Page 1 of 17

European Journal of Applied Sciences – Vol. 11, No. 6

Publication Date: December 25, 2023

DOI:10.14738/aivp.116.15700

Partom, Y. & Lindenfeld, A. (2023). Dynamic Response of Ductile Materials from Split Hopkinson Tension Bar Tests. European

Journal of Applied Sciences, Vol - 11(6). 23-39.

Services for Science and Education – United Kingdom

Dynamic Response of Ductile Materials from Split Hopkinson

Tension Bar Tests

Y. Partom

18 HaBanim, Zikhron Ya'akov 3094017, Israel

A. Lindenfeld

18 HaBanim, Zikhron Ya'akov 3094017, Israel

ABSTRACT

We use a SHTB (= Split Hopkinson Tension Bar) system to test the response of

aluminum 6061-T651 samples under dynamic tension, and to calibrate a strength

model for them. Interpretation of SHTB tests cannot be done with the so-called

classical analysis, because: 1) the samples are long (relative to compression

samples) and are not in dynamic equilibrium; 2) a neck is usually formed towards

the end of the test; and 3) the strain to failure is quite high (around 50%). We

therefore interpret the tests (calibrating from them a strength model and validating

it), using computer simulations of the entire test system. Our entire system

simulations are able to reproduce with good fidelity: 1) the response of the SHTB

gauges on the two bars; 2) the time evolution, shape and location of the neck; and

3) the strain to failure measured after the test.

INTRODUCTION

High strain rate material characterization in compression, for strain rates of the order of 102-

104/s, is commonly carried out with a SHPB (also called Kolsky bar) [1]. With this method the

specimen is compressed between input and output bars, and the resulting stress-strain curve

is deduced from strain measurements on both bars, at some distance from the sample. The

equationsused are known as the classical analysis. It has been shown [1,2], that the validity of

the classical analysis depends on a uniform stress distribution along the specimen (called stress

equilibrium).

Several modifications to the SHPB configuration have been developed for tests in tension, and

they are all called SHTB (= Split Hopkinson Tension Bar). Some of these use special specimen

geometries [3], and others use special loading arrangements [4-6]. With all these configurations

it’s not possible to use the classical analysis because: 1) the samples are usually long (relative

to compression samples), and are generally not in dynamic equilibrium; 2) a neck is usually

formed towards the end of the test; and 3) the strain to failure is quite high (around 50%) [7-

9].

Various techniques have been suggested to overcome this difficulty. Rajendran and Bless [10]

use Bridgman’s approximation [11] to obtain a stress-strain curve in dynamic tension beyond

the point of necking. To this end they measure the geometry of the neck, using high speed

photography. They find this approach to have different levels of accuracy for different

materials. Rodriguez et al. [7], using finite element simulations, suggest using an ‘effective

Page 2 of 17

Services for Science and Education – United Kingdom 24

European Journal of Applied Sciences (EJAS) Vol. 11, Issue 6, December-2023

length’ for which uniform strain can be assumed to hold in the specimen, and they determine

this effective length from the simulations. Rotbaum et al. [12] suggest using long specimen, to

keep the stress and strain constant in a large part of the specimen prior to necking. In this way

they are able to use the classical analysis with minimal error, but only up to 3% plastic strain.

Mirone [13] uses a correction to the classical analysis, which enables to reconstruct the stress- strain curve up to failure, but with an error of 15-25%.

Here we abandon the classical analysis and use instead computer simulations of the entire

SHTB system. With nowadays computer capabilities it’s not difficult to run simulations of the

entire test system including the bars, all the way to the specimen failure. From the simulations

we obtain time histories of particle velocities, and stresses and strains everywhere in the test

system. This includes details of neck formation, and strains (or particle velocities) measured on

the bars.

As shown in [14-17] and in [26], the final neck location depends on the boundary (or loading)

conditions, and is not sensitive to imperfections in the sample material or geometry. We

therefore use the final neck location as additional diagnostics of the test.

As mentioned above, to reach specimen failure in a dynamic tension test, we need a relatively

long loading pulse. For an existing Hopkinson bar system, this is not generally possible by using

a long striker bar, as these systems are usually designed for compression tests. As shown later,

we achieve a long enough loading pulse by using a technique that allows for wave reflections

from the input and output bars.

In section 2 we describe the experimental setup of our SHTB system, using circular DB (= Dog

Bone) specimens. In section 3 we describe our computational scheme. In section 4 we calibrate

a strength model in both compression and tension. Finally, in section 5 we extract the strain to

failure relation from the tests and from the simulations.

EXPERIMENTAL

The material we use in the tests is aluminum 6061-T651, which fails in tension at an average

strain of about 55%. We first calibrate the strength model for this material in compression. We

then use this calibration as the starting point for the calibration in tension.

For the compression tests we use our standard SHPB system with hardened C300 maraging

steel bars. The striker and the input and output bar lengths are 40, 399 and 250cm respectively,

and their diameter is 25.4mm (Fig. 1a). In both compression and tension tests we use our PDV

(= Photonic Doppler Velocimetry) diagnostics system [19], instead of the traditional strain

gauges diagnostics, at 1m from the specimen on the input and output bars. This PDV diagnostics

system, developed by our colleagues, was shown in [19] to be advantageous over the traditional

strain gauges diagnostics.

For the tension tests we use aluminum 7075 bars, to reduce the impedance mismatch between

the bars and the specimen. The striker and the input and output bar lengths are 75, 300 and

220cm, respectively. The input pulse goes mainly through the sleeve into the output bar. A small

part of it goes through the first thread into the specimen (which stays elastic), and then through

Page 3 of 17

25

Partom, Y. & Lindenfeld, A. (2023). Dynamic Response of Ductile Materials from Split Hopkinson Tension Bar Tests. European Journal of Applied

Sciences, Vol - 11(6). 23-39.

URL: http://dx.doi.org/10.14738/aivp.116.15700

the other thread back into the output bar. The input pulse is reflected from the far end of the

output bar, and then travels back to load the specimen in tension through the thread shown at

the bottom part of Fig. 1. This loading arrangement in tension was suggested by Nicholas [4].

When the first reflected pulse in not long enough to fail the specimen, the loading in tension

continues by a similar second reflection of the same pulse. In our tests the specimen has always

failed before the end of the second pulse. In Fig. 2 we show an x-t diagram of the various waves

in the input and output bars, including the second reflected wave. To recover the specimen after

it fails without additional damage, we use a sand box located behind the output bar.

Figure 1: Experimental setup. Top: Compression and tension. Bottom: Tension.

Figure 2: x-t diagram of the SHTB test including the first and second pulse reflections from the

far end of the output bar; blue=compression, red=tension.

Interferometer 2 Interferometer 1

Output bar

Specime

n

Input

bar

30

o

Striker

Region shown below

Page 4 of 17

Services for Science and Education – United Kingdom 26

European Journal of Applied Sciences (EJAS) Vol. 11, Issue 6, December-2023

For the compression tests we use disk specimens with L0/D0=0.5, with diameters, striker

velocities, and average strain rates, as shown in Table 1.

Table 1: Summary of the experimental conditions for the compression tests.

Specimen no. L0/D0(mm/mm) Vstriker (m/sec) (1/sec)

1 6/12 8.6 560

2 5/10 11.0 1500

3 7/14 18.5 1700

4 6/12 19.0 2500

5 7/14 30.0 4000

Both compression and tension specimens are made of the inner core of a 25.4mm diameter rod

with an average grain size of 30μm.

In Fig. 3 we show our tension specimen. We see that the diameter is 7.5mm and the gauge

lengths are 22.5, 30 and 37.5mm, so that their L0/D0 are 3, 4 and 5, respectively.

Figure 3: Geometry and dimensions of the SHTB specimens.

COMPUTER SIMULATIONS

As outlined in the introduction, we use here computer simulations of the entire SHTB system

to calibrate the strength model parameters in tension from test diagnostics, and we do this also

for the compression tests with the SHPB system. This enables us to: 1) to take into account the

early part of the output signals, when the specimen is not yet in dynamic equilibrium; and 2) to

take into account the effect of TS (=Thermal Softening), which is small but not negligible, and is

usually ignored by the classical analysis.

We perform the computer simulations with an in-house 2D Lagrangian hydrocode [20-22]. Our

hydrocode uses J2 plasticity with Wilkins’ radial return method for elastic-viscoplastic flow

[23]. There are many strength models in the literature to represent elastic-viscoplastic

response, and many of them show ability to reproduce test results. We could choose any of

them to demonstrate our approach to strength model calibration from SHTB tests via computer

simulations. We choose to use the JC (= Johnson-Cook) strength model [24], which is widely

used by far. The JC model is given by:

Page 5 of 17

27

Partom, Y. & Lindenfeld, A. (2023). Dynamic Response of Ductile Materials from Split Hopkinson Tension Bar Tests. European Journal of Applied

Sciences, Vol - 11(6). 23-39.

URL: http://dx.doi.org/10.14738/aivp.116.15700

(1)

where A, B, n, C, m are material parameters, is a reference strain rate taken as 1/s, T0 is the

initial temperature (room temperature here) and Tm is the melting temperature.

For the compression tests, the striker and bars properties are: (density)=8.1g/cc, G (shear

modulus) =71.8GPa and K (bulk modulus) =128.3GPa, resulting in a bar velocity of 4734m/s.

The cell size along the bars is 1.5mm, and it reduces to 0.3mm toward the specimen. The

average cell size in the specimen is 0.1mm, and the average computing time is several hours on

four processors.

The geometry of the tension tests computations is shown in Fig. 4. The striker and bars

properties are: =2.8g/cc, G=26.7GPa, and K=75.8GPa, resulting in a bar velocity of 5056m/s.

The threaded connection between the specimen and the bars consists of two toroidal rings,

with end angles of 60o on both sides (see Fig. 4). This simplified representation of the threads

is adequate, as it enables radial motion between the specimen and the bars, and keeps the

threads in the elastic range. The meshing is similar to that of the compression tests. We define

two computational Lagrangian gauges at the ends of the specimen gauge section, for the

evaluation of the average strain rate in the specimen.

Figure 4: Geometry of the tension tests simulations: a) specimen region; b) thread region

(enlarged).

Page 6 of 17

Services for Science and Education – United Kingdom 28

European Journal of Applied Sciences (EJAS) Vol. 11, Issue 6, December-2023

TEST RESULTS AND STRENGTH MODEL CALIBRATION

In Fig, 5 we show the experimental and computed signals from test 5 of Table 1. We focus on

the transmitted pulse as it includes less noise. The calibrated parameters of the JC model are

shown in Table 2, with Tm=900K and m=1. We see from Fig. 5-bottom (red line) that a better

agreement is obtained when the simulation includes thermal softening (TS). A similar level of

agreement is obtained for all other tests (not shown here).

Figure 5: Compression tests, comparison between test and simulation for specimen 5 in Table

1. Top: all signals. Bottom: transmitted signal. (The red line drop of the incident wave signal is

an artifact of the recording system and should be ignored. Similar drops appear in the output

signals of several other figures).

Table 2: Material parameters of the JC model, calibrated from the compression tests.

A (MPa) B (MPa) C n Tm (°K) m

339 180 0.015 0.36 900 1

We check the sensitivity of the above calibration by changing the value of A by ±7MPa, which is

about ±2%, and we show the comparison in Fig. 6. We see from Fig. 6 that the simulations are

rather sensitive to small changes in the parameter A. We find a similar sensitivity (not shown

here) to the other strength model parameters.

Page 7 of 17

29

Partom, Y. & Lindenfeld, A. (2023). Dynamic Response of Ductile Materials from Split Hopkinson Tension Bar Tests. European Journal of Applied

Sciences, Vol - 11(6). 23-39.

URL: http://dx.doi.org/10.14738/aivp.116.15700

In Table 3 we summarize the specimen geometries, loading conditions, and average strain rates

of the tension tests. The average strain rate is defined for the gauge section as the ratio of the

velocity difference of its two ends to its current length.

Table 3: Summary of the tension tests. D0=7.5mm.

Specimen no. L0/D0 Vstriker (m/sec) (1/sec) Failure

1 3 10.0 230 No

2 3 13.8 400 No

3 3 19.6 670 Yes

4 3 20.4 700 Yes

5 3 26.5 950 Yes

6 3 27.0 1000 Yes

7 4 15.5 350 No

8 4 15.0 340 No

9 4 20.5 500 Yes

10 4 19.5 470 Yes

11 4 28.0 800 Yes

12 4 28.0 800 Yes

13 5 10.6 150 No

14 5 9.3 120 No

15 5 22.5 470 Yes

16 5 19.5 400 Yes

17 5 29.5 670 Yes

18 5 28.0 620 Yes

Specimens 1-6: L0=22.5mm. Specimens 7-12: L0=30mm. Specimens 13-18: L0=37.5mm.

Table 4: Summary of experimental and simulation results of the tension tests.

Specimen

No.

Final gauge length

(elongation), mm

Loading wave

failure

Normalized neck

location

Strain to

failure

Exp., ±0.125 Sim., ±0.1 Exp. Sim. (in the neck)

1 24.25 (1.75) 23.85 (1.35) N/A N/A N/A N/A

2 26.50 (4.0) 26.67 (4.17) N/A 0.46 0.46 N/A

3 27.50 (5.0) 27.23 (4.73) 2

nd 0.3 0.44 0.6

4 27.50 (5.0) 27.44 (4.94) 1

st 0.3 0.44 0.6

5 28.0 (5.5) 27.53 (5.03) 1

st 0.31 0.4 0.6

6 28.25 (5.75) 27.23 (4.73) 1

st 0.34 0.39 0.7

7 35.0 (5.0) 34.77 (4.77) N/A 0.32 0.41 N/A

8 34.5 (4.5) 34.75 (4.75) N/A 0.28 0.39 N/A

9 n.m. 35.99 (5.99) 2

nd n.m. 0.33 0.5

10 n.m. 35.90 (5.90) 2

nd n.m. 0.36 0.55

11 36.25 (6.25) 35.88 (5.88) 1

st 0.23 0.27 0.6

12 35.75 (5.75) 35.88 (5.88) 1

st 0.22 0.27 0.6

13 39.50 (2.0) 39.59 (2.09) N/A N/A N/A N/A

14 38.75 (1.25) 38.65 (1.15) N/A N/A N/A N/A

15 44.0 (6.5) 44.43 (6.93) 2

nd 0.28 0.23 0.5

16 44.25 (6.75) 44.65 (7.15) 2

nd 0.27 0.26 0.55

Page 8 of 17

Services for Science and Education – United Kingdom 30

European Journal of Applied Sciences (EJAS) Vol. 11, Issue 6, December-2023

17 44.25 (6.75) 43.67 (6.17) 1

st 0.18 0.2 0.6

18 44.25 (6.75) 43.82 (6.32) 1

st 0.17 0.2 0.55

Neck location is normalized by the gauge length. n.m. means not measurable.

In Fig. 6 we show the recorded signals for specimens 8, 10 and 11 of Table 3.

a) No failure

b) Second wave failure

Page 9 of 17

31

Partom, Y. & Lindenfeld, A. (2023). Dynamic Response of Ductile Materials from Split Hopkinson Tension Bar Tests. European Journal of Applied

Sciences, Vol - 11(6). 23-39.

URL: http://dx.doi.org/10.14738/aivp.116.15700

c) Third wave failure

Figure 6: Recorded signals from the tension tests with specimen number: a) 8, b) 10, and c) 11.

The wave for which failure occurs is circled on the output bar signal. (See bracketed remark

under Fig. 5).

In Fig. 7 we compare the experimental and computed signals for specimens 8, 10 and 11, and it

can be seen that agreement is good. Similar agreement is obtained for the other tests (not

shown here). Another parameter that we use to validate the simulation results (beside the neck

location, see section 5) is the elongation of the gauge section. This parameter represents the

integral behavior of the specimen during the whole deformation process, including necking. In

Table 4 we compare the measured and computed gauge section elongation, and see that except

for specimen 1, agreement is within 9%. As we’ve been using in tension the same strength

model parameters calibrated for compression, we conclude that for aluminum 6061-T651

there’s no significant difference between the strength model parameters in uniaxial tension and

compression.

Input bar Output bar

a)

Page 10 of 17

Services for Science and Education – United Kingdom 32

European Journal of Applied Sciences (EJAS) Vol. 11, Issue 6, December-2023

b)

c)

Figure 7: Comparison between recorded and computed signals for tension tests: a) test 8, b)

test 10 and c) test 11. (See bracketed remark under Fig. 5).

In Fig. 8 we show the neck and the EPS (= Effective Plastic Strain) evolution in specimen 10

from our simulation, all the way until just before failure.

In Fig. 9 we show EPS profiles along the gauge section from simulations of tests 6, 11 and 18 at

four times before failure.

We see that EPS profiles become non-uniform as the neck develops, in accordance with

previous experimental [8, 9] and computational work [7, 15]. Also, maximum EPS is on the axis,

in accordance with previous numerical work [7, 24].

Page 11 of 17

33

Partom, Y. & Lindenfeld, A. (2023). Dynamic Response of Ductile Materials from Split Hopkinson Tension Bar Tests. European Journal of Applied

Sciences, Vol - 11(6). 23-39.

URL: http://dx.doi.org/10.14738/aivp.116.15700

Figure 8: Time evolution of the neck, and color maps of effective plastic strain, in specimen 10;

a) to d): first tension wave, e) to h): second tension wave.

Page 12 of 17

Services for Science and Education – United Kingdom 34

European Journal of Applied Sciences (EJAS) Vol. 11, Issue 6, December-2023

Fig. 9/a

Fig.9/b

Page 13 of 17

35

Partom, Y. & Lindenfeld, A. (2023). Dynamic Response of Ductile Materials from Split Hopkinson Tension Bar Tests. European Journal of Applied

Sciences, Vol - 11(6). 23-39.

URL: http://dx.doi.org/10.14738/aivp.116.15700

Fig. 9/c

Figure 9: Effective plastic strain profiles at different times from the simulations of tests: a) 6, b)

11 and c) 18. Specimens loading starts at t1500μs.

FAILURE PREDICTION

When the specimen fails, a release wave is generated at the failure location, which then travels

in both directions. This release wave is represented in both bar signals as a velocity change, as

seen in Figs. 6b and 6c. It decreases the velocity of the input bar, and increases the velocity of

the output bar. The velocity change in the bars caused by specimen failure is:

spec spec

failure

bar bar bar bar

F .A

V

Z .C .A

 = =

(2)

where F and spec are the force and the longitudinal stress in the specimen prior to failure

respectively, Zbar is the bar mechanical impedance, bar is the bar density, Cbar is the elastic bar

velocity, and Aspec, Abar, are the cross-section areas of the specimen and the bar, respectively. For

a typical value of this velocity change, we use spec ≈ 450MPa from Fig. 7, bar = 2800kg/m3,

Cbar=5056m/s, Abar  5.0710-4m2 and Aspec = 2.8310-5m2 , for a representative neck diameter of

6 mm (measured from the failed specimens), and we get vfailure = 1.8 m/sec. Hence, an abrupt

velocity increase of ~2 m/sec in the experimental signals marks the time of failure.

We use the time of this velocity change to calibrate the EPS (effective plastic strain) value at

failure. In Fig. 10 we show an example of such a calibration for specimen 11. We see there that

for higher values of EPS at failure, the abrupt velocity change occurs later. We show the

calibrated EPS at failure (StF) in Table 4, and we see there that the average StF is 0.58, with a

standard deviation of 0.05. There is a significant difference between the StF in the neck and the

average strain in the gauge section at failure, as also pointed out in [15].

Page 14 of 17

Services for Science and Education – United Kingdom 36

European Journal of Applied Sciences (EJAS) Vol. 11, Issue 6, December-2023

In Table 4 we compare neck location between tests and simulations, and find good agreement.

We conclude, in agreement with [15, 17] that: 1) neck location is insensitive to small

perturbations; 2) neck location depends on geometry and loading conditions; and 3) neck

location moves towards the impacted side for a higher impact velocity. In Fig. 11 we show the

effect of loading velocity on neck location.

a)

b)

Page 15 of 17

37

Partom, Y. & Lindenfeld, A. (2023). Dynamic Response of Ductile Materials from Split Hopkinson Tension Bar Tests. European Journal of Applied

Sciences, Vol - 11(6). 23-39.

URL: http://dx.doi.org/10.14738/aivp.116.15700

c)

Figure 10: StF (strain to failure) calibration for test 11; a) StF=0.4, b) StF=0.5, c) StF=0.6.

(See bracketed remark under Fig. 5).

a) b)

Figure 11: Specimens before and after impact. a) specimen 6 after failure; b) effect of impact

velocity on neck location. (Impact is on the left, and u=velocity difference between specimen

two sides).

SUMMARY

We focus here on the response of ductile materials to uniaxial dynamic tensile loadings by

means of a SHTB system. We use a combined experimental and numerical approach, our

representative material is aluminum 6061-T651, we use circular dog-bone specimens with

gauge length to diameter ratio of 3, 4 and 5, and we use the JC strength model.

Page 16 of 17

Services for Science and Education – United Kingdom 38

European Journal of Applied Sciences (EJAS) Vol. 11, Issue 6, December-2023

We do not use the so-called classical analysis to interpret our tests and to calibrate the strength

model. Instead, we use direct computer simulations of the entire test system.

From the tests and simulations, we get: 1) high level overall accuracy, predictions and test data

differ by no more than 2%; 2) accurate prediction of stress-strain curves even for less than 5%

strain; 3) inherent and simple prediction of thermal softening; 4) accurate prediction of StF

(strain to failure) in the neck; 5) accurate prediction of the neck location; and 6) accurate

prediction of the elongation of the gauge section, which can serve as an additional calibration

parameter.

References

[1] Kolsky H (1949), An investigation of the mechanical properties of materials at very high rates of loading,

Proc Phys Soc B 62:676-700.

[2] Bertholf LD, Karnes CH (1975) Two-dimensional analysis of the Split Hopkinson Pressure Bar system, J Mech

Phys Solids 23:1-19.

[3] Lindholm US, Bessey RL, Smith GV (1971) Effect of Strain Rate on Yield Strength, Tensile Strength, and

Elongation of Three Aluminum Alloys, J Mater 6:119-133.

[4] Nicholas T (1981) Tensile Testing of Materials at High Rates of Strain, Exp Mech 62:177-185.

[5] Staab GH, Gilat A (1991) A direct-tension split Hopkinson bar for high strain rate testing, Exp Mech

313:232-235.

[6] Nemat-Nasser S, Isaacs JB, Starrett JE (1991) Hopkinson techniques for dynamic recovery experiments, Proc

R Soc Lond A 435:371-391.

[7] Rodríguez J, Navarro C, Sánchez-Gálvez V (1994) Numerical assessment of the dynamic tension test using

the Split Hopkinson Bar, J Test Eval 22:335-342.

[8] Gilat A, Schmidt TE, Walker AL (2009) Full Field Strain Measurement in Compression and Tensile Split

Hopkinson Bar Experiments, Exp Mech 49:291-302.

[9] Besnard G, Hild F, Lagrange JM, Martinuzzi P, Roux S (2012) Analysis of necking in high-speed experiments

by stereo correlation, Int J Impact Eng 49:179-191.

[10] Rajendran AM, Bless SJ (1986) Determination of tensile flow stress beyond necking at very high strain rate,

Exp Mech 26:319-323.

[11] Bridgman PW (1994) The stress distribution at the neck of a tension specimen, Trans ASM 32:553-574.

[12] Rotbaum Y, Rittel D (2014) Is there an optimal gauge length for dynamic tensile specimens? Exp Mech

54:1205-1214.

[13] Mirone G (2013) The dynamic effect of necking in Hopkinson bar tension tests, Mech Mater 58:84-96.

[14] von Karman T, Duwez P (1950) The Propagation of Plastic Deformation in Solids, J Appl Phys 21:987-994.

[15] Osovski S, Rittel D, Rodríguez-Martínez JA, Zaera R (2013) Dynamic tensile necking: influence of specimen

geometry and boundary conditions, Mech Mater 62:1-13.

Page 17 of 17

39

Partom, Y. & Lindenfeld, A. (2023). Dynamic Response of Ductile Materials from Split Hopkinson Tension Bar Tests. European Journal of Applied

Sciences, Vol - 11(6). 23-39.

URL: http://dx.doi.org/10.14738/aivp.116.15700

[16] Rittel D, Rotbaum Y, Rodríguez-Martínez JA, Sory D, Zaera R (2014) Dynamic necking of notched tensile

bars: an experimental study, Exp Mech 54:1099-1109.

[17] Vaz-Romero A, Rodríguez-Martínez JA, Arias A (2015) The deterministic nature of the fracture location in

the dynamic tensile testing of steel sheets, Int J Impact Eng 86:318-335.

[18] Gerlach R, Kettenbeil C, Petrinic N (2012), A new split Hopkinson tensile bar design, Int J Impact Eng 50:63-

67.

[19] Avinadav C, Ashuach Y, Kreif R (2011) Interferometry-based Kolsky bar Apparatus, Rev Sci Instrum 82

073908.

[20] Firstenberg O, Ashuach Y, Partom Y (2006) A simple model for dynamic shear failure of stainless steel, Proc

8th Int Conf Mech Phys Behavior Mater Under Dynamic Loading, Dijon, France, J Phys IV 134:191-196.

[21] Lovinger Z, Ashuach Y, Firstenberg O (2009) Measuring dynamic strength at low plastic strains using a hat- shaped specimen, Proc 9th Int Conf Mech Phys Behavior Mater Under Dynamic Loading, Brussels, Belgium,

DYMAT 1649-1655.

[22] Lovinger Z, Rikanati A, Rosenberg Z, Rittel D (2011) Electro-magnetic collapse of thick-walled cylinders to

investigate spontaneous shear localization, Int J Impact Eng 38:918-929.

[23] Wilkins ML (1964), Calculation of elastic-plastic flow, In: Alder B, Fernbach S, Rotenberg M (Eds) Methods

in computational physics, vol. 3, Academic Press pp 211-263.

[24] Johnson GR, Cook WH (1983), A constitutive model and data for metals subjected to large strains, high

strain rates and high temperatures, Proc 7th Int Symp Ballist, The Hague, Netherlands.

[25] Glema A, Łodygowsky T, Perzyna P (2000) Interaction of deformation waves and localization phenomena in

inelastic solids, Comput. Methods Appl Mech Eng 183:123-140.

[26] Partom, Y, Lindenfeld A (2017) 20th Shock compression of condensed matter Conference, AIP Conf. Proc.

1979, 070023.