materials

Article
Fracture Parameters of Cement Mortar with Different
Structural Dimensions Under the Direct Tension Test
Inkyu Rhee 1, * , Jun Seok Lee 2 and Young-Sook Roh 3, *
1 Department of Civil Engineering, Chonnam National University, Gwangju 61186, Korea
2 Institute of Bio-Housing, Chonnam National University, Gwangju 61186, Korea; jslee2080@jnu.ac.kr
3 School of Architecture, Seoul National University of Science and Technology, Seoul 01811, Korea
* Correspondence: rheei@jnu.ac.kr (I.R.); rohys@seoultech.ac.kr (Y.-S.R.)




Received: 29 April 2019; Accepted: 5 June 2019; Published: 7 June 2019


Abstract: In this paper, we measured the fracture properties of cement mortar—which is composed
of sand and has a nearly constant diameter—using a direct tension test. Four double-notched mortar
bar specimens with different structural dimensions were assessed. The failure load, load-crack mouth
opening displacement, and elongation of the gauge length were measured under direct displacement
control. The fractured surfaces were scanned and measured so that we could calculate the tensile
strength accurately and determine the fracture energy and characteristic length. The average ratio of
total fracture energy (GF ) to specific fracture energy (Gf ) was 1.94; this was lower than the typical
value for concrete, of 2.5. The direct tension test showed that the double-notched mortar specimens
had a smaller fracture processing zone after the initiation of tensile cracks, so the tail portion of
the softening branch was small. This decreased the GF /Gf ratio. We verified this result based on
a nonlinear fracture mechanics simulation and found that it agreed well with our experimental
results. We also investigated the size effects of four different scaled specimens while holding the
ratio of structural dimension, d, and notch length, a is constant, so that there was no shape effect.
The traditional linear elastic fracture mechanics (LEFM) prediction and Bažant’s size effect law yield
a gradient closer to 1/2 in the case of relatively large specimens. In the case of our cement mortar
specimens, this prediction was not supported, where the value of the slope was 1/0.727. This was
unexpected because LEFM predicts strong size effects. One possible explanation for this result is that
the size effects of concrete are most often evaluated using a bending test; also, concrete has a larger
maximum aggregate size than mortar. Due to the random heterogeneities in aggregate distribution,
higher tail energies may be seen for concrete, leading to differences in the GF /Gf ratio. At the same
time, the peak tensile stress could be affected by the relationship between structural dimensions and
aggregate size.

Keywords: cement mortar; fracture energy; fractured area; size effect; characteristic length

1. Introduction
A wide variety of experiments have been carried out to explore the fracture properties of concrete,
including brittle crack initiation and propagation [1–3]. The size effect on geometrically similar
structures can be measured in terms of the nominal strength of structure with the effect of the
characteristic structure size. In early stage, the classical Weibull theory was adopted for explaining a
statistical size effect caused by randomness of material strength. If the size of a quasi-brittle structures
becomes sufficiently large compared to material inhomogeneities, the structure becomes perfectly
brittle, and if the size becomes sufficiently small, the structure becomes non-brittle because the fracture
process zone (FPZ) extends over the whole cross section of the structure [2]. The basic size effect laws
that are power laws in terms of structural dimension. In linear elastic fracture mechanics (LEFM),

Materials 2019, 12, 1850; doi:10.3390/ma12111850 www.mdpi.com/journal/materials

Materials 2019, 12, 1850 2 of 15

this exponent was estimated as −1/2 when the geometrically similar structures with geometrically
similar cracks or notches were considered [2]. The law of deterministic size effect provided a way
of bridging two different powers applicable into two adjacent size range. One is strength limit and
the other is LEFM limit. According to crack band theory, the FPZ is influenced by the aggregate
content and maximum aggregate size [3–5]. The structure size at which this bridging transition occurs
represents a characteristic size. Extensive research works have been made for last four decades in order
to estimate this transition size effect [2]. Since Walsh [2] introduced the doubly logarithmic plot of
nominal strength versus size and observed this transitional zone, Bažant, Kim, and Carpenteri [3,6,7]
suggested the modified exponent of that of the LEFM theory. On the other hand, the formation of an
FPZ would have a significant influence on the total fracture energy (GF ) of concrete [5].
As a binding material, the fracture behavior of cement mortar may be of interest. It will enable us
to determine basic fracture properties by canceling out the inclusion effect. Numerous experimental
bending tests have been carried out [1–3,5,8–22] to assess the fracture energy and fracture toughness
of cement mortar and concrete. Cementitious materials exhibit a wide range of properties, from
relatively ductile to brittle, depending on the water-to-cement ratio, aggregate-to-cement ratio, curing
time, environmental condition (wet or dry condition of specimen), boundary condition of fixture,
eccentricity of loading [3,5,10,13–15]. When investigating the presence of size effect in the material
itself experimentally, it seems preferable to carry out uniaxial tension tests. For uniaxial tension tests,
this boundary influence has been extensively investigated for rectangular and cylindrical specimens of
different sizes using rotating boundary condition [10].
These was an effort to decouple a mix of material behavior and structural effects when fixed
boundary condition was applied. This wise experimental setup using rotating boundary condition
were also proven by a series of lattice analysis for test specimen [10]. The size effect of concrete and
sandstone was reported by van Vliet and van Mier [10,14,15]. The tests were carried out on specimens
of six different sizes in a scale range of 1:32. In the smallest specimens, failure stress was governed by
the aggregates. With increasing size of the specimen, the role of the statistical strength distribution
increased as well. The presence of flaws in the specimens determined the ultimate stress if stress/strain
gradients of a structural is not overwhelmed [10,14,15]. The basic hypothesis of the cohesive crack
model is that for mode I, an FPZ of finite width can be described by a fictitious line crack that transmits
normal stress, where this stress is a function of the separation, w. This hyper-elastic type of softening
first shows a very steep descending curve and then, at roughly 0.15~0.33 ft ’ , a gentler slope [4,5,9].
The tail of the descending curve is very long. This poses severe problems for the measurement of
GF , which corresponds to the area under the entire curve [5]. As this measurement is designed for
concrete material, this tail of the descending branch differs to that of mortar specimens due to the lack
of major inclusion.
When discussing scaling and size effects, it is important to consider that the fracture energy and
material strength imply the existence of a characteristic fracture length as a material property. The
expression of the characteristic length, lch , in the context of the FPZ, was introduced by Irwin [3,5].
The ratio of the GF and the Gf is approximately 2.5 for concrete, and crack bandwidth, wf , can be
defined as 14Gf /ft ’ or 5.6GF /ft ’ , and wo = (1/7)wf for the intercept of the initial tangent with axis of w [5].
These three fracture parameters, GF , Gf , and ft ’ (or lch , l1 , and ft ’ ) are sufficient to generate a cohesive
model of concrete material. Similar measurements for evaluating the fracture energy and lch have been
reported by many researchers [23–31].
In this paper, since the fracture parameter measure of cement mortar mostly have been done
under three-point bending test, we attempted to determine these three fracture parameters for cement
mortar under direct tension test. To this end, it is also important to measure the fracture surface
area when estimating tensile strength, rather than using the overall surface area of the specimen.
According to van Vliet et al. [3,15], the size effect on strength cannot be fully understood unless the
material composition and the specimen geometry, as well as the presence and magnitude of stress/strain
gradients is considered. In our case, the statistical aggregate distribution that causes stress/strain
Materials 2019, 12, 1850 3 of 15

gradients, material composition issue was relatively minimized by using cement mortar specimen
(material inhomogeneity only considered aside from the specimen shape, load eccentricity, and rotating
boundary condition of fixture devices). This may lead that material size effect could not be overruled
by these possible influences. Some experimental studies [32] have attempted to use the fracture
surface area to calculate fracture energy more precisely. This is essential to quantify the three fracture
parameters discussed above. The purpose of this paper is to devise a basic model of cement mortar
fractures that considers size effects and to assess fracture properties using uniaxial tensile tests. To this
end, we carried out direct tensile tests of cement mortar bar specimens with different structural sizes.
The load-crack mouth opening displacement (CMOD) was recorded, in addition to the elongation
of the gauge length, using linear variable differential transformers (LVDTs). We then assessed the
size effect, based on the tensile strength and fracture energy, in terms of the final CMOD at failure.
To accurately measure the fracture energy of the test specimens, each failure surface was scanned
using three-dimensional (3D) scanning equipment. This enabled us to quantify the lch , which can be
expressed as the surface energy on the cracked surface with respect to the elastic strain energy on the
crack band volume.

2. Uniaxial Tension Test of Size Effects

Each specimen was prepared with a 50% water-to-cement ratio and 53% cement-to-sand ratio. A
12 min mixing process was used: 7 min of dry-mixing, the addition of water over 4 min, and then
dynamic compaction for 1 min. The specimens were demolded after 1 day of curing under room
temperature and placed in water for 27 days. All specimens were tested using the direct tension setup
illustrated in Figure 1. The experiments were performed on a hydraulic testing machine with a capacity
of 50 kN. The channels for loading both ends of the specimen are made of hardened steel and have
polished surfaces. A double-notched tension bar was used to measure the failure load associated with
the CMOD. The purpose of this tension test was to quantify the fracture energy (mode I fracture).
We also investigated the size dependency of the scaled specimens (T-10, T-20, T-30, and T-40). The
ratio of initial notched length to the maximum dimension of a specimen, a/d (d = w2 ), was set to 0.25,
as indicated in Table 1. Four group specimens, with ligament areas of 10, 20, 30, and 40 mm2 , were
cast. Each specimen group contained nine sub-specimens. Two LVDTs and two clip gauges (range:
±2.5 mm) were installed to measure the total displacement of the bar and the CMOD at the notched
length. The axial deformation referred to in this paper is the average value of the two LVDTs. The
CMOD values on either side of the notched part of the specimen showed significant bias when cracks
developed in the notched sections, probably due to the instant rotation of the specimen when the crack
developed. Therefore, the CMOD referred to in this paper represents the superior result between the
two CMODs measured. The average compressive strengths of specimens T-10, T-20, T-30, and T-40
were 66.6, 55.3, 52.4, and 56.7 MPa, respectively, after 28 days of curing. Each specimen was in the
form of a 50-mm cube. Failure stress-CMOD plots are depicted in Figure 2 for the T-series specimens.
The stresses were calculated based on the initial test area, given by Ao = w2 ·b.
 Materials 2018, 12, x FOR PEER REVIEW 4 of 15
Materials 2019, 12, 1850 4 of 15
Materials 2018, 12, x FOR PEER REVIEW 4 of 15
w0 b

w0 b
Channel 1: Load cell
h1
Channel 1: Load cell
h1
h2 w1 Channel 3: LVDT( back)
Channel 2: LVDT( front)

h2h3 w1 Channel 3: LVDT( back)

Channel 2: LVDT( front)

h0
h3 Channel 5: clip gage
h4 w2 ( Right)
h0
Channel 5: clip gage
h4 w2 ( Right)

Channel 4: Clip gage

( Left)

Channel 4: Clip gage

( Left)

(a) (b)
(a)Generic geometry of the specimens(b)
Figure 1. (a) and (b) test apparatus.

Figure
Figure1. (a)Generic
1. (a) Genericgeometry
geometryof
ofthe
thespecimens
specimensand
and(b)
(b)test
testapparatus.
apparatus.
Table 1. Specimen dimensions in mm (Note: structural dimension, d = w2).
Table 1. Specimen dimensions in mm (Note: structural dimension, d = w ).
Specimen Table 1. Specimen dimensions in mm (Note: structural dimension, d = w22).
h0 h1 h2 h3 h4 w0 w1 w2 a b Ao (mm2)
label Specimen Label h0 h1 h2 h3 h4 w0 w1 w2 a b Ao (mm2 )
Specimen
T-10 h
165.0
T-10
0 h42.5
1165.0 h
20.0
42.5
2 h
17.5
20.03 17.5h5.0
4 5.0 w30.0
30.0
0 w 1 10.0w
15.0
15.0 10.0
2 2.5 a2.5
10.0 b 100 Ao (mm
10.0 100 2)
label
T-20 330.0
T-20 85.0 40.0 40.0
330.0 85.0 35.035.0 10.010.0 60.0
60.0 30.020.0 20.0
30.0 5.0 5.0
20.0 20.0
400 400
T-10 165.0 42.5 20.0 17.5 5.0 30.0 15.0 10.0 2.5 10.0 100
T-30 495.0
T-30 127.5 60.0 60.0
495.0 127.5 52.552.5 15.015.0 90.0
90.0 45.030.0 30.0
45.0 7.5 7.5
30.0 30.0
900 900
T-20 330.0 85.0 40.0 35.0 10.0 60.0 30.0 20.0 5.0 20.0 400
T-40 T-40
660.0 660.0 170.0 80.0
170.0 80.0 70.0 20.0 70.0 20.0 120.0 60.0 40.0 10.0
120.0 60.0 40.0 10.0 40.0 40.0 1600 1,600
T-30 495.0 127.5 60.0 52.5 15.0 90.0 45.0 30.0 7.5 30.0 900
T-40 660.0 170.0 80.0 70.0 20.0 120.0 60.0 40.0 10.0 40.0 1,600

Figure2.2.Plot
Figure of the
Plot Piola-Krichhoff
of the (PK) stress
Piola-Krichhoff versusversus
(PK) stress the crack
themouth
crackopening
mouth displacement (CMOD).
opening displacement
(CMOD).
3. Uniaxial
Figure 2.Cauchy
Plot ofStress According to(PK)
the Piola-Krichhoff 3D Image Analysis
stress versus the crack mouth opening displacement
(CMOD).
As the fractured
3. Uniaxial area, A
Cauchy Stress n , differs from
According to 3Dthe initial
Image test area, Ao , for each fractured specimen, 3D
Analysis
image analyses were carried out to measure the actual An . A light-emitting diode (LED)-type remote
3. Uniaxial
As theCauchy Stress
fractured According
area, A n, differstofrom
3D Image Analysis
the initial test area, A o, for each fractured specimen, 3D
3D scanner (smartSCAN; AICON, 2016, Breuckmann, Germany) based on the miniaturized projection
image analyses were carried out to measure the actual An. A light-emitting diode (LED)-type remote
As thewas
technique fractured area, three
used with A n, differs from
different the initialangles.
projection test area,
TheAAo, for
n ofeach
eachfractured specimen,was
T-series specimen 3D
3D scanner (smartSCAN; AICON, 2016, Breuckmann, Germany) based on the miniaturized
image analyses
measured were carried
on a 5.0-M outdigital
resolution to measure
imagethe actual
with An. A light-emitting
a minimum precision of 7diode (LED)-type
µm. The remote
actual fractured
projection technique was used with three different projection angles. The A n of each T-series
3D scanner
surfaces were(smartSCAN;
larger than the AICON, 2016, Breuckmann,
initial section, so the fracture Germany)
stresses were based on the
calculated miniaturized
according to the
specimen was measured on a 5.0-M resolution digital image with a minimum precision of 7 μm. The
projection technique was used with three different projection angles. The A
Cauchy stress principle, as indicated in Figure 3. It is important to quantify the mode I fracture
n of each T-series
actual fractured surfaces were larger than the initial section, so the fracture stresses were calculated
specimen was on
energy based measured on a 5.0-M
experimental resolution
test data. Smallerdigital imageshow
specimens withrelatively
a minimum precision
larger of 7 μm.
differences The
between
according to the Cauchy stress principle, as indicated in Figure 3. It is important to quantify the
actual fractured
fractured surfaces
and initial wereas
sections, larger than the
indicated initial3e,f.
in Figure section, so the fracture stresses were calculated
according to the Cauchy stress principle, as indicated in Figure 3. It is important to quantify the
Materials 2018, 12, x FOR PEER REVIEW 5 of 15

mode I2019,
Materials fracture energy
12, 1850 based on experimental test data. Smaller specimens show relatively larger
5 of 15
differences between fractured and initial sections, as indicated in Figure 3e,f.

(a) (b)

(c) (d)
10000 T10 (10mm x 10mm) 0.6
T20 (20mm x 20mm)
Fractured Area (mm2), Log-scaled

T30 (30mm x 30mm)

Deviation Rato from An

T40 (40mm x 40mm) 0.5

T10 (scanned area)
T20 (scanned area)
T30 (scanned area) 0.4
T40 (scanned area)

1000 0.3

0.2

0.1

0
100
T10 T20 T30 T40
0 2 4 6 8 10
Specimen Size (mm)
No. of Sample
(e) (f)

n =
Figure3.3. Scanned
Figure Scannedfractured
fractured sections.
sections. (a–d)
(a)–(d) Scanned
Scanned images;
images; (a).T-10
(a). T-10(A(A n =144.67
144.67mm
mm 2 );
2); (b). T-20
(b). T-20
(Ann == 554.05 mm ); (c). T-30
(A 2 T-30 (Ann == 1158.09 2 T-40 (Ann ==2,217.54
1158.09 mm );); (d). T-40 2
2,217.54 mm );); (e)
2
(e) measured fractured
measured fractured
surfaceareas
surface areasby
byspecimen
specimensize;
size;(f)
(f)deviation
deviationratios
ratiosbetween
betweenfractured
fracturedand andinitial
initialsections.
sections.

4.4. Size
Size Effect
Effectof
ofMortar
MortarSpecimens
Specimensunder
UnderUniaxial
UniaxialTension
Tension
The
Thefracture
fracturestress-CMOD
stress-CMODgraphgraphobtained
obtainedbased
based ononthetheinitial
initialsection
sectionininFigure
Figure22waswas replotted
replotted
in
inFigure
Figure44forforaafractured
fracturedsection.
section. The
The peak
peak values
values ofof the
the fractured
fractured surfaces
surfaces were
were adjusted
adjusted to to yield
yield
lower
lower values. The gauge lengths of specimens T-10, T-20, T-30, and T-40 were 35, 70, 105, andmm,
values. The gauge lengths of specimens T-10, T-20, T-30, and T-40 were 35, 70, 105, and 140 140
respectively. The tensile
mm, respectively. The strength-CMOD relationships,
tensile strength-CMOD based on the
relationships, (a) on
based clipthe
gages
(a) and
clip(b) LVDTs
gages andwith
(b)
gauge length L , are plotted in Figure 4. The results were similar, indicating that
LVDTs with gauge length Lg, are plotted in Figure 4. The results were similar, indicating that the
g the FPZ was located
nearby
FPZ was to the notched
located zone.
nearby to The peak strain
the notched ranged
zone. from strain
The peak 10−5 to 8.81
4.18 × ranged 10−5×. 10
from×4.18 The failure
−5 to 8.81 strain
× 10−5.
finally saturated −3
× 10 . The CMODs
The failure strainatfinally
approximately
saturated1.25
at approximately 1.25 × 10 at−3failure were 0.0333,
. The CMODs 0.091,were
at failure 0.145, and
0.0333,
0.245 mm for specimens T-10, T-20, T-30, and T-40, respectively.
0.091, 0.145, and 0.245 mm for specimens T-10, T-20, T-30, and T-40, respectively.
Materials 2018, 12, x FOR PEER REVIEW 6 of 15

Materials 2019, 12, 1850 6 of 15

Materials 2018, 12, x FOR PEER REVIEW 6 of 15

(a) (b)
Figure 4. Direct tensile strength-CMOD relationship based on the (a) clip gages and (b) linear
variable differential transformers (LVDTs).

(a) (b)
The experimental results for the peak Piolar-Kirchhoff (PK) stresses and Cauchy stresses, for
different
Figure 4.4. Direct
Directtensile
structural tensile strength-CMOD
dimensions, relationship
d (fromrelationship
strength-CMOD 10–40 mm), based
are
based shown
on on(a)
the the (a)gages
inclip
Figureclip5a,b.
gages
and Toand
(b) (b)
measure
linear linear
the size
variable
variable
[4,5,9],differential
effectdifferential
we transformers
transformers
linearized (LVDTs).
Equation(LVDTs).
(1) so that the projected reference strength, f t *  Bf t ' could be
the for
2
form of f t  (PK)
*
The experimental
evaluated by intersecting results N
the peak
-axis Piolar-Kirchhoff
in the 1/ Cstresses and Cauchy
. Furthermore, stresses, for
the characteristic
different
measure structural
of specimen size, o can
dimensions, d be
(from 10–40 mm),
calculated areThe
as C/A. shown
resultsin Figure 5a,b. To
of this linear fittingmeasure
of thethePK size
and
*∗ = B f '0 could be
effect [4,5,9], we linearized Equation (1) so that
effect [4,5,9], we linearized Equation (1) so that the projected the projected
'
reference
reference strength,
strength, f ft 
√ and  = 7.55t mm fort PK stress;Bf t could be
Cauchy stress data are plottedthe σin Figure 5c,d, where Bf ∗= 2.24 Mpa
−2
evaluated by intersecting N 2 -axis in the form of ftt *= 1/ C. Furthermore, o the characteristic
evaluated
Bft ' = 3.16
measure ofby
Mpaintersecting
and 
specimen theλo can
size, -axis
be in retrieved.
the form
calculated ofThe
as C/A. ftThe  1/results
C . stress
Furthermore,
of thisvalue
linearfor the characteristic
fitting ofgroup
the PK
o = 5.24 mm were minimum each of
N

and
measureCauchyof stress
specimen data are
size, plotted
can be in Figure
calculated 5c,d,
as where
C/A. The B f 0 = 2.24 Mpa and λ = 7.55 mm for PK
results
specimens was used in the calculations of these parameters based on experimental data. The
o t of this linear fitting
o of the PK and
stress; B ft0 =and
3.16average
Mpa and λo =data
5.24tend
mm were retrieved. TheCminimum stress valuerational
for each group of
maximum
Cauchy stress data stress
are plotted in Figure to have
5c,d, where Bft ' = 2.24
a negative value, which
Mpa and ois=not
7.55 mm forin PKthe case
stress;
specimens
of tensile was
stress.used in the calculations of these parameters based on experimental data. The maximum
Bft ' average
and = 3.16 Mpa datatend
stressand o = 5.24 mma were
to have negativeretrieved.
C value,The
which minimum stress in
is not rational value for of
the case each group
tensile of
stress.
1 1
 N −22  1 of2 these 1parameters
specimens was used in the calculations   C based
 A on experimental data. The
    (1)
2
t0 2 o
σN tend
maximum and average stress data = Bf '
to have
 + aBf '
negative λC
= value,
C + Aλwhich is not rational in the case
(1)
t0 2

of tensile stress. B ft B ft λo

1 1
 N 2     C  A
 Bft 
' 2
 Bft '  o 6 (1)
2
6

5 5
Stress (MPa)
PK Stress (MPa)

64 64

53 53
(MPa)
Cauchy
PK Stress (MPa)

42 42
Cauchy Stress

31 31

20 20
0 10 20 30 40 50 0 10 20 30 40 50
1 Specimen size, d (mm) 1 Specimen size, d (mm)

0 (a) 0 (b)
0 10 20 30 40 50 0 10 20 30 40 50
Figure 5. Cont.
Specimen size, d (mm) Specimen size, d (mm)
(a) (b)
 Materials 2019, 12, 1850 7 of 15
Materials2018,
Materials 2018,12,
12,xxFOR
FORPEER
PEERREVIEW
REVIEW 77of
of15
15

22 1.5
1.5
yy==0.0265x
0.0265x++0.2
0.2

(MPa-1/2))
(MPa-1/2))

R²==0.8521
0.8521
1.5 R²

-1/2
1.5
-1/2

11

-1/2 (MPa
-1/2 (MPa

11
0.5

σσNN-1/2
σσNN-1/2

0.5
0.5
0.5
yy==0.0191x
0.0191x++0.1
0.1
R²==0.7218
0.7218
00 00 R²
00 10
10 20
20 30
30 40
40 50
50 00 10
10 20
20 30
30 40
40 50
50
Specimensize,
Specimen size,dd(mm)
(mm) Specimensize,
Specimen size,dd(mm)
(mm)

(c)
(c) (d)
(d)
Figure
Figure5.5.5.Peak
Figure Peak stresses
Peakstresses versus
stressesversus specimen
versusspecimen size:
specimensize: (a)
size:(a) PK,
(a)PK, and
PK,and (b)
and(b) Cauchy
(b)Cauchy stresses,
Cauchystresses, and
stresses,and linearized
andlinearized stresses
linearizedstresses
stresses
versus
versus different
different specimen
specimen sizes:
sizes: (c)
(c)PK,
PK,and
and (d)
(d) Cauchy
Cauchy
versus different specimen sizes: (c) PK, and (d) Cauchy stresses. stresses.
stresses.

B ft0Bf '' λ  , a log-scale

t t and oo , a double log-scale plot of the
Afterdetermining
After
After determining the
determining theempirical
the empirical
empirical parameters,
parameters,
parameters, and
Bf o , a double
and plot of theplot
double log-scale strength
of theby
structural
strength bysize was generated,
structural wasas
sizewas shown inas
generated, Figure
shown 6. inIn Figure
the case6.ofInthe theCauchy
caseof stress,
ofthe the gradient
theCauchy
Cauchy was
stress,the
the
strength by structural size generated, as shown in Figure 6. In the case stress,
slightly less
gradient was than 1:2
was slightly based
slightly less on
less than linear
than 1:2 elastic
1:2 based
based on fracture
on linear mechanics
linear elastic
elastic fracture(LEFM) for
fracture mechanics the 10–30-mm
mechanics (LEFM)
(LEFM) for specimens.
for the
the 10–
10–
gradient
However, an
30-mmspecimens. abruptly
specimens.However, descending
However,an branch
anabruptly was
abruptlydescending observed
descendingbranch in
branchwas the case
wasobservedof the
observedin 40-mm
inthe
thecasespecimen
caseof ofthe group.
the40-mm
40-mm
30-mm
Thus, a significant
specimen group. dropa in
Thus, strength occurred
significant drop in in the case
strength of the 40-mm
occurred in the specimens;
case of the this can
40-mm be seen in
specimens;
specimen group. Thus, a significant drop in strength occurred in the case of the 40-mm specimens;
Figure
this can5c,d. Thein
be seen
seen reduction was extremely
Figure 5c,d.
5c,d. largewas
The reduction
reduction in the context of
extremely linear
large fracture
in the mechanics,
the context
context linearwhere
of linear the
fracture
this can be in Figure The was extremely large in of fracture
gradient decreases
mechanics, wherethe proportionally
the gradientdecreasesto the square
decreases root.
proportionally tothe
thesquare
squareroot.
root.
mechanics, where gradient proportionally to

StrengthCriteria
Strength Criteria
00
00 0.2
0.2 0.4
0.4 0.6
0.6 0.8
0.8 11
-0.1
-0.1

22
/Bft t)')

-0.2
-0.2
'

11
(σNN/Bf

-0.3
-0.3
Log(σ
Log

CauchyStress
Stress LEFM
LEFM
-0.4 Cauchy
-0.4
PKStress
PK Stress

-0.5
-0.5

-0.6
-0.6
(λ/λoo))
Log(λ/λ
Log
Figure 6. Double log-scale plot of the size effects.
Figure6.6.Double
Figure Doublelog-scale
log-scaleplot
plotof
ofthe
thesize
sizeeffects.
effects.
Based on the work of Bažant [5,9], a linear softening branch was used to formulate the GF . Herein,
Based
a bilinear
Based on on the
the work
softening work of Bažant
branch
of Bažant [5,9],
was used byaamodifying
[5,9], linear softening
linear softening branch
the area was used
underneath
branch was used to
theto formulate the
stress-strain
formulate the GGFFin
curve ..
Herein,
Figure 7.
Herein, a bilinear softening
Thus, GFsoftening
a bilinear branch
can be rewritten was used
as follows:
branch was by modifying the area underneath the stress-strain
used by modifying the area underneath the stress-strain
curve in Figure 7. Thus, G F can be rewritten as follows:
curve in Figure 7. Thus, GF can be rewritten as follows:
 N
ft '
1 k 1

Materials 2019, 12, 1850 Et1 8 of 15

Materials 2018, 12, x FOR PEER REVIEW wc  nd a 8 of 15
 ft '
a a N Ec
Et2
'
ft 0
1 k
p i f
1

(a) Et1 (b)

wc  nd a
Figure 7. Fracture process. (a) Fracture processzone
ft ' (FPZ) and (b) bilinear softening model for
mortar. Ec
a a Et2

0 as Equations (2) and (3), and are used to obtain

The equations of Bažant [4,5] can be rewritten
 p i f
the elastic stored energy in the FPZ shown in Figure 7b:
(a) (b)
(a)
WFracture (FPZ)  (b)(b)
 Nand bilinear
Wbilinearsoftening 2
Figure 7. Fracture
Figure7. Fractureprocess.
process.(a)  2(2process
Fracture
1a  nd a )b 
zone
kprocess zone(FPZ) and
,  GFsoftening
b model
modelforfor
mortar.
(2)
mortar. a  c
2 E  a
The equations of Bažant [4,5] can be rewritten as Equations (2) and (3), and are used to obtain the
elasticThe stored energyof
equations inBažant
the FPZ[4,5]
shown
can in  Figure
be 7b: asEEquations
rewritten 2 Ec  f t
(2) and'2
(3), and are used to obtain
the elastic stored energy in the FPZ G  w
F shown 1  (1 
c  in Figure  2
)
 7b:
c
   (3)
 E
σ 2 E  2 E

∂W t 1  ∂W t 2 c
= 2(2k1 a + nda )b N2 ,
 
= GF b (2)
 2E
 c  ∂a W

W∂a
where, k1 and n are the empirical  2(2constants
k1a  nd aderived
)b  N in  GF b According to Bažant (2)
 ,the experiments. [4,9],
a  2 Ec  !a 02
k1 is close to 1 and n is 3~5, depending on theEmaterial c Ecused ft (3 for concrete; 5 for rocks and
GF = wc 1 − (1 − α2 ) − α2 (3)
ceramics). a and d a are the notch length  and maximum
2 Ec
Et1
2 Ec  f t
Egrain
t2 2Esize,
c
'2 respectively. In our case, a
GF  wc 1  (1   )   theAccording (3)
variedk1among
where, and n are the the
specimens
empirical listed in Table
constants 1. d was experiments. 0.5 mm Et 2 in case of thetosand particles
[4,9], kused
 derivedainEthe t1  2 Ec Bažant 1 is
close to 1 and n is 3~5, depending on the material used (3 for concrete;
in the experiment. b is the thickness of the specimen, as listed in Table 1. W is the work done by the 5 for rocks and ceramics). a
where, k
and da are1 the notchand n length and maximum grain size, respectively. In our case, a varied among the'
are the empirical constants derived in the experiments. According to Bažant [4,9],
FPZ, which is shown in Figure 7b.  N is the estimated tensile stress under the size effect and f t is
specimens listed
k1 is close to 1 and in Table n 1.isda3~5,
wasdepending
0.5 mm in the on case
the of the sand
material particles
used (3 forused in the experiment.
concrete; 5 for rocks b and is
thethethickness
referenceoftensile
the specimen, c , Et 1in
strength.asElisted TableE1.
, and t2 W areisthetheinitial
work doneelasticbymodulus
the FPZ,and firstisand
which shownsecond
in
ceramics). a and d a are the notch length and maximum grain size, respectively. In our case, a
descending
Figure 7b. σN softening branches,
is the estimated respectively,
tensile stress under and thesize is theeffect and ft0 retention
strength factortensile
is the reference at the strength.
inflection
Evaried
point
c , Et1 , among
in
and theE t2
the
are specimens
descending
the initial listed
branch.
elasticwin Table
is
modulus the1. d
crack
and a was
band,
first 0.5
and mm
which in
second the
can case
also
descending of
be the sand
expressed particles
softening as nd aused
. The
branches,
c
and α is W isdescending
point, 
respectively, wc
ininflection
the experiment. fthe
b is strength
' the thicknessretention factor at the
of the specimen, asinflection point 1.
listed' in Table
t in Figure 7a ranges from 0.22–0.6 f t depending on the
in the the specimen branch.
work done by the
size; the
FPZ, which is shown in Figure 7b. ' N is the estimated tensile stress under thet size effect and f t is
is the crack band, which can also be expressed as nd a . The inflection point, α f 0 in Figure 7a ranges '
average
from experimental
0.22–0.6 ft0 depending valueonisthe f t (Figure
0.4specimen 8).the average experimental value is 0.4 f 0 (Figure 8).
size; t
the reference tensile strength. Ec , Et 1 , and Et 2 are the initial elastic modulus and first and second
descending softening branches, 3 respectively, and  is the strength retention factor at the inflection
point in the descending branch. wc is the crack band, which canT10also be expressed as nd a . The
2.5 T20
inflection point,  f t in Figure 7a ranges from 0.22–0.6 f t depending
Tensile Strength (MPa)

' '
T30 on the specimen size; the
average experimental value is20.4 f t ' (Figure 8). T40

1.5
3
T10
1 T20
2.5
Tensile Strength (MPa)

T30
0.5
2 T40

0
1.5
0 0.05 0.1 0.15 0.2
CMOD* (mm)
1
Figure 8. Simplified stress-strain relationship based on the average tensile stress-inelastic CMOD for
0.5energy (Gf ) and total fracture energy (GF ).
measuring specific fracture

0
0 0.05 0.1 0.15 0.2
CMOD* (mm)
 Figure 8. Simplified stress-strain relationship based on the average tensile stress-inelastic CMOD for
measuring specific fracture energy (Gf) and total fracture energy (GF).

The basic expression of B in Equation (1) can be rewritten as B* using Equations (4) and (5): 9 of 15
Materials 2019, 12, 1850
'
E E ft
B  1  (1   ) c   2 c ,
* 2
ft *  ,  N  B ft
* *2
The basic expression of B in Equation
Et1 (1)
Et 2 can be rewritten as 
B* using Equations (4) and (5):(4)
1
r
Ec Ec ft0 o
B∗ = 1 − ( 1 − α2 ) − α2 , ft∗ = q , σN = B∗ ft∗2 (4)
2 Et1 Et2 λ
 ft '
 1  1+ λo
   *2  *2 , Y  C  A   (5)
N f 0 !2 B 1 B o (λ )
t
= + , Y = C + A(λ)λ (5)
1 B∗2 σ
1N B∗2 λo (λ)
where, b  *2 , a ( )  *2 , O     91.354  1.9073  0.0109 , B =3.29
2 *

where, b = B1
B∗2
, a(λ) = B∗2 λB (λ) O, λO
1 (λ) = 91.354 − 1.9073λ + 0.0109λ , B = 3.29.
2 ∗
O
earlier, λ constant. Instead, we used the function λ to fit the experimental
AsAsmentioned o is not constant. Instead, we used the function  to fit the
mentioned earlier, o is not
test results more accurately. Figure 9 and Table 2 show the inverse analysis results obtained using the
experimental test results more accurately. Figure 9 and Table 2 show the inverse analysis results
size effect law defined in Equations (4) and (5), when the reference tensile strength was set to that of
obtained using the size effect law defined in Equations (4) and (5), when the reference tensile
specimen T-40. The results are acceptable, being that they are between those given by the size effect
strength was set to that of specimen T-40. The results are acceptable, being that they are between
law and the experimental results. However, they violated the linearized relationship of Equation (1)
those given by the size effect law and the experimental results. However, they violated the linearized
and a constant value of λo .
relationship of Equation (1) and a constant value of o .
35 1.2
Exp data Exp data
30 1 Proposed Eq.

25
0.8
(f't/σN)2

20
0.6
λo(λ)

15
0.4
10

5 y = 0.0109x2 - 1.9073x + 91.354 0.2

R² = 1
0 0
20 40 60 80 100 0 20 40 60 80 100
λ=d/da λ=d/da
(a) (b)
Figure
Figure9.9.(a)(a)
The
The material characteristic
material length,
characteristic λoλwith
length, respect
o with to to
respect thethe
structural size,
structural λ; λ;
size, (b)(b)
the size
the size
effect
effect whenλoλ(λ).
when o (λ).

Table 2. Strength variation by specimen.

Table 2. Strength variation by specimen.
Specimen Label
Specimen label Avg.Avg. Cauchy Stress (MPa)
Cauchy Estimated Tensile Stress (MPa)
Stress (MPa) Estimated Tensile Stress (MPa)
T-10
T-10 2.577
2.577 2.327
2.327
T-20
T-20 1.591
1.591 1.583
1.583
T-30
T-30 1.088
1.088 1.085
1.085
T-40
T-40 0.719
0.719 0.723
0.723

Bažant [4,5] described the size effect in terms of the member size, maximum aggregate size, FPZ,
Bažant [4,5] described the size effect in terms of the member size, maximum aggregate size, FPZ,
and crack width, wc, by first defining k1 and n. The size reduction rate depends on the 1:2 slope of the
and crack width, wc , by first defining k1 and n. The size reduction rate depends on the 1:2 slope of
size radical, as given by Equation (1). As this law was intended for use in models of concrete and
the size radical, as given by Equation (1). As this law was intended for use in models of concrete
rock, it must be modified empirically for application to mortar. As shown in Figure 10, failure
and rock, it must be modified empirically for application to mortar. As shown in Figure 10, failure
strength decreased with a gradient of ½ when n = 3, k1 = 1 and m = 2. However, there are
strength decreased with a gradient of 12 when n = 3, k1 = 1 and m = 2. However, there are discrepancies
between this strength reduction and that shown by the experimental test data for the mortar specimen,
where the slope value was 1/0.727. As mortar contains sand grains with a maximum size of 0.5 mm, wc
would be 1.5 mm if n = 3, as proposed by Bažant and Oh [8]. Then, the constants m and k1 have to be
 mortar specimen, where the slope value was 1/0.727. As mortar contains sand grains with a
maximum size of 0.5 mm, wc would be 1.5 mm if n = 3, as proposed by Bažant and Oh [8]. Then, the
constants m and k1 have to be changed from 2 and 1 to 0.7 and 0.273, respectively, in accordance with
the measured fracture energy (mode I fracture), as illustrated in Figure 7a. Here, o can be defined
as o   n / 2k1  d / a  .
Materials 2019, 12, 1850 10 of 15

 N  in
changed from 2 and 1 to 0.7 and 0.273, respectively, B accordance
ft *
(6)
with the measured fracture energy
(mode I fracture), as illustrated in Figure 7a. Here, λo can be defined as λo = (n/2k1 )(d/a).
 n  d  Ec
ft ' o     σ B=B · 1f ∗
ft *   2k1   a  N t  Et (6)
where, , , , wc  nd a

f0 1    q
Ec
where, ft∗ = q tm , λo = n d
m
1+ λλo o 2k1 a ,B= 1+ −Et , wc = nda .

0.1
0.727
0 1
0 0.2 0.4 0.6 0.8
-0.1

-0.2
Log (σN/σo)

2.35
1
-0.3 2
1
-0.4
PK stress (Exp)
-0.5 Cauchy stress (Exp)
Bazant (n = 3, k1 = 1, m = 2)
-0.6 LEFM
Modified Bazant (n, k1 = varied, m = 2)
Modified Bazant (n = 3, k1 = 0.375, m = 0.727)
-0.7
Log (d/d0)
Figure 10.Doubly
Figure10. Doublylog-scale
log-scaleplot
plotofofthe
thesize
sizeeffect
effectofofthe
theT-series
T-seriesmortar
mortarspecimens.
specimens.

As n = 3 and d/a = 4 is constant in λo , the acceptable range of tensile strength can be calculated with
As n = 3 and d/a = 4 is constant in  , the acceptable range of tensile strength can be calculated
λo = 16; then, k1 is close to 0.375 rather thano ~1, as in the case of concrete [4,5]. This provides a better
with oof
estimate = tensile
16; then, k1 is close
strength withtorespect
0.375 to
rather than ~1,
specimen size,asasinshown
the case of concrete
in Table 3. If k1 [4,5].
is held This provides
constant at 1, a
better
then estimate
n should of tensiletostrength
be changed withthis
8. However, respect
valuetofarspecimen
exceeds the size,range
as shown
of 3~5in forTable 3. Ifand
concrete k 1 isrock.
held
constant
The at 1, then
lower value of k1nindicates
should be changed
that the FPZto in8. However,
the vicinity ofthis thevalue
notchedfar crack
exceedswouldthe berange of 3~5 as
narrower, for
concrete and
illustrated rock. 7a.
in Figure TheThis
lower value
could be of k1 indicates
explained that the
by using FPZelement
a finite in the vicinity
smeared ofcrack
the notched
model with crack
would
mode be narrower,
I fracture. as illustrated
In Figure 11b, the FPZ in has
Figure 7a. width
a finite This could
(in redbecolor)
explained
beforeby using a finite
developing into aelement
major
smeared
crack in thecrack model
direction with
of the mode length,
shortest I fracture. In Figure
as shown 11b, the
in Figure 11c.FPZ has a finite
Regarding width (inmesh
the minimum red color)
size,
before
hef (>wc ),developing
Bažant andintoOh a[8]major crackthat
assumed in the
the direction
stress-strainof the shortest length,
relationship in the FPZ as shown inas
is linear, Figure
follows: 11c.
Regarding the minimum mesh size, hef (>wc), Bažant and Oh [8] assumed that the stress-strain
2GF 1 1 −1
 
relationship in the FPZ is linear, as follows: wc = 0 2 − (7)
ft Ec Et 1
2GF  1 1 
w 
2G c 1    !−1 (7)
2 Ec E 2 Ec
'2
− (t1 −α )c
f E −t α
F
wc = 0 2
(8)
ft Ec Et1 Et2
Materials 2018, 12, x FOR PEER REVIEW 11 of 15

1
2G  1 2 Ec E 
wc  '2F   (1   )  2 c  (8)
ft  Ec Et1 Et 2 
Materials 2019, 12, 1850 11 of 15

Table 3. Log-scale data of the size effect of the T-series mortar specimens.
Table 3. Log-scale data of the size effect of the T-series mortar specimens.
Specimen PK Cauchy Bažant Modified Bažant
Specimen CauchyLEFM Bažant
label Stress Stress Stress
PK Stress ( n = 3, k1 = 1, m = 2)
LEFM ( n = 3, Modified
k = 0.375, Bažant
m = 0.727)
Label (n = 3, k1 = 1, m = 2) (n = 3, k11 = 0.375, m = 0.727)
T-10T-10 3.89 3.89 2.58 2.58 3.16 3.16 4.09 4.09 2.78
2.78
T-20T-20 2.15 2.15 1.59 1.59 2.20 2.20 3.21 3.21 1.51
1.51
T-30T-30 1.39 1.39 1.09 1.09 1.80 1.80 2.73 2.73 0.99
0.99
T-40T-40 0.94 0.94 0.72 0.72 1.56 1.56 2.41 2.41 0.72
0.72

(a) (b) (c)

11. Llustration
Figure 11. of the
llustration of the FPZ
FPZ and
and failure simulation
simulation performed
performed with ATENA software: (a) FE FE
model ofof T-series
T-seriesspecimen
specimenwithwiththe
therigid fixtures
rigid at the
fixtures ends,
at the (b,c)(b),
ends, crack
(c) distribution and FPZ
crack distribution across
and FPZ
the notches
across with different
the notches crack width.
with different crack width.

5. Measurement of
5. Measurement of Mode
Mode II Fracture
Fracture Energy
Energy Based
Based onon the
the Cauchy
Cauchy Stress
Stress
We
We developed
developeda acrack
crackmodel based
model on the
based on crack-opening law and
the crack-opening lawfracture energy. energy.
and fracture This is suitable
This is
for modeling the propagation of cracks through concrete and was used in conjunction
suitable for modeling the propagation of cracks through concrete and was used in conjunction withwith
the
crack band.
the crack band.TheThe
crack opening
crack openingfunction
function waswasderived
derivedexperimentally
experimentallyby byHordijk
Hordijk[33],
[33],as
as shown
shown in in
Equation (9).
Equation (9).
GF Gf GF
w f = 5.14 0 or w f ' 2.06 0 where, ' 2.5 (9)
GF ft G
ft f G f GF
w f  5.14 ' or w f 2.06 ' where, 2.5 (9)
In our case, as shown in Table f4,t the Gf is not affected ft by specimen G fsize. The GF /Gf ratio showed
that the mortar specimens were less brittle than other types of materials. Typically, this ratio is
In our case, as shown in Table 5, the Gf is not affected by specimen size. The GF/Gf ratio showed
approximately 2.5 in the case of concrete [3,5]. The value of α, which links wf and GF , is approximately
that the mortar specimens were less brittle than other types of materials. Typically, this ratio is
5.14 for concrete, whereas for mortar it is 3.71, and the average GF /Gf ratio is 1.94, as shown in Equation
approximately 2.5 in the case of concrete [3,5]. The value of α, which links wf and GF, is
(10) and Table 4. Figure 12 shows the post-failure behavior of mortar according to specimen size.
approximately 5.14 for concrete, whereas for mortar it is 3.71, and the average GF/Gf ratio is 1.94, as
The definition of wf proposed in Equation (10) was validated based on a nonlinear finite element
shown in Equation (10) and Table 4. Figure 12 shows the post-failure behavior of mortar according
simulation with a smeared crack model generated in ATENA software [34], as shown in Figure 13.
to specimen size. The definition of wf proposed in Equation (10) was validated based on a nonlinear
Good agreement regarding the force and CMOD was seen between the simulation and experiment.
finite element simulation with a smeared crack model generated in ATENA software [34], as shown
in Figure 13. Good agreement
GF regardingGtheF forceG and
F CMODGwas F seen between the simulation and
w f = α 0 = 2.55 ∼ 4.51 0 ' 3.71 0 where, = 1.65 ∼ 2.23 ' 1.94 (10)
experiment. ft ft ft Gf

GF GF GF GF
wf    2.55 4.51 3.71 where,  1.65 2.23 1.94 (10)
ft ' ft ' ft ' Gf
Materials 2019, 12, 1850 12 of 15
Materials
Materials2018, 12,12,
2018, x FOR PEER
x FOR REVIEW
PEER REVIEW 12 12
of 15
of 15

Table 4. Value of parameter α used for calculating crack bandwidth, wf .

Table
Table4. 4.
Value of of
Value parameter α used
parameter forfor
α used calculating crack
calculating bandwidth,
crack wf.wf.
bandwidth,
Specimen
Specimen
’
ft (MPa) wf (mm) Gf (N/mm) GF (N/mm) GF /Gf α
Specimen
Label ft’(MPa) wf (mm) Gf (N/mm) GF (N/mm) GF/ Gf αα
label
’ ft (MPa) wf (mm) Gf (N/mm) GF (N/mm) GF/ Gf
labelT-10 2.58 0.0333 0.02 0.0337 1.685 2.55
T-10
T-10T-20 2.58
2.58 1.590.0333
0.0333 0.0910 0.02
0.02 0.01945 0.0337
0.0337 1.685 2.55
0.0321 1.650 1.6854.51 2.55
T-20
T-20 1.59
1.59 0.0910
0.0910 0.01945
0.01945 0.0321
0.0321 1.650
1.650 4.51
4.51
T-30 T-30 1.09 1.090.1450 0.14500.01946 0.01946 0.0434
0.0434 2.230
2.230 3.64 3.64
T-30 1.09 0.1450 0.01946 0.0434 2.230 3.64
T-40 T-40 0.72 0.720.2450 0.24500.01962 0.01962 0.0427
0.0427 2.176
2.176 4.13 4.13
T-40 0.72 0.2450 0.01962 0.0427 2.176 4.13

0.05
0.05
ft 'ft '

Fracture Energy (N/mm)

0.04

Fracture Energy (N/mm)

0.04
0.03
0.03
0.02
GFGF 0.02
0.01
0.01
GfGf 00 T10T10 T20T20 T30T30 T40T40
00 GF-Gf
GF-Gf 0.013733
0.013733 0.01262
0.01262 0.0239284
0.0239284 0.0230448
0.0230448
ww
f f Gf Gf 0.020.02 0.01945
0.01945 0.0194586
0.0194586 0.0196152
0.0196152
(a)(a) (b)(b)
Figure
Figure 12. 12.12.
Figure Fracture
Fracture energies:
energies:
Fracture (a)(a)
energies: specific
specific
(a) and
and
specific total
total
and fracture
fracture
total energies,
energies,
fracture GG
energies, f and
Gf fand
GFG
and , (b)
G measured
F,, (b)
(b) fracture
measured
measured fracture
fracture
energies
energies fromfrom
energies T-series
T-series
from specimens.
specimens.
T-series specimens.

2000 2000
2000
2000
T10-2 T20-2
T20-2
T10-2
T10-3 T20-4
T20-4
T10-3 T20-6
T10-4 T20-6
T10-4 1500 T20-8
1500
1500 T10-6 1500 T20-8
T10-6 T20-9
T20-9
T10-9
T10-9 Atena (T-20)
Atena (T-20)
Force (N)
Force (N)

Atena (T-10)
Force (N)

Atena (T-10)
Force (N)

1000 1000
1000
1000

500500 500500

0 0 0 0
0 0 0.10.1 0.20.2 0.30.3 0 0 0.10.1 0.20.2 0.30.3
CMOD (mm)
CMOD (mm) CMOD (mm)
CMOD (mm)
(a)(a) (b)(b)
2000
2000 2000
T30-3 2000
T30-3 T40-1
T40-1
T30-4
T30-4 T40-2
T30-5 T40-2
T30-5 T40-3
T40-3
1500
1500 T30-7
T30-7 1500
1500 Atena (T-40)
Atena (T-40)
T30-8
T30-8
Force (N)

T30-9
Force (N)
Force (N)

T30-9
Force (N)

Atena (T-30)
Atena (T-30)
1000
1000 1000
1000

500500 500500

0 0 0 0
0 0 0.10.1 0.20.2 0.30.3 0 0 0.10.1 0.20.2 0.30.3
CMOD
CMOD(mm)
(mm) CMOD
CMOD(mm)
(mm)
(c)(c) (d)(d)

Figure 13. ATENA simulation results using the measured values of Gf and GF for specimens (a) T-10,
(b) T-20, (c) T-30, and (d) T-40.
Materials 2018, 12, x FOR PEER REVIEW 13 of 15

Figure
Materials 2019,13.
12, ATENA
1850 simulation results using the measured values of Gf and GF for specimens (a) T-10,
13 of 15
(b) T-20, (c) T-30, and (d) T-40.

Hillerborg [35]
Hillerborg [35] and
and Bažant
Bažant [36]
[36] calculated
calculated llch based on
ch based on the
the G and G
GFF and Gff.. As
As given
given byby Equation
Equation (9),
(9),
the ratio between G and G is approximately 1.94 in the case of our mortar specimen.
the ratio between GFF and Gff is approximately 1.94 in the case of our mortar specimen. These two These two
characteristic lengths
characteristic lengths have
have the
the same
same change
change with
with the
the factor
factor of
of 1.94
1.94 in
in Equation
Equation (9).
(9). These
These measures
measures of
of
llch represent the brittleness of the specimen in terms of its elastic energy, U with respect
ch represent the brittleness of the specimen in terms of its elastic energy, Uo with respect to the
o to the surface
energy released
surface by crack by
energy released development, Ws , as indicated
crack development, by Equation
Ws, as indicated by (11). In Figure
Equation (11).14,
In we can see
Figure the
14, we
decrease
can in decrease
see the brittlenessin with larger with
brittleness specimen
largersize.
specimen size.

EEccG GFF EEc GcG ff Ws

W
llchch = 
= 1.94
1.94 = s (11)
(11)
ff t0'22 ft0f2'2 U
Uo o
t t

1200
L1_Hillerborg
Characteristic Length, Lch (mm)

1000 Lch_Bazant

800

600

400

200

0
T-10 T-20 T-30 T-40

Figure 14. Characteristic length of the T-series specimens.

Figure 14. Characteristic length of the T-series specimens.
6. Conclusions
6. Conclusion
In this study, we investigated the fracture energy and size effect of mortar specimens under direct
In this
uniaxial study, we investigated the fracture energy and size effect of mortar specimens under
tension.
direct uniaxial tension.
1. The fractured surfaces were scanned using a 3D scanner and we calculated Cauchy stresses
1. The fractured
to evaluate thesurfaces
fracturewere energy scanned
preciselyusing a 3D scanner
by using and wesurface.
final fractured calculated
TheCauchy stresses
actual fractured
to evaluate
surfaces werethe fracture
larger energy
than the initialprecisely
section, sobytheusing final
fracture fractured
stresses were surface.
calculatedThe actual
according
fractured surfaces were larger than the initial section, so the
to the Cauchy stress principle. It is important to quantify the mode I fracture energy based fracture stresses were
on
calculated
experimental according
test data. to Smaller
the Cauchy stress group
specimen principle.
shows It is important
relatively to quantify
larger thebetween
differences mode I
fracture
fracturedenergy based
and initial on experimental test data. Smaller specimen group shows relatively
sections.
2. larger
The average GF /Gf between
differences ratio for thefractured
mortar and initial was
specimens sections.
1.94, which is lower than the typical value
2. The average G /G ratio for the mortar specimens
for concrete, of 2.5. The direct tension test on the double-notched
F f was 1.94, which
mortarisspecimens
lower than theno
with typical
major
value for yielded
inclusion concrete, of 2.5.
a small The
FPZ direct
after tension
tensile cracktest on thesodouble-notched
initiation, mortar
that the tail portion specimens
of the softening
branch
with nowas alsoinclusion
major relativelyyielded
small. This led toFPZ
a small a decrease in the G
after tensile F /Gf ratio.
crack We verified
initiation, so thatthisthevia
taila
nonlinear
portion of fracture mechanics
the softening branchsimulation,
was alsowhich agreed
relatively wellThis
small. withledourtoexperimental
a decrease in results.
the GF/Gf
3. ratio.
We investigated the size effect for four different specimens with
We verified this via a nonlinear fracture mechanics simulation, which agreed d/a held constant, to eliminate
well
shapeour
with effects. The traditional
experimental results.LEFM-based prediction and the Bažant size effect law predict a
3. We gradient of 1/2 in the
investigated the casesize of relatively
effect for fourlargedifferent
specimens. Typical laboratory-scaled
specimens with d/a held constant,specimens to
usually exhibit gradients below 1/2 in the case of concrete; thus, nonlinear
eliminate shape effects. The traditional LEFM-based prediction and the Bažant size effect fracture mechanics
should
law be useda togradient
predict estimate tensile
of 1/2strength
in theincasetermsof of structural
relativelydimensions. In the case
large specimens. of our
Typical
mortar specimens, the slope value, 1/0.727, violated either the LEFM
laboratory-scaled specimens usually exhibit gradients below 1/2 in the case of concrete; theory and Bažant size
effect law, which was unexpected because LEFM predicts a strong size effect. More observations
thus, nonlinear fracture mechanics should be used to estimate tensile strength in terms of
are required to explore this according to variations in d/a. In order to explore this discrepancy,
Materials 2019, 12, 1850 14 of 15

different direct tensile test specimen could be applied; (a) un-notched specimen, (b) one-side
single-notched specimen, and (c) double-notched specimen. Notch sensitivity may lead to
different extent of FPZ and tensile strength change in accordance with specimen size.

Author Contributions: Conceptualization, I.R.; Methodology, Y.-S.R.; Software, I.R.; Validation, I.R., J.S.L. and
Y.-S.R.; Investigation, J.S.L.; Data Curation, J.S.L.; Writing—Original Draft Preparation, I.R.; Writing—Review &
Editing, I.R., Y.-S.R.
Funding: This paper was funded by National Research Funds in Korean Ministry of Education (grant no.:
NRF-2016R1D1A3B04930497).
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Van Vliet, M.R.A. Size Effect in Tensile Fracture of Concrete and Rock. Ph.D. Thesis, TU-Delft, Delft,
The Netherlands, 2000; p. 192.
2. Rilem, T.Q.F.S. Quasibrittle fracture scaling and size effect-Final report. Mat. Struct. 2004, 37, 547–568.
[CrossRef]
3. Van Mier, J.G.M. Concrete Fracture: A Multiscale Approach; CRC Press: Boca Raton, FL, USA, 2013; pp. 1–357.
4. Bažant, Z. Size effect in blunt fracture: Concrete, Rock, Metal. ASCE J. Eng. Mech. 1984, 110, 518–535.
[CrossRef]
5. Bažant, Z.; Planas, J. Fracture and Size Effect in Concrete and other Quasibrittle Materials; CRC Press: Boca Raton,
FL, USA, 1997; pp. 1–640.
6. Bažant, Z.; Kim, J.-K.; Pfeiffer, P.A. Nonlinear fracture properties from size effect tests. J. Strut. Eng. 1986,
112, 289–307. [CrossRef]
7. Karihaloo, B.L.; Carpintreri, A.; Elices, M. Fracture mechanics of cement mortar and plain concrete. Adv. Cem.
Based Mat. 1993, 1, 92–105. [CrossRef]
8. Bažant, Z.; Oh, B.H. Crack band theory for fracture of concrete. Mat. Constr. 1983, 16, 155–177. [CrossRef]
9. Bažant, Z. Concrete fracture models: Testing and practice. Eng. Fract. Mech. 2002, 69, 165–205. [CrossRef]
10. Van Vliet, M.R.A.; Van Mier, J.G.M. Size effect of concrete in uniaxial tension. Trans. Eng. Sci. 1996, 13,
823–830.
11. Bažant, Z.; Daniel, I.; Li, Z. Size effect and fracture characteristics of composite laminates. J. Eng. Mat. Tech.
1996, 118, 317–324. [CrossRef]
12. Kozul, R.; Darwin, D. Effects of Aggregate Type, Size, and Content on Concrete Strength and Fracture Energy;
SM Report No. 43; University of Kansas: Lawrence, KS, USA, 1997; pp. 1–98.
13. Nishikawa, T.; Ito, S.; Awaji, H. Fracture mechanics and microstructure of cement mortar. In Proceedings of
the FRAMCOS-3, Gifu, Japan, 12–16 October 1998; pp. 223–232.
14. Van Vliet, M.R.A.; Van Mier, J.G.M. Effect of strain gradients on the size of concrete in uniaxial tension.
In Fracture Scaling; Springer: Dordrecht, The Netherlands, 1999; Volume 95, pp. 195–219.
15. Van Vliet, M.R.A. Size effect of concrete and sandstone. Heron 2000, 45, 91–108.
16. Zheng, W.; Kwan, A.K.H.; Lee, P.K.K. Direct tension test of concrete. Mat. J. 2001, 98, 63–71.
17. Zhu, Y.; Xu, S.L. Fracture properties of cement paste and mortar: An experimental investigation.
In Proceedings of the International Association of Fracture Mechanics for Concrete and Concrete Structures,
FRAMCOS-6, Catania, Italy, 17–22 June 2007; pp. 1–7.
18. Xu, S.; Zhu, Y. Experiment studies on fracture energy of cement paste and mortar. In Key Engineering Materials;
Trans Tech Publications: Zürich, Switzerland, 2007; Volume 348–349, pp. 169–172.
19. Aliha, M.R.M.; Ayatollahi, M.R. Brittle fracture evaluation of a fine grain cement mortar in combined
tensile-shear deformation. Fatigue Fract. Eng. Mat. Struct. 2009, 32, 987–994. [CrossRef]
20. Wang, J.J.A.; Liu, K.C.; Naus, D. A new test method for determining the strength and fracture toughness of
cement mortar and concrete. In Proceedings of the ASME Pressure Vessels and Piping Conference, Prague,
Czech Republic, 26–30 July 2009; American Society of Mechanical Engineers: New York, NY, USA, 2009;
Volume 6, pp. 13–19.
21. Chen, X.; Wu, S.; Zhou, J. Strength values of cementitious materials in bending and tension test methods.
J. Mat. Civ. Eng. 2014, 26, 484–490. [CrossRef]
Materials 2019, 12, 1850 15 of 15

22. Zhou, J.; Qian, P.; Chen, X. Stress-strain behavior of cementitious materials with different sizes. Sci. World J.
2014, 2014, 919154. [CrossRef] [PubMed]
23. Del Viso, J.R.; Carmona, J.R.; Ruiz, G. Size and shape effects on the compressive strength of high strength
concrete. Cem. Concr. Res. 2008, 38, 386–395. [CrossRef]
24. Chupanit, P.; Roesler, J.R. Fracture energy approach to characterize concrete crack surface roughness and
shear stiffness. J. Mat. Civ. Eng. 2008, 20, 275–282. [CrossRef]
25. Volokh, K.Y. Characteristic length of damage localization in concrete. Mech. Res. Comm. 2013, 51, 29–31.
[CrossRef]
26. Cifuentes, H.; Alcalde, M.; Medina, F. Comparison of the size-independent fracture energy of concrete
obtained by two test methods. In Proceedings of the FRAMCOS-8, Toledo, Spain, 10–14 March 2013; pp. 1–6.
27. Fernandez-Canteli, A.; Castanon, L.; Nieto, B.; Lozano, M. Determining fracture energy parameters of
concrete from the modified compact tension test. Frattura Integrità Strut. 2014, 30, 383–393. [CrossRef]
28. Toward a Rational Understanding of Shear in Beams and Slabs; CEB-FIP Bulletin: Zürich, Switzerland, 2016;
Volume 85, pp. 214–218.
29. Uday, N.P. Experimental determination of fracture energy by RILEM method. Int. J. Eng. Sci. 2017, 6,
106–115. [CrossRef]
30. To, T.; Celarie, F.; Roux-Langlois, C.; Bazin, A.; Gueguen, Y.; Orain, H.; Le Fur, M.; Burgaund, V.; Rouxel, T.
Fracture toughness, fracture energy and slow crack growth of glass as investigated by the single-edge
precracked beam (SEPB) and Chevron-notched beam (CNB) methods. Acta Mat. 2018, 146, 1–11. [CrossRef]
31. Ko, S.; Chan, K.; Hawkins, R.; Jayaram, R.; Lynch, C.; El Mamoune, R.; Nguyen, M.; Pekhotin, N.; Stokes, N.;
Wu, D.; et al. Experimental and numerical characterization of the intra-laminar fracturing behavior in
discontinuous fiber composite structures. In Proceedings of the 33th ASC Conference, Seattle, WA, USA,
24–26 September 2018; pp. 1–15.
32. Werner, S.; Kustermann, A.; Thienel, C. Analyzing the influence of maximum aggregate size upon the
concrete behavior under projectile impact by means of fracture mechanic and surface analytic methods.
In Proceedings of the 9th fib International Ph.D. Symposium in Civil Engineering, Karlsruhe, Germany,
22–25 July 2013; pp. 1–8.
33. Hordiijk, D.A. Local Approach to Fatigue of Concrete. Ph.D. Thesis, TU-Delft, Delft, The Netherlands, 1991.
34. Atena Theory Manual; Cervenka Consulting: Praha, Czech Republic, 2013.
35. Hillerborg, A. The theoretical basis of a method to determine the fracture energy GF of concrete. Mat. Struct.
1985, 18, 291–296. [CrossRef]
36. Bažant, Z.; Pfeiffer, P.A. Determination of fracture energy from size effect and brittleness number. ACI Mat. J.
1987, 84, 463–480.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).