Design Considerations

The Role of Ductility

in Seismic Design
design, however, ductility plays a fundamental
role in determining the level of force for which
If the technical bases of building a structure must be designed. Even though
code provisions for seismic ductility is essential to code-based seismic
design, this fact is rarely explicitly stated, or
design loads are not thoroughly sufficiently explained, in building codes.
understood, the use of the code While not denying the usefulness of a
will not necessarily result in a detailed analysis of structural response to
earthquake loads, a qualitative examination of /
sound structural design. Some the dynamic behavior of some very simple
level of ductility must be structural models can help in achieving a "gut
provided. feeling" for the seismic behavior of structures,
including the role of ductility in limiting
seismic design loads. The use of response
DAVID 0. KNUITUNEN spectra as a simple, approximate means of
dynamic structural analysis for seismic load-

E
ARTHQUAKE loadings on structures ings presents a more quantitative example of
are complex, and the dynamic, inelastic the fundamental importance of ductility
response of real structures to intense provisions in code-based seismic design.
earthquake shaking tends to be extremely
difficult to analyze and understand. Building The Nature of Seismic Loadings
codes, which usually present seismic design The behavior of structures subjected to earth-
provisions in terms of pseudo-static loadings quake loads is complex. Earthquake ground
and elastic analysis, insufficiently emphasize motion typically contains many cycles of
the technical bases of these provisions. In rapidly varying acceleration, each with a
particular, the importance of ~eismic detailing different amplitude and a different duration
requirements, which are intended
I
to ensure (see Figure 1). The intensity of shaking during
the ductility of a structure, can be easily strong earthquakes is sufficient to push most
misunderstood by those who must interpret · structures into the inelastic range, which
the codes. Engineers are used to thinking of greatly complicates their analysis. Inelastic
ductility as a desirable attribute of structural time-history analyses of realistic three-dimen-
systems, but secondary in importance to such sional structures are rarely undertaken in a
attributes as strength and stiffness. In seismic practical design-office environment, primarily

CIVIL ENGINEERING PRACTICE SPRING 1987 17

 150 .-----,------.----.----.-~---.---"T---T""""--....----.----.----,

--1
100 - Duration of Strong
Ground Shaking

50

N
u
Q)
~
E
u
.!:: -50

Double
~ Frequency Content
Amplitude

-150

-200

-250 .___ ____,_ _........._ ___.__ _.....___ _...__ _..___ _..______.___ __.,_ ___.__ ___,
0.0 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00 22.00
Time in Seconds Ref. 8

FIGURE 1. An earthquake ground acceleration record.

due to the modeling complexities and the elements of the structure.3,4

immense amounts of computer time that are Unfortunately, this form of code-based
involved. Even if such an analysis were under- analysis tends to obscure the dynamic and
taken, it would only furnish the response of a inelastic characteristics of the response of real
building to a particular earthquake. A future structures to major earthquakes. In particular,
earthquake may prove much more damaging. the role of inelastic behavior and ductility in
Because of the practical impossibility of determining static loadings, although funda-
exact seismic load analysis, earthquake- mental, is rarely explicitly stated in code
resistant design is presently based on various formulations.
approximate methods. One such method uses
response spectra, which are derived from basic Simplified Models
physical principles and from empirical for Structural Response
observations of the effects of various earth- The qualitative analysis of some very simple
quakes on structures of different natural structures, such as those illustrated in Figures
periods. 1•2 However, the more common 2 through 6, can be very helpful in achieving
approach, which is currently implemented in a "gut feeling" for the response of structures
most building codes, is to use some fairly to earthquake loading. In these examples, a
simple formulas to determine a design base simple idealization of a single-story portal
shear and to distribute this shear vertically to frame is analyzed. The girder is assumed to be
the various levels of the building. Methods of rigid, so that all bending occurs in the
elastic analysis are then used to distribute columns. Also, the mass of the entire structure
these pseudo-static forces to the various is assumed to be lumped at the floor level.

18 CIVIL ENGINEERING PRACTICE SPRING 1987

 I
I
r---
1----
I
I
. ma

- - Rigid
I I
I I V=ma
~ ~

FIGURE 2. A rigid structure subject to base acceleration.

This type of structure is presented since its determined by:

behavior is very easy to visualize. However,
the principles derived from its analysis can
also apply to braced frame or shearwall build-
ings. In their most general terms, these where:
principles can apply to more realistic, multi-
story structures as well. E = the modulus of elasticity

Response of Elastic Structures !col = the moment of inertia of the column

to Ground Accelerations
If the columns of a portal frame were com- h = the height of the column
pletely rigid, and the structure were subjected
to an acceleration of its base, the shearing Restated, the formula for the shearing force is:
force in the structure would be equal to the
mass of the structure times the acceleration of
the base (see Figure 2). If the direction (sign)
of the base acceleration changed, the direction The mass of the structure is subjected to
of the inertial force would change instantan- an acceleration, a:
which is initially less than
eously, and the V = ma relationship would the base acceleration, a. Equilibrium requires
remain true. Thus, the maximum force exper- that the inertial force be equal to the shear in
ienced by a rigid structure would always be the supporting structure:
exactly equal to the maximum base accelera-
tion times the mass of the structure. ma'= kt:.
When a more realistic, elastic structure is
subjected to a base acceleration, its supports Initially, kt:. is less than the mass times the
will flex and the mass will initially lag behind ground acceleration.
the base (see Figure 3). The initial shearing For an earthquake-type ground motion,
force in the supports is equal to the displace- a, a' and t:,. change with time. When the direc-
ment of the mass relative to the base times tion of the base motion changes, the mass of
the stiffness of the supports. For example, for the elastic structure tends to keep moving in
the portal frame shown in Figure 3: the original direction, producing a whiplash
effect. Due to this sort of dynamic response,
V= kt:. the maximum elastic forces experienced by
the structure can be much higher than the
The elastic stiffness of the system, k, is product of the mass and the maximum accel-

CIVIL ENGINEERING PRACTICE SPRING 1987 19

 .
A

I .....,.,,,,,.....,,,,,,,.,,,,,.....,,,,,,,... 1 a' ..
r-
ma~
1- -
I
I
I
I
~
V=kA
.
a
..
FIGURE 3. An elastic structure subject to base acceleration.

eration of the ground. connections. Since the damping force always

For most structural models, some degree tends to resist the motion of the structure
of damping should be included (see Figure 4). relative to its base, the presence of damping
The damping effect adds another term to the reduces the maximum value of structural
balance of forces equation: response to seismic load.
The amount of damping present in a
ma'= kLi -cv structure is usually expressed as the ratio of
the actual damping coefficient to the critical
where: damping coefficient. Typical structural damp-
ing ratios used in seismic design vary from 5
c = the coefficient of damping, to 10 percent, although values from 1 to 20
percent may be appropriate in certain cases. 2
v = the velocity of the mass of the struc-
ture, which has a positive sign if it is Limitation of Seismic
in the same direction as a~ Forces Due to Yielding
An elasto-plastic structure is one that behaves
The damping coefficient is a measure of like an elastic structure up to a determined
the energy dissipation that occurs in a struc- yield limit. At that point, plastic deformation
ture due to such factors as internal friction in takes place, and the system begins to deflect

I
r- ma
1- -
I
I
I

.
I
~

a
..
V=kA

. CV

FIGURE 4. An elastic structure with damping.

20 CIVIL ENGINEERING PRACTICE SPRING 1987

 V Rupture

Plastic Hinge

- V A
~i~-r-~~_::;-,:t- Inelastic Deformation, Ad
ATotal :S: Ar

Ductilityµ= A/ Ay

FIGURE 5. An elasto-plastic structure.

continuously without any additional load. This ture must be ensured to be at least equal to
deformation continues until the structure the ductility required, or demanded, by the
eventually collapses or the material ruptures earthquake:
or buckles. The available ductility, or ductility
capacity, of the structure can be defined as
the total deflection at collapse or rupture
divided by the elastic deflection at first yield For a given earthquake ground motion,
(see Figure 5). the force that the structure must be able to
An elasto-plastic structure subjected to a resist elastically is inversely related to the
base acceleration initially responds like an available ductility, which is the basis for code
elastic system. However, if the deflection, !J., provisions that prescribe a lower design force
exceeds the yield deflection, !J.Y, the shear for structures with higher levels of ductility.
levels off at the yield limit (see Figure 6). However, the converse of this rule is also true.
Larger forces cannot be developed in the For a given earthquake, the lower the available
structure. Put another way, it is not necessary to resistance (strength) of the structure, the
design the structure for larger forces. higher the level of ductility that will be
The structure shown in Figures 5 and 6 demanded if the structure is not to collapse.
becomes unstable after the plastic hinges form. For more complex structures and for
If the acceleration continued in the same actual, reversible earthquake loadings, while
direction for an extended period of time, the the concept of ductility becomes much more
structure would collapse. However, earth- complex to define with precision, 5 the basic
quake accelerations do not remain constant, result is the same. Seismic design forces can
but change direction rapidly. When the accel- be reduced where high levels of inelastic
eration reverses, the deflection of the mass deformation are possible. However, the struc-
relative to the ground lessens and the struc- tural detailing must be consistent with the
ture unloads as indicated in Figure 6. level of ductility that is assumed, explicitly or
For the structure to remain safe after implicitly, in the design.
yielding has commenced, the maximum in- Structures designed to have extremely
elastic deflection that occurs in the structure high levels of ductility, with correspondingly
during the earthquake (the response) must be low elastic resistance, will experience large
less than the deflection required to cause the inelastic deflections when an earthquake
structure to collapse. If the required ductility, occurs. Since these deflections can cause
µreq' is defined as the maximum actual deflec- extensive damage to non-structural compon-
tion response divided by the· deflection at ents, they must be taken into account in the
yield, then the available ductility in the struc- design of highly ductile structures. Large

CIVIL ENGINEERING PRACTICE SPRING 1987 21

 a) 6;:,: 6y
a'
..
r-
I- -
I
I
I
- ma'

r--1,
.
I

I
V = Vy

.
,.J,

b) V
/ Rupture

Vy ---➔.{

- ----------~
Ay Ad

FIGURE 6. An elasto-plastic system subject to base acceleration (a); and unloading of an

elasto-plastic system (b).

inelastic deflections can also contribute to The analysis of simple oscillators is useful
significant P-A effects, which should be taken for developing an intuitive feel for dynamic
into account in the structural design. behavior. In addition, the responses of a
number of simple oscillators can be added
Multi-Degree-of-Freedom Structures together to derive the response of a more
The simple, one-story structures presented so complicated multi-degree-of-freedom system
far are examples of single-degree-of-freedom using the method of mode superposition (see
structures, or simple oscillators (see Figure 7). Figure 8).
Their dynamic characteristics can be com- Any displacement pattern of a structure
pletely described in terms of a single fre- with N degrees of freedom can be expressed
quency or period of vibration (which is simply in terms of N modal displacement patterns,
a function of the mass and stiffness of the which represent possible free-vibration
structure), and for damped systems by a single responses of the structure. Each modal dis-
damping ratio. placement pattern has a fixed shape in which

22 CIVIL ENGINEERING PRACTICE SPRING 1987

 A
. m
A
.
Undamped Oscillators

A
..
.
C
m

A
C
----K

Damped Oscillators

. ..
Period, T

Undamped Oscillator Damped Oscillator

Free Vibration Response

FIGURE 7. Illustration of typical simple oscillators.

the displacements of each degree-of-freedom mode of the real structure can be treated as
remain in constant ratios. For any given an independent simple oscillator, with equiv-
dynamic loading, the response of the structure alent stiffness and mass that are functions of
in each mode is completely defined by the the actual stiffnesses and masses in the struc-
mode shape and a single time-varying modal ture and the mode shape. The time-varying
amplitude factor. In dynamic analysis, each responses of all modes can then be summed

CIVIL ENGINEERING PRACTICE SPRING 1987 23

 dividing the mass of the oscillator to produce
a quantity having units of acceleration, and
referred to as pseudo or pseudo spectral accelera -
tion. Typical response spectra for earthquake
loadings are shown in Figure 9.
A given response spectrum can be plotted
for only a single ground motion time-history
(earthquake acceleration record) and for only
a single damping ratio. Typically, as in Figure
9, a family of curves is plotted for a particular
I- earthquake acceleration pattern, one each for
I a range of damping values. Response spectra
I can also be plotted that take into account
I inelastic behavior. In this case, each curve also
I assumes some specific ductility ratio.

~-- Spectra are often plotted on 3-way log

paper as shown in Figure 10. With such a plot,
it is possible to directly read either the
maximum deflection or the maximum force
(usually normalized to pseudo acceleration).
Actual earthquake response spectra
typically appear jagged as revealed in Figures
9 and 10, because small differences in the
fundamental (undamped, natural) period can
lead to rather large differences in response to
FIGURE 8. A multi-degree of freedom system. an earthquake motion. Smoothed spectra such
as those shown in Figure 11 have been
to furnish the total response of the struc- developed using statistical techniques to
ture.1,6,7 average out the irregularities so that the more
Quite often, for complex structures, sum- large-scale influence of period on response
ming the earthquake responses of only a few can be readily observed.2,s,9 Techniques have
critical modes is sufficient to produce a fair also been developed for constructing design
estimate of the total response of the structure. spectra from a few basic properties of the anti-
cipated ground motion. 2,9 An example of one
Response Spectra such design spectrum is shown in Figure 12.
It is possible to compute the maximum Design spectra such as that shown in Figure
response of a simple oscillator of any specified 12 are made up of linear segments when
period and damping coefficient from an earth- plotted on 3-way log paper, and are derived
quake acceleration record. A response spec- by applying amplification factors to the
trum is the plot of the maximum responses of maximum effective values of ground accelera-
a large number of oscillators of different tion, velocity and displacement. The amplifica-
periods to the same dynamic input (the earth- tion factors are determined by a statistical
quake acceleration record). The response analysis of the actual, computed response
spectrum is a convenient way to represent spectra. For a given set of amplification factors,
maximum structural response to a dynamic there is a calculable probability that the
loading. A response spectrum may be used maximum response predicted by the design
for design by entering with the period of the spectrum will not be exceeded.
structure and reading off the response, usually Amplification factors for elastic design
either the maximum displacement or the response spectra are functions of the assumed
maximum elastic force in the structure. damping level in the structure. Techniques
Maximum force is usually normalized by have been suggested for constructing inelastic

24 CIVIL ENGINEERING PRACTICE SPRING 1987

 5.0

4.0

3.0
C: ~
0 ·;;:
·.;::; ~
...
~
Q) lJ
,._ 2.0
aJ
u
0
u C:
<( 0
•;:;
~... ... ~
Q)
ti
Q) aJ
u
c.. u
Vl
<(
II 1.0
0.9
als: 0.8
II
u 0.7 No Damping
w
C:
Q)
0.6
·o
:£:Q) 0.5
0
u 0.4
Q)
...u0
u..
~ ... 0.3
-
Q)
~
....J

0.2

0.1 - - - - - - - -..........~i.....i.....a.-'-......,__._..._..._._,_1....,1_.....-'--'
0 0.5 1.0 1.5 2.0 2.5
Undamped Natural Period, T, sec
Ref. 20

FIGURE 9. Typical earthquake response spectra.

design response spectra from the elastic ment and force patterns using the usual
spectra for assumed levels of available methods of dynamic analysis by mode super-
ductility. 2 · position. However, since the maximum
To use a response spectrum for the design responses of the structure in different modes
of multi-degree-of-freedom structures, the generally do not occur at the same time, the
designer enters the spectrum with the period modal responses from the response spectrum
of vibration and damping ratio of each signifi- are usually not summed directly. Instead, the
cant vibration mode of the structure, and modal responses are summed by approximate
reads off the desired quantity: design force, or methods based on the principle of mode
maximum deflection. These quantities are then su perposition. 1•2•6
converted into appropriate modal displace- The principle of mode superposition

CIVIL ENGINEERING PRACTICE SPRING 1987 25

 200

100 ,__..,..._...............
80 ,__..._._..,._____
60 I---+<-,_ _ _ __

u
QJ 10
"'
'- 8
·=i, 6
·o0
Q)
> 4
E
::,
E
·x
2"'
2

1
0.8

0.4

0.2

0.1
0.04 0.06 0.1 0.2 0.4 0.6 1 2 4 6 8 10 20

Undamped Natural Period of Simple Oscillators, sec Ref. 8

FIGURE 10. Earthquake response spectra on a 3-way log plot. These curves are for a different
earthquake acceleration pattern from Figure 9. Each curve represents a single damping value.

strictly applies only to linear elastic structures. incidence of yielding is well distributed
However, because of its convenience, the throughout the structure.2
method is used often in conjunction with The ground motion inputs to a design
inelastic response spectra to estimate the response spectrum (the maximum ground
response of structures in the inelastic range. acceleration, maximum ground velocity and
This approximation is reasonably accurate for maximum ground displacement) may be taken
low levels of required ductility, providing the from a single historical earthquake record.

26 CIVIL ENGINEERING PRACTICE SPRING 1987

 22

20
I
I
18

16
, I

, I
14

12

10

8
, I

I /
I /
I
6
I,
II '
4
1t,'.,,,
,_
""""-
2

0 ...__..__ _.__.,___,__ __,__....___,__ _.__ _.____._ _,___...__.___,__ _,

0 0.4 0.8 1.2 1.6 2.0 2.4 2.8
Period, sec

50

20

10

u(1)
.,,
'.!: 2
·oi-
0
Qi 1
>

0.5

0.2

0.02 0.05 0.1 0.2 0.5 2 5 10

Period, sec Ref. 8

FIGURE 11. Smoothed earthquake response spectra.

CIVIL ENGINEERING PRACTICE SPRING 1987 27

 200

100

50
u
Q)
V,
.......
E
u

·oi-
0
20
Qj
>
10

1 ---~""'--~~"--~--...
0.1 0.2 0.5 1
'-'-~----~~---;.i,,_~---"--J
2 5 10 20 50 100
Frequency, Hz Ref. 2

FIGURE 12. An elastic design response spectrum.

However, no specified earthquake will the spectrum is an inelastic response

represent the most critical possible loading for spectrum.
all types of structures. For this reason, spectra
are often plotted for composite design earth- Response Spectra &
quakes. Such spectra attempt to combine the Code Seismic Design Forces
portions of the response spectra of various Figure 13 shows the plots of two approximate
historical earthquakes that are most likely to design spectra derived from the ground
cause high levels of damage to structures in motion characteristics of the 1940 El Centro
each given frequency range. Therefore, they earthquake in California. While the method
supply a much safer design than the spectrum used to plot the spectra follows the method
of any specific actual earthquake. 2,8,10 that was suggested by Newmark and Hall,2
It is important to remember that a the plots have been "delogged" since it is
response spectrum must be plotted: easier to view the relative values of the pseudo
acceleration ordinates on a straight plot than
• for a specific earthquake ground motion on a 3-way log plot. The two spectra that
(either an actual, historical earthquake or a have been plotted are an elastic design
composite "design earthquake"); spectrum for 5 percent damping and the
• for a given level of damping; and corresponding inelastic design spectrum for a
• for a given level of assumed ductility, if ductility factor of 5 times the yield deforma-

28 CIVIL ENGINEERING PRACTICE SPRING 1987

 0.9

0.8
-
00
0
"'Q) 0.7
El Centro: Elastic Design
Response Spectrum (5% Damping)
C.
·.;::;
:i 0.6 El Centro: Inelastic
~ Design Response Spectrum
C: (Ductility Factor of
0
·.;::;
0.5 5 times yield deformation)
...
n:I
Q)
UBC: K=1.33
a3
u 0.4
u
<
0
0.3 UBC: K=0.67
"O
:J
Q)

"'
Q,,
0.2

0.1

0
0 1 2 3 4
Period of Vibration (sec)

FIGURE 13. A comparison of UBC base shear (ultimate loads) with design response spectra.

tion. A 5 percent damping ratio is the level C = the seismic response factor, determined
that might pertain, for example, to a reinforced by the estimated elastic properties of the
concrete or .a bolted steel structure within structure
working stress levels, or a welded steel
structure stressed to near the yield point. 2 S = a soil/structure interaction factor
On the same figure, the pseudo accelera-
tions implied by the 1985 Uniform Building W = the total weight of the building
Code (UBC) for K factors of 1.33 and 0.67 have
been plott.ed. The UBC formula for base shear Since the pseudo acceleration is the
is: 3 maximum shear in the structure divided by
the mass and expressed as a multiple of g, the
V= ZIKCSW pseudo acceleration implied by the UBC base
shear formula is simply derived by:
where:
PSa = ZIKCS
Z = the zone factor, reflecting the likelihood of
strong earthquake shaking at the site In Figure 13, Z, I and S have been
assumed to equal 1. C, as defined by the UBC,
I= the importance factor, determined by the is a function of the period, T, specifically, C =
intended function of the building 1/(15*T·2). Because the design spectra represent
forces at ultimate levels of stress, the curves
K = a factor reflecting the probable level of based on the UBC have been multiplied by a
ductility, determined by the type of struc- load factor of 1.4. Typical safety factors used
tural system in practice for seismic design range from about

CIVIL ENGINEERING PRACTICE SPRING 1987 29

(1.33 x 0.6)-1 = 1.25 for steel structures to 1.4 for would produce design forces 3 to 5 times
concrete structures. 11,12 Therefore, the use of a those specified by the UBC.
1.4 factor minimizes the difference between Low code forces can be justified only if a
the UBC and the spectral forces. high level of ductility is present in the struc-
Pseudo accelerations can be thought of as ture. However, due to the shape of the Code
recommended design forces, expressed as a curves, it is not possible to determine the
fraction of the dead weight of the structure. exact ductility level assumed in the Code.
For any given period, T, the ordinate of one of Most notable and, perhaps, surprising to many
the design response spectra yields the exact designers are the high ductility levels that are
design force for a single-degree-of-freedom obviously necessary even for K = 1.33
structure with a period equal to T. The buildings.
ordinate also approximates the design force
for a multi-degree-of-freedom structure with a Ductility Capacity in Structures
fundamental period equal to T. The ordinate The determination of appropriate levels of
of the elastic response spectrum represents ductility to assume in the design of complex
the maximum expected force that the struc- structures is difficult and controversial,2,5,13,14
ture would be required to resist if yielding Overall ductility depends, for example, not
were not permitted. The ordinate of the only on the ductility capacity of the individual
inelastic spectrum represents the resistance structural members, but also on such factors
that would be required if yielding were as the sequence of plastic hinge formation in
permitted up to a ductility ratio of 5. The the structure, which in turn depends on the
corresponding ordinates of the UBC curves particular time-history of the loading. In fact,
indicate the minimum capacity, or resistance, the common practice of assigning a single
that would be expected to be found in a box number to the ductility of complex structures
type structure and a ductile moment frame of is at best an approximation, even though it is
the same period if they were designed in one that is essential for seismic analysis using
accordance with the UBC. inelastic design response spectra. The theore-
Two observations demand immediate tical shortcomings of this practice have been
attention when comparing the Code curves to pointed out, but it is widely accepted as being
the empirical design spectra. First, the general adequate for many practical types of
shape of the Code curves is not in particularly buildings. 2,5,14 Some specific suggestions have
close agreement with that of the design been made for typical ductility levels that can
spectra. The tails of the Code curves, which be expected in various types of structures.
vary proportionally to 1/T 0.s are much flatter Representative values are 4 to 6 for carefully
than the empirical curves, which vary with detailed concrete buildings; up to perhaps 8
1/T or 1/T2. Second, the Code forces are far for properly detailed steel structures.2
lower than those given by the elastic design Ductility ultimately depends on the
spectrum. proper detailing of the structure, and in
For example, for a moment frame building particular the joints and connections. 15,16,17,18
in California with a period of 2 seconds, the For a concrete structure, this detailing means
UBC would specify an (ultimate) base shear of such things as paying careful attention to
0.044 times the building's weight. The elastic minimum and maximum reinforcing ratios,
response spectrum suggests that had the thorough detailing of splices and anchorages,
building been subjected to the El Centro and confining concrete by closely spaced
earthquake and remained elastic, it might have hoops or spirals in areas where inelastic
experienced forces equal to about 0.24 times behavior can be expected to occur. In a steel
its weight, or more than 5 times the level of structure, members must be proportioned to
the UBC design forces. Similarly, an elastic avoid local buckling and joints must be
analysis of a mid-rise concrete shearwall stiffened to prevent stress concentrations that
building, for which the fundamental period could lead to the premature fracture of welds.
might be on the order of 0.3 to 05 seconds, Connections that might not have sufficient

30 CIVIL ENGINEERING PRACTICE SPRING 1987

 11 11=1
(Structure Ductility)

7h
h

0 req=1/h
(Member Ductility)

FIGURE 14. Lower member ductility requirements for a frame with plastic hinges in the
girders.

ductility must be strong enough to force yield- only if the code detailing provisions, and other
ing to occur in the main members. Members ductility-related provisions, are followed. If
subjected primarily to axial stress should the required level of ductility is not obtained,
ideally be redundant, so that the buckling of the "correct" seismic design loading would be
one or several members will not cause the significantly higher, and it is very likely that
collapse of the entire structure. wind loading would no longer control.
Code provisions mandating special detail-
ing for elements of the seismic resisting system Conclusion
are thus generally intended to ensure that the There is a clear need to provide for ductility
required levels of ductility are present in the in seismic design and, in particular, in designs
structure. based on code earthquake forces. Ductility is
Proportioning of members can also influ- not an auxiliary requirement in seismic design,
ence the amount of available ductility. For it is a primary factor in determining the level of
example, a frame in which plastic hinges are loading for which a building should be designed. If
allowed to form in the columns will require codes did not require structures to be ductile,
much higher member ductilities to achieve it would be necessary to design for forces that
the same overall ductility than a similar frame could easily be 4 or 5 times larger than those
in which plastic hinges are forced to occur in presently specified.
the girders (see Figure 14).16,17 In principle, of course, it should always
The designer must realize that ductility be possible to design structures to respond
requirements apply in a seismic zone even if elastically to an earthquake, and therefore not
wind forces control a particular design. The require ductility. In seismic engineering
designer determines whether wind forces literature, it has been traditionally assumed
control by comparing the wind shears and that designing a structure to remain elastic
moments induced in the building with those would never be economical because the
derived from code formulas for seismic load. required design force levels would be so high.
However, those code seismic forces assume a This assumption may not necessarily always
certain level of ductility that will be attained be true. In areas where seismic forces are

CIVIL ENGINEERING PRACTICE SPRING 1987 31

expected to be relatively low, reducing tute Monograph Series, no date.
ductility requirements with a corresponding 3. Uniform Building Code, Whitier, California: International
increase in design load may pose an econ- Conference of Building Officials, 1985.
omically preferable option.19 However, even in 4. Massachusetts State Building Code, Boston: Secretary of
such circumstances, there is a strong argument State, Commonwealth of Massachusetts, 1986.
for maintaining some degree of ductility in 5. Mahin, S.A, and Bertero, V.V., "Problems in Establishing
structures in seismic areas as a buffer against and Predicting Ductility in -a Seismic Design," Proceedings
the uncertainties inherent in seismic design. of the International Symposium on Earthquake Structural
Engineering, St. Louis, Missouri, August 1976.
Seismic loadings are probably the most
highly uncertain loadings known to structural 6. Clough, R.W., and Penzien, J., Dynamics of Structures,
New York: McGraw-Hill, 1975.
engineering. It is impossible to predict the
maximum likely level of seismic excitation of 7. Biggs, J.M., Introduction to Structural Dynamics, New York:
McGraw-Hill, 1975.
a structure with any certainty. Ductile struc-
tures are more forgiving than non-ductile 8. Housner, G.W., and Jennings, P.C., Earthquake Design
Criteria, Berkeley: Earthquake Engineering Research
structures of mistakes in loading assumptions Institute Monograph Series, no date.
or in elastic analysis. Ductile structures can
9. Newmark, N.M. and Rosenblueth, E., Fundamentals of
redistribute forces during local overloads Earthquake Engineering, Englewood Cliffs, New Jersey:
where a non-ductile structure might collapse. Prentice-Hall, 1971.
For this reason, given the uncertainty of 10. Bertero, V.V., Class notes for course CE 227 Design for
predicting seismic loads, some level of ductility Dynamic Loads, University of California, Berkeley, 1979.
should be provided in all buildings in seismic 11. American Institute of Steel Construction, Manual of
zones, no matter what level of force is used in Steel Construction.
design. 12. American Concrete Institute, ACI 318: Building Code
Requirements for Reinforced Concrete.
ACKNOWLEDGEMENTS - The author thanks Guy 13. Bertero, V.V., "Strength and Deformation Capacities of
Nordenson, Kim Roddis and Alan Simon for their Buildings Under Extreme Environments," in Pister, KS.,
encouragement and their valuable criticism of the ed., Structural Engineering and Structural Mechanics, Engle-
wood Cliffs, New Jersey: Prentice-Hall, 1980.
early drafts of this article; to Weidlinger Associates
for encouragement and technical support; and to 14. Biggs, J.M., Hansen, R.J., and Holley, M.J., Jr., "On
Methods of Structural Analysis and Design for Earth-
Frank Nigro for his expert CAD-drafting of the quakes," in Hall, W.J., ed., Structural and Geotechnical
figures. Mechanics, Englewood Cliffs, New Jersey: Prentice-Hall,
1977.
=.,,...,,.,___,.,...,,,,,,.,, DAVID 0. KNUTTUNEN received his 15. Popov, E.P., "Seismic Behavior of Structural Sub-
B.S. in Civil Engineering from the assemblages," Journal of the Structural Division, American
Society of Civil Engineers, Vol. 106, No. ST7, July 1980.
Massachusetts Institute of Technology
in 1978, and his Master's Degree from 16. Park, R., "Ductility of Reinforced Concrete Frames
under Seismic Loading," New Zealand Engineering, Vol. 23,
the University of California, Berkeley, No.11, November 1968, pp. 429-435 (cited in Reference 10).
in 1979. He has worked as a structural engineer in 17. Paulay, T., "Deterministic Seis.mic Design Procedures
the Boston area since 1980 with Souza, True and for Reinforced Concrete Buildings," Engineering Structures,
Partners and, more recently, with Weidlinger Vol. 5, No. 1, January 1983.
Associates. Currently, he is taking some time off to 18. Paulay, T., "A Critique of the Special Provisions for
work on some computer software development Seismic Design of the Building Code Requirements for
projects. Reinforced Concrete (ACI 318-83)," Journal of the American
Concrete Institute, Vol. 83, March/April 1986.
19. Gergely, P., "Earthquake Resistant Design of Concrete
Structures in the Eastern United States," paper delivered
REFERENCES at the ASCE Met Section lecture series, New York, copy
received courtesy of G. Nordenson, Weidlinger Associates,
1. Chopra, AK, Dynamics of Structures: A Primer, Berkeley: 1986.
Earthquake Engineering Research Institute Monograph
Series, no date. 20. Veletsos, AS., "Earthquakes and Their Effects - An
Overview," A Northeast Regional Seminar on Earthquake
2. Newmark, N.M., and Hall, W.J., Earthquake Spectra and Fundamentals, Earthquake Engineering Research Institute
Design, Berkeley: Earthquake Engineering Research Insti- Seminar held in Boston, October 1985.

32 CIVIL ENGINEERING PRACTICE SPRING 1987