Professional Documents
Culture Documents
Bond of Deformed Reinforcing Steel Bars Embedded in Steel Fiber
Bond of Deformed Reinforcing Steel Bars Embedded in Steel Fiber
Number of pages : 65
Number of tables : 66
Number of figures : 44
Number of appendices : -
Keverling Buismanweg 4
Postbus 69
2600 AB Delft
015-2693793
015-2693799
info@delftcluster.nl
www.delftcluster.nl
Tijdens de uitvoering van het project bestond de Delft Cluster-groep van thema Grond en
Constructie uit:
Naam Organisatie
Thema Trekker dr. P. van den Berg GeoDelft
Naam Organisatie
dr.ir. A.J. Bigaj-van Vliet TNO Bouw
dr.ir. C.R. Braam TU Delft, CITG
Dipl.-Ing. P. Schumacher TU Delft, CITG
dr.ir. A.H.J.M. Vervuurt TNO Bouw
Prof.dr.ir. J.C. Walraven TU Delft, CITG
1 INTRODUCTION..............................................................................................7
2.1 Pull-out tests with short embedment length (RILEM Standard RC6 or similar
test set-up) ................................................................................................................... 12
2.1.1 S. Cattaneo [6], S. Cattaneo & G. Rosati [7] ...................................................12
2.1.2 F.S. Rostásy & K.Hartwig [27]........................................................................15
2.1.3 M.H. Harajli [12] .............................................................................................18
2.1.4 M.H. Harajli, M. Hout & W. Jalkh [14]...........................................................21
2.1.5 F. De Bonte [9] ................................................................................................25
2.1.6 P. Soroushian, F. Mirza & A. Alhozaimy [32] ................................................28
2.1.7 G. A. Plizzari [25]............................................................................................31
2.1.8 S. Hota & A.E. Naaman [16] ...........................................................................34
2.2 Pull-out tests with short embedment length (Danish Standard DS2082 or similar
test set-up) ................................................................................................................... 38
2.2.1 A. Samen Ezeldin, P.N. Balaguru [30], [31]....................................................38
2.2.2 N. Krstulovic-Opara, K.A. Watson & J.M. LaFave [19] .................................41
2.2.3 C.V. Nielsen, J.F., Olesen & B.K. Aarup [22], B. Aarup & B. Chr. Jensen [1],
B. Aarup, J. Karlsen & G. Lindström [2] ........................................................45
2.2.4 G. Heshe & C.V. Nielsen [27], C.V. Nielsen [21]...........................................48
2.4 Beam tests (RILEM Standard RC5 [28] or similar test set-up) ............................. 58
2.4.1 M.H. Harajli [12] .............................................................................................58
5 REFERENCES ..............................................................................................64
The use of steel fiber reinforced concrete (SFRC) in civil engineering structures has been
widely discussed in recent years. In the early stage of development it was hoped that the
tensile strength of the concrete could be increased significantly through the addition of fibers.
However, only an insignificant increase in strength could be achieved with the amounts of
fibers customarily used in practice. The load bearing capacity of reinforced and prestressed
concrete proved not to be increased markedly by steel fiber addition within the practical range
i.e. a volume fraction Vf ≤ 3%. With fibers, however, the behavior of solid structural
components was found to be improved as far as ductility is concerned, in particular in the
zones where tensile stresses occur in the concrete. After initiation of cracks, the crack-
arresting effect of fibers provides a redistribution of the stresses within the structure and
consequently, the overall structural toughness is enhanced. Furthermore, fibers durably and
noticeably improve the behavior under service loads: addition of steel fibers significantly
reduces both deformation and crack width. Note, that when using the Eurocode 2, the amount
of steel reinforcement is to a high degree determined by the serviceability limit state (SLS)
checks [34]. Another major - and primarily industrial – incentive in using fibers is to reduce
the production costs by shortening the construction time and, where possible, reducing the
amount of conventional reinforcement.
An overview of the practical applications of SFRC shows that depending on the type of
structure, the use of the steel fibers can either reduce the required amount of conventional steel
reinforcement or in some cases replace it altogether, while maintaining satisfactory
performance of the structure [33]. Steel-concrete composite floors, reinforced concrete floors
supported by columns or walls and floors on an elastic foundations belong to the category of
structural elements in which the conventional steel reinforcement can be partially replaced by
the use of SFRC. In these cases the use of steel fibers is intended to reduce opening of creep
and shrinkage cracks and to increase the speed of construction works. Steel reinforcement is
still needed there to guarantee sufficient deformation capacity and load carrying capacity at the
supports. Besides the traditional use of fibers for controlling cracks in e.g. slabs and toppings,
examples can be given of fiber application for load-bearing purposes. Research affirms the
possibility of using fibers for structural repairs, ductile beam-column connections [32] or shear
reinforcement, e.g. in order to replace conventional (web) reinforcement in I-shaped girders
[23]. Also in case of prefabricated tunnels, it is possible to eliminate conventional (bending)
reinforcement if SFRC is used, provided that the bending moments remain low. However,
with respect to force distribution in tunnel structures it is important to note that, under some
geological circumstances or exceptional loading situations, it is possible to find sections where
stresses due to bending dominate the stress distribution and even absence of a normal force is
possible. In such cases it is not feasible to apply steel fibers as main reinforcement. Therefore,
in order to provide a general structural solution for future tunnel planning it is often suggested
to combine the best properties of both steel fibers and ordinary steel reinforcement. This
approach results in a combination of steel fiber reinforcement and conventional steel
reinforcement. In such case bending moments in the tunnel lining are resisted by the
reinforcing steel, while splitting forces due to thrust jacking and coupling forces resulting from
the girder behavior of the tunnel lining are withstand solely by the SFRC. In case of extrusion
tunnels, a multiaxial state of stress is present: tension is the result of truss forces, bending
results from soil settlements and compression follows from the soil (and water) pressure. Also
in this case the use of SFRC may result in a reduction of the amount of conventional steel
reinforcement.
The addition of steel fibers to concrete mixtures puts additional demands to the design of the
mix composition. A number of adaptations need to be made with respect to the aggregate
grading curve when fibers are added to the concrete mix. Firstly, continuous aggregate grading
gives a lower probability of fiber balling than discontinuous grading. Secondly, from a
workability point of view it is recommended to limit the maximum grain size to 1/2 - 1/3 of
the fiber length [17, 34]. As the maximum grain size decreases, the fibers are more randomly
distributed and orientated in the mix. Thirdly, the ratio of finest/total aggregate volume needs
to be adjusted to attain an optimum packing density. As the fiber volume Vf increases, the
packing density decreases. A similar effect is found when the fiber’s aspect ratio (Lf/df) is
increased. It is generally agreed that the Lf/df ratio should not be higher than 60. If higher Lf/df
ratios are used in combination with higher fiber volumes (e.g. in tunnel linings), it is advised
The American Concrete Institute (ACI) developed a guideline to simplify the mix design of
SFRC [5]. Though it does not take into account the effect and importance of additives such as
superplasticizers, it is a very good indication when verifying the quality of the mix
composition.
Generally, the workability of concrete gets worse if fibers are added to the matrix. With
increasing aspect ratio and volume of fibers, the workability of the concrete mix decreases.
The probability of fiber balling increases when the maximum fiber volume is reached for a
given maximum grain size and aspect ratio or when the volume of coarse aggregate is too high
with respect to the total aggregate volume. Moreover, production aspects such as elongation of
the mixing time and time-discontinuity when adding fibers contribute to decreasing
workability of the concrete. Addition of suitable superplasticizers shall help to improve the
workability of the concrete mix to the desired level. Also compaction has an influence on fiber
orientation. An internal vibrator for compacting the concrete will disturb the orientation of the
fibers locally, creating weak spots. With the application of an external vibrator the fibers will
tend to orientate in a direction perpendicular to the vibration direction. Yet, the compacting
method by means of external vibrators is preferable. Good workability in combination with
proper compaction is necessary to guarantee a homogeneous and isotropic structure of SFRC.
As a result of non-uniform density and inhomogeneous orientation of the fibers the properties
of the SFRC material after hardening may not be the same in every place and in every
direction. Nevertheless, one must not forget the unavoidable deviations from random fiber
orientation, which occur as a result of the geometrical influences (formwork, free surfaces,
conventional reinforcement). Furthermore, the pouring direction as well as the flow direction
tends to orient the steel fibers a well.
Under increasing pull-out forces, the bond behavior in plain (non-fibrous) concrete is marked
by: 1) the initiation of inclined cracks at contact points between the steel lugs and concrete at
relatively low stresses (see Fig. 1.1a); 2) crushing of concrete in front of the lugs (see
Fig. 1.1b); 3) shearing-off of an increasingly larger part of concrete keys between the lugs
until the keys are fully sheared off (see Fig. 1.1c), after which some frictional bond resistance
is left. The gradual shearing-off of the concrete keys is possible only in well-confined
Since the tensile strength of concrete is low, cracking in concrete occurs at relatively low
loads. At the onset of cracking, the role of the randomly orientated steel fibers in SFRC is to
retard the propagation of micro cracks in the matrix. It is conceivable, that the steel fibers are
able to bridge the so-called Goto-microcracks arising at the ribs of the bar (see Fig. 1.2).
Figure 1.2: Effect of steel fibers in cracked reinforced concrete section [27].
Once cracking occurs, fibers strengthen the matrix by transmitting a substantial tensile force
during fiber slipping, with a sufficient margin against fracture, prevent further opening of
cracks and resist additional tensile forces, which the concrete matrix itself cannot sustain.
Hence, while the fibers may not much delay the formation of the first crack, they may keep
2.1 Pull-out tests with short embedment length (RILEM Standard RC6 or
similar test set-up)
In the experimental investigation presented in [6] and [7] the bond strength, failure
development and failure type (splitting or pull-out) in case of ribbed steel bars embedded in
high performance fiber reinforced concrete were studied. Figure 2.1 shows test set-up and
geometry of the test specimens used.
Figure 2.1: Scheme of the testing device (left) and geometry of test specimens (right) [6].
Figure 2.2: Bond stress – displacement curves for plain concrete and SFRC [6]:
(a) BD0-1455: plain concrete fccm=147 MP, ds = 14 mm, c/ds = 1.5
BD1-1455: SFRC fccm=155 MPa, ds = 14 mm, c/ds = 1.5
(b) BD0-1870: plain concrete fccm=147 MP, ds = 18 mm, c/ds = 1.5
BD1-1870: SFRC fccm=155 MPa, ds = 18 mm, c/ds = 1.5
The experimental investigation presented in [27] was intended to answer the question why
crack widths are reduced in reinforced bending elements made of SFRC when compared to
reinforced elements made of normal weight concrete without fibers. The possible
improvement of bond of the embedded deformed bars due to fiber addition was seen as an
explanation for this finding. Besides, the extent of longitudinal cracking and occurrence of
bond splitting failure in case of SFRC was investigated. Figure 2.5 shows the types of test
specimens and Tables 2.5, 2.6 and 2.7 give the main characteristics of this test program.
Figure 2.6: Bond stress – slip curves for plain concrete and SFRC (hooked fibers) [27]:
(left) concentric and edge bar position
(right) corner bar position
Figure 2.7: Mean bond stress at (theoretical) occurrence of first longitudinal cracking in the
concrete cover, related to fiber content and bar position [27].
Typical bond stress - displacement curves ( - curves) are given in Figure 2.6. Figure 2.7
shows the bond stress at first longitudinal cracking, defined as the stress at 0.15‰ transverse
strain, a value that is assumed to correspond to the mean failure strain of fiber reinforced
An experimental investigation of local bond stress - slip behavior of ribbed bars embedded in
SFRC is discussed in [12]. Whereas the primary emphasis was on the monotonic local bond-
slip relationship, two tests are carried out under large slip reversals in order to study the effect
of fibers on the cyclic bond-slip behavior. A part of this investigation (i.e. beam tests) is
discussed in chapter 2.4.1. Figure 2.8 shows the type of specimens used in the pull-out test
series. Tables 2.9, 2.10 and 2.11 give the main characteristics of this part of the test program.
Since the concrete strength was not intended to be a parameter in this investigation, the results
of the individual specimens were normalized to fc’ = 22 MPa, assuming that the bond
resistance is proportional to (fc’)0.5. Note that it is not quite clear how the strength of concrete
in the specimens was deduced, since only in some cases concrete cylinders for testing concrete
compressive strength were taken after adding fibers and in general no more than the
compressive strength of the unreinforced matrix was determined before adding fibers. In case
of mixes with larger fiber aspect ratio (50 / 0.5), a decrease of the concrete compressive
strength after adding fibers is reported, which suggests that workability of the mix and
compaction of the specimens were very poor. Test results and major conclusions from this
investigation are presented in Figure 2.9 and in Table 2.12.
Figure 2.9: Mean bond stress – slip relationships for different reinforcement and fiber parameters
[12].
The experimental investigation presented in [14] was intended to extend the basis for a
formulation of an analytical model suitable for describing the bond stress – slip response of
reinforcing bars embedded in SFRC. In particular the experimental results of tests defined to
investigate splitting-type bond failure were reported. Furthermore constitutive equations
describing the characteristic bond properties were proposed, further discussed in chapter 2.5.
Figure 2.10 shows the types of specimens used in this test series. Tables 2.13, 2.14 and 2.15
give main characteristics of the test program. Typical bond stress - slip curves obtained in this
test series are given in Figures 2.11.
Figure 2.10: Pull-out test series: specimen dimensions and reinforcement details [14].
Since the concrete strength was not intended to be a parameter in this investigation, the results
for individual specimens were normalized to fc’ = 30 MPa (i.e. the average strength of
concrete matrix before adding fibres), assuming that the bond resistance is proportional to
(fc’)0.5. Just as in [12], it is not quite clear in which way the actual strength of the concrete in
the specimens was deduced.
5.5
Before conclusions from this investigation are presented, some issues must be discussed.
Firstly, it is stressed that in some specimens the splitting mode of specimen failure was
introduced by placing a thin plastic sheet in the plane of the longitudinal axis of the bar, i.e. by
making a notch. This type of specimens are absolutely unsuitable for determining the
resistance of concrete to splitting forces originating from bond action of a pulled bar: in
abovementioned specimens, due to specimen geometry (notch) and boundary conditions
(rubber pad covered with grease acting as friction-free or outward-directed-shear-force
support) additional splitting forces are acting. These splitting forces are superimposed on
forces due to the splitting action of the embedded bar and, the consequently, splitting process
is speed-up in comparison with cases where no notch is present, discussed in [12] and [14]
(compare case P1-unconfined, ds = 25 mm, Vf = 0%, 1%, 2% in Fig. 2.9 [12] and case P1A,
ds = 25 mm, Vf = 0%, 1%, 2% in Fig. 2.11 [14]; compare case P2-unconfined, ds = 25 mm,
Vf = 0%, 0,9% and case P2A, ds = 25 mm, Vf = 0%, 0,9% in Fig. 2.11 [14]). For this reason
test results of series of specimens with inserted plastic sheet should be excluded from the
evaluation.
Secondly, unexpected differences are found in the response of specimens with nearly identical
geometry, as described in [12] and in [14]. While case P1-unconfined, ds = 25 mm, Vf = 0.9%
in Fig. 2.9 [12] shows the typical response for a pull-out type of bond failure, in case P1A,
ds = 25 mm, Vf = 0%, 1%, 2% in Fig 2.11 [14] a sudden drop of bond resistance is found,
which denotes occurrence of splitting cracks. Such differences in behavior can hardly be
linked to increase of embedded length of the bar from 3.5 ds to 5 ds, or to increases in concrete
compressive strength from 22 MPa to 30 MPa, in the former and in the latter case,
respectively. Most likely this are test conditions, i.e. boundary conditions or eccentricities, that
lead to different behavior in test series reported in [12] and [14]. Note that while in case of
Figure 2.11: Local bond stress – slip relationships for different specimens and fiber parameters
[14].
Nevertheless, main conclusions with respect to bond behavior of ribbed steel bars in SFRC are
summarized in Table 2.8, as formulated in [14].
The experimental investigation presented in [9] focused on determining the influence of the
fiber content and of concrete cover thickness on bond strength of ribbed bars and on bond
failure propagation in SFRC. Figure 2.12 shows the types of test specimens and Tables 2.17,
2.18 and 2.19 give main characteristics of this test program.
Note that the effective rib area was most probably not measured and only the minimum value
guaranteed by the producer is provided. Furthermore, the dimensions of the test specimens are
small (in some cases even very small) compared to the fiber length. In particular in case of
specimen type 40/40 it is very likely that the fiber orientation is restrained by the geometrical
boundaries of the specimen and by concentrically placed bar. It would not be surprising if in
such a case fibers tend to orientate parallel to the bar and are not very effective in arresting
bond and splitting cracks. This aspect is, however, not addressed in the test report.
Figure 2.13: Bond stress – slip curves for specimen type 40/40:
(left) single test results
(right) mean test results
Figure 2.14: Bond stress – slip curves for specimen type 75/40:
(left) single test results
(right) mean test results
Typical bond stress - displacement curves are given in Figures 2.13 –2.15. Main conclusions
from this investigation are summarized in Table 2.20.
The experimental investigation reported in [32] deals with the effects of steel fibre
reinforcement on the local bond behavior of deformed bars under conditions similar to the
ones in a of beam-column connection. Figure 2.16 shows the test set-up and geometry of the
specimen used in the pull-out test series. Tables 2.21, 2.22 and 2.23 give main characteristics
of this test program.
Note that the effective rib area of the bars is estimated on the basis of very limited geometrical
data. Flexural toughness was measured on specimens with cross-section 102 x 102 mm. With
respect to the geometry of test specimens for pull-out tests few issues must be discussed. It is
reported that during casting a plastic film sheet was placed in plane of the longitudinal bar axis
to represent the splitting crack caused by the pullout of a bar in actual joint conditions. It is
debatable if such type of pre-defined crack is suitable for modeling cracks in SRFC. However,
it is not of such a critical importance as in the tests discussed in Chapter 2.1.4, considering the
amount of confining reinforcement provided and the fact that due to this confinement bond
failure is caused by bar pull-out and not by splitting. Yet, there is another aspect related to the
placing of the plastic sheet that needs further consideration.
Figure 2.13: Correlation of bond strength with flexural toughness, defined as the area underneath
the load-deflection curve determined on 102 x 102 x 356 mm prisms loaded at one-
third points on a span of 303 mm, up to a mid-span deflection equal to the test beam
span divided by 150 (regression line and 95% confidence interval) [32].
Figure 2.15: Effect of steel fiber aspect ratio (Lf / df) on local bond behavior [32]:
(a) mean bond stress – slip relationships for various fiber aspect ratio (Lf / df)
(b) bond stress at peak versus fiber aspect ratio (Lf / df) (regression analysis)
(c) slip at peak versus fiber aspect ratio (Lf / df) (regression analysis)
As can be seen in Fig. 2.12, the opening in the plastic sheet is very small with respect to the
bar diameter (1 ds x 4 ds on each side of the bar). It is hard to judge how this will influence the
orientation of the fibers and their local concentration, in particular in the situation when the
In [25] results from pull-out tests on short anchorages of deformed bars, simulating the
influence zone of one stirrup, are discussed. In these tests particular attention is given to the
influence of the transverse reinforcement area on local bond behavior and on splitting crack
opening. Figure 2.15 shows the specimen geometry and arrangement of measuring devices,
9.5
Figure 2.16: Bond stress – slip curves for specimens without stirrup reinforcement
Figure 2.17: Bond stress – slip curves for specimens with 8 mm stirrups
The purpose of the experimental study presented in [16] was to investigate the bond
characteristics of reinforcing bars embedded in SFRC under various types of loading, i.e.
monotonic, unidirectional cyclic and reversed cyclic loading. Four types of matrix were tested,
i.e. SIFCON, fiber reinforced concrete, confined concrete and plain concrete. Figure 2.19
shows the loading setup and specimen geometry. Tables 2.29, 2.30 and 2.31 give main
characteristics of this test program. Relative rib area of the reinforcing bar is not given and
cannot be estimated due to too limited data on rib geometry. In this review attention is given
mainly to results of monotonic pull-out tests.
Test results and major conclusions from this investigation are presented in Figures 2.20 and in
Table 2.32. Note that in case of plain concrete specimens the tests were stopped directly after
(a) (b)
(c)
The experimental study presented in [30] and [31] was designed to evaluate the bond strength
and bond – slip behavior of reinforced concrete containing steel fibers. The tests were
conducted using modified pull-out specimen in which the concrete surrounding the rebar was
in tension. Such configuration was intended to simulate the behavior of the bars in the tension
zone of beams and beam-columns. The investigated loading patterns included monotonic, half
cyclic and reverse cyclic pull-out load. In this review, however, only part of research
concerning monotonic loading is considered. 18 types of matrix were tested, varying silica
fume content, fiber content and fiber aspect ratio. Figure 2.21 shows loading setup and
specimen geometry. Tables 2.33, 2.34 and 2.35 give main characteristics of this test program.
Note that for none of the bar diameters used the relative rib area is given. For the interpretation
of test results it is important to note that the geometry of the test specimens is not related to the
(variable) bar diameter. In particular, for each bar diameter an other ratio between concrete
cover and bar diameter is chosen (i.e. other confining conditions). Furthermore, the
embedment length of the bars varies with bar diameter from 50mm to 180mm (Lb / ds varies
from 5.3 to 8). Such a large variation of embedment length in combination with the
assumption of uniformly distributed bond stresses does not enable objective comparison of
(average) bond stress values. Test results and major conclusions from this investigation are
presented in Figures 2.22 and in Table 2.36.
Figure 2.21: Loading set-up (left) and geometry of test specimen (right) [30].
It is remarkable that despite the in general large ratios between concrete cover on the bar and
bar diameter, frictional pull-out failure was obtained only for bars with diameter ds = 9 mm
(c/ds = 6). Unfortunately no information is provided with respect to deformations on the bars
(ribs and/or indentations), which would allow judging the magnitude of radial splitting forces
introduced in the specimen due to the wedging action of pulled bars. One could presume that
bars with a very high effective rib areas were used. For all but ds = 9 mm splitting bond failure
was observed.
(b)
(c)
Figure 2.22: Normalized load – slip curves for reinforcing bar ds = 19 [30]:
(a) comparison for different fiber content
(b) comparison for different fiber aspect ratio
(c) comparison for different silica fume content
Note that splitting bond failure does not provide the pull-out bond strength and that obtained
splitting bond capacity is directly linked to concrete confinement (which is not consistently
chosen in the discussed test series, as mentioned above).
The goal of this research reported in [19] was to determine the effect of a change in the
material tensile strength and toughness on the pull-out response of deformed bars. Four
material systems were therefore selected; plain concrete, two types of SFRC and high
performance SFRC (HPSFRC), representing brittle, pseudo-brittle and pseudo-ductile material
behavior.
- curves
concrete in compression c/ds
Lb / ds
Figure 2.23 shows the layout of the pull-out specimen used in this study. Tables 2.37, 2.38 and
2.39 give main characteristics of this test program. Test results and major conclusions from
this investigation are presented in Figures 2.24 and 2.25 and in Table 2.40. A modified version
of the Danish Standard DS2082 pull-out test specimen was used, in which the concrete
surrounding the rebar was in tension. Note that for none of the bar diameters used the relative
rib area is given.
Figure 2.25: Change of the average bond strength with respect to compressive cylinder strength
[19]: (a) bar ds = 10 mm, (b) bar ds = 25 mm.
For the interpretation of test results it is important to note that the geometry of the test
specimens is not related to the (variable) bar diameter. In particular, constant cover thickness
is chosen (i.e. for each bar diameter an other ratio between concrete cover and bar diameter is
taken which results in different confining conditions). Furthermore, the embedment length of
the bars is constant, i.e. Lb / ds varies. Variation of embedment length in combination with the
assumption of uniformly distributed bond stresses does not enable objective comparison of
(average) bond stress values. Note that splitting bond failure does not provide the pull-out
bond strength and that obtained splitting bond capacity is directly linked to concrete
confinement (which is not consistently chosen in the discussed test series, as mentioned
above).
The goal of this research reported in [22], [1] and [2] was to investigate bond behavior of
ribbed bars and prestressing strands in high strength concrete, with emphasis on the effect of
adding steel fibers. In this review only results concerning the behavior of ordinary reinforcing
steel are evaluated. Figure 2.26 shows the layout of the pull-out specimen used in this study.
Tables 2.41, 2.42 and 2.43 give main characteristics of this test program. A modified version
of the Danish Standard DS2082 pull-out test specimen was used, in which the concrete
surrounding the rebar was in tension. Note that for none of the bar diameters used relative rib
area is given. No information is provided with respect to the tensile properties of the steel fiber
reinforced COMPRESIT matrix.
1
1.4
2.6
6
16 4 67 3, 4.4, 5.6 0 0
The test program included, among others, variation in embedment length. Concerning
interpretation of test results with respect to this influential parameter, one must remember that
a strong effect of boundary conditions on pull-out resistance is expected for the chosen
geometry of the test specimen (i.e. pull-out of the concrete cone around the bar rather than
pull-out of the bar). Furthermore, decrease of average bond strength with increasing
embedment length is partly inherit, when uniform bond stress distribution is assumed. Test
results and major conclusions from this investigation are presented in Figures 2.27, 2.28 and
2.29 and in Table 2.44.
The tests described in [27] and [21] deal with bond between reinforcement bars and
COMPRESIT, ultra high strength compact reinforced composite. The purpose of the tests was
to investigate the anchorage capacity of deformed reinforcement bars embedded in the matrix,
taking into account the influence of splitting along an anchored deformed bar, transverse
reinforcement and lateral compression. Results of cyclic load tests, in particular discussed in
[21], are not included in this review. Figure 2.30 shows the layout of the pull-out specimens
used in this study. Tables 2.45, 2.46 and 2.47 give main characteristics of this test program. A
modified version of the Danish Standard DS2082 pull-out test specimen was used, in which
the concrete surrounding the rebar was in tension. Note that relative rib area of the bar used is
not given and that tensile properties of the COMPRESIT matrix are not specified. Test results
and major conclusions from this investigation are presented in Figure 2.31 and in Table 2.48.
Figure 2.30: Geometry of the test specimens without and with transverse reinfrocement[27].
Figure 2.31: Examples of experimental tensile load – displacement relationships for test specimens
with Lb / ds = 3.75 (σ / fc indicates level of lateral compressive normal stress) [27].
The experimental study presented in [23] and [24] intends to describe the global behavior of
tie elements based on a more explicit representation of the concrete behavior in tension. By
that bond of reinforcing bars to concrete is indirectly investigated. In this study, plain and
fibrous concrete are used for which the relevant tensile properties are determined, which is
often not the case in other tests mentioned in this review.
Figure 2.32 shows the loading setup and specimen geometry. Table 2.49, 2.50 and 2.51 give
main characteristics of this test program. Note that the relative rib area of the reinforcing bar is
not specified. Test results and major conclusions from this investigation are presented in
Figures 2.33 and in Table 2.52.
The experimental study presented in [10] was performed to study the post-cracking behavior
of steel bars embedded in SFRC in uniaxial tension. Figure 2.34 shows the geometry of the
specimen and details of the steel bar. Note that the bar was provided with strain gauges placed
in channels on two opposite sides of the bar. Concrete strain was measured by means of strain
gauges glued on the concrete surface. Neither the locations of the strain gauges on the bar nor
the position of the strain gauges on the surface of specimen are specified. Most probably they
were located somewhere midlength of the specimen. Table 2.53, 2.54 and 2.55 give main
characteristics of this test program.
Note that no information is provided with respect to concrete strength and toughness. The
relative rib area of the reinforcing bar is not specified. With regard to the measuring technique
and interpretation of test data serious doubts are rising. Pondering relatively long embedment
length assuming uniform strain distribution along the bar is obviously very simplistic. Since
concrete cover on the bar is relatively small, it is almost inevitable that longitudinal splitting
cracks must have developed – this is however not mentioned in the report. In general it may be
stated that although major tendencies indicated by these tests may be qualitatively correctly
captured, due to the simplistic approach and insufficient background information, results
2.3.3 H.H. Abrishami & D. Mitchell [3], [4] and D. Mitchell, H.H. Abrishami & S.
Mindess [20]
The experimental study presented in [3], [4] and [20] is part of a major research project to
investigate the influence of concrete strength, bar size, epoxy coating on the reinforcement,
and steel fibers on the response of tension members. [4] and [20] focus on the effect of steel
fibers on the behavior of reinforced concrete elements subjected to pure tension, including the
tension stiffening characteristics and the crack control. Part of this research devoted to epoxy-
coated bars is not discussed in this review. Figure 2.36 shows the geometry and
instrumentation for a typical specimen. Table 2.57, 2.58 and 2.59 give main characteristics of
this test program. Note that the relative rib area of the reinforcing bar is not specified. Test
results and major conclusions from this investigation are presented in Figures 2.37, 2.38, 2.39
and 2.40, and in Table 2.60.
4.8
Figure 2.39: Effect of steel fibers on the tension response of concrete [4].
The experimental investigation of local bond stress - slip behavior of ribbed bars embedded in
SFRC, presented in [12], is already partly discussed in 2.1.3. Figure 2.41 shows the type of
specimens used in the beam test series. Tables 2.61, 2.62 and 2.63 give main characteristics of
this part of test program.
Since the concrete strength was not intended to be a parameter in this investigation, the results
of individual specimens were normalized to fc’ = 22 MPa, assuming that the bond resistance is
proportional to (fc’)0.5. Note that it is not quite clear how the strength of concrete in the
specimens was deduced. Test results and major conclusions from this investigation are
presented in Figures 2.42 and in Table 2.64. The bond stress was calculated using moment
equilibrium at the beam mid-span section between the extremely applied load and the internal
force couple provided solely by the bottom and top reinforcing bars.
Figure 2.42: Mean bond stress – slip relationships for different confinement and fiber content [12].
Moving from the constitutive to the structural behavior, it is of primary interest to study steel-
concrete bond, because it affects the overall response. With reference to judging safety and
durability, modeling of bond is obviously a principal issue. Yet, modeling of bond in SFRC is
not very well advanced. Few investigations in this direction only touch various aspects,
approaches and techniques related to bond modeling. It is obviously much to early to be ably
to come with a well-established numerical model, which is able to serve as engineering tool
accurate in various bond-related questions.
Fundamental studies of Cattaneo [6] allowed developing limit analysis model to evaluate the
ultimate bond capacity by maximizing bar pressure with respect to crack extension. This
refined theoretical model allows finding the solution with the static and energetic methods.
The description of the splitting-pullout failure transition is in good agreement with
experimental evidence. It is however not sufficient for practical engineering application to be
able to predict bond strength only. A very simple approach has been taken by Kstulovic et al
[19], who proposed an empirical equation to predict ultimate bond strength. Using regression
analysis a function has been derived, which relates ultimate bond strength to matrix tensile
strength and toughness. As with every empirical relationship, its validity is limited due to
limited (own-test-based) database used in the analysis.
Local bond stress – slip models offer more application relatedpossibilities. Two of such
models can be mentioned: the empirical relationship proposed by Vandewalle [35] and the
analytical model proposed by M.H. Harajli [13] and M.H. Harajli et al [14]. Model proposed
by Vandewalle [35] allows evaluating the ultimate bond capacity based on hydraulic pressure
analogy (elastic-cracked model). The local bond stress – slip relationship can be predicted
using Coulomb criteria solely in the initial ascending branch. In case of SFRC sufficient
accuracy is reached only in case of splitting bond failure, as shown by De Bonte [9]. De Bonte
[9] analyzed also the accuracy of the empirical model of Harajli [13] and M.H. Harajli et al
[14], consisting of a monotonic envelope curve and a reduced bond resistance curve
corresponding to pull-out and splitting mode bond failure, respectively. With respect to this
model it is stressed that serious doubts exist with respect to the relevancy of test results that
serve as the basis for model formulation and correctness of test interpretation, as discussed in
preceding sections. De Bonte [9] concludes that the accuracy of model predictions is not quite
satisfactory, both with respect to ultimate bond strength and local bond-slip relationship.
Besides local bond stress – slip models, analytical models are devised to predict directly the
load-deformation response of a tie element. Though few such models exist, actually only the
model of Noghabai [24] explicitly accounts for tension softening of concrete. The relatively
simple approach is taken, where the tension element is modeled by parallel “springs”. The
behavior of the tie element is described as the addition of the response of the bare rebar and
the tension stiffening of the concrete that takes place at predefined crack planes. Also, a
condition for compatibility between the rebar and concrete has to be met. The obvious
deficiency of the model is the need of assuming the average crack spacing before tension
stiffening effect can be estimated. Where this assumption is explicitly based on the observed
crack patterns, the model seems to fit the experimental evidence quite well (however it must
be stressed that verification is limited to the own tests). Contrary to the refined fracture-
mechanics based model of Noghabai [24], Abrishami & Mitchell [3] proposed a simple
approach to model tension stiffening. However, neglecting the influence of splitting cracks
and assuming (unverified) orientation of fibers in the cross section limit the applicability of the
model to the pull-out bond failure mode and shows its oversimplification.
The pull-out behavior is a function of the bar characteristics (geometry and steel type),
characteristics of the surrounding matrix, and the level of lateral confinement (e.g. cover
thickness or the presence of additional active or passive confinement). Results from the
investigations discussed in this review may serve to outline the whole spectrum of factors
influencing bond behavior, however at current stage available knowledge is clearly far from
being complete. This is partly due to lack of consistency and systematic in studies, partly due
to contradictions in conclusions from different investigations, and partly due to lack of
understanding of links between concrete technology, basic material characteristics and
composite behavior of reinforced concrete. Typical for SFRC diversity of properties of
concrete matrix components (in particular shape and size of fibers) as well as differences in
composition and quality of the SFRC composite make it very complicated to compare
quantitatively results from different investigations, in particular when reference parameters
describing general tensile characteristics of the matrix are missing or when non-representative
specimen types are used in the investigations. Nevertheless, in the following tables
conclusions from various researches are gathered and synthesized.
As mentioned repeatedly in earlier sections, changes in bond quality have implications for the
performance of reinforced concrete structures. The primary contribution of fibers for structural
applications is in the area of ductility. The enhancement of ductility can be effectively utilized
in all three modes of loading: bending, shear, and torsion. In this respect bond properties of
reinforcement embedded in SFRC affect the structural behavior of members and structures
from the point of view of structural safety and stiffness reduction caused by splitting.
An important aspect with respect to contribution of fibers concerns the structural durability.
While (transverse) flexural or tensile cracks expose a very limited bar length to the
environment, bond splitting cracks develop longitudinally along the bar so that a considerable
length of the reinforcement can be exposed to aggressive agents. As far as the structural
durability is concerned, the presence of fibers is again particularly useful since the fibers allow
a reduction of the splitting crack width. Note that durability may in turn affect structural safety
as well. While it is known that the addition of steel fibers leads to a reduction of crack width it
is not established whether this effect is only due to the proven transfer of tensile forces by the
fibers across cracks or also because of an improvement of bond of the embedded deformed bar
reinforcement by fibers.
In general it may be concluded that effects of steel fiber reinforcement on the interaction
between deformed bars and concrete have not been investigated thoroughly and
systematically. While extensive experimental and analytical studies have focused on the bond
characteristics of reinforcing bars embedded in conventional concrete, experimental data on
the bond properties of reinforcing bars embedded in SFRC is very limited: there are only few
reports dealing with this question; their statements are partly contradictory. Since in the tests
discussed above both the reinforcing bars and the fibers had different geometrical and
mechanical properties and different concrete mixes were used, the results are hardly
comparable. Although tendencies observed may be in general correctly captured, reference
parameters are often missing, which hinders reliable evaluation of bond test results. Note that
such a reference property could be the shape of softening diagram under tensile load, since it
represents the effectiveness of fibers in transmitting tensile stresses – also radial tensile
stresses due to splitting bond forces.
Hence, it necessary to further study the influence of steel fibers on the bond behavior of
deformed bars. A program of additional systematic experimental study should be developed
keeping in mind the following aspects directly and indirectly related to bond behavior of
deformed bars in SFRC:
• Choosing type of bond tests relevant for structural models used to analyze members and
RC structures in intended field of application (local bond stress-slip relationship, tension
stiffening model for tie elements etc.).
• Adjusting geometry of test specimens on one hand to intended field of application and on
the other hand to homogeneity level (i.e. quality) of SFR composite (fiber length, aggregate
size, workability etc.).
• Limiting variation of material characteristics of steel and concrete to that relevant for
intended field of application.
• Providing necessary reference information with respect to the actual matrix behavior.
It is evident that short pull-out tests provide direct information with regard to bond
performance of a bar embedded in a concrete matrix. However, for full understanding of the
basic principles governing the behavior of reinforced concrete, more than only knowledge of
local bond stress - slip response under specific conditions is required. In particular crack
control and tension stiffening in reinforced members deserve more attention. In case of SFRC
it must be further clarified to what extend improved (local) bond performance in combination
with ability of SFRC to transmit tension across cracks contribute to improvement of the
cracking behavior and the tension stiffening effect in reinforced concrete members. Although
reinforced concrete members subjected to pure tension (tie elements) do not often occur in
practice, studying this simple case forms crucial step in understanding the response of
biaxially or triaxially stresses reinforced elements.
With respect to the homogeneity of SRFC one must not forget the relation between volume the
fraction of fibers and the level of matrix compaction (i.e. quality of concrete casting). Though
it seems to be obvious, it is frequently forgotten that when no particular attention is paid to the
improvement of mix packing, an increase in the fiber volume fraction may detrimentally affect
the composite quality. Concrete mix workability and compactability as well as the actual
quality of specimen casting are in fact parameters that directly influence bond behavior of a
rebar embedded in SFRC matrices. The influence of this and other parameters on the basic
matrix behavior can be captured if information is provided with respect to basic material
characteristics of the concrete matrix. Knowledge of matrix compressive strength is certainly
not sufficient. It is indispensable to possess information concerning the actual matrix tensile
behavior, such as matrix tensile strength and toughness, if one aims at developing a basis for
any type of systematic analysis or model formulation.
5 References
1 Aarup B. & Jensen B.Chr., Bond Properties of High Strength Fiber Reinforced
Concrete, ACI Special Publication SP-180 - Bond and Development of Reinforcement,
1998, pp.459-472
2 Aarup B., Karlsen J. & Lindström G., Fiber reinforced high performance concrete for
in-situ cast joints, ACI Fall Conference, Orlando, Sept. 2000, pp.9
3 Abrishami H.H & Mitchell D., Influence of Splitting Cracks on Tension Stiffening, ACI
Structural Journal, Vol. 93, No. 6, 1996, pp. 703-710
4 Abrishami H.H & Mitchell D., Influence of Stel Fibers on Tension Stiffening, ACI
Structural Journal, Vol. 94, No. 6, 1997, pp. 769-776
5 ACI Commitee 544, Design considerations for Steel Fiber Reinforced Concrete, ACI
Structural Journal, Vol. 85 (6), 1988, pp. 563-580
6 Cattaneo S., Ultra High-Performance Concrete: Constitutive Behavior and Structural
Applications, PhD Thesis, Politecnico di Milano, 2000, pp.185
7 Cattaneo S. & Rosati G., Bond and splitting in High Performance Fiber Reinforced
Concrete, Proceedings of Fifth RILEM Symposium on Fiber-Reinforced Concretes
(RC), Lyon, Ed. P.Rossi and G.Chanvillard, 2000
8 DANSK STANDARD DS-2082, Armeringsstål. Prøvning af forankrings-evnen (Steel
for reinforced concrete – Pull-out test), 1980, pp.2
9 De Bonte F., Hechtsterkte bij staalvezelbeton, MSc Thesis, Katholieke Universiteit
Leuven, 2000, pp.140