Sie sind auf Seite 1von 13

Engineering Fracture Mechanics 135 (2015) 113–125

Contents lists available at ScienceDirect

Engineering Fracture Mechanics


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

Determining the impact behavior of concrete beams through


experimental testing and meso-scale simulation: II. Particle
element simulation and comparison
Mingxin Wu a, Chuhan Zhang a,⇑, Zhenfu Chen b
a
State Key Laboratory of Hydroscience and Engineering, Tsinghua University, Beijing 100084, China
b
School of Urban Construction, University of South China, Hengyang 421001, China

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

Article history: A complete model of the hammer–transducer–concrete beam system using meso-scale
Received 28 June 2014 particle elements is proposed. Drop-weight tests are performed with a constant weight
Received in revised form 27 November 2014 and varying drop heights. The numerical results are compared with those from the impact
Accepted 6 December 2014
tests in Part I. The results of impact loads, deflections, and strains exhibit satisfactory
Available online 30 December 2014
agreement. The validated model is then used to predict dynamic behavior under higher
loading rates. Different fracture modes are observed in the simulation. The strain-fracture
Keywords:
energy concept is introduced to particle element modeling and the energy conversion
Meso-scale particle element model
Drop-weight tests
between kinetic and strain-fracture energy components are discussed for different strain
Rate effects rates.
Fracture mode Ó 2014 Elsevier Ltd. All rights reserved.
Energy conversion

1. Introduction

Experimental tests and numerical simulation are widely performed to study the dynamic characteristics of concrete.
Dynamic impact tests and analysis have been particularly preferred. The finite element method (FEM) is employed to study
the effects of loading rate on structural response [1–3], failure mode [4,5], and fracture process [6–8]. Rate-dependent con-
stitutive relationships, such as the cohesive model, are commonly used in FEM simulation. The discrete element method
(DEM) [9] and nonlinear mass-spring model [10] are also used for the numerical study of the dynamic behavior of normal
and fiber-reinforced concrete beams subjected to impact.
Numerical simulation using meso-scale models is a logical tool for studying the complex failure mechanisms of concrete
as a result of its heterogeneity. Previous studies have proposed various meso-scale models, such as FEM by Wittmann et al.
[11], truss model by Bažantet al. [12], lattice model by Schlangen and van Mier [13], and DEM by Azevedo and Lemos [14].
Meso-scale models have recently been adopted for extensive studies but with a main focus on static load simulation. The
particle element method (PEM) has been proven to be a powerful tool for explaining the fracture mechanisms of pseudo-
brittle material [15] and has thus been adopted in the dynamic meso-scale model of concrete for tensile and compressive
simulations [16,17].
A complete system of numerical simulation for drop-weight tests using meso-scale PEM is proposed in this study. The
method consists of not only the specimen but also the loading system, including the hammer, pad, and transducer. A series

⇑ Corresponding author. Tel.: +86 10 62787216.


E-mail address: zch-dhh@tsinghua.edu.cn (C. Zhang).

http://dx.doi.org/10.1016/j.engfracmech.2014.12.020
0013-7944/Ó 2014 Elsevier Ltd. All rights reserved.
114 M. Wu et al. / Engineering Fracture Mechanics 135 (2015) 113–125

Nomenclature

a vertical acceleration of particle


A area of the bond-cross section
CIF capacity increase factor
DIF dynamic increase factor
Ef strain-fracture energy in simulation
Efric frictional energy in simulation
Ek kinetic energy in simulation
Es strain energy in simulation
Et total energy in simulation
F bond force
Fn normal bond force
Fs shear bond force
I inertia memont
n
k normal bond stiffness per unit area
s
k shear bond stiffness per unit area
m mass of particle
M bond moment
N number of particles
P_ force rate
Pd generalized deforming force
Pi generalized inertial force
Pt total impact force
R bond radius
u vertical displacement of particle
Un normal displacement
Us shear displacement
u_ y velocity in the y-direction in test
vx particle velocity in the x-direction
vy particle velocity in the y-direction
Wf strain-facture energy in test
Wt total energy of the beam in test
Wk kinetic energy in test
d deflection at mid-span of the beam
e_ strain rate
ec compressive strain
et tensile strain
h rotation
q particle density
r c normal bond strength
sc shear bond strength
x rotational velocity of particle

of loading cases with a constant weight of 4 kg and varying drop heights from 0.5 m to 2.0 m that correspond to the drop-
weight tests are simulated. The time histories of impact forces, deflections at the mid-span of the beam, as well as tensile and
compressive strains all support the experimental test results. The validated model is then used to predict the dynamic
behavior of concrete beams at high loading rates. In the failure process, high capacity and different cracking profiles are
observed. The energy partition among the different portions of the complete system, as well as the conversion among kinetic
and strain-fracture energy components are discussed. The study concludes that the rate effects of concrete can be explained
by the large energy demand attributed to the inertial effect and fracture consumption at high strain rates.

2. Modeling of the complete test system

2.1. Brief description of PEM

PEM was originally proposed by Cundall and Strack [18] for modeling the mechanical behavior of granular materials. In
this method, the computational domain is discretized into rigid particles connected by normal and shear springs. The model
is capable of simulating a continuum system, assuming that the elements in contact are bonded together. Newton’s second
law is applied to each particle, and Force–displacement law is used at its contact with other elements, alternately. The
M. Wu et al. / Engineering Fracture Mechanics 135 (2015) 113–125 115

Fig. 1. Solution scheme of PEM.

Fig. 2. Parallel bond model of particle elements.

computational system cycles to maintaining equilibrium or approaching failure. The solution scheme of PEM is illustrated in
Fig. 1.
The parallel bond model is adopted in this study. As shown in Fig. 2, the two particles in contact, namely, particles A and
B, are bonded by a series of parallel springs that can transform both bond force F and moment M. When the particles exhibit
relative displacement DU n or DU s and rotation Dh, the increments in the bond force and moment carried by the parallel bond
are given by the following linear relationships:

DFn ¼ kn ADU n ð1Þ

DFs ¼ ks ADU s ð2Þ

DM  n I Dh
 ¼ k ð3Þ

where k n and k
s are the normal and shear bond stiffness per unit area, respectively; A ¼ 2R  is the area of the bond cross-sec-
tion, where R represents the bond radius; I ¼ 2R  3 =3 is the inertia moment.
When the normal or shear stress calculated by beam theory exceeds the corresponding strength r  c or s
c , the parallel bond
will break, indicating crack initiation in the material. The parallel bond model of particle elements are introduced in detail in
Refs. [19,20].

2.2. Pre-process approach and inverse method for meso-scale concrete modeling

The proposed model adopts a previously developed meso-scale model by PEM for three-phase concrete, i.e., aggregate,
mortar, and interface [16]. A pre-process scheme for generation of the specimen of multiphase concrete and an inverse
method for determination of the meso-mechanical parameters are briefly introduced below.
The pre-process approach consists of two steps, i.e., the generation of geometric coordinates for aggregate and the iden-
tification of particles and bonds for the three phases. The detailed approach may be found in Refs. [16,21].
The meso-parameters can be calculated by using the inverse method based on the typical macro-mechanical parameters
from laboratory tests. The relationship between meso- and macro-mechanical parameters has been discussed in [15,16].
Correlation analysis is found to be satisfactory between the stiffness in meso modeling and the elastic modulus in macro-
scale, while the proportion of shear and normal stiffnesses is found to be related to the Passion ratio. The procedures of
inverse method are reported in [20].
116 M. Wu et al. / Engineering Fracture Mechanics 135 (2015) 113–125

Fig. 3. Model of the complete system.

Fig. 4. Model of the loading system.

2.3. Complete system modeling

A complete system composed of both the beam specimen and the loading system, including the hammer, pad, and trans-
ducer, is designed in numerical modeling to simulate the experimental tests illustrated in Part I. As shown in Fig. 3, the
dimensions of the beam specimen are 100 mm  400 mm with a span of 300 mm, whereas the hammer has a dimension
of 80 mm  120 mm with a simplified tup shaped like an arc. The force transducer, aluminum pad, and support are all dis-
cretized into particles and assigned with corresponding parameters.

2.4. Calibrations of meso-scale parameters

All particle elements are set to have radii of 1.2 mm–1.8 mm and are assigned a frictional coefficient of 0.5. The radius of
bond R is the same as the minimum radius of the two particles in contact. Other parameters, such as bond stiffness and the
strength of each concrete phase, are obtained through the inverse method. The complete system model contains different
portions, such that the parameters were calibrated separately.
First, preparatory tests using aluminum tup and aluminum-alloy pad with rigid foundation were used to calibrate the
parameters of the loading system. The discretization of the loading system is shown in Fig. 4, in which the hammer is
dropped to impact. The pad between the hammer and the transducer also has to be modeled. The equivalent bond stiffnesses
are determined through comparison of the simulated impact load history with the test results. The hammer and force
M. Wu et al. / Engineering Fracture Mechanics 135 (2015) 113–125 117

Table 1
Equivalent meso-mechanical parameters for the complete system.

Category Particle density q


 Normal bond stiffness per unit area k 
 =k Normal bond strength Shear bond strength
n s
(kg/m3) n (GPa/m2)
k r c (MPa) sc (MPa)
Loading Hammer 3500 2500 1.0 200 200
system Pad 2000 2500 1.0 20 20
Force 2000 2500 1.0 200 200
transducer
Concrete Aggregate 2800 30,000 1.9 36 54
beam Mortar 2800 25,000 2.1 24 36
Interface 2800 12,500 2.3 12 18
Support 9000 40,000 1.0 200 200

transducer are assumed to be linear elastic given their sufficiently high strengths, whereas the pad is assigned with lower
strength considering bond damage and slipping to simulate the irreversible deformation of the pad. With the use of the trial
approach, the parameters are determined and listed in Table 1. The results of the numerical simulation and experimental
tests are compared in Fig. 5. The load histories between the test and simulation are in agreement, indicating that the cali-
brated equivalent meso-parameters are reasonable and can be used in simulation.
After parameter calibration for the loading system, the concrete beam parameters for the complete system shown in Fig. 3
are validated. The meso-scale model of concrete is assigned with different meso-mechanical parameters, including stiffness
and strength, for different phases of concrete. Several steps are conducted in the inverse process. The details may be found in
Ref. [20]. The parameters obtained are listed in Table 1. The test with a 4 kg hammer dropped from 0.5 m, was chosen for the
parameter calibration. The remaining cases of loading can be conducted on the basis of the validated parameters and do not
need any rate-dependent assumptions.

3. Results comparisons and prediction

The dynamic tests, with the hammer weighting 4 kg and drop heights at 0.5 m, 1.0 m, 1.5 m, and 2.0 m, are simulated by
the complete system model of PEM to ensure simplicity without losing generality. A comparison between the numerical sim-
ulation and corresponding tests is presented.

3.1. Force histories

Part I highlighted that the total concentrated force measured by the transducer can be divided into two parts: the general-
ized inertial force Pi and the deforming force Pd. The total force Pt in the numerical simulation is calculated by the sum of
forces acting on the contact surface between the hammer and transducer, whereas Pi is analyzed by using Eq. (4) according
to the principle of virtual work.
X
mi ati ðuti  uitDt Þ
N
Pi ðtÞ ¼ ð4Þ
dt  dtDt
where N is the number of particles of the beam; mi denotes the mass of particle i; ati and uti are the vertical acceleration and
displacement (y-direction) of particle i at time t, respectively; and dt represents the deflection at mid-span of the beam at
time t.
The comparison of force histories of the four loading cases are illustrated in Fig. 6. The force histories, including Pt, Pi, and
Pd, resulting from the meso-scale model satisfactorily match the drop-weight tests. Both dynamic capacity and inertial
effects increase with increasing drop height. Although the ascent velocities and peak values of impact forces between the
numerical and test results are similar, the time durations of loading appear to be slightly longer in the simulation. Note that
the inertial forces Pi in the test are calculated assuming a polynomial distribution for acceleration; whereas it is calculated
according to the real displacements of particle elements from Eq. (4) in the numerical analyses. The agreement proves that
the assumptions in the test analysis are reasonable.

3.2. Vertical deflections

Fig. 7 illustrates a comparison of vertical deflections at the mid-span of the beam. Both the numerical and test results
show a tendency for increasing deflections with increasing loading velocities. The results are in excellent agreement before
0.3 ms, but when the impact forces descend (see Fig. 6), the two curves begin to present discrepancies. The curves also begin
to straighten starting at approximately 0.3 ms. These phenomena may be caused by the initiation and propagation of cracks
occurring at the time instant.
118 M. Wu et al. / Engineering Fracture Mechanics 135 (2015) 113–125

Fig. 5. Comparison of the numerical and test results for loading system.

Fig. 6. Comparison of numerical and test results of force histories in cases with varying drop heights, (a) 0.5 m, (b) 1.0 m, (c) 1.5 m, and (d) 2.0 m.

3.3. Tensile and compressive strains

Average strains were obtained on the basis of the displacement of particles within 50 mm in length at mid-span on both
the top and bottom edges of the beam, where the strain gauges are glued in the experimental tests. The region generally
considers the effects of cracking initiation, propagation, and the entire process of damage-cracking-failure development.
As shown in Fig. 8, good agreement was achieved between the simulation and tests, especially in terms of the strain
M. Wu et al. / Engineering Fracture Mechanics 135 (2015) 113–125 119

Fig. 7. Comparison of numerical and test results of vertical deflection histories.

Fig. 8. Comparison of numerical and test results of strain histories in cases with varying drop heights, (a) 0.5 m, (b) 1.0 m, (c) 1.5 m, and (d) 2.0 m.

velocities of both tension and compression. The histories of tensile and compressive strains behave almost symmetrically at
the beginning of loading. When a crack begins at the bottom edge, the tensile strain increases rapidly. The compressive strain
initially increases and then becomes tensioned when the cracks propagate to the top edge of the beam, indicating the com-
plete failure of the structure. The figures show that the rates of tensile strain in simulation can be divided into three phases,
i.e., an initial phase with a small slope, a second phase related to beam deformation, and finally a rapid increasing phase
related to cracking. The strain rate is defined according to the second phase of tensile strain history.
120 M. Wu et al. / Engineering Fracture Mechanics 135 (2015) 113–125

_
Fig. 9. CIF versus force rate, P.

Fig. 10. DIF versus strain rate, e_ .

3.4. Predicted results for higher rates

The analysis on impact behavior of concrete beams for higher loading velocity is conducted through numerical simula-
tions to address the limitations of the test facility. Without loss of generality, five cases of impact force rates from
500 MN/s to 1000 MN/s are applied by using varying hammer impact velocities from 8 m/s to 16 m/s. The results are dis-
cussed in the following sections.

4. Discussions

4.1. Rate effects

As discussed in Part I, the rate effects can be defined by two categories of increase factors. The capacity increase factor
(CIF) is the ratio of total dynamic force (Pt) to the maximum static force, which describes the enhancement of structural
capacity as a result of the increased capability to absorb kinetic and strain-fracture energies when the force rate increases.
The dynamic increase factor (DIF) is the ratio of dynamic deforming force (Pd) to the static force, which reflects strength
enhancement of fracture resistance versus strain rates. The CIF vs. P_ and DIF vs. e_ for both numerical simulation and
drop-weight tests are compared in Figs. 9 and 10.
As shown in Fig. 9, the CIF values are in good agreement between the test and numerical results. The results show more
pronounced increases in CIFs with values from 10 to 15 when the force rates increase from 500 to 1000 MN/s. The relation-
ship of CIF vs. P_ is nearly linear in both experimental tests and numerical simulation.
Fig. 10 provides typical DIFs from the current study and also gives the empirical curve of the modified Comité Euro-Inter-
national du Béton (M-CEB) equation by Malvar and Ross [22]. The DIF values from empirical equation are smaller, which may
come from the extensive experiments worldwide using different facilities and specimens. The increase in DIF evidently indi-
cates significant strain rate effects.
M. Wu et al. / Engineering Fracture Mechanics 135 (2015) 113–125 121

Fig. 11. Crack profiles in tests and numerical simulations, (a) test with drop height of 0.5 m, (b) simulation with drop height of 0.5 m (strain rate of 1.4 s1),
(c) test with drop height of 2.0 m, and (d) simulation with drop height of 2.0 m (strain rate of 3.3 s1).

Fig. 12. Cracking process of loading case at strain rate of 6.9 s1.

4.2. Cracking profiles

The rate effects of concrete are further studied with respect to cracking profiles. A single crack is observed in current tests
within the medium-to-high range of strain rates (0.1–4 s1). The tensile and compressive stain histories clearly show that
the crack initiates at the bottom edge of beam and develops and propagates to the top edge. The cracking profiles in dropping
heights at 0.5 m and 2.0 m are compared in Fig. 11. Note that in the case with drop height of 2.0 m, the specimen is damaged
into two pieces with a tiny piece breaking from the beam. The corresponding crack profile in simulation appears a similar
bifurcation pattern (Fig. 11d).
Meanwhile, the cases at higher strain rates predicted by PEM present different crack profiles, as shown in Figs. 12 and 13.
122 M. Wu et al. / Engineering Fracture Mechanics 135 (2015) 113–125

Fig. 13. Cracking process of loading case at strain rate of 10 s1.

Fig. 14. Partition of energy among different portions of the system in cases with varying drop heights, (a) 0.5 m, (b) 1.0 m, (c) 1.5 m, and (d) 2.0 m.

In the load case with a drop velocity of 10 m/s, the corresponding strain rate is 6.9 s1, and the dynamic capacity of the
specimen is more than nine times the static one. As denoted by point A in Fig. 12, the crack initiates from the center of the
bottom edge and propagates to the middle of the beam. When the impact force reaches the maximum value and begins to
descend, a crack occurs at the top of beam and propagates to the bottom (point B). A pattern of multiple cracks is also found
M. Wu et al. / Engineering Fracture Mechanics 135 (2015) 113–125 123

Fig. 15. Energy consumed during PEM simulation and by the test (drop height of 0.5m), (a) simulation, and (b) test.

Fig. 16. Schematic of energy components in the (a) simulation and the (b) test.

at point C; that the cracks at the center are attributed to tension, whereas the others are attributed to shear. Similar phenom-
ena were observed by Sukontasukkul and Mindess in impact tests [23] and by Ozblot et al. in an FEM simulation [5].
When subjected to a high drop velocity of 14 m/s with a corresponding strain rate of 10 s1, the crack profile becomes
more complicated (Fig. 13). Two cracks are found at the beginning of damage attributed to tension (point A0 ). At point B0 ,
the left crack inclines because of the tensile shear effect and shear crack occurs in upper-right edge. Meanwhile, compres-
sion-shear cracks appear at the loading point as a result of the stress concentration. At last, the cracks propagate and inter-
sect to break the beam into pieces (point C0 ).

4.3. Partition of energy among different portions of the system

The energy partition among different portions of the complete system is analyzed through simulation. Part I pointed out
that the concrete beam absorbs only partial energy from impact. As shown in Fig. 14, the simulation clearly shows the par-
tition of energy. For all loading rates, the kinetic energy of the hammer just before impact comprises the total energy in the
system and will be transferred to the pad and then to the concrete beam. However, with the increase in drop height, the
energy absorbed by the pad increases and becomes partly irreversible because of the breakage of contact springs, which
is equivalent to the residual deformation in experimental tests. The total energy remains nearly unchanged during impact,
confirming that the energy equilibrium is maintaining.

4.4. Conversion of beam energy

As shown in Fig. 15, particle element modeling (PEM) differs from the test analysis method in terms of energy
consumption mechanism of the concrete beam. In meso-scale particle modeling, total energy Et can be categorized as kinetic
component Ek and strain component Es if we ignore the frictional consumption Efric temporally. Es eventually reverts to Ek
when the springs are broken. In the numerical simulation, the concept of fracture energy Ef is not introduced into PEM.
In drop-weight test analysis, however, total energy Wt is divided into inertial (kinetic) component Wk and strain-fracture
energy Wf. Both total energy values are close and can be presented as
W t  Et ð5Þ
124 M. Wu et al. / Engineering Fracture Mechanics 135 (2015) 113–125

Fig. 17. Comparison of the numerical and test results of energy partition in cases with varying drop heights, (a) 0.5 m, (b) 1.0 m, (c) 1.5 m, and (d) 2.0 m.

which yields

W f þ W k  Es þ Ek ð6Þ

The main difference in the partition of energy between the simulation and the tests presented in Fig. 15 is that Ek is much
greater than Wk. As shown in the schematic in Fig. 16, Ek is presumably the sum of inertial energy induced by motion in the
y-direction (vy), x-direction (vx), and in the rotational counterpart (x), whereas Wk is calculated based only on the vertical
motion during testing. Thus, part of strain fracture energy Ef is equivalent to a portion of inertial energy given the velocities
of vx and x in PEM. Based on the discussion above, PEM is revised by considering the strain-fracture energy component
Ef in simulation. Ef can be computed by totaling strain energy Es, the partial kinetic energy generated by movement at
horizontal x and rotational velocities, and frictional dissipation Efirc if it is significant under compression-shear failure at high
strain rates, i.e.,

1X
Ef ¼ Es þ ðmi v 2ix þ Ii x2i Þ þ Efirc ð7Þ
2 N

in which N is the number of the particles and mi, vix, Ii, and xi are the mass, x-direction velocity, rotational inertia, and rota-
tional velocity of particle i, respectively. Ef is actually considered to be the primary energy consumed during the initiation of
cracking on surfaces. This variable is neglected in traditional PEM.
In the revised PEM modeling process, the energy components are calculated and compared with the corresponding test
results, as displayed in Fig. 17. The calculated results agree with those of the test; however, the discrepancy between the
simulation and the tests is significant after 0.3 ms because of the variation in impact force curves once cracking has been
initiated. Thus, parameters for concrete beam and loading systems must be optimized further. In both the numerical simu-
lation and the experimental tests, the values of all of the energy components increase significantly when drop height
increases.
M. Wu et al. / Engineering Fracture Mechanics 135 (2015) 113–125 125

5. Conclusions

In this study, drop-weight tests for concrete beams are numerically simulated using the meso-scale PEM model. The sim-
ulation results are discussed and compared with those of the corresponding tests in terms of impact capacity, vertical deflec-
tion, strain evolution, and energy partition. The drop-weight tests are then conducted under high loading rates using the
validated model. The following conclusions may be drawn from the study:

1. The results of the numerical and experimental studies match, and the model is validated. The method of complete system
modeling confirms that the PEM model can simulate not only the dynamic characteristics of pseudo-brittle materials, but
also the impact failure between two solid bodies. The meso-scale PEM does not need rate-dependent parameters in the
simulation. This advantage essentially comes from the fact that the damage-cracking process can be explicitly illustrated
through the breaking and slipping of bonds and particles. Thus, the dynamic constitutive relationship and strain rate
effects can be obtained directly through numerical simulation. The meso-scale PEM model can potentially simulate
experimental tests and the actual structures of different kinds of concrete, such as high-strength, normal, and fiber-rein-
forced concrete.
2. In both experimental and numerical studies, the rate effects of concrete are significant as dynamic loading capacity CIF
and fracture resistance DIF are enhanced. Both CIF and DIF are attributed to the energy consumption mechanism, which
involves inertial effects in addition to strain-fracture energy. We propose the energy perspective criterion as a logical tool
to explain the mechanism of rate effects and to safely evaluate concrete structures in place of the stress–strain criterion.
3. In contrast to bending mode failure at low strain rate, which is associated with a single crack across a concrete beam, the
profiles at high strain rates display mixed-modes of multiple cracks as a result of simultaneous tension, shear, and com-
pression-shear events.
4. A revised version of the PEM is developed in this study. It considers the strain-fracture energy component during crack-
ing, which accounts for the strain, frictional, and partial components of kinetic energy in the original PEM. Thus, the
meso-scale particle model is correlated with traditional fracture mechanics. Furthermore, we can compare the energy
consumed by the experimental tests and by PEM using a unified framework.

Acknowledgements

Appreciation is expressed to Professors Peter A. Cundall, Jin Feng and Guo Changqing for providing valuable suggestions.
Sincere thanks also to Professors Wang Jinting and Xu Yanjie for their discussions. This work was supported by the National
Natural Science Foundation of China (Grant Nos. 51479098, 51239006 and 91215301).

References

[1] Dube JF, Pijaudiercabot G, Laborderie C. Rate dependent damage model for concrete in dynamics. J Engng Mech-ASCE 1996;122(10):939–47.
[2] Georgin JF, Reynouard JM. Modeling of structures subjected to impact: concrete behaviour under high strain rate. Cem Concr Comp
2003;25(1):131–43.
[3] Cotsovos DM, Pavlovic MN. Modelling of RC beams under impact loading. P I Civil Engng-Str B 2012;165(2):77–94.
[4] Reinhardt HW, Ozbolt J. Fracture of concrete beams at different loading rates. In: 6th International conference on fracture mechanics of concrete and
concrete structures-FRAMCOS VI. Catania (Italy): Taylor & Francis; 2007, p. 563–9.
[5] Travas V, Ozbolt J, Kozar I. Failure of plain concrete beam at impact load: 3D finite element analysis. Int J Fracture 2009;160(1):31–41.
[6] Yon JH, Hawkins NM, Kobayashi AS. Numerical-simulation of mode-1 dynamic fracture of concrete. J Engng Mech-ASCE 1991;117(7):1595–610.
[7] Ruiz G, Pandolfi A, Ortiz M. Three-dimensional cohesive modeling of dynamic mixed-mode fracture. Int J Numer Meth Engng 2001;52(1–2):97–120.
[8] Rosa AL, Yu RC, Ruiz G, Saucedo L, Sousa J. A loading rate dependent cohesive model for concrete fracture. Engng Fract Mech 2012;82:195–208.
[9] Magnier SA, Donze FV. Numerical simulations of impacts using a discrete element method. Mech Cohes-Frict Ma 1998;3(3):257–76.
[10] Habel K, Gauvreau P. Response of ultra-high performance fiber reinforced concrete (UHPFRC) to impact and static loading. Cem Concr Comp
2008;30(10):938–46.
[11] Wittmann FH, Roelfstra PE, Sadouki H. Simulation and analysis of composite structures. Mat Sci Engng a-Struct 1985;68(2):239–48.
[12] Bazant ZP, Tabbara MR, Kazemi MT, Pijaudiercabot G. Random particle model for fracture of aggregate or fiber composites. J Engng Mech-ASCE
1990;116(8):1686–705.
[13] Schlangen E, Vanmier J. Simple lattice model for numerical-simulation of fracture of concrete materials and structures. Mater Struct
1992;25(9):534–42.
[14] Azevedo NM, Lemos JV. A generalized rigid particle contact model for fracture analysis. Int J Numer Anal Met 2005;29(3):269–85.
[15] Potyondy DO, Cundall PA. A bonded-particle model for rock. Int J Rock Mech Min 2004;41(8):1329–64.
[16] Qin C, Zhang CH. Numerical study of dynamic behavior of concrete by meso-scale particle element modeling. Int J Impact Engng 2011;38(12):1011–21.
[17] Wu MX, Qin C, Zhang CH. High strain rate splitting tensile tests of concrete and numerical simulation by mesoscale particle elements. J Mater Civil
Engng-ASCE 2014;26(1):71–82.
[18] Cundall PA, Strack O. A discrete numerical model for granular assemblies. Géotechnique 1979;29(1):47–65.
[19] Itasca, Particle Flow Code in 2 Dimensions: theory and background. particle flow code in 2-Dimensions version 4.0 user’s manual. Minneapolis, MN
(USA): Itasca Consulting Group Inc; 2008.
[20] Wu MX, Zhang CH. Influence of static pre-loading on the dynamic bending strength of concrete with particle element modeling. Sci China Technol Sc
2014. http://dx.doi.org/10.1007/s11431-014-5671-5.
[21] Qin C, Guo CQ, Zhang CH. A pre-processing scheme based on background grid approach for meso-concrete mechanics. J Hydra Engng
2011;8(42):941–8.
[22] Malvar LJ, Ross CA. Review of strain rate effects for concrete in tension. ACI Mater J 1998;95(6):735–9.
[23] Sukontasukkul P, Mindess S. The shear fracture of concrete under impact loading using end confined beams. Mater Struct 2003;36(260):372–8.

Das könnte Ihnen auch gefallen