NON-LINEAR ANALYSES OF REINFORCED CONCRETE ELEMENTS

WITH PRONOUNCED SLIPS OF LONGITUDINAL BARS

Franco BRAGA
Title Professor
University or Affiliation Dept. Of Structural Engineering, University “Sapienza” of Rome
Address via Eudossiana 18, 00184 Rome (Italy)
email address [email protected]

Rosario GIGLIOTTI
Title Assistant Professor
University or Affiliation Dept. Of Structural Engineering, University “Sapienza” of Rome
Address via Eudossiana 18, 00184 Rome (Italy)
email address [email protected]

Michelangelo LATERZA
Title Associate Professor
University or Affiliation Dept. Of Structural Engineering, University of Basilicata
Address C. da Macchia Romana, 85100 Potenza (Italy)
email address [email protected]

Michele D’AMATO
Title Research Fellow
University or Affiliation Dept. Of Structural Engineering, University of Basilicata
Address C. da Macchia Romana, 85100 Potenza (Italy)
email address* [email protected]

Sashi KUNNATH
Title Professor
University or Affiliation Dept. Of Civil and Enviromental Engineering, University of California Davis
Address 95616 Davis (California)
email address [email protected]

Abstract
The paper shows results of two non-linear cyclic analyses performed with OpenSees where
slippage phenomenon of longitudinal bars is pronounced. Bond-slips have modeled by using
the simplified model proposed by Braga et al. which assigns an equivalent stress-strain
relationship to longitudinal steel accounting for both material elongation and relative slip. The
model is also capable to take into account anchorages at bars ends (such as bends or hooks)
and has been developed mainly for assessing older existing buildings where, because of the
presence of reinforcing plain bars, bond-slips are particularly pronounced. Unlike many other
refined models published in literature, the analytical formulation of the proposed model
requires a minimum computational effort avoiding any nested iterations loop in the context of
fiber model discretization of a section.
Keywords: Bond-slip, concrete elements, existing buildings, non-linear analyses, smooth
bars, stress-strain relations.
1. Introduction
Slippage phenomenon may play a central role in the response of a reinforced concrete (RC)
structure when it is subjected to lateral loads. It has been demonstrated by many experimental
tests that bond-slip reduces the stiffness and hysteretic dissipation capacity of a structure and,
therefore, makes inadequate the classical assumption of full-bond of longitudinal bars.

Page 1 of 8
In older RC buildings significant slips verify because of plain bars applied and they are
amplified by poor confining action due to absence of detailing rules within elements ends, by
poor bond strength of concrete and by an insufficient lap splice of bars. Bond-slip also
regards seismic response of concrete buildings reinforced with deformed bars because of
micro-crushing of surrounding concrete owing to bar ribs.
This paper is aimed to show two analytical results of non-linear analyses performed on RC
sub-structures when bond-slips are pronounced. The simplified model proposed by Braga et
al. [1,2] has been applied for accounting bond-slips of longitudinal bars. It is capable of
modeling straight and hooked bars and of defining an equivalent stress-strain relationship
including steel elongation and relative slips with respect to the surrounding concrete.
2. Modified steel bar model accounting for bond-slips
The simplified model proposed by Braga et al. [1] has been developed on the following
assumptions (Figure 1):
- bond-slip field u(x) is linear along the bar;
- bond-stress relationship is elastic-perfectly plastic;
- any anchorage at bar end (such as bend or hook) is described by a linear function of
the displacement u0 at end.

Figure 1. a) Longitudinal bar anchored in a concrete block and b) reference scheme [1]
The second assumption made becomes particularly appropriate in the case of smooth bars.
For this kind of longitudinal reinforcements the bond strength peak sharply degrades and
reaches a constant residual value owing to the friction between concrete and smooth bar. The
constant value, moreover, is quite stable under cyclic loads [1]. As example, Figure 2
compares the assumed law with the one proposed by the ModelCode 90 [3].
Bond-stress τ

Model Code 90
Approximated law

τ=τmax(u/u1)α
τd

u1 u2 u3 axial slip u

Figure 2. Bond stress –axial slip relationship proposed by ModelCode 90 [3].

Page 2 of 8
It is demonstrated that the model uses a highly efficient approach to model bond-slip and
requires a minimum computational effort without any nested iterations loop in the context of
fiber model discretization of a section [2]. Analytical relationships for defining the curve are
summarized in tabular form in [2].
Starting from the assumptions made the proposed model provides a non-linear monotonic
tensile stress-slip relationship (σ-uL) of a longitudinal bar embedded in a concrete block
(Figure 1). The derived (σ-uL) law accounts for both steel elongation and its relative slips
with respect to concrete. Some examples of the (σ-uL) relationship by varying the bar
diameter are reported in (Figure 3).

Figure 3. Different tensile stress-slip relationships obtained by varying the bar diameter
In a non-linear analysis program requiring the material relationship in the form of stress-strain
law, the uL can be intended as the integrated axial displacement of a steel fiber along the
weighted length Li of the elements ends, where Li is the element part in the adopted
integration scheme on which the response is constant. Li may be intended as the plastic
hinging region Lpl of the element. Starting from this assumption [2]:
u L ,tot
ε= (1)
L pl
where uL,tot is the total relative axial displacement of a longitudinal bar calculated with the
proposed model, and Lpl is the plastic hinge length.
The so-derived axial strain allows us to determine, starting from a (σ-uL) law, a (σ-ε) law
useful in a fiber-section state determination which is based on a section strain distribution.
The total displacement uL,tot is given by:
uL ,tot = uL , A + uL , B (2)

where uL,A and uL,B are the relative axial displacement including the bond-slip related to
adjacent concrete blocks (Figure 4).

Figure 5 reports different (σ-uL) relationships obtained by varying the anchorage length in a
concrete block (expressed as ratio L/D).
When in both concrete blocks an adequate anchorage length is available in the two schemes
the steel yielding is reached and:

(σ , u ) = (σ , u ) = (σ , u )
L, A L, B L
(3)

Therefore, Eq. (2) becomes:

Page 3 of 8
 (σ , u ) = (σ , 2u )
L,tot L
(4)

Figure 4. a) Total relative slip at the concrete crack and b) two anchorage schemes considered.

Figure 5. Tensile stress-slip relationships obtained by varying the ratio L/D.

As far as the modeling of the hooked end is concerned, the simplified model describes any
anchorage as an elastic relationship referred to its initial section. Therefore, the stiffness of the
hook is the only one parameter required in defining hooks or bends at the bar end. The
authors propose to calculate the end stiffness kh* as the stiffness at the incipient pull-out
condition [1].
In Figure 5 are reported some comparisons regarding hooked and straight bars having the
same anchorage length ratio L/D. It is easy to note that the higher L/D ratio the closer the
curves of a straight and hooked bar with the same length. Moreover, for ratios L/D higher
than (L/D)’ the both relationships are coincident.

Figure 6. Comparisons between hooked and straight bars by varying the L/D ratio.

Page 4 of 8
3. NUMERICAL SIMULATIONS
In performing non-linear analyses with OpenSees the non-linear (σ-ε) relationship has been
applied by considering its equivalent relationship based on the energy principle. In particular,
one can apply a bilinear or a trilinear equivalent relationship by using, for example, the
uniaxial hysteretic material (uniaxialMaterial Hysteretic without any pinching factor (Figure
7).

Figure 7. Example of trilinear law defined on the energy principle and assigned to longitudinal bars.
Hereafter, results regarding two non-linear analyses performed with OpenSees and
comparisons with experimental results are shown. More in detail, simulations regard an
internal RC beam-column joint reinforced with smooth bars [4] and a RC cantilever with
deformed bars [5]. The confined concrete model proposed by Braga et al. (BGL model, [6])
has been used for the section core. The BGL model has been implemented into OpenSees
with the uniaxialMaterial interface and named ConfinedConcrete01 [7, 8].
3.1 Braga et al., Spec. C11 [4]
An internal beam-column joint (Spec. C11) reinforced with smooth bars was tested under a
vertical load of 270 kN (axial load ratio of 16 %) without P-Δ effect. The specimen was
designed only for vertical loads without any detailing rule for lateral forces in according to
designing criteria adopted in Italy during ‘60s and ‘70s. The test was performed by applying a
lateral displacement both in positive and negative direction at the top of the upper column.
More details about the experimental program and results are reported in [4].
For numerical simulations beams and columns have been modeled with BeamWithHinges
elements (Figure 8a). Detailed description of the analytical model is reported in [2]. Instead,
in Figure 8b is reported the stress-strain relationship assigned to steel of longitudinal bars
including material elongation and slippage phenomenon (red line). Slips reduce the stiffness
and delay the yielding of longitudinal bars with respect to the full-bond design assumption
(dashed line). Figure 8b also plots the equivalent trilinear relationship established with the
energy principle assigned to steel fibers by using the trilinear hysteretic material
(uniaxialMaterial Hysteretic) without any pinching factor. In according to the experimental
measurements, the plastic hinge length for all elements has been assumed equal to h/3, where
h is the depth of the section [2]. Rigid end offset at each element has been assigned for
modeling the panel region.
In Figure 9 comparisons with experimental results are reported. Analytical simulations have
been carried out by assuming the classical assumption of full-bond for longitudinal steel
(Figure 9a) and by taking into account bond-slips with the proposed model (Figure 9b).

Page 5 of 8
 Figure 8. a) Analytical model adopted and b) stress-strain relationship assigned to longitudinal bars of
column taking into account bond-slips.

Figure 9. Comparisons with experimental results by considering a) full-bond for longitudinal bars or b)
the proposed model accounting for bond-slips.
3.2 Saatcioglu and Ozcebe, Specimen U4 [5]
A cantilever was subjected to a lateral displacement at the top and to a constant axial load
(axial load ratio of 18%, Figure 10). The cantilever was reinforced with deformed bars and
failed in flexure [5].
The examined cantilever has been modeled in OpenSees with a BeamWithHinges element
whose hinge length Lp is assumed equal to the section depth h of the column. Non-linear
analyses take into account P-Δ effect by means of P-Delta Coordinate Transformation
(geomTransf PDelta) [7].
In Figure 11a the confined concrete law calculated with the BGL model [6] and assigned to
the section core is reported. Whereas, in Figure 11b the stress-strain relationship obtained
with the proposed model for accounting the bond-slips is plotted. In the analyses the Steel02
material is applied to reproduce the stress-strain for steel obtained.
In Figure 12 comparisons between experimental and analytical responses are depicted. Figure
12a shows the concrete relationship of the cover and of the section core. Stress-strain
relationship for longitudinal bars is shown in Figure 12b where is also reported the steel
material constitute law (blue dashed curve). An equivalent stress-strain relationship by using
the Steel02 material is applied in this analysis without the Bauschinger effect.

Page 6 of 8
 Figure 10. Cantilever model considered.

Figure 11. a) Confined concrete law obtained with the BGL and b) stress-strain relationship for
longitudinal steel including bond-slip.

Figure 12. Comparisons with experimental tests by considering a) full-bond for longitudinal bars or b) the
proposed model accounting for bond-slips.
4. Conclusions
The simplified model proposed by Braga et al. [1, 2] has been applied for simulating
slippages between longitudinal bars and surrounding concrete. The model has been developed
mainly for assessing RC existing structures where smooth bars have been usually applied and,
unlike many other refined models published in literature, it does not require any nested
iterations loops. With a minimum computational effort the model furnishes an equivalent steel
stress-strain relationship accounting for material elongation and relative displacement
between the longitudinal bar and concrete.
It is easy to recognize that the full-bond assumption implies fuller cycles with an
overestimation of stiffness, resistance and, therefore, of energy dissipated under lateral loads.

Page 7 of 8
The two comparisons carried out show a good agreement with the experimental results
especially when system deformations become significant and the bars slippages dominate the
global response either with plain bars or with deformed bars.
5. References
[1] Braga, F., Gigliotti, R., Laterza, M., D’Amato, M., Kunnath S., “A modified steel bar
model incorporating bond-slip for seismic assessment of concrete structures”, Journal
of Structural Engineering, ASCE (accepted for publication, in press).
[2] D’Amato, M., Braga, F., Gigliotti, R., Laterza, M., Kunnath, S., “Validation of a
modified steel bar model incorporating bond-slip for seismic assessment of concrete
structures”, Journal of Structural Engineering, ASCE, 2012 (accepted for publication, in
press).  
[3] ModelCode 1990. Design Code. Comitè Euro-International du Beton, 1991.  
[4] Braga, F., Gigliotti, R., Laterza, M., “RC existing structures with smooth bars:
experimental behavior of beam-column joints subjected to cyclic lateral loads”, The
Open Construction & Building Technology Journal, Bentham Open, 2009, 3, pp. 52-67.
[5] Saatcioglu, M., and Ozcebe, G., “Response of reinforced concrete columns to simulated
seismic loading”, Structural Journal, ACI, 1989, 86, pp. 3-12.
[6] Braga, F., Gigliotti, R., and Laterza, M., “Analytical stress-strain relationship for
concrete confined by steel stirrups and/or FRP jackets”, Journal of Structural
Engineering ASCE, 2006, 132(9), pp. 1402-1416.
[7] D’Amato, M., Braga, F., Gigliotti, R., Kunnath, S., and Laterza, M., “A numerical
general-purpose confinement model for non-linear analysis of R/C members”,
Computers and Structures Journal, 2012, (accepted for publication, in press).
[8] Braga, F., Gigliotti, R., Laterza, M., D’Amato, Kunnath, S., “The OpenSees BGL
model for non-linear analyses of confined concrete elements ”, OpenSees Days
Conference, Rome, May 24th-25th 2012.

Page 8 of 8