Professional Documents
Culture Documents
Abstract: To evaluate the effects of beam specimen depth and aggregate size on the fracture energy of lightweight concrete (LWC), different
Downloaded from ascelibrary.org by Tokyo Univ Seisan Gijutsu on 05/18/15. Copyright ASCE. For personal use only; all rights reserved.
beam specimens designated into 32 notations were tested under three-point bending. In each of the all-lightweight concrete and sand-
lightweight concrete groups, the maximum aggregate size varied between 4 and 19 mm; the beam depth ranged from 150 to 600 mm
in each ready-mixed concrete batch with the same mix proportions. Based on experimental observations and verification of prior empirical
models, simple closed-form equations were proposed to generalize the influence of the concrete unit weight on the size effect for the fracture
energy of concrete. Test results clearly showed that when the maximum aggregate size is larger than 8 mm, the aggregate size in LWC has an
insignificant effect on fracture parameters such as the fracture energy, crack opening mouth displacement, and characteristic length due to
crack propagation through the lightweight aggregate particles. The fracture energy of LWC was lower than that of normal-weight concrete,
indicating that the size-dependence of the fracture energy increases with decreasing concrete unit weight. A comparison between the pre-
dicted and experimental fracture energies revealed that the reliability of existing models significantly depends on the concrete type and
ligament depth of the beam specimen, whereas the proposed model generally gives better agreement with the test data; it consistently predicts
the trend of the size effect, regardless of the concrete unit weight. DOI: 10.1061/(ASCE)MT.1943-5533.0000884. © 2014 American Society
of Civil Engineers.
Author keywords: Lightweight concrete; Fracture energy; Size effect; Aggregate size; Concrete unit weight.
Introduction The fracture energy of concrete is also dependent on the size and
strength of aggregates and on the interfacial cohesion between ag-
Concrete is a typical quasibrittle material that exhibits nonlinear gregates and paste because the width of the fracture process zone is
behavior even at the ascending branch and softening characteristics significantly affected by aggregate interlock action (Elices and
after peak stress owing to crack growth. To explain the effect of Rocco 2008). The prediction equation specified in the CEB-FIP
crack propagation on the structural response of concrete elements, provision is relevant to the aggregate size and concrete compressive
fracture parameters, including the fracture energy, fracture tough- strength. Hillerborg (1985) noted that GF tends to increase pro-
ness, and fracture process zone, need to be established (Gettu et al. vided that the maximum aggregate size (da ) increases from 8 to
1990). The Comité Euro-International du Beton (CEB-FIP) code 20 mm; although this tendency shows a large scatter. Elices and
provision (1999) specifies an empirical equation for evaluating Rocco (2008) conducted 44 notched beam tests to ascertain the in-
the fracture energy (GF ) of concrete. However, there is still debate fluence of the matrix, aggregate, and matrix-aggregate interface
on the issue of concrete fracture parameters, particularly with re- properties on the fracture parameters and showed that GF increases
gard to size dependence. Zhang and Wu (1999) pointed out that GF by a rate of 120% for a fourfold increase in da . Choi et al. (2003)
is dependent on the member size because of the existence of the revealed that the rate of increase in GF against da becomes more
fracture process zone in front of an extending crack in concrete. prominent with increasing concrete compressive strength. How-
Carpinteri and Chiaia (1995, 1996) reviewed extensive test results ever, the variation in GF against the strength and interfacial cohe-
and concluded that GF increases with the member size up to a cer- sion capacity of aggregate is still uncertain owing to the lack of
tain limit, beyond which it remains constant; considering these ob- experimental evidence.
servations, they proposed a simple size-effect law procedure that The extensive experimental research reviewed above for GF was
includes the extrapolation of infinite size. conducted using normal-weight concrete (NWC) (Gettu et al. 1990;
1
CEB-FIP 1999; Zhang and Wu 1999; Carpinteri and Chiaia 1995,
Dept. of Architectural Engineering, Kyonggi Univ., Suwon, 1996; Elices and Rocco 2008; Hillerborg 1985; Choi et al. 2003;
Kyonggi-do 443-760, Republic of Korea.
2
Professor, Dept. of Plant Architectural Engineering, Kyonggi Univ.,
RILEM 1985; Bažant et al. 1991; Bažant and Kazemi 1991; Zhang
Kyonggi-do 443-760, Republic of Korea (corresponding author). E-mail: et al. 2010; Karihaloo and Nallathambi 1990; Guo and Gilbert
yangkh@kgu.ac.kr 2000; Barragan et al. 2001; Giaccio et al. 1993). Artificial light-
3 weight aggregates are commonly weaker and have smoother sur-
Professor, School of Architecture and Building Science, Chungang
Univ., Seoul 156-756, Republic of Korea. faces than normal-weight aggregates. These properties of
4
Associate Professor, Dept. of Civil and Environmental Engineering, lightweight aggregates result in lower tensile strength of light-
Inha Technical College, Incheon 402-752, Republic of Korea. weight concrete (LWC), lower cohesion between aggregate par-
Note. This manuscript was submitted on September 2, 2012; approved
ticles and pastes, and a more brittle descending branch after
on June 24, 2013; published online on June 26, 2013. Discussion period
open until October 1, 2014; separate discussions must be submitted for in- peak stress compared with NWC of the same compressive strength.
dividual papers. This paper is part of the Journal of Materials in Civil Furthermore, cracks in LWC propagate mostly by passing through
Engineering, Vol. 26, No. 5, May 1, 2014. © ASCE, ISSN 0899-1561/ the aggregates; this significantly reduces the aggregate interlock
2014/5-845-854/$25.00. capacity along the crack planes. Hence, the width of the fracture
Table 1. Details of Ready-Mixed Concrete Batches and Measured Mechanical Properties of Concrete
Unit volume weight (kg=m3 )
Concrete Concrete da W=C f c0 ft Ec ρc
mixa type (mm) (%) W C FL FS GL (MPa) (MPa) (MPa) (kg=m3 )
A4 ALWC 4 30 200 667 725 — — 47.3 2.4 17,505 1,650
A8 8 240 455 48.3 2.6 17,520 1,620
A13 13 264 418 44.9 2.5 16,754 1,607
A19 19 255 437 42.2 2.5 16,251 1,617
S4 SLWC 4 35 571 — 1,423 — 37.6 2.5 21,181 2,234
S8 8 520 455 44.3 2.6 17,654 1,786
S13 13 566 418 38.5 2.5 16,387 1,794
S19 19 549 437 39.1 2.6 16,539 1,797
Note: da = maximum aggregate size, W=C = water-to-cement ratio by weight, FL = lightweight fine aggregate, FS = natural normal-weight sand, GL =
lightweight coarse aggregates, fc0 = compressive strength measured from a cylinder with 150 mm diameter and 300 mm height, f t = direct tensile strength
measured from a prism with square section of 100 mm, Ec = modulus of elasticity recorded from the cylinder, and ρc = unit weight of hardened concrete.
a
In the concrete mix notation, the first and second parts indicate the type of concrete and maximum aggregate size, respectively. For example, A8 indicates an
all-lightweight ready-mixed concrete with a maximum aggregate size of 8 mm.
revealing that the propagation and shape of the crack are not af- Fig. 3 plots typical load-CMOD curves measured from ALWC
fected by D and da . In addition, observation of the failure surfaces beams. The slope at the ascending branch of CMOD curves was
clearly demonstrated that the crack mostly penetrates through the independent of D and da . When a visible crack occurred at the
lightweight coarse aggregate particles. Hence, aggregate interlock tip of the notch, CMOD sharply increased; this shows that the rate
action along the crack can be expected to be reduced. of increase was negligibly affected by da owing to the extremely
reduced aggregate interlock along the crack. This response after
cracking immediately influenced the amount of wc . Hence, the
amount of wc was found to be almost irrelevant to da , as given
7000
d a = 4 mm
No1
in Table 3. After reaching peak load, the crack opening gradually
6000 d a = 8 mm
grew, and the applied load dropped sharply, showing a larger rate of
No2
decrease with increasing D.
5000 d a = 13 mm
No3
Applied load (N)
d a = 19 mm
No4
4000 Load-Deflection Response
3000 The load-deflection response observed in all beams aligned to the
findings of load-CMOD curves, as shown in Fig. 4. Until the oc-
2000 currence of cracks at the tip of the notch, all specimens exhibited
1000
linear behavior. The rate of increase in displacement slightly de-
creased as D increased up to 450 mm; beyond this, it was indepen-
0 dent of D. The rate of decrease in the applied load at the descending
0 0.2 0.4 0.6 0.8 1 1.2
branch became slightly sharp as D increased, i.e., the flexural
(a) Crack mouth opening displacement, CMOD (mm) behavior of beams tended to become more brittle with increasing
D. In addition, when the beam was larger, the amount of deflections
7000
D
No4
= 150 mm
at peak load and at final fracture increased, as given in Table 3.
6000 D = 300 mm
No4
These observations of the ascending and descending branches were
D = 450 mm
No4
independent of da , unlike the general tendency (Man and van Mier
5000 D = 600 mm
No4 2008) revealed for NWC specimens, wherein the fracture behavior
Applied load (N)
2000
Effect of d a on G F
1000
Fig. 5 shows the fracture energy of LWC measured from beams
0 with a D of 150 mm according to da . The same figure plots the
0 0.2 0.4 0.6 0.8 1 1.2 GF of NWC beams (Elices and Rocco 2008; Hillerborg 1985; Choi
(b) Crack mouth opening displacement, CMOD (mm) et al. 2003; Karihaloo and Nallathambi 1990; Guo and Gilbert
2000) with a similar depth for comparison. In general, GF mea-
Fig. 3. Typical load-CMOD curves of ALWC beams
sured from the present LWC beams was lower than that of NWC.
d a = 8 mm
No2
4000 d a = 13 mm
No3
0.10
3000 d a = 19 mm
No4
(a)
Deflection at mid-span (mm) Depth of ligament, d l (mm)
No4
D = 450 mm
4000 No4
D = 600 mm Fig. 6 shows GF according to the variation in dl . NWC data com-
piled from Hillerborg’s tests (1985) are also provided in the figure
3000 to examine the influence of the concrete type on the size effect on
GF . The differences in GF between SLWC and ALWC slightly nar-
2000 rowed as dl increased. The fracture energy of concrete generally
1000
increased with an increase in dl , showing that the rate of increase
was minimally affected by da. When dl increased from 90 to
0 360 mm, the average rates of increase in GF were evaluated to
0 0.2 0.4 0.6 0.8 1 1.2 be 228% and 197% for ALWC and SLWC, respectively. The rate
(b) Deflection at mid-span (mm) of increase in GF for NWC was 115% as dl increased from 90 to
300 mm. The rate of increase in GF was slightly higher in ALWC
Fig. 4. Typical load-deflection curves of ALWC beams than in SLWC, which in turn was higher than in NWC. Hence, the
size dependence of GF becomes stronger with a decrease in the
concrete unit weight. The tensile resistance capacity of LWC is
The fracture energy of SLWC was slightly higher by an average of generally lower than that of NWC of the same compressive
1.08 times than that of the companion ALWC, as given in Table 3. strength, indicating that the ratio of tensile to compressive strengths
For NWC, GF tended to increase with increasing da . GF measured is higher in LWC than in NWC. This lower tensile strength due to
in LWC beams with a da above 8 mm was almost constant, the decrease in concrete unit weight increases the amount of
although a value approximately 166% higher was obtained when CMOD and the effective length of the fracture process zone, which
da increased from 4 mm to 8 mm. The effect of da on GF of LWC results in a larger variation in energy dissipation against dl in LWC
was independent of dl . As a result, when da is above 8 mm, the rate than in NWC. This is one reason for the increased size effect of GF
of increase in GF against da in LWC was small enough to be on LWC than on NWC. The rate of increase in GF was found to be
ignored compared with the results observed in the best-fit curve mitigated when dl was larger than 270 mm. Based on an extensive
review of NWC test data, Wittmann et al. (1990) concluded that the
for NWC. This indicates that the aggregate interlock action of
fracture process zone remains constant after reaching the highest
LWC with da above 8 mm can be expected to have a negligible
length; as a result, an asymptotic value of GF can be obtained
effect on GF .
beyond a certain limit of dl . In the present tests, the value of GF
increased by an average of 109% only when dl increased from 270
0.4 to 360 mm. This clearly indicates the asymptotic size-dependence
NWC - D = 100 mm (d l /D = 0.5) of GF in LWC as well.
ALWC
Fracture energy,GF (N/mm)
D = 150 mm (d l /D = 0.6)
0.3 SLWC
Proposal of Simple Model
Best fit curve for NWC
0.2 The fracture energy is usually considered to be a material constant
representing the fracture process, including aggregate interlock-
ing and the dissipation mechanism ahead of the crack tip. As the
0.1 energy dissipation process is complicated and somewhat ambigu-
ous (Carpinteri and Chiaia 1995, 1996), numerous experimental
and analytical approaches (Gettu et al. 1990; Carpinteri and Chiaia
0.0 1995; Bažant and Oh 1983; Shah 1990) have been investigated to
0 5 10 15 20 25 30 35 define and determine GF according to a standard method. As a re-
Maximum aggregate size, d a (mm) sult, the RILEM recommendation (1985) of not considering the
size effect has predominantly governed the experimental determi-
Fig. 5. Effect of da on GF
nation of GF since the 1990s. However, the measured value of GF
Based on the multifractal scaling effect law plotted in Fig. 7, F value for NWC increases with the increase of da , while that
Carpinteri and Chiaia (1995, 1996) concluded that the nominal for LWC increases up to da of 8 mm, beyond which it almost re-
fracture energy of concrete can be fundamentally deduced as mains constant. To reflect these trends into a simple equation for
follows: G∞ 0
F , influencing parameters such as f c and da were adjusted by
nonlinear multiple regression analysis until a high correlation co-
∞ lch −1=2 efficient (R2 ) was obtained, as shown in Fig. 8. The different effects
GF ¼ GF 1 þ ð1Þ
dl of da according to the type of concrete were identified by introduc-
ing the parameter of ξ a . Overall, G∞
f can be reasonably expressed
The characteristic length (lch ) was assumed to be a function of in the following form:
the maximum aggregate size, namely, B1 da . The exponent −1=2
G∞ 0 0.5
F ¼ 0.03 ln½ξ a ðf c Þ þ 0.135 ð3Þ
representing the maximum slope of the size effect law was deter-
mined from the hypothesis of a Brownian microscopic disorder of where ξ a is da for NWC, and ðda Þ1.25 ≤ 13.45 for LWC, indicating
the microstructure. Overall, the format of Eq. (1) is very similar to that da for LWC is limited to be below 8 mm. Note that the deter-
that of the size effect equation generalized by Bažant and Kazemi mination of G∞ f using Eq. (1), as proposed by Carpinteri and Chiaia
(1991). This may be because those models were fundamentally de- (1996) usually gives a tedious procedure through regression analy-
rived under the hypothesis that the potential energy release rate at sis of a couple of datasets. However, the value of G∞ f straightfor-
the fracture is dependent on the width of the crack band zone wardly determined from Eq. (3) is similar to that obtained from
formed from smeared microcracks and on the length of the fracture Eq. (1), as shown in Fig. 7.
process zone. Bažant and Kazemi also assumed that the width of
the crack band zone is proportional to da . On the other hand, Yang
Determination of Experimental Constants
and Sim (2011) showed that the length of the fracture process zone
is minimally affected by da, especially in LWC, whereas it is sig- A comprehensive database on GF compiled from the available lit-
nificantly affected by ρc. This fact was also observed in the present erature (Elices and Rocco 2008; Hillerborg 1985; Choi et al. 2003;
test results. Similar to the analogy between Eq. (1) and the size Karihaloo and Nallathambi 1990; Guo and Gilbert 2000) was es-
effect law of Bažant and Kazemi, a closed-form equation of GF tablished to determine the experimental constants of B1 and C1 in
can thus be formulated using the size effect equation generalized Eq. (2). As it is extremely difficult to find published data for LWC
by Yang and Sim (2011) in the following form (see Fig. 7): except for the three LWC beam specimens tested by Tang et al.
(2008), the database only includes NWC specimens with a total
B1 ρc C1 −1=2 of 108 datasets. The distribution of main parameters in the
GF ¼ G∞ 1 þ ð2Þ
F
d l ρ0 NWC database is plotted in Fig. 9. The NWC beam specimens were
mainly from low- or medium-strength concrete, although a small
where B1 and C1 are experimental constants, and ρ0 (¼ 2; 300 minority of specimens had an f c0 of more than 80 MPa. The overall
kg=m3 ) is the reference value for the concrete unit weight. Note depth of beam specimens ranged between 63 and 258 mm, but the
that dl in Eq. (2) is in millimeters. The above equation represents majority were around 100 mm. This indicates that test data to
Frequency
Frequency 25 19 19
20 30
15
20
10 11
3 10 4
5 1
0 0
30 30~40 40~60 60~80 80 100 100~150 150~200 200~250 250
(a) Compressive strength of concrete, f' c (mm) (b) Overall depth of beam, D (mm)
60 80
Downloaded from ascelibrary.org by Tokyo Univ Seisan Gijutsu on 05/18/15. Copyright ASCE. For personal use only; all rights reserved.
50 55 70
67
60
40 34
Frequency
Frequency
50
30 40
30
20 25
11 20
10 6 9
2 10 5
2
0 0
5 5~15 15~25 25~30 30 50 50~100 100~150 150~200 200
(c) Maximum aggregate size, d a (mm) (d) Depth of ligament, d l (mm)
70 60
59
60 50 54
50
40
Frequency
Frequency
38
40 29
30
30 19
20
20
7 10 4
10 1 3 2
0 0
0.5 0.5~0.6 0.6~0.7 0.7~0.8 0.8 5 5~10 10~20 20~30 30
(e) d l / D ratio (f) d l / d a ratio
ascertain the size effect on GF are rare, even for NWC. The dl =D with Carpinteri and Chiaia’s (1996) equation and the size effect
ratios were limited to a specific range between 0.5 and 0.9; as a model (SEM) proposed by Shah (1990), 41 NWC beams (Gettu
result, dl varied from 41 to 233 mm. The maximum aggregate size et al. 1990; Elices and Rocco 2008; Karihaloo and Nallathambi
of NWC ranged between 5 and 32 mm; the D=dl and dl =da ratios 1990; Guo and Gilbert 2000) and the current 32 LWC beams were
varied from 3.1 to 42.7 and from 1.6 to 38.2, respectively. used only because those models require a series of specimens with
Using the reference value of G∞f determined by Eq. (3), the op- different depths to determine GF . The statistical comparison includ-
timum values of the unknown quantities in the basic formula of ing the mean (γ cs:m ), standard deviation (γ cs:s ), and coefficient of
Eq. (2) were fitted by regression analysis using the current test data variation (γ cs:ν ) of the ratios (γ cs ¼ ðGF ÞExp: =ðGF ÞPre: ) for the mea-
and 108 NWC data points (Elices and Rocco 2008; Hillerborg sured to predicted fracture energies are also given in the same figure
1985; Choi et al. 2003; Karihaloo and Nallathambi 1990; Guo for each concrete type. The previous model set used for the com-
and Gilbert 2000). In this regression analysis, the unit weights
parisons includes four separate formulas: the design equations
and da of NWC were assumed to be 2,300 kg=m3 and 19 mm, re-
specified in the CEB-FIP provision (1999), the empirical equations
spectively, for the specimens with no data available. Ultimately,
constants B1 and C1 were approximated as 900 and −0.35, respec- developed by Carpinteri and Chiaia (1996), the equations by
tively, from the fitting procedure. Bažant and Oh (1983), and the SEM proposed by Shah (1990).
Appendix summarizes examples to determine GF using these
models for comparison. The important findings that emerged from
Comparison of Experimental Results and Predictions the comparisons are described below.
The fracture energy of concrete was predicted using different As the design equations specified in the CEB-FIP provision
equations, including that of both previous and current models; are primarily formulated as a function of fc0 and da , based on
these results were compared with the experimental results. Separate NWC test data, they give a higher value for γ cs:m in LWC than in
comparisons were made for existing NWC data (Elices and NWC. The value of γ cs:m obtained from the CEB-FIP equations
Rocco 2008; Hillerborg 1985; Choi et al. 2003; Karihaloo and increases for LWC with an increase in dl ; as a result, the equations
Nallathambi 1990; Guo and Gilbert 2000) and the current LWC become unconservative for beams with a dl of less than 200 mm
beams, as plotted in Fig. 10. It should be noted that for comparison [Fig. 10(a)], regardless of the concrete type. Carpinteri and Chiaia’s
4 4
(GF )Exp./(GF )Pre.
2 2
1 1
0 0
0 100 200 300 400 0 100 200 300 400
(a) Depth of ligament, d l (mm) (c) Depth of ligament, d l (mm)
Downloaded from ascelibrary.org by Tokyo Univ Seisan Gijutsu on 05/18/15. Copyright ASCE. For personal use only; all rights reserved.
5 5
4 4
(GF )Exp./(GF )Pre.
2 2
1 1
0 0
0 100 200 300 400 0 100 200 300 400
(b) Depth of ligament, d l (mm) (d) Depth of ligament, d l (mm)
4
(GF )Exp./(GF )Pre.
0
0 100 200 300 400
(e) Depth of ligament, d l (mm)
Fig. 10. Comparison of predicted GF with experimental results against dl (a) CEB-FIP (1999); (b) Carpinteri and Chiaia (1996); (c) Bažant and Oh
(1983); (d) SEM; (e) this study
equation requires a series of specimens with different depths to than in LWC. The SEM also produces scattered results, regardless
empirically determine the values of G∞ F and lch . Using the value of the concrete type [Fig. 10(d)]. Similar to Carpinteri and Chiaia’s
of G∞ F determined from Eq. (3), lch varied between 13.3 and equation, the SEM should experience a tedious regression pro-
256 da . Meanwhile, lch determined from regression analysis of test cedure with regard to beams having different depths. Note that the
data using Eq. (1) ranged between 53 and 256 da for LWC and 13.3 empirical equations of the CEB-FIP provision, and Bažant and Oh,
and 114 da for NWC. This indicates that the value of lch Carpinteri and the SEM do not consider the size effect on GF ; they give a
and Chiaia’s equation is extremely variable-dependent on fc0 , constant value of GF for concrete with the same fc0 and mix pro-
ρc , and dl . Carpinteri and Chiaia’s equation has lower values for portions, regardless of D. The predictions obtained from the model
γ cs:s and γ cs:ν than the design equations of the CEB-FIP provision. proposed in this study agreed better with the test results than the
The value of GF determined from Carpinteri and Chiaia’s equation other models did, regardless of the concrete type [Fig. 10(e)]. The
generally underestimates the test results of NWC, although the values of γ cs:m , γ cs:s , and γ cs:ν were 1.01, 0.104, and 0.103, respec-
considered size effect is reliable [Fig. 10(b)]. Bažant and Oh tively, for LWC and 1.24, 0.437, and 0.352, respectively, for NWC.
empirically estimated GF according to the tensile strength of con- In addition, the variation in the ratios (γ cs ) was marginally depen-
crete. Bažant and Oh’s equation considerably underestimates GF , dent on dl , indicating that the size effect on GF is properly reflected
regardless of dl [Fig. 10(c)]; and gives lower values of γ cs in NWC in the proposed model.
5. Proposed model Choi, S. H., Kim, H. W., and Kim, W. J. (2003). “The fracture energy of
Using Eq. (3), G∞
f can be straightforwardly obtained as
concrete according to the maximum size of coarse aggregate by 3PBT
0.27 N=mm. Substituting this value into Eq. (2), GF for each and WTT.” J. Archit. I. Korea, 19(4), 97–104.
beam can then simply be calculated to be 0.0769, 0.1046, Elices, M., and Rocco, C. G. (2008). “Effect of aggregate size on the frac-
ture and mechanical properties of a simple concrete.” Eng. Fract.
0.1235, and 0.1379 N=mm for a dl of 90, 180, 270, and
Mech., 75(13), 3839–3851.
360 mm, respectively. Gettu, R., Bažant, Z. P., and Karr, M. E. (1990). “Fracture properties and
brittleness of high-strength concrete.” ACI Mater. J., 87(6), 608–618.
Giaccio, G., Rocco, C., and Zerbino, R. (1993). “The fracture energy (GF )
Acknowledgments of high-strength concretes.” Mater. Struct., 26(7), 381–386.
Guo, X. H., and Gilbert, R. I. (2000). “The effect of specimen size on
This work was supported by the Basic Science Research Program the fracture energy and softening function of concrete.” Mater. Struct.,
through the National Research Foundation of Korea (NRF: 2009- 33(5), 309–316.
0067189) and by a Nuclear Research and Development program of Hillerborg, A. (1985). “Results of three comparative test series for
the Korea Institute of Energy Technology Evaluation and Planning determining the fracture energy GF of concrete.” Mater. Struct., 18(5),
407–413.
(KETEP) grant funded by the Korea Government Ministry of
Karihaloo, B. L., and Nallathambi, P. (1990). “Size-effect prediction
Knowledge Economy (2011T100200161). from effective crack model for plain concrete.” Mater. Struct., 23(3),
178–185.
Mai, Y. W., and Cotterell, B. (1985). “Effect of specimen geometry on
References the essential work of plane stress ductile fracture.” Eng. Fract. Mech.,
21(1), 123–128.
Barragan, B. E., Giaccio, G. M., and Zerbino, R. L. (2001). “Fracture and Man, H. K., and van Mier, J. G. M. (2008). “Size effect on strength and
failure of thermally damaged concrete under tensile loading.” Mater. fracture energy for numerical concrete with realistic aggregate shapes.”
Struct., 34(5), 312–319. Int. J. Fract., 154(1), 61–72.
Bažant, Z. P., Gettu, R., and Kazemi, M. T. (1991). “Identification of RILEM Technical Committee 50. (1985). “Determination of the fracture
nonlinear fracture properties from size effect tests and structural analy- energy of mortar and concrete by means of three-point bend tests on
sis based on geometry-dependent R-curves.” Int. J. Rock Mech. Min. notched beams, draft recommendation.” Mater. Struct., 18(4), 287–290.
Sci., 28(1), 43–51. Shah, S. P. (1990). “Size-effect method for determining fracture energy and
Bažant, Z. P., and Kazemi, M. T. (1991). “Size dependence of concrete process zone size of concrete.” Mater. Struct., 23(6), 461–465.
fracture energy determined by the RILEM work-of-fracture method.” Tang, W. C., Lo, T. Y., and Chan, W. K. (2008). “Fracture properties of
Int. J. Fracture, 51(2), 121–138. normal and lightweight high-strength concrete.” Mag. Concrete Res.,
Bažant, Z. P., and Oh, B. H. (1983). “Crack band theory for fracture of 60(4), 237–244.
concrete.” Mater. Struct., 16(93), 155–177. Wittmann, F. H., Mihashi, H., and Nomura, N. (1990). “Size effect on frac-
Carpinteri, A., and Chiaia, B. (1995). “Multifractal nature of concrete frac- ture energy of concrete.” Eng. Fract. Mech., 35(1), 107–115.
ture surfaces and size effects on nominal fracture energy.” Mater. Yang, K. H., and Sim, J. I. (2011). “Modeling of the mechanical properties
Struct., 28(8), 435–443. of structural lightweight concrete based on size effects.” Technical Rep.,
Carpinteri, A., and Chiaia, B. (1996). “Size effects on concrete fracture Dept. of Architectural Engineering, Kyonggi Univ., Korea.
energy: Dimensional transition from order to disorder.” Mater. Struct., Zhang, D., and Wu, K. (1999). “Fracture process zone of notched three-
29(5), 259–266. point-bending concrete beams.” Cem. Concr. Res., 29(12), 1887–1892.
Comité Euro-International du Beton (CEB-FIP). (1999). Structural con- Zhang, J., Leung, C. K., and Xu, S. (2010). “Evaluation of fracture param-
crete: Textbook on behavior, design and performance, International eters of concrete from bending test using inverse analysis approach.”
Federation for Structural Concrete (FIB), Lausanne, Switzerland. Mater. Struct., 43(6), 857–874.