Bond deterioration between reinforcing bars and concrete subject

to cyclic loads and cracks

C. Mahrenholtz1, R. Eligehausen1, J. Hofmann1, and S. Pampanin2

1
University of Stuttgart, Stuttgart, Germany
2
University of Canterbury, Christchurch, New Zealand

ABSTRACT: The bond characteristic between reinforcing bars and concrete can have a
significant influence on the specific deflection and cracking of the entire structure. In case of a
seismic event, the structure and thus the main reinforcement is cyclically loaded. At the same
time, cracks along the reinforcement anchorage open and close due to the cyclic response of the
structure. Both phenomena progressively damage the bond strength. The effect of simultaneous
acting cyclic load and crack was experimentally investigated at the Laboratory of the University
of Stuttgart to determine a bond stress-slip model.

1 INTRODUCTION

1.1 Motivation

Seismically induced cyclic loading of the total structure causes a cyclic loading of the
reinforcing bars and a cyclic variation of the crack widths in the concrete. At joints of structural
elements, e.g. column-to-foundation connections, both phenomena act simultaneously. The
resulting effect is most pronounced for post-installed anchorages due to the absence of both
anchorage hooks and confining reinforcement. Figure 1 illustrates this situation schematically.
The cyclic opening and closing of cracks is referred to as cyclic cracks in the following.
(a) Load (b) Reversed load

Tension of reinforcing bar Compression of reinforcing bar

Opening of cracks Closing of cracks

Detail
Figure 1. Schematic illustration of a column-to-foundation connection with a short, straight anchorage
length typical for post-installed reinforcing bars: Cyclic loading of the longitudinal reinforcement of the
column and cyclic variation of the crack width in the concrete foundation for (a) loading stroke and (b)
reversed loading stroke.

Depending on the design code, for seismic load cases the anchorage length of a straight bar is
specified as up to 55 times the bar diameter to allow for sufficient redistribution of bond stress if
the bond is locally damaged by load and crack cycling. In practise, engineering judgement on a
case by case basis is used to justify reduced anchorage lengths for post-installed bars, often
being inevitable due to the limited concrete member depth.
The long-term aim of current research is to develop models for the load transfer mechanism in
column-to-foundation connections subject to cyclic loads and cracks in order to increase the
range of application for post-installed reinforcing bars, e.g. for seismic retrofit measures.
Numerical simulations of column-to-foundation connections subjected to seismic excitation will
help to achieve this aim. This requires an applicable and verified bond model.

1.2 Background

Earlier investigations of the influence of cyclic load on the bond, e.g. in Monti et al. (1993), has
shown that the behaviour of various reinforced concrete applications can be predicted by
numerical simulation based on the model proposed in Ciampi et al. (1982).
The bond stress-slip relation for cyclic loads on reinforcing bars is defined in Ciampi et
al. (1982) by a model as shown in Figure 2(a). The ultimate bond stress of each cycle n is
denominated τ1(n). The damage parameter
τ 1( n )
d = 1− (1)
τ 1( n =1)

is introduced to describe the damage on the bond. The magnitude of the damage parameter
depends on the energy hysteresis. It can be estimated according to Eligehausen et al. (1983) by
a2
d = 1 − e − a1 ( E / E0 ) , (2)
with the tuning parameters
a1 = 1.2 , a 2 = 1.1 . (3)
E is the dissipated energy. E0 is the energy consumed under monotonic loading which is used to
normalize the exponent. Figure 2(b) depicts the damage function schematically.

(a) τ (b)
τ+1(n=1) 1.00
Ω=1–d

Progressive
Monotonic envelope
τ +
0.80 deterioration
1(n=2) d

0.60 Deterioration
Reduced envelope depends on n,
d = 1 – τ1(n)/τ1(n=1)

0.40 smax, ∆wmax, fc', φ

smin smax s
0.20 Ω
E for τ-1(n=1)
+ E for τ-1(n=2) 0.00
+
τ 1(n=1) E0 0.0 0.5 1.0 1.5 2.0 2.5
E/E0

Figure 2. (a) Model for the bond stress-slip relation, (b) damage parameter d and reduction parameter Ω
as a function of the relative consumed energy after Eligehausen et al. (1983).

The bond deterioration is reflected in the bond stress-slip relation by a reduced bond stress-slip
curve which alters continuously during the cycling. The approach made in Ciampi et al. (1982)
is to define the reduced bond stress-slip curve as the initial bond stress-slip curve of the
monotonic reference test factorized by Ω = 1 – d. The reduction parameter Ω as a function of
the relative consumed energy E/E0 is shown schematically in Figure 2(b).
Though different testing conditions, e.g. number of cycles and peak slip values, result in
different bond deterioration per cycle, the deterioration can be described by the same function.
Thus, this approach predicts the deterioration of bond under generalized excitations for standard
bar diameters and concrete strengths. However, the bond model is not yet verified for crack
cycling. Such a situation exists in joints, e.g. column-to-foundation connections, as explained
above.

2 EXPERIMENTS

2.1 Research Approach

In reality, bond stress τ and crack width ∆w vary along the anchorage length. When analyzing
the bond stress-slip relation at element level both, bond stress τ and crack width ∆w are
approximately constant. This assumption is considered to be valid up to a bond length of five
times the reinforcing bar diameter φ , the specific embedment depth used for the tests discussed
in this paper.
The diagram in Figure 3(a) depicts crack width ∆w and bond stress τ as a function of the slip s,
reflecting the above explained idealization that the crack opens parallel to the tension loading of
the reinforcing bar and closes parallel to the compression loading of the reinforcing bar (Figure
1). To establish a base line for reference, the test program also included tests on cyclically
loaded reinforcing bars in constant cracks, yielding a diagram as shown in Figure 3(b).
These tests consisted of two phases. Phase I, cycling: The reinforcing bar was cyclically loaded
displacement-controlled by the peak slip values smax and smin = – smax. In parallel, the crack was
opened and closed cyclically between ∆wmax and ∆wmin (Figure 3(a)) or kept constant
(Figure 3(b)). Phase II, pull-out: To determine the residual load capacity, the crack was opened
to ∆wmax and the reinforcing bar pulled out monotonically.
Another test series was conducted for which the reinforcing bars were loaded monotonically
while the crack width was kept constant at ∆wmax. The load-displacement curve of this test series
is indicated as a dashed curve in Figure 3(a) and (b).
(a) τ (b) τ

smin smax s smin smax s

∆w ∆w
∆wmax ∆wmax

smin smax s smin smax s

Legend: Phase I: cycling Monotonic reference test, uncracked concrete

Phase I: pull-out Monotonic reference test, cracked concrete

Figure 3. Test scheme of cyclic load tests with (a) cyclic cracks and (b) constant cracks.
2.2 Test Program

The majority of tests were carried out using a reinforcing bar diameter φ = 16mm. However, to
determine the influence of the bar diameter, test series with bar diameters φ = 12mm and
φ = 25mm were also conducted. All reinforcing bars were cast-in concrete with a nominal
strength of fc’ = 20MPa.
Three different maximum crack widths ∆wmax were tested: A hairline crack width of
∆wmax = 0.1mm, a moderate crack width of ∆wmax = 0.4mm and an extreme crack width of
∆wmax = 0.8mm. The minimum crack width ∆wmin was defined by compressing the test member
with a stress equivalent to 15% of the concrete compressive strength. The bond strength remains
almost constant even if the confining pressure is further increased (Lowes et al. (2004)).
As peak slip smax = – smin , two multiples of the slip su at the ultimate load of the monotonic
reference test were applied, 0.5su and 1.0su, Table 1 comprises the test program.

Table 1. Test program

Number ∆wmax φ smax
Series name Test condition
of tests (mm) (mm) = – smin
w0.1-d16-f20-s0.5-cyc 2 0.1 16 0.5su
w0.1-d16-f20-s1.0-cyc 2 0.1 16 1.0su
w0.4-d16-f20-s0.5-cyc 2 0.4 16 0.5su cyclic load and
w0.4-d16-f20-s1.0-cyc 2 0.4 16 1.0su cyclic crack
w0.8-d16-f20-s0.5-cyc 2 0.8 16 0.5su opening
w0.8-d16-f20-s1.0-cyc 2 0.8 16 1.0su and closing
w0.4-d12-f20-s0.5-cyc 2 0.4 12 0.5su
w0.4-d25-f20-s0.5-cyc 2 0.4 25 0.5su
w0.1-d16-f20-s0.5-con - 0.1 16 0.5su
w0.1-d16-f20-s1.0-con 2 0.1 16 1.0su
w0.4-d16-f20-s0.5-con 2 0.4 16 0.5su
w0.4-d16-f20-s1.0-con 2 0.4 16 1.0su cyclic load and
w0.8-d16-f20-s0.5-con 2 0.8 16 0.5su constant crack
w0.8-d16-f20-s1.0-con 2 0.8 16 1.0su
w0.4-d12-f20-s0.5-con 2 0.4 12 0.5su
w0.4-d25-f20-s0.5-con - 0.4 25 0.5su
w0.4-d16-f20-mon 2 0.4 16 –
monotonic load,
w0.4-d12-f20-mon 2 0.4 12 –
constant crack
w0.4-d25-f20-mon 2 0.4 25 –

2.3 Test Setup

The bars were physically debonded at top and bottom, and cast-in in a horizontal position
(Figure 4, Detail). The debonding did not affect the stress distribution and made a free slip of
the reinforcing bar along the bond free length possible. The bonded length was always 5 φ. For a
given and constant specimen height, the top and bottom debonded length varied. For tests with a
bar diameter of 16mm, the debonded length was 5φ; for tests with a bar diameter of 12mm, the
debonded length was increased, for tests with a bar diameter of 25mm reduced accordingly.
Each concrete test member with the dimension L/W/H = 600mm/400mm/240mm was
reinforced by stirrups, and longitudinally placed high strength tie rods protruded at both ends
(Figure 4). Two pilot holes and two thin metal sheets were provided in the centre at both sides
to aid the crack formation. The tie rods were debonded at both sides of these crack inducers to
enable large cracks.
Debonding Section A-A Detail Reinforcing bar
C
Reinforcing bar

B B

5φ
Pilot holes

Crack inducers
C
Section B-B Section C-C
Longitudinal reinforcement
(high strength tie rod)

A A H
Stirrups
(standard reinforcement)
L W
Figure 4. Geometry of concrete test member.

After putting the concrete test member on a sliding support, the tie rods were connected with a
fixed bearing on one side and with the member loading actuator on the other side. Next, a
loading fixture was attached to the reinforcing bar, and the supports for the reinforcing bar
loading actuator were placed on the member. A pair of transducers measuring the actual crack
width were installed on each side face of the test member. The servo control system used the
input signal of these transducers to control the crack width. Another transducer measured the
actual reinforcing bar displacement at the unloaded end as the input signal for its displacement
control (Figure 5). A detailed description of tests with cyclic cracks can be found in
Mahrenholtz et al. (2011).

Actuator for reinforcing

bar loading
Reinforcing bar Reinforcing bar displace-
ment controlled loading and
unloading of reinforcing bar
Concrete member
Crack width controlled
loading and unloading of
Crack transducer
concrete member

Actuator for concrete

Reinforcing bar transducer member loading
Figure 5. Schematic side view of test setup.
 2.4 Results and Evaluation of Experiments

Figure 6(a) shows exemplary bond stress-slip curves for cyclic load and cyclic crack tests,
cyclic load and constant crack tests, and monotonic load and constant crack tests. Figure 6(b)
depicts the damage factor d versus normalized dissipated energy E/E0 for all tests on 16mm
reinforcing bar diameter. For better readability, only the values after the first, the fifth and the
tenth cycle are given.
In addition, the analytical descriptions of the bond deterioration of cast-in reinforcing bars
subject to load cycling for different testing conditions are drawn in Figure 6(b): The analytical
approach after Eligehausen et al. (1983) expressed by Equation (2) is indicated by a solid line.
In Simons (2007) the following tuning parameters for Equation (2) were suggested: a1 = 1.2 and
a2 = 0.5 for cyclic loads in uncracked concrete, indicated by a dash-dotted line, and a1 = 1.5 and
a2 = 0.25 for cyclic loads in cracked concrete, indicated by a dashed line.
6.0 1.0
(a) 0.9 (b)
3.0 0.8
0.7
0.0 0.6
τ [N/mm²]

0.5
d

-3.0 0.4 Eligehausen (1983)

w0.4-d16-f20-s0.5-cyc 0.3 Simons (2007), uncracked
-6.0 w0.4-d16-f20-s0.5-con 0.2 Simons (2007), cracked
0.1 Mahrenholtz (2011)
w0.4-d16-f20-s0.5-mon
-9.0 0.0
-1.5 0.0 s [mm] 1.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5
E /E 0

Figure 6. (a) Examples of measured bond stress-slip curves. (b) Damage factor d vs. normalized energy
dissipation E/E0, triangles indicate tests in crack width ∆wmax = 0.1mm, circles tests in crack width
∆wmax = 0.4mm, and squares tests in crack width ∆wmax = 0.8mm.

The following basic findings can be derived from Figure 6(b):

• The damage behaviour can be described for both tests with constant and tests with cyclic
cracks, by Equation (2) by the tuning parameters a1 = 2.5 and a2 = 1.0. The corresponding
curve is indicated as a dotted line in Figure 6(b). this approach predicts a larger damage
progression compared to the other analytical descriptions.
• The damage factor d follows roughly the same curve for all tests carried out within the tests
program given in Table 1, irrespective of the testing parameters smax, φ and ∆wmax.
• In the contrary, the bond deterioration evaluated for the tests in constant cracks carried out
within the tests program given in Table 1 and the bond deterioration predicted according to
Simons (2007) (dashed line) are different.
It may be reasonably assumed that the difference is caused by the geometry and material
properties of the specific reinforcing bars and concrete. This assumption is supported by the fact
that also the prediction of the bond damage for tests in uncracked concrete according to
Eligehausen et al. (1983) (solid line) and Simons (2007) (dash-dotted line) show a significant
difference.
In the following, the influence of crack width, peak slip and bar diameter is explained.
Figures 7(a) to (d) depicts the influence of various values for the peak slip and maximum crack
width. A comparison of Figures 7(a) and (b) with Figures 7(c) and (d) illustrates that the energy
consumption and thus the damage is larger for larger peak slips. Further, the bond damage
increases with larger crack widths. This effect is more pronounced in cyclic cracks (Figure 7(a)
and (c)) than in constant cracks (Figure 7(b) and (d)).
In general, the energy consumption is higher in cyclic cracks (Figure 7(a) and (c)) compared to
the energy consumption in constant cracks (Figure 7(b) and (d)). This demonstrates that the
bond degradation progress in cyclic cracks is more pronounced than in constant cracks.
1.0 1.0
(a) 0.9
(b)
0.9
0.8 0.8
0.7 0.7
0.6 w0.1-d16-f20-s0.5-cyc 0.6 w0.1-d16-f20-s0.5-con
0.5 w0.4-d16-f20-s0.5-cyc 0.5 w0.4-d16-f20-s0.5-con

d
d

0.4 w0.8-d16-f20-s0.5-cyc 0.4 w0.8-d16-f20-s0.5-con

0.3 Eligehausen (1983)
1983 0.3 Eligehausen (1983)
1983
0.2 Simons (2007),
2007 (cracked)
cracked 0.2 Simons (2007),
2007 (cracked)
cracked
0.1 (2011)
Mahrenholtz 2011 0.1 Mahrenholtz 2011
(2011)
0.0 0.0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5
E /E 0 E /E 0
1.0 1.0
0.9 (c) 0.9 (d)
0.8 0.8
0.7 0.7
0.6 w0.1-d16-f20-s1.0-cyc 0.6 w0.1-d16-f20-s1.0-con
0.5 w0.4-d16-f20-s1.0-cyc 0.5 w0.4-d16-f20-s1.0-con
d

0.4 w0.8-d16-f20-s1.0-cyc 0.4 w0.8-d16-f20-s1.0-con

0.3 Eligehausen (1983)
1983 0.3 Eligehausen (1983)
1983
0.2 Simons (2007),
2007 (cracked)
cracked 0.2 Simons (2007),
2007 (cracked)
cracked
0.1 Mahrenholtz 2011
(2011) 0.1 (2011)
Mahrenholtz 2011
0.0 0.0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5
E /E 0 E /E 0

Figure 7. Influence of peak slips smax and crack widths ∆wmax on damage factor d vs. normalized energy
dissipation E/E0 for tests in (a), (c) cyclic cracks and (b), (d) constant cracks.

1.0 1.0
0.9 (a) 0.9 (b)
0.8 0.8
0.7 0.7
0.6 w0.4-d16-f20-s0.5-cyc 0.6 w0.4-d16-f20-s0.5-con
0.5 w0.4-d12-f20-s0.5-cyc 0.5 w0.4-d12-f20-s0.5-con
d
d

0.4 w0.4-d25-f20-s0.5-cyc 0.4 w0.4-d25-f20-s0.5-con

0.3 Eligehausen (1983)
1983 0.3 Eligehausen (1983)
1983
0.2 Simons (2007), cracked
2007 (cracked) 0.2 Simons (2007), cracked
2007 (cracked)
0.1 Mahrenholtz 2011
(2011) 0.1 Mahrenholtz 2011
(2011)
0.0 0.0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5
E /E 0 E /E 0

Figure 8. Influence of reinforcing bar diameter φ on damage factor d vs. normalized energy dissipation
E/E0 for tests in (a) cyclic cracks and (b) constant cracks.
Figures 8(a) and (b) clearly show that the influence of reinforcing bar diameter on the damage
factor is statistically insignificant for the tested range between 12mm and 25mm. This is true for
both the bond deterioration due to cyclic load in cyclic cracks (Figure 8(a)) and the bond
deterioration due to constant cracks (Figure 8(b)). It can be assumed that also the influence of
bar diameters larger and smaller than tested so far will be small.
Further, it can be seen that the bond deterioration progresses in opening and closing cracks
faster than in constant cracks.

2.4.1 Conclusions

The cyclic load tests on reinforcing bars in cyclic cracks and constant cracks resulted in
different bond stress-slip curves (Figure 6(a)). However, the approach to describe the
deterioration of bond resistance by the hysteresis energy proved to be applicable also in case of
crack cycling which is characterized by a more accelerated damage. For both cases the same
tuning parameters a1 = 2.5 and a2 = 1.0 can be used to predict the observed damage
irrespectively to the reinforcing bar diameter, peak slip, crack width, and condition of concrete
(uncracked, constant cracked, cyclic cracked concrete).
The general applicability of the analytical approach has been shown by Eligehausen et
al. (1983) for various displacement histories and various bond conditions for standard bar
diameters and concrete strengths. According to Simons (2007), the analytical approach is also
satisfactorily predicting the bond deterioration of post-installed reinforcing bars. The tuning
parameters a1 and a2 determined therein for an epoxy mortar and a hybrid mortar have been
identical to those determined for cast-in reinforcing bars. Therefore, for standard bar diameters
and concrete strengths it is assumed that the analytical approach is generally providing
sufficiently accurate results for reinforcing bars post-installed with epoxy mortars and hybrid
mortars. This is also true if cyclic crack opening and closing accelerate the damage on the bond.
The analytical model has been implemented in a finite element program which is now used for
numerical studies of the column-to-foundation connections. For benchmarking, large scale test
specimens are currently being tested in the Structural Laboratory of the University of
Canterbury. Cast-in hooked and straight reinforcing bars, as well as post-installed reinforcing
bars are investigated.

2.4.2 References
Ciampi, V., Eligehausen, R., Bertero, V.V. and Popov, E.P. (1982) “Analytical model for concrete
anchorages of reinforcing bars under generalized excitations”, Report No. UCB/EERC 82-23,
Earthquake Engineering Research Center, University of California, Berkeley, California
Gambarova, P.G. and Rosati, G.P. (1996) “Bond and splitting in reinforced concrete: test results on bar
pull-out”, Materials and Structures, Vol. 29, 1996, 267-276
Eligehausen, R., Popov, E.P. and Bertero, V.V. (1983) “Local bond stress-slip relations of deformed bars
under generalized excitations”, Report No. UCB/EERC 83-23, Earthquake Engineering Research
Center, University of California, Berkeley, California, Oct. 1983
Lowes, L.N., Moehle, J.P. and Govindjee, S. (2004) “Concrete-Steel Bond Model for Use in Finite
Element Modeling of Reinforced Concrete Structures”, ACI Structural Journal, July 2004, V. 101,
No. 4, 501-511
Mahrenholtz, C. (2011) “Cyclic Displacement and Cyclic Crack Tests on Reinforcing Bars”, IWB,
University of Stuttgart, Stuttgart, Germany, not published
Mahrenholtz, C., Eligehausen, R. and Hofmann, J. (2011) “Testing Anchors in Cyclic Cracks”, Concrete
International, publication 2011
Monti, G., Spacone, E. and Filippou, F. (1993) “Model for anchored reinforcing bars under seismic
excitations”, Report No. UCB/EERC 93-08, Earthquake Engineering Research Center, University of
California, Berkeley, California, Dec. 1993
Simons, I. (2007) “Verbundverhalten von eingemörtelten Bewehrungsstäben (Bond behaviour of post-
installed reinforcing bars)”, Dissertation, IWB, University of Stuttgart, Stuttgart, Germany