Content uploaded by Marc Koschemann
Author content
All content in this area was uploaded by Marc Koschemann on Aug 03, 2022
Content may be subject to copyright.
INFLUENCE OF THE TEST SETUP ON LOCAL BOND BEHAVIOR OF RIBBED STEEL
BARS
Marc Koschemanna, Manfred Curbacha, Steffen Marxa
a Technische Universität Dresden, Institute of Concrete Structures, Dresden, Germany
ABSTRACT
In the past decades, there has been an increasing effort to describe the local bond behaviour with a bond
stress-slip model. Such bond models are often used in studies of long bond lengths, numerical
calculations and crack width determination. An example of this is the approach of fib Model Code 2010.
However, this approach is based on results of pull-out tests with a bond length of 5 times the bar diameter
ds, which does not adequately reflect the bond conditions in real components.
Current investigations at the Technische Universität Dresden (TUD) deal with the bond behaviour and
the bond stress distribution of ribbed steel bars in normal and high-strength concrete. The test program
includes systematic investigations of the influence of the bond length on the ultimate bond stress using
pull-out, and beam-end specimens with bond lengths from 1 ds up to 4 ds. The samples differ with regard
to the concrete cover, the transverse reinforcement as well as the orientation of the ribs within the bond
zone. Using different levels of confinement, it is shown how the development of splitting cracks and
internal stress conditions influence the bond behavior.
This article deals with the experimental and instrumental setup, test results and the procedure of
evaluation. Furthermore, it is discussed how different test setups represent the bond conditions of real
components best.
INTRODUCTION
Bond test setup and influence of bond length
Since the beginning of research on reinforced concrete, efforts have been made to quantify the bonding
effect between reinforcing bar and concrete because this affects anchorage, crack distribution and, not
least, the deformation behaviour of structural elements. To describe the bond behavior, the relationship
between the bond stress τ as the sum of the individual bond mechanisms adhesion, friction and
mechanical interlocking and the relative displacement, the so called slip s0, is used. For the experimental
determination of this relationship, bond tests are carried out, where the bond stress is the pull-out force
F related to the bond area Ab (1).
=
=
4
(1)
However, according to Lindorf [1], more than 30 parameters influence the bond behavior in the
experiment. In addition to the material and geometry properties of the concrete and the bar as well as
the type of loading, also the test setup or specimen play a major role in this. Thus, different test
specimens result in different bond stress-slip relationships. The suitability of test setups to investigating
bond behaviour has been the subject of controversial discussion since the beginning of the 20th century
[2].
The most used test to evaluate the bond behaviour of steel reinforcement is the pull-out test (PO)
according to RILEM [3]. It is known for the comparatively simple setup and easy way of specimen
production. According to Alvarez [4], the determination of bond stress-slip relationships in pull-out tests
is based on the notion that the pull-out body can reproduce the conditions at the infinitesimal bond
element and that the influences of the test specimen and loading geometry are eliminated. The
determined relationship would thus represent a kind of fundamental material law of bond, which could
be modified in a further step to suit the conditions in real components. However, in reality the bond
stress-slip relationship determined in the pull-out test is strongly dependent on the specific condition of
the test specimen, and it is not possible to transfer this principles to the conditions in structural elements
without further ado.
The main weaknesses of the RILEM pull-out test are the large concrete cover and the arch effect of the
reacting forces (Figure 1 left). This effect causes a transverse pressure in the bond zone, what increases
the bond resistance and subsequently leads to an overestimation of the bearing capacity. However, the
load capacity increase cannot been quantified by a study so far. Due to the minimum edge length of the
test specimen of at least 200 mm or 10.0 times the bar diameter ds, the minimum concrete cover is at
least 90 mm, respectively ≥ 4.5·ds. The large concrete cover represents a high level of confinement,
which is necessary to achieve the highest possible bond resistance, the so called pull-out failure. For real
structural elements, the concrete cover is usually in the range between 20 to 55 mm. Depending on the
bar diameter, this normally corresponds to 1.0 to 4.0·ds. According to Vandewalle [5], a concrete cover
of 2.5 to 3.5·ds is required to achieve a failure by bar pull-out. Therefore, the concrete cover of most of
the structural elements does not comply this criterion and the reinforcement has an insufficient level of
confinement to reach the maximal bond resistance, which can be achieved by pull-out tests.
Figure 1 - Schemes of pull-out test (left) and of beam-end test (middle) and modes of bond failure (right)
The lower concrete cover leads to a change of the failure mode towards a more or less sudden splitting
failure. Depending on the transverse reinforcement, both failure modes appear in combination (Figure
1 right), which may be defined as “splitting-induced pull-out failure”. Anyway, an insufficient level of
confinement leads to a premature failure and a reduction of the bond resistance. However, the pull-out
test is suitable for testing the influence of individual material parameters such as concretes strength and
rib geometry of the bar. The finding that reliable bond stress-slip relationships can only be determined
under boundary conditions adapted to the conditions in the component, led to the development of
numerous types of test specimens.
In contrast, test configurations such as the beam test and the beam-end test (BE) represent the bond
conditions of real RC components in a better way (Figure 1 middle). The force flow within the specimen
does not cause any transverse stresses in the bond zone and the concrete cover is adjustable. However,
the associated guidelines and standards specify a bond length of 10 ds instead of 5 ds in the pull-out test
[4], [6]. The comparison of results of bond tests with differently specified specimen types therefore
includes the influence of different bond lengths [7].
As early as 1905, Bach [8] determined the decrease in the length-related bond resistance τ with
increasing bond length on the basis of tests with plain steel bars. Bach saw the cause of this behavior in
the elasticity of the pull-out bar. As the tensile force decreases, the strain along the embedded bar also
decreases. The assumption of a uniform distribution of the bond stress along the bond length is therefore
not applicable for longer bond lengths. Mörsch [9] found a plausible explanation for this phenomenon.
Figure 2 middle shows the tension stress σs of the bar and the bond stress distribution along the bond
length for different load stages. The local bond stress maximum shifts from the loaded to the unloaded
end of the bond length as further the load reaches the bond resistance of the embedded bar. Close to the
load limit (Stage III), the load is mainly transmitted by the rear part of the bond length. Areas closer to
the loaded end are already damaged, but can still transfer minor load due to friction between the bar and
concrete. Depending on the local slip, each point within the bond length is at a different bond stress
state. Consequently, the mean value calculated under the assumption of a uniform distribution is smaller
than the local maximum. Therefore the average bond stress decreases with increasing bond length.
Figure 2 - Typical bond stress to slip curve (left), bond stress distributions at different loading stages
according [9], taken from [1] (middle) and results derived from strain gauges by Mains [10] (right)
These findings were confirmed by the investigations of Mains for ribbed bars [10]. By means of strain
gauges distributed along the pull-out bar of long test specimens, Mains could reveal the non-uniform
distribution of bond stresses for different load levels. From then on, bond tests with short bond lengths
are usually carried out to describe local failure criteria for the bond between reinforcement and concrete.
Bond models and specifics of tests with short bond length
According to Model Code 2010 [11] it is possible to consider an average local bond to local slip
relationship for short bond lengths. For the definition of a short bond length lb the magnitude of lb ≤ 5 ds
has been established in scientific practice, which is in accordance with the standard for the pull-out test
by RILEM [3]. Nevertheless, the experimental results of [12] and [13], among others, indicate that the
length-related bond stress continues to increase even for shorter bond lengths. Therefore, for a bond
length of 5 ds, the bond stress is not uniformly distributed along the bar and the bond behavior cannot
be considered local.
The Model Code 2010 provides a mean bond-stress to slip relationship, which is based on the research
by Eligehausen et al. [14]. As a conclusion from the results of 125 pull-out tests with a bond length lb =
5·ds and two normal strength concretes they introduced recommendations for a bond model for different
boundary conditions. In contrast to this, the investigations of Huang & al. [15] revealed much higher
bond stresses than expected by the bond model of MC2010. Based on their results of 28 pull-out tests
with a bond length of 2,5·ds, they introduced a linear approach to describe the maximal bond stress in
dependence on the concrete strength in case of pull-out failure for normal and high strength concrete.
Figure 3 shows the equations and parameters of both models for good bond conditions and well confined
concrete.
()=
(
);
<
<
<
Parameter
Model Code
Huang et al.
NSC
HSC
s1 [mm]
1.0
1.0
0.5
s2 [mm]
3.0
3.0
1.5
s3 [mm]
cli
cli
cli
α
0.4
0.4
0.3
τ
ult
2.5
0.45·fcm 0.45·fcm
τ
f
/τ
ult
0.40
0.40
0.40
Figure 3 - Equations and parameters of the local bond-slip relationships by [11] and [15]
When carrying out tests with short bond lengths, some effects have to be taken into account, which are
usually not relevant for bond lengths of 5 ds and even more. In general, for short bond lengths, different
arrangements and imperfections in the bond zone have a bigger impact on the results, which is mostly
reflected in a larger scatter.
1200
1000
800
600
400
200
0
0 1 35 7 9 11 13 15 17 19 21
Position [inch]
Bond stress [psi]
F
Figure 4 - Bond stress-slip curves for different rib arrangements [16]
By means of numerical simulations, Zobel [16] shows that the exact position of the ribs within the bond
zone have considerable influence on the results of tests with short bond lengths. Depending on the
arrangement, two or three ribs are involved in the force transmission for a bond length of 2 ds. Therefore,
the ultimate bond stress can vary up to 25% from the average (Figure 4), which covers the scatter of the
experimental results from [19] well.
Another influencing factor is the diameter of the bond breaker. Usually plastic sleeves are used to define
the bond length within the specimen and to ensure a free pre-length of at least 5 ds. However, the plastic
sleeve is also an interference for the load transfer, which leads to an early cone-typed failure of the first
concrete key. Zobel [16] investigated the influence of the ratio of diameters between bond breaker and
bar dh/ds with the result, that an increase of this ratio causes a decrease in bond strength (Figure 5). As
bigger the diameter of the bond breaker is, as further in reaches the failure cone and pre-damages the
bonded zone. This affects the test results with very short bond lengths more than for longer bond lengths.
Figure 5 - Bond stress-slip curves for different ratios dh/ds [16] (left) and bond zone after pull-out failure
(right, © Marc Koschemann)
EXPERIMENTAL METHODS
General
Past and current investigations at the Technische Universität Dresden are dealing with the bond behavior
under cyclic loading and long-term loading. Within the scope of consecutive two projects, about 300
individual bond tests are carried out, with two-thirds were already executed. Nevertheless, in order to
evaluate the results under various loading scenarios, precise knowledge of the static bond behavior is
required. Accordingly, tests on the influence of the specimen, the bond length and the arrangement of
the bond zone have been and are being carried out. In the following, the findings of tests with monotonic
loading and different test configurations carried out so far are presented and discussed. More information
about other loading scenarios can be find in [17, 18].
Material properties and test setups
Different kinds of concrete are used, reaching from a mean uniaxial strength of fcm = 30 MPa up to fcm =
120 MPa. In the test carried out so far, three normal strength concrete (NSC, named C20, C35 and C40)
and two high strength concretes (HSC, named C80 and C120) were used. These two are self-compacting
concretes, whereas the others must be compacted. All kinds of concrete have a maximum grain size of
16 mm. The compressive strength and splitting tensile strength were determined on 10 cm cubes. The
cylinder compressive strength was calculated with a conversion factor, which was determined in advance
for each concrete using 6 standard cylinders (d =15 cm). Table 1 shows the properties of the concretes
after a minimum test age of 28 days.
Table 1 - Concrete properties for an age of at least 28 days
Type
f
c,cube100
[MPa]
cal. f
c,cyl
[MPa]
f
ct,sp,cube100
[MPa]
f
c,cube100
/f
ct,sp,cube100
[–]
E-Modul
[MPa]
C20
31.2
29.7
2.9
10.9
33200
C35
67.3
54.7
4.4
15.3
38700
C40
56.6
49.2
3.9
14.5
34000
C80
111.9
95.6
5.5
20.3
40800
C120
137.7
119.7
7.1
19.4
51900
The nominal bar diameter ds is 16 mm in all tests and the pull-out bar is made of B500 B (yield strength
fyk = 500 MPa). The bars were provided with a pressed-on sleeve to apply the loading. The bond breakers
are plastic tubes, with a usual diameter of dh = 25 mm, what means a ratio of dh/ds = 1.56. In some tests
smaller or bigger diameters dh are used to verify the findings by Zobel [16].
The predominant number of experiments were conducted with the beam-end test. To compare the
performance of this test setup classical pull-out test were carried out, too. The pull-out tests were
configured according to the specifications from RILEM [3] with 200 mm cubes and a cover of 92 mm
(c = 5.75·ds), but with a lead length of 120 mm. The standard bond length was set to lb = 2·ds, but
different bond length from 1·ds up to 10·ds were tested too. Unlike the pull-out test, there is no defined
standard for the beam-end test and the specimen dimensions. The ASTM Guideline A944-10 [20]
provides some guidance on the principles of this test setup, but recommends dimensions that are
inappropriate for the specific purpose of this examination. The used beam-end specimen configuration
was designed in dependence on the guideline, but was also modified. Figure 6 shows the chosen
geometry and standard configuration for lb = 2·ds.
Figure 6 – Standard configuration of beam-end specimen for lb = 2·ds.
The concrete cover was set to 2·ds or 32 mm, what is within the range of construction practice range. To
prevent sudden failure due to splitting, transverse reinforcement in the form of two Ø6 stirrups was
placed within the bond zone. Another two stirrups were positioned at both ends of the specimen. For
some tests no stirrups or a bigger cover was used.
The two longitudinal bars Ø12 ensure the load transfer within the specimen in case of transverse
cracking. The two Ø8 bars are used exclusively for assembly purposes of the stirrups. In addition, the
standard arrangement of the ribs was carried out according to the detail in Figure 4, with ribs facing
towards the sides and compression struts towards the center of the specimen. In individual test series,
the orientation or the position of the ribs was changed by rotation or shifting of the bar. It remains to be
said that with the beam-end test there are considerably more variation possible and thus more influences
can be examined. Table 2 gives an overview about the monotonic loaded tests carried out so far.
Table 2 - Test program for monotonic tests (lb and c in relation to ds)
Configuration
c= 2.0 (BE)/5.75 (PO); lb=
lb=2.0; c=
lb=2.0
lb=2.0; rotated
lb=2.0
Σ
Concrete
Test
1.0
2.0
2.5
3.0
4.0
3.0
4.0
no stir.
90°
180°
shifted
C20
BE
-
3
-
-
-
-
-
-
-
-
-
3
PO
-
3
-
-
-
-
-
-
-
-
-
3
C35
BE
4
3
-
-
-
7
PO
4
3
-
-
-
7
C40
BE
4
8
-
3
4
3
3
3
3
3
3
37
PO
-
3
-
-
3
-
-
-
-
-
-
6
C80
BE
4
18
-
3
3
3
3
3
3
3
3
46
PO
4
6
-
3
3
-
-
-
-
-
-
16
C120
BE
3
13
2
3
3
3
7
3
3
3
3
46
PO
4
6
-
3
-
-
-
-
-
-
-
13
The specimens are made in series of 3 or 4 samples out of the same batch. The bar is placed in horizontal
position during casting and the beam-end specimens are concreted upside-down to ensure good bonding
conditions. After demolding the samples are covered with moist cloths for 6 days and then stored indoor
until testing. The monotonic tests are executed path-controlled and are also used as reference for tests
with cyclic and long-term loading. By default, the slip at the unloaded end of the bar is measured by
contact-free displacement transducers (LVDT). For beam-end tests another LVDT is attached at the
loaded end and two more LVDTs are placed right above the bond zone, one longitudinal and one in
transverse direction, to record the growth of cracks on the upper surface. The measurement distance is
100 mm for both of them. The force is measured with a load cell, which is connected with the hydraulic
cylinder. All signals were sampled with a rate of 5 Hz. In addition, distributed optical fiber sensors have
already been used in some tests and will be used in all upcoming series. More information about this
and the test stand can be found in [17, 18]. Further investigation are conducted on tensile tie specimens
with an embedment length of 60·ds and used to compare the bond behavior for short and long bond
length as well as crack growth under long term loading.
RESULTS AND DISCUSSION
Influence of bond length
The results and the evaluation of the tests with different bond lengths are provided in [18]. In BE tests,
a doubling of the bond length from 2 to 4·ds was observed to reduce the bond strength by approx. 30%.
With the PO setup, this effect is less significant. The reason for this are probably the higher absolute
pull-out forces for longer bond lengths, which increase the splitting effect and lead to larger crack widths
in BE specimens. After the splitting is occurred, only the two stirrups ensure a certain level of
confinement, which depends on the ring tensile forces. In other words, the selected BE configuration
has a certain resistance to splitting and crack growth independent of the bond length. Therefore,
statements about the performance of PO and BE tests must always be made under consideration of the
bond length. Thus, for the bond length of lb = 1·ds, quasi no differences in bond behavior were observed
between PO and BE tests, because there was no cover cracking of the BE samples.
Influence of specimen and concrete strength
Looking at the experimental results for monotonic loading and a bond length of lb = 2·ds (Figure 7 left),
it can first be seen that an increase of concrete compressive strength results in higher bond strength. All
test results are above the approach according to MC 2010 [11]. On the other hand, there is good
agreement with the linear approach according to [15], although this applies especially to the PO setup.
For the two high-strength concretes and the C35, lower bond strengths were achieved with the BE
specimen. To separate the influence of the specimen, the average bond stress-slip curves were related to
a respective average concrete strength using (Table 1) the approach according to [15] for each concrete
grade. Comparing these courses (Figure 7 middle), no significant influence of specimen type can be
observed for all concretes up to a slip of about 0.1 mm, except the C40. The results of the PO tests for
the C40 are below those of the BE specimen, and the rising branch is less steep. It can be assumed that
the concrete was not properly compacted or that the bond was pre-damaged. However, up to a slip of
0.1 mm the BE specimens provide a sufficient confinement effect due to the cover of 2·ds, whereby the
bonding conditions are comparable to those in the PO specimen. The further increase in slip was
especially for both HSC and the C35 associated with damage to the surrounding concrete in the BE
specimens, which was followed by transverse and longitudinal cracking (Figure 7 right) and a rapid
decrease in bond resistance (Figure 7 middle). The behavior can be categorized as splitting induced pull-
out failure. In the PO tests, failure occurred due to shearing the concrete between the ribs, which was
documented with the typical bond stress-slip curves with plateau part.
Figure 7 - Relationship of bond strength τ
ult
to compressive strength f
cm
(left), average curves of the bond
stress
-
slip relationships (middle) and longitudinal crack width to slip curves (right); in brackets:
Number of individual tests; * bond stress in respect to the average strength for each concrete
For the C20 concrete, the curves for the PO and BE tests differ almost not at all. On the surface of the
samples no cracks were detected. Also for the C35 and the C40, the crack development was usually not
visible on the surface, but a much larger deformations wl was measured compared to the C20. The type
of failure is characterized by the slow decrease of the bond stress-slip curve after exceeding the
maximum. The reason for the different behavior towards the HSCs is most likely due to the higher
relative tensile strength of the NSCs (Table 1). Thus, it can be noted that the influence of the specimen
type is not equal for all concretes, and the tensile strength of the concrete is a significant factor in the
level of confinement.
Level of confinement
One argument used against the pull-out test is that the stress state in the specimen does not correspond
to that in a real component and that internal transverse pressure increases the bond effect. To verify and
quantify this hypothesis, BE tests with different levels of confinement were conducted. Table 3 and
Figure 8 gives an overview of the results for different configurations. The results for each concrete grade
are related to an associated average concrete strength (see Table 1).
Table 3 – Mean values for tests with different level of confinement
Measured variable
PO
BE
BE
BE
BE
config-
uration
cover
[ds]
5.75
4
3
2
2
stirrups
[-]
0
2
2
2
0
C40
τ0.10
[MPa]
12.5
18.4
17.8
18.5
18.0
τult
[MPa]
21.4
26.1
26.1
24.6
23.6
s0,ult
[mm]
0.74
0.79
0.65
0.50
0.45
C80
τ0.10
[MPa]
32.1
36.7
35.8
33.1
31.0
τ
ult
[MPa]
41.5
40.6
39.6
37.4
34.0
s0,ult
[mm]
0.71
0.28
0.26
0.25
0.24
C120
τ0.10
[MPa]
40.7
38.4
37.6
38.2
41.8
τult
[MPa]
52.1
50.2
50.3
44.0
47.2
s0,ult
[mm]
0.96
0.60
0.46
0.28
0.21
0
10
20
30
40
50
60
Bond strength τult [MPa]
Mean compressive strenght fcm [MPa]
fib MC2010 [11]
Huang et al. [15]
0 30 60 90 120 150
0
10
20
30
40
50
60
Bond stress τ*[MPa]
Slip s
0
[mm]
0 0.4 0.8 1.2 1.6 2.0
Longitudinal crack width wl[mm]
Slip s0[mm]
0 0.2 0.4 0.6 0.8 1.0
0.10
0.08
0.06
0.04
0.02
0.00
C20 - PO(3) C35 - PO(3) C40 - PO(3) C80 - PO (12) C120 - PO (6)
C20 - BE(3) C35 - BE(3) C40 - BE (8) C80 - BE (18) C120 - BE (11)
Figure 8 – Bond stress-slip curves for different levels of confinement and different types of concrete
Comparing first the τ0.10-values for the individual configurations, the results for the C120 and the C40,
with the exception of the PO tests, show no influence by the concrete cover. The results for the C80
show a slightly larger scatter, with mainly the mean value for a cover c = 4·ds being noticeable. With
further slip increase, the specimens with a c = 4·ds experience damage or cracking in the surrounding
concrete, which results in premature failure with small values for s0.ult. Especially for the samples without
stirrups, a significant decrease in bond resistance was observed after cracking, with C40 again standing
out due to its comparatively high tensile strength. The bond strength τult differs only slightly for a cover
of 3·ds and 4·ds for all concretes, which indicates a sufficient confinement effect for both cases. In all
these samples, no cracks were found on the surface or only minimal deformations were measured on
those. If the values τult of these BE configurations are now compared with those from the PO tests, an
increase in bond strength of approx. 3 to 5% is obtained for the concretes C80 and C120. Therefore, this
corresponds to the increase that occurs with the PO test setup due to the internal transverse pressure.
Orientation of ribs within the bond zone
With the BE test setup, there is also the question of how the ribs of the bar are oriented in relation to the
outside of the specimen. To investigate this aspect, the pull-out bar was rotated in this way that the
compression struts of the ribs point towards a side face (90°) or outwards (180°). In addition, in
individual series the position of the ribs within the bond zone was shifted by half rib spacing. Table 4
and Figure 9 gives an overview of the results for different configurations.
Table 4 – Mean values for tests with different rib orientation and arrangement
Measured variable
BE
BE
BE
BE
configu-
ration
rotation
[°]
0
90
180
0
shift
[-]
no
no
no
yes
C40
τ0.10
[MPa]
18.5
19.8
17.2
15.2
τult
[MPa]
24.6
25.1
23.4
18.8
s0,ult
[mm]
0.50
0.44
0.50
0.4
C80
τ0.10
[MPa]
33.1
32.0
28.8
34.3
τult
[MPa]
37.4
36.0
34.6
36.4
s0,ult
[mm]
0.25
0.30
0.28
0.20
C120
τ0.10
[MPa]
38.2
37.8
37.6
42.4
τult
[MPa]
44.0
45.1
42.5
46.4
s0,ult
[mm]
0.28
0.28
0.28
0.24
0
5
10
15
20
25
30
Bond stress τ[MPa]
Slip s
0
[mm]
0 0.4 0.8 1.2 1.6 2.0
C40
0
8
16
24
32
40
48
Bond stress τ[MPa]
Slip s
0
[mm]
0 0.4 0.8 1.2 1.6 2.0
C80
0
10
20
30
40
50
60
Bond stress τ[MPa]
Slip s
0
[mm]
0 0.4 0.8 1.2 1.6 2.0
C120
PO BE c= 4ds BE c= 3ds BE c= 2ds BE c= 2ds & no stirrups
BE c = 4·d
s
BE c = 3·d
s
BE c = 2·d
s
BE c = 2·d
s
& no stirrups
25
30
35
40
45
50
Bond stress τ[MPa]
Slip s
0
[mm]
0 0.2 0.4 0.6 0.8 1.0
C120
Figure 9 – Bond stress-slip curves for different rib arrangements and different types of concrete
The results obtained with 90° rotation of the bar do not show a clear trend. Slightly higher bond strengths
were achieved for the C40 and C120 compared to the standard configuration, and lower ones for the
C80. However, if the compression struts point outward (180° rotation), a decrease in bond resistance
was observed for all concretes. On average, the strengths achieved are about 5% below those of the
standard configuration, with failure occurring at similar slip values. The compression struts cause
splitting forces transverse to the bar, which in the standard configuration occur on the side of the bar
opposite to the cover side (inside the specimen). If the bar is rotated 180°, the splitting forces act in the
concrete cover and thus apparently reduce the confinement effect.
In the case of the shifted rib arrangement, quite different results were obtained for the three concretes.
For the C40, a significantly lower bond strength was achieved, suggesting that these specimens were
not optimally compacted. This is because high pull-out forces were achieved for both the C80 and C120,
some of which exceeded those of the standard configuration. In addition, an increase of the τ0.10-values
and a decrease in the s0.ult-values were measured for both HSCs. Consequently, this arrangement leads
to a higher bond stiffness, which is probably due to the better participation of the first rib. Because the
first rib is slightly further away from the front bond breaker, the compression struts can form better and
are not affected by the bond breaker as in the standard arrangement.
CONCLUSION
In this article, recent investigations on the influence of test setup, concrete cover and rib arrangement
on the bond behavior of ribbed steel bars were presented. Based on a literature review and the results of
over 180 individual tests under monotonic load, it is shown how the level of confinement and the
arrangement of the ribs within the bond zone affect the bond behavior in experimental investigations.
For this purpose, tests were carried out on pull out and beam end specimens. Up to a slip of s0 = 0.1 mm,
no influence of the specimen type on the bond performance was observed. Larger slip values usually
caused splitting cracks in the beam-end tests, which resulted in lower bond strengths. However, the
relative tensile strength plays an important role in this, which is higher for low strength concretes and
thus no cracking was observed with these. In comparison to beam-end specimens with the same level of
confinement, pull-out tests achieved approximately 3 to 5% higher bond strengths, which may be
explained by the internal transverse pressure in this test setup. In addition, the test results showed that
the orientation and position of the ribs within the bond zone also influence both the stiffness and strength
of the bond, especially for short bond length. With unfavorable rib orientation, a 7% reduction in bond
resistance was observed compared to the standard configuration.
In summary, it was demonstrated up to which point the pull-out test covers the bond conditions in real
components and what influence internal stress states in the specimens have on the bond behavior.
Furthermore, it was shown that seemingly minor changes in the setup of tests with short bond lengths
significantly affect the results.
12
15
18
21
24
27
Bond stress τ[MPa]
Slip s0[mm]
0 0.2 0.4 0.6 0.8 1.0
C40
20
24
28
32
36
40
Bond stress τ[MPa]
Slip s
0
[mm]
C80
0 0.2 0.4 0.6 0.8 1.0
BE c= 4dsg ggg g BE c= 3ds ggggg BE c= 2dsgg ggg BE c= 2ds & no
BE standard config.
BE rotation 90°
BE rotation 180°
BE shifted
ACKNOWLEDGEMENT
The presented studies are funded by the German Federal Ministry of Economic Affairs and Climate
Action (BMWi, project No. 0324016B and project No. 1501601). In addition, a thank goes to the Otto-
Mohr-Laboratory, Technische Universität Dresden, for carrying out the test and the good cooperation.
REFERENCES
[1] Lindorf, A.: Ermüdung des Verbundes von Stahlbeton unter Querzug. Doctoral Thesis (2011),
Technische Universität Dresden.
[2] Cairns, J.; Plizzari, G. A.: Towards a harmonized European bond test. Materials and Structures
36(262) (2003), p. 498–506.
[3] RILEM: Essais portant sur l’adhérence des armatures du béton.” Matériaux et Constructions 3
(1970), p. 169–178.
[4] Alvarez, M.: Einfluss des Verbundverhaltens auf das Verformungsvermögen von Stahlbeton.
Doctoral Thesis (1998), Eidgenössische Technische Hochschule Zürich, IBK Bericht 236
[5] Vandevalle, L.: Theoretical prediction of the ultimate bond strength between a reinforcement bar
and concrete. Proceedings of Bond in concrete: International conference bond in concrete from
research to practice. Vol. 1, Riga, Latvia (1992), October 15th – 17th, pp. 1/1-1/8
[6] ASTM international: „ASTM A944-10 - Standard test method for comparing bond strength of
steel reinforceing bars to concrete using beam-end specimens.” West Conshohocken (2015)
[7] Wildermuth, A.: Untersuchung zum Verbundverhalten von Bewehrungsstäben mittels
vereinfachter Versuchskörper. Deutscher Ausschuss für Stahlbeton 609 (2013), Beuth, Berlin
[8] Bach, C.: Versuche über den Gleitwiderstand einbetonierten Eisens. Mitteilungen über
Forschungsarbeiten 22 (1905), p. 1–41.
[9] Mörsch, E.: Der Eisenbetonbau. Stuttgart (1920), Konrad Wittwer
[10] Mains, R. M.: Measurement of the distribution of tensile and bond stresses along reinforcing bars.
Journal of the American Concrete Institute 23 (1951), p. 225–252.
[11] fib Bulletin 65: Model Code 2010 – Final draft, Volume 1. Federation international du béton (fib),
Lausanne (2012)
[12] Müller, F. P.; Eisenbiegler, W.: Ermittlung der Verbundspannungen an gedrückten einbetonierten
Betonstählen. Deutscher Ausschuß für Stahlbeton, Vol. 319, Beuth Verlag (1981), Berlin
[13] Rostasy, F. S.; Scheuermann, J.: Verbundverhalten einbetonierten Betonrippenstahls bei extrem
tiefer Temperatur“, Deutscher Ausschuß für Stahlbeton, Vol. 380, pp. 43–105, Beuth Verlag
(1987), Berlin
[14] Eligehausen, R.; Popov, E. P.; Bertero, V. V.: “Local bond stress-slip relationships of deformed
bars under generalized excitations”, Earthquake Engineering Center, University of California
(1983), Berkeley
[15] Huang, Z.; Engström, B.; Magnusson, J.: Experimental and analytical studies of the bond
behaviour of deformed bars in high strength concrete. 4th International Symposium on Utilization
of High-strength/High-performance Concrete, De Larrard, F. (Ed.), Proceedings of Symposium,
Paris, France (1996), May 29th – 31st, pp. 1115–1124
[16] Zobel, R.: Verbundmodellierung von Beton- und Spannstahl unter Querzug. Doctor Thesis
(2018), Technische Universität Dresden.
[17] Koschemann, M., Kühn, T., Speck, K. and Curbach, M.: Bond behaviour of reinforced concrete
under high cycle fatigue pull-out loading. Foster, F.; Gilbert, R.; Mendis, P.; Al-Mahaidi, R.;
Millar, D. (Eds.), Proceedings of 5th international fib Congress, Melbourne, Australia (2018),
November 7th –11th, p. 2834–2847.
[18] Koschemann, M. and Curbach, M.: Bond stress distribution of ribbed steel bars in reinforced
concrete with short bond length under various loading conditions. Gatuingt, F. and Torrenti, J.-
M. (Eds.), 13th fib International PhD Symposium in Civil Engineering, Proceedings of
Symposium – 2021 Session, Paris, France (2021), July 21st – 22nd, p. 176–183.
[19] Ritter, L.: Der Einfluss von Querzug auf den Verbund zwischen Beton und Betonstahl. Doctoral
Thesis (2014), Technische Universität Dresden.
[20] ASTM international: ASTM A944-10 - Standard test method for comparing bond strength of steel
reinforceing bars to concrete using beam-end specimens. West Conshohocken (2015)
