Measurement Science and Technology

PAPER You may also like

- Sensor calibration of polymeric Hopkinson
Design of a split Hopkinson pressure bar with bars for dynamic testing of soft materials
Milena Martarelli, Edoardo Mancini,
partial lateral confinement Barbara Lonzi et al.

- Size Effect in the Split Hopkinson Pressure

Bar Experiment
To cite this article: Andrew D Barr et al 2016 Meas. Sci. Technol. 27 125903 Hailiang Nie, Weifeng Ma, Junjie Ren et al.

- A measurement of wave propagation in

the split Hopkinson pressure bar
C R Siviour
View the article online for updates and enhancements.

This content was downloaded from IP address 193.140.109.205 on 10/01/2023 at 12:11

 Measurement Science and Technology

Meas. Sci. Technol. 27 (2016) 125903 (8pp) doi:10.1088/0957-0233/27/12/125903

Design of a split Hopkinson pressure

bar with partial lateral confinement
Andrew D Barr1, Sam D Clarke1, Sam E Rigby1, Andrew Tyas1,2
and James A Warren1,2
1
  Department of Civil and Structural Engineering, The University of Sheffield,
Mappin Street, Sheffield, S1 3JD, UK
2
  Blastech Ltd., The BioIncubator, 40 Leavygreave Road, Sheffield, S3 7RD, UK

E-mail: [email protected]

Received 20 July 2016, revised 13 September 2016

Accepted for publication 5 October 2016
Published 25 October 2016

Abstract
This paper presents the design of a modified split Hopkinson pressure bar (SHPB) where
partial lateral confinement of the specimen is provided by the inertia of a fluid annulus
contained in a long steel reservoir. In contrast to unconfined testing, or a constant cell pressure
applied before axial loading, lateral restraint is permitted to develop throughout the axial
loading: this enables the high-strain-rate shear behaviour of soils to be characterised under
conditions which are more representative of buried explosive events. A pressure transducer
located in the wall of the reservoir allows lateral stresses to be quantified, and a dispersion-
correction technique is used to provide accurate measurements of axial stress and strain.
Preliminary numerical modelling is utilised to inform the experimental design, and the
capability of the apparatus is demonstrated with specimen results for a dry quartz sand.

Keywords: experimental design, partial confinement, split Hopkinson pressure bar

(Some figures may appear in colour only in the online journal)

1. Introduction varying moisture contents [8, 9], initial densities [10, 11] and
particle size distributions [12], but cannot be used to describe
Widespread use of improvised explosive devices in current the yield surface, as shear failure cannot occur.
conflicts has driven a need to understand the role of soils in Several authors have developed methods which allow the
buried explosive events. The design of effective protective lateral confinement of a SHPB specimen to be modified to
solutions requires accurate predictions of the blast loading provide a triaxial stress state, and a number of these are based
produced in these events, which has been shown in large-scale on an adaptation of the rigid confinement condition. Pierce
field tests to be greatly influenced by the properties of the sur- and Charlie [13] investigated the wave speed of partially-
rounding soil [1–5]. The ability to define yield surfaces at high saturated sand at confining stresses of 0 kPa and 310 kPa
pressures and strain rates is a key component in developing using a steel tube lined with a membrane. While the steel tube
a robust constitutive model of soil behaviour under extreme prevented the development of lateral strains, water pressure
loading, and so a reliable means of testing soils triaxially applied between the tube and membrane provided an addi-
under these conditions is required. tional confining stress, which was also applied along the pres­
The split Hopkinson pressure bar (SHPB) is commonly sure bars through use of a piston assembly on the transmitter
used to investigate the response of materials at strain rates of bar. Bailly et al [14] made use of brass confining rings which
102 s−1 to 104 s−1, and tests on soils using the SHPB are typi- approximated elastic—perfectly-plastic behaviour at high
cally carried out by confining a soil specimen in a rigid tube or strain rates. Tests specimens deformed approximately in uni-
ring, which restricts lateral deformation. These uniaxial strain axial strain until the radial stress reached the yield stress in
tests are useful for characterising the compaction response the ring, after which the specimen was permitted to deform
of soil at varying strain rates [6–8] and comparing soils with laterally at a quasi-constant confining stress.

0957-0233/16/125903+8$33.00 1 © 2016 IOP Publishing Ltd  Printed in the UK

Meas. Sci. Technol. 27 (2016) 125903 A D Barr et al

Figure 1.  Schematic of the partially confined split Hopkinson pressure bar apparatus with a soil specimen: (a) bar and reservoir
arrangement, (b) water reservoir section with axial/radial axis convention.

Table 1.  Material model and equation of state parameters for water (SI units) [19].

mat_null
ρ0
1000
eos_linear_polynomial
C0 C1 C2 C3 C4 C5 C6 E0
0.0 2.190e9 9.224e9 8.767e9 0.4934 1.3937 0.0 205.36e3

Other authors have adapted the use of the triaxial cell to lateral confinement, which is also influenced by the inertia
high-strain-rate testing. Christensen et al [15] performed tri- of the surrounding soil. Quantification of this inertial effect
axial tests on sandstone to confining stresses of 207 MPa using in a soil specimen with a developing stress state requires a
a long pressure vessel. The pressure vessel contained the spec- departure from the CTC techniques which are currently avail-
imen and pressure bars, with an opening at one end to allow able, and so a new apparatus has been developed to allow a
loading of the incident bar, which was secured with a collar. confining stress to develop passively during high-strain-rate
Similar experiments were carried out on basalt by Lindholm axial loading. This partial lateral confinement falls between
et al [16], who used a shorter pressure vessel around the rock unconfined testing, where the specimen deforms under uni-
specimen in conjunction with a hydraulic actuator on the end axial stress, and CTC testing, where a constant cell pressure is
of the transmitter bar to apply hydrostatic stresses of up to maintained throughout.
690 MPa. Frew et  al [17] adapted the triaxial SHPB further
to use pressure vessels around both the specimen and trans-
mitter bar end. This apparatus was used by Martin et al [18] 2.  Experimental setup
to characterise the shear response of sand at confining stresses
between 25 MPa and 150 MPa, where a method was also The modified SHPB consists of a typical pressure bar arrange-
developed to record the axial and radial deformation of the ment with the addition of a 600 mm long steel water reservoir,
specimen during hydrostatic loading. which is mounted in linear bearings and centred around the
The methods employed by Christensen et al [15], Lindholm specimen as shown in figure  1(a). With the pressure bars in
et al [16] and Martin et al [18] have been successfully used position an annular void exists along the length of the reser-
to perform dynamic conventional triaxial compression (CTC) voir, which is filled with water at atmospheric pressure, as in
tests, where hydrostatic loading is followed by a high-strain- figure 1(b). Saturation of a soil specimen is prevented through
rate deviatoric phase. The current work seeks to understand use of a 0.4 mm thick latex membrane, which is sealed against
the shear behaviour of soils in explosive events, where a the pressure bars using o-rings. When the specimen is loaded
soil may deform uniaxially before developing significant axially, radial deformation is resisted by the inertia of the

2
Meas. Sci. Technol. 27 (2016) 125903 A D Barr et al

The water was modelled using the linear polynomial equa-

tion of state:
P = C0 + C1µ + C2µ2 + C3µ3 + (C4 + C5µ + C6µ2 )E
(1)
where C0, C1, C2, C3, C4, C5 and C6 are constants, µ = ρ /ρ0 − 1,
ρ and ρ0 are the current and initial densities of the fluid, and
E is the specific internal energy of the fluid. The null mat­
erial properties (of which only density is required) and equa-
tion of state parameters used for water in this study are shown
in table  1. The initial internal energy of the water, E0, was
specified as 205.36 kPa to pressurise the water to atmospheric
conditions (101 kPa atmospheric pressure).
The model was discretised using axi-symmetric Lagrangian
shell elements with 0.5 mm side lengths, with the mesh size
Figure 2.  Effect of reservoir length on pressure at wall of reservoir. informed by the results of a mesh sensitivity analysis which
is not included here for brevity. Automatic 2D surface-to-sur-
water annulus: this radial resistance develops in response to
face contact was specified between all parts. The reservoir was
the behaviour of the specimen, and is measured by a pressure
varied in length between 100 mm and 1000 mm and was varied
transducer positioned in the wall of the reservoir.
in internal radius between 17.5 mm and 37.5 mm. For simplicity,
To ensure that the recorded pressure is representative of the
the steel reservoir was modelled as a rigid boundary, assuming
radial stress on the surface of the specimen, numerical model-
that the fluid pressures generated will not be large enough to
ling of the reservoir was used to select the length and internal
cause significant radial strains in the reservoir. The o-ring seals
diameter of the reservoir, as described in section 3. The res-
were also modelled as a rigid boundary, as edge reflections will
ervoir length was chosen so that the time for a stress wave
not affect the pressure transducer measurements until after the
initiated at the specimen surface to travel to and from the end
axial loading is completed given a sufficiently long reservoir.
of the reservoir exceeds the loading duration in the specimen,
ensuring that inward-travelling waves from the boundary do
not interfere with the pressure measurements. This also allows 3.2.  Effect of reservoir length
simplification of the seals between the reservoir and the pres­
sure bars, which are only required to retain the water at atmos- To test the effect of the length of reservoir, three analyses were
pheric pressure. run with a reservoir radius of 17.5 mm and lengths of 100 mm,
500 mm and 1000 mm, centred around the sample. Results of
these analyses are shown in figure  2. For the 100 mm long
3.  Preliminary modelling reservoir, inward-travelling waves from the remote ends begin
to contaminate the results at approximately 370 μs. The traces
3.1.  Model setup for the 500 mm and 1000 mm reservoirs are not affected by
Numerical analyses of the arrangement shown in figure  1 these reflected waves, and are almost identical. Assuming a
were conducted using the explicit FE code LS-DYNA [20] in wavespeed of 1484 m s−1 in the water, a reservoir length of
order to test the efficacy of measuring the dynamic fluid pres­ 500 mm gives the potential to record a loading event with a
sure as an indicator of the radial confining stress acting on the duration of 337 μs before any inward-travelling waves from
sample, and to investigate the effect of reservoir length and the remote boundary of the reservoir can affect the recordings.
inner radius (i.e. thickness of fluid annulus). As the purpose of For the current pressure bar setup the reservoir is therefore
the numerical modelling is to prove the concept of the appa- required to be at least 500 mm in length.
ratus and to explore the required geometry of the reservoir,
it is sufficient to simulate scenarios with indicative sample 3.3.  Effect of reservoir inner radius
material properties in order to inform the experimental design
and allow more sophisticated material model parameters to Following on from this, the reservoir length was set at 500 mm
eventually be gathered. and analyses were run with inner radii of 17.5 mm, 22.5 mm
The steel pressure bars were modelled as linear elastic with and 37.5 mm, giving distances from the sample edge to the
a density, Young’s modulus and Poisson’s ratio of ρ = 7850 wall of the reservoir (fluid annulus thickness) of 5 mm, 10 mm
kg m−3, E  =  200 GPa and ν = 0.3 respectively. The geometry and 25 mm respectively. Results of these analyses are shown
of the pressure bars was as shown in figure 1(a). The striker in figure  3. For the 17.5 mm inner radius reservoir, there is
bar was given an impact velocity of 5 m s−1 and a sample a near-perfect match (r2  =  0.93) between the pressure in the
length of 5 mm was chosen for all analyses. The sample was elements at the sample edge and the at the reservoir wall. The
also modelled as a linear elastic material, with ‘typical’ rubber agreement is less precise, but still acceptable (r2  =  0.85),
material properties, i.e. ρ = 1000 kg m−3, E = 100 MPa and for the 22.5 mm case, where the increased fluid thickness
ν = 0.5, in order to generate large radial strains within the reduced the confining stress generated in the specimen. For
sample and induce a large fluid pressure within the reservoir. the 37.5 mm case, the time taken for the pressure wave to

3
Meas. Sci. Technol. 27 (2016) 125903 A D Barr et al

Figure 4.  Water reservoir methodology: (a) specimen installation,

Figure 3.  Effect of reservoir width on difference between water (b) specimen measurement, (c) water reservoir alignment, (d) water
pressure at sample face and water pressure at wall of reservoir for reservoir seals and instrumentation.
inner radii of (a) 17.5 mm, (b) 22.5 mm and (c) 37.5 mm.
accurate, the inner radius of the reservoir will be no larger
propagate from the sample to the edge of the reservoir is no than 22.5 mm.
longer negligible compared to the loading duration, leading
to a delay in the development of the pressure at the reservoir
3.4.  Reservoir dimensions
wall (figure 3(c)). The pressure recorded at the wall of the res-
ervoir cannot therefore be said to be an accurate representa- Following these analyses, a 600 mm length, 20 mm inner
tion of the pressure directly adjacent to the sample in this case radius and 33 mm outer radius were adopted for the reservoir.
(r2  =  0.37). To ensure that the pressure transducer reading is As the reservoir is longer than 500 mm and has an inner radius

4
Meas. Sci. Technol. 27 (2016) 125903 A D Barr et al

Figure 5.  Lathe-cut benchmarks used to measure the initial

specimen length.

Figure 7.  Relationship of phase velocity to frequency for the first

mode of propagation of a longitudinal wave.

Figure 6.  Variation of strain rate during a partially-confined

experiment on dry quartz sand.

less than 22.5 mm, the fluid pressure recorded at the pressure

gauge embedded in the inner wall of the reservoir will be an Figure 8.  Incident bar stress calculated using one-dimensional
accurate measure of the sample confining stress. A sufficiently wave theory and a dispersion-corrected method.
large wall thickness (13 mm) was selected to ensure the reser- gauge, enabling the initial density the soil to be calculated.
voir behaves as an effectively rigid boundary. The precise location of the end of a pressure bar is difficult
to discern through a microscope, and so shallow grooves
4.  Sample methodology were cut into the circumference of each bar as benchmarks,
as shown in figure 5. The difference in length between the
To demonstrate the capability of the partially-confined SHPB, benchmarks before and after the specimen was installed
and to verify that the selected design leads to accurate mea- was used to infer the initial length of the specimen, and so
surements of fluid pressure, an experiment was carried out on the benchmarks were cut using a chisel-edged bit on a lathe
a dry soil specimen. The soil was a uniform fine and medium so that they provided well-defined edges.
sand (EN ISO 14688–1:2002), with a mean particle size of 5. The water reservoir was translated into position, and the
250μm and a coefficient of uniformity, Cu of 2.2. Installation pressure port was aligned with the centre of specimen
of the specimen into the water reservoir was as follows, where using an additional set of lathed grooves.
numbers refer to the annotations in figure 4. 6. The incident bar bearing was returned to its test position
and bolted down.
1. The o-ring seals for the water reservoir were fitted to the
7. The o-rings were inserted into the water reservoir to seal
incident and transmitter bars.
its ends.
2. The water reservoir was installed by temporarily
8. The water reservoir was filled with water using the filling
unbolting and translating the bearing on the incident bar.
port and sealed by fitting the pressure transducer and
3. A sand specimen was installed between the incident and
filling port bolt. The transducer used in this experiment
transmitter bars. In this experiment the specimen was
was a Kulite HKM—375–2500, calibrated by the manu-
held in position, and isolated from the fluid annulus, by
facturer to perform linearly to a pressure of 25 MPa.
a thin polythene membrane: future experiments will use
a bespoke latex membrane. The specimen was prepared The method then proceeded as in a conventional SHPB
with a nominal length of 5 mm and a diameter of 25 mm. experiment. Loading was provided by firing a steel striker bar
4. Once installed, the specimen length was measured accu- from a gas gun to impact the incident bar (vstriker ≈ 20 m s−1),
rately using a travelling microscope fitted with a digital dial and signals from the pressure bar strain gauges and pressure

5
Meas. Sci. Technol. 27 (2016) 125903 A D Barr et al

Figure 9.  Partially-confined SHPB experiment on dry quartz sand: (a) axial and radial stresses, (b) axial stress–strain response.

transducer were recorded using a TiePie Handyscope four- 3. A correction is applied to the amplitude of each frequency
channel digital oscilloscope, using 14-bit resolution and a component using the factors M1 and M2, which account
sampling rate of 1.562 MHz. for the variation of strain and Young’s modulus across the
The inertial forces which dominate the early stages of bar cross section, respectively. These are derived from
loading are a key topic of interest in the current study, and Davies’ analysis of the radial effects in a cylindrical pres­
so no attempt was made to modify the incident stress wave sure bar [24].
through the use of pulse shapers [18]. This necessarily means 4. The signal is transformed back into the time domain
that the strain rate varies during the experiment, as shown in using the inverse FFT.
figure 6, where the strain rate increases from zero to 3400 s−1
This dispersion correction is particularly important in
over approximately 50 μs.
inferring the stress transmitted into the specimen from the
incident bar, as it is calculated from the sum of the incident
5.  Signal processing and reflected waves, which both contain significant high-
frequency components. Using one-dimensional wave theory
In processing the signals from SHPB experiments it is often the incident and reflected stress waves measured at the inci-
assumed that longitudinal stress waves in the pressure bars dent bar strain gauge are assumed to maintain their shape as
propagate one-dimensionally at a common velocity c0, and they are translated along the time axis, while in the corrected
so measurements taken at the strain gauges are often simply method the dispersion associated with 1000 mm of travel in
translated to the end of the bar using a suitable time delay [21]. the bar is added to the incident wave and removed from the
In reality, stress waves propagate at a specific phase velocity, reflected wave. This is illustrated in figure 8, where the dis-
cP, which is a function of frequency and the bar’s diameter, persion-corrected method reduces the amplitude of the stress
one-dimensional wave speed and Poisson’s ratio, as shown in wave and removes a large initial oscillation in stress, which
figure  7 [22]. Phase velocity decreases as the frequency of a could have otherwise led to erroneous conclusions on the
wave increases, leading to dispersion of a signal as it propagates behaviour of the specimen.
down the bar. Dispersion of the stress pulse is accompanied by a
frequency-dependent variation in stress and strain across the bar
6.  Sample results
cross-section, so that a signal recorded on the surface of the bar
at some distance from the specimen will not accurately describe
The recorded axial (σ1) and radial (σ3) stresses are shown in
the stresses the specimen was subjected to, and hence cannot be
figure 9(a), where the axial stress is the mean of the stresses
used to accurately determine the specimen response.
acting on each specimen face. The transit time from the specimen
To ensure that the inferred measurements of axial stress
to the pressure transducer through the water annulus (5.1 μs,
and strain accurately represent the specimen behaviour, the
assuming a wavespeed in water of 1482 m s−1) was taken into
pressure bar signals were processed using an implementation
account when analysing the radial stress in the specimen. The
of the dispersion-correction method described by Tyas and
pressure transducer provided excellent measurements of radial
Pope [23]. In this method:
stress, which correspond well with the features in the recorded
1. The time-domain strain signal is converted into the fre- axial stress. Of particular note is the first 40 μs of the axial
quency domain using the fast Fourier transform (FFT). stress pulse, where the specimen deforms without any mea-
2. A correction is applied to the phase angle of each fre- sured radial stress on the surface of the specimen, resulting in
quency component to account for the dispersion over the a peak in the stress difference (σ1–σ3) in figure 9(b). A similar
distance between the strain gauge and the bar end, arising lag in measurements of radial stress has been observed in con-
from the relationship shown in figure 7. fined SHPB experiments [8], and was attributed to the effects

6
Meas. Sci. Technol. 27 (2016) 125903 A D Barr et al

of radial inertia in the sand specimen. This behaviour and the quartz sand, and excellent measurements of axial and radial
effects of inertia will be investigated in more detail in future stress were recorded.
test series using the water reservoir. This apparatus will enable the shear response of soils
in buried explosive events to be characterised under more
indicative stress states than can be achieved using existing
7. Discussion
techniques, and so the resulting increase in the accuracy of
It was noted in section 3.3 that the radial confinement gener- numerical models of these events will be of great benefit to the
ated in the specimen is a function of the internal radius of development of systems to protect against buried explosive
the reservoir: as the thickness of the fluid annulus increases, threats.
the radial confinement decreases. The apparatus can therefore
be used to investigate the shear behaviour of soils at a range
Acknowledgments
of confining stresses by using a number of reservoirs with
varying internal diameters. This research was supported by the Engineering and Physical
The preliminary modelling indicated that the maximum Sciences Research Council, grant EP/L011441/1.
internal radius for reliable confining stress measurements
was 22.5 mm. The minimum internal radius is limited by the
requirement to move the reservoir into position without dis- References
turbing the specimen, and is likely to be on the order of 15 mm.
The magnitude of the confining stress using these geometries [1] Clarke S D, Fay S D, Warren J A, Tyas A, Rigby S E,
will depend on the behaviour of a particular specimen under a Reay J J, Livesy R and Elgy I 2015 Geotechnical causes for
specific axial loading. variations in output measured from shallow buried charges
Int. J. Impact Eng. 86 274–83
The ability to analyse the partially-confined behaviour of [2] Clarke S D, Fay S D, Warren J A, Tyas A, Rigby S E,
soils in this way bridges the gap between nominally uncon- Reay J J, Livesy R and Elgy I 2016 Predicting the role of
fined experiments (using a thin friable membrane) and fully- geotechnical parameters on the output from shallow buried
confined experiments (using a rigid steel ring), and these three explosives submitted for publication in Int. J. Impact Eng.
experimental methods can be used together to provide a full [3] Rigby S E, Fay S D, Clarke S D, Tyas A, Reay J J, Warren J A,
Gant M and Elgy I 2016 Measuring spatial pressure
picture of the behaviour of soils during high-strain-rate blast distribution from explosives buried in dry Leighton
and impact events. Buzzard sand submitted for publication in Int. J. Impact
Future test series using this new apparatus will be used Eng. 96 89–104
to define high-strain-rate yield surfaces for a series of soils, [4] Fox D M, Huang X, Jung D, Fourney W L, Leiste U and Lee J S
which will enable any strain-rate dependent behaviour to be 2011 The response of small scale rigid targets to shallow
buried explosive detonations Int. J. Impact Eng. 38 882–91
identified, including the quantification of radial inertial effects. [5] Fox D M, Akers S A, Leiste U H, Fourney W L, Windham J E,
This will be aided by comparison of the observed behaviour Lee J S, Ehrgott J Q and Taylor L C 2014 The effects of air
with results from high-pressure quasi-static triaxial tests [25]. filled voids and water content on the momentum transferred
The effects of soil parameters such as moisture content, initial from a shallow buried explosive to a rigid target Int. J.
density, particle size distribution and angularity on this shear Impact Eng. 69 182–93
[6] Bragov A M, Lomunov A K, Sergeichev I V, Tsembelis K
behaviour will also be investigated, enabling mechanisms to and Proud W G 2008 Determination of physicomechanical
be proposed for the observed variation in loading from buried properties of soft soils from medium to high strain rates
explosive events. Int. J. Impact Eng. 35 967–76
[7] Song B, Chen W and Luk V 2009 Impact compressive
response of dry sand Mech. Mater. 41 777–85
8. Summary [8] Barr A D, Clarke S D, Petkovski M, Tyas A, Rigby S E,
Warren J and Kerr S 2016 Effects of strain rate and
A methodology for partially-confined SHPB experiments has moisture content on the behaviour of sand under one-
been presented where the specimen is restrained laterally by dimensional compression Exp. Mech. 56 1625–39
[9] Martin B E, Chen W, Song B and Akers S 2009 Moisture
the inertia of a fluid annulus contained in a long steel reser- effects on the high strain-rate behavior of sand Mech.
voir. As lateral restraint is allowed to develop in response to Mater. 41 786–98
the axial stress pulse rather than being applied as a quasi-static [10] Lu H, Luo H and Komaduri R 2009 Dynamic compressive
cell pressure, this approach is more appropriate than existing response of sand under confinements Proc. of the SEM
dynamic conventional triaxial compression experiments for Annual Conf. (Albuquerque, NM, USA) pp 9–15
[11] Luo H, Lu H, Cooper W L and Komanduri R 2011 Effect of
representing explosive events in soils. mass density on the compressive behavior of dry sand under
A pressure transducer in the wall of the reservoir is used confinement at high strain rates Exp. Mech. 51 1499–510
to measure the radial stress in the specimen. Numerical mod- [12] Huang J, Xu S and Hu S 2013 Effects of grain size and
elling was performed to select the dimensions of the reser- gradation on the dynamic responses of quartz sands
voir to ensure that internal reflections do not affect the stress Int. J. Impact Eng. 59 1–10
[13] Pierce S J and Charlie W A 1990 High-intensity compressive
measurements, and so the radial stress on the surface of the stress wave propagation through unsaturated sands PhD
specimen will be accurately represented. The use of the par- Thesis Geotechnical Engineering Program, Colorado State
tially-confined SHPB was demonstrated on a dry specimen of University, USA

7
Meas. Sci. Technol. 27 (2016) 125903 A D Barr et al

[14] Bailly P, Delvare F, Vial J, Hanus J L, Biessy M and Picart D [19] Shin Y S, Lee M, Lam K Y and Yeo K S 1998 Modeling
2011 Dynamic behavior of an aggregate material at mitigation effects of watershield on shock waves Shock Vib.
simultaneous high pressure and strain rate: SHPB triaxial 5 225–34
tests Int. J. Impact Eng. 38 73–84 [20] Hallquist J O 2006 LS-DYNA Theory Manual (Livermore
[15] Christensen R J, Swanson S R and Brown W S 1972 Split- Software Technology Corporation, CA, USA)
Hopkinson bar tests on rock under confining pressure Exp. [21] Gray G T III 2000 Classic split-Hopkinson pressure bar testing
Mech. 12 508–13 ASM Handbook vol 8, ed D Medlin and H Kuhn (Materials
[16] Lindholm U S, Yeakley L M and Nagy A 1974 The dynamic Park, OH: ASM International)
strength and fracture properties of dresser basalt Int. J. Rock [22] Bancroft D 1941 The velocity of longitudinal waves in
Mech. Min. Sci. Geomech. Abstr. 11 181–91 cylindrical bars Phys. Rev. 59 588–93
[17] Frew D J, Akers S A, Chen W and Green M L 2010 [23] Tyas A and Pope D J 2005 Full correction of first-mode
Development of a dynamic triaxial Kolsky bar Meas. Sci. Pochhammer–Chree dispersion effects in experimental
Technol. 21 105704 pressure bar signals Meas. Sci. Technol. 16 642–52
[18] Martin B E, Kabir M E and Chen W 2013 Undrained [24] Davies R M 1948 A critical study of the Hopkinson pressure
high-pressure and high strain-rate response of dry bar Phil. Trans. R. Soc. A 240 375–457
sand under triaxial loading Int. J. Impact Eng. [25] Barr A D 2016 Strain-rate effects in quartz sand PhD Thesis
54 51–63 The University of Sheffield