929

High Rates an
33. High Rates and Impact Experiments

Kaliat T. Ramesh

33.1 High Strain Rate Experiments ................ 930

Experimental techniques for high-strain-rate
33.1.1 Split-Hopkinson or Kolsky Bars ...... 931
measurements and for the study of impact-
33.1.2 Extensions and Modifications
related problems are described. An approach
of Kolsky Bars .............................. 937
to classifying these experimental techniques is 33.1.3 The Miniaturized Kolsky Bar........... 941
presented, and the state-of-the-art is briefly de- 33.1.4 High Strain Rate Pressure-Shear
scribed. An in-depth description of the basis for Plate Impact ................................ 942
high-strain-rate experiments is presented, with 33.2 Wave Propagation Experiments ............. 945
an emphasis on the development of a range of 33.2.1 Plate Impact Experiments.............. 946
strain rates and a range of stress states. The is-
33.3 Taylor Impact Experiments .................... 949
sues associated with testing metals, ceramics
and soft materials are reviewed. Next, experi- 33.4 Dynamic Failure Experiments ................. 949
33.4.1 Void Growth
mental techniques that focus on studying the
and Spallation Experiments........... 950
propagation of waves are considered, including
33.4.2 Shear Band Experiments ............... 952
plate impact and shock experiments. Experiments
33.4.3 Expanding Ring Experiments ......... 952
that focus on the development of dynamic fail- 33.4.4 Dynamic Fracture Experiments ....... 953
ure processes are separately reviewed, including 33.4.5 Charpy Impact Testing................... 953
experiments for studying spallation and dynamic

Part D 33
33.5 Further Reading ................................... 953
fracture.
References .................................................. 954

From a mechanics viewpoint, the consequences of an developed during the early times after a relatively high
impact are threefold. First, stress waves or shock waves velocity impact.
are propagated inside the impacted bodies, and the It is useful to develop a sense of the range of strain
propagation of these waves must be understood. Sec- rates developed during typical impact problems (note
ond, large inelastic deformations may be developed, that the strain rates developed as a result of an im-
typically at high rates of deformation. Third, the en- pact are always functions of time, and therefore we
tire impacted structure may be excited by the impact, focus here on the peak strain rates developed). Thus
leading to structural dynamics and vibration problems, during an asteroid impact on the Earth, the peak strain
typically at long times. This chapter explicitly ignores rates that are developed are likely to be of the or-
the last of these consequences. Further, if the impact der of 108 s−1 (this results from hypervelocity impact,
velocity is sufficiently small, all of the stress waves i. e. impact velocities above 5 km/s). For impacts cor-
propagated inside the impacted bodies will be elas- responding to typical velocities from defense-related
tic. The measurement of elastic wave propagation is terminal ballistics (≈ 1–2 km/s), the peak strain rates
discussed in other chapters, notably those on ultrason- developed are of the order of 105 s−1 to 106 s−1 . All
ics and photoacoustic characterization. Thus, our focus strain rates below the peak strain rate are likely to be de-
is on the experimental techniques associated with the veloped during the event at sufficiently long times, and
propagation of nonelastic waves, and with the meas- their significance to the problem must be determined on
urement of high-strain-rate behavior. Our emphasis is a case-by-case basis. In both the planetary impact and
therefore on the measurement of the phenomena that are ballistic impact cases, for example, substantial parts of
930 Part D Applications

the deformation occur at lower strain rates (perhaps as • High-strain-rate experiments, that is, experiments
low as 102 s−1 ). Further, significant parts of the damage designed to measure the high-strain-rate properties
that occur as a result of these events might be a re- of a material.
sult of the lower strain rate deformations (depending • Wave-propagation experiments, that is, experiments
on the damage growth kinetics). This is because the designed to understand the characteristics of wave
high-strain-rates are typically sustained for much longer propagation within the material or structure; these
times than the ultra-high-strain-rates, and it takes time may develop high strain rates as well, but the
for many damage mechanisms (for example, the nucle- high rate deformations vary in both space and
ation and growth of voids) to develop. time.
Experimental techniques that are colloquially de- • Dynamic failure experiments, that is, experiments
fined as impact experiments can have dramatically designed to understand the processes of dynamic
different objectives. Since the design of the experi- failure within a material or structure.
mental technique is critically influenced by the final • Direct impact experiments, that is, experiments
objective of the experiment, it is important to first designed to understand or discover broad impact
decide what information one wishes to extract from phenomena (such as cratering efficiency experi-
the experiment. Typically, what are commonly called ments or vehicle crash experiments).
impact experiments fall into one of four different cate-
gories in order of increasing complexity in terms of the We examine each of these types of experiments in de-
dynamics. tail.

33.1 High Strain Rate Experiments

One of the defining features of impacts that occur at high strain rates. Conventionally, strain rates at or below
velocities sufficiently large to cause inelastic (and par- 10−3 s−1 are considered to represent quasistatic defor-
Part D 33.1

ticularly plastic) deformations is that most of these mations, and strain rates below 10−6 s−1 are considered
deformations occur at high strain-rates. These de- to be in the creep domain.
formations may also lead to large strains and high Creep experiments are typically performed at rela-
temperatures. Unfortunately, we do not understand the tively high temperatures, and a variety of specialized
high-strain-rate behavior of many materials (often de- machines exist for these kinds of loading; dead loads
fined as the dependence σf (ε, ε̇, T ) of the flow stress are often of particular interest. Quasi static experiments
on the strain, strain rate and temperature), and this is are typically accomplished through a variety of ser-
particularly true at high strains and high temperatures. vohydraulic machines, and ASTM standards exist for
A number of experimental techniques have therefore
been developed to measure the properties of materials
at high strain-rates. In this section we focus on those
Pressure-shear plate impact
Conventional Kolsky bars

experimental techniques that develop controlled high

Servohydraulic machines

Miniaturized Kolsky bar

rates of deformation in the bulk of the specimen, rather

Specialized machines

than those in which high-strain-rates are developed just

behind a propagating wave front.
The primary experimental techniques associated
with the measurement of the rate dependent properties
of materials are described in Fig. 33.1 (note that the
stress states developed within the various techniques are
not necessarily identical). An excellent and relatively
10–6 10– 4 10–2 100 102 104 106 108
recent review of these methods is presented by Field
Strain rate (s–1)
et al. [33.1]. For the purposes of this discussion, strain
rates above 102 s−1 are classified as high strain rates, Fig. 33.1 Experimental techniques used for the devel-
strain rates above 104 s−1 are called very high strain opment of controlled high strain rate deformations in
rates, and strain rates above 106 s−1 are called ultra materials
 High Rates and Impact Experiments 33.1 High Strain Rate Experiments 931

most of these experiments. Most servohydraulic ma- Interface 1 Interface 2

chines are unable to develop strain rates larger than
100 s−1 repeatably, but some specialized servohydraulic
machines can achieve strain rates of 101 s−1 . Finally,
the strain rates in the intermediate rate domain (be- Input bar Specimen Output bar
V
tween 100 s−1 and 102 s−1 ) are extremely difficult to
study, since this is a domain in which wave-propagation
is relevant and cannot be easily accounted for (how-
ever, the strain rates in this domain are indeed of
interest to a number of machining problems). The pri- Fig. 33.2 Schematic of the compression Kolsky bar. The input and
mary approach to testing in this strain rate range uses output bars are designed to remain elastic, as is the pacting projec-
drop towers or dropweight machines [33.2], and great tile
care must be exercised in interpreting the data because
of the coupling between impact-induced wave prop- signed to remain elastic throughout the test. These
agation and machine vibrations [33.3–7]. There are bars sandwich a small specimen (usually cylindrical),
existing ASTM standards for some drop-weight tests, which is expected to develop inelastic deformations.
including ASTM E208-95a (2000) for measuring the The bars are typically made of high-strength steels such
ductility transition temperature of steels, ASTM E436- as maraging steel, with a very high yield strength and
03 for performing drop-weight tear testing of steels and substantial toughness. Other bar materials that have
ASTM E680-79 (1999) for measuring the impact sensi- been used include 7075-T6 aluminum, magnesium al-
tivity of solid phase hazardous materials. loys and PMMA (for testing very soft materials) and
In this section we focus on experimental techniques tungsten carbide (for testing ceramics). One end of the
for the higher strain rates (greater than 102 s−1 ), includ- input bar is impacted by a projectile made of a material
ing the high (102 –104 s−1 ), very high (104 –106 s−1 ) identical to that of the bars; the resulting compressive
and ultra-high-strain-rate (> 106 s−1 ) domains. pulse propagates down the input bar to the specimen.
The wave propagation is shown in the Lagrangian wave

Part D 33.1
33.1.1 Split-Hopkinson or Kolsky Bars propagation diagram of Fig. 33.3, where time t = 0 cor-
responds to the instant of projectile impact. Several
The now-classical experimental technique in the high- reverberations of the loading wave occur within the
strain-rate domain is the Kolsky bar or Split-Hopkinson specimen; a transmitted pulse is sent into the output bar
pressure bar (SHPB) experiment for determining the and a reflected pulse is sent back into the input bar. Typ-
mechanical properties of various materials (e.g. [33.8– ically, resistance strain gages are placed on the input and
11] in metals, ceramics and polymers respectively) in output bars for measuring
the strain rate range of 10+2 –8 × 10+3 s−1 . The terms
1. the incident pulse generated by the impacting pro-
Split-Hopkinson pressure bar and Kolsky bar are often
jectile,
used interchangeably. However, one should note that the
2. the reflected pulse from the input bar/specimen in-
term Split-Hopkinson pressure bar implies the perfor-
terface and
mance of compression experiments, whereas the term
3. the transmitted pulse through the specimen to the
Kolsky bar is more general and may include compres-
output bar.
sion, tension, torsion or combinations of all of these.
Since the fundamental concept involved in this tech- Note that the strain gauge locations are ideally such
nique – that of determining the dynamic properties of that the incident and reflected pulses do not overlap, as
materials using two long bars as transducers, with the shown in Fig. 33.3. The strain gage signals are typically
specimen size much smaller than the pulse length – measured using high-speed digital oscilloscopes with
was developed by Kolsky [33.12], we will use the term at least 10-bit accuracy and preferably with differential
Kolsky bar in the rest of this chapter. inputs to reduce noise.
We begin by describing the compression Kolsky Let the strain in the incident pulse be denoted by εI ,
bar (an extensive description is presented in [33.13]). that in the reflected pulse by εR , and that in the trans-
A schematic of the typical experimental apparatus is mitted pulse by εT (these are bar strains as measured by
shown in Fig. 33.2. The device consists of two long the strain gages). Examples of the raw signals measured
bars (called the input and output bars) that are de- on the input and output bars are presented in Fig. 33.4
932 Part D Applications

by
t
v1 − v2 cb
ės = = (εI − εR − εT ) , (33.3)
l0 l0
εT where l0 is the initial specimen length. Since the strains
εR
in the elastic bars are known, we can compute the bar
stresses and thence the normal forces at the two speci-
εI men/bar interfaces as
x P1 = E b (εI + εR ) Ab (33.4)
12
Fig. 33.3 Wave propagation (also called time-distance or and
Langrangian) diagram showing the stress waves progagat-
P2 = E b (εT ) Ab (33.5)
ing through space and time in the Kolsky bar arrangement.
The strain gauge locations are also shown. Speciman at the specimen/input-bar and specimen/output-bar in-
length is exaggerated to show the wave reverberation in the terfaces respectively. The mean axial stress s̄s in the
specimen specimen is then given by

for a test on vanadium. Then the characteristic relations P1 + P2 1 E b Ab

associated with one dimensional elastic wave propaga- s̄s (t) = = (εI + εR + εT )
2 As0 2 As0
tion in the bar tell us that the particle velocity at the (33.6)
specimen/input-bar interface is given by
  where As0 is the initial cross-section area of the speci-
v1 (t) = cb εI − εR , (33.1) men.
It is instructive to examine the wave propagation
and that at the specimen/output-bar interface is given by within the specimen in some detail (Fig. 33.3). When
Part D 33.1

the incident pulse arrives at the specimen a compres-

v2 (t) = cb εT , (33.2) sive loading wave is generated inside the specimen.
√ The compressive loading wave propagates through the
where cb = E b /ρb is the bar wave speed, with E b the specimen and arrives at the specimen output-bar inter-
Young’s modulus and ρb the density of the bar material. face. By design, the specimen’s impedance is smaller
The mean axial strain rate in the specimen is then given than the impedance of the bars surrounding the spec-
imen on both sides. Since the output bar has a higher
impedance than the specimen itself, the wave that re-
Volts
0.08
flects from the specimen/output-bar interface remains
a loading wave, resulting in an even higher compressive
0.06 stress. This wave now arrives at the specimen/input-bar
0.04 interface, again sees a higher impedance and again re-
flects as a loading wave, resulting in a further increase
0.02
in the compressive stress. This process continues until
0 the stress within the specimen reaches a value that is
–0.02 sufficiently high to generate inelastic strains, resulting
Input bar signal in finite plastic flow of the specimen under the com-
–0.04 Output bar signal pressive loading. Once substantial plastic flow of the
–0.06 specimen material has commenced, we may neglect fur-
–0.08 ther wave propagation within the specimen, since the
0 0.1 0.2 0.3 0.4 amplitude of the subsequent wavefronts will be very
Time (ms)
small. Thus at these later times the stress within the
Fig. 33.4 Examples of strain gauge signals obtained from specimen is essentially uniform; the stress is said to
the input and output bars for a high strain rate test on pure have equilibrated. If the boundary conditions are fric-
vanadium tested at 600 ◦ C tionless, the specimen stress is also uniaxial.
 High Rates and Impact Experiments 33.1 High Strain Rate Experiments 933

Once an equilibrium condition has been achieved, specimen are then given by [33.14]
we have P1 = P2 , and (33.4) and (33.5) then imply that
εs (t) = − ln [1 − es (t)] , (33.11)
εI + εR = εT . (33.7)
and
This condition, together with (33.1), (33.2), (33.4) and ės (t)
(33.5), substantially simplifies equations (33.3) and ε̇s (t) = . (33.12)
1 − es (t)
(33.6) for the specimen strain rate and specimen stress.
That is, by assuming stress equilibrium, uniaxial stress Note that compression is defined as positive. Assum-
conditions in the specimen and 1-D elastic stress wave ing plastic incompressibility, the true stress σs in the
propagation without dispersion in the bars (discussed et specimen is obtained as
seq.), the nominal strain rate ės , nominal strain es and σs (t) = ss (t) [1 − es (t)] . (33.13)
nominal stress ss (all in the specimen) can be estimated
using From (33.9) and (33.10), a stress-strain curve is ob-
tained at a strain rate defined by an average taken over
2cb
ės (t) = − εR (t) , (33.8) the strain rate history obtained in (33.8); an example
l0 of such a stress-strain curve is presented in Fig. 33.5
t (obtained on vanadium using the pulses presented in
es (t) = ės (τ) dτ , (33.9) Fig. 33.4). We note here that the correction to a true
0
strain rate represented by (33.12) has often not been
used by researchers in the field, resulting in the peculiar
and situation that papers will present true stress versus true
E b Ab strain curves obtained at an engineering (rather than
ss (t) = εT (t) , (33.10)
As true) strain-rate. The magnitude of this correction can
where As and l0 are the initial cross-sectional area and be considerable if the final strain is of the order of 20%.
length of the specimen respectively. Note the negative
sign in (33.8). This arises because the strain in the re- The Design of Kolsky Bar Experiments

Part D 33.1
flected pulse has the opposite side of the strain in the The design of a specific Kolsky bar experiment essen-
incident pulse; the latter is compressive and is conven- tially comes down to the design of the specimen (i. e.,
tionally considered positive in these experiments. Thus the choice of specimen length l0 and diameter d0 ) and
the specimen strain rate expressed by equation (33.8) is the choice of the projectile impact velocity so as to at-
a positive strain-rate, i. e., in this convention, a compres- tain a specific strain rate in the specimen. Other design
sive strain-rate. The true strain and true strain rate in the parameters such as the bar length L and the bar diam-
eter D are more relevant to the design of the Kolsky
bar system rather than the specific experiment. The most
True stress (MPa)
300
important design parameters are the three ratios: L/D,
D/d0 and l0 /d0 . Typically, L/D is of order 100, D/d0
250 is of order 2 to 4, and l0 /d0 is 0.6 to 1. The choice of bar
diameter itself depends on the material being tested, the
200 typical specimen sizes that are available, and the degree
of wave dispersion that is allowable. Although using
150 a large value of D/d0 provides a large area mismatch
and thus allows for the testing of much harder specimen
100
materials, the expectation that the specimen-bar inter-
50
face can be approximated as a plane places limits on the
Vanadium (600 °C) at 4500 s–1 value of this ratio (that is, one needs to avoid the punch
0 problem). The majority of Kolsky bars in common use
0 0.05 0.1 0.15 0.2 0.25 in the world are between 7 mm and 13 mm in diameter,
True strain
although very large bar diameters have been used for
Fig. 33.5 True stress versus true strain curve obtained us- experiments on concrete (e.g., [33.15]; very large bar
ing the signals presented in Fig. 33.4. The material is diameters have a specific disadvantage in that launching
vanadium tested at 600 ◦ C at a strain rate of 4500 s−1 the projectile may require substantial gun facilities).
934 Part D Applications

A systematic way to perform experimental design is experimental configuration force any observed stresses
presented below. The strain in the incident pulse is given and strain rates in the specimen to satisfy (33.17). The
by result is shown as the straight line in Fig. 33.6, which
V is a plot of the stress versus the strain-rate (hereafter
εl = , (33.14) called a response diagram). Note that in this figure com-
2cb
pressive stresses are presented along the positive Y-axis,
where V is the projectile velocity. Using this and the following our convention. The straight line in Fig. 33.6
equilibrium condition (33.7) in (33.10), we can write the represents the locus of all points in stress versus strain-
specimen stress entirely in terms of the reflected strain rate space that can be reached using a compression
and the impact velocity: Kolsky bar, a specific specimen geometry, and a specific
E b Ab E b Ab projectile impact velocity. Figure 33.6 also shows the
ss (t) = εT (t) = (εI + εR ) curve corresponding to the constitutive response of the
As As
  specimen material (typically the flow stress increases
E b Ab V
= + εR . (33.15) with strain-rate). The stresses and strain rates actually
As 2cb observed in a given test correspond to the intersection
However, (33.8) allows us to write the strain in the re- of the test response line with the constitutive curve, as
flected pulse in terms of the specimen strain rate shown in Fig. 33.6. Given that the constitutive response
l0 of the material is not controllable, a specific strain-rate
εR = − ės . (33.16) can only be achieved by moving the test response line;
2cb
that is, in order to sample other points along the con-
Using this in (33.15), we obtain the relation between the stitutive curve, we would need to move the straight line
specimen stress and the specimen strain rate that must corresponding to the experimental parameters. Equation
be satisfied during any given test using a specific Kol- (33.17) indicates that the maximum strain rate that can
sky bar, a specific specimen, and a particular choice of be observed in a Kolsky bar test is given by
impact velocity
   V
Ab V ės |max = , (33.18)
ss (t) = E b
Part D 33.1

l0
A 2c
s   b 
Ab l0 while the maximum compressive stress that can be ob-
− Eb ės . (33.17) served in such a test is given by
As 2cb
  
This last result represents what we call a test char- Ab V
ss |max = E b . (33.19)
acteristic response: that is, the constraints of the As 2cb

Specimen stress

Increasing
area ratio
ss|max = Eb ⎛ b⎛ ⎛ ⎛
A V
⎝As⎝ ⎝2cb ⎝
Test response line

Observed
spec. stress
Increasing V

Specimen material Decreasing l0

response curve
Fig. 33.6 Response diagramm for the
Observed compression Kolsky bar, showing the
V
spec. e· e·s|max = intersection of the material response
l0
Specimen strain rate curve (dashed) with the test response
line
 High Rates and Impact Experiments 33.1 High Strain Rate Experiments 935

Figure 33.6 shows that the maximum strain rate can est is critical for the successful use of any Kolsky bar
only be approached an actual Kolsky bar test if the ma- technique. The typical approach to checking equilib-
terial response is exceedingly soft, so that the observed rium involves a comparison of the force histories on the
specimen stress is very close to zero. Conversely, the two sides of the specimen [33.13]. Note that the axial
maximum specimen stress can only be observed in the stress distributions in the specimen cannot be measured
test if the specimen strain rate is very close to zero. This in Kolsky bar experiments, since only the forces at the
is a mathematical manifestation of the well-known re- two ends of the specimen are experimentally accessible.
sult that a given test procedure will generate lower strain For some situations (such as an experiment on a soft
rates on strong materials, while very soft materials can specimen material or at very high strain rates) even
be deformed at very high strain rates in an identical the force at the input-bar/specimen interface is difficult
test. As Fig. 33.6 shows, it is apparent that the easi- to obtain accurately, because the measure of this force
est way to get a higher maximum strain rate would be involves a subtraction of two pulses with similar magni-
to increase the impact velocity or to reduce the spec- tudes that have suffered wave dispersion. For the more
imen length. Similarly, an increase in the maximum complex experiments, involving very soft specimens or
observable stress (e.g. when studying very strong mater- extremely high-strain-rates, it is sometimes necessary to
ials) can be obtained by increasing the impact velocity perform a full computational analysis of the wave prop-
or by increasing the area mismatch Ab /As . However, agation within the specimen to obtain an estimate of the
these parameters cannot be varied arbitrarily; restric- time required for equilibration [33.16].
tions are usually imposed by the requirements for test
validity (see, for example, Jia and Ramesh [33.16], out- Friction Effects. Friction at the specimen/bar inter-
lined next). Other restrictions arise from such issues faces causes the state of stress to deviate from the
as the yielding of the bar material (which limits im- uniaxial stress condition and leads to spuriously stiff
pact velocity) and the sensitivity of the strain gages results [33.18, 19]. As Fig. 33.5 shows, decreasing the
and signal-to-noise ratio of instrumentation (which de- specimen length can greatly increase the accessible
termines the smallest stresses that can be effectively strain rate range for a given experiment. However, as
measured). the l0 /d0 ratio of the specimen becomes smaller, the

Part D 33.1
effects of friction at the interfaces become substantial
Requirements for Valid Kolsky Bar Experiments if the coefficient of friction at the specimen/bar in-
Several basic requirements must be satisfied in order terfaces is not sufficiently low, leading to inaccurate
to ensure that the measured response in a Kolsky bar measures of the stress [33.19]. For quasi-static com-
experiment is truly the constitutive behavior of the spec- pression tests, specimen l0 /d0 ratios of 1.5 to 3.0 are
imen material. We discuss each of these requirements recommended in various ASTM standards [33.20], with
briefly below. proper use of lubricant. In contrast, for conventional
compression Kolsky bar tests, l0 /d0 ratios of 0.5–1.0
Equilibrated Stresses. The nature of the Kolsky bar are widely used, following Bertholf and Karnes [33.19].
experiment is such that the loading arrives first at one The interfacial friction depends on the smoothness
side of the specimen. Because stress waves propagate of the end surfaces, the bar and specimen materials
in the specimen at finite speeds, equilibration of the and lubricant used, impact velocity and temperature.
stresses takes a non-zero time. Equilibration of the Commonly used lubricants in Kolsky bar experiments
stresses is required to ensure that the stress measured include MoS2 lubricants and lithium greases. Several
from the output bar side of the specimen represents researchers have measured the friction of lubricants at
the average stress in the specimen. In terms of experi- various strain rates using ring specimens (e.g. [33.21]).
mental design, equilibration of the stresses will occur These measurements have generally shown that the fric-
if the loading time of interest is long in comparison tion decreases at high strain rates.
with the specimen characteristic time (τs = csp /l0 , the
time for the compressive wave to traverse the length Dispersion Effects. Longitudinal waves in elastic bars
of the specimen). Davies and Hunter estimated that suffer from geometric dispersion, so that the inci-
three reverberations of the loading wave in the spec- dent, reflected and transmitted pulses change as they
imen are required for stress equilibration [33.17] for propagate along the input and output bars; however,
ductile metal specimens. The ability to check that the (33.1–33.13) essentially ignore dispersion. Wave dis-
axial stress has in fact equilibrated at the time of inter- persion affects the measured strain pulses, which in
936 Part D Applications

turn affect the measured stress-strain response. Disper- second and the third terms, the relative error (r) in flow
sion effects can be rigorously included through elastic stress measurement due to inertia is
wave propagation computations (e.g. [33.22]), but two    
consequences of dispersion remain important. First, dis- σs,2 − σy ρd 2 1 1 l0 2 2
=r = + ε̇ .
persion inevitably induces superimposed oscillations σy σy 64 6 d0
in the loading of the specimen. Second, shorter stress (33.21)
pulses with sharper rise times cause greater dispersion
because of the greater high-frequency content and the It is apparent that (for a given strain rate) the error
wider range. It is now possible to routinely correct for drops significantly when the specimen size (length or
the effects of dispersion using elastic-wave propagation diameter, given l0 /d0 ) is reduced (this is one driver for
calculations, and indeed this must be done for some miniaturizing the Kolsky bar arrangement). It is also
experiments where information that is present early in apparent that tests on materials with high flow stress
the stress-time signal is important. One approach to and low density are less prone to such inertial errors,
minimizing the effects of dispersion is to use so-called whereas very high density materials and very soft ma-
buffer materials or shaper materials that are placed be- terials will be problematic. A useful way to consider
tween the impacting projectile and input bar [33.23]. inertial effects is to examine the limiting strain rate that
This results in the shaping of the incident wave with can be attained while keeping the inertial error below
a longer rise time and thus lower dispersion. some chosen value. For a given value of the relative
error r in stress and a given ratio of l0 /d0 , the lim-
Inertial Effects. The stresses associated with axial and iting strain rate can be expressed as a function of the
radial inertia should be small compared with the flow specimen length and the specific strength σy /ρ as
stress of the material under investigation. The stress
⎛ ⎞
wave loading that is used in high strain rate experiments
1 σy 192
causes inertia to have an influence on the measured ε̇lim = r⎝  ⎠ , (33.22)
properties, particularly at very high strain rates [33.24]. l0 ρ 32 + 3
β2
The magnitude of the inertial contribution to the appar-
Part D 33.1

ent stress also depends on the density and size of the where β = l0 /d0 .
specimen [33.16]. Gorham [33.24] (following [33.25]) This gives us, for example, ε̇lim = 1/l0 [(48r)/11]1/2
developed an approach to estimate the inertial effect in [(σy )/(ρ)]1/2 when β = l0 /d0 is assumed to be 0.5. Fig-
Kolsky bar experiments that can be rewritten to empha- ure 33.7 shows this latter strain rate limit (due to inertia)
size specimen size and l/d ratio effects as follows:
   
1 1 l0 2 Limiting strain rate (s–1)
σs,2 − σy = ρd ε̇
2 2
+
64 6 d0 20 000
    Copper Alpha- Al alloy
1 1 l0 2 ρl v̇ titanium 6061-T651
− ρd ε̈
2
− − ,
32 6 d0 2 15 000
(33.20) L = 1 mm

where σs,2 and σy are the flow stress measured from the
output bar and the actual yield stress of the material re- 10 000
spectively, ρ is the density of the specimen material and
v is the velocity of the interface between the specimen L = 2 mm
and the output bar. This rearrangement is useful because 5000
the l/d ratio is constrained to remain within a fairly nar- L = 4 mm
row range (0.5 ≤ l/d ≤ 1), and so (33.20) allows us to
focus on the effect of specimen size through the speci- 0
0 5 000 10 000 15 000 20 000
men diameter. In any experiment with a nearly constant
Specific strength (m2/s2)
strain rate, the magnitude of the second term is much
smaller than that of the first term, and the magnitude of Fig. 33.7 Limiting strain rates in compression Kolsky bar
the third term is generally small for materials which do experiments due to inertial errors in flow stress meas-
not exhibit a very high strain hardening. Ignoring the urement
 High Rates and Impact Experiments 33.1 High Strain Rate Experiments 937

as a function of the specific strength for a value of error or a pulse-shaper material at the projectile/impact-bar
r of 3% (variations smaller than 3% are typically not ob- interface (this material is typically a soft metal, for
servable in Kolsky bar experiments). Three typical spec- example copper) [33.28]. The question of when equili-
imen sizes and three commonly used materials are also bration occurs remains a matter of some debate. A rule
shown in the figure [33.16]. This is a useful approach of thumb that is commonly used is that at least five re-
to estimating the limiting strain rate in very-high-strain- verberations of the elastic wave should occur before the
rate experiments for a given specimen size and a known difference in the stress in the two sides of the ceramic
specific strength of the material. This is also helpful for specimen becomes sufficiently small to be ignored. This
clarifying the sometimes dramatic reported changes of also generates limits on the effective strain rates that can
flow stresses in the range of strain rates of ≈ 1 × 103 – be developed for valid Kolsky bar testing [33.29]. Note
≈ 5 × 104 s−1 , by helping to distinguish valid from in- that for the harder hot-pressed ceramics, the specimen
valid tests. strain is extraordinarily difficult to measure because of
indentation of the ceramics into the platens.
33.1.2 Extensions and Modifications The specimen design also has significant impact on
of Kolsky Bars the results obtained in Kolsky bar experiments on ce-
ramics. In a sense, it is possible to view experiments in
A very large number of extensions and modifications ceramics in one of two modes. One wishes to measure
of the traditional Kolsky bar system have been devel- the properties of the ceramic itself – this is often the
oped over the last five decades. Most of these are listed original intent. However, one may instead be examining
in Table I of the review by Field et al. [33.1], who in- the properties of the ceramic specimen as a structure,
clude an exhaustive literature set. We discuss some of since the stress concentrations at the corners and the
the key modifications in the rest of this section, focusing edges can have such a strong influence on the fail-
on some particular challenges. ure process. In the quasi static mechanical testing of
ceramics, researchers have gone to great lengths to opti-
Modifications for Ceramics mize the specimen design. This degree of optimization
There are three major difficulties associated with test- has not yet been accomplished in the dynamic testing

Part D 33.1
ing ceramics in the conventional compression Kolsky of ceramics. Several specimen designs have been pro-
bar. First, these materials are extremely hard, and thus posed and used, e.g. [33.30]. It appears that substantial
are likely to cause damage to the bar ends. Second, gains in understanding can be obtained by juxtaposing
these materials do not develop substantial plastic strain, high-speed photography (Fig. 33.8) of specimen defor-
so that failure will be caused essentially in the elas- mation during loading with the recorded stress-time
tic domain and specimen strain can be very difficult curves [33.27]. The figure demonstrates that, in this
to measure. Third, since these materials are extremely polycrystalline ceramic:
brittle, the specimen design can easily introduce defects
that will cause premature failure, raising the question 1. axial splitting is not the mode of compressive failure
of whether the material properties can be measured. and
An excellent summary of these issues is presented in 2. the interaction of flaws is of great importance in
the ASM Handbook article by Subhash and Ravichan- dynamic brittle failure under compression.
dran [33.26]. The first of these difficulties can be
somewhat mitigated by using special inserts (called Modifications for Soft Materials
platens) at the end of the bars that are made of even The testing of very soft materials (such as polymers
harder ceramics, e.g. [33.27]. The second and third dif- and soft tissues) in the Kolsky bar represents the al-
ficulties are much harder to address. The fact that failure ternate extreme of difficulty in comparison to testing
occurs immediately after the elastic domain means that ceramics [33.31]. The two primary difficulties here are
it becomes very important to ensure that stress equili-
1. the transmitted stress is very small, and so can be
bration has occurred in the specimen, and this cannot
measured only with great difficulty, and
be done with the traditional trapezoidal pulse. The best
2. equilibration of the stress in the specimen can take
way to resolve this issue is to use a controlled rise time
a very long time [33.32].
for the incident pulse, and to ensure that equilibration
occurs well before failure develops. The incident pulse There have been substantial improvements in the test-
shape is controlled [33.23] by using a buffer material ing of soft materials over the last two decades, primarily
938 Part D Applications

1 2 3 4
5.2 mm

2.3 mm

5 6 7 8

Fig. 33.8 High speed photographs of the development of microcracks in a transparent polycrystalline ceramic (AION)
under dynamic compression along the horizontal axis [33.27]. The interframe time is 1 μs and each exposure time is 20 ns

involving the use of polymer or magnesium bars with arrangement (for conductive specimens) is induction
small elastic moduli, so that small stresses can still heating [33.37].
generate large enough strains to measure [33.33]. The A second, more modern approach to running high
use of semiconductor strain gages instead of foil gages temperature experiments in the Kolsky bar is to heat
has also provided significant improvement [33.33]. The the specimen alone, and then to bring the bars into
equilibration question is addressed sometimes by the contact with the specimen just before impact [33.39].
direct measurement of force using piezoelectric quartz This setup has the advantage of reducing temperature
Part D 33.1

gages [33.34]. All of these approaches require great care gradients in the bars, and the added benefit of requir-
with both experimental setup and data analysis [33.35]. ing relatively small energy inputs to heat the specimen.
One particularly important question that must be ex- The major design issue in such a system is the cold
plicitly addressed for each material is that of material contact time – the duration of time over which the
incompressibility. input and output bars are in contact with the speci-
men before the arrival of the incident wave. During
Modifications for High Temperature Testing this contact time, thermal gradients develop within the
The most common approach to running high temper- specimen, and the overall specimen temperature drops
ature Kolsky bar experiments is to simply heat the by an amount that is typically not recorded. Vari-
specimen while it is in contact with the input and output ous arrangements have been used to reduce this cold
bars (typically inside a tube furnace), allowing a strong contact time, e.g. [33.39] and [33.40]. Approaches to
temperature gradient to develop within the bars. If an heating the specimen include RF heating and radiative
estimate of the temperature gradient can be made, its heating [33.39]. Conductive heating and pulse heating
effect can be compensated for by making the proper approaches have also been used to develop very large
adjustments to the wave-speeds due to the temperature- heating rates in metallic specimens [33.41].
dependent elastic modulus of the bar material [33.36].
Such techniques have limitations on the attainable spec- Tension Kolsky Bars
imen temperature that are imposed by heat conduction A version of the Kolsky bar system that is effective in
in the bars and by the properties of the bar mater- tension was developed in the 1960s [33.42]. The basic
ials. One solution to this problem has been to provide principle of this system is identical to the compression
an insulating layer between the specimen and bars in system, except that a method for generating a tensile
the form of a short length of impedance-matched ce- wave must be used, and special tension grips are re-
ramic bar [33.37]. A variant of this technique has been quired for the specimen. There are two basic approaches
used to perform high temperature torsion tests [33.38]. to generate a tensile wave in the input bar. The first
One method of providing rapid heating in such an approach, known as an impact-tension approach, con-
 High Rates and Impact Experiments 33.1 High Strain Rate Experiments 939

sists of firing a tubular projectile at a flange at the end pulse provides the strain rate history (which can then
of a bar, thereby generating a tensile wave within the be integrated over time to obtain the specimen strain).
bar. A variant of this approach uses a reaction mass and Multiple strain gauge stations may be needed in order
causes the reflection of a compressive wave as a tensile to separate out the incident and reflected pulses. The
wave [33.43]. The second, and less common, approach equations are identical to those used for the compres-
consists of storing a large tensile strain within a section sion Kolsky bar, with the specimen length replaced by
of a bar and then suddenly releasing it is so as to gen- a gauge length on the tensile specimen. A full tensile
erate a tensile wave (this is known as the direct-tension stress-strain curve is obtained at fairly high strain rates
approach) [33.44]. (on the order of 102 –103 s−1 using this system).
The gripping of tensile specimens requires some However, the traditional approach to obtaining the
care, because otherwise the spurious wave reflections specimen strain in the tension Kolsky bar provides an
from the grips would make it difficult to interpret the inaccurate measure of strain, in general. This is be-
experiments. Commonly used grips include threaded cause it measures the relative displacement of the two
ends and adhesively bonded ends. A general rule of specimen-bar interfaces, and converting this to speci-
thumb is that if the gripped ends are made of the same men strain requires an assumption of a specimen gauge
material as the bar and are no longer than the bar diam- length (this is very well known in compression, but
eter, then the effect on the wave propagation will be is not well defined in tension). Since fillets are re-
minimal. An example of the direct-tension Kolsky bar quired in tensile specimens in order to avoid failure
is shown in Fig. 33.9 [33.45]. A tensile force is ap- due to stress concentrations at the specimen flanges,
plied to a section of the input bar using the hydraulic the non-uniform deformation developed at the flanges
actuator system, while the rest of the input bar is re- contributes to the total deformation of the specimen.
strained using a friction clamp. When the friction clamp It is therefore difficult and at times impossible to de-
is suddenly released, a tensile pulse propagates down termine the effective gauge length to be used in the
the input bar and loads the specimen in tension; part calculation of specimen axial strains [33.42]. Ideally
of this incident pulse is transmitted into the output bar, one should have a strain-gage mounted directly on
and part is reflected back to the input bar. Strain gages the specimen to provide measures of strain through-

Part D 33.1
are mounted on the input and output bars to measure out the experiment. One alternative approach is to use
elastic strain in the bars. Using these strain gages, the a high-speed camera to directly measure the specimen
incident, reflected and transmitted pulses in the bars can diameter and length during an experiment. A sec-
be measured. As before, the transmitted pulse provides ond alternative is to use the Laser Occlusive Radius
the stress history of the specimen, while the reflected Detector (LORD) technique to directly determine the

L0

D1-20-UNF D0
R

L2 L1 L2

Strain gauge stations

Axial actuator Input bar Clamp Specimen Output bar

1 2 3

3048 1524

Fig. 33.9 Schematic of the direct-tension Kolsky bar. The inset shows an example of a threaded specimen
940 Part D Applications

local specimen strains during high-strain-rate experi- fact this is one of the most common applications of the
ments [33.45]. This approach requires the assumption torsional bar system [33.48].
of incompressibility in the specimen in order to es- A significant experimental concern in the setup of
timate the axial specimen strain (an assumption that the torsional Kolsky bar system is the avoidance of
is typically not valid once necking begins). However, bending waves. Flexural waves in a long bar may have
it is capable of measuring fairly large plastic strains, signal content propagating at speeds similar to that of
which are generally not measurable using standard on- the torsional wave, resulting in some difficulties in wave
specimen strain gauges because they debond during the analysis. It is particularly important that the friction
deformation. clamp design not generate flexural waves within the
bars – note in particular that the use of heavy friction
Torsional Kolsky Bars clamps is not an advantage. Careful placement of the
The torsional Kolsky bar is an adaptation of the com- supports and specific initial measurement of the propa-
pression Kolsky bar that is used to generate high strain gation of bending waves within the bar is warranted.
rate shearing deformations within typically tubular The difficulties associated with specimen gripping
specimens [33.46]. An excellent summary of the tech- are also an issue with the torsional Kolsky bar design.
nique is presented by Hartley et al. [33.47]. A schematic Typical approaches to gripping the specimen include
of such a device is presented in Fig. 33.10. The principle the use of adhesives, asymmetric flanges, and various
of operation of this bar is identical to that at the tension types of compression grips. Specimen design is critical,
Kolsky bar, except that a torsional wave is now prop- with the usual questions associated with the definition
agated down the input bar towards the specimen. One of specimen gauge length and definition of fillet radii. It
advantage of the torsional Kolsky bar is that primary- is recommended that a full finite deformation analysis
mode torsional waves are not dispersive in the bar, so of the plastic deformation of a specimen be performed
that dispersion corrections are not necessary. A second before using the given specimen design with an experi-
advantage of this system is that (in principle) very large ment.
shearing strains can be sustained and so material be- The equations associated with the use of the tor-
havior can be studied over very large strain regimes. sional Kolsky bar system are different in detail, but
Part D 33.1

That is, the difficulties associated with maintaining fric- not in character, from those used for the compression
tion during large deformations (in the compression bar) Kolsky bar system. A full description of the relevant
and avoiding necking (in the tension bar) are avoided equations is presented by Hartley et al. [33.47]. The pri-
in the torsional bar system. Further, radial expansion mary distinctions arise because of the torsional mode:
and contraction are not significant in torsion of isotropic one must consider torques and angular velocities rather
materials, so that many of the inertial and end effects than axial forces and translational velocities. We note
present in the tension and compression tests are absent that a response diagram very similar to Fig. 33.6 may
in the torsional Kolsky bar experiment. However, shear also be constructed for the torsional Kolsky bar, and
bands may develop during torsional deformations: in may be used in a very similar way to accomplish

Torquing
device
Clamp Input bar Specimen Output bar

Fig. 33.10 Schematic of the torsional Kolsky bar. Note the location of the friction clamp with respect to the supports
 High Rates and Impact Experiments 33.1 High Strain Rate Experiments 941

experimental design. Examples of shear stress versus Gorham [33.51] obtained strain rates of 4 × 104 s−1
shear-strain curves obtained using the torsional Kolsky in direct impact using a high-speed camera for strain
bar are presented in Fig. 33.15 for the aluminum alloy measurement, and even higher rates were obtained in
6061-T6. a similar configuration by Safford [33.52]. Gorham
The torsional Kolsky bar system has also been mod- et al. [33.53] and Shioiri et al. [33.54] obtained the
ified to perform high-temperature testing at high shear strain in direct impact experiments by assuming that
strain rates [33.38]. In this approach the necessity of stress equilibrium is satisfied throughout the deforma-
using grips on the specimen can be turned into an ad- tion; strain rates of 2–4 × 104 s−1 are claimed. Kamler
vantage, with the grips acting as thermal insulators and et al. [33.55] developed a very small (1.5 mm dia.) di-
reducing the heating of the bars themselves. rect impact Kolsky bar and performed experiments on
copper at very high strain rates of 6 × 103 to 4 × 105 s−1 ;
33.1.3 The Miniaturized Kolsky Bar again, stress equilibrium was assumed.
In the third approach, several fully miniaturized
Attempts to push the Kolsky bar arrangement to higher versions of the compression Kolsky bar have been de-
strain rates have led to three modifications: decrease veloped (e.g., see Jia and Ramesh [33.16]). In the Jia
of the specimen length, direct impact on the specimen and Ramesh version (see Fig. 33.11), the bars are on
(thus increasing effective velocity), and miniaturization the order of 3 mm in diameter and 30 cm long, and may
of the entire system. The first approach, with smaller be made of maraging steel or tungsten carbide. Sample
specimen sizes, is typically limited by frictional effects. sizes are cubes or cylinders on the order of 1 mm on
The second approach is the so-called direct impact tech- a side; cube specimens are used when the failure mode
nique: a projectile directly impacts a specimen placed in must be imaged using a high-speed camera, or when
front of an elastic bar (which operates like the output bar the amount of material is so small that only cuboidal
in the regular version of the Kolsky bar). Since there is specimens can be cut.
no input bar, the impact velocity can be very high. How- Very high strain rates (up to 5 × 104 s−1 ) can be
ever, because of the absence of an input bar, there is no attained in miniaturized systems, while retaining the
reflected signal from which the strain rate and strain in ability to study materials at strain rates as low as

Part D 33.1
the specimen can be extracted. Strain measurement in 1.0 × 103 s−1 (the maximum achievable strain-rate is
the direct impact approach has been addressed in several limited by an inertial correction and varies with the
ways, including material being tested). Both computational and experi-
mental results have shown that this extended capability
1. assuming the projectile is effectively rigid [33.49] can be attained without violating the requirements for
and valid high-rate testing, and indeed while improving the
2. measuring the velocity of the back surface of the quality of the experimental results in terms of preci-
projectile [33.50]. sion and accuracy [33.16]. The technique is simple, and
the entire system can be designed to fit on a desktop –
the technique is therefore sometimes referred to as the
desktop Kolsky bar.
Equilibration of stresses can be rigorously checked
from both sides of the specimen in the experiment
even during the highest rate deformations that can be
attained. Both experiments and simulations show that
rapid equilibration of the stress can be achieved even
at very high strain rates. However, effective lubrication
remains critical. The miniaturized system makes it pos-
sible to reach a very high strain rate in a small sample
with a relatively high l/d ratio, thus minimizing the in-
fluence of friction. Inertial errors in stress measurement
are significantly reduced with the smaller specimens.
A comparison of the response measured by this
Fig. 33.11 Photograph of a miniaturized Kolsky bar (total miniaturized technique with the results of the tradi-
length of 30  ) tional compression Kolsky bar system is presented in
942 Part D Applications

Flow stress (MPa) Flyer

600 Anvil
Desktop Kolsky bar (l /d = 0.55)
500 Desktop Kolsky bar (l /d = 1.1)
Conventional Kolsky bar (l /d = 0.55)
V0
400 u0 υ0 TDI
Incident beam
300
NVI
TDI
200
Specimen

100 Fig. 33.13 Schematic of the high-strain-rate pressure-shear

plate impact experiment. The specimen thickness is graetly
0 exaggerated for clarity
103 104 105
Strain rate (s–1)
on a projectile which is launched down the barrel of
Fig. 33.12 Comparison of results obtained using the minia- a gas gun towards a stationary target or anvil plate. The
turized and traditional compression Kolsky bars. The projectile velocity (V0 ) at impact may be decomposed
material is 6061-T6 aluminium alloy into a normal component u 0 = V0 cos θ and a transverse
component v0 = V0 sin θ, where θ is the angle between
Fig. 33.12 for 6061-T6 aluminum alloy. This figure the normal to the plates and the direction of flight of the
shows the rate-sensitivity over an extended range of projectile.
strain rates that is not attainable with the traditional The target is positioned in a special fixture (known
technique. Even higher-strain-rates can be attained as the target-holder) within an evacuated chamber. The
using the pressure-shear plate impact experiment, de- flyer and the anvil plates are aligned before impact
scribed et seq. using an optical technique described by Klopp and
Part D 33.1

The miniaturized Kolsky bar or desktop Kolsky bar Clifton [33.56]. Rotation of the projectile is prevented
experimental technique is particularly effective when it by a Teflon key in the projectile, which glides within
comes to understanding the properties of nanostructured a matching keyway machined in the barrel. The mis-
materials, or of materials with very fine grain sizes. alignment (tilt) between the two impacting plates is
Note, however, that the small specimen sizes can make obtained by measuring the times at which the flyer
it difficult to use this approach with materials (such as makes contact with four pins that are mounted flush
solders) that have large grain sizes or a large microstruc- with the impact face of the target. Tilts better than
tural length scale.

33.1.4 High Strain Rate Pressure-Shear t Thin specimen

Plate Impact Flyer Anvil

The high strain rate pressure-shear (HSRPS) plate TDI

impact technique was developed to study the shear- window
ing behavior of materials undergoing homogeneous NVI
window
shearing deformations at extremely high shear rates
of 104 –106 s−1 , while under superimposed hydrostatic
pressures of several GPa. A detailed description of this
technique is provided by Klopp and Clifton [33.56].
The experiment involves the impact of plates that
are flat and parallel but inclined relative to their direc-
x
tion of approach (Fig. 33.13). The specimen consists of
a very thin (of the order of 100 μm thick), and very flat Fig. 33.14 Wave propagation diagram for high-strain-rate
plate of the material under investigation. This thin spec- pressure-shear plate impact, with the specimen carried on
imen is bonded to a hard plate (the flyer), itself carried the flyer
 High Rates and Impact Experiments 33.1 High Strain Rate Experiments 943

1 milliRadian are routinely achieved in these experi- volved and a one dimensional analysis is both sufficient
ments. and rigorously correct. Like most plate impact experi-
At impact, plane longitudinal (compressive) and ments, this is a uniaxial strain experiment in that no
transverse (shear) waves are generated in the specimen transverse normal strains can occur during the time of
and the target plate (Fig. 33.14) propagating at the lon- interest.
gitudinal wave speed cl and the shear wave speed cs . Measurements of the particle velocities at the free
These waves reverberate within the specimen, result- surface of the target plate are made using laser interfer-
ing in a buildup of the normal stress and of the shear ometry off a diffraction grating that is photodeposited
stress within the specimen material. Information on the onto the rear surface. The normal velocity and the
stress levels sustained by the specimen material is car- transverse displacement at the center of the rear sur-
ried by the longitudinal (normal) and transverse waves face of the target are measured using a normal velocity
propagating into the target plate. Since the target re- interferometer (NVI) and a transverse displacement
mains elastic, there is a linear relationship between the interferometer (TDI), e.g. [33.57]. The laser used is typ-
stresses and the particle velocities in the target plate. ically either an ion laser or a diode-pumped, frequency
Thus it is sufficient to measure the normal and trans- doubled, Nd:YAG laser. The interferometric signals are
verse particle velocities in the target plate to deduce the captured by high speed silicon photodiodes (with rise
stress state and deformation state within the specimen. times < 1 ns) and recorded on high-speed digitizing os-
The entire experiment is completed before any unload- cilloscopes, using typical sampling rates of 1 ns/point
ing waves from the periphery of the plates arrives at the and bandwidths of 500 MHz – 1 GHz.
point of observation, so that only plane waves are in- The longitudinal and transverse waves generated
on impact (Fig. 33.14) reverberate within the specimen,
remaining loading waves since the impedance of the
a) Normal stress σ material of the flyer and target plates is (by design)
higher than that of the specimen materials. The develop-
ment of the stress in the specimen is most easily seen in
the characteristics diagram of Fig. 33.15. The character-

Part D 33.1
Anvil Flyer
istic relations (σ ± ρcl u = const. or τ ± ρcs v = const.,
states states with σ and τ the normal stress and shear stress and u
and v the normal and transverse particle velocities re-
ρ cl spectively; ρ is the density of the plate material) in the
linear elastic plates give us
ρ cl
−σ + ρcl u = ρcl u 0 ,
τ + ρcs v = ρcs v0 , (33.23)
u0
b) Shear stress τ in the flyer and
−σ + ρcl u = 0 ,
τ + ρcs v = 0 , (33.24)

in the anvil. Thus all states in the flyer will lie on the
Anvil Flyer solid line passing through u 0 (for the normal stress,
states ρ cs states
Fig. 33.15a) or v0 (for the shear stress, Fig. 33.15b),
τf while all states on the anvil will lie on the solid line
passing through the origin. In the case of the normal
stress (Fig. 33.15a), the normal stress in the specimen
must always lie along the characteristic lines shown by
the dashed lines. The state of the specimen at any given
υA υfs υF υ0
time is given by the intersection of the characteristic
Fig. 33.15a,b Characteristic diagram for high-strain-rate line of specimen with either the characteristic line of
pressure-shear plate impact. (a) Development of normal the anvil or the characteristic line of the flyer. If there
stress and (b) shear stresses in the specimen is a difference in the normal velocities on the two sides
944 Part D Applications

of the specimen (the flyer side and the anvil side), then The shear stress history in the specimen is obtained
the specimen will continue to be compressed in the nor- from the transverse particle velocity history using the
mal direction. However, the plate impact experiment elastic characteristics of the target plate (Fig. 33.14b)
is a uniaxial strain experiment rather than a uniaxial
1
stress experiment like the Kolsky bar. The continued τ(t) = ρcs vA = ρcs v fs (t) , (33.28)
compression along the normal direction thus implies 2
a change in volume. Due to the finite compressibility of where ρcs is the shear impedance of the anvil plate. The
the specimen, the normal stress in the specimen attains shear stress and the shear strain in the specimen can
an equilibrium value (corresponding to the intersection now be cross-correlated to give the shear stress-shear
of the flyer and anvil characteristics) given by strain curve (Fig. 33.16) for the material tested at an
essentially constant shear rate corresponding to (33.26).
1
σeqm ∼
= ρcl V0 cos θ , (33.25) Examination of (33.25)–(33.28) shows that the
2 critical measurements that must be made during the
where ρcl is the acoustic impedance of the steel, V0 experiment are of the projectile impact velocity V0 and
is the projectile velocity and θ is the skew angle. the transverse free-surface particle velocity vfs . The
At equilibrium, the hydrostatic pressure in the speci- latter particle velocity is measured using the transverse-
men can be approximated to the normal compressive displacement-interferometer [33.57]. The projectile im-
stress, differing from it by at most the strength of the pact velocity can be measured in a number of different
specimen material. Note that the difference in normal ways. These include measuring the times when the
velocities on the two sides of the specimen contin- projectile comes in contact with specific pins placed
uously decreases. Thus the normal strain-rate in the a known distance apart; measuring the times when the
high-strain-rate pressure-shear plate impact experiment projectile crosses two magnetic or radiofrequency sen-
is initially high but tends towards zero, actually approx- sors or laser beams that are spaced a known distance
imating zero when normal stress equilibrium has been apart; or using a Laser Line Velocity Sensor [33.58].
achieved. This is again a result of the uniaxial strain Finally, the misalignment (tilt) between the flyer plate-
condition. specimen combination and the anvil plate must be
Part D 33.1

The shear stress in the specimen also increases measured for every test, both for purposes of diagnos-
with each reverberation of the shear wave within the tics and to remove small components of the normal
specimen (Fig. 33.15b), until the specimen starts flow- velocity that may show up within the transverse velocity
ing at a stress level τf (the flow stress). Thereafter, signal due to the misalignment.
a finite difference in the transverse velocity can be main-
tained across the two surfaces of the specimen (unlike
Shear stress (MPa)
the normal velocity situation, it is possible to sustain 400
a difference in transverse velocities across the spec-
imen because the uniaxial strain condition does not 350 High-strain-rate
present a constraint in this case). The velocity differ- pressure-shear plate impact
300
ence vF − vA also corresponds to (see Fig. 33.15b) the
velocity difference v0 − v fs , where v fs is the transverse 250
free-surface velocity measured at the rear surface of the 200
target using the TDI. The nominal shear strain rate in Torsional Kolsky bar
the specimen is then given by 150
650 s–1
V0 sin θ − v fs 100 850 s–1
γ̇ = , (33.26) 9.5×104 s–1
h 50 Al 6061-T6: Shear 2.3×105 s–1
where h is the specimen thickness. The nominal shear 0
strain rate in the specimen can be integrated to give the 0 0.05 0.1 0.15 0.2
Shear strain
nominal shear strain history in the specimen:
Fig. 33.16 Shear stress versus shear strain curves for
t 6061-T6 aluminium alloy obtained using high-strain-rate
γ (t) = γ̇ (τ) dτ . (33.27) pressure-shear plate impact, compared with those obtained
0 usinf the torsional Kolsky bar
 High Rates and Impact Experiments 33.2 Wave Propagation Experiments 945

The high strain rate pressure-shear plate im- obtained in a given test is the intersection of the ma-
pact technique is capable of achieving shear rates terial response curve with the test response line, as
of 8 × 104 to 106 s−1 , depending on the specimen in Fig. 33.6.
thickness, [33.57–59]. A version of this experiment The superimposed hydrostatic pressures that can be
that is designed to allow recovery of the speci- exerted during the high-strain-rate pressure-shear plate
men (for microstructural examination) after a single impact experiment may be as high as 10 GPa, depend-
high-strain-rate shear loading has also been devel- ing on the impedances of the flyer and target plates and
oped [33.60]. The primary equations (33.25)–(33.28) the projectile velocity. The superimposed hydrostatic
can be rewritten to provide a locus of test re- pressures must always be remembered when compar-
sponse, as presented for the Kolsky bar in Fig. 33.6. ing high-strain-rate pressure-shear plate impact data
Here we find that the test requires that the shear with data obtained using the other techniques shown in
stress and the shear strain rate obey the relation- Fig. 33.1, since all of the other techniques can generate
ship τ = 0.5ρcs (V0 sin θ − γ̇ h), representing the test essentially uniaxial stress states, typically correspond-
response line. The maximum shear strain rate that ing to low hydrostatic pressures. In particular, while the
can be attained is γ̇max = V0 sin θ/h (and corresponds effect of pressure on the flow stress of most metals is
to τ = 0), while the maximum shear stress that negligible in comparison with the effect of strain-rate,
can be attained is τmax = 0.5ρcs V0 sin θ (and corre- the effect of pressure on the strength of polymers and
sponds to the apex of the characteristic triangle in amorphous materials may be substantial (even in com-
Fig. 33.15). The specific shear stress and shear rate parison with the effect of strain-rate).

33.2 Wave Propagation Experiments

Experiments designed to study the propagation of a fundamental feature of large amplitude wave propa-
large amplitude stress waves within materials consti- gation in materials: in the timescales associated with

Part D 33.2
tute a very broad class of impact experiments. Note the wave propagation, it is typically not possible to
that we do not include in this category those experi- observe the far-field stress state, and so locally one
ments that are designed to study the propagation of is typically exploring the uniaxial strain condition. In
waves within structures - such experiments are bet- other words, some local confinement is an inherent
ter considered in discussions of structural dynamics characteristic of large amplitude wave propagation in
or of elastic wave guides (see photoacoustic charac- materials as a result of impact. This can make it diffi-
terization chapter). Our interest here is in experiments cult to compare results obtained at ultra-high strain rates
that examine the interactions of waves with materials, (typically obtained with uniaxial-strain experiments)
particularly exciting inelastic modes such as plastic- with results obtained at high and very high strain rates
ity, cracking or other kinds of damage. In contrast (typically obtained with uniaxial stress experiments),
to the previous section, the experiments in this sec- particularly if the material has pressure dependent prop-
tion all generate strain rates and stress states that vary erties.
in both space and time, and the wave propagation is The strain rates developed in large-amplitude wave
fundamentally dispersive because of material behav- propagation experiments (where shocks are developed)
ior. can be of the order of 106 –108 s−1 , but exist only for
In broad terms, wave-propagation experiments of a short time behind a propagating wave front, and be-
this type fall into the same two categories considered cause of inelastic dissipation (as well as reflections from
in the previous section: bar wave experiments [33.61– surfaces), the strain rates will vary with position within
66] and plate impact experiments, or more specif- an impacted plate. The temperatures behind the wave
ically, uniaxial-stress wave propagation experiments front may be substantial, and must be accounted for as
and uniaxial-strain wave propagation experiments. The well. Comparisons of material properties estimated us-
plate impact experiments are far more commonly used, ing wave propagation experiments and high-strain-rate
since they can explore a wider range of the phenom- experiments (the distinction made in this chapter) can
ena that arise in impact events, and so we focus on therefore require careful parsing of experimental condi-
such experiments here. It is perhaps worth pointing out tions.
946 Part D Applications

33.2.1 Plate Impact Experiments speed. Significantly higher impact velocities will result
in the propagation of shock waves into the plates. The
The basic concept of the plate-impact experiment has propagation of plastic waves within materials is a topic
been touched on in the description of the high-strain- of some interest; however, the majority of plate impact
rate pressure-shear plate impact experiment above. experiments are conducted at velocities sufficiently high
A flyer plate is launched down a gun barrel towards to generate shock waves within the plates, and so we
a stationary target plate, typically using a gas gun for will describe shock wave experiments in some detail.
launch. The flyer plate is normally carried on a pro- It should be noted that the strain state in normal
jectile known as a sabot, and is launched at velocities plate impact experiments is the uniaxial strain state only
ranging from a few tens of meters per second up to for a finite time. Once release waves from the bound-
several kilometers per second using a variety of gas ary of the plates arrive at the point of interest, the strain
and powder guns. The flyer plate may impact the tar- state is no longer uniaxial-strain, and much more com-
get plate at normal incidence, resulting in what is called plex analyses are necessary. However, before the release
a normal plate impact experiment. An oblique plate waves arrive at the point of interest (for instance, the
impact experiment consists of an impact that occurs location at which particle surface velocities and dis-
at an angle as described in the pressure-shear plate placements are being measured using interferometry),
impact experiment. The vast majority of plate impact the strain state is rigorously one-dimensional and can
experiments that are performed today are normal plate be analyzed much more easily. Longer experimental
impact experiments, largely because oblique plate im- times in the uniaxial-strain condition can be obtained
pact experiments typically require a keyed gun barrel by using larger diameter plates, but the increased cost of
to prevent rotation of the projectile during flight. We performing such experiments with large diameter plates
consider only normal plate impact experiments in the is substantial. Typical experiments are performed with
remainder of this section. plate diameters on the order of 50 mm.
A schematic of the simplest form of the normal plate
impact experiment is shown in Fig. 33.17. The mini- Shock Wave Experiments
mum diagnostics that are typically obtained consist of In any uniaxial strain wave, there is always a differ-
Part D 33.2

measurements of the projectile impact velocity and the ence between the normal stress σa (the stress along
rear surface particle velocity. Once the impact occurs, the direction of wave propagation) and the transverse
normal uniaxial-strain compressive waves will be prop- stress σt (the stress transverse to the direction of wave
agated in both the flyer and the target. The case where propagation, which arises because of the uniaxial-strain
these waves are elastic is not interesting. Typically the constraint). The difference between these two stresses
target plate is expected to deform inelastically during can be considered to be a shear stress and is of the
the experiment, and in most experiments both flyer and order of magnitude of the deviatoric strength (yield
target will do so. If the impact velocity is sufficiently strength for metals) of the material. The difference in
high, one or both of the propagating waves within the the amplitude σa of the stress wave and the propagating
plates will be inelastic. At velocities just above the ve- pressure ( p = (σa + 2σt )/3, all stresses defined as pos-
locity required to cause yield of the target material, both itive in compression) is (2/3)(σa − σt ). Since this latter
elastic and plastic waves will propagate into the target, quantity is of the order of magnitude of the flow stress,
with the elastic precursor propagating at a higher wave the difference between the longitudinal stress and the
pressure is of the order of the strength of the material.
When the amplitude of the stress wave is extremely
Projectile velocity υ Target plate
large in comparison to the strength of the material, the
stress wave may be approximated as a shock wave,
and treated essentially as a problem in hydrodynamics.
Interferometry
A shock wave is a sharp discontinuity in pressure, tem-
perature energy and density (more specifically, a shock
wave is a traveling wave front across which a discon-
Flyer tinuous adiabatic jump in state variables occurs). There
Sabot
are several excellent reviews of shock wave propagation
Fig. 33.17 Schematic of the normal phase impact experiment, in- ([33.67, 68]). The application of the balance of mass,
cluding the sarbot balance of momentum and balance of energy conditions
 High Rates and Impact Experiments 33.2 Wave Propagation Experiments 947

to the shock wave gives us the so-called Rankine– from the biannual meetings of the American Physical
Hugoniot jump conditions across the shock [33.67]. Society Topical Group on Shock Compression of Con-
These are: densed Matter published by the American Institute of
Physics [33.70–74] and a series on the Shock Com-
Mass : ρ0 Us = ρ1 (Us − u p ) , (33.29)
pression of Solids published by Springer [33.75–79].
Momentum : P1 − P0 = ρ0 Us u p , (33.30) Experimental details are often only presented in these
Energy : e1 − e0 = 1/2(P1 + P0 ) conference proceedings, and so the reader is advised to
× (v0 − v1 ) . (33.31)
examine these books carefully.
The intent of most shock wave experiments is to
Here the initial state is denoted by the subscript 0 and measure a so-called Hugoniot curve for a material.
the final state is denoted by the subscript 1; v repre- A Hugoniot curve (Fig. 33.18) is the locus of all pos-
sents the specific volume (the reciprocal of the density). sible final states (e.g., in a pressure-specific volume
In addition, an equation of state must be prescribed for space) attainable by a single shock from a given ini-
each material that connects the specific internal energy tial state. We note that such a curve couples the material
of the material to the pressure, temperature and density. response with the thermodynamics, and does not repre-
A shock wave generated during a plate-impact experi- sent a true thermodynamic path. The shock speed Us
ment propagates at a shock speed Us that varies with is given by the slope of the Rayleigh line shown in
the particle velocity u p , and it is commonly observed Fig. 33.18, and is thus clearly dependent on the particu-
that these two variables are related linearly or nearly so: lar final state.
As an example, consider the shock wave analysis
Us = U0 + su p , (33.32)
of the normal impact of two identical plates in the
where U0 and s are material-specific parameters (the schematic experiment shown in Fig. 33.17. We assume
first is essentially the sound wave speed in the material). that the impact velocity is V0 and that there is no unload-
Large numbers of experiments have been performed ing from the edges during the time of interest. At the
to determine these parameters in various materials moment of impact shock waves are generated in both
(e.g., a summary of such data is presented by Mey- the target and the flyer plates. What are the pressure and

Part D 33.2
ers [33.68] Table 4.1). Another useful reference is that velocity states in the two plates after the shock? The
by Gray [33.69]. initial conditions are that the target plate is initially at
The shock wave propagation literature is extensive, rest, while the flyer plate is initially moving at the im-
including a large number of conference proceedings pact velocity (Fig. 33.19). The interface condition is one
of traction continuity and velocity continuity at the A-B
interface between the two plates. Applying (33.30) and
Pressure (33.32) on each side of the interface, we obtain
Hugoniot curve
PB = ρ0B U0B u pB + ρ0B sB u 2pB (33.33)
Final state in the target and
P
PA = −ρ0A (U0A + sA u pA + V0 )(u pA − V0 ) (33.34)

Rayleigh line
a) Flyer Target b) t

Initial state
P0

A B
υ υ0 u0 =V0 u0 = 0
Specific volume x
Fig. 33.18 Schematic of a Hugoniot curve from a shock Fig. 33.19 (a) Schematic of symmetric impact shock wave
experiment showing how the initial and final states are experiment. (b) First phase of shock wave propagation in
connencted by a Rayleigh line impacted plates
948 Part D Applications

in the flyer, with the subscripts A or B used to indi- sonnel. Most of these facilities are either gas guns, light
cate variables in the target or flyer respectively. Thus the gas guns or powder guns, and for the higher velocities
states at A and B (the pressure and velocity states im- multistage guns are typically required. Since kinetic en-
mediately after the shock has been generated) are given ergy increases with the square of the velocity, reaching
by the solutions of (33.33) and (33.34). The solution is higher velocities typically requires the use of lower-
shown graphically in Fig. 33.20; note the similarity to mass sabots and flyers. Velocities greater than 10 km/s
the characteristics diagram in Fig. 33.15 for the high- have been achieved with ≈ 1 g flyers using multi-stage
strain-rate pressure-shear plate impact experiment. This guns at some national laboratories [33.83]. The second
approach is sometimes called the impedance matching of these issues, the accurate measurement of projectile
method for determining the solution. Problems involv- velocities, has already been dealt with in the section
ing asymmetric impact can be solved in exactly the on high-strain-rate pressure-shear plate impact. The last
same way. We note that the release waves (reflections two issues, the measurement of the stress state and the
from the free surfaces) are somewhat more complicated, measurement of the particle velocity at these very short
since release does not occur along the Rayleigh line but time scales are discussed in this section.
along the isentrope instead. Detailed approaches to the There are several quantities that are of interest in
solution of specific shockwave problems are presented a shock wave plate impact experiment. These are the
in the books by Drumheller [33.67], Meyers [33.68] and normal stress, the transverse stress, the shock speed,
Zukas [33.80]. the particle velocity in the direction of wave propa-
The major experimental issues associated with gation (not all other particle velocity components are
shockwave plate impact experiments are zero), the pressure and the temperature. Of the normal
stress, the transverse stress and the pressure, only two
1. the development of gun launching facilities at the of these have to be measured independently. Note that
appropriate velocities; if the shock impedance of the material is known, then
2. the accurate measurement of projectile velocity; measurements of the particle velocity can be used to
3. the measurement of the stress state within the spec- infer at least one of the stresses. Thus one needs in
imen; and general to measure two stresses, the shock speed, one
Part D 33.2

4. the measurement of the particle velocities in the tar- particle velocity and the temperature. All of these need
gets. to be measured with nanosecond resolution over time
frames of several microseconds. Table 33.1 (adapted
We do not discuss the first of these issues in any detail: from [33.1]) lists some typical ways in which these
such gun-launch facilities are typically extremely spe- quantities can be measured.
cialized facilities made by a small number of companies Excellent summaries of experimental techniques for
and national laboratories, and extraordinary precautions the measurement of these quantities are provided in the
must be taken to ensure the safety of all laboratory per- recent review by Field et al. [33.1] and in a chapter by
Meyers [33.68].
The installation of stress gauges within specimens
Pressure (stress gauges must of necessity be placed within speci-
mens rather than on free surfaces) is a difficult task to
perform consistently, and individual laboratories have
developed specialized ways of performing these instal-
A/B lations. Some of the difficulties associated with the use

Target states Flyer states Table 33.1 Measurement approaches (after [33.1])
(33) (34)
Variable of interest Measurement approach
Normal stress Stress gauges
Transverse stress Stress gauges (lateral gauges)
A or B υ0 Particle velocity Interferometry (e.g. VISAR)
Particle velocity Shock speed Velocity history or stress history
Fig. 33.20 Solving for the initial state in the symmetric Temperature Pyrometry [33.81],
impact shockwave experiment spectroscopy [33.82]
 High Rates and Impact Experiments 33.4 Dynamic Failure Experiments 949

of stress gauges are described by Gupta [33.84]. The use gated through a transparent window material. The most
of lateral stress gauges to measure transverse stresses common approach to performing such interferometry
appears to be particularly difficult, and remains un- is using the velocity interferometer system for any re-
common. However, excellent measurements of strength flector (VISAR) developed by Barker et al. [33.90–92].
have been obtained using this technique [33.85]. Versions of the system that use fiber optic probes, and
The stress gauges that are used are either of the versions that can provide data from multiple points
piezoresistive or piezoelectric types. Typical piezoresis- (the so-called Line VISAR) have been developed; the
tive gages are manganin [33.86] or ytterbium [33.87] latter can be used to look at heterogeneous wave
stress gauges. Manganin gauges are perhaps the most structures [33.93]. Most shockwave experiments are de-
commonly used, being cheap and commercially avail- signed to provide a one-dimensional strain state, so
able for use up to 40 GPa. Typical piezoelectric stress that single-point measurements should be representa-
gauges include quartz [33.88] and PVDF [33.89], al- tive of the events. However, some of the phenomena that
though PZT and lithium niobate gauges are also used. develop during shock wave propagation lead to hetero-
PVDF gages are typically used for stress levels up to geneous structures. These include the development of
1 GPa, while quartz gages may be used up to 5 GPa. failure waves and spallation. In such circumstances, it
The particle velocity on the rear surface of the target is a distinct advantage to be able to obtain full field in-
is typically measured using laser interferometry. The formation using high-speed photography (this is most
rear surface may be a free surface, or may be interro- useful for transparent materials) [33.94].

33.3 Taylor Impact Experiments

The most commonly used direct impact experiment sired velocity, and the ability to measure the deformed
is the Taylor impact experiment. The experiment in- shape of the rod. However, the experiment has the
volves the impact (at normal incidence) of a rod onto disadvantage that it essentially represents an inverse

Part D 33.4
a plate (assumed to be nearly rigid). Substantial dy- problem: one must try to determine material proper-
namic deformations are developed at the site of the ties on the basis of the macroscopic measurement of
impact, leading to a local expansion of the rod and an inhomogeneous deformation. Several improvements
the development of a final shape that depends on the have been made to this experiment, including the use of
impact velocity and on the properties of the material high-speed photography to measure the process of de-
of the rod. An alternative version of the Taylor im- formation rather than simply the final deformation, and
pact experiment involves a symmetric impact of two techniques for measuring the internal hardness variation
rods. The Taylor impact experiment has the advan- within sections of the deformed bar. An excellent review
tage that it is extremely simple to perform. All that of the Taylor impact experiment is provided by Field
is needed is a facility for launching the rod at the de- et al. [33.1].

33.4 Dynamic Failure Experiments

One must distinguish between the kinds of experiments neous, and a small number of measurements (typically
needed to understand the deformation of a material and single point or area averaged measurements) are suffi-
the kinds of experiments needed to understand a failure cient to obtain information about the deformation and
process within the material or structure. The former are stress fields of interest. However, once a failure pro-
most easily studied with experimental techniques that cess begins in the material, the deformation field rapidly
develop homogeneous stress fields, homogeneous strain becomes localized, and the experimental techniques
fields and well-controlled loading. In such experiments, used to extract information using single-point or area-
as in the high-strain-rate experiments described earlier averaged measurements no longer provide adequate
in this chapter, as the deformation evolves, the stress information. Understanding the failure process typically
fields and deformation fields remain nearly homoge- requires a different suite of experimental measurements,
950 Part D Applications

and often requires a completely different experimental because these complex failure processes may feature
design. interactions between the failure process, the dynamic
A simple example of this can be seen in the case loading, nearby free surfaces, and high strain rate mater-
of fracture: the uniaxial tension experiment is very well ial properties, a theoretical understanding of the failure
developed for the measurement of material behavior un- process can be a great asset in designing appropri-
der uniaxial tension, but the utility of this experiment to ate experiments. Early experiments in this area were
determine the ability of the material to resist the propa- primarily concerned with defining the phenomenol-
gation of cracks is minimal. Instead, specially designed ogy of these failure processes, helping to determine
fracture experiments are required. This is one reason which mechanisms should be included in the model-
why, although material scientists have a great affection ing. The development of improved analyses and much
for the simple tension experiment, they find it difficult to greater computational power now affords us the op-
extract useful information about the toughness of a ma- portunity to perform experiments with the appropriate
terial by examining the elongation of the specimen in measurements to be able to extract parameters that char-
tension. A similar situation arises when understanding acterize the failure process. Indeed, an early decision
the dynamic failure process: specialized experiments that must be made in experimental design is whether
are required to measure the dynamic failure process, and the objective is to determine the micromechanisms as-
normal high-strain rate experiments must be interpreted sociated with the dynamic failure process or to provide
with care when failure processes are occurring. characteristic parameters that define the sensitivity of
The experimentalist in the dynamics of deformable the material to this particular failure process.
solids does have one advantage over the experimentalist
in the traditional mechanics domain. Most failure pro- 33.4.1 Void Growth
cesses are inherently dynamic, and he/she may already and Spallation Experiments
have access to instruments with sufficient time resolu-
tion to resolve the failure process. However, there is also At this time, analytical and computational papers on the
a potential complication. The failure process, itself dy- dynamic growth of voids outnumber experimental pa-
namic, may be initiated by the arrival of a dynamic load. pers by more than 10 to 1, attesting to the difficulty
Part D 33.4

Thus, both slowly varying and rapidly varying loads of performing controlled dynamic experiments on void
may result in the development of dynamic failures. growth. Most experiments that consider the dynamic
There are three primary modes of failure in most growth of voids are plate impact experiments, although
engineering materials: a few experiments have considered dynamic tension
(for example using the tension Kolsky bar) with vary-
1. void nucleation, growth and coalescence,
ing degrees of triaxiality [33.97] produced by inserting
2. crack nucleation, growth and coalescence, and
controlled notches. Dynamic tension experiments with
3. shear band development.
controlled triaxiality are essentially dynamic versions
The first and third of these are typically associated of similar experiments that have been performed to
with ductile failure, while the second is associated with examine the quasi-static process of void growth and
brittle failure. However, all three modes may be ob- coalescence [33.98]. Understanding these experiments
served simultaneously during a ductile failure event. requires a coupled computational and experimental ap-
Most experimental techniques in the dynamic fail- proach (sometimes called a hybrid approach), since
ure literature are designed to examine the growth of the stress state is deliberately nonuniform. The util-
a pre-existing void or crack, although the nucleation ity of the experiments is greatly improved if they are
process may be of particular importance during dy- coupled with high-speed photography of the develop-
namic failure [33.95]. The late stages of any failure ing deformation, as well as the use of digital image
process typically involve coalescence processes and can correlation [33.99–105]. Comparison of computational
be the most difficult to resolve in an experiment. We and experimental results is particularly useful in ex-
begin by discussing experimental techniques associ- amining such tension experiments [33.106]; we note,
ated with the measurement of void growth and shear however, that such experiments alone are not sufficient
band development under dynamic loading. We do not for determining the mechanics associated with the fail-
discuss experimental techniques for dynamic fracture ure processes that occur inside the specimen. Rather,
in any detail, since this topic is heavily discussed in such experiments help bracket the parameters and thus
the literature [33.96]. It is important to remember that constrain the assumed failure model. Quantitative meas-
 High Rates and Impact Experiments 33.4 Dynamic Failure Experiments 951

and shear waves [33.112]. The technique has the advan-

Flyer Target t
tage that the location of the prior crack is known, and
Rarefaction fan the subsequent failure process can be studied using both
the measured diffracted waves and the fracture surface
morphology [33.113].
The second type of plate impact experiment of in-
terest here is the spallation experiment [33.114, 115].
u0 = 0 The basic approach is identical to that in the shockwave
Spall plane x experiments described in Fig. 33.19, with the differ-
ence that the phenomenology of interest is developed
Fig. 33.21 Schematic and wave propagation diagrams for when the release waves interact (Fig. 33.21). The flyer
the spallation experiment. Note the location of the spall is normally thinner than the target, and may be made
plane with respect to the two rarefaction fans of a different material. The wave propagation within
this arrangement is shown in Fig. 33.21. When the com-
urements of the internal deformations associated with pressive shock waves reach the free surfaces of the
the dynamic growth of voids would be ideal, but such flyer and the target plates, they are reflected as release
measurements are not available in solids except with waves. However, release waves unload along the isen-
very expensive stereoscopic x-ray equipment. The first trope rather than along the Rayleigh line (Fig. 33.18),
steps in this direction have been taken by the PCS group so that a rarefaction fan is developed in each case.
at the Cavendish [33.107]. Very high tensile stresses are developed very rapidly
Plate impact experiments that examine void growth at the location where the two rarefaction fans inter-
are of two types. In the first type, pioneered by Clifton sect, and the corresponding plane is called the spall
and his group at Brown [33.109–112], a pre-fatigued plane. Voids nucleate, grow and coalesce along the spall
crack is generated within a target plate that is sub- plane, resulting in the separation of a piece of the tar-
jected to impact, such that a reflected tensile wave loads get plate. This is called a spall failure or spallation
the crack. The existence of the prior crack generates (note the identical experiment can be performed for

Part D 33.4
a well-defined region within which the void growth relatively brittle materials such as hardened steels). The
can commence. This approach has been used with development of the spall failure results in specific sig-
both brittle materials [33.109, 110] and ductile mater- natures in the particle velocity at the rear surface of
ials [33.111], and the loading has included both tensile the target plate which can be interpreted in terms of

Critical stresses (GPa)

7
Maraging steel
Non-thermal critical stress
Adiabatic critical stress
6 Range of spall strengths
4340 steel
5

4 α-Ti

Tantalum
3 Ti–6Al –4V
Annealed Cu
Cold 7075-T6
drawn
2 CU Fe
Mild steel Al
1
6061-T6 2024-T4

0 Fig. 33.22 Spall strength of a num-

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Strain hardening exponent n ber of metals, as listed by Wu
et al. [33.108]
952 Part D Applications

the spall strength of the material [33.116]. An excel- band literature, since they can be considered to be sim-
lent review of spallation experiments, and a discussion ple shear experiments (at least until the onset of the
of the approaches to interpreting the results is presented instability). However, the technique has been shown to
by Antoun et al. [33.117]. be extremely sensitive to defects in the dimensions of
The spall strengths of a large number of materials the specimen [33.123] and defects in the material, as
have been determined (Fig. 33.22), and a wide range of with all dynamic failure processes.
values is quoted for some cases. One of the key diffi- To counter this, a predefined defect can be intro-
culties in this experiment is that the rate of hydrostatic duced into a thin-walled tube that is then subjected to
tensile loading (or the tensile strain-rate) is difficult to dynamic torsion (note such experiments are essentially
control accurately, since this quantity depends both on designed to study the failure process). Duffy and his
the process of void growth and on the local hydrostatic coworkers introduced dimensional defects by hand us-
tension (which cannot be measured directly). However, ing local polishing. Deltort [33.124] and Chichili and
Wu et al. [33.108] have shown that a lower approxima- Ramesh [33.125] introduced standardized notches as
tion to the spall strengths of most metals can be obtained defects in known locations using machining. Chichili
using a critical pressure model. and Ramesh [33.126] also developed a recovery tech-
nique, which allowed the recovery of specimens within
33.4.2 Shear Band Experiments which an adiabatic shear band had been grown under
a single known torsional pulse, with a superimposed
The initiation and development of adiabatic shear hydrostatic compressive load. A similar capability has
bands in materials have been reviewed by Bai and been developed by Bai and co-workers [33.127]. Such
Dodd [33.118] and more recently by Wright [33.119]. approaches are critical for the development of an un-
Adiabatic shear bands represent thermo-viscoplastic in- derstanding of the deformation mechanisms [33.125]
stabilities, and are often observed in ductile metals associated with adiabatic shear bands under well-
subjected to high rates of loading, because there is in- defined dynamic loads.
sufficient time to conduct away heat during the event. A number of other experimental techniques have
Such shear bands can be the dominant mode of fail- been developed, in which the emphasis is on the de-
Part D 33.4

ure in an impact event, and are particularly common velopment of adiabatic shear bands in a repeatable
in impacts involving ductile metals subjected to overall manner so that the deformation micromechanisms can
compression and also in the perforation and punching be studied. These include hat specimens deformed
of sheets and plates. Adiabatic shear bands are also in a compression Kolsky bar [33.128], dynamic per-
observed in applications such as high-speed machin- foration or punching experiments [33.129], explosive
ing, where they limit the speed of the manufacturing loading [33.130] and shear compression susceptibil-
process. ity experiments [33.131]. A very interesting approach
The classical experimental technique used to study is to consider the asymmetric impact of prenotched
adiabatic shear bands is the dynamic torsion of thin- plates [33.132]. In all of these cases, the simultaneous
walled tubes in a torsional Kolsky bar experiment, as use of high-speed photography is a great asset. Further,
developed by Marchand and Duffy [33.48]. In this ap- given that the temperature is a critical component of
proach, an initially homogeneous torsional deformation the thermoviscoplastic problem, direct dynamic meas-
of a thin-walled tube evolves into a strongly localized urements of the temperature of the shear band tip are of
deformation with the development of an adiabatic shear great interest [33.133]. Unfortunately, a complete char-
band somewhere along the length of the tube during acterization of the deformation state and temperature
continued loading. Simultaneous recording of the stress state at the shear band tip continues to be difficult, and
history and the end displacement history together with is an area of continuing research.
time-resolved high-speed photography of the continu-
ing deformation provides a direct measure of the time of 33.4.3 Expanding Ring Experiments
occurrence of the adiabatic shear band. This technique
has been used to study adiabatic shear band develop- The tensile instability that is analogous to shear band
ment in a variety of metals, including steel [33.120], development is the onset and growth of necking.
Ti-6Al-4V [33.121] and tungsten heavy alloys [33.122]. However, the dynamic measurement of the necking in-
Experiments of this type have guided much of the mod- stability is very difficult. The difficulties that arise are
eling that has been performed in the adiabatic shear identical to the difficulties described in the high rate
 High Rates and Impact Experiments 33.5 Further Reading 953

tension experiment earlier in this chapter: the end con- static fracture testing, dynamic fracture testing does
ditions required to develop the tension typically modify not have an accepted set of testing standards. Thus
the failure mode. The most effective way to gener- there are a wide variety of specimen geometries, con-
ate tension and study the necking instability is to use straints, and loading conditions in use [33.96]. Most
an expanding ring experiment. The dynamic expansion of these have in common the generation of a fatigued
of the ring produces an initially axisymmetric tension pre-crack and an attempt to measure the K-dominated
mode, and the onset of the necking instability repre- crack-tip stress, strain or deformation fields, typically
sents the breakage of symmetry within the experiment. using optical diagnostics and/or high-speed photog-
There are three fundamental approaches to develop- raphy. The specimens are often impulsively loaded
ing dynamic expansion of rings. The first is to use using a Kolsky bar type configuration, or by di-
an explosive expansion, the second to use an electro- rect impact in either 3-point bend, 1-point bend or
magnetically driven expansion, and the third to use asymmetric impact configurations (the Kalthoff experi-
the axial motion of a wedge to develop a radial ex- ment [33.149]). Useful results can also be obtained
pansion of an enclosing ring. In practice, the first and with crack opening measurements and properly located
third approaches are very difficult to control. Thus the strain gauges [33.150], but it can be extremely diffi-
primary experimental method available for expanding cult to identify the onset of crack propagation from
rings is that of electromagnetic launch [33.134–139]. the signals. The more commonly used optical diagnos-
This approach is often used to examine the develop- tics include photoelasticity [33.151], caustics [33.152],
ment of tensile fragmentation in metallic systems, and dynamic moiré [33.153], and coherent gradient sens-
has recently been coupled with high-speed photogra- ing [33.154].
phy to provide unprecedented detail of the dynamic
necking process [33.140]. Although these experiments 33.4.5 Charpy Impact Testing
are extremely difficult to perform, the analysis of such
experiments is very attractive to the theoretician and Relatively rapid measures of the dynamic toughness of
has the benefit of providing a great deal of insight (re- a material can be obtained using the ASTM standard
cent analyse are presented by Zhou et al. [33.141] for Charpy impact test (ASTM E23-06). The technique

Part D 33.5
the brittle case and by Guduru and Freund [33.142] for essentially uses a notched bar for a specimen and
the ductile case). Related experiments involve explosive a pendulum machine as an impacting device. Commer-
loading to generate expanding cylinders [33.143]. cial pendulum machines are available to perform such
experiments. There is also a NIST recommended guide
33.4.4 Dynamic Fracture Experiments for maintaining Charpy impact machines, NIST Spe-
cial Publication 960-4. Versions of these devices have
We do not discuss experimental techniques for dy- been developed for subscale testing, as well as for the
namic fracture in any detail, since this topic is heavily testing of polymers. A recent reference book on pen-
discussed in the literature [33.144–148] and there is dulum impact machines has also been published by
a very recent book on the subject [33.96]. Unlike quasi ASTM [33.155].

33.5 Further Reading

• T. Antoun, L. Seaman, D.R. Currar, G.I. Kanel, • J.E. Field: Review of experimental techniques for
S.V. Razorenov, A.V. Utkin: Spall Fracture high rate deformation and shock studies, Int. J. Imp.
(Springer, New York 2004) Eng. 30, 725–775 (2004)
• Y.L. Bai, B. Dodd: Adiabatic Shear Localization: • G.T. Gray III: Classic Split-Hopkinson pressure bar
Occurrence, Theories and Applications (Pergamon, testing. In: ASM Handbook, Vol. 8, ed. by H. Kuhn,
Oxford 1992) D. Medlin, (ASM Int., Materials Park Ohio 2000)
• D.S. Drumheller: Introduction to Wave Propagation pp. 462–476
in Nonlinear Fluids and Solids (Cambridge Univ. • K.A. Hartley, J. Duffy, R.H. Hawley: The torsional
Press, Cambridge 1998) Kolsky (Split-Hopkinson) bar. In: ASM Handbook,
954 Part D Applications

Vol. 8, ed. by H. Kuhn, D. Medlin, (ASM Int., Ma- • T. Nicholas, A.M. Rajendran: Material characteri-
terials Park Ohio 1985) pp. 218–228 zation at high strain-rates. In: High Velocity Impact
• R.W. Klopp, R.J. Clifton: Pressure-shear plate im- Dynamics, ed. by J.A. Zukas 1990, (Wiley, New
pact testing. In: ASM Handbook, Vol. 8, ed. by York 1990) pp. 127–296
H. Kuhn, D. Medlin, (ASM Int., Materials Park • K. Ravi-Chandar: Dynamic Fracture (Elsevier, Am-
Ohio 1985) pp. 230–239 sterdam 2004)
• M.A. Meyers: Dynamic Behavior of Materials (Wi- • T.W. Wright: The Mathematical Theory of Adia-
ley Interscience, New York 1994) batic Shear Bands (Cambridge Univ. Press, Cam-
bridge 2002)
References

33.1 J.E. Field, W.G. Proud, S.M. Walley, H.T. Goldrein: 33.13 G.T. Gray III: Classic split-Hopkinson pressure bar
Review of experimental techniques for high rate testing. In: ASM Handbook, Vol. 8, ed. by H. Kuhn,
deformation and shock studies. In: New Experi- D. Medlin (ASM Int., Materials Park Ohio 2000)
mental Methods in Material Dynamics and Impact, pp. 462–476
ed. by W.K. Nowacki, J.R. Klepaczko (Institute 33.14 K.T. Ramesh, S. Narasimhan: Finite deformations
of Fundamental Technological Research, Warsaw and the dynamic measurement of radial strains
2001) p. 109 in compression Kolsky bar experiments, Int. J. Sol.
33.2 C.S. Coffey, V.F. DeVost: Drop weight impact Struct. 33(20), 3723–3738 (1996)
machines: a review of recent progress, JANNAF 33.15 T.S. Lok, X.B. Li, D. Liu, P.J. Zhao: Testing and
Propulsion Systems Hazards Subcommittee Meeting response of large diameter brittle materials sub-
CPIA Publ., 1(446), 527–531 (1986) jected to high strain rate, J. Mat. Civil Eng. 14(3),
33.3 J.D. Winkel, D.F. Adams: Instrumented drop weight 262–269 (2002)
impact testing of cross-ply and fabric composites, 33.16 D. Jia, K.T. Ramesh: A rigorous assessment of the
Composites 16(4), 268–278 (1985) benefits of miniaturization in the Kolsky bar sys-
33.4 N. Banthia, S. Mindess, A. Bentur, M. Pigeon: tem, Exp. Mech. 44(5), 445–454 (2004)
Impact testing of concrete using a drop-weight 33.17 E.D.H. Davies, S.C. Hunter: The dynamic compres-
impact machine, Exp. Mech. 29(1), 63–69 (1989) sion testing of solids by the method of the split
Part D 33

33.5 P.R. Sreenivasan, S.K. Ray, S.L. Mannan, P. Ro- Hopkinson pressure bar (SHPB), J. Mech. Phys.
driguez: Determination of K(Id) at or below NDTT Solids 11, 155–179 (1963)
using instrumented drop-weight testing, Int. J. 33.18 J.F. Bell: An experimental diffraction grating study
Fract. 55(3), 273–283 (1992) of the quasi-static hypothesis of the SHPB experi-
33.6 C.S. Coffey, V.F. Devost: Impact testing of explosives ment, J. Mech. Phys. Solids 14, 309–327 (1966)
and propellants, Propellants Explosives Pyrotech- 33.19 L.D. Bertholf, C.H. Karnes: Two dimensional anal-
nics 20(3), 105–115 (1995) ysis of the split Hopkinson pressure bar system, J.
33.7 H.M. Hsiao, I.M. Daniel, R.D. Cordes: Dynamic com- Mech. Phys. Solids 23, 1–19 (1975)
pressive behavior of thick composite materials, 33.20 ASTM: ASTM Standard E9 3.01 102: Standard test
Exp. Mech. 38(3), 172–180 (1998) method of compression testing of metallic mater-
33.8 T. Nicholas, A.M. Rajendran: Material characteriza- ials at room temperature (ASTM International, West
tion at high strain-rates. In: High Velocity Impact Conshohocken 1999)
Dynamics, ed. by J.A. Zukas (Wiley, New York 1990) 33.21 F. Wang, J.G. Lenard: An experimental study of
pp. 127–296 interfacial friction-hot ring compression, Trans.
33.9 S.M. Walley, J.E. Field: Strain rate sensitivity of ASME: J. Eng. Mater. Technol. 114, 13–18 (1992)
polymers in compression from low to high strain 33.22 R. Skalak: Longitudinal impact of a semi-infinite
rates, DYMAT J. 1, 211–228 (1994) circular elastic bar, J. Appl. Mech. 24, 59–64 (1957)
33.10 G. Ravichandran, G. Subhash: A micromechanical 33.23 D.J. Frew, M.J. Forrestal, W. Chen: Pulse shaping
model for high strain rate behavior of ceramics, techniques for testing brittle materials with a split
Int. J. Sol. Struct. 32, 2627–2646 (1995) Hopkinson pressure bar, Exper. Mech. 42, 93–106
33.11 D. Jia, K.T. Ramesh, E. Ma: Effects of nanocrystalline (2002)
and ultrafine grain sizes on constitutive behavior 33.24 D.A. Gorham: Specimen inertia in high strain-rate
and shear bands in iron, Acta Mater. 51(12), 3495– compression, J. Phys. D 22, 1888–1893 (1989)
3509 (2003) 33.25 J.Z. Malinowski, J.R. Klepaczko: A unified analytic
33.12 H. Kolsky: An investigation of the mechanical prop- and numerical approach to specimen behaviour in
erties of materials at very high rates of loading, the SHPB, Int. J. Mech. Sci. 28, 381–391 (1986)
Proceedings of the Physical Society, Vol. 62B (Lon- 33.26 G. Subhash, G. Ravichandran: Split-Hopkinson bar
don, 1949) p. 676 testing of ceramics. In: ASM Handbook, Vol. 8, ed.
 High Rates and Impact Experiments References 955

by H. Kuhn, D. Medlin (ASM Int., Materials Park Ohio microsecond time-resolved pyrometry, Int. J. Ther-
2000) pp. 497–504 mophys. 25(2), 561–574 (2004)
33.27 H. Wang, K.T. Ramesh: Dynamic strength and frag- 33.42 J. Harding, E.O. Wood, J.D. Campbell: Tensile test-
mentation of hot-pressed silicon carbide under ing of materials at impact rates of strain, J. Mech.
uniaxial compression, Acta mater. 52(2), 355–367 Engng Sci. 2, 88–96 (1960)
(2004) 33.43 S. Nemat-Nasser, J.B. Isaacs, J.E. Starrett: Hopkin-
33.28 G. Subhash, G. Ravichandran: Mechanical be- son techniques for dynamic recovery experiments,
haviour of a hot pressed aluminum nitride under Proc. Royal Soc. London A20, 371–391 (1991)
uniaxial compression, J. Mater. Sci. 33, 1933–1939 33.44 G.H. Staab, A. Gilat: A direct-tension split Hop-
(1998) kinson bar for high strain-rate testing, Exp. Mech.
33.29 G. Ravichandran, G. Subhash: Critical appraisal of 31(3), 232–235 (1991)
limiting strain rates for compression testing of ce- 33.45 Y. Li, K.T. Ramesh, E.S.C. Chin: Comparison of the
ramics in a split Hopkinson pressure bar, J. Amer. plastic deformation and failure of A359/SiC and
Ceram. Soc. 77, 263–267 (1994) 6061-T6/Al2O3 metal matrix composites under dy-
33.30 W. Chen, G. Subhash, G. Ravichandran: Evalua- namic tension, Mat. Sci. Eng. A 371(1-2), 359–370
tion of ceramic specimen geometries used in split (2004)
Hopkinson pressure bar, DYMAT J. 1, 193–210 (1994) 33.46 W.E. Baker, C.H. Yew: Strain-rate effects in the
33.31 W. Chen, F. Lu, D.J. Frew, M.J. Forrestal: Dynamic propagation of torsional plastic waves, Trans.
compression testing of soft materials, Trans. ASME: ASME: J. Appl. Mech. 33, 917–923 (1966)
J. Appl. Mech. 69, 214–223 (2002) 33.47 K.A. Hartley, J. Duffy, R.H. Hawley: The torsional
33.32 G.T. Gray III, W.R. Blumenthal: Split-Hopkinson Kolsky (split-Hopkinson) bar. In: ASM Handbook,
pressure bar testing of soft materials. In: ASM Vol. 8, ed. by H. Kuhn, D. Medlin (ASM Int., Mater-
Handbook, Vol. 8, ed. by H. Kuhn, D. Medlin (ASM ials Park Ohio 2000) pp. 218–228
Int., Materials Park Ohio 2000) pp. 462–476 33.48 A. Marchand, J. Duffy: An experimental study of
33.33 C.R. Siviour, S.M. Walley, W.G. Proud, J.E. Field: Are the formation process of adiabatic shear bands in
low impedance Hopkinson bars necessary for stress a structural steel, J. Mech. Phys. Solids. 36(3), 251–
equilibrium in soft materials?. In: New Experi- 283 (1988)
mental Methods in Material Dynamics and Impact, 33.49 C.K.H. Dharan, F.E. Hauser: Determination of
ed. by W.K. Nowacki, J.R. Klepaczko (Inst. Fund. stress-strain characteristics at very high strain
rates, Exper. Mech. 10, 370–376 (1970)

Part D 33
Technol. Res., Warsaw 2001) pp. 421–427
33.34 W. Chen, F. Lu, B. Zhou: A quartz-crystal- 33.50 G.L. Wulf, G.T. Richardson: The measurement of
embedded split Hopkinson pressure bar for soft dynamic stress-strain relationships at very high
materials, Exper. Mech. 40, 1–6 (2000) strain rates, J. Phys. E 7, 167–169 (1974)
33.35 D.T. Casem, W. Fourney, P. Chang: Wave separa- 33.51 D.A. Gorham: Measurement of stress-strain prop-
tion in viscoelastic pressure bars using single-point erties of strong metals at very high strain rates,
measurements of strain and velocity, Polymer Test- Inst. Phys. Conf. Ser. 47, 16–24 (1980)
ing 22, 155–164 (2003) 33.52 N.A. Safford: Materials testing up to 105 s−1 using a
33.36 J.L. Chiddister, L.E. Malvern: Compression-impact miniaturised hopkinson bar with dispersion cor-
testing of aluminum at elevated temperatures, Ex- rections, Proc. 2nd Intl. Symp. on Intense Dynamic
per. Mech. 3, 81–90 (1963) Loading and its Effects, ed. by G. Zhang, S. Huang
33.37 Z. Rosenberg, D. Dawicke, E. Strader, S.J. Bless: (Sichuan Univ. Press, Chengdu 1992) pp. 378–383
A new technique for heating specimens in Split- 33.53 D.A. Gorham, P.H. Pope, J.E. Field: An im-
Hopkinson-bar experiments using induction coil proved method for compressive stress-strain
heaters, Exper. Mech. 26, 275–278 (1986) measurements at very high strain rates, Proc. R.
33.38 A. Gilat, X. Wu: Elevated temperature testing with Soc. Lond. A 438, 153–170 (1992)
the torsional split Hopkinson bar, Exper. Mech. 34, 33.54 J. Shioiri, K. Sakino, S. Santoh: Strain rate sensi-
166–170 (1994) tivity of flow stress at very high rates of strain.
33.39 A.M. Lennon, K.T. Ramesh: A technique for meas- In: Constitutive Relation in High/Very-High Strain
uring the dynamic behavior of materials at high Rates, ed. by K. Kawata, J. Shioiri (Springer, Berlin
temperatures, Int. J. Plast. 14(12), 1279–1292 (1998) Heidelberg 1995) pp. 49–58
33.40 S. Nemat-Nasser, J.B. Isaacs: Direct Measurement 33.55 F. Kamler, P. Niessen, R.J. Pick: Measurement
of Isothermal Flow Stress of Metals at Elevated of the behaviour of high-purity copper at very
Temperature and High Strain Rates with Appli- high rates of strain, Canad. J. Phys. 73, 295–303
cation to Ta and Ta-W Alloys, Acta Mater. 45(3), (1995)
907–919 (1997) 33.56 R.W. Klopp, R.J. Clifton: Pressure-shear plate im-
33.41 D. Basak, H.W. Yoon, R. Rhorer, T.J. Burns, T. Mat- pact testing. In: ASM Handbook, Vol. 8, ed. by
sumoto: Temperature control of pulse heated H. Kuhn, D. Medlin (ASM Int., Materials Park Ohio
specimens in a Kolsky bar apparatus using 1985) pp. 230–239
956 Part D Applications

33.57 K.J. Frutschy, R.J. Clifton: High-temperature 33.71 S.C. Schmidt, R.D. Dick, J.W. Forbes, D.G. Tasker
pressure-shear plate impact experiments on OFHC (Eds.): Shock Compression of Condensed Matter -
copper, J. Mech. Phys. Sol. 46(10), 1723–1743 1991 (Elsevier, Amsterdam 1992)
(1998) 33.72 S.C. Schmidt, W.C. Tao: Shock Compression of Con-
33.58 K.T. Ramesh, N. Kelkar: Technique for the con- densed Matter - 1995 (Am. Inst. Phys., Woodbury
tinuous measurement of projectile velocities in 1996)
plate impact experiments, Rev. Sci. Instrum. 66(4), 33.73 S.C.D.P. Schmidt Dandekar, J.W. Forbes: Shock
3034–3036 (1995) Compression of Condensed Matter - 1997 (Am. Inst.
33.59 S. Yadav, D.R. Chichili, K.T. Ramesh: The mechan- Phys., Woodbury 1998)
ical response of a 6061-T6 Al/Al2 O3 metal matrix 33.74 M.D. Furnish, L.C. Chhabildas, R.S. Hixson: Shock
composite at high rates of deformation, Acta met- Compression of Condensed Matter - 1999 (Am. Inst.
all. mater. 43(12), 4453–4464 (1995) Phys., Melville 2000)
33.60 D. Jia, A.M. Lennon, K.T. Ramesh: High-strain- 33.75 L. Davison, R.A. Graham: Shock compression of
rate pressure-shear recovery: a new experimental solids, Phys. Rep. 55, 255–379 (1979)
technique, Int. J. Sol. Struc. 37(12), 1679–1699 33.76 J.R. Asay, M. Shahinpoor: High-Pressure Shock
(2000) Compression of Solids (Springer-Verlag, New York
33.61 A. Bekker, J.C. Jimenez-Victory, P. Popov, 1993)
D.C. Lagoudas: Impact induced propagation of 33.77 L. Davison, D.E. Grady, M. Shahinpoor: High-
phase transformation in a shape memory alloy rod, Pressure Shock Compression of Solids II: Dynamic
Int. J. Plast. 18, 1447–1479 (2002) Fracture and Fragmentation (Springer, New York
33.62 J.U. Cazamias, W.D. Reinhart, C.H. Konrad, 1996)
L.C. Chhabildas, S.J. Bless: Bar impact tests on alu- 33.78 L. Davison, Y. Horie, M. Shahinpoor: High-Pressure
mina (AD995). In: Shock Compression of Condensed Shock Compression of Solids IV: Response of Highly
Matter - 2001, ed. by M.D. Furnish, N.N. Thadhani, Porous Solids to Shock Compression (Springer, New
Y. Horie (Am. Inst. Phys., Melville 2002) pp. 787– York 1997)
790 33.79 L. Davison, M. Shahinpoor: High-Pressure Shock
33.63 L.C. Chhabildas, M.D. Furnish, W.D. Reinhart, Compression of Solids III (Springer, New York 1998)
D.E. Grady: Impact of AD995 alumina rods. In: 33.80 J.A. Zukas (Ed.): High Velocity Impact Dynamics
Shock Compression of Condensed Matter - 1997, ed. (Wiley, New York 1990)
Part D 33

by S.C. Schmidt, D.P. Dandekar, J.W. Forbes (Am. 33.81 M.F. Gogulya, A.Y. Dolgoborodov, M.A. Brazhnikov:
Inst. Phys., Woodbury 1998) pp. 505–508 Investigation of shock and detonation waves by
33.64 F.G. Díaz-Rubio, J. Rodríguez Pérez, V. Sánchez optical pyrometry, Int. J. Imp. Eng. 23(1), 283–293
Gálvez: The spalling of long bars as a reliable (1999)
method of measuring the dynamic tensile strength 33.82 G.I. Pangilinan, Y.M. Gupta: Time-resolved raman
of ceramics, Int. J. Impact Eng. 27, 161–177 (2002) measurements in nitromethane shocked to 140
33.65 K.G. Holland, L.C. Chhabildas, W.D. Reinhart, Kbar, J. Phys. Chem. 98(13), 4522–4529 (1994)
M.D. Furnish: Experiments on CERCOM SiC rods un- 33.83 W.D. Reinhart, L.C. Chhabildas, D.E. Carroll,
der impact. In: Shock Compression of Condensed T.K. Bergstresser, T.F. Thornhill, N.A. Winfree:
Matter - 1999, ed. by M.D. Furnish, L.C. Chhabil- Equation of state measurements of materials using
das, R.S. Hixson (Am. Inst. Phys., Melville 2000) a three-stage gun to impact velocities of 11 km/s,
pp. 585–588 Int. J. Imp. Eng. 26(1-10), 625–637 (2001)
33.66 R. Russell, S.J. Bless, T. Beno: Impact induced fail- 33.84 Y.M. Gupta: Use of piezoresistance gauges to
ure zones in Homalite bars:. In: Shock Compression quantify the stress state in a shocked solid. In:
of Condensed Matter - 2001, ed. by M.D. Furnish, Experimental Techniques in the Dynamics of De-
N.N. Thadhani, Y. Horie (Am. Inst. Phys., Melville formable Solids, ed. by K.T. Ramesh (ASME, New
2002) pp. 811–814 York 1993) pp. 89–101
33.67 D.S. Drumheller: Introduction to Wave Propagation 33.85 R. Feng, G.F. Raiser, Y.M. Gupta: Material strength
in Nonlinear Fluids and Solids (Cambridge Univ. and inelastic deformation of silicon carbide under
Press, Cambridge 1998) shock wave compression, J. Appl. Phys. 83(1), 79–
33.68 M.A. Meyers: Dynamic Behavior of Materials (Wiley, 86 (1998)
New York 1994) 33.86 Z. Rosenberg, D. Yaziv, Y. Partom: Calibration of
33.69 G.T. Gray III: Shock wave testing of ductile mater- foil-like manganin gauges in planar shock-wave
ials. In: ASM Handbook, Vol. 8, ed. by H. Kuhn, experiments, J. Appl. Phys. 51(6), 3702–3705 (1980)
D. Medlin (ASM Int., Materials Park Ohio 2000) 33.87 Y.M. Gupta: Analysis of manganin and ytterbium
pp. 462–476 gauge data under shock loading, J. Appl. Phys.
33.70 S.C. Schmidt, J.N. Johnson, L.W. Davidson: Shock 54(9), 6094–6098 (1983)
Compression of Condensed Matter - 1989 (North 33.88 R.A. Graham, F.W. Neilson, W.B. Benedick: Piezo-
Holland, Amsterdam 1990) electric current from shock-loaded quartz -
 High Rates and Impact Experiments References 957

a submicrosecond stress gauge, J. Appl. Phys. 36(5), 33.103 P.E. Magnusen, E.M. Dubensky, D.A. Koss: The ef-
1775–00 (1965) fect of void arrays on void linking during ductile
33.89 T. Obara, N.K. Bourne, Y. Mebar: The construc- fracture, Acta Metallurgica 36(5), 1503–1509 (1988)
tion and calibration of an inexpensive PVDF stress 33.104 E.M. Dubensky, D.A. Koss: Void pore distributions
gauge for fast pressure measurements, Meas. Sci. and ductile fracture, Metallurg. Mat. Trans. A 18(9),
Technol. 6(4), 345–348 (1995) 1887–1895 (1987)
33.90 L.M. Barker: The accuracy of VISAR instrumentation. 33.105 R.J. Bourcier, Koss D.A., R.E. Smelser, O. Richmond:
In: Shock Compression of Condensed Matter - 1997, The influence of porosity on the deformation and
ed. by S.C. Schmidt, D.P. Dandekar, J.W. Forbes fracture of alloys, Acta Metallurgica 34(12), 2443–
(Am. Inst. Phys., Woodbury, 1998) pp. 833–836 2453 (1986)
33.91 L.M. Barker: The development of the VISAR and 33.106 T. Pardoen, I. Doghri, F. Delannay: Experimental
its use in shock compression science. In: Shock and numerical comparison of void growth models
Compression of Condensed Matter - 1999, ed. by and void coalescence criteria for the prediction of
M.D. Furnish, L.C. Chhabildas, R.S. Hixson (Am. ductile fracture in copper bars, Acta Mater. 46(2),
Inst. Phys., Melville 2000) pp. 11–17 541–00 (1998)
33.92 L.M. Barker, R.E. Hollenbach: Laser interferome- 33.107 H.J. Prentice, W.G. Proud: 3 Dimensional dy-
ter for measuring high velocities of any reflecting namic deformation measurments using stereo-
surface, J. Appl. Phys. 43, 4669–4675 (1972) scopic imaging and digital speckle photography.
33.93 J. Edwards, K.T. Lorenz, B.A. Remington, S. Pol- In: Shock Compression of Condensed Matter, Vol. 2,
laine, J. Colvin, D. Braun, B.F. Lasinski, D. Reisman, ed. by M.D. Furnish, M. Elert, T.P. Russell, C.T. White
J.M. McNaney, J.A. Greenough, R. Wallace, H. Louis, (AIP, Baltimore 2005) p. 1275
D. Kalantar: Laser-driven plasma loader for shock- 33.108 X.Y. Wu, K.T. Ramesh, T.W. Wright: The effects of
less compression and acceleration of samples in thermal softening and heat conduction on the dy-
the solid state, Phys. Rev. Lett. 92(6) (2004) namic growth of voids, Int. J. Sol. Struct. 40(13),
33.94 N.K. Bourne, Z. Rosenberg, J.E. Field: High-speed 4461–4478 (2003)
photography of compressive failure waves in 33.109 G. Ravichandran, R.J. Clifton: Dynamic fracture un-
glasses, J. Appl. Phys. 78, 3736–3739 (1995) der plane-wave loading, Int. J. Fract. 40(3), 157–201
33.95 B. Paliwal, K.T. Ramesh, J.W. McCauley: Direct ob- (1989)
servation of the dynamic compressive failure of a 33.110 V. Prakash, L.B. Freund, R.J. Clifton: Stress wave ra-

Part D 33
transparent polycrystalline ceramic (AlON), J. Am. diation from a crack tip during dynamic initiation,
Ceramic Soc. 89(6), 2128–2133 (2006) J. Appl. Mech.-Trans. ASME 59(2), 356–365 (1992)
33.96 K. Ravi-Chandar: Dynamic fracture (Wiley, New 33.111 M. Zhou, R.J. Clifton: Dynamic ductile rupture un-
York 2005) der conditions of plane strain, Int. J. Imp. Eng.
33.97 E. El-Magd, M. Brodmann: Influence of precipitates 19(3), 189–206 (1997)
on ductile fracture of aluminium alloy AA7075 at 33.112 Z. Zhang, R.J. Clifton: Shear band propagation from
high strain rates, Mat. Sci. Eng. A. Struct. Mat. Prop. a crack tip, J. Mech. Phys Sol. 51(11-12), 1903–1922
Microstruct. Proces. 307(1-2), 143–150 (2001) (2003)
33.98 D.M. Goto, D.A. Koss, V. Jablokov: The influence of 33.113 R. Godse, G. Ravichandran, R.J. Clifton: Mi-
tensile stress states on the failure of HY-100 steel, cromechanisms of dynamic crack-propagation in
Metallurg. Mat. Trans. A 30(9), 2835–2842 (1999) an aisi-4340-steel, Mat. Sci. Eng. A 112, 79–88
33.99 J. Kang, M. Jain, D.S. Wilkinson, J.D. Embury: Mi- (1989)
croscopic strain mapping using scanning electron 33.114 J.N. Johnson: Dynamic fracture and spallation in
microscopy topography image correlation at large ductile solids, J. Appl. Phys. 52(4), 2812–2825 (1981)
strain, J. Strain Anal. Eng. Design 40(5), 559–570 33.115 D.E. Grady: The spall strength of condensed matter,
(2005) J. Mech. Phys Sol. 36(3), 353–00 (1988)
33.100 W. Tong, H. Tao, X.Q. Jiang, N.A. Zhang, M.P. Marya, 33.116 G.I. Kanel, S.V. Razorenov, K. Baumung, J. Singer:
L.G. Hector, X.H.Q. Gayden: Deformation and Dynamic yield and tensile strength of aluminum
fracture of miniature tensile bars with resistance- single crystals at temperatures up to the melting
spot-weld microstructures, Metallurg. Mat. Trans. point, J. Appl. Phys. 90(1), 136–143 (2001)
A 36A(10), 2651–2669 (2005) 33.117 T. Antoun, L. Seaman, D.R. Curran, G.I. Kanel,
33.101 B. Wattrisse, A. Chrysochoos, J.M. Muracciole, S.V. Razorenov, A.V. Utkin: Spall Fracture (Springer,
M. Nemoz-Gaillard: Kinematic manifestations of New York 2004)
localisation phenomena in steels by digital image 33.118 Y.L. Bai, B. Dodd: Adiabatic Shear Localization:
correlation, Europ. J. Mech. A 20(2), 189–211 (2001) Occurrence, Theories and Applications (Pergamon,
33.102 B. Wattrisse, A. Chrysochoos, J.M. Muracciole, Oxford 1992)
M. Nemoz-Gaillard: Analysis of strain localization 33.119 T.W. Wright: The mathematical theory of adiabatic
during tensile tests by digital image correlation, shear bands (Cambridge Univ. Press, Cambridge
Exp. Mech. 41(1), 29–39 (2001) 2002)
958 Part D Applications

33.120 J.H. Giovanola: Observation of adiabatic shear 33.136 S. Dujardin, G. Gazeaud, A. Lichtenberger: Dynamic
banding in simple torsion. In: Impact Loading and behavior of copper studied using the expanding
Dynamic Behaviour of Materials, ed. by C.Y. Chiem, ring test, J. Physique 49(C-3), 55–62 (1988)
H.D. Kunze, L.W. Meyer (DGM Informationsge- 33.137 W.H. Gourdin: Analysis and assessment of elec-
sellschaft, Oberursel 1988) pp. 705–710 tromagnetic ring expansion as a high-strain-rate
33.121 M.G. daSilva, K.T. Ramesh: The rate-dependent test, J. Appl. Phys. 65(2), 411–422 (1989)
deformation and localization of fully dense and 33.138 W.H. Gourdin: Constitutive properties of copper
porous Ti-6Al-4V, Mat. Sci. Eng. A 232(1-2), 11–22 and tantalum at high-rates of tensile strain - ex-
(1997) panding ring results, Inst. Phys. Conf. Ser. 102,
33.122 K.T. Ramesh: On the localization of shearing defor- 221–226 (1989)
mations in a tungsten heavy alloy, Mech. Materials 33.139 W.H. Gourdin, S.L. Weinland, R.M. Boling: De-
17, 165–173 (1994) velopment of the electromagnetically launched
33.123 A. Molinari, R.J. Clifton: Analytical characterization expanding ring as a high-strain-rate test tech-
of shear localization in thermoviscoplastic mater- nique, Rev. Sci. Instr. 60(3), 427–432 (1989)
ials, J. Appl. Mech.-Trans. ASME 54(4), 806–812 33.140 H. Zhang, K. Ravi-Chandar: On the dynamics of
(1987) necking and fragmentation - I. Real-time and
33.124 B. Deltort: Experimental and numerical aspects of post-mortem observations in Al6061-O, Int. J.
adiabatic shear in a 4340 steel, J. Phys. IV France Fracture 142, 183 (2006)
Colloq. C8 (DYMAT 94) 4, 447–452 (1994) 33.141 F.H. Zhou, J.F. Molinari, K.T. Ramesh: Analysis of
33.125 D.R. Chichili, K.T. Ramesh, K.J. Hemker: Adiabatic the brittle fragmentation of an expanding ring,
shear localization in alpha-titanium: experiments, Comp. Mat. Sci. 37(1-2), 74–85 (2006)
modeling and microstructural evolution, J. Mech. 33.142 P.R. Guduru, L.B. Freund: The dynamics of multi-
Phys Sol. 52(7), 1889–1909 (2004) ple neck formation and fragmentation in high rate
33.126 D.R. Chichili, K.T. Ramesh: Recovery experiments extension of ductile materials, Int. J. Sol. Struct.
for adiabatic shear localization: A novel experi- 39(21-22), 5615–5632 (2002)
mental technique, J. Appl. Mech.-Transact. ASME 33.143 M.J. Forrestal, B.W. Duggin, R.I. Butler: Explosive
66(1), 10–20 (1999) loading technique for the uniform expansion of
33.127 Q. Xue, L.T. Shen, T.L. Bai: Elimination of loading 304 stainless steel cylinders at high strain rates, J.
reverberation in the split Hopkinson torsional bar, Appl. Mech.-Transact. ASME 47(1), 17–20 (1980)
Part D 33

Rev. Sci. Instr. 66(9), 5298–5304 (1995) 33.144 J.F. Kalthoff: On the measurement of dynamic frac-
33.128 M.A. Meyers, G. Subhash, B.K. Kad, L. Prasad: Evo- ture toughnesses - a review of recent work, Int. J.
lution of microstructure and shear-band formation Fract. 27(3-4), 277–298 (1985)
in alpha-hcp titanium, Mech. Mat. 17(2-3), 175–193 33.145 R.A.W. Mines: Characterization and measurement
(1994) of the mode 1-dynamic initiation of cracks in met-
33.129 Z.Q. Duan, S.X. Li, D.W. Huang: Microstructures and als at intermediate strain rates - a review, Int. J.
adiabatic shear bands formed by ballistic impact Imp. Eng. 9(4), 441–454 (1990)
in steels and tungsten alloy, Fat. Fract. Eng. Mat. 33.146 A.J. Rosakis: Application of coherent gradient sens-
Struct. 26(12), 1119–1126 (2003) ing (cgs) to the investigation of dynamic fracture
33.130 Q. Xue, M.A. Meyers, V.F. Nesterenko: Self- problems, Optics Lasers Eng. 19(1-3), 3–41 (1993)
organization of shear bands in titanium and 33.147 K. Ravi-Chandar: Dynamic fracture of nominally
Ti-6Al-4V alloy, Acta Mater. 50(3), 575–596 (2002) brittle materials, Int. J. Fract. 90(1-2), 83–102 (1998)
33.131 D. Rittel, S. Lee, G. Ravichandran: A shear- 33.148 A. Shukla: High-speed fracture studies on bimate-
compression specimen for large strain testing, Exp. rial interfaces using photoelasticity - a review, J.
Mech. 42(1), 58–64 (2002) Strain Anal. Eng. Design 36(2), 119–142 (2001)
33.132 M. Zhou, A.J. Rosakis, G. Ravichandran: Dynam- 33.149 J.F. Kalthoff: Modes of dynamic shear failure in
ically propagating shear bands in impact-loaded solids, Int. J. Fract. 101(1-2), 1–31 (2000)
prenotched plates. 1. Experimental investigations 33.150 D.D. Anderson, A. Rosakis: Comparison of three real
of temperature signatures and propagation speed, time techniques for the measurement of dynamic
J. Mech. Phys Sol. 44(5), 981–1006 (1996) fracture initiation toughness in metals, Eng. Fract.
33.133 P.R. Guduru, A.J. Rosakis, G. Ravichandran: Dy- Mech. 72(4), 535–555 (2005)
namic shear bands: an investigation using high 33.151 V. Parameswaran, A. Shukla: Dynamic fracture of
speed optical and infrared diagnostics, Mech. Mat. a functionally gradient material having discrete
33(6), 371–402 (2001) property variation, J. Mat. Sci. 33(10), 3303–331
33.134 D.E. Grady, D.A. Benson: Fragmentation of metal (1998)
rings by electromagnetic loading, Exp. Mech. 23(4), 33.152 A.J. Rosakis, A.T. Zehnder, R. Narasimhan: Caustics
393–400 (1983) by reflection and their application to elastic-
33.135 W.H. Gourdin: The expanding ring as a high-strain plastic and dynamic fracture-mechanics, Optic.
rate test, J. Met. 39(10), A65–A65 (1987) Eng. 27(7), 596–610 (1988)
 High Rates and Impact Experiments References 959

33.153 K. Arakawa, R.H. Drinnon, M. Kosai, A.S. Kobayashi: on crack-tip field-measurements using coherent
Dynamic fracture-analysis by moire interferome- gradient sensing, Exp. Mech. 31(3), 243–251 (1991)
try, Exp. Mech. 31(4), 306–309 (1991) 33.155 T.A. Siewert, M.P. Monahan (Eds.): Pendulum
33.154 H.V. Tippur, A.J. Rosakis: Quasi-static and dynamic Impact Testing: A Century of Progress (ASTM Inter-
crack-growth along bimaterial interfaces - a note national, Conshohocken 2000)

Part D 33