Download as pdf or txt
Download as pdf or txt
You are on page 1of 23

Research paper

Earthquake Spectra
Post-earthquake assessment 2020, Vol. 36(1) 299–321
Ó The Author(s) 2020
of moderately damaged Article reuse guidelines:
sagepub.com/journals-permissions

reinforced concrete plastic DOI: 10.1177/8755293019878192


journals.sagepub.com/home/eqs

hinges

Kai Marder, M.EERI1, Kenneth J. Elwood, M.EERI1,


Christopher J. Motter, M.EERI2 and G. Charles Clifton1

Abstract
Modern reinforced concrete buildings are often designed to dissipate energy during
strong earthquakes by permitting the controlled formation of plastic hinges. Plastic
hinges require assessment of residual capacity in post-earthquake situations. However,
few past studies have investigated this topic, and results from experiments focused on
undamaged structures are not always transferable to post-earthquake situations. Data
from an experimental program, in which both cyclic and earthquake-type loadings were
applied to nominally identical reinforced concrete beams, are used to investigate the
relationship between residual crack widths and rotation demands. Assessment of the
peak deformation demands incurred during a damaging earthquake is critical for post-
earthquake assessments, but residual crack widths are shown to be dependent on sev-
eral factors in addition to the peak rotation demand. Non-dimensional metrics captur-
ing the distribution of cracking are proposed as a more informative alternative. The
reduction in stiffness that occurs as a result of earthquake-induced plastic hinging dam-
age was also investigated. A proposed model is shown to give a lower-bound estimate
of the residual stiffness following arbitrary earthquake-type loadings.

Keywords
post-earthquake, reinforced concrete, damage assessment
Date received: 28 October 2018; accepted: 1 July 2019

1
Department of Civil and Environmental Engineering, The University of Auckland, Auckland, New Zealand
2
Department of Civil and Environmental Engineering, Washington State University, Pullman, WA, USA

Corresponding author:
Kai Marder, M.EERI, Department of Civil and Environmental Engineering, The University of Auckland, Auckland, New
Zealand.
Email: kmar702@aucklanduni.ac.nz
300 Earthquake Spectra 36(1)

Introduction
The principles of ductile detailing and capacity design have long been incorporated into
the seismic provisions of many reinforced concrete design codes (e.g. American Concrete
Institute, 2014; European Committee for Standardization, 2004; Standards New Zealand,
2006). This strategy of developing controlled plastic hinging is effective in preventing
structural collapse, but does not guarantee the post-earthquake usability or reparability of
reinforced concrete buildings. Severe plastic hinging damage, evidenced by crushing of the
confined core or buckling of reinforcement, will typically necessitate full reinstatement of
the concrete component or post-earthquake demolition, but such damage is rare for mod-
ern construction except under very strong ground motions. A ‘‘moderate’’ level of plastic
hinging damage, characterized by flexural cracking, yielded longitudinal reinforcement,
and possible spalled cover concrete, has been more commonly observed in buildings
designed to modern codes following recent earthquakes in New Zealand (Henry et al.,
2017; Kam et al., 2011). Improved understanding of the effects of such damage is required
in order to make informed decisions in post-earthquake situations.
Two guidelines for the detailed post-earthquake evaluation of damaged reinforced con-
crete buildings have previously been published: Federal Emergency Management Agency
(FEMA) 306 (Applied Technology Council, 1998) in the United States and the ‘‘Guideline
for post-earthquake damage evaluation and rehabilitation’’ (Japan Building Disaster
Prevention Association (JBDPA), 2014) (described in English by Nakano et al. (2004)) in
Japan (herein referred to as the JBDPA Guideline). FEMA 306 is targeted at wall build-
ings, while the JBDPA Guideline covers both wall and moment frame structures. Both
documents cover buildings with flexural- or shear-dominant failure mechanisms and all
ranges of damage severity. This study focuses on moderate flexural damage due to the
impracticalities of attempting to quantify the residual capacity of structures exhibiting
non-ductile or severe flexural damage. Such buildings are likely to be demolished or sub-
jected to major retrofit actions, regardless of the assessed residual capacity.
When assessing the post-earthquake residual capacity of a damaged building, it is criti-
cal to understand the deformation mechanism of the building during the initial damaging
earthquake. There is little rationale in conducting a detailed post-earthquake assessment if
the structural behavior of the building in its undamaged state cannot be reasonably pre-
dicted. FEMA 306 recognizes this by calling for the expected lateral mechanism of the
building to be compared against the observed damage early in the assessment procedure.
Improved knowledge of the relationship between observable damage and the deformation
demand is required for such a comparison to be carried out. This research investigates the
extent to which damage observations (specifically residual crack widths) in beam plastic
hinges are indicative of the demands incurred during the damaging earthquake.
Post-earthquake assessment also requires determination of the capacity of the damaged
building. FEMA 306 and the JBDPA Guideline differ from ‘‘pre-earthquake’’ seismic
assessment guidelines and standards (American Society of Civil Engineers, 2017; JBDPA,
2001) in that the structural capacity is modified to account for the observable damage.
Both documents use ‘‘reduction factors’’ to alter the capacity of damaged components,
relative to what would be calculated for equivalent undamaged components. The JBDPA
Guideline uses a single component-level reduction factor, while FEMA 306 uses different
reduction factors for stiffness, strength, and deformation capacity, allowing compatibility
with non-linear static assessment procedures. A similar definition of component residual
capacity as that from FEMA 306 is used in this study, as shown in Figure 1. Given the
Marder et al. 301

Figure 1. Parameters used to define the residual capacity of reinforced concrete ductile plastic hinges.

focus on moderate plastic hinging damage, which is not associated with a loss of flexural
strength, the residual strength is assumed to be unreduced from that of an equivalent unda-
maged component. Previous research (e.g. Acun and Sucuoglu, 2010; El-Bahy et al., 1999;
Marder et al., 2018a; Nojavan, 2015) on flexure-controlled beams and columns, which cov-
ers a range of longitudinal reinforcement ratios, axial load ratios, and shear span ratios,
has also shown that variations in loading at or below 2% member drift (i.e. chord rotation)
are not typically associated with a reduction in deformation capacity. The research pre-
sented herein, therefore, focusses on methods for estimating the loss of stiffness due to
plastic hinging damage.
Both FEMA 306 and the JBDPA Guideline use the maximum residual crack width as
the primary quantifiable damage metric upon which reduction factors for damaged com-
ponents are based. In both documents, the reduction factors have been derived using
experimental data from reversed-cyclic tests with progressively increasing deformation
demands. Such tests necessarily exhibit progressively increasing damage states, which
makes it difficult to isolate the effects that a particular damage state has on the component
residual capacity. In addition, in reversed-cyclic laboratory experiments, damage observa-
tions are typically made at or near the peak deformation demand, but the observable dam-
age in a post-earthquake situation reflects the residual state of the building rather than the
peak demands incurred. Experiments are required that can better reflect the damage states
after earthquake loadings and isolate the effect of these damage states on plastic hinge
response. The analysis conducted as part of this study makes extensive use of such data
from a test program conducted by the authors (Marder et al., 2018b).
It is emphasized that this research is targeted at moderate plastic hinging damage, up to
and including cracking, longitudinal reinforcement yielding, and cover concrete spalling.
Severe damage, characterized by bar buckling or crushing of core concrete, is considered
beyond the scope of this study. This article also focuses on the performance of damaged
plastic hinges prior to repair, critical for the assessment of the performance of a concrete
building in aftershocks and for post-earthquake repair decisions (epoxy-injection repair of
such damage is considered in Marder et al. (in press)).
302 Earthquake Spectra 36(1)

Experimental program overview


Data from a set of reinforced concrete beams tested by the authors are used to investigate
the objectives of this study. The experimental program has been described in full and the
data made publicly available in the Earthquake Spectra data paper (Marder et al., 2018b).
This section provides a brief discussion on aspects of the experimental program relevant to
this article. The reader is referred to Marder et al. (2018b) for more information.
Seventeen nominally identical reinforced concrete beams were tested under various
loading protocols and boundary conditions. The beams were tested in a cantilever config-
uration and subjected to uniaxial loading. All specimens were nominally identical, with a
rectangular cross-section of 720 mm 3 320 mm, a shear span length of 2580 mm, and a
longitudinal reinforcement ratio of 0.6%, which was approximately 1.25 times the mini-
mum longitudinal reinforcement requirement in both American Concrete Institute (ACI)
318-14 and New Zealand Standards (NZS) 3101:2006. Beams were symmetrically rein-
forced and developed a reversing plastic hinge, as is common design practice in New
Zealand. They were detailed to meet the criteria for ductile plastic regions in NZS
3101:2006 (Standards New Zealand, 2006) and also meet all requirements for special
moment frame beams in ACI 318-14 (American Concrete Institute, 2014).

Table 1. Test matrix of selected specimens from Marder et al. (2018b)


Specimen name Earthquake peak Earthquake Loading protocol Elongation
drift (%) loading rate type (Figure 2) restraint

MONO N/A N/A N/A None


CYC N/A N/A CYC None
CYC-NOEQ N/A N/A CYC-NOEQ None
P-1 1.4 Dynamic P-1 None
P-2 2.2 Dynamic P-2 None
P-2-S 2.2 Static P-2 None
LD-1 1.4 Dynamic LD-1 None
LD-2 2.2 Dynamic LD-2 None
LD-2-S 2.2 Static LD-2 None
CYC-ER N/A N/A CYC Heavy (40 kN/mm)
LD-2-ER 2.2 Dynamic LD-2 Heavy (40 kN/mm)
CYC-LER N/A N/A CYC Light (13 kN/mm)
LD-2-LER 2.2 Dynamic LD-2 Light (13 kN/mm)

Figure 2. Loading protocols applied to the beam specimens. Earthquake loading shown in red, and
cyclic loading shown in black.
Marder et al. 303

Table 1 provides a test matrix for the 13 beam specimens considered in this study. Figure 2
shows the various loading protocols applied to the test specimens, which included cyclic,
monotonic, and earthquake-type displacement histories. The earthquake displacement his-
tories were derived using non-linear response history analysis on a model of a 10-story ductile
moment frame building subjected to either pulse-type (P) or long duration (LD) ground
motions. As identified in Table 1, both dynamic and quasi-static rates of loading were used in
the earthquake loadings. Various levels of axial spring stiffness, which partially restrained the
axial elongation of the beams, were also applied to selected specimens to simulate the resis-
tance to beam elongation that may be present in moment frame buildings (Bechtoula et al.,
2006; Matthews, 2004). The specimen name suffixes LER and ER correspond to lower
(;13 kN/mm) and higher (;40 kN/mm) levels of axial spring stiffness, respectively.
During the experimental program, both total and maximum residual crack widths were
measured, where total refers to the sum of all crack widths over 0.2 mm. The residual crack
widths were measured at a state of zero lateral force following each half-cycle of cyclic
loading, as well as following the initial earthquake loadings. For inclined cracks, crack
width was measured in the beam longitudinal direction rather than perpendicular to the
crack. In all specimens, residual crack width data collection continued until the point of
cover concrete spalling. Some specimens exhibited spalling during the initial earthquake
loading; collection of crack data was not possible in those cases.

Residual crack widths and deformation demands


This section investigates the degree to which residual crack width metrics are representa-
tive of the deformation demands incurred during a damaging earthquake. The demand
parameter of greatest interest is the peak inter-story drift, a commonly used deformation
metric in the seismic design and assessment of ductile buildings. Peak inter-story drift is
also used as a key engineering demand parameter in fragility curves for performance-based
earthquake engineering (Federal Emergency Management Agency, 2012) and thus is
assumed to strongly correlate with component damage. This section makes use of data
from the previously described cantilever beam tests. The chord rotation of the beams is
taken as a proxy metric for inter-story drift in a frame with strong columns and weak
beams; this is reasonable given the lack of column or panel zone cracking observed in
buildings in Christchurch (Kam et al., 2011). It is emphasized that this analysis is focused
on flexural cracks, representative of plastic hinging behavior. The reported drift from the
tests was based on the total measured lateral deformation in the beams. (Base rotation
and translation were measured and found to be negligible in all specimens.)
The total residual crack width of a plastic hinge is dependent on the residual elongation
and rotation of the hinge. (For this study, a ‘‘plastic hinge’’ is defined as the physical region
where non-hairline cracks were observed.) Figure 3 illustrates the theoretical relationship
between total residual crack width and elongation for four residual deformation states fol-
lowing an earthquake. When no residual rotation is present, the two sides of the beam have
identical total residual crack widths equaling the elongation of the hinge at the centroid of
the cross-section. When residual rotation is present, one side has a larger total residual
crack width to account for the residual longitudinal strain gradient across the cross-section,
and the total residual crack width on the tension side is greater than the elongation of the
hinge centroid. The ratio of total residual crack width on the tension side to elongation
reduces if the cracks are unable to fully close in compression.
304 Earthquake Spectra 36(1)

Figure 3. Theoretical total residual crack widths for four possible residual deformation states of a
plastic hinge: (a) cracks fully closed in compression, no residual drift; (b) cracks do not fully close in
compression, no residual drift; (c) cracks fully closed in compression, residual drift; and (d) cracks do not
fully close in compression, residual drift.

The theoretical relationships shown in Figure 3 are validated by data from the experi-
mental program. Figure 4 shows the total and maximum residual crack widths and the
drift versus elongation responses for six beam specimens. The relationship between total
residual crack width and elongation is readily apparent. In specimens CYC-ER and
Marder et al. 305

Figure 4. Residual crack widths versus elongation for six beam specimens: (a) CYC, (b) CYC-LER,
(c) MONO, (d) CYC-ER, (e) LD-1, and (f) P-2.
306 Earthquake Spectra 36(1)

MONO, the total crack widths on the tension side grew to exceed the elongation by fac-
tors of 1.5–2 due to the compression side having limited longitudinal strain (the case illu-
strated by Figure 3c). Comparatively, in specimens CYC-LER and CYC, the ratio of total
crack width to elongation was less than 1.5 due to residual elongation in the compression
steel (the case illustrated by Figure 3d). Specimens LD-1 and P-2 had similar residual
crack widths as a result of similar residual deformations, despite having been subjected to
different peak drift demands (1.4% and 2.2%, respectively).
Data in Figure 4e to f demonstrate that the total residual crack width can, at best, only
inform estimates of the peak demands placed on a plastic hinge to the same degree that
the residual deformations can help in estimation of peak demands. In plastic hinges that
are free to elongate, the magnitude of residual elongation is dependent on both the ampli-
tude between the peak positive and negative drifts and the number of loading cycles (Lee
and Watanabe, 2003), and is not an accurate metric for estimating the peak drift demands
previously incurred. It is important to note that this is a different result than would be
obtained if only analyzing data from reversed-cyclic or monotonic tests (as is common in
most simulated seismic loading studies), where an approximately linear relationship
between total residual crack width and peak drift demand exists (Figure 4a to d).
In cases where some restraint to elongation is present (Figure 4b and d), the usefulness
of the total residual crack width to provide information about the prior demands on the
hinge is further complicated. Reductions in residual elongation (and therefore total resi-
dual crack width) of over 50% were observed in the restrained specimens, relative to the
equivalent unrestrained specimen shown in Figure 4a. The elastic restraint system
employed in the experimental program is not an accurate representation of the restraint to
beam elongation in moment frame structures, which is highly non-linear. Nonetheless, the
results are relevant, as tests on indeterminate frames (e.g. Bechtoula et al., 2006;
Matthews, 2004) have shown that differences in residual elongation on the order of 50%
can occur in beams subjected to identical drift histories, depending on the location of the
beam within the structure. Both floor systems and columns can contribute to the restraint
of beam elongation in ductile reinforced concrete moment frames. The beams considered
in this study are symmetrically reinforced without a monolithic floor slab. Fenwick and
Megget (1993) discuss how the presence of non-symmetrical reinforcement and monolithic
slabs influence the progression of axial elongation. While further experimental study may
be warranted, the finding that crack widths are a function of residual deformations, not
peak demands, is expected to be applicable in cases with a monolithic floor slab or non-
symmetric reinforcement.
The preceding analysis focused primarily on total residual crack width. However, the
data presented in Figure 4 also demonstrate that the maximum residual crack width tends
to increase with elongation, but with a more unpredictable, non-linear progression than
the total residual crack width. Figure 5 shows peak drift demand versus maximum residual
crack width data for all beams where such data were available (11 specimens). The non-
linearity of the relationship results in an increasing range of possible maximum residual
crack widths for a given drift demand as the peak drift demand increases. These data indi-
cate that maximum residual crack width is also a poor metric for the purposes of informing
estimates of peak drift demand. The above conclusions suggest that the focus on maximum
crack widths as the primary damage metric in FEMA 306 and the JBDPA Guideline may
be inappropriate.
Marder et al. 307

Figure 5. Peak drift demand versus maximum residual crack width.

Another possible use of residual crack data is to ignore the specific crack widths and
instead use non-dimensional metrics that capture the distribution of cracking. Figure 6 shows
the peak drift demand versus (a) the ratio of maximum to total residual crack width, and (b)
the number of residual cracks wider than 0.2 mm (herein referred to as ‘‘non-hairline
cracks’’), for the same 11 beam specimens. The linear trends in Figure 6 are evident, showing
an increase in the distribution of non-hairline cracks with increasing peak drift demands.
This behavior is expected, as sufficient rotations must be applied to a hinge to induce strain
hardening in the longitudinal reinforcement and allow spreading of the yield zone to occur.
Many plastic hinge zones were observed to have only a single wide crack following the
2010–2011 Canterbury earthquakes in New Zealand (e.g. as shown in Figure 7). Concerns
were raised about these crack patterns being indicative of a shorter than expected plastic
hinge length and a concentration of strain in the longitudinal reinforcement crossing the
crack (Morris et al., 2014). The data in Figure 6 demonstrate that a plastic hinge exhibit-
ing a single wide crack may simply not have experienced large enough rotations to force
additional cracks to develop. It is here recommended that the observation of a single non-
hairline residual crack, or similar limited cracking, only be treated as evidence of a high
localized reinforcement strain in situations where such behavior would otherwise be
expected (e.g. when the cracking moment is higher than the yield moment, or due to cur-
tailing of reinforcement, as discussed in Opabola et al. (2018)). Low reinforcement ratios
were identified as an important parameter in causing limited distribution of cracking in
reinforced concrete walls during the Canterbury earthquakes (Lu et al., 2016; Sritharan et
al., 2014). The beam specimens tested in this experimental program had a longitudinal
reinforcement ratio of 0.6%, approximately 1.25 times the minimum requirement in both
ACI 318-14 and NZS 3101:2006. A drift demand of 0.8% was sufficient to induce a sec-
ondary non-hairline residual crack in all cases (Figure 6b). These results demonstrate that
an increase in distributed cracking with drift demand occurs even in lightly reinforced
beams.
308 Earthquake Spectra 36(1)

Figure 6. Peak drift demand versus residual crack width metrics: (a) ratio of maximum to total crack
width, and (b) number of non-hairline cracks.

While Figure 6 shows a linear trend between the distribution of cracking metrics and
peak drift demand, the high degree of dispersion in the data, represented by the 5th and
95th percentile prediction intervals, means that these metrics cannot be used to obtain a
reliable estimate of the peak deformation demand. The 95th percentile line from Figure 6b
may be conservatively used to estimate peak drift demand for the beam specimens dis-
cussed here, but these results are not necessarily representative of beams in general. The
Marder et al. 309

following qualitative observations based on the distribution of cracking may be more


appropriate for general use in post-earthquake situations:

 A lack of non-hairline residual cracks is an indicator that the longitudinal reinforce-


ment has not yielded.
 A single non-hairline crack is evidence of modest inelastic rotation demands, not
sufficient to induce strain hardening in the longitudinal reinforcement. This does
not apply when anchorage of longitudinal reinforcement is inadequate or when the
cracking moment exceeds the yielding moment.
 Multiple non-hairline cracks distributed along a plastic hinge is evidence of more
significant inelastic rotation demands, sufficient to induce strain hardening in the
longitudinal reinforcement. Figure 6 can be used to estimate the range of likely peak
drift demands experienced.
 Further confidence in peak drift demand estimates can be achieved through analysis
of the building system, using a representative recorded ground motion, if the lateral
mechanism identified in the model adequately captures the distribution of damage
(e.g. location of beam hinging) throughout the building. This peak drift demand
from the model should be compared with the range of estimates from Figure 6 based
on the observed component damage.

It is noted that these findings do not apply for columns subjected to axial compression,
where a lack of non-hairline residual cracks is not necessarily indicative of a lack of yield-
ing. Cracks may close entirely if the hinge returns to a state of zero rotation, regardless of
the peak rotation demands. It may be possible to infer the demands incurred in the col-
umns of moment frames indirectly, based on the estimated demands in the beams and an
assumed lateral mechanism of the frame. Further system-level research investigating this
potential relationship is required.

Assessment of plastic hinge residual stiffness


A number of factors contribute to the degradation of stiffness that occurs in moderately
damaged reinforced concrete plastic hinges, including crack closure, bond degradation,
concrete degradation, inelastic shear deformations, and the Bauschinger effect.
Quantifying the effects of each of these factors on the residual stiffness is not feasible for
the purposes of a post-earthquake assessment. Distinction between unloading and reload-
ing stiffness is also impractical. Instead, a single degraded stiffness parameter that is pre-
sented as a fraction of the initial secant stiffness to yield, such as the stiffness reduction
factors used in FEMA 306, is desirable. This section investigates simple methods for
approximating the residual stiffness of moderately damaged plastic hinges.
The definition of secant stiffness used in this study is illustrated in Figure 8. The stiffness
is taken as the secant line between the point at which the loading commences and the first
point at which a force threshold (shown as Mn in Figure 8) is reached in either the positive
or negative loading direction. Cycles that occur below the force threshold are considered to
have no effect on the secant stiffness. This definition applies for both the initial and reload-
ing secant stiffness, with the difference being whether the point at which the loading com-
mences is the origin (in the case of initial secant stiffness) or the residual deformation (in
the case of reloading stiffness).
310 Earthquake Spectra 36(1)

Figure 7. Single wide residual crack in a beam plastic hinge zone following the 2011 Christchurch
earthquake.

One method of assessing residual stiffness is as a function of the observable damage.


Figure 9 compares total and maximum residual crack width data against reloading stiff-
ness data for the beam specimens of the experimental program (listed in Table 1). The
reloading stiffness is taken as the secant line between the residual deformation (the defor-
mation at which the crack widths were measured) and the first instance of reaching the
nominal moment strength Mn, calculated using the provisions of NZS 3101:2006. The
reloading stiffness is expressed in terms of both kN/mm and using the appropriate ratiopofffiffiffiffiffi
EcIg with the theoretical flexural stiffness 3EI/L3, where Ec was calculated as 4700 f 0c
(MPa units). Data corresponding to the first half-cycle following the initial earthquake
loadings are shown in red and data corresponding to cyclic loading are shown in black.
The residual crack width metrics, and particularly the total crack width (Figure 9a), are
inversely correlated with the reloading stiffness during cyclic loading, regardless of the
prior loading protocol or the level of axial restraint. However, the inverse relationship
does not hold for the reloading stiffness immediately after initial earthquake loadings,
where higher stiffness values for a given crack width were observed. This implies that a
model calibrated on total residual crack width data from cyclic tests may be able to pro-
vide a lower-bound residual stiffness following arbitrary earthquake-type loadings.
Marder et al. 311

Figure 8. Definition of secant stiffness (red line) used in this study, shown here using data from the
initial earthquake loading of one example beam specimen (LD-2).

Figure 10 presents a least-squares power regression model relating total residual crack
width to the reloading stiffness of the beam specimens. The model was derived using all
total residual crack width data collected during cyclic loading (i.e. all black data points
from Figure 9a). The total residual crack width is presented as a fraction of the beam depth
and the reloading stiffness is presented using the appropriate fraction of EcIg in order to
give a non-dimensional relationship. The model has a coefficient of determination of 0.93.
A simplified version of the model, with less implied precision, is also overlaid on the data.
Despite the strong correlation between total residual crack width and stiffness degrada-
tion, residual crack width metrics have limitations for informing post-earthquake assess-
ments of residual stiffness. A lack of non-hairline residual cracks does not necessarily
indicate that a component was not pushed into the inelastic range. Cracks may close in
components subjected to axial compression. In these situations, using the residual crack
widths alone as a metric for predicting the residual stiffness would be inappropriate, as
zero residual crack width would imply no reduction in the secant stiffness to yield. Such
an assessment would be clearly inaccurate, as factors such as bond degradation, concrete
degradation, and the Bauschinger effect can affect the residual stiffness regardless of the
residual crack width. Furthermore, spalling of cover concrete can preclude the measure-
ment of cracks, making residual crack widths an unusable metric in certain situations.
Finally, Figure 10 shows that a large proportion of the stiffness degradation occurred due
to relatively small total residual crack widths, on the order of 0.5% of the beam depth.
Identification of such minor cracking in all plastic hinges of a moment frame building
312 Earthquake Spectra 36(1)

Figure 9. Reloading stiffness versus (a) total residual crack width and (b) maximum residual crack width.
Marder et al. 313

Figure 10. Least-squares regression power model for total residual crack width versus reloading
stiffness.

may be impractical in post-earthquake situations. Distinguishing earthquake-induced


cracking from cracks due to other causes (e.g. shrinkage) may also present problems.
An alternate method of assessing residual stiffness is to use a function of the estimated
deformation demands during the damaging earthquake. Use of deformation demand para-
meters to define the residual stiffness is a commonly employed method in non-linear hys-
teretic models. Figure 11 illustrates the peak-oriented reloading stiffness model, which has
found widespread use in modeling of reinforced concrete plastic hinges (e.g. Clough and
Johnston, 1966; Takeda et al., 1970).
Figure 12a shows that the reloading stiffness response of the beam specimens was
strongly correlated with the difference between the residual drift and the prior peak drift
in the reloading direction (i.e. the deformation parameters that define a peak-oriented
reloading stiffness model). These results support the use of peak-oriented stiffness models;
however, obtaining sufficient knowledge of the deformation history in order to calculate a
peak-oriented stiffness is not feasible in post-earthquake situations. Figure 12b shows that
when the maximum prior drift demand is used as a deformation metric, the reloading stiff-
ness degradation of the beam specimens during cyclic loading followed a consistent pat-
tern, but this pattern did not hold for the first half-cycle immediately following the initial
earthquake loadings (similar to the finding for the residual crack width metrics). Data
points corresponding to the reloading stiffness measured immediately after earthquake-
type loadings have notably higher stiffness values for a given peak drift demand. Again,
314 Earthquake Spectra 36(1)

Figure 11. Peak-oriented reloading stiffness model for an elastic-perfectly plastic element.

this implies that while cyclic stiffness data may not be representative of residual stiffness
following arbitrary earthquake loadings, development of a lower-bound stiffness model by
calibrating to cyclic test data may be appropriate.
The data presented in Figure 12b came from nominally identical beam specimens, and
therefore similar results would be obtained using any lateral deformation metric. Marder
(2018) identified that for components with varying yield rotations, displacement ductility
provided a better correlation with reloading secant stiffness than did peak drift demand.
Di Ludovico et al. (2013) previously used a database of 23 reversed-cyclic column tests to
derive the expression Kr/Ky = 1 – (1.07 – 1.15 m–0.92), where Kr is the peak-to-peak resi-
dual stiffness, Ky is the initial secant stiffness to yield, and m is the displacement ductility.
The empirical model proposed by Di Ludovico et al. (2013) is very similar to the expres-
sion Kr/Ky = 1/m, which can be derived from the peak-to-peak stiffness of an elastic-
perfectly plastic system subjected to symmetric reversed-cyclic loading, as demonstrated in
Figure 13. Kr/Ky = 1/m is preferred here due to its simplicity and rational derivation.
In order to investigate the applicability of the Kr/Ky = 1/m model as a lower-bound
stiffness following arbitrary earthquake loadings, additional relevant experimental data
were required for validation. The following information was deemed necessary in each
experiment: (1) an initial arbitrary earthquake-type loading, (2) a subsequent loading to
allow assessment of residual stiffness, and (3) a limited residual drift (\1.0%) after the
Marder et al. 315

Figure 12. Reloading stiffness versus (a) the deformation demand metrics used to define a peak-
oriented model, and (b) peak drift demand.
316 Earthquake Spectra 36(1)

Figure 13. Relationship between origin-to-peak stiffness degradation and ductility demand.

initial loading. The limit on residual drift is to prevent the direction of reloading from
unduly affecting the measured residual stiffness. As shown in Figure 11, increased residual
drifts can result in increased differences between ‘‘unloading’’ and ‘‘reloading’’ stiffness
values in plastic hinges exhibiting a peak-oriented reloading stiffness. Reloading due to
earthquake shaking can occur in either direction (i.e. either the ‘‘unloading’’ or ‘‘reloading’’
stiffness could occur), and the measured residual stiffness is therefore sensitive to the resi-
dual drift. The intent of the Kr/Ky = 1/m model is to give an approximate value of residual
stiffness that neglects the distinction between unloading and reloading behavior. Data used
for validation of this model, therefore, require low residual drifts prior to measurement of
residual stiffness.
Shake table tests that involve the application of multiple earthquake ‘‘runs’’ to a test
specimen satisfy each of these three criteria. A dataset of relevant shake table tests on duc-
tile reinforced concrete columns was compiled, consisting of four test programs (Arias
Acosta, 2011; Hachem et al., 2003; Laplace et al., 1999; Schoettler et al., 2015), eight speci-
mens, and 34 runs. If any specimen exhibited more than moderate damage (e.g. crushing
of core concrete or buckling of longitudinal reinforcement), all future runs on that speci-
men were omitted from the analysis. All specimens were cantilever columns with a circular
or oval cross-section and spiral transverse reinforcement, typical of bridge column con-
struction in seismically active areas of the United States.
Figure 14 shows the normalized stiffness versus the absolute maximum ductility demand
reached in any preceding run for all columns in the dataset. (Estimation of the normalized
stiffness based on plotted data was required in cases where numerical data were not avail-
able.) The stiffness data immediately following earthquake loadings for the beam
Marder et al. 317

Figure 14. Proposed stiffness degradation relationship versus experimental stiffness values normalized
against (a) an empirical initial secant stiffness and yield rotation, and (b) an initial secant stiffness as a
function of EcIg and a theoretical yield rotation.

specimens tested by the authors (previously shown as the red data points in Figure 12) are
also included in Figure 14. The stiffness was defined as the secant line between the residual
drift and the first instance of reaching 80% of the maximum base moment. In Figure 14a,
the residual stiffness values were normalized against the initial secant stiffness to 80% of
the maximum base moment, which is used here as the measured secant stiffness to yield
(similar to the approach used by Paulay and Priestley (1992) and Elwood and Eberhard
(2009)). The displacement ductility of the columns was calculated using a yield rotation
318 Earthquake Spectra 36(1)

that was approximated as the rotation corresponding to the first instance of reaching 80%
of the maximum base moment. (The yield rotation of the beam specimens was visually esti-
mated, as sharp changes in the load-deformation response occurred at the first yield of
extreme tension steel.)
In practice, a measured initial secant stiffness to yield is not available to normalize
against. In Figure 14b, the residual stiffness values are normalized against an initial secant
stiffness calculated as a ratio of the gross flexural stiffness, as per the provisions of
American Society of Civil Engineers (ASCE) 41-17 (note that 0.3EcIg was used in all cases,
as all the test specimens had initial axial load ratios of approximately 0.1 or below). A
measured yield rotation is also not available for calculating the displacement ductility in
practical situations; a theoretical formulation must instead be used. The displacement duc-
tility shown in Figure 14b was determined using a yield rotation calculated from Equation
1. (It is noted that this formulation significantly under-predicted the yield rotation of the
beam specimens.)

My L
uy = ð1Þ
3Ec Ieff

where uy is the yield rotation, My is the yield moment (taken as the nominal moment Mn,
calculated as per NZS 3101:2006 using lower characteristic strengths),
pffiffiffiffiffi L is the shear span
length, Ec is the Young’s Modulus of the concrete (taken as 4700 f 0c ), and Ieff is the effec-
tive moment of inertia, taken as a fraction of the gross section moment of inertia based on
the recommendations of ASCE 41-17.
Figure 14 shows that the proposed Kr/Ky = 1/m equation gives a lower-bound predic-
tion of the residual stiffness of the test specimens following arbitrary earthquake-type load-
ings. The exception to this is at low displacement ductility demands (less than 2.0), likely
due to the variability involved in quantifying the yield deformation of reinforced concrete
columns. Furthermore, the model is sensitive to estimated ductility demand at low values
of ductility (i.e. stiffness reduction varies by a factor of 2 between ductility of 1 and 2).
Given the uncertainty involved in estimating the displacement ductility in post-earthquake
situations, the stepwise function shown in Figure 14 may be considered a more appropriate
option for estimating a lower-bound residual stiffness in this low ductility region. This
approach suggests that any component which is estimated to have been subjected to a dis-
placement ductility greater than 1.0 should be considered to have a residual stiffness that
does not exceed 50% of its initial secant stiffness to yield. Thus, a lower-bound estimate of
residual stiffness as a function of displacement ductility is given by Equation 2.
8
Kr <
1:0, m\1:0
= 0:5, 1:0<m<2:0 ð2Þ
Ky :
1=m, m ø 2:0

It is important to note that the data in Figure 14 do not account for the full
variability that exists in post-earthquake situations. While variations in the prior loading
history and the theoretical versus empirical initial stiffness and yield rotation are
accounted for, the variation in the estimated versus actual peak drift demand are not.
Estimation of the ductility demand involves considerable uncertainty as it relies on the
judgment of the assessor or a numerical model of the building. The use of a displacement
ductility of 1.0 as the onset of the stepwise function is useful, as non-hairline residual crack
Marder et al. 319

widths can be used as indicators that yielding of the longitudinal reinforcement has
occurred. However, as previously discussed, a lack of non-hairline residual crack widths
does not necessarily indicate a lack of yielding in columns, or other members subjected to
axial compression.

Conclusion
Various factors related to the post-earthquake assessment of moderately damaged rein-
forced concrete plastic hinges were investigated. ‘‘Moderate’’ damage here refers to damage
states up to and including cracking, longitudinal reinforcement yielding, and cover con-
crete spalling, but does not include bar buckling or crushing of core concrete. The key con-
clusions of this study are summarized as follows.
Residual crack widths are dependent on the residual deformations of the component being
considered. Residual deformations are a function of the load history and axial load, and are
not always indicative of the peak deformation demands imposed on a plastic hinge during a
damaging earthquake. Crack width data taken from standard cyclic tests may show trends
that are artifacts of the loading protocols rather than causal relationships. Given these limita-
tions, residual crack widths may be best used as a qualitative metric. In beams, metrics cap-
turing the distribution of cracking can indicate whether or not sufficient inelastic rotations to
induce strain hardening in the longitudinal reinforcement are likely to have occurred.
Rather than focusing solely on component damage, peak drift demand should be esti-
mated through analysis of the building system and validated against the distribution of
damage observed in the building. This peak drift demand from the model can be com-
pared with the range of peak demand estimates based on the extent of plastic hinge
damage (Figure 6).
The residual stiffness of a plastic hinge can be approximated as a function of the total
residual crack width, but there are a number of limitations with using crack widths in such
a manner. An alternative method is to calculate the residual stiffness as a function of the
estimated maximum deformation demand during the damaging earthquake. A simple
inverse relationship between the residual stiffness (normalized against the initial secant
stiffness to yield) and the displacement ductility was found to yield a lower-bound esti-
mate, based on a dataset of reinforced concrete components where the residual stiffness
was measured after application of arbitrary earthquake-type loadings.
It is emphasized that these conclusions are not applicable in cases of severe damage,
such as longitudinal reinforcement buckling or crushing of core concrete. The focus of this
study was also on modern ductile components, and hence the recommendations do not
cover non-flexural damage states, such as wide diagonal cracks, or anchorage failure.

Declaration of conflicting interests


The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/
or publication of this article.

Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or
publication of this article: Much of the experimental data used in this paper came from a testing pro-
gram funded by the Natural Hazards Research Platform and QuakeCoRE, with additional contribu-
tions from Atlas Tilt Slab, Sika New Zealand, and BBR Contech (QuakeCoRE publication number:
320 Earthquake Spectra 36(1)

0331). The first author also received support from the Commonwealth Scholarship and Fellowship
Plan.

References
Acun B and Sucuoglu H (2010) Performance of reinforced concrete columns designed for flexure
under severe displacement cycles. ACI Structural Journal 107(3): 364–371.
American Concrete Institute (2014) Building Code Requirements for Structural Concrete (ACI 318-
14). Farmington Hills, MI: American Concrete Institute.
American Society of Civil Engineers (2017) Seismic Evaluation and Retrofit of Existing Buildings
(ASCE 41-17). Reston, VA: American Society of Civil Engineers.
Applied Technology Council (1998) Evaluation of Earthquake Damaged Concrete and Masonry Wall
Buildings (FEMA 306). Washington, DC: Federal Emergency Management Agency.
Arias Acosta J (2011) Seismic performance of circular and interlocking spirals R/C bridge columns
under bi-directional shake table loading, PhD Thesis, University of Nevada, Reno, Reno, NV.
Bechtoula H, Sakashita M, Kono S, et al. (2006) Cyclic performance of lower stories of mid-rise
reinforced concrete frame buildings. ACI Structural Journal 103(4): 513–521.
Clough RW and Johnston SB (1966) Effects of stiffness degradation on earthquake ductility
requirements. In: Proceedings of Japan earthquake engineering symposium, Tokyo, Japan.
Di Ludovico M, Polese M, d’Aragona MG, et al. (2013) A proposal for plastic hinges modification
factors for damaged RC columns. Engineering Structures 51: 99–112.
El-Bahy A, Kunnath SK, Stone WC, et al. (1999) Cumulative seismic damage of circular bridge
columns: Benchmark and low-cycle fatigue tests. ACI Structural Journal 96(4): 633–641.
Elwood KJ and Eberhard MO (2009) Effective stiffness of reinforced concrete columns. ACI
Structural Journal 106(4): 476–484.
European Committee for Standardization (2004) Eurocode 8: Design of Structures for Earthquake
Resistance—Part 1: General Rules, Seismic Actions and Rules for Buildings (EN 1998-
1:2004+A1:2013). Brussels: European Committee for Standardization.
Federal Emergency Management Agency (2012) Seismic Performance Assessment of Buildings
(FEMA P58). Washington, DC: Federal Emergency Management Agency.
Fenwick RC and Megget LM (1993) Elongation and load deflection characteristics of reinforced
concrete members containing plastic hinges. Bulletin of the New Zealand Society for Earthquake
Engineering 26(1): 28–41.
Hachem MM, Moehle JP and Mahin SA (2003) Performance of Circular Reinforced Concrete Bridge
Columns Under Bidirectional Earthquake Loading. Berkeley, CA: Pacific Earthquake Engineering
Research Center, University of California, Berkeley.
Henry RS, Dizhur D, Elwood KJ, et al. (2017) Damage to concrete buildings with precast floors
during the 2016 Kaikoura earthquake. Bulletin of the New Zealand National Society for
Earthquake Engineering 50(2): 174–186.
Japan Building Disaster Prevention Association (JBDPA) (2001) Standard for Seismic Evaluation of
Existing Reinforced Concrete Buildings, version 1. Tokyo, Japan: JBDPA (in English).
Japan Building Disaster Prevention Association (JBDPA) (2014) Guideline for Post-Earthquake
Damage Evaluation and Rehabilitation. Tokyo, Japan: JBDPA (in Japanese).
Kam WY, Pampanin S and Elwood K (2011) Seismic performance of reinforced concrete buildings
in the 22 February Christchurch (Lyttelton) earthquake. Bulletin of the New Zealand National
Society for Earthquake Engineering 44(4): 239–278.
Laplace PN, Sanders D, Saiidi MS, et al. (1999) Shake Table Testing of Flexure Dominated
Reinforced Concrete Bridge Columns. Reno, NV: Center for Earthquake Engineering Research,
University of Nevada, Reno.
Lee J-Y and Watanabe F (2003) Predicting the longitudinal axial strain in the plastic hinge regions
of reinforced concrete beams subjected to reversed cyclic loading. Engineering Structures 25(7):
927–939.
Marder et al. 321

Lu Y, Henry RS, Gultom R, et al. (2016) Cyclic testing of reinforced concrete walls with distributed
minimum vertical reinforcement. Journal of Structural Engineering 143(5): 04016225.
Marder K (2018) Post-earthquake residual capacity of reinforced concrete plastic hinges. PhD
Dissertation, The University of Auckland, Auckland, New Zealand.
Marder K, Elwood K, Motter C, et al. (in press) Quantifying the effects of epoxy repair of reinforced
concrete plastic hinges. Manuscript Submitted for Publication.
Marder K, Motter C, Elwood K, et al. (2018a) Effects of variation in loading protocol on the
strength and deformation capacity of ductile reinforced concrete beams. Earthquake Engineering
& Structural Dynamics 47(11): 2195–2213.
Marder K, Motter C, Elwood K, et al. (2018b) Testing of seventeen identical ductile reinforced
concrete beams with various loading protocols and boundary conditions. Earthquake Spectra
34(3): 1025–1049.
Matthews J (2004) Hollow-core floor slab performance following a severe earthquake, PhD Thesis,
University of Canterbury, Christchurch, New Zealand.
Morris G, Bull D and Bradley B (2014) Reviewing the uncertainties in seismic experimentation
following the unexpected performance of RC structures in the 2010-2011 Canterbury
Earthquakes. In: New Zealand society for earthquake engineering conference, Auckland, New
Zealand.
Nakano Y, Maeda M, Kuramoto H, et al. (2004) Guideline for post-earthquake damage evaluation
and rehabilitation of RC buildings in Japan. In: Proceedings of the 13th world conference on
earthquake engineering, Vancouver, BC, Canada, 1–6 August.
Nojavan A (2015) Performance of full-scale reinforced concrete columns subjected to extreme
earthquake loading. PhD Thesis, University of Minnesota, Minneapolis, MN.
Opabola EA, Best T, Elwood KJ, et al. (2018) Deformation capacity of reinforced concrete beams
with single crack. In: New Zealand society for earthquake engineering conference, Auckland, New
Zealand.
Paulay T and Priestley MJN (1992) Seismic Design of Reinforced Concrete and Masonry Buildings.
Hoboken, NJ: John Wiley & Sons.
Schoettler M, Restrepo J, Guerrini G, et al. (2015) A Full-Scale, Single-Column Bridge Bent Tested
by Shake-Table Excitation. Berkeley, CA: Pacific Earthquake Engineering Research Center,
University of California, Berkeley.
Sritharan S, Beyer K, Henry RS, et al. (2014) Understanding poor seismic performance of concrete
walls and design implications. Earthquake Spectra 30(1): 307–334.
Standards New Zealand (2006) Concrete Structures Standard, NZS 3101:2006, Incorporating
Amendment no. 1, 2, and 3. Standards New Zealand: Wellington, NZ.
Takeda T, Sozen MA and Nielsen NN (1970) Reinforced concrete response to simulated
earthquakes. Journal of the Structural Division 96(12): 2557–2573.

You might also like