E.

Cosenza (ed), Eurocode 8 Perspectives from the Italian Standpoint Workshop, 129-144, © 2009 Doppiavoce, Napoli, Italy

CAPACITY MODELS OF RC MEMBERS WITH EMPHASIS

ON SUB-STANDARD COLUMNS WITH PLAIN BARS

Edoardo Cosenza a, Gaetano Manfredi b, Gerardo M. Verderame c,

Paolo Ricci d, Giovanni De Carlo e, Angelo Masi f
a
DIST, University Federico II, Naples, Italy, cosenza@unina.it
b
DIST, University Federico II, Naples, Italy, gamanfre@unina.it
c
DIST, University Federico II, Naples, Italy, verderam@unina.it
d
DIST, University Federico II, Naples, Italy, paolo.ricci@unina.it
e
DIST, University Federico II, Naples, Italy, giovanni.decarlo@unina.it
f
DiSGG, University of Basilicata, Potenza, Italy, angelo.masi@unibas.it

ABSTRACT
Current code prescriptions allow to evaluate the ultimate rotational capacity from hybrid
(mechanical-empirical) or empirical formulations, for R.C. members with deformed bars and
seismically detailed. These formulations can be extended to non-conforming elements by
applying correction coefficients calibrated on experimental data. These coefficients, for
members with plain bars, imply a reduction of 40% at least; this reduction increases if lapping
of longitudinal bars is present. The experimental campaign conducted at the University of
Naples on 16 real-scale columns with plain bars allowed to extend the experimental database
for this typology. Experimental results highlight the excessive conservativeness of the current
code proposal. Based on these results, new correction coefficients are proposed.

KEYWORDS
Existing RC building, non conforming, ultimate capacity, plain bars, lapping.

1 INTRODUCTION

The present Italian technical regulations (D.M. 2008), on a level with the most modern of
international codes (CEN, 2005), allow us to determine the seismic capacity of existing RC
buildings with recourse to non-linear analysis methodologies. The use of such methods of
analysis, however, requires knowledge of the real post-elastic rotational capacities of each
element of the construction (beams, columns) both in monotonic field, for non-linear static
analysis, and in cyclical field, for non-linear dynamic analysis. In monotonic field, a series of
parameters (yielding, peak resistance, ultimate state) has to be defined, in order to define the
response curve of the element. In cyclic field, hysteretic rules and strength and stiffness
degradation models have to be defined; they significantly influence the assessment of ultimate
rotational capacity. Nevertheless, these rules are not easy to define, due to the number of
geometrical and mechanical parameters and to the uncertainties involved. For example, the
type of loading influences in a not negligible way the response of the r.c. element. Most of the
code prescriptions only define the deformation capacity at the elastic limit (yielding) and at
ultimate (collapse); therefore, based on these prescriptions, it is not possible to completely
130 E. Cosenza, G. Manfredi, G.M. Verderame, P. Ricci, G. De Carlo, A. Masi

define the strength degradation of the monotonic envelope, nor the hysteretic behaviour
through appropriate rules.
Generally, deformation at yielding is evaluated as a chord rotation, accounting for different
contributions corresponding to bending, shear and fixed-end rotation deformation
mechanisms.
The rotational capacity is generally evaluated referring to a fixed strength decay (20%)
respect to the peak resistance, evaluated on the envelope force-displacement curve. It is clear
that this definition is strongly influenced by the maximum resistance condition, as well as the
post-peak degradation, monotonic or cyclic. It is difficult to define a relationship between the
element parameters and the rotational capacity, due to the complex phenomena influencing
the post-elastic deformation behaviour and to the natural variability affecting this phenomena.
The code, consistently with the methodologies developed in literature, proposes two main
approaches: a mechanical-empirical approach, based on plastic hinge length concept, and a
purely empirical approach.
Referring to the purely empirical formulation proposed in (CEN, 2005), in the present work,
based on experimental data, the applicability of this formulation to under-designed elements
with plain bars is evaluated. In particular, correction coefficients applied to the code
formulation are proposed for elements with plain bars, with or without lapping of longitudinal
reinforcement.

2 EVALUATION OF ULTIMATE CHORD ROTATION

In this section, the theoretical background of current European code (CEN, 2005) formulas
for the ultimate rotational capacity of reinforced concrete members is presented. Principles
and methodologies standing behind the two main approaches to the assessment of this value
(mechanical and empirical) are introduced.

2.1 Code formulas (EC8 part 3.3)

Eurocode 8 – Part 3 at section A.3.2.2 (Limit state of near collapse) provides expressions for
the evaluation of ultimate element capacity of R.C. elements. The value of total chord rotation
capacity under cyclic loading, following a mechanical approach, is given by [EC8 - Eq.
(A.1)]:

⎛ f ⎞
⎜ αρsx yw
⎡ max(0.01;ω′) ⎤
0.225 0.35 ⎟
1 ⎛ LV ⎞ ⎜ fc ⎟
θ um = 0.016 ⋅ (0.30 ν ) ⎢ fc ⎥ ⎜ ⎟ 25 ⎝ ⎠
(1.25100ρd ) (1)
γ el ⎣ max(0.01;ω) ⎦ ⎝ h ⎠

where γel,  equal  to  1.5  for  primary  seismic  elements  and  to  1.0  for  secondary  seismic 
elements,  is  meant  to  convert  mean  values  of  chord  rotation  to  mean‐minus‐one‐
standard‐deviation ones. The code also provides another expression for the evaluation 
of the plastic part of the ultimate chord rotation [EC8 - Eq. (A.3)]: 

⎛ f ⎞
⎜ αρsx yw
⎡ max (0.01;ω′) ⎤
0.3 0.35 ⎟
1 ⎛L ⎞ ⎜ fc ⎟
θ pl = θ um − θ y = 0.0145 ⋅ (0.25 ν ) ⎢ ⎥ f c 0.2 ⎜ V ⎟ 25 ⎝ ⎠
(1.275100ρd ) (2)
⎣ max(0.01;ω ) ⎦
um
γ el ⎝ h ⎠
Capacity Models of RC Members with Emphasis on Sub-Standard Columns with Plain Bars 131

In this expression the coefficient γel equal  to  1.8  for  primary  elements  and  to  1.0  for 
secondary  ones.  To  evaluate  the  total  chord  rotation,  the  plastic  part  calculated 
according to this formula should be added to the yielding rotation [EC8 - Eq. (A.10)].
The values of chord rotation calculated according to (1) and (2) apply to elements with
deformed bars, seismically detailed and without lapping of longitudinal bars in the vicinity of 
the end region where yielding is expected (plastic hinge region). 
The correction coefficient applied to members with deformed bars without seismic detailing is
equal to 0.825 for both formulas. If the longitudinal deformed bars are lapped, expressions (1)
and (2) should be applied doubling the mechanical compression reinforcement ratio ( ω′ ). 
Moreover, if the lap length is less than the minimum value  l ou ,min :

⎡ f yw ⎤
l ou ,min = d bL f yL / ⎢(1.05 + 14.5α1ρ sx ) fc ⎥ (3)
⎣ fc ⎦

another reduction factor equal to (l o / l ou ,min ) should be applied, calibrated only for expression
(2), that is only for the plastic part of chord rotation. Corrections applied to the yielding chord
rotation are given at section A.3.2.4(3) of the code; they are omitted here for the sake of
brevity.
In elements with plain bars the chord rotation evaluated according to (1) should be multiplied
by 0.575, while the plastic part of chord rotation given by (2) should be multiplied by to
0.375. It’s worth noting that both coefficients already include the reduction factor equal to
0.825, accounting for the lack of seismic detailing. If longitudinal bars are lapped in members
with plain bars, another coefficient has to be adopted, depending on the lap length ( l o ) and
the shear span ( L V ). For total chord rotations, it is given by:

0.0025(180 + min(50, l o / d bL ))(1 − l o / L V ) (4)

while for the only plastic part it is:

0.0035(60 + min(50, l o / d bL ))(1 − l o / L V ) (5)

Moreover, shear span in expressions (1) and (2) should be reduced by the lap length l o ,
assuming that the ultimate condition is controlled by the region right after the end of the lap.
The ultimate rotation may also be calculated following an equivalent mechanical approach
through the evaluation of the ultimate section curvature, assumed to be constant over the
plastic hinge length, which is empirically calibrated. Hence, the ultimate rotational capacity
may be evaluated according to [EC8 - Eq. (A.4)]:

1 ⎛⎜ ⎛ L ⎞⎞
θ um = θ + ( φ − φ ) L ⎜⎜1 − 0.5 pl ⎟⎟ ⎟ (6)
γ el ⎜⎝ L V ⎠ ⎟⎠
y u y pl
⎝

Section curvatures at ultimate and at yielding are calculated based on the first principles, with
the constitutive relationships given by Eurocode 2 (CEN, 2004). If the concrete confinement
model given in 3.1.9 in Eurocode 2 is assumed, the plastic hinge length is equal to [EC8 - Eq.
(A.5)]:
132 E. Cosenza, G. Manfredi, G.M. Verderame, P. Ricci, G. De Carlo, A. Masi

d bL f y
L pl = 0.10L V + 0.17 h + 0.24 (7)
fc

If the confinement model proposed by Eurocode 8 – part 3 is adopted, better representing the
effects of confinement under cyclic loading, the plastic hinge length is given by:

LV d bL f y
L pl = + 0.20h + 0.11 (8)
30 fc

For expressions (7) and (8) no correction factor accounting for the above mentioned
deficiencies is given. Therefore, they should only be applied to members with deformed bars,
seismically detailed and without lapping of longitudinal bars.

2.2 Mechanical approach: background theory

From a phenomenological standpoint, the plastic hinge region can be identified with the zone
of the element where yielding of reinforcement and concrete crushing take place. The plastic
hinge length used in the evaluation of the element rotational capacity is, instead, purely
conventional. It only represents the length over which ultimate section curvature, assumed to
be constant, is integrated, following an equivalent bending approach, to calculate the effective
chord rotation including shear and fixed-end rotation contributions to the overall
deformability of the member; the curvature is calculated based on Bernoulli’s plane section
assumption.
The plastic hinge length can not be evaluated based on a purely mechanical approach. As a
matter of fact, based on section equilibrium conditions and full-interaction hypothesis, in a
post-peak phase the curvature should increase only at the base section of the element (“zero
length hinge problem)” (Daniell et al., 2008; Haskett et al., 2009). Moreover, a purely
mechanical approach, leading to the evaluation of flexural deformability, would not account
for other deformation mechanisms such as shear deformability and slippage of reinforcing
bars from the connection element. These contributions are not negligible at all. Shear
mechanisms may contribute in the overall post-elastic member deformability up to 30 %
(Fenwick et al., 1993), whilst the end rotation due to the slippage of reinforcing bars may
contribute up to 40 % (Sezen, 2002).
Therefore, researchers over years have empirically calibrated the plastic hinge length over
which theoretical ultimate section curvature is integrated, aiming at achieving the best
agreement with experimental values of ultimate chord rotation.
Following this approach, rotational capacity of an element may be expressed as:

θ u = θ y + (φ u − φ y )L pl (9)

where length Lpl is made up of three terms, corresponding to different deformation

mechanisms:

L pl = L pl,flex + L pl,shear + L pl,slip (10)

Table 1 reports main formulations that have been proposed over years, starting from the first
fundamental work by Baker (Baker et al., 1956). These expressions show that the shear span
LS and the section depth h are the major variables influencing the plastic hinge length, while
Capacity Models of RC Members with Emphasis on Sub-Standard Columns with Plain Bars 133

the term corresponding to fixed-end rotation is generally proportional to diameter and

yielding strength of longitudinal reinforcement bars. First proposed formulations are mainly
calibrated based on experimental tests on beam elements, therefore the fixed-end rotation
contribution is not clearly evaluated. In recent formulations, calibrated also on column
elements, this contribution is clearly represented instead.
Moreover, in (9) the ultimate condition is given in terms of curvature φ u , depending, based
on plane section hypothesis, on steel or concrete failure. Nevertheless, the evaluation of
ultimate curvature is not easy or univocal, due to the influence of some aspects as concrete
confinement, spalling of the concrete cover or buckling of compressive reinforcing bars. For
example, the use of different confinement models may significantly influence the
determination of the ultimate curvature, therefore the plastic hinge length can assume very
different values.
The plastic hinge formulation proposed in (Panagiotakos et al., 2001) is the most interesting
among the expressions presented in literature. It is based on an extensive experimental
database, which will be discussed in the next paragraph.

Table 1. Empirically derived hinge lengths.

Reference Hinge Length (Lpl)
14
(Baker et al., 1956) k1k 2 k3 ⋅ ( z d ) ⋅d

d ⎡⎢ ⎛ z ⎞ ⎛⎜ ⎛ q − q' ⎞ d ⎞⎟ ⎤⎥
(Mattock, 1964) ⋅ 1+ ⎜ 1.14 −1⎟ 1− ⎜ ⎟
2 ⎢ ⎝ d ⎠ ⎜ ⎜ qb ⎟ 16.2 ⎟ ⎥
⎣ ⎝ ⎝ ⎠ ⎠⎦
d z
(Corley, 1966) + 0.2
2 d
d
(Mattock, 1967) + 0.05 z
2
(Park, 1982) 0.4h
(Priestley et al., 1987) 0.08 L + 6d
v b
(Paulay et al., 1992) 0.08 L + 0.022 d f
v b y
0.12 L + 0.014α d f for cyclic loading
v sl b y
(Panagiotakos et al., 2001)
0.18 L + 0.021α d f for monotonic loading
v sl b y
0.09 L + 0.2 h for cyclic loading
v
(Fardis, 2007)
0.04 L + 1.2 h for monotonic loading
v

The ultimate chord rotation is given by:

φyLV ⎛ L pl ⎞
θu = + (φ u − φ y )L pl ⎜⎜1 − 0.5 ⎟ (11)
3 ⎝ L V ⎟⎠

and the plastic hinge length L pl is given as a linear function of shear span L V (bending
contribution) and of the product (f y d bL ) (fixed-end contribution):

L pl = αL V + β(f y d bL ) (12)
134 E. Cosenza, G. Manfredi, G.M. Verderame, P. Ricci, G. De Carlo, A. Masi

Coefficients α =0.12 e β =0.0014 are derived from a regression analysis on experimental data
from cyclic tests. The ultimate curvature φ u is evaluated accounting both for the concrete
confinement and for the spalling of the concrete cover. The mean and median of the
experimental-to-predicted ratio for expression (11), using (12), are equal to 1.23 and 0.99
respectively, with a Coefficient of Variation (CoV) of 83%.
The last plastic hinge expression proposed by (Fardis, 2007), based on a more extensive
experimental database, is depending not on the shear span L V but also on the height h of the
section. Moreover, the fixed-end rotation contribution is evaluated with a separate term:

⎛ L pl ⎞
θ u = θ y + (θ u ,slip − θ y,slip ) + (φ u − φ y )L pl ⎜⎜1 − 0.5 ⎟⎟ (13)
⎝ L V ⎠

with:

L pl = 0.09L V + 0.20h (14)

where:

φyLV ⎛ h ⎞ φ y d bL f y
θy = + 0.0013⎜⎜1 + 1.5 ⎟⎟ + (15)
3 ⎝ LV ⎠ 8 fc

φ y d bL f y
θ y,slip = (16)
8 fc

φ u d bL f y
θ u ,slip = (17)
16 f c

The use of the illustrated relationships, together with the confinement model showed in the
same work, leads to an experimental-to-predicted ratio with mean and median, on a database
of 1307 experimental tests, equal to 1.105 and 0.994 respectively, with a CoV of 53.6%.
Expressions (11) e (13), although providing a different evaluation of the fixed-end
contribution, present the same control variables of the code expression (6), which directly
shows, in the calculation of plastic hinge length, the dependence on all the above mentioned
parameters.

2.3 Empirical approach: background theory

Formulas for the evaluation of rotational capacity can also be obtained with a totally empirical
approach, based on experimental data, with pure numerical regression analyses. Different
empirical expressions are proposed in literature (Rossetto, 2002; Zhu et al, 2007); among
them, the expression proposed in (Panagiotakos et al., 2001) is certainly based on the most
extensive database. Therefore, it is the most representative and it represents a reference for
code formulas (CEN, 2005).
This experimental database consists in 633 cyclic tests and 242 monotonic tests on beams,
columns and walls, which do not present brittle failure mechanisms. The relationship is a
linear regression of the logarithm of θ u on the control variables or their logarithms, without
Capacity Models of RC Members with Emphasis on Sub-Standard Columns with Plain Bars 135

coupling. Only control variables which turn out to be statistically significant for the prediction
of θ u are retained. Separate regression analyses for monotonic tests and for cyclic ones are
performed. To obtain a more representative experimental database, with particular regard to
members with unsymmetric reinforcement well represented in monotonic tests, another
regression analysis on all 875 tests is performed, carrying to the following expression:

⎛ α ⎞⎛ α ⎞
θ u = α st ⋅ α cyc ⋅ ⎜1 + sl ⎟⎜1 − wall ⎟
⎝ 2.3 ⎠⎝ 3 ⎠
⎛ f ⎞ (18)
⎜ 100 αρsx yw
⎡ max(0.01;ω′) ⎤
0.275 0.45 ⎟
⎛ LV ⎞ ⎜ fc ⎟
(0.20 ν ) ⎢ fc ⎥ ⎜ ⎟ 1.1 ⎝ ⎠
(1.30100ρd )
⎣ max(0.01;ω ) ⎦ ⎝ h ⎠

where αcyc is a binary coefficient given equal to 1 for monotonic loading and equal to 0.6 for
cyclic loading. The ratio between the experimental ultimate rotation and the numerical value
provided by (18) has mean equal to 1.06, median equal to 1.00 and CoV of 47%.
During years, together with the extension of the experimental database, the coefficients in this
expression have been slightly modified. The last proposal, given in (Fardis, 2007), is based on
1307 monotonic and cyclic tests:

⎛ α ⎞⎛ 3 ⎞
θ u = α st ⋅ (1 − 0.43α cyc ) ⋅ ⎜1 + sl ⎟⎜1 − α wall ⎟
⎝ 2 ⎠⎝ 8 ⎠
⎛ f ⎞ (19)
⎜ αρsx yw
⎡ max(0.01;ω′) ⎤
0.225 0.35 ⎟
⎛ LV ⎞ ⎜ fc ⎟
(0.30 ν ) ⎢ fc ⎥ ⎜ ⎟ 25 ⎝ ⎠
(1.25100ρd )
⎣ max(0.01;ω ) ⎦ ⎝ h ⎠

where α st is equal to 0.0185 for hot-rolled ductile steel, 0.0115 for heat-treated (tempcore)
steel, and 0.0090 for cold-worked steel. The mean value of the ratio between the experimental
ultimate rotation and the numerical value provided by (19) is 1.05, the median is equal to
0.995 and the CoV is of 42.8%. The comparison between the coefficients of variation clearly
shows the better prediction capacity of (19), given by the growth of experimental knowledge
state.
In the same work, a regression analysis for the only plastic part is also presented, which was
already proposed in (CEB-FIB Bulletin 24, 2003) based on 1100 experimental tests. The
expression is:

⎛ α sl ⎞
θ pl ⎟(1 − 0.4α wall )
pl
u = α st ⋅ (1 − 0.52α cyc ) ⋅ ⎜1 +
⎝ 1 .6 ⎠
⎛ f ⎞ (20)
⎜ αρsx yw
⎡ max(0.01;ω′) ⎤
0.30 0.35 ⎟
0.20 ⎛ L ⎞ ⎜ fc ⎟
(0.25 ν ) ⎢ ⎥ fc ⎜ V ⎟ 25 ⎝ ⎠
(1.275100ρd )
⎣ max(0.01;ω ) ⎦ ⎝ h ⎠

The mean value of the ratio between the experimental ultimate rotation and the corresponding
numerical prediction is 1.05, the median is equal to 0.995 and the CoV is of 42.7%, against
the 47% in the first proposal (see Eq. 18).
136 E. Cosenza, G. Manfredi, G.M. Verderame, P. Ricci, G. De Carlo, A. Masi

Expressions (1) and (2) proposed in EC8 almost perfectly agree with (19) and (20), assuming
α cyc =1 (cyclic loading), α sl =1 (with slip), α wall =0 (only beams and columns) e α st e α stpl
equal to 0.0185 (hot-rolled ductile steel).
Consistently with the characteristic of tests included in the experimental database, the
proposed expressions for the ultimate rotational capacity should be applied only to members
with deformed bars, with seismic detailing and without lapping of longitudinal bars in the
vicinity of plastic hinge region, that is, to members which are not representative of existing
buildings. Authors define correction coefficients allowing to extend the use of these
expressions to members with different characteristics. These coefficients are calibrated to
counterbalance the mean error evaluated through the comparison between values from
expressions (19) and (20) and results of experimental tests on under-designed members, not
included in the original (primary) database. This approach, certainly approximated, is
necessary because of the small number of experimental data for these members. Because of
the low number of these data, it seems to be allowed to suppose that their inclusion in the
database would have not led to any significant change in the regression expression. Moreover,
applying the primary expression to members of different typologies, only using a
multiplicative coefficient, is the same as postulating that the ultimate rotation depends on the
control parameters by the same way, independently on the specific characteristics of
considered elements. Nevertheless, the assumed methodology seems to be the only one that
can be followed, due to the few experimental data now available for this kinds of elements. A
higher reliability can be obtained only by extending the experimental database, so that a wider
range of loading conditions and geometrical and mechanical characteristics can be covered. In
Table 2 correction coefficients and the extension of the corresponding experimental database
used for calibrations are reported.

Table 2. Correction factors for non-detailed members.

# of Mean - Median Reference
Element Type Correction Factor
Data CoV
w/o seismic detailing and 0.81 - 0.85 (Panagiotakos et al., 2002;
0.85 27
continuous ribbed bars 42% CEB-FIB Bulletin 24, 2003)
w/o seismic detailing w/
hooked plain bars and w/ or
w/o lap-splicing over
(
0.015 ⋅ 10 + min ( 40;lo d b ) ) 15
1.07 - 0.975
32%
(Fardis, 2006)
plastic-hinge length

2.4 Critical review

The expressions for the ultimate rotational capacity, as clearly shown in the previous
paragraphs, are necessarily calibrated on experimental data, due to the complex nature of
mechanisms affecting the post-elastic behaviour of reinforced concrete members and their
interaction.
Both the approaches presented in literature and in Code are characterized by high values of
the coefficient of variation of the experimental-to-predicted capacity ratio.
The high CoV affecting expressions (19) and (20) – or (1) and (2) - is not only due to the
natural experimental variability, but also to the difficulty in completely modelling with a
simple formulation the interaction between the complex phenomena influencing the post-
elastic deformation behaviour of reinforced concrete element. Panagiotakos e Fardis in
(Panagiotakos et al., 2001), based on the analysis of subgroups of tests, homogenous for
geometrical and mechanical characteristics and for loading conditions, quantify the CoV
associated with the only natural variability in 12.5 %.
Capacity Models of RC Members with Emphasis on Sub-Standard Columns with Plain Bars 137

The limited prediction capacity of these expressions is also due to impossibility of introducing
in the control variables some parameters which certainly affect the rotational capacity. The
major among these parameters is the load path, that is the energy dissipated in hysteretic
cycles. This aspect has been experimentally investigated by (Pujol et al., 2006), who analyzed
the influence of displacement history on the decay of element resistance capacity. The
experimental tests show that, given equal the geometrical and mechanical characteristics and
the applied axial load (that is, all the input parameters of code and literature regression
formulations), it is possible to predetermine the value of element chord rotation
corresponding to a conventional drop of 20 % of peak resistance, by imposing a given load
path (cfr. Figure 1).

Figure 1. Influence of displacement history on ultimate chord rotation (Pujol et al., 2006).

Panagiotakos and Fardis, in the above mentioned work, try to explicitly account for the effect
of cyclic loading by another regression, where the type of loading is evaluated with a variable
expressing the equivalent number of inelastic imposed cycles ( ∑ | θi | θ u ), instead of the
coefficient ( α cyc ). Nevertheless, contrary to expectations, the inclusion of this parameter
makes worse the prediction capacity of the formulation. The coefficient of variation (CoV) of
the ratio between the experimental and the predicted value, in fact, increases up to 51 %. On
the other hand, the usual structural modelling approaches do not allow to introduce the
dissipated energy in the control variables.
A critical analysis of expressions (19) and (20), based on mechanical considerations regarding
the absence of a direct relationship between the median estimation of the ultimate rotation and
some parameters that certainly influence the member capacity, seems to be without
foundation. Due to the purely statistical nature of the expression, in fact, the retaining of these
variables turns out to be not significant because of their strong correlation with other
parameters, already present in the formulation (Panagiotakos et al., 2001).
It’s worth noting that the higher coefficient of variation affecting the hybrid mechanical-
empirical formulation (plastic hinge length) with respect to the purely empirical one is
probably related to the difficulty in expressing the ultimate rotation as a function of element
characteristics based on a statistical regression analysis restrained to a mechanical
relationship.
138 E. Cosenza, G. Manfredi, G.M. Verderame, P. Ricci, G. De Carlo, A. Masi

3 DEFORMATION CAPACITY OF RC MEMBERS WITH PLAIN BARS

Plain reinforcing bars have been widely used in the construction of European reinforced
concrete buildings. In Italy and in the whole Mediterranean area, their use was widely spread
up to 1970s, in north-American countries and in New Zealand construction with plain bars are
present until 1950. The high spreading of reinforced concrete buildings with plain bars among
existing buildings can be deduced if it is considered that 50 % of Italian existing buildings has
been constructed between earliest 1940s and latest 1970s, when reinforced concrete structures
with plain bars were the prevailing construction typology.
The correct evaluation of deformation capacity of R.C. elements has to account for the
effective bond capacities between reinforcing bars and the surrounding concrete. For
members with plain bars, low bond capacities directly influence the three main deformation
mechanisms: bending, shear and fixed-end rotation.
As shown by experimental evidence, the scarce capacities of load transfer between the
reinforcing bars and the surrounding concrete makes the deformation contribution associated
with the fixed-end rotation effect very important. This contribution, in fact, due to the cyclic
and post-elastic decay of bond capacities, may represent up to 80-90 % of overall
deformability of the element (Verderame et al., 2008a; Verderame et al., 2008b).
Bond capacities also influence the development of cracks along the element. A lower number
of wider cracks is observed when bond decreases. This greatly influences both shear and
bending deformability, reducing the former and increasing the latter.
Therefore, formulations able to provide a reliable assessment of ultimate deformation capacity
of elements with plain bars are of a particular interest for assessment of existing buildings.
The ultimate rotational capacities for members with plain bars, according to code, as already
shown at paragraph 2.1, is evaluated by applying a correction coefficient to the capacity
formulations calibrated on members with deformed bars and seismically detailed. In the
following a new calibration of these coefficients, which result to be too conservative, is
proposed, based on an experimental database of columns with plain bars extended with recent
experimental results from tests executed in the laboratory of the Department of Structural
Engineering at the University of Naples “Federico II”, in the research project ReLUIS-DPC
2005-2008 Linea 2.

3.1 Experimental data set

Most of literature data about the experimental behaviour of R.C. elements comes from test
executed on members with deformed bars. During last years, the need for a reliable
assessment of seismic capacity of existing structures has produced an increasing number of
experimental campaigns aimed at the study of behaviour of under-designed elements. This
allowed to extend the experimental database for the calibration of correction coefficients
applied to the regression relationships for the evaluation of ultimate rotational capacity.
Coefficients proposed in (CEN, 2005), reported at 2.1, are calibrated on very few
experimental tests. Expression proposed in (Fardis, 2006), reported in Table 2, are, instead,
based on 15 experimental tests; 9 of them without lapping ( l o / d bL = 100 ) and 6 with a lap
length l o varying between 15, 25 and 40 times the diameter d bL of longitudinal reinforcing
bars. The comparison between the code correction coefficient and the latest one proposed in
(Fardis, 2006) shows the considerable conservativeness of the prescription proposed by
Eurocode 8.
In recent times, in the Department of Structural Engineering at the University of Naples
“Federico II”, a great attention has been addressed to the experimental study of members with
Capacity Models of RC Members with Emphasis on Sub-Standard Columns with Plain Bars 139

plain bars, both through test aimed at the characterization of bond capacities in cyclic
(Verderame al., 2009a; Verderame et al., 2009b) and post-elastic field (Verderame et al.,
2008c) and through tests on real-scale columns elements under monotonic and cyclic loading.
The first phase of the experimental activity 6 monotonic test e 6 cyclic ones have been
performed, on elements with square section (300×300)mm2, for different values of the applied
axial load. In this phase particular attention has been addressed to the detail of longitudinal
bars, by executing tests on elements without lapping of longitudinal bars at the base of the
element and on elements with a lap length l o equal to 40 times the diameter d bL of
longitudinal bars.
The second phase of the experimental campaign, just finished and still unpublished, is
focused on the comparison between rotational capacity and deformation mechanisms of R.C.
elements with plain and ribbed bars. In particular, 4+4 tests have been executed on elements
equal for the geometry of the transverse section, the geometric ratio of longitudinal and
transverse reinforcement, the axial load level and the load path, varying the geometry of the
transverse section. The characteristics of tested elements are reported in Table 3, where the
drift corresponding to the 20 % decay of the peak resistance is also given

Table 3. Geometrical and mechanical characteristics of tested elements.

b LV/h P/Agfc ρl reinforcement θu, exp
n test Reference lo/dbL loading
[mm] [%] Type (20%)
[%]
1 300 5.23 0.12 0.8 40 Plain cyclic 6.23
2 300 5.23 0.12 0.8 40 Plain cyclic 5.82
3 300 5.23 0.12 0.8 100 Plain cyclic 6.49
4 300 5.23 0.24 0.8 40 Plain cyclic 3.72
5 300 5.23 0.24 0.8 100 Plain cyclic 3.81
6 300 5.23 0.24 0.8 100 Plain cyclic 2.77
Verderame et al., 2008
7 300 5.23 0.12 0.8 40 Plain monotonic 6.83
8 300 5.23 0.12 0.8 40 Plain monotonic 6.88
9 300 5.23 0.12 0.8 100 Plain monotonic 10.72
10 300 5.23 0.12 0.8 100 Plain monotonic 7.87
11 300 5.23 0.24 0.8 40 Plain monotonic 7.87
12 300 5.23 0.24 0.8 100 Plain monotonic 4.29
13 300 5.00 0.2 1.0 100 Plain monotonic 8.53
14 300 5.00 0.2 1.0 100 Plain cyclic 5.43
15 300 3.00 0.1 0.9 100 Plain cyclic 5.27
16 500 5.00 0.1 0.9 100 Plain cyclic 6.23
Reluis (2005-2008)
17 300 5.00 0.2 1.0 100 Ribbed monotonic 6.86
18 300 5.00 0.2 1.0 100 Ribbed cyclic 3.87
19 300 3.00 0.1 0.9 100 Ribbed cyclic 3.65
20 500 5.00 0.1 0.9 100 Ribbed cyclic 4.66

By adding these tests, the database for the evaluation of the correction coefficient applied to
the ultimate rotational capacity of elements with plain bars consists of 26 tests, 7 of which
monotonic. It’s worth noting that tests (#1,2,3,5,6) were already included in the database used
by Fardis for calibrating expressions reported in Table 2 (Fardis, 2006).

3.2 Calibration of correction factor

In this paragraph the correction coefficient applied to code expressions for the ultimate
rotation of members with plain bars is calibrated, based on experimental data introduced at
3.1.
140 E. Cosenza, G. Manfredi, G.M. Verderame, P. Ricci, G. De Carlo, A. Masi

The correction coefficients will be calibrated according to the methodology already illustrated
at 2.3, with regard to the following expression:

θ u = θ y + θ pl
u (21)

with:

⎛ f ⎞
⎜ αρsx yw
⎡ max(0.01;ω′)⎤
0.3 0.35 ⎟
0.2 ⎛
L ⎞ ⎜ fc ⎟
θ pl ν
= 0.03 ⋅ (1 − 0.52α cyc )(0.25 ) ⎢ ⎥ fc ⎜ V⎟ 25 ⎝ ⎠
(1.275100ρd ) (22)
⎣ max(0.01;ω) ⎦
u
⎝ h ⎠

It is to be noted that a factor accounting for the type of loading ( α cyc ) has been added to the
code expression (2) for the plastic part of the ultimate rotation θ pl u . This assumption is
considered to be allowed because of the almost perfect agreement between the code
expression and the one proposed in (Fardis, 2007), as already shown at 2.3.
Table 4 reports, for all experimental tests in the database, the ratios between experimental
ultimate rotation and the corresponding theoretical value ( θ u ,exp / θ u ), according to (22).

Table 4. Ratios between experimental ultimate rotations and corresponding theoretical values.
n test Reference loading lo/dbL θu,exp/θu,m
1 University of Patras cyclic 15 0.33
2 University of Patras cyclic 15 0.62
3 University of Patras cyclic 25 0.39
4 University of Patras cyclic 25 0.41
5 University of Patras cyclic 100 0.58
6 University of Patras cyclic 100 0.60
7 Other sources cyclic 100 0.54
8 Other sources cyclic 100 0.74
9 Other sources cyclic 100 0.83
10 Other sources cyclic 100 1.25
11 University of Naples (Verderame et al., 2008) cyclic 40 1.26
12 University of Naples (Verderame et al., 2008) cyclic 40 0.83
13 University of Naples (Verderame et al., 2008) cyclic 40 0.60
14 University of Naples (Verderame et al., 2008) cyclic 100 1.21
15 University of Naples (Verderame et al., 2008) cyclic 100 1.13
16 University of Naples (Verderame et al., 2008) cyclic 100 0.81
17 University of Naples (Verderame et al., 2008) monotonic 100 0.69
18 University of Naples (Verderame et al., 2008) monotonic 100 0.70
19 University of Naples (Verderame et al., 2008) monotonic 100 0.92
20 University of Naples (Verderame et al., 2008) monotonic 40 1.09
21 University of Naples (Verderame et al., 2008) monotonic 40 0.80
22 University of Naples (Verderame et al., 2008) monotonic 40 0.50
23 University of Naples (Reluis) monotonic 100 1.20
24 University of Naples (Reluis) cyclic 100 1.41
25 University of Naples (Reluis) cyclic 100 1.42
26 University of Naples (Reluis) cyclic 100 1.76

The ratio ( θ u ,exp / θ u ) for members without lapping of longitudinal bars (conventionally
reported as l o / d bL = 100 ) has mean equal to 0.99 and median equal to 0.87, with a CoV of
37%. Hence, based on the experimental database, it can be deduced that the assessment of the
Capacity Models of RC Members with Emphasis on Sub-Standard Columns with Plain Bars 141

ultimate rotation with (22) overestimates the median value of rotational capacity by about
13%.
The use of expression (22) for members with lapping of longitudinal bars overestimates even
more the experimental rotational capacity. A linear regression performed on the ratio
( θ u ,exp / θ u ) gives the following expression for the correction coefficient:

k= 0.020 min(45, l o / d bL ) (23)

This coefficient, applied also to elements without lapping, allows to account for the
overestimate of the rotational capacity given by (22); in particular, the rotational capacity
evaluated according to (22) is reduced by 10%. The ratio [ θ u ,exp /(kθ u ) ], calculates on all
tests in the experimental database, has mean equal to 1.10 and media equal to 1.01, with a
CoV of 37%.
Figure 2a reports, for each experimental test, both cyclic and monotonic, the ratio between the
experimental ultimate rotation and the corresponding theoretical value ( θ u ,exp / θ u ), together
with the correction coefficient given by (23), which should be applied to (22).
It is noted that including monotonic tests in the evaluation of the correction factor (k) is the
same as postulating that the reducing of rotational capacity due to cyclic loading, evaluated in
(22) by the coefficient (1 − 0.52α cyc ) , is, on average, not depending on bond capacities. As a
matter of fact, this coefficient, as previously illustrated, is calibrated on a database made up of
members with deformed bars; therefore, the evaluation of the correction coefficient (k) has
been executed supposing that the reduction given by (1 − 0.52α cyc ) can also be extended to
members with plain bars.
Nevertheless, from a theoretical standpoint, a member with deformed bars, given equal the
geometrical and mechanical characteristics, should show a higher cyclic degradation with
respect to a member with plain bars, because of the micro-cracking of the concrete
surrounding the reinforcing bar due to the higher bond performances, which emphasizes the
strength degradation of concrete alternatively in compression and in tension.

2.0 2.0
θu,exp / θu cyclic tests θu,exp / θu
1.8 1.8
# 26 tests monotonic tests
1.6 1.6 # 19 tests

1.4 1.4
1.2 1.2
1.0 1.0
0.8 0.8
[0.02·min(45,lo/dbL)] [0.02·min(50,lo/dbL)]
0.6 0.6
0.4 0.4
0.2 0.2
lo/dbL lo/dbL
0.0 0.0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
(a) (b)
Figure 2. Proposed correction factor: (a) cyclic and monotonic tests, (b) only cyclic tests.
142 E. Cosenza, G. Manfredi, G.M. Verderame, P. Ricci, G. De Carlo, A. Masi

However, due to the uncertainties related to the inclusion of monotonic tests, the correction
coefficient will now be calibrated based on the only cyclic tests. For these tests, the ratio
( θ u ,exp / θ u ) for elements without lapping of longitudinal bars has mean equal to 1.02 and
median equal to 0.98, with a CoV of 39%. Therefore, based on the experimental tests,
expression (22) shows a very good agreement with the cyclic rotational capacity of elements
with plain bars without lapping of longitudinal reinforcement.
A linear regression performed on the ratio ( θ u ,exp / θ u ), for elements with lapping of
longitudinal reinforcement, gives the following expression for the correction coefficient:
k= 0.020 min(50, l o / d bL ) (24)
Figure 2b reports, for each cyclic experimental result, the ratio between the experimental
ultimate rotation and the corresponding theoretical value ( θ u ,exp / θ u ), together with the
correction coefficient given by (24), applied to (22).

3.3 Discussion of results

The extension of the experimental database allowed to re-calibrate the correction coefficients
applied to the assessment of the ultimate rotational capacity of elements with plain bars, with
or without lapping of longitudinal reinforcement.
The choice between the correction coefficient calibrated only on cyclic tests or on the whole
experimental database is not easy. The numerical results, in fact, are rather different.
Although the inclusion of monotonic tests in the database is affected by the previously
highlighted uncertainties, it is to be noted that the consideration of the only cyclic tests would
imply a further reduction in the extension of the experimental database, which is already
limited.
However, both the expressions of correction coefficients proposed in the present work
highlight the conservativeness of EC8 proposal, which is based on very few experimental
tests. Moreover, EC8 assumes that, when lapping of longitudinal reinforcement is present, the
ultimate condition is controlled by the region right after the end of the lap, so that the shear
span and, therefore, the rotational capacity are further reduced, but this assumption is not
confirmed by the experimental results. The highest plastic demand, in fact, always
concentrates at the base section of the element.
Despite the difficulties in the choice of the most reliable expression for the correction
coefficient, recent experimental results clearly highlight the higher rotational capacity of
members with plain bars with respect to ones with deformed bars, equal for structural
characteristics and details. As a matter of fact, the comparison between the ultimate rotations
of the elements from the second phase of the experimental campaign, briefly illustrated at 3.1,
highlight that the capacity of members with plain bars are higher, on average, by 35 %
compared with the corresponding members with deformed bars (see Table 3).
From a mechanical standpoint, the higher ultimate rotational capacity of columns with plain
bars may be explained with the comparison between two opposite mechanisms: the increase
of deformability caused by the fixed-end rotation mechanism, particularly exalted due to the
low bond capacities; on the other hand, the higher degradation of global resistance due to the
increase of deformation demand on concrete in compression, localized at the base of the
element and associated with the concentrated rotation (“rocking effect”). According to
experimental evidence, the former seems to prevail on the latter, leading to an overall increase
of ultimate rotational capacity compared with members with higher bond capacities.
Capacity Models of RC Members with Emphasis on Sub-Standard Columns with Plain Bars 143

4 CONCLUSIONS

In this work, the theoretical background to code formulas for the assessment of ultimate
rotational capacity of reinforced concrete members has been briefly presented. Most recent
literature contribution, together with advantages and deficiencies of the approaches to the
calibration of this relationships, have been illustrated.
Special attention has been addressed to the calibration of correction coefficients used for the
assessment of ultimate rotational capacity of under-designed elements, with emphasis on
members with plain bars.
Main conclusions drawn form this work are:
• The evaluation of post-elastic deformation capacity of r.c. elements may only be based
on experimental data; any mechanical approach would not allow to evaluate accurately
the complex interaction phenomena influencing the deformability of the element.
• The reliability of regression expressions proposed in literature, some of which have
been adopted in code, is a direct result of the extension and the correct sorting of the
database.
• The estimate of rotational capacity of under-designed elements is strongly influenced
by the low number of experimental data related to this typologies.
• Recent experimental tests on columns with plain bars, executed at the University of
Naples (DIST), allow to extend significantly the database used for the calibration of
correction coefficients applied to the assessment of these elements, with or without
lapping of longitudinal reinforcement.
• The re-calibration of correction coefficients, even within the limits of the adopted
methodology, has allowed to highlight the excessive conservativeness of current code
prescriptions for elements with plain bars; this is confirmed by the experimental
evidence, showing that the ultimate rotation of members with plain bars is higher
compared with members with deformed bars, on average, by 35 %, given equal the
structural characteristics and details.

5 ACKNOWLEDGEMENTS

This work has been developed under the research program of "ReLUIS – Rete dei Laboratori
Universitari di Ingegneria Sismica", with the financial support of the Italian Department of Civil
Protection – Executive Project 2005-2008.

6 REFERENCES

Baker A.L.L. (1956). “Ultimate load theory applied to the design of reinforced and prestressed concrete
frames”, Concrete Publications Ltd., London, UK.
CEB-FIB Bulletin 24 (2003). “Seismic assessment and retrofit of reinforced concrete buildings”,
International Federation for Structural Concrete, Task Group 7.1.
CEN (2004). “European standard EN1992-1-1. Eurocode 2: Design of concrete structures - Part 1-1:
General rules and rules for buildings”, European Committee for Standardisation, Brussels.
CEN (2005). “European standard EN1998-3. Eurocode 8: Design provisions for earthquake resistance of
structures. Part 3: Assessment and retrofitting of buildings”, European Committee for
Standardisation, Brussels.
Corley W.G. (1966). “Rotational Capacity of Reinforced Concrete Beams”, Journal of the Structural
Division, ASCE, Volume 92, No. ST5, 121-146.
144 E. Cosenza, G. Manfredi, G.M. Verderame, P. Ricci, G. De Carlo, A. Masi

Daniell J. E., Oehlers D. J., Griffith M. C., Mohamed Ali M.S., Ozbakkaloglu T. (2008). “The softening
rotation of reinforced concrete members”, Engineering Structures, Volume 30, No. 11, 3159-3166.
Fenwick R.C., Megget L.M. (1993). “Elongation and load deflection characteristics of reinforced
concrete members containing plastic hinges”, Bulletin of New Zealand Society for Earthquake
Engineering, Volume 26, No. 1.
Fardis M.N. (2006). “Design rules for FRP retrofitting according to Eurocode 8 and their background”,
Lecture to Fib Course 2006 “Retrofitting of concrete structures through externally bonded FRPs with
emphasis on seismic applications”, Mexico.
Fardis M.N. (2007), LESSLOSS – Risk mitigation for earthquakes and landslides. “Guidelines for
displacement-based design of buildings and bridges”, Report n°5/2007, IUSS Press, Pavia, Italy.
Haskett M., Oehlers D.J., Mohamed Ali M.S., ChengqingWu (2009). “Rigid body moment–rotation
mechanism for reinforced concrete beam hinges”, Engineering Structures, in press.
Mattock, A.H. (1964). “Rotational capacity of hinging regions in reinforced concrete beams” Flexural
Mechanics of Reinforced Concrete, SP-12, American Concrete Institute, Farmington Hills.
Mattock A.H. (1967). “Discussion of rotational capacity of hinging regions in reinforced concrete
beams”, Journal of the Structural Division, ASCE, Volume 93, No. ST2, 519-522.
Panagiotakos T.B., Fardis M.N. (2001). “Deformation of reinforced concrete members at yielding and
ultimate”, ACI Structural Journal, Volume 98, No. 2, 135-148.
Panagiotakos T.B., Kosmopoulos A.J., Fardis M.N. (2002). “Displacement-based seismic assessment and
retrofit of reinforced concrete buildings”, Proceedings of the 1st fib Congress, Osaka, Japan, 13-19
October.
Park R., Priestley M.J.N., Gill W.D. (1982), “Ductility of square-confined concrete columns”, Journal of
Structural Division, ASCE, Volume 108,No. ST4, 929-950.
Paulay T., Priestley, M.J.N. (1992). “Seismic design of reinforced concrete and masonry buildings”, John
Wiley and Sons, New York.
Priestley M.J.N., Park R. (1987). “Strength and Ductility of Concrete Bridge Columns Under Seismic
Loading”, ACI Structural Journal, Volume 84, No. 1, 61-76.
Rossetto T. (2002). “Prediction of deformation capacity of non-seismically designed reinforced concrete
members”, 7th U.S. National Conference on Earthquake Engineering, Boston, USA.
Sezen, H. (2002). “Seismic behavior and modeling of reinforced concrete building columns.” PhD
Dissertation, Dept. of Civil and Environmental Engineering, Univ. of California at Berkeley,
Berkeley, Calif.
Verderame G.M, Fabbrocino G, Manfredi G. (2008a), Seismic response of r.c. columns with smooth
reinforcement. Part I: monotonic tests. Engineering Structures, Volume 30, No. 9, 2277-2288.
Verderame G.M, Fabbrocino G, Manfredi G. (2008b), Seismic response of r.c. columns with smooth
reinforcement. Part II: cyclic tests. Engineering Structures. Volume 30, No. 9, 2289-2300.
Verderame G.M., Ricci P., Manfredi G., Cosenza E. (2008c), La diffusione della deformazione plastica
nella risposta di elementi armati con barre lisce, Proc. of the ReLuis Workshop“Valutazione e
riduzione della vulnerabilità sismica di edifici esistenti in c.a.”, Roma, 29-30 maggio 2008 (in
Italian).
Verderame G.M., De Carlo G., Ricci P., Manfredi G.(2009a), Cyclic bond in elastic field of plain round
bars. Part I: Experimental results, Building & Construction Materials (submitted).
Verderame G.M., Ricci P., De Carlo G., Fabbrocino G. (2009b), Cyclic bond in elastic field of plain round
bars. Part II: Modelling, Building & Construction Materials (submitted).
Zhu L., Elwood K.J., Haukaas T. (2007). “Classification and seismic safety evaluation of existing
reinforced concrete columns”, ASCE Journal of Structural Engineering, Volume 133, No. 9, 1326-13.