Professional Documents
Culture Documents
International Journal of Impact Engineering: X Xjun Feng XX, X Xwenjin Yao XX, X Xweixin Li XX, X Xwenbin Li XX
International Journal of Impact Engineering: X Xjun Feng XX, X Xwenjin Yao XX, X Xweixin Li XX, X Xwenbin Li XX
TAGEDPA R T I C L E I N F O TAGEDPA B S T R A C T
Article History: This paper numerically investigates the large-caliber hard projectile perforation of plain concrete slab with
Received 3 March 2017 steel culvert confinement by using the recently developed Lattice Discrete Particles Model (LDPM). With
Revised 22 May 2017 mesoscale constitutive laws governing the interaction between adjacent particles, LDPM simulates concrete
Accepted 26 May 2017
features like cohesive fracture, strain hardening in compression and compaction due to pore collapse. The
Available online 1 June 2017
perforation simulation model is established with LDPM for concrete, elastoplastic model for steel culvert
and penalty contact for projectile-concrete interaction. Simulation of 5 shots concrete perforation tests are
TagedPKeywords:
carried out to validate the numerical model, and the numerical results are in good agreement with the
Concrete perforation
Steel culvert
experimental data in terms of residual velocities as well as target damage mode. Extensive simulations are
Lattice discrete particle model further performed to investigate the effects of aggregate discretization, steel culvert confinement and diam-
Residual velocity eter size on perforation responses. Comparative numerical study indicates that with culvert confinement,
Damage mode the projectile residual velocity is significantly reduced except for perforation case with 0.3 m thickness tar-
get. Meanwhile, crack fills the 0.6 m diameter target which poses less resistance to projectile impact.
© 2017 Elsevier Ltd. All rights reserved.
1. Introduction TagedPmotivated by the high cost of undertaking impact concrete tests and
the ease of material properties effects evaluation.
TagedPCharacterized with high compressive strength, easily shaping, TagedPThe presence of extreme material distortion and failure still
considerable durability and low cost, concrete material has been remains a challenge in computational mechanics. Lagrangian struc-
widely used in most of human civilization, including buildings, high- tured-mesh methods, for example the finite element method,
way bridges, offshore platforms and civil defense etc. During their exhibit mesh sensitivity and subjectivity in the presence of severe
service life, these concrete structures might be exposed to high material deformation during penetration modeling. Techniques like
intensity extreme loadings due to projectile impact and explosion. element erosion have been developed as a non-physical treatment
Therefore, it is of great importance to understand the behavior of [33]. In contrast, Eulerian finite difference and finite volume meth-
concrete structures subjected to impact loadings. Extensive perfora- ods are more resilient in modeling large deformation and flow [34],
tion and penetration tests [17] have been performed to explore the however their fixed grid generally produces error because they fail
mechanism of the impact response on the concrete target. Based on to accurately identify transient contact surfaces and free surface for-
which various empirical, semi-empirical and analytical formulae mation. This leads to significant limitations when material breakage
[813] have been developed to analyze the deep penetration and or fracture is dominant. The complex constitutive description of con-
perforation problems. As computation becomes increasingly power- crete material under high pressure, high stain rate and high temper-
ful, numerical simulation tools such as finite element method (FEM) ature is also an important aspect issue. Target material undergoes
[1417], extended finite element method [18], discrete element fracture, compaction and high confinement shear etc. Advancements
method [19,20], smoothed particle hydrodynamics (SPH) [2123], are being made toward multiscale descriptions of material failure,
material point method [24,25], reproducing kernel particle method where behavior is described according to the fundamental mecha-
(RKPM) [26,27] and various meshfree methods [2832] are nisms at finer scales [35,36].
TagedPDiscrete element methods, in contrast to traditional mesh-based
* methods, do not rely on a structured mesh topology to compute the
Corresponding author.
E-mail addresses: fengjunandrew@gmail.com (J. Feng), njyaowj@163.com discrete numerical solution. Consequently, large deformation as well
(W. Yao), w.li@u.northwestern.edu (W. Li), lwb2000cn@njust.edu.cn (W. Li). as crack initiation and propagation can be naturally captured
http://dx.doi.org/10.1016/j.ijimpeng.2017.05.017
0734-743X/© 2017 Elsevier Ltd. All rights reserved.
40 J. Feng et al. / International Journal of Impact Engineering 109 (2017) 3951
TagedPwithout requiring the non-physical treatments like element dele- 2. Review of the Lattice Discrete Particle Model (LDPM)
tion. By modeling concrete through a discrete system of beams, van
Mier and coworkers [3739] assigned different material properties TagedPAccording to Ref. [44], LDPM model generation procedure and
to the lattice elements. The mesoscale lattice models provide realis- governing constitutive equations are explained in the following four
tic simulations of concrete cracking and fracture propagation subsections.
[4043]. Building on the successful developments of lattice model
and discrete element method, Cusatis et al. [44] proposed a compre- 2.1. LDPM model construction
hensive mesoscale discrete model entitled Lattice Discrete Particle
Model (LDPM). Capable of simulating the tensile fracture and dam- TagedPConcrete meso-scale structure is modeled by LDPM through the
age in compression, the mesoscale discrete model is then exten- following steps:
sively calibrated and validated in quasi-static regime [45]. Later,
TagedP Spherical aggregate generation is the first step which is carried
Cusatis [46] extended the LDPM to include the strain-rate effects
out assuming that each aggregate piece can be approximated as
which has been applied to dynamic simulations including Hopkin-
a sphere. Under this assumption, the following spherical aggre-
son bar test, perforation and penetration analysis of high perfor-
gate size distribution function proposed by Stroeven [49] is con-
mance fiber reinforced concrete [20,47]. Therefore, LDPM is utilized
sidered
in this work to numerically study the response of hard projectile
perforation of plain concrete target. qdq0
f ðdÞ ¼ ð1Þ
TagedPUnder projectile impact loading, concrete target experiences ½1ðd0 =da Þq dqþ1
extreme high strain rates which requiring an accurate numerical
model with a better understanding of the time-dependent mechani- in which da and d0 are the maximum and minimum spherical aggre-
cal behavior of concrete. In most cases, the rate effect is quantified gate size, respectively, and q is a material parameter. It can be shown
through the Dynamic Increase Factor (DIF) defined as the ratio [49] that Eq. 1 is associated with a sieve curve (percentage of spheri-
between the dynamic properties of interest and the associated cal aggregate by weight retained by a sieve of characteristic size d)
quasi-static value. The DIF curves are generally obtained by inter- in the form
ploting between experimental data with huge scatter. Moreover, the nf
d
rate effect incorporation with constitutive equations is implemented f ðdÞ ¼ ð2Þ
da
by multiplying DIF curve function by the model parameters. The
method however is not correct since it fails to distinguish between where nf ¼ 3q. For nf ¼ 0:5; Eq. (2) corresponds to the classical
‘intrinsic’ phenomena included in the constitutive laws, and ‘appar- Fuller curve [50] which for its optimal packing properties, is exten-
ent’ phenomena which are structure features of the response sively used in concrete technology. Considering concrete cement
[15,46,48]. The most important component leading to the concrete content c, water-to-cement ratio wa/c, specimen volume, maximum
strain-rate effect behavior attributes to the inertia effects. As pointed da and minimum d0 spherical aggregate size along with the consid-
by Cusatis [46], as long as inertia is correctly accounted for in the ered distribution function Eq. (1), the spherical aggregate system
interpretation of experimental results and the numerical simula- can be generated using a random number generator.
tions, no inertia effects should be included in the constitutive equa- TagedP By using a try-and-error random procedure, spherical aggregate
tions. pieces are introduced into the concrete volume from the largest
TagedPWith reference to perforation tests conducted by Li et al. [6], this to the smallest size. Fig. 1(a) shows the spherical aggregate sys-
article briefly introduces the LDPM formulations and parameter tem generated for a typical dogbone specimen.
meanings which is then used in the numerical models together with TagedP Delaunay tetrahedralization of the spherical aggregate piece
penalty contact. As a consequence, 5 sets of large-caliber hard pro- centers is employed to define the interactions of the spherical
jectile perforation simulations are performed and validated by com- aggregate system (Fig. 1(b)).
parison of residual velocities and target damage mode. Further TagedP Finally, a three-dimensional domain tessellation anchored to the
simulations are carried out to analyze the effects of aggregate discre- Delaunay tetrahedralization is carried out to create a system of
tization, steel culvert confinement and diameter scale on the perfo- polyhedral particles interacting through triangular facets, and a
ration responses. lattice system composed of the line segments connecting the
Fig. 1. LDPM polyhedral particles and cell distretization for a typical dogbone specimen.
J. Feng et al. / International Journal of Impact Engineering 109 (2017) 3951 41
TagedPspherical aggregate centers. Fig. 1(c) shows the final polyhedral TagedP eyond pore collapse (ɛV ɛc1 ). In this case one has s bc ¼ s c1 ðrDV Þ exp
b
particle discretization of the dogbone specimen. ½ðɛV ɛc1 ÞHc ðrDV Þ=s c1 ðrDV Þ and s c1 ðrDV Þ ¼ s c0 þ ðɛc1 ɛc0 ÞHc ðrDV Þ.
TagedPThe ogival nose projectile, with 64 mm caliber and 3 caliber- rTagedP ing FEM nodes are also constrained to the aggregate particles with
radius-head (CRH) as dimensioned in Fig. 2, was launched to acquire penalty-function contact.
a normal striking velocity of about 400 m/s. As mentioned by
Li et al. [6], the projectiles were machined from heat treated 3.2. Selected LDPM parameters
35CrMnSiA alloy which is supposed to have high strength and
toughness. TagedPAccording to the information in Table 1, the concrete mixture
proportion by weight is characterized by cement content
3.1. Perforation simulation scheme c ¼ 470 kg/m3, water to cement ratio wa =c ¼ 0:44; aggregate to
cement ratio a=c ¼ 3:71; minimum aggregate size 5 mm, maxmum
TagedPSince the high strength steel alloy projectiles suffered a negligible aggregate size 15 mm, Fuller coefficient nF ¼ 0:5 which are input
deformation during the perforation process as reported in Ref. [6]. information for LDPM particle generation. Take simulation of test
This work chooses to model the projectile as a rigid body for the No. 1 in Table 2 for example, a total number of 422,890 particles and
sake of computational cost. The foregoing introduced LDPM is 2,515,260 tetrahedrons are employed to discretize the concrete tar-
adopted herein to describe the concrete material response. While get where the surface contains 124,060 triangular facets.
the steel culverts confining the concrete targets are modeled by FEM TagedPWith reference to the parameters provided by Cusatis et al. [45],
via an elastoplastic material law with Young’s modulus of 200 GPa the listed LDPM parameters are justified to meet the 34.3 MPa
and yield stress of 500 MPa. Meshed by hexahedral blocks of 20 £ for 15-cm-cube concrete specimen: normal elastic modulus
10 £ 3 mm dimension, the 0.3 m thick culvert ring is divided into E0 ¼ 35000 MPa, densified normal modulus Ed ¼ E0 ; shear-normal
9450 elements. coupling parameter a ¼ 0:25; tensile strength s t ¼ 3:2 MPa, yielding
TagedPThis work adopts the penalty-function contact model to compute compressive stress s c0 ¼ 60 MPa, shear strength s s ¼ 2s t ; tensile
the normal contact force between the concrete particle system and characteristic length lt ¼ 200 mm, softening exponent nt ¼ 0:2; ini-
the rigid projectile as illustrated in Fig. 3. The penalty force Fn is tial hardening modulus Hc0 ¼ 0:4E0 ; transitional strain ratio kc0 ¼ 4;
assumed to be proportional to the normal penetration Pn: Fn ¼ K Pn initial internal friction coefficient m0 ¼ 0:17; internal asymptotic
where K represents the linear spring stiffness [54]. The contact force friction coefficient m 1 ¼ 0; transitional stress s N0 ¼ 200 MPa,
is always normal to the contact pair surface and resists the projectile deviatoric-to-volumetric strain ratio kc1 ¼ 1; and kc2 ¼ 5.
moving forward. The time step Dt is used in conjunction with the TagedPThe hydrostatic compression response, shown in Fig. 4(a), is
minimum mass mmin of the nodes participating in the contact to obtained by LDPM simulation with a cylinder specimen of 150 mm
define the penalty spring stiffness K ¼ mDmin t2
. The explicit algorithm diameter and 300 mm height. In the case of dydrostatic compres-
used in LDPM requires the time step must satisfy Dt < 2/vhf where sion, three transitions can be observed in the stress-strain curve. The
vhf is the highest natural frequency of the computational system behavior is elastic up to the first transition point corresponding to
[44]. Here we choose Dt ¼ 3 107 s; and the penalty spring stiffness the beginning of the voids collapse. Then the nonlinear response
is of the order of magnitude of 108 N m1 which is sufficient to avoid develops up to the second transition that corresponds to the fully
overlapping problem in impact simulation. Similarly, the culvert compaction [55]. After that, the stress exponentially increases due to
Table 2
Comparisonof perforation tests and numerical predictions.
No. Projectile mass (kg) Target thickness (m) Striking velocity (m/s) Measured Residual Velocity (m/s) LDPM prediction (m/s)
TagedPthe comminution response. By keeping the lateral confinement pres- tTagedP he projectile exits with approximately 30 m/s residual velocity. For
sure constant, the triaxial compression response is achieved with this impact condition, 405 m/s striking velocity is very closed to the
axial displacement control in numerical modeling. For confining ballistic limit which is defined as the minimum velocity needed to
pressures of 50 MPa, 100 MPa, 150 MPa and 200 MPa, the axial perforate the target plate with no residual velocity [20,57]. Cavity
stress versus axial strain curves are plotted in Fig. 4(b). Since the expansion theory indicates that the penetration resistance depends
shear yield criterion of concrete material is pressure dependent, the on the expansion velocity projected from penetration velocity and
higher confining pressure will induce the axial response to exhibit the triaixal mechanical response of target material [9,58]. As the
strain-hardening. This pressure-dependent inelastic property is a scabbing on the rear surface occurs, the resistance acting on the pro-
typical response for concrete like granular material which is primar- jectile nose becomes smaller as a result of free surface boundary
ily controlled by the heterogeneity of grain skeleton [56]. reflected tensile stress degradation on the concrete material. More-
over, the low penetration velocity (approximately 30 m/s) further
3.3. LDPM simulation validation decrease the resistance. These two main factors both contribute to
the long tail in the velocity versus time curve for 0.6 m target perfo-
TagedPIn this section, we evaluate the numerical performance of LDPM ration case. With reference to Wu et al. [59], the energy profile
for the simulation of plane concrete perforation via the crack pattern obtained from the impact simulation is analyzed hereby. As shown
and residual velocity. The numerical results of projectile residual in Fig. 6, the system total energy, the projectile kinetic energy and
velocities are listed in Table 2 for 5 different impact tests. In general, the internal energy, including elastic energy stored by the target and
LDPM predicted projectile residual velocities after perforation tar- dissipated energy due to plastic work, are depicted with normaliza-
gets with different thickness match the experimental measured data tion to the original total energy. The loss in the total energy is limited
with less than 10% error. It is to be noted that both experimental and within 3%, therefore the energy conservation is satisfied in this sim-
numerical results show nonperforation for shot test No. 5. Although ulation system.
the rear surface shear plugging occurs, the projectile remained TagedPThe cross section views of crack opening distribution inside the
embedded in the target with approximately 0.57 m depth of pene- concrete targets are plotted in Fig. 7 with color bar ranging from 0.1
tration predicted by LDPM. Ref. [6] did not give the penetration to 2 mm for visualization. Projectile crush is very obvious for targets
depth of test No. 5, but the rear crater was measured about 0.27 m with 0.3 m, 0.4 m and 0.5 m targets since the residual velocities are
and most of the projectile nose tip was exposed from the back view. relatively high. Meanwhile, numerical results of perforation cases
Therefore, we may estimate the shot No. 5 experimental depth of with 0.6 m and 0.7 m reveal that the concrete target rear surface
penetration is about 0.54 m which agrees well with the predicted fragments are plugged by the low velocity projectile punch. In other
value by LDPM simulation. words, the projectile may move forward together with the concrete
TagedPThe curves of projectile deceleration versus time for perforation fragmented pieces (Fig. 7(d)) and plugged blocks (Fig. 7(e)).
simulations with thickness 0.3 m, 0.4 m, 0.5 m and 0.6 m are plot- TagedPFront and rear surface damage comparison of experimental and
ted in Fig. 5 with 4 different colors. Due to the similar striking veloc- numerical results are plotted in Fig. 8. Since the damage mode is
ity and mass, the peaks of deceleration curves are almost identical symmetric, we merge the snapshots of front and rear surface
even though the target thickness is different. For thicker perforation obtained from perforation experiment and simulation in Fig. 8(a)
cases, the penetration time lasts much longer, especially for the case (f) for comparison. Numerical results match with the experimental
with 0.6 m thickness target. After perforation through 0.6 m target, damage in terms of radial cracks distribution as well as crater shape
TagedPand size. For 0.3 m and 0.4 thickness target perforation scenarios, TagedP ith 1.4 m diameter and 0.8 m thickness. Except the concrete
w
both experimental and numerical results indicate that the rear crater strength, the projectile scale and target dimension of experiment
contour size is larger than the front one, as shown in Fig. 8(a)(d). 1:3 in Ref. [2] are similar to that of shot No. 5 investigated in this
This may due to the reason that the rear surface is damaged mainly work. To compare Rabczuk and Belytschk’s meshfree method with
by crush and plug while the front surface crater is generated by nob- LDPM, the numerical achieved crack patterns are plotted together in
bing. Fig. 8(e) shows an asymmetrical crater shape for experiment, Fig. 9. The Lemaitre model, with isotropic damage depending on the
which occurs might due to the heterogeneity. What is notable is that effective strain, was employed by Rabczuk and Belytschk to describe
the projectile fails to perforate 0.7 m thickness target the plug zone the concrete material. So the cracks are treated as damaged zones in
is featured by an irregular shape as shown in Fig. 8(g). Fortunately, Fig. 9(a) which is different from the crack openings in LDPM derived
LDPM constructs the meso-structure of concrete target by randomly from the gaps between facets. Because the aggregate anisotropy is
distributing the aggregate particle inside the body which accounts accounted for, LDPM numerical result reveals the unsymmetric crack
for the anisotropy in mesoscale. The rear surface plug zone numeri- pattern which is more realistic. Moreover, both the impact surface
cally obtained is quite closed to the shape of experimental results by crater area and radial crack distributions matches well with each
rotating 90° in counter-clockwise direction. other.
TagedPRabczuk and Belytschko [31] proposed a meshfree method using
Lagrangian kernels for initial simulation stage and switching to 3.4. Further analyses of LDPM simulation
Eulerian kernels for cracking and large deformation. Capturing mate-
rial fracture behavior including crack branching, the combined TagedPIt is important to have an inside view of the stress distribution of
meshfree method is applied to model perforation and penetration the target to better understand the impact response. With 0 to 40
tests performed by Unosson and Nilsson [2] whereby the 75 mm MPa range, Fig. 10 demonstrates the internal stress as well as cracks
diameter ogival nose projectile were fired at concrete cylinder target distributions of the target at 1st millisecond where the topside and
rearside view with the central plane cut of the target is plotted. It is
observed from Fig. 10(a) and (d) that the high pressure is mainly
concentrated near the projectile nose zone due to the resistance
against penetration. While the Mises stress distribution in Fig. 10(b)
and (e) gives a good approximation of the damaged zone owing to
the shear failure criterion. The crack openings are plotted in Fig. 10
(c) and (f) where the fragments are more realistic. It is worth noting
that the localized cracks in red colors correspond to the contour of
Mises stress distribution.
TagedPSince the front and rear cratering is a typical phenomenon
observered in perforation test, we further discuss the perforation
process with the LDPM numerical simulation of shot test No. 4. The
typical sequences of the perforation event are shown in Figs. 11 and
12 at every 1 ms. In cross section view, Fig. 11 plots the snapshots of
concrete target internal crack openings (ranging from 0.1 to 5 mm),
meanwhile Fig. 12 depicts concrete material velocities distribution
(ranging from 0 to 10 m/s) evolution history. At the first millisecond,
the projectile totally penetrate into the concrete target body, and
Fig. 6. Energy profile during perforation response.
Fig. 11(a) shows that the front crater has not formed yet which
J. Feng et al. / International Journal of Impact Engineering 109 (2017) 3951 45
TagedPagrees with the initial cratering region captured by high speed cam- TagedPtest No. 5, the target is not perforated by the projectile, this section
era captured photograph in Ref. [60]. Fig. 12(a) suggests that the only concentrates on perforation cases with 0.3 m, 0.4 m, 0.5 m
conical zone near the front surface has high spalling velocity to form and 0.6 m thickness targets.
the front crater and the concrete material surrounding the projectile
nose is driven by the projectile to obtain high speed velocities. At 4.1. Effect of discretization
the second millisecond, the cracks propagate to the rear surface as
well as steel culvert boundary as shown Fig. 11(b) and Fig. 12(b) TagedPDifferent from particle of Galerkin method [59] or element of
gives a more obvious description of both front crater and back crater FEM, the aggregate particle sphere in LDPM is not only used for dis-
formation tendency. Due to the low velocity, the projectile moves cretization but also contains the information of concrete meso-
forward ‘slowly’ to perforate the target. At the fifth millisecond, rear structure which is related to the fracture energy and softening mod-
crater segments are apparently separated from the target in Fig. 11 ulus of constitutive laws. In order to evaluate the aggregate size as
(e). Together with material velocity distribution in Fig. 12(e), we can well as its distribution, we analyze the particle size diameter d0, da
figure out the front and rear crater zones of LDPM perforation simu- and the Fuller coefficient nF with the simulation of shot No. 1. Table 3
lation. lists 6 sets of discretization schemes and the corresponding curves of
cumulative distribution of Eq. (1) are plotted in Fig. 13.
4. Perforation response discussion TagedPWith different discretizations, the numerical predicted projectile
residual velocities of shot No. 1 vary as listed in Table 3. Discretiza-
TagedPWith the validated numerical model for perforation simulation, tions Nos. 13 are featured by aggregate particles of different maxi-
this section further explores the perforation response influenced by mum and minimum sizes. With cylinder specimens of 150 mm
factors like steel culverts and cylinder diameter size. For penetration diameter and 300 mm height, the triaxial compression simulations
46 J. Feng et al. / International Journal of Impact Engineering 109 (2017) 3951
Table 3
Shot No. 1 LDPM simulations with different
discretizations.
1 5 15 0.5 265
2 5 10 0.5 268
3 10 15 0.5 265
4 5 15 0.35 270
5 5 15 0.65 260
6 8 12 0.5 275
Fig. 11. Cross section view of crack opening distribution at each 1 ms.
J. Feng et al. / International Journal of Impact Engineering 109 (2017) 3951 47
Fig. 12. Cross section view of concrete material velocity distribution at each 1 ms.
TagedPresponse (see Fig. 14(b)) and penetration resistance (see Fig. 15(b)). 4.2. Effect of steel culverts (SC)
To be specific, concrete targets with smaller Fuller coefficient con-
crete targets can better resist the impact loadings. This phenomenon TagedPThe steel culverts in the perforation test were used to cast the
is contributed by the discretization size effect on the softening mod- concrete material as a container and also confine the cylinder target
ulus. Discretization Nos. 1 and 6 are characterized with different size lateral surfaces in radial direction. For each thickness, 2 sets of simu-
ranges but same medium value of d0 and da. From Fig. 14(c) and lations are performed herein: confined by 3 mm thickness steel cul-
Fig. 15(c), it is interesting to find that the more uniform discretiza- vert and without any steel culvert. Then, this section compares and
tion shows a softer triaxial compression response which may stems further analyzes the damage mode as well as residual velocities of
from the size effect. As shown in Table 3, the residual velocity, corre- cylinder target perforations caused by the culvert confinement
sponding to discretization 6, increases by 15 m/s for the interested effect.
perforation. TagedPThe lateral steel culverts confinement alter the drag force
acting on the projectile nose to some extent. The confinement
effect is evaluated through the projectile deceleration curves
during the perforation process, as shown in Fig. 16. It is con-
cluded that the steel culvert poses negligible effect on the
projectile deceleration for perforation simulation with 0.3 m
thickness. Meanwhile, the steel culvert effect on the drag force
is apparent for perforation cases with 0.4 m, 0.5 m and 0.6 m
thickness. This phenomenon may stems from the penetration
response time duration. The thinner the concrete target, the less
time duration will last which ends even before the boundary
reflected stress waves start to affect the local impact zone mate-
rial. This may also account for the significant difference occur-
ring to deceleration after the peak of perforation simulations
with and without steel culverts. Table 4 provides numerically
obtained perforation residual velocities for targets with different
conditions. There is slight difference for residual velocities for
0.3 m thickness target perforations with and without a steel
Fig. 13. Discretization scheme.
Fig. 15. Shot No. 1 projectile velocity evolutions of simulations with different aggregate discretizations.
Fig. 16. Numerical comparison of steel culvert effect on the projectile deceleration.
TagedPculvert. On the contrary, 0.4 m, 0.5 m and 0.6 m thickness target TagedPTo analyze the steel culvert effect on the concrete target damage
perforation simulation results indicate that the residual velocity mode, this work compares the front and rear surface damage of
rises by 36 m/s, 56 m/s and 119 m/s due to the lack of effect of 0.6 m thickness cases as plotted in Fig. 17. As revealed in Fig. 17(a)
steel culvert reinforcement. and (b), the front surface damage of target without steel culvert is
Table 4
Numerical comparison of perforation with different target conditions.
No. Target thickness (m) 1.2 m diameter without steel culvert 1.2 m diameter with steel culvert 0.6 m diameter with steel culvert
TagedPdominated by localized radial cracks meanwhile the main cracks TagedPresidual velocities for 4 sets of perforations with 0.6 m and
occurring to the steel culvert confined target is in circumferential 1.2 m target thickness. Due to the concrete degradation caused
direction. For the concrete perforation cases without steel culvert, by localized pulverization damage, projectile residual velocities
the free boundary condition allows the cylinder target to expand in corresponding to targets with 0.6 m thickness get increased by
radial direction which causes radial cracks initiation when circum- 9 m/s, 19 m/s, 37 m/s and 66 m/s, respectively.
ferential tensile stress reaches fracture or tensile yield strength. The TagedPFor targets with 0.4 m thickness, Fig. 19 shows front and rear sur-
circumferential cracks may result from the reflected stress provided face damage mode difference caused by diameter scale. The crater
by the steel culvert confinement. Similarly, the rear surface damage zones in Fig. 19(a) and (c) almost fill the whole front and rear target
mode is also mainly featured by radial cracks and circumferential surfaces due to the localized reflected and superimposed stress
cracks for target with and without steel culvert. The steel culvert waves. Although, the crater sizes of 0.6 m diameter targets are
confinement poses more resistance on the projectile body, thus the reduced owing to the smaller surface. The damage however is more
residual velocity is 146 m/s and 27 m/s as listed in Table 4. The rear server comparing Fig. 19(a) and (b).
crater is caused by low velocity projectile plug for steel culvert con- TagedPAs depicted in Figs. 20 and 21, the cross section diagrams of the
fined target while the rear crater of free lateral target is formed by 0.4 m thickness targets are comparaed in terms of crack distribu-
projectile crush since the residual velocity is much higher. tions and concrete material velocity distributions. Obvious deforma-
tion occurring to the steel culverts of 0.6 m diameter targets can be
observed from Fig. 20(a) and Fig. 21(a). For 0.6 m diameter cases, the
4.3. Effect of slab diameter size opening cracks almost fill the whole concrete target cross section
due to the comminution damage and the rear crater zone reaches
TagedPTo investigate the diameter effect on the perforation response, the steel culvert boundary. The reduction of diameter size will cause
we perform perforation simulations with 0.6 m and 1.2 m diameter the concrete target to suffer structure damage and become less pro-
cylinder for each slab thickness. The comparative analysis is then tective against projectile impact.
performed to shed some light on the target scale effect on the perfo-
ration response.
TagedPNumerical results of projectile deceleration versus penetration 5. Conclusion
depth for 4 thickness cases are plotted in Fig. 18. It is notable that
the projectile deceleration data has not decreased to zero when the TagedPThe recently developed Lattice Discrete Particle Model (LDPM) is
projectile penetration depth reaches the plate thickness. Rather the adopted in the present article for the simulation of the hard projec-
drag force vanishes until the penetration depth is near target thick- tile perforation of the plain concrete slab confined by a steel culvert.
ness plus 100 mm which corresponds to the position where projec- Based on the results of the LDPM numerical simulations, the follow-
tile nose fully perforate the target. The LDPM simulation of projectile ing conclusions can be drawn out. (1) The employed mesoscale dis-
deceleration curves indicate that the rare crater is generated by the crete element model is able to realistically predict the projectile
projectile crush for perforation cases. In other words, the con- residual velocities and crater contour for all 5 slab thicknesses. (2)
crete fragments behind the target rear surface may still pose The rear crater fragments are crushed by high velocity projectile,
resistance on the projectile until the whole projectile nose part while the low residual velocity projectile poses a plug effect on the
fully exits the target. Table 4 gives the detailed projectile rear target separated blocks. The front crater is generated after the
Fig. 18. Numerical comparison of target diameter effect on the projectile deceleration.
TagedP[42] Schlangen E, Qian Z. 3D modeling of fracture in cement-based materials. J Multi- TagedP[52] Cusatis G, Rezakhani R, Schauffert EA. Discontinuous cell method (DCM) for the
scale Model 2009;1(02):245–61. simulation of cohesive fracture and fragmentation of continuous media. Eng
TagedP[43] Savija B, Pacheco J, Schlangen E. Lattice modeling of chloride diffusion in sound Fract Mech 2017;170:1–22.
and cracked concrete. Cem Concr Compos 2013;42:30–40. TagedP[53] Bazant ZP. Creep and damage in concrete. Mater Sci Concr 1995;IV:355–89.
TagedP[44] Cusatis G, Pelessone D, Mencarelli A. Lattice discrete particle model (LDPM) for TagedP[54] Pelessone D. Mars: modeling and analysis of the response of structures-users
failure behavior of concrete. i: theory. Cem Concr Compos 2011;33(9):881–90. manual. ES3, Beach (CA), USA 2009.
TagedP[45] Cusatis G, Mencarelli A, Pelessone D, Baylot J. Lattice discrete particle model TagedP[55] Ara oz G, Luccioni B. Modeling concrete like materials under sever dynamic pres-
(LDPM) for failure behavior of concrete. ii: calibration and validation. Cem Concr sures. Int J Impact Eng 2015;76:139–54.
Compos 2011;33(9):891–905. TagedP[56] Ashari SE, Buscarnera G, Cusatis G. A lattice discrete particle model for pressure-
TagedP[46] Cusatis G. Strain-rate effects on concrete behavior. Int J Impact Eng 2011;38 dependent inelasticity in granular rocks. Int J Rock Mech Min Sci 2017;91:49–
(4):162–70. 58.
TagedP[47] Smith J, Cusatis G, Pelessone D, Landis E, O’Daniel J, Baylot J. Discrete modeling of TagedP[57] Børvik T, Dey S, Clausen A. Perforation resistance of five different high-strength
ultra-high-performance concrete with application to projectile penetration. Int J steel plates subjected to small-arms projectiles. Int J Impact Eng 2009;36
Impact Eng 2014;65:13–32. (7):948–64.
TagedP[48] O zbolt J, Sharma A. Numerical simulation of dynamic fracture of concrete TagedP[58] Feng J, Li W, Wang X, Song M, Ren H, Li W. Dynamic spherical cavity expansion
through uniaxial tension and l-specimen. Eng Fract Mech 2012;85:88–102. analysis of rate-dependent concrete material with scale effect. Int J Impact Eng
TagedP[49] Stroeven P. A stereological approach to roughness of fracture surfaces and tortu- 2015;84:24–37.
osity of transport paths in concrete. Cem Concr Compos 2000;22(5):331–41. TagedP[59] Wu C, Wu Y, Crawford JE, Magallanes JM. Three-dimensional concrete impact
TagedP[50] Fuller WB, Thompson SE. The laws of proportioning concrete. Trans Am Soc Civil and penetration simulations using the smoothed particle Galerkin method. Int J
Eng 1907;59(2):67–143. Impact Eng 2017;106:1–17.
TagedP[51] Rezakhani R, Cusatis G. Asymptotic expansion homogenization of discrete fine- TagedP[60] Warren T, Fossum A, Frew D. Penetration into low-strength (23 MPa) concrete:
scale models with rotational degrees of freedom for the simulation of quasi-brit- target characterization and simulations. Int J Impact Eng 2004;30(5):477–503.
tle materials. J Mech Phys Solids 2016;88:320–45.