International Journal of Impact Engineering 86 (2015) 111–123

Contents lists available at ScienceDirect

International Journal of Impact Engineering

j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / i j i m p e n g

Explosive driven shock tube loading of aluminium plates:

experimental study
M.A. Louar a,b,*, B. Belkassem b, H. Ousji a,b, K. Spranghers a, D. Kakogiannis b, L. Pyl a,
J. Vantomme b
a Department of Mechanics of Materials and Constructions, Vrije Universiteit Brussel (VUB), Pleinlaan 2, B-1050 Brussels, Belgium
b Department of Civil and Materials Engineering, Royal Military Academy (RMA), Av. De la Renaissance 30, B-1000 Brussels, Belgium

A R T I C L E I N F O A B S T R A C T

Article history: To accurately identify the material characteristics in material testing, the control of the initial and bound-
Received 16 April 2015 ary conditions is very important. Among these conditions, obtaining a repeatable and known loading is
Received in revised form 4 June 2015 a sensitive point especially for tests in the dynamic and impulsional range. Several techniques are used
Accepted 26 July 2015
to generate blast wave loads. This paper focuses on the use of an explosive driven shock tube (EDST).
Available online 5 August 2015
The results of a series of tests using an EDST are discussed, with regard to the pressure, the impulse, the
wave planarity and the repeatability of the loading. The EDST and free air blast are used to generate blast
Keywords:
loading on fully clamped square aluminium plates. The response from both loads is measured with a
Shock tube
Free air explosion high-speed 3D digital image correlation (DIC) system and compared to each other. The DIC measure-
DIC ments are compared to strain gauge measurement for validation purposes. It is found that EDST allows
Strain gauge to reach higher levels of pressure and impulse than the free air blast leading to higher deformations and
Full-field measurements strain rates. It is also observed that the EDST tests show a higher repeatability and symmetry relative to
the free air blast tests.
© 2015 Elsevier Ltd. All rights reserved.

1. Introduction case. This would result in a more predictable loading that can also
be reproduced more realistically in a numerical model.
The study of the behaviour of structures and materials under blast The blast response of simple structural components such as plates
loading is essential to determine the efficiency of protective mea- is investigated in order to study the behaviour of both materials and
sures and the survivability of personnel and equipment in case of structures under high strain rate conditions. In the literature, several
intentional or accidental explosions. The use of numerical tools to types of materials are considered such as composites [5–13], and
achieve this purpose is undeniably efficient, but a numerical model metals [4,14–20]. The blast load is generated using different tech-
needs reliable material models and boundary conditions (among niques: free air detonation of a high explosive (HE) [4,5,12,13,18,19],
other points) in order to accurately predict the behaviour of a struc- using a gas driven shock tube [6–11,14] or an explosive driven shock
ture under blast loading. These material properties can only be tube (EDST) [6,15,20]. Several practical problems related to the use
derived and validated by experimental testing which can be usual of free air explosions are reported in Refs. 4, 18, and 19 regarding
tests in tension, compression, etc., or tests with more complex stress/ the repeatability of the tests and the distribution of the generated
strain fields and using an inverse method [1–4]. The free air blast wave. This explains why gas driven shock tubes are widely used.
detonation of a high-explosive (HE) is a widely used way to gen- The use of an EDST is an alternative for the two previous tech-
erate a blast wave. However, for small scale tests, small variation niques which allows to obtain well selected pressure–impulse
in the charge (shape, mass, position, etc.) can significantly affect the combinations. Moreover, the confinement and the multiple reflec-
loading and hence, bias the test results. The use of an explosive driven tions permit to obtain relatively important pressures and impulses
shock tube (EDST) to channel the blast wave from the detonation with small charges in comparison with free-air blasts.
of a HE is a way to reduce the effects of these variations in the load In order to experimentally analyse the response of specimens
under an impulsive loading, several methods are applied such as
high-speed imaging with digital image correlation (DIC) to measure
the full-field deformations [4–10,17–19], strain gauges [9], ballis-
* Corresponding author. Department of Mechanics of Materials and Constructions,
Vrije Universiteit Brussel (VUB), Pleinlaan 2, B-1050 Brussels, Belgium. Tel.:
tic pendulum (for impulse measurement) [12,15,20], macroscopic
+ 3224414012; Fax: + 3224439186. and/or microscopic examination [5–12], and residual compressive
E-mail address: [email protected] (M.A. Louar). strength measurements [7,11].

http://dx.doi.org/10.1016/j.ijimpeng.2015.07.013
0734-743X/© 2015 Elsevier Ltd. All rights reserved.
112 M.A. Louar et al./International Journal of Impact Engineering 86 (2015) 111–123

Fig. 1. Profile of characteristic pressure wave [21].

200 mm
Outer tube
Steel fiber
80 mm
20 mm reinforced
concrete

Inner tube
84.1 mm

Axis

Explosive
1200 mm
charge

(a)

EDST

1 3

2 4

Sensors
Top view Front view

(b)

Fig. 2. Schematic representation of (a) the explosion driven shock tube (EDST) and (b) the position of the pressure sensors on the EDST.
 M.A. Louar et al./International Journal of Impact Engineering 86 (2015) 111–123 113

Fig. 3. Experimental set-up: (a) EDST with frame and (b) position of the pressure gauges on the plate.

This paper reports a series of experiments with EDST gener-

3 ated blast loading and free air blast loading of thin square aluminium
Raw signal
Filtered signal plates. The following aspects of the EDST generated loading are in-
vestigated: the reproducibility of the loading, the pressure–
2.5
impulse combinations obtained and the planarity of the wave. The
analysis of the aluminium plates responses by means of high-
speed DIC allows to draw some conclusions about the repeatability
2
of the tests and the symmetry of the dynamic response of the speci-
mens. As a validation for the DIC measurements, strain gage
1.5 measurements are performed as well during some tests.
Pressure (MPa)

2. The loading
1
In this section, the blast loading generated by the EDST is
characterised in terms of both incident and reflected pressure,
0.5 impulse and wave planarity. The repeatability of the tests is also
discussed.

0
2.1. Free air blast

When detonation of explosive charge occurs, a sudden release

-0.5
of energy takes place in the surrounding medium (usually air). This
results in a shock wave which propagates radially in all directions
-0.5 0 0.5 1 1.5 2 2.5
Time (ms) at a supersonic speed with respect to the initial medium. The front
of the shock wave has a greater pressure than the area immedi-
Fig. 4. Application of a low-pass filter to the measured incident pressure signal to ately in front of the wave; this overpressure decreases exponentially
reduce the noise. as the shock wave propagates in the surrounding environment. The

11 6
Raw reflected pressure signal Reflected Sensor 1
10 pressure Sensor 2
5
9
EDST
8
4
7 Charge
Pressure (MPa)
Pressure (MPa)

6
3
5 Incident
pressure
4 2

2 1

1
0
0
0 1 2 3 4 5 6 7 8 0.5 1 1.5 2
Time (ms) Time (ms)

(a) (b)

Fig. 5. Typical raw signals (a) from a sensor on the aluminium plate and (b) from two sensors on the EDST.
114 M.A. Louar et al./International Journal of Impact Engineering 86 (2015) 111–123

2500 700

600
2000

Incident specific impulse (Pa.s)

500

Incident pressure (kPa)

1500
400

300
1000

200
500
100

0 0
5 10 15 20 30 40 5 10 15 20 30 40
Charge masse (g) Charge masse (g)

(a) (b)

Fig. 6. (a) Peak incident pressures and (b) specific incident impulses for the charges used with an indication of the variability in the measurements.

shape of the characteristic pressure–time function of a shock wave When a blast wave propagates in the medium where it was gen-
consecutive to a detonation depends on the energy released by the erated (in this case air), it is referred to as the incident wave. When
explosive charge and the distance from the centre of the explo- this wave encounters an obstacle, a reflected wave is created. This
sion to the point of interest. reflection amplifies the blast wave. The ratio between reflected and
The pressure–time profile of a far-field blast (Fig. 1) can be de- incident peak pressures is indicated by the reflection coefficient CR
scribed using the following parameters. The shock wave reaches the [22] and can be calculated by Equation 2:
point of interest at ta producing a sudden rise in pressure to a finite
value ΔP+. The duration t+ is called positive phase duration and the ΔPr+ 8ΔPi+ + 7P0
Cr = = (2)
area under the curve is the positive impulse I+. Beyond time ta + t+ ΔPi+ ΔPi+ − 5P0
shocked gas pressure is below ambient pressure creating a vacuum.
The pressure continues to decrease to a finite value ΔP− then rises where P0 is the atmospheric pressure and the subscripts i and r refer,
gradually to ambient pressure at time ta + t+ + t−. Similarly to the pos- respectively, to incident and reflected.
itive phase, the area between the pressure curve and the abscissa It is assumed in general that the free air detonation of a spher-
between ta + t+ and ta + t+ + t− represents the negative impulse I−. ically shaped HE charge leads to a spherical blast wave. At a lab scale
Many functions are available to model the pressure–time signal (few tens of grams) and at mid-range, this is generally not the case.
measured at a certain distance of the detonation point using a com- Small variations in weight, shape and local density of the charge
bination of the aforementioned parameters. The Friedlander equation affect the evolution of the detonation wave and thus the blast wave.
[22] is used (see Equation 1) most of the time to describe it:
2.2. Experimental set-up for the EDST
− αt
⎛ t⎞
ΔP (t ) = ΔP + ⎜ 1 − ⎟ e t+ (1)
⎝ t+ ⎠ The EDST used in this study is a cylindrical steel tube with an
inner diameter of 168.2 mm, a thickness of 4.5 mm and a length
where α is positive and is called the waveform parameter and of 1200 mm. At the entrance, where the pressure is the most sig-
depends on the overpressure ΔP+ [22]. nificant, the tube is reinforced over the first 200 mm with another

12000 1800

1600
10000
Reflected specific impulse (Pa.s)

1400
Reflected pressure (kPa)

8000 1200

1000
6000
800

4000 600

400
2000
200

0 0
5 10 15 20 30 40 5 10 15 20 30 40
Charge masse (g) Charge masse (g)

(a) (b)

Fig. 7. (a) Peak reflected pressures and (b) specific reflected impulses for the charges used with an indication of the variability in the measurements.
 M.A. Louar et al./International Journal of Impact Engineering 86 (2015) 111–123 115

14000 3. Experimental results

Test 1
Test 2 The incident and reflected pressure–time signals are measured
12000 Test 3 for charges of 5, 10, 15, 20, 30 and 40 g of C4 using piezoelectric
Test 4 pressure sensors. The set-up consists of an EDST, a steel frame and
10000 Friedlander equation a 15 mm thick aluminium plate that includes pressure sensors
(Fig. 3); the thickness is rather high because of the wanted rigidi-
ty for the reflected pressure measurements. For the incident pressure
8000
Pressure (kPa)

measurements, the aluminium plate is removed and only the sensors

on the tube are used. The aim of these tests is to measure the pres-
6000 sures and impulses that will be applied on future test specimens
and also to check their distribution on the loaded area. Each test
is repeated three times in order to have an estimation of the range
4000 of variability. The explosive charges are placed at the entrance of
the tube.
2000
3.1. The experimental measurements
0
It is observed that the signals from the sensors on the tube
contain a significant level of noise that alters the pressure values.
−2000 This noise is due to the vibrations from the explosion in the tube
0.5 1 1.5 2 2.5
Time (ms) wall. The vibrations are transmitted to the sensors and are notice-
able in the signals. Since this noise is mainly a high frequency signal,
Fig. 8. Pressure–time signals from the sensor positioned at the centre of the plate a butterworth filter is used in Matlab to reduce it. This filter acts,
for a charge of 20 g for four different tests illustrating the repeatability of the loading in this case, as a low-pass filter. Fig. 4 compares the raw signal with
and comparing the profiles to the Friedlander waveform.
the filtered signal for the incident pressure signal. The treated signal
is found to be acceptable and all the signals measured by the trans-
ducers on the tube are filtered using the same filter function. The
steel tube of 193.7 mm inner diameter and 4.5 mm thickness. The pressure and impulse measurements are also found to have a good
radial space between the two tubes is filled with steel fibre rein- repeatability with an average relative standard deviation of about
forced concrete as illustrated in Fig. 2a. At the end of the tube, four 10% each. Fig. 5a shows a typical signal from a sensor on the target
pressure transducers are mounted to measure the pressure at two plate; it can be seen that no filtering is required.
different sections of the tube and two different angular positions. Fig. 5b gives typical signals measured by the sensors placed on
The distance between two sensors is 80 mm and the distance the tube. Among the advantages of having transducers on the EDST
between the last transducer and the end of the tube is 20 mm is that the incident and reflected pressures can be measured for each
(Fig. 2b). The pressure sensors used in these tests are high frequen- test.
cy pressure transducers PCB QUARTZ ICP 102B04 with a sampling The incident pressures are now discussed. Fig. 6 presents the
frequency of 5 MHz. average peak-pressure and specific impulse values from the

Fig. 9. Pressure–time signals from three sensors: sensor 1 at the centre of the plate and sensors 2 and 3 at radial distances of 45 mm and 65 mm, respectively, illustrating
the planarity of the blast wave.
116 M.A. Louar et al./International Journal of Impact Engineering 86 (2015) 111–123

Fig. 10. Distribution of the pressure (a) and the impulse (b) on the plate.

measurements for the different charges. The average variability might be due to intensity of the shock and the pressure level which
(based of three tests) in the incident pressure measurements is 8.5% is close to the saturation pressure of the sensors (13 MPa). This gives
and for the impulse, it is 11.93% which represents an improve- an idea of the loading that can be generated by the EDST with rel-
ment relatively to what is observed in Ref. 4 for free air explosions. atively small charges. As a comparison, the detonation of a charge
The higher variability in the impulse is due to the aforementioned of 20 g of C4 at 1.2 m gives a reflected pressure Pr = 105.9 kPa and
vibrations in the tube. a reflected impulse of Ir = 30.77 Pa.s. The same charge at 250 mm,
The measurement results for the reflected pressure and the which means close range explosion, gives Pr = 7986 kPa and
impulse are shown in Fig. 7 in which the measurements from the Ir = 204.8 Pa.s , whereas with the same charge, the EDST generates a
tests with 30 and 40 g of C4 are not provided due to the saturation blast wave with Pr = 10, 742 kPa and Ir = 1641Pa.s this demonstrates
of the sensors. A higher variability in the 20 g tests is observed. This the effect of confinement on the blast wave.

Fig. 11. Distribution of the pressure on the plate at different times.

 M.A. Louar et al./International Journal of Impact Engineering 86 (2015) 111–123 117

Table 1 Concerning the planarity of the blast wave, Fig. 9 shows the
EN AW-1050A H24 elastic and Johnson–Cook material parameters and thermal prop- pressure–time signals from three transducers on the loaded area.
erties from Ref. 19.
The sensors are situated at different radial distances from the centre
E [GPa] A [GPa] B [GPa] n [−] C [−] m [−] Tf [K] of the plate, which coincides with the centre of the tube. It can be
69 0.11 0.15 0.36 0.014 1 918.15 seen that the three signals are very similar.
Another aspect treated is the distribution of the pressure and
the impulse over the aluminium plate. Fig. 10 shows that the loading
3.2. The repeatability and distribution of the loading is mainly concentrated on a circular section corresponding to the
EDST for the used distance between the end of the tube and the
Fig. 8 shows the pressure–time signal measured by the sensor plate (5 mm). Fig. 11 shows the time evolution of the pressure dis-
positioned at the centre of the plate as a result of the detonation tribution on the plate and it is concluded that the loading can be
of a charge of 20 g of C4 at the entrance of the EDST. This gives good considered planar and localised in the cross-sectional area of the
indications regarding the reproducibility of the loading. It is also tube all over the loading time.
observed that the pressure–time profiles are similar to the ones ob- From these observations, it can be assumed for future numeri-
tained with the Friedlander equation for free air blast. cal work that only the area of the EDST is loaded.

Fig. 12. Experimental set-up: (a) high-speed cameras in stereovision configuration and (b) high contrast speckle pattern.

a− (5,5) b− (15,15) c− (25,25)

Displacement (mm)

60
EDST with 20g
45 EDST with 15g
Free air blast
20 with 25g

0 0 0
0 2 4 6 0 2 4 6 0 2 4 6
Time (ms)
Maximum principal
true strain ε1 (%)

8,5
6

1.5
0 0 0
0 2 4 6 0 2 4 6 0 2 4 6
Time (ms)
500
Strain rate (s−1)

300

80
0 0 0
0 2 4 6 0 2 4 6 0 2 4 6
Time (ms)
Fig. 13. Comparison between measurements for free air blast and for EDST loading at three different points on the target plate.
 118 M.A. Louar et al./International Journal of Impact Engineering 86 (2015) 111–123

400 mm 25
0.5 ms

Displacement (mm)
20 0.625 ms
0.75 ms
15 0.875 ms
y
10 1 ms

400 mm
(-150,0) (0,0) (150,0)
x 5

0
−100 −50 0 50 100
x (mm)
Fig. 16. Displacement profiles along the x-axis of a panel under free air blast with
a charge weight of 25 g at 250 mm.

Fig. 14. Schematic of the specimen with the axes.

4.2. Experimental set-up

4. High-speed 3D-DIC measurements In order to evaluate the dynamic response of aluminium (EN AW-
1050A H24, Table 1) plates under blast loading, two sets of tests are
4.1. Digital image correlation performed. In the first set, the loading is generated by the EDST;
in the second, it is generated by a free air explosion. The square speci-
The DIC is a full-field optical, and thus non-contact, measure- mens with a total surface of 400 × 400 mm2 and a thickness of
ment technique. It can be used for 2D (one camera) or 3D (2 cameras 2.5 mm are clamped to the steel frame using bolts (see Fig. 2) leaving
or more) measurements. In the case of local DIC, it uses the grey a free area of 300 × 300 mm2. On the face opposite to the loading,
value digital images from the camera(s) to follow the displace- a high-contrast speckle pattern is applied on the specimens (Fig. 12b)
ment and the deformation of small areas from a pattern on the target and the neighbouring area of the steel frame. Facing this side, two
named subsets. This is done by minimising the square difference PhotronFastcam SA5 high-speed digital cameras (Fig. 12a) are placed
of grey value over the neighbouring area of the considered subset. in a stereovision configuration at a distance of 1.8 m from the frame.
Using the intrinsic and extrinsic parameters of the sensors, the 3D This distance allows to record 6 ms before the blast wave reaches
position of each point can be obtained. More details on principals the cameras. The aperture of both lenses (Nikon D80, 50 mm focal
and application of the DIC can be found in Ref. 23. length f) is set to f/5.6 to keep the plate in focus during the test.
The accuracy of the measurements of the displacements can be The specimens are illuminated with 4 high intensity spotlights to
assessed through different methods. In this paper, the standard de- balance the effect of the reduced aperture. To avoid variations in
viation calculated during the correlation process is used to estimate the measurements due to temperature (heat from the spotlights)
the accuracy of the results. Assuming a Gaussian distribution of the the lights are on until stabilisation of the ambient temperature
noise in the measurement signal, the calculated standard devia- (between aeration and spotlights) but the specimens are covered
tion σ represents the confidence interval for the match process. The before the beginning of the tests. The recording of the images at a
obtained values for σ in the experiment are always smaller than frame rate of 25,000 fps is triggered using a light intensity trigger
0.025 mm. Similar values are obtained in Ref. 18 in which a differ- oriented towards the explosive (Fig. 12a). The observed area is about
ent method to estimate the accuracy is used. 400 × 400 mm2 and is imaged with 512 × 512 pixels, which gives a

50
True maximum principal strain ε1 (−)

0.014
0.475 ms 0.5 ms
0.5 ms
Displacement (mm)

0.012 40 0.625 ms
0.525 ms
0.01 0.55 ms 0.75 ms
0.575 ms 30 0.875 ms
0.008
0.6 ms 1 ms
0.006 0.625 ms 20
0.004 0.65 ms
0.675 ms 10
0.002 0.7 ms
0 0.75 ms 0
−100 −50 0 50 100 −100 −50 0 50 100
x (mm) x (mm)
Fig. 15. Maximum principal true strain profiles along the x-axis of a panel under Fig. 17. Displacement profiles along the x-axis of a panel under EDST load with a
free air blast for a charge weight of 25 g at 250 mm. charge weight of 15 g.
 M.A. Louar et al./International Journal of Impact Engineering 86 (2015) 111–123 119

one with the free air explosion with a charge of 25 g. The displace-
True maximum principal strain ε1 (−)
0.475 ms ment, strain and the strain rate of three different points located at
0.06 0.5 ms the coordinates (5,5), (15,15) and (25,25) relatively to the axis il-
0.525 ms lustrated in Fig. 14 are compared. The measurements show that with
0.55 ms the EDST, even smaller charges than the free air blast lead to higher
0.04 displacements, deformations and significantly higher strain rates.
0.575 ms This is due to the higher impulse generated by the EDST.
0.6 ms
0.02 0.625 ms 4.3.1. Structural response of the panels
0.650 ms Due to the nature of the pressure distribution of the spherical
blast wave in the free air explosion tests, the response of the fully-
0.675 ms
clamped panel is composed of three stages. First, a compressive
0 0.7 ms through thickness stress wave is generated by the interaction
0.75 ms between the blast wave and the panel. Reaction forces are induced
−100 −50 0 50 100 in the fixed boundaries. At this point, the panel is still static. Then,
x (mm) momentum and kinetic energy are transferred to the panel. The re-
action forces generate a stress wave that propagates from the
Fig. 18. Maximum principal true strain profiles along the x-axis of a panel under boundaries to the centre of the panel (Fig. 15). While propagating,
EDST load with a charge weight of 15 g. this stress wave creates bending and shear deformations. This causes
the panel to move in the z-direction having, first, a square-like shape
magnification factor of about 0.78 mm/pixel. An area of interest (AOI) (due to the square clamping), than with a curved profile as shown
of 200 × 200 mm2 is considered for the correlation calculations. For in Fig. 16, when the maximum deflection point reaches the centre
correlation, a subset size of 21 × 21 pixels is used with a subset of the plate. Finally, the panel elastically vibrates and stabilises in
spacing of 3 pixels and a strain filter of 9. its final deformed shape.
It has to be noted that several in-plane deformations are ob- From a practical point of view, it can be seen in Fig. 16 that the
tained from the DIC, and even if only one of the strains is shown maximum deflection point is not always in the centre of the plate.
at a time for brevity, the results are similar for the rest. And from Fig. 15, it is clear that the strain fields are not symmet-
ric. This is due to the asymmetry of the loading discussed previously.
4.3. Experimental results But, with the multiple reflections of the stress waves, the final state
is symmetric.
Fig. 13 shows a comparison between three measurements with With the EDST load, the response is different from the previ-
two tests performed with the EDST and charges of 15 and 20 g, and ous one. The planar wave is concentrated in the cross-sectional area

−3 −3
x 10
−3 −3 x 10 x 10
x 10 100 4 100 4
100 2 100 2
2 2
y (mm)

y (mm)
0 0
y (mm)

y (mm)

0 0 0 0 0 0
−2 −2 −2 −2

−100 −100 −4 −100 −4 −100 −4

−4 −100 0 100 −100 0 100
−100 0 100 −100 0 100
x (mm) x (mm) x (mm) x (mm)
(a) 0.96 ms (a) 0.96 ms

−3 −3
x 10
−3
x 10
−3 x 10 x 10
100 100 100 100
4 4
5 5
2 2
y (mm)

y (mm)
y (mm)

y (mm)

0 0 0 0
0 0 0 0
−5 −5 −2 −2
−10 −10 −100 −4 −100 −4
−100 −100 −100 0 100 −100 0 100
−100 0 100 −100 0 100
x (mm) x (mm) x (mm) x (mm)
(b) 1.08 ms
(b) 1.08 ms
−3 −3
100 100 x 10 x 10
0.01 0.01 100 4 100 4
2 2
y (mm)

y (mm)

0 0
y (mm)

y (mm)

0 0 0 0 0 0
−0.01 −0.01
−2 −2
−100 −0.02 −100 −0.02
−100 0 100 −100 0 100 −100 −4 −100 −4
−100 0 100 −100 0 100
x (mm) x (mm) x (mm) x (mm)
(c) 1.20 ms (c) 1.20 ms

Fig. 19. True strain ε xy fields from two EDST tests with 15 g of C4 at different time Fig. 20. True strain ε xy fields from two free air blast tests with 25 g of C4 at 250 mm
steps. at different time steps.
120 M.A. Louar et al./International Journal of Impact Engineering 86 (2015) 111–123

a− (10,10) b− (40,40) c− (70,70)

50 50 50

Displacement (mm)
EDST
40 40 40 Free air blast
30 30 30
20 20 20
10 10 10
0 0 0
0 2 4 6 0 2 4 6 0 2 4 6
Time (ms)

0.06 0.06 0.06

EDST
principal strain (−)
True maximum

Free air blast

0.04 0.04 0.04

0.02 0.02 0.02

0 0 0
0 2 4 6 0 2 4 6 0 2 4 6
Time (ms)
Fig. 21. The displacement and maximum principal true strain ε1 histories at different points.

of the tube and hence, the loading time is higher than with the free to be noted that the discontinuities in the fourth quadrant of the
air blast. Due to these two points and to the relative softness of the measurements from the EDST loading in Fig. 19 are due to the re-
aluminium used for the tests, a local indentation is observed on an flection of light that saturates the sensor locally and hence the
area corresponding to the cross-section of the EDST. This response correlation cannot be achieved correctly in that area. These reflec-
is also observed in Ref. 24. First, a circular area starts to move in tions are caused by a change in the curvature of the plates during
the z-direction. Then the motion propagates towards the edges and a test that makes the light from the spotlights reflect towards one
the centre of the plate similarly to an impact loading (Fig. 17). From of the cameras and thus, it is sometimes hard to predict. Never-
a stress/strain point of view, a stress wave is generated at the edges theless, it is generally not a major problem because of the important
of the loaded area then propagates towards the centre and the edges amount of data obtained by the DIC.
of the plate (Fig. 18). The higher momentum applied makes the plate Figs. 16 and 17, and Figs. 19 and 20 (more data available) illus-
reach the maximum displacement faster than in the case of a free trate the better reproducibility of the test conducted with the EDST
air blast and also less vibrations are generated due to high plastic in comparison to the free air loading in terms of distribution and
deformations. It is clearly seen from Figs. 17 and 18 that the dis- value. Fig. 21 presents the displacement histories and the maximum
placement and strain fields on the plate remain symmetric from the principal true stain ε1 histories at three points with coordinates
beginning to the end. (10,10), (40,40) and (70,70) from the tests with both loads. The vari-
ability in the measurements from these points is resumed in Fig. 22
4.3.2. Repeatability of the tests which gives the average variability for each loading and each one
Figs. 19 and 20 allow to make a comparison of the strain ε xy mea- of the previous points. The higher variability in the measure-
sured for the EDST loading and for the free air blast loading. It has ments from the free air loading is mainly due to the uncertainties

60 60
Variability in the displacement (%)

Variability in the true maximum

EDST
50 Free air blast 50
principal strain (%)

40 40

30 30

20 20

10 10

0 0
(10,10) (40,40) (70,70) (10,10) (40,40) (70,70)
Point position Point position
Fig. 22. Comparison between the variability in the measurements (left) for the displacement and (right) for the true maximum principal strain ε1 .
 M.A. Louar et al./International Journal of Impact Engineering 86 (2015) 111–123 121

2000 of the plates should be also symmetric relatively to the same axis.
Test 1 This symmetry, if reached, offers the advantage to limit the record-
Test 2
1800 ing to only one quadrant of the specimens and, hence, to reduce
Test 3
the size of the pixel which increases the accuracy of the measure-
1600 ments and, in a numerical model, only one quarter of the plate has
to be modelled which reduces the calculation time and allows a finer
1400 mesh.
In Fig. 24, it is globally noticed that the EDST tests offer a better
symmetry in comparison with the free air blast tests. Fig. 24 shows
Incident pressure (kPa)

1200
the displacement and the true strain ε yy histories of three points
with the coordinates (5,5), (40,40) and (75,75) and their symmet-
1000
ric points in the three other quadrant of the plate i.e. for the point
(5,5) the three other point plotted are (−5,5), (−5,−5) and (5,−5). It
800
can be seen that the EDST tests have a better symmetry than the
free air blast tests in Fig. 25.
600

400 4.3.4. Validation of the DIC measurements

With the purpose of validating the DIC measurements, three tests
200 are performed using the same conditions described in Section 4.3
with an EDST loading. The specimens are made of a different alu-
0
minium alloy (EN AW-2017A) than the previous one but the
dimensions remain the same. Each plate is partially provided with
0 0.05 0.1 0.15 0.2 0.25 a high-contrast speckle pattern (Fig. 26). In the first quadrant of the
Time (ms)
specimens a strain gauge (FRA-3-23) is placed at the coordinate
Fig. 23. Comparison between pressure histories from three free air blast tests. (75,75) relatively to the axes presented in Fig. 14. The strain gauges
are fixed on the panels following the instructions of the provider
and using a cyanoacrylate adhesive.
in the loading itself related to the imperfect shape of the charge and Fig. 27a shows a comparison between the true strain ε yy mea-
the errors in the positioning of the detonator and the explosive. The surements from the strain gauge at the coordinate (75,75) and the
variability in the loading in the free air blast tests is illustrated in DIC at the coordinates (75,−75) and (−75,−75). The strain gauge mea-
Fig. 23 that shows the (raw) incident pressure measured by a blast surements are in good agreement with the DIC results for the ε yy
pencil placed at 250 mm from the centre of the charge. history. Regarding the strain rate (Fig. 27b), the results from the DIC
show a reasonably good agreement with the strain gauge results.
4.3.3. Symmetry of the response of the plates
Theoretically, in the EDST tests and the free air blast tests, the 5. Conclusion
loading is symmetric relatively to the perpendicular central axis of
the frame. And, since the boundary condition and the specimens The use of an EDST as a loading tool is assessed in this paper.
are symmetric relatively to the axis shown in Fig. 14, the response First, the blast waves generated by an EDST are examined by

a− Point (5,5) b− Point (40,40) c− Point (75,75)

50 50 50
Displacement (mm)

EDST
40 40 40 Free air blast
30 30 30
20 20 20
10 10 10
0 0 0
0 2 4 6 0 2 4 6 0 2 4 6
Time (ms)

0.06 0.06 0.06

True strain εyy (−)

EDST
Free air blast
0.04 0.04 0.04

0.02 0.02 0.02

0 0 0
0 2 4 6 0 2 4 6 0 2 4 6
Time (ms)

Fig. 24. The displacement and true strain ε yy histories at different groups of symmetric points in the four quadrant of the plates.
122 M.A. Louar et al./International Journal of Impact Engineering 86 (2015) 111–123

EDST

Variability in the symmetry of

40 40

Variability in the symmetry of

Free air blast

the displacement (%)

the true strain εyy (%)

30 30

20 20

10 10

0 0
(5,5) (40,40) (75,75) (5,5) (40,40) (75,75)
Point position Point position
Fig. 25. Comparison between the variability in the symmetry of the measurements (left) for the displacement and (right) for the true strain ε yy .

measuring the incident and reflected pressures and impulses with

pressure transducers. The tests show a good repeatability with an
average standard deviation of 10–15%. Then, the dynamic re-
sponse of thin aluminium plate under an EDST generated blast wave
and free air blast are compared. This is performed by means of high-
speed DIC measurements which are also compared to strain gauge
measurements for validation.
With an EDST, the load on the target plate is concentrated onto
an area corresponding to the section of the EDST. The measure-
ments show that the blast wave may be considered planar
during the loading time. It is also found that the pressure–time
history of generated waves can be described with a Friedlander
equation.
The use of an EDST to focus the blast wave and to guide it results
in higher impulses than with free air blast, allowing to reach higher
deformation and strain rate levels. The maximum strain rate ob-
tained with the free air blast load for 25 g of C4 at 250 mm is 80 s−1,
while with the EDST, for 20 g C4 at the entrance of a tube of 1.2 m
length, the strain rate reaches 500 s−1.
Two experimental set-ups are examined: the EDST and the free
air blast. The DIC measurements show differences in the structur-
al response of the panels between the free air blast and the EDST
Fig. 26. Test specimens with a strain gauge. loads. For the free air blast loading, three phases are observed. First

−3
x 10
10 60
DIC − (75,−75) DIC − (75,−75)
DIC − (−75,−75) DIC − (−75, −75)
Strain gauge 40 Strain gauge
8

20
6
True strain εyy (−)

Strain rate (s−1)

0
4
−20

2
−40

0
−60

−2 −80
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Time (ms) Time (ms)

(a) (b)

Fig. 27. Comparison between (a) the true strain ε yy and (b) the strain rates from the DIC and the strain gauge.
 M.A. Louar et al./International Journal of Impact Engineering 86 (2015) 111–123 123

a linear elasto-plastic deformation followed by damped elastic [6] Tekalure SA, Shukla A, Ruggerio P, 2006, Blast loaded thin composite plates –
an experimental study, SEM Annual Conference and Exposition on Experimental
vibrations and, finally, a stabilisation in a deformed state. For the
and Applied Mechanics.
EDST loading, the elastic vibrations are barely obtained due to the [7] Hebert M, Rousseau C, Shukla A. Shock loading and drop weight impact response
high plastic deformation. The stress/strain wave propagation pat- of glass reinforced polymer composite. Comp Struct 2008;84:199–208.
terns are also different. More localised effects result from the EDST [8] Tekalure SA, Shukla A, Shivakumar K. Blast resistance of polyurea based layered
composite materials. Comp Struct 2008;84:271–81.
load. And, as a result, a stress/strain wave is generated at the borders [9] Leblanc J, Shukla A. Dynamic response and damage evolution in composite
of the loaded area and propagates towards the edges and the centre materials subjected to underwater explosive loading: an experimental and
of the plate while, for the free air blast, the stress/strain wave is gen- computational study. Comp Struct 2010;92(10):2421–30.
[10] Pankow M, Justusson B, Salvi A, Waas AM, Yen CF, Ghiorse S. Shock response
erated at the edges and propagates towards the centre. of 3D woven composites: an experimental investigation. Comp Struct
The tests conducted with an EDST show a better reproducibil- 2011;93:1337–46.
ity than the ones with a free air blast. The response of the specimens, [11] Leblanc J, Shukla A, Rousseau C, Bogdanovich A. Shock loading of three-
dimensional woven composite materials. Comp Struct 2007;79:344–55.
in the first case, is also more symmetric than the second one which [12] Langdon GS, Chi Y, Nuricj GN, Haupt P. Response of GLARE panels to blast
offers the possibility to reduce the recorded area and increase the loading. Eng Struct 2009;31:3116–20.
accuracy of the measurements. [13] Soutis C, Mohamed G, Hodzic A. Modelling the structural response of GLARE
panels to blast load. Comp Struct 2011;94:267–76.
Finally, for future work, the full-field DIC measurements in con- [14] Stoffel M, Schmidt R, Weichert D. Shock wave loaded plates. Int J Solids Struct
junction with numerical simulations can be used to estimate the 2001;38:7659–80.
material parameters through an inverse method. An EDST gener- [15] Langdon GS, Nurick GN, Cantwell WJ. The response of fibre metal laminate
panels subjected to uniformly distributed blast loading. Eur J Mech A
ates a more controllable blast wave than a free air blast at a lab-
2008;27:107–15.
scale. Thus, the loading can be accurately modelled numerically [16] Børvik T, Hanssen AG, Langseth M, Olovsson L. Response of structures to planar
which enhances the estimation procedure. blast loads – a finite element engineering approach. Comp Struct 2009;87:507–
20.
[17] Wang E, Shukla A. Analytical and experimental evaluation of energies during
References shock wave loading. Int J Impact Eng 2010;37:1188–96.
[18] Spranghers K, Vasilakos I, Lecompte D, Sol H, Vantomme J. Full-field deformation
measurements of aluminium plates under free air blast loading. Exp Mech
[1] Blaysat B, Florentin E, Lubineau G, Moussawi A. A dissipation gap method for 2012;52:1371–84.
full-field measurement-based identification of elasto-plastic material [19] Spranghers K, Vasilakos I, Lecompte D, Sol H, Vantomme J. Numerical simulation
parameters. Int J Numer Methods Eng 2011;91:685–704. and experimental validation of the dynamic response of aluminium plates under
[2] Moussawi A, Lubineau G, Florentin E, Blaysat B. The constitutive compatibility free air explosions. Int J Impact Eng 2013;54:83–95.
method for identification of material parameters based on full-field [20] Henchie TF, Chung Kim Yuen S, Nurick GN, Ranwaha N, Balden VH. The response
measurements. Comput Methods Appl Mech Engrg 2013;265:1–14. of circular plates to repeated uniform blast loads: an experimental and
[3] Blaysat B, Florentin E, Lubineau G, Moussawi A. Identification of the parameters numerical study. Int J Impact Eng 2014;74:36–45.
of an elastic material model using the constitutive equation gap method. [21] Baker WE. Explosions in air. Austin: University of Texas Press; 1973.
Comput Mech 2010;46:521–31. [22] Smith PD, Hetherington JG. Blast and ballistic loading of structures. Oxford:
[4] Spranghers K, Vasilakos I, Lecompte D, Sol H, Vantomme J. Identification of the Butterworth-Heinemann Ltd; 1994.
plastic behavior of plates under free air explosions using inverse methods and [23] Sutton MA, Orteu J-J, Schreier HW. Image correlation for shape, motion and
full-field measurements. Int J Solids Struct 2014;51:210–26. deformation measurements. US: Springer; 2009 ISBN 978-0-387-78746-6.
[5] Dang X, Chan PC. Design and test of a blast shield for Boeing 737 overhead [24] Kumar P, Leblanc J, Stargel DS, Shukla A. Effect of plate curvature on blast
compartment. Shock Vib 2006;13:629–50. response of aluminum panels. Int J Impact Eng 2012;46:74–85.