Construction and Building Materials 188 (2018) 633–644

Contents lists available at ScienceDirect

Construction and Building Materials

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

Compressive behavior of UHPFRC under quasi-static and seismic strain

rates considering the effect of fiber content
Pablo Augusto Krahl ⇑, Gustavo de Miranda Saleme Gidrão, Ricardo Carrazedo
School of Engineering of São Carlos, University of Sao Paulo, Av. Trabalhador Saocarlense, 400, 13566-590 Sao Carlos-SP, Brazil

h i g h l i g h t s

DIF values for mechanical properties of UHPFRC under seismic strain rates are presented.

Dynamic modulus of elasticity measured from three different technics are presented.

Stress-strain equations for static and dynamic compressive behaviors are presented.

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

Article history: The high ductility and toughness of ultra-high performance fiber-reinforced concrete (UHPFRC) make the
Received 11 January 2018 composite a promising material to be applied in earthquake-resistant structures. For this reason, the
Received in revised form 14 June 2018 knowledge of its constitutive behavior on seismic strain rates (102 s1) becomes of great importance,
Accepted 18 August 2018
which was not reported for compressive loading yet. Thus, the purpose of this paper is present experi-
Available online 28 August 2018
mental results and propose a uniaxial stress-strain equation that predicts the behavior of UHPFRC under
compression at seismic strain rate considering different fiber contents. The experimental tests were per-
Keywords:
formed with cylindrical specimens in a servo-hydraulic machine under seismic and quasi-static strain
Ultra-High Performance Fiber Reinforced
Concrete
rates for comparison and to determine the dynamic increase factor (DIF). Also, nondestructive tests were
UHPC performed to determine the dynamic modulus of elasticity. From experiments, UHPFRC was rate sensitive
Strain rate mainly for peak stress and toughness. The DIF observed for peak strength were 1.12 and 1.11 for UHPFRC
Dynamic constitutive behavior, dynamic with 1% and 2.5% of fibers. For toughness, the DIF was 1.21 for the composite with 1.0% of fibers. Peak
increase factor strain and modulus of elasticity instead were not rate sensitive. Also, high standard deviations character-
ized the test results. From the analytical study, the proposed constitutive model predicted the dynamic
experimental results accurately for different fiber contents.
Ó 2018 Elsevier Ltd. All rights reserved.

1. Introduction
Ultra-high performance fiber reinforced concrete (UHPFRC) has
a high capacity of energy absorption under severe loading due to
the optimal combination of a well graded matrix with surface
treated steel fibers. The excellent interaction between matrix and
fibers makes UHPFRC a ductile composite with high toughness.
The superior mechanical performance compared to other concretes
attracted the attention of researchers to apply UHPFRC under
seismic conditions of loading, [1–6].
As is well accepted, the mechanical properties of cement-based
materials are sensitive to strain rates higher than quasi-static [7,8].
Also, the sensibility is not the same for tensile and compressive
⇑ Corresponding author.
E-mail addresses: pablokrahl@usp.br (P.A. Krahl), carrazedo@sc.usp.br
(R. Carrazedo).
loading [9], and the volume of fibers can present an essential role
in the behavior [9,10]. For UHPFRC under compression, most of
the researches have been developed for high strain rates, for exam-
ple, tests with split-Hopkinson pressure bar (SHPB) [11,12]. No
results were found for UHPFRC under compression at seismic
strain rates, which are comprised in the range between 5  103
and 5  101 s1 [10,13]. Consequently, besides the tests results,
the constitutive modeling for this material is also required for fur-
ther applications of UHPFRC in research and construction at seis-
mic areas.
Recently, some of the major finds for UHPFRC under compres-
sive dynamic loading were reported and deeply discussed [9]. In
short, UHPFRC is less rate-sensitive than normal-strength concrete,
but the compressive strength, strain capacity, and toughness are all
improved by the increase in strain rate. Furthermore, it is worth
highlighting that the effect of fiber content on the dynamic behav-
ior of UHPFRC is more significant on toughness than on the other

https://doi.org/10.1016/j.conbuildmat.2018.08.121
0950-0618/Ó 2018 Elsevier Ltd. All rights reserved.
634 P.A. Krahl et al. / Construction and Building Materials 188 (2018) 633–644

mechanical properties [11]. Also, several researches have been 100

raised hypothesis about the possible physical mechanisms that
make cement-based materials rate-sensitive, but there is no a uni- Fine sand
versally accepted explanation for the phenomena [10]. Lu et al. 80 Cement

Percentage pass (%)

Silica fume
[14] presented a detailed discussion, from a macroscopic view,
Quartz powder
about the rate sensitivity, and in brief, they attributed the improve-
60
ment of mechanical properties to changes in the nature of fracture
process due to high strain rates, also to the viscous behavior and
inertial effects. 40
In the field of constitutive modeling, a conventional approach to
take into account the dynamical effects is through the dynamic
increase factor (DIF) [15]. The DIF is the ratio between the dynamic 20
and quasi-static properties. Su et al. [16] determined DIF values for
UHPFRC under compressive and tensile stress states. For UHPFRC 0
with 3% of fibers submitted to a strain rate of 90 s1, DIF was 0.1 1 10 100
1.54 for compression strength, when 3% of nano-CaCO3 was added Particle size (μm)
to the mix. The authors emphasized the need for experimental
results in the range of 0 to 40 s1, which concern the seismic strain Fig. 1. Particle size distribution of the UHPFRC constituents.
rate.
Furthermore, the research on constitutive modeling of UHPFRC
is in current development. Few studies were reported even for
homogeneous mixture was obtained. Then, water and superplasti-
compressive quasi-static response [17–20] and no studies for seis-
cizer were added, and concrete was mixed during additional
mic strain rates. The fib Model Code [15] recommends that the
10 min to get the desired workability. Finally, steel fibers were
same constitutive model utilized for quasi-static case can be used
included, and concrete was mixed for more 5 min to obtain a
to simulate the dynamic behavior. The changes to be made are on
homogeneous distribution of fibers. Concrete was then molded
the model parameters that must be calculated considering the DIF
on a vibrating table to improve densification. Following the casting
values correspondent to the strain rate. For quasi-static conditions,
process, all samples were stored in a moist chamber for 28 days.
there exists many analytical models to describe the compressive
After this, the samples were submerged in water for heat treat-
behavior of fiber-reinforced concrete, [21–23]. Besides, some stud-
ment for seven days at 70 °C. Nine specimens were cast for each
ies for UHPFRC under high rates of loading utilized the continuum
fiber content in which six were tested in quasi-static strain rate
damage mechanics framework where the damage variable is
and three in dynamic strain rate.
described by a Weibull function [23,24]. In this methodology,
experimental results are required to calibrate the parameters of
2.2. Test setup and loading procedure
the function that describes the damage variable. Thus, both of
the mentioned methodologies have a statistical base to determine
The compression tests were performed on cylindrical speci-
the dynamical response.
mens of 50 mm in diameter and 100 mm in length. The test was
From the discussion above, the current research focused on pre-
carried out under displacement control in a universal servo-
sent experimental results for UHPFRC under compressive seismic
hydraulic testing machine with a maximum load capacity of
strain rates. From the stress-strain curves, peak stress, peak strain,
1500 kN. Besides, an axial hinge was installed on the top of the
modulus of elasticity, toughness are reported. Quasi-static results
specimen. Two clip gages were attached to the UHPFRC samples
are also reported for comparison and the determination of the DIFs.
to measure axial displacement with a gauge lenght of 50 mm, as
Furthermore, analytical stress-strain models are proposed for both
presented in Fig. 2. Also, two axial LVDTs (Linear Variable Differen-
cases of strain rate. Also, measurements of dynamic modulus of
tial Transformer) were fixed between the steel plates of the test
elasticity are determined through standardized acoustic and ultra-
machine. The LVDTs allow to register the post-peak stress versus
sonic tests.

2. Experimental program

2.1. Materials, mix proportions and procedure, casting and curing

process

The materials used to produce a self-compacting UHPFRC were

768 kg/m3 of type III Portland cement, 192 kg/m3 of silica fume
(specific surface of 20000 m2/kg and density of 2220 kg/m3),
844 kg/m3 of fine sand with grain size less than 0.42 mm,
384 kg/m3 of quartz powder (density of 2670 kg/m3), 69 kg/m3 of
polycarboxylate-based superplasticizer (density of 1060 kg/m3)
and 154 kg/m3 of water. UHPFRC with five different fiber volumes
(1%, 1.5%, 2%, 2.5% and 3%) were casted with copper-coated steel
fibers with 13 mm of length and 0.2 mm of diameter. The fibers
had tensile strength of 2850 MPa and modulus of elasticity of
200 GPa. Fig. 1 presents the particle size distribution of the mix
constituents.
High shear pan mixer for mixing UHPC was utilized. Firstly, all
dry components, excepting fibers, were mixed for 5 min until a Fig. 2. Compression test set-up.
 P.A. Krahl et al. / Construction and Building Materials 188 (2018) 633–644
Fig. 3. Clip gage to measure longitudinal strain.
strain relationship. However, they measure strains from the hinge
of the test machine that must be corrected by the measures of the
clip gages, Eq. (1). The modulus of elasticity was determined with
the clip gage measurements at a stress level of 40% of the ultimate
load, as recommended by ASTM C469 [25].
Fig. 4 and Fig. 3 present a more detailed view of the clip gage.
Two points of each clip are fixed at the sample with the distance
between them of 50 mm that configures the gauge length. One of
these points is the end of a flexible steel beam where strain gages
were fixed to register the displacement correspondent to the gauge
length.
The displacement applied at quasi-static compression tests had
the speed of 0.005 mm/s. For the gauge length of the LVDTs that is
100 mm corresponds to the strain rate of 5  105 s1. The speed
for dynamic tests was 1 mm/s whose strain rate was 102 s1.
The inclination of the ascending branch of the stress-strain
curves obtained from clip and LVDT are different because the
LVDTs measurements account for accommodations between the
machine plates and specimen. Eq. (1) enables to correct such effect
and adjust the strains of the pre-peak and post-peak measured by
the LVDT for the stiffness resitered on the clip gage, [26],


Eclip  ELVDT
e1 ¼ e1;LVDT  r1
Eclip ELVDT
Where e1, LVDT is the average axial strain measured by the LVDTs, r1
is the axial stress, Eclip and ELVDT are the modulus of elasticity
obtained from the clip and LVDTs readings. Fig. 5 presents the con-
version performed by Eq. (1) for the effective stiffness of the sample.
Dynamic modulus of elasticity E0,d was also determined for the
UHPFRC samples through acoustic and ultrasonic test technics. The
Fig. 4. Details of the clip gage.
635
180
160
140
120 Ajusted curve
Stress (MPa)
100 LVDT
Vf = 2 %
80
60
40
20
0
0.00 0.01 0.02 0.03 0.04 0.05
Strain
Fig. 5. Conversion of the LVDT measurements by Eq. (1).
procedure for the acoustic test is established by ASTM C215 [27]
and ASTM E1876-01 [28]. The ultrasonic tests were conducted in
the cylindrical samples by direct transmission approach, according
to ACI 228.2R-13 [29].
3. Test results and discussions
The dynamic increase factor (DIF) is quantitative measurement
of the strain rate effect. DIF shows the degree of enhancement of
any mechanical property C due to the influence of strain rate e_ d
relative to the quasi-static strain rate e_ s , Eq. (2).
Cðe_ d Þ
DIF ðe_ d Þ ¼ ð2Þ
Cðe_ s Þ
This parameter is utilized in the present research to character-
ize dynamic peak stress, peak strain, modulus of elasticity, tough-
ness, and the stress at a strain of 0.035. Fig. 6 (a) and (b) present the
experimental stress-strain curves for UHPFRC under quasi-static
and dynamic loading, respectively. Furthermore, Fig. 6 (b) presents
ð1Þ the comparison between static and dynamic responses in a
reduced window.
Fig. 6 (b) shows in the comparison inside the reduced windows
that UHPFRC is rate sensitive. Furthermore, three distinct stages
were registered for the entire stress-strain curves under static
and dynamic loads. Both, quasi-static and dynamic stress-strain
curves presented linear ascending branch until near peak strain
(80% to 90% of the peak). First cracking and strain localization near
the peak load characterized the pre-peak branch. Then, the first
post-peak range was characterized by a sharp drop in load due to
the appearance of macrocracks. The high decay in load is associ-
ated to the slip between fiber and matrix until a level that the
developed friction prevails under the reduced applied stresses
and consequently the rate of slip reduces. This stage ranges from
the peak stress to a characteristic inflection point. Finally, the sec-
ond post-peak range presents a gradual decrease in load where
prevails frictional slip and fiber dowel action in the main cracks.
The fiber content governs the whole post-peak range and therefore
the ductility and toughness of the composite.
Table 1 summarizes the average modulus of elasticity and peak
stress of UHPFRC with different fiber contents subjected to quasi-
static and dynamic strain rates. Also, the DIF values and the
standard deviation for each variable is presented. Toughness is
considered here as the area under the stress-strain diagram until
636 P.A. Krahl et al. / Construction and Building Materials 188 (2018) 633–644

180 180

150 Experiments 150

Average curve
Stress (MPa)

Stress (MPa)
120 120
Vf = 1.5 %
90 90
Vf = 1.0 %
60 60

30 30

0 0
0 1 2 3 0 1 2 3
Strain (%) Strain (%)
180 180

150 150

Stress (MPa)
120
Stress (MPa)

120
Vf = 2.0 % Vf = 2.5 %
90 90

60 60

30 30

0 0
0 1 2 3 0 1 2 3
Strain (%) Strain (%)
180 180 180

1% 150

150 1.5 %
150 120
2%
90
2.5 %
Stress (MPa)

120 120 3%
Stress (MPa)

60
Vf = 3.0 % 30

90 90 0
0.0 0.2 0.4 0.6 0.8 1.0

60 60

30 30

0 0
0 1 2 3 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Strain (%) Strain (%)
(a)
Fig. 6. (a) Quasi-static stress-strain relations and (b) Dynamic stress-strain relationships.

the strain of 0.035. Each quasi-static test result was related to each
dynamic test result to obtain the average DIF for each fiber content.
It can be seen from the results that seismic strain rate had no
significant effect on the modulus of elasticity of UHPFRC by observ-
ing the standard deviations. For peak stress, besides all DIFs were
greater than 1, only UHPFRC with 1.5% and 2.5% of fibers presented
a significant increase due to dynamic effects based on the standard
deviations. Thus, besides significative rate effects were observed
for some cases, no conclusions can be drawn for the fiber content
for the presented variables. Table 2 presents the results of peak
strain and toughness for UHPFRC with different fiber content.
Peak strain suffered no significant effect due to seismic strain
rates. In the case of toughness, UHPFRC with 1% of fibers presented
a high increase due to dynamic effects. Again, the fiber content did
not present a trend relative to the strain rate. It can be seen from
the results of both tables that the DIF values for all variables pre-
sented high dispersion.
3.1. Modulus of elasticity
Modulus of elasticity is an essential property for predictions of
structural behavior. Fig. 7 presents the average experimental
results for quasi-static and dynamic modulus of elasticity for dif-
ferent fiber contents and the linear approximations based on
regression analysis of the experimental data. The results of mea-
surements from three distinct technics for dynamic modulus are
presented (from clip gages, acoustic and ultrasonic technics).
As observed in Fig. 7, the modulus of elasticity E0 increased with
the increase in fiber content. For quasi-static condition, by compar-
ing UHPFRC with 1% and 3% of fibers, the increase was 7.2%. This
 P.A. Krahl et al. / Construction and Building Materials 188 (2018) 633–644 637

180 180 180 180

150 150

120
Dynamic Dynamic
150 150 120
Quasi-static Quasi-static
90 90

60 60
120

Stress (MPa)
120
Stress (MPa)
30 30

0 0
90 0 1 2 3 90 0 1 2 3

60 Vf = 1.0 % 60 Vf = 1.5%
30 30

0 0
0 1 2 3 0 1 2 3
Strain (%) Strain (%)
180 180 180 180

150 150
Dynamic Dynamic
120 120
150 Quasi-static 150 Quasi-static
90
90
60
60

Stress (MPa)
Stress (MPa)

120 120 30
30
0
0 0 1 2 3
0 1 2 3 90
90

Vf = 2% 60
Vf = 2.5%
60

30 30

0 0
0 1 2 3 0 1 2 3
Strain (%) Strain (%)
180 180 180 180

150
Dynamic 1% 150
120 Quasi-static 1.5 % 120
150 150
90 2% 90
60 2.5 %
60
Stress (MPa)

120 3%
Stress (MPa)

120 30
30
0
0 1 2 3 0
90 90 0.0 0.5 1.0 1.5

Vf = 3%
60 60

30 30

0 0
0 1 2 3 0 1 2 3
Strain (%) Strain (%)
(b)
Fig. 6 (continued)

Table 1
Rate effect on modulus of elasticity and peak stress of UHPFRC.

Mixture ID Strain rate Modulus of elasticity E0 Peak stress f0

(s1) Avg. (GPa) Avg. DIF r Avg. (MPa) Avg. DIF r
UHPFRC – 1% 0.00005 38.85 134.34
0.01 38.52 0.99 0.05 149.91 1.12 0.06
UHPFRC – 1.5% 0.00005 39.72 146.15
0.01 41.37 1.04 0.03 153.03 1.05 0.06
UHPFRC – 2% 0.00005 40.21 148.77
0.01 41.08 1.02 0.03 156.84 1.06 0.08
UHPFRC – 2.5% 0.00005 41.33 148.47
0.01 41.38 1.00 0.04 164.67 1.11 0.05
UHPFRC – 3% 0.00005 41.67 151.71
0.01 42.78 1.03 0.05 157.17 1.04 0.05
638 P.A. Krahl et al. / Construction and Building Materials 188 (2018) 633–644

Table 2
Rate effect on peak strain and toughness of UHPFRC.

Mixture ID Strain rate Peak strain e0 Toughness

(s 1
) Avg. (%) Avg. DIF r Avg. (MPa) Avg. DIF r
UHPFRC – 1% 0.00005 0.405 1.40
0.01 0.414 1.03 0.08 1.70 1.21 0.11
UHPFRC – 1.5% 0.00005 0.451 1.60
0.01 0.411 0.91 0.05 1.66 1.04 0.12
UHPFRC – 2% 0.00005 0.409 1.80
0.01 0.421 1.04 0.11 1.96 1.10 0.15
UHPFRC – 2.5% 0.00005 0.429 1.78
0.01 0.453 1.06 0.11 2.13 1.21 0.23
UHPFRC – 3% 0.00005 0.426 2.10
0.01 0.445 1.04 0.08 2.24 1.09 0.18

50 sensitive and its constitutive behavior must be studied for the

Quasi-static material application. Fig. 8 presents the peak strain and peak stress
Modulus of elasticity (GPa)

Dynamic, clip
of UHPFRC under dynamic and quasi-static loading for different
48 Sonelastic long.
Sonelastic flex. fiber contents.
Ultrasound For peak strain, the maximum DIF was 1.06 for UHPFRC with 2%
46 of fibers. However, a high standard deviation was observed for this
variable, as discussed previously. In the case of peak stress, the
44 maximum DIF was 1.12 for UHPFRC with 1% of fibers. Pyo et al.
[31] had a similar found for the first cracking of UHPFRC with 2%
42 of fibers (25 mm of length and 0.2 mm in diameter) under tensile
loading. The DIF was 1.15 for the increase in strain rate from
40 104 s1 to 102 s1.
It can be seen from the Fig. 8 that the addition of fibers
increased the peak stress under quasi-static and dynamic loading.
38
1.0 1.5 2.0 2.5 3.0 The microcracking evolution is more inhibited for higher fiber con-
Fiber content (%) tents. Thus, the localization is delayed as a consequence, and the
peak load is greater for the higher fiber contents. Under dynamic
Fig. 7. Static and Dynamic modulus of elasticity for UHPFRC. loads, inertial effects are added together giving lateral confinement
against the lateral expansion [32], that makes the damage distribu-
tion more diffuse enhancing the energy demand to propagate
increase was expected due to the addition of the fibers that have a cracks. Such behavior dissipates more energy than in quasi-static
greater modulus of elasticity. However, the increase in load rate loading, and for this reason, the composite presents greater peak
did not reflect a variation in DIF relative to fiber content, that is, stress. Besides, other effects [14], as viscous behavior, can take
the DIF was similar for all fiber contents. The maximum DIF found place.
for E0 from the clip gages measurements was 1.04. For the acoustic Linear fittings for peak stress as a function of fiber content are
test with longitudinal impact and transversal impact, the maxi- presented for static and dynamic loading, Eqs. (8) and (9).
mum DIFs were respectively 1.09 and 1.10, for the ultrasonic test
was 1.18. The higher velocity of the propagation of ultrasonic f 0;s ¼ 130:12 þ 7:48 V f ð8Þ
waves in UHPFRC compared to the other technics can explain the
f 0;d ¼ 143:94 þ 5:68 V f ð9Þ
higher values of modulus of elasticity.
From Fig. 7, the Eqs. (3) to (7) present the approximations for The addition of fibers did not present a conclusive tendency on
the static and dynamic modulus of elasticity obtained from regres- the peak strain at the quasi-static condition. Thus, the average
sion analysis as a function of fiber content. result 0.00424 was adopted for the analytical model. For dynamic
E0;s ¼ 37:46 þ 1:45 V f ð3Þ strain rates, Eq. (10) determines the peak strain as a function of the
volume of fibers Vf.
E0;d;clip ¼ 37:61 þ 1:71 V f ð4Þ e0;d ¼ 4:02  103 þ 1:59  104 V f ð10Þ

E0;d;sl ¼ 41:95 þ 0:59 V f
E0;d;sf ¼ 41:77 þ 1:00 V f
E0;d;u ¼ 43:92 þ 1:12 V f
3.2. Compressive strength and strain at peak
This section presents the results for peak strain and peak stress
for UHPFRC with different fiber contents and the two studied strain
rates. Chen et al. [30] already found that UHPFRC is less sensitive to
the increase in strain rate than normal-strength concretes, mainly
due to the denser matrix. However, the composite is still rate
ð5Þ Another point that is commonly discussed about the constitu-
tive behavior of UHPFRC is the linearity of pre-peak branch of
ð6Þ the compressive stress-strain response [33]. According to Graybeal
[33], the linearity of the stress-strain curve until peak stress can be
ð7Þ evaluated by the ratio between the initial modulus of elasticity E0
obtained according to ASTM C469 [25] and the secant modulus for
peak stress Esec. Then, higher the ratio E0/Esec, the higher is the non-
linearity of the pre-peak branch. Thus, it is expected that the ratio
be smaller for UHPFRC than normal strength concretes. From the
present results, this ratio also shows the influence of fiber content
and strain rate on the pre-peak nonlinear behavior. Graybeal [33]
measured 1.1 for E0/Esec, for a steam treated UHPFRC with 2% of
fiber content under quasi-static loading. Fig. 9 presents static and
dynamic E0 / Esec for different fiber contents of the present research.
 P.A. Krahl et al. / Construction and Building Materials 188 (2018) 633–644 639

0.50 170
0.48

Peak Stress (MPa)

160

Peak strain (%)

0.46
0.44 150
0.42
140
0.40
Static
Dynamic
0.38 130
Linear fit static
0.36 Linear fit dynamic
120
1.0 1.5 2.0 2.5 3.0 1.0 1.5 2.0 2.5 3.0
Fiber content (%) Fiber content (%)

Fig. 8. Peak strain and peak stress for UHPFRC with different fiber contents.

1.30 2.4

1.25 Static 2.2

Toughness (MPa)
Dynamic
2.0
E0 / Esec

1.20

1.8
1.15
1.6
1.10
1.4

1.05
1.0 1.5 2.0 2.5 3.0
1.2
1.0 1.5 2.0 2.5 3.0
Fiber content (%) Fiber content (%)
Fig. 9. Ratio between initial tangent modulus of elasticity and secant modulus. Fig. 10. Quasi-static and dynamic toughness of UHPFRC.

From the UHPFRC with 1% of fibers, under quasi-static condi-
tions, the increase in fiber volume fraction always decreased the
value of E0/Esec which means that the nonlinear pre-peak behavior
is more pronounced for the composite with lesser fiber content. In
the case of seismic strain rate, this effect follows the opposite
trend. The higher E0/Esec value was found for UHPFRC with 3% of
fibers.
3.3. Toughness of UHPFRC
UHPFRC under very high strain rates presents a remarkable
increase in toughness compared to quasi-static strain rates,
[11,12,34]. Besides, Lai and Sun [11] showed that the increase in
fiber volume fraction also reflects an increase in dynamic effects
compared to UHPFRC with lower fiber contents. Fig. 10 presents
the results of toughness for UHPFRC under seismic strain rates
compared to quasi-static results.
The strain rate and fiber effects were apparent on increasing
toughness. However, as opposed to the results for high strain rates,
[11], the DIF did not increase with an increase in fiber content. The
maximum DIF was 1.21 for UHPFRC with 1% of fibers. The presence
of fibers retains the typical sudden failure of UHPC, and its quantity
governs how much energy the composite can dissipate in the post-
peak regime, as observed in Fig. 10. Also, inertial effects cause lat-
eral confinement on the samples under dynamic load, which in
turn increase toughness. The confinement can increase the fric-
tional stresses developed between fiber and matrix [35] that make
the damage propagation more diffuse due to the lower fiber slip
and consequent greater stress transfer between crack faces.
Furthermore, a characteristic inflection point (eip, rip) was
observed in static and dynamic responses in the post-peak range.
This point separates the branch of sudden drop in load and the
more gradual softening after stabilization. To evaluate the increase
in energy dissipation capacity between peak load and the inflection
point due to dynamic load and the increase in fiber content, the
area under this range of the stress-strain curve was calculated
for both strain rates and denominated as T1. A regression analysis
in the post-peak range of the stress-strain curve with a bilinear
function allowed to determine numerically the inflection point.
Fig. 11 (a) presents the values for T1 against fiber content and
Fig. 11 (b) presents the strain correspondent to the inflection point
against the energy T1 of static and dynamic loading.
Fig. 11 (a) showed that T1 is rate sensitive and the increase in
fiber content presented a tendency for DIF. The maximum DIF for
T1 was 1.72 for UHPFRC with 2.5% of fibers. Besides, the energy
T1 presented simultaneously an excellent correlation with the
strain eip at the inflection point for static and dynamic load.
Eqs. (11)–(13) presents the approximations respectively for sta-
tic and dynamic T1 as a function of Vf and eip which addresses static
and dynamic cases.
640 P.A. Krahl et al. / Construction and Building Materials 188 (2018) 633–644

1.1 1.6

1.0 1.5 Static

Dynamic
0.9
Static 1.4 Dynamic
Linear fit
0.8 1.3
T1 (MPa)

εip (%)
0.7 1.2

0.6 1.1

0.5 1.0

0.4 0.9

0.3 0.8
1.0 1.5 2.0 2.5 3.0 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1
Fiber content (%) T1 (MPa)
(a) (b)
Fig. 11. Approximations for static and dynamic T1 and eip.

44
4. Analytical solutions
Static
40 Dynamic
Linear fit static 4.1. Quasi-static stress-strain equations
36 Linear fit dynamic
From the experimental observations, the current analytical
σu (MPa)

32 proposition divides the stress-strain curve into three parts. First,

a nonlinear ascending branch until peak. Then, a bilinear descend-
28 ing branch divided by the characteristic inflection point identified
in the experiments. For the first part, the model of Carreira and Chu
24 [36] enables to determine the nonlinear behavior for UHPFRC with-
out the need of curve fitting, Eq. (16).
20 2 3
6 bee 7
16 f ¼ f 04 0 b 5
1.0 1.5 2.0 2.5 3.0 b1þ ee
0
ð16Þ
Fiber content (%)
b¼ 1
f0
1e
Fig. 12. Ultimate stress versus fiber content. 0 E0

Where f0, e0, E0 are the peak stress, peak strain and initial modulus
T 1;s ¼ 0:42 þ 5:26  103 e1:4V f ð11Þ of elasticity, respectively, which are known in advance. For the
descending branch of the stress-strain curve, a bilinear model is
T 1;d ¼ 0:55 þ 20:09  103 e1:08V f ð12Þ considered [37], with the two linear branches are separated by
the statistically obtained inflection point (rip,eip), Eq. (17).
eip ¼ 5:39  103 þ 8:51  103 T 1 ð13Þ (

mðe  e0 Þ þ f 0 ; e0 < e < eip
r¼

 ð17Þ
Also, the point correspondent to the strain of 0.035 was consid- n e  eip þ rip ; e > eip
ered as the ultimate state of stress and strain (eu, ru). Fig. 12 pre-
sents the average experimental stress ru versus fiber content.
Under dynamic load, the second range of the post-peak pre-
sented a faster rate of degradation of bearing capacity. So, the
stress at 3.5% was typically smaller for UHPFRC under dynamic
load. The dynamic effect decreased at higher rates the bearing
capacity of UHPFRC at this strain level, except for UHPFRC with
1% of fibers. As shown in Fig. 6 (b), the inclination of the softening
branch, after the inflection point, was higher for UHPFRC under
dynamic load. The average inclination for the seismic strain rates
was -783.98 MPa and -431.58 MPa for the quasi-static cases. Thus,
the mechanisms involved in the damage process under dynamic
load reduced at higher rates the bearing capacity of UHPFRC after
the inflection point. Eqs. (14) and (15) present linear approxima-
tions respectively for static and dynamic ru.
ru;s ¼ 15:48 þ 7:61 V f ð14Þ

ru;d ¼ 16:95 þ 5:07 V f ð15Þ

Fig. 13. Representation of the variables for the present constitutive model.
 P.A. Krahl et al. / Construction and Building Materials 188 (2018) 633–644 641

where m and n are the inclinations of the two straight lines that 2T 1
rip ¼  f0 ð18Þ
compose the post-peak range until the considered ultimate strain eip  e0
of 0.035. Zhou et al. [37] proposed values for rip and eip as a per-
The energy dissipated from the inflection point until the ulti-
centage of the peak stress and strain. In the present study, rip
mate strain did not present a good correlation with the fiber vol-
and eip are determined as a function of the energy T1, Eq. (11), dis-
ume fraction as opposed to T1. However the stress ru can be
sipated between peak strain e0 and the inflection point eip, see
determined as a function the Vf with a reasonable approximation,
Fig. 13 that also presents the variables of the constitutive model.
Eqs. (14) and (15). Fig. 14 shows the comparison of the proposed
Eqs. (11) and (12) allows to determine T1 for static and dynamic
constitutive model and the envelop curves of the experimental
cases, respectively, and eip by Eq. (13). The area correspondent to
results for quasi-static loading.
the energy T1 is considered as a trapezium. Therefore, Eq. (18)
As seen from Fig. 14, the proposed model predicted the test
enables to determine the stress rip at the inflection point.
results with great accuracy. As the parameters rip, eip, ru, and eu

180 180

150 150
Experimental
Stress (MPa)

120 Analytical

Stress (MPa)
120 Vf = 1.5 %

90 90
Vf = 1.0 %

60 60

30 30

0 0
0 1 2 3 0 1 2 3
Strain (%) Strain (%)
180 180

150 150
Stress (MPa)

120 Vf = 2.5 %
Stress (MPa)

120
Vf = 2.0 %
90 90

60 60

30 30

0 0
0 1 2 3 0 1 2 3
Strain (%) Strain (%)

180

150
Stress (MPa)

120
Vf = 3.0 %
90

60

30

0
0 1 2 3
Strain (%)
Fig. 14. Comparison between quasi-static analytical and experimental results.
642 P.A. Krahl et al. / Construction and Building Materials 188 (2018) 633–644

are dependent on fiber content and do not account for fiber constitutive models as for quasi-static loading but considering
aspect ratio, they must be calibrated to be applied in the prediction the parameters of the model increased by the DIF. For compression,
of the stress-strain behavior of UHPFRC different from the present the Model Code recommends DIFs for normal strength concrete,
study. but there are no recommendations for UHPFRC. In the present
study, DIF is adopted as a function of fiber content and was deter-
4.2. Stress-strain relation for dynamic loading mined statistically from the present experimental results for
modulus of elasticity, peak stress, peak strain, toughness T1 and
It was found that UHPFRC is sensitive to seismic strain rates ru. Eqs. (19) to (23) present the results
(102 s1). Then, the dynamic increase factor (DIF) approach was  0:96
E0;d e_ d
chosen to model the dynamic behavior of this cement-based com- DIF E0 ¼ ¼ 1 þ Vf  ð19Þ
posite. The fib Model Code [15] recommends to utilize the same E0;s e_ s

180 180

150 Experimental
150
Analytical 1
120
Analytical 2

Stress (MPa)
120
Stress (MPa)

Vf = 1.5%
90 Vf = 1.0 % 90

60 60

30 30

0 0
0 1 2 3 0 1 2 3
Strain (%) Strain (%)
180 180

150 150
Stress (MPa)

120
Stress (MPa)

120
Vf = 2% Vf = 2.5%
90 90

60 60

30 30

0 0
0 1 2 3 0 1 2 3
Strain (%) Strain (%)

180

150
Stress (MPa)

120 Vf = 3%
90

60

30

0
0 1 2 3
Strain (%)
Fig. 15. Comparison between dynamic analytical and experimental results.
 P.A. Krahl et al. / Construction and Building Materials 188 (2018) 633–644
 0:74
1 f 0;s e_ d
¼ ¼ 0:89 þ V f 
DIF f 0 f 0;d e_ s
 0:60
e0;d e_ d
DIF e0 ¼ ¼ 0:94 þ V f 
e0;s e_ s
 0:54
T 1;d e_ d
DIF T 1 ¼ ¼ 1:30 þ V f 
T 1;s e_ s
 0:49
1 ru;s e_ d
¼ ¼ 0:99 þ V f 
DIF ru ru;d e_ s
With the DIF values and the model developed for quasi-static
behavior, the dynamic stress-strain behavior can be predicted. This
approach is called here as Analytical 1. Another approach consid-
ered is by utilizing directly the statistical results obtained from
dynamic tests, which is called as Analytical 2, which is the same
approach utilized for quasi-static modeling. Both are compared
to the experimental envelops, Fig. 15.
As observed, the analytical models predicted well the experi-
mental dynamic response. However, the solution Analytical 1,
obtained with DIF values, resulted in a tougher but acceptable
post-peak regime compared to the solution Analytical 2. The pre-
peak portion was similar for both analytical solutions. It should
be pointed out that besides the DIF values have been calculated
for peak strain and modulus of elasticity, these values were low
rate sensitive.
5. Conclusions
The present research aimed to present experimental results of
UHPFRC under compressive load at strain rate correspondent to
seismic loading. Besides, the effect of the steel fiber content was
also investigated including tests for quasi-static strain rate. From
the study the main finds follow:
- UHPFRC under compression was rate sensitive to seismic strain
rates (102 s1) for peak stress and toughness. The DIF values
observed for peak strength were 1.12 and 1.11 for UHPFRC with
1% and 2.5% of fibers. For peak strain and modulus of elasticity
can be concluded that UHPFRC was no rate sensitive. For tough-
ness, the DIF value was 1.21 for UHPFRC 1.0% of fibers. Also, the
results were characterized by high standard deviations. Thus,
besides DIFs were greater than 1 for several cases, no statisti-
cally based conclusions can be made for some fibers contents.
Furthermore, the fiber content associated to the rate effect did
not presented a trend in the DIF. Also, many studies attribute
the rate effect observed in cement-based materials to the
time-dependent nature of fracture processes, the free water
movement (viscous behavior) and the inertial effect that arise
with the increase in strain rate.
- It was observed that the fiber content affected the pre-peak
behavior when comparing static and dynamic strain rates. The
relation E0/Esec evaluated this effect. For quasi-static strain rate,
the increase in fiber content from 1% to higher values always
decreased the ratio E0/Esec. By contrast, under seismic strain
rates, the increase in fiber content increased the ratio E0/Esec.
Thus, it can be concluded that the rate effect makes the pre-
peak behavior more nonlinear as the fiber content increase as
opposed to the quasi-static condition.
- For the three distinct stages registered for the entire stress-
strain curves under static and dynamic loads, the second stage
was the most influenced by rate effects. The maximum DIF
found for the energy T1 dissipated at this stage was 1.72 for
UHPFRC with 2.5% of fibers. Furthermore, it was also observed
643
that the rate of decrease in load capacity from the inflection
ð20Þ
point to the ultimate strain of 0.035 was greater for seismic
strain rates.
- The static and dynamic experimental constitutive behaviors
ð21Þ were well predicted by the proposed analytical model. In addi-
tion, the dynamic behavior was predicted by the statistically
based approach and by the approach recommended by the fib
Model Code, which is based on the utilization of the DIF. The
ð22Þ
last approach resulted in tougher stress-strain curves, but
acceptable for design.
- The dynamic modulus of elasticity E0,d was also determined by
ð23Þ
acoustic and ultrasonic nondestructive technics. The high veloc-
ity of the ultrasonic wave propagation in UHPFRC resulted in
the higher E0,d, followed in sequence by acoustic tests and
mechanical tests. For UHPFRC with 1% fibers, the difference
between quasi-static modulus and dynamic modulus deter-
mined by ultrasonic technic was 15%.
Conflict of interest
There is no conflict of interest.
Acknowledgements
The authors gratefully acknowledge the financial support (grant
ID – 141993/2014-6) provided by National Counsel of Technologi-
cal and Scientific Development (CNPq) from Brazil.
References
[1] S. Balaji, G.S. Thirugnanam, Study on exterior RC beam-column joints upgrade
with SIFCON in joint core under reversed cyclic loading, KSCE J. Civ. Eng. 21
(2016) 1–7, https://doi.org/10.1007/s12205-016-0643-x.
[2] A. Elmenshawi, T. Brown, Deformation capacity of ultra-high strength concrete
flexural elements subjected to inelastic load reversals, Struct. Des. Tall Spec.
Build. 24 (2014) 421–439, https://doi.org/10.1002/tal.
[3] C. Hung, C. Chueh, Cyclic behavior of UHPFRC flexural members reinforced
with high-strength steel rebar, Eng. Struct. 122 (2016) 108–120, https://doi.
org/10.1016/j.engstruct.2016.05.008.
[4] C.-C. Hung, H. Li, H.-C. Chen, High-strength steel reinforced squat UHPFRC
shear walls: cyclic behavior and design implications, Eng. Struct. 141 (2017)
59–74, https://doi.org/10.1016/j.engstruct.2017.02.068.
[5] G. Palacios, Performance of Full-Scale Ultra-High Performance Fiber Reinforced
Concrete Column Subjected to Extreme Earthquake Type Loading and Effect of
Surface Preparation on the Cohesion and Friction Factors of the AASHTO
Interface Shear Equation, University of Texas, 2015.
[6] S. Xu, C. Wu, Z. Liu, K. Han, Y. Su, J. Zhao, J. Li, Experimental investigation of
seismic behavior of ultra-high performance steel fiber reinforced concrete
columns, Eng. Struct. 152 (2017) 129–148, https://doi.org/10.1016/j.
engstruct.2017.09.007.
[7] P. Bischoff, S. Perry, Impact behavior of plain concrete loaded in uniaxial
compression, J. Eng. Mech. 121 (1995) 685–693, https://doi.org/10.1061/
(ASCE)0733-9399(1995) 121:6(685).
[8] P.H. Bischoff, S.H. Perry, Compressive behaviour of concrete at high strain
rates, Mater. Struct. 24 (1991) 425–450, https://doi.org/10.1007/BF02472016.
[9] D.Y. Yoo, N. Banthia, Mechanical and structural behaviors of ultra-high-
performance fiber-reinforced concrete subjected to impact and blast, Constr.
Build. Mater. 149 (2017) 416–431, https://doi.org/10.1016/
j.conbuildmat.2017.05.136.
[10] R.J. Thomas, A.D. Sorensen, Review of strain rate effects for UHPC in tension,
Constr. Build. Mater. 153 (2017) 846–856, https://doi.org/10.1016/
j.conbuildmat.2017.07.168.
[11] J. Lai, W. Sun, Dynamic behaviour and visco-elastic damage model of ultra-
high performance cementitious composite, Cem. Concr. Res. 39 (2009) 1044–
1051, https://doi.org/10.1016/j.cemconres.2009.07.012.
[12] Z. Rong, W. Sun, Y. Zhang, Dynamic compression behavior of ultra-high
performance cement based composites, Int. J. Impact Eng. 37 (2010) 515–520,
https://doi.org/10.1016/j.ijimpeng.2009.11.005.
[13] L. Soufeiani, S.N. Raman, M.Z. Bin Jumaat, U.J. Alengaram, G. Ghadyani, P.
Mendis, Influences of the volume fraction and shape of steel fibers on fiber-
reinforced concrete subjected to dynamic loading – A review, Eng. Struct. 124
(2016) 405–417, https://doi.org/10.1016/j.engstruct.2016.06.029.
[14] D. Lu, G. Wang, X. Du, Y. Wang, A nonlinear dynamic uniaxial strength
criterion that considers the ultimate dynamic strength of concrete, Int. J.
Impact Eng. 103 (2017) 124–137, https://doi.org/10.1016/j.
ijimpeng.2017.01.011.
644 P.A. Krahl et al. / Construction and Building Materials 188 (2018) 633–644
[15] FIB (Fédération Internationale du Béton), (2010). Model Code 2010—Final
draft, Volume 1.
[16] Y. Su, J. Li, C. Wu, P. Wu, Z.X. Li, Influences of nano-particles on dynamic
strength of ultra-high performance concrete, Compos. Part B Eng. 91 (2016)
595–609, https://doi.org/10.1016/j.compositesb.2016.01.044.
[17] K.V. Harish, J.K. Dattatreya, M. Neelamegam, Experimental investigation and
analytical modeling of the r – e characteristics in compression of heat-treated
ultra-high strength mortars produced from conventional materials, Constr.
Build. Mater. 49 (2013) 781–796, https://doi.org/10.1016/j.conbuildmat.2013.
08.068.
[18] P.R. Prem, Influence of curing regime and steel fibres on the mechanical
properties of UHPC, Mag. Concr. Res. 67 (2015) 988–1002.
[19] 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.
[20] Z. Wu, C. Shi, W. He, D. Wang, Uniaxial compression behavior of ultra-high
performance concrete with hybrid steel fiber, J. Mater. Civ. Eng. 28 (2016) 1–7,
https://doi.org/10.1061/(ASCE)MT.1943-5533.0001684.
[21] M.A. Mansur, M.S. Chin, T.H. Wee, Stress-strain relationship of high-strength
fiber concrete in compression, J. Mater. Civ. Eng. 11 (1999) 21–29.
[22] D.A. Fanella, A.E. Naaman, Stress-strain properties of fiber reinforced mortar in
compression, ACI J. Proc. 82 (1985) 475–483.
[23] J. Zhou, X. Chen, Stress-strain behavior and statistical continuous damage
model of cement mortar under high strain rates, J. Mater. Civ. Eng. 25 (2013)
120–130, https://doi.org/10.1061/(ASCE)MT.1943-5533.0000570.
[24] X. Hou, S. Cao, Q. Rong, W. Zheng, G. Li, Effects of steel fiber and strain rate on
the dynamic compressive stress-strain relationship in reactive powder
concrete, Constr. Build. Mater. 170 (2018) 570–581, https://doi.org/10.1016/
j.conbuildmat.2018.03.101.
[25] . C469, /C469M-14, Standard Test Method for Static Modulus of Elasticity and
Poisson’s Ratio of Concrete in Compression, (2014) ASTM International, West
Conshohocken, PA
[26] E. Osorio, J.M. Bairán, A.R. Marí, Lateral behavior of concrete under uniaxial
compressive cyclic loading, Mater. Struct. 46 (2013) 709–724.
[27] A.S. for T.M. ASTM -02, Standard Test Method for Fundamental Transverse,
Longitudinal, and Torsional Resonant Frequencies of Concrete Specimens,
(2003) 1–7.
[28] A.S. for T.M. ASTM -01, Standard Test Method for Dynamic Young’ s Modulus ,
Shear Modulus , and Poisson’s Ratio by Impulse Excitation of Vibration, (2001).
[29] ACI 228.2R-98, Nondestructive test methods for evaluation of concrete in
structures, ACI Comm. Rep. (1998) 62.
[30] X. Chen, L. Xu, S. Wu, Influence of pore structure on mechanical behavior of
concrete under high strain rates, J. Mater. Civ. Eng. 28 (2016) 04015110,
https://doi.org/10.1061/(ASCE)MT.1943-5533.0001380.
[31] S. Pyo, K. Wille, S. El-Tawil, A.E. Naaman, Strain rate dependent properties of
ultra high performance fiber reinforced concrete (UHP-FRC) under tension,
Cem. Concr. Compos. 56 (2015) 15–24, https://doi.org/10.1016/j.
cemconcomp.2014.10.002.
[32] H. Hao, Y. Hao, J. Li, W. Chen, Review of the current practices in blast-resistant
analysis and design of concrete structures, Adv. Struct. Eng. 19 (2016) 1193–
1223, https://doi.org/10.1177/1369433216656430.
[33] B.A. Graybeal, Compressive Behavior of Ultra-High-Performance Fiber-
Reinforced Concrete, 146 (2007).
[34] Z. Wu, C. Shi, W. He, D. Wang, Static and dynamic compressive properties of
ultra-high performance concrete (UHPC) with hybrid steel fiber
reinforcements, Cem. Concr. Compos. 79 (2017) 148–157, https://doi.org/
10.1016/j.cemconcomp.2017.02.010.
[35] A.C. McSwain, K.A. Berube, G. Cusatis, E.N. Landis, Confinement effects on fiber
pullout forces for ultra-high-performance concrete, Cem. Concr. Compos. 91
(2018) 53–58, https://doi.org/10.1016/j.cemconcomp.2018.04.011.
[36] D.J. Carreira, K. Chu, Stress-strain relationship for plain concrete in compression,
ACI J. Proc. 82 (1985) 797–804 (http://www.concrete.org/Publications/
ACIMaterialsJournal/ACIJournalSearch.aspx?m=details&ID=10390).
[37] J. Zhou, J. Pan, A.M. Asce, C.K.Y. Leung, F. Asce, Mechanical Behavior of Fiber-
Reinforced Engineered Cementitious Composites in Uniaxial Compression, J.
Mater. Civ. Eng. 27 (2015) 1–10, https://doi.org/10.1061/(ASCE)MT.1943-
5533.0001034.