Engineering Structures 102 (2015) 409423

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Structural performance of ultra-high-performance concrete

beams with different steel fibers
Doo-Yeol Yoo a, Young-Soo Yoon b,
a
b

Department of Civil Engineering, The University of British Columbia, 6250 Applied Science Lane, Vancouver, BC V6T 1Z4, Canada
School of Civil, Environmental and Architectural Engineering, Korea University, 145 Anam-ro, Seongbuk-gu, Seoul 136-713, Republic of Korea

a r t i c l e

i n f o

a b s t r a c t

Article history:
Received 13 June 2015
Revised 22 August 2015
Accepted 25 August 2015
Available online 5 September 2015

In this study, ten large ultra-high-performance concrete (UHPC) beams reinforced with steel rebars were
fabricated and tested. The experimental parameters included reinforcement ratio and steel fiber type.
Two different reinforcement ratios (q = 0.94% and 1.50%) and steel fiber types (smooth and twisted steel
fibers) were adopted. In addition, three different fiber lengths (Lf = 13, 19.5, and 30 mm) for the smooth
steel fibers and one fiber length (Lf = 30 mm) for the twisted steel fiber were considered. For a control
specimen, a UHPC matrix without fiber was also considered. Test results indicated that the addition of
steel fibers significantly improved the load carrying capacity, post-cracking stiffness, and cracking
response, but it decreased the ductility. Specifically, with the inclusion of 2% by volume of steel fibers,
approximately 2754% higher load carrying capacity and 1373% lower ductility were obtained. In
addition, an increase in the length of smooth steel fibers and the use of twisted steel fibers led to the
improvements of post-peak response and ductility, whereas no noticeable difference in the load carrying
capacity, post-cracking stiffness, and cracking response were obtained according to the fiber length
and type. Sectional analysis incorporating the suggested material models was also performed based on
AFGC/SETRA recommendations, and the ratios of flexural capacities obtained from experiments and
numerical analyses ranged from 0.91 to 1.19.
2015 Elsevier Ltd. All rights reserved.

Keywords:
Ultra-high-performance concrete
Flexure
Steel fiber
Ductility
Sectional analysis
Fiber orientation

1. Introduction
Ultra-high-performance concrete (UHPC) has been developed in
many countries [14], and many new studies have been performed
in recent years. Because UHPC has a very low water-to-binder ratio
(W/B), high-fineness admixtures, and high volume contents of steel
fibers (mostly 2% by volume), it exhibits excellent performance in
terms of mechanical properties (compressive strength >150 MPa
and tensile strength >8 MPa), energy absorption capacity, fatigue
performance, and durability [5,6]. In particular, owing to its unique
strain-hardening and multiple cracking behaviors, UHPC has been
attractive for use in civil infrastructures subjected to tensile and
bending loads.
Graybeal [7] and Chen and Graybeal [8] studied the structural
behaviors of full-scale prestressed UHPC I- and Pi-girders under
flexure based on experiments and numerical simulations. Yuguang
et al. [9] also performed a feasible study for applying UHPC in
bridge decks, and Yang et al. [10,11] carried out several structural
Corresponding author. Tel.: +82 2 3290 3320; fax: +82 2 928 7656.
E-mail addresses:
(Y.-S. Yoon).

dyyoo2@gmail.com

(D.-Y.

http://dx.doi.org/10.1016/j.engstruct.2015.08.029
0141-0296/ 2015 Elsevier Ltd. All rights reserved.

Yoo),

ysyoon@korea.ac.kr

tests of reinforced UHPC beams under flexure for attaining a fundamental understanding. Fujikake et al. [12] and Yoo et al. [13] performed a drop-weight impact test for UHPC beams reinforced with
prestressing tendons and steel rebars and suggested analytical
models for predicting their deflection responses. In addition, Astarlioglu and Krauthammer [14] numerically analyzed the blast resistance of normal-strength concrete and UHPC columns, and they
reported that the UHPC columns exhibited a lower mid-span
deflection and sustained more than four times the impulse than
those of the normal-strength concrete columns. Likewise, many
researchers [716] have performed structural tests for various
elements made of UHPC containing short smooth steel fibers.
For the material level, numerous studies [5,1722] have been
performed to investigate the effect of fiber properties (i.e., fiber
type, geometry, dosage, orientation, etc.) on the mechanical performance of UHPC. Kim et al. [20] investigated the effects of different
macro steel fibers on the flexural behaviors of hybrid UHPC. Based
on the test results, they noted that the uses of longer hooked-end
steel fibers and twisted steel fibers provide better flexural performance including flexural strength, deflection capacity, and energy
absorption capacity than that of straight steel fibers. Yoo et al. [21]
experimentally and numerically estimated the mechanical and

410

D.-Y. Yoo, Y.-S. Yoon / Engineering Structures 102 (2015) 409423

fracture properties of UHPC according to the steel fiber content and

reported that the flexural performance and fracture energy of
UHPC were almost linearly increased with the fiber content even
though an insignificant difference in the first cracking properties
was observed. In addition, Barnett et al. [22] investigated the effect
of placement method on the biaxial flexural performance of UHPC.
From their test results, the panels poured from the center were significantly stronger than the panels poured from the edge and randomly, since the alignment of fibers led to more fiber bridging
against the radial cracks formed.
The application of UHPC to real structures has been limited thus
far because of its high cost. One of the methods reducing the production cost is to use coarse aggregate, since by using the coarse
aggregate, the amount of powder can be reduced [23]. Thus, some
researchers have recently developed the UHPC with coarse aggregate and investigated its properties. Ma and Orgass [23] performed
comparative investigations on the mixing process and mechanical
and shrinkage properties of UHPC with and without coarse aggregate. In their study, several findings were obtained; (1) the UHPC
with coarse aggregate was easier to be fluidized and reduced the
mixing time, (2) there was no distinct difference in the compressive strength according to the existence of coarse aggregates,
whereas higher elastic modulus and lower strain capacity were
obtained for the UHPC with coarse aggregate, and (3) autogenous
shrinkage of UHPC with coarse aggregate was approximately 60%
of its counterpart. Wille et al. [24] also investigated the effect of
coarse aggregate on the compressive strength of UHPC based on
the database from the International Symposium on Ultra High
Performance Concrete. They reported that the UHPC containing
coarse aggregate with the maximum size ranged from 7 to
16 mm reached a slightly higher strength of 178 MPa on average
than its counterpart without coarse aggregate, which reached a
strength of 162 MPa on average. On the contrary, the information
with regard to the tensile performance of UHPC containing coarse
aggregate, which is one of the most important parameters for
designing the structures, is very limited, and even though various
properties of UHPC might be changed according to the type of
coarse aggregate [23], no specific recommendation for the type
or performance of coarse aggregate exists. Therefore, further
extensive research is required to be done for using UHPC containing coarse aggregate.
Another method is to reduce the steel fiber content, since the
price of high strength steel fibers is considerably expansive compared to other drying components for making the matrix material.
For instance, 2% by volume of smooth steel fibers occupies
approximately 33% of total cost of UHPC used in this study. Therefore, to reduce the steel fiber content without deteriorating its
tensile performance or to improve its tensile performance without
changing the steel fiber content at the material level is one of the
key challenges for UHPC technology. Yoo et al. [5] and Aydn and
Baradan [25] very recently reported that the flexural performance
and fracture energy are noticeably improved with an increase in
the length of smooth steel fibers (or the aspect ratio), which is
attributed to an increase in the effective bonding area between
the fiber and matrix. Furthermore, Wille et al. [17] noted that the
use of twisted steel fibers (Lf/df = 30/0.3 mm/mm) improves the

tensile strength and strain capacity, although lower fiber contents

are adopted compared to that of the short smooth steel fibers (Lf/
df = 13/0.2 mm/mm) (herein, Lf is the fiber length and df is the fiber
diameter). These results indicate that the fiber content required to
satisfy a certain tensile or flexural performance at the material
level can be decreased by increasing the length of smooth steel
fibers and by using the twisted steel fibers. However, to the best
of authors knowledge, no published study exists that is related
to structural applications of these findings to verify the possibility
of improving the performance by increasing the length of smooth
fibers and by using the twisted fibers. Thus, investigations on the
structural performance of UHPC elements subjected to tension or
flexure are required for different fiber types and lengths.
Accordingly, this study investigated the flexural behavior of
steel bar-reinforced beams made of UHPC containing no coarse
aggregate with two different reinforcement ratios. To evaluate
the effects of the type and length of steel fiber on the flexural
behaviors of UHPC beams, two different fiber types (smooth and
twisted steel fibers) and three different lengths of smooth steel
fibers (Lf = 13, 19.5, and 30 mm) were considered. As control specimens, the beams made of UHPC matrix without fiber were also
fabricated and tested. Lastly, to predict their flexural behaviors,
sectional analysis incorporating suggested material models from
experiments and inverse analyses was performed and verified
through comparison with the test data.

2. Experimental program
2.1. Materials and mixture proportions
The mixture proportions used in this study are summarized in
Table 1. Type 1 Portland cement and silica fume were included
as cementitious materials. The chemical and physical properties
of used cementitious materials used can be found in a previous
study [26]. Silica flour including 98% SiO2 with a diameter of
2 lm was added as a filler, and silica sand with a grain size less
than 0.5 mm was used as a fine aggregate. In order to improve
the homogeneity, a coarse aggregate was excluded in the mixture,
similar to the classic UHPC from previous studies [6,12,17]. To
investigate the effects of the length and type of steel fibers on
the mechanical and structural performances, three different fiber
lengths of 13, 19.5, and 30 mm [5] for smooth steel fibers (S) and
one 30-mm-long twisted steel fiber (T) having a triangular section
and three ribs within the fiber length were considered at 2% (by
volume), which led to four series of test specimens. The shape of
the twisted steel fiber is shown in Fig. 1 [18]. Since several design
guidelines of UHPC [4,27] proposed the tension-softening model
based on the UHPC mixture including 2 vol.% of steel fibers, most
of previous studies [7,1013] on the structural response of UHPC
used 2 vol.% of steel fibers. Therefore, the fiber volume fraction of
2% was also adopted in this study for data consistency. Based on
our preliminary mixing [28], the UHPC mixture containing 2 vol.
% of smooth steel fibers exhibited no significant difference of fluidity according to the fiber aspect ratio ranging from 65 to 100,
whereas the mixture with a fiber aspect ratio higher than 100

Table 1
Mix proportions.
Unit weight (kg/m3)

UHPC w/o fiber

UHPC w/fiber

Water

Cement

Silica fume

Silica flour

Silica sand

EA

SRA

SP (%)

165.5
160.3

786.6
788.5

196.7
197.1

236.0
236.6

865.3
867.4

59.0

7.9

1.7
2.0

Note: EA = expansive admixture, SRA = shrinkage reducing admixture, and SP = superplasticizer.

411

D.-Y. Yoo, Y.-S. Yoon / Engineering Structures 102 (2015) 409423

Fig. 1. Twisted steel fiber [18]; (a) shape of fiber, (b) cross-section of fiber.

Table 2
Properties of steel fibers.
Type of fiber

df
(mm)

Lf
(mm)

Aspect
ratio
(Lf/df)

Density
(g/cm3)

Tensile
strength
(MPa)

Elastic
modulus
(GPa)

Smooth steel
fiber

0.2
0.2
0.3
0.3

13.0
19.5
30.0
30.0

65.0
97.5
100.0
100.0

7.9
7.9
7.9
7.9

2788
2500
2580
2428

200
200
200
200

Twisted steel
fiber

10-mm LVDTs

Note: df = diameter of fiber and Lf = length of fiber.

exhibited slight formation of fiber ball, leading to insufficient fiber

dispersion. Therefore, the maximum aspect ratio of steel fibers
used in this study was determined by 100, as given in Table 2.
The specimens are generally divided into two series; (1) the specimens with smooth steel fibers (SX) and (2) the specimens with
twisted steel fibers (TX). The letters S, T, and X denote the smooth
steel fiber, twisted steel fiber, and length of fiber. For instance, S30
indicates the UHPC specimen including smooth steel fibers with a
length of 30 mm. The physical and geometrical properties of the
steel fibers used are listed in Table 2. In addition, for a control
specimen, a matrix having mixture proportions similar to that of
UHPC but without fiber (hereafter called NF) was also investigated. Owing to the extremely high free shrinkage and low tensile
strength of NF, restrained shrinkage cracking was observed at an
early age without external load. Therefore, to prevent the cracking
caused by the restraint of shrinkage from the inner steel rebar, 7.5%
expansive admixture (EA) and 1% shrinkage-reducing admixture
(SRA) were incorporated into the NF mixture, as indicated in
Table 1. According to previous test results performed by Yoo
et al. [29], approximately 3942% of free shrinkage strain was
reduced at 7 days by including 7.5% EA and 1% SRA with slightly
higher strengths. By using the NF mixture with 7.5% EA and 1%
SRA, no shrinkage cracking was observed until the testing day.
Since UHPC has an extremely low W/B and no coarse aggregate,
the mixing sequence is different from that for conventional
concrete. First, all drying components such as cement, silica fume,
silica flour, and silica sand were premixed for approximately
10 min. Subsequently, water premixed with polycarboxylate
superplasticizer having a density of 1.01 g/cm3 and dark brown
color was added in the dry state and mixed for further 10 min.
Once the mixture became flowable, steel fibers were dispersed
and mixed for an additional 5 min.
2.2. Compression and flexure tests (ASTM C 39 and JCI-S-0022003)
A total of fifteen cylindrical specimens (three cylinders for each
variable) having a diameter of 100 mm and length of 200 mm were

Cylindrical
Specimen
( 100 200 mm)

Compressometer

Fig. 2. Compression test.

fabricated and tested as per ASTM C 39 [30]. The detailed test setup
is shown in Fig. 2. A uniaxial load was applied using a universal
testing machine (UTM) with a maximum capacity of 3000 kN
through displacement control at a rate of 0.1 mm/min. In order
to obtain the average stressstrain curve along with elastic modulus and strain capacity, a compressometer equipped with three
linear voltage differential transformers (LVDTs) was installed.
A total of fifteen prismatic beams (three beams for each variable) with a cross-sectional area of 100 mm
100 mm and length
of 400 mm were fabricated and tested as per JCI-S-002-2003 standard [31]. A clear span of 300 mm was used, and a mid-span notch
of 30 mm (30% of the beam height) with a width of 4 mm was
applied, as shown in Fig. 3. A uniaxial load was applied using a
UTM with a maximum capacity of 3000 kN under displacement
control. To exclude the support settlement from the mid-span
deflection, two frames equipped with two LVDTs were installed
at the middle of the beam height on both sides. In addition, a clip
gage with a maximum capacity of 10 mm was installed at the midspan notch to measure the crack mouth opening displacement
(CMOD).
Since UHPC exhibits flowable and self-consolidating characteristics, in most cases, the specimens have been fabricated by placing
concrete at a certain point and allowing it to flow [5,32]. However,
according to the placement method, the flexural performance was
significantly changed because of the different fiber orientation and
dispersion. Therefore, in order to provide similar fiber orientation
and dispersion, all test specimens in this study were identically
fabricated by placing concrete at the end and allowing it to flow.
All test specimens (both cylinders and prisms) were covered
with plastic sheets immediately after concrete casting and cured
at room temperature for the first 48 h, prior to demolding. After
demolding, the specimens were heat cured (90 2 C) with steam

412

D.-Y. Yoo, Y.-S. Yoon / Engineering Structures 102 (2015) 409423

P
LOAD CELL

50

50

150

150

100

SPECIMEN

LVDT
100

100
Frame for LVDT

30 mm notch
Clip gage
Roller
Sitting

TEST MACHINE FIXED SUPPORT

(a)

(b)

Fig. 3. Three-point flexure test; (a) test picture, (b) geometry and test setup (unit: mm).

for 3 days and then stored in a laboratory at room temperature

until the testing day.
2.3. Details of structural test specimen and instrumentation
A total of ten large UHPC beams were fabricated and tested under
four-point bending load. The details of sectional geometry and reinforcement are shown in Fig. 4. All test beams have a cross-sectional
area of 150 mm
220 mm and length of 2500 mm. Because the
steel fibers included in the mixture can substitute a part of longitudinal steel rebars, all test beams were designed to have reinforcement ratios below 2%, similar to what was done in a previous
study [10]. Based on the test results performed by Yoo [33], steel

b = 150

and glass fiber-reinforced polymer (GFRP) bar-reinforced UHPC

beams without shear reinforcement exhibited both flexural and
shear failure modes even though their reinforcement ratio was less
than 2%. Therefore, to provide a flexural failure mode for all test
beams, shear reinforcement was designed on the basis of NF specimens (without fiber) and consisted of closed stirrups made from
steel rebar with a diameter of 9.53 mm. The beams were also reinforced with one layer of steel rebars having two different diameters
[ds = 12.7 mm (D13) and 15.9 mm (D16)], leading to two different
reinforcement ratios (q = 0.94% and 1.50%) calculated by the equation of As/bd. Herein, As is the area of the rebar, b is the width of
the beam, and d is the effective depth of the beam. The properties
of the steel rebars used are summarized in Table 3. Thus, the beams

b = 150

d = 177

170
D13
(SD400)

25

D10
(SD400)

h = 220

D10
(SD400)

h = 220

d = 179

25

170
D16
(SD400)

25
25

100

25
25

25

Reinforcement ratio ( = 0.94%)

100

25

Reinforcement ratio ( = 1.50%)

(a)
150

900

P/2

400

P/2

900

150

15 30

T1
T2
T3

60
110
190

Strain gage

D10 @ 80 mm
LVDT

B1

(b)

(c)

Fig. 4. Details of test program (unit: mm); (a) cross-section details of test beams, (b) details of test setup, (c) locations of strain gages.

413

D.-Y. Yoo, Y.-S. Yoon / Engineering Structures 102 (2015) 409423

Table 3
Properties of deformed steel rebars.

Longitudinal reinforcement

Transverse reinforcement

Name

ds (mm)

As (mm2)

fy (MPa)

Es (GPa)

ey

eu

D16
D13
D10
D10

15.90
12.70
9.53
9.53

198.6
126.7
71.3
71.3

510
495
491
491

200
200
200
200

0.0026
0.0025
0.0025
0.0025

0.19
0.19
0.20
0.20

Note: ds = diameter of rebar, As = area of rebar, fy = yield strength, Es = elastic modulus, ey = yield strain, and eu = ultimate strain.

are divided into two series: a beam series reinforced with steel rebar
having a reinforcement ratio of 0.94% and a beam series reinforced
with steel rebar having a reinforcement ratio of 1.50%. Then, the
final designation system for UHPC beams consisted of the fiber type
and reinforcement ratio. For example, the specimen S13-0.94%
denotes UHPC beams including smooth steel fibers with a length
of 13 mm and steel rebars with a reinforcement ratio of 0.94%. In
recent years, Yoo et al. [19] have reported that the deformed steel
rebar (SD400) embedded in UHPC yields at an embedment length
of two times the rebar diameter before pullout. Therefore, the
requirement for development length was satisfied for all test beams
without the end hook of longitudinal steel rebars. Because of the
shear reinforcement, it was hard to place UHPC at the end of the
beam and allow it to flow, and thus, all the steel bar-reinforced
beams were similarly fabricated by placing the concrete back and
forth along the beam length.
Four-point loading was monotonically applied using a UTM
with a maximum capacity of 2000 kN, and the loads, deflections,
and strains were simultaneously recorded. Owing to its strainhardening behaviors, the maximum crack width was significantly
lower nearly up to the peak load, similar to what was observed
in a previous research [33]. Therefore, at each loading stage (10
or 20 kN interval), only the number of cracks and average crack
spacing were recorded. In order to obtain the net mid-span deflection, the support settlements were subtracted from the measured
mid-span deflection by using LVDTs. In addition, strain gages were
attached to the center of all longitudinal reinforcements to measure the strains in the reinforcements, and five strain gages were
attached to the side surface of the beam to determine the neutral
axis depth, as shown in Fig. 4(c).
3. Experimental results and discussion
3.1. Material properties (compression and flexure)
The average compressive stressstrain curves for all test series
are shown in Fig. 5(a), while the average parameters such as

Table 4
Summary of mechanical test results.
Name

NF
S13
S19.5
S30
T30

Load (kN)

150
1000
100
S13
S19.5
S30
T30

0.2

0.4

dMOR
(mm)

CMODMOR
(mm)

45265.0
(2464.2)
46732.5
(469.5)
46880.5
(169.9)
46772.9
(669.1)
46971.6
(2704.8)

8.18
(0.43)
19.26
(2.28)
30.69
(8.87)
31.91
(5.86)
32.24
(1.81)

0.0034
(0.0033)
0.54
(0.053)
0.75
(0.32)
1.57
(0.11)
1.06
(0.11)

0.028
(0.0015)
0.66
(0.063)
0.94
(0.40)
2.06
(0.17)
1.36
(0.15)

200.9
(11.515)
211.8
(5.710)
209.7
(2.846)
209.7
(1.482)
232.1
(8.503)

0.00453
(0.000175)
0.00484
(0.000281)
0.00458
(0.000236)
0.00528
(0.000249)

Ec

0:4
f c  f 1
e2  0:00005

where fc0 is the ultimate compressive strength, f1 is the stress corresponding to a longitudinal strain of 50
106, and e2 is the longitudinal strain produced by stress at 40% of fc0 .
Due to the catastrophic failure of NF immediately after reaching
the peak load, its strain capacity and average stressstrain curve
were not measured. A compressometer was only used until the
compressive stress reached approximately 50% of the strength for
obtaining the elastic modulus, and the ultimate strength was measured without a compressometer. The specimens with steel fibers
showed a slightly higher compressive strength and elastic modulus

0.6

50
0

40

0.006

0.012

0.018
NF
S19.5
T30

30

0.024
36

S13
S30

27

20

18

10

Displacement (mm)

Deflection (mm)

(a)

(b)

Flexural stress (MPa)

200

fMOR
(MPa)

Normalized deflection, /L

1500

Ec (MPa)

ec0 (mm/

compressive strength, elastic modulus, and strain capacity (strain

at the peak load) are listed in Table 4. The elastic modulus was calculated from the stressstrain curve according to ASTM C 469M as
follows [34]:

0.006
250

500

fc0 (MPa)

Note: fc0 = compressive strength, ec0 = strain at the peak load, Ec = elastic modulus,
fMOR = flexural strength (modulus of rupture), dMOR = deflection at the peak load,
CMODMOR = CMOD at the peak load, and (x.xxx) = standard deviation.
a
Data is not available.

Load (kN)

0.004

Compressive stress (MPa)

2000

0.002

Flexure test (JCI-S-002-2003

[31])

mm)

Strain (mm/mm)
0

Compression test (ASTM C 39

[30])

Fig. 5. Mechanical test results; (a) average compressive stressstrain curve, (b) average flexural loaddeflection curve.

414

D.-Y. Yoo, Y.-S. Yoon / Engineering Structures 102 (2015) 409423

160

160

Load (kN)

120

(b) = 1.50%

NF
S13
S19.5
S30
T30

120

80

80

40

40

20

40

60

80

0
100

20

40

60

80

Load (kN)

(a) = 0.94%

0
100

Deflection (mm)

3.2. Structural behavior of reinforced UHPC beams under flexure

Fig. 6. Loaddeflection curves of steel bar-reinforced UHPC beams.

than that without fiber. The highest compressive strength and

strain capacity were obtained for T30. For example, the average
compressive strength of T30 was 232.1 MPa, which is approximately 15.5% higher than that of NF. No noticeable difference in
elastic modulus was obtained according to the length and type of
fiber (average elastic modulus of UHPC with steel fibers was found
to be 46839.4 MPa). All cylinders showed a linear stressstrain
curve up to failure and then failed in a brittle manner (the load
suddenly dropped to zero immediately after reaching the peak).
However, insignificant fragmentation was obtained for UHPC with
steel fibers, owing to the fiber confinement effect [35].
The average flexural loaddeflection curve is illustrated in Fig. 5
(b), and the average parameters are given in Table 4. The flexural
stress of a notched beam under center-point loading was
calculated using Eq. (2) as follows:

found to be between S30 and S19.5. In addition, the specimen T30

showed a steeper decrease in the load carrying capacity after the
peak load (a lower post-peak ductility) compared to that of the
specimen S30 having identical fiber length and aspect ratio. The
results of higher strength and lower post-peak ductility for twisted
steel fibers are consistent with the findings of Kim et al. [20]. Based
on these observations, it can be concluded that the use of longer
smooth steel fibers and twisted steel fibers are effective in
improving the flexural performance of UHPC at the material level.

3PL

2bh  a0

where P is the applied load, L is the span length, b is the beam

width, h is the beam depth, and a0 is the notch depth.
The flexural performance including the strength, deflection
capacity, and CMOD at the peak were significantly improved by
using both smooth and twisted steel fibers. This improvement is
attributed to the deflection-hardening behavior resulting from fiber
bridging at the crack surfaces. For example, the specimen T30
exhibited the highest flexural strength of 32.2 MPa, which is
approximately 4 times higher than that of NF (without fiber). In
addition, the flexural performance improved with an increase in
the length of smooth steel fiber, because on using longer fibers with
higher aspect ratios, the effective bonding area between the fiber
and matrix increased [5,25]. The highest flexural strength was
obtained for the specimen T30, and its deflection capacity was

3.2.1. Load versus deflection response

The loaddeflection curves for all test specimens are shown in
Fig. 6. In addition, the loads and deflections at the first cracking,
steel rebar yield, and peak load are summarized in Table 5. Because
UHPC including 2 vol.% of steel fibers shows strain-hardening
response with multiple micro-cracks [36], it is difficult to detect
the first cracking point with the naked eye. Therefore, in order to
determine the first cracking point precisely, a total of six strain
gages with a length of 60 mm were attached underneath the
beams in the maximum moment region. The first cracking point
was defined as the point at which the strain measured from the
strain gages suddenly changed with the first crack occurrence.
The first cracking load and the corresponding deflection
decreased by including the steel fibers, probably because of the
inhomogeneous distribution of the steel fibers within the matrix
by the internal steel rebars. In addition, the first cracking load
slightly decreased with an increase in the reinforcement ratio.
For example, approximately 718% lower first cracking loads were
obtained by increasing the reinforcement ratio from 0.94% to
1.50%. This is caused by the fact that a higher residual tensile stress
obtained in concrete prior to the external load due to the restraint
of shrinkage from the internal steel rebars was generated with an
increase in the reinforcement ratio. This is consistent with the findings from Yoo et al. [37] that higher reinforcement ratios led to
higher residual stresses caused by the restraint of autogenous
shrinkage of UHPC.
The flexural stiffness before the occurrence of first cracking was
insignificantly influenced by the steel fibers, whereas the postcracking stiffness and peak load were substantially improved by
including 2 vol.% smooth and twisted steel fibers. In particular,
the specimen NF-1.50% exhibited the peak load of 97.9 kN, which
is similar to those (87.396.6 kN) of the UHPC beams including
steel fibers with a lower reinforcement ratio of 0.94%. This result
is inconsistent with the findings of a previous study [38] on
conventional steel-fiber-reinforced concrete (SFRC) exhibiting

Table 5
Summary of flexural test results for reinforced UHPC beams.
Name

NF
S13
S19.5
S30
T30

Reinforcement ratio q (%)

0.94
1.50
0.94
1.50
0.94
1.50
0.94
1.50
0.94
1.50

First cracking

Yielding state

Peak state

Ultimate state

Ductility index

Pcr (kN)

Dcr (mm)

Py (kN)

Dy (mm)

Pp (kN)

Dp (mm)

Du (mm)

Dp /Dy

Du/Dy

36.6
30.6
26.6
23.3
18.0
16.7
21.3
18.7
18.0
14.7

1.12
1.09
0.75
0.67
0.82
0.63
1.12
0.61
0.78
0.51

46.0
77.9
80.6
109.9
78.0
103.3
79.9
105.3
77.9
111.9

9.15
12.06
11.96
12.73
11.54
12.29
11.33
13.01
11.03
13.22

62.6
97.9
87.3
124.1
93.3
125.2
95.9
124.6
96.6
133.9

94.53
73.03
28.41
20.30
30.51
43.35
30.46
45.28
36.22
43.64

94.63
73.13
52.61
51.43
50.68
65.63
79.81
72.90
65.74
81.88

10.33
6.06
2.38
1.59
2.64
3.53
2.69
3.48
3.28
3.30

10.34
6.06
4.40
4.04
4.39
5.34
7.04
5.60
5.96
6.20

Failure mode

Flexurea
Flexurea
Flexureb
Flexureb
Flexureb
Flexureb
Flexureb
Flexureb
Flexureb
Flexureb

Note: Pcr = first crack load, Dcr = deflection at the first crack, Py = load at steel rebar yield, Dy = deflection at steel rebar yield, Pp = peak load, Dp = deflection at the peak, and
Du = ultimate deflection.
a
Flexural failure occurs with concrete crushing.
b
Flexural failure occurs with rebar rupture.

415

D.-Y. Yoo, Y.-S. Yoon / Engineering Structures 102 (2015) 409423

(1) Load = 40 kN

(1) Load = 40 kN

(2) Load = 60 kN

(2) Load = 60 kN

(3) Load = 80 kN

(3) Load = 80 kN

(4) Load = 90 kN

(4) Load = 120 kN

(5) At failure

(5) Load = 130 kN

(6) At failure
Fig. 7. Effect of steel fibers on cracking response; (a) NF-1.50%, (b) T30-1.50%.

160

160

Load (kN)

120

80

80

40

40

0
0

20

40

60

80

100
0

Number of cracks

Load (kN)

NF
S13
S19.5
S30
T30

120

0
10

Ave. crack spacing (mm)

(a)
160
NF
S13
S19.5
S30
T30

120

120

80

80

40

40

30

60

90

120
0

Number of cracks

Load (kN)

160

Load (kN)

deflection-softening response. The previous study reported that an

addition of steel fibers up to 1% to a beam with stirrups exhibited
an insignificant increase in the flexural capacity because the steel
fibers did not substantially increase the tensile resistance after
the steel rebar yielding. However, in the case of UHPC, a much
improved flexural capacity was obtained by including steel fibers
due to the excellent fiber bridging capacity at the crack surfaces,
leading to strain-hardening response. Therefore, it can be noticed
that the use of strain-hardening UHPC is more effective in improving the flexural capacity of reinforced beams than the use of
deflection-softening SFRC.
The load carrying capacity increased with an increase in the
reinforcement ratio. By increasing the reinforcement ratio from
0.94% to 1.50%, approximately 3056% higher peak loads were
obtained. An increase in the load carrying capacity of UHPC beams
by increasing the reinforcement ratio was already reported in previous studies [10,33]. On the other hand, no noticeable differences
in the load carrying capacity and post-cracking stiffness were
observed in accordance with the length and type of fiber, regardless of the reinforcement ratio (the maximum difference of the
peak load was only 10%). This is dissimilar to the results of material
flexural tests using the beams with size of 100
100
400 mm (in
the previous section), in which the use of longer fibers (S19.5 and
S30) and twisted fibers (T30) exhibited approximately 64% (on
average) higher flexural strengths than that of the specimen S13.
This is caused by the fact that the contribution of steel fibers on
the post-cracking tensile performance was offset by the longitudinal steel rebars and the orientation and dispersion of the fibers
with longer lengths became poorer with the interruption of

0
10

Ave. crack spacing (mm)

(b)
Fig. 8. Number of cracks and average crack spacing of steel bar-reinforced UHPC
beams; (a) q = 0.94%, (b) q = 1.50%.

416

D.-Y. Yoo, Y.-S. Yoon / Engineering Structures 102 (2015) 409423

Fig. 9. Cracking behaviors of 100
100
400 mm sized material UHPC beams tested as per ASTM C 1609; (a) S13, (b) S30, (c) T30.

Shin et al. [39] also suggested the following equation for evaluating deflection ductility when the RC beams continued to sustain
the load well beyond the peak flexural load:

150
T2

T1

T3

Load (kN)

120

B
T1

T3

T2

90

C
B

Du
Dy

60
30
0
-3000

-1500

1500

3000

4500

6000

7500

9000

Strain ()
Fig. 10. Typical loadstrain curves of NF-1.50% and T30-1.50%.

stirrups and longitudinal steel rebars compared to those of the

fibers with shorter lengths. The second reason will be further discussed in a later section. Nevertheless, a comparison of the flexural
responses of the reinforced UHPC beams with different lengths and
types of fiber clearly shows that an increase in the fiber length and
the use of twisted fibers lead to improvements in the post-peak
response and ductility. For example, the specimen S30-1.50% was
able to sustain a deflection of 72.9 mm before failure (a sudden
load drop caused by the rupture of the steel rebar), which is
approximately 42% higher than the deflection sustained by the
specimen S13-1.50%.
3.2.2. Ductility
The ductility of RC beams can be quantitatively evaluated using
the ductility index, which is expressed in terms of deflection, curvature or rotational ductility. Among them, in this study, deflection
ductility was adopted to characterize the ductility of all test beams
because of its simplicity. The ductility index is obtained by dividing
the mid-span deflection at the peak load by the mid-span deflection at the steel rebar yielding as follows:

lp

lu

Dp
Dy

where Dp is the mid-span deflection at the peak and Dy is the

mid-span deflection at the steel rebar yielding.

P
(unit: mm)

100
30 mm
Notch
50

150
Fig. 11. Finite element modeling.

where Du is the mid-span deflection at the ultimate state.

Various definitions of deflection at the ultimate state have been
adopted by different researchers. Shin et al. [39] and Hadi and
Elbasha [40] defined the ultimate deflection as the deflection corresponding to 80% of the peak load along the descending branch
of the loaddeflection curve. On the other hand, Yang et al. [10]
and Bernardo and Lopes [41] defined the ultimate state as the point
at which the loaddeflection curve starts to decrease sharply. However, the former definition was not appropriate to some of the test
specimens in this study (NF-0.94%, NF-1.50%, S19.5-1.50%, and
T30-1.50%), because the load carrying capacity suddenly dropped
before reaching a deflection corresponding to 80% of the peak load
in the descending branch. In addition, the latter definition is subjective and ambiguous as Bernardo and Lopes [41] reported. Therefore, in this study, the ultimate state was defined as the point at
which the load carrying capacity suddenly dropped caused by
the steel rebar rupture or concrete crushing at the top (compressive) fiber.
The ductility indices obtained from Eqs. (3) and (4) are summarized in Table 5. In the case of the NF-series, the ductility index
decreased with an increase in the reinforcement ratio. It is very
interesting to notice that lower ductility indices were obtained
by including steel fibers. In addition, the magnitude of decrease
in ductility by including steel fibers increased with the application
of a lower reinforcement ratio. This is because the deformation of
longitudinal steel rebar was localized at the point where a specific
single crack was widened (at times, two cracks were widened) due
to the superb fiber bridging capacity and crack localization properties of UHPC, and thus, the steel rebar ruptured at a relatively smaller mid-span deflection compared to that for the beams without
fiber. Thus, their strain-hardening property was unfavorable to
ultimate deflection capacity of reinforced UHPC beams. Higher
ductility was observed when longer fibers were used because the
deflection at the ultimate state increased. Accordingly, it is clearly
indicated that the use of longer fibers is effective for improving the
post-peak response and ductility of reinforced UHPC beams. In the
case of the beams containing steel fibers, there was no distinct
trend in the ductility index according to the reinforcement ratio.
3.2.3. Crack and failure patterns
A comparison of typical crack patterns for the beams with and
without steel fibers (NF-1.50% and T30-1.50%) is shown in Fig. 7.
After the cracking moment, vertical flexural cracks were formed
perpendicular to the maximum principle stress direction. At a
low applied load (P = 40 kN), the cracks in the beams without fiber
(NF-series) propagated more deeply into the compression zone
than the cracks in the beams with steel fibers (S- and T-series)
did. This indicates that the neutral axis depth in the beams without

417

30
15

10

1.5

4.5

Deflection (mm)

Crack width (mm)

15

Exp.
Ave. Exp.
Pred.

0
0

Crack width (mm)

1.5

4.5

Deflection (mm)

1.5

4.5

45

15

Load (kN)

30

Tensile stress (MPa)

Load (kN)

Tensile stress (MPa)

45

Exp.
Ave. Exp.
Pred.

(b)

15

15

Crack width (mm)

(a)

10

30

45

15

Load (kN)

10

Exp.
Ave. Exp.
Pred.

Tensile stress (MPa)

45

15

Load (kN)

Tensile stress (MPa)

D.-Y. Yoo, Y.-S. Yoon / Engineering Structures 102 (2015) 409423

10
5

30
15

Exp.
Ave. Exp.
Pred.

0
0

Deflection (mm)

(c)

1.5

4.5

Deflection (mm)

Crack width (mm)

(d)

Fig. 12. Inverse analysis and three-point bending test results (Left: TSC and Right: experimental and numerical results); (a) S13, (b) S19.5, (c) S30, (d) T30.

15
K=1
K=1.25

10

0
0

Tensile stress (MPa)

Tensile stress (MPa)

15

K=1
K=1.25

10

0
0.02

0.04

0.06

0.02

0.04

Strain (mm/mm)

Strain (mm/mm)

(a)

(b)

0.06

15
K=1
K=1.25

10

Tensile stress (MPa)

15

Tensile stress (MPa)

fiber is further shifted up near the compression fiber. In contrast,

the beams with steel fibers exhibited flexural cracks with very
short depths because the propagation of cracks was effectively
restrained by fiber bridging at the crack surfaces.
Beyond an applied load of approximately 60 kN, the number of
cracks in the beams without fiber marginally increased, but the
crack widths continuously increased with the applied load.
However, the beams including steel fibers exhibited a continuous
increase in the number of cracks until near the peak load along
with an insignificant increase in the crack widths. In addition, most
of the cracks gradually propagated into the compression zone and
the cracks were not visually widened. This observation indicates
that the UHPC is able to redistribute the tensile stresses before
the fiber pullout.
The beams without fiber were failed by the concrete crushing at
the compression zone. This is caused by the fact that since the neutral axis depth was quite deeply moved to the compression fiber,
substantially high compressive stress was generated through force
equilibrium. In contrast, for the beams with steel fibers, the steel
fibers at one or two specific cracks began to pull out at near the
peak load, and then the width of this specific crack increased significantly compared to those of other cracks (this phenomenon is
well known as crack localization). Thus, the deformation of longitudinal steel rebar was localized at the point of crack localization,
and consequently, the steel rebar was ruptured once it reached the
ultimate state along with a sudden drop of the load. Therefore, all
test specimens including steel fibers exhibited flexural failure
mode with a rupture of steel rebars, as summarized in Table 5.
Fig. 8 shows the load as a function of the number of cracks and
average crack spacing. The number of cracks in the beams with
steel fibers increased with an increase in the load. The number of
cracks in the beams without fiber also increased with the load,
but the increase was much less than those in the beams with steel
fibers. The average crack spacings in the beams with steel fibers
rapidly decreased with an increase in the load at the low load stage
and became stable after reaching approximately 50% of the peak
load. On the other hand, the beams without fiber showed almost
identical average crack spacings with an increase in the load,
implying that new cracks occurred outward rather than between
the former cracks. The cracking response and crack distribution
were insignificantly influenced by the length and type of fiber for

K=1
K=1.25

10

0
0

0.02

0.04

Strain (mm/mm)

(c)

0.06

0.02

0.04

0.06

Strain (mm/mm)

(d)

Fig. 13. Tensile stressstrain models by AFGC/SETRA recommendation; (a) S13, (b)
S19.5, (c) S30, (d) T30.

the structural beams with steel rebars. This is inconsistent with

the findings for cracking response in the material beams without
steel rebar, as shown in Fig. 9. The number of cracks tangibly
increased with an increase in the fiber length, and the specimen
T30 produced more micro-cracks than the specimens with smooth
steel fibers (S-series) did. Possible explanations for the inconsistency are as follows: (1) as the tensile stress generated by the
external load was resisted by both the steel fibers and rebars, the
effect of steel fibers on cracking response was relatively decreased,
and (2) the fiber orientation and dispersion were poorer for the
reinforced beams with longer steel fibers owing to the interruption
of internal steel rebars.

418

D.-Y. Yoo, Y.-S. Yoon / Engineering Structures 102 (2015) 409423

Stress

Stress

fc'

fc'

Ec
lim

1%

Ec

0.3 e

1%

lim

f1%
fbt
ftj

c'

0.3

Strain

f1%
ftj
fbt

(a)

c'

Strain

(b)

Fig. 14. Schematic description of material models for UHPC; (a) strain-softening (S-series), (b) strain-hardening (T-series).

top
Fc

c
d
hi
hi+1

i
i+1

i
i+1
s

hs

h
Ft
Fs

bottom
Fig. 15. Schematic description of stress and strain distributions in cross-section.

3.2.4. Load versus strain response

The typical loadstrain curves of the specimens NF-1.50% and
T30-1.50% are shown in Fig. 10. The strains were measured using
the strain gages attached to the side surfaces of the beams (see
Fig. 4(c)). Negative strains indicate compressive strains, while positive strains indicate tensile strains. In addition, the dark solid lines
denote the strains in the specimen NF-1.50%, while the light solid
lines denote the strains in the specimen T30-1.50%.
Before the occurrence of first cracking, the positions T1, T2, and
T3 exhibited compressive strains because these were higher than
the centroid of the cross-section. Immediately after the first cracking, the strain at position T3 for NF-1.50% suddenly changed from
compression to tension, whereas the compressive strain at position
T3 for T30-1.50% was maintained far beyond the first cracking
point (up to nearly P = 116 kN). This is because the fibers at the
crack surfaces inhibited the propagation and widening of cracks,
and consequently, the sudden rise of the neutral axis with cracking
was controlled by including the steel fibers. As position C was
located on the neutral axis during the early stages (before the first
cracking), the strain values were close to zero for all test beams.
However, tensile strains at position C were obtained after cracking
due to the rise of the neutral axis. The tensile strains at positions C
and B significantly increased in the specimen NF-1.50% immediately after the occurrence of cracking. In contrast, the magnitude
of tensile strain increased after the initiation of cracks was considerably reduced by including the steel fibers because of the fiber
bridging effect, which leads to multiple micro-cracks. This observation is consistent with the findings from Yang et al. [10] indicating
that no abrupt change in bottom strains was obtained even after
cracking for reinforced UHPC beams.

4. Numerical analysis for predicting the flexural response of

reinforced UHPC beams
4.1. Material modeling
In order to predict the flexural behavior of UHPC beams reinforced with steel rebars, sectional analysis was performed. For this,
material models of UHPC under compression and tension were
suggested according to the length and type of fiber. As shown in
Fig. 5(a), all cylindrical specimens made of UHPC exhibited a linear
compressive stressstrain relationship up to the peak value and a
sudden load drop immediately after the peak value. Therefore,
the compressive behavior was simply modeled by a linear stress
strain curve up to the peak value. In addition, once the compressive
stress reached the peak value, the stress was modeled to drop to
zero. The UHPC without fiber was also modeled by a linear tensile
stressstrain curve considering compressive elastic modulus up to
the peak value, followed by zero stress. In contrast, the pre- and
post-cracking tensile behaviors were separately modeled for UHPC
with steel fibers. The pre-cracking tensile behavior was modeled
by a linear stressstrain relationship using the elastic modulus
measured from the compression test. On the other hand, for the
post-cracking tensile modeling, inverse analysis was performed
based on a previous study [42]. The cohesive stress and crack opening displacement relations were obtained from a crack growth
analysis on the basis of a fictitious crack model [43]. In addition,
the finite element modeling obtained by the program DIANA [44]
were incorporated into the analysis, as shown in Fig. 11. Only half
of the beam was modeled because of the specimens symmetry,
and to evaluate the crack propagation precisely, the center of the

D.-Y. Yoo, Y.-S. Yoon / Engineering Structures 102 (2015) 409423

where ee is the elastic strain, ftj is the tensile strength, and Ec is the
elastic modulus of concrete.
The strains at crack widths of 0.3 mm and 1% of the height of
the prism specimen are, respectively, expressed by the following
two equations:

Start
Determination of material properties

Assuming curvature at initial stage

e0:3

f tj
w0:3

lc
cbf Ec

Assuming neutral axis

e1%

f tj
w1%

lc
cbf Ec

Determining strain distribution by assuming

that plane section remains plane
Calculating stress distribution using stress-strain
models under compression and tension
Calculating sectional compressive and tensile forces

Checking force equilibrium condition

F=0 ?

NO

Reassume neutral axis (iteration)

Increase of curvature (k+k)

Compressive stress-strain model

Tensile stress-strain model

where w0.3 is the crack width of 0.3 mm, e0.3 is the strain at the
crack width of 0.3 mm, lc is the characteristic length (2/3
heights
of rectangular and T-beams), cbf is the partial safety factor (cbf = 1),
w1% is the crack width corresponding to 0.01H: H is the height of the
prism specimen, e1% is the strain at the crack width corresponding
to 0.01H.
The ultimate tensile strain is expressed by

elim

Calculating moment

f bt
Checking s < u ?
NO

End
Fig. 16. Algorithm for sectional analysis.

beam was modeled with dense meshing. The detailed procedure of

inverse analysis can be found in a previous study [21].
The calculated polylinear tension-softening curves (TSCs) and a
comparison of experimental and numerical results are presented in
Fig. 12. The average loaddeflection curves were obtained as follows. The deflection was first determined to increase in 0.01 mm
increments, and the load was calculated based on a linear interpolation from the test data. Subsequently, the loads at an identical
deflection point were averaged. The same averaging method was
adopted in a previous study [45]. As shown in Fig. 12, higher tensile
strengths and stress resisting capacities at large crack widths were
obtained using longer smooth steel fibers and twisted steel fibers.
In addition, the numerical analyses incorporating polylinear TSCs
exhibited good agreement with the experimental results for all test
series. The difference between experimental and numerical peak
loads was less than 0.8%.
Because the TSC is composed of the stress and crack width, it
needs to be converted to the stressstrain curve for sectional analysis. For this, the AFGC/SETRA recommendation [3] was adopted in
this study. Based on the previous studies by Yang et al. [11] and
Yoo [33], the analytical results based on the AFGC/SETRA recommendation with polylinear TSCs obtained from inverse analysis
exhibited good agreement with the test data of reinforced UHPC
beams with micro steel fibers (Lf/df = 13/0.2 mm/mm).
The tensile stressstrain curve can be obtained using
Eqs. (5)(10) based on the obtained polylinear TSCs. First, the
elastic tensile strain of UHPC is expressed as follows:

f tj
Ec

f w0:3
K cbf

f w1%
K cbf

10

f 1%

Determining moment-curvature curve

ee

Lf
4lc

where elim is the ultimate tensile strain.

The stresses at two different characteristic points (at w0.3 and
w1%) are expressed as follows:

YES

YES

419

where fbt is the stress at the crack width of 0.3 mm, K is the fiber
orientation coefficient, and f1% is the stress at the crack width corresponding to 0.01H. Since the fiber orientation in the structural
beams was difficult to be determined through image analysis [5]
and predicted because of the interruption of internal steel rebars,
two different fiber orientation coefficients K = 1, which is adopted
by several previous studies [11,13], and K = 1.25, which is suggested
by AFGC/SETRA recommendation for all loading other than local
effects, were assumed to calculate the tensile stress parameters.
The calculated tensile stressstrain curves are shown in Fig. 13.
The tensile models are strongly influenced by the length and type
of fiber: the tensile strength and ultimate strain (strain at zero tensile stress) substantially increased with an increase in the fiber
length. This is attributed to the improvement of pullout capacity
of fibers with an increase in the fiber length [36]. The AFGC/SETRA
recommendation [3] classifies the tensile responses using two
different lawsthe strain-softening law (ftj > fbt) and the strainhardening law (ftj < fbt)as shown in Fig. 14. The tensile stress at
a crack width of 0.3 mm in the specimen T30 was higher than
the tensile strength, classified by strain-hardening law, whereas
the tensile stress at a crack width of 0.3 mm in the specimens
S13, S19.5, and S30 was lower than the tensile strength, classified
by strain-softening law. Therefore, the specimen T30 showed a
tensile stressstrain model with a different shape compared with
those of the specimens with smooth steel fibers. In addition, the
specimen S30 exhibited higher tensile stress at the crack width
corresponding to 0.01H than the tensile strength, and thus, unusual shape of tensile stressstrain curve was obtained. Lower tensile stresses at given crack widths were obtained for all test
specimens by using K = 1.25, whereas there was no change in the
ultimate strain. A schematic description of the complete material
models for UHPC under compression and tension is illustrated in
Fig. 14, while a bilinear stressstrain curve for the steel rebar
was simply assumed.

D.-Y. Yoo, Y.-S. Yoon / Engineering Structures 102 (2015) 409423

100
50

30

60

90

Deflection (mm)

20

60

100
50

20

40

20

Deflection (mm)

(f)

80

(c)

100
50
0

60

(b)

20

20

Deflection (mm)

60

60

80

50

20

20

60

100
50
0

80

60

80

(e)

100
50
Exp.
Pred.

Exp.
Pred.

40

40

Deflection (mm)

150

20

Deflection (mm)

(g)

(d)

100

40

Deflection (mm)

Exp.
Pred.

40

50
Exp.
Pred.

150

Exp.
Pred.

60

40

150

Exp.
Pred.

50

100

Exp.
Pred.

Deflection (mm)

Load (kN)

Load (kN)

Load (kN)

40

150

150

100

Exp.
Pred.

Deflection (mm)

(a)

50

Load (kN)

100

Exp.
Pred.

Exp.
Pred.

Load (kN)

50

Load (kN)

100

150

150

150

Load (kN)

150

Load (kN)

Load (kN)

150

Load (kN)

420

40

60

80

(h)

20

40

60

80

Deflection (mm)

Deflection (mm)

(i)

(j)

Fig. 17. Comparison of experimental and numerical results (K = 1); (a) NF-0.94%, (b) S13-0.94%, (c) S19.5-0.94%, (d) S30-0.94%, (e) T30-0.94%, (f) NF-1.50%, (g) S13-1.50%, (h)
S19.5-1.50%, (i) S30-1.50%, (j) T30-1.50%.

4.2. Comparison of experimental and numerical results from sectional

analysis
Owing to its simplicity, sectional analysis has been adopted by
several researchers to analyze the flexural response of UHPC beams
and conventional SFRC beams [11,13,46]. For this, the cross-section
of the beam was first divided into numbers of layers along the
height. The compressive and tensile stresses at each layer were
then calculated based on the assumption of linear strain distribution (plane section remains plane) at a given curvature, as
illustrated in Fig. 15. The neutral axis depth was calculated from
the force equilibrium condition in the cross-section, and finally
the moment was calculated. The calculation was repeated until
the ultimate strain of the steel rebar was reached. The algorithm
of the sectional analysis is illustrated in Fig. 16, and the details of
the procedure can be found elsewhere [33].
The strains at the top and bottom faces (etop and ebottom,
respectively) are given by

etop /c

11

ebottom /h  c

12

where / is the curvature, c is the neutral axis depth, and h is the

beam height.
The strain at each layer was obtained as illustrated in Fig. 15.
After determining the strain distribution, the stress at each layer
in the cross-section was calculated using the predefined
stressstrain material models for UHPC. The strain in the steel
rebar was considered as the strain in the surrounding concrete at
the steel level. The stress in the steel rebar was calculated from
the predefined bilinear material model.
The force in the cross-section calculated based on the stress at
each layer should satisfy the equilibrium condition, which is
expressed as follows:

Z
Ac

Table 6
Comparison of experimental and sectional analysis results.
Name

NF
S13
S19.5
S30
T30

As

f s dAs 0

13

where fc is the stress at each layer of UHPC and fs is the stress in the
steel rebar.
In order to satisfy Eq. (13), iterative calculations were performed to obtain an appropriate neutral axis depth. The internal

Peak load (kN)

Exp.
(1)

Pred.
(2)a

Pred.
(3)b

0.94
1.50
0.94
1.50
0.94
1.50
0.94
1.50
0.94
1.50

62.6
97.9
87.3
124.1
93.3
125.2
95.9
124.6
96.6
133.9

54.3
82.6
88.9
114.5
113.9
139.5
118.7
144.8
119.7
145.6

54.3
82.6
80.8
106.6
101.2
127.0
105.3
131.5
106.4
132.5

Average
Standard deviation
a
b

Exp./pred.
(1)/(2)a

Exp./pred.
(1)/(3)b

1.15
1.19
0.98
1.08
0.82
0.90
0.81
0.86
0.81
0.92
0.95
0.144

1.15
1.19
1.08
1.16
0.92
0.99
0.91
0.95
0.91
1.01
1.03
0.110

Fiber orientation coefficient, K = 1.

Fiber orientation coefficient, K = 1.25.

resisting moment M at a given curvature was then calculated by

using the following equation:

Z
M

Z
f c ydAc

Ac

f s ydAs

14

As

where y is the distance from the neutral axis.

In order to convert the calculated momentcurvature curve
obtained from the sectional analysis to the loaddeflection curve,
the following Eqs. (15) and (16) were simply adopted. The relationship between the curvature and the deflection was obtained from
the linear elastic theory [47].

PL1
2

15

M
24
d

Ec I 3L2  4L21

16

Z
f c dAc

Reinforcement
ratio q (%)

where P is the applied load, L1 is the length between the support

and the loading point (=900 mm), I is the moment of inertia, L is
the span length (=2200 mm), and d is the mid-span deflection.

421

1.5

200
Exp. ( =0.94%)
Pred. ( =0.94%)
Exp. ( =1.50%)
Pred. ( =1.50%)

Numerical peak load (kN)

Peak load ratio (based on S13)

D.-Y. Yoo, Y.-S. Yoon / Engineering Structures 102 (2015) 409423

0.5

150

100

50

S13

S19.5

S30

K=1
K=1.25

50

T30

100

150

200

Experimental peak load (kN)

Specimen
Fig. 20. Comparison of peak loads obtained from experiments and sectional
analyses.

Fig. 18. Ratio of peak load based on S13 specimen.

Figs. 19 and 20 and summarized in Table 6. On the whole, the analyses using K = 1.25 exhibited quite good agreement with the experimental results and better prediction, compared to those without
consideration of fiber orientation coefficient (K = 1). In addition,
the use of K = 1.25 provided an average ratio of experimental and
numerical peak loads closer to the value of 1 and a lower standard
deviation than that of K = 1, as given in Table 6. The ratios were ranged from 0.91 to 1.19, lower than those in the previous study [11]
for prestressed UHPC beams, and thus, it was concluded that the
use of fiber orientation coefficient of K = 1.25 is appropriate to predict the flexural response of reinforced UHPC beams with stirrups.

The comparison of experimental and numerical results for all

test specimens (with the assumption of K = 1) is shown in Fig. 17
and summarized in Table 6. When no fiber orientation was considered (K = 1), the loaddeflection curves of the specimens without
fiber and with short steel fibers (NF and S13) were quite well predicted using the sectional analyses. Although a similar shape of
loaddeflection curve was obtained from the analyses for the specimens with longer steel fibers (S19.5, S30, and T30), the numerical
results showed stiffer loaddeflection response after the first
cracking and higher load carrying capacity than the experimental
results. This is probability due to the difference in fiber distribution
characteristics (i.e., fiber orientation, fiber dispersion, and number
of fibers per unit area) between the small prism specimens used for
inverse analysis and the large structural reinforced beams. To support this explanation, the ratios of experimental and numerical
peak loads based on that in the specimen S13 were investigated,
as shown in Fig. 18. Regardless of the reinforcement ratio, the peak
load ratios obtained from numerical analyses were much higher
than those obtained from experiments. This clearly indicates that
the orientation and dispersion of longer steel fibers were more
disturbed by the internal rebars than the specimen S13, leading
to overestimation of the numerical predictions.
The comparison of experimental and numerical results
considering fiber orientation coefficient of K = 1.25 is illustrated in

100
50

60

90

Deflection (mm)

20

40

60

60

Deflection (mm)

(f)

80

60

20

40

60

80

20

40

60

80

(b)

(c)

(d)

(e)

100
50

150

100
50

20

40

Deflection (mm)

(g)

150

100
50

60

20

40

60

80

Deflection (mm)

(h)

50
Exp.
Pred.

100

Exp.
Pred.

Exp.
Pred.

Exp.
Pred.

Deflection (mm)

0
40

40

Deflection (mm)

Exp.
Pred.

0
20

20

50

Exp.
Pred.

150

Exp.
Pred.

100

Deflection (mm)

Load (kN)

Load (kN)

Load (kN)

50

50

Exp.
Pred.

150

100

100

Deflection (mm)

(a)
150

150

Load (kN)

30

50

Load (kN)

100

Exp.
Pred.

Exp.
Pred.

150

Load (kN)

50

(1) The addition of steel fibers resulted in a slight increase in the

compressive strength and elastic modulus and a significant
improvement in the flexural performance including flexural

150

Load (kN)

100

This study investigated the flexural response of reinforced

UHPC beams with different steel fibers. A total of ten large
reinforced UHPC beams were fabricated and tested. In order to
predict the flexural response, sectional analyses incorporating the
suggested material models were performed. From the results, the
following conclusions are drawn.

Load (kN)

150

Load (kN)

Load (kN)

150

5. Conclusions

20

40

60

80

Deflection (mm)

(i)

20

40

60

80

Deflection (mm)

(j)

Fig. 19. Comparison of experimental and numerical results (K = 1.25); (a) NF-0.94%, (b) S13-0.94%, (c) S19.5-0.94%, (d) S30-0.94%, (e) T30-0.94%, (f) NF-1.50%, (g) S13-1.50%,
(h) S19.5-1.50%, (i) S30-1.50%, (j) T30-1.50%.

422

D.-Y. Yoo, Y.-S. Yoon / Engineering Structures 102 (2015) 409423

strength and deflection capacity. The use of longer smooth

steel fibers and twisted steel fibers were more effective in
improving the flexural performance at the material level
than the shorter smooth steel fibers, whereas the compressive performance was insignificantly affected by the length
and type of fiber.
(2) The first cracking load and corresponding deflection for reinforced UHPC beams were slightly reduced by including steel
fibers because of inhomogeneous fiber dispersion. On the
other hand, the post-cracking stiffness and load carrying
capacity were significantly improved with the addition of
steel fibers because of the superb fiber bridging capacity at
the crack surfaces, resulting in a strain-hardening response.
The increase in the length of smooth steel fibers and the use
of twisted steel fibers resulted in the improvements of postpeak response and ductility, whereas there was no noticeable difference in the load carrying capacity and postcracking stiffness according to the length and type of fiber.
(3) A lower ductility index was obtained for the beams including steel fibers than for those without fiber, owing to the
crack localization properties. The beams without fiber failed
by crushing of concrete at the compression zone, whereas
the beams with steel fibers failed by rupture of steel rebars.
The crack propagation was effectively inhibited by fiber
bridging at the crack surfaces, but the cracking response of
reinforced UHPC beams was insignificantly affected by the
length and type of fiber.
(4) To predict the flexural response of reinforced UHPC beams,
material models were first suggested based on experimental
tests and inverse analyses. Subsequently, sectional analyses
incorporating the suggested material models were performed based on AFGC/SETRA recommendation. The use of
fiber orientation coefficient (K = 1.25) showed quite good
agreement with the experimental results and better prediction, compared to that without consideration of fiber orientation coefficient (K = 1). The ratios of flexural capacities
obtained from experiments and numerical analyses ranged
from 0.91 to 1.19. Thus, it was concluded that the use of fiber
orientation coefficient of K = 1.25 is proper for predicting the
flexural response of reinforced UHPC beams with stirrups.

Acknowledgements
This research was supported by a grant from a Construction
Technology Research Project 13SCIPS02 (Development of impact/
blast resistant HPFRCC and evaluation technique thereof) funded
by the Ministry on Land, Infrastructure, and Transport.
References
[1] Kim SW, Park JJ, Kang ST, Ryo GS, Koh KT. Development of ultra high
performance cementitious composites (UHPCC) in Korea. In: Proceedings of
the fourth international IABMAS conference, Seoul, Korea; 2008. p. 110.
[2] FHWA. Material property characterization of ultra-high performance
concrete. US Department of Transportation, Federal Highway Administration;
2006.
[3] AFGC/SETRA. Ultra high performance fibre-reinforced concretes. Interim
recommendations. Bagneux, France: SETRA; 2002.
[4] JSCE. Recommendations for design and construction of ultra-high strength
fiber reinforced concrete structures (Draft). Tokyo, Japan: Japan Society of Civil
Engineers; 2004.
[5] Yoo DY, Kang ST, Yoon YS. Effect of fiber length and placement method on
flexural behavior, tension-softening curve, and fiber distribution
characteristics of UHPFRC. Constr Build Mater 2014;64:6781.
[6] Graybeal B, Tanesi J. Durability of an ultrahigh-performance concrete. J Mater
Civil Eng 2007;19(10):84854.
[7] Graybeal BA. Flexural behavior of an ultrahigh-performance concrete I-girder. J
Bridge Eng 2008;13(6):60210.

[8] Chen L, Graybeal BA. Modeling structural performance of second-generation

ultrahigh-performance concrete pi-girders. J Bridge Eng 2011;17(4):63443.
[9] Yuguang Y, Walraven J, Uiji JD. Study on bending behavior of an UHPC overlay
on a steel orthotropic deck. In: Proceedings of 2nd international symposium on
ultra high performance concrete; 2008. p. 63946.
[10] Yang IH, Joh C, Kim BS. Structural behavior of ultra high performance concrete
beams subjected to bending. Eng Struct 2010;32(11):347887.
[11] Yang IH, Joh C, Kim BS. Flexural strength of large-scale ultra high performance
concrete prestressed T-beams. Can J Civil Eng 2011;38(11):118595.
[12] Fujikake K, Senga T, Ueda N, Ohno T, Katagiri M. Study on impact response of
reactive powder concrete beam and its analytical model. J Adv Concr Technol
2006;4(1):99108.
[13] Yoo DY, Banthia N, Kim SW, Yoon YS. Response of ultra-high-performance
fiber-reinforced concrete beams with continuous steel reinforcement
subjected to low-velocity impact loading. Compos Struct 2015;126:23345.
[14] Astarlioglu S, Krauthammer T. Response of normal-strength and ultra-highperformance fiber-reinforced concrete columns to idealized blast loads. Eng
Struct 2014;61:112.
[15] Xia J, Mackie KR, Saleem MA, Mirmiran A. Shear failure analysis on ultra-high
performance concrete beams reinforced with high strength steel. Eng Struct
2011;33(12):3597609.
[16] Wu C, Oehlers DJ, Rebentrost M, Leach J, Whittaker AS. Blast testing of ultrahigh performance fibre and FRP-retrofitted concrete slabs. Eng Struct 2009;31
(9):20609.
[17] Wille K, Kim DJ, Naaman AE. Strain-hardening UHP-FRC with low fiber
contents. Mater Struct 2011;44(3):58398.
[18] Ryu GS, Kim SH, Ahn GH, Koh KT. Evaluation of the direct tensile behavioral
characteristics of UHPC using twisted steel fibers. Adv Mater Res
2013;602:96101.
[19] Yoo DY, Shin HO, Yang JM, Yoon YS. Material and bond properties of ultra high
performance fiber reinforced concrete with micro steel fibers. Compos Part BEng 2014;58:12233.
[20] Kim DJ, Park SH, Ryu GS, Koh KT. Comparative flexural behavior of hybrid ultra
high performance fiber reinforced concrete with different macro fibers. Constr
Build Mater 2011;25(11):414455.
[21] Yoo DY, Lee JH, Yoon YS. Effect of fiber content on mechanical and fracture
properties of ultra high performance fiber reinforced cementitious composites.
Compos Struct 2013;106:74253.
[22] Barnett SJ, Lataste JF, Parry T, Millard SG, Soutsos MN. Assessment of fibre
orientation in ultra high performance fibre reinforced concrete and its effect
on flexural strength. Mater Struct 2010;43(7):100923.
[23] Ma J, Orgass M, Dehn F, Schmidt D, Tue NV. Comparative investigations on
ultra-high performance concrete with and without coarse aggregates. In:
Proceedings international symposium on ultra high performance concrete
(UHPC), Kassel, Germany; 2004.
[24] Wille K, Naaman AE, Parra-Montesinos GJ. Ultra-high performance concrete
with compressive strength exceeding 150 MPa (22 ksi): a simpler way. ACI
Mater J 2011;108(1):4654.
[25] Aydn S, Baradan B. The effect of fiber properties on high performance alkaliactivated slag/silica fume mortars. Compos Part B-Eng 2013;45(1):639.
[26] Yoo DY, Park JJ, Kim SW, Yoon YS. Early age setting, shrinkage and tensile
characteristics of ultra high performance fiber reinforced concrete. Constr
Build Mater 2013;41:42738.
[27] Kim JS, Kim BS, Jeon SJ, Joh CB. Development of design guidelines of ultra high
performance concrete. J Korean Soc Civil Eng 2013;61(2):618.
[28] Yoo DY, Zi G, Kang ST, Yoon YS. Biaxial flexural behavior of ultra-highperformance fiber-reinforced concrete with different fiber lengths and
placement methods. Cem Concr Compos 2015. http://dx.doi.org/10.1016/j.
cemconcomp.2015.07.011 [in-press].
[29] Yoo DY, Shin HO, Lee JY, Yoon YS. Enhancing cracking resistance of ultra-highperformance concrete slabs using steel fibers. Mag Concr Res 2015;67
(10):48795.
[30] ASTM C 39/39M. Standard test method for compressive strength of cylindrical
concrete specimens. West Conshohocken, PA: ASTM International; 2014. p. 17.
[31] JCI-S-002-2003. Method of test for load-displacement curve of fiber reinforced
concrete by use of notched beam. Japan Concrete Institute Standard; 2003.
[32] Ferrara L, Ozyurt N, di Prisco M. High mechanical performance of fibre
reinforced cementitious composites: the role of casting-flow induced fibre
orientation. Mater Struct 2011;44(1):10928.
[33] Yoo DY. Performance enhancement of ultra-high-performance fiber-reinforced
concrete and model development for practical utilization. PhD thesis. Seoul,
Korea: Korea University; 2014. 586 p.
[34] ASTM C 469M-10. Standard test method for static modulus of elasticity and
Poissons ratio of concrete in compression. American Society of Testing and
Materials; 2010. p. 15.
[35] Yoo DY, Kim J, Zi G, Yoon YS. Effect of shrinkage-reducing admixture on biaxial
flexural behavior of ultra-high-performance fiber-reinforced concrete. Constr
Build Mater 2015;89:6775.
[36] Yoo DY, Kang ST, Lee JH, Yoon YS. Effect of shrinkage reducing admixture on
tensile and flexural behaviors of UHPFRC considering fiber distribution
characteristics. Cem Concr Res 2013;54:18090.
[37] Yoo DY, Park JJ, Kim SW, Yoon YS. Influence of reinforcing bar type on
autogenous shrinkage stress and bond behavior of ultra high performance
fiber reinforced concrete. Cem Concr Compos 2014;48:15061.
[38] Aoude H, Belghiti M, Cook WD, Mitchell D. Response of steel fiber-reinforced
concrete beams with and without stirrups. ACI Struct J 2012;109(3):35968.

D.-Y. Yoo, Y.-S. Yoon / Engineering Structures 102 (2015) 409423

[39] Shin SW, Ghosh SK, Moreno J. Flexural ductility of ultra-high-strength concrete
members. ACI Struct J 1989;86(4):394400.
[40] Hadi MN, Elbasha N. Effects of tensile reinforcement ratio and compressive
strength on the behaviour of over-reinforced helically confined HSC beams.
Constr Build Mater 2007;21(2):26976.
[41] Bernardo LFA, Lopes SMR. Neutral axis depth versus flexural ductility in highstrength concrete beams. J Struct Eng 2004;130(3):4529.
[42] Uchida Y, Kurihara N, Rokugo K, Koyanagi W. Determination of tension
softening diagrams of various kinds of concrete by means of numerical
analysis. In: Wittmann FH, editor. Fracture mechanics of concrete
structures. Freiburg, Germany: Aedificatio Publishers; 1995. p. 1730.

423

[43] Hillerborg A, Modeer M, Petersson PE. Analysis of crack formation and crack
growth in concrete by means of fracture mechanics and finite elements. Cem
Concr Res 1976;6(6):77382.
[44] DIANA Users Manual Release 9.2, Delft, The Netherlands; 2007.
[45] Zhao Z, Kwon SH, Shah SP. Effect of specimen size on fracture energy and
softening curve of concrete: Part I. Experiments and fracture energy. Cem
Concr Res 2008;38(8):104960.
[46] Kooiman AG. Modelling the post-cracking behavior of steel fibre reinforced
concrete for structural design purposes. Heron 2004;45(4):275307.
[47] Gere JM. Mechanics of materials. 6th ed. Belmont: Brooks/Cole; 2003.