Construction and Building Materials 172 (2018) 10–18

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

A comparison of existing analytical methods to predict the flexural

capacity of Ultra High Performance Concrete (UHPC) beams
Mohamadreza Shafieifar ⇑, Mahsa Farzad, Atorod Azizinamini
Department of Civil and Environmental Engineering, Florida International University, United States

h i g h l i g h t s

Small-scale UHPC beams were tested to evaluate the ultimate moment capacity.

The experimental results were used to validate a Finite Element (FE) model.

Using validated FE model, a series of large-scale UHPC beams were simulated.

The results of existing analytical methods and the FE model were compared.

a r t i c l e i n f o a b s t r a c t

Article history: Ultra High-Performance Concrete (UHPC) shows enhanced performance and ductility compared to con-
Received 5 February 2018 ventional concrete which can be beneficial in the construction industry. Many researchers have per-
Received in revised form 19 March 2018 formed experimental studies on the structural behavior of the UHPC beams to establish a reliable
Accepted 23 March 2018
analytical method to calculate the flexural capacity of the section. However most of these studies were
Available online 30 March 2018
performed through limited specimens due to the high cost of UHPC. The objective of this paper is to com-
pare the accuracy of well-known existing methods to calculate the moment capacity of a reinforced UHPC
Keywords:
beam through a parametric study. To that aim, several small-scale beams were constructed and tested to
UHPC
UHPFRC
evaluate the flexural behavior and ultimate moment capacity of the UHPC beams. The performance of the
Ultra High Performance Concrete tested specimens is discussed regarding the moment capacity, load-deflection curves, crack development
Concrete Damage Plasticity and the modes of failure.
CDP The obtained results through the experiments were, then, used to validate the Finite Element (FE)
Abaqus model. Comparing the numerical and experimental results indicates that generally, the proposed numer-
Moment capacity ical model can predict the structural behavior of the UHPC beams reasonably. Hence, the validated FE
model was employed as a reference point to evaluate the existing analytical approaches to calculate
moment capacity of UHPC beams. A series of large-scale beams with different geometries and reinforcing
details were numerically simulated, and the results were compared with the results obtained through the
analytical methods. The results showed that some of existing methods can predict the ultimate moment
capacity of the UHPC beam with an acceptable accuracy.
Ó 2018 Elsevier Ltd. All rights reserved.

1. Introduction French Association of Civil Engineering (AFGC) [1] specification

Ultra High-Performance Fiber Reinforced Concrete (UHPFRC) or
known as UHPC is a unique cementitious based material with fine
aggregates, silica fume, fibers, superplasticizer and low water/
cement ratio. UHPC exhibits an exceptional strength, ductility,
and durability that can be an alternative for new constructions.
This material has been developed over recent decades, and there
are different definitions for UHPC based on various classifications.
⇑ Corresponding author.
E-mail address: mshaf017@fiu.edu (M. Shafieifar).
defines UHPC as concrete with a compressive strength of more
than 21 ksi (150 MPa) and a maximum of 36 ksi (250 MPa).
Several bridges need rehabilitation and using UHPC can
increase their durability [2]. The US Federal Highway Administra-
tion (FHWA) has been considering the use of UHPC in bridges since
2001. Currently, several bridges (more than 180), in which UHPC
was used mostly to connect precast elements, are open to traffic
in North America [3,4]. Also using recycled materials in such
advanced concrete can improve the sustainability characteristic
of UHPC [5]. Considering exceptional properties of UHPC and
extending the use of this material in buildings and bridge industry

https://doi.org/10.1016/j.conbuildmat.2018.03.229
0950-0618/Ó 2018 Elsevier Ltd. All rights reserved.
 M. Shafieifar et al. / Construction and Building Materials 172 (2018) 10–18
Notation
a depth of a rectangular stress block
As area of steel rebar in tension
b width of the beam
c depth to the neutral axis
d depth from extreme compressive fiber to centroid of re-
bar steel
df diameter of the fibers
Efs modulus of elasticity of fibers
EUHPC modulus of elasticity of UHPC
[6–8], in-depth knowledge is required to understand the approach
which aids in calculating the moment capacity of UHPC sections.
The moment, shear and compression capacity of a normal
strength reinforced concrete is well understood based on which
the codes have developed procedures to obtain the moment capac-
ity. Although several experimental and numerical tests have been
performed on the flexural behavior of the reinforced UHPC beams,
and are documented as a technical reports, there is no general
design guideline for the material in the US standards.
The tensile capacity of UHPC is achieved through cement matrix
and steel fibers that stitch the cracks. The ductility and tensile
strength resulting from the fibers is a characteristic material
behavior of UHPC that cannot be ignored. These characteristics
change the behavior of the UHPC beams compared to normal
strength reinforced concrete ones. Accordingly, the procedures
developed in codes for normal strength concrete beams cannot
be straightforwardly used for UHPC sections without
modifications.
The previous researches on UHPC have dominantly been
focused on material properties, and less studies have been under-
taken on flexural behavior and moment capacity of a UHPC sec-
tions. Several studies are available to summarize the recent
UHPC material investigations [9,10]. Some experimental and ana-
lytical studies on the structural behavior of the UHPC sections
can be found in Refs. [11–16].
Due to the complicated character of developing a Finite Element
(FE) model, a simplified analytical procedure can facilitate the
design process. The developed analytical procedure should provide
basic assumptions to calculate the moment capacity of the UHPC
beams.
The objective of this research is to evaluate the existing meth-
ods to calculate the ultimate moment capacity and understand
the flexural behavior of a reinforced UHPC beam section. To do
so, several small-scale UHPC beams with different percentage of
longitudinal reinforcement and effective depth were tested. The
results of these experimental tests were used to validate the FE
models. The material properties of UHPC adopted in this model
were based on the material properties used in an earlier research
[17]. The FE model was able to predict the behavior of the tested
beams including load-deflection curves, ultimate capacity, and
mode of failure with a good agreement. This model was able to
predict the behavior of UHPC specimens with different geometries,
loading conditions, and reinforcing details with a reasonable accu-
racy, and was considered as a reference point. Then it was used as a
benchmark for a parametric study on large-scale beams to evaluate
the existing analytical method.
2. Experimental program
Twelve small-scale UHPC beams were fabricated and tested
under three-point loading. Construction of the specimens was
11
fc compression strength of UHPC
ft the tensile stress of UHPC
fy yield strength of steel rebar
h height of the beam
lf length of fibers
qs rebar percentage
rfy fiber yielding stress
sf fiber-concrete bond strength
b1 stress block parameter
undertaken in several stages. In the first stage, the forms were oiled
and the steel bars were placed in their positions in the forms. Ply-
wood blocks with holes drilled in them were used to support the
steel bars. As the casting direction may affect the fiber orientation,
all specimens were cast horizontally and trowel finished [18–20].
After casting, the specimens were covered with polythene sheets
for 72 h. and then de-molded. The specimens were kept moist for
one week after casting to control the rate of moisture loss and
hence prevent premature shrinkage cracking. Then they were
cured in an air-dry condition until the test. Prior to test, the beams
were painted in white to facilitate tracing of the cracks.
All specimens designated as Sh  b- qs  d/h (D) where: h and b
specify the height and the width of the beam in inch, respectively.
qs shows the percentage of rebar in tension and d/h presents the
ratio of effective depth to the height of the section (see Fig. 1). As
some specimens had the same geometry and rebar percentage
and were cast as alternate specimens, (D) at the end of the speci-
men’s names shows the duplicate specimen results. For example,
S6  6-1.7-0.85 shows the specimen with the width and height
of 6 in. (15.2 cm), 1.7% of rebar ratio and d/h = 0.85 (d = 5.1 in.).
Also, Specimen S6  6-0-0 and S6  6-0-0D show the specimens
without rebar reinforcement. Based on previous studies [21], the
development length in UHPC is much less than regular concrete
and there were no need for the mechanical anchorage.
The total length and load span of the beams were 20 in. (50.8
cm) and 18 in. (45.7 cm), respectively. To compare the results, all
tested specimens had similar material and width and longitudinal
reinforcement provided reinforcement ratios of 0–2.6%. The speci-
mens were designed to show flexure and shear behavior under
center-point bending test loading approach for tensile based on
AFGC [1].
2.1. Materials
The UHPC used in this study was composed of a blended premix
powder, water, superplasticizer, and 2% steel fibers by volume
which is the most common ratio suggested by commercial UHPC
companies in North America. The premix powder included cement,
silica fume, ground quartz, and sand. The fibers were 0.5 in. (13
mm) long with a diameter of 0.2 mm respectively, with a tensile
strength of 400 ksi (2800 MPa). Flow table test was performed
according to ASTM C1437 [22], to obtain the rheology of the UHPC.
Static and dynamic flowability of UHPC was measured 8 in. (20
cm), and 10 in. (25 cm), respectively.
The compressive and tensile strength of UHPC were obtained
through testing cylinder specimens (3  6 in. (7.5  15 cm)) and
dog-bone test, respectively. The mean compressive (fc) and tensile
(ft) strength of the tested beams were 21 ksi (145 MPa) and 1.4 ksi
(9.6 MPa) respectively.
All steel reinforcements were from one heat in manufacturing.
Tension tests performed on three representative specimens
12 M. Shafieifar et al. / Construction and Building Materials 172 (2018) 10–18
Fig. 1. Designation of the beam45.7 (Sh  b- qs  d/h) (units: in.).
Fig. 2. Load setup.
resulted in an average yield strength of 68 ksi (469 MPa) and ulti-
mate strength of 113 ksi (780 MPa).
2.2. Testing procedure and loading
The test was conducted at the age of 150 days after casting. The
load was applied constantly at the middle of the beam. Fig. 2 illus-
trates the loading setup. Load cells and pressure transducers were
used to measure the load at each level of loading. The deflection
was measured by the potentiometers installed at the mid-span of
the beams. The applied load, deflection, and crack tracing were
recorded after each load increment. To observe the post-peak
behavior of the specimens, the loading was continued up to either
the failure of the beam or 1-inch (2.54 cm) deflection.
3. Analysis and discussion of experimental results
The behavior of the specimens including crack patterns, mode
of failure, load-deflection relation, and ductility are discussed in
the following sections.
3.1. Cracking and modes of failure
The failure patterns of the tested beams are shown in Fig. 3.
Varying reinforcement ratio and effective height of the section
changed the ultimate load capacity and initial stiffness of the
beams. No significant cracks were observed in the early stages of
loading. Different mode of failure including flexure, shear-flexure,
and shear failure were observed for the specimens. The flexure
cracks propagated at the middle of the beams and were followed
by shear cracks near the supports in the shear zone. In general, fail-
ure of the UHPC beams started when the fiber began to pull out of
the concrete matrix.
Specimen S6  6-0.6-0.85 having 0.85% rebar percentage was
failed by fracture of the rebars before reaching high ductility.
This can be attributed to the short length of the plastic hinge
due to the high bonding of UHPC and the stress concentration
in the rebar.
3.2. Load – displacement curve of the tested beams
The mid-span deflection curves throughout the loading of the
tested beams are presented in Fig. 4. Generally, using fibers min-
imizes the cracking and increases the ductility of the beams. The
D
ductility ratio was defined as (Duy ) where Dy was the ideal yield-
ing displacement and Du was the displacement associated to
0.85 of the ultimate capacity after the peak load. For example,
the ductility ratio of specimen S6  6-2.6-0.85 and S6  6-1.7-
0.85 were more than 6 and 5, respectively. The average ductility
of specimens having no rebar (S6  6-0-0) was 3.7. Table 3
reports the corresponding load to the ultimate flexural capacity
of the tested beams.
4. Analytical equations
There are some analytical solutions to calculate the flexural
capacity of UHPC beams [1,11,15,23,24]. These analysis use inter-
nal stresses based on the stress-strain distribution to estimate
the moment capacity.
An approach to calculate the moment capacity of a UHPC sec-
tion can be the same as normal strength concrete considering its
compressive strength and ignoring its tensile strength. In this
 M. Shafieifar et al. / Construction and Building Materials 172 (2018) 10–18 13

S6×6-2.6-0.85 S6×6-2.6-0.85D

S6×6-1.7-0.85 S6×6-1.7-0.85D

S6×6-1.7-0.65 S6×6-1.7-0.65D

S4×6-3.9-0.75 S4×6-3.9-0.75D

S2×6-1.8-0.50 S6×6-0.6-0.85

S6×6-0-0 S6×6-0-0D
Fig. 3. Failure crack patterns of the specimens after the end of the test.

approach, UHPC stresses in compression can be represented by an from equations related to moment capacity of normal concrete
equivalent rectangular stress block which is proposed for high beams suggested by ACI 318 [25]. This code provides minimum
strength concrete. This approach may cause conservative result requirements for the material, analysis, design, and detailing of
for designing flexural elements. This procedure can be adopted normal concrete.
14 M. Shafieifar et al. / Construction and Building Materials 172 (2018) 10–18

120 80 350
S6x6-2.6-0.85 500 S4x6-3.9-0.75
S6x6-2.6-0.85D 70 S4x6-3.9-0.75D
100 S6x6-1.7-0.85 S6x6-0.6-0.85 300
S6x6-1.7-0.85D 400 S2x6-1.8-0.50
S6x6-1.7-0.65 60 S6x6-0-0
80 250
S6x6-1.7-0.65D S6x6-0-0D

Load (kN)
50
Load (kips)

Load (kips)
300

Load (kN)
200
60 40
150
200 30
40
20 100
20 100
10 50

0 0 0 0
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
Deflection/Span Length Deflection/Span Length

Fig. 4. Normalized deflection (D/L) of the tested beams (L = 18 in. (45.7 cm)).

a) Strain distribution b) ACI 318[25] c) ACI 544 [26] d) FHWA [11]

Fig. 5. Stress distribution assumption of previously suggested methods.

a section. The coefficient b1 is equal to 0.65 for concrete strengths

M n ¼ As f y d  ð1Þ
2 more than 8 ksi (55 MPa). Based on the literature, sf can be taken
as 0.6 ksi (4.15 MPa) [16,27,28] for fiber-reinforcement concrete.
Another approach can be adopted from ACI 544 [26] which is an Another method of calculating the flexural capacity of rectangu-
available standard of fiber reinforced concrete. According to this lar UHPC beams is based on one of the Federal Highway Adminis-
code, the flexural capacity of a Steel Fiber Reinforced Concrete tration (FHWA) reports [11]. In this model, using the equilibrium
(SFRC) section with a rectangular h  b cross-section is calculated equation and strain compatibility, the moment capacity of a UHPC
by the following equation: beam is estimated when either the extreme compression or ten-

 
a hþea sion strain reaches its limiting value (see Fig. 5). In this method,
M n ¼ As f y d  þ f t bðh  eÞ ð2Þ when the tension limit state controls, the following equation can
2 2
be used for estimating the moment capacity of the section.

As f y þ f t h
 
c¼ ð3Þ c 3h  c
ft
ef þ0:003
þ 0:85b1 f c M n ¼ As f y d  þ f t bðh  cÞ ð8Þ
0:003 3 6
!
ef þ 0:003 qsfy þf t
e¼ c ð4Þ c¼   ð9Þ
0:003 f t þ 0:0035EUHPC c
hc

a ¼ b1 c ð5Þ
Note that estimating the neutral axis depth can be done by solv-
  ing a quadratic equation or an iterative process. To estimate the
lf
rfs ¼ 2sf 6 rfy ð6Þ modulus of elasticity of UHPC some other equations are provided
df by previous studies [1,9,29].
rfs qffiffiffiffi qffiffiffiffi
ef ¼ ð7Þ
EUHPC ¼ 50000 f c ðpsiÞ ¼ 4200 f c ðMPaÞ ðSritharanÞ ð10Þ
Efs
In this method, the tensile stress of concrete assumed to be uni- qffiffiffiffi qffiffiffiffi
formly distributed over an area with a height of (h–e). The distance EUHPC ¼ 46200 f c ðpsiÞ ¼ 3840 f c ðMPaÞ ðGraybealÞ ð11Þ
from the extreme compression line to the top of the uniform tensile
qffiffiffiffi qffiffiffiffi
block (e) is calculated by the equations (Eqs. (2)–(7)). The depth of
EUHPC ¼ 260000 f c ðpsiÞ ¼ 9500 f c ðMPaÞ ðAFGC2013Þ ð12Þ
3 3

the neutral axis (c) is calculated by equilibrium equation of the

 M. Shafieifar et al. / Construction and Building Materials 172 (2018) 10–18 15

Table 1
The employed material parameters of UHPC for the FE computation which set according to the results of the experimental results [17].

Elastic parameter
Young’s modulus 7500 ksi (54710 Mpa) Poisson’s ratio 0.18
Concrete Damage Plasticity (CDP) parameters
Plasticity Compressive behavior Tensile behavior
Dilation angle 56 Yield stress Inelastic strain Yield stress Cracking strain
Eccentricity 0.1 18 ksi (124 MPa) 0 0.1 ksi (0.7 MPa) 0
fb0/fc0 1.1 20 ksi (138 MPa) 0.001 1.4 ksi (9.7 MPa) 0.0035
k 0.66 11 ksi (76 MPa) 0.009 0.3 ksi (2.1 MPa) 0.05
Viscosity Parameter 0

Mesh Type: C3D20R (A 20-node Quadratic brick, reduced integration).

Table 2 truss element (T3D2) and embedded in the UHPC with a perfect
The employed material parameters of steel reinforcement bars for bond. The supports and loading plate were idealized as square
the FE computation. prisms which were modeled with an 8 node brick element
Elastic parameter (C3D8) taking material properties of steel [30,31].
Young’s Modulus 29000 ksi (200000 MPa) To observe the crack propagation, the damage parameter in ten-
Poisson’s Ratio 0.30 sion can be also assumed to activate when the peak tensile
Plastic parameters strength is achieved. This parameter can be calculated through
Yield stress Plastic strain the Eq. (12), recommended by Mahmud et al. [32].
68 ksi (470 MPa) 0
r
68 ksi (470 MPa) 0.0127 dt ¼ 1  ð13Þ
113 ksi (780 MPa) 0.0877 f t
113 ksi (780 MPa) 0.1177

Mesh Type: T3D2 (3D truss elements). 6. Numerical and analytical results of the tested specimens

6.1. Numerical results

5. Finite element modeling
One option to predict the behavior of UHPC beams is FE models.
The FE software, Abaqus, has several models developed for con-
crete. For the present study, Concrete Damage Plasticity (CDP)
was used for modeling the beams. The validation of UHPC material
properties used in FE model for pure compression and tension, was
done in previous study [17]. UHPC behavior in tension and com-
pression was assumed as multi-linear stress-strain and the effect
of the fibers was assumed to be uniformly distributed. Table 1
shows all the parameters for CDP modeling used in the research.
Also, a multi-linear stress-strain curve was used for modeling the
steel reinforcing bars as shown in Table 2.
A three-dimensional FE simulation was used to model the fail-
ure process of the experimentally tested beams. Three steel loading
plates were used as supports. The mesh size of the tested beams
was considered around 0.5 in. (12 mm) in height and length, and
1 in. (25 mm) in width of all the specimens. The beams were ana-
lyzed using 20 node brick elements (C3D20R) to model the UHPC.
The rebar reinforcements were modeled using a 2-node linear 3-D
The load-displacement curves of both numerical and experi-
mental tests are plotted in Fig. 7. The model was able to predict
the initial stiffness, peak load and corresponding displacement of
the tested beams as well as mode of failure with an acceptable
accuracy. Fig. 6 shows the stress distribution of the specimen
(S6  6-2.6-0.85) at the peak load as an example.
6.2. Comparison between the models
Experimental results have been employed to validate the
results of the analytical and finite element methods. Maximum
loads measured from the experiments, FE results, and existing ana-
lytical methods are compared as illustrated in Table 3. In these
methods the length and the diameter of the fibers were 0.5 in.
(13 mm) and 0.2 mm, respectively. The nominal fiber’s yield
strength (rfy) was assumed 400 ksi (2800 MPa). Therefore, based
on ACI544 [26] analytical model (Eq. (6)), rfs was calculated 78
ksi (538 MPa). The tension strength (ft), compression strength
(fc), modulus of elasticity of the UHPC and steel properties were

Fig. 6. (a) Stress distribution (b) exaggerated deformed model (c) tensile damage representing cracks from FE model for specimen S6  6-2.6-0.85.
16 M. Shafieifar et al. / Construction and Building Materials 172 (2018) 10–18

120 80
S6x6-2.6-0.85 500 S4x6-3.9-0.75
S2x6-1.8-0.50 300
S6x6-1.7-0.85 70
100 S6x6-0.6-0.85
S6x6-1.7-0.65 400 250
60 S6x6-0-0
80
50

Load (kN)
200

Load (kips)
Load (kips)

Load(kN)
300
60 40
150
200 30
40 100
20
20 100 50
10

0 0 0 0
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
Deflection/Span Length Deflection/Span Length

Fig. 7. Load – Normalized deflection (D/L) curve of the tested beams (doted) and FE model predictions (Solid).

Table 3
Results of the maximum load capacity of the tested beams compared with different approaches.

Specimen S6  6-2.6-0.85 S6  6-1.7-0.85 S6  6-1.7-0.65 S4  6-3.9-0.75 S2  6-1.8-0.50 S6  6-0.6-0.85 S6  6-0-0 kips Mean Absolute
kips kips kips kips kips kips (kN) Percentage Error
(kN) (kN) (kN) (kN) (kN) (kN) (MAPE)
Experimental 109.2 84.45 53.57 65.17 6.54 45.7 23.76 –
(486) (376) (238) (290) (29) (203) (106)
FE 101.20 82.75 61.07 57.31 5.13 49.70 27.73 11.8%
(450) (368) (272) (255) (23) (221) (126)
FHWA [11] 87.74 67.03 55.94 46.44 5.42 41.37 27.19 16.3%
(390) (298) (249) (2 07) (24) (184) (121)
ACI318 [25] 67.72 45.28 34.18 38.02 3.03 16.41 – 46.7%
(301) (201) (152) (169) (13) (73)
ACI544 [26] 89.78 70.59 59.49 45.72 5.80 45.67 31.40 16.9%
(399) (314) (265) (203) (26) (203) (140)

assumed the same as the numerical model mentioned in Tables 1

and 2.
To measure the accuracy of each method, Mean Absolute
Percentage Error (MAPE) with the following equation was used.

n
100 X
Ei  Pi
MAPE ¼ ð14Þ
n i¼1 Ei

where Ei is the experimental result and Pi is the predicted value.

Comparing the numerical and experimental results show that
FE models reasonably well predicted the responses of the ultimate
moment capacity of UHPC beams for which the MAPE was 12.7%.
The mean percentage error of initial stiffness of numerical model
was reported 9.5% as well. As shown in Table 3, FHWA [11] and
ACI544 [26] could calculate the ultimate moment capacity with a
reasonable accuracy, while the ACI 318 [25] gives way less value
than the experimental results. This is attributed to ignorance of
the tensile capacity of the concrete in this approach. It should be
noted that although some of the tested beams failed in shear/
shear-flexural manner, still the experimental results may show
higher capacity than the one calculated through the existing equa-
tions for flexural capacity. This might be attributed to the conser-
vative nature of analytical methods. Moreover, based on the
results, the outcome of the suggested CDP finite element model
is fairly reliable to calculate the ultimate moment capacity of the
reinforced UHPC beams as it was for UHPC samples behavior in
pure compression and tension [17]. Therefore, this numerical
model can be used as a benchmark for existing analytical models,
making it possible to do more parametric study on UHPC beams.
Using this, a parametric study have been done on UHPC beams to
consider the effect of different variables on the accuracy of the
Fig. 8. Crack formation of the simulated beams (a) S24  12-3-0.9, (b) S12  6-3-
0.9 and (c) S8  4-3-0.9.
existing analytical method. These variables include height, longitu-
dinal reinforcement ratio, and cover to total height ratio. The
details and results of this study is discussed in the following
section.
7. Parametric study
Considering the cost of the UHPC material, constructing and
testing several large scale beams is not economically possible;
therefore a series of large-scale beams with different height, rebar
percentage and cover to total height ratio was numerically mod-
eled and compared with the analytical results obtained through
existing analytical methods. The bar ratio and the height of the
beam can considerably affect the beam capacity. The cover can
affect the cracking pattern and consequently the capacity of the
beam as well as the effective height of the beams. The designation
of the specimens were the same as illustrated in Fig. 1. The length
 M. Shafieifar et al. / Construction and Building Materials 172 (2018) 10–18 17

Table 4
Load capacity of the beams with different approaches (h = 24 in. (61 cm), b = 12 in. (30.5 cm) and L = 288 in. (732 cm)).

Specimen S24  12-1-0.9 S24  12-3-0.9 S24  12-5-0.9 S24  12-7-0.9 S24  12-1-0.95 S24  12-3-0.95 S24  12-5-0.95 S24  12-7-0.95 Mean Absolute
kips kips kips kips kips kips kips kips Percentage Error
(kN) (kN) (kN) (kN) (kN) (kN) (kN) (kN) (MAPE)
FE 118.2 221.1 309.5 382.3 122.0 231.6 327.2 408.4 –
(526) (983) (1377) (1700) (543) (1030) (1455) (1817)
FHWA [11] 101.8 200.8 297.5 392.4 105 210.6 313.9 415.2 14.6%
(453) (893) (1323) (1745) (467) (937) (1396) (1847)
ACI318 [25] 57.45 164.5 261.1 347.3 60.7 174.3 277.4 370.1 50.9%
(256) (732) (1161) (1545) (270) (775) (1234) (1646)
ACI544 [26] 109.9 203.4 286.2 358.3 113.2 213.2 302.5 381.2 19.1%
(489) (905) (1273) (1594) (504) (948) (1346) (1696)

Table 5
Load capacity of the beams with different approaches (h = 12 in. (30.5 cm), b = 6 in. (15 cm) and L = 144 in. (366 cm)).

Specimen S12  6-1-0.9 S12  6-3-0.9 S12  6-5-0.9 S12  6-7-0.9 S12  6-1-0.8 S12  6-3-0.8 S12  6-5-0.8 S12  6-7-0.8 Mean Absolute
kips kips kips kips kips kips kips kips Percentage Error
(kN) (kN) (kN) (kN) (kN) (kN) (kN) (kN) (MAPE)
FE 26.8 53.7 73.5 88.2 24.2 46.6 62.2 74.9 –
(119) (239) (327) (392) (108) (207) (277) (333)
FHWA [11] 26.2 50.9 75.0 98.7 24.6 46.0 66.8 87.2 6.0%
(117) (226) (334) (439) (109) (205) (297) (388)
ACI318 [25] 14.4 41.1 65.3 86.8 12.7 36.2 57.1 75.4 20.1%
(64) (183) (290) (386) (56) (161) (254) (335)
ACI544 [26] 28.3 51.4 71.9 89.8 26.6 46.5 63.8 78.3 3.9%
(126) (229) (320) (399) (118) (207) (284) (348)

Table 6
Load Capacity of the beams with different approaches (h = 8 in. (20 cm), b = 4 in. (10 cm) and L = 96 in. (244 cm)).

Specimen S8  4-1-0.9 S8  4-3-0.9 S8  4-5-0.9 S8  4-7-0.9 S8  4-1-0.8 S8  4-3-0.8 S8  4-5-0.8 S8  4-7-0.8 Mean Absolute
kips kips kips kips kips kips kips kips Percentage Error
(kN) (kN) (kN) (kN) (kN) (kN) (kN) (kN) (MAPE)
FE 12.2 22.7 29.3 35.4 11.0 19.2 25.2 29.2 –
(54) (101) (130) (157) (49) (85) (112) (130)
FHWA [11] 11.6 22.6 33.3 43.9 10.9 20.4 29.7 38.8 12.5%
(52) (101) (148) (195) (48) (91) (132) (173)
ACI318 [25] 6.4 18.3 29.0 38.6 5.7 16.1 25.4 33.5 19.6%
(28) (81) (129) (172) (25) (72) (113) (149)
ACI544 [26] 12.6 22.9 32.0 39.9 11.8 20.7 28.3 34.8 9.1%
(56) (102) (142) (177) (52) (92) (126) (155)

60
Mean Absolute Percentage Error (%)

FHWA ACI544 ACI318 calculated with a four-point loading test, to be in consistent with
50 the calibrated model, three point loading was used for the para-
metric study. As an example, Fig. 8 shows the crack formation of
40 beams S24  12-3-0.9, S12  6-3-0.9 and S8  4-3-0.9. This figure
shows the mesh sizes, crack formation, deformed shape of the
30 specimens and supports conditions.
Tables 4–6 show the results of the FE analysis compared to the
20 analytical results obtained through the existing methods. In the
first table, the ultimate moment capacity of the modeled beams
10 with 24 in. (61 cm) height is tabulated while the second and third
one show the similar results regarding the beams with 12 in. (30.5
0 cm) and 8 in. (20 cm) height respectively.
h= 24 in. h=12 in h=8 in.
Fig. 9 shows the summary of the results compared to FE model.
(61 cm) (30.5 cm) (20 cm)
The results indicate that both ACI544 [26] and FHWA [11] method
Fig. 9. Mean absolute percentage error of load capacity of the beams with existing can be used to calculate the ultimate moment capacity of the UHPC
method compared with calibrated FE results. beams with a reasonable accuracy. Further large-scale experiments
may be required to verify the accuracy of the listed methods. It is
also worth to mention that designing UHPC beams based on
of the spans of the specimens were considered 12 h and the width ACI318 [25], which is suggested for normal strength concrete,
of the beams (b) were taken 0.5 h, where h is the height of the leads to conservative design results and underestimation of the
beam. Although the moment capacity of the sections are usually moment capacity of UHPC elements, especially for large beams.
18 M. Shafieifar et al. / Construction and Building Materials 172 (2018) 10–18
8. Summary and conclusions
This work aimed at calculating the ultimate moment capacity
of Ultra High Performance Concrete (UHPC) beams. Several
small-scale beams were tested experimentally and the results
were used to validate the numerical model. The UHPC material
properties for Concrete Damage Plasticity (CDP) was previously
used to predict the behavior of the material in pure tension
and compression. Comparison of numerical and experimental
results of the test demonstrated that the numerical model can
fairly predict the behavior of the UHPC beams including load
capacity, initial stiffness, deflection at the peak, and post-peak
behavior. The results show that the calibrated Finite Element
(FE) model with a maximum error of 12% can be used to predict
the behavior of the UHPC reinforced beams as well as beams
without reinforcement, when the actual material properties are
not available.
Considering the accuracy of the FE model to predict the
behavior of the tested beams, this model was used as a reference
point to evaluate the existing analytical models. Comparing the
results of these methods showed that ACI544 [26] and FHWA
[11] methods can predict the ultimate moment capacity of UHPC
beams with a maximum error of 12%. Furthermore, using ACI318
[25], suggested for normal strength concrete, leads to overdesign
results.
Conflicts of interest
The authors state that there is no conflict of interest.
Acknowledgments
The research study, results of which reported in this paper was
partially sponsored by Accelerated Bridge Construction University
Transportation Center (ABC-UTC) at Florida International Univer-
sity. ABC-UTC is a Tier 1 UTC funded by U.S. DOT. Authors would
like to acknowledge and thank the sponsors for their support.
Opinions and conclusions expressed in this paper are of the
authors and do not necessarily represent those of sponsors. The
authors are also thankful to DuctalÒ by LafargeHolcim, for provid-
ing the UHPC material.
References
[1] AFGC/SETRA Groupe de travail BFUP, Betons fibres a ultra-hautes
performances Ultra High Performance Fibre-Reinforced Concretes
Recommendations, 2013 359.
[2] Rehmat, Sheharyar e, Amir Sadeghnejad, Saiada Fuadi Fancy, Alireza Valikhani,
Brian Chunn, Alireza Mohammadi, Ramin Taghinezhadbilondy, et al. ‘‘Non-
Destructive Testing (NDT) of a Segmental Concrete Bridge Scheduled for
Demolition, with a Focus on Condition Assessment and Corrosion Detection of
Internal Tendons”.
[3] F. Toutlemonde, J. Resplendino, Designing and Building with UHPFRC 2013,
10.1002/9781118557839.
[4] Federal Highway Administration Research and Technology, https://www.
fhwa.dot.gov/research/resources/uhpc/bridges.cfm.
[5] Y. Park, A. Abolmaali, Y.H. Kim, M. Ghahremannejad, Compressive strength of
fly ash-based geopolymer concrete with crumb rubber partially replacing
sand, Constr. Build. Mater. 118 (2016) 43–51, https://doi.org/10.1016/J.
CONBUILDMAT.2016.05.001.
[6] Mohamadreza Shafieifar, Mahsa Farzad, Atorod Azizinamini, Alternative ABC
connection utilizing UHPC. No. 17-03398, 2017 Transportation Research Board
Annual Meeting Proceedings, Washington, D.C., 2016.
[7] M. Farzad, M. Shafieifar, A. Azizinamini, Accelerated Retrofitting of Bridge
Elements Subjected to Predominantly Axial Load Using UHPC Shell, (2018).
https://trid.trb.org/view/1496558 (accessed March 8, 2018), Transportation
Research Board Annual Meeting Proceedings, Washington, D.C.
[8] M. Farzad, A. Mohammadi, M. Shafieifar, H. Pham, A. Azizinamini,
Development of Innovative Bridge Systems Utilizing Steel-Concrete-Steel
Sandwich System, 2017. https://trid.trb.org/view.aspx?id=1437942 (accessed
January 15, 2018), Transportation Research Board Annual Meeting
Proceedings, Washington, D.C.
[9] G. Benjamin A., Material Property Characterization of Ultra-High Performance
Concrete, 2006 186.
[10] M. Baqersad, E. Sayyafi, H.B.-C.E. Journal, undefined 2017, State of the Art:
Mechanical Properties of Ultra-High Performance Concrete, Civilejournal.com.
(n.d.). https://civilejournal.com/index.php/cej/article/view/306 (accessed
16.01.18).
[11] S. Aaleti, B. Petersen, S. Sritharan, Design Guide for Precast UHPC Waffle Deck
Panel System, including Connections, FHWA-HIF-1, 2013, 127. http://trid.trb.
org/view.aspx?id=1256734.
[12] M.M. Kamal, M.A. Safan, Z.A. Etman, R.A. Salama, Behavior and strength of
beams cast with ultra high strength concrete containing different types of
fibers, HBRC J. 10 (2014) 55–63, https://doi.org/10.1016/j.hbrcj.2013.09.008.
[13] M. Singh, A.H. Sheikh, M.S. Mohamed Ali, P. Visintin, M.C. Griffith,
Experimental and numerical study of the flexural behaviour of ultra-high
performance fibre reinforced concrete beams, Constr. Build. Mater. 138 (2017)
12–25, https://doi.org/10.1016/j.conbuildmat.2017.02.002.
[14] D.Y. Yoo, N. Banthia, Y.S. Yoon, Flexural behavior of ultra-high-performance
fiber-reinforced concrete beams reinforced with GFRP and steel rebars, Eng.
Struct. 111 (2016) 246–262, https://doi.org/10.1016/j.engstruct.2015.12.003.
[15] B. Bae, H. Choi, C. Choi, Flexural strength evaluation of reinforced concrete
members with ultra high performance, Concrete 2016 (2016).
[16] A.N. Dancygier, Z. Savir, Flexural behavior of HSFRC with low reinforcement
ratios, Eng. Struct. 28 (2006) 1503–1512, https://doi.org/10.1016/j.
engstruct.2006.02.005.
[17] Mohamadreza Shafieifar, Mahsa Farzad, Atorod Azizinamini, Experimental and
numerical study on mechanical properties of ultra high performance concrete
(UHPC), Constr. Build. Mater. 156 (2017) 402–411.
[18] W. Pansuk H. Sato Y. Sato R. Shionaga, 2008, Tensile behaviors and fiber
orientation of UHPC, in Proceedings of second international symposium on
ultra high performance concrete, Kassel, Germany (pp. 161–168)., (n.d.).
[19] S.W. Kim, S.T. Kang, J.J. Park, G.S. Ryu, 2008. Effect of filling method on fibre
orientation and dispersion and mechanical properties of UHPC. In Proceedings
of second international symposium on ultra high performance concrete,
Kassel.
[20] Ashkan Bozorgzad, Consistent distribution of air voids and asphalt and random
orientation of aggregates by flipping specimens during gyratory compaction
process, Constr. Build. Mater. 132 (2017) 376–382.
[21] J. Yuan, B.A. Graybeal, Structural behavior of ultra-high performance concrete
prestressed i-girders, Eng. Struct. 1 (2011) 228–235, https://doi.org/10.1159/
000331754.
[22] ASTM, C. ‘‘1437-01, 1437-01. Standard Test Method for Flow of Hydraulic
Cement Mortar, ASTM International, Pennsylvania, 2001.
[23] B.A. Graybeal, Compression Testing of Ultra-High-Performance Concrete, 4,
2017, 102–112. doi: 10.1520/ACEM20140027.
[24] D.Y. Yoo, N. Banthia, Y.S. Yoon, Predicting the flexural behavior of ultra-high-
performance fiber-reinforced concrete, Cem. Concr. Compos. 74 (2016) 71–87,
https://doi.org/10.1016/j.cemconcomp.2016.09.005.
[25] American Concrete Institute Committee 318, An ACI Standard and Report 318-
14, 2014.
[26] American Concrete Institute Committee 544, ‘‘Design considerations for steel
fiber reinforced concrete,” International Concrete Abstracts Portal, vol. 85, no.
5, pp. 563–579, 1988.
[27] M. Imam, L. Vandewalle, F. Mortelmans, Shear-moment analysis of reinforced
high strength concrete beams containing steel fibres, Can. J. Civil. Eng. 22
(1995) 462–470.
[28] S.A. Al-Ta’an, J.R. Al-Feel, Evaluation of shear strength of fibre-reinforced
concrete beams. Cem. Concr. Comp. 1990, 12,2, 87–94.
[29] S. Sritharan, B.J. Bristow, V.H. Perry, Characterizing an Ultra-High Performance
Material For Bridge Applications Under Extreme Loads, in: 3rd Int. Symp. High
Perform. Concr. 2003.
[30] S. Momenzadeh, O. Seker, M. Faytarouni, J. Shen, Seismic performance of all-
steel buckling-controlled braces with various cross-sections, J. Constr. Steel
Res. 139 (2017) 44–61, https://doi.org/10.1016/J.JCSR.2017.09.003.
[31] J. Shen, O. Seker, B. Akbas, P. Seker, S. Momenzadeh, M. Faytarouni, Seismic
performance of concentrically braced frames with and without brace buckling,
Eng. Struct. 141 (2017) 461–481, https://doi.org/10.1016/J.
ENGSTRUCT.2017.03.043.
[32] G.H. Mahmud, Z. Yang, A.M.T. Hassan, Experimental and numerical studies of
size effects of Ultra High Performance Steel Fibre Reinforced Concrete
(UHPFRC) beams, Constr. Build. Mater. 48 (2013) 1027–1034, https://doi.org/
10.1016/j.conbuildmat.2013.07.061.