EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS

Earthquake Engng Struct. Dyn. 2003; 32:2407–2424 (DOI: 10.1002/eqe.334)

E
ects of reinforcement slippage on the non-linear response

under cyclic loadings of RC frame structures

Suchart Limkatanyu1;∗; † and Enrico Spacone2;3

1 Department of Civil Engineering; Faculty of Engineering; Prince of Songkla University; Hadyai;
Songkla 90110; Thailand
2 PRICOS Department; University of Chieti; viale Pindaro 42; 65127 Pescara; Italy
3 Department of CEAE; University of Colorado; Boulder; CO 80309-0428; U.S.A.

SUMMARY
This paper discusses the importance of including the bond-slip e
ects in assessing the response under
cyclic loads of reinforced concrete frames. The discussion is based on analyses performed using numer-
ical models which are simple, computationally ecient and capable of representing the salient features
of reinforced concrete frames under both static and dynamic loads. The numerical models comprise a
displacement-based, reinforced concrete frame element with bond-slip and a rigid beam–column joint
element with bond-slip. Two applications illustrate the model accuracy and show the importance of
including bond-slip. The rst application considers a reinforced concrete beam–column subassemblage
experimentally tested under cyclic loads. The second application considers the shaking table test of a
two-story one-bay reinforced concrete frame. In both cases the analytical results correlate well with
the experimental results in terms of strength, displacement demands and hysteretic energy dissipation.
Furthermore, the paper shows how the analyses that include bond-slip yield a better correlation with
the experimental results with respect to the analyses that assume a perfect bond. Copyright ? 2003
John Wiley & Sons, Ltd.

KEY WORDS: frames; reinforced concrete; bond-slip; hysteretic models; cyclic loads; seismic loads;
nite elements; non-linear analysis

INTRODUCTION

The accurate representation of the bond-slip e
ects is crucial in predicting the response of
reinforced concrete (RC) frames subjected to both static and dynamic loadings. Under the
assumption of a perfect bond between the concrete and the steel rebars, the sti
ness of RC

∗ Correspondence to: Suchart Limkatanyu, Department of Civil Engineering; Faculty of Engineering; Prince of
Songkla University; Hadyai; Songkhla 90110; Thailand.
† E-mail: lsuchart@ratree.psu.ac.th

Contract/grant sponsor: Royal Thai Fellowship

Contract/grant sponsor: National Science Foundation; contract/grant number: CMS-9804613

Received 25 December 2002

Copyright ? 2003 John Wiley & Sons, Ltd. Accepted 14 April 2003
2408 S. LIMKATANYU AND E. SPACONE

structures is overestimated, as is the hysteretic energy dissipated during cyclic loads. Exper-
imental tests on RC subassemblages have shown large xed-end rotations at the structural
member ends. These xed-end rotations are caused by slippage of the rebars passing through
the joints or being anchored into the footings. Under cyclic loads, the bond gradually deteri-
orates, and additional exibility leads to the characteristic pinched hysteretic loops observed
in several tests. The inclusion of the bond-slip e
ects into numerical models is a crucial
step towards the development of accurate non-linear techniques for the analysis of RC frame
structures.
Major achievements in modeling the static and dynamic responses of RC structures have
been accomplished in recent years, especially in the development of ber frame models. The
main advantage of the ber-section model is that it automatically couples the axial and bending
e
ects. Fiber-section models can be combined with any frame element framework, including
displacement-based and force-based formulations.
Most ber-section frame models available in the published literature are typically based
on the assumptions that plane sections remain plane and that there is strain-compatibility
between the concrete and the steel rebars, thus neglecting the bond-slip e
ects. This results in
an overestimation of the initial sti
ness and of the hysteretic energy of RC structures [1; 2].
The simplest way to account for the bond-slip e
ects in frame elements is to add non-linear
springs at the member ends [3]. Although simple, this approach requires the formulation of an
ad-hoc phenomenological moment-rotation relation, and disrupts the continuity of the bers
between adjacent elements. Monti and Spacone [4] explicitly include the bond-slip e
ects into
the ber section without relaxing the strain-compatibility between the concrete and the rebar.
The total steel ber strain is the sum of two components: the strain in the rebar, plus a strain
equivalent to the rebar slip. The element is ecient and accurate, but rather complicated to
implement because it combines the force-based element state determination [1] with the steel
ber iteration loop needed to determine the contribution of the steel strain and of the bond-slip
to the total ber strain [4; 5].
The authors have recently developed RC models that explicitly account for the bond-slip
e
ects: one is a displacement-based RC frame element with bond-slip [6], the other is a rigid-
panel joint element with bond-slip [7]. The main objective of this paper is to use these models
to investigate how bond-slip a
ects the cyclic response of RC structures through the study
of two structural specimens previously tested in the lab at other research institutions. After a
brief presentation of the formulations of the two RC models with bond-slip, correlation studies
between numerical and experimental results for the two structural specimens (a beam–column
joint and a two-story one-bay frame) are presented. The force-based ber frame element with
perfect bond proposed by Spacone et al. [1] is used to compare the accuracy of the ber
elements with and without bond-slip in assessing the response of the experimentally tested
specimens.

DISPLACEMENT-BASED RC FRAME ELEMENT WITH BOND-SLIP

The two-node displacement-based RC frame model with bond-interfaces used in this paper
is presented in Reference [6] and is shown in Figure 1. The element formulation is briey
reviewed hereafter. The RC frame element consists of the following components: a two-node
beam, plus n two-node bars, which are allowed to slip with respect to the beam. The nodal

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:2407–2424
 EFFECT OF REINFORCEMENT SLIPPAGE ON NON-LINEAR RESPONSE 2409

y
U 41 u1 (x) U 42
U 31 U 32
U 21 v B (x) U 22 x
1
z
U 1
uB (x) U 2
1

U 31+i U 32+i
ui ( x )
Element DOFs Fiber section

σc
σs Db
unconfined

f c' confined
fy Db1
bond
'
concrete steel interface
3
0.2 f c D b

ε co ε cu εc εs ub3
ub
f t' ub1 ub2
Constitutive laws

Figure 1. RC frame element with bond-slip: degrees of freedom, ber section

discretization and uniaxial constitutive laws.

degrees of freedom of the beam and of the rebars are di
erent to permit reinforcement slip.
Figure 1 shows the nodal displacements and the displacement elds along the element. The
element nodal displacements U = {U11 U21 U31 · · · U3+n 1
| U12 U22 U32 · · · U3+i
2
}T and the section
displacements u(x) = {uB (x)vB (x) | u1 (x) · · · un (x)} are also shown in Figure 1. The section
T

deformations are grouped in the vector d(x) = {
B (x) B (x) |
1 (x) · · ·
i (x) · · ·
n (x)}T , where

B (x) is the beam axial strain, B (x) is the beam curvature, and
i (x) is the axial strain of bar
i. Using the Euler–Bernoulli beam theory, plane sections are assumed to remain plane and
normal to the element longitudinal axis. The element section-deformations are related to the
element section-displacements through the compatibility relations:
B = d uB = d x, B = d vB2 = d x2 ,
and
i = d ui = d x, where uB (x) is the beam axial displacement, vB (x) is the beam transverse
displacement, and ui (x) is the axial displacement of bar i. The section forces conjugate of the
section deformations d(x) are D(x) = {NB (x)MB (x) | N 1 (x) · · · N i (x) · · · Nn (x)}T , where NB (x)
and MB (x) are the beam-section axial force and bending moment, respectively, and N i (x) is
the axial force in bar i. Finally, the slip ubi (x) between the beam and bar i is ubi (x) = ui (x) −
uB (x) + dvdx
B (x)
y i , where d vB (x)= d x is the beam section rotation and y i is the distance of bar i
from the beam reference axis.
The element displacements are written as functions of the nodal displacements U through
the displacement shape functions Nu (x)


NuB (x)
u(x) = U = Nu (x)U (1)
Nu(x)

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:2407–2424
2410 S. LIMKATANYU AND E. SPACONE

where NuB (x) is the array of the displacement shape functions of a two-node beam element.
They describe a linear axial displacement eld and a cubic vertical displacement eld. Nu(x)
contains the displacement shape functions of the bars with bond slip, which are linear in the
present two-node formulation.
The beam and bond compatibility equations are both satised point-wise. The section de-
formations d(x) and bond deformations db (x) are directly related to the nodal displacements
U through the following equations:
d(x) = Bu (x)U db (x) = Bb (x)U (2)
where Bu (x) = @Nu (x), Bb (x) = @b Nu (x) and @, @b are linear di
erential operators [6].
The equilibrium equations are satised in the integral sense. Application of the virtual
displacement principle, substitution of Equation (2) and subsequent elimination of the virtual
nodal displacements U, yield the following integral form of equilibrium:
 
BuT (x)D(x) d x + BbT (x)Db (x) d x = P (3)
L L

where P is the nodal force vector conjugate of U and Db (x) is an array containing the
bond-interface stresses. If Equation (3) is expressed in incremental form, the matrix form of
equilibrium becomes
KU = P − P0 (4)
where K is the element sti
ness matrix, computed as
K = K B + Kb (5)
 
where KB = L BuT (x) kB (x)Bu (x) d x, Kb = L BbT (x)kb (x)Bb (x) d x are the beam and the bond
contributions to the element sti
ness, respectively, kB (x) is the beam-section sti
ness and
kb (x) is the bond-interface sti
ness. P is the array containing the element forces
P = P B + Pb (6)
  T
where PB = BT (x)DB (x) d xPB ,
L u
Pb = L Bb (x)Db (x) d x are the beam and the bond contri-
butions to the element forces, respectively. P0 = PB0 + Pb0 is the array containing the element
initial forces.
The ber-section model is used to compute the response of the beam section. The loading
envelopes of the uniaxial cyclic constitutive laws for the concrete, steel and bond laws are
schematically shown in Figure 1. The Kent and Park law [8] is used for the concrete. The
Menegotto and Pinto law [9] is used for the steel rebars. For the bond-interface between the
concrete and the rebar, the Eligehausen et al. law [10] is used.

PLANE RIGID-PANEL JOINT ELEMENT WITH BOND-SLIP

The four-faced plane joint element with bond-slip proposed by Limkatanyu [7] is shown in
Figure 2. The joint element comprises a beam-panel, a column-panel, and a rigid-link member.
These two panels are assumed to be independent of one another and are connected together
through a rigid-link member to prevent spurious rigid body modes. The same rotation is

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:2407–2424
 EFFECT OF REINFORCEMENT SLIPPAGE ON NON-LINEAR RESPONSE 2411

Rigid Panel

=
Rebar
Column Panel
Rebar

Beam Panel
Rebar

+ +
Rebar

Rigid Link

Figure 2. Rigid-panel joint element with bond-slip.

imposed at each face of the joint element. The slippage of the rebars passing through the
joint is explicitly accounted for.
The rigid-panel assumption is reasonable when the shear deformations of the joint are
negligible. This assumption is substantiated by the research conducted by Pantazopoulou and
Bonacci [11], who carried out the database analysis of 143 specimens of exterior and interior
beam–column joints tested over a span of 35 years in the US, Canada, Japan, and New
Zealand. Pantazopoulou and Bonacci [11] state that joints should be designed with sucient
hoop reinforcement so that their load carrying capacity exceeds that of the adjacent members.
Furthermore, they observe that none of the specimens analyzed failed because of joint shear
failure, even when poorly reinforced, because of the connement provided by the transverse
beams framing into the joint, which represent the typical conguration in internal joints.
Consequently, the proposed joint element is best suited for the aforementioned situation and
is intended to account mostly for bond-slip e
ects, while shear panel failures are not of
concern in this study. All the numerical models used in this study are implemented in the
general purpose nite element program FEAP [12].

APPLICATIONS

The results from two experimental tests are used to verify the accuracy of the models with
bond-slip and to discuss the e
ects of reinforcement slippage on the cyclic response of RC
frames. The rst application considers the cyclic test of a RC beam–column subassemblage
where large xed-end rotations are observed at the column faces due to the slippage of the

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:2407–2424
2412 S. LIMKATANYU AND E. SPACONE

1372 mm
beam-column joint
610 mm Pwest , ∆ west Peast , ∆east

west beam east beam

1372 mm

2210 mm 457 mm 2210 mm

457 mm
610 mm

457 mm
356 mm

beam section column section

Figure 3. Geometry and loads of specimen B11 [13].

rebars passing through the joint. The second application considers an RC frame dynamically
tested on the shaking table at the University of California, Berkeley.

Beckingsale et al. [13]: Specimen B11

Beckingsale et al. [13] experimentally tested a series of interior RC subassemblages. These

specimens represented 2/3 models of a beam–column system in a typical building in the
range of 10 to 15 stories. One of these subassemblages, labeled Specimen B11, is used for
this study. The conguration of Specimen B11 is shown in Figure 3. Following the original
work by Beckingsale et al. [13], the units for specimen B11 are given in the SI system.
The column was subjected to a constant axial compression of 335 kN (approximately equal
to 5% fc
Ag ) and cyclic vertical displacements at both beam ends. The beam and column
cross-sections are also shown in Figure 3. For the beam exural reinforcement, the yield
strength is fy = 298 MPa and the elastic modulus is Es = 200 GPa. For the column exural
reinforcement, fy = 423 MPa and Es = 200 GPa. For the concrete, the unconned ultimate stress
is fc
= 36 MPa. The above data is provided by Beckingsale et al. [13]. Using the formula
proposed by Scott et al. [14], the conned concrete ultimate strength in the beam and in
the column are computed at 39:6 MPa and 55 MPa, respectively. The tensile strength of the

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:2407–2424
 EFFECT OF REINFORCEMENT SLIPPAGE ON NON-LINEAR RESPONSE 2413

0.10
18 20 22

12 14

West-End Displacement (m)

0.05
6 8

2 4 10
16

0.00
1 3 9 15

5 7
-0.05

11 13

17 19 21
-0.10

Figure 4. Cyclic loading sequence for west-end beam of specimen B11.

concrete, though considered in the analyses, does not a
ect the results, except at the very
early loading stages, before the section cracks. As for the bond stress–slip relation, only the
ascending branch of the constitutive law of Figure 1 is of interest here, because no bond
failure was observed either in the experiments or in the analyses. The initial sti
ness used is
21 MPa= mm for the deformed D19 (19 mm diameter) bar and 18 MPa= mm for the deformed
D22 (22 mm diameter) bar. These values are based on a set of formulas developed by Monti
et al. [15] from the regression analysis of a number of pullout tests available in the published
literature.
In the numerical model, both western and eastern beams are discretized with 20 elements,
while both top and bottom columns are discretized with 15 elements. Five Gauss–Lobatto
integration points are used for all elements. The beam and column elements are connected
through the beam–column joint element. In spite of the large number of elements, the simplic-
ity of the models used leads to rather short computational times even for the complex analyses
presented. The pattern of the imposed vertical displacements at the western beam-end is il-
lustrated in Figure 4. After two elastic cycles, larger end displacements were progressively
imposed with two full cycles applied at the same end displacement level.
Figure 5(a) compares the west-end experimental response with the numerical response ob-
tained with the model with bond-slip, while Figure 5(b) compares the experimental response
with the numerical response obtained with the model with the perfect bond proposed by Spa-
cone et al. [1]. Similar graphs are obtained for the east-end response and are not shown
here in the interest of space [7]. Positive beam loads and positive beam-end displacements
are associated with sagging bending moment while negative beam loads and negative beam-
end displacements are associated with hogging bending moment. As expected, both models
succeed in predicting the beam strength. However, the model with bond-slip yields a much
better prediction in terms of sti
ness and hysteretic energy dissipated. The beam-end response
obtained with the model with bond-slip clearly shows the patterns of slightly spindle-shaped
hysteretic loops also visible in the experimental responses.

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:2407–2424
2414 S. LIMKATANYU AND E. SPACONE

200
P

west east
100

West-End Force (kN)

0

-100
Model with Bond-Slip
Experiment
-200
-0.10 -0.05 0.00 0.05 0.10
(a) West-End Displacement (m)

200
P

west east
100
West-End Force (kN)

-100
Model without Bond-Slip
Experiment
-200
-0.10 -0.05 0.00 0.05 0.10
(b) West-End Displacement (m)

Figure 5. Experimental and numerical responses of specimen B11 (west-end beam):

(a) model with bond-slip; and (b) model without bond-slip.

Bond-slip mainly a
ects the shape of the unloading–reloading branches. During unloading,
initial unloading is followed by closing of the cracks at the beam–column interface, reloading,
and yielding of the steel in tension. With the model with bond-slip, when the beams unload,
closing of the beam–column interface cracks is accompanied by slip of the rebars passing
through the beam–column joint. Crack closure is indicated by an abrupt change of sti
ness
in the specimen response. This phenomenon results in a xed-end rotation at the beam–
column faces that leads to a more exible response. The di
erent response in the positive and
negative loading directions is due to the fact that the area of top reinforcement in the beams
is twice that of the bottom reinforcement, as shown in Figure 3. Therefore, yielding of the
beam bottom rebars in tension cannot induce yielding of the beam top rebars in compression.

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:2407–2424
 EFFECT OF REINFORCEMENT SLIPPAGE ON NON-LINEAR RESPONSE 2415

400
yield strength joint region
300
11 Cycle 6

Stress in Top Bar (MPa)

200

100

-100 22 21

-200 west beam end east beam end

-300
yield strength
-400
0 1 2 3 4 5
(a) Distance along Beam (m)

400
yield strength joint region
300
Stress in Bottom Bar (MPa)

200 Cycle 6

100

-100 11

-200 west beam end 21 22 east beam end

-300
yield strength
-400
0 1 2 3 4 5
(b) Distance along Beam (m)

Figure 6. Steel bars stress distributions along beam for specimen B11: (a) top bar; and (b) bottom bar.

This implies that the top beam–column interface cracks do not close when the beam-end
forces induce compressive stresses in the top bars. On the other hand, the bottom beam–
column interface cracks close every time the beam-end forces induce compressive stresses in
the bottom bars. By the time the beam–column interface cracks open throughout the section
depth, the Bauschinger e
ects in the bottom rebars also reduce the beam-section sti
ness.
The formation of cracks in the beam–column joint does not a
ect the integrity of the
joint due to the sucient amount of joint reinforcement, even though the density of cracks
observed during the experimental tests increased during the loading history [13]. Consequently,
the contribution of the joint panel to the inelastic behavior of the specimen is insignicant
and the assumption of rigid concrete in the joint panel seems reasonable.
Figure 6(a) and (b) shows the stress distributions of the outer-layer top and bottom bars
along the beam, respectively, at di
erent cycles (the cycle numbers refer to the labeling of

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:2407–2424
2416 S. LIMKATANYU AND E. SPACONE

0.70
joint region

Cycle 6
0.35

Slip in Top Bar (mm)

22

0.00
11

-0.35 west beam end east beam end

21

-0.70
0 1 2 3 4 5
(a) Distance along Beam (m)

0.70
joint region

11
Slip in Bottom Bar (mm)

0.35 21

0.00
Cycle 6

-0.35 west beam end east beam end

22
-0.70
0 1 2 3 4 5
(b) Distance along Beam (m)

Figure 7. Steel bar bond-slip distributions along beam for specimen B11:
(a) top bar; and (b) bottom bar.

Figure 4). The rebar stress distributions clearly show the push–pull phenomenon for the rebars
passing through the joint. The rebars are in compression at one side and in tension at the
opposite side. This results in stress gradients much steeper along the joint than along the
beams framing into the joint. This implies higher bond demands in the joint region. The yield
penetration towards both beam-ends from the column faces is about 300 mm for the beam
rebars. The column rebars did not yield during the analyses [7]. The same observation is
reported in the discussion of the experimental results [13]. Specimen B11 was designed based
on a strong column–weak beam design philosophy.
Figure 7(a) and (b) shows the bond-slip distributions of the outer-layer top and bottom bars
along the beam, respectively. These distributions are reported for the same cycles for which

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:2407–2424
 EFFECT OF REINFORCEMENT SLIPPAGE ON NON-LINEAR RESPONSE 2417

the steel stresses are shown in Figure 6. In general, the slips of the rebars inside the joint
are much higher than along the beams, due to the high rebar stress gradients shown in Figure
6. These large slips inside the joint cause large xed-end rotations at the beam–column faces
that were also observed in the test [13].

Clough and Gidwani [16] Specimen RCF2

The accuracy of the frame element with bond-slip in predicting the non-linear dynamic re-
sponse of RC frames is further validated by the correlation study with a two-story one-bay RC
frame specimen that was tested on the shaking table at the University of California, Berkeley.
The frame, referred to as specimen RCF2, represents a 0.7 scale model of a typical two-story
oce building. The details of the test specimen and of the test set-up on the shaking table
are described by Clough and Gidwani [16]. The design of specimen RCF2 was based on the
1970 UBC Code [17] and the 1971 ACI Code [18].
During the shaking table test the specimen RCF2 was subjected to three consecutive ground
motions. The characteristic of each ground motion corresponds to the N69W Taft record from
the Arvin–Tahachapi earthquake of 21 July 1952. The peak ground acceleration of each ground
motion was scaled to 0:095 g, 0:57 g, and 0:65 g. The three ground motions are referred to as
W1, W2, and W3, respectively. In this study, only the ground-motion W2 is of interest because
the shear cracks in the girders are minimal and the beam–column joints still maintain integrity,
thus implying that the behavior of the structure is dominated by the exure associated with
yielding and slippage of the reinforcement bars. This behavior is best suited for validating
the proposed frame and joint models.
Figure 8 shows the dimensions of the girder and of the column cross-sections. Following
the original work by Clough and Gidwani [16], the units for specimen RCF2 are given in
the American system. Based on the information available in Clough and Gidwani [16], the
unconned concrete compression strength fc
is 4:4 ksi. The conned concrete compression
strength was evaluated based on the work by Scott et al. [14]. For the top and bottom
girders, the conned compression strength is computed at 5.2 and 6:0 ksi, respectively. For
the columns, the conned compression strength is 5:0 ksi. The concrete tensile strength is
neglected, because the sections were already cracked following the application of ground
motion W1 [16]. For the girder exural reinforcement [16], the yield strength fy and the elastic
modulus Es for a #3 bar (bar diameter = 0:375 in.) are 52:0 ksi and 28400 ksi, respectively.
For a # 4 bar (bar diameter = 0:5 in:) fy and Es are 56:1 ksi and 28000 ksi, respectively. For
a # 5 bar (bar diameter = 0:625 in.) fy and Es are 41:5 ksi and 29800 ksi, respectively. For
the wire-mesh used in the girder, fy and Es are 64:7 ksi and 29400 ksi, respectively. As for
the bond stress–slip relation, in the test bond it did not fail or reach the plateau shown in
Figure 1; thus, only the ascending branch of the bond-slip envelope is of interest. Based on
the work by Monti et al. [15], the initial sti
ness of the bond-slip curve is 98 ksi= in for a #3
bar, 95 ksi= in for a #4 bar, and 91 ksi= in for a #5 bar.
Specimen RCF2 consists of two identical parallel frames supporting the girders with two
applied masses, as shown in Figure 8. The specimen is symmetric along the longitudinal axis,
and the ground motion direction is applied along this axis. Using symmetry, only one plane
frame is modeled. The top and bottom girders are each discretized into 24 elements while
the top and bottom columns are each discretized into 15 elements. The girder and column
elements are connected together through the joint element. There are seven elements that

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:2407–2424
2418 S. LIMKATANYU AND E. SPACONE

Concrete Block

67.75” 11.38” 2.9”

8 kips

Concrete Block
16 kips
11.38”

Bottom Story Girder

67.75”
16.0”

144.0” 72.0”

Front Elevation Side Elevation

18’’ 18’’ 18’’ 18’’
2.88’’

2.88’’
8.50’’

8.50’’
4#3 6#3
wire mesh 9 bars wire mesh 9 bars
per layer 5.75’’ per layer 5.75’’
2#4 2#5

Top Story Girder Section Bottom Story Girder Section

5.75’’

8.50’’ 2#5

Column Section

Figure 8. Girder and column cross-sections for specimen RCF2.

represent the 16-inch anchorage zone downward into the column base. In the original test
specimen, the rebars were hooked 16 in. into the foundation. To simulate the e
ect of the
hooks, the bar nodes are assumed fully anchored (and thus xed) 16 in. into the foundation.
Before the start of the rst shaking table test W1, a snap test was performed by Clough
and Gidwani [16] and the resulting measured natural frequencies were 3.80 and 9:80 Hz for
the rst and second mode of vibration, respectively. Before the start of the second shaking
table test W2, the snap test was repeated and showed natural frequencies of 3.13 and 8:70 Hz.
The frequency drop was mostly due to concrete cracking and some damage following test
W1 [16]. The second set of natural frequencies is used as a reference for calibrating the
analytical model. The natural frequencies of the modal analysis performed on the analytical
model before imposing the ground-motion W2 yields frequencies equal to 3.05 and 7:43 Hz for
the rst and second mode of vibration, respectively. 3% Rayleigh damping ratios are assumed
for both modes of vibration.
Figure 9(a) and (b) shows the comparison between experimental and analytical displacement
histories of the top and bottom story, respectively. From the response histories it is clear
that the rst mode of vibration dominates the frame response. Figure 9 shows overall good

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:2407–2424
 EFFECT OF REINFORCEMENT SLIPPAGE ON NON-LINEAR RESPONSE 2419

3
Experimental Result

Top Story Displacement (in)

2 Model with Bond-Slip

-1
δ

-2

-3
0 4 8 12 16
(a) Time (sec)

3
Bottom Story Displacement (in)

Experimental Result
2 Model with Bond-Slip

-1

-2 δ

-3
0 4 8 12 16
(b) Time (sec)

Figure 9. Experimental and analytical displacement response for specimen

RCF2: (a) top story; and (b) bottom story.

15
Experimental Result
Total Base Shear (kips)

10 Model with Bond-Slip

-5

-10

-15
0 4 8 12 16
Time (sec)

Figure 10. Experimental and analytical base shear response for specimen RCF2.

agreement between experimental and numerical results. The model with bond-slip succeeds
not only in matching the frequency content of the response and the general waveform of
the response but also in predicting the maximum displacements. Figure 10 compares the

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:2407–2424
2420 S. LIMKATANYU AND E. SPACONE

15

Total Base Shear (kips)

10
5 V V

0
-5
-10
-15
-2 -1 0 1 2
(a) Bottom Story Drift (in)
Total Top Story Shear (kips)

10

5 V V

-5

-10
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
(b) Top Story Drift (in)

Figure 11. Shear force-interstory drift response of specimen RCF2 obtained with model
with bond-slip: (a) bottom story; and (b) top story.

experimentally measured and the analytically predicted total base shear histories. Similarly
to the conclusions drawn from Figure 9, the numerical model with bond-slip is successful
both in representing the structural frequencies and the general waveform of the response and
in predicting the maximum base shear. Figure 11(a) plots the bottom story shear versus the
corresponding drift. It shows a pattern of a slightly spindle-shaped hysteretic loop. The sti
ness
degradation results mainly from crack opening and concrete crushing at the column bases while
the pinching is caused by slippage of the column rebars anchored in the foundations, which
leads to a column base rotation. Figure 11(b) presents the relation between top story shear
and top story drift. Figure 11(b) does not show yielding of the column reinforcement and
the small amount of hysteretic damping is mainly caused by the hysteretic behavior of the
concrete. Pinching is still present in the response due to the slippage of the column rebars
passing through the beam–column joints. No major inelasticity aside from concrete cracking
was observed in the girders.
In order to show the importance of bond-slip in assessing the response of the frame, the
analyses were repeated with the model with a perfect bond proposed by Spacone et al. [1]. The
concrete and steel material properties are those used in the analysis with bond-slip. The modal
analysis performed before imposing the ground-motion W2 yields the natural frequency values
of 3.25 and 7:85 Hz for the rst and second modes of vibration, respectively. As expected,

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:2407–2424
 EFFECT OF REINFORCEMENT SLIPPAGE ON NON-LINEAR RESPONSE 2421

3
Model without Bond-Slip

Top Story Displacement (in)

Model with Bond-Slip
2

-1
δ
-2

-3
0 4 8 12 16
(a) Time (sec)

3
Bottom Story Displacement (in)

Model without Bond-Slip

Model with Bond-Slip
2

-1
δ
-2

-3
0 4 8 12 16
(b) Time (sec)

Figure 12. E
ects of bond-slip on displacement response of specimen RCF2:

(a) top story; and (b) bottom story.

these natural frequency values are slightly larger than those obtained from the modal analysis
of the model with bond-slip, indicating a sti
er model. 3% Rayleigh damping ratios are used
for the rst and second modes of vibration. Figure 12(a) and (b) shows the e
ects of bond-
slip on the response histories of the top and bottom story displacement, respectively. For the
bottom story displacement, the maximum displacement obtained with the model with bond-
slip is approximately 51% larger than that obtained with the model without bond-slip. For the
top story displacement, the maximum displacement obtained with the model with bond-slip is
approximately 65% larger than that obtained with the model without bond-slip. Much of this
di
erence derives from bond-slip in the foundations and in the beam–column joints. Figure
12 clearly shows a sti
er response in the case of the model without bond-slip.
Figure 13 compares the base shear response histories with and without bond-slip. While
the two histories are di
erent, the inuence of bond-slip on the maximum base shear force is
insignicant. This happens because the base shear is related to the product of the structural
lateral sti
ness and of the structural interstory drift. When the e
ects of rebar slippage are
taken into account they cause a decrease in the lateral sti
ness and at the same time an increase
in the lateral displacements. It seems that in this particular application these two factors cancel
each other out. The same observation is also reported by D’Ambrisi and Filippou [19].

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:2407–2424
2422 S. LIMKATANYU AND E. SPACONE

20
Model without Bond-Slip
Model with Bond-Slip

Total Base Shear (kips)

10

-10

-20
0 4 8 12 16
Time (sec)

Figure 13. E
ects of bond-slip on base shear response for specimen RCF2.

15
Total Base Shear (kips)

10
5 V V

0
-5
-10
-15
-2 -1 0 1 2
(a) Bottom Story Drift (in)
Total Top Story Shear (kips)

10

5 V V

-5

-10
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
(b) Top Story Drift (in)

Figure 14. Shear force-interstory drift response of specimen RCF2 obtained with model
without bond-slip: (a) bottom story; and (b) top story.

Finally, Figure 14(a) and (b) shows the relation between story shear and interstory drift
displacements for the bottom and top story, respectively, for the analyses without bond-slip,
and should be compared to Figure 11. Similarly to the case of the model with bond slip, most
of the inelastic action is concentrated at the bottom story columns while the top story response

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:2407–2424
 EFFECT OF REINFORCEMENT SLIPPAGE ON NON-LINEAR RESPONSE 2423

is basically elastic. Compared to the responses of Figure 11, the pinching characteristic is not
observed in Figure 14 due to the perfect-bond assumption. Comparing Figure 11(a) and Figure
14(a), the hysteretic-energy dissipation in the case of the model without bond-slip is much
more pronounced than in the case of the model with bond-slip. This is due to the column
base rotation caused by the rebars slippage in the foundations.

SUMMARY AND CONCLUSIONS

This paper applies two recently developed RC models that explicitly account for the bond-
slippage of the steel rebars to the study of two experimentally tested RC structural specimens,
a beam–column joint and a two-story frame. The models are a ber frame element and a
beam–column joint element. The main objective of the study is to analyze the importance of
modeling reinforcement slippage in assessing the response of RC frame structures subjected
to cyclic loads.
The study of the RC beam–column subassemblage validates the model accuracy and shows
how including the e
ects of bond-slip leads to a slightly spindle-shaped hysteretic loop and to
a good assessment of the amount of hysteretic-energy dissipation, while excluding the bond-
slip e
ects overestimates the amount of hysteretic-energy dissipation. The bond-slip e
ects do
not a
ect the loading capacity of the structure since no slip failure was observed in either the
experimental test or the analytical results. The stress and bond-stress distributions of the rebars
indicate that the bond demand along the joint is critical and that the rebar slippage inside
the joint results in large xed-end rotations at the beam–joint interface. In this application
the assumption of rigid concrete in the joint-panel is reasonable due to the sucient joint
reinforcement provided.
In the second application, the study of the shaking-table test of a two-story one-bay RC
frame, very good correlation is observed when bond-slip is included, while the model with
a perfect bond leads to an overestimation of the hysteretic damping of the frame and an
underestimation of the structural exibility and of the displacement demands. Similarly to the
rst application, most of the slip e
ects are observed inside the joints and in the footings and
result in xed-end rotations that shorten the structural frequencies.
Overall, the analytical models used are quite simple and can be easily implemented in a non-
linear code and are general enough to be applied to other problems involving slip. The authors
are using the same models to study the response up to failure of RC beams strengthened with
Fiber Reinforced Polymers (FRP). Other applications in the eld of concrete structures are
the study of regions with insucient lap-splices (a problem not uncommon in older European
bridge piers built in seismic areas) and the model extension to concrete structures prestressed
with both steel and FRP cables. Finally, a further enhancement of the joint element should
model the e
ects of joint shear failure, that may be important in the study of external joints
that lack the benecial connement of the cross beams found in interior joints.

ACKNOWLEDGEMENTS

This study was supported by the Royal Thai Fellowship and by Grant CMS-9804613 from the National
Science Foundation. This support is gratefully acknowledged. Any opinions expressed in this paper are

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:2407–2424
2424 S. LIMKATANYU AND E. SPACONE

those of the authors and do not reect the views of the sponsoring agencies. The authors would also
like to thank Prof. P. B. Shing and Prof. K. J. Willam of the University of Colorado, Boulder, for their
fruitful discussions on theoretical issues.

REFERENCES

1. Spacone E, Filippou FC, Taucer FF. Fibre beam–column model for non-linear analysis of R/C frames: Part I.
Formulation. Earthquake Engineering and Structural Dynamics 1996; 25:711–725.
2. Spacone E, Filippou FC, Taucer FF. Fibre beam–column model for non-linear analysis of R/C frames: Part II.
applications. Earthquake Engineering and Structural Dynamics 1996; 25:727–742.
3. Rubiano-Benavides NR. Predictions in the inelastic seismic response of concrete structures including shear
deformations and anchorage slip. Ph.D. dissertation, Department of Civil Engineering, University of Texas,
Austin, 1998.
4. Monti G, Spacone E. Reinforced concrete ber beam element with bond-slip. Journal of Structural Engineering
(ASCE) 2000; 126:654–661.
5. Monti G, Filippou FC, Spacone E. Finite element for anchored bars under cyclic load reversals. Journal of
Structural Engineering (ASCE) 1997; 123:614–623.
6. Limkatanyu S, Spacone E. Reinforced concrete frame element with bond interfaces. Part I: Displacement-based,
force-based, and mixed formulations. Journal of Structural Engineering (ASCE) 2002; 128:346–355.
7. Limkatanyu S. Reinforced concrete models with bond-interfaces for the nonlinear static and dynamic analysis of
reinforced concrete frame structures. Ph.D. dissertation, Department of Civil, Environmental, and Architectural
Engineering, University of Colorado, Boulder, 2002.
8. Kent DC, Park R. Flexural members with conned concrete. Journal of the Structural Division (ASCE) 1971;
97:1964–1990.
9. Menegotto M, Pinto PE. Method of analysis for cyclically loaded reinforced concrete plane frames including
changes in geometry and inelastic behavior of elements under combined normal force and bending. IABSE
Symposium on Resistance and Ultimate Deformability of Structures Acted on by Well-De
ned Repeated
Loads, Final Report, Lisbon, 1973.
10. Eligehausen R, Popov EP, Bertero VV. Local bond stress-slip relationships of deformed bars under generalized
excitations: Experimental results and analytical model. UCB/EERC Report 83-23, Earthquake Engineering
Research Center, University of California, Berkeley, 1983.
11. Pantazopoulou SJ, Bonacci JF. On earthquake-resistant reinforced concrete frame connections. Canadian Journal
of Civil Engineering 1994; 21:307–328.
12. Taylor R. FEAP: A Finite Element Analysis Program. Users Manual: Version 7.4. Department of Civil and
Environmental Engineering, University of California, Berkeley, http://www.ce.berkeley.edu/∼rlt/feap/, 2002.
13. Beckingsale CW, Park P, Paulay T. Post elastic behavior of reinforced concrete beam-column joints. Research
Report 80-20, Department of Civil Engineering, University of Canterbury, New Zealand, 1980.
14. Scott BD, Park R, Priestley MJN. Stress–strain behavior of concrete conned by overlapping hoops at low and
high strain rates. ACI Structural Journal 1982; 79:13–27.
15. Monti G, De Sortis A, Nuti C. Problemi di Scala nella Sperimentazione Pseudodinamica di Pile da Ponte in
C.A. (in Italian). Proceedings, Workshop Danneggiamento, Prove Cicliche e Pseudodinamica, (Damage, Cyclic
Tests and Pseudodynamic Testing), Napoli, Italy, 1994.
16. Clough RW, Gidwani J. Reinforced concrete frame 2: testing and analytical correlation. UCB/EERC Report
76-15, Earthquake Engineering Research Center, University of California, Berkeley, 1976.
17. Uniform Building Code, International Conference of Building Ocials, Whittier, California, 1970.
18. ACI Committee 318, Building Code Requirements for Reinforced Concrete, American Concrete Institute,
Detroit, MI, 1971.
19. D’Ambrisi A, Filippou FC. Correlation studies on an RC frame shaking-table specimen. Earthquake Engineering
and Structural Dynamics 1997; 26:1021–1040.

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:2407–2424