Documente Academic
Documente Profesional
Documente Cultură
A R T I C LE I N FO A B S T R A C T
Keywords: Ceramic materials have excellent mechanical properties such as light weight, great hardness and high com-
Ceramic balls pressive strength. In this paper, a numerical study is conducted to investigate the response of ceramic balls
UHPC protected ultra-high performance concrete (UHPC) targets against the high-velocity rigid projectile impact using
SPH-FE method the coupled smoothed particle hydrodynamics-finite element (SPH-FE) method in LS-DYNA. Based on the va-
Projectile impact
lidated numerical models, parametric studies are performed to explore the effect of diameter, spatial arrange-
ment and material type of ceramic balls as well as the impact position on the dynamic performance of UHPC
targets, and then perforation and ballistic limits of ceramic balls protected UHPC targets are obtained. Compared
with other UHPC slabs at the striking velocities from 500 m/s to 850 m/s, UHPC slabs protected with 6-layer hex-
pack arranged ceramic balls with the diameter of 20 mm is most effective in terms of reducing the depth of
penetration (DOP). In addition, the utilization of ceramic balls is economical in protective structures since the
damaged ceramic balls can be replaced and undamaged ceramic balls are reusable.
1. Introduction Finite element (FE) method has been widely adopted to analyse the
solid mechanics, but it is not capable of handling large distortions of
In conventional protective design, concrete is one of the most ex- elements when subjected to extreme loading conditions, which may
tensively used materials to resist impact loadings caused by projectiles. lead to computational overflow or inaccurate predictions. In order to
In recent years, resulted from the increasingly growing demand for avoid computational overflow caused by element distortions, “erosion
advanced concrete materials with an ultra-high strength and an ex- algorithm” is often used to delete elements that experience large dis-
cellent toughness, ultra-high performance concrete (UHPC) has been tortions. However, the deletion of elements breaches the conservation
developed [1–3] and its behaviour against the high-velocity projectile of mass and momentum and fails to obtain the debris information.
impact was tested [4–8]. Although UHPC has been a promising material In order to overcome these shortcomings caused by FE method,
in terms of the impact resistance which can be directly used in the various mesh-free approaches such as smoothed particle hydro-
structural components in civil and military constructions, optimal dynamics (SPH) [14,15], element free Galerkin (EFG) [16], etc. have
structural designs with different reinforcement configurations are under been developed in recent years and summarized in Refs [17–19]. These
extensive investigation to maximize the impact resistance of UHPC standard mesh-free methods have been successfully improved and used
components. in failure and fragmentation of brittle materials under dynamic load-
The outstanding mechanical properties of ceramic panels such as ings by researchers. For example, Rabczuk and Eibl [20,21] used SPH
light weight, great hardness and high compressive strength are widely and modified moving least square SPH methods to simulate dynamic
exploited as the reinforcement or shielding for the metal structures behaviours of concrete under blast and impact loadings. Rabczuk and
[9–12] or concrete structures [13] to resist projectile impact. However, Belytschko [22,23] also developed a robust and efficient cracking
some disadvantages may limit its use in ballistic design when the ex- particle method with the EFG concepts to model discrete cracks of
pensive panel is penetrated or perforated by the projectile even in a concrete under dynamic loadings, where this method is based on a
small area and the whole panel becomes non-repairable and requires Lagrangian kernel and it can eliminate artificial fracture caused by
complete replacement. Therefore, a novel protective structure with the tensile instabilities [24]. However, the Lagrangian kernel is not effec-
reusable and economical properties is necessary to be developed. tive in large deformation problems. Based on the cracking particle
⁎
Corresponding authors.
E-mail addresses: Jian.Liu-4@student.uts.edu.au (J. Liu), chengqing.wu@uts.edu.au (C. Wu).
https://doi.org/10.1016/j.ijimpeng.2018.11.006
Received 11 June 2018; Received in revised form 7 November 2018; Accepted 16 November 2018
Available online 20 November 2018
0734-743X/ © 2018 Elsevier Ltd. All rights reserved.
J. Liu et al. International Journal of Impact Engineering 125 (2019) 143–162
method, Rabczuk and his co-workers proposed a new method based on is defined and the gradient of a function is given by:
a mixed Lagrangian–Eulerian kernel formulation [23] as well as other
∇f (x ) = ∇f (x ) − f (x ) ∇1(x ) (4)
approaches [25,26] for treating crack growth to simulate concrete
fracture and fragmentation under blast and impact loadings. More re- where 1 is the unit function.
cently, dual-horizon peridynamics [27,28] in mesh-free methods were Combining Eqs. (1) and (4), the particle approximation to the
proposed and used in crack modelling to avoid erratic crack paths. gradient of a function can be redefined as:
Mesh-free approaches have shown wide applications and excellent
N
performance as mentioned above. However, besides the tensile in- mj
Πh∇f (x i ) = ∑ ρj
[f (x j) Aij − f (x i ) Aji ]
stabilities, there still have some other limitations such as higher com- j=1 (5)
putational cost than traditional FE methods and boundary condition
problems to hinder their achievements in dynamic problems. Therefore,
hybrid FE and mesh-free methods were developed to make full use of
where Aij =
1
hd + 1
θ′ ( xi − xj
h )
Several forms of discrete conservation equations can be derived. In
the good properties of both methods. An overall review of several order to accept the conservative equations in the SPH method in nu-
methods for coupling mesh-free particle methods especially the EFG merical simulations, weak forms are used [31–34] to define the con-
method and the SPH method with the FE method can be seen in Ref servations of mass, momentum and energy in solid mechanics. The SPH
[29]. governing equations in terms of conservations of mass, momentum and
In the present study, the impact resistance of thin and thick UHPC energy are respectively given by:
components that equipped with uniformly distributed ceramic balls as a
N
shielding structure is investigated against the rigid projectile impact at dρ
dt
(x i ) = ∑ mj [v (x j) − v (x i)] Aij
velocities between 500 m/s and 800 m/s. The discrete ceramic balls can j=1 (6)
be replaced after damage which makes the shielding structure reusable
after impact. Another advantage of ceramic balls is that the impact N α, β σ jα, β (x j)
dv α ⎡ σi (x i ) ⎤
surface of ceramic balls is cambered so the ceramic balls could provide (x i ) = ∑ mj ⎢ Aij − Aji ⎥
dt ρi2 ρj2
the asymmetric resistance force against the projectile impact and hence j=1 ⎣ ⎦ (7)
changes the ballistic trajectory. The SPH-FE method is used with the
N
assistance of LS-DYNA to conduct the simulations of high-velocity dE ρ
(x i ) = − 2i ∑ mj [v (x j) − v (x i)] Aij
projectile impacts. A parametric study is conducted on thin UHPC slabs dt ρi j=1 (8)
to investigate the influence of impact location, diameter, spatial ar-
rangement and material properties of ceramic balls, and then perfora- where mi, ρi, σiα, β
and v(xi) are mass, density, stress tensor and velocity
tion and ballistic limits of ceramic balls protected thin UHPC slabs are associated with particle i; α and β are the space indices.
obtained. Furthermore, the depth of penetration (DOP) values of thick The Monaghan type artificial viscosity [35] is used to smooth the
UHPC slabs protected with 6-layer ceramic balls with the diameter of shock and to prevent the penetration of adjacent particles. To consider
20 mm in a hex-pack arrangement are obtained at three striking velo- the artificial viscosity, an artificial viscous pressure term Πi, j is added to
cities of 550 m/s, 675 m/s and 800 m/s, and then compared with other the momentum conversation. The equation to represent Πi, j is given by:
UHPC slabs from the previous tests [4,7,8,30].
1 ⎛
Πi, j = ¯i, j + βμi,2 j⎞⎟
⎜ − αμ i, j c
144
J. Liu et al. International Journal of Impact Engineering 125 (2019) 143–162
this concrete model are obtained and then verified based on projectile subjecetd to projectile impacts [12,53,54]. It consists of three parts
impact tests on concrete targets with the compressive strength ranging including strength, damage and pressure models. The normalized
from 60 MPa to 157 MPa [46]. equivalent stress (von Mises stress) for the strength is expressed by the
As tested in Ref [4], static test results on uniaxial compression tests following relation:
indicated the uniaxial compressive strength of the present UHPC ma-
terial (containing 3 Vol-% steel fibres) was 140 MPa, and a flexural σ * = σi* − D × (σi* − σf*) (13)
strength of 28.5 MPa was also obtained based on four-point bending where σi* is the normalized intact equivalent stress; σf* is the normal-
tests. Regarding the uniaxial tensile strength (ft) and the ultimate ten- ized fracture stress; D is the damage parameter. The normalized intact
sile strain (ɛtu) of the present UHPC material, an inverse evaluation equivalent stress and the normalized fracture stress are derived from:
based on four-point bending tests is conducted according to Japan
Concrete Institute JCI-S-003-2007 [50] as shown in Eqs. (11) and (12). σi* = A × (P * + T *) N × (1 + C × lnɛ̇*) (14)
In this method, the flexural bending test results are transferred to
uniaxial tensile characteristics for the ductile fibre reinforced ce- σf* = B × (P *) M × (1+C × lnɛ̇*) ≤ SFMAX (15)
mentitious composites and the stress distribution is assumed uniform in
where P* is the normalized pressure; T* is the normalized tensile
the tensile zone. This method has been successfully adopted in evalu-
strength; ɛ̇* is the normalized strain rate; SFMAX is the maximum
ating the uniaxial tensile strength of UHPC material as reported by Li
fracture strength. These parameters are all normalized by the equiva-
et al. [51].
lent stress at the Hugoniot Elastic Limit (HEL); A, B and C are constants
ɛtu = ϕu × D × (1 − x nl ) (11) of intact strength, fracture strength and strain rate, respectively. D is the
damage parameter representing the accumulated damage, which is
E × ϕu × D × x nl 2 calculated from:
ft =
2 × (1 − x nl ) (12) Δɛ P
D = ∑
ɛ Pf (16)
where x nl = 2cos(θ /3) − 1; θ = arccos(6m* − 1) ; m* = Mmax/E × ϕu ×
B × D3; x nl = x n / D ; ɛtu is the ultimate tensile strain; ϕu is the curvature where ΔɛP is the increment in plastic strain per cycle of integration; ɛ Pf is
at the maximum load expressed by ϕu = 216δ /(23L2) in which δ is the the plastic strain to fracture, which is given by:
mid-span deflection and L is the clear span between the supports
(= 300 mm); D is the depth of test specimen (= 100 mm); B is the ɛ Pf = D1 × (P * + T *) D2 (17)
width of test specimen (= 100 mm); xn is the distance from compres-
sive edge to neutral axis (mm); E is the modulus of elasticity (N/mm2); where D1 and D2 are the damage coefficients ranging from 0 to 1. The
Mmax is the maximum moment = Pmax /2 × L/3 (N × mm); Pmax is the hydrostatic pressure for the undamaged material is given by:
maximum load at a given point. P = K1 × μ + K2 × μ2 + K3 × μ3 (18)
Through conducting the four-point bending test [4], the lateral force
displacement curve is obtained as shown in Fig. 1(a), where the mid- where K1 is the bulk modulus; K2 and K3 are the pressure coefficient;
span deflection is 2.6 mm and the maximum load is 95 kN. After sub- μ = ρ / ρ0 − 1, in which ρ is the current density and ρ0 is the initial
stituting all the parameters into Eqs. (11) and (12), the uniaxial tensile density. The mechanical properties and model parameters for SiC and
strength and ultimate tensile strain are 10.4 MPa and 0.022, respec- Al2O3 ceramic balls from Refs [12,54,55] are listed in Table 2.
tively, as shown in Fig. 1(b). All the updated parameters of Johnson The steel case is modelled by Plastic Kinematic model (Mat_3). From
Holmquist Concrete model based on the previous studies [44,46] for the previous tests [4,6–8,30,56], the abrasion and damage of the steel
the present 140 MPa UHPC slabs are listed in Table 1. ogive-nosed projectiles after impact were negligible, so the same pro-
Johnson Holmquist Ceramics model (Mat_110), namely, JH-2 jectiles adopted in the present study are considered as the rigid body (<
model, [52] is adopted in the present study to simulate the brittle be- 800 m/s impact) for conservative predictions on the results. All the
haviour of silicon carbide (SiC) and alumina (Al2O3) ceramic balls. This material parameters for projectile and steel cases are summarized in
model is one of the most widely used material models for ceramics Table 3.
Fig. 1. Tensile characteristics of UHPC in the present study (a) Lateral forced displacement curve (b) Predicted uniaxial tensile stress strain curve.
145
J. Liu et al. International Journal of Impact Engineering 125 (2019) 143–162
Table 1
Parameters of Johnson Holmquist Concrete model for UHPC material [44,46].
Parameters Magnitude Parameters Magnitude
3
Density (kg/m ) 2500 Crushing pressure, Pcrush (MPa) 47
Sher modulus (GPa) 16.6 Crushing volumetric strain, μcrush 0.0021
Normalized cohesive strength, A 0.3 Locking pressure, Plock (MPa) 800
Normalized pressure hardening, B 1.73 Locking volumetric strain, μlock 0.1
Strain rate coefficient, C 0.007 Damage constant, D1 0.04
Pressure hardening exponent, N 0.79 Damage constant, D2 1
Uniaxial compresive strength (MPa) 140 Pressure constant, K1 (MPa) 8.5E4
Maximum tensile hydrostatic pressure (MPa) 10.4 Pressure constant, K2 (MPa) −1.71E5
Normalized maximum strength, Smax 7 Pressure constant, K3 (MPa) 2.08E5
Table 2
Mechanical properties and parameters in JH-2 model for ceramic balls [12, 54, 55].
Mechanical properties of ceramic material
146
J. Liu et al. International Journal of Impact Engineering 125 (2019) 143–162
147
J. Liu et al. International Journal of Impact Engineering 125 (2019) 143–162
Fig. 3. Configurations of ceramic balls (a) Hex-pack, D60, 2 layers (b) Checkerboard, D40, 3 layers (c) Hex-pack, D40, 3 layers (d) Hex-pack, D20, 6 layers.
region of UHPC targets after impact, the choice of 160 mm × 160 mm 3.3.2. Effect of SPH particle distance
SPH domain size is reasonable in the current study. It shall be noted SPH particle distance is another important factor that influences
that the larger diameter of ceramic balls may lead to the deviation of both computational efficiency and accuracy. Apparently, an increase of
ballistic trajectory, e.g. D60 ceramic balls as shown in Fig. 2(c), a larger SPH particle distance improves computational efficiency but decreases
SPH domain size of 230 mm × 230 mm for the UHPC slab is chosen the accuracy of numerical results. Structured and unstructured particle
herein. arrangements were compared as reported in Ref [26], where the results
indicated the unstructured discretization almost coincided with the
148
J. Liu et al. International Journal of Impact Engineering 125 (2019) 143–162
Fig. 5. Single ceramic ball (a) D60 (b) D40 (c) D20.
149
J. Liu et al. International Journal of Impact Engineering 125 (2019) 143–162
Table 4
Contact type for each part.
Contact type Master part Slave part
Fig. 10. Setup of high-velocity impact test on thick UHPC slabs [4].
Fig. 9. Schematic diagram of high-velocity impact test on thin UHPC slabs [6].
particle nano-composites based on six uncertain parameters. Zhu et al.
3.3.5. Effect of the sensitivity of material parameters [59] discussed the sensitivity of 7 parameters in the ceramic material
Besides the major numerical parameters in the SPH method as dis- model (JH-2) which are not “physical” on numerical results, in which it
cussed above, there are numerous input parameters of material models was indicated that N had the greatest influence while B and M had the
and not all of them are “physical”. It is necessary to discuss their sen- smallest influence.
sitivity on the numerical results. Vu-Bac et al. [64] introduced a set of In the present study, the similar method as the Ref [59] is used to
Matlab functions to analyse the probabilistic sensitivity of uncertain investigate the sensitivity of 7 parameters which are not “physical” in
input material parameters on model outputs. Hamdia et al. [65] used a Johnson Holmquist Concrete model for UHPC material, i.e. A, B, C, N,
methodology for stochastic modelling of the fracture in polymer/ Smax, D1 and D2. In addition to the initial value for each parameter, four
150
J. Liu et al. International Journal of Impact Engineering 125 (2019) 143–162
Table 5
Comparison of numerical results for different parameters.
Variable Amount of particles Vr(m/s) in Test 1–1 Error DOP (mm) in Test 2–1 Error
SPH domain size (mm × mm) 100 × 100 62,500 166 −2.9% 127 −1.6%
160 × 160 160,000 167 −2.3% 128 −0.8%
240 × 240 360,000 167 −2.3% 128 −0.8%
SPH particle distance (mm) 1 640,000 166 −2.3% 128 −0.8%
2 160,000 167 −2.3% 128 −0.8%
4 40,000 190 +11.1% 243 +88.4%
FORM parameter 0 160,000 167 −2.3% 128 −0.8%
1 160,000 158 −7.6% 127 −1.6%
7 160,000 Error termination − Error termination −
8 160,000 Error termination − Error termination −
CSLH parameter 1.05 160,000 162 −5.3% 121 −6.2%
1.2 160,000 167 −2.3% 128 −0.8%
1.3 160,000 170 −0.6% 131 +1.6%
values including 5% less, 3% less, 3% more and 5% more than the in- influence on Vr and D1 has the smallest influence on Vr. B has the
itial value are set to calculate their corresponding outputs and the re- greatest influence on DOP and D2 has the smallest influence on DOP.
lative change with respect to the output of the initial value. By fitting
these 5 points with a linear equation, the slope can be obtained through
dividing the relative change of the output value by the relative change 3.3.6. Determination of other parameters
of the input value to evaluate the sensitivity of certain input para-
meters. For the remaining numerical parameters of the SPH method in LS-
The sensitivity of 7 material parameters in terms of numerical DYNA, such as scale factor for the smooth length (Hmin and Hmax),
projectile impacts in Test 1–1 and Test 2–1 is given in Table 6. The number of cycles between particle sorting (NCBS), interaction of par-
negative slope means that the increase of the input parameter decreases ticles (CONT) and neighbour particles search (INI), they all are con-
the result, and the positive slope means that the increase of the input sidered as default values.
parameter increases the result. From this table, B has the greatest
Fig. 11. Influence of major parameters (a) SPH domain size (b) SPH particle size (c) SPH particle approximation theory (d) smooth length constant.
151
J. Liu et al. International Journal of Impact Engineering 125 (2019) 143–162
Table 6
Sensitivity of material parameters of UHPC on Vr in Test 1–1 and DOP in Test 2–1.
A B C N Smax D1 D2
Test 1–1, Slope 1 = ΔVr /ΔInput −0.233 −0.745 0.209 0.325 −0.092 0.033 −0.185
Test 2–1, Slope 2 = ΔDOP/ΔInput −0.073 −0.535 −0.021 0.155 −0.123 0.013 −0.009
Fig. 12. Numerical and experimental projectile impact process with fragmentation (a) 70 mm thick UHPC slab, 348 m/s (Test 1–6) [6] (b) 700 mm thick UHPC
target, 808 m/s (Test 2–3) [4].
152
J. Liu et al. International Journal of Impact Engineering 125 (2019) 143–162
Table 7
Comparison of experimental and numerical results.
Tests on thin UHPC slabs [6] Simulation
Test H (mm) M (g) V0 (m/s) Vr (m/s) M′ (g) Vr (m/s) Error
red means the ceramic balls suffer more severe damages. It is clear that response of the steel case on front and rear surfaces under the same
only a small number of ceramic balls close to the impact region are loading environment when reaching the yielding strength at 270 µs.
damaged with fragments while many ceramic balls far away from the The front surface is subjected to more severe effective stress than the
impact region experience minor damages. Fig. 16 shows the impact rear surface and some stress areas further away from the impact region
Fig. 13. Comparison between experimental and numerical results (a) Residual velocity of projectile versus thickness of slab at ∼350 m/s (b) Residual velocity of
projectile versus striking velocity of projectile for 50 mm thick UHPC slabs (c) DOP versus striking velocity of projectile for thick UHPC slabs.
153
J. Liu et al. International Journal of Impact Engineering 125 (2019) 143–162
Table 8
Numerical results for thin and thick UHPC slabs protected with ceramic balls.
Test V0 (m/s) Arrangement Material D (mm) N Tc (mm) Tu (mm) P Vr (m/s) Vr, y (m/s) α DOP (mm)
are caused by the contact force between the ceramic balls and the steel including the kinetic energy of the projectile, the kinetic energy of
case, which confirms the movement of ceramic balls after impact is ceramic balls, the internal energy of ceramic balls and the internal
confined and the flying of ceramic fragments is minimized by the steel energy of the UHPC slab in Test 20. Starting from 109.3 kJ, the initial
case. kinetic energy of the projectile is partly consumed by the fracture of
Fig. 17(a) presents the impact resistance provided by the target ceramic balls and simultaneously transferred to the internal energy of
acting on the projectile in Test 20, where T1, T2 and T3 corresponds ceramic balls (31.4 kJ) and then continues to transfer to the internal
with the three stages of the perforation process as shown in Fig. 14. energy of the UHPC slab (8.2 kJ). The energy absorbed by ceramic balls
During the projectile perforation of ceramic balls, oscillations in impact increases more rapidly than the UHPC slab and its internal energy
resistance occur because of the gaps among ceramic balls. When the (31.4 kJ) constitute over 50% of the dissipated kinetic energy (62.3 kJ),
projectile goes through these gaps, the impact resistance significantly proving ceramic balls are effective in consuming energy subjected to
decreases until the projectile strikes the next ceramic ball. After that, the projectile impact. During the impact, the total dissipated energy
the impact resistance will start to rise again before the projectile passes caused by both fracture and friction is 21.8 kJ, which takes about 20%
through the next gap. In Fig. 17(a), the area under the impact re- of the initial kinetic energy of the projectile. It shall be noted that the
sistance displacement curve represents the work done by impact re- kinetic energy of ceramic balls during impact is small (2.1 kJ) and its
sistance of targets. Fig. 17(b) shows the energy evolution during impact value decreases after 100 µs. These observations further prove the
Fig. 14. Projectile perforation process into D40 ceramic balls protected thin UHPC slab in the hex-pack arrangement at impact position 3 at 800 m/s (Test 20).
154
J. Liu et al. International Journal of Impact Engineering 125 (2019) 143–162
Fig. 15. Localized damages of the hex-pack arrangement of D40 ceramic balls (Test 20) per layer at 800 m/s at 270 µs.
Fig. 16. Impact response of the steel case (Test 20) at 270 µs at 800 m/s (a) Front surface (b) Rear surface.
movement of ceramic balls and ceramic fragments is effectively re- the hex-pack arrangement as shown in Fig. 14. Fig. 19(a) presents the
strained by the steel case. impact resistance displacement curves for these two types of arrange-
ments, e.g. Tests 20 and 24. When the projectile penetrating into
ceramic balls from 0 to 110 µs, the work done by the impact resistance
4.1.2. Effect of arrangement of ceramic balls
of hex-pack arranged ceramic balls is much larger than the checker-
The effect of the arrangement of ceramic balls on the residual ve-
board arrangement, which leads to the reduction of the residual velo-
locity and the ballistic trajectory of the projectile is the top priority to
city of the projectile by 10.9% as shown in Fig. 19(b). This is because
be considered. Fig. 18 shows the projectile perforation into the
the hex-pack arranged ceramic balls have the lower porosity than the
checkerboard arrangement of 3-layer D40 ceramic balls protected on
checkerboard arranged ceramic balls resulted from the configuration
UHPC slabs at impact position 3 at 800 m/s (Test 24). Comparing with
setup as shown in Fig. 3.
Test 20, it is observed that the ballistic trajectory of the projectile
changes more obviously in the checkerboard arrangement than that in
Fig. 17. Impact response of D40 ceramic balls protected thin UHPC slab in the hex-pack arrangement at impact position 3 at 800 m/s (Test 20) (a) Impact resistance
(b) Energy evolution.
155
J. Liu et al. International Journal of Impact Engineering 125 (2019) 143–162
Fig. 18. Projectile perforation into D40 ceramic balls protected thin UHPC slab in the checkerboard arrangement at impact position 3 at 800 m/s (Test 24).
Fig. 19. Impact response of D40 ceramic balls protected thin UHPC slab in various arrangements at impact position 3 at 800 m/s (a) Impact resistance (b) Residual
velocity of projectile.
4.1.3. Comparison between Al2O3 and SiC ceramic balls ceramic balls. Several studies have been conducted to investigate the
Fig. 20 shows the comparison between Al2O3 (Test 34) and SiC influence of material properties on the ballistic behaviour of ceramics
(Test 30) ceramic balls with the diameter of 20 mm protected on UHPC [11,55,66–71]. Some parameters involving density and porosity, me-
slabs in terms of the impact resistance and the residual velocity of the chanical strength, hardness, Young's modulus, fracture toughness and
projectile. In this comparison, the projectile strikes the impact position HEL are of importance in resisting the projectile impact. However, any
3 of the ceramic balls at 800 m/s. The SiC ceramic balls protected UHPC single property does not directly have an effect on the ballistic beha-
slab does much more work to resist the projectile impact and hence viour since the fracture mechanics of ceramic materials during impact is
reduces the residual velocity of the projectile by 36.3% compared with complicated [55]. For example, ceramics should have the high com-
the Al2O3 ceramic balls protected UHPC slab, proving SiC ceramic balls pressive strength, hardness and Young's modulus to resist the projectile
have a superior performance in resisting the projectile than Al2O3 impact [55,68,70], but any increase of the single value only provides
Fig. 20. Comparison between D20 Al2O3 and SiC ceramic balls at impact position 3 at 800 m/s (a) Impact resistance (b) Residual velocity of projectile.
156
J. Liu et al. International Journal of Impact Engineering 125 (2019) 143–162
Fig. 21. Impact response of D60 ceramic balls protected thin UHPC slab at various impact positions at 800 m/s (a) Impact resistance (b) Residual velocity of
projectile.
limited help [67] and crack propagation becomes intensive when the velocity from 500 m/s to 800 m/s through taking Tests 4–9 for con-
ceramic hardness is very high [55]. The increase of ceramic toughness siderations, in which the maximum ballistic deviation angle can reach
generally reduces the crack propagation, however, ceramic hardness 17°.
decreases simultaneously, which means the improved toughness may Taking Tests 9, 20 and 30 for considerations, the comparison of
have advantages and disadvantage over the anti-penetration process D60, D40 and D20 ceramic balls protected UHPC slabs in terms of the
[55]. On the contrary, some previous reports [69,71] indicated that the impact resistance and the residual velocity of the projectile at impact
toughened ceramic provides higher ballistic resistance of targets. position 3 at 800 m/s is shown in Fig. 26. Comparing D20 and D40
Therefore, more careful considerations should be taken to the me- ceramic balls protected UHPC slabs, it is observed that coarser ceramic
chanical properties of ceramic materials in its evaluation of the ballistic balls contribute less to the impact resistance force and work due to the
performance. In the present study, the effects of ceramic hardness and lower compaction. Also, the fluctuation of impact resistance during
compressive strength overweight the material toughness, and SiC perforation becomes more evident for the larger diameter ceramic balls.
ceramic balls perform better than Al2O3 against the same projectile However, when increasing the diameter to 60 mm, the accumulated
penetration. work done by the impact resistance of ceramic balls is quite close to
that of D40 ceramic balls. The residual velocity of the projectile for the
4.1.4. Effect of impact positions D20 ceramic balls protected target reduces by 29.6% and 30.9% re-
The effect of impact positions is considered important because the spectively compared with D40 and D60 ceramic balls protected targets.
surface of the ceramic balls is cambered that different impact positions However, the ballistic trajectory is less altered when the size of the ball
may induce an asymmetric resistance and deviate the ballistics trajec- reduces especially when the diameter is smaller than 60 mm. It shall be
tory of projectiles. Figs. 21–23 respectively show the impact response noted that the ballistic trajectory deviation will lead to the reduction of
of UHPC slabs protected by D60 (Tests 2, 3 and 9), D40 (Tests 13, 14 the projectile residual velocity in the perpendicular direction to the
and 20) and D20 (Tests 25, 26 and 30) ceramic balls under various target and the reduction reaches 4.3% compared with the resultant
impact positions at 800 m/s. Taking D60 ceramic balls for example, as value in this case.
shown in Fig. 21, although the projectile striking at impact position 2 is
subjected to the largest impact resistance provided by the ceramic balls, 4.1.6. Perforation and ballistic limits of protected targets
the total work done by ceramic balls is almost the consistent, so the The relations between the residual velocity of the projectile and the
residual velocities of projectile after perforation are quite close under layers of ceramic balls as well as the perforation limit for protected thin
these three scenarios. Furthermore, seen from Figs. 22 and 23, the UHPC slabs are investigated as shown in Fig. 27. In this figure, it can be
various impact positions have even less effect on the impact responses seen that the safe layers to prevent perforation of D60, D40 and D20
of targets when the diameter is reduced to 40 mm and 20 mm. This is ceramic balls protected thin UHPC slabs are 5, 6 and 8, respectively. In
attributed to the lower void ratio and better compaction for the finer the case of UHPC slabs protected with 5-layer D60 ceramic balls, the
ceramic balls. Therefore, a conclusion can be drawn that with the re- maximum ballistic deviation angle reaches 24°.
duction in ceramic ball size, the influence of the impact location is Additionally, a series of numerical simulations at various striking
reduced. When the diameter of ceramic balls is sufficiently small, the velocities from 500 m/s to 800 m/s are conducted to investigate the
residual velocity of the projectile is not affected by the impact position. ballistic limit for D20 (Tests 27–30), D40 (Tests15-20) and D60 (Tests
4–9) ceramic balls protected UHPC slabs. Furthermore, the relations
4.1.5. Effect of diameter of ceramic balls among the residual kinetic energy, the kinetic energy dissipation and
The effect of the diameter of ceramic balls on its impact response is the initial kinetic energy of projectiles are analysed under such varying
analysed in this section. Figs. 24 and 25 respectively show the typical striking velocities as shown in Fig. 28. Through curve-fitting the data as
projectile perforations of UHPC slabs with D20 (Test 30) and D60 (Test labelled in the primary horizontal and vertical axis, it is indicated that
9) ceramic balls in the hex-pack arrangement at impact position 3 at there has a linear relationship between the initial and residual kinetic
800 m/s. Compared with the D40 ceramic balls protected UHPC slab, energies of projectiles after perforation. Based on the equations to re-
the ballistic trajectory within the D20 ceramic balls protected UHPC present the trendlines, the limit striking velocities to perforate D20,
slab is rarely altered, but the ballistic trajectory within the D60 ceramic D40 and D60 ceramic balls protected UHPC slabs are 677 m/s, 541 m/s
balls protected UHPC slab is significantly deviated and the deviation and 506 m/s, respectively. The relations between the kinetic energy
angle has an increasing tendency with the increase of the striking dissipation and the initial kinetic energy of projectiles after perforating
157
J. Liu et al. International Journal of Impact Engineering 125 (2019) 143–162
Fig. 22. Impact response of D40 ceramic balls protected thin UHPC slab at various impact positions at 800 m/s (a) Impact resistance (b) Residual velocity of
projectile.
Fig. 23. Impact response of D20 ceramic balls protected thin UHPC slab at various impact positions at 800 m/s (a) Impact resistance (b) Residual velocity of
projectile.
Fig. 24. Projectile perforation process into D20 ceramic balls protected thin UHPC slab in the hex-pack arrangement at impact position 3 at 800 m/s (Test 30).
are further shown in the secondary horizontal and vertical axis in this of dissipation increases with the larger striking velocity of the pro-
figure. It is fairly clear that the smaller size of ceramic balls will help jectile.
consume the initial kinetic energy of the projectile and the consumption
increases with the increasing striking velocity of the projectile (the
4.2. DOP analysis of ceramic balls protected thick UHPC slabs
mass of the projectile is assumed as constant). The above phenomenon
gives the strong evidence that finer ceramic balls are of better capability
In order to evaluate the performance of ceramic balls as a protecting
in resisting the projectile impact due to its higher compaction especially
structure on the UHPC slabs, DOPs of thick UHPC slabs with 6-layer
when the diameter is 20 mm. Also, the finer ceramic balls are more
D20 ceramic balls in the hex-pack arrangement are investigated at three
effective in dissipating the kinetic energy of the projectile and the value
striking velocities of 550 m/s, 675 m/s and 800 m/s (Tests 35–37). The
158
J. Liu et al. International Journal of Impact Engineering 125 (2019) 143–162
Fig. 25. Projectile perforation process into D60 ceramic balls protected thin UHPC slab in the hex-pack arrangement at impact position 3 at 800 m/s (Test 9).
Fig. 26. Comparison of projectile perforation of thin UHPC slabs protected with D20, D40 and D60 ceramic balls (a) Impact resistance (b) Residual velocity of
projectile.
Fig. 28. The relations among residual kinetic energy, kinetic energy dissipation
Fig. 27. Residual velocity of projectile versus layers of ceramic balls at 800 m/ and initial kinetic energy of projectiles.
s.
and 30 Vol-% 5–15 mm basalt aggregates (UHP-BAFRC-2) [7],
impact responses of the target under such three striking velocities when 110.7 MPa UHPC with 2 Vol-% steel fibres and 30 Vol-% 5–20 mm
the projectiles come to rest are shown in Fig. 29, in which DOPs at corundum aggregates (UHP-CAFRC-1) [8] and 128.8 MPa UHPC with 2
these three scenarios are 83 mm, 125 mm and 189 mm, respectively. Vol-% steel fibres and 45 Vol-% 5–20 mm corundum aggregates (UHP-
As shown in Fig. 30, the numerical results are compared with other CAFRC-2) [8]. At the striking velocity of 550 m/s, DOP for the ceramic
UHPC slabs as reported in the previous studies, including 100 MPa re- balls protected UHPC target is 37.1% smaller than the RPC-SWM target
active powder concrete (RPC) with 44-layer steel wire meshes (RPC- at 539 m/s, 40.7% smaller than the UHPC-PF target at 549 m/s and
SWM) [30], 120 MPa UHPC with 3 Vol-% ultra-high molecular weight 35.7% smaller than the UHPC-SF target at 553 m/s. Compared with
polyethylene (UHMWPE) fibres (UHPC-PF) [4], 140 MPa UHPC with 3 other UHPC targets under the lower impact velocity of ∼500 m/s, e.g.
Vol-% steel fibres (UHPC-SF) [4], 114 MPa UHPC with 3 Vol-% steel UHP-BAFRC-1, UHP-BAFRC-2, UHP-CAFRC-1, UHP-CAFRC-2, DOP for
fibres (UHP-BAFRC-1) [7], 125.2 MPa UHPC with 2 Vol-% steel fibres the ceramic balls protected UHPC target at 550 m/s decreases in a range
159
J. Liu et al. International Journal of Impact Engineering 125 (2019) 143–162
Fig. 29. Projectile penetration of thick UHPC slabs with 6-layer hex-pack arranged D20 ceramic balls at different striking velocities (a) 550 m/s (b) 675 m/s (c)
800 m/s.
5. Conclusions
1) When the rigid projectile strikes the ceramic balls protected UHPC
slabs, only a small number of ceramic balls close to the impact re-
gion are damaged with fragments while the ceramic balls far away
from the impact region seem not to experience such severe brittle
Fig. 30. DOP versus striking velocity of projectile. damage, which demonstrates that the utilization of ceramic balls is
economical in protective structures since the undamaged ceramic
balls are reusable;
from 17% to 34.1%. At the striking velocity of 675 m/s, DOP of the 2) Although the checkerboard arranged ceramic balls protected UHPC
ceramic balls protected UHPC target respectively decrease by 26%, slab tends to change the ballistic trajectory more possibly than the
47.7% and 24.7% when compared with the RPC-SWM target at 679 m/ hex-pack arranged ceramic balls protected UHPC, the latter has the
s, the UHPC-PF target at 692 m/s and the UHPC-SF target at 683 m/s. better impact performance regarding the residual velocity of the
Additionally, despite the impact velocities for these two targets are projectile than the former because the hex-pack arrangement has the
different, e.g. the ceramic balls protected UHPC target under 675 m/s, lower void ratio;
UHP-CAFRC-1 targets under 708 m/s-722 m/s, the former target has a 3) If the diameter of ceramic balls is not much larger than the rigid
smaller value of DOP (decrease by 50%) than the latter one. At the projectile diameter, the residual velocity of the projectile is insig-
striking velocity of 800 m/s, although the ceramic balls protected UHPC nificantly affected by the various impact positions and the extent of
target remarkably decrease DOP by 31%, 32% and 26.5% respectively the effect will be reduced with the finer ceramic balls;
through comparing with the UHPC-PF target at 806 m/s, UHP-BAFRC-1 4) The finer ceramic balls are more effective in the impact resistance
and UHP-BAFRC-2 targets at 850 m/s, it has a marginal improvement to due to the higher compaction. The finer ceramic balls are easier to
reduce DOP in comparison with the RPC-SWM target at 807 m/s (de- consume the kinetic energy of the projectile and the value of the
crease by 6.4%), the UHPC-SF target at 808 m/s (decrease by 9.1%) and energy consumption rises with the larger striking velocity of the
UHP-CAFRC-1 targets at ∼840 m/s (almost the same). It is worth projectile. On the contrary, when the diameter of ceramic balls is
noting that the deformation and abrasion of projectiles are not con- much larger than the rigid projectile diameter, the ballistic trajec-
sidered in the present study so that conservative numerical results with tory is easier to be altered;
larger values of DOP are obtained especially for high-velocity projectile 5) Ceramic balls contribute significantly to the dissipation of the ki-
penetrations. The above observations indicate the superior impact netic energy of the projectile. Compared with the existing advanced
performance of UHPC slabs with 6-layer D20 ceramic balls in the hex- UHPC targets, 6-layer D20 ceramic balls in the hex-pack arrange-
pack arrangement than other UHPC structures. ment protected on the thick UHPC slabs can improve the
160
J. Liu et al. International Journal of Impact Engineering 125 (2019) 143–162
161
J. Liu et al. International Journal of Impact Engineering 125 (2019) 143–162
Software 2016;100:19–31. [69] Nastic A, Merati A, Bielawski M, Bolduc M, Fakolujo O, Nganbe M. Instrumented
[65] Hamdia KM, Silani M, Zhuang X, He P, Rabczuk T. Stochastic analysis of the and Vickers indentation for the characterization of stiffness, hardness and toughness
fracture toughness of polymeric nanoparticle composites using polynomial chaos of zirconia toughened Al2O3 and SiC armor. J Mater Sci Technol
expansions. Int J Fract 2017;206(2):215–27. 2015;31(8):773–83.
[66] Clayton J. Penetration resistance of armor ceramics: dimensional analysis and [70] Rosenberg Z, Ashuach Y, Yeshurun Y, Dekel E. On the main mechanisms for de-
property correlations. Int J Impact Eng 2015;85:124–31. feating AP projectiles, long rods and shaped charge jets. Int J Impact Eng
[67] Kaufmann C, Cronin D, Worswick M, Pageau G, Beth A. Influence of material 2009;36(4):588–96.
properties on the ballistic performance of ceramics for personal body armour. Shock [71] Zhang X, Li Y. On the comparison of the ballistic performance of 10% zirconia
Vib 2003;10(1):51–8. toughened alumina and 95% alumina ceramic target. Mater Des
[68] Krell A, Strassburger E. Order of influences on the ballistic resistance of armor 2010;31(4):1945–52.
ceramics and single crystals. Mater Sci Eng 2014;597:422–30.
162