Original article

Proc IMechE Part L:

J Materials: Design and Applications
A review of the integrity of metallic 2019, Vol. 233(1) 73–94
! IMechE 2018
vehicle armour to projectile attack Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/1464420718759704
journals.sagepub.com/home/pil

Donncha Lenihan1, William Ronan2, Padraic E O’Donoghue3

and Sean B Leen1

Abstract
This article presents a review of engineering and design aspects relevant to the mechanical and structural integrity of
military vehicular armour, including materials-related technologies. Theoretical, experimental and numerical techniques
for assessment are discussed and evaluated. A number of prominent material constitutive models are comparatively
assessed. The Johnson–Cook model is shown to be particularly consistent in terms of agreement with experimental data,
identification of material constants and ease of application. The article also discusses different numerical codes used and
their relevance over time. Finally, it is argued that there is a need for a materials design tool for military vehicular armour.

Keywords
Armour, projectile, terminal ballistics, impact, protection, military vehicles

Date received: 13 December 2017; accepted: 25 January 2018

Introduction
i.e. roughly 500–2000 m/s. In addition to this, much
An armoured fighting vehicle (AFV) is a military vehi- interest is also focused on the behaviour of shaped
cle with a hull specifically designed to resist penetra- charge jets, which impact with velocities between
tion from enemy weapons, thus allowing it to move 2000 and 8000 m/s.4 This is obviously a very broad
more freely in the battlespace. AFVs are used for a spectrum and there is a wealth of more specific ana-
wide variety of roles, from full-scale assault on enemy lyses of each of the velocity regions mentioned.
positions to quickly laying down temporary bridges so This review is organised as follows: (i) The classes
that friendly vehicles can cross a river. Since World of materials, mostly metals, in use for armour and
War I, when AFVs were first developed and used,1 the projectile applications are introduced. (ii) The three
development of their armour has gone hand-in-hand main types of armour defeat mechanisms are
with the development of anti-armour weapons. The described: rigid penetration, hydrodynamic penetra-
crew and other personnel in an AFV, while protected tion and spalling. There are other ways in which
from the majority of battlefield threats, are neverthe- armour is defeated, e.g. blast waves. However, these
less in danger due to being more conspicuous, less are considered to be outside the scope of this review.
mobile and representing a high-value target for the (iii) Some key experimental techniques used for ter-
enemy. Attacks on vehicle-mounted troops have a dif- minal ballistics are considered, focusing in detail on
ferent profile of wounding and lethality than attacks the effects of spalling. (iv) Material models used for
on dismounted troops,2 and in recent conflicts have analytical and numerical quantification of the deform-
accounted for the majority of casualties.3 Vehicular ation, damage and subsequent failure of armour
armour is required to provide a high level of protec- materials during impact events are reviewed. (v) The
tion but with minimum effect on vehicle size and numerical aspects of terminal ballistics and various
weight. It is incumbent, therefore, on the designer of techniques and specific analysis codes used to simulate
armour to ensure that the response of the armour impact events are reviewed. The increasing possibility
material to various battlefield threats can be quanti-
fied and evaluated effectively. 1
Mechanical Engineering, NUI Galway, Newcastle, Galway, Ireland
2
This paper provides an overview of the engineering Biomedical Engineering, NUI Galway, Newcastle, Galway, Ireland
3
aspects of armour design, by focusing on the most Civil Engineering, NUI Galway, Newcastle, Galway, Ireland
common threats to AFVs and the mechanisms of fail-
Corresponding author:
ure. The study of high-speed projectile impact on Donncha Lenihan, National University of Ireland Galway, New
armour, known as terminal ballistics, deals primarily Engineering Building, Newcastle, Galway H91 HX31, Ireland.
with impacts that occur in the ‘ordnance range’, Email: [email protected]
74 Proc IMechE Part L: J Materials: Design and Applications 233(1)

in more recent years to model at smaller scales is also processing techniques and material composition.12
discussed. This allows the capture of the microstruc- A technique that combines the advantageous ductility
tural behaviour of materials, thus providing add- and toughness of RHA (i.e. an ‘absorber’ material)
itional insights on the armour failure process. (vi) with the ‘disruptive’ qualities of HHA is dual hard-
Finally, future trends in armour design and analysis ness armour, which is created by roll-bonding two
are discussed. separate plates together into a single plate of
armour (e.g. a plate of HHA bonded to the front of
a plate of RHA). This technique can produce armour
Armour materials which is 78% more effective (using a metric such
There is no worldwide standard for general armour as the material mass efficiency5) per unit weight
steels or materials.5 Much of the technical informa- than RHA.9
tion associated with military armour is proprietary or Aluminium is also used in AFVs. The first gener-
classified for security reasons. There are, however, ation of aluminium-hulled AFVs (1960s) used the
standard levels of armour protection. North Atlantic 5083 alloy, whose minor elements are magnesium
Treaty Organization (NATO), for example, have and manganese. More modern aluminium AFVs
developed a STANAG (STANdard AGreement) use a 7039-type alloy, which contains zinc and mag-
document describing these levels,6 which was recently nesium.12 The latter material has a higher specific
updated to take account of the increasing lethality of strength (see Table 1), but poorer corrosion resist-
some forms of kinetic energy ammunition.7 It is ance.13 Both materials, however, are considerably
important to note that metallic materials (including weaker than RHA. To provide the same level of pro-
those used as armour) behave differently under tection, therefore, the plates need to be thicker than
dynamic loading conditions than under static condi- RHA or HHA. One consequence of the increased
tions. For example, the flow stress is more sensitive to plate thickness is greater resistance to bending,
the strain rate observed, and thermal softening weak- which means that aluminium-hulled AFVs can often
ens the material due to adiabatic heating.8 As a result, dispense with the stiffening elements common in steel-
the optimal metal austenising and tempering tempera- hulled AFVs.12 Newer processing techniques such as
tures for ballistic performance differ from those for friction stir processing have been shown to improve
static strength and hardness.5 The primary function the ballistic performance of the high-strength alumin-
of armour material is to resist the various mechanisms ium alloy by introducing hard ceramic particles to the
of defeat, which will be discussed further in the next surface of the alloy, creating a composite material on
section. the surface.14
A distinction can be made between ‘energy- The other metal used occasionally for armour is
disruptive’ and ‘energy-absorbing’ armour materials.9 titanium, alloys of which are stronger than aluminium
The ‘disruptive’ material is intended to induce frag- but less dense than steel (Table 1). This gives a high
mentation in the penetrator, thus dispersing its kinetic specific strength compared to steel and aluminium
energy, while the ‘absorber’ transfers the kinetic competitor materials. Although it is an effective
energy of the projectile to heat.10 In general, ceramics armour material, its high price renders it unsuitable
and metals with a high hardness are disruptive and for most AFVs,9,12 nevertheless, it is used in some
metals with large ductility and toughness are aircraft (especially ground-attack aircraft that are vul-
absorbing. nerable to small-arms fire), where weight is a more
Traditionally, and continually, armoured vehicles important consideration.
have been fabricated from high-strength steels, gener- Non-metallic armour usually refers to composite
ally with 0.25–0.4% carbon. The principal alloying materials or ceramics. Composites have the advantage
elements are chromium (which increases hardness)
and nickel (which increases toughness).11 The tensile
strength of such steels has varied from 850 MPa to Table 1. Prominent metals used for vehicle armour, along
1700 MPa. Although it can be cast into complete with selected properties.
shapes, such as vehicle turrets, amour is usually pro-
duced in the form of rolled homogeneous armour Yield Specific
(RHA).12 The tempering process used after the heat Density strength strength
Material (kg/m3 ) (MPa) (MN.m.kg1) References
treatment gives the armour a uniform microstructure,
hence the ‘homogeneous’ designation.9 An alternative Al 5083 2660 156 0.0586 (15), (16)
to RHA is homogeneous high-hardness armour Al 7039 2740 337 0.1230 (15), (17)
(HHA), which uses low-alloy steels, heat-treated to RHA 7850 933 0.1189 (18)
give increased hardness. The tempering temperatures HHA 7850 1250–1500 0.1592–0.1911 (19)
for this material are lower than for RHA.9 Although
Ti-6Al-4V 4550 990 0.2176 (20)
the resulting material is very brittle, which initially
Mild steel 7800 325 0.0417 (20)
caused problems even without projectile impacts,
this problem has been overcome through improved RHA: rolled homogeneous armour; HHA: high-hardness armour.
Lenihan et al. 75

of a relatively low density and an ability to dissipate

the energy of incoming projectiles (energy absorp-
tion), while ceramic materials make use of their high
hardness and compressive strengths (energy disrup-
tion). Some ceramic materials in use for armour pur-
poses are boron carbide (B4C), silicon carbide (SiC)
and titanium di-boride (TiB2).21 A detailed overview
of such materials, their applications, advantages and
disadvantages is given by Hazell.22 Despite the advan-
tages conveyed by ceramics and composite multi-
layers, however, the use of such materials remains
confined to more specialised and expensive vehicles.
Steel armours continue to be used for the majority of
AFVs, as they offer protection with a large degree of
cost efficiency.23,19

Defeat mechanisms and

projectile behaviour
In this section, we consider projectile impact and dif-
ferent defeat mechanisms under three different cate-
gories or regimes: (i) low-speed penetration, where
the projectile does not deform significantly; (ii) high-
Figure 1. Different failure modes for armour.4 (a) Brittle
speed penetration, where there is extensive plastic
fracture, (b) Ductile hole growth, (c) Radial fracture,
deformation of the projectile (the elastic contribution
(d) Plugging, (e) Fragmentation, (f) Petaling.
is not the first-order effect and the projectile and target
behave in a fluid-like manner, and are considered as a
hydrodynamic interaction) and (iii) spalling, where the
projectile is arrested by the armour system but frag-
ments of the armour material are ejected into the vehi-
cle at significant velocities. While projectile penetration
of the vehicle does not necessarily occur in the latter,
the fragments may still cause significant damage.

Penetration – Low speed

Small-arms ammunition includes calibres up to
20 mm.24 Small-arms ammunition is generally fired
at velocities of 600–1000 m/s.4 Two sample experi-
mental studies that investigate these impacts are
Børvik et al.25 and Kılıç et al.26 Key factors that influ-
ence the nature of the damage and whether penetra-
tion will occur are: (i) projectile velocity, (ii) projectile
material, (iii) projectile nose shape, (iv) armour type/
thickness and (v) armour material.
A vital parameter for any weapon–target combin- Figure 2. Comparison of penetration failure in plates of
ation is the so-called ballistic limit velocity, which varying thickness, from finite element analysis (FEA). The
marks a ‘threshold’ velocity above which penetration greater deformation of thinner plates prior to failure (dishing)
will occur.4 Accordingly, and unlike the theory of can be clearly seen (H ¼ 1.5 mm), as can the characteristic
hydrodynamic penetration (discussed below), the ini- bulge of deformed material from ductile hole enlargement
tial projectile velocity is always a crucial factor. There (thickest plate) (H ¼ 3 mm). Figure reprinted from Rosenberg
are several ways in which armoured targets can be and Dekel,27 with permission from Elsevier.
penetrated, depending on the parameters listed
above, and this is illustrated in Figure 1. thus absorbing much of the kinetic energy of the
The thickness of the armour plate is a key variable projectile.4 In this way, thinner plates can be more
controlling the mode of penetration. Thick plates tend effective per unit mass at stopping projectiles than
to fail by the methods shown in Figure 1(a) to (e), thicker ones. This means that spaced armour consisting
whereas thinner plates tend to bend and stretch of multiple layers of thinner plates offset from one
around the area of impact (see Figures 1(f)) and 2), another can be a mass-effective means of protection.
76 Proc IMechE Part L: J Materials: Design and Applications 233(1)

However, this has the disadvantage of adding width than a specific ‘transition’ velocity, generally greater
and greater geometric complexity to a vehicle hull. than 1 km/s.15 This includes the ordnance range of
Dishing is a failure mechanism evident in thin velocities, variously defined as 1.4–1.8 km/s,38
plates28 in which the impact causes a global bending 1.5 km/s39–41 and <2.3 km/s18; these studies all con-
of the plate as well as a local indentation or perfor- cern kinetic energy projectiles. The theory, in essence,
ation.29 Figure 2 shows the effect of transitioning from treats both the target and projectile materials as fluids,
dishing (thin plates) to ductile hole enlargement neglecting their strengths as the pressure exerted by
(thicker plates). Iqbal et al.30 carried out a numerical the projectile is orders of magnitude greater than typ-
study which demonstrated that sharper-nose projectiles ical armour strengths.42 Hydrodynamic penetration
tend to cause failure through ductile hole enlargement theory has also been shown to be applicable to the
(Figure 1(b)), while blunter projectiles tended to cause impact of shaped charge jets, which travel at speeds
failure through plugging (shear). This is an important of 2.5–10 km/s43 (Li et al. report on a value of
consideration, as many studies focus on target response 6.5 km/s44 and Jia et al.45 use 4 km/s in their
to blunt projectiles (Figure 1(d)). Yunfei et al.31 study). An explosively formed projectile travels at
demonstrated experimentally that ductile hole enlarge- comparable but slightly lower speeds: Li et al.46 use
ment causes the target material to deform radially, speeds of roughly 2900 m/s, while Wu and Liu47 oper-
forming a distinctive bulge (see Figure 2). ate with speeds of approximately 1500 m/s.
Petalling is the phenomenon of radial cracks Shaped charge jets are the penetrating mechanism
causing distinctive ‘petals’ of deformed metal on the used by most anti-armour munitions, generally taking
penetrated rear surface of the target (Figure 1(f)). the form of high-explosive anti-tank (HEAT)
For plates of a moderate thickness, Wei et al.32 rounds.48 The modern discovery of the shaped
showed that shear plugging dominates. For plugging charge munition is generally attributed to Charles
(Figure 1(d)), an important mechanism is adiabatic Munroe in 1888.49 Munroe observed that an imprint
shear fracture,8 where intense heat is generated in a block of explosive was reflected on an adjacent
during the impact event. Because of the short time- piece of steel after detonation. The imprinted letters
scales involved, the heat cannot be conducted away created an accelerating jet of material, increasing the
and the material temperature rises, thus diminishing penetrating capability of the explosive.
its strength.21 Masri33 showed that this effect is more A shaped charge uses the same principle and con-
pronounced when the target is thin enough to deform sists of a cylinder of explosive into which a cone-
in a ductile hole enlargement mode, but not so thin as shaped cavity has been cut, the apex of the cavity
to fail by dishing. Discing is another failure mechan- facing away from the target. The cavity in the explo-
ism in which a region of very high radial strain causes sive can be lined with metal or unlined. The shaped
a circumferential crack leading to the complete separ- charge jet consists of the liner material and travels at
ation of a ‘disc’ of material from the rear face of the speeds of many kilometres per second, having very
target.34 Rosenberg and Dekel4 note that discing is a good penetration characteristics. For example,
tensile failure phenomenon and highlight the results Jia et al.45 studied a standard shaped charge of diam-
by Wingrove,35 showing that projectiles with an eter 56 mm, which could penetrate up to 160 mm
ogive-shaped nose tend to cause this effect in high- of RHA.
strength materials. The ogive is the familiar ‘bullet’ A variant of the shaped charge weapon is the
shape. The main material component of most small explosively formed penetrator/projectile (EFP). This
arms projectiles is the lead core. Sometimes steel is utilises the Misznay–Schardin effect, which is a char-
used instead of lead as a core material in order to acteristic of flat sheets or discs of explosive material.
reduce costs. Military small arms ammunition must When detonated, the explosive blast expands in the
have a jacket covering the tip of the round – this direction perpendicular to the plane of the explosive.50
was specified in the Hague Convention of 1899.36 An EFP consists of a specially shaped charge of high
This jacket is generally made from copper alloy or explosive placed next to a shallow plate of metal,
steel. Other materials can be found inside specialised forming it into a quickly moving projectile upon det-
rounds. For example, 12.7mm armour-piercing onation.51 EFPs are projected towards their targets at
ammunition has a core made from tungsten carbide.37 speeds of roughly 2 km/s,4 and are formed from the
Clearly, there are a wide variety of materials pre- same material as the munition metal plate. Tantalum
sent in even the simpler forms of anti-armour ammu- is one of the most effective choices, because of its high
nition. Each of these materials must be modelled density and ductility at high strain rates.52 In their
accurately in any numerical or theoretical analysis physical characteristics, EFPs differ from shaped
of armour impact events. charges in that the strains and strain rates involved
in forming them are smaller (strain rates not exceed-
ing 104 s1, and true strain 44, compared to strain
Penetration – High-speed hydrodynamic rates between 104 and 105 s1 for shaped charges, and
The theory of hydrodynamic penetration applies in true strains of 10).53 In their effects, EFPs tend to be
the case of impact events involving velocities greater more lethal since they have a larger diameter than
Lenihan et al. 77

shaped charges, causing a wider impact crater and

more fragmentation and spalling.4
Another projectile type that travels at speeds on the
hydrodynamic scale is the ‘long rod penetrator’. This
is a type of kinetic energy warhead (i.e. it is inert) with
a small cross section and high muzzle velocity, which
ensures impact at velocities above the hydrodynamic
transition.4,54 The high density (they are often made
from materials such as tungsten or depleted uranium)4
and small cross-sectional area of such projectiles
maximise penetrative energy on the target. Long-rod
penetrators are the main ammunition of most modern
main battle tanks (MBTs), and are referred to by
acronyms based on their properties, e.g. armour
piercing fin-stabilised discarding sabot, tracer
(APFSDS-T).48,55
Although hydrodynamic penetration is quite
common and well studied, as is low-speed ‘rigid’
penetration, the transition zone between the two is
not well understood. Lou et al.56 attribute this to
the complex deformations of the rigid projectile at
higher speeds, making it difficult to apply the analyt-
ical framework of lower speed impacts. Penetrator
erosion, which is the basis for hydrodynamic theory,
does not begin to occur until the impact velocity of
the projectile exceeds the plastic wave speed of the Figure 3. General diagram of the spalling process showing
target.57 the impacting projectiles A (the ‘flyers’), the target material B
and the resulting spall fragment or fragments C.

No penetration – Spalling
In some instances, it is possible to defeat an AFV even
when full penetration of the armour does not occur. next section. Chen et al.59 observed spall at impact
Spalling is a failure mechanism associated with velocities of 129–500 m/s. Vogler and Clayton61
impact, whereby material (from the armour system) observed that spall severity increases with increasing
is ejected at high velocity. This can cause significant impact velocity, from 250 m/s to 450 m/s.
damage or loss of life. The phenomenon of material spallation is best
Spalling is caused by the interaction of shock waves, understood from a shock physics perspective, and a
which creates a region of high tension within a com- comprehensive introduction to the topic is given by
ponent.58 These wave interactions arise when a com- Carlucci and Jacobsen.21 Even without full penetra-
pression pulse, caused by a high-rate impact on one tion of armour from an anti-tank weapon, spall frag-
surface of a body, propagate through the body and ments can be lethal to vehicle crews. Some
then reflect as tensile waves off the far, stress-free sur- armaments, such as high-explosive squash head
face of the body. The tensile stress field develops at a (HESH) ammunition,62 are designed specifically to
very high rate, with typical strain rates in the region of maximise pressure waves and spall fragmentation,
104106 s1.58 This tension field causes the nucleation, rather than for penetration of the armour. Spalling
growth and coalescence of microscopic voids in the is also of interest in the design and protection of struc-
material,59 which leads to fracture and acceleration tures using cementitious materials, e.g. concrete struc-
of fragments from the free end of the body (i.e. the tures subjected to air blast detonation,63 although the
side opposite to the impact surface). Spalling, also focus of this review is on ductile metal scenarios.
referred to as ‘scabbing’, can be thought of as failure
due to cavitation resulting from excessive tensile stres-
Experimental techniques
ses within the material.60 Two different kinds of spal-
ling are shown in Figure 3, one in which a large and Experimental techniques are discussed for both
intact segment of armour is detached, the other in low- and high-speed tests; the methods examined
which many small particles are generated. Whether include both material tests and performance testing
spalling occurs in a component depends on the tensile of armour systems. Finally, particular attention is
strength of the material and the velocity of the inter- given to experimental methods that induce spalling,
acting shock waves, which depends on the velocity given its importance in the efficacy of vehicle
of the projectile. This is discussed further in the armour systems.
78 Proc IMechE Part L: J Materials: Design and Applications 233(1)

Low-speed impact and material testing

Drop testing is perhaps the simplest form of impact
mechanics experiment. The projectile is held directly
above the target and released, with gravity providing
the energy of the impact. A simple and typical meth-
odology for a drop test is described by Aryaei et al.64
Drop testing is often used for low-velocity impacts,65
but limitations on the maximum velocities obtainable
give it limited use in the field of terminal ballistics.
A drop-testing experimental methodology is used by
Antoinat et al.,66 for example, to investigate the per- Figure 4. Schematic diagram of Split Hopkinson pressure bar
foration of thin aluminium plates, but this is only apparatus.
applicable to thin targets. The maximum striking vel-
ocity in this case is under 10 m/s.66 Drop testing can
achieve strain rates in the range of 10–1000 s1, which
is too low to be relevant for terminal ballistics, apart
from rare scenarios as outlined above (strain rates for
impact events range from 104 to 108 s1).67
In order to test the ballistic capabilities of mater-
ials, the velocities involved in testing must be similar
to the flight velocities of munitions. In 1914,
Hopkinson68 presented a methodology for analysing
ballistic impacts. In his setup, the velocity of an
impacting bullet causes a pressure wave in a steel
bar when it strikes the end of the bar. The bar is
split into two sections, and by carefully varying the
ratio of their lengths, it is possible to infer the dur-
ation and maximum pressure of the impact event. Figure 5. A wide range of strain rates, measured in units
This methodology was modified by Kolsky69 in 1949 of s1, and the various experimental techniques that are valid.
to allow for the measurement of dynamic stress–strain Reprinted from Field et al.67 with permission from Elsevier.
response in specimens contained within a split section
of the pressure bar. A schematic and explanation of review of experimental impact techniques by Field
this method are shown in Figure 4, and is explained et al.,67 and covering 16 orders of magnitude of
further by Bobbili et al.70 The striker hits the incident strain rate. However, the focus of the review is
bar, sending a stress wave towards the specimen. Once firmly towards the right-hand side of the spectrum.
the stress wave reaches the end of the incident bar, a At the highest end of Figure 5 is plate impact. The
portion of this wave is transmitted through the speci- methodology involved is discussed briefly by Field
men to the transmitter bar, while the remainder et al.67 – a disc of material is fired at a target at
reflects off the specimen and travels back towards high speed, producing shock waves in the projectile
the striker. By instrumenting both bars with strain and target upon impact. A typical velocity for such an
gauges, the magnitude of the stress waves can be cal- experiment is 184 m/s.73 The loading involved in this
culated and thus the stress in the specimen inferred. kind of impact approximates 1D strain in the initial
The stain in the specimen is assumed to be uniform stages.67 Impacts of this speed can lead to spallation
and is inferred from the strain in the bars. of the materials, as discussed in greater detail above.
The Taylor impact test simply involves firing a At the lower half of Figure 5, conventional cross-head
cylindrical specimen of test material at a large, rigid, devices are suitable for experimental purposes. These
stationary anvil and measuring the subsequent plastic also have an important role to play in impact mech-
deformation of the cylinder in order to quantify the anics; they can be used alone74 or in conjunction with
material’s dynamic yield stress.71 The test dates from higher strain-rate tests19 as a means of obtaining lim-
1946, and although more accurate methods have since ited but nevertheless useful empirical data for deter-
been developed, the Taylor test remains useful as a mining constants in models such as the Johnson–
benchmark for material models used in finite element Cook (JC) strength model.
analysis (FEA).72 The main advantage of the Taylor
test is its ability to characterise materials at strain
rates of 104106 s1 using one simple, uniform and
High-speed and ballistic testing
easily repeatable test.71 Impacts at higher speeds entail using more elaborate
The test methods mentioned above are shown test setups than those described in the previous sec-
together in Figure 5, adapted from a wide-ranging tion. In many cases, full-scale ammunition and
Lenihan et al. 79

weapons are used. A further difficulty arises in the and two-stage gas guns are shown in Figure 6.
proper instrumentation of high-speed testing – the In the two-stage gas gun, the piston (2) is initiated
test is pointless if the desired parameter cannot be by combustion of a powder charge in the chamber
measured properly. This section discusses the test (7). This rapidly compresses a light gas (e.g. Helium)
methods and methods for data and result collection. in the pump tube, bursting the diaphragm (4) at a set
NATO STANAG 4659,6 mentioned above, defines pressure and accelerating the projectile (5).
levels of armour protection. The test procedures for A two-stage gas gun, shown schematically in
determining performance are outlined in similar detail Figure 6, can launch a projectile at speeds of up to
in STANAG 4164.75 These documents describe how 10 km/s, as described by Rosenberg and Dekel.4
to get useful data from test firings of anti-armour A single-stage gas gun is simpler (see Figure 6), com-
ammunition. NATO define penetration of armour to prising a pressure vessel or powder chamber, a valve
have occurred (for calibres up to 40 mm) when a hole, and a barrel. Field et al.67 report an upper velocity
through which light is visible, is made in a ‘witness limit of 1.2 km/s for a single-stage light-gas gun
plate’ (i.e. a plate of thin sheet metal) located behind (LGG). Various recent studies with single-stage guns
the target. The US Dept. of the Army76 define com- have operated with velocities well below this
plete penetration in the same way for calibres up to limit.39,57,79,80 Two-stage guns, although capable of
20 mm. Using a witness plate in this way allows higher speeds than single ones, are limited with
experimental observation of the extent of the respect to projectile diameter (due to the necessarily
behind-armour effects.77 A second plate can also be smaller diameters in the launch tube) for an equiva-
placed in front of the target (with a pre-made hole for lent size and cost of machine.67 These can be
the projectile to pass through) in order to observe improved even further: for example, Zhao et al.81
fragmentation on both sides.78 report on the case where a two-stage LGG accelerated
The simplest form of physical test, of course, is to the projectile up to 7 km/s, and a secondary explosive
simply fire an anti-armour weapon or charge at a apparatus attached to the muzzle accelerated it to
sample of target material on a firing range (e.g. 10 km/s. Other recent examples of two-stage LGG
Mishra et al.55 and Wu and Liu47). This has the studies may be found in Yunfei et al.,31 Arnold and
advantage of being the most faithful representation Schäfer77 and Nishida et al.78
of ‘real-life’ loading conditions for the armour. An important requirement for high-speed impact
On the other hand, there are many variables that experiments is the ability to visualise and measure
cannot be controlled well or at all at such a scale – the results. Prentice et al.29 presented a non-contact
accuracy of firing and meteorological conditions, for methodology for measuring displacements of target
example. For this reason, and because of the cheaper materials at high strain rates, using high-speed stereo-
cost and simplicity of setup involved, laboratory gas scopic digital speckle photography. Such optical tech-
guns are a common means of testing armour mater- niques are useful compared to strain gauges, for
ials.4 These allow for better control of accuracy and example, because they provide field rather than
projectile velocity. Schematics of both single-stage point measurement and do not affect the

Figure 6. Simple schematic of a gas-gun test facility, including instrumentation for gathering data. Two different kinds of gas-guns are
shown.
80 Proc IMechE Part L: J Materials: Design and Applications 233(1)

Figure 7. Generic plot of free-surface velocity as a function

of time following an impact, adapted from Ikkurthi and
Chaturvedi.82 Note where the two curves clearly separate –
this is the time at which spall occurs.

measurement.67 For further information on the

instrumentation of impact experimental techniques,
Field et al.67 have conducted an extensive review on
the topic.

Tests for spalling and fragmentation Figure 8. Typical experimental setup and plotted results from
Since spalling occurs near the free surface, one way to a spallation experiment.58 Not shown, but usually included in
characterise the behaviour of the material is to plot the setup, is a means of measuring the flyer velocity. This usually
the velocity of that surface as a function of time. This consists of a pair of light beams which the projectile breaks, or
a set of high-speed cameras.86 Figure reprinted from Antoun
is demonstrated in Figure 7. The velocity of the
et al.58 with permission from Springer.
impacting projectile is an important parameter, since
it directly affects the frequency of the shock wave and
thus the free-surface velocity response. Two different favourably to newer techniques, such as that
responses are shown in Figure 7, one representing a described by Strand and Whitworth in 2007.87
target material without spalling, and the other one Along with the typical experimental setup, Figure 8
with spalling. The surface velocity of the un-spalled shows a set of results of such an experiment.58 These
material reduces to near-zero as the pressure wave in results measured the axial stress as a function of time,
the material rebounds. In the spalled material, the for different parallel planes within the material exam-
mean velocity remains higher, since the detached ined (compressive stress on the positive axis). Since
piece of material is travelling away from the original spalling is caused by the rapid growth of a tensile
surface. Knowledge of the free-surface velocity allows stress field, it is clear that this is also an important
for the inference of stress–time history in a material.67 measure.
Practical tests to induce spalling in ductile metals Rosenberg and Dekel4 discuss the difference
must, due to the nature of the failure mechanism, between full and incipient spall, i.e. the latter consist-
involve a reasonably complex setup involving a ing of multiple small fragments of material, as
high-speed flyer (projectile), a non-contact means of opposed to a complete separation of a large fragment,
measuring the velocity of the flyer before impact and the result of ‘complete coalescence of the voids, lead-
the fitting of certain instrumentation to the target ing to a clear opening inside the specimen’ (p. 4).
material in order to measure strain response at small Deshpande et al.62 also found that the material
time intervals. Figure 8 shows a typical experimental tends to break up into as many as five pieces in the
setup for examining and measuring the region close to the axis of the projectile, but with
phenomenon.58 reduced damage closer to the sides, and this has the
The most common and versatile method for effect of maintaining material integrity (Figure 9).
obtaining the free-surface velocity data is the velocity It was postulated that the phenomenon of multiple
interferometer system for any reflector (VISAR).68,83 spallation occurs due to the interaction of multiple
This technique uses an interferometer to measure the waves and their reflections. As the outer free surface
difference betweeen the optical phase of reflected light of the material detaches due to a stress wave exceed-
from a coherent (i.e. laser) light source which illumin- ing a critical material tensile strength, a new free sur-
ates the surface of interest. The technique, its back- face is created. Stress waves reflect off this new surface
ground, theoretical basis and limitations are described and interact with secondary stress waves, causing the
in detail by Dolan.84 Barker85 compares VISAR growth of voids deeper in the material if the resultant
Lenihan et al. 81

of these waves still exceeds the material strength. This reproduce experimentally observed uniaxial
process continues deeper and deeper into the material behaviour.
until the stress waves and their interactions fade in
magnitude below the critical material strength.62
Constitutive models
A number of different material models have been
Material models developed, which allow for the calculation of the
Commonly used constitutive models for various flow stress of a material being subjected to large
armour materials are presented in the first section strain rates and strains in the plastic region, of
and subsequently assessed in terms of their ability to which terminal ballistics is an example. Such models
are useful in the design and evaluation of armour
materials, since they can be used in conjunction with
numerical analysis software to accurately predict the
response of a material to a given threat. A summary
of some of the most widely used models is given in
Table 2.

Johnson–Cook (JC). One of the most popular constitu-

tive models used in impact situations dates from the
1980s and is known as the JC model.88 It takes into
account the effects of strain rate and temperature on
flow stress:
 
Figure 9. Schematic diagram, adapted from Deshpande fl ¼ ½A þ B"n ½1 þ C ln "_  1  T m ð1Þ
et al.,62 showing how a greater amount of fragmentation occurs
near the axis of the projectile, while a smaller amount of where " is the equivalent plastic strain, "_ is the non-
fragmentation away from this axis has the effect of maintaining dimensionalised strain rate, T* is the homologous
the integrity of the material. temperature (Note 1), A, B and n are strain-hardening

Table 2. Summary of strength models analysed in this review.

Model Year Description References

Johnson–Cook (JC) 1983 Flow stress is a function of strain, strain rate and temperature, three (88) (89)
empirical constants and two empirical exponents.
Revised Johnson–Cook 1998 Two additional empirical constants are added, taking account of sudden (90)
strengthening which takes place in some metals at high strain rates
(>104 s1)
Zerilli–Armstrong (ZA) 1987 Makes a distinction between materials with BCC lattice structures (91) (92)
(e.g. a-iron) and FCC structures (e.g. copper). Flow stress is a func-
tion of strain, strain rate, temperature and six material constants.
Tanimura–Mimura 2009 Materials organised into characteristic groups. A number of material (92)
constants based on this group and the material’s quasi-static stress–
strain curve are required.
Khan–Huang–Liang 2003 Developed using experimental data of Titanium alloys. Increased strain (93)
rates lead to decreased work hardening.
Physical base (PB) 1957/ Breaks the total stress into two contributions: an athermal component (94–98)
1975 which is independent of temperature and strain rate, and a thermally
activated second component which takes these factors into account.
Nemat–Nasser–Li 1998 Similar to the PB model, i.e. based on addition of athermal and thermally (99)
activated components of stress. Requires at least six material
constants.
Rusinek–Klepaczo 2001 Similar to the PB model, i.e. based on addition of athermal and thermally (100) (101) (94)
activated components of stress. Requires up to ten material constants.
Voyiadjis–Abed 2005 Has different expressions for flow stress depending on FCC or BCC, as (102)
with the ZA model, but claims to be more accurate since it uses an
exact value rather than an approximation which was claimed to be
valid in the ZA model.
FCC: face-centred cubic; BCC: body-centred cubic; ZA: Zerilli–Armstrong.
82 Proc IMechE Part L: J Materials: Design and Applications 233(1)

constants, C and "_0 (the reference strain rate) are where

strain rate-hardening constants and m is a tempera-  
ture-softening constant. Despite its lack of sophistica- ln "_
 ¼ 0 1  ð5Þ
tion, this model remains popular as:89 ln "_

. It is easy to understand and implement in FEA and

codes.  
. It is intuitive and constants can be approximated ln "_
 ¼ 0 1  ð6Þ
without loss of function. ln "_
. Many existing materials have been characterised by
this equation, and their constants are widely where er is a reference strain, and C0, C1, C2, 0, 0, "_
available. and "_ are material constants. This formulation for
the ZA model is used in the following section, when
Such is the popularity of the JC model that many different strength models are compared.
revisions and additions have been presented over the Xu and Huang109 compare the ZA model to six
years; e.g. Zhang et al.,103 Tan et al.,104 Song et al.,105 other constitutive models for experiments on pure
Wang et al.106 and Hou and Wang107). For example, copper, finding that it had the highest errors with
Rule and Jones presented a revised form of the respect to the experimental data. It did, however,
model90 to incorporate two additional empirical con- require the determination of the least number of
stants enhancing the strain rate sensitivity, for higher material parameters (four, compared to nine for a dif-
strain rates, i.e. above 104 s1. The resulting revised ferent model), which is an important advantage.
Johnson–Cook (RJC) model is:
   Tanimura–Mimura (TM 2009). The consititutive model
n  1 1
 fl ¼ ½C1 þ C2 "  1 þ C3 ln "_ þ C4  developed by Tanimura and Mimura also differenti-
C5  ln "_ C5 ates between different materials based on their crystal
 
 1  T m structure, like the ZA model. The latest iteration of
ð2Þ this model, presented in 2009, is known as the TM
2009 model,92 and has the following form for the
The JC model and its variants are the most com- flow stress:
monly used methods of analysing experimental results    
for new or modified armour materials. It is relatively 
m1  s "_p
fl ¼ s þ   "p þ  1  ln
simple to obtain accurate material parameters. For CR "_sp
example, Trajkovski et al.74 present a methodology  

"_p m2
for obtaining JC constants for a new material þ B "p
"_uni
(PROTAC 500 steel), empirically, through quasi-
static temperature-controlled testing. ð7Þ

Zerilli–Armstrong (ZA). The ZA model91,92 distinguishes where "_sp is the lower limit of the strain rate, "_p is the
between metals with body-centred cubic (BCC) and plastic strain rate and "_uni is a unit strain rate. The
face-centred cubic (FCC) crystal lattice structures. term rs is the stress at the lower strain rate "_sp , and ,
Pure iron is BCC, although various treatment pro- , rCR, m1, m2 and B are material parameters. The
cesses for steels can affect this. Aluminium, on the authors of this model found that it matched experi-
other hand, is a FCC metal. Table 1 lists some of mental data much closer than the JC and ZA models,
the materials used for armour purposes. The compact although these results related to vehicle-crushing data,
form of the model is described by Meyer,108 which rather than ballistic impact, and so involved signifi-
gives the flow stress as: cantly lower strain rates (1 s1 to 103 s1). Another
pffiffiffi
advantage of the TM 2009 model is that, although
fl ¼ C0 þ C1 þ C2 " expfC3 þ C4 ln "_gT þ C5 "n there is a large number of material parameters,
ð3Þ many of these are constant across an entire material
group (e.g. ferrous metal). Because of this, the only
where C0, C1, C2, C3, C4, C5, and n are material con- information required for a material is the quasi-static
stants. For FCC metals C1 ¼ C5 ¼ 0, whereas for BCC stress–strain curve. The model described in equation
metals C2 ¼ 0. A modified version of the flow stress is (7) does not contain a temperature term – the results
presented by Meyer108 for hexagonal close-packed given in Tanimura et al.92 do not mention or account
(HCP) metals, such as Ti-6Al-4V, and is given as for material temperatures.
 
fl ¼ C0 þ C1 exp T Khan–Huang–Liang (KHL). The modified KHL model93
  
 ð4Þ was developed based on experiments carried out on a
þ C2 "r 1  exp "="r expfTg
range of titanium alloys, notably Ti-6Al-4V. The form
Lenihan et al. 83

of the model is Short-range barriers such as vacancies, self-interstitials,

interstitials and substitutionals caused by alloying
    C  
ln "_ n1 p n0 "_ Tm  T m elements can be overcome by thermal activation,
fl ¼ A þ B 1  ð" Þ whereas long-range barriers such as grain boundaries
ln Dp0 "_ Tm  Tr
cannot and are therefore athermal.99 The model devel-
ð8Þ
oped (this example is based on OFHC (Note 2) copper,
a FCC metal) is:
where ep is the plastic strain, ",
_ "_ are the current
(    )1:5
strain rate and reference strain rate, similar to the kT "_
0:5
0:5
JC model (equation (1)), and Tr is a reference tem- fl ¼ 0 1   ln þ ln 1 þ aðTÞ"
G0 "_0
perature. The parameter Dp0 is an arbitrary strain rate,

chosen in Khan et al.93 as 106 s1. The constants A, B,  1 þ aðTÞ"0:5 þ a0 "n1
n1, n0, C and m depend on the material. This model ð12Þ
was found to correlate better than the JC model with
experimental data on dynamic (split Hopkinson pres- where
sure bar) impact of Ti-6Al-4V. The effects of thermal

softening and a reduction in work hardening at high aðTÞ ¼ a0 1  ðT=Tm Þ2 ð13Þ

strain rates are captured by the KHL model.
In the model above, a0 is a constant reflecting the
Physical base (PB). The PB model is explained by initial average dislocation spacing, k/G0 is the ratio of
Klepaczko et al.,94 and also described by Nemat- the Bolzmann constant to the Gibbs energy and has
Nasser and Isaacs,95 Nemat-Nasser and Guo.96,97 units of K1, "_0 is a reference strain rate, r0 is a char-
It developed from observations reported by Seeger.98 acteristic thermally activated stress, a0 is a character-
It breaks the total stress into two contributions, an istic athermal stress and n1 is a material-dependent
athermal component, which is independent of tem- exponent. It was found it to be in good agreement
perature and strain rate, and a thermally activated with split Hopkinson bar experimental results over
second component which takes these factors into temperatures of 77–1100 K, strain rates of 103 to
account. The rationale behind this approach is 8000 s1 and strains of greater than 100%.99
explained well by Voyiadjis and Abed.102 The flow
stress takes the form Rusinek–Klepaczko (RK). Rusinek and Klepaczko94,99
presented a model which, in a similar manner to the
fl ¼ a þ   ð9Þ PB model above, is based on the addition of athermal
and thermally activated stresses:
where ra is the athermal part and r* is the thermally   
EðTÞ _

nð"_ , TÞ T
activated component, giving a total equivalent stress fl ¼ B "p , T "0 þ "p p þ 0 1  D1
r. A power–law approximation gives ra: E0 Tm
!  #
m

n "_max
 log
a ¼ a1 "p 1 ð10Þ "_p

In equation (10), "p is the plastic strain, a1 is a ð14Þ

material constant and n1 is a strain hardening expo- The Macaulay operator sets the expression inside
nent. The second part of equation (9) is given as the brackets to zero if it is less than zero, thus elim-
2 inating the instantaneous rate sensitivity of the model
!1q 3  at high temperatures and low strain rates.94 The elas-
kT "_p 5 1
  ¼ ^ 41   ln ð11Þ tic modulus, plastic modulus and strain hardening
G0 "_r p exponent functions are:
   
where ^ is a thermally activated threshold strength, T  Tm
EðTÞ ¼ E0 1 exp  1  ð15Þ
k/G0 is a ratio characterising the sensitivity of the Tm T
material to temperature effects (in terms of the refer-
  !!v
ence Gibbs energy at a temperature of absolute zero
T "_max
G0 and Boltzmann’s constant k), "_r is a characteristic B "_p , T ¼ B0 log ð16Þ
Tm "_p
strain rate and p and q are material constants.
(   )
Nemat-Nasser and Li (NNL). The model developed by
T _p
"
n "_p , T ¼ n0 1  D2 log , n50 ð17Þ
NNL99 is also physically based and also breaks the Tm "_min
flow stress into thermal and athermal components, as
in the PB model. The physical basis for this distinction E0 is the Young’s modulus at T ¼ 0 K, h* is the char-
is based on the barriers to dislocation motion. acteristic homologous temperature of the material
84 Proc IMechE Part L: J Materials: Design and Applications 233(1)

(a material constant, h* & 0.5 for ferritic steels), e0 is constants. The material chosen for the purposes of
the strain at yield in quasi-static loading, 0 is the comparison was Ti-6Al-4V, since it is one of the
effective stress at T ¼ 0 K, D1 is a material constant, better characterised materials. The three models com-
"_min and "_max are the limits for strain rates in the pared were the JC, ZA and KHL models. The material
model, m* is a strain rate temperature dependency parameters and their sources are presented in Table 3.
constant, B0 is a material constant modulus of plasti- The data in Table 3 were used to predict the uni-
city, D2 is a material constant and all other symbols axial tensile stress–strain response for each material at
have the same meanings as in previous models. a range of temperatures and strain rates and com-
A modified version of the RK model was presented pared with experimental results on the same material,
in Rusinek et al.,101 with a specific application to FCC as presented by Chen et al.110 The results of this ana-
metals, namely copper. lysis are shown in Figure 10.
The results presented span a wide range of strain
Voyiadjis–Abed (VA). The VA model, in a similar way to rates (1  104 to 1  104 s1) and temperatures
the ZA model, also distinguishes between BCC and (room temperature and 700  C). At room temperature
FCC metals, and is physically based, with flow stress and 1  104 s1, the JC and KHL models both per-
split into thermally activated and athermal compo- form well, whereas the ZA model significantly overpre-
nents. The equation for the flow stress of BCC dicts stress for low strain (<0.15) and underpredicts it
metals is given as for higher strains. At a higher strain rate of 1  102
s1, the JC and KHL models again show good correl-

1=q 1=p ation to the test data at both temperatures, although
fl ¼ Y^ 1  1 T  2 T ln "_p þ B"np þ Ya
they slightly overpredict stress at room temperature and
ð18Þ show too much hardening at higher strains at the higher
temperature. The ZA model again performs relatively
where Ya is the initial athermal yield stress, Y^ is the poorly at room temperature, overpredicting stress for
threshold yield stress of the Peierls barrier to disloca- yield and low strains and underpredicting for higher
tion motion, p and q are material constants strains; this model shows negligible hardening, but cap-
(0 4 p 4 1.0, 1.0 4 q 4 2.0), 1 and 2 are material tures the mean stress reasonably well overall. For the
parameters, B is a hardening constant and n is a much larger strain rate of 2500 s1, all three models
material exponent. For FCC metals, the flow stress is overpredict stress at both temperatures, most signifi-
cantly at yield and lower strains. Both the JC and the

1=q 1=p KHL models perform well at the higher strains for
fl ¼ B"np 1  1 T  2 T ln "_p þ Ya ð19Þ
700  C. At the highest strain rate of 1  104 s1, all
three models overpredict stress, particularly (i.e. by
and all the symbols have the same definitions as in 30–50%) at yield and low strain for room temperature,
equation (18). but perform significantly better at higher strains. At
700  C, for this rate, the JC and KHL models signifi-
cantly overpredict hardening (and hence stress) at mod-
Comparison of strength models erate to high strains (>0.05); the KHL model gives
It is of interest to explore the relative performance of closer results to the test data, however.
these material models. Hence, some of the key models In some cases, the ZA model results are not shown
discussed above are compared here for uniaxial condi- at the higher temperature due to extremely high
tions based on available published values for material (apparently non-physical) stress levels. Meyer

Table 3. Published material properties for the Johnson–Cook (JC), Zerilli–Armstrong (ZA) and Khan–Huang–Liang (KHL) models
for Ti-6Al-4V.

Model Material properties Source

A B n C m
(MPa) (MPa) – – –
JC 1104 1036 0.6349 0.0139 0.7794 (93)

C0 C1 C2 "r "_ "_ 0 0

(MPa) (MPa) (MPa) (s1) (s1) (K1) (K1)
ZA 1217 139 3955 0.1877 1.6  104 8.25  105 1.59  102 7.55  103 (108)

A B n1 n0 C m D0
(MPa) (MPa) (s1)
KHL 1069 874.8 0.5456 0.4987 0.02204 1.3916 106 (93)
Lenihan et al. 85

Figure 10. Comparison of the Johnson–Cook (JC), Khan–Huang–Liang (KHL) and Zerilli–Armstrong (ZA) constitutive models for
Ti-6Al-4V via uniaxial stress–strain curves. The models are compared with experimental results presented by Chen et al.112 There is
no particular significance to the temperatures chosen, apart from highlighting the effect of temperature softening on the material.
(a) T ¼ 300 K; "_ ¼ 0.0001/s. (b) "_ ¼ 0.01/s. (c) "_ ¼ 2500/s. (d) "_ ¼ 10 000/s

modified the ZA model to capture the behaviour of amenable to analytical solution. Thus, simulation of
HCP materials such as Ti-6Al-4V. However, Meyer these impact events generally requires advanced
considered a wide range of strain rates, up to numerical techniques. Rosenberg and Dekel4 discuss
4.64  104 s1, and the original calibration also per- the advantages of finite element (FE) and finite differ-
formed poorly at strain rates below 2.20  104 s1. ence methods in ballistic ‘hydrocode’ simulations
Xu and Huang,109 who carried out tests on OFHC where they also explain the relative merits of
Cu reached the same conclusion for the ZA model Lagrangian versus Eulerian analysis techniques.
with FCC materials. In summary, based on the data Lagrangian analyses couple the material points on
in Figure 10, the ZA is considered to be the least the deforming solid to the nodes of the mesh, whereas
accurate of the models analysed, showing the greatest the Eulerian approach fixes the mesh as an unchan-
departure from the experimental data. Of the other ging reference frame and allows material points and
two models, KHL is marginally closer to a full agree- surfaces to deform within that mesh. A more detailed
ment with the experimental data than the JC model, explanation of both is given by Donea et al.111 It is
although the former requires an additional two mater- concluded that an ideal compromise in a ballistics
ial constants compared to the latter. simulation is to model the projectile in a Lagrangian
mesh (assuming it to be hard and non-deforming) and
the target in a Eulerian grid (assuming it to be softer
Numerical considerations and deforming).4
Another way of achieving the advantages of a
Analysis codes used Lagrangian approach whilst allowing for large
Due to the complex nature of the dynamic loading, deformation is through the use of smooth particle
geometry and boundary conditions inherent in impact hydrodynamics (SPH).112 The application of this
mechanics with application to ballistics, very few computational method to impact problems began in
problems that have a practical application are the early 1990s and initial descriptions, algorithms
86 Proc IMechE Part L: J Materials: Design and Applications 233(1)

and examples are presented by Petschek and comparisons with experimental data. The other sig-
Libersky,113 Libersky et al.114 and Stellingwerf and nificant change is that the growth in computing
Wingate.115 It has now been implemented in commer- power has facilitated the analysis of more meaningful
cial FE codes such as Abaqus and Ansys. problems having greater complexity.
Johnson et al.116 conducted an early review of the
application of SPH to ballistics problems. A number of
algorithms and examples were presented, with the Future trends
long-term goal of developing a method in which a
standard Lagrangian FE grid could be drawn for any
Advanced armours
impact problem, with the code automatically changing Layered armour is a technique that is gaining in popu-
highly deformed elements of the material to SPH larity over recent years and is becoming standard with
nodes.116 In subsequent work, Johnson89 presented the heaviest and most advanced classes of AFVs.9 One
examples of simulations where this had been achieved, such example in use by British and US forces is
thus demonstrating the value of linking FE and SPH Chobham armour, which is composed of tiles or
methods. A further advantage of the mesh-free nature cells of ceramic materials in a metallic matrix.
of SPH, as explained by Batra and Zhang,117 relates to Although the details of Chobham armour remain
the simulation of crack propagation which is more classified, there is a wealth of published literature on
realistic, as it is dictated by physics and not the location ceramic armour in general. For example, Hohler
and density of the mesh at any particular point. et al.122 evaluated a number of ceramic materials in
The popularity of four major numerical codes over a layered system with metallic plates in order to deter-
the past 25 years is shown in Figure 11. The results are mine the ideal ceramic and the ideal ratio of ceramic
based on a search of keywords in a leading journal of to metallic material. Bürger et al.123 carried out an
impact mechanics, and the codes considered are FEA simulation of a ceramic composite armour,
EPIC,;89,118 Ansys AUTODYN;119 Abaqus/ including detail on the brittle material models used
Explicit and LS-DYNA.121 This is coupled with a
120
in the code. Hogan et al.124 present a detailed experi-
significant growth over the same time period in the use mental analysis of ceramic impact events, measuring
of FE methods to address problems in impact mech- the dispersion of spall material.
anics. The proportion of articles on impact mechanics Sandwiching together materials with different
which include FEA has increased from less than 10% properties is a cost-effective way to optimise
to nearly 50% of all articles published. In the past 15 armour, as it allows for different layer properties to
years, LS-DYNA and Abaqus/Explicit have sup- have varying effects on a projectile as it penetrates.23
planted EPIC and, to a lesser extent, Ansys Worsham et al.125 present a novel system of armour
Autodyn as the most popular codes. An example of (shown in Figure 13), which uses a number of differ-
results from EPIC is shown in Figure 12. The growth ent layers, each of which is specifically designed to
in the level of numerical analysis is reflective of the have a different negative effect on a penetrator.
maturity of the analysis packages and that there is In Corvid’s system (Figure 13), the angled plates
greater confidence in their results as a result of greater of hardened steel break up incoming projectiles and

Figure 11. Chart showing the popularity of four leading FEA codes used for impact simulations over time. The primary axis pertains
to the bar chart, and is simply the number of articles for each type of code, stacked cumulatively.
Lenihan et al. 87

Figure 12. Comparison of experimental and numerical simulation of a tungsten projectile impacting a steel target with a velocity of
1020 m/s. The graph on the left shows the close correlation between simulation (EPIC code) and experiments. The image on the right,
from the numerical simulation, shows the complete penetration of the target and the large cloud of spall material. Reprinted from
Johnson89 with permission from Elsevier.

the AFV’s main armour, with a gap between the

main armour and the ERA. When a shaped
charge interacts with the explosive the latter deton-
ates, causing the steel plates to accelerate forwards
and backwards. The plates interact with the shaped
charge jet, moving obliquely and continuously pre-
senting new material which the jet erodes. This
serves to diminish the power of the jet and break
it up, in turn reducing the penetrating potential it
has on the main armour of the vehicle. 126 ERA can
reduce the effect of a shaped charge (i.e. the depth
of penetration into the main armour) by up to
75%, although the effect is significantly reduced
or even eliminated if the impact occurs at right
angles.44 There are multiple disadvantages of
Figure 13. A proposed spaced armour system, in which dif- ERA, such as:
ferent materials and geometries are spaced at different layers in
order to have different effects on a penetrator as it passes
through. Reprinted from Worsham et al.125 with permission
. Each module of ERA can absorb one strike, after
from Elsevier. which it is destroyed. Some complex anti-tank war-
heads exploit this by using a ‘tandem’ warhead
with axially stacked shaped charges.
impart a rotation to them, reducing the component of . The fragments of ERA have significant kinetic
velocity normal to the face of the armour. The alu- energy, posing a serious hazard for ground troops
minium plates serve to hold these hardened steel in support of AFVs or unprotected non-comba-
plates in the correct orientation, especially with mul- tants in the area.
tiple impacts. The air gap (in Figure 13, this is filled . There is a challenge in designing modules of ERA,
with foam) gives space for projectiles and debris to which can operate as designed without causing
disperse somewhat. The hardened steel plate provides adjacent modules to sympathetically detonate.
the main armour protection, and the composite The challenges in creating an effective pattern are
materials behind it serve as a spall liner, catching discussed by De Rosset.127
debris and preventing it from entering the crew
compartment. More recent developments in this field include
Explosive reactive armour (ERA) was developed attempts to design ERA, which will defeat the threat
specifically to counteract the threat posed by from kinetic energy projectiles (see Penetration –
shaped-charge weapons. It consists of a ‘sandwich’ High-speed hydrodynamic section). As with the
of high-explosive material held between two steel defeat of a shaped charge, a lot of the variability in
plates. The whole unit is attached to the outside of the performance of the armour is dependent on the
88 Proc IMechE Part L: J Materials: Design and Applications 233(1)

angle of impact, with a more oblique angle producing the power of supercomputers, to improve the imaging
a better effect.41 resolution of VISAR (mentioned in Tests for spalling
and fragmentation section) to observe the motions of
individual grains of a material.67 This is becoming
Vulnerability models
more important as new forming and manufacturing
Germany’s Universelles Verwundbarkeitsmodell techniques for improved armours involve more care-
(Universal Vulnerability Model, shortened to ful control of material microstructures. For example,
UniVeMo) is used by the German Army as the Wielewski et al.131 hypothesise a method for improv-
standard for predicting the lethality of weapons ing the spall strength of Ti-6Al-4V based on careful
and the vulnerability of vehicles, structures and control of certain microstructural features of the
troops.128,129 Rather than simulating only one spe- material, such as the number of plastically hard or
cific impact event in isolation, this model tries to soft grain boundaries. Hazell et al.132 subjected
quantify the global effects of anti-armour weapons Elektron 675, a magnesium alloy, to a number of
on vehicles and troops and ultimately on the force’s spall experiments. It was found that long striations
combat effectiveness. An earlier incarnation of the of small grains, which formed along the extrusion dir-
model was presented by Arnold and Paul,130 which ection of the material, provided an opportunity for
focused on the behind-armour debris cloud (i.e. cracks to grow at right angles to the spall plane.
spalling) as a continuation of Arnold and Schäfer’s Barrett et al.,133 Golden et al.134 and Li et al.135
work on this topic77 (see High-speed and ballistic developed a multiscale viscoplastic model of P91 steel,
testing Section). incorporating it into the macro-scale FEA analyses
The dearth of published material on equivalent vul- using a representative volume element (RVE).
nerability models for other military forces is not A schematic representation of the complex grain struc-
necessarily indicative of their absence, since there ture of this material is shown in Figure 14.
are legitimate security considerations for keeping The hierarchical grain structure with strengthening
such information under close control. Nevertheless, mechanisms shown in Figure 14, although taken from
through the use of improved numerical models and a study on cyclic viscoplasticity in chrome steels with
computational resources, such vulnerability models small strain ranges,133 is a useful example of the kind of
are likely to improve significantly in capability and microstructural considerations which would feed into a
prevalence in the near future. multiscale model for ballistic impact.
Jacques et al.136 presented a finite-element model
which incorporates the role of micro-inertial effects
Multi-scale modelling due to void growth for ductile fracture. Their
Techniques for evaluating and simulating the role of models take account of void nucleation and growth
microstructure in the response of materials to impact by modelling such voids within a larger RVE – the
loading are becoming increasingly prevalent. For formulation of this is also discussed in greater detail
example, Field et al. note that progress has been by Sartori et al.137 and Molinari and Mercier.138 The
made in using advanced techniques, coupled with RVE contains a number of spherical voids of varying

Figure 14. Schematic representation of the hierarchical grain structure of 9-12Cr steel (P91), showing various strengthening
mechanisms. Reprinted from Barrett et al.133 with permission from Elsevier.
Lenihan et al. 89

radii embedded in spherical shells of matrix material.

In a comprehensive review of their work in this
field,139 the authors explain the relationship between
the macro (component) level, the mesoscopic level, in
which an RVE is constructed for each integration
point on the macro level, and the micro level, which
consists of a single void. The behaviour of the
unit cells and RVEs was governed by an analytical
relationship which was verified by FEA calculations
at the void level. The work carried out by Jacques,
Mercier, Molinari and Sartori related to plate impact
tests on tantalum; the results140 showed a very good
correlation between experimental data and the micro-
and macro-level model. If such methods could be
extended to a general class of metallic materials, this
would facilitate the development of a material design
tool for terminal ballistics, in which microstructural Figure 15. An illustration of the performance of a physically-
variables, instead of empirically derived coefficients, based model. The RK constitutive model, presented by
Klepaczko et al.94 and described in Rusinek–Klepaczko (RK)
could be used to predict and therefore optimise mater-
section, is compared to experimental stress-strain data at a
ial performance. variety of strain rates. The material used is DH-36 steel.
Reprinted from Klepaczko et al.94 with permission from
Material design tool Elsevier.

A recent study141 analysed a wealth of experimental field before adequate material design tools are
data for a range of materials to develop a powerful available.
means of predicting ballistic limit velocities and pene-
tration thicknesses. It was also noted, however, that it
Conclusion
should soon be possible to predict these quantities
from information regarding material chemistry and The integrity of vehicular armour is clearly of the
heat treatment alone. utmost importance for designers and military plan-
Rusinek at al.101 distinguish between phenomeno- ners. The threat profile facing armoured vehicles is
logical (i.e. empirical) material models and physically quite diverse and so there are multiple requirements
based models. While the former, if implemented cor- for armour materials.
rectly, can have fewer material constants, they suffer This review first listed the main categories of material
from lack of applicability and flexibility. The latter used in vehicular armour, and compared their attributes,
are more universal and have greater flexibility, but strengths and weaknesses. It was found that among
require a large number of material constants. They metals, specialised steels continue to be the most preva-
can provide more accurate results: Figure 15 demon- lent material, and metallic armour in general is much
strates the accuracy of the RK model, described in more pervasive and widespread than ceramic- or compo-
Rusinek–Klepaczko (RK) section, when compared site-based armour. This is mainly due to cost and manu-
with experimental data. Although the materials used facturability considerations. Following on from that was
in Figures 10 and 15 are different, the better agreement a summary of the main defeat mechanisms for armoured
with experimental data in the latter example is obvious. vehicles. It was shown that with low-speed penetration, a
A micromechanical model, i.e. one which was variety of failure mechanisms prevail based on projectile
based on a fundamental analysis of the physical material, geometry and impact speed. Projectiles them-
activity at the grain and crystal scale, would in selves vary in their size and composition, highlighting the
theory allow for the useful generality of the physic- importance of proper material information when doing
ally based model with the simplicity of the phenom- theoretical or numerical modelling. This kind of model-
enological model. Nemat-Nasser and Li referred to ling must be rigorously tested and ‘reality-checked’, on
these micromechanical models in 1998,99 pointing account of the variety of failure mechanisms.
out that such models always require some degree With high-speed penetration, a different mode of
of empirical evaluation of certain material param- analysis, one based on fluid hydrodynamics, becomes
eters in order to be useful. Dunne and Petrinic142 important. The next section of the review described
also mention the challenges inherent in modelling various experimental techniques in both the low- and
crystal plasticity computationally, although it high-speed testing regions, and charted their popular-
must be borne in mind that exponential improve- ity over time. It was shown that certain popular
ments in computing power constantly provide new impact test methods such as drop testing can be inad-
opportunities and impetus for such research. It is equate for the higher strain rates involved in terminal
clear that much work remains to be done in this ballistics.
90 Proc IMechE Part L: J Materials: Design and Applications 233(1)

The material models section explained eight differ- 3. Wolf SJ, Bebarta VS, Bonnett CJ, et al. Blast injuries.
ent material constitutive models and compared three Lancet 2009; 374: 405–415.
in terms of their flexibility and adaptability to differ- 4. Rosenberg Z and Dekel E. Terminal ballistics. New
ent strain, strain rate and temperature regimes. The York: Springer, 2012.
5. Maweja K and Stumpf W. The design of advanced per-
results illustrate a challenge that faces designers – the
formance high strength low-carbon martensitic armour
variability of many of the material models. They dem-
steels. Part 1. Mechanical property considerations.
onstrate narrow ranges of applicability, outside of Mater Sci Eng 2008; 485: 140–153.
which they fail to predict real-life results with suffi- 6. NATO Military Agency for Standardisation. STANAG
cient accuracy. Since quasi-static testing is the most 4569: protection levels for occupants of logistic and
practicable way of obtaining stress–strain data from a light armour vehicles. Brussels: NATO.
previously uncharacterised material, this creates a dif- 7. Ogorkiewicz RM. NATO airs new vehicle protection
ficulty and makes it difficult to calibrate data. standards. Jane Int Defence Rev 2013; 46.
However, quasi-static testing or dynamic testing at 8. Molinari A, Mercier S and Jacques N. Dynamic failure
lower strain rates is often used to provide material of ductile materials. Procedia IUTAM 2014; 10:
data for these models; either more extensive testing 201–220.
9. Hazell PJ. The development of armour materials. Milit
is required or more universal material models
Technol 2006; 30: 54–62.
must be developed. Without accurate material con-
10. Hazell PJ, Edwards MR, Longstaff H, et al. Penetration
stants, the usefulness of the numerical methods of a glass-faced transparent elastomeric resin by a
described in the numerical considerations section is lead-antimony-cored bullet. Int J Imp Eng 2009; 36:
severely limited. 147–153.
In conclusion, despite the extensive literature 11. DeGarmo E. Materials and processes in manufacturing.
reporting experimental data and the many material Hoboken, NJ: John Wiley, 2003.
models available, design of metallic military vehicular 12. Ogorkiewicz RM. Technology of tanks (volume II).
armour is a task complicated by the lack of univer- London: Jane’s Information Group, 1991.
sally applicable material models and the extensive 13. Pérez-Bergquist SJ, Gray GTR, Cerreta EK, et al. The
testing and calibration requited for current models. dynamic and quasi-static mechanical response of three
aluminum armor alloys: 5059, 5083 and 7039. Mater Sci
From the designer’s perspective, a design tool that
Eng A 2011; 528: 8733–8741.
describes the response of a variety of metallic
14. Sudhakar I, Madhu V, Reddy GM, et al. Enhancement
armour materials to various impacts is a panacea; of wear and ballistic resistance of armour grade
such a tool would require only the microscale material AA7075 aluminium alloy using friction stir processing.
structure (e.g. grain sizes and crystal structures) and Defence Technol 2015; 11: 10–17.
the chemical composition in order to predict material 15. The Aluminum Association, Inc. International alloy des-
behaviour from the quasi-static to the hypervelocity ignations and chemical composition limits for wrought
range. As computational capabilities continuously aluminum and wrought aluminum alloys. Arlington,
grow, coupled with more competent material VA: The Aluminum Association Inc, 2015.
models, this objective becomes a realistic target. 16. Bauri R, Yadav D, Shyam Kumar CN, et al. Tungsten
particle reinforced Al 5083 composite with high
strength and ductility. Mater Sci Eng A 2015; 620:
Declaration of conflicting interests 67–75.
The author(s) declared no potential conflicts of interest with 17. Heilig G, Durr N, Sauer M, et al. Mesoscale analysis of
respect to the research, authorship, and/or publication of sintered metals fragmentation under explosive and sub-
this article. sequent impact loading. In: The 12th hypervelocity
impact symposium 2013; Baltimore, MD:
Funding Hypervelocity Impact Society, 58: 653–662.
The author(s) received no financial support for the research, 18. Grace FI. Ballistic limit velocity for long rods from
authorship, and/or publication of this article. ordnance velocity through hypervelocity impact. Int J
Imp Eng 1999; 23: 295–306.
19. Nilsson M. Constitutive model for Armox 500T and
Notes
Armox 600T at low and medium strain rates. Report
1. T ¼ TTm , where Tm is the absolute melting temperature for Swedish Defence Research Agency. Tumba,
and T is temperature (K). Sweden: Swedish Defence Research Agency, 2003.
2. Oxygen-free high [thermal] conductivity. 20. Srivathsa B and Ramakrishnan N. Ballistic perform-
ance maps for thick metallic armour. J Mater Process
Technol 1999; 96: 81–91.
References 21. Carlucci DE and Jacobson SS. Ballistics – theory and
1. Greenhalgh E. Technology development in coalition: design of guns and ammunition. Boca Raton, FL: CRC
the case of the first world war tank. Int History Rev Press, 2008.
2000; 22: 806–836. 22. Hazell PJ. Advances in ceramic armour. Milit Technol
2. Ramasamy A, Harrisson S, Lasrado I, et al. A review of 2009; 33: 118–126.
casualties during the Iraqi insurgency 2006 – a British 23. Brown N. Lighten up: tracking innovations in passive
field hospital experience. Injury 2009; 40: 493–497. armour solutions. Jane Int Defence Rev 2013; 46: 62–65.
Lenihan et al. 91

24. Joint Education and Doctrine Division J-7, US Armed 43. Poole C. Penetration of a shaped charge. PhD Thesis,
Forces. DOD Dictionary of Military and Associated University of Oxford, UK, 2005.
Terms. [Online]. Available at: http://www.dtic.mil/doc- 44. Li Xd, Yang Ys and Lv St. A numerical study on the
trine/dod_dictionary/?zoom_query=small+arms disturbance of explosive reactive armors to jet penetra-
(2015, accessed 5 May 2015). tion. Defence Technol 2014; 10: 66–75.
25. Børvik T, Dey S and Clausen AH. Perforation resistance 45. Jia X, Huang ZX, Zu XD, et al. Theoretical analysis of
of five different high-strength steel plates subjected to the disturbance of shaped charge jet penetrating a
small-arms projectiles. Int J Impact Eng 2009; 36: 948–964. woven fabric rubber composite armor. Int J Impact
26. Kılıç N, Bedir S, Erdik A, et al. Ballistic behavior of Eng 2014; 65: 69–78.
high hardness perforated armor plates against 7.62 mm 46. Li W, Wang X and Li W. The effect of annular multi-
armor piercing projectile. Mater Design 2014; 63: point initiation on the formation and penetration of an
427–438. explosively formed penetrator. Int J Imp Eng 2010; 37:
27. Rosenberg Z and Dekel E. Revisiting the perforation of 414–424.
ductile plates by sharp-nosed rigid projectiles. Int J 47. Wu J and Liu JDY. Experimental and numerical study
Solids Struct 2010; 47: 3022–3033. on the flight and penetration properties of explosively-
28. Lee YW and Wierzbicki T. Fracture prediction of thin formed projectile. Int J Imp Eng 2007; 34: 1147–1162.
plates under localized impulsive loading. Part I: dishing. 48. DCSINT. OPFOR: worldwide equipment guide.
Int J Imp Eng 2005; 31: 1253–1276. TRADOC DCSINT Threat Support Directorate. Ft.
29. Prentice HJ, Proud WG, Walley SM, et al. Optical tech- Leavenworth, KS: US Army Training and Doctrine
niques for the investigation of the ballistic impact of Command, 1999.
thin plates. Int J Imp Eng 2011; 38: 849–863. 49. Walters W. A brief history of shaped charges. In: 24th
30. Iqbal MA, Gupta G, Diwakar A, et al. Effect of pro- International symposium on ballistics, New Orleans, 2008.
jectile nose shape on the ballistic resistance of ductile 50. Thurman JT. Practical bomb scene investigation. 2nd ed.
targets. Eur J Mech A/Solids 2010; 29: 683–694. Boca Raton, LA: CRC Press, 2011.
31. Yunfei D, Wei Z, Yonggang Y, et al. Experimental 51. Hallquist JO, Goudreau GL and Benson DJ. Sliding
investigation on the ballistic performance of double- interfaces with contact-impact in large-scale
layered plates subjected to impact by projectile of high Lagrangian computations. Comput Methods Appl
strength. Int J Imp Eng 2014; 70: 38–49. Mech Eng 1985; 51: 107–137.
32. Wei Z, Yunfei D, Sheng CZ, et al. Experimental inves- 52. Cardonne SM, Kumar P, Michaluk CA, et al.
tigation on the ballistic performance of monolithic and Tantalum and its alloys. Int J Refract Metals Hard
layered metal plates subjected to impact by blunt rigid Mater 1995; 13: 187–194.
projectiles. Int J Imp Eng 2012; 49: 115–129. 53. Murr LE, Ferreyra ET, Pappu S, et al. Novel deform-
33. Masri R. The effect of adiabatic thermal softening on ation processes and microstructures involving ballistic
specific cavitation energy and ductile plate perforation. penetrator formation and hypervelocity impact and pene-
Int J Imp Eng 2014; 68: 15–27. tration phenomena. Mater Charact 1996; 37: 245–276.
34. Lee YW and Wierzbicki T. Fracture prediction of thin 54. Shin H and Lee W. Material design guidelines for
plates under localized impulsive loading. Part II: discing explosive confinements to control impact shock-
and petalling. Int J Imp Eng 2005; 31: 1277–1308. induced detonations based on shock transmission/
35. Wingrove AL. The influence of projectile geometry on reflection analysis. Int J Imp Eng 2003; 28: 465–478.
adiabatic shear and target failure. Metallurg Trans 55. Mishra B, Jena PK, Hazarika B, et al. An experimental
1973; 4: 1829–1833. study on the shattering behavior of a high strength
36. Scott JB. The Hague peace conferences of 1899 and armour steel under blast and long rod penetrator
1907: a series of lectures delivered before the Johns impact. Mater Design 2010; 31: 3971–3981.
Hopkins University in the Year 1908, Baltimore, MD, 56. Lou Jf, Zhang Yg, Wang Z, et al. Long-rod penetra-
USA: The Johns Hopkins Press, 1909. tion: the transition zone between rigid and hydro-
37. Børvik T, Dey S and Olovsson L. Penetration of granu- dynamic penetration modes. Defence Technol 2014;
lar materials by small-arms bullets. Int J Imp Eng 2015; 10: 239–244.
75: 123–139. 57. Krishna Teja Palleti HN, Gurusamy S, Kumar S, et al.
38. Rosenberg Z and Dekel E. On the role of material Ballistic impact performance of metallic targets. Mater
properties in the terminal ballistics of long rods. Int J Design 2012; 39: 253–263.
Imp Eng 2004; 30: 835–851. 58. Antoun T, Curran DR, Razorenov SV, et al. Spall frac-
39. Lynch NJ. Constant kinetic energy impacts of scale size ture. New York: Springer, 2003.
KE projectiles at ordnance and hypervelocity. Int J Imp 59. Chen D, Tan H, Yu Y, et al. A void coalescence-based
Eng 1999; 23: 573–584. spall model. Int J Imp Eng 2006; 2006: 1752–1767.
40. Partom Y and Littlefield DL. Validation and calibra- 60. Danian C, Yuying Y, Zhihua Y, et al. A modified
tion of a lateral confinement model for long-rod pene- Cochran–Banner spall model. Int J Imp Eng 2005;
tration at ordnance and high velocities. Int J Imp Eng 2005: 1106–1118.
1995; 17: 615–626. 61. Vogler TJ and Clayton JD. Heterogeneous deformation
41. Paik SH, Kim SJ, Yoo YH, et al. Protection perform- and spall of an extruded tungsten alloy: plate impact
ance of dual flying oblique plates against a yawed long- experiments and crystal plasticity modeling. J Mech
rod penetrator. Int J Imp Eng 2007; 34: 1413–1422. Phys Solids 2008; 2008: 297–335.
42. Anderson CE and Orphal DL. An examination of devi- 62. Deshpande PU, Prabhu VD and Prabhakaran KV.
ations from hydrodynamic penetration theory. Int J Impulsive loading of armour by high explosive squash
Imp Eng 2008; 2008: 1386–1392. head munition. Defence Sci J 2003; 53: 357–365.
92 Proc IMechE Part L: J Materials: Design and Applications 233(1)

63. Li J and Hao H. Numerical study of concrete spall 81. Zhao SC, Song ZF and Zhoa XP. A potential approach
damage to blast loads. Int J Imp Eng 2014; 2014: 41–55. to launch hypervelocity projectiles up to 10km/s based
64. Aryaei A, Hashemnia K and Jafarpur K. Experimental on two-stage gas gun facilities. Proc Eng 2013; 58:
and numerical study of ball size effect on restitution 98–109.
coefficient in low velocity impacts. Int J Imp Eng 82. Ikkurthi VR and Chaturvedi S. Use of different damage
2010; 37: 1037–1044. models for simulating impact-driven spallation in metal
65. Starratt D, Sanders T, Cepuš E, et al. Efficient method plates. Int J Imp Eng 2004; 2004: 275–301.
for continuous measurement of projectile motion in bal- 83. Barker LM and Hollenbach RE. Laser interferometer
listic impact experiments. Int J Imp Eng 2000; 24: for measuring high velocities of any reflecting surface.
155–170. J Appl Phys 1972; 43: 4669–4775.
66. Antoinat L, Kubler R, Barou JL, et al. Perforation of 84. Dolan DH. Foundations of VISAR analysis.
aluminium alloy thin plates. Int J Imp Eng 2015; 75: Albuquerque, NM and Livermore, CA: Sandia
255–267. National Laboratories, 2006.
67. Field JE, Walley SM, Proud WG, et al. Review of 85. Barker LM. VISAR vs PDV. [Online]. Available at:
experimental techniques for high rate deformation and http://www.valynvisar.com/VISAR_vs_PDVLynn_M.
shock studies. Int J Imp Eng 2004; 30: 725–775. pdf (2011, accessed 19 May 2015).
68. Hopkinson B. A method of measuring the pressure pro- 86. Strassburger E, Hunzinger M, Patel P, et al. Analysis of
duced in the detonation of high explosives or by the the fragmentation of AlON and spinel under ballistic
impact of bullets. Philosoph Trans R Soc Lond 1914; impact. J Appl Mech 2013; 80: 031807.
213: 411–413. 87. Strand OT and Whitworth TL. Using the heterodyne
69. Kolsky H. An investigation of the mechanical proper- method to measure velocities on shock physics experi-
ties of materials at very high rates of loading. Proc Phys ments. In: 15th APS topical conference on shock com-
Soc 1949; 62: 676–700. pression of condensed matter, Waikoloa, HI, 2007.
70. Bobbili R, Ramakrishna B, Madhu V, et al. Prediction 88. Johnson G and Cook W. A constitutive model and data
of flow stress of 7017 aluminium alloy under high strain for metals subjected to large strains, high strain rates
rate compression at elevated temperatures. Defence and high temperatures. In: 7th International symposium
Technol 2015; 11: 93–98. on ballistics, The Hague, The Netherlands, 1983.
71. Chapman DJ, Radford DD and Walley SM. A history 89. Johnson GR. Numerical algorithms and material
of the Taylor test and its present use in the study of models for high-velocity impact computations. Int J
lightweight materials. Design Use Light-Weight Mater Imp Eng 2011; 38: 456–472.
2005; 465: 12–24. 90. Rule WK and Jones SE. A revised form for the
72. Forde LC, Proud WG and Walley SM. Symmetrical Johnson–Cook strength model. Int J Imp Eng 1998;
Taylor impact studies of copper. ProcR Soc A: Math 21: 609–624.
Phys Eng Sci 2009; 465: 769–790. 91. Armstrong RW and Zerilli FJ. Dislocation mechanics
73. Tong W. Pressure-shear stress wave analysis in plate based analysis of material dynamics behavior. Le
impact experiments. Int J Imp Eng 1997; 19: 147–164. Journal de Physique Colloques 1988; 49: 529–534.
74. Trajkovski J, Kunc R, Pepel V, et al. Flow and fracture 92. Tanimura S, Tsuda T, Abe A, et al. Comparison of
behavior of high-strength armor steel PROTAC 500. rate-dependent constitutive models with experimental
Mater Design 2015; 66: 37–45. data. Int J Imp Eng 2014; 69: 104–113.
75. NATO Military Agency for Standardisation. STANAG 93. Khan AS, Suh YS and Kazmi R. Quasi-static and
4164 land (edition 2) – test procedures for armour per- dynamic loading responses and constitutive modeling
foration test of anti-armour ammunition, Brussels; of titanium alloys. Int J Plast 2003; 20: 2233–2248.
NATO, 1998. 94. Klepaczko JR, Rusinek A, Rodrı́guez-Martı́nez JA,
76. United States Department of the Army. Detail specifi- et al. Modelling of thermo-viscoplastic behaviour of
cation: armour plate, steel, wrought, homogeneous (for DH-36 and Weldox 460-E structural steels at wide
use in combat vehicles and for ammunition testing), ranges of strain rates and temperatures, comparison
Aberdeen Proving Ground, MD: US Army Research of constitutive relations for impact problems. Mech
Laboratory, 2009. Mater 2009; 41: 599–621.
77. Arnold W and Schäfer FK. Behind armor blast (BAB) 95. Nemat-Nasser S and Isaacs JB. Direct measurement of
caused by shaped charges. Int J Imp Eng 1999; 23: isothermal flow stress of metals at elevated tempera-
13–25. tures and high strain rates with application to Ta and
78. Nishida M, Kato H, Hayashi K, et al. Ejecta size dis- Ta-W alloys. Acta Materialia 1997; 45: 907–919.
tribution resulting from hypervelocity impact of spher- 96. Nemat-Nasser S and Guo W. Flow stress of commer-
ical projectiles on CFRP laminates. Proc Eng 2013; 58: cially pure niobium over a broad range of tempera-
533–542. tures and strain rates. Mater Sci Eng A 2000; 284:
79. Cenna AA, Page NW, Kisi E, et al. Single particle 202–210.
impact tests using gas gun and analysis of high strain- 97. Nemat-Nasser S and Guo W. Thermomechanical
rate impact events in ductile materials. Wear 2011; 271: response of DH-36 structural steel over a wide range
1497–1503. of strain rates and temperatures. Mech Mater 2003; 35:
80. Kolopp A, Rivallant S and Bouvet C. Experimental 1023–1047.
study of sandwich structures as armour against 98. Nemat-Nasser S and Li Y. Flow stress of F.C.C. poly-
medium-velocity impacts. Int J Imp Eng 2013; 61: crystals with application to OFHC Cu. Acta
24–35. Materialia 1998; 46: 565–577.
Lenihan et al. 93

99. Rusinek A and Klepaczko JR. Shear testing of a sheet 115. Stellingwerf RF and Wingate CA. Impact modeling
steel at wide range of strain rates and a constitutive with smooth particle hydrodynamics. Int J Imp Eng
relation with strain-rate and temperature dependence 1993; 14: 707–718.
of the flow stress. Int J Plast 2001; 17: 87–115. 116. Johnson GR, Stryk RA and Beissel SR. SPH for high
100. Rusinek A, Rodrı́guez-Martı́nez JA and Arias A. A velocity impact computations. Comput Methods Appl
thermo-viscoplastic constitutive model for FCC Mech Eng 1996; 139: 347–373.
metals with application to OFHC copper. Int J Mech 117. Batra RC and Zhang GM. Modified smoothed par-
Sci 2010; 52: 120–135. ticle hydrodynamics (MSPH) basis functions for mesh-
101. Voyiadjis GZ and Abed FH. Microstructural based less methods, and their application to axisymmetric
models for bcc and fcc metals with temperature and Taylor impact test. J Computat Phys 2008; 227:
strain rate dependency. Mech Mater 2005; 37: 1962–1981.
355–378. 118. G.R. Johnson and R.A. Stryk User Instructions for
102. Seeger A. The mechanism of glide and work hardening the EPIC-2 Code, Brooklyn Park, MN: Honeywell
in face-centered cubic and hexagonal close-packed Inc., 1986.
metals. In: Dislocations and mechanical properties of 119. Ansys Inc. Ansys Autodyn: explicit software for non-
crystals. New York: Wiley, 1957. linear dynamics. [Online]. Canonsburg, PA. Available
103. Zhang DN, Shangguan QQ, Xie CJ, et al. A modified at: http://www.esss.com.br/pdf/autodyn-11.pdf (2007,
Johnson–Cook model of dynamic tensile behaviors for accessed 03 June 2015).
7075-T6 aluminum alloy. J Alloys Comp 2015; 619: 120. Dassault Systemes. Abaqus/Explicit 6.14 Data Sheet.
186–194. [Online]. Available at: http://www.3ds.com/fileadmin/
104. Tan JQ, Zhan M, Liu S, et al. A modified Johnson– PRODUCTS/SIMULIA/PDF/datasheets/simulia-
Cook model for tensile flow behaviors of 7050-T7451 abaqus-explicit-datasheet.pdf (2014, accessed 03 June
aluminum alloy at high strain rates. Mater Sci Eng A 2015).
2015; 631: 214–219. 121. Hallquist J.O. LS-DYNA theory manual. Livermore,
105. Song W, Ning J, Mao X, et al. A modified Johnson- CA: Livermore Software Technology Corporation,
Cook model for titanium matrix composites reinforced 2015.
with titanium carbide particles at elevated tempera- 122. Hohler V, Weber K, Tham R, et al. Comparative ana-
tures. Mater Sci Eng A 2013; 576: 280–289. lysis of oblique impact on ceramic composite systems.
106. Wang X, Huang C, Zou B, et al. Dynamic behavior Int J Imp Eng 2001; 26: 333–344.
and a modified Johnson-Cook constitutive model of 123. Bürger D, Rocha De Faria A, De Almeida SFM, et al.
Inconel 718 at high strain rate and elevated tempera- Ballistic impact simulation of an armour-piercing pro-
ture. Mater Sci Eng A 2013; 580: 385–390. jectile on hybrid ceramic/fiber reinforced composite
107. Hou QY and Wang JT. A modified Johnson-Cook armours. Int J Imp Eng 2012; 43: 63–77.
constitutive model for Mg-Gd-Y alloy extended to a 124. Hogan JD, Spray JG, Rogers RJ, et al. Dynamic frag-
wide range of temperatures. Computat Mat Sci 2010; mentation of natural ceramic tiles: ejecta measure-
50: 147–152. ments and kinetic consequences. Int J Imp Eng 2013;
108. Meyer HW. A modified Zerilli-Armstrong constitutive 58: 1–16.
model describing the strength and localizing behavior of 125. Worsham MJ, Treadway SK and Shirley AD. Defeat
Ti-6Al-4V. Report by the Army Research Laboratory. of high velocity projectiles by a novel spaced armor
Aberdeen, MD, US Army Research Laboratory, 2006. system. Procedia Eng 2013; 58: 338–347.
109. Xu Z and Huang F. Comparison of constitutive 126. Held M, Mayseless M and Rototaev E. Explosive
models for FCC metals over wide temperature and reactive armor. In: Proceedings of the 17th inter-
strain rate ranges with application to pure copper. national symposium on ballistics, Midrand, RSA, 1998.
Int J Imp Eng 2015; 79: 65–74. 127. De Rosset WS. Patterned armor performance
110. Chen G, Ren C, Qin X, et al. Temperature dependent evaluation for multiple impacts. Report for the
work hardening in Ti–6Al–4V alloy over large Army Research Laboratory. Aberdeen, MD: US
temperature and strain rate ranges: experiments and Army Research Laboratory, 2003.
constitutive modeling. Mater Design 2015; 83: 128. Dorsch H, Bongartz A and Rossberg D. Modelling of
598–610. ballistic effects with regard to urban operations. In:
111. Donea J, Huerta A, Ponthot J, et al. Arbitrary 27th International symposium on ballistics. Freiburg,
Lagrangian – Eulerian methods. In: Encyclopedia of Germany: International Ballistics Society, 2013;
computational mechanics. New York: John Wiley, 2013: 591–601.
1999. 129. Dorsch H, Doerr A, Hoellinger R, et al. Modelling of
112. Moxnes JF, Prytz AK, Frøyland Ø, et al. Strain rate behind wall debris effects. In: 28th International sym-
dependency and fragmentation pattern of expanding posium on ballistics, Atlanta, GA: International
warheads. Defence Technol 2015; 11: 1–9. Ballistics Society, 2014, 1462–1470.
113. Petschek AG and Libersky LD. Cylindrical smoothed 130. Arnold W and Paul W. Behind armor debris investi-
particle hydrodynamics. J Computat Phys 1993; 109: gation and their application into a new vulnerability
76–83. model. Int J Imp Eng 2001; 26: 21–32.
114. Libersky LD, Petschek AG, Carney TC, et al. High 131. Wielewski E, Appleby-Thomas GJ, Hazell PJ, et al.
strain Lagrangian hydrodynamics. J Computat Phys An experimental investigation into the micro-
1993; 109: 67–75. mechanics of spall initiation and propagation in
94 Proc IMechE Part L: J Materials: Design and Applications 233(1)

Ti-6Al-4V during shock loading. Mater Sci Eng A 137. Sartori C, Mercier S, Jacques N, et al. Constitutive
2013; 578: 331–339. behavior of porous ductile materials accounting for
132. Hazell PJ, Appleby-Thomas GJ, Wielewski E, et al. micro-inertia and void shape. Mech Mater 2015; 80:
The influence of microstructure on the shock and 324–339.
spall behaviour of the magnesium alloy, Elektron 138. Molinari A and Mercier S. Micromechanical model-
675. Acta Materialia 2012; 60: 6042–6050. ling of porous materials under dynamic loading.
133. Barrett RA, O’Donoghue PE and Leen SB. A disloca- J Mech Phys Solids 2001; 49: 1497–1516.
tion-based model for high temperature cyclic visco- 139. Molinari A, Mercier S and Jacques N. Dynamic failure
plasticity of 9-12Cr steels. Computat Mater Sci 2014; of ductile materials. Procedia IUTAM 2013; 10:
92: 286–297. 201–220.
134. Golden B, Li D, O’Dowd N, et al. Microstructural 140. Czarnota C, Jacques N, Mercier S, et al. Modelling of
modelling of P91 martensitic steel under uniaxial load- dynamic ductile fracture and application to the simu-
ing conditions. J Pressure Vessel Technol 2013; 136: lation of plate impact tests on tantalum. J Mech Phys
1–6. Solids 2008; 56: 1624–1650.
135. Li DF, Golden BJ and O’Dowd NP. Multiscale mod- 141. Anderson CE and Riegel JP. A penetration model for
elling of mechanical response in a martensitic steel: a metallic targets based on experimental data. Int J Imp
micromechanical and length-scale-dependent frame- Eng 2015; 80: 24–35.
work for precipitate hardening. Acta Materialia 142. Dunne F and Petrinic N. Introduction to computational
2014; 80: 445–456. plasticity. 2nd ed. Oxford, UK: Oxford University
136. Jacques N, Mercier S and Molinari A. A constitutive Press, 2005.
model for porous solids taking into account microscale
inertia and progressive void nucleation. Mech Mater
2015; 80: 311–323.