This paper investigates the flexural fatigue performance of self- SCC possesses good fluidity and deformability, making
consolidating concrete (SCC) and self-consolidating fiber-reinforced it more suitable for the addition of fibers as compared to
concrete (SCFRC) containing round corrugated steel fibers with a NVC and allows for much easier construction, resulting in
size of 1 x 30 mm (0.04 x 1.18 in.) in different 0.5, 1.0, and 1.5% a more reliable quality in concrete placement and a more
volume fractions. Approximately 250 flexural fatigue tests and homogeneous material structure.14 SCC reinforced with
195 complementary static flexural tests were executed on beam
steel fibers enhances its applications because the mechanical
specimens with a size of 100 x 100 x 500 mm (3.94 x 3.94 x 19.7 in.)
under four-point flexural loading. The fatigue-life data show that the
performance of concrete is improved. Self-consolidating
probabilistic distribution of fatigue life of SCC/SCFRC at a given fiber-reinforced concrete (SCFRC) is more ductile and
stress level can approximately be modeled by the two-parameter tougher than conventional SCC and has demonstrated higher
Weibull distribution. Three different methods were used to obtain residual strengths.15 The workability of SCFRC is directly
the Weibull parameters. A single-log fatigue equation was used to influenced by the type and content of fibers used, as well as
analyze the flexural fatigue performance of SCC/SCFRC with a the SCC matrix. A higher aspect ratio and volume concen-
10% probability of failure. The results show significantly improved tration of fibers improve the performance of SCFRC in the
fatigue performance of SCFRC with enhanced sensitivity of fatigue hardened state but also affects its workability. Thus, studies
lives to the change of applied stress. Theoretic fatigue lives for SCC/ were conducted to obtain optimum fiber-reinforced concrete
SCFRC were estimated that exhibit an increase to a different extent. (FRC) mixtures with required self-consolidating proper-
ties.16-18 Dhonde et al.17 revealed that SCFRC could be made
Keywords: fatigue life; self-consolidating fiber-reinforced concrete; stress
level; Weibull distribution. with satisfactory filling and passing ability using short fibers
(L 30 mm [1.2 in.] long), as these did not influence its
INTRODUCTION slump flow or stability. Researchers investigated whether
Self-consolidating concrete (SCC) is an innovative concrete SCFRC shows either similar or improved performance in
that does not require vibration for placing and compaction. terms of compressive strength, flexural strength, splitting
It is able to flow under its own weight, completely filling tensile strength, elastic modulus, creep and shrinkage, and
formwork, and encapsulate the reinforcement, achieving full shear and pullout behavior compared to SCC and normally
compaction, even in the presence of congested reinforce- vibrated fiber-reinforced concrete (NVFRC) under statically
ment.1 The hardened SCC is dense and homogeneous and has applied loads.11,14,15,17,19-21 The microstructure around the
improved engineering properties and durability compared to matrix, the distribution, and the orientation of the fibers are
different in SCC than in conventional concrete. Entrapped
normally vibrated concrete (NVC). The improved construc-
air and neighboring fibers affect the performance of a fiber
tion practice and performance, combined with the health and
in NVFRC more than in SCFRC. The steel fibers, due to
safety benefits, make SCC a very attractive solution for both
the lack of any mechanical vibrations in SCFRC, are more
precast concrete and civil engineering construction.2
favorably aligned into the direction of the flow, thereby
Due to its substantial engineering applications and
improving its bending characteristics. In SCC, the fibers are
commercial benefits, SCC has generated tremendous interest fully embedded in the matrix, thereby imparting better bond
among researchers, engineers, and concrete technolo- or pullout strength.13,21 Thus, it is expected that SCC and
gists.3,4 Numerous research studies have shown that it is prac- SCFRC, as in the case of their mechanical properties, such
tical to make a flowable yet stable SCC tailored for any appli- as compressive and flexural strength under statically applied
cation.5,6 A number of investigations related to the rheological, loads, may exhibit better fatigue characteristics.
mechanical, and structural behavior of SCC under statically The global thrust on construction of bridges and highway
applied loads have been reported in literature that substantiate pavements for infrastructure development has fascinated
the better performance of SCC compared to NVC.7-10 many researchers,3,4,7,8 leading to investigations of the
The importance of the homogeneity of the material is fatigue behavior of concrete. The bridges and pavements
evident for any application because it will affect the material were expected to resist millions of cycles of repeated axle
properties.11 SCC contains large proportions of finer parti- loads during their intended life. Considering fatigue strength
cles and does not need mechanical vibrators for compaction, an important parameter in the design of these structures,
which results in a denser and more homogenous concrete
compared to NVC. The denser structure of SCC dimin-
ishes the presence of air voids so better bonding between
the concrete and reinforcing materials is achieved; this ACI Materials Journal, V. 109, No. 5, September-October 2012.
MS No. M-2011-244.R1 received November 28, 2011, and reviewed under Institute
could be beneficial and lead to better results in terms of the publication policies. Copyright 2012, American Concrete Institute. All rights reserved,
mechanical behavior of the constituents compared with that including the making of copies unless permission is obtained from the copyright
proprietors. Pertinent discussion including authors closure, if any, will be published in
of conventional concrete.12,13 the July-August 2013 ACI Materials Journal if the discussion is received by April 1, 2013.
EXPERIMENTAL investigation
Materials and mixture proportions
the majority of research in the literature on the fatigue of
The concrete mixtures were prepared with Grade 43 ordinary
NVC and NVFRC has been focused on its behavior in
portland cement conforming to Indian Standard (IS) 8112 and
flexure.22-25 The research investigations were carried out
fly ash (Class F). The mixtures were prepared using well-
to suggest relationships between stress level S, which is
graded crushed stone coarse aggregate with a nominal size
the ratio of maximum fatigue stress fmax to the modulus
of 12.5 mm (0.49 in.) and locally available coarse sand with
of rupture fr and the number of load cycles N that causes
a fineness modulus of 2.85. A polycarboxylic-ether-based
failure. One of the extensively used fatigue equations is a
high-range water-reducing admixture (HRWRA) and a
single-log equation, as shown in the following22,25-27
polycarboxylate-polymer-based viscosity-modifying agent
(VMA) were used to achieve the flowable yet cohesive SCC
fmax and SCFRC mixtures. Corrugated steel fibers were 30 mm
S= = a b log10 ( N ) (1) (1.18 in.) in length and 1 mm (0.04 in.) in diameter in all
fr the SCFRC mixtures. Table 1 shows the proportions of all
four mixtures of SCC and SCFRC used in this investiga-
where a and b are experimental coefficients. The fatigue tion. The mixture with no steel fibersthat is, the SCC
test data of NVC and NVFRC are random in nature and mixturewas taken as the control mixture. Three different
show considerable scatter; even under carefully controlled SCFRC mixtures contained steel fibers in volume fractions
test procedures; thus, it becomes essential to introduce of 0.5, 1.0, and 1.5%. The dosage of HRWRA and VMA
probabilistic concepts to ensure adequate fatigue resis- was adjusted to obtain the required workability for all the
tance. Oh,27,28 Singh and Kaushik,25,29 and Mohammadi and SCC and SCFRC mixtures. All the mixtures were mixed in
Kaushik30 conducted experimental and theoretical studies to a 100 L (0.1 m3) drum mixer in the laboratory. First, the fine
investigate the fatigue-life distributions of NVC and NVFRC and coarse aggregates were fed into the mixer and mixed for
at different stress levels. From the test data, it was observed approximately 1 minute. The cement and fly ash were added
that the statistical distribution of the fatigue life of NVC and to the aggregates and the ingredients were mixed in a dry
NVFRC can be approximately described by the two-param- condition for approximately 30 seconds. Subsequently, two-
eter Weibull distribution.28-30 A number of investigations thirds of the water was added to the dry mixture and mixing
have been made to study the fatigue behavior of NVC and was allowed for the next 60 seconds. HRWRA premixed
NVFRC; the fatigue characteristics of SCC/SCFRC have with the remaining one-third of the water was added to the
yet to be investigated despite the fact that SCC/SCFRC has wet mixture and mixing continued for another 150 seconds.
been widely accepted for the construction of bridge deck, In the case of the SCFRC mixtures, the steel fibers were
bridge piers, and pier caps4 and possible future applications added to the wet mixture by uniformly sprinkling them into
in highway and airfield concrete pavements, wherein fatigue the drum, and then the remaining one-third of the water
is the predominant mode of loading. premixed with HRWRA and VMA was added. Mixing was
allowed for another 60 seconds for the SCFRC mixtures.
RESEARCH SIGNIFICANCE The SCC mixture did not show any sign of bleeding but the
The literature available on SCC and SCFRC reported SCFRC mixtures were unstable and bleeding was observed
studies on their mechanical properties under statically during the filling of the molds; thus, a polycarboxylate-
applied loads, but to the best of the authors knowledge, polymer-based VMA was used to improve the stability of
there is practically no information available on the fatigue the SCFRC mixtures.
Table 3Compressive strength test results for SCC and SCFRC mixtures
28-day average* compressive strength, MPa (psi)
Batch No. SCC SCFRC0.5 SCFRC1.0 SCFRC1.5
1 35.9 (5210) 36.0 (5225) 41.1 (5965) 42.0 (6096)
2 35.2 (5109) 39.9 (5791) 39.1 (5675) 39.8 (5776)
3 35.0 (5080) 36.2 (5254) 39.6 (5747) 44.0 (6386)
4 34.9 (5065) 36.9 (5355) 41.6 (6038) 39.9 (5791)
5 36.6 (5312) 39.6 (5747) 41.2 (5980) 42.9 (6226)
6 35.7 (5181) 37.7 (5472) 39.3 (5704) 40.8 (5922)
7 36.7 (5326) 38.5 (5588) 39.8 (5776) 42.7 (6197)
8 35.2 (5109) 36.1 (5239) 41.6 (6038) 43.6 (6328)
9 36.8 (5341) 39.4 (5718) 39.8 (5776) 41.8 (6067)
10 36.5 (5298) 36.6 (5312) 42.2 (6125) 43.6 (6328)
11 35.8 (5196) 39.9 (5791) 38.8 (5631) 42.9 (6226)
12 36.5 (5297) 36.6 (5312) 40.8 (5922) 43.8 (6357)
13 36.3 (5268) 38.2 (5544) 39.0 (5660) 42.5 (6168)
14 35.5 (5152) 38.8 (5631) 38.9 (5646) 41.9 (6081)
15 36.2 (5254) 40.1 (5820) 41.0 (5951) 43.8 (6357)
16 35.4 (5139) 38.3 (5559) 40.7 (5907) 41.7 (6052)
Average 35.9 (5210) 38.1 (5530) 40.3 (5849) 42.4 (6154)
*
Average of three specimens.
Note: 1000 psi = 6.89 MPa.
Workability and casting of specimens use of vibrator. The specimens were demolded 36 hours after
All the workability tests were conducted after approx- casting and moist-cured under laboratory conditions. For
imately 1 minute of final mixing; namely, slump flow, ascertaining the quality of each batch of SCC and SCFRC,
V-funnel, J-ring, and L-box tests were carried out for SCC and compressive strength tests were conducted on cube specimens
SCFRC mixtures per the guidelines of EFNARC.2 The results after 28 days of curing. The beam specimens were cured
of the workability tests conducted randomly on five batches for 75 days and thereafter stored under laboratory conditions
each of SCC and SCFRC mixtures are presented in Table 2. for approximately 2 months to minimize the effect of strength
Standard beam specimens with a size of 100 x 100 x 500 mm gain during the course of fatigue testing, which in itself is
(3.94 x 3.94 x 19.7 in.) for static flexural and flexural fatigue extended in nature. To further eliminate the effect of strength
tests and cube specimens with a size of 150 x 150 x 150 mm gain, if any, the testing was done batch-wise, wherein the static
(5.9 x 5.9 x 5.9 in.) for compressive strength tests were cast flexural strength tests on a particular batch were conducted
in different batches. Each batch contained seven beam and just prior to the flexural fatigue testing of the same. The results
three cube specimens. The mixture was poured into the of the compressive strength tests on SCC and SCFRC speci-
specimen molds from a height of approximately 450 mm mens are reported in Table 3. The average 28-day compressive
(17.7 in.) in a single layer. Molds were filled without any strength for all batches of SCC, SCFRC0.5, SCFRC1.0, and
SCFRC1.5 was 35.90, 38.10, 40.30, and 42.40 MPa (5210, the fatigue stress ratio R (R = fmin/fmax), kept constant at
5530, 5849, and 6154 psi), respectively. 0.10 throughout the investigation, as has been done in
previous studies.29-31 Constant-amplitude sinusoidal loads
Fatigue test program were applied at a frequency of 10 Hz to complete the test
The flexural fatigue testing of SCC and SCFRC was the in a reasonable period of time. Because fatigue testing is a
primary objective of this investigation. The maximum and time-consuming and expensive process and a large number
minimum load limits are required to be defined to initiate of specimens were proposed to be tested in this investiga-
a fatigue test. These load limits were obtained for each tion, an upper limit of 2 million cycles of fatigue loading
batch of specimens by testing three beam specimens from was adopted. The test was terminated when the failure of the
a particular batch in static flexure. The beams were simply specimen occurred or this upper limit was reached, which-
supported over a span of 450 mm (17.7 in.) and loaded at ever was earlier. For each SCC and SCFRC mixture, the
third points, thus leading to a four-point bending test. The numbers of cycles to failure for the specimen under different
average static flexural strength fr for each batch of SCC and load conditions were noted as fatigue life N.
SCFRC was obtained just before the fatigue tests, the results
of which are presented in Table 4. The static flexural tests ANALYSIS OF FATIGUE-LIFE DATA
were carried out with a 100 kN (22.2 kip) servo-controlled The fatigue test data obtained for the SCC and SCFRC
actuator run in the displacement control mode at a loading mixtures in this study shows considerable variability, even
rate of 0.5 mm/minute (0.02 in./minute). The static flex- at a given stress level. Thus, before initiating the analysis
ural strength taken as an average of all the batches of SCC, process, some data points may deserve consideration for
SCFRC0.5, SCFRC1.0, and SCFRC1.5 was 4.85, 6.05, 7.20, rejection as outliers. Chauvenets criteria32 was applied
and 9 MPa (704, 878, 1045, and 1308 psi), respectively. A to the fatigue-life data at different stress levels tested in
considerable increase in the peak loads over the first crack this investigation and points meeting this criterion for
loads was observed for SCFRC specimens, particularly for outliers were identified and excluded from further analysis.
mixtures containing 1.0 and 1.5% fiber-volume fractions. Batson et al.,33 Singh and Kaushik,29 and Mohammadi and
The increment in peak load may be attributed to the contri- Kaushik30 used the same criterion in their work on the
bution of fibers after the cracking of the matrix. flexural fatigue of plain NVC and NVFRC.
After the static flexural strength of a particular batch of
SCC or SCFRC was established, the remaining beam speci- Fatigue-life distributions of SCC and SCFRC
mens were tested in flexural fatigue. The loading conditions The fatigue test data of concrete is known to exhibit great
were kept the same for both static flexural and flexural variability, which becomes enhanced in the case of FRC and
fatigue tests. The flexural fatigue tests were conducted at thereby necessitates introducing the probability concepts
stress levels S (S = fmax/fr, fmax is maximum fatigue stress, and in the design to secure the adequate fatigue resistance of
fr is static flexural strength), ranging from 0.90 to 0.65 with concrete structures such as bridges, highway pavements, and
q = ua (8)
k
ni ln ( ni )
*
1 1 k
i =1
k
= ln ( ni ) (9)
ni
*
* k i =1
i =1
1 k *
* = ni (10)
k i =1
Fig. 5Comparison of shape parameters for fatigue life of Fig. 6Comparison of shape parameters for fatigue life of
SCC, NVC, SCFRC, and NVFRC with Vf = 0.5% of steel fibers. SCFRC and NVFRC with Vf = 1.0% of steel fibers.
Goodness-of-fit test
As shown in the preceding section, the fatigue-life distri-
butions of SCC and SCFRC at different stress levels can
Fig. 8Fatigue curves of SCC and SCFRC corresponding approximately be described by the two-parameter Weibull
to 10% probability of failure (Pf = 0.1) using single-log distribution. Further, the values of the correlation coefficient
failure equation. at each stress level also substantiated this. In addition, the
Kolmogorov-Smirnov test was applied as goodness of fit to
ability in the distribution of fatigue life of SCC and SCFRC the fatigue-life data at each stress level tested in this inves-
compared to NVC and NVFRC, respectively. For example, tigation; it was observed that the model was acceptable at a
at stress level S = 0.85, the average value of the shape param- 5% significance level.29,30,32 The calculations to this effect
eter for the fatigue life of SCC, obtained by the different are not given.
methods of analysis in this investigation, is 4.3471 compared
with 3.8920 and 3.5457 reported by Oh28 and Mohammadi Flexural fatigue performance of SCC and SCFRC
and Kaushik,30 respectively, for NVC, as plotted in Fig. 5. In the preceding sections, the flexural fatigue-life data
This trend is established at all other stress levels tested in of SCC and SCFRC were shown to follow the two-param-
this study. In particular, the shape parameter of SCC is eter Weibull distribution at different stress levels. This can
higher than that of NVC by 17%, 36%, 34%, 41%, and 36% further be used to calculate the fatigue lives corresponding
at stress level S = 0.85, 0.80, 0.75, 0.70, and 0.65, respec- to different failure probabilities Pf.29
tively. A maximum decrease of approximately 15% in the Substituting 1 Pf = LN in Eq. (3), the following relation
coefficient of variation in the fatigue-life data of SCC was is obtained
observed as compared to NVC.
Similar trends have been observed for fatigue life of
SCFRC with 0.5, 1.0, and 1.5% volume fractions of steel 1
fibers. The shape parameters calculated by all three aforemen- ln ln = ln ( n) ln(u) (11)
tioned methods for SCFRC were found to be greater than the 1 Pf
shape parameters for NVFRC with the same volume fractions
of steel fibers at different stress levels tested in the investiga- Rearranging Eq. (11)
tion. For example, the average value of the shape parameter
for the fatigue life of SCFRC with 0.5% volume fractions of
steel fibers at stress level S = 0.80 was 2.0985, as compared 1
with 1.5448 reported by Singh and Kaushik25 for NVFRC ln ln + ln(n)
with 0.5% steel fibers. The average value of the shape param- 1 Pf
eters for SCFRC with 1.0 and 1.5% volume fraction of steel N = ln 1 (12)
fibers is 1.6904 and 1.6634 at stress level S = 0.80, compared
to 1.2385 and 1.4376 reported by Singh and Kaushik29 and
Mohammadi and Kaushik,30 respectively, for SFRC. The
maximum increase of 37%, 42%, and 55% in the shape param-
eter for SCFRC was observed compared to NVFRC with Equation (12) can be used to calculate the fatigue life N
0.5%, 1.0%, and 1.5% of steel fibers, respectively, reported by for a particular probability of failure Pf. Using the average
Singh and Kaushik29; at the same time, a maximum decrease values of the parameters of the Weibull distribution for
of 21%, 25%, and 28% in the coefficient of variation for the fatigue-life data at a given stress level S as listed in Table 5,
fatigue-life data of SCFRC was observed with 0.5%, 1.0%, Eq. (12) is used to calculate the fatigue lives for SCC and
and 1.5% volume fraction of steel fibers, respectively. It can SCFRC with 0.5, 1.0, and 1.5% of volume fractions of steel
also be observed from Fig. 5 to 7 that the shape parameter fibers, corresponding to a failure probability of 10%that is,
decreases as the fiber-volume fraction increases, resulting in Pf = 0.1. These calculated values of fatigue lives are plotted
higher variability in the fatigue life of SCFRC at higher fiber- in Fig. 8 to obtain fatigue curves for SCC and SCFRC with
volume fractions, even at the same stress level. In general, for 0.5, 1.0, and 1.5% of volume fraction of steel fibers for a
all mixtures of SCFRC with different volume fractions of steel failure probability of 0.1.
Table 7Theoretic fatigue lives (number of cycles) for SCC and SCFRC calculated by single-log fatigue
equation corresponding to 10% probability of failure
Stress level, S 0.90 0.85 0.80 0.75 0.70
Mixture
SCC Theoretic fatigue life 323 1463 6630 30,039 136,100
Theoretic fatigue life 1068 4234 16,783 66,522 263,665
SCFRC0.5
Enhanced extent, % 230 189 153 121 93
Theoretic fatigue life 3259 14,590 65,329 292,513 1,309,747
SCFRC1.0
Enhanced extent, % 908 897 885 873 862
Theoretic fatigue life 1101 4747 20,460 88,192 380,145
SCFRC1.5
Enhanced extent, % 240 224 208 193 179
In this study, a single-log fatigue equation, Eq. (1), was five different stress levels. It can be seen that the theoretic
used to evaluate the performance of SCC and SCFRC corre- fatigue lives of SCFRC containing 0.5, 1.0, and 1.5% steel
sponding to a probability of failure Pf = 0.1. The single-log fibers increased to a different extent. With the increasing
fatigue equation is commonly used by researchers to describe stress level, the enhanced extent of the theoretic fatigue lives
the relation between stress level S and fatigue life N.22,25,37 In of the SCFRC mixtures increased, which indicates that the
Eq. (1), the fatigue performance is dependent on the two SCC containing steel fibersthat is, SCFRChas an excel-
important coefficients/parameters a and b. The parameter a lent fatigue performance, particularly at higher stress levels
reflects the height of the fatigue curve. The larger the param- (corresponding to heavy traffic load) compared with SCC.
eter a, the higher the fatigue curve. The parameter b reflects It can be seen from Table 7 that the enhanced extent of
the steep degree of the fatigue curve. The larger the parameter the theoretic fatigue life of SCFRC with 1.0% of steel fibers
b, the steeper the fatigue curve, and the fatigue life of concrete is the highestthat is, 908%at a stress level of 0.90. A
is more sensitive to the change in stress.37 The parameters a similar increase was observed for SCFRC with 1.0% of steel
and b of Eq. (1) are obtained from regression analysis for fibers at all the other stress levels, indicating a superior flex-
the fatigue curves of SCC and SCFRC plotted in Fig. 8. The ural fatigue performance of SCFRC1.0 compared to SCC and
estimated fatigue equations and the corresponding correla- other SCFRC mixtures used in the investigation. It can also
tion coefficients are also presented in Table 6. The param- be seen from the results that the fatigue performance of SCC
eters a and b of the single-log equation and their enhanced is improved with the addition of steel fibers. It may, however,
extent generated for SCC and SCFRC with 0.5, 1.0, and 1.5% be noted that the fatigue performance herein is represented
of volume fraction of steel fibers for a failure probability of in terms of applied maximum fatigue stress expressed as a
10%that is, Pf = 0.1are also listed in Table 6. It can be percentage of corresponding static flexural stressthat is, in
seen that the regression parameters of fatigue equations of all terms of stress level S. In contrast, when the fatigue perfor-
SCFRC mixtures are increased to a different extent, and all mance is examined in terms of applied maximum fatigue
correlation coefficients are higher than 0.97. The increase in stress fmax, the ranking differs. Increasing the fiber content
the value of a indicates that the flexural fatigue performance from 0.5 to 1.5% seems to improve the fatigue performance
of all SCFRC mixtures is significantly improved. The value in terms of applied maximum fatigue stress. Similar trends
of b increased for all SCFRC mixtures, indicating an increase were obtained by earlier investigators38,39 while studying the
in the sensitivity of their fatigue lives to change of stress. flexural fatigue performance of NVFRC.
Compared with the SCC and other SCFRC mixtures, SCFRC It may be noted that the results, such as parameters of the
with 1.0% of steel fibers has the largest enhanced extent of Weibull distribution and theoretic fatigue life of SCFRC
7.30% for a and the smallest enhanced extent of 0.78% for b, reported in this paper, are applicable to the type, size, and
thus indicating that the fatigue performance of SCFRC with volume fraction of the fibers used; therefore, additional work
1.0% of steel fibers is improved to the largest extent and the is required to generate more data for other types, sizes, and
sensitivity of its fatigue life to change in stress is increased to volume fraction of fibers.
the smallest extent.
Table 7 shows the theoretic fatigue lives of various SCC CONCLUSIONS
and SCFRC mixtures calculated by a single-log fatigue 1. The probability distributions of fatigue life of SCC
equation corresponding to a 10% probability of failure at and SCFRC, at different stress levels, can be approximately