Professional Documents
Culture Documents
22 (3) 0155
22 (3) 0155
22 (3) 0155
R. Park*
ABSTRACT
Definitions for the required and available ductility used in seismic design are
discussed. Methods for estimating the yield deformation and the maximum available
deformation are described and suggestions are made for appropriate definitions. Examples
are given of different imposed histories of inelastic displacement which have been used in
the experimental testing of structures and structural assemblages in which cycles of
quasi-static loading are applied. A quasi-static procedure for establishing the available
ductility factor of a subassemblage by laboratory testing is recommended.
BULLETIN OF THE NEW ZEALAND NATIONAL SOCIETY FOR EARTHQUAKE ENGINEERING, Vol. 22, No.3, September 1989
156
7~
H,,.t1
l Deflected
shape ::c
Ideal elasto-plastic
behaviour
Plastic -c;j
hinge g
..J
/
Displacement,
t
ductility factor will be limited by the stiffness, and more complex stiffness
available (ultimate) curvature ductility degrading idealizations which closely
factor. The relationship between the follow the actual loop shapes. Several
displacement ductility factor of a investigators have studied the influence of
reinforced concrete structure and the the shape of the hysteresis loops on the
curvature ductility factors at the plastic response of structures to severe
hinges can be determined considering the earthquakes.
geometry of the deformation of the
structure, providing that the equivalent Of particular interest is the effect on the
plastic hinge length, over which the response of significant stiffness
ultimate curvature can be considered degradation when the structure is cycled in
constant [1,2], is known. In recent the inelastic range. On average, the
reinforced concrete column tests [2] the difference between the ductility demands
equivalent plastic hinge length, taking for elasto-perfectly plastic single-degree-
into account the spread of plasticity due of-freedom systems and stiffness degrading
to bond deterioration and diagonal tension systems with the same initial strength,
cracking, was found experimentally to be on initial stiffness and viscous damping, when
average close to i = 0.5h, where his the responding to severe earthquakes found by
column depth. T~ plastic hinge rotation Mahin and Bertero [8] ahd Moss, et al [9]
is given by BP= (¢u - ¢y)2p. were small, except perhaps for short period
structures where the ductility demand of
It is evident that there can be significant the degrading stiffness system may be
numerical differences between the larger. Degrading stiffness systems were
magnitudes of the required displacement, found to dissipate hysteretically about the
rotation and curvature ductility factors. same amount of energy as elasto-perfectly
This is because once yielding has commenced plastic systems, even though they do not
in a structure the deformations concentrate reach their full strength as often [8].
in the yielding regions. For example, for This is because energy is dissipated
reinforced concrete moment resisting frames hysteretically by the elasto-perfectly
the required¢ /¢ at the plastic hinges plastic system only when the full strength
may be severalma{i~s the required 6 /~ is reached, but for the stiffness degrading
for the structure [l]. max Y system (for example, shown as real
behaviour in Fig.l) energy is dissipated
3. EFFECT OF HYSTERESIS LOOP SHAPE due to non-linear behaviour in almost all
ON RESPONSE cycles after first yield.
Fig.l illustrates that the load-deformation However Mahin and Bertero [8] have found
behaviour of real members can vary that bilinear hysteretic loops with even a
significantly from ideal elasto-perfectly small negative post-yield slope (-5%) can
plastic behaviour. Fig.2 shows a range of substantially increase the ductility
typical measured experimental lateral load- demand, particularly for short period
displacement loops for subassemblages of structures and long duration earthquakes.
structural concrete, masonry, steel and It should be noted though that the bilinear
timber. A number of shapes of hysteresis model is not typical of real behaviour of
loops have been used to model the cyclic structural members. Stiffness degrading
load-deformation behaviour of structures of models are more typical and the reduction
different materials to be utilized in of strength seen in hysteresis loops
inelastic time-history dynamic analyses, generally occurs as an overall reduction in
such as bilinear with variable post-yield strength (as in Fiq.2) or only at the end
157
LATERAL
DISPLACEMENT
~~s!?eetp~1t-y~~ldh!~:d-~:~~~m~~l~~ ~I:~~~:
It is co:1cluded that the important property
required in seismi.c design is ''adequate
perfectly plastic hysteresis loops that ductility", which means the ability of the
strength degradation to 80% of the initial structure to undergo large amplitude cyclic
stren~th dur~ng se~ere seismic excitacior1 deformations to the required maximum
in the inelastic range did not displacement in the inelastic range without
significantly influence the displacement a substantial reduction in strength.
response.
4, DEFINI'l"IONS OF /.\VAIL.ABLE DUCTILITJ
Pre.stressed concrete members have
significantly narrower moment-curvature The ductility required ot a structure
hysteresis loops (see Fig.2c), and hence during response to a severe earthquake
very much lo;•ler hysteretic energy needs to be matched by the available
dissipation, than reinforced concrete or ductility of tl1e structure. Definitions
structural steel menbers. 'l'he ma2Uffcm: which can be used to estimate the available
displacements reached by code-designed ductility factor are considered below.
prestressed concrete single-degree-of-
freedom systems has been found to be on 4.1 Definition of the Yield Deformation
average approximately 30% greater than
reinforced concrete systems of similar When calculating ductility factors the
initial strength, initial stiffness and definition of the yield deformation
viscous ::lamping, when responding to severe (displacement, rotation or curvature) often
earthquakes [10]. causes difficulty since the load-
deformation relation may not have a well
For reinforced concrete structures defined yield point~ This may o6cur, for
siqnificant inelastic deforrn.ations due to example, due to nonlinear behaviour of the
sh~ar or bond mechanisms lead to severe materials or due to yielding in different
dearadation of strength and. stiffness a.nd parts of a structure commencing at
to_, pinched hysteresis loops \-,ri th reduced different load levels9
enei:uv di.ss ioation" Pig,, 2a and 21:1 .shows
typi;~l m~asured experimental load- Various alternative definitions which have
displacement hysteretic behaviour of two been used by investigators to estimate the
reinforced concrete beam-column assemblies yield displacement are illustrated in
[31 one controlled bv ductile flexural Fig.3. These are:
pl~~tichinging in the.,beams (Fig.,2a) and
the. other controlled eventually by slip of Fig.3a The displacement when yielding
lonqitudinal beam bars through the joint f Lest occurs in the sys ten.
core due to bond deterioration (Fig.2b).
Kitayama, et al [11] have investigated the Fig,,Jb The yield displacement of the
inelastic dvnamic response to severe equivalent elasto-plast:ic system
earthquake motions of 4, 7 and 16 storey with the same elastic stiffness
moment resi.sting frames \4ith the plastic and ultimate load as the real
hinqe behaviour in the beams modelled by system,,
stiifness degrading hysteresis loops with
and without pinching behaviour ca.used by F.ig,,Jc The yield displacement of the
bond deterioration. The effect of equivalent elasto-plastic system
significant pinching of the hysteresis with the same energy absorption as
loops on the response was found to be the real system [BJ.
relatively small and it was concluded that
some bond deterioration of beam bars within Fig.:;d The yield displacement of the
a beam-column joint may be tolerable. equivalent elasto-plastic system
with reduced stiffness found as
similarly, Dean et al [6] found that the the secant stiffness at either
first yield or at 0.75 of the
:t:~~~~~:::1t wf~~and r~~:tf~;~~;dinr o:ti!~~:;; ultimate lateral load Hu,
dissipating capacity, such as plywood whichever is less [2]. The non-
sheathed timber shear w~lls (see Fig~2e), linear elastic behaviour before
to .severe earthquake motions is not first yield or 0.75 H is due to
significantly greater than that for cracking in the case a¥ reinforced
structures responding elasto-plastically. concrete~
It is evident that in the past there has The definition illustrated in Fig~3d is
been excessive emphasis on the desirability considered to be the most realistic sj_nce
of achieving Jn design structures which,, it applies generally to structures of
when subjected to cyclic deformations in concrete, masonry, steel and timber.
the. inelastic range dtie to severe
earth.quake 1oadingf displ2(/ load~ 4.2 Definition of the Maximum
deformation hystereis loops. It liv11,,i.J_able (Ultimate) Deformation
realized that some variation in hvstersis
1.oop shape will not have a major i;fluence The maxrnmm available (ultimate)
on the inelastic dynamic response cf deformation has also been estimated using
structures when subjected to severe various assumptions by i.nvesti.gators in the
earthquake excitation. That is, hysteresis past. Some possible definitions for the
loops showing some pi.nching or stiffness maximum available displacement are shown i.n
degradation will not lead to significantly Fig.4. These are :
larger :inelastic displacements, providing
that the structure has so1ne damping of Fig«4a The disolacement corresponding to
viscous type and is capable of some further a partlcular limiting value for
damping by hysteretic energy dissipation. tlle compressive strain. (For
159
DISPLACEMENT !iy
DISPLACEMENT
(a) Based on First Yield ( b) Based on Equivalent
Elasto-P..fastic Yield
Equal
areas
Limiting
compression strain
Hmax - - - ---:::.:--=----4r---=:.::::.:::::::-_
c:::i a
6
...J
6...J
Hmax -
I A small reduction
in load
- - - :;:::.:::=--------
of an element
T-- ---------
c:::i
6
...J
DISPLACEMENT DISPLACEMENT 6u
vibration, except in the very short period compared in terms of their application to
range. Yielding was found to occur about the design of structures for earthquake
the same number of times in each direction resistance.
but the maximum displacement was general!~
larger in one direction than in the other. 5.4 Examples of Quasi-Static Cyclic
For stiffness degrading single-degree-of- Loading in Terms of Displacement
freedom-systems, designed tor a seismic !)u_c::_t:i l i t_y
force level corresponding to a displacement
ductility factor of 4, responding to severe As an example of a quasi-static loading
earthquake ground motions recorded on firm history, the Commentary of the current
ground at moderate epicentral distances, (1984) SANZ Code for general design and
such as the 1940 El Centro earthquake, the loadings for buildings (15] recommends, as
number of yield reversals in each direction an approximate criterion for the adequate
did not generally exceed 4. For the ductility of moment resisting frames, that
unrealistic elasto-perfectly plastic the structure should be able to undergo 4
systems the number of yield excursions in loading cycles to a displacement ductility
each direction varied from about 15 for a factor of 4 in each direction without the
period of 0.2 seconds to about 3 for a horizontal load carrying capacity reducing
period of about 2 seconds. It is to be by more than 20%.
noted that the destructive earthquake which
occurred in Mexico City in September 1985 A quasi-static loading pattern which has
had most of its energy in relatively long been used for tests at the Construction
period ground motions and the duration of Technology Laboratories, Skokie, USA [16]
the strong earthquake shaking was and at the Public Works Research Institute
exceptionally long. Hence the number of Ministry of Construction, Japan [14], and
yield excursions for that Mexico City at several other organizations, is shown in
earthquake would be expected to be several Fig.5a. The displacement 6 has been taken
times that of the 1940 El Centro as the displacement cor~sponding first
earthquake. yield of the outer longitudinal reinforcing
bars. The ductility level is increased
Quasi-static load testing gives step-wise and the number of symmetrical
conservative estimates of the real strength loading cycles at each ductility level has
of the structure or structural assemblage, been n = 2 to 10, typically n = 2 in the
since real earthquake loads are dynamic and United States [16], and n = 10 in Japan
an increase in the strain rate results in [ 14] .
an increase in the strength of the
materials. However significant differences A quasi-static loading pattern which has
between the shapes of the hysteresis loops been used for many years at the University
obtained from quasi-static and dynamic of Canterbury [2] is shown in Fig.5b. The
loading tests may not be observed. For yield displacement is found using the mean
example Iwasaki, et al (14) concluded that measured secant stiffness at either 0.75 of
the effect of loading velocity on the the theoretical ultimate load or at first
energy dissipation of reinforced concrete yielding of steel, whichever is least, as
columns was not significant for illustrated in Fig.Jct. Again the
displacements of 4 times that at first ductility level is increased step-wise and
yield of the longitudinal reinforcement, normally two symmetrical loading cycles
but at higher displacements the energy have been applied at each level. Sometimes
dissipation capability was appreciably the ductility levels have been increased
larger when the loading velocity was 100 in steps of 2 cycles toµ= ±1, ±2, ±3
cm/sec than when the loading velocity was etc., if limited ductility is expected. In
10 cm/sec. New Zealand the commonly used strength
criterion is that the reduction in strength
In quasi-static load testing the should not exceed 20% of the initial
displacement history does not follow in strength.
detail the complex response of a structure
to an actual earthquake. Instead a more A more detailed quasi-static loading
simple displacement history is applied to history used for seismic load tests
enable an assessment to be made as to involving bi-directional earthquake loading
whether the structure is tough enough to be is that agreed to by the principal
likely to perform satisfactorily during a investigators of the United States-New
severe earthquake. Unfortunately, Zealand-Japan-China collaborative research
investigators in the past have used a range project on the seismic design of reinforced
of displacement histories, and various concrete beam-column joints [17). Again
definitions of yield and ultimate the yield displacement is determined from
deformation, which have made the comparison the secant stiffness measured at 0.75 of
of results of different investigations the theoretical ultimate load. The
difficult. As a result, values for displacement controlled loading history
ductility factor obtained from experimental imposed is illustrated in Fig.6 for the
tests have sometimes been misused in first 12 cycles. Obtaining that
judging the likely performance of international agreement was a major step
structures during severe earthquakes. forward and will permit proper comparison
or the performance of the structures tested
Agreement is needed for appropriate in the four countries.
definitions of the main parameters
describing inelastic behaviour for quasi- The above quasi-static loading histories
static load testing, so that performance used in New Zealand and in the US-NZ-Japan-
obtaine~ from analytical and experimental China collaborative research project are
investigations can be properly assessed and suitable for earthquakes of typical
162
8 ·--------
:::1.
[jj
:::,.;
6
lu
...J 4
::,...
'-
:::; 2 DISPLACEMENT
i::: CYCLES
tJ 0.75
5.._ 0.75
<: -2
~
~ -I.
':( Fig.5 EXAMPLES OF DISPLACEMENT HISTORIES
it -6 USED FOR QUASI-STATIC CYCLIC
~
LOADING TESTS OF COLUMNS AND
STRUCTURAL ASSEMBLAGES
-8 ----------
( bl Tests at the University of Canterbury
N N N
b c b c
s s s
£y_cles 1 and 2 J;y_cle 4 [;_y_cles 5 and 6
NS loading to 0.SHu EW loading NS loading to 2/J.y
<;y_cle 3 to liy
NS loading to liy
N N N
b c c_ _""'d
w E w a d E w E
h b g b h
g
g
s
t;'{.cles 7 and 8 ~y_cles 9 and 10 ~y_cles 11 and 12
Bidirectional NS direction to Bidirectional loading
loading to 2/J.y 4/J.y or 0,02drift to 4/J.y or 0.02 drift
direction, passing through the origin and cumulative structure ductility factor E11 at
the displacement points at first yield or which the lateral load sustained has
three quarters of the ideal strength, reduced to 80% of the maximum applied
whichever is less. These lines are lateral load strength measured during the
extended until they intersect the load test. The available structural ductility
corresponding to the ideal strength values. factor is then taken as l'a = Eµ/8. This
The mean of the displacements corresponding procedure assumes that four load cycles to
to these intersection points in the two µa (tha~ is, eight load excurs~ons) will
directions of loading is taken as the yield result in the same reduction in applied
displacement~ (see Fig.7). This in fact load as the history of displacement cycles
is the yield ~isplacement of an equivalent shown in Fig.8 applied to the appropriate
elasto-plastic system with reduced elastic 'f:.1,, where Eµ = a1,a.
stiffness to take into account the effect
of cracking and other non-linear elastic For example, if in the test the applied
effects. lateral load in load excursion "m" (see
Fig.8) is for the first time less than 80%
of the maximum measured lateral load during
Displacement controlled test the whole load excursion, the available
cycles cumulative ductility is E1, = 36. Then the
available structure ductility factor is
In the second part of the test the imposed
structure (displacement) ductility factor I' ''a = 36/8 = 4.5.
is increased step-wise in cycles, where As a further example, if in the test the
,, = ~ /~ and L\n is the maximum applied lateral load reduces for the first
displ~~?~mEt'nt imposiJ in the loading time to 80% of the maximum measured lateral
excursion. First, two cycles toµ= ±2 are load strength atµ +3 in load excursion
applied, followed by two cycles toµ= ±4 "m" (see Fig.8), the available cumulative
and two cycles to µ = ±6. Beyond this ductility factor is Eµ = 36 + 3 = 39. Then
level further cycles toµ= ±6 or ±8 may be the available structure ductility factor is
applied. The displacement history is shown ,, a = 3 9/ a 4 •9 •
in Fig.8. In the figure the cumulative
structure ductility factor Bµ undergone
during the cycles of displacements is shown Note that in this test the main concern is
at the peak of each loading run. For the level of ductility reached before
example, a subassemblage which has been significant reduction in applied load
subjected to two cycles to I' = ±2 followed occurs, rather than the energy dissipated
by one cycle toµ= ±4 would have undergone by the assemblage as given by the area
a cumulative structure ductility factor of within the hystersis loops. This is
E1, = 16. because, as discussed in Section 3, the
energy dissipated by the structure is a
The measured load-displacement record of less important factor in seismic design
the subassemblage is then assessed to than the displacement capacity.
determine the available structure
(displacement) ductility factor µ . 'rhis
is carried out by determiniig the
LATERAL
LOAD
Ideal Strength
L --,.---- 8
Stage 1-Load control to establish 6.y
Stage 2- ·
Displacement Control
___________ ----'---Lµ at
q peak of
-2
-I. ---- h j
16 21.
I n
-6 --------
r
-8 ~--------_36 ~8- p
7. CONCLUSIONS ACKNOWLEDGEMENTS
1. Ductility factors of structures, Informative discussions with
expressed as the maximum deformation colleagues, particularly Professor T.
divided by the corresponding deformation at Paulay of the University of Canterbury,
yield, are useful nondimensional indices of Professor M.J.N. Priestley of the
inelastic deformations. Ductility factors University of California at San Diego and
can be defined in terms of the required Dr R.C. Fenwick of the University of
ductility during severe earthquakes and the Auckland, are gratefully acknowledged.
available ductility, and can be expressed
in terms of displacements, rotations and
curvatures. REFERENCES
----- -------