5
l.O
1.5
Per1ocl
2.0
(secJ
2.5
3.0
0.0
0.5
1.0
2.0
1.5
Per.1od
(secJ
J.L (~
= 5%)
2.5
3.0
..
,
118
Hollister
0 DEG
0.8
 J.L = 2
 J.L ... 3
 J.L = 4
0.6
   J.L = 6
0.4
0.(
0.2
O.J
o.o
o.o
o.s
9 0 D E:G
.u=l
l. 0
p.6
0.6
Capitola
l.:l
p. = 1
 J.L = 2
   J.L .... 3
 J.L  4
p. .... 5
l.O
Cy
.u = 5
o.o
l.O
l.S
2.0
Period (5ec:)
Corralitos
2.5
0.6
1.5
2.0
Period (sec:)
1. 0
Santa Cru:z;
=a
=
=
....
....
   J.L ....
o.a
o.s
0.0
DE:G
   p.
 p.
    p.
 p.
p.
l.O
3.0
J.S
3. 0
90 DEG
.u=l
1
2
3
4
=2
=3
= 4.
J.L = 5
l.O
5
6
o.a
 J.L
   J.L
 J.L
0.6
   J.L
0. 4
0,(
0,2
0.2
0.0
0.0
0,0
o.s
1.0
l.S
2.0
Period (sec)
J.S
o.s
o.o
3.0
Emeryville
l.S
2.0
Period (sec)
l.O
350 DE:G
   J.L =
 J.L =
   J.L =
 J.L
l.. 0
o.a
1
2
3
4
 J.L = 5
J.L ... 6
0.6
0.4
0.2
o.o
o.o
o.s
l.. 0
1.5
::.o
2.5
3.0
Period (sec)
2.5
3.0
RJL
1940
EL
NS
C~NT!tO
   1' = 2
 1' = 4
 1' = 6
12
10
,.
.'
I ,, \
,', , ,
....,  .._,....,
,
,a. ;: I :... ,, ,.
; ':
.. './
(.
..~
J.1"""2
i i
.'\..\
....
90 DE:G
10
"
Capitola
RJ.L
14
.,.
.........  .............. . . .
! : ~\
I
! ... i
! :. ,
... ...'.: ..
....
~ I
...,.
1::::1 . .
 J.L
1::1
,, '
.. :'i .:,....,,
... .:~
.....................
.
....
~.
,'
J.L
! ~ ,., _.'\
_,
. \ \ .' . . ,:
, , ,
"'
'
............
! ,., \
,,
\
! ."
......
0. 5
0.0
1.0
2.0
1.5
0. 0
3.0
2.5
2.0
1.0
1.5
Period (sec)
0.5
Period (:sec)
~.s
3.0
I
1985 MEXICO
RJ.L
EW
SC~
Santa Cruz
90 DEG
\0
I
14
12
JL=2
. Jl = 4
 f.L
10
_,,_
'
~
'
J.L2
..
.~
.~
'.
.. .. \
. .. ...
.... ...,
..
.' ,
J.L
 f.!
'
Cl ..
1::1
,4
I I
,. \/\' ' .,
::.:;
; i ,,
',I
,
,
,
. '._ ......
.:;._.,..... __..... ,,
...........
2
0
0
00
o.s
1.0
1.5
2.0
2.5
3.0
o.o
0.5
1.0
1.5
2.0
Period (sec)
Period (:sec)
2.5
3.0
llolllster
RJ.L
DEG
10
Emeryville
,,
, '
,,.,..
'
.. ~'
'
\\
\\
~ ~
:
; .:
,.,.I
.....
. .
.1'!
,..'""'',
! \ ... ,. ,_,
,...
'
;' !
... ..
....,
..
I
'
......\
 ....,,
....................... .
 J.l = 4
'\  J.l = 6
~'. \
..
\ '\
p=2
. J.L=4
,.,.,  J.l = 6
10
J.L=2
J50 DEG
.'..,',,,
0.0
0. 5
2.0
1.0
1.5
Period (sec)
2.5
0.5
0. 0
J.O
1.0
::z.o
1.5
Period (sec)
2.5
J.O
.....
I
Corralitos
0 DEG
Oakland llarbor
   J.L = 2
........ fl = 4
 J.l = 6
,. ,
J.L=2
J.l = 4 _, .......... .
i i
I
,
,. i  J.L = 6 ;
,
b.
_i
\
I
I \ 1 ,,
.. \~.
.j : ,
!~:
I
' ..... ...... ........
;, :
~.
;; ..
'
.
. ! .
.
'... .... :
1'.,,
. ..
.
. ..... .
0I
JS DEG
. .....
1.,,i ....:
0.0
0. 5
2.0
1.0
1.5
Period (sec)
2. 5
J.O
i':
I.
0. 0
0.
1.0
1.5
2.0
Period (sec)
2.5
J.O
Resistance Coef.
c;
1. 0 ,               . .
Chile
0.8
L O r      Pcol
,
 t 1 Cent r o
...! ......
Resistance Coef.
Hlco
0.8
Tart
0.6
0.6
ATC (R 8, Cd 5. 5)
0.4
    Hlya9l
0.4
0.2
0.2
.  ........
0. 0
0.0
0.5
1.0
1.5
2.0
.. 2.5
3.0
o. o L2:::::~;;;;:;;:~~
0.0
Period (sec)
0.5
1.0
1.5
2.0
2.5
3.0
Period (sec)
....
....N
I
OVF!,(.t)
6..
Chile
OVF(IIf'd)
Centro
Pcol"
Derived Pacol1u
Taft
    Hlya91
2
0 i.LL<'.LLJ.:LU:'.L.Lt:.ZLL:LL.t:."LLL.'LLLLL.'LALLLL:LJ
0.0
0.5
1.0
1.5
2.0
Period (sec)
2.5
3.0
o.o
0.5
1.0
1.5
2.0
2.5
Period (sec)
Fig. 6.6 REQUIRED SEISMIC COEFFICIENT (Cy) AND OVERSTRENGTH FACTOR (OVF) FOR SDOFS
DESIGNED IN ACCORDANCE WITH THE ATC FOR S1 AND ASSUMING 1' = 5.5 AND ~ 5%
3.0
122CORRALITOS
CllAN.
c=s%
2.0
1.5
1.0

o.o
0.0
0.5
1.0
l.S
2.0
PERIOD (sec)
HOLLISTER
CIIAN.
2.5
J.O
c=s%
1.2
l.O
0.8

0.<(
0.2
0.0
o.o
o.s
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
123
liS
.1.! ==
!2
10
f.l.6
Capitcla
90 D&C
ooooooooJ.,; 21
806
=~
 fJ.
lO
R..,..
.u""
=~,
806
,,
/ f\ \
 p. 
0'\..,0.~;/./ '\..ooooo/to<:::~o.
<::::::
::z
o.o
1.5
:.a
?ed.cci (sec)
:z.s
l.Q
0.5
3.0
3.a
0.0
Pericci {aec)
R..,.
Sa~t&
'0 D&C
Cru:
117
:z
0
o.a
a.5
R"
Uollister
10
0 DEC
806
o.o
3.0
1.5
Pez:icci {sec:)
0.5
l.O
:z.o
Period {sac)
&maryville
R"'
l.S
3.0
lSO DEC
10
.u 2
\ \
0IJ. .. 6
':.
"
\.\
\ ",
.... '.
................""'.

~:
OoO
0.5
1.0
loO
Pericd (sac)
C:or:~lito
10
l.5
o.o
J.O
0.5
l.O
1.5
l.O
J. 0
Period (sac)
R..,.
D!:C
10
8.6
.u 2
!==......... fl. .. 4
Oakland uarbor
J5
DEC
8.6
.us
0. 0
1.0
lo 0
l.5
Pe:icd (sac:)
J.O
0.0
0 5
0
1.0
1.5
;.o
:.s
J.O
Fig. 6.8 VARIATION OF RJl AND OVERSTRENGm ru(8.6 RJl)Wim T AND J.t
124PROBLEM:
GIVEN:
REQUIRED: Assuming that the SDOFS has been designed on the basis of a
(SIDRS)yeoDE , find relationship between the reduction factor due to
overstrength, Rovs , and the overall design reduction factor, Ry , and the
reduction factor corresponding to the energy dissipated by plastic
deformation, RSL
Defining
R~'
and R 5 as follows:
(LERS)ccM=R~'[IRS)ccM=R,.JRs(SlDRS),coDE] ( 6 l)
Then as overstrength spectra,
= (IRS)cGM  (SIDRS)ycODE
(OVS)spectra is (OVS)spectra
and defining
OJ'S
Defining
(OVS)~crra
(SIDRS)
R,
yCODE
  (6 2)
Rs 1
(LERS)CGM (6.3)
(SlDRS)yCODE
T (sees)
,'
Rovs
RyR"
(6. 5)
R"
(SIDRS)y CODE
= (SIDRS).,. CODE
1(SIDRS~~~D"E      __ .
(1.4)
Then
T (sees)
125
NORlvfALIZED C
4
(SLEDRS)
code
soil rype 54
0.0
0.5
1.0
1.5
2.0
2.5
3.0
T (SECONDS)
Fig. 6.10  CO:MPARISON OF STRENGTH SPECTRA FOR ROCK AND FIRM
ALLUVIUM SITES
NOR.lv4LIZED C
4
(SLEDRS) code
soil type S4
0.0
1.0
2.0
3.0
T (SECONDS)
4.0
5.0
6.0
11
fl
v
(LEOS )t , (IRS )l=( Cy
)t , or C5 = W
~ED
c..r
mOXC..I
2 .S
2.0
1.5
0\
I
1.0
0.5
Cs'
0.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Period (Second)
127DISP.[in)
11
~5%
p.=l
 J.! = 2
 J.! = 4
 p. == 6
12
10
DISP.[inj
NS
C:E:N'l:RO
1940
,,
,_ .......
........
90 DEG
Jl1
~5%
 Jl  2
 ;.t. 4
 Jl 6
12
10
CAPITOLA
'
....
, ...
'
2
0
1.5
l.O
0.5
0.0
2.0
2.5
. 3.0
2.0
1.0
1.5
PERIOD (sec)
0.5
0.0
Period (sec:)
DIS?.[ir.]
50
DISP.[inj
p.=l
 p. = 2
40
 J.!
 J.!
30
=
=
.u=1
12
 fJ. 2
;.t. 4
0 DEG
~5%
Jl6
10
4
6
CAPITOLA
l4
3. 0
2. 5
,,
,'
6
20
10
s%
2
0
0. 0
l.S
l.O
0.5
!.0
2.5
3.t:
0.5
0. 0
?e:iod (sec:)
.u=l
12
;;.
 iJ 
12
.u= 6
10
35 DEG
10
, 'I
2. 5
1.0
1.5
2.0
PERIOD (sec)
3. 0
JOS.Or::G
J.'=l
 iJ. = 2
 ;: = 4
.u=o
,,
"\.~,
.:...! !
.,.;.:. ..:
0
0. 0
0.5
l.O
1.5
2.0
PERIOD (sec:)
2. 5
3. 0
0. 0
0. 5
1.0
2. 0
1.5
PERIOD (sec)
2. 5
J. 0
DISP .[inJ
14
CORRALITOS
10
DISP.(inJ
DEG
~
JL=l
......... I' = 2
 Jl = 1
 Jl = 6
12
90
= 5%
HOLLISTER
14
_ _ ,, = l
12
 Jt
90
=2
 Jl = 1
10
, ..
..,a
 Jl
=6
~'
.........
.,'"
\.
'~
.....
, ;
,' _,
, ,' ........ .
. .
' .
,'
~I
~'
DEG
...
'
~'
2
0
0.0
CORRALITOS
DISP.(inJ
11
2. 5
2. 0
1.5
1.0
PERIOD (sec)
0.5
=2
12
10
 Jl = 6
 Jl = (
=5%
14
12
,
,,
10
2.0
1.0
1.5
PERIOD (sec)
0. 5
DISP .(in J
0 DEG
p=l
 1'
3. 0
0
0.0
HOLLISTER
p=1
 J1
2
" Jl """ 1
p=6
1::1
2. 5
3. 0
0 DEG
"
,'1
;'\ /.
.....
N
00
I
(=a 5%
,
8
.,,,,..
'!
.~::
0.0
0.5
2.0
1.5
1.0
PERIOD {sec)
2. 5
J. 0
0
0.0
0.5
2. 0
1.5
1.0
PERIOD (sec)
2. 5
J .0
129
IDir...
1940
IDir...
NS
EL C!:N'l'RO
CAPITOLA
0.025
0.025
.u=l
 ,t..t = 2
 ,t..t = 4
 ,t..t = G
0.020
f~=1
 JJ = 2
 J.l.
 J.l.
0.02a
0.015
0.015
0.010
0.010
0. 0 OS
0.005
o.o
90 DEG
c:"
=6
o.a
0.0
l.O
IDIL
1.5
2.0
Period (sec)
0.5
1985
KEXICO
2.5
SC~
0.0
3.0
IDIL
EW
,t..t=l
.  ,t..t = 2
 ,t..t = 4
 ,t..t = 5
0.015
1.0
1.5
2.0
PERIOD (sec)
CAP ITO!.A
0.025
0. 025
0.020
0.5
0 020
jJ      jJ
3.a
0 DEG
{5%
JJ1
 JJ  2

2.5
4
6
o. 015
a.01a
0.010
0.005
~
.~
. .
a.oos
~S%
0. 0
0. 0
0. 0
o.s
l.O
1.5
?e:i..cc
IDIT
o.o:;:5
2.0
 j J 
1
 J..l
=a
IDir...
JS .OEG
J..I=i
~
0.0
:; 0
1.0
1.5
2.a
PERIOD (sec)
{=S%
OAKLAND HARBOR
0.025
0.020
0. 015
a.010
0.010
0.005
305
JJ=l
 f.!
2
 ~ = 4
 J.l. = 6
0.015
0. 0
0. 0
a.5
(sec:)
OAKLAND HA?..:SO?..

0.020
1.5
2.5
J.O
DEG
{s%
0.005
~
a.o
0.5
1.0
1.5
2.0
PERIOD (sec)
2.5
3. 0
a. a
a. 5
2.0
1.0
1.5
PERIOD (sec)
2.5
J. 0
IDIL
CORRALITOS
0.025
90
..
p=l
J.1. = 2
 J.1.  4
p=6
0.020
IDIL
DEG
(=5%
HOLLISTER
0.025
90
DEG
=5%
p=l
JL = 2
 JL = 4
 JL = 6
0.020
0.015
0.015
0.010
0.010
,,
, ' ,~:\:......
I,.'" ....
...,._,
0.005
,;, ...
1
,.~
/\
"
..
, \"\._
,.
~....
'":..~::.::
0.005
!
0. 0
0. 0
0.0
0. 5
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
o.o
0.5
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
....
I
IDIL
CORRALITOS
0.025
0. 020
IDIL
0 DEG
( =5%
J.l=1
J.1. = 2
0.025
0.020
 J.1. = 6
0.015
0.015
0.010
0.010
0.005
0.005
0.0
0.0
0. 5
0 DEG
s%
p=l
J.1. = 2
 Jt = 4
0. 0
HOLLISTER
1.0
1.5
2.0
2.5
3.0
 JL = 4
I'= 6
0.0
0.5
PERIOD (sec)
Fig. 6.14. IDI' SPECfRA (5% DAMPING) [36]
1.0
1.5.
2.0
PERIOD (sec)
2.5
3.0
131
7.1 SUMMARY
The stateofthepractice and oftheart in the use of the concepts of deformation, ductility,
ductility ratio, drift, and interstory drift indices for attaining efficient Earthquake Resistant
Design (EQRD) of buildings structures have been reviewed.
After discussing the advantages of using an energy approach for the EQRD of structures,
and pointing out the differences between deformation, ductility and ductility ratio, the needs
for providing structures with the largest ductility economically feasible and for controlling
the interstory drift index were discussed in detail. The need for establishing more reliable
design criteria for EQRD of structures was also discussed.
The stateofthepractice and oftheart ofEQRD of buildings were reviewed, beginning with
a review of the problems in design and construction of EQresistant structures, followed by
a review of present Building Seismic Codes, with emphasis on how the concepts of Ductility
Displacement Ratio, p,, and Interstory Drift Index, IDI, are used, and how they could be
used, to improve the stateofthepractice according to present knowledge.
The review
covers the building seismic codes of the U.S.,Japan, New Zealand, and Europe (ECCS and
CEB).
Based on a review of the problems encountered in the design and construction of EQresistant buildings, research, development and educational
7. 2 CONCLUSIONS
From the studies conducted and the results presented in this report, the following main
observations can be made regarding the use of ductility and drift limits in EQRD:
132
(EJ
limiting the lateral interstory drift have been recognized in the literature, their
implementation, particularly their reliable quantification, has not been accomplished
fully in present seismic design codes;
While it is possible to use the concept of ductility in a vague manner in discussing the
philosophy of ductilitybased design, when such philosophy has to be applied in the
EQRD of structures the philosophy has to be quantified, and it is therefore necessary
to use unambiguous parameters;
Although displacement ductility factors, J1,, provide good indications of structural
damage, they usually do not adequately
components. To produce safe and economical structures, seismic design methods must
incorporate drift (damage) control, in addition to lateral displacement ductility, as a
design constraint;
methodologies fall short of realizing the objectives of the general philosophy. While
the statement of the general philosophy points out the need to consider three different
limit states (criteria for levels of earthquake, i.e.:
functional; and safety, or survival), in practice, design is typically only carried out for
one criterion (usually safety), on the assumption that the other two would be satisfied
automatically;
I:
133The growing concern over the costs of earthquake damages (direct, functional, and
indirect) points out the need that more attention be given to control of serviceability
and functionality, i.e., control of damage;
Achievement of reliable and efficient EQRD requires satisfaction not only of the
criterion for strength and toughness, but also the criteria for deformation
and
repairability.
are
Strength,
toughness,
deformation
control
and
repairability
The following three main problematical areas have been identified in the earthquakeresistant design of structures: (1) Establishment of reliable critical earthquake input
(design earthquakes); (2) determination of the demands on the entire soilfoundationsuperstructure and nonstructural components system; and (3) prediction of the real
supplies to the building at the moment that an earthquake strikes;
 While a sound preliminary design and reliable analysis of this design are necessary,
they do not ensure an efficient earthquakeresistant
of a structure depends not only on how it has been designed, but also on how it has
been constructed and maintained (monitored and preserved) up to the moment that
the earthquake occurs. There is a need to improve the construction and maintenance
practices of structures;
for the
estimation of demands, which can be grouped into two categories: (1) specified seismic
forces; and (2) methods used to estimate response to these seismic forces;
Strength Demands. For regular buildings up to a certain height (240 ft. in the U.S.),
most of the codes in the world recommend the use of equivalent (static), lateral
seismic forces, which are expressed as a base shear V =(C.,IR)W where:
seismic coefficient equivalent to a SLEDRS
(Smoothed
c. is the
134Response Spectra) for acceleration, S/g, and R is the reduction factor. Although in
most codes the value of R is given without any explicit reference
to global
Structural response is usually estimated using linear elastic analyses of the effects
induced by the equivalent static forces or by these forces multiplied by load factors,
depending on whether the design will be performed using allowable (service or
working) stress, or the strength (load and resistance factor) design method;
There are few countries in which codes recommend the use of limit analysis and limit
design methods (plastic design methods);
Stiffness and Drift Demands. Most seismic codes address design for lateral stiffness
and for drift at service level. Only a few codes explicitly require that the contributions
of torsion should be considered in estimating the maximum lateral drift, and very few
give any guidelines regarding how to deal with the effect of multicomponents of
seismic excitations.
or recommendations
procedures for estimating the demands regarding the stability effects at ultimate limit
states;
Strength Supplies. Most of the Reinforced Concrete (RC) EQRD codes require that
the supplied strength be estimated using the strength method, in which the required
strength of critical sections are evaluated as a function of just the minimum specified
strength of the materials, and then reduced by a strength (resistance) factor. There
are a few codes in which the design and detailing of the critical regions of the
structure are based on the probable supplied strength capacity to the members and to
their connections and, therefore, to the entire structure. The stateofthepractice as
reflected by most present EQRD codes for RC buildings does not appear to include
135the use of the concept of energy dissipation capacity in a rational and reliable way
through the use of the
~,;
Present practice emphasizes the use of strength as the primary criterion for
preliminary EQRD.
justified where serviceability controls, it cannot be accepted in cases where the design
is contrplled by the ul.timate (safety) limit state where plastic deformation is accepted.
At safety limit state (mechanism formation and mechanism movement), base shear is
insensitive to variation of deformation and, therefore, to damage. Although there have
been some proposals to base preliminary design on only lateral damage, i.e.,on only
controlling the interstory drift, a practical method of this type of design has yet to be
developed. A more rational approach is one which not only recognizes the importance
of strength and stiffness (control of deformation), but also recognizes that while these
two factors are strongly interrelated in the case of elastic response, they are less
strongly interrelated
To control inelastic
}
136The future of EQRD is an energy approach in which the concept of ~' is used in the
derivation
reduction factor R. The values of R must take into account the reductions due to:
hysteretic behavior (J..L 6 ); changes in damping and in the fundamental
vibration of the whole building system; and the real overstrength.
period of
The R should be
Ideally, the use of either of the above methods should be complemented with time
history nonlinear dynamic analyses of the response of the preliminarily designed
building system to the predicted Maximum Credible Earthquake ( MCEQ), ground
motions that can occur at the site. If this is not possible, the least that should be
conducted is a static nonlinear analysis of the building under monotonically increasing
lateral loads;
should be based on a realistic 3D model which considers properly the effect of torsion
under multicomponents of ground motions;
acceptable maximum IDI to control damage varies with the type of structure and its
function, usually varying from 0.01 to 0.03. The IDI spectra demands can be estimated
based on the IDRS for strength for the adopted p. 6
In the case of RC
estimating the lateral stiffness of buildings, the real lateral stiffness varies with the
level of deformation;
Most of the practical methods that have been recommended for designing considering
IDI have been based on the assumption that the nonlinear displacement response is
equal to the linear response spectral values provided that the system has certain
minimum
yielding strength.
Recent
displacements are very sensitive to the dynamic characteristics of the ground motions,
and in some cases the displacement can be significantly higher than those computed
from a linear elastic response.
Seismic components and their input direction can significantly affect the response of
a torsionally sensitive structural system. Ground components applied at the structural
reference axes may remarkably underestimate the response because the structural
maximum response is dependent on the seismic input direction and its magnitude;
Code Comparison. In judging the results obtained from the comparison of different
codes, it is necessary to keep in mind that it is not enough , just to analyze the code
requirements of the seismic forces and minimum stiffness or maximum acceptable IDI
to be used in the design. The designed structure and the seismic behavior of the
138
actual structure are not solely the result of specified seismic forces and IDI, but are
governed by the overall design philosophy and the complex combination of the forces
and IDI with many other factors such as:
requirements;
and nonstructural
components
system.
Furthermore, the seismic forces in the code of one country reflect the seismicity as
well as the seismic risk of that country, and these factors vary considerably not only
from one country to another, but even from one region to another within a country;
Except for UBC, all the codes reviewed herein consider that portions of the live loads
are seismically reactive and are included in the computation of the seismic forces;
For strength (ultimate or capacity) design there are significant differences in the values
specified by the different codes for the load factors as well as the ways that the loads
are combined;
The codes reviewed herein are strengthbased rather than ductility and damage
controlbased, and with the exception of the Japanese BSL, advocate a single level
design;
Although the UBC and New Zealand, NZS, code recognize in their material
specification the possibility for overstrength, the only code that explicitly recognizes
and accounts for "maall structural overstrength" due to inelastic redistribution of
forces is the ECCS;
Although most of the seismic codes that have been reviewed permit damage that will
not jeopardize human life, none explicitly defines what constitutes acceptable damage.
Most of the codes recognize that the level of acceptable damage has to be different
for different types of facilities depending
139importance or risktolife factor. However, the values adopted for this factor seem to
be very low, and it appears to be incompatible with the fact that essential facilities and
those housing very hazardous materials should remain practically elastic. The values
for the occupancy factor, specified by the different codes reviewed herein, varied from
1 to 2;
Code Specified SLEDRS. For buildings with a fundamental period ofT ;;:: 2 sees. and
located on firm soil, the U.S. and Japan have similar required SLEDRS which are
somewhat smaller (up to 20% forT = 3.0 sees.) than the
>
.c.~zs.
2.0 sees. and up to T=4 sees. located on very soft soil (soft clay, UBC type S4 or
Zone III of Mexico City), the UBC specifies the most severe SLEDRS, and the CEB
has the least demanding SLEDRS;
Use of
~.
to Reduce SLEDRS to SIDRS. All codes except the Mexico Code use a
up to T = 3.0 sees., the SIDRS specified by the Japanese BSL is more than 33%
higher than any one of the other SIDRS.
Site with Soft Soils (Type S3
):
which is 8.6, and the smallest is that specified by the Japanese BSL (3.3). For tall
buildings with a T > 1. 7 sees. and up to T = 3.0 sees., the yielding strength required
by BSL exceeds by more than 30%, 82% and 121% those specified by the Mexican
D.F., CEB and NZS codes respectively. The yielding strength required by UBC for
tall buildings having T > 2.0 sees. is the lowest one of all the codes considered herein;
Use of IDI Limitations in EQRD. Although all of the seismic codes reviewed herein
have regulations limiting the maximum IDI for limit sta.tes, none of these codes have
140recommendations regarding how the limitations should be directly introduced into the
preliminary EQRD of a building structure.
should result in better damage control under service EQs. This is specifically true in
cases when nonstructural elements can be damaged: IDI :S 0.0006 which is 1/2, 1/4
and 1/6 of those specified by CEB, BSL and UBC respectively.
sees.):
Long T (T
> 1.6
In the case of buildings located on firm soils, the results regarding the
maximum acceptable IDI limits are similar to those for short T. For buildings located
on soft soil, the Mexican D.F. code requirements become as severe as the NZS;
explicitly specifies that the maximum IDI should not exceed the values of 0.006 and
0.012 depending upon whether the nonstructural components can or cannot be
damaged.
The UBC implicitly specifies that the IDI shall not exceed the values of
1.5% in the case of buildings less than 65 feet in height and 1.125% for buildings
greater in height. Although the Japanese BSL does not specify any limit for the IDI
at the Safety Level, in practice the Japanese designers limit the IDI to 0.01. These
limits are a consensus judgment from experience based on observations and analyses
conducted during previous Eqs. Compliance with these limits will ensure not only
human safety, but also damage control, provided that these limits are connected with
a minimum required yielding strength. The minimum UBC required yielding strength
seems to be too low. Thus, the design of tall buildings that attempts to provide only
this minimum strength will undergo significantly larger IDI than the maximum
acceptable by the code in case of severe EQ ground motions;
141Efficient EQRD.
necessary to start with an efficient preliminary EQRD. To carry out this preliminary
design, it is necessary first to develop (establish) reliable design Eqs;
and then
survivability and control of damage under a rare but possible severe (extreme) EQ
ground motion;
Because reliable measured data on EQ ground motions at different sites (soil profile
and topography) was scarce until 1987, design spectra are currently formulated using
inadequate statistical information.
SIDRS for Strength, C1 For any given site, the ideal solution is to derive the SIDRS
directly from statistical and probabilistic analyses of the IRS corresponding to all
recorded motions at the selected site or at similar sites located in tectonically similar
regions and even of records derived through the use of theoretical considerations;
The shape of the IRS (i.e., the variation of Cr with T) varies significantly depending
on the predominant frequency (or period T,) of the recorded ground motion which
in turn depends on the site conditions (soil profile and topography) from which the
record was obtained;
142
There is significant reduction (deamplification) of the LERS (i.e. ,for JJ. = 1) produced
by yielding (JJ. > 1) for structures with a T coinciding with or very close to the
predominant period (TJ of the ground motion. The longer the T,, the larger seems
to be the deamplification;
The degree of reduction of the LERS due to JJ. > 1 decreases as T deviates from T,
and tends to zero as T tends to zero;
Because of the uncertainties in estimating the values ofT, and T, caution should be
taken in applying in practice the observed reduction of the LERS due to
JJ.
= 1;
For sites on firm or medium stiff soils (types S1 and SJ, there are already several
recorded ground motions whose IRS exceeds the SIDRS adopted by the codes
reviewed herein. This is true even in cases of JJ.
= 6 which
to achieve (supply), but also very difficult to justify its possible use because of the
damage that will be involved;
For soft soil sites (soil profile S3 or S4), particularly with soft clays whose depth exceeds
40 ft., from the IRS corresponding to recorded ground motions which can resist and
transfer ground acceleration of 0.30 g to the structure foundation, it appears that the
SIDRS corresponding to the Cy adopted by all codes will be exceeded even when a JJ.
= 6 could
be supplied and used. The only exception is the SIDRS specified by the
Code Procedures to Determine SIDRS for C,.. The SIDRS for Cy specified by codes
are obtained by deamplifying (reducing the LERDS) through the use of a reduction
or behavior factor.
143The values recommended by the UBC (i.e., R.,) appear too high, particularly for
structures with a T < T, if the designer attempts to design the structure with the
strength required by the code: The value for the reduction factor should be tied to
other requirements besides the value of J.L. The values of the reduction factor should
be affected by the real strength capacity, i.e., the overstrength above the yielding
strength specified by the code;
For structures designed according to UBC, the required overstrength depends on the
J.L,
In the case of structures located on rock or firm alluvium, the required normalized
overstrength has the largest values for T in the range of 0.1 to 0.5 sees. and varies
from 0.47 for J.L
= 2 to
= 6.
In the case of very soft soils, the longer the value of the predominant period of the
ground motions, T,, the larger is the range of the period of the structures, T, for which
significant overstrength is required. The normalized overstrength for a T of 0.9 sees.
can vary from 1.23 for J.L
3.84 to 1.81. The Rov for aT of 2.0 sees. can vary from 6.77 for
J.L
= 2 to
1.78 for
J.L
= 6;
U.S. lowrise buildings usually have large lateral seismic overstrength with respect to
that required by U.S. codes. The taller the building, the smaller the overstrength is.
Thus, it appears that the mediumrise buildings (particularly those located on sites with
very soft soils) are the ones that have to be suspected of becoming a serious threat to
life and/or incurring large economic loss in case of a major EQ.
1447. 3 RECOMMENDATIONS
Develop more reliable methods for estimating the values of the reduction factor; This
requires more precise definition of this factor. Although the values of the reduction
factor are affected by several parameters, the main two are the energy dissipated
through hysteretic behavior (damping ratio
overstrength;
The ideal solution is to attain reliable SIDRS directly from the recorded and/or
analytically derived ground motions. This will eliminate the need for specifying R ...
Therefore, for the proper use of these SIDRS, what remains is to calibrate the real
strength (overstrength) of structures that are designed according to present code;
There is a need to consider in the inelastic design of structures the effects of the
duration of strong motions which include the accumulative ductility and number of
yielding reversals. This can be accomplished through the use of an energy approach
estimating the critical required Hysteretic Energy,
Ea;
There is a need to find reliable factors that will permit the use of the computed
SIDRS for SDOF systems to design MDOF systems;
As it is very difficult to design MDOF structures that will develop uniform story 1J. 6
throughout its height, there is a need to investigate a possible concentration
of
required 1J. 6 at one or more stories and to establish the yielding overstrength required
to limit the maximum
system.
J..1. 6
1457. 3. 2
Recommendations
Nonlinear displacements are very sensitive to the dynamic characteristics of the ground
motions and of the structure, and they can be significantly different from those obtained
based on nonlinear behavior.
For ground motions with long T,, the nonlinear displacement can be significantly
smaller (nearly 50% smaller) than the linear displacement for structures with T
On the other hand, for values T
=T,.
higher. The smaller the TIT, ratio, the larger the difference is, and it tends to be
proportional to the value of J.L.
Based on derived SIDRS for strength of SDOF systems, formulate SIDRS for
displacement of SDOF systems for different
and J.L.
Based on the derived SIDRS for the displacement of SDOF systems, obtain lower and
upper bounds for the IDI of MDOF systems.
147
REFERENCES
1.
2.
3.
4.
5.
Frame," Proceedings,
Workshop
on
EarthquakeResistant
6.
7.
Third
International
Microzonation
Conference,
University
of
1488.
Bertero,
from Recent
Earthquakes
and Research
9.
Conference
on Earthquake
10.
11.
12.
Naeim, F., "The Seismic Design Handbook," Van Nostrand Reindhold, New York,
1989, Chapter 6, pp. 171209.
13.
Bertero, V. V. and Bresler, B., "Design and Engineering Decisions: Failure Criteria
(Limit State)," Proceedings, Sixth World Conference on Earthquake Engineering,
New Delhi, India, January 1977.
14.
Dowrick, D. J., "Earthquake Resistant Design for Engineers and Architects," John
Wiley & Sons, Ltd., Sussex, England, 1987.
15.
Cheng, F. Y. and Ger, J. F., "The Effect of Multicomponent Seismic Excitation and
Direction on Response Behavior of 3D Structures," Proceedings, Fourth U.S.
149National Conference on Earthquake Engineering, Palm Springs, California, May 2024, 1990, Vol. 2, pp. 314.
16.
17.
18.
19.
20.
Department of the Army, the Navy and the Air Force, "Seismic Design Guidelines
for Essential Buildings," February 1986.
21.
22.
Biggs, J .M., "Introduction to Structural Dynamics," McGraw Hill, Inc., New York,
1964.
23.
Newmark, N. N. and Hall, W. J., "Procedures and Criteria for Earthquake Resistant
Design," Building Practices for Disaster Mitigation, Building Science Series 46, U.S.
Department of Commerce, National Bureau of Standards, Washington, DC, February
25.
Engineering Research
26.
Research
Design," Report
No.
UCB/EERC88118,
Earthquake
November 1988.
27.
Mahin, S. A. and Bertero, V. V., "An Evaluation of Inelastic Seismic Design Spectra,"
Journal, Structural Division of the American Society of Civil Engineers, 107,No. ST9,
September 1981, pp. 11771195.
28.
29.
Concrete
15130.
in Earthquake
Resistant
Design of
Freeman,
32.
33.
34.
Shirnazaki, K., "Strong Ground Motion Drift and Base Shear Coefficient for RIC
Structures," Proceedings, Ninth World Conference on Earthquake
Engineering,
35.
36.
Bertero, V. V. and Miranda, E., "Analysis of the Implications of the Ground Motions
Recorded
15237.
S~ismic
38.
39.
Conference on
Earthquake Engineering, May 2024, 1990, Palm Springs, California, Vol. 2, pp. 279288.
40.
41.
Spectra," Journal of
42.
U.S. Army, "Seismic Design Guidelines for Upgrading Existing Buildings," Technical
Manual:
43.
Deirlein, G. C. and Hsieh, S.H., "Seismic Response of Steel Frames with Semirigid
Connections
15344.
Applied Technology
~Council,
the United States and Japan," Report ATC15, Palo Alto, California, 1984.
45.
Code,"
46.
47.
v, pp.
48.
11071112.
Tahara, T. and Kamei, T. T., "Comparative Design of 10Story Steel Building using
ATC 306, Los Angeles City Code and Current Japanese Code," Proceedings, Ninth
World Conference on Earthquake Engineering, August 29, 1988, TokyoKyoto,
Japan, Vol. V, pp. 11131118.
49.
Hisatoku, T. et al., "Comparative Design of 19Story Steel Building using ATC 306,
UBC 1982 and Current Japanese Code," Proceedings, Ninth World Conference on
Earthquake Engineering, August 29., 1988, TokyoKyoto, Japan, Vol. V, pp. 11191124.
15450.
51.
52.
by
Helmut Krawinkler
Aladdin Nassar
Mohsen Rahnama
and
U.S. and Kajima Project Research Team
June 1991
TABLE OF CONTENTS
1. INTRODUCTION
3
3
4
31
33
35
35
10
16
16
17
20
21
21
23
26
28
39
39
42
47
47
49
52
6. ACKNOWLEDGEMENTS
53
REFERENCES
54
TABLES
56
FIGURES
62
1.
INTRODUCTION
Seismic design is simply an attempt to assure that strength and deformation capacities
of structures exceed the demands imposed by severe earthquakes with an adequate margin
of safety. This simple statement is difficult to implement because both demands and
capacities are inherently uncertain and dependent on a great number of variables. A
desirable longrange objective of research in earthquake engineering is to provide the basic
knowledge needed to permit an explicit yet simple incorporation of relevant demand and
capacity parameters in the design process. To this end, much work needs to be done.
Identification and evaluation of relevant ground motion and seismic demand parameters is
one aspect of this work. A demand parameter is defined here as a quantity that relates
seismic input (ground motion) to structural response. Thus, it is a response quantity,
obtained by filtering the ground motion through a linear or nonlinear structural filter. A
simple example of a demand parameter is the acceleration response spectrum, which
identifies the strength demand for elastic single degree of freedom (SDO F) systems.
Considering that most structures behave inelastically in a major earthquake, it is evident that
this parameter alone is insufficient to describe seismic demands. Relevant demand
parameters include, but are not limited to, ductility demand, inelastic strength demand, and
cumulative damage parameters such as energy demands.
The term damage potential is used here to denote the potential of ground motions to
inflict damage to manmade structures. This potential depends on the "severity" of ground
shaking as well as the ability of the structure to resist this shaking. Thus, both demands
and capacities need to be considered in assessing the damage potential. In this report the
emphasis is on demand evaluation, but capacity issues are considered in several instances,
particularly in the evaluation of the damage potential of ground motions recorded during the
Lorna Prieta earthquake.
Three sets of ground motions are utilized in this study; a set of 15 ground motions
representative of nearsource rock and firm soil motions recorded in Western U.S.
earthquakes, a set of 51 motions recorded during the October 17, 1989 Lorna Prieta
earthquake, and a set of 8 ground motions from Japanese earthquakes provided by Kajima
Corporation. Several seismic demand parameters are evaluated, using various types of
SDOF and MDOF (multidegree of freedom) models.
The shortrange objectives of the work summarized here are to illustrate the feasibility
of assessing seismic demand parameters and damage potential with simplified analytical
models and to evaluate the sensitivity of seismic response to ground motion and structure
characteristics. The longrange objective is to demonstrate that simple yet rational
demand/capacity models can be used to replace the present empirical code design approach
with a more transparent approach based on fundamental principles.
2.1
Introduction
Present code design is based on elastic "strength demand spectra" (e.g., the product
ZCW in the U.S. 1988 UBC), which are scaled to design base shear spectra by means of
system dependent but usually period independent reduction factors (Rw in the 1988 UBC
orR in the ATC3 approach). The elastic strength demand spectra are modified versions of
smoothened SDOF response spectra, the primary modification being the raising of long
period values to account for multimode effects and provide more safety for multistory
structures. Even though designs based on this approach appear to be satisfactory, there are
two conceptual problems with this approach. Firstly, it is well established that the
relationship between system ductilities and Rfactors is nonlinear, particularly for short
period structures (data will be shown later). Secondly, because of variable overstrength
that comes from many sources, present designs will result in lateral strengths that vary
widely, dependent on the type of structural system and the number of stories (natural
period). As a consequence, the ductility demands that code designed structures may
experience in a severe earthquake have little relation to the relative Rfactors and to the
design base shear, and the design process cannot provide consistent protection for different
structural systems ..
Examples of predicted local (member) ductility demands for three families of code
designed (1988 UBC or SEAOC) steel structures are shown in Fig. 2.1 (Osteraas and
Krawinkler, 1990). All structures have the same plan and story height (72'x120'x12') and
24'x24' bays, and the number of stories is increased till the code period reaches 4 seconds
(except for braced frames which are permitted only to a height of 160', i.e., 13 stories).
All structures are designed by code, but many sources of overstrength are included in the
estimate of structure strength. The local ductilities shown in the figure are estimated from
the strength of the structures, statistical RJlT relationships discussed later, and
relationships between structure and member strengths. Although the true ductilities may
differ from the shown values, two patterns are clearly established. Firstly, the ductility
demands vary significantly with period (i.e., with the number of stories), and secondly, the
braced frames have about the same local ductility demand as the full moment frames (every
frame moment resisting) and perimeter frames, even though they are designed for 50%
more base shear (Rw = 8 versus Rw = 12). This example is shown to document the
arguments made in the previous paragraph.
For the quoted reasons, and others discussed later, there appears to be a need for a
different and more transparent design approach that permits better tuning of the design to
the ductility capacities of different structural systems and the elements that control seismic
behavior. In addition, the issue of damage control deserves special consideration and
should be separated from that of safety against collapse. The following section summarizes
a proposed design approach that forms the motivation for this study on damage potential
and seismic demands imposed by ground motions.
2.2
safety against collapse are treated as separate design levels associated with earthquakes of
different probabilities of occurrence (damage and collapse threshold earthquakes, see
Osteraas and Krawinkler, 1990). Although the designs for these two limit states are
separated, both designs follow the same basic concept and a reconciliation of generated
design constraints is performed in the process. A flow chart illustrating the proposed dual
level design process is shown in Fig. 2.2.
The basic concept on which both designs are based is an explicit capacity/demand
concept in which both capacity and demand parameters are considered explicitly and the
objective is to design structures in which the capacities exceed the demands with a
sufficient margin of safety. Thus, both capacity and demand parameters must be well
defined, due consideration must be given to the uncertainties inherent in capacities and
demands, and, as a final result, a transparent reliability based design approach needs to be
developed. There is much research work to be done in order to fulfill this objective, and
the study summarized here is nothing but a small step in providing some of the basic
concepts and data needed for this purpose.
This study is not concerned with the issue of damage control. It focuses only on
design for safety against collapse, and there only on the problem of ductility capacities and
demands and the development of strength criteria that result in acceptable capacity/demand
ratios.
In the design for safety against collapse it is postulated that the ductility capacity of
the important lateral load resisting elements is the basis for seismic design. For "brittle"
elements (e.g., many types of connections, vertical load carrying columns that may buckle,
etc.) the ductility capacity is 1.0, and for "ductile" elements it is a quantity to be determined
from expenmental/analytical studies. For the latter elements the ductility capacity is a
function of the deformation history, i.e., of the number, magnitude and cumulative damage
effect of all inelastic excursions, the duration of strong motion, and other cumulative
parameters (Krawinkler et. al., 1983). It is proposed that in the design procedure the
element ductility capacity be modified (weighted) to account for anticipated cumulative
damage effects. This points out the importance of strong motion duration, frequency
content of ground motions, and period of the structural systems, since they all affect the
number and magnitudes of inelastic excursions, which in tum determine the cumulative
damage experienced by a structural component.
In order to derive strength requirements, the element ductility capacities have to be
transformed into story ductility capacities (often a straight forward geometric
transformation), which are then used to derive "inelastic strength demands" for design
(discussed later). The so derived strength demands denote the required strength of the
structure. Accepting the fact that the design profession prefers to perform elastic rather
than plastic design, the structure strength level has to be transformed to a member strength
level in order to perform conventional elastic strength design. Pilot studies have shown
that this transformation is usually not difficult but may require an iteration (Osteraas and
Krawinkler, 1990). After this preliminary design an important step is design verification
through a static incremental nonlinear analysis to verify that the required structure strength
is achieved and to assure that "brittle" elements are not overloaded (ductility less than 1.0).
Clearly, there are many issues in this design approach that have not been addressed
and that may complicate the process considerably. But the proposed approach has been
shown to work in simple examples (Osteraas and Krawinkler, 1990), and deserves further
study to explore its potential. In order to implement the approach, much fundamental
information needs to be developed on the following aspects:
1. Ductility capacities of ductile elements (not part of this study).
2. Statistical data on cumulative damage demand parameters needed to modify (weigh)
ductility capacities.
3. Statistical data on inelastic strength demands for prescribed ductility capacities,
using SDOF systems.
4. Statistical data on the effects of higher modes in MDOF systems, needed to modify
the inelastic strength demands derived from SDOF systems.
In this study several sets of ground motion records are utilized to evaluate the
demand parameters listed in 2. to 4. by performing inelastic time history analysis on
various SDOF and MDOF models. The study is rather comprehensive for nearsource firm
soil and rock motions, which are reasonably uniform in their characteristics and permit a
statistical evaluation. For soft soil motions, characteristic patterns are investigated but no
quantitative conclusions can be drawn without a much more comprehensive parameter
study of soil characteristics. The seismic demand parameters used in this study are
summarized in the next section.
2.3
Elastic Strength Demand, Fy,e This parameter defines the yield strength
required of the structural system in order to respond elastically to a ground
motion. For SDOF systems the elastic acceleration response spectra provide the
needed information on this parameter.
Ductility Demand, Jl This parameter is defined as the ratio of maximum
deformation over yield deformation for a system with a yield strength smaller
than the elastic strength demand Fy,e
Inelastic Strength Demand, F y(JJ.). This parameter defines the yield
strength required of an inelastic system in order to limit the ductility demand to a
value of Jl.
in elastic strength that will result in a ductility demand of J.L. Thus, Ry(J.L) =
Fy.efFy(J.L). This parameter is often denoted as R.
Energy and Cumulative Damage Demands.
TDE=DE+HE
Many cumulative damage models have been proposed in the literature, the
simplest one being of the form
N
D = C
(.1Dpi)c
(2.1)
i= 1
where D
cumulative damage
C,c
N
.18pi
earthquake
= the plastic deformation range of excursion i (see Fig. 2.4)
For bilinear systems this expression reduces to the total hysteretic energy if the
coefficient Cis taken as the yield strength Fy and the exponent cis taken as 1.0.
For components of steel structures the exponent c was found to be in the range
of 1.5 to 2.0 (Krawinkler, et al., 1983). The coefficient C varies widely and
depends strongly on the performance characteristics of the structure. In the
comparative study performed here, C can be eliminated and the plastic
deformation ranges t18pi can be normalized with respect to the yield deformation
D =
(t18pJ8y)c
(2.2)
i=l
In this study this expression is evaluated using exponents of c = 1.0, 1.5, and
2.0.
The plastic deformation ranges t18pi in Eqs (2.1) and (2.2) are not the ranges as
they appear in the time history response. For the purpose of cumulative damage
evaluation these ranges have to be reordered because small excursions have to
be considered as interruptions of bigger ones. This can be accomplished with
one of several available cycle counting methods. For lowcycle fatigue damage
the rainflow cycle counting method was found to be the best suited one
(Krawinkler, ei al., 1983). The plastic deformation ranges so identified,
together with the number of inelastic excursions, N, provide basic information
needed for cumulative damage modeling.
There are many other cumulative damage models that have been proposed in the
literature specifically for reinforced concrete structures and elements. Chung et
al., 1987, provides a comprehensive summary of these models. One of these
models is utilized in a pilot study discussed in Section 3.6.
The list of seismic demand parameters enumerated here is by no means complete.
But for conceptual studies much can be learned by using these parameters to assess the
damage potential of ground motions. In the following chapter these parameters are
evaluated for various SDOF systems for closely spaced periods in order to permit a
representation in terms of spectra, using a period range from 0. f sec. to 4.0 sec. In
Chapter 4 the strength and ductility demands are evaluated for three types of MDOF
systems, using six discrete periods covering a range from 0.22 to 2.05 seconds.
3.1
Introduction
This chapter is concerned with an evaluation of seismic demands imposed by ground
motions on SDOF systems. The use of these simplified models for seismic demand
assessment for real structures is limited, as many important effects present in complex multi
degree of freedom structures cannot be represented adequately. Nevertheless, these simple
models are a very useful tool for conceptual studies, as they provide basic information,
show trends and patterns, and permit comprehensive and statistically acceptable parameter
studies. It is clear' that much of the derived information has to be modified in order to be of
relevance to MDOF structures. Some of these modifications are discussed in Chapter 4.
Iri the context of design, the following questions are addressed in this study.
How can important structural characteristics, including deterioration in
strength and stiffness, be represented in SDOF hysteresis models?
What are representative elastic and inelastic strength demand spectra for firm
soil and rock sites?
What is the relationship between elastic and inelastic strength demand
spectra, which is represented by the strength reduction factor Ry(p).
What consistent patterns exist in parameters that are useful for cumulative
damage evaluation, such as the hysteretic energy dissipation?
How sensitive are the seismic demand parameters to characteristics of the
hysteresis models used to represent structures?
How do seismic demand parameters attenuate with distance from the fault
rupture?
To what extent are the seismic demand parameters affected by site soil
conditions?
How can the damage potential of ground motions be assessed from
simplified SDOF models?
3.2
Hysteresis Models
The SDOF models used in this study are intended to represent global response
characteristics of structures. As such, they should be capable of representing all important
structural characteristics that may significantly affect the response to ground motions. In
addition to basic elastic stiffness and yield strength, the following characteristics deserve
consideration:
Degradation in stiffness.
Deterioration in strength.
parameters, a general hysteresis model was developed. The characteristics of this general
model are shown in Fig. 3.1, and the types of specific hysteresis models that can be
generated through rule modifications are summarized in Table 3.1. The specific hysteresis
models that can be represented by the general model are of the following three types.
1.
Nondeteriorating models:
These models are the basic bilinear and trilinear models, which are well established
2.
largest positive and negative excursions of past loading cycles. The following two well
known models are included in this category:
10
Experimental evidence makes it necessary to refine these models further since history
dependent stiffness degradation and strength deterioration may occur in elements and
complete structures.
11
excursions is usually small and probably negligible in the context of seismic behavior.
Thus, only inelastic excursions need to be considered, and from those the large ones cause
significantly more damage than the smaller ones. Also, the cumulative damage may not
lead to a noticeable deterioration until several cycles have been executed This is illustrated
in Fig. 3.2(a), which shows the loaddisplacement response of a steel beam that failed by
crack propagation aild fracture at a beamtocolumn flange weld. Such elements have a
large deterioration threshold, but deteriorate rapidly once deterioration is noticeable, as is
illustrated in Fig. 3.2(b). For other elements deterioration may be evident early in the
loading history (small deterioration threshold) and may occur at an almost constant rate.
The response of such an element is shown in Fig. 3.3(a), and the mode of deterioration is
illustrated in Fig. 3.3(b ).
These examples show that it is necessary to model history dependent deterioration in
a general hysteresis model that is supposed to capture all important phenomena that affect
seismic response.
being the plastic deformation range (see Fig. 2.4) and the hysteretic energy dissipation in
an excursion. The latter parameter is used here because hysteretic energy is a well defined
quantity in all cases, whereas the plastic deformation range is an ambiguous quantity for
stiffness degrading systems. It is assumed that the hysteretic energy dissipation capacity is
a known quantity and that it is independent of the loading history. The latter assumption is
hard to justify, but it has to be made in order to make a parameter study manageable. The
deterioration per excursion is assumed to be defined by a deterioration parameter f3, given
by the following expression:
(3.1)
where
f3
Ei
Er
LE}
c
12
Strength Deterioration:
Deterioration in strength is modeled through a modification of the yield strength F y
and a corresponding translation of the strain hardening stiffness Kh. The modification of
the yield strength is given by
(3.2)
where
13
assumed to be 100Fy8y (i.e., a= 100), and the exponent cis taken as 1.0 in Fig. 3.4 and
2.0 in Fig. 3.5, respectively. As can be seen, the rate of deterioration is very different in
the two cases because of this difference in c.
where
K u,n
K u,o
f3u
(3.3)
= (1  /3) > 0, with the value of f3 given by Eq. (3.1), using the
where
(3.4)
8r,n
8r,o
I4
f3k = (1 + {3), with the value of {3 given by Eq. (3.1), using the
appropriate parameters to model accelerated degradation of the
loading stiffness.
The initial value for 81,0 is 8y. and the modification is applied after each inelastic excursion.
When ~Ej approaches Er. the target displacement approaches infinite which causes a
complete loss of stiffness.
The hysteresis response for constant amplitude cycling of a peak oriented model with
accelerated stiffness degradation is shown in Fig. 3.7(a), using E 1 = 100Fy8y (i.e., a=
100) and c = 1.0. When this response is compared to that of the peak oriented model with
strength deterioration, see Fig. 3.7(b), the deterioration patterns appear to be similar.
However, this comparison is misleading as no strength deterioration is associated with the
model shown in Fig. 3.7(a). If in anyone of the repeated cycles the loading would be
continued beyond the displacement of 48y, the reloading stiffness would be maintained
until the original skeleton curve is reached, whereas in the strength deterioration model
shown in Fig. 3. 7 (b) the stiffness would change to the strain hardening stiffness at a
displacement of 48y. The differences between the two models are evident in the two
histories shown in Fig. 3.8, which consist of 4 cycles with amplitude 48y followed by 4
cycles with amplitude 88y.
A computer program has been developed that incorporates all the models identified in
Table 3.1, as well as any desired combination of strength deterioration and stiffness
degradation. The program can trace the hysteretic response for prescribed arbitrary
displacement histories, and is presently being implemented for dynamic time history
analysis. A parameter study is being performed to evaluate the effects of different
deterioration characteristics on the important seismic demand parameters of SDOF systems.
At this time, parameter studies have been performed for bilinear models and peak
oriented models. Referring to Fig. 3.1, these models are described by the yield strength
Fy. the elastic stiffness Ke, and the strain hardening stiffness Kh. The latter is expressed
more conveniently in terms of the strain hardening ratio a= KhfKe. Strain hardening ratios
of 0.0 (elasticplastic system), 0.02 and 0.1 are used in the parameter study, and all
systems are assumed to have 5% damping.
Time history analyses of these SD 0 F systems were performed to predict the
discussed seismic demand parameters. The natural period of the systems was varied at
15
closely spaced intervals to derive spectral information with a sufficient degree of accuracy
to capture all important trends.
All results discussed in this chapter are based on systems whose yield levels are
adjusted for each record to produce preselected ductility demands of J.l = 1 (elastic
response), 2, 3, 4, 5, 6, and 8. The corresponding yield levels are, therefore, the strength
demands Fy,e and Fy(2) to Fy(8). Thus, all the information generated pertains to systems
that may have different yield levels but equal ductility demands. In the context of the
proposed design approach this is the relevant information, as the available ductility capacity
is the basis for design and the required strength is a quantity derived from the criterion that
the ductility capacity should exceed the ductility demand.
The process of determining the yield level for specified ductilities has to be done very
carefully and may involve several trials because the relationship between yield level and
ductility demand is not necessarily a monotonic function. This is illustrated for one specific
case in Fig. 3.9, taken from Krawinkler and Nassar, 19901. For this record and the
selected structural period of 0.9 seconds, there exist at least three yield levels that result in a
ductility demand of 3.0. Clearly, it is only the highest of the three yield levels that defines
the strength demand F y(3). Interpolation between widely spaced F J.L data points could
have led to very erroneous results. Figure 3.9(b) illustrates that the three systems with
different yield levels have indeed the same ductility demand in the time history response.
The following sections discuss the results obtained from the SDOF study, using three
different sets of ground motion records.
3.3
3.3.1
should prove useful for direct application to design of structures located on rock or firm
soils. Ground motions at such sites depend on seismological/geological parameters such as
earthquake magnitude, stress drop, sourcesite distance, and travel path through geologic
medium, but are considered to be not greatly affected by local surface geology. Thus, their
frequency characteristics are rather similar in most cases (unless the motions are far from
the source) and, in average, can be represented by a smoothened elastic response spectrum
16
of the shape given in the ATC3 document for soil types SJ. Thus, 15 records were
selected whose elastic response spectra resemble that of the smoothened ATC3 ground
motion spectrum. Clearly, there are significant differences in the spectra for each record,
but in average these spectra are a reasonable match of the ATC3 spectrum, at least in the
range from 0 to 1.0 sec. as will be discussed later.
Basic properties of the 15 records selected for this study are presented in Table 3.2.
The records are from Western U.S. earthquakes ranging in magnitude from 5.7 to 7.7.
The unsealed PGA and PGV values are listed in Table 3.2, together with values obtained
by scaling the PGA to 0.4g and the PGV to 30 em/seconds. The scaling issue is discussed
later. Most of the results presented here are obtained from unsealed records and are
presented in terms of ratios, which makes scaling unnecessary. The strong motion
durations D sm listed in the table are based on the definition given by McCann and Shah,
1979. They define the end of strong motion as the point in time when the time derivative of
the cumulative root mean square function is positive for the last time. The beginning of the
strong motion is obtained similarly by applying the same procedure to the reversed time
history of the record.
Seismic demand parameters for different SDOF systems were computed for each
record and evaluated "statistically". The statistical evaluation is limited to simple arithmetic
mean and mean cr calculations.
3.3.2
The basic seismic demand parameters computed for each system are the elastic
strength demand, Fy,e. and the inelastic strength demands, Fy(f.J), which define the required
yield strengths of systems for prescribed ductilities. Since these parameters depend on the
severity of the ground motion, they can be evaluated statistically only once the records are
scaled to a common severity. This will be done later. In nondimensional form, the
inelastic strength demand for constant ductilities can be expressed by the strength reduction
factor Ry(p), which is the ratio of elastic strength demand, Fy,e over inelastic strength
demand, Fy(J.L). This reduction factor can be thought of as an effectiveness factor that
shows by how much the elastic strength demand of a given SDOF system (represented by
the acceleration response spectrum) can be reduced by allowing the system to behave
inelastically within the limits of a predefined ductility ratio.
17
The mean and meana values of these Rfactors, obtained from the 15 records, for
different ductilities and plotted versus period are shown in Figs. 3.10 and 3.11 for systems
with 0% and 10% strain hardening, respectively. RJ.LTrelationships of this type, together
with mean elastic response spectra, can be used to evaluate the inelastic strength demands
for SDOF systems on firm soils. (Modifications applicable to MDOF systems are discussed
in Chapter 4). It is evident that these RJ.LT relationships are highly nonlinear, giving low
R values (i.e., small permissible reductions in yield strength) in the short period range,
peaking in the range of medium periods, and decreasing again for long periods.
These figures show clearly that the often used relationship of R = J.L does not apply
for short period systems but may be a reasonable approximation for medium and long
periods, at least for mean values. It would be easy to show from the data that the often
used relationship R = (2J.L 1)1/2 is a poor approximation for the short period range. The
data presented here can be used to derive more realistic RJ.LT relationships, accounting
also for the effect of strain hardening a, which is not negligible ..
The meancr diagrams are shown for two reasons. Firstly, a comparison with the
mean diagrams gives an indication of the scatter involved in the data. Secondly, and
perhaps more important, they may be more appropriate for design purposes than the mean
diagrams. The mean a values of the Rfactors correspond to the mean+ a values of the
inelastic strength demands Fy(J.L). Considering the variations in the frequency content of
ground motions, which are responsible for the deviations in strength demands, it may be
prudent to use mean+ crvalues rather than mean values of Fy(J.L). The corresponding mean
a of the Rfactors
are significantly smaller than the means and are smaller than J.L in almost
all cases.
Cumulative damage demands, which affect the ductility capacity, can be represented
in different ways as was discussed in Section 2.3. In most cases they are represented by
energy terms, most appropriately the hysteretic energy dissipation demand, and the number
of inelastic excursions. Statistical data on these parameters are presented in Figs. 3.12 to
3.15.
Normalized hysteretic energy dissipation demands (HE!Fy8y) for different ductilities
and plotted versus period are shown in Figs. 3.12 and 3.13 for bilinear systems with 0%
and 10% strain hardening .. For bilinear systems, normalized hysteretic energy is equal to
L1op/ 8y. where L18pi is the cumulative plastic deformation, a parameter that has been used
for damage modeling by Krawinkler et. al., 1983. If linear damage accumulation is
18
assumed, Fig. 3.13 indicates that for 10% strain hardening systems with equal ductility the
damage in a 0.2 sec. structure is about twice that in a long period structure. If no strain
hardening is present, the differences in normalized hysteretic energy demands between
short and long period structures is much smaller, see Fig. 3.12. Information of this type
can be used to weigh ductility capacities.
As Figs. 3.14 and 3.15 show, for any given ductility the number of inelastic
excursions decreases much more rapi~y with period than the normalized hysteretic energy
demands. Since for bilinear systems the normalized hysteretic energy demand is equal to
(the summation is performed over the number of inelastic excursions), this
observation leads to the conclusion that for equal ductilities the average (mean) normalized
plastic deformation range (I:(Llop/oy)/N) for long period systems must be significantly
larger than for short period systems. This is verified from the results shown in Fig. 3.16.
I:Llopi/Sy,
Some researchers advocate the use of input energy as a basic parameter for assessing
the seismic demand of ground motions. Since for inelastic systems the maximum input
energy occurs usually at the end of the motion (except for short period structures in some
cases) and is then equal to the total dissipated energy (TDE =HE+ DE), it is of interest to
evaluate the contribution of HE to the total dissipated energy TDE. The fraction of TDE
dissipated by HE is shown in Fig. 3.17. From all the parameters investigated in this study
the ratio of HEITDE was the most stable one. The scatter of the data is very small (mean +
a values not shown) and the ratio follows a very consistent pattern, particularly for long
period structures.
In all the results presented here it is evident that the strain hardening ratio a has some
effect on the seismic demand parameters. In almost all cases the inelastic strength demand
for elasticplastic systems (a = 0) was larger than that of the corresponding strain
hardening system. The reason is that elasticplastic systems drift more than strain
hardening systems, and a larger yield strength is required in order to limit the ductility to
the same value. The larger Fy(p) for elasticplastic systems is reflected in smaller reduction
factors. This is documented in Fig. 3.18, which shows the ratio of Rfactors of elasticplastic over strain hardening systems. The effect of strain hardening is particularly
pronounced in the hysteretic energy demands for very short period systems, as is illustrated
in Fig. 3.19.
The information presented so far was in normalized form, making scaling of the
records unnecessary. The presentation of actual average strength demand spectra,
19
however, requires scaling of the records before a statistical evaluation can be attempted.
Unfortunately, there is no unique and "best" way to scale records; scaling to common
PGA, or PGV, or equal area under the elastic response spectra has been reported in the
literature, all resulting in different ordinates and shapes of the demand spectra. For the finn
soil records used in this study we have decided to give priority to scaling to a common
PGA of0.4g.
The means and mean+u of the elastic and inelastic strength demand spectra scaled to
this PGA are shown in Figs. 3.20 and 3.21. Superimposed, in dashed lines, is the ATC3
ground motion spectrum for S 1 soil sites and Av =A a = 0.4. Comparing the elastic
strength demand spectra with this ground motion spectrum, it is noticed that the mean
spectrum matches the ground motion spectrum well in the period range below 1.0 sec.,
~ut
poorly for longer periods. This comes as no surprise as the ATC3 ground motion
spectrum consists of two bounds, one associated with an effective PGA of 0.4g (the
straight portion) and one associated with a period independent effective PGV of 30 em/sec.
(the curved portion). It is unlikely that a specific ground motion approaches both bounds.
It is probably incidental, but curious, that the mean+uelastic spectrum matches the constant
velocity region of the ATC3 ground motion spectrum very well.
The important observation to be made from Figs. 3.20 and 3.21 is that the inelastic
strength demand spectra are rather smooth and, for periods larger than about 0.2 sec., can
be represented closely by a smooth curve. Thus, it is quite feasible to develop equations
for inelastic strength demand spectra directly, rather than taking the detour of representing
the elastic spectrum by equations and employing strength reduction factors.
Attempts were also made to scale all records to a common PGV of 12 in./sec. (30
em/sec.). This scaling did not match well with the ATC3 ground motion spectrum as is
evident from Fig. 3.22.
3.3.3
All the analyses summarized in Section 3.3.2 were repeated for stiffness degrading
system, using the peak oriented (modified Clough) model for stiffness degradation and
keeping all other parameters the same as in the bilinear system. The results were evaluated
for all demand parameters in terms of ratios of degrading to bilinear systems. The most
important result is presented in Fig. 3.23, which shows an example of ratips of strength
reduction factors. Consistently, and for all cases investigated, this ratio was close to 1.0
20
except for very short period systems. The simple conclusion is that for strength design it
makes little or no difference whether the hysteretic behavior of a structure resembles that of
a bilinear system or that of a system with stiffness degradation of the type represented by
the peak oriented model. This is not to say that other types of stiffness degradation (see
Section 3.2) and particularly strength deterioration may not have a significant effect on
strength demands. This issue will be investigated in detail with the models discussed in
Section 3.2.
For cumulative damage parameters, stiffness degradation plays a more important role.
This is illustrated in Fig. 3.24, which shows the ratios for hysteretic energy dissipation
demands. But even here the effect of stiffness degradation is not very large except for
systems with very short periods. One can only come to the conclusion that this type of
stiffness degradation is not a very important issue, except if refined damage models are
used that are very sensistive to the hysteretic energy dissipation.
3.4.1
agencies and private organizations, the primary sources being the CDMG and USGS. The
motions recorded by the latter two agencies are well documented (e.g., CDMG, 1989, and
USGS, 1989), many of the records have been digitized and corrected in final format (e.g.,
CDMG, 1990, and USGS, 1990), and several discussions on ground motion issues have
been published already (e.g., Boore et al., 1989, and Competing Against Time, 1990).
21
In the study summarized here, the digitized records made available by CDMG and
USGS were utilized for a global evaluation of strength and ductility demands for bilinear
single degree of freedom (SDOF) systems, with due consideration given to attenuation
characteristics and local site conditions. Records from 51 stations were used for this
purpose. The overriding consideration in the selection of records was that each record
could be viewed as a "freefield" record. Thus, only records from instrument shelters or
one to two story buildings were considered in order to avoid records that could be
considerably contaminated by structural feedback.
The records were classified as either rock, alluvium, or soft site records. This
classification was based in most cases on the information provided in CDMG, 1989, and
USGS, 1990. However, because of the ambiguity of some of the information given in
these references, and in view of the objectives of this study, several judgmental decisions
had to be made in this soil classification. Rock and alluvium site records were used
primarily to study attenuation patterns, and therefore, records which are known to be
uncharacteristic for these two categories were eliminated. This applied to four records that
were classified in CDMG, 1989, as "alluvium" records, but which are known to be soft
soil records (see section on Strength Demands for Ground Motions in Soft Soils) and to the
"rock" motion recorded at the San Francisco Presidio.
Properties of the records classified as rock, alluvium, and other (mostly soft soil site)
records are summarized in Tables 3.3 to 3.5. For, each station only the horizontal
component with the larger PGA value was used. With very few exceptions this component
was also the record with the larger PGV and Arias intensity. The sets of motions for rock
and alluvium sites are rather small for the later described attenuation study, but we had to
limit ourselves to records which were available in digitized form at the time of writing.
The next two sections focus on demand information obtained from time history
response analyses, using these records as input to bilinear SDOF systems with 5%
damping and 10% strain hardening. The emphasis is on strength demand spectra. In the
subsequent section these strength demands are compared with strength capacities of code
designed structures in order to assess the damage potential of the Lorna Prieta ground
motions.
22
3.4.2
Attenuation characteristics are known to follow well established patterns for ground
motions in rock and firm soil (Joyner and Boore, 1988). Thus, the two sets of records
classified as rock and alluvium records were used to evaluate attenuation characteristics of
the peak ground acceleration PGA as well as of the aforementioned strength demands. For
each parameter a regression analysis was performed, using the following relationship
between the attenuating parameter y and the distance r:
logy = a + d log r + kr
where
r0
(3.5)
= (ri + h2)0.5
h
a,d,k
= a depth parameter
= regression parameters.
This equation is of the form proposed by Joyner and Boore, 1988, without
consideration of a site soil correction factor. Joyner and Boore set the value of d to 1.0,
whereas in this study d was used as a free regression parameter. The only constraint on the
regression parameters was that k had to be negative and was taken as zero if it turned out to
be positive in an unconstrained regression analysis. In Joyner and Boore, 1988, the depth
parameter his obtained from a search procedure that minimizes the sum of the squares of
the residuals, and is taken as 8.0 km for attenuation of PGA. This value of 8 km is used
here for attenuation of PGA as well as of strength demand spectra.
Attenuation of PGA
The record sets listed in Tables 3.3 and 3.4 were used in separate least square
regression analyses to evaluate the attenuation of peak ground acceleration PGA. The
regression curves and data points are shown in Fig. 3.25, together with attenuation
relationships obtained by using the Joyner and Boore prediction for a moment magnitude
6.9 earthquake (Boore et al., 1989). As was pointed out already by others (Boore et al.,
1989, and Competing Against Time, 1990), and is confirmed in Fig. 3.25, the
accelerations are higher than predicted by the Joyner and Boore equation and attenuate
slower, particularly at large distances from the fault rupture. A clear difference is also
evident in the attenuation of PGA for rock and alluvium sites, with similar PGA values
23
close to the fault rupture but much slower attenuation for alluvium sites at larger distances,
see Fig. 3.26.
24
The significant changes in shapes of the rock spectra with distance from the fault
rupture may not be critical for the design of structures located on rock sites, since the
design likely will be governed by spectra derived from nearsite earthquakes. However,
there is a critical consequence to be considered. The elastic spectra shown in Figs. 3.27
and 3.28 represent frequency characteristics of the ground motions, and a broad plateau
with relatively high spectral values at longer periods clearly indicates that soft soils over
rock will be much more excited than anticipated from a typical nearsource rock spectrum
with peaks at short periods. There is no doubt that the slow attenuation of long period
components of the ground motion, reflected in the wide plateaus of the farsource rock
spectra (80 km), has contributed significantly to the amplification of motion in soft soils in
and around San Francisco and Oakland. This conclusion is not new, but little
documentation existed so far on the importance of this effect. As Fig. 3.28(a) shows, the
spectral accelerations at long periods in farsource rock spectra may be easily three times as
large as anticipated from nearsource rock spectra.
Figures 3.29 and 3.30 show results similar to those of Figs. 3.27 and 3.28, but for
inelastic strength demand spectra for Jl = 4. As was stated previously, these spectra
identify the required strength of SDOF systems whose ductility is limited to a value of 4.0.
The shapes of these spectra are relatively smooth and are very different from the elastic
ones, indicating that the inelastic strength demands are not related to the elastic ones by a
period independent reduction factor (Rfactor) as is assumed in present design codes. The
variation in spectra shapes with distance is illustrated in Fig. 3.30 and shows only minor
variations for alluvium spectra, but great variations for rock spectra based on the much
slower attenuation of inelastic strength demands for longer periods.
Figure 3.31 shows regressed strength demand spectra for Jl = 1 (elastic), 2, 3, and 4
at a distance to the fault rupture of 10 km. Again, the nonlinear relationship between elastic
and inelastic spectra is evident, particularly in the rock spectrum where the peak in the
elastic spectrum disappears already in the Jl = 2 spectrum.
25
one of the records (SLAC) can be viewed as a freefield ground motion, the other three
motions were recorded in the basement or ground level of multistory buildings. The
SLAC station is located on tertiary sedimentary rock, whereas the other three stations are
on alluvial soils.
The elastic strength demand spectra for the four records are shown in Fig. 3.32.
With all the differences in peak ground motion values and site conditions it comes as some
surprise that the elastic strength demand (acceleration response) spectra are so similar over
the whole range of periods. One may want to conclude that the engineering content of the
ground motions is similar and the seismic demands on the Stanford campus and its vicinity
are rather uniform.
3.4.3
It is well established by now that records at soft soil sites (particularly fill on bay
mud) show great amplifications of PGA values compared to nearby firm soil records.
However, it would be misleading to proportion seismic demands imposed on structures
according to PGA alone, without regard to frequency characteristics of the ground motions.
The following discussion focuses on an evaluation of the damage potential of Lorna Prieta
ground motions that were recorded on soft soil, utilizing the aforementioned elastic and
inelastic strength demand spectra. Results from six individual ground motions (see Table
3.7) are discussed and illustrated, since the records are site specific and no statistical
evaluation is possible. Five of the six ground motions were recorded on deep soft soils
whereas .the Stanford/SLAC record is representative of motion in firm ground. CDMG,
1989, classifies four of the five soft soil records under "alluvium," but all these stations are
on soft soils and, in addition, at one site (Oakland) several feet of fill material were placed
on top of finegrained sand.
Elastic as well as inelastic strength demand spectra for each of the records are shown
in Fig. 3.33. Superimposed on the spectra are two sets of curves that relate the spectra to
code design considerations. The dashed curves (S2  S4 ) represent the 1988 UBC code
design coefficients ZC for soil types S 2 , S3 , and S4 , and are a measure of the elastic
strength demand envisioned by the code. The dasheddotted curves (SMF88 and CMF68)
will be discussed in the next section. An inspection of this figure as well as of Figs. 3.34
to 3.36 lead to the following conclusions on the seismic demands imposed by ground
motions on soft soil.
26
1. The shapes of the elastic strength demand (acceleration) spectra, which are shown in
Fig. 3.34 (spectra normalized with respect to PGA, or dynamic amplification factors,
DAF) are very different for all six records. Except for the Stanford/SLAC motion,
which was recorded on firm ground, the elastic spectra show a clear signature of the
soft and deep soil on which the motions were recorded. This signature is most evident
in the large DAF of the Foster City I Redwood Shores record at a period of 0.7 seconds
(DAF = 4.0) and in the wide humps of several of the other spectra, extending to and
27
Shores record for p. =2, 3, and 4. The peaks and valleys of these plots coincide with
those of the corresponding elastic strength demand spectrum. The consequence of this
phenomenon is that the peaks evident in the elastic spectra diminish and ultimately
disappear in inelastic strength demand spectra for increasing p. values.
6. The observation made in the previous paragraph has significant implications for design.
For structures with small ductility capacity the required strength will be high and the
elastic strength demand spectrum will be an important design parameter. For structures
with large ductility capacity the inelastic strength demand spectra, which are very
different from the elastic spectra, should control the design. This implies that more
knowledge. needs to be acquired on the magnitude and shape of inelastic strength
demand spectra. The data discussed here show that the shape is site dependent, but in a
different manner than the elastic spectra. This can be seen by comparing the normalized
elastic strength demand spectra of Fig. 3.34 with the normalized inelastic strength
demand spectra for p. = 4 shown in Fig. 3.36.
7. The elastic strength demand imposed on structures on soft soil may be easily 6 times as
high as that on nearby structures built on rock. This factor is obtained as the product of
PGA ratios and DAF ratios of softsite to rocksite records.
The work discussed here led to an interesting observation worthy of reporting. The
six records used in this section were initially handdigitized four days after the Lorna Prieta
earthquake from enlarged photocopies of the analog traces published by USGS and
CDMG. No filtering and baseline corrections were applied to the handdigitized records
and results similar to those shown in Fig. 3.33 were available five days after the
earthquake. It is of some interest to note that the spectra derived from these cursory
digitized records are not much different from those presented in Fig. 3.33. An illustration
of typical differences is shown in Fig. 3.37. For reasons that would deserve explanations
the differences are larger for the inelastic spectra than for the elastic ones, and are larger in
the long period domain than in the short period one.
3.4.4
strength capacities of structures must be compared to the strength demands imposed by the
ground motions. These capacities must be estimated with due consideration given to the
28
real strength of structures. For many reasons structures are stronger, and sometimes much
stronger, than is implied by the seismic base shear coefficients used in code design. The
overstrength above the design force level may come from many sources, and depends
primarily on the type of structural system and the number of stories (period) of the
building. The strength of five types of code designed buildings, considering most but not
all sources of overstrength, was estimated by Osteraas and Krawinkler, 1990, and Nassar
and Krawinkler, 1991, and is illustrated in Fig. 3.38. Four of these five types of
structures were designed in conformance with the 1988 SEAOC Bluebook (SBF88 = steel
braced frames, SMF88 = steel moment frames, SPF88 = steel perimeter frames, and
CMF88 =concrete moment frames), and one type was designed in conformance with the
1968 SEAOC Bluebook (CMF68 =concrete moment frames).
The strength capacity curves of the strongest and weakest structure types (SMF88
and CMF68, respectively) provide a range of the expected strength of modern code
designed structures and are used here to assess the damage potential of the Lorna Prieta
ground motions. These two capacity curves are superimposed on the strength demand
spectra illustrated in Figs. 3.27, 3.29, 3.31, and 3.33, and are left to the reader for
interpretation, with only a few observations summarized here.
Figure 3.27 indicates that even for code designed structures located on rock, small
inelastic deformation demands have to be expected in some cases as far away as 80 km
from the fault rupture, considering that the CMF68 curve falls partly below the elastic
strength demand spectra for this distance. Figure 3.29 indicates that global ductility
demands of 4 are anticipated as far away as 20 km from the fault rupture. Figure 3.31
indicates that at a distance from the fault rupture of 10 km the ductility demands could have
exceeded the value of 4 considerably for CMF68 structures with period between 0.2 and
1.0 seconds and located on alluvial soils. Clearly, these observations are more qualitative
than quantitative, since the spectra are obtained from a regression analysis and represent
average rather than site specific conditions, and since significant deviations from the
estimated values of structure strength have to be anticipated.
Figure 3.33 shows clearly that in soft soils, where the shaking was severe even at
large distances, the ductility demands for modern structures could have been substantial.
This holds true particularly for early70 vintage low and midrise reinforced concrete
frame buildings (CMF68), except for 2story buildings. The global ductility demand for
modern steel moment frames (SMF88) was small for lowrise and highrise construction
and moderate for midrise buildings.
29
The demand and capacity curves shown in Figs. 3.27, 3.29, 3.31, and 3.33 illustrate
the variation of expected damage (assuming that ductility is an acceptable measure of
damage) with sourcesite distance and site soil conditions  from the viewpoint of demands,
and types of structural system and number of stories (period)  from the viewpoint of
capacities. In none of the cases illustrated, the ductility demand is severe enough to justify
collapse of a modern structure. But relatively high ductility demands are evident in some
cases even though the ground motions were significantly smaller (except close to the
source) than anticipated by present code philosophy. This indicates that excessive ductility
demands may have to be anticipated in some code designed structures if the ground
motions approach the level envisioned in design codes. This holds true particularly for
structures located on soft soils. It is understood that soft soil PGA amplification decreases
as the rock PGA increases (Idriss, 1990, estimates that at rock PGAs exceeding 0.4g no
soft soil PGA amplification occurs), but there is good reason to believe that significantly
larger soft soil strength demands can be generated in a severe earthquake than those shown
in Fig. 3.33 and, therefore, the ductility demands for modern structures may be much
higher than those shown in this figure.
In summary, the Lorna Prieta earthquake has again demonstrated the sensitivity of
ground motions to sourcesite distance and orientation, travel path through geologic media,
and local site conditions. Moreover, the earthquake has also demonstrated the great
dependence of the elastic as well as inelastic structural response on the frequency
characteristics of the ground motion. The simple conclusion is that the demands imposed
by an earthquake on a structure, which should be the basis for protective design, must be
evaluated with due consideration given to all the aforementioned factors. This conclusion
points towards the need for detailed microzonation and raises questions on the
appropriateness of presently employed global seismic zoning. It is recognized that detailed
microzonation is a long way from reality because much of the needed information is not yet
available. However, the Lorna Prieta earthquake provided an excellent opportunity to
assess seismic demands based on recorded ground motions, as a first step in identifying the
sensitivity of the demands to known site conditions and structural response characteristics.
30
31
only important one for systems of less than 0.8 sees., the Sendai 030 records maintains
some relevance for systems of medium periods, and the Nagoya 306 record takes on
increasing importance for medium to long period systems. The range of importance of the
Hachinohe NS record is pushed to very long period systems. Again, all other records do
not control over a significant period range. Even though these observations are based on
SDOF strength demand spectra, which pay no attention to multimode effects, there is little
doubt that some of the other records can be eliminated from considerations for strength
design.
A different picture of relative importance of the records is obtained when cumulative
damage parameters are evaluated and the duration of ground motion becomes an important
consideration. Let us again assume that normalized hysteretic energy dissipation demand
(HE!Fyay) is the relevant parameter for this purpose. Plots of this parameter for the 8
records and J.L = 2 and 4 are shown in Fig. 3.41. Different records gain now on
importance, particularly the two Hachinohe records (NS and EW), which are characterized
by long strong motion durations and significant frequency content at long periods. On the
other hand, the Tokyo 101 record, which is of short duration, causes very little cumulative
damage even for short period systems where it controlled the strength demand.
In the interpretation of these results it must be considered that Fig. 3.41 shows
hysteretic energy dissipation normalized by Fy8y A very different picture is obtained for
the absolute values of energy dissipation, as is illustrated in Fig. 3.42. This figure shows,
for the same cases as illustrated in Fig. 3.41, the hysteretic energy dissipation demands per
unit mass. Only two records control this demand, the Sendai 030 motion for periods less
than 1. 7 seconds, and the Hachinohe NS motion for larger periods. The differences
between Figs. 3.41 and 3.42 come from the fact that in these respective period ranges these
two records demand high strength as well as many inelastic cycles if the ductility is to be
limited to the specified values for which the plots apply.
The seismic demand assessment for the 8 Japanese records discussed in this section
is intended to put the importance of these records in perspective, in the context in which
these records are believed to be used in Japan. The records have vastly different frequency
contents, which in part are caused by greatly different site soil effects, and are very unlikely
to occur at the same site. As far as frequency content is concerned, they appear to bracket a
wide range of different conditions. The selection of specific records for particular designs
will depend on how much is known about the local conditions. The presented information
is intended to assist in assessing the importance of the individual records for use in design.
32
3.6
Experiences from past strong earthquakes have shown the vulnerability of reinforced
concrete buildings to strong ground motions. In order to assess the seismic safety of
reinforced concrete buildings, quantitative analysis of structural damage under earthquakes
is essential. Several damage parameters for reinforced concrete buildings have been
proposed for this purpose. The objective of this study is to evaluate the damage potential
of recorded earthquake ground motions by analyzing one of the damage parameters of
single degree of freedom systems.
Damage Index
Under earthquake loading, reinforced concrete structures are generally damaged by
both repeated stress reversals and high stress excursions. The damage index proposed by
Park and Ang is expressed as a linear combination of the maximum deformation and the
hysteretic energy:
in which
Om
Ou
Qy
dE
=
=
33
El Centro
Hachinohe
Taft
The acceleration time histories, Fourier amplitude spectra, and linear elastic response
spectra of these earthquakes are shown in Figures 3.44 to 3.46.
SDOF Models (see Fig. 3.43)
The following SDOF models are utilized in the analysis:
a)
b)
kp
h
8y
SDOF system
~
Ou
damage index
Qy
8y:
Oemax:
O.lk0
0.05
0.58emax
0.05
4.08y
k0 8y
Results
Damage indexes for these SDOF systems for El Centro, Hachinohe, and Taft are
shown in Figure 3.47 to 3.49, respectively. Though the yield deformation is assumed as
half of the maximum elastic deformation of the SDOF system (i.e., each SDOF system has
a strength reduction factor of 2.0), the damage indexes are different. This shows the
difference of the damage potential among earthquakes. For example, among these
earthquakes, Hachinohe has comparatively high damage potential around 1.5 second of
natural period.
34
The previous chapter provided information on seismic demands for inelastic SDOF
systems. This information is relevant for relative assessment of damage potential of
ground motions, but needs to be modified to become of direct use for design of real
structures, which are mostly multidegreeoffreedom (MDOF) systems governed by
several translational and torsional modes. For elastic MDOF systems, the combination of
modal responses using SRSS, CQC, or other approaches, provides reasonable estimates of
peak dynamic response characteristics. For inelastic MDOF systems, modal superposition
cannot be applied with any degree of confidence and different techniques have to be
employed in order to predict strength or ductility demands that can be used for design.
Story ductility ratio, defined as the maximum dynamic interstory displacement
normalized by the interstory yield displacement, is used here as the basic deformation
parameter and, therefore, the strength demand associated with the a specific story ductility
ratio becomes the basic design parameter. This strength demand differs from that of the
corresponding SDOF system because of the higher mode and torsional effects, and
depends on many other structural characteristics such as the member stiffness and strength
distribution over the height of the structure, structural system redundancy and modes of
failure of critical structural elements.
frames with widely spaced elastic modal periods were utilized to represent the behavior of
three distinct structural models. The torsional effects of 3dimensional structures are
neglected and the results do not apply to structures with closely spaced modal periods.
The three models used are illustrated in Fig. 4.1. The "beam hinge" (strong columnweak beam) model, referred to as BH model from here on, represents structures (e.g.,
35
special moment resisting frames) in which plastic hinges will form in beams only (as well
as supports). The "column hinge" (weak column strong beam) model, referred to as CH
model from here on, represents structures (e.g., braced frames and moment resisting
frames designed to accommodate plastic hinges in columns) in which plastic hinges will
form in columns only. The "weak story" model, referred to as WS model from here on,
represents structures in which plastic hinges will form in columns of the first story only.
This model represents the behavior of frames with a strength (not stiffness) discontinuity in
the first story. Note that structures with infill walls in all but the first story will have both
strength and stiffness discontinuities, which is not represented by the WS model.
The three models were designed to develop the structure mechanisms shown in Fig.
4.1 under the 1988 UBC equivalent static lateral load pattern, i.e., relative member
strengths were tuned so that all shown plastic hinges develop simultaneously under this
specific lateral load pattern. Note that the BH model cannot develop a story mechanism,
the CH can develop one in any story and theWS can develop one only in the first story.
The only difference between the CH and WS models is the strength discontinuity above the
first story for the WS model, in which all stories remain elastic throughout the entire
dynamic time history analysis discussed later.
The elastic member stiffnesses in each story are tuned so that, under the 1988 UBC
equivalent static load pattern, the interstory drift in every story is identical, resulting in a
straight line deflected shape. Furthermore, the stiffnesses are tuned such that the first mode
period of each structure is equal to that given by the UBC as T = 0.02hn3!4 sec, where
hn = total frame height in feet. As a consequence, the first mode shape for the three models
is close to a straight line. This tuning is done for the three models in order to permit a
direct comparison of dynamic analysis results. Only structures with 2, 5, 10, 20, 30 and
40 stories (story height h = 12'), ranging in period from T = 0.217 to 2.051 sec, were
studied. Pertinent data for the first five elastic modes of the CH (and WS) models are
presented in Table 4.1. The modal periods for the BH models are the same for the first
mode but deviate slightly for higher modes due to difficulties in precise stiffness tuning,
which also led to slight deviation in modal masses.
A bilinear momentrotation (M8) hysteresis model with strain hardening ratios a= 0,
2 and 10% is assumed at each plastic hinge location. Therefore, under an incremental
("pushover") 1988 UBC load pattern, each structure will have a bilinear response identical
to that of a bilinear SDOF system, which was employed as one of the SDOF models in
Chapter 3.
36
The base shear strength, Vy, is varied for each structure and ground motion record so
that it is identical to the inelastic strength demand Fy(J.Lt) of the corresponding SDOF system
with the first mode period and the same strain hardening ratio, for SDOF target ductility
ratios of J..Lt = 1, 2, 3, 4, 5, 6 and 8. Tuning the base shear capacity of MDOF systems to
the strength of their corresponding SDOF systems subjected to the same ground motion
record, is critical in order to compare results for SDOF and MDOF systems. Because
modal superposition is hardly feasible for inelastic MDOF systems, it is desirable to utilize
SDOF demand estimates to predict MDOF demands. The inelastic strength demands can be
evaluated for SDOF systems as discussed in Chapter 3. The question to be answered is
how different are the ductility demands fo:r MDOF systems if their base shear strengths are
identical to the inelastic strength demands of the SDOF systems for the prescribed target
ductility ratio f.Lt Or more relevant for design, the question becomes: how different should
the base shear strength of the MDOF system be (assuming the code prescribed seismic load
pattern over the height of the structure), compared to that of the corresponding SDOF
system, in order to limit the maximum story ductility ratio in the MDOF system to the
specified target ductility ratio J..Lt (see Fig. 4.2).
The following assumptions were made in the design and analysis of the MDOF models :
Even though the 1988 UBC uses different equations for estimating the natural
periods of different structures, the same equation, T
The Pdelta effect of gravity loads is not included explicitly and is assumed to be
accounted for through a reduction in the strain hardening ratio a.
The effect of gravity load moments on plastic hinge formation is not included.
This simplifies the nonlinear structural response from a multilinear to a bilinear
one.
37
Axial load  bending moment (PM) interaction is not considered, i.e., the
bending strength at each plastic hinge location is assumed to be constant,
irrespective of the axial load carried by a column.
The overturning moment at a given story is assumed to be the integral area of the
story shears from the top of the structure to the base of the columns of that story
at a given instance. No allowance was made for moments at the base of columns.
Therefore, the calculated overturning moments are the externally applied moments
irrespective of the type of model used.
5% damping was assumed for the frrst two modes in the time history analysis.
The BH and CH models were designed and utilized in a previous study to assess
MDOF strength and ductility demands for the 1987 Whittier Narrows ground motions
(Krawinkler and Nassar, 1990). In the study summarized here these two models as well as
the WS model are utilized to derive statistical (mean and mean+cr) values for MDOF
strength and ductility demands for S1 type ground motions, utilizing the 15 S1 ground
motions for which SDOF results are available. All model structures, with their strength
tuned as discussed previously, were subjected to the 15 S1 motions using a modified
version of the DRAIN2D analysis program. The results of the analyses were evaluated for
strength and ductility demands as well as for overturning moment characteristics.
A total of 5,670 nonlinear time history analyses were performed, using 15 ground
motion records, 3 types of structures (BH, CH, and WS), 6 different fundamental periods
(number of stories = 2, 5, 10, 20, 30, and 40), 7 yield levels (corresponding to SDOF
yield strength for target ductility ratios J..L1 = 1, 2, 3, 4, 5, 6, and 8), and 3 strain hardening
ratios (a= 0.0, 0.02, and 0.10). In addition, elastic time history analyses (J..L =e) were
performed as needed. Note that an MDOF system designed for a base shear equal to that
imposed on an equivalent elastic SDOF will not necessarily remain elastic, therefore the
distinction between J..Lt = 1 (elastic SDOF) and J..L = e (elastic MDOF). From each dynamic
time history analysis the following information was stored for evaluation :
overturning moments resulting from maximum story shears since the story shear maxima
38
may not occur at the same instance. Figures. in which the distribution of the
aforementioned parameters are shown over the height of the structure. represent envelopes
of story maxima for these parameters. irrespective of the time at which they happen.
Results presented in the following sections are mean values obtained from the 15 S1
records, a record set designated from here on as 15s.
resulting interstory ductility ratios would be equal to the SDOF target ductility ratio Jlt
However, due to higher mode participation in MDOF systems, this is usually not the case.
The question is how do the MDOF interstory ductility ratios compare to those of the
equivalent SDOF system. Or more important, how much do we need to modify the base
shear capacity of the MDOF system, compared to that of the equivalent SDOF system, to
limit the maximum interstory ductility ratio to the same prescribed value (see Fig. 4.2).
4.2.1
20story BH structures, for an SDOF target ductility ratio of J.lt = 8 and strain hardening a
= 0% for the 15s data set (thus the notation 15s.bh208.00). Part (b) of the figure shows
the corresponding information for the CH model.
Fig. 4.4 shows the mean story ductility ratios for the three models (BH, CHand WS,
respectively) with 0% strain hardening, base shear capacities equal to the inelastic strength
demand for SDOF systems with a target ductility ratio Jlt
= 2.
4.4 except that the target ductility ratio J.lt is equal to 8. Fig. 4.6 is similar to Fig. 4.5
except for a= 10%.
If the MDOF structures would respond like SDOF systems, the story ductility ratio in
every story would be equal to the target ductility ratio Jlt Clearly, this is not the case, as
expected, because of higher mode participation and differences in mechanisms in the three
structure types. The results in Figs. 4.4 to 4.6 show consistent trends that can be
summarized as follows :
39
The pattern of the story ductility demand distribution over the height of the structure is
similar for the BH and CH structures. The distribution is rather uniform over height
for 2 to 10 story structures; for taller structures the ductility demands are high in the
lower and upper stories (except for very large ductility ratios (p.1 = 8)), and decrease
significantly in the middle stories. This pattern is due to the participation of higher
modes whose elastic periods (see Table 4.1) coincide with the peaks of the elastic and
inelastic strength demand spectra.
The ductility demands for the CH structures are consistently higher than the demands
for the BH structures, particularly in the lower stories, even though both structures are
designed for the same base shear strength. This is because a story mechanism can
readily develop in a CH model, resulting in more drift.
The deviation of story ductility ratios from those predicted by the equivalent SDOF
systems increases with the number of stories, the target ductility ratio, and decreasing
strain hardening. The largest deviation is associated with the WS model and the least
with the BH model.
The strain hardening ratio a has a significant effect on the ductility demands, as can be
seen by comparing the results for a = 0 and 10%. The elasticperfectly plastic
structures (a= 0%) create much larger ductility demands, particularly in the lower
stories of tall structures, even though the base shear strength, Vy
is greater than that for a = 10%.
= Fy(Jl.r), for a= 0%
The relative story strengths, which are in conformance with the 1988 UBC seismic load
pattern, do not cause an equal ductility demand in each story. (It probably was not
intended to do so).
In most cases, the maximum story ductility ratio occurs in the first story, indicating that
the 1988 UBC seismic load pattern generally provides good protection against
excessive ductility demands in upper stories. There are notable exceptions, particularly
for 5 to 20 story structures and lower ductility ratios, where the story ductility demands
in the upper stories are sometimes higher than in the first story. This indicates the need
for some (but not major) revisions of the code seismic load pattern. In the evaluation of
the results discussed later it is assumed that the first story ductility ratio governs for
design and that the required base shear strength is determined from this first story
ductility ratio.
40
The ductility demands for theWS model demonstrate the great problems inherent in
these structures. Only the first story ductility ratios are actual ductility demandsl, and
they are much larger than those of the corresponding CH structures. It should be
emphasized that in the first story, the WS and CH structures are identical, i.e., they
have the same strength and story mechanism. Differences start above the first story,
insofar that all other stories in the WS structures are "strong" and remain elastic
whereas the strength of their counterparts in the CH structures are tuned to the first
story strength and the code seismic load pattern. The high ductility demand imposed on
the WS structures is attributed to the concentration of energy dissipation in the first
story. The demand is even higher for higher ductility ratios and no strain hardening.
Even though the shear transfer in the WS structure is limited by the shear capacity of
the first story, the elastic vibration of the upper stories can attract severe shear forces
(not limited by any yield level) that transmit high axial forces to the first story. It is
worth noting that the stories above the first do not vibrate as a rigid body, and that their
interstory drift demand increases with strain hardening in the first story.
Strain hardening helps distribute the seismic energy throughout the entire height of the
BH and CH models. Comparing Figs. 4.5 and 4.6 shows that the double curvature
noted in the story ductility demand distribution of tall structures with no strain
hardening, almost disappears. The upward transfer of shear, and consequently seismic
energy, increases with strain hardening.
Figure 4.7 shows the story ductility demands in the first story, J.LJ, of the MDOF
systems for SDOF target ductility ratios J.Lt = 2 and 8 and strain hardening ratios a= 0 and
10%, plotted against the first mode period of the MDOF systems. The target ductility
ratios, indicated by the horizontal dashed lines, are the ductility ratios of the SDOF systems
whose strength, Fy(J.Lc), was used as base shear strength Vy in the design of the MDOF
systems. The following can be noted from Fig. 4.7:
1 "Ductility" values larger than 1.0 are shown in the upper stories of the
WS structures
to illustrate the increase in elastic forces, attracted in these stories, beyond the force
level based on the code seismic load pattern. For instance, as shown in Fig. 5.5 (c), the
elastic shear forces in the upper stories of the 40~story WS structure (with target
ductility J.Lt= 8 and a= 0%), is about 6 times that required by the code load pattern,
even though the shear transfer to the upper stories is limited by the strength capacity of
the first story. This increase is mainly caused by increased elastic vibrations above the
first story.
41
The deviations from the target ductility ratios exhibit consistent trends, being small for
lowrise BH and CH structures and increase with the number of stories. However, the
WS structures exhibit very large ductility demands. Also, the difference in ductility
demands for CHand BH models appears to be period independent (almost parallel lines
in the figure). The maximum ductility demand for the 40 story structures, may be as
high as 10 times the target ductility ratio for theWS model, 3 times for the CH model
and 2 times for the BH model.
The significant effect of strain hardening is evident by comparing the results for strain
hardening a= 0 and 10%. Systems with strain hardening (a= 10%) give consistently
smaller ductility demands than those without (a= 0%). This is attributed to the fact
that systems without strain hardening drift more and especially so for systems that can
develop separate story mechanisms, i.e., WS and CH models.
=2
to 8 and strain
Working in the dimensionless RJ..L domain, shown in Fig. 4.8, the modifications
needed for the MDOF base shear capacities, in order to limit the story ductility demands at
the base (assumed to be the maximum) to the prescribed target ductility ratio J..Lr. can be
readily evaluated. For a given number of stories (period) and strain hardening ratio, the
RJ..Lrelationship for any of the three types of MDOF systems and their equivalent SDOF
system can be plotted as shown in Fig. 4.8. Point 1 refers to the strength capacity
associated with an SDOF target ductility ratio f..Lt An MDOF system designed for that
strength capacity will give a higher ductility demand J..L. (shown as Point 2). In order to
limit the MDOF ductility demand to J..Lt, linear interpolation between data points similar to
Point 2 is used to obtain Point 3. The ratio of the reduction factors at Points 1 and 3 is the
reverse of the ratio of the strength capacities of the SDOF and MDOF systems that will give
the same target ductility ratio f..Lt.
Fig. 4.9 illustrates the extent to which the SDOF Rfactors need to be modified to
limit the ductility demands in the first stories of the three types of MDOF systems to
specified target ductility ratios. The figure shows the R J..L relationships for 2 and 40 story
structures with 0 and 10% strain hardening. Data points on which the curves are based are
obtained by using, as ordinates, the mean strength reduction factors for f..Lt = 1, 2, 3, 4, 5,
6, and 8 of the SDOF system with the first mode period of the MDOF systems (these
values defme the strength of both the SDOF and MDOF systems used in the analysis), and,
as abscissa, the mean ductility ratios of the SDOF and MDOF systems. The data points
shown on solid curves for the MDOF systems correspond to Point 2 in Fig. 4.8.
Connecting the data points for each system with straight lines gives approximate RJ..L
relationships for the SDOF as well as MDOF systems. As mentioned before, the ductility
demand parameter used here for the MDOF systems is the story ductility demand in the first
story of the structure, which in most cases is the maximum overall story ductility ratio.
The following observations are made from the Rm relationships in Fig. 4.9:
The R J..L relationships for the three MDOF models are relatively smooth curves with no
apparent irregularities, justifying the use of linear interpolation between the R J..L data
points.
The BH model has consistently the closest RJ..L relationships to that of the SDOF
system, especially for strain hardening systems (a= 10% ). Short period BH structures
(2 to 10 stories high) have ductility demands that are even smaller than those predicted
by SDOF systems.
43
I.
The WS model shows the greatest deviation from the SDOF RJJ. relationships,
particularly for systems with no strain hardening (a= 0%).
The deviation of the MDOF R JJ. relationships from that of the SDOF system increases
with the number of stories (period). The effect of strain hardening is greatest for the
~ 2.5 Aa
1.2 Av !f
= 213
1.22 Av !f
2.5 Aa
(4.1)
According to the Commentary of the ATC306 document (ATC306, 1978), this ratio is
intended to provide more safety for long period structures and account for higher mode
44
effects. For periods exceeding 0.488 sec. this ratio amounts to 0.984Tl 13. Note that this
modification is independent of the target ductility ratio since ATC3 assumes a constant
reduction factor, R (or Rw in the 1988 UBC), irrespective of the expected ductility demand
for the given structure.
Fig. 4.11 shows the same information as Fig. 4.1 0, but plotted vs. the target ductility
ratio J.lt instead of period, for 2 and 40 story structures with 0 and 10% strain hardening.
The following observations can be made from Figs. 4.10 and 4.11 :
The required strength modifications for MDOF effects depend strongly on the target
ductility ratio, the fundamental period of the system, the type of structural system, and
the strain hardening ratio a. For BH structures, i.e., structures that involve beam
mechanisms in every story, the modification factor is smallest and is mostly in good
agreement with the ATC3 modification only if there is significant strain hardening (a=
10%). Even here the caveat exists that, for longer period structures and higher target
ductility ratios, somewhat larger modifications are required than indicated by the ATC3
approach. For short period BH structures, the base shear strength demand is
consistently lower than the corresponding SDOF strength demand, indicating that
MDOF effects are not important for this range.
For CH structures, i.e., structures in which story mechanisms form within each story
(e.g., plastic hinging in columns, or mechanisms in bracing systems), the MDOF
strength demands are higher than for BH structures. Quantitative information can be
obtained from the graphs. It can be seen that the required increase in strength compared
to the BH structures is about the same regardless of fundamental period. The figures
also clearly demonstrate that WS structures, i.e., structures with a weak first story, are
indeed a great problem. Such structures require strength capacities that may be more
than twice those required in BH structures in order to limit the story drift to the same
target ductility ratio.
A comparison between results for 0 and 10% strain hardening shows that matters get
considerably worse if the structural system has no strain hardening, particularly for CH
structures and longer periods. Systems without strain hardening drift more, and larger
strength is required in order to limit the drift to a prescribed target ductility ratio. This
effect is actually larger than indicated in the graphs, since the SDOF strength demands,
which are used as normalizing factors, are already larger for elasticplastic SDOF
systems than for strain hardening SDOF systems (see Figs. 3.10 and 3.11). The
45
differences between the results shown for 0 and 10% strain hardening in Figs. 4.10
and 4.11 come only from multimode effects.
In general, the modification factor for MDOF vs. SDOF strength demands increases
with target ductility ratios (except for 2story structures which show an increase then
decrease in modification factors for higher target ductility ratios) and strain hardening.
For the regular structures studied here, elastic MDOF systems attract lower base shears
than that predicted by the equivalent SDOF systems.
These results clearly demonstrate that an estimation of the required strength in the
design process needs to be based on criteria that consider the fundamental period and
distinguish between different "failure" mechanisms and anticipated strain hardening. These
factors may significantly affect the ductility demand the structures will experience in an
earthquake. When drawing quantitative conclusions from the presented data it needs to be
emphasized that the presented results are based on mean values obtained from 15 firm soil
records and, in their final presentation as shown in Figs. 4.10 and 4.11, are derived from
straight line interpolation between discrete ductility values and discrete periods. These
straight line interpolations may lead to considerable approximations for systems for which
the SDOF and MDOF ductility demands differ significantly (e.g., WS structures with high
target ductility ratios).
The foregoing discussion focused on a procedure that can be employed to derive
design strength demands for MDOF systems from inelastic strength demand spectra of
SDOF systems. The presented numerical results apply only within the constraints
identified in this chapter and cannot be generalized without a much more comprehensive
parametric study. The parameters that need to be considered include the frequency content
of the ground motions (which may be greatly affected by local site conditions), the
hysteretic characteristics of the structural models (strain hardening, stiffness degradation,
strength deterioration, etc.), and the dynamic characteristics of the MDOF models (periods,
mode shapes and modal masses of all important modes, as well as stiffness and strength
discontinuities). Furthermore, it must be pointed out that the RJ.l relationships for MDOF
systems developed here are based on ductility demands in the first story. In several cases
the maximum story ductility demand did not occur in the first story, but close to the top of
the structure. This issue is not considered here, but can be addressed through
modifications to the distribution of design story forces over the height for structures
46
from a base shear equal to the strength demand of the corresponding SDOF system and the
1988 UBC seismic load pattern. All results in this section are presented for specific SDOF
target ductility ratios Jlt; the corresponding MDOF story ductility ratios can be obtained
from Figs. 4.4 to 4.6.
= 2.
Figure 4.15 is for 0% strain hardening and SDOF target ductility ratio Jlt
47
= 8 and
Fig. 4.16 is for 10% strain hardening and SDOF target ductility ratio Jlt = 8. The following
can be noted from Figs. 4.14 to 4.16:
For CH structures all stories yield, resulting in uniform ratios of 1.0 over the entire
height of the structures for elasticperfectly plastic systems (a= 0%), or ratios of 1.0 +
a (p,i 1) at story i for systems with strain hardening ratio a (see Figs. 4.4 to 4.6). At
any given story, once the dynamic story shear reaches its yield capacity, the story
yields and develops a story mechanism, thus preventing any further transfer of shear
forces across this story unless there is some strain hardening. This is evident by
comparing Fig. 4.15 with 4.16 parts (b) for CH structures for 0 and 10% strain
hardening.
For BH structures, the ratios are always greater than those of the CH structures and
greater than 1.0 even for a= 0%. This is due to the inability of the BH model to
develop a story mechanism, thus attracting more story shear forces. For a= 0%, the
BH model can attract an additionall020% story shears due to this effect for Jlr
= 2 and
up to 100% more for Jlt = 8. Note that this would not occur if the BH model were
incrementally loaded with the 1988 UBC load pattern, since a complete structure
mechanism would develop.
The large story shear forces attracted above the first weak story of the WS structures
are illustrated in parts (c) of Figs. 4.14 to 4.16. In contrast to the CH model, making
the stories above the first strong enough to remain elastic attracts much larger story
shear forces, especially in the upper stories. These large shear forces may not pose a
problem by themselves, since for these structures the cause of the weak story is
excessive shear strengths in all upper stories, however, as will be discussed later, they
generate very large overturning moments that have to be resisted in the first story. The
first story will have to resist high axial forces on its columns, besides having to
dissipate most of the seismic energy imparted to the structure. The Pdelta effect may
become of major concern for these structures. It is worth noting that the maximum
story shears in the upper stories can reach up to 2 (or 6) times the shears obtained from
the 1988 UBC load pattern, for 0% strain hardening in the first story and Jlt = 2 (<;>r 8),
and even more for 10% strain hardening.
Figures 4.17 and 4.18 show shear forces for 40story structures with SDOF target
ductilitiy ratios of 2 and 8, normalized with respect to the design base shears. The thin
48
lines refer to the UBC load pattern normalized to a value of 1.0 at the base. The following
observations can be made from these figures :
For inelastic BH and CH structures, the governing dynamic load pattern is closer to a
rectangular (from the almost constant slope shear force diagram) rather than an inverted
triangular shape. This is attributed to the full development of story shear capacities as
illustrated in Fig. 4.14 to 4.16. On the other hand, the WS structures can be
characterized by a 3step load pattern with maximum intensity at the upper stories and
minimum at the lower stories.
The deviation of dynamic story shears from the design values increases with strain
hardening, target ductility ratio, and towards the bottom of the structures. The
deviation greatly depends on the structural system. It is largest for the WS structures
and smallest for the CH structures.
49
Figure 4.20 shows the mean distribution of overturning moments, normalized with
respect to design overturning moments at each story, for 2 to 40story elastic MDOF
systems. For elastic systems, the moments are close to the design values at the upper
stories (usually not of much importance) and significantly smaller at the lower stories. This
may justify the use of overturning moment reduction factors. However, for elastic
systems, it needs to be emphasized that there are two reasons for the relatively small values
shown at the base of all structures. First, is that the base shears attracted by the elastic
MDOF systems are smaller than the design base shears (see Fig. 4.13), which were based
on the corresponding SDOF systems. Second, the dynamic overturning moments are
smaller than those based on the code seismic load pattern. The contribution of the latter can
be estimated by comparing Figs. 4.13 and 4.20. For example, the mean base shear of
40story structures is about 0.94 the design base shear, and the mean overturning moment
at the base is about 0.72 the design overturning moment. Thus, the actual reduction in
overturning moments is 0.72/0.94
= 0.77.
For inelastic systems, a much discussed code design issue is whether it is necessary
to design structures for overturning moments that are produced by the story shear
capacities, or whether it is justifiable to use reduced overturning moments assuming that
not all stories yield simultaneously.
Figures 4.21 to 4.23 show the maximum dynamic story overturning moments for an
SDOF target ductility ratio Jlt
10%, respectively.
correspond to that on shear forces in Figs. 4.14 to 4.18. The following observations can
be made from Figs. 4.21 to 4.23:
The presented results provide much evidence that for structures whose story strengths
are tuned to the design level (all BH and CH structures), simultaneous yielding in most
stories is vezy likely and. therefore. no reductions in overturning moments should be
applied. In the presented graphs, a value of 1.0 implies no overturning moment
reduction, and, as the figures show, this value is reached or exceeded in most cases.
For systems without strain hardening, the CH structures develop the maximum design
overturing moments throughout the entire height of the structure (computed values
close to 1.0), and the BH structures develop even higher demands in the upper stories,
which increases with the target ductility Jlt and strain hardening a. For systems with
strain hardening, the dynamic overturning moments at the base of the structure may be
in the order of 50% higher than design values, for both the BH and CH structures.
50
For inelastic weak story (WS) structures, the dynamic overturning moments become
very large compared to the design overturning moments. For these types of structures,
the ductility demands are very high (Figs. 4.4 to 4.6), the story shear forces are much
larger than the shear strength of the first story (Figs. 4.14 to 4.16), and the overturning
moments exceeds those based on first story strength and code seismic shear distribution
by a large margin (Figs. 4.21 to 4.23). All these issues need to be considered in the
design of weak story structures.
The results presented here give a clear indication that overturning moments in inelastic
structures can be very large. If the story strengths are tuned to the code seismic load
pattern and base shears equal to that predicted by the equivalent SDOF systems for a
prescribed target ductility ratio (as was done in all BH and CH structures), it should be
assumed that all stories will yield simultaneously and the maximum overturning moments
should be based on the shear strengths of all stories above with due consideration given to
strain hardening. No overturning moment reduction factors should be applied. In most
real designs, the story strengths cannot be tuned exactly to the code seismic load pattern
and individual stories may have a shear strength larger than required. In such cases the
overturning moments will further increase. This study provides no information on the
magnitude of this increase, as it depends on the relative strength of each story and cannot
be generalized.
Extreme strength discontinuities, such as those in the WS structures, should be
avoided, whenever possible, as they lead to excessive ductility and overturning moment
demands that may be greatly amplified by the portions of the structure that remain elastic.
51
5.
SDOF inelastic strength demand in order to account for higher mode effects. A
52
procedure for implementing this modification for regular structures is discussed in this
report. Quantitative conclusions are drawn on the effects of different "failure"
mechanisms on the required structure strength.
The assessment of overturning moments for MDOF systems has shown that in inelastic
systems simultaneous yielding of all stories is a very likely event if the strength
distribution over the height follows the code prescribed load pattern. No overturning
moment reduction factors should be applied, except for structures that respond elastically
in an earthquake. For structures with a weak story (WS structures), the overturning
moments may be much larger than anticipated from the strength of the weak story.
6.
ACKNOWLEDGEMENTS
The work summarized here was in part supported by a grant given by Kajima
Corporation and administered by CUREe, the California Universities for Research in
Earthquake Engineering. Additional support was provided by the John A. Blume
Earthquake Engineering Center at Stanford University, and the Stanford/USGS Institute
for Research in Earthquake Engineering and Seismology. The support from all these
sources is gratefully acknowledged.
53
REFERENCES
ATC 306. (1978). "Tentative Provisions for the Development of Seismic Regulations for
Buildings," Applied Technology Council, June 1978.
Boore, D.M., Seekins, L., and Joyner, W.B. (1989). "Peak Accelerations from the 17
October 1989 Lorna Prieta Earthquake," Seismological Research Letters, SSA, Vol. 60,
No.4, OctoberDecember 1989.
CDMG (1989). "CSMIP StrongMotion Records from the Santa Cruz Mountains (Lorna
Prieta), California Earthquake of 17 October 1989," California Department of
Conservation, Division of Mines and Geology, Office of Strong Motion Studies, Report
No. OSMS 8906.
CDMG (1990). "CSMIP Processed StrongMotion Records from the Santa Cruz
Mountains (Lorna Prieta) Earthquake of 17 October 1989," California Department of
Conservation, Division of Mines and Geology, Office of Strong Motion Studies, Report
No. OSMS 9005, August 1990.
Chung, Y.S., Meyer, C., and Shinozuka, M. (1987) "Seismic Damage Assessment of
Reinforced Concrete Members," Report NCEER870022, National Center for Earthquake
Engineering Research, State University of New York at Buffalo, October 1987.
Competing Against Time, (1990). The Governor's Board of Inquiry on the 1989 Lorna
Prieta Earthquake, George W. Housner, Chairman, May 1990.
ldriss, I.M. (1990). "Response of Soft Soil Sites During Earthquakes," Proceedings, a
memorial symposium to honor Professor H.B. Seed, University of California, Berkeley,
May 1990.
Joyner, W.B., and Boore, D.M. (1988). "Measurement, Characterization, and Prediction
of Strong Ground Motion," Proceedings of Earthquake Engineering & Soil Dynamics II,
GT Div/ASCE, Park City, Utah, June 2730, 1988.
Krawinkler, H., and Nassar, A. (1990). "Strength and Ductility Demands for SDOF and
MDOF Systems Subjected to Whittier Narrows Earthquake Ground Motions," SMIP90,
Proceedings of the Seminar on Seismological and Engineering Implications of Recent
StrongMotion Data, California Department of Conservation, Sacramento, California, June
8, 1990.
Krawinkler, H., and Nassar, A. ( 19901 ). "Damage Potential of Whittier Narrows Ground
Motions," preliminary report submitted to California Department of Conservation, April
1990.
Krawinkler, H., Zohrei, M., Lashkari, B., Cofie, N ., and Hadidi, H. (1983).
"Recommendations for Experimental Studies on the Seismic Behavior of Steel Components
and Materials," John A. Blume Earthquake Engineering Center, Report No. 61,
Department of Civil Engineering, Stanford University.
Mahin, S.A., and Bertero, V.V. (1975). "An Evaluation of Some Methods for Predicting
the Seismic Behavior of Reinforced Concrete Buildings," EERC Report 75/5, Earthquake
Engineering Research Center, University of California, Berkeley, 1975.
54
55
Table 3.1 Specific Hysteresis Models Generated from the General Model Shown in Fig. 3.1
SKELETON PARAMETEilS
MODEL
Fy
Fs
HYSTERESIS PARAMETERS
Ke
Kt
Kh
Ke
Kt
Kh
Fy
Fs
Ke
Basic Bilinear
Ke
Kh
Fy
Ke
Kh
Fy
Ke
Kh
l'y
Clough Model
basic Stiffness Degradation
Ke
Kh
ry
Ke
Ke
Kh
l'y
Ke
Kh
Fy
Ke
Kh
Ke
Ke
Ke
'
Basic Trilinear
Ku
Kr
... j)
B,
Kp
... d
Ke
Ke
Ke
Ke
Ke
ps rd 0
'
Rule
omax
Ke
Rule
I\ 0 to
.
Ke
Rule
Omax
l'y
Pu Ku,o
Rule
Omax
Kh
Fy
Ke
Rule
Rule
Rule
Om ax
Kh
Fy
Ke
Rule
Rule
Rule
1\ ot.()
Kh
Fy
Ke
Rule
Hule
Rule
0 max
Pu K11,0 puKu,o
ps rd 0
ps rd,o
Table 3.2 Properties of the 15 S 1 Records from Western U.S. Eanhquakes Used in Statistical Study
175.9
17:7
392.4
30.0
156.8
20.5
392.4
30.0
.5
392.4
30.0
30.0
30.0
VI
~
6.
27.0
5.6
192.7
6.3
5.
12.0
6.5
39.1
6.4
29.3
392.4
59.
15.8
392.4
39.7
296.2
30.0
44.0
392.4
70.1
167.9
30.0
32.3
364.5
.0
228.1
24.9
178.7
14.7
392.4
392.4
Table 3.3. Lorna Prieta Rock Site Records Used in this Study
Name
Comp.
Number
57007
47379
58065
1601
58127
47189
47377
58219
58130
58338
58151
58163
58133
58131
58132
58043
0
90
0
360
90
261
0
90
90
45
90
90
90
270
90
297
Distance (km)
PGA (cm/sec. 11 2) PGV (em/sec.)
Epicenter to Rupture
7
1
618
55.2
29
15
434
33.8
16
16
433
21.2
51
33
282
28.4
55
36
14.7
80
54
39
71
10.3
44
49
69
3.3
71
53
7.4
83
73
92
111
14.3
74
93
81
9.2
76
95
89
11.6
77
95
89
11.6
78
97
91
9.6
78
97
14.3
60
99
80
106
21.0
104
85
71
13.6
Table 3.4. Lorna Prieta Alluvium Site Records Used in this Study
Name
Comp.
0
0
0
0
0
90
255
0
0
250
0
90
0
310
280
Distance (km)
Epicenter to Rupture
27
9
14
9
16
30
18
31
40
25
40
28
30
45
31
37
32
32
46
34
55
39
56
40
71
53
73
55
108
89
58
41.3
36.1
33.3
34.5
30.9
16.3
36.6
12.5
39.1
15.8
10.2
10.8
13.6
11.8
17.1
Table 3.5. Other Lorna Prieta Records (Mostly Soft Soil Sites)
Name
Comp.
67
90
90
360
0
43
90
90
205
290
35
90
90
260
90
270
0
Distance (km)
PGA (em/sec. "2) PGV (em/sec.)
Epicenter to Rupture
29
15
349
28.9
28
16
280
43.5
24
35
167
13.9
43
215
33.4
48
362
33
62.8
272
63
53.1
44
63
278
45.4
79
60
326
29.3
84
65
103
8.2
92
73
238
37.9
76
281
95
40.8
79
98
195
33.5
79
98
156
33.4
97
255
41.1
114
99
80
22.0
100
239
35.5
138
119
158
18.5
59
Name
Source
Ep. Dist.
(km)
PGA
(cm/sec2)
PGV
(em/sec)
282
28
USGS #1601
51
USGS
51
,_.).)
33
USGS #1227
47
378
USGS #1230
54
288
.:+0
24
Source
Name
(km)
PGV
Site Cond.
PGA
(cm/sec2) (em/sec)
CDM.G #47524
48
362
63
"alluvium
63
278
45
"alluvium"
CDMG#58223
79
326
29
"alluvium"
92
238
38
"alluvium''
97
255
41
fill on mud
USGS #1601
51
282
28
sed. rock
Stanford, SLAC
Record
Magnitude
. Date
2.14.1956
3.27.1963
3.27.1963
6.12.1978
6.12.1978
4.30.1962
5.16.1968
5.16.1968
5.9
6.9
6.9
7.4
7.4
6.5
7.9
7.9
Unsealed
PGA cm/sec."2
74.00
19.00
10.00
258.23
240.90
45.00
224.95
182.93
60
6.08
2.23
1.29
36.15
36.32
4.73
33.27
40.43
PGA cm/sec.":Z
608.55
426.00
387.59
357.16
331.63
475.68
338.07
226.23
PG V em/sec.
50.00
50.00
:5~J.oo
50.00
50.00
50.00
50.00
50.00
Table 4.1 Modal Periods and %Mass for MDOF Structures Used in this Study
2STORY
Mode
#
1
2
3
4
5
(sec)
Mass
%
0.217
0.089
90
10
5STORY
10STORY
20STORY
30STORY
40STORY
(sec)
Mass
%
(sec)
Mass
%
(sec)
Mass
%
(sec)
Mass
%
(sec)
Mass
%
0.431
0.176
0.111
0.082
0.064
82
11
4
2
1
0.725
0.288
0.181
0.133
0.106
80
11
4
2
1
1.220
0.475
0.294
0.214
0.169
78
11
4
2
1
1.653
0.636
0.391
0.282
0.221
78
11
4
2
1
2.051
0.781
0.479
0.345
0.269
78
11
4
2
1
61
MOMENT FRAME
PERIMETER FRAME
8
BRACED FRAME
 
0 0~
1
4
2
3
ELASTIC PERIOD (seconds)
Fig. 2.1. Estimates of Local Ductility Demands for Code Designed Steel Frame Structures
Designed According to the 1988 UBC (SEAOC Approach)
(from Osteraas and Krawinkler, 1990)
62
DAMAGE THRESHOLD
COLLAPSE THRESHOLD
Identify acceptable
deflection and strength
reduction factors
Identify member
ductility capacities
Local
ductility
capacities
Global
ductility
capacity
t
Global
ductility
capacity
Global
strength
demand
I
Derive local
strength and stiffness
demands
'
Global
strength
demand
Local
strength
demand
/
Compare local strength demands
Design tor controlling condition
Revise design
!
Static, nonlinear analysis
I
Revise assumptions
with targets
and assumptions
Final design
Fig. 2.2 Flow Chart of Ductility and Drift Based Dual Design Approach
(from Osteraas and Krawinkler, 1990)
63
RESISTANCE
RESISTANCE
DISPL
64
Ku
I
I
r
L
Fs
Ke = Elastic Stiffness
Kt =Transition Stiffness
Kh =Hardening Stiffness
F y = Yield Strength
F s = Structure Strength
Ku =Unloading Stiffness
Kr =Reloading Stiffness
Kp =Pinching Stiffness
F d =Yield Strength after Strength Deterioration
F p = Pinching Target Resistance
ot
= Target Displacement
65
SPEC.
..
E
Bt<;
i.'?
12 00
__/
...J
~.)7,)9
I
L
0
..
e.co<~
DETERIORATION
TIIRESIIOLD
4.8"
"1
r'
s
.00
11
E<
V1
H
V1
 ee
(a)
e .ee_
1.59
c .....
I.
5"
EN!:' DEF't.EC:TION IN
J.eCI
NUHBER OF CYCLES
c
.
E
121
12 . .
..
...
IC
4 ..
II
I.
0
D.T
DETERIOHATTON
~r
!I
.ee
4 . . .
(b)
 ..
_,
....
a ...
NlJNI.IER OF CYCJ.lS
66
DET.
67
2
4
c:2
l
i
0.5
= 100
c:2.0
1.2
1.0
0.8
o
u.
0.6
0.4
0.2
0.0
0
10
Number of Cycles
68
12
C:1
C:2.0
69
C:1.0
=100
C:1.0
I
I
i
1 ~
4
70
c:1.0
c:1.0
1.5 ..;..
3 ,__ _ _ _ _ _ _~~~~~
CIS
E
Q)
>.
~2+=:::......,.,..l
:l
1 ~.~~
0.00
0.04
0.08
0.12
Time {sec)
(b) Displacement Time Histories for the Three SDOF Systems with J.L 3 in Fig. 3.9(a)
Fig. 3.9 Problems in the Determination of Inelastic Strength Demands for Given J.L
(from Krawinkler and Nassar, 1991)
72
12
10
8
......
''>
a:
6
4
0.0
0.5
1.0
1.5
2.5
2.0
3.0
3.5
4.0
T (sec)
(a) Mean Values
~~~~_j~===~==========~====~
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
. T {sec)
(b) Means minus Standard Deviation
Fig. 3.10 Strength Reduction Factors for Bilinear Systems with no Strain Hardening
73
.::;.6
a:
2
J.l = 2, 3. 4, 5, 6. 8 (thin
~thick lines) ~
0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
T (sec)
.~~~,~~,,
!
I
~thick lines) ~
0~,~~++~==~===r====r===~
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
T (sec)
(b) Means minus Standard Deviation
Fig. 3.11 Strength Reduction Factors for Bilinear Systems with 10% Strain Hard~ning
74
=0%
 mean
80 ~~~~~~~,~~~
)1='2.3,J,5,6,8 (thin thick :ines)
30 ~~~~~!~~~
>>
:.c
~ JO ~~~~~~~~~~~~~
UJ
....
0.0
1.0
0.5
1.5
2.5
2.0
3.0
4.0
3.5
T (sec)
A ,._
40
20
1'4 \
V\
V'
I
I
 
0
0.0
.........
j1
1
~
,.___ I
1.0
I
l
0.5
 mean+<r
i.
:I
~
_\j ~ ~"" ~
./""'k
'oj
=0%
1.5
2.0
2.5
'
'
.
I
'
3.0
'
I
3.5
4.0
T (sec)
75
=10%
 mean
80 ~~~~~~
0.0
0.5
, .0 .
, .5
2.0
2.5
3.0
3.5
4.0
T (sec)
100 ,_~~~~++~~~~~
o~~~~~==~==t===C===~=====d
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
T (Set?)
(b) Means plus Standard Deviation
Fig. 3.13 Norm. Hysteretic Energy Demands for Bilinear Systems with 10% Strain Hardg
76
=0%
 mean
40 ,~~~~.~~
~o<=2.3.4..S.6.S
!
i
30 ~F;~~~~~
0 ~.~~~~~~
0.0
0.5
1.0
.5
2.5
2.0
3.0
3.5
4.0
T (sec)
strain hardening
= 0%
 mean+cr
60 ,~~~~~~~~
~2.3.4.5.6.8
50 ~7r~~~'~~~
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
T (sec)
77
= 10%
 mean
40 ~~~,~~
~2.3.1.5.6.3
0 ~~~~~~~~
4.0
3.5
3:J
2.5
2.0
1.5
1.0
0.5
0.0
T (sec)
=10%
 mean+cr
60 .~~~~~
~2.3,4,5,6,8
50 ,_H+~~~~+~~~~
0 ~~~~~~~~
1.5
0.0
1.0
4.0
0.5
2.0
2.5
3.0
3.5
T (sec)
Fig. 3.15 Number of Inelastic Excursions for Bilinear Systems with 10% Strain Hardg
78
5 ~~~~~~~~~~~~
J.l"'2.3,4,5,6,3 (thin thick lines)
4+~'~~~r7~+~
~3 +~~~~~~~~~~====~
:.c
0~~~~~~~~~
0.0
, .0
0.5
1.5
2.0
2.5
3.0
3.5
4.0
T (sec}
=10%
 mean
5 ~~~~~~~~~~~
!J"'2.3,4.S,6.8 (tl'lin illicit lines)
r1
~I
~rr~;;:;;;;;;;;:;=:
~3+~1~~~~~~==~~+~
:.c

~~2~~~~~~~~~~~~f=~:r~====~
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0.
T (sec}
79
=0%
strain hardening
1.0
UJ
Q
,..,.
';
y~
~ 
0.5
.....
UJ
.,...
/""',
OA
I
'
'
'
0.2
i'
\
I!
!I
''
ii
i
i
!
!
0.0
I
I
I
i
0.8
iI
 mean
iI
;
3.0
3.5
'
0.0
0.5
1.Q
1.5
2.0
2.5
4.J
T (sec)
I
i
0.8
0.5
;
:
;'
UJ
::r: 0.4
../
0.2
i
I
'
,
I
I
~i
.....
~2.3.4.5.6.8
II
'
0.0
0.5
I
1.0
1.5
I
I
I
I
2.0
2.5
3.0
3.5
0.0
!~
I
I
J'V
 mean
'
~
UJ
Q
= 10%
4.0
T (sec)
80
1.4
....
0
.!.
.0
i
I
1.2
!
'
1.0
.......
.....
;"'
0.6
0.4
;
I
i
>o
0.2
0.0
J..L
i'
::::::::...::::.
'
!
I
;
I
I
i'
iI
II
2.0
2.5
3.0
3.5
I
'
~
~
I
II
I!
i!
!
I
!I
""'
''
a:
~.....
.!.
.0
~!~
";? 0.8
g
0.5
0.0
1.0
1.5
4.0
T (sec)
,..
l..o.
1.0
"'. K:J
,,r
~ 0.8
0.6
s.
w
0.4
I
I
'(
w
:::c:

I
I
I J.I"'2.3.4.S.6.8 (thin 
:z:::
0.2
0.5
1.0
1.5
2.0
2.5
3.0
3.5
0.0
0.0
ltlic:k lines)
4.0
T (sec}
81
1.2 ,__.;.;:~;::iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiil
ATC5 1
,.1
,.=2
,.3
,..6
1.0
j.ia4
J.l"S
0.8
J.1.=8
Cl
"";.. 0.6
u..
0.4
0.2
0.0
J.:.~.:===~~~~~~~~~;~~~~~;~
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
T (sec)
1.6
1.4
J.ll
1.2
1'2
J.1.3
J.L4
J.LS
J.L6
1.0
,. .. s
C)
"";.. 0.8
u..
0.8
0.4
0.2
0.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
35
4.0
T (sec)
82
1.2 r........:.::~;iiiiiiiiiiiiiiiii;iiiiiiiiiiiii;l
  ATCS 1
~,.1
1.0
~,.2
~ .. 3
~ .. 4
~
0.8
.. s
~=6
~=8
C)
;.. 0.6
u.
0.4
0.2
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
T (sec)
~~~ffi
ATCS 1
~o~l
1.4
1''"2
~o~3
1.2
~o~a4.
~o~aS
~o~6
1.0
~o~8
;.. 0.8
LL.
0.8
0.4
0.2~~~~~
0.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
T (sec)
=0.4g
83
..
.;iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiJ
ATC.S,
~&1
1.0
~&2
~&3
~&4
~&S
0.8
~&6
).l8
CD
;. 0.6
.....
0.4
0.2
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
T (sec)
1.4
~&1
~&2
~&3
1.2
~&4
~&S
1.0
~&6
J.LI
CD
"";. 0.8
.....
0.8
0.4
0.0
05
1.0
1.5
2.0
2.5
3.0
3.5
4.0
T (sec)
84
.....
0
I
)\
.0
:::::: 1.0
'
"'h.

0.8
I
i
C)
:::::: 0.4
:::::t.
>
a:
0.2
0.0
l.::
I
i
i
0.5
1.0
1.5
2.0
2.5
0.0
I
I
1.
I
!
'0
II
~ 0.6
vi
I
I
k,
.!..
rl"
1.2
II
!I
3.0
3.5
4.0
T (sec)
04
0
J.1 =2, 3, 4,
;I
I
I
iI
.!..
.0
'
w3
:1:
9C) 2
z
'0
:1:
I
I
I:
I
I
1\
Ii
~~!
0.0
0.5
1.0
1.5
2.0
2.5
3.0
iI
I
I
I
3.5
4.0
T (sec)
85
ATTENUATION OF PGA
700
Cll
<
"
a..
'b0
u
0
500
e
Cll
400
I
\
300
''
200
100
... ......
<
"a..
..~
00
600
Alluvium Sites
700
600
....
e
ATTENUATION OF PGA
Rock Sites
\
\
400
I
I
''
300
200
''
100
500
...........
04~~.~4
o
~
~
oo
ao
100
ATTENUATION OF PGA
Alluvium & Rock Sites
700
Alluvium Sites
Roc:!< S~es
600
'b0
Ill
500
400
<
300
"
a..
200
 
100
so
40
eo
100
Fig. 3.26 Attenuation Relationships Derived from Rock and Alluvium Site Records
86
1.4
r0 = 10 km
r0 = 20 km
1.2
r0 =40 km
r0 = 80 km
1.0
SMF88
CMF68
0.8
.9
.......
a)
:>:
u..
.......
0.6
.............
0.4

0.2

0.
0.
0.5
1.0
1.5
2.0
T {sec)
1.4
r0
.9
1.2
1.0
r0 = 40 km
r0 =80 km
""
"
0.8
a)
:>:
u..
= 10 km
= 20 km
r0
SMF88
.......
CMF68
.......
.......
......
......
0.6
0.4
0.2
    

0.
0.
0.5
1.0
1.5
T {sec)
87
2.0
ro a 20 km
ro a 40 km
3.0
u.
ct
r0 80km
2.0
1.0
0.
0.5
1.0
1.5
2.0
T (sec)
3.0
u.
ct
Q
2.0
1.0
0.
0.5
1.0
1.5
2.0
T (sec)
88
0.6
1.1.
=4
I
I
I
I
I
I
0.5
r0 10km
r0 20 km
r0 40 km
r0 80 km
I
I
I
I

0.4
I
I
......,
.:;
''
0.3
''
::...
""
SMF88
CMF68
',, ,,
0.2
...........
0.1
,,
 
0.
0.
0.5
1.0
1.5
2.0
T (sec)
0.6
=4
I
\
0.5
r0 10km
r0 20km
r0 40km
r0 SOkm
SMF88
CMF68
I
I
I
I

I
I
0.4
......,
.:;
''
0.3
::...
""
''
''
,
,,
,,

0.2
 ......... _

0.1
0.
0.
0.5
1.0
1.5
2.0
T (sec)
89
1.0
r0 10km
r0 20 km
r0 .. 40 km
r0 80 km
"tJ
c
ca
E
Q)
0.8
...
.c
en
c
Q)
"2
0.6
UJ
0.4
.!::!
16
...0
0.2
0.
0.
1.0
0.5
1.5
2.0
T (sec)
r0 10km
r0 20km
r0 40km
r0 80km
"C
c
ca
E
Q)
0.8
.c
a
cQ)
i...
en
0.6
0.4
iii
...E
0
0.2
0.
0.
0.5
1.0
1.5
2.0
T (sec)
90
1.4
=10 km
J.l=l
J.1=2
J.1=3
J.l=4
1.2
1.0
s
SMF88
CMF68
0.8
G)
LL.
0.6
0.4
0.2
0.
0.
1.5
1.0
0.5
2.0
T (sec)
1.4
J.l=l
J.1=2
  J.l=3
J.l=4
1.2
1.0

0.8
LL.
0.6
C)
=10 km
SMF88
CMF68
CD
0.4
0.2
0.
0.
0.5
1.0
1.5
2.0
T (sec)
91
1.4
r,

SLAC
_sPG
1.2
VAMP
  VAPA
1.0
0.8
~
II.
0.8
0.4
0.2
o.
0.5
1.0
1.5
2.0
T (sec)
Fig. 3.32 Elastic Response Spectra of Motions Recorded at or Near Stanford University
92
1.5 ,...._ _ _
___
_ _Str.
__
__...,;(e
__
_ _ _,
Hollister,
South
& Pine
48_
km)
1'=1
1'=2
1'=3
1'=4
ZC (UBC 1988)
1.0
).'=I
ZC (USC 1988)
5 2 54
"\~,
1.0
5MF88
', ',
CMF68
,.
',
,.
.........
u.
u.
',,
........
f'=4
__
CMF68
_ ,____  
.... ....
..........................
....
0.5
0.0
5MF88
.....
_____ 
 

+::~~~
0.0
0.5
1.0
2.0
1.5
T (sec)
T (sec)
ZC (UBC 1968)
1.5 ...,..._ _ _
_ _International
_ _ _ _ _Airport
_...;__;....;.,
...:.,._ _..,
SFO,
(e_ _
79 _
km)
1'=1
1'=2
1'=3
1'=4
52 54
1.0
5MF88
CMF68
0.5
1.5
2.0
T (sec)
T (sec)
1.5
Emeryville,
6363
Christie
Av._(e_=_
97_km)
___
__
___
_ _ _,
....~.....;
I'= I
1'=2
1'=3
1'=4
1.0
5MF88
CMF68
,.
u.
0.5
0.5
1.0
1.5
2.0
T (sec)
T (sec)
Fig. 3.33 Strength Demand Spectra for Six Lorna Prieta Ground Motions
93
~4.0
Q.
" '0
; 3.0
...
.::: 2.0
CD
cC),
U) 1.0
_~
0.0
0.5
1.0
1.5
2.0
T (sec)
Fig. 3.34 Normalized Elastic Strength Demand Spectra
8.0
6.0
4.0
2.0
0.0
+r,r,...
..
1.5
2.0
0.5
1.0
0.0
T (sec)
Fig. 3.35 Strength Reduction Factors for Foster City I Redwood Shores Record
94
).1.=4
1.0 . ,     :       . . . . . ; . .          ....
<
(!)
Emeryville
Foster City
Hollister
Oakland
SFO Airport
Stanford
0.8
.c, 0.6
c.
ca
ID
Q
..

.c 0.4
Cl
c
ID
...
tn 0.2
0.0
'
tr........,.......................,....................
0.0
0.5
1.0
1.5
2.0
T (sec)
Fig. 3.36 Normalized Inelastic Strength Demand Spectra (for f.1 =4)
STRENGTHDEMANDSPECTRA
stanford, SLAC (e =51 km)
1.5,.....;..~~
Corrected
Hand Digitized
1.0
T (sec)
Fig. 3.37 Spectra Obtained from HandDigitized and Corrected Stanford/SLAC Record
95
0.5
SMF88

SPF88
0.4
>
CMF68
C)
0.3
CMF88
\
\
u.
0.2
,.
' ..... , ........ ......  ... ,
......... ......... ... .... ..._
.... ____
. 
0.1
==~~~~
0.0
0.0
1.0
0.5
1.5
2.0
T (sec)
Fig. 3.38 Strength Capacity of Five Types of Code Designed Building Structures
96
Tokyo
(NS) _ _ _ _ _ _ ____,
_101
_.....;..___;_
.:~___;;
j..L=l
j..L=2
j..L=3
j..L=4
, .0
j..L=5
j..L=6
Ci
j..L=8
>.
0.5
0.
JL~~~~_:~~~~~~~~~~~~~~~;;~~
0.
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
T (sec)
, .5 .......;....___;__ _ _ _ _ _ ____,
j..L=l
j..L=2
j..L=3
j..L=4
1L=S
1.0
j..L=6
c;
j..L=8
>.
0.5
0.
0.5
, .0
1.5
2.0
2.5
3.0
3.5
4.0
T (sec)
97
2.5 r......!.=_.:;_~.,
Hachinohe EW
Hachinohe NS
2.0
Nagoya 306 NS
Osaka 205 EW
Sendai 030 EW
1.5
Sendai 038 EW
Sendai 501 EW
Tokyo 101 NS
u.>1.0
s
0.5
0.
0.
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
T (sec)
1.0r~~~
Hachinohe EW
Hachinohe NS
Nagoya 306 NS
Osaka 205 EW
Sendai 030 EW
c;
.._..
u?
Sendai 038 EW
Sendai 501 EW
Tokyo 101 NS
0.5
0.
0.5
1.0
1.5
2.0
2.5
3.0
3.5
T (sec)
98
4.0
1.0r~~
C)
.._.
Hachinohe EW
Hachinohe NS
Nagoya 306 NS
Osaka 205 EW
Sendai 030 EW
Sendai 038 EW
Sendai 501 EW
Tokyo 101 NS
0.5
u.?
0. LI~~~~~~~::~~====~::~~~~~~~;;~~
0.
0.5
1.0
2.0'
1.5
2.5
3.0
3.5
4.0
T (sec)
1.0~~~~
Hachinohe EW
Hachinohe NS
Nagoya 306 NS
Osaka 205 EW
Sendai 030 EW
Sendai 038 EW
Sendai 501 EW
Tokyo 101 NS
0.
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
T (sec)
99
30.0r~~~~
Hachinohe EW
Hachinohe NS
Nagoya 306 NS
Osaka 205 EW
Sendai 030 EW
Sendai 038 EW
Sendai 501 EW
Tokyo 101 NS
20.0
c.o>>
u.
UJ
::z:::
10.0
0.
0.
0.5
2.0
1.0
2.5
3.0
3.5
4.0
T (sec)
70.0
=4
Hachinohe EW
Hachinohe NS
Nagoya 306 NS
Osaka 205 EW
Sendai 030 EW
Sendai 038 EW
Sendai 501 EW
Tokyo 101 NS
60.0
50.0
. >> 40.0
c.o
u.
UJ
J:
30.0
20.0
10.0
0.
o.
0.5
1.0
1.5
2.0
2.5
3.0
3.5
T (sec)
100
4.0
HYSTERETIC ENERGY
Inelastic Systems J.L = 2
25000.0
J\
20000.0
( ,)
Q)
15000.0
tn
(,)
UJ
J:
Hachinohe EW
Hachinohe NS
Nagoya 306 NS
Osaka 205 EW
Sendai 030 EW
Sendai 038 EW
Sendai 501 EW
Tokyo 101 NS
10000.0
I
/'\
5000.0
I\
I \
I \
I \
I
\
I
\
0.
0.
1.0
1.5
2.0
2.5
3.0
3.5
4.0
T (sec)
HYSTERETIC ENERGY
Inelastic Systems J.L = 4
20000.0r~~,
Hachinohe EW
Hachinohe NS
Nagoya 306 NS
Osaka 205 EW
Sendai 030 EW
Sendai 038 EW
Sendai 501 EW
Tokyo 101 NS
15000.0
~

( ,)
Q)
10000.0
(,)
UJ
J:
5000.0
0.
0.5
1.0
1.5
2.0
2.5
3.0
3.5
T (sec)
101
4.0
.___ \.. kp
y
 
Qy
.. .a..............
..... ....
I
........
102
100
~. ~~~~~~~ll'.~
<
lOlJ
Cm/S e
c:)
,,~~
20
l 5
10
5. 0
1. 0
2. 0
3. 0
PERIOO<SECl
.,.,,
1000
500
0
0
1. 0
2. 0
PERIOD<SECJ
(c) Elastic Response Specrtum
3. 0
100
u
u
<
HI.J
(Cm/Sec)
20
..~~
1~
10
5. 0
PER[OD<SECJ
c CoIl
I 5 00
r.,,,
1000
500
0
0
2. 0
1. 0
PER[OD<SECJ
3. 0
'"
,,
I
l5
l0
PERlOD<SECl
1 500
!000
~
500
lr~
r.._
l. 0
2. 0
PERlOD<SECl

3. 0
..:
X
w
0
z
w
''
<
<
:::E
0
.;
"':,,:::;::::~
''
1.0
o.'
J.O
NA7URAL PER!OD<SECl
peak Ot"ientedl
degrading type
..:
X
"'z
0
"'''<
:::E
<
.;
0. 0
I 0
' 0
NATURAL PER!OD<SECl
106
J. 0
I
I
z
w
''
<
::0:
<
c
o.o
107
l 0
Strength Capacity
;r~~
____ _,
I
I
I
I
I
SDOFI
MDOF
I
I
I
. "1
1
J.L'
Ductility Ratio, ~
1.0 ;;:;r:,:r...;...,~
'
:1
j
0.8 +:...,.._....___)rf;1!    ;      +       +      1
~
C1)
0.6
J:
'l
CD
.2:
c;
+___;,.t+t~++1
0.4
G)
a:
.,
~
L11
+!.,floolr'o~++1
:~'::
!._:
0.2
':
',
'
:
++~::__'_;r'\  . __
    +         !        1
''
''
c\ :}_ 't"j
10
15
20
25
1.0
''
~ \
t:
'
'
':
0.8
r
'
C)
J:
: ~~J
'
.c
G)
(15s.ch208~00)
0.6
CD
>
:;::
as 0.4
G)
a:
(.
'
'
.~~LL
~
'
0.2
_,
'
.
~'1
'
I
:
:
:
I
:
0.0
0
10
..:
15
20
25
109
0.8
C)
0.6
Q)
>
2,5,10,20.30,40 stories
(thick +thin lines)
~ 0.4
Q)
a:
0.2
_ _ _ _ _ _ ___.l
=odyn,1/ oy,l
(a) BH Model
DYNAMIC STORY DUCTILITY DEMANDS, Jli (15s.ch2.00)
1.0
0.8
C)
Q) 0.6
:t:
Q)
>
2,5,10,20,30,40 stories
(thick+ thin lines)
~ 0.4
Q)
c::
0.2
0.0
=odyn,l/ oy,l
(b) CH Model
DYNAMIC STORY DUCTILITY DEMANDS, Jli  (15s.ws2.00)
1.0
~~
=
~
0.8
C)
0.6
c;
o.4
Q)
2,5,10,20,30,40 stories
(thick + thin lines)
a:
0.2
_I
_l_
0.0
0
10
12
(c) WS Model
Fig. 4.4 Dynamic Story Ductility Demands for J.lt(SDOF) = 2, a= 0% (Mean for 15s.Recordsl
110
1.0
r~~rrr__;...:.......;._
_ _ _.....:...,
0.8
.C)
::
~
0.6
Cl)
>
~ 0.4
a;
a:
0.2
0.0 +._,....U........L...J.1.,.....;;.;;l.~,1
5
10
15
20
25
0
30
=odyn,i I Oy,i
(a) BH Model
DYNAMIC STORY DUCTILITY DEMANDS, J.li (15s.ch8.00)
1.0
0.8
. ::
C)
0.6
Cl)
>
~ 0.4
a;
a:
0.2
0.0
5
10
15
25
20
30
=odyn,l/ oy,l
(b) CHModel
DYNAMIC STORY DUCTILITY DEMANDS, J.li (15s.ws8.00)
1.0
!
h:
i)
0.8
.C)
::
0.6
CD
>
~ 0.4
a;
a:
0.2
0.0
0
10
20
30
40
50
60
70
80
=Sdyn,1t ay,1
(c) WS Model
Fig.4.5 Dynamic Story Ductility Demands for J.lt(SDOF) = 8, a= 0% (Mean for 15s Records)
111
1.0
~~rr
0.8
.:
0
0.6
Cl)
>
2,5,10.20,30,40 stories
(thick+ thin lines)
~ 0.4
Qi
a:
0.2
0.0 +.....L....I..L...1........=_,............J
5
10
15
20
25
30
0
J..1.i
= odyn,l I 5y,l
(a) BH Model
DYNAMIC STORY DUCTILITY DEMANDS,
J..1.i
(15s.ch8.10)
1.0
0.8
.:
C)
G) 0.6
:I:
Cl)
.::
'(; 0.4
Qi
a:
0.2
0.0
5
10
15
25
20
30
=odyn,l I Oy,J
(b) CHModel
DYNAMIC STORY DUCTILITY DEMANDS, J..1.i (15s.ws8.10)
1.0
!/ !;.
0.8
.:
0.6
Cl)
>
0.4
Qi
a:
0.2
2,5,10,20,30,40 stories
(thick +thin lines)
II
0.0
10
20
30
40
50
J..1.i
60
70
80
=8dyn,l I 5y,J
(c) WS Model
Fig. 4.6 Dynamic Story Ductility Demands for ~(SDOF) = 8, a= 10% (Mean for 15s Records)
112
..12
r/v
:1
    SOOF
 e  MOOF BH Model
o MOOF CH Model
e MDOF WS Model
g10
ca
E
~8
>
.'1:::
:56
u
::s
i!4
02
/
v
.12
r
::1..
c~10
    SDOF
. MDOF BH Model
o MDOF CH Model
e MDOF WS Model
~8
>
:1:::
su
::s
..Q
_._____ ...
i!4
...n
~'"V"
U) 2
u::
U:::o
0.5
0.0
1.5
1.0
2.5
2.0


0
0.0
0.5
1.0
T (sec}
(a)
).1 1(SDOF)
=2, a= 0%
::1..
c
70
c
    SDOF
. MOOF BH Model
o MOOF  CH Model
 t 1  MDOF WS Model
ca
E 60
~50
~
~ 40
.s 30
i!'
0
u;
i?
u:
20
10 ~
'"'
0
0.0
_/
.v
(b)
....
1.5
1.0
T (sec)
2.5
2.0
./
d' 70
:ij
E 60
f
lO
1..
.
2.0
~t 1 (SDOF)
= 2, a= 10%
(1558.00)
1 ~
_.,;;K. _ _ _
0.5
_1V
1.5
T (sec)
... 80
Lo
 
f!
La

2.5
    SDOF
. MDOF  BH Model
o MDOF CH Model
SO
   MDOF  WS Model
40
~k~ 
~~ ~'~
=[~ === ==I.J""\1~
u:
10

1 . _;_;_

'"'
o~~~4+~
0.0
0.5
1.0
1.5
2.0
2.5
T (sec)
SDOF
Ry(JJ.)(MDOF)
Ry(J..Lt)(SDOF) _ Vy(J..Lt)
Ry(J..Lt)(MDOF)  Fy(J..Lt)
Fy.e.
Vy(IJ.)
Ry(J..Lt}(MDOF) = Fy.e
Vy(JJ.t)
1.0
1.0
Fig. 4.8 RJJ. Relationships for SDOF and MDOF Systems (Explanation)
114
1~
    SDOF
1 t1+~t_._ MDOF BH Modal
o MDOF CH Model
11 MDOF WS Model
    SDOF
+_.~:.....++!1_._ MDOF BH Modal
o MDOF CH Model
11 MDOF WS Model
o~~4+~=+====~====~
ot~~4~=+====~==~~
2
......
12
10
12
10
Ductility Ratio, 11
Ductility Ratio, 11
......
Ul
12
10 _ _ , __ .
8
:i
~6~+~~~~~~~~l
>,6
a:
a:
10
20
30
40
50
60
70
+
I
I
    SDOF
__._ MDOF BH Modal
o MDOF CH Model
11 MDOF WS Model
Ductility Ratio, 11
I
I
II
I
    SDOF
__._ MDOF  BH Modal
o MDOF  CH Model
a MDOF WS Model
80
10
20
30
40
50
60
Ductility Ratio, Jl
70
80
BASE SHEAR MODIFICATION FOR MDOF EFFECTS ' VVI(MDOF) IFY'(SDOF) (1552 00)
3.0
   ATCS 1
  .  MDOF BH Model
o MDOF CH Model
11 MDOF WS Model
iL 2.5
0
0
,..
u..
U)
2.0
 1.5
~~
"' ~
iL
0
0
:IE
1.0
> o.5
BASE SHEAR MODIFICATION FOR MDOF EFFECTS V (MDOF) IF. Y'(SDOF) (155210)
'
3.0
iL
   ATCS 1
  .  MDOF BH Model
o MDOF CH Model
11 MDOF WS Model
2.5
0
0

r
Cl)
2.0
1.5
;t
k>
~ ~
iL
0
0
:E
~
.,
1.0
lo(
......
> o.5
0.0
_ ..... I"'\...:::::::::: ~
C

0.5
1.0
1.5
2.0
0.5
0.0
2.5
1.0
T (sec)
(a) J.l,
=2, a= 0%
'
3.0
2.5
Cl)
:::::. 1.5
"0
0
:IE
""S.
>
   ATCS 1
 e  MDOF BH Model
o MDOF CH Model
1.0
~
::::: .,
vr .......... 
2.5
!o
'
3.0
lL
U)
2.5
2.0
::::.. 1.5
"0
0
:E
""S.
>
0.5
0.0
0.0
2.0
1.5
T (sec)
BASE SHEAR MODIFICATION FOR MDOF EFFECTS V (MDOF) IFY'(SDOF) (15s3 00)
jl_ _____
0.0
0.0
li:'
:a
1.0
0.5
    ATCS 1
MDOF BH Model
o MDOF CH Model
   MDOF WS Model
e
~
..'
.()_
~ ~ r
. 
""' 7"""
/'
~
0.0
0.5
1.0
1.5
T (sec)
(c) 11, = 3, a= 0%
2.0
2.5
00
0.5
1.5
1.0
T (sec)
(d) ~1 1
= 3, a= 10%
Fig. 4.10 Base Shear Modification for MDOF Effects vs. Period ( 15s Records)
2.0
2.5
    ATCS1
(LS
(LS
MDOF CH Model
__.,_ MDOF WS Model
(/) 4
3
:e
>.
lL
1
LL'
;>1
___...
~ __. ,.._ fa
_,..,_
~
    ATCS 1
 .  MOOF  BH Model
o MDOF CH Model
    MDOF WS Model
o
0
cn4
._ 3
......
lL
0
0
:E
>
>1
....
0.0
0.5
1.5
1.0
.....
2.0
2.5
0.5
0.0
T (sec)
(0 J1 1 = 4, a= 10%
    ATCS 1
G:'5
0
o
MDOF CH Model
__.,_ MDOF WS Model
(f)4
3
LL'
:e
~1
./

    ATCS 1
MDOF BH Modol
MDOF CH Modol
    MDOF WS Model
+o
(LS
>.
lL
/
   
_.
_.v
lL
0
0

:E
2 
>
>1

::::.3
0
2.5
2.0
(/) 4
v
....
1.5
1.0
T (sec)
BASE SHEAR MODIFICATION FOR MDOF EFFECTSI V'y1(MDOF) IF'y1(SDOF) (1558 00)
n;:

......
.....
t
~
'S;;!"' 
a:
.....
........       ..
...~

.
     ~
0.0
0.5
1.0
1.5
2.0
2.5
0.0
0.5
1.0
1.5
T (sec)
T (sec)
(h) J1 1 = 8, a= 10%
(g) Jl 1 =8, a=O%
hg. 4.10 (Cont'd) Base Shear Modification for MDOF Effects vs. Paiod (15s Records)
2.0
2.5
BASE SHEAR MODIFICATION FOR MDOF EFFECTS I V'y1(MDOF) IFy(SDOF) (15s 0200)
1.5
...
~.
tL
0
0
U)
~
..
tL
c
U)
~
.._
'
:e
0.5
~
a
2
.
'
    ATCS 1
__.__ MDOF BH Model
o MDOF CH Model
 t 1  MDOF  WS Model
0
c
tn4
~
a
i:L
1"\..
c
:e
>.
>1   
BASE SHEAR MODIFICATION FOR MDOF EFFECTS, V, Y'(MDOF) IF (SDOF) (155 40.10)
(LS
~
;:1
(a) 2 Stories, a= 0%
/
/~
(L
    ATCS 1
    MDOF BH Model
o MDOF CH Model
 t t  MDOF WS Model
0
c
tn4
(L6
    ATCS 1
__.._ MDOF  BH Model
o MDOF CH Model
    MDOF WS Model
0.0
. 0
BASE SHEAR MODIFICATION FOR MDOF EFFECTS, V'y1(MDOF)/ F (SDOF) (155 40.QO)
:e
F==
c 0.5
:e
>>
    ATCS 1
__.._ MDOF  BH Model
o MDOF CH Model
 MDOF WS Model
0.0
1.0
...
;:.
.
.........
tL
tL
1.5
>
1.0
BASE SHEAR MODIFICATION FOR MDOF EFFECTS Vy(MDOF)/ Fy(SDOF) (155 0210)
~
:a
.
4
~
I
   
1.0
0.8
t;.,.'rY+r:~.....:.:.... .,.1
C1)
I
I
0.4
a:
++~~....:...'1
I
\ :.. ... II
0.2
 t       :        :      : .~1:i~.J       f
0.0
:1
+~~~~1
0.5
0.0
1.5
1.0
Vdyn,i I Vdes,i
Fig. 4.12 Variation in Maximum Dynamic Story Shears for Elastic 20Story Structures (15s Records)
0.8
.r:.
C)
C1)
::z::
0.6
f})
lr:f ~
C1)
>
.;
0.4
..5!
C1)
a:
0.2
0.0
II~
1
0.0
0.5
1.0
1.5
Vdyn,i IVdes,i
Fig. 4.13 Maximum Dynamic Story Shears for Elastic Structures (Mean for 15s Records)
119
0.8
.c
0
~ 0.6
Q)
>
2,5,10,20,30.40 stories
(thick .thin lines)
~ 0.4
Q)
a:
0.2
+,..,.L.;....J
0.0
1.0
0.5
0.0
1.5
2.0
V d~n,l IV des,l
(a) BH Model
1.0
0.8
.c
0
~ 0.6
Q)
>
0.4
Q)
a:
0.2
0.0 +,..&.....~!
1.5
1.0
0.5
2.0
0.0
Vdyn,l I Vdes,l
(b) CH Model
MAX. DYNAMIC STORY SHEARS, V d n 1 / V des 1 (15s.ws2.00)
1.0 ~:::F:::t:==f~~~~,
c;
0.8
.c
0
~ 0.6
a;
0.4
a:
0.2
0.0
0.5
1.5
1.0
2.0
3.0
vdyn,l/ vdea,l
(c) WS Model
Fig. 4.14 Max. Dyn. Story Sheazs (Nonnalized by Vdes) for J.Lt(SOOF) = 2, a= 0% (Mean for 15s Records l
120
.C)
=.
~
0.6
>
~ 0.4
41
G)
a:
II
\(
(thick
~thin
lines)
0.2
l._____,....____ll.t~=~~.......~
0.0
0.0
0.5
1.0
1.5
2.5
2.0
3.0
Vdyn,i I Vdes,l
(a) BH Model
MAX. DYNAMIC STORY SHEARS, Vdvn 1/ Vdes 1 (15s.ch8.00)
1.0~r~~~~~~~
!
0.8
.=.
C)
0.6
41
>
~ 0.4
Q)
a:
thin lines)
0.2
0.0~~~~~~~
0.0
0.5
1~
15
2~
2.5
3.0
Vdyn,l/ Vdes,l
(b) CH Model
MAX. DYNAMIC STORY SHEARS, V n 11 Vdes 1 (15s.ws8.00)
1.0 1,.'::::::l;e~;;;;:=~~==~~,
0.8
.=.
C)
~ 0.6
41
>
a;
0.4
(thick
&!
~thin
lines)
0.2
0.0 +.E=::;=:::=::::::::!..._...,~
0
2
4
6
8
10
V dyn,l/ Vdes,l
(c) WS Model
Fig. 4.15 Max. Dyn. Story Shears (Normalized by Vdes) for ~(SDOF) = 8, a= 0% (Mean for 15s Records)
121
0.8
.r:.
C)
2,5,10,20,30,40 stories
(thick + thin lines)
~ 0.6
G)
>
:; 0.4
;
a:
0.2
0.0
0.5
1.0
1.5
2.0
2.5
3.0
V dyn,l/ V des,l
(a) BH Model
1.0
0.8
.r:.
C)
~ 0.6
:; 0.4
;
a:
02
0.0 +.,:~"'~....1..._;.;;;;:.__ _l
3.0
0.5
1~
1~
2~
2.5
0.0
Vdyn,l/ Vdes,l
(b) CH Model
1.o
1TIr~~~~~~1
0.8
.r:.
C)
~ 0.6
G)
>
a; o.4
;
a:
02
o.o+~~~~~~~~~~
10
v dyn,l/ vdes,l
(c) WS Model
Fig. 4.16 Max. Dyn. Story Shears (Normalized by V~)for J.tt(SDOF) = 8, a= 10% (Mean for 15s Records)
122
0.8
~
~ 0.6
Q)
>
:i 0.4
Q)
a:
0.2
0.0 +...,~'"~...,....!
0.5
1.0
1.5
2.5
0.0
2.0
v,t vdes,base
(a) BHModel
MAX. OYN. & DESIGN STORY SHEARS, Vi IV des base (15s.ch402)
1.0
  Vdes.i I Vdes.base
  Vdyn.i IVdes.base (Ct = 0%)
V n.;l Vdes.base (Ct = 10%)
0.8
0')
~ 0.6
Q)
>
'i
0.4
"ii
a:
0.2
o.o~,~~~,,!
0.0
0.5
1.0
2.0
1.5
2.5
Vi I Vdes,base
(b) CH Model
  VdesJ IVdes.base
  VdynJIVdu,base(et= 0%)
V .IIVdes.base(Ct= 10%)
0.8
0')
~ 0.6
Q)_
>
'i
0.4
"ii
a:
0.2
0.0 +....~===r~=r~
0.0
0.5
1.0
1.5
2.0
2.5
v,IVdes,base
(c) WS Model
Fig. 4.17 Max. Dyn. & Design Story Shears (Nonnalized by V~for Jlt(SOOF) 2, 40 Stories (Mean for 15s Records)
123
MAX. DYN. & DESIGN STORY SHEARS, V1/ Vdes base  (15s.bh408)
1"'1:!~
. . ;:;;;~~~~;;;;;ii;iil
1.0
:!
......
  VdesJ IVdes,base
Va,nJIVdes,bllse (a= 0%)
V ,;IVdes.bllse(a=IO%)
0.8
.c
"\.'1...
0.6
G)
>
0.4
G)
a:
0.2
0.0
.L.   . ,     . . . . : . __ _
0.0
0.5
_r:;===:_.,.I::=:J
1.5
1.0
2.0
2.5
3.0
Vi IV des,base
(a) BH Model
MAX. PYN. & DESIGN STORY SHEARS, v, I Vdes.base (15s.ch408)
1.o ro:;;::;:::;;~~~~~.,
  Vdes,i IVdes,base
Vctvn.l/ V aes.ba.Se (a= 0%)
0.8
.C)
c
~ 0.6
G)
>
(; 0.4
G)
a:
02
0.0 +._ _ _...,...._ _ _....__ _
0.0
0.5
~:.....;;;;;'i'L1
1~
1~
2~
2.5
3.0
v,l v des,base
(b) CH Model
MAX. DYN. & DESIGN STORY SHEARS, V,l Vdes base (15s.ws408)
1.0
nc:::==~~==~~~~~~iiiiiil
  Vdesj IVcles,base
Vc:~ynj/Vdes,base(a= 0%)
V ,;IVcles.basa(a=IO%)
0.8
.C)
c
:!
0.6
G)_
>
a;
a;
a:
0.4
02
0.0 !_..i::::==;::===~,~===1
0
2
4
6
8
v,lvdea,base
Fig. 4.18 Max. Dyn. & Design Story Shears (Nonnalized by Vda.b;asc) for ~(SDOF) =8, 40 Stories (Mean for 15s Records)
124
0.8
0.6
:t:
i!
I
ii
>
t tl
0.4
..
a:
:
0.2
0.0
..
1/
!l
i/
l
'
.c
en
I
i
0.0
0.5
1.0
1.5
MoT(dyn,i) I MoT(des,i)
Fig. 4.19 Variation in Ma"<imum Dynamic Story Overturning Moments for Elastic 20Story Structures (15s Records)
~
'
0.8
.c
en
'Q; 0.6
:t:
~
>
ttl
a;
0.4
a:
0.2
0.0
~'i
~/
tv~
'/
II
0.0
0.5
1.0
1.5
MoT(dyn,i) I MoT{des,f)
Fig. 4.20 Maximum Dynamic Story Overturning Moments for Elastic Structures (Mean for 15s Records)
125
(15s.bh2.00)
1.0
:!
0.8
.s:
_E)
0.6
G)
>
2,5,10,20,30,40 stories
0.4
C)
a:
0.2
0.0 ..1...~.:..L&..,..J
0.5
1.0
1.5
2.0
0.0
MoT(dyn,l) I MoT(des,l)
(a) BH Model
MAX. 0YNAMIC STORY OVERTURNING MOMENTS, Morrdvn.il/ Morrdes.il" (15s.ch2.00)
0
0.8
.s:
C)
~ 0.6
G)
>
~ 0.4
C)
a:
0.2
2,5,10,20,30,40 storiesl
(thick + thin lines)
0.0 ..l..~......1
1.0
1.5
2.0
0.5
0.0
MoT(dyn,l) I MoT(des,l)
(b) CHModel
MAX. DYNAMIC STORY OVERTURNING MOMENTS, M0
10
:!
,::::F::c::r:c=:L..=.;=L......;...,
0.8
.s:::.
C)
0.6
GL
>
iii
0.4
C)
a:
0.2
0.0
+r..,.....JI....L.It..L...,......1
0.0
0.5
1.0
1.5
2.0
2.5
3.0
MoT(dyn.l) I MoT(dH,i)
(c) WS Model
Fig. 4.21 Max. Dyn. Story Ove~g Mom. (Normalized by Mandes,i)) for ~SOOF} =2, a~ 0%, (Mean for 15s Records)
126
0.8
.:::.
0
~ 0.6
C)
>
~ 0.4
a;
a:
0.2
+......a!... . . .._1
0.0
0.0
1.0
0.5
1.5
MoT(dyn,i)
2.0
2.5
3.0
I MoT(des,l)
(a) BH Model
MA:X. ~YNAMIC STORY OVERTU~NING MOMENTS, Morcdvn.il/ Morcdes.il" (15s.ch8.00)
1
0.8
.:::.
Cl)
~ 0.6
C)
>
~ 0.4
a;
a:
2,5,10,20,30,40 stories
(thick+ thin lines)
0.2
0.0+~~~~~
0.0
1.5
1.0
0.5
2.0
2.5
3.0
I MoT(des,l)
(b) CHModel
MoT(dyn,l)
1.0 ...,........................~='.:::.:..l:=L...,;.___,
!
0.8
.:::.
Cl)
0.6
G)
>
2,5,10,20,30,40 stories
(thick + thin lines)
~ 0.4
a;
a:
0.2
0.0+LL~~~~~~~
MoT(dyn,l)
10
I MoT(des,l)
(c) WS Model
Fig. 4.22 ~ Dyn. Story Overturning Mom. (Nonnalized by Mor<des.iY for ~(SOOF) =8, a= 0%, (Mean for 15s Reccres;
127
0.8
.c
0
~ 0.6
Q)
>
c;
C)
a:
0.4
0.2
0.0
0.5
1.0
1.5
MoT(dyn,l)
2.0
2.5
3.0
I MoT(des,l)
(a) BH Model
MAX. DYNAMIC STORY OVERTURNING MOMENTS, Morrd
1.0~r~~~~~~~~~~~
0.8
.c
0
~ 0.6
Q)
>
0.4
Q)
a:
0.2
0.0
1~
0.5
~0
1.5
2.5
3.0
I MoT(des,l)
(b) CH Model
MoT(dyn,l)
0.8
.c
0
~ 0.6
CD
>
t; 0.4
Ci)
a:
2,5,10,20,30.40 stories
(thick thin lines)
0.2
0.0+~L~~~~~~~~
10
MoT(dyn,J) I MoT(des,l)
(c) WS Model
Fig. 4.23 Max. Dyn. Story Overturning Mom (Norrnaliztd by Morcdes.l"V for ~(SDOF) = 8, a= 10%, (Mean for 15s Records)
128
by
Vitelmo V. Bertero
Eduardo Miranda
and
The CUREe and The Kajima Research Teams
This report summarizes studies which have been conducted by Eduardo Miranda under the direct
supervision of Vitelmo Bertero as a part of a research project on design guidelines for ductility
and drift limits, which is carried out with the ultimate goal of developing an improved method
for earthquakeresistant design of structures. These studies, which are part of Task 2 of the
above research project, focus on the evaluation of the damage potential of recorded ground
motions and the implications of the results obtained regarding the design of structures, with
particular emphasis on the structural response modification factor for strength and the
establishment of limits for lateral drift. The studies can be considered as a complement of the
studies conducted by Helmut Krawinkler, et alia, at Stanford University and reported in
"Evaluation of Damage Potential of Recorded Ground Motions," June 1991.
The research project is supported by a grant provided by Kajima Corporation and administered
by CUREe (California Universities for Research in Earthquake Engineering). This financial
support is gratefully acknowledged. Appreciation is also expressed to the California Department
of Conservation, Division of Mines and Geology, Strong Motion Instrumentation Program
(SMIP), for providing most of the processed data of recorded ground motions used in these
studies, as well as partial support for conducting them. Several researchers of the U. S. and
Kajima project research team have provided valuable input to these studies.
discussions deserve special gratitude.
Their critical
TABLE OF CONTENTS
1. IN'I'R.ODUCTION . . . . . . . . . . . . . . . . .
1. 1 STATEMENT OF PROBLEM . .
1. 1. 1
Earthquake Input:
Earthquakes . . . . . . . . .
1. 2 OBJECTIVES AND SCOPE . . . .
............
............
Establishment
............
............
...
...
of
...
...
.......
.......
Reliable
.......
.......
......
......
Design
......
......
..
..
1
1
..
..
1
2
. 5
. 5
. 5
. 6
. 11
. 15
. 29
. 29
.....
.....
DUE
.....
.....
. 30
. 33
39
39
40
43
1. INTRODUCTION
1. 1 STATEMENT OF PROBLEM
One of the most effective ways to mitigate the destructive effects of earthquakes is to improve
existing methods of:
resistant structures (manmade facilities in general); and upgrading (retrofitting) and maintaining
existing seismic hazardous facilities. As discussed in detail in the reports on Task 1 [1] and Task
2 [2], the principal issues that remain to be solved in order to improve seismic design of new
structures and seismic upgrading of existing structures are related to the three following basic
elements: Earthquake Input to the Foundation of the Structure, Demands Imposed by This
Input on the Structure, and the Supplied Capacities to the Structure, Which Should Exceed
the Demands. Therefore, it is obvious that the essential data needed to start any reliable design
of a new structure or upgrading of an existing facility is the reliable establishment of the
earthquake input, or in other words the establishment of design earthquakes.
In practice, the application of this simple concept meets with serious difficulties,
because, firstly, there are great uncertainties in predicting the dynamic characteristics of ground
motions that have yet to occur at the building site, and secondly, even the critical response
parameter of a specific structural system may vary according to the various limit states that could
control the design. Because in most cases the design is controlled by the safety limit state which
involves damage to the building as a result of inelastic deformations, to establish the design
earthquake at this level (limit state), it is necessary to estimate the damage potential of the
different earthquake ground motions that can occur at any given site. One of the most promising
ways, not only for estimating the damage potential of earthquake ground motions but in general
for improving the EarthquakeResistant Design (EQRD) of structures and particularly for
improving establishment of design earthquakes for the limit states involving damage (inelastic
behavior), is through the use of an energy approach.
2
As discussed in Ref. 1, and in more detail in Refs. 3 and 4, a promising parameter for improving
)
r
selection of proper design earthquakes for safety limit states is through the concept of Energy
Input, Ey , and Associated Parameters. In Refs. 3 and 4, it is shown that for defining properly
the Safety (Survival) Level design earthquake it is necessary to consider the following spectra
simultaneously: the E 1 ; the Inelastic Design Response Spectra (IDRS) for strength and
displacement; and the Energy Dissipation,
including the cumulative ductility J.L~ , and Number of Yielding Reversals (NYR) spectra.
Examples of evaluation of the E1 and IDRS (particularly for strength, i.e., the yielding strength
spectra or its equivalent yielding seismic coefficient, Cy , spectra) has been given and discussed
in Ref. 1. In Ref. 2, Professor Krawinkler and his associates address in detail not only the
problem of evaluation of the IDRS for strength, but also other important issues of seismic design
for ductility capacity, including the effects of cumulative damage.
Normalized Hysteretic
Energy Dissipation is used as the basic cumulative damage parameter. A total of 15 Western
U. S. ground motion records were considered in the statistical studies presented in Ref. 2. All
of the records are from sites corresponding to soil type S1 (rock or stiff soils). To complement
the studies conducted by Professor Krawinkler and his research associates, the authors decided
to conduct similar statistical studies of recorded ground motions, but considering also ground
motions recorded on soft soil, and obtaining not only the inelastic strength spectra but also the
inelastic deformation spectra. These studies were conducted with the following main objectives.
The ultimate goal of the studies which are being conducted at Berkeley on the evaluation of the
damage potential of ground motions has been to improve the establishment of design earthquakes
and, consequently, to improve the earthquakeresistant design of new structures and the seismic
upgrading of existing hazardous facilities.
earthquake ground motions recorded on various soil conditions ranging from rock to very soft
soil have been evaluated, and the implications of the results obtained on the reliable establishment
of design earthquakes have been assessed [5].
3
The main objectives of this report are to summarize: first, the main results obtained in the
above studies, focussing on the normalized elastic and inelastic strength spectra, as well as on
the displacement spectra; and, second, the implications of these results regarding the
establishment of design earthquakes, particularly with reference to the use of present values for
the structural response factpr, R or Rw, and the code methodology for specifying limitation on
the interstory drift limits.
In the studies conducted, the main results of which are reported herein, the emphasis is placed
on the effects of soil, particularly on soft soils, of the seismic demands of strength and
deformation.
5
2. STATISTICAL STUDY OF INELASTIC RESPONSE SPECTRA
2. 1 INTRODUCTORY REMARKS
To improve the tools for establishing reliable design earthquakes, constant displacement ductility
ratio, Jl6
fl,
linear (p. = 1) and particularly nonlinear (p. > 1) spectra were obtained by computing
the response of a family of Single Degree of Freedom (SDOF) systems with the use of the
computer program NLSPECfRA, which was developed specifically for this purpose [5].
Constant ductility spectra were computed for 124 earthquake ground motions on various soil
conditions ranging from rock to very soft soils. The following values of ductility ratios were
selected for this study: 1 (elastic), 2, 3, 4, 5, and 6. The spectra were computed for a set of 50
periods between 0.05 and 3.0 seconds. Due to the large number of records and the computational
effort involved in calculating constant ductility nonlinear spectra, the study was limited to bilinear
systems with a postelastic stiffness of 3% of the elastic stiffness and with a damping ratio of 5%
of critical. It has been shown that the use of elastoplastic or bilinear hysteretic models can
provide very good estimates of strength and deformation demands of structures that show elastic
stiffness degradation under cyclic loading. Note that if the hysteretic behavior is of the elasticdeformation softening type, then the deformation demands can be considerably larger.
produced more records than the total number of records obtained in the Western United States
between 1933 and 1984. For this study 124 records were selected, with emphasis on those
recorded in California and on those recorded during the last six years. Of the total number of
records, 96 (77%) were recorded in the last six years and 90 (73%) were recorded in California.
The ground motions were classified into three groups according to the soil conditions at the
recording station. These groups were rock, alluvium and very soft soil. Tables 2.1, 2.2 and 2.3
list the selected ground motions recorded on rock, alluvium and soft soil sites, respectively.
(1)
Despite several studies conducted in this area, there is no agreement on which parameter
should be used to scale nonlinear spectra.
(2)
Most seismic hazard studies (i.e., seismic hazard curves) and attenuation relations are
based on PGA.
(3)
Using both parameters provides an opportunity to compare the results obtained with the
results of previous studies that have used the same normalization parameters.
(4)
The results can be combined with readily available information (such as maps of peak
accelerations from maximum credible earthquakes or maps of effective peak acceleration)
that is based on these parameters.
For ground motions recorded on rock or alluvium sites, nonlinear spectra were computed for a
fixed set of 50 periods between 0.05 and 3.0 seconds. In the case of ground motions recorded
on very soft soil, spectra were computed for a fixed set of 50 ratios of Tffg , where Tg is the
predominant period of the ground. The reason for using Tffg instead ofT is that Tg can have
large variations depending on the shear wave velocity of the soil and the depth of the soft
deposits. For instance, in Mexico City, where the soft clay deposits have approximately the same
characteristics throughout the city, the predominant period of motion can vary anywhere from
0.6 seconds to more than 3.8 seconds, depending on the depth of these deposits, as well as on
the type of earthquake ground motion. For statistical analyses of spectra it makes no sense to
average spectral ordinates at a certain period for ground motions with significantly different
7
predominant periods. For structural design purposes, it is important to characterize the demands
on structures with periods shorter than, longer than, or near to the predominant period.
In this study the predominant period was computed as the period corresponding to the maximum
spectral velocity ordinate. It can be shown that essentially the same predominant period would
be obtained if the Fourier amplitude spectrum or the input energy spectrum were used instead
of the velocity spectrum, because of the relationship between these three spectra. Figure 2.1
shows six examples of how, using the velocity spectra, it is possible to identify predominant
periods, Tg, for sites in the San Francisco Bay Area, Romania, and Mexico City.
The average (mean) nonlinear strength spectra of 38 ground motions recorded at rock sites
normalized using PGA and EPA are shown in Figs. 2.2 and 2.3. The spectra are plotted for
displacement ductilities of 1, 2, 3, 4, 5 and 6 (from top to bottom). For a given ground motion,
the PGA is usually larger than the EPA, hence the spectral ordinates of the average nonlinear
spectra normalized by EPA (Fig. 2.3) are larger than those of the average nonlinear spectra
normalized by PGA (Fig. 2.2).
Mean inelastic strength demand spectra of 62 ground motions recorded on alluvium are shown
in Figs. 2.4 and 2.5, where PGA and EPA, respectively, have been used as normalizing
parameters. The maximum amplification for alluvium sites is larger than that observed at rock
sites. Mean inelastic strength demand spectra of 24 ground motions recorded on soft soil are
shown in Figs. 2.6 and 2.7.
From analysis of the plots of Figs. 2.2 to 2.7, it is clear that the spectral shapes for the inelastic
demands (J.l > 1) differ significantly from elastic (J.l
= 1) spectral shapes.
Furthermore, from
analyses of the results obtained from soft soil records it was observed that for T < Tg there is
small difference between the strength demand for ductilities between 2 and 6. This implies that
a small change in the yielding strength of a structure with T < Tg may produce a large change
in the ductility demands.
8
By comparing the average spectra of the three different soil conditions it can be seen that the
largest dynamic amplification for elastic response (fl
These
results are different .:rom those reported in Ref. 7, where larger amplifications for rock and
alluvium sites than for soft soil sites are shown. Moreover, the maximum amplification (with
respect to the PGA) computed in that reference is nearly 30% smaller than the maximum
amplification computed here. For rock and alluvium sites, the maximum amplifications computed
in this study are practically the same as those reported in Ref. 7, with a smaller set of ground
motions.
Regardless of the type of local site conditions, for ductility demands equal to or larger than 5
strength demands decrease monotonically with increasing period.
While mean spectra provide information on the most probable demands of the structure, it is
important to take into account the dispersion of these demands. One method of measuring the
dispersion is to compute the coefficient of variation (COV), which is defined as the ratio of the
standard deviation to the mean. Figures 2.8 to 2.13 show the COV of the spectral ordinates for
the three types of soil conditions, with PGA and EPA used as normalizing parameters. The use
of acceleration parameters to normalize the spectra produces an increase in dispersion with
increase in period. Relatively large COVs are produced in the long period region; however,
strength demands are usually very small in this region. For soft soil sites, a local increase in
COV is produced for periods near the predominant period (i.e., T{f8
=1).
ground displacement as the normalizing parameter would produce an opposite trend (i.e., small
COVs in the long period range and large COVs in the short period range). By comparing COVs
of spectra normalized by PGA and EPA, it can be seen that except for periods between 0.1 and
0.5 second, COVs are larger for spectra normalized using EPA than for spectra normalized using
PGA. An important observation is that, regardless of the type of soil, COVs are nearly the same
for different levels of ductility, which means that dispersion in strength demands does not
increase with increasing ductility demands.
Mean plus one standard deviation inelastic strength demands for the three types of soil conditions
.l
9
are shown in Figs. 2.14 to 2.19.
Spectral amplifications between 0.1 and 0.5 second for ground motions recorded on soft soil are
usually much smaller than the 2.5 factor which is used for defining EPA. Therefore, the use of
EPA (which was primarily developed for firm soil records) to normalize strength demands on soft
soil sites produces unusually large amplifications (see Figs. 2.7 and 2.19) For this type of soil
condition, the use of PGA is probably more appropriate than the use of EPA.
11
STATION NAME
GEOLOGY
EAR11lQUAKE
DATE
SAN FRANCISCO
SIDceous
San Francisco
sandstone
PARKFIELD
Cholame Shandon No.2
CASTAIC
Old Ridge Road
LLOLLEO
VALPARAISO
LA UNION
LAVIWTA
ZIHUATANEJO
Rock
Sandstone
Pari<filed
June 27, 1966
San Fernando
February 9, 1971
Sandstone &
Central Chile
volcanic rock
March 3, 1985
Volcanic
Central Chile
rock
March 3, 1985
Metavolcanic
Michoacan
Rock
Gabbro
Michoacan
Rock
Tuna lite
Michoacan
Rock
NATL GEOGR.
Balsamo
San Salvador
INSTITUTE
Formation
October 10,1986
INST. URBAN
Ruviate
San Salvador
CONSTRUCTION
Pumice rock
GEOTECH. INVEST.
Ruviate
San Salvador
CENTER
Pumice rock
October 10,1986
MTWILSON
Quartz
WhittierNarrows
dionte
October 1, 1987
CORRAUTOS
Landslide
Lema Prieta
deposits
SANTA CRUZ
Lema Prieta
ucsc
Umestone
SAN FRANCISCO
Franciscan
Lema Prieta
Cliff House
sandstone
SAN FRANC!SCO
Franciscan
Lema Prieta
Pacific Heights
sandstone
SAN FRANCISCO
Presidio
Serpentine
Lorna Prieta
October 17, 1989
SAN FRANCISCO
Franciscan
Rincon Hill
sandstone
YERBA BUENA
Franciscan
Lema Prieta
ISLAND
sandstone
Lema Prieta
I MAGN.
EPICTR.
DlST.[km]
5.3(MJ
11
5.6(MJ
6.5(MJ
29
7.8(Ms)
45
7.8(Ms)
84
8.1(Ms)
84
8.1 (Ms)
44
8.1 (Ms)
135
5.4(Ms)
5.7
5.4(Ms)
5.3
5.4(Ms)
4.3
6.1(MJ
19
7.1(Ms)
7.1 (Ms)
16
7.1(Ms)
99
7.1(Ms)
97
7.1(Ms)
98
7.1(Ms)
95
7.1(Ms)
95
PGA
PGV
[g's]
[c:nlsec]
N10E
0.08
4.91
S80E
0.11
4.61
N55E
0.48
78.09
N21E
0.32
17.16
N69W
0.27
27.82
DIRECTION
N10E
0.67
23.70
S80E
0.43
43.60
N70E
0.18
16.30
S20E
0.16
9.70
NOOE
0.17
20.34
N90E
0.15
11.70
NOOE
0.13
i 6.11
N90E
0.12
10.51
N90W
0.10
15.86
SOOE
0.16
18.34
270
0.53
72.70
180
0.39
56.10
90
0.38
39.20
180
0.67
55.50
180
0.42
61.80
90
0.68
80.00
90
. 360
0.19
4.09
0.13
4.32
90
0.47
47.50
360
0.62
55.20
90
0.41
21.20
360
0.43
21.20
90
0.11
21.00
0.07
11.20
360
0.05
9.88
270
0.06
14.30
90
0.20
33.50
0.10
13.30
90
0.09
11.50
360
0.08
7.34
90
0.06
14.70
360
0.03
4.61
12
STATION NAME
ELCENTRO
ln1gallon District
TAFT
Uncoln School Tunnel
FIGUEROA
445 Figueroa St.
HOLLYWOOD
FrEH~
Field
Alh.Nium
Alwium
AJII.Nium
Alii.Nium
EARTHQUAKE
DATE
Imperial Valley
May 18, 1940
Kern County
July 21' 1952
February 9, 1971
San Fernando
February 9, 1971
AVE.STARS
San Fernando
layers
February 9, 1971
Kokutetsu Bldg.
MELOLAND
Interstate 8 Overpass
BONDS CORNER
Highways 98 & 115
JAMES ROAD
El Centro Array 11 5
IMPERIAL V. COLLEGE
El Centro Array il 7
ELALMENDRAL
ZACATULA
Alii.Nium
Alluvium
Alluvium
Allwium
AJiwium
BURBANK
Cal. Fed. Savings Bldg.
Imperial Valley
October 15, 1979
Imperial Valley
October 15, 1979
Imperial Valley
October 15, 1979
Imperial Valley
October 15, 1979
fill
March 3, 1985
Alh.Nial
Central Chile
sand
March 3, 1985
Allwium
ALTADENA
Eaton Canyon Park
Central Chile
ALHAMBRA
Freemont School
MlyagiKe~Oki
Compacted
Allwium
Alluvium
Alluvium
MAGN.
s.3<Mu
7.7(Ms)
San Fernando
GEOLOGY
Mlchoacan
Sect. 19,1985
WhittierNarrows
October 1, 1987
E?ICTR.
DIST.[km]
a
56
s.s<Mu
41
s.s<Mu
35
s.s<Mu
38
7.4(Ms)
110
s.s(Mu
21
6.6(MJ
6.6(MJ
22
6.s(Mu
21
7.B(Ms)
84
7.8(Ms)
8.1 (Ms)
sa.
49
6.1(MJ
s.1(Mu
13
6.1(MJ
26
WhittierNarrows
October 1, 1967
WhittierNarrows
October 1, 1987
PGA
PGV
[g's]
[em/see]
S90W
0.21
36.92
SOOE
0.34
33.45
DIRECTION
N21E
0.15
15.72
SSSE
0.17
17.71
N52!:
0.15
17.38
S38W
0.12
17.31
N90E
0.21
21.14
scow
0.17
16.50
N4SN
0.14
9.65
S44W
0.15
16.74
N90W
0.44
57.01
NOOE
0.24
36.16
360
0.31
71.65
270
0.30
90.94
S40E
0.58
43.63
S50W
0.77
44.07
S40E
0.52
43.99
ssow
0.37
86.56
S40E
0.33
44.56
S5ow
0.45
107.80
N50E
0.29
26.90
S40E
0.16
19.80
N70W
0.23
27.70
S20W
0.36
33.20
SOOE
0.26
30.39
N90W
0.18
13.96
270
0.40
16.98
180
0.30
21.63
90
0.16
4.92
360
0.31
10.43
130
0.22
12.61
40
0.17
9.63
'
13
STATION NAME
GEOLOGY
EARTHQUAKE
DATE
DOWNEY
Deep alhN!um
WhittierNarrows
October 1, 1987
INGLEWOOD
Terrace
WhittierNarrows
deposits
October 1, 1987
LOS ANGELES
Terrace
WhittierNarrows
116th Sl School
deposits
October 1, 1987
LOS ANGELES
Allt.Mum
WhittierNarrows
Baldwin Hills
over shale
October 1, 1987
LOS ANGELES
Hollywood Storage FF
LOS ANGELES
Obregon Parle
LONG BEACH
Rancho Los Cemtos
SAN MARINO
Southwestern Academy
TARZANA
Cedar Hill Nursery
WHITTIER
7215 Blight Tower
ALBA
900 S. Fremont
CAPITOLA
Fire Stabon
HOLLISTER
South & Pine
OAKLAND
2StOI"f Office Bldg.
STANFORD
Parking Garage
Allt.Mum
Alluvium
Allt.Mum
Alluvium
Alluvium
Alluvium
Alluvium
Allt.Mum
Alluvium
Alluvium
Allt.Mum
WhittierNarrows
October 1, 19 87
WhittierNarrows
October 1, 1987
WhittierNarrows
October 1, 19 87
WhittierNarrows
October 1, 1987
WhittierNarrows
October 1, 1987
WhittierNarrows
October 1, 1987
WhittierNarrows
October 1, 1987
Lema Prieta
October 17, 1989
Lema Prieta
October 17, 1989
Lema Prieta
October 17, 1989
Lema Prieta
October 17, 1989
MAGN.
EPICTR.
DIST.(km]
6.1(MJ
17
6.1(MJ
25
6.1(MJ
22
6.1(MJ
27
6.1(MJ
25
6.1(MJ
10
6.1(MJ
27
6.1(MJ
6.1(MJ
44
6.1(MJ
10
6.1(MJ
7.1 (Ms)
7.1 (Ms)
48
7.1(Ms)
92
7.1 (Ms)
51
PGA
PGV
(g's]
[em/sec]
270
0.16
12.76
180
0.20
29.27
DIRECTION
90
0.23
16.28
360
0.27
8.65
360
0.40
18.65
270
0.29
17.83
90
0.17
6.64
360
0.15
7.51
90
0.12
6.94
360
0.21
8.99
360
0.44
22.07
270
0.45
12.98
90
0.25
18.48
360
0.15
~6.65
360
0.20
12.87
270
0.15
4.82
90
0.63
24.20
360
0.46
19.27
90
0.63
27.10
360
0.43
26.60
90
0.29
10.84
360
0.25
20.64
90
0.39
30.70
360
0.46
36.10
90
0.17
30.90
360
0.36
62.80
290
0.24
37.90
200
0.19
20.00
360
0.26
33.18
90
0.22
21.30
14
STATION NAME
GC:OLOGY
EARTHQUAKE
MAGN.
DATE
BUCHAREST
Building Research lnst.
SCT
Stia. de
Soft
Soft
Romania
March 4 ,19n
clay
CENTRAL DE ABASTOS
Soft
Michoac~n
Frigor1fico
clay
CENTRAL DE ABASTOS
Soft
Mlchoa~n
Ofidna
clay
Sept19, 1985
COLONIA ROMA
EMERYVILLE
Free Field South
Soft
Acapulco
clay
Say mud
Say mud
OAKLAND
Outer Harbor Wharl
Say mud
Fill
SAN FRANCISCO
International Airport
Say mud
Lorna Prieta
bay mud
FOSTER CITY
Say mud
174
8.1(Ms)
8.1(Ms)
8.1 (Ms)
6.9(Ms)
385
389
389
7.1(Msl
97
7.1 (Ms)
97
7.1 (Ms)
95
7.1 (Ms)
98
Lorna Prieta
SAN FRANCISCO
Redwood Shores
7.1(Ms)
Lorna Prieta
October 17, 1989
Lorna Prieta
October 17, 1989
(cmlsacj
'2N
0.17
32.62
SN
0.20
75.11
N9CJN
0.17
60.50
SOOE
0.10
38.74
99.53
0.10
34.57
n.52
0.08
24.85
76.50
0.08
41.86
67.95
0.07
34.98
N90W
0.06
11.90
SOOE
0.05
10.92
350
0.21
21.50
260
0.26
41.06
Lorna Prieta
October 17, 1989
TREASURE ISLAND
Naval Base
Lorna Prieta
(g's]
DIST.[km]
Lorna Prieta
October 17, 1989
EMERYVILLE
Free Field North
PGV
DIRECTION
Michoac~n
Comunic. y Transport.
PGA
EPICTR.
7.1(Ms)
79
7.1(Ms)
$5
7.1(Ms)
63
350
0.20
15.74
260
0.22
37.94
305
0.27
42.30
125
0.29
40.80
90
0.15
33.40
360
0.10
15.60
90
0.33
29.25
350
0.23
26.45
980
0.13
17.11
350
0.15
15.76
90
0.28
45.40
0.25
31.80
15
VELOCITY (in/sec)
VELOCI1Y On/sec)
~0
25
FOSTER CITY 0
T; t.tS
40
20
30
15
20
10
10
0
0.0
1.0
0.5
1.5
2.0
2.5
3.0
0.0
0.5
PERIOD (sec)
1.0
1.5
2.0
2.5
3.0
2.5
3.0
PERIOD (sec)
VELOCITY (in/sec)
VELOCI1Y (in/sec)
60
70
60
50
50
40
40
30
30
20
20
10
10
0.0
1.0
0.5
1.5
2.0
2.5
3.0
0.0
0.5
1.0
1.5
2.0
PERIOD (sec)
PERIOD (se:c)
VELOCITY (In/sec)
VELOCI1Y On/sec)
140
25
SCT EW
Tr
120
20
100
15
eo
60
10
40
5
20
0
0
0.0
0.5
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
0.0.
0.5
1.0
1.5
2.0
PERIOD (sec)
Figure 2.1 Predominant ground period for various soft soil sites.
2.5
3.0
16
11
3.0
ROCK SITES
2.5
= 1
2.0
1.5
1.0
0.5
0.0
0.0
0.5
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
Figure 2.2 Mean strength demands of ground motions recorded on rock when normalized using PGA (!1=1 ,2,3,4,5,6}.
Cy
EPNg
3.0
ROCK SITES
2.5
2.0
1.5
1.0
0.5
0.0
0.0
0.5
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
Figure 2.3 Mean strength demands of ground motions recorded on rock when normalized using EPA (J.1=1 ,2,3,4,5,6).
17
'1
3.0
ALLUVIUM SITES
2.5
= 1
2.0
1.5
1.0
0.5
0.0
0.0
Figure 2.4
0.5
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
Cy
EPNg
3.0
ALLUVIUM SITES
2.5
2.0
1.5
1.0
0.5
0.0
0.0
0.5
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
Figure 2.5 Mean strength demands of ground motion recorded on alluvium when normalized using EPA (!1=1 ,2,3,4,5,6).
18
11
5
2
1
0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
T /Tg
Figure 2.6 Mean strength demands of ground motions recorded on soft soil when
normalized using PGA (!l=1,2,3,4,5,6).
Cy
EPNg
5
SOFT SOIL SITES
0.0
0.5
1.0
1.5
2.0
2.5
3.0
T /Tg
Figure 2.7
19
cov
cov
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
J.1=2
!l = 1
0.0
0.0
0.0
0.5
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
0.0
cov
cov
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.5
1.5
2.0
1.0
PERIOD (sec)
2.5
3.0
0.2
IJ.=4
J.1=3
0.0
0.0
0.0
0.5
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
0.0
cov
cov
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.5
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
0.2
1J.=6
IJ.=S
0.0
0.0
0.0
0.5
Figure 2.8
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
0.0
0.5
2.0
1.5
1.0
PERIOD (sec)
2.5
3.0
20
cov
cov
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
j.L
=1
j.L=2
0.0
0.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.0
0.5
PERIOD (sec)
1.0
1.5
2.0
2.5
PERIOD (sec)
cov
cov
1.0
1.0
IJ.8
0.8
0.6
0.6
0.4
0.4
0.2
3.0
0.2
j.L=3
j.L=4
0.0
0.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
PERIOD (sec)
PERIOD (sec)
cov
cov
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
j.L=5
j.L=6
0.0
0.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.0
PERIOD (sec)
0.5
1.0
1.5
2.0
2.5
3.0
PERIOD (sec)
21
cov
cov
1.2
1.2
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
!.1 = 1
J..l=2
0.0
0.0
0.0
0.5
1.0
1.5
2.0
2.5
0.0
3.0
0.5
cov
cov
1.2
1.2
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
1.0
1.5
2.0
2.5
3.0
PERIOD (sec)
PERIOD (sec)
0.2
J..l=3
J..l=4
0.0
0.0
0.0
0.5
1.0
1.5
2.0
2.5
0.0
3.0
0.5
cov
cov
1.2
1.2
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
1.0
1.5
2.0
2.5
3.0
2.5
3.0
PERIOD (sec)
PERIOD (sec)
0.2
J..l=S
0.0
0.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.0
PERIOD (sec)
0.5
1.0
1.5
2.0
PERIOD {sec)
22
cov
cov
, .2
1.2
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
1.1=1
1.1=2
0.0
0.0
0.0
0.5
1.0
1.5
2.0
2.5
0.0
3.0
0.5
PERIOD (sec)
1.5
2.0
2.5
3.0
2.5
3.0
2.5
3.0
PERIOD (sec)
COV
cov
1.2
1.2
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
1.0
0.2
1.1=3
0.0
0.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.0
0.5
PERIOD (sec)
1.0
1.5
2.0
PERIOD (sec)
cov
cov
1.2
1.2
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
1.1=5
0.0
0.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.0
PERIOD (sec)
0.5
1.0
1.5
2.0
PERIOD (sec)
23
cov
cov
1.2
1.2
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
!..l = 1
!..l=2
0.0
0.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.0
0.5
T /Tg
1.0
1.5
2.0
2.5
T /Tg
cov
cov
1.2
1.2
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
!..l=3
0.0
3.0
!..l=4
0.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.0
0.5
1.0
T /Tg
1.5.
2.0
2.5
3.0
T /Tg
cov
cov
1.2
1.2
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0.0
!..l=6
0.0
00
0.5
1.0
1.5
T /Tg
2.0
2.5
3.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
T /Tg
24
cov
cov
i.2
1.2
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
1.1=2
0.0
0.0
0.0
0.5
1.0
1.5
T/Tg
2.0
2.5
0.0
3.0
cov
cov
1.2
1.2
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.5
1.0
1.5
T /Tg
2.0
0.2
1.1=3
0.0
2.5
3.0
!J.=4
0.0
0.0
0.5
1.0
1.5
2;0
T/Tg
2.5
0.0
3.0
cov
0.5
1.0
1.5
2.0
T /Tg
2.5
3.0
cov
1.2
1.2
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
!J.=S
0.0
!J.=6
0.0
0.0
0.5
1.0
1.5
2..0
2.5
3.0
0.0
T /Tg
0.5
1.0
1.5
2.0
2.5
3.0
T /Tg
25
ROCK SITES
=1
0
0.5 .
0.0
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
Figure 2.14 Mean plus one standard deviation strength demands (normalized by PGA)
for rock sites.
Cy
EPNg
4
!J.
=1
ROCK SITES
!J.
0.0
0.5
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
Figure 2.15 Mean plus one standard deviation strength demands (normalized by EPA)
for rock sites.
26
11
4
ALLUVIUM SITES
I.A.
= 1
0
0.0
0.5
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
Figure 2.16 Mean plus one standard deviation strength demands (normalized by PGA)
for allwium sites.
cy
EPNg
4
ALLWIUM SITES
0.0
0.5
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
Figure 2.17 Mean plus one standard deviation strength demands (normalized by EPA)
for allwium sites.
27
i1
8
SOFT SOIL SITES
2
J.l.
0
0.0
=6
0.5
1.0
1.5
2.0
2.5
3.0
T /Tg
Figure 2.18 Mean plus one standard deviation strength demands (normalized by PGA)
for soft soil sites.
Cy
EPNg
8
SOFT SOIL SITES
2
J.l.
0.0
=6
0.5
1.0
1.5
2.0
2.5
3.0
T /Tg
Figure 2.19 Mean plus one standard deviation strength demands (normalized by EPA)
for soft soil sites.
29
3. STRENGTH REDUCTION SPECTRA
3. 1 GENERAL REMARKS
Current earthquakeresistant design philosophy accepts structural and nonstructural damage in
the event of severe earthquake ground motions. A consequence of this philosophy is that code
design forces are smaller, and in some cases much smaller, than those required to maintain elastic
behavior.
As discussed in Ref. 1, in present U.S. seismic codes the reduction in forces is done through the
use of structural modification factors (R or Rw). In the commentaries on these codes it is stated
that these R and Rw factors should account for at least dissipation of energy through plastic
behavior (p.), increase in damping, and the inherent overstrength in the structure. Nevertheless,
there is no rational procedure from which the recommended values have been derived, and, as
presently specified, they are empirical and only a function of the structural system, thus assuming
that the R or Rw are the same regardless of the values of the period of vibration, T 1 , or the local
site conditions. Moreover, the conglomeration of so many variables into one factor obscures the
underlying intent, and greatly impedes the development of improved understanding.
Improvements in the design process can be achieved if the factors contributing to the reduction
of seismic forces are taken into account explicitly. As discussed in Ref. 1, the R factor can be
defined as
(3. 1)
where:
R~
R;
the reduction due to the real strength of the constructed structure at the
time that the earthquake occurs compared with the code required
design strength: Rs
= Rovs
+1
Rovs the reduction due to the overstrength between the real strength and the
code required strength.
30
The study reported herein was devoted to obtaining reliable data on the quantification of R11
paying particular attention to how the values of R11 are affected by local soil conditions, including
very soft soil.
3. 2
DISSIPATION OF ENERGY
In this study an attempt has been made to estimate the reduction in strength that is produced by
allowing inelastic behavior.
(referred to in this study as R) is given by the ratio of the elastic demand to the strength demand
on an inelastic system undergoing a certain ductility demand, 1Ji ,
R"
= C1 (~=1)
(3. 2)
Cy(~=~i)
This ratio was obtained for a total of 31,000 different SDOF systems (the product of 124 ground
motions, 50 periods and 5 levels of displacement ductility). Figure 3.1 shows the mean of the
strength demand reductions due to nonlinear behavior for ground motions recorded on rock. It
can be seen that, unlike codespecified reduction factors, force reductions are period dependent.
In general, the reductions are small in the shortperiod range, increase to values larger than p, for
periods between 0.8 and 2.5 seconds, and approach p, for periods greater than 2.5 seconds.
Maximum reductions are produced at a period of approximately 1.3 seconds.
Reductions
corresponding to mean minus one standard deviation are shown in Fig. 3.2. Strength reduction
factors for alluvium and soft soil sites are shown in Figs. 3.3 and 3.4. Strength reductions for
alluvium sites follow, in general, the same trend as strength reductions for rock sites. For periods
shorter than 1.2 seconds, reductions for large p, (p. > 2) computed for alluvium sites are larger
than those computed for rock sites.
For soft soils and periods close to the predominant period of the site, very large reductions are
produced. For the same displacement ductility ratio (larger than 4) these reductions are nearly
twice the maximum reductions computed for rock or alluvium sites. For periods greater than two
times the predominant site period (T!fg > 2.0) strength reductions are approximately equal top,.
31
ForT< 0.9Ts, the reductions are small, smaller than fl. The sharp decrease of
R~
with T < Ts
compared toT> Tg points out clearly the importance of estimating as accurately as possible the
Ts and the T of the structure. Because of the uncertainties involved in such estimations, it will
be necessary to consider a band of values of Tg as well as ofT.
In order to study the dispersion of the computed strength reduction factors, coefficients of
variation (COV) were computed for each type of soil condition and each different displacement
ductility ratio. Dispersion of computed values of R.u are shown in Figs. 3.7, 3.8 and 3.9, where
COVs are presented for each type of soil condition and four levels of displacement ductility.
Unlike normalized strength demands, COVs of R.u do not show important variations with changes
in period, and increase slightly with increasing displacement ductility.
33
RJ.L
8
J.L=6
ROCK SITES
0
0.0
0.5
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
Figure 3.1 Mean of strength reductions due to nonlinear behavior for rock sites.
R!!
8
ROCK SITES
0.0
Figure 3.2
0.5
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
Mean minus one standard deviation of strength reductions due to nonlinear behavior for rock sites.
34
R~
8
ALLUVIUM SITES
Jl=6
!.1=5
Jl=4
!.1=3
!J.=2
0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
PERIOD (sec)
Figure 3.3 Mean of strength reductions due to nonlinear behavior tor alluvium sites.
R~
8
ALLWIUM SITES
0.0
Figure 3.4
0.5
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
Mean minus one standard deviation of strength reductions due to nonlinear behavior for alluvium srtes.
35
RJ.L
i4
'12
'10
8
6
0.0
0.5
1.0
1.5
2.0
2.5
3.0
T /Tg
Figure 3.5
Mean of strength reductions due to nonlinear behavior for soft soil sites.
'12
10
14
6
4
2
0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
T /Tg
Figure 3.6
Mean minus one standard deviation of strength reductions due to nonlinear behavior for soft soil sites.
36
cov
cov
, .0
1.0
!l=2
!l=3
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0.0
0.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.0
0.5
PERIOD (sec)
1.0
1.5
2.0
2.5
3.0
PERIOD (sec)
cov
cov
1.0
1.0
!l=4
!l=5
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0.0
0.0
0.0
0.5
1.0
1.5
2.0
PERIOD (sec)
Figure 3.7
2.5
3.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
PERIOD (sec)
37
cov
cov
1.0
1.0
!1=3
!J.=2
0.8
0.8
0.6
0.6
0.4
0.4
0.2
~~
0.0
0.2
0.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.0
0.5
PERIOD (sec)
1.0
1.5
2.0
2.5
3.0
PERIOD (sec)
cov
cov
1.0
1.0
!1=5
!J.=4
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0.0
0.0
0.0
0.5
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
PERIOD (sec)
38
cov
cov
1.0
1.0
J..L=2
J..L=3
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0.0
0.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.0
0.5
1.0
T /Tg
1.5
2.0
2.5
3.0
T/Tg
cov
cov
1.0
1.0
J..L=4
J..L=5
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0.0
0.0
0.0
0.5
1.0
1.5
2.0
T /Tg
2.5
3.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
T/Tg
39
4. INELASTIC DISPLACEMENT DEMAND SPECTRA
4. 1 INTRODUCTORY REMARKS
In Ref. 1, the importance and the need for controlling the relative displacements (drift) of a
structure when subjected to seismic excitations is discussed in detail. This reference contains the
following statements:
"While displacement ductility factors generally provide a good indication of structural
damage, they do not usually adequately reflect the damage to nonstructural elements.
This is an important limitation in seismicresistant design since a significant portion of
the hazard to occupants and of the total cost of repairing earthquake damage is a
consequence of nonstructural damage. Nonstructural damage is more dependent on the
relative displacements (drift) than on the overall displacements. To obtain a reliable
measure of nonstructural damage, maximum drifts must remain unnormalized or be
divided by the value of drift corresponding to the damage threshold.
Nonstructural
For example,
nonstructural damage for relatively rigid structures may be small even for large values of
displacement ductility since the yield displacement may be well below the nonstructural
damage threshold. On the other hand, the nonstructural damage and lateral displacements
for flexible structures may become intolerable large even before significant yielding
develops.
To produce safe and economical structures, seismic resistant design methods must
incorporate drift (damage) control in addition to lateral displacement ductility as design
constraints."
"The growing concern over the costs of earthquake damage (direct, functional and
indirect) and the difficulty of repairing much postyield damage, points out the need that
more attention should be given to control of damage and repairability at the design stage.
These needs have been clearly emphasized by the 1989 Lorna Prieta earthquake. The
control of damage will, of course, also help to improve human safety, which is the
40
In summary, achievement of reliable and efficient EQRDs requires satisfaction not only
of the criterion for strength and toughness, but also the criteria for deformation and
repairability.
As discussed in Ref. 1, it is believed that one of the most practical methods for including the
criterion of IDI control in the preliminary design is to use displacement spectra for the different
levels of earthquakes that are included in the design. For the case of two levels of design
earthquakes, service and safety (survival) levels, a procedure has been suggested in Ref. 8. The
method is based on the use of Smoothed Design Response Spectra (SDRS) in tripartite
logarithmic scale, as discussed in Ref. 9. For the serviceability limit state, the displacements can
be read directly from the SLEDRS for strength. For the safety limit state, although there is a
need for two spectra (one for the strength and one for the displacement), the IDRS for
displacement can be obtained directly from the IDRS for strength for any given ductility ratio,
p., by multiplying their ordinates by the corresponding p.. This approach is discussed below.
From the point of view of present practice, it was considered of interest to try to find out to what
extent inelastic displacement demands can be predicted using elastic analysis. Thus, the ratio of
inelastic to elastic deformation was computed for each timehistory analysis, and for a set of 50
periods between 0.05 and 3.0 seconds.
be considerably larger than elastic demands. Previous studies have recommended fixed periods
(independent of p.) to specify spectral regions in which elastic analysis can be used to estimate
41
inelastic displacement demands. However, it can be seen that this period is clearly dependent
on the ductility level.
approximately the same as elastic displacements for periods greater than 0.5 second, while for
a ductility ratio of 6 such an assumption is only valid for periods greater than about 1.0 second.
Using the normalized strength demands in Figs. 2.2 to 2.7, elastic and inelastic displacement
demands can be obtained using the expression
(4. 1)
or
= (CjEPA)ILT (EPA)
(4. 2)
41t2
Figure 4.2 is an example of a displacement demand spectra which has been computed assuming
an EPA of 0.4g, consistent with that assumed by ATC 306 [6] for the regions of highest seismic
risk in the U.S. Even though inelastic displacement demands can be considerable larger than
elastic demands for very short periods (T < 0.2 second), displacement demands in this region are
very small.
Displacement ratios for ground motions recorded on alluvium are shown in Fig. 4.4.
Displacement demands assuming an EPA of 0.4g are shown in Fig. 4.4. In this case, for a
ductility ratio of 2 inelastic displacements are approximately the same as elastic displacements
for periods greater than 0.3 second.
While Figs. 4.2 and 4.4 give the most probable displacement demands for rock and alluvium sites
(because they are based on mean values), caution should be exercised in the use of these spectra
or any other mean displacement spectra which have been computed from spectra normalized by
acceleration parameters (i.e., PGA or EPA) which result in large dispersion in the longperiod
42
range.
Mean ratios of inelastic to elastic displacements for ground motion records on soft soil are shown
on Fig. 4.5. For periods near the predominant period of the site, inelastic displacements are 30%
to 40% smaller than the corresponding elastic displacements. This observation has important
design implications. For instance, the use of energy dissipation devices can be particularly
effective in this situation, where not only very large force reductions may be achieved, but also
(and more importantly for damage control) significant reductions may be obtained in
displacement demands. However, these reductions should be used cautiously for design, because
they depend on accurate estimations of T and Tg.
For periods shorter than the predominant site period, inelastic displacement demands can be
significantly larger than elastic demands, which for sites with very long periods, T g , will have
an effect on a very large range of periods, T. An example of this situation is shown in Fig. 4.6,
where mean displacement demands have been computed for a soft soil site with a predominant
period of 1.5 seconds and an effective peak acceleration of 0.25g. In this case, displacement
demands based on elastic analysis can significantly underestimate inelastic demands for buildings
or structures with fundamental periods as long as 1.3 seconds.
43
ll.:nel as tic
fl. elastic
4 ~~
ROCK SITES
       1L = 1
................ 1L = 2
!1=3
1L =4
1L = 5
0
0.0
Figure 4.1
1.0
0.5
1.5
2.0
PERIOD (sec)
2.5
3.0
DISPLACEMENT (in)
16
...
......
/ ...,.=
!L=2
12
___
ROCK SITES
!!= 1
14
~L=4
10
1L
=6
6
4
2
EPA= 0.4g
0.0
Figure 4.2
0.5
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
44
.6.inelastic
.6.elastic
~,
ALLUVIUM SITES
\\
3


1~.
,\~.
\':\
\\ ,.,
......' ...................
''':::;:. ~.
. . . . . .
~ =__,~,
!l = 1
!l =2
!l =3
!l = 4
!.1=5
!.1=6
=...
~
0
0.0
Figure 4.4
0.5
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
DISPLACEMENT (in)
16
ALLWIUM SITES
!.1=1
14
!l = 2
12
J.L=4
10
    !l = 6
6
4.
2
EPA= 0.4g
0
0.0
Figure 4.3
0.5
1.0
1.5
2.0
PERIOD (sec)
2.5
3.0
Mean of displacement demands in elastic and inelastic systems for alluvium sites assuming an EPA of 0.4g.
45
~inelastic
Llstastic
\\
\\
\
11=1
 t.l = 2
       t.l =3
     t!=4
t!=S
     t!=6
\'
\~\ \\_
'...\
.. \"' "\\'.
'''
.... . '~'":=~
... :..::: ~
0.5
0.0
1.0
1.5
2.0
2.5
3.0
T /Tg
Figure 4.5 Mean of ratio of inelastic to elastic displacement demands for soft soil
sites.
DISPLACEMENT (in)
20
SOFT SOIL SITES
J.1=1
............... !J. = 2
!J=4
!J=6
15
10
Tg = 1.5 sec
PGA= 0.25g
0.0
0.5
1.0
1.5
2.0
2.5
3.0
PERIOD (sec)
Figure 4.6 Mean of displacement demands in elastic and inelastic systems for a soft
soil site with a predominant site period of 1.5 sec and assuming a PGA of
0.25g.
47
5. SUMMARY AND CONCLUSIONS
5.1 SUMMARY
Nonlinear response spectra for 124 earthquake ground motions recorded on various soil
conditions, ranging from rock to very soft soils, were computed and analyzed statistically to
'
provide engineers with improved tools to estimate strength and displacement demands on new
and existing buildings. For each record, responses were computed for 50 different periods
between 0.05 and 3.0 seconds, and for 6 displacement ductility ratios, 1, 2, 3, 4, 5 and 6. The
study was limited to computing the responses of SDOF bilinear systems with postelastic stiffness
of 3% of the elastic stiffness and with a damping ratio of 5% of critical. Average (mean) and
mean plus one standard deviation inelastic strength demand spectra were computed for rock (38
records), alluvium (62 records) and soft soil sites (24 records). These spectra provide adequate
tools with which to estimate strength demands in a deterministic framework.
Based on the results obtained for the strength demand spectra, a comprehensive statistical study
of the strength reduction factor due to hysteretic energy dissipation, R.u , i.e., displacement due
to displacement ductility ratio p., was conducted. The main purpose of this study was to obtain
reliable data on which to judge the reliability of present code recommended values for the
strength reduction factors R and
affecting these values. Emphasis was given to studying how the values of R14 are affected by soil
conditions, including the effects of very soft soils.
Displacement seismic demand spectra were obtained using the normalized strength demand
spectra.
Present practice for checking against lateral displacement is based on the assumption that the
inelastic displacement demands for severe earthquake ground motions can be based on estimation
of the elastic demands and multiplication of such elastic demands by an empirical coefficient
which is independent of soil conditions.
computation of the ratio of inelastic to elastic displacement demands for each of the 124 ground
motions.
48
5. 2 CONCLUSIONS
From analysis of the results obtained, the following observations can be made.
Spectral shapes for inelastic strength demands (p. > 1) differ significantly from elastic
(p.
= 1) spectral shapes.
The largest dynamic amplification for elastic response (p. = 1) is induced by soft soil sites.
However, these large amplifications are significantly reduced when inelastic behavior
occurs (p. > 1). For ductility ratios larger than 3, inelastic strength demands decrease
monotonically with increasing period.
For soft soil records and periods smaller than the predominant period of the site
(T!f8 < 1), there is little difference between the strength demand for ductility ratios
between 2 and 6. This implies that small changes in the yielding strength of structures
with T < T 8 may produce large changes in the ductility demands.
Spectral amplifications between 0.1 and 0.5 seconds for ground motions recorded on soft
soil are usually much smaller than the 2.5 factor which is used to define the Effective
Peak Acceleration (EPA). For this type of soil condition, the use of PGA is probably
more appropriate than the use of EPA.
period of vibration, the level of displacement ductility ratio, and the local soil conditions.
49
The dispersion of strength reduction factor R.u was observed to be nearly independent of
the period of vibration, and to increase with increasing ductility ratio.
For soft soil conditions, the values of R,u are characterized by small values for
Ttrg < 1 and by very large reductions for periods close to the Tg . The R,u values are
approximately equal to f.l for Ttrg greater than 2.5. This means that estimation of the
predominant period, Tg , of the site . is of particular importance when designing or
upgrading structures on soft soil sites, where inelastic strength demands are strongly
influenced by the Ttrg ratio.
Values of the reduction factor R,u which are based on the assumption that the maximum
displacement of an inelastic system is the same as that for an elastic system, i.e., R,u
=f.l,
are unconservative for structures with short T. Values of R,u which are based on the
assumption that energy absorption in inelastic systems is equal to energy absorption in
elastic systems, i.e., R,u = [2f.l  1]112, are also unconservative for structures with short T.
The mean values of the ratio of inelastic deformation to elastic deformation show that for
structures with short T the inelastic displacement demands can be considerable larger than
the elastic demands.
The range of the values of the structural period for which elastic analysis can be used
directly to estimate the inelastic displacement demand is dependent on the ductility ratio
level and the soil conditions.
For soft soil sites and for values of Ttrg very near to 1, inelastic displacements can be
up to 40% smaller than the corresponding elastic displacements. For values of Tffg<0.8,
the inelastic displacement demands can be significantly larger than the elastic demands,
so that for sites with very long Tg , the displacement demands based on elastic analysis
can significantly underestimate inelastic displacement demands of structures having T as
large as 1.5 seconds, or even larger, depending on the value of Tg .
50
From the above observations it is obvious that the studies reported herein clearly indicate the
importance of having a reliable estimation of the predominant period of the site, particularly in
case of soft soils.
51
6. REFERENCES
[1]
Bertero, V.V., et al., "Design Guidelines for Ductility and Drift Limits: Review of the
StateofthePractice and oftheArt on Ductility and DriftBased EarthquakeResistant
Design of Buildings," A CUREeKajima Report, July 1991.
[2]
[3]
Bertero, V.V., and Uang, CM., "Issues and Future Directions in the Use of an Energy
Approach for SeismicResistant Design of Structures," Proceedings of the Workshop on
[4]
[5]
Miranda, E., "Seismic Evaluation and Upgrading of Existing Buildings," Ph.D Thesis,
Civil Engineering Dept. of the University of California at Berkeley, California, May,
1991. (To be published in two separate EERC reports).
[6]
[7]
Seed, H. B., et alia, "Site Dependent Spectra for Earthquake Resistant Design, 11 Report No
52
[8]
[9]
Ridell, R., and Newmark, N.M., "Statistical Analysis of the Response of Nonlinear
Systems Subjected to Earthquakes," Structural Research Series No. 468, Department of
Civil Engineering, University of Illinois, UrbanaChampaign, August 1979.