Construction and Building Materials 134 (2017) 563573

Contents lists available at ScienceDirect

Construction and Building Materials

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

Tensile properties of concrete at very early ages

Duy H. Nguyen a, Vinh T.N. Dao a,, Pietro Lura b,c
a
School of Civil Engineering, The University of Queensland, Australia
b
Empa, Swiss Federal Laboratories for Materials Science and Technology, Switzerland
c
Institute for Building Materials, ETH Zurich, Switzerland

h i g h l i g h t s

Testing for reliable tensile properties of very early-age concrete is reviewed.

The current dearth of such reliable data is emphasised.

Key features of an improved direct tensile testing system are presented.

Newly-collected reliable data and their analysis are reported.

Fundamental tensile properties of concrete at very early ages are presented.

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

Article history: Proper knowledge of tensile properties of concrete from very early ages is essential for effective control of
Received 14 July 2016 not only early-age cracking but also of residual stress due to restrained early-age deformation as well as
Received in revised form 11 November 2016 for demolding and handling of young concrete members in the precast industry. Despite significant past
Accepted 28 December 2016
research, such knowledge is currently inadequate and based on experimental data with questionable reli-
Available online 4 January 2017
ability, due mainly to the considerable challenges in testing early-age concrete.
This paper first highlights the challenge and importance of collecting reliable test data on full tensile
Keywords:
stress-deformation curves for very early-age concrete. Through identifying and effectively addressing
Early-age concrete
Direct tensile test
critical drawbacks in previous test setups, an improved direct tensile testing system that can reliably cap-
Tensile properties ture simultaneously stress and deformation of concrete from the age of several hours after mixing has
Fracture properties been successfully developed. Key features of the improved setup, including the air-bearing box for fric-
Cyclic loading tion minimisation and digital image correlation for non-contact full-field deformation capturing, are then
Tensile relaxation reported in the paper. Based on the newly collected data, fundamental tensile properties of concrete at
Digital image correlation very early ages are re-assessed and presented. Such properties include tensile strength, Youngs modulus,
strain at peak stress, fracture energy, performance under cyclic loading and tensile relaxation.
2016 Elsevier Ltd. All rights reserved.

1. Introduction well as critical structures such as concrete containments of nuclear

Early-age cracking may occur in concrete structures from as
early as several hours after casting. Very often, these early-age
cracks would further propagate and render these structures unser-
viceable at later stages due to subsequent shrinkage and loading.
As a result, the performance characteristics and service life of
affected concrete structures can be severely compromised. The
effective control of such cracking is thus of particular concern for
highway pavements, bridges, tunnels, liquid reservoirs [13], as
Corresponding author.
E-mail address: v.dao@uq.edu.au (V.T.N. Dao).
power plants and storage facilities for hazardous substances [3,4].
The basic underlying mechanism for such cracking is the tensile
stress (or strain) due to restrained early deformation in the con-
crete reaching its tensile capacity. Importantly, even when early-
age cracking does not occur, the residual stress due to restrained
early-age deformation can substantially reduce the remaining
tensile-carrying capacity of concrete. Such effect, if not properly
accounted for, may critically compromise the performance of con-
crete structures in service loads and environment [5].
Accordingly, a proper understanding and control of the above-
mentioned early-age cracking risk and residual stress from
restrained early-age deformation in concrete structures requires,
among other things, an adequate knowledge of tensile and fracture

http://dx.doi.org/10.1016/j.conbuildmat.2016.12.169
0950-0618/ 2016 Elsevier Ltd. All rights reserved.
564 D.H. Nguyen et al. / Construction and Building Materials 134 (2017) 563573
properties of concrete from very early ages. Such knowledge of
early-age tensile properties is also highly beneficial in precast
industry to optimize the timing for prestressing, demoulding and
handling of precast members [6,7].
Having recognised the significance of such knowledge, different
research groups have attempted to measure the relevant tensile
properties of concrete since the early 1970s, with useful but still
rather limited reported outcomes [814]. In [13], using a tensile
wedge splitting setup, the load-displacement up to peak stress of
young concrete specimens could be captured However, the
post-peak behaviour could not be obtained, rendering the determi-
nation of fracture properties challenging. In fact, the full stress-
strain curves have been reported by only a few groups [9,10]. This
is believed to be due mainly to the considerable practical chal-
lenges to be overcome when testing early-age concrete to obtain
reliable stress-deformation curves, which requires reliable captur-
ing of both stress and deformation at the same time. Specifically:

Due to the low tensile strength to be measured, friction
between the test specimen and its supporting surface must be
minimised if reliable loading is to be recorded. Various mea-
sures have been attempted [8,9], including the use of a mercury
bath [15], roller bearing [8] or materials with a low coefficient
of friction such as Teflon [16]. The best available method is pos-
sibly the inclusion of an air-bearing box which essentially
floats the test specimen on a thin layer of air, thus effectively
eliminating all friction The system was first proposed by Han-
nants group [9] and subsequently further improved by Dao
et al. [10].

Due to its fragility, disturbance to test specimens must be min-
imised throughout the moulding, handling and testing pro-
cesses. In particular, such fragility of early-age concrete has
important implications on reliable deformation measurement
Methods for deformation measurement in previous research
can be categorised into three groups:

Strain gauges are attached to the mould [10,17], giving total
deformation of the whole specimen Deformation over the
region of interest has to be estimated from the measured
total deformation, typically by finite element modelling. This
estimation procedure inevitably has introduced additional
uncertainties that have not been appropriately taken into
account, raising doubts over the reliability of obtained
results.

Strain gauges are attached to the posts cast into concrete
[18,19] The likely movement of the posts in young con-
crete and the resulting disturbance may have significantly
compromised the measurement accuracy; but again, the
effect of such movement and disturbance has not been ade-
quately quantified.

Digital Image Correlation (DIC): DIC uses high definition pro-
fessional cameras for reliable, non-contact capturing of the
required displacement fields. DIC has been used successfully
to track the free surface of flowing concrete [20] and capture
surface deformation of mortar specimens [21]. However,
there seem limited reported studies in which the reliable
deformation of concrete surface captured by DIC is synchro-
nised with reliable loading to provide the required load-
displacement curves. Roziere et al. [16] reported one such
study on concrete specimens from the age of 7 h after cast-
ing Accordingly, this technique may not be appropriate
for concrete of earlier ages due to the weaker, wetter and
softer nature of the concrete surface at this stage. Nguyen
and Dao [22] appear to be the first to have reported a suc-
cessful use of DIC for direct tensile testing of concrete from
3 h after mixing.
This paper reports key initial outcomes of a research program
ongoing at The University of Queensland aimed at addressing the
above-mentioned knowledge gaps, including:

An improved direct tensile test that enables the reliable and
simultaneous capturing of deformation and stress of concrete
specimens from very early ages.

On the basis of newly-collected tensile stress-deformation
curves, improved knowledge of such important tensile proper-
ties of early-age concrete, such as: tensile strength, Youngs
modulus, fracture energy, as well as concrete performance
under cyclic loading and tensile relaxation.
2. Experimental study
2.1. Direct tensile testing setup
The setup for direct tensile testing of concrete test specimens
used in this study, as schematically shown in Fig. 1, is based on
its earlier version [10] with significant improvement to allow reli-
able simultaneous capturing of stress and deformation.
During testing, the test apparatus is placed on the horizontal
platform of a displacement-controlled Instron 5985 testing frame.
The test apparatus itself comprises an air-bearing box and a
lever arm attached to a small steel frame . The lever arm is
pin-connected to the frame and is self-balanced in the test posi-
tion, enabling the direction but not the magnitude of the force
applied through the Instron loading machine to be altered.
The test specimen is placed on top of the air-bearing box with
one end pin-connected to the load cell and the other to the lever
arm. The test specimen and its dimensions are given in Fig. 2:
The minimum dimension of 70 mm is greater than the average
width of the fracture process zone, which is about three times
the maximum aggregate size used [23]. The curved transitions
aim to promote failure in the middle section while eliminating sig-
nificant stress concentrations that could initiate undesirable crack-
ing within the transitions [10].
The two notable features of this unique test setup include:

Air-bearing box with the top plate having 32 holes of 1 mm
in diameter symmetrically distributed under the test specimen:
During testing, an air pressure of 140 kPa is supplied to float the
test specimen on a thin layer of air, effectively eliminating the
friction between the test specimen and the supporting base.
Fig. 1. Schematic illustration of the direct tensile test setup.
 D.H. Nguyen et al. / Construction and Building Materials 134 (2017) 563573
Fig. 2. Test mould and concrete specimen with dimensions and patterned AOI.

Digital Image Correlation (DIC) with two high resolution pro-
fessional cameras (, in Fig. 1): This enables the non-
contact capturing of the desired deformation over the whole
area of interest (AOI, Fig. 2) without any disturbance to the
test specimen. Both cameras are synchronized to work
together with the loading process by a software developed
in-house at The University of Queensland. The angle formed
by each camera axis and the vertical line perpendicular to
the top surface is within the recommended optimal range of
between 10 and 20 for best analysis quality [24]. Four refer-
ence pattern areas (Fig. 2) are also applied to the steel mould
at four corners of the AOI to facilitate subsequent system
verifications.
2.2. Pattern material and application
The principle of Digital Image Correlation (DIC) technique is to
capture images of the AOI with a surface pattern of speckles (Fig. 2)
at a predetermined time interval. The deformation over the AOI is
then obtained through analysing the change in position of these
speckles by comparing the captured images. As a result, to enable
reliable capturing of required deformation in this study, the
applied surface pattern must:

stick to the concrete surface to ensure no relative displacement
between the speckles and the concrete;

have appropriate quality and contrast to ensure suitably good
images for subsequent image analysis; and,

dry as quickly as possible to facilitate the testing process.
Following a number of trials of different potential pattern
materials, RustOleum industrial fast drying spray paint was found
to satisfy all three desirable performance characteristics as men-
tioned above. During the application of paint, due care was taken
to ensure the paint was very thin and not continuous. The quality
of applied patterns was then evaluated both visually and analyt-
ically: Based on recommended criteria for the mean intensity
gradient [25] and subset entropy [26], the optimal subset size
was found to range between 40 and 60 pixels which was con-
sistently achieved for all test specimens. In addition, due to the
fast drying nature of the chosen paint [27] and testing started
within 10 min after spraying, the influence of paints chemical
substances on measured properties of concrete was deemed
negligible.
The application of chosen surface pattern as in Fig. 2 enabled
the deformation capturing by DIC of:

the AOI region of concrete surface, and

the four corners of the two steel mould halves.
565
2.3. Reliability of deformation captured by DIC
To assess the reliability of the deformation captured by DIC, the
measured deformation over the AOI region of concrete surface and
of the steel mould halves was compared with the input displace-
ment rate of 8.35  104 mm/s in Fig. 3.
Prior to the formation of major cracking, the average relative
displacement rate between two mould halves (Fig. 3b) and
between two AOIs edges (Fig. 3a) were much smaller than the
input displacement rate. This was also observed in previous studies
[16] and was possibly due to the combined effect of various factors,
including system slacks, distributed microcracking and slippage
between concrete and steel mould.
After breakage following major cracking, the relative displace-
ment between the two specimens halves should equal the applied
displacement. Indeed, the relevant displacement of the steel mould
halves (Fig. 3b) and of the AOI region of concrete surface (Fig. 3a),
obtained from DIC analysis, were 8.35  104 mm/s, which was
consistent with the input displacement rate. This clearly confirms
the reliability of deformation captured by DIC in this study for very
early-age concrete.
Accordingly, the improved test setup (Fig. 1) successfully
enables the reliable and simultaneous capturing of deformation
and loading of concrete specimens from very early ages. Basing
on the data collected using this setup, improved knowledge of
important early-age tensile properties of concrete can be obtained,
which are reported in the subsequent sections.
2.4. Concrete mix design and test procedure
A typical concrete mix design with a slump of 80 mm and a 28-
day characteristic compressive strength of 32 MPa was used
throughout the test series. Details of the mix design are given in
Table 1. The aggregates used were locally available Hornfels aggre-
gates with the particle size distributions given in Table 2. The
material batch sheet for cement, which was a commercially avail-
able general purpose Portland cement, is as shown in Table 3.
Significant effort was made to ensure the consistency of all
ingredients, mix design, curing and test conditions for all speci-
mens throughout the test duration: (i) All ingredients were stored
in sufficient quantity for the whole study; (ii) Moisture condition
of aggregates were regularly checked to allow adjustment of mix-
ing water accordingly; and (iii) The mixing, placement and curing
of concrete were kept the same as far as practically possible to
ensure similar mix quality.
All mix ingredients were stored in the laboratory condition for
at least 7 days prior to usage. The concrete was mixed in a 70-
litre rotating mixer and placed directly into the steel moulds,
which was then compacted via a vibrating table. Test specimens
566 D.H. Nguyen et al. / Construction and Building Materials 134 (2017) 563573

(a) Displace ment rate of AOIs edges. (b) Displacement rate of two mould halves.
Fig. 3. Verification of the reliability of deformation captured by DIC.

Table 1
Concrete mix design.

Ingredients Portland cement 20 mm aggregate 10 mm aggregate Coarse sand Fine sand Water reducer Water
(kg) (kg) (kg) (kg) (kg) (l) (l)
Quantity (/m3) 310 790 280 495 366 1.09 195

Table 2
Particle size distributions of aggregates.

Sieve size (mm) Cumulative passing (%)

20 mm aggregate 10 mm aggregate Coarse Sand Fine Sand
40 100 100 100 100
19.5 98.5 100 100 100
14 59 100 100 100
9.5 9.6 86.8 100 100
4.75 1.5 6.8 99.5 97.4
2.36 1.2 2 75.5 83.1
1.18 1.1 1.7 46.3 68.7
0.6 1 1.5 26.9 44.7
0.3 0.9 1.3 14.1 16.5
0.15 0.8 1.1 7.3 4.5
0.075 0.5 0.4 3.3 1
0 0 0 0 0

Table 3
Material batch sheet for Portland cement.

SiO2 Al2O3 Fe2O3 CaO MgO SO3 Loss on Ignition

19.7% 5.4% 2.8% 63.7% 1.2% 2.8% 3.2%
C3S C2S C3A C4AF Fineness Na2O equivalent
61.8% 9.9% 9.6% 8.5% 380 m2/kg 0.54%

were then covered by damp cloth to prevent moisture loss and
subsequently transferred to the test room with temperature and
humidity controlled at 20 2 C and 60 5%, respectively. Concrete
age was measured from the time all mixing water was added.
For each specimen, approximately 15 min before testing, the
damp cloth was removed, exposing the AOI for pattern application.
Flat RustOleum industrial fast drying spray paint was chosen for
the pattern material due to its fast drying capacity as well as neg-
ligible gloss: White paint was applied first, followed by black paint
after some waiting time for the white paint to dry out. The optimal
waiting time (approximately 510 min) was determined based on
the adhesive capacity of the paint to the concrete surface.
The specimen was then carefully placed and connected into test
position, followed by the installation of the DIC camera system.
The camera capturing frequency was set at 2 s. During testing,
the two sides of the test specimen were moved apart at a constant
rate of 8.35  104 mm/s. This rate is the same as that in [10], and
is an order of magnitude smaller than that of between 0.3 and
1.2 mm/min in [8] and 0.75 mm/min in [9]. The testing rate was
chosen to ensure the capturing of the post-peak tensile stress-
strain curves in most cases. A slower rate, though giving a better
resolution in the post-peak region, would lead to undesirably long
test duration for each specimen. For the applied displacement rate
of 0.05 mm/min, test duration for a specimen was typically
between 10 and 40 min for monotonic loading and 40 to 60 min
for cyclic loading. The nominal age used in subsequent sections
refers to the age of first loading during a test. Undoubtedly, due
to ongoing hydration, certain changes in concrete properties
 D.H. Nguyen et al. / Construction and Building Materials 134 (2017) 563573
during a test duration would occur and should be kept in mind
when interpreting test data. It should also be noted that any
shrinkage due to surface drying or hydration was included in the
deformation captured by DIC in the test setup.
3. Experimental results and discussion
Based on the data recorded by DIC and load cells, the strain can
be calculated over a gauge length of 70 mm of the AOI while the
stress can be determined based on the recorded tensile load and
the gross cross-sectional area of 70 mm by 100 mm of the middle
region (Fig. 2). Typical stress-strain curves obtained for concrete at
different ages from mixing are shown in Fig. 4: For concretes of
later ages, the reduced number of data points after peak stress
was mainly due to the faster crack propagation. A higher image
capturing frequency would provide improved measurement of
the post-peak behaviour.
From these stress-strain curves, a number of important con-
crete properties at very early ages can be derived. Such properties
include tensile strength, Youngs modulus, tensile strain capacity
and fracture energy. Importantly, the performance under cyclic
loading and tensile relaxation of early-age concrete in direct ten-
sion are also investigated.
3.1. Compressive strength at 28 days
The 28-day compressive strength was determined for concrete
of each batch in accordance with the relevant Australian Standards
[28,29] and plotted in Fig. 5. The 28-day compressive strengths
(a) Monotonic loading.
(b) Cyclic loading.
Fig. 4. Typical tensile stress-strain relationships.
567
were found to vary over only a small range of within 6% of the
mean value and have a standard deviation of 0.6 MPa, giving con-
fidence to the consistency of concrete quality across all batches.
3.2. Direct tensile strength
The development of the direct tensile strength with age from
mixing is plotted in Fig. 6a. It can be observed that the tensile
strength of the concrete under investigation was negligibly small
during the first 3 h or so but accelerated quickly in the following
few hours.
Since wet curing was provided for every specimen up until the
start of their tensile testing, the contribution of matric suction in
such increase in tensile strength should be negligible, if not zero.
The increase in measured tensile strengths was thus due almost
entirely to the ongoing hydration of cementitious material. The
rapid early-age strength growth also suggests that every additional
hour of effective curing of concrete that can be afforded on the con-
struction site is highly beneficial in ensuring sufficiently high ten-
sile capacity, thereby reducing the cracking risk.
The tensile strengths obtained in this study were plotted
together with results from other studies [8,10,17] in Fig. 6b. A sim-
ilar trend of tensile strength development with time is clearly evi-
denced, indicating a rather high consistency in the order of
magnitude of direct tensile strength of early-age concrete mea-
sured in different studies. The slight differences in reported tensile
strengths among studies are possibly due to the combined effects
of the following:

Mix designs and curing procedures used [8,17,30];

Effectiveness of measures to minimise friction between test
specimens and supporting surfaces, as briefed in Section 1;

Accuracy of load recording devices; and,

Assumed times of zero for determining the age of concrete:
While this would not affect the comparison of data within a
given study, such comparison among studies is slightly affected.
Fig. 6a also presents a comparison of the obtained direct tensile
strengths with corresponding values as given by relevant models in
currently available standards (Table 4). The significant deviation
between the models predictions and measured values (Figs. 6
and 7) clearly highlights the rather limited predictive capability
of these models. Such poor capability is due mainly to the current
dearth of reliable data on tensile properties of very early-age con-
crete, prompting the need to generate more reliable data as high-
lighted in Section 1 Introduction.
3.3. Youngs modulus
In this study, the Youngs modulus was determined by a linear
fit to the initial ascending part of stress-strain curves between 5%
and 40% of the tensile strength, following relevant Australian stan-
dards [33].
The obtained results are plotted in Fig. 7. Similar to the direct
tensile strength, the Youngs modulus was found to be negligibly
small during the first 5 h or so but then started to increase quickly
thereafter. There also exists significant deviation between the mea-
sured Youngs moduli and corresponding values predicted by dif-
ferent available models. Furthermore, the Youngs modulus
obtained in this study is considerably less scattered than that in
past research [10].
The above results are then plotted against comparable Youngs
modulus from a previous study [10] in Fig. 7b. In [10], similar con-
cretes were also tested for full stress-strain behaviour using the
same test frame. However, only the relative displacement between
the two steel mould halves (DLMould_Avg) was measured using two
568 D.H. Nguyen et al. / Construction and Building Materials 134 (2017) 563573

Fig. 5. Compressive strength at 28 days.

(a) This study (b) This study versus previous ones

Fig. 6. Tensile strength development with age from mixing.

Table 4
Models for prediction of tensile strength and Youngs modulus.

BS EN 1992-1-1:2004 [31] JSCE: 2007 [32]

Tensile strength (MPa): fctm(t) = bcc(t) fctm (Eq. (1)) For ordinary Portland cement at up to 3 days of age:
p p
Youngs modulus (MPa): Ecm(t) = [fcm(t)/fcm]0.3 Ecm (Eq. (2)) Tensile strength (MPa): ftk(t) = c f0 c(t) = 0.44 f0 c(t) (Eq. (4))
n h qio p p
(Eq. (3)) Youngs modulus (MPa): Ee(t) = U(t) 4.7  103 fc(t) = 3431 f0 c(t) (Eq. (5))
Age coefficient: bcc t exp s 1  28 t
s = 0.25 for class N cement (this study); where
s = 0.20 for class R cement; fc(t) concretes compressive strength at age t;
s = 0.38 for class S cement. t concrete age in days;
where c a coefficient varied with the degree of drying of concrete;
t concrete age in days; U(t) early creep factor, taken as 0.73 at up to 3 days of age.
fcm, fctm and Ecm 28-day compressive strength, tensile strength
and Youngs modulus;
fcm(t) = bcc(t) fcm concretes compressive strength at age t.

LVDTs attached to the steel mould halves. The required
deformation over the AOI region (DLAOI) was then estimated from
the measured deformation (DLMould_Avg). On the basis of finite
element modelling [10], the ratio DLAOI/DLMould_Avg was assumed
as:

0.50 before peak stress was reached, and

0.90 after peak stress.
To assess whether such estimation is acceptable, the ratio
DLAOI/DLMould_Avg was determined using reliable deformation
captured by DIC in this study (Section 2.3) and plotted in Fig. 8:
It can be observed that:

In the post-peak period following major cracking, the ratio
DLAOI/DLMould_Avg was fairly constant at around 0.95, which is
in rather good agreement with the assumption in [10];

Prior to the formation of major cracking, the ratio varied over a
large range initially before stabilising at around 0.2. This was
significantly different from the assumed value of 0.5 in [10].
This provides at least a partial explanation for the significantly
more scattered distribution of Youngs modulus in [10] com-
pared to that in this study.

DLAOI: is the relative displacement between the two extreme
edges of AOI region of concrete surface (line ).

DLMould_Avg: is the average relative displacement between the
two steel mould halves (line , being the average of lines
and ).
3.4. Relationship between direct tensile strength and Youngs modulus
The direct tensile strength versus Youngs modulus from this
study and available standards are plotted in Fig. 9. It can be
observed that:
 D.H. Nguyen et al. / Construction and Building Materials 134 (2017) 563573 569

(a) This study (b) This study versus previous ones

Fig. 7. Youngs modulus development with age from mixing.

(a) Relative displacements, L (b) LConcrete /LSteel mould

Fig. 8. Assessing the reliability of deformation reported in a typical previous study [10].

Fig. 9. Direct tensile strength versus Youngs modulus.

There was a very strong linear relationship between the direct
tensile strength and Youngs modulus for studied concretes dur-
ing the first 10 h after mixing. The ratio of Youngs modulus
(MPa) to direct tensile strength (kPa) appeared to be indepen-
dent of concrete age and was estimated by linear best fitting
to be a constant of 7.88. A similar observation was also reported
in [10].

Such linear relationship between direct tensile strength and
Youngs modulus of early-age concrete can be conveniently
used to determine either property if the other property is
known. Further work is needed to confirm such relationship
for other types of cement, mix designs and curing regimes.

The prediction by JSCE:2007 [32] based on the square root of
compressive strength at the respective age is in good agreement
with results of this study; while that given by BS EN 1992-1-1
[31] is not. Whereas this correct prediction of the linear rela-
tionship between the two properties by JSCE:2007 is acknowl-
edged, the need to improve its currently poor capability in
predicting individual early-age concrete properties (as evi-
denced in Figs. 6 and 7) remains.
570 D.H. Nguyen et al. / Construction and Building Materials 134 (2017) 563573

3.5. Strain at peak stress

The evolution of strain at peak stress over time recorded in this

research and major past studies [3,810,16] is presented in Fig. 10,
from which the following observations can be made:

The strain at peak stress appeared to decrease over time within

the test age range, and possibly reaching a minimum value at
around 89 h of age. This corresponds well with the general
trend reported in the literature [8,3438], which suggests an
initial sharp decrease of strain at peak stress from thousands
of microstrain before reaching a stable value of between
100 le and 200 le.

Compared to previous studies, the strain at peak stress in this
study appears to be consistent with a clear evolution over time,
which agrees well with Kasai et al. [8]. Test results from Dao
et al. [10], on the other hand, suffer large variation (Fig. 10b)
due to the difficulty in displacement recording as discussed in
Fig. 11. Fracture energy development with age from mixing.
Section 3.3. With limited test results, Hannant et al. [18] and
Roziere et al. [16] reported lower strain at peak stress, possibly
due to different mix designs and/or cement types employed.

The tensile strain capacity value of 70 le suggested in CIRIA
C660 [3] for crack control of early-age concrete seems appropri-
ate (Fig. 10b).

3.6. Fracture characteristics

Using the obtained test data, fracture properties can be calcu-

lated based on a stress-crack opening curve following the fictitious
crack model proposed by Hillerborg [39]. The obtained develop-
ment over time of fracture energy GF is plotted in Fig. 11:

The fracture energy GF was observed to increase with time at an

increasing rate over the observed period of 10 h after mixing.

There appears a good agreement between the time evolution of
GF in this study and in [10], despite the difference in methods to Fig. 12. Fracture energy versus tensile strength and Youngs modulus.

determine deformation (Section 3.3). This agreement was possi-

bly due to the governing influence of the tensile strength and
post-peak deformation, which were similar in both studies. ever, their relationships are only weakly linear as evidenced by
The effect of the variation in pre-peak deformation (Section 3.3) the relatively small R-square values of 0.82 and 0.9, compared to
was negligible. stronger linear relationships reported in [10].

The fracture energy GF is plotted against the corresponding ten-

sile strength ft and Youngs modulus E in Fig. 12, together with 3.7. Early-age concrete under cyclic loading
their best-fit linear trendlines. It can be seen that the fracture
energy increases with increasing values of tensile strength and The behaviour of concrete, both in its hardening and hardened
Youngs modulus, similar to observations reported in [10]. How- states, is dependent upon its load history. While the concrete

Fig. 10. Strain at peak stress development with age from this and major past studies.
 D.H. Nguyen et al. / Construction and Building Materials 134 (2017) 563573 571

performance under cyclic loading in its mature state is relatively

well-researched [40], our understanding of such performance at
very early ages is very limited. An improved knowledge of beha-
viour of hardening concrete under cyclic loading is therefore
required to provide a better basis for proper assessment of early-
age cracking risk.
Accordingly, using the improved direct tensile setup, four spec-
imens were subjected to cyclic loading to assess its effect on tensile
Youngs modulus of early-age concrete. For each load cycle, the test
specimen was subjected to a constant displacement rate of
0.05 mm/min up to about 40% of the tensile strength at that age,
estimated from tensile strength development over time of parallel
series of specimens (Fig. 7). A typical stress-strain curve in cyclic Fig. 14. Testing scheme for concrete relaxation under constant deformation.
loading is shown in Fig. 4b and a summary of the results is pre-
sented in Fig. 13.
Table 5
As evidenced in Fig. 13, when a single test specimen was subject Age, tensile strength and initial tensile stress of relaxation specimens.
to repeated tensile load cycles in this study, the Youngs modulus
in subsequent load cycles were found to be higher than those in Age (h:min) ft (kPa) r0 (kPa) r0/ft
preceding cycles, which is in stark contrast to the typical decreas- 5:05 39.8 17.2 0.43
ing trend reported for mature concrete [41,42]. Such difference is 6:40 122.2 52.4 0.43

possibly due to the combined effects of the following: Note: r0 applied stress at the beginning of relaxation test.
ft direct tensile strength obtained from parallel specimens.

On the one hand, due to the ongoing hydration during this early
stage, there were considerable increases in Youngs modulus
over the time intervals between load cycles, as evidenced in
Fig. 7. In contrast, for mature concrete, such ongoing hydration
is deemed negligible.

On the other hand, due to the microcracking and damage accu-
mulated over the load cycles, Youngs modulus was compro-
mised. The accumulated microcracking affects the properties
of both early-age and mature concretes.

3.8. Relaxation behaviour of early-age concrete

Adequate knowledge of creep and relaxation behaviour of early-

age concrete is required for the proper assessment of performance
of early-age concrete, including its cracking risk [12]. To trial the
capability of the current direct tensile setup for relaxation testing
and subsequently to complement currently available data
[12,43], in this part of the study, the variations over time of stresses
in two test specimens subject to constant deformation were inves-
tigated (Fig. 14). The imposed deformations were chosen to ensure Fig. 15. Tensile relaxation of very early-age concrete.
similar initial tensile stresses r0 in all specimens, which were
approximately 43% of the corresponding tensile strengths at start
test [44,45].) Test duration for each specimen was approximately
of the relaxation tests (Table 5). (The ratios of stress to tensile
60 min from the time t0 that r0 was reached.
strength would become increasing smaller during a relaxation test
The obtained relaxation behaviour is plotted in Fig. 15. For both
due to the increasing tensile strength over time as a result of ongo-
specimens, the stress in the concrete was found to initially
ing hydration but remain within the typical range for this type of
decrease over time at increasingly diminishing rates, which is con-
sistent with typical relaxation behaviour of concrete [44,45]. This
suggests the appropriateness of this test setup for the study of
relaxation behaviour of early-age concrete.
The stress in specimen 2, however, increased slightly after
about 10 min, due possibly mainly to the effect of ongoing hydra-
tion. Further work is ongoing to generate more data to form the
basis for improved understanding of such relaxation behaviour of
young concrete, taking due account of the effects of loading and
other mechanisms.

4. Summary and conclusions

This paper first highlights the current dearth of reliable data on

tensile properties of very early-age concrete, despite their signifi-
cance in assessing cracking risk and residual stress as well as in
precast industry. This is argued as a result of the considerable
Fig. 13. Youngs modulus of early-age concrete under cyclic loading. practical problems to be overcome when testing very early-age
572 D.H. Nguyen et al. / Construction and Building Materials 134 (2017) 563573
concrete due to its weak nature. Through identifying and effec-
tively addressing critical drawbacks in previous test setups, an
improved direct tensile testing system that can reliably capture
simultaneously stress and deformation of concrete from the first
several hours after mixing has been successfully developed and
reported. Such capturing has been achieved for concrete from the
age of 3 h in this study, but this starting age would be earlier or
later depending on the mix design and curing.
Two key features of this setup are:

Air-bearing box to effectively eliminate the friction between the
test specimen and supporting base.

Digital Image Correlation (DIC) to capture the required defor-
mation over the whole desired area of interest. The reliability
of DIC capturing of very early-age concrete is clearly demon-
strated through comparing with input displacement rate.
The improved test setup is then used to generate data on tensile
stress-strain curves to form the basis for improved knowledge of
such important tensile properties of early-age concrete as tensile
strength, Youngs modulus, fracture energy:

The overall development trends over time are found to be in
general agreement with limited available data from previous
studies. All tensile strength, Youngs modulus and fracture
energy are negligibly small during the first 4 h or so after mix-
ing but increase quickly afterwards. The highly linear relation-
ship between direct tensile strength and Youngs modulus is
particularly noted.

Importantly, the Youngs modulus obtained in this study is con-
siderably less scattered than that in past research. This is due
mainly to the more reliable capturing of deformation using
DIC in this study.
It is also found that the improved direct-tensile test setup can
be conveniently used to study the behaviour of early-age concrete
under direct-tensile cyclic loading and relaxation. Initial results
suggest that:

Different from its general decrease due to cyclic loading
observed in mature concrete, Youngs modulus of early-age con-
crete specimen in subsequent loading cycles is found to
increase but at a rate substantially lower than for the case of
no accumulated microcracking. This is possibly due to the com-
bined and counteracting effects of ongoing hydration of cemen-
titious materials and accumulation of micro-damage as a result
of cyclic loading.

The tensile relaxation of early-age concrete is generally consis-
tent with typical relaxation behaviour reported in literature.
However, further data is needed to better understand such
relaxation behaviour under the influence of loading, hydration
and other mechanisms.
With the reliability and capability of the improved direct tensile
testing setup clearly established, further work is ongoing to pro-
vide a better basis for rational design of concrete structures taking
due account of the effect of early-age performance characteristics.
Acknowledgements
Thanks are due to Renee Shi, Denton Liu, Qiule Qu, Guanwei
Zhang, Yang Xu, and Technical staff within the School of Civil
Engineering at the University of Queensland for their help with
the experimental work.
References
[1] A.K. Schindler, B.F. McCullough, The importance of concrete temperature
control during concrete pavement construction in hot weather conditions, J.
Transp. Res. Board 2002 (1813) 310.
[2] G. Bertagnoli, F. Tondolo, G. Mancini, Early age cracking of massive concrete
piers, Mag. Concr. Res. 63 (10) (2011) 723736.
[3] P.B. Bamforth, Early-Age Thermal Crack Control in Concrete (CIRIA C660),
CIRIA, London, UK, 2007.
[4] F. Benboudjema, J.M. Torrenti, Early-age behaviour of concrete nuclear
containments, Nucl. Eng. Des. 238 (10) (2008) 24952506.
[5] Cost Action TU1404, Towards the next generation of standards for service life
of cement-based materials and structures, 2016.
[6] H.D. Hernandez, Time-Dependent Pretress Losses in Pretensioned Concrete
Construction, University of Illinois at Urbana-Champaign, Ann Arbor, 1975.
185185.
[7] M. Levitt, Production Specific Processes, Precast Concrete, Spon Press, 1990.
[8] Y. Kasai, K. Yokoyama, I. Matsui, Tensile properties of early-age concrete, in:
Mechanical Behaviour of Materials Proceedings of the International
Conference on Mechanical Behaviour of Materials, The Society of Materials
Science, Kyoto, Japan, 1972, pp. 288299.
[9] D.J. Hannant, J. Branch, M. Mulheron, Equipment for tensile testing of fresh
concrete, Mag. Concr. Res. 51 (4) (1999) 263267.
[10] V.T.N. Dao, P.F. Dux, P.H. Morris, Tensile properties of early-age concrete, ACI
Mater. J. 106 (6) (2009) 483492.
[11] J. Byfors, Plain Concrete at Early Ages, Swedish Cement and Concrete Research
Institute, Stockholm, 1980.
[12] . Bjntegaard, T.A. Martius-Hammer, M. Krauss, H. Budelmann, RILEM
Technical Committee 195-DTD Recommendation for Test Methods for AD
and TD of Early Age Concrete: Round Robin Documentation Report: Program,
Test Results and Statistical Evaluation, Springer, Netherlands, Dordrecht, 2015.
[13] K. Raoufi, T. Their, W.J. Weiss, J. Olek, T. Nantung, Saw-Cutting Guidelines for
Concrete Pavements: Examining the Requirements for tIme and Depth of Saw-
Cutting, 2009, p. 813.
[14] L. Buffo-Lacarrire, S. Baron, F. Barr, D. Chauvel, A. Darquennes, J.P. Dubois, J.
Gayete, F. Grondin, B. Kolani, H. Lanon, A. Loukili, G. Moreau, C. Rospars, A.
Sellier, J.M. Torrenti, Restrained shrinkage of massive reinforced concrete
structures: results of the project CEOS.fr, Eur. J. Environ. Civil Eng. (2015) 124.
[15] D.M.F. Orr, G.F. Haigh, An apparatus for measuring the shrinkage
characteristics of plastic mortars, Mag. Concr. Res. 23 (74) (1971) 4348.
[16] E. Roziere, R. Cortas, A. Loukili, Tensile behaviour of early age concrete: new
methods of investigation, Cement Concr. Compos. 55 (2015) 153161.
[17] J. Abel, K. Hover, Effect of water/cement ratio on the early age tensile strength
of concrete, Transp. Res. Rec. 1610 (1998) 3338.
[18] D.J. Hannant, J. Branch, M. Mulheron, Equipment for tensile testing of fresh
concrete, Mag. Concr. Res. (1999) 263267.
[19] J. Branch, D.J. Hannant, M. Mulheron, Factors affecting the plastic shrinkage
cracking of high-strength concrete, Mag. Concr. Res. (2002) 347354.
[20] Z. Shan, Z. Yu, J. Shi, Experimental investigation of flow of fresh self-
compacting concrete in improved L-box, Constr. Build. Mater. 84 (2015) 30
38.
[21] A. Messan, P. Ienny, D. Nectoux, Free and restrained early-age shrinkage of
mortar: influence of glass fiber, cellulose ether and EVA (ethylene-vinyl
acetate), Cement Concr. Compos. 33 (3) (2011) 402410.
[22] D. Nguyen, V. Dao, A novel method for tensile testing of very early-age
concrete, in: S.T. Smith (Ed.), The 23rd Australasian Conference on the
Mechanics of Structures and Materials (ACMSM23), Byron Bay, Australia,
2014, pp. 4752.
[23] Z.P. Bazant, B.H. Oh, Crack band theory for fracture of concrete, Mater. Struct.
16 (93) (1983) 155177.
[24] B.L. Karihaloo, Fracture Mechanics and Structural Concrete, Longman, Harlow,
England, 1995.
[25] B. Pan, Z. Lu, H. Xie, Mean intensity gradient: an effective global parameter for
quality assessment of the speckle patterns used in digital image correlation,
Opt. Lasers Eng. 48 (4) (2010) 469477.
[26] S. Yaofeng, J.H.L. Pang, Study of optimal subset size in digital image correlation
of speckle pattern images, Opt. Lasers Eng. 45 (9) (2007) 967974.
[27] Technical Data Professional Enamel Sprays, in: R.-O. Corporation (Ed.) PRO-
01, USA, 2015.
[28] Standards Australia, AS 1012.8.1-2014: Method for making and curing
concrete Compression and indirect tensile test specimens, 2014.
[29] Standards Australia, AS 1012.9-1999: Determination of the compressive
strength of concrete specimens, 1999.
[30] G. De Schutter, Applicability of degree of hydration concept and maturity
method for thermo-visco-elastic behaviour of early age concrete, Cement
Concr. Compos. 26 (5) (2004) 437443.
[31] British. Standard, Eurocode 2: Design of Concrete Structures, Part 11: General
Rules and Rules for Buildings, British Standard, UK, 2004, p. 230.
[32] JSCE Concrete Committee, English Version of Standard Specification for
Concrete Structures, Design, Japan Society of Civil Engineers, Japan, 2007, p.
503.
[33] Standards Australia, Methods of testing concrete AS1012.17, Method 17:
Determination of the static chord modulus of elasticity and Poissons ratio of
concrete specimens, Australia, 1997.
 D.H. Nguyen et al. / Construction and Building Materials 134 (2017) 563573
[34] Comite Euro-International du Beton, Durable Concrete Structures, Telford,
Great Britain, 1992.
[35] S. Swaddiwudhipong, H.-R. Lu, T.-H. Wee, Direct tension test and tensile strain
capacity of concrete at early age, Cem. Concr. Res. 33 (12) (2003) 20772084.
[36] A.M. Neville, J.J. Brooks, Concrete Technology, Longman Scientific & Technical,
New York, Harlow, Essex, UK, 2007.
[37] Z.-H. Guo, X.-Q. Zhang, Investigation of complete stress-deformation curves for
concrete in tension, ACI Mater. J. 84 (4) (1987) 278285.
[38] W. Brameshuber, H.K. Hilsdorf, Development of strength and deformability of
very young concrete, in: Swartz Shah (Ed.), SEM/RILEM International
Conference on Fracture of Concrete and Rock, Springer-Verlag, Houston,
Texas, 1987, pp. 409421.
[39] A. Hillerborg, The theoretical basis of a method to determine the fracture
energy GF of concrete, Mater. Struct. 18 (4) (1985) 291296.
573
[40] B.Y. Bahn, C.-T.T. Hsu, Stress-strain behavior of concrete under cyclic loading,
ACI Mater. J. 95 (2) (1998) 178193.
[41] H.W. Reinhardt, H.A.W. Cornelissen, D.A. Hordijk, Tensile tests and failure
analysis of concrete, J. Struct. Eng. 112 (11) (1986) 24622477.
[42] F. Aslani, R. Jowkarmeimandi, Stress-strain model for concrete under cyclic
loading, Mag. Concr.Res. 64 (8) (2012) 673685.
[43] H. Morimoto, W. Koyanagi, Estimation of stress relaxation in concrete at early
ages, in: R. Springenschmid (Ed.), Thermal Cracking in Concrete at Early Ages
Proceedings of the International RILEM Symposium, E&FN Spon, Munich, 1994,
pp. 95102.
[44] A. Switek-Rey, E. Denari, E. Brhwiler, Early age creep and relaxation of
UHPFRC under low to high tensile stresses, Cem. Concr. Res. 83 (2016) 5769.
[45] Y. Wei, W. Hansen, Tensile creep behavior of concrete subject to constant
restraint at very early ages, J. Mater. Civil Eng. 25 (9) (2013) 1277.