Engineering Fracture Mechanics: Sciencedirect

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Engineering Fracture Mechanics 209 (2019) 317–332

Contents lists available at ScienceDirect

Engineering Fracture Mechanics


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

Mesoscopic numerical simulation of dynamic size effect on the


T
splitting-tensile strength of concrete

Liu Jin, Wenxuan Yu, Xiuli Du , Wangxian Yang
Key Laboratory of Urban Security and Disaster Engineering of Ministry of Education, Beijing University of Technology, Beijing 100124, China

A R T IC LE I N F O ABS TRA CT

Keywords: Under static loadings, concrete has been demonstrated to exhibit obvious size effect due to the
Concrete heterogeneity of mesoscopic composition. Under dynamic loadings, the strain rate effect has a
Size effect significant influence on the failure of concrete. It is therefore of great importance to explore the
Splitting-tensile size effect of concrete under dynamic loadings. In this study, the focus is on the size effect in
Strain rate effect
dynamic splitting-tensile strength of concrete. A mesoscopic numerical model for the simulation
Mesoscopic simulation
of the splitting-tensile failure and size effect at different strain rates (10−5/s ∼ 200/s) was es-
tablished. The mesoscopic simulation results indicate that the size effect on the dynamic splitting-
tensile strength of concrete has an obvious discrepancy with the static one. There is a critical
strain rate in dynamic splitting-tensile strength of concrete. As the applied strain rate below the
critical strain rate the size effect behavior is restrained and weakened gradually with the addition
of strain rate. As the applied strain rate exceeds the critical strain rate the dynamic strength
enhances linearly as the structural size increases. Furthermore, considering the contribution of
strain rate on the dynamic size effect, a Static and Dynamic unified Size Effect Law (i.e. SD-SEL)
for splitting-tensile strength of concrete was developed. The proposed SD-SEL was verified by the
numerical results and the available test data.

1. Introduction

The size effect means that the mechanical properties vary as the structural size increases. Concrete materials have the size effect
behavior and their global mechanical properties show the nonlinear behavior due to the heterogeneity of mesoscopic composition
[1,2]. With the development of society and economy, the engineering structure continues to grow in size and performance in-
creasingly, resulting that the size effect problem of concrete has become the focus of research. Since the discovery of the size effect
behavior of concrete, there has been large quantity of studies on the size effect behavior in physical experiments, numerical simu-
lations and theoretical frameworks. The studies on size effect of concrete structures have fallen into two major directions [3,4]: (1)
size effect of concrete material, and (2) size effect of concrete members. The size effect of concrete material is a complex problem of
material science, mainly focusing on the influence of internal components [3].
The tensile strength is one of the main parameters in the design of concrete structures and it mainly influences the cracking
properties and durability of concrete materials. It is important to conduct efforts on the size effect of tensile strength in concrete.
However, due to limitations of test equipment, it is hard to carry out the tensile tests directly and lots of researchers measured the
splitting-tensile behavior instead in laboratory experimental tests. For instance, Ross et al. [5] and Hasegawa et al. [6] carried out the
tensile-splitting tests with cylindrical concrete specimens having small sizes, the test results indicated that the strength depended on a


Corresponding author.
E-mail address: duxiuli@bjut.edu.cn (X. Du).

https://doi.org/10.1016/j.engfracmech.2019.01.035
Received 4 December 2018; Received in revised form 14 January 2019; Accepted 28 January 2019
Available online 02 February 2019
0013-7944/ © 2019 Elsevier Ltd. All rights reserved.
L. Jin et al. Engineering Fracture Mechanics 209 (2019) 317–332

characteristic dimension and the strength decreases slightly with the increasing size. Similar conclusions were drawn in the efforts of
Ince et al. [7]. They conducted splitting-tensile tests on nine series of cubic and diagonal cubic specimens having three different
structural sizes and proposed prediction formulas based on the results of that the nominal strength of specimens decrease with the
addition of specimen size. Bažant et al. [8] designed cylinder specimens having different diameters (the maximum diameter was
600 mm) for splitting-tensile tests to investigate the size effect. The research results showed that the concrete has a size effect
behavior on the splitting-tensile strength before the critical size and the strength increases slightly when the diameter of cylinder
specimen exceeds the critical size. Conversely, Zhou et al. [9] hold different opinions according to the size effect tests on the splitting-
tensile strength of high-strength concrete. In their findings, the splitting-tensile strength of concrete enhances as structural sizes
improve and the concrete shows a distinct size effect behavior. In general, these studies above are concentrated in the static size effect
of concrete while few studies involve size effect behavior under dynamic loadings.
Concrete materials also have rate-dependent behavior (i.e. the so-called strain rate effect), this means that the mechanical
properties and the failure behavior of concrete show a significant discrepancy compared with those of concrete under static loads
[10–12]. Due to the limitations of the laboratory condition, only few efforts have been conducted on the mechanical behavior of
concrete under the coupled effect of size effect and strain rate effect. Krauthammer and Elfahal et al. [13,14] performed the dynamic
size effect tests of normal-and-high strength concrete with concrete cylinders (the maximum diameter was 600 mm) under the strain
rates of 0.014/s–3.03/s and found that the dynamic compressive strength decreases as the specimen size increases. Bindiganavile and
Banthia [15] designed concrete beams with three kinds of sizes (the maximum size was 150 × 150 × 450 mm) for three-points
bending impact tests under the impact velocity of 1.98–4.43 m/s, and discovered that the dynamic size effect of concrete is more
significant and the compressive response is more size-dependent than flexure response. However, some distinct viewpoints were
proposed by other scholars. Recently, Wang et al. [16] carried out the compressive tests with cylindrical concrete specimens (the
maximum size was 100 mm) on the size effect of layered roller compacted concrete under high-strain-rate loadings and the results
indicated that, the nominal compressive strength enhances with increasing the specimen size. Furthermore, the split Hopkinson
pressure bar test results on the dynamic compressive failure of cylindrical concrete specimens (the maximum size was 50 × 100 mm)
by Li et al. [17] show that the specimen size had an ignorable effect on the dynamic compressive strength while the concrete strength
grade greatly influences the dynamic compressive strength.
As mentioned efforts above, the investigations on dynamic size effect of concrete is far from consummate and the production
mechanism has not formed a unified cognition. Moreover, studies on the size effect law under dynamic loadings are almost blank.
Due to the limitations of test equipment and techniques, it is unrealistic to explore the dynamic size effect of concrete only by physical
tests, especially dynamic tests with large structural sizes under high strain rates. In the past decades, with the rapid development of
computer, numerical simulations have been applied by many researchers to investigate the mechanical behavior and static size effect
behavior of concrete [18–23]. Furthermore, mesoscopic modelling methods have also been employed to the simulations of me-
chanical properties of concrete and it has made some achievements [1,22–25].
The focus of the present study is on the dynamic size effect in splitting-tensile strength of concrete. The objectives are to: (1)
explore the size effect in dynamic splitting-tensile strength of concrete under different strain rates (10−5/s ∼ 200/s), and (2) propose
a Static and Dynamic unified Size Effect Law (i.e. SD-SEL) for splitting-tensile strength based on the numerical results.

2. Mesoscopic simulation method

Concrete materials have the size effect behavior and their global mechanical properties show the nonlinear behavior due to the
heterogeneity of mesoscopic composition [3,26]. Accordingly, the mesoscopic characteristics of concrete materials should be defi-
nitely described in the numerical studies. In this part, a 3D mesoscopic modeling method for exploring the dynamic size effect of
splitting-tensile strength of concrete would be presented in details.

2.1. Generation of 3D mesoscopic concrete models

In the mesoscopic simulations, taking account of the influence of concrete heterogeneities, the concrete was regarded as a tri-
phase composite material consisting of coarse aggregate, mortar matrix and the Interfacial Transition Zones (ITZs) between the
former two phases [18–22]. Referring to the previous efforts of many scholars [1,25,27], the random aggregate model which has been
widely demonstrated to research the failure process and deformation in fracture mechanics of concrete was also utilized in this study.
Herein, the coarse aggregate particles were assumed as spheres with random spatial distribution. The “take-and-place” method was
employed to generate coarse aggregate particles and the Fortran programming was selected to deliver aggregates to the matrix
randomly without overlapping based on the Monte Carlo method, which was also adopted in the efforts of [25,27,28]. The ITZs were
regarded as thin layers wrapping around the aggregate particles and they were endowed with an independent material (i.e. weakened
mortar) properties. In reality, the real typical thickness of ITZ is about 20–50 μm. However, at mesoscopic the computational amount
will be huge if the actual thickness of ITZ is utilized, especially for 3D models. Similar to the work [25], considering the calculation
amount, the thickness of ITZs was set as 2 mm without much loss of accuracy in 3D simulations. The 3D mechanical models were
divided by hexahedral linear isoparametric elements, and aggregates, mortar matrix and the ITZs were connected with each other by
sharing node.
The 3D mesoscopic cubic concrete models are depicted in Fig. 1. As seen in Fig. 1, mesoscopic models contain aggregates having
two kinds of equivalent grain sizes, namely, middle stone particles (diameter d = 30 mm) and small stone particles (d = 12 mm). The
concrete geometrical models involve about 40% of aggregate particles per volume.

318
L. Jin et al. Engineering Fracture Mechanics 209 (2019) 317–332

Fig. 1. 3D mesoscopic simulation model of concrete specimens and loading schematic diagram for splitting-tensile loadings.

The corresponding loading and boundary conditions for the mesoscopic models of concrete were introduced as follows:

(1) The vertical motion of the bottom edge of concrete models was constraint.
(2) In order to prevent the local crush, the rigid plate (whose width is a, thickness is 0.1a, and length is equal to the specimen length
b. a/b = 0.06.) was set on the middle area of the upper surface of concrete specimen. The contact surface of the rigid plate and
specimen was set as the coupling boundary.
(3) The upper surface of the rigid plate was set as loading area and was subjected to vertical compression by velocity v controlling.
The applied velocity v divided by the side length b of specimen is defined as nominal strain rate ε ̇ (ε ̇ = v/b). The other edges of
concrete models were left free.

2.2. Constitutive models

The plastic damaged model, which was first established by Lubliner [29] and improved by Lee and Fenves [30], is widely
employed in describing the static and dynamic mechanical behavior of concrete [24]. Tensile cracking and compressive crushing are
the main failure mechanisms of the concrete, which is the core of this model. The isotropic damage variables are conducted to
characterize the stiffness degradation of concrete caused by tensile and compressive failure. Due to the limited length of the paper,
more detailed description of this model can be found in the previous work of Du et al. [1,25].
Grote et al. [31] have proposed that the mechanical properties of mortar are similar to that of concrete. Due to the lack of test
results, the properties of the ITZ are still not well understood. It is well accepted that the ITZ has large heterogeneity, high porosity
and this is similar with mortar. Their mechanical properties could be described by the weakened mechanical parameters of mortar
[32]. Accordingly, referring to the works [27], herein the mechanical properties of aggregate, mortar and ITZs can be characterized
by this plastic damaged constitutive model. In addition, to discuss the influences of mash sizes on numerical models, concrete cubic
specimens having different mash sizes were simulated under static loadings and the numerical results are given in Fig. 2. It can be
seen that the mash size of mesoscopic model basically has no effect on the damage strength of specimen and models having small
mash sizes have meticulous failure patterns. Therefore, the mesh sizes of all models in this study were set as 1 mm.
It should be noted that for the purpose of preserving the well posedness of the boundary value problem and include a char-
acteristic length of microstructure for simulations of a deterministic size effect, a nonlocal theory should be employed as a reg-
ularization technique, which is similar to the effort [33]. Herein the characteristic length is lacking in the constitutive model. The

Fig. 2. Numerical results of concrete cubic specimens (structural size b = 100 mm) having different mesh sizes under static loadings.

319
L. Jin et al. Engineering Fracture Mechanics 209 (2019) 317–332

tensile post-failure behavior is given in terms of a fracture energy cracking criterion by specifying a stress-displacement curve instead
of a stress-strain curve to avoid or relax unreasonable mesh sensitive results [1,34]. However, the use of a fracture energy approach
only provides mesh-independent results for load-displacement diagrams.

2.3. Strain rate effect

Dilger et al. [35] and Bischoff et al. [36] have carried out the tests on dynamic mechanical properties of concrete materials. Their
test results have indicated that, comparing with tensile and compressive strengths, the other mechanical parameters like Young’s
modulus, fracture energy and Poisson’s ratio are less strain rate sensitivity. Similar to the treatment in [19,20,25], herein the strength
enhancement behavior of the meso-components, which was expressed by dynamic increase factor (i.e. the DIF, dynamic strength/
quasi-static strength), was the only consideration. In addition, the strain rate effect for the meso-components on material strength was
incorporated into the previous plastic damaged constitutive model.
In addition, the code by the Comite Euro-international du Beton (CEB) [37] has various values of the DIF on the compressive and
tensile strengths under different strain rates based on many available experiment results. Although the CEB gave an empirical
formula, it underestimated the DIF of concrete tensile strength. Malvar and Ross [38] developed the empirical formula and it can be
expressed as:
δ
ftd ε̇
TDIF = = ⎛ d ⎞ for εḋ ≤ 1s−1
⎜ ⎟

fts ⎝ tṡ ⎠
ε (1a)

1/3
ftd ε̇
TDIF = = β⎛ d⎞
⎜ ⎟ for εḋ > 1s−1
fts ⎝ εtṡ ⎠ (1b)

in which TDIF = tensile dynamic increasing factor, ftd and fts respect respectively the dynamic uniaxial tensile strength at the
strain rate εḋ and the quasi-static uniaxial tensile strength, εtṡ = 10−6 s−1 (static strain rate), logβ = 6δ − 2, δ = 1/(1 + 8fc′/fc0′), fc0′
= 10 MPa, and fc′ = the quasi-static uniaxial compressive strength of concrete (in MPa).
It should be informed that the strain rate effect for the aggregate, mortar matrix and the ITZ are not the same in reality. However,
some available test results in [31] discovered that the DIF for mortar is similar to that for concrete. The limited researches [39] also
demonstrated that the DIF of aggregate can be in the similar range of that of concrete. Due to the lack of relative tests, the ITZ is also
assumed to have the same DIF. Therefore, here in the strain rate sensitive model, the concrete strengths are assumed to be the static
strengths multiplied and the DIFs of the meso-components are referred by the corresponding empirical DIFs in Eq. (1). Similar
treatments can be found in [19,20,27].

2.4. Verification of the 3D mesoscopic models

Du et al. [40] have conducted the tests on the splitting-tensile size effect of concrete cubic specimens under static loadings
recently. In this part, the splitting-tensile tests were numerically simulated at mesoscopic for the validation of the above mesoscopic
modelling approach. In the tests of Du et al. [40], the macroscopic mechanical behavior of concrete cubic specimens having three
kinds of structural sizes (b = 150, 250 and 350 mm) under static loadings were investigated, and the splitting-tensile strengths were
obtained. The Young’s modulus, compressive strength and tensile strength, as well as the fracture energy of aggregate particles and
mortar matrix are listed in Table 1. Actually, it is difficult to define the mechanical parameters of ITZs. Similar to the treatment in
literatures [25,27], the mechanical parameters of ITZs were determined by the inversion method in this study. The inversion method
is a repeated trial algorithm. It means that a series of mechanical parameters for ITZs were involved to simulate the macroscopic
behavior of concrete and compare with that tested in the experiment. Then the parameters with which the closest results can be
obtained were adopted. After large quantity of repeated numerical trials by the inversion method, when the mechanical parameters
shown in Table 1 were adopted in simulations, the failure patterns and nominal splitting-tensile strength of concrete cubic specimens
having different structural sizes under static loadings were obtained. Topically, the comparison of the numerical and test results of

Table 1
Mechanical parameters of the concrete meso-components used in the study.
Parameter Aggregate particle Mortar matrix ITZ

Young’s modulus, E (GPa) 50a 30a 26b


Poisson ratio, ν 0.16 0.20 0.22
Dilatant angle, ψ (°) 30 30 30
Eccentricity, η (%) 0.1 0.1 0.1
Stress ratio, fb0 = fc0 1.16 1.16 1.16
K 0.667 0.667 0.667
Fracture energy Gc (J/m2) 60 50 30
Compressive strength, σc (MPa) 130a 30a 28.4c
Tensile strength, σt (MPa) 10a 3a 2.8c

Note: aData were obtained from the tests [40], bData were got from Jin and Du [1,25], and cData were obtained by the inversion method.

320
L. Jin et al. Engineering Fracture Mechanics 209 (2019) 317–332

Fig. 3. Comparison of the numerical and test results of the concrete cubic specimens.

the concrete cubic specimens are present in Fig. 3. It can be seen that the main crack took place in the middle line of cubic specimen
in simulations and this main crack was in good agreement with the crack in tests. As shown in Fig. 3(b), on the failure surface there
were some aggregate particles split by cracks, which were similar with the failure surface in the tests. In addition, the splitting-tensile
strengths of numerical models having different structural sizes were approaching those of test specimens, which was described in
Fig. 3(c). These comparisons of the numerical test results indicate that the above-mentioned meso-mechanical approach can well
describe the splitting-tensile behavior of concrete material and the validated 3D mesoscopic modelling method would be applied in
the following numerical study to explore the dynamic size effect of concrete splitting-tensile strength.

3. Simulation results and discussions

In order to explore the influence of strain rates on the dynamic splitting-tensile failure behavior and size effect of dynamic
splitting-tensile strength, the failure behavior of concrete cubic specimens having four different structural sizes (the side length
b = 100, 150, 300 and 450 mm) under different strain rates (the applied strain rates areε ̇ = 10−5/s (quasi-static), 10−3/s, 10−1/s, 1/
s, 10/s, 100/s and 200/s) were simulated by the above validated mesoscopic simulation method. To avoid the discreteness of
numerical simulations as much as possible, taking into account the amount of calculation, three cubic specimens for simulation in
every case were conducted and a total of 84 concrete cubic models were simulated. In the mesoscopic simulations, the mechanical
parameters of the meso-components, the aggregate content and the maximum size of aggregate particles were kept as identical.

3.1. Failure process

Fig. 4 depicts the splitting-tensile damage process of concrete cubic specimen (structural size b = 100 mm) under different strain
rates in simulations. It can be clearly seen that with the increase of strain rate the time required for the specimen reaching damage is
getting shorter. The damage generated firstly at the internal region having relatively weak mechanical properties (the ITZs), and the
damage propagated to mortar matrix or aggregates as time goes on. Under the strain rates ofε ̇ = 10−3/s and ε ̇ = 1/s, the crack started
to appear along the central line below load areas and then extended continuously up and down. Finally, many cracks existed

321
L. Jin et al. Engineering Fracture Mechanics 209 (2019) 317–332

Fig. 4. Splitting-tensile damage process of concrete cubic specimen (structural size b = 100 mm) under different strain rates.

throughout the whole specimen and separated the concrete specimen into approximately equal halves. Furthermore, some cracks
occurred on the surfaces parallel with the central line later under strain rate ε ̇ = 1/s in the end.

3.2. Final failure pattern

Typical final failure patterns of concrete cubic specimen (structural size b = 100 mm) under different strain rates in simulations
are exhibited in Fig. 5. From Fig. 5(a) it can be seen that the area near upper and lower surfaces (Area A) is in a state of triaxial
compression while Area B is in a State of tension-compression, which resulting that much damage appeared in Area B first and
specimens begin to fracture in this area. As given in Fig. 5(b), damage gradually develops from the outside of specimen to the inside
and the damage outside the specimen is more serious than the inside. Viewed from the lateral view, the number and width of cracks
near central line increase with the increasing strain rate. After the strain rate ε ̇ reaches 1/s, a fracture belt is formed gradually in the
middle of specimen. The cracks begin to appear on both sides of the specimen and continue to expand. One can find from Fig. 5(d)
that under low(er) strain rates few aggregate particles gets damaged while large amount of aggregate particles begins to be pulled off
when the value of strain rate reaches 1/s and exceeds 1/s, this may be because that the cracks could not bypass the aggregates in a
short time and extend along the shortest path under high strain rates. Fig. 6 presents typical failure patterns of concrete cubic
specimens having four kinds of structural sizes under different strain rates. One can note that the failure patterns are closely similar
for the specimens having different sizes under the same strain rate. The fracture belt composed of many cracks near central line is
widening by degrees with the increasing strain rate. This large splitting-tensile fracture belt has also been observed in the available
tests [41], as also presented in Fig. 6. In addition, due to large vertical deformation in the middle region of the specimens, bending-
tensile cracks also occurred on the side surfaces parallel with the central line when the magnitude of the strain rate is larger than 1/s.
Damage occurred in most areas of the concrete specimen, the overall stiffness of specimen softens, and even local comminuted
damage shows up finally under high strain rates.

3.3. Load-displacement relations

Fig. 7 plots Load(P) -displacement(δ) curves of concrete specimens having different structural sizes under different strain rates.
The load is the force applied to the loading area and displacement(δ) is the deformation at the loading area. It can be seen from Fig. 7

322
L. Jin et al. Engineering Fracture Mechanics 209 (2019) 317–332

Fig. 5. Typical final failure patterns of concrete cubic specimen (structural size b = 100 mm) under different strain rates.

that with the increase of structural size the peak load increases obviously under different strain rates. As shown in Fig. 8, the peak
load also enhances and the displacement corresponding to the peak load grows as strain rates increases. Furthermore, the higher
strain rate is, the greater stiffness of specimens at early stage of loadings is.

4. Dynamic size effect on splitting-tensile strength

4.1. Dynamic splitting-tensile strength

To obtain the nominal splitting-tensile strength, according to Chinese Standard for Test Method of Mechanical Properties of
Ordinary Concrete [42], the measured load of loading area can be converted to the splitting-tensile stress fst (nominal strength) as
2Pu
fst =
πA (2)
2
where Pu = the applied peak load (N); and A = the area of splitting failure surface (mm ).
The average splitting-tensile strengths of concrete cubic specimens having different structural sizes under different strain rates are
also listed in Fig. 9. Fig. 9 describes the relationship between splitting-tensile strength of concrete having different structural sizes
under different strain rates. One can find that the splitting-tensile strength of specimens enhances obviously with increasing the strain

323
L. Jin et al. Engineering Fracture Mechanics 209 (2019) 317–332

Fig. 6. Typical failure patterns of concrete cubic specimens having four kinds of structural sizes under different strain rates.

Fig. 7. Load(P) – displacement(δ) curves of loading area of concrete specimens having four kinds of structural sizes under different strain rates.

324
L. Jin et al. Engineering Fracture Mechanics 209 (2019) 317–332

Fig. 8. Load(P) – displacement(δ) curves of loading area of concrete specimen (structural size b = 100 and 300 mm) under different strain rates.

Fig. 9. Relationship between splitting-tensile strength of concrete having different structural sizes under different strain rates.

rate, especially under high strain rates. This shows the significant behavior of strain rate effect. In addition, concrete specimens with
large(r) structural sizes are more sensitive to the strain rate than those with small(er) structural sizes.
In order to verify the rationality of the splitting-tensile strengths obtained from simulations, the comparison of the TDIFs (herein,
the TDIF represents the ratio of dynamic splitting-tensile strengths and static splitting-tensile strengths in the same structural size)
between numerical results and the available test results [12,37,43–48] for concrete specimens having different structural sizes under
different strain rates is made in Fig. 10. It can be seen that the data points of TDIF obtained from the present simulations are basically
within the range of the available test results, which further illustrates the rationality of mesoscopic modelling approach and the
selected mechanical parameters of meso-components. Furthermore, the TDIF increases slowly under low strain rates while the TIDF
shows a remarkable growth after the strain rate exceeds 1/s. That is to say, after the strain rate exceeds 1/s the strain rate hardening
effect is obviously improving, resulting in great enhancement in dynamic splitting-tensile strength. These findings are consistent with
the studies of Antoun [44] and Birkimer et al. [46]. Herein the strain rate ε ̇ = 1/s is taken as the critical strain rate εċ r , the strain
ratesε ̇ < 1/s are defined as low strain rates and the strain rates ε ̇ > 1/s are defined as high strain rates.

4.2. Dynamic size effect

Fig. 11 reports the change of dynamic splitting-tensile strength of concrete specimens having different structural sizes under
different strain rates. One can note that under low strain rates with the addition of structural size the splitting-tensile strength
gradually decreases and the concrete shows a size effect behavior. However, the decrease speed of the splitting-tensile strength
(described by the linear fitting slope k) is getting slower as the strain rate increases, which illustrates that the size effect behavior of
splitting-tensile strength is restrained and weakened gradually. Even under the critical strain rate εċ r the slope k approaches to zero,
indicating that the corresponding size effect behavior is totally suppressed. Moreover, under high strain rates, the size effect behavior
of concrete shows up again and it is completely different with that under low strain rates. The splitting-tensile strength enhances as
the strain rate improves and the discrepancy in splitting-tensile strength of specimens having different structural sizes tends to
increase gradually. Similar conclusions were also reported in the investigations of [16,27].

325
L. Jin et al. Engineering Fracture Mechanics 209 (2019) 317–332

Fig. 10. Comparison of TDIF between numerical results and existing experimental results.

Fig. 11. Relationship between dynamic splitting-tensile strength and structural size of concrete having different strain rates.

4.3. Preliminary analysis on dynamic size effect

According to above discussion, the dynamic splitting-tensile strength of concrete material shows significantly different size effect
behavior under different strain rates. Here in this part, the mechanism of dynamic size effect is under preliminary analysis. As known,
the characteristic of micro- and meso-structure has an important influence on the static size effect. That is to say, the heterogeneity of
mesoscopic composition leads to the nonlinearity of mechanical properties [3,26].
Under low strain rates, especially under quasi-static condition, internal stress distribution of concrete is relatively uniform under
loadings. The weakest region of mechanical properties (like the ITZs) first reaches the ultimate strength and cracks appear. Then
cracks have enough time to expand by bypassing aggregates and splitting-tensile damage cracks through the specimen form finally.
The larger the specimen size is, and the larger region of poor mechanical properties in concrete is, causing the obvious size effect.

326
L. Jin et al. Engineering Fracture Mechanics 209 (2019) 317–332

With the strain rate increases, the pore-water pressure of concrete enhances and the viscous effect of free water slows down the
propagation and expansion of cracks. The lateral restraint effect also enhances gradually and puts concrete specimens be in a complex
stress state, leading to suppressing or delaying the expansion of macro cracks. In addition, more and more cracks form and expend.
The crack path and number increase as the specimen size improves. The stress distribution within concrete becomes uneven pro-
gressively and the effect of internal defects on the dynamic splitting-tensile strength gradually declines. At same time, the effect of
inertial force on the enhancement of splitting-tensile strength gradually increases. These factors commonly cause the piecemeal
suppression of dynamic size effect behavior of concrete material under low strain rates. Even size effect behavior disappears at the
critical strain rate.
Under high strain rates, concrete specimens receive a huge impact and enough energy in a short time, making that the failure
pattern has great changes. The high-speed impact brings more energy, which needs to be released in a short time and internal cracks
of concrete have not enough time to develop. The restriction effect of concrete significantly enhances, making concrete in a highly
constrained, highly complex state of stress. Therefore, the splitting-tensile strength improves obviously as the strain rate exceeds the
critical strain rate. Moreover, the internal defects of concrete basically have no effect on the dynamic size effect and the inertial force
of concrete greatly enhances as the increasing strain rate. The magnitude of the inertial force is related to the structural size of
specimens, and the larger the size, the stronger the inertia. That is to say, the inertial force of concrete plays a leading role in the
influence of dynamic size effect, resulting that with the addition of structural size the splitting-tensile strength gradually enhances
under high strain rates.

5. Dynamic size effect Law

There are many theoretical analyses on size effect of concrete material based on different study frameworks [26,49–51], which
are widely approbated by lots of scholars. Among these theories, the size effect law (SEL) according to Bažant [26,51] is often
employed to well describe size effect behavior of concrete duo to their simplicity.

5.1. Contrast with Bažant’s size effect Law

In many papers, Bažant et al. have proposed the existence of three types of size effect in concrete structures, classified as Type I,
Type II, and Type III, respectively [26,51,52]. According to Bažant [51], Type I size effect occurs in structures of positive geometry
having no notches or preexisting large cracks while Type II size effect applies to geometrically similar specimens with similar deep
notches or to structures in which a large crack (similar for different sizes) forms stably before the maximum load is reached. In the
present study, the cracks occurring in concrete specimens under splitting-tensile loadings are significantly larger than the fracture
process zone. Therefore, here in this part, the Type II size effect is expected to dominate and the size effect law proposed by Bažant
[26,51] based on the linear elastic fracture mechanics was applied to characterize the size effect behavior of splitting-tensile strength
obtained by simulations. The SEL according to Bažant has the following form:

Bf st'
σNu =
1 + D / D0 (3)
where σNu represents the static nominal strength of concrete; D represents the structural size; and fst′ represents the splitting-tensile
strength of concrete (herein fst′ = σ0' ). B represents the dimensionless geometry-dependent parameter and D0 represents the size-
dependent parameter called transitional size. These two empirical parameters (B and D0) are identified by regression analysis based
on numerical results.
Eq. (3) can be written by mathematical transformation as:
' 2
⎛ f st ⎞ D 1
⎜ σN u ⎟ = D0 B2 + B2
⎝ ⎠ (4)
Eq. (4) can be depicted as a linear form:
Y = AX + C (5)
where Y = (fst′/σNu)2, X = D, A = 1/D0B2, and C = 1/B2. A and C are obtained from a regression analysis based on numerical results.
The determined parameters of Bažant’s SEL under different strain rates are listed in Table 2. It can be seen that the parameter B shows
a slow growth under low strain rates while it strengthens rapidly after the applied strain rate exceeds 1/s. This finding is similar with
the variety of TDIF discussed previously. Moreover, the parameter D0 improves with the enhancement of strain rate and it becomes a
negative number under high strain rates, illustrating that Bažant’s SEL is no longer suitable for the size effect behavior of splitting-
tensile strength under high strain rates.
Fig. 12 plots the comparison between the splitting-tensile strength obtained from simulation and Bažant’s SEL. As shown in
Fig. 12, the data points of numerical splitting-tensile strength under strain rates and critical strain rate are approaching the curve of
Bažant’s SEL. The data points of numerical splitting-tensile strength under quasi-static loading (ε ̇ = 10−5/s) are the closest to the
Linear Elastic Fracture Mechanics (i.e. LEFM), showing an obvious size effect behavior. With the increasing strain rate, the data points
gradually close to the strength criterion curve (the horizontal line), indicating the size effect behavior is suppressed by degrees.
However, the data points of numerical splitting-tensile strength under high strain rates is far from the strength criterion curve

327
L. Jin et al. Engineering Fracture Mechanics 209 (2019) 317–332

Table 2
. Parameters of B and D0 under different strain rates.

Strain rates ε ̇ (s−1) A C Bfst′ (MPa) B D0(mm)

10−5 2.68 × 10−4 1.45 × 10−1 2.62705 1.07139 541


10−3 1.86 × 10−4 1.49 × 10−1 2.59312 1.05755 798
10−1 9.22 × 10−5 1.29 × 10−1 2.78164 1.13444 1402
1 6.31 × 10−6 9.11 × 10−2 3.31371 1.35143 14,441
10 −7.40 × 10−7 9.39 × 10−3 7.73917 3.15627 −12698
100 −3.48 × 10−7 1.09 × 10−3 22.75794 9.28138 −3122
200 −2.60 × 10−7 6.91 × 10−4 28.53094 11.63578 −2656

Note: fst′ is the splitting-tensile strength of the concrete specimen (sized by 100 mm) under the strain rate of 10−5/s, fst′ = 2.4519 MPa.

Fig. 12. Comparison between the splitting-tensile strength obtained from simulation and Bažant’s SEL.

gradually, demonstrating that the size effect behavior opposed to that under low strain rates exhibits progressively.
Based on the effort of Jin [27] previously and refer to the SEL of Bažant, here in this part the curve well describing the size effect
behavior under high strain rates was determined. The expression of Bažant’s SEL is the double logarithmic curve
(lg(σNu/ Bf c' )=lg(1 + D / D0)−1/2 ) and the data points of splitting-tensile strength under low strain rates vary nearby this curve. When
applied strain rate approaches the critical strain rate the strengths concentrate in the nearby area of the strength criterion curve.
When applied strain rate is very small (like quasi-static), the strengths close to the strength limit (LEFM, its slope is −1/2). Similarly,
when applied strain rate exceeds the critical strain rate, as the strain rate continues to grow the strengths are getting closer to another
strength limit and the data points may also vary along a certain double logarithmic curve, which can well describe the size effect
behavior opposed to that under low strain rates.
Herein, under the assumption of that the strengths close to the strength limit with a slope of m at the ultimate strain rate, the
double logarithmic curve under high strain rates can be written as:

lg(σNu/ Bf c' )=lg(1 + D / D0)m (6)

where m represents the approximation coefficient reflecting the slope of the strength limit, with which the curve is asymptotic at
the ultimate strain rate and it needs to be obtained by repeated trials and comparisons with numerical data points. As depicted in
Fig. 12(b), when m = 0.5, the data points of numerical splitting-tensile strength under strain rates agree well with the double
logarithmic curve. That is to say, the double logarithmic curve (lg(σNu/ Bf c' )=lg(1 + D / D0)1/2 ) can accurately characterize the size
effect behavior (the splitting-tensile strength enhances with the increasing sizes). The m is defined as 0.5, which is also the in the
range recommended in the work of Jin [27]. Eq. (6) can be expressed as

σNu = Bf c' · 1 + D / D0 (7)

5.2. Static and dynamic unified size effect Law

According to the numerical results and discussion above, the strain rate effect has a palpable influence on the size effect behavior
of splitting-tensile strength in concrete and Bažant’s SEL behaves its limitation under high strain rates. A Static and Dynamic unified
Size Effect Law (i.e. SD-SEL) is urgent to be established taking account of the strain rate effect.
The influence of strain rate on the size effect is reflected in two aspects: 1) the enhancement of splitting-tensile strength. It can be
characterized by the coefficient of strength enhancement φε .̇ Hereφε ̇ = TDIF. 2) Suppression and reverse enhancement of size effect.
the influence coefficient βε ̇ of strain rate was applied to describe this behavior. Based on these two aspects, the Static and Dynamic

328
L. Jin et al. Engineering Fracture Mechanics 209 (2019) 317–332

Fig. 13. Fitting curve for the increase of dynamic splitting-tensile strength.

unified Size Effect Law can be given as:

Bf st'
σNε ̇ u = ·φε ·̇ βε ̇
1 + D / D0 (8)
where σNε ̇ u
represents the nominal splitting-tensile strength, B, D0 represent two empirical parameters obtained by regression analysis
under quasi-static loading, f st' represents the splitting-tensile strength of concrete standard specimens (sized by 100 mm herein) under
quasi-static loading, and the coefficient φε ̇can be determined by the TDIF fitting formula of concrete standard specimens.
As depicted in Fig. 13, One can obtain the fitting segmented formula of the splitting-tensile strength of concrete standard spe-
cimens (sized by 100 mm) as following:
TDIFfit = 0.0058(lgε )̇ 2 + 0.06391lgε ̇ + 1.1740, (ε ̇ ≤ εcṙ = 1/s) (9)

TDIFfit = 2.0615(lgε )̇ 2 − 0.1571lgε ̇ + 1.3385, (ε ̇ > εcṙ = 1/s) (10)

5.3. The influencing coefficient βε ̇ of strain rate effect

The emphasis in the SD-SEL is the determination of the influencing coefficient βε ̇ of strain rate effect. The relationship between
βε and
̇ lg(ε )̇ is given in Fig. 14. From Fig. 14 it can be seen that as the applied strain rate below the quasi-static loading corresponding
strain rate (ε ̇ ≤ εstatic
̇ = 10−5/s ), the size effect behavior is not affected by the strain rate, therefore βε ̇ = 1.
At the critical strain rate εċ r (εcṙ = 1/s ), the size effect behavior is entirely suppressed. Thus, the nominal splitting-tensile strength
σNε ̇ u should be written as

σNε ̇ u = f st' ·φε ̇ (11)


Simultaneous Eq. (8) and Eq. (11), the coefficient βε ̇ (> 1)under the critical strain rate εċ r is expressed as:

1 + D / D0
βε ̇ |ε =̇ εcṙ =
B (12)
As the applied strain rate ε ̇ is between εstatic
̇ and εċ r , linear or nonlinear curves can be applied to describe the variety of βε .̇ Similar
to the work of Jin [27], the comparison result indicates that the quadratic function curve was more accurate and the theoretical value

Fig. 14. Determination of the coefficient βε̇.

329
L. Jin et al. Engineering Fracture Mechanics 209 (2019) 317–332

is closer to the simulated results. There, a quadratic function curve was supposed and employed, and the coefficient βε ̇ can be defined
as:
βε ̇ = 1, ε ̇ ≤ 10−5/s (13a)

β =
ε̇
( 1 + D / D0
B )
− 1 ·(lgε )̇ 2 + 10· ( 1 + D / D0
B )
− 1 ·lgε ̇ + 25·(
1 + D / D0
B
− 1)
+ 1, 10−5/s < ε ̇ ≤ εcṙ
(lg(εcṙ ))2 + 10lg(εcṙ ) + 25 (13b)
Herein the critical strain rate εċ r = 1/s . Thus, Eq. (13b) simplifies to:

β =
ε̇
( 1 + D / D0
B )
− 1 ·(lgε )̇ 2 + 10· ( 1 + D / D0
B )
− 1 ·lgε ̇ + 25·(
1 + D / D0
B
− 1)
+ 1, 10−5/s < ε ̇ ≤ 1/s
25 (13c)
As the applied strain rate closes to the ultimate strain rate εlim
̇ , according to Eq. (7), the data points of dynamic splitting-tensile
strength close to the strength limit with a slope of 1/2 and the size effect behavior is not influenced by the strain rate. The coefficient
βε ̇ infinitely approached (1 + D / D0) . Furthermore, as the applied strain rate ε ̇ is between εċ r and εlim
̇ , the coefficient βε ̇ can be set as a
curve asymptotic to (1 + D / D0) . As depicted in Fig. 14, a hyperbolic tangent function was utilized to describe the effect of strain rate
temporarily and based on Eq. (12), the coefficient βε ̇ can be expressed as:

1 + D / D0 1 + D / D0
βε ̇ = [(1 + D / D0) − ]·tanh(α·lgε )̇ + , (ε ̇ > 1/ s )
B B (14)
where α represents the adjustment coefficient, indicating the change speed of βε ̇ with the strain rate. In fact, Eq. (14) is a
transcendental equation and it is hard to determine its analytical solution. A trial algorithm was employed to approximate the
numerical solution in the following part.

5.4. Validation of the static and dynamic unified size effect law

The numerical results obtained previously were adopted to calibrate the proposed SD-SEL on splitting-tensile strength of concrete.
At the strain rate ε ̇ = 10−5/s (quasi-static), these two empirical parameters B and D0 (B = 1.07139, D0 = 541) are determined and
utilized. The expression of coefficient φε ̇ is given according to Eq. (9) and Eq. (10) and the coefficient βε ̇ is obtained by Eq. (13c) and
Eq. (14). Based on these conditions, after a large number of trials when the adjustment coefficient α is defined as 0.05, one can predict
the magnitude of the dynamic splitting-tensile strength of concrete having different structural sizes. Fig. 15 describes the predicted
results based on the proposed SD-SEL against with the numerical data points. It can be obviously noted that, most of the numerical
strength points fall on the surface of the theoretical splitting-tensile strength, illustrating that the proposed SD-SEL can well char-
acterize the size effect behavior on the splitting-tensile strength of concrete material under different strain rates.
In order to compare with the test results and the proposed SD-SEL, experimental data in Chen et al.’s study [53] were adopted to
verify the theory. The comparison of test results in [53] and theoretical results is described in Fig. 16. For convenience, fst/fstR ratio
was applied, where fst denotes the splitting-tensile strength of specimens having different sizes under different strain rates, and fstR
denotes the splitting-tensile strength of referent specimens under quasi-static loadings (fstR = the splitting-tensile strength of referent
specimen sized by 30 mm under static loadings in Chen et al.’s study [53] while in theoretical results fstR = the splitting-tensile
strength of cubic specimen with side length of 30 mm calculated by the proposed SD-SEL). It can be seen from Fig. 16 that, although
some test data in [53] are slightly below the theoretical surface based on the proposed SD-SEL, most tests results approach or fall on
the theoretical surface, which preliminarily verifies the proposed SD-SEL.

Fig. 15. Comparison of the dynamic splitting-tensile strengths obtained from simulations and dynamic size effect theory.

330
L. Jin et al. Engineering Fracture Mechanics 209 (2019) 317–332

Fig. 16. Comparison of the dynamic splitting-tensile strengths obtained from tests in [53] and dynamic size effect theory.

It should be mentioned that comparing the theoretical result with one group of experimental data is far from enough. There are
still lacking of efficacious test results in dynamic size effect on the splitting-tensile strength of concrete, especially under high strain
rates and large structural sizes. Therefore, more relevant tests on the dynamic size effect of concrete should be carried out to further
verify the proposed SD-SEL on the splitting-tensile strength in the future studies.

6. Conclusions

Concrete materials have the size effect behavior and their global mechanical properties show the nonlinear behavior due to the
heterogeneity of mesoscopic composition. Considering the effect of concrete heterogeneity and the strain rate effect, a mesoscopic
mechanical model of concrete cubic specimen for the splitting-tensile failure and size effect at different strain rates (10−5/s ∼ 200/s)
was set up and applied. Furthermore, based on the numerical results and combined the influence of the strain rate on size effect
behavior, a static and dynamic unified size effect law was proposed. The conclusions can be drawn as following:

1) The size effect behavior on the splitting-tensile strength of concrete under dynamic loadings has an obvious discrepancy compared
with that under static loadings.
2) There is a critical strain rate (εcṙ = 1/s ) in dynamic splitting-tensile strength of concrete. As the applied strain rate below the
critical strain rate the size effect behavior is restrained and weakened gradually with the addition of strain rate. As the applied
strain rate exceeds the critical strain rate the dynamic strength enhances linearly as the structural size increases and the enhanced
trend of the strength increases as the strain rate improves.
3) Considering the contribution of strain rate on the dynamic size effect, the proposed SD-SEL can well characterize the size effect
behavior on the splitting-tensile strength of concrete material under different strain rates.

Acknowledgement

This research was supported by the National Key Basic Research and Development Program of China (No. 2018YFC1504302; No.
2016YFC0701104) and the National Natural Science Foundation of China (No. 51822801; No. 51421005). The support is gratefully
acknowledged.

References

[1] Du XL, Jin L, Ma GW. A meso-scale analysis method for the simulation of nonlinear damage and failure behavior of reinforced concrete members. Int J Damage
Mech 2013;22(6):878–904.
[2] Häfner S, Eckardt S, Luther T, Könke C. Mesoscale modeling of concrete: geometry and numerics. Comput Struct 2006;84(7):450–61.
[3] Du XL, Jin L, Li D. A state-of-the-art review on the size effect of concretes and concrete structures (I): concrete materials. J Chin Civil Eng 2017;9:28–45.
[4] Du XL, Jin L, Li D. A state-of-the-art review on the size effect of concretes and concrete structures (II): RC members. J Chin Civil Eng 2017;50(11):24–44.
[5] Ross CA. Split-Hopkinson Pressure-Bar Tests on concrete and mortar in tension and compression. J ACI Mater 1989;86(5):475–81.
[6] Hasegawa T, Shioya T, Okada T. Size effect on splitting tensile strength of concrete. JCI conference. Japan Concrete Institute; 1985. p. 309–12.
[7] Ince R, Gör M, Eren ME, Alyamaç KE. The effect of size on the splitting strength of cubic concrete members. Strain 2015;51(2):135–46.
[8] Bažant ZP, Kazemi MT, Hasegawa T, Mazars J. Size effect in Brazilian split-cylinder tests. Measurements and fracture analysis. J ACI Mater 1991;88(3):325–32.
[9] Zhou FP, Balendran R, Jeary A. Size effect on flexural, splitting tensile, and torsional strengths of high-strength concrete. Cem Concr Res 1998;28(12):1725–36.
[10] Li QM, Meng H. About the dynamic strength enhancement of concrete-like materials in a split Hopkinson pressure bar test. Int J Solids Struct 2003;40(2):343–60.
[11] Zeng S, Ren X, Li J. Triaxial behavior of concrete subjected to dynamic compression. J Struct Eng 2013;139(9):1582–92.
[12] Yan DM, Lin G. Dynamic properties of concrete in direct tension. Cem Concr Res 2006;36(7):1371–8.
[13] Krauthammer T, Elfahal MM, Lim J, Ohno T, Beppu M, Markeset G. Size effect for high-strength concrete cylinders subjected to axial impact. Int J Impact Eng

331
L. Jin et al. Engineering Fracture Mechanics 209 (2019) 317–332

2003;28(9):1001–16.
[14] Elfahal MM, Krauthammer T. Dynamic size effect in normal-and high-strength concrete cylinders. J ACI Mater 2005;02(2):77.
[15] Bindiganavile V, Banthia N. Size effects and the dynamic response of plain concrete. J Mater Civil Eng 2006;18(4):485–91.
[16] Wang XH, Zhang SR, Wang C, Song R, Shang C, Fang X. Experimental investigation of the size effect of layered roller compacted concrete (RCC) under high-
strain-rate loading. Constr Build Mater 2018;165:45–57.
[17] Li M, Hao H, Shi Y, Hao Y. Specimen shape and size effects on the concrete compressive strength under static and dynamic tests. Constr Build Mater
2018;161:84–93.
[18] Hao Y, Hao H, Li ZX. Numerical analysis of lateral inertial confinement effects on impact test of concrete compressive material properties. Int J Protective Struct
2010;1(1):145–67.
[19] Zhou XQ, Hao H. Modelling of compressive behaviour of concrete-like materials at high strain rate. Int J Solids Struct 2008;45(17):4648–61.
[20] Zhou XQ, Hao H. Mesoscale modelling of concrete tensile failure mechanism at high strain rates. Compu Struct 2008;86(21–22):2013–26.
[21] Snozzi L, Caballero A, Molinari JF. Influence of the meso-structure in dynamic fracture simulation of concrete under tensile loading. Cem Concr Res
2011;41(11):1130–42.
[22] Pedersen RR, Simone A, Sluys LJ. Mesoscopic modeling and simulation of the dynamic tensile behavior of concrete. Cem Concr Res 2013;50:74–87.
[23] Wang L, Bao J. Mesoscale computational simulation of the mechanical response of reinforced concrete members. Comput Concr 2015;15(2):305–19.
[24] Grassl P, Grégoire D, Solano LR, Pijaudier-Cabot G. Meso-scale modelling of the size effect on the fracture process zone of concrete. Int J Solids Struct
2012;49(13):1818–27.
[25] Du XL, Jin L, Ma GW. Numerical simulation of dynamic tensile-failure of concrete at meso-scale. Int J Impact Eng 2014;66:5–17.
[26] Bažant ZP, Planas J. Fracture and size effect in concrete and other quasi-brittle materials. CRC Press; 1997.
[27] Jin L, Yu WX, Du XL, Zhang S, Li D. Meso-scale modelling of the size effect on dynamic compressive failure of concrete under different strain rates. Int J Impact
Eng 2019;125:1–12.
[28] Wriggers P, Moftah SO. Mesoscale models for concrete: homogenisation and damage behavior. Finite Elem Anal Des 2006;42(7):623–36.
[29] Lubliner J, Oliver J, Oller S, Onate E. A plastic-damage model for concrete. Int J Solids Struct 1989;25(3):299–326.
[30] Lee J, Fenves GL. Plastic-damage model for cyclic loading of concrete structures. J Eng Mech 1998;124(8):892–900.
[31] Grote DL, Park SW, Zhou M. Dynamic behavior of concrete at high strain rates and pressures: I. experimental characterization. Int J Impact Eng
2001;25(9):869–86.
[32] Kim SM, Al-Rub RKA. Meso-scale computational modeling of the plastic-damage response of cementitious composites. Cem Concr Res 2011;41(3):339–58.
[33] Bažant ZP, Jirasek M. Nonlocal integral formulations of plasticity and damage: survey of progress. J Eng Mech 2002;128(11):1119–49.
[34] Huang YJ, Yang ZJ, Chen XW, Liu GH. Monte Carlo simulations of meso-scale dynamic compressive behavior of concrete based on X-ray computed tomography
images. Int J Impact Eng 2016;97:102–15.
[35] Dilger WH, Koch R, Kowalczyk R. Ductility of plain and confined concrete under different strain rates. J Proc 1984;81(1):73–81.
[36] Bischoff PH, Perry SH. Compressive behaviour of concrete at high strain rates. Mater Struct 1991;24(6):425–50.
[37] Euro-International Committee for Concrete. CEB-FIP Model Code 1990. London: Thomas Telford Services Ltd; 1991.
[38] Malvar LJ, Ross CA. Review of strain rate effects for concrete in tension. J ACI Mater 1998;95(6):735–9.
[39] Wang QZ, Li W, Song XL. A method for testing dynamic tensile strength and elastic modulus of rock material using SHPB. Pure Appl Geophys
2006;163:1091–100.
[40] Du M, Jin L, Li D, Du X. Mesoscopic simulation study of the influence of aggregate size on mechanical properties and specimen size effect of concrete subjected to
splitting tensile loading. Eng Mech 2017;34(09):54–63.
[41] Wu M, Qin C, Zhang C. High strain rate splitting tensile tests of concrete and numerical simulation by mesoscale particle elements. J Mater Civil Eng
2014;26(1):71–82.
[42] Standard for Test Method of Mechanical Properties of Ordinary Concrete, GB/T 50081-2002, Beijing, China; 2002.
[43] Ross CA, Tedesco JW. Split-Hopkinson pressure-bar tests on concrete and mortar in tension and compression. J ACI Mater 1989;86(5):475–81.
[44] Antoun TH. Constitutive/failure model for the static and dynamic behaviors of concrete incorporating effects of damage and anisotropy. Dayton OH: The
University of Dayton; 1991.
[45] Körmeling HA, Reinhardt HW, Zieliński AJ. Experiments on concrete under single and repeated uniaxial impact tensile loading. Stevin Report Delft 1980.
5–80-3.
[46] Birkimer DL, Lindemann R. Dynamic tensile strength of concrete materials. J ACI 1971;68(1):47–9.
[47] Wu H, Zhang Q, Huang F, Jin Q. Experimental and numerical investigation on the dynamic tensile strength of concrete. Int J Impact Eng 2005;32(1–4):605–17.
[48] Ranade R, Li VC, Heard WF. Tensile rate effects in high strength-high ductility concrete. Cem Concr Res 2015;68:94–104.
[49] Weibull W. A statistical theory of the strength of materials. Proc R Swedish Inst Eng Res 1939;151:1–45.
[50] Carpinteri A, Ferro G. Size effects on tensile fracture properties: a unified explanation based on disorder and fractality of concrete microstructure. Mater Struct
1994;27(10):563–71.
[51] Bažant ZP. Probability distribution of energetic-statistical size effect in quasibrittle fracture. Probabilistic Eng Mech 2004;19:307–19.
[52] Bažant ZP, Pang SD. Activation energy based extreme value statistics and size effect in brittle and quasibrittle fracture. J Mech Phys Solids 2007;55:91–131.
[53] Chen J, Xiang D, Wang Z, Wu G, Wang G. Dynamic tensile strength enhancement of concrete in split Hopkinson pressure bar test. Adv Mech Eng 2018;10(6):1–7.

332

You might also like