Вы находитесь на странице: 1из 10

International Journal of Impact Engineering 94 (2016) 13–22

International Journal of Impact Engineering 94 (2016) 13–22 Contents lists available at ScienceDirect International

Contents lists available at ScienceDirect

International Journal of Impact Engineering

journal homepage: www.elsevier.com/locate/ijimpeng On the problem of bare-to-cased charge equivalency Hezi

On the problem of bare-to-cased charge equivalency

Hezi Grisaro, Avraham N. Dancygier *

charge equivalency Hezi Grisaro, Avraham N. Dancygier * Faculty of Civil and Environmental Engineering, Technion –

Faculty of Civil and Environmental Engineering, Technion – Israel Institute of Technology, Technion City, Haifa 32000, Israel

ARTICLE

INFO

Article history:

Received 6 January 2016 Received in revised form 27 February 2016 Accepted 4 March 2016 Available online 9 March 2016

Keywords:

Cased charge

Blast wave

Casing effect

Equivalent charge

Numerical simulations

Scaling laws

ABSTRACT

Many explosives are covered with a steel casing. The fragmentation process of the casing dissipates part of the detonation energy and therefore cased charges yield lower overpressures and impulses than the same charges without a casing (bare charges). It is often required to assess the mass of an equivalent bare charge, which will produce similar impulses (at the same distances) to those of a given cased charge. Another pertinent parameter is the cased-to-bare impulse ratio, which is a direct measure of the effect of the casing on the resulted impulse. This paper deals with several aspects of the problem of a cased charge equivalency. A review of available models for the assessment of the ratio between the masses of the equivalent bare and cased charges is presented. The current study proposes a procedure to assess the mass ratio, which consists of relatively simple numerical simulations and of the blast waves scaling laws. The simulations are verified against experimental data and their results are compared with avail- able models for the mass ratios. A relation between the mass ratio and the impulse ratio is also presented. Finally, examination of the effect of the casing material properties indicates that the casing-to-charge mass ratio is a key parameter in the assessment of the mass of an equivalent charge. © 2016 Elsevier Ltd. All rights reserved.

1. Introduction

The performance of an explosive charge may be quantified by its pressure–time and impulse–time curves. The impulse is the in-

tegration of the pressure with respect to time and it is considered as a very important parameter in the study and design of protec- tive structures. A common source of explosive and impact load is

a charge with a metal casing. After detonation, the casing expands

and ruptures into many fragments. At this time, the gases are dis- charged through the spaces in the casing and propagate in the air.

Experimental data show that there is a significant difference between the blast wave parameters of cased and bare charges [1–4]. When the charge is cased, part of the detonation energy is dissipated through the expansion and rupture of the casing. As a result, the blast wave parameters, and especially the peak impulse, will be lower than those that are caused by the same charge without casing (bare charge). Analysis of blast-wave parameters that are caused by bare charges is commonly done with experimentally verified numerical simu- lations. Simulations of cased charges, however, are cumbersome and very expensive in terms of computational resources and time. This

is because the casing has to be modeled with a very fine mesh that

* Corresponding author. Faculty of Civil and Environmental Engineering, Technion

– Israel Institute of Technology, Technion City, Haifa 32000, Israel. Tel.: + 972 4 8292487; Fax: +972 4 8295697.

E-mail address: avidan@technion.ac.il (A.N. Dancygier).

http://dx.doi.org/10.1016/j.ijimpeng.2016.03.004

0734-743X/© 2016 Elsevier Ltd. All rights reserved.

requires a very small time step. Yet, the propagation of the blast wave in the air takes relatively a much longer time than the frag- mentation of the casing, and as a result, it is very hard and sometimes impossible to be simulated. This limitation can be overcome in per- tinent analyses by finding a corresponding or ‘equivalent’ bare charge (without casing) that will produce a blast wave with the same pa- rameters as that of the cased charge. As in other engineering problems, scaling a phenomenon is a common and helpful practice [5]. It is generally employed to resolve budgetary limitations of full-scale experiments, and particularly those that involve explosions. The most common scaling form for the latter type of experiments is the “Hopkinson scaling” or “cube root scaling” [5–7], which is based on the “Buckingham π theorem”. The scaling laws can also be used to evaluate the “equivalent charge” (as will be shown in the following text). According to these laws, two similar blast waves (caused by charges that have similar geometries with different dimensions, at the same atmosphere) will produce similar scaled parameters (e.g., impulse) at the same “scaled distance”. The scaled distance is defined by the ratio R/C 1/3 , where R is the dis- tance and C is the mass of the charge. The scaling laws described above are applied in this study in the interpretation of results from a numerical investigation, for the comparison of the blast param- eters of bare and cased charges. It is evident that two similar bare charges with different masses will not produce the same impulse at the same distance. However, the scaling laws show that they will cause impulses that converge to the same scaled-impulse versus scaled-distance curve. When cased charges are considered, a common definition for an equivalent bare

14

H. Grisaro, A.N. Dancygier/International Journal of Impact Engineering 94 (2016) 13–22

charge mass, C e , is that at given scaled distances it produces scaled impulses that are equal to those caused by a given cased charge, C c . I.e., C e satisfies, at any given standoff distance R, the following equations:

I ( C c ) = 13 C e
I ( C
c ) =
13
C
e
⎛ R ⎞ = I ( C e ) → ⎟ 13 ⎠ 13 C
R ⎞
= I ( C
e
) →
13
13
C e
C
e

f ⎝ ⎜

(

IC

e

)=

(

IC

c

)

(1)

where f(R/C e 1/3 ) is a function of the scaled impulse–distance curve. Hutchinson argued that this definition could be confusing [8] and thus he suggested another definition for the estimation of the impulse reduction due to the effect of the casing, i.e., Hutchinson suggested to define the ratio I(C c )/I(C b = C c ), where it is noted that C b is a bare charge. Hutchinson further argued that both ratios, the impulse ratio and the mass ratio (C e /C c ), are equal [8]. The current work examines ways to evaluate the equivalent bare- to-cased mass ratio or the cased-to-bare impulse ratio. It consists of a calculation procedure, which is based on numerical simula- tions and employment of the scaling laws of blast waves. First, we review available models for an equivalent charge, C e . These models are then examined through the proposed numerical procedure. The numerical calculations of this procedure are verified against pub- lished experimental data for bare charges. Then, the procedure is used to evaluate the equivalent bare charge masses and the impulse ratios for particular cases, for which there are also experimental data that allow further verification of the proposed procedure. The re- lation between the impulse ratio I(C c )/I(C b = C c ) and the equivalent bare charge-to-cased charge mass ratio C e /C c is derived from the nu- merical results in order to check Hutchinson assumption (that they are equal). Finally, the effect of the casing material (and therefore, of its different mechanical properties) was examined through the numerical procedure.

2. Available models of an equivalent bare charge

Gurney, in his classic work, showed that the fragments veloci- ty of cylindrical and spherical casings can be estimated by using simple energy balance of the charge and the casing [9]. Fano, as quoted by Fisher [4], followed Gurney’s assumptions and assumed non uniform velocity of the gases (in a cylindrical charge), being zero at the center of the charge and increasing linearly up to a certain value at the charge–metal interface. Using this assumption and a suitable energy balance, he obtained a formula for the equivalent charge mass. In addition, he assumed a factor of 0.8 “to account for the fraction of the total detonation energy belonging to the gases and the case as kinetic energy at time of rupture of the case” [4]. In summary, Fano proposed the following equation for the equiv- alent bare charge:

C 0 . 8 e = 0 . 2 + C 1 + 2 MC
C
0 . 8
e
=
0
. 2
+
C
1
+
2
MC
c
c

(2)

where M is the mass of the casing. Fisher modified Fano’s formula by making a different assump- tion of uniform gases velocity and obtained the following expression

[4]:

C 0 . 8 e = 0 . 2 + C 1 + MC c
C
0 . 8
e
=
0
. 2
+
C
1
+
MC
c
c

(3)

In the same report [4], Fisher modified his own formula and pro- posed the following empirical equation, which agrees better with experimental data:

C 1 + MC ( 1 − M ′) e c = C 1 +
C
1
+ MC
( 1 −
M
′)
e
c
=
C
1
+ M C
c
c

(4)

where M′ is a non-dimensional coefficient, equal to the minimum of M/C c and 1.0 (note that with M′ = 0.8, Eq. (4) converges to Eq. (3)). Hutchinson noted on the above that using the factors “0.8” and “0.2” is redundant, because the derivation already included con- sideration of the kinetic energy that goes to the fragments [10]. He further rightfully noted that according to the Fano and Fisher for- mulas, for very large M/C c ratios (M/C c ∞), the equivalent bare charge does not converge to zero, as expected [11]. Hutchinson proposed another approach to evaluate the equiv- alent bare charge, according to which, the equivalent bare charge should be derived based on the conservation of momentum rather than energy. This approach yielded the following formula for an equivalent cylindrical bare charge:

C

e

C

c

=

0 . 5 0 . 5 + MC c
0
. 5
0 . 5 +
MC
c

(5)

The above models depend only on the casing-to-charge mass ratio. In the same work [11], Hutchinson followed Crowley’s ap- proach [12] to consider also the casing material and explosive type in the estimation of the equivalent bare charge, as follows:

C

e

C

c

=− 1

1

0 . 5 ⎞ ⎟ 0 . 5 + MC ⎠ c
0
. 5
0 . 5 +
MC ⎠
c

f

m

(6)

where f m is a factor that takes into account the casing material yield stress and the explosive type. Hutchinson derived an analytical ex- pression for the factor f m [11], which was not in good agreement with experimental data. Still, he showed that for each set of results with the same casing material and explosive type there is a unique value of f m that yields good agreement. In a later work, Hutchinson changed his approach for the equiv- alent bare charge [8]. He argued that the definition of an equivalent bare charge that will produce the same impulse at the same dis- tance (see Eq. (5)) is confusing, and suggested the use of the ratio between the peak blast impulses from two charges with the same mass – cased and bare, I c /I b (i.e., I(C c )/I(C b = C c )). He used his previ- ous formula (Eq. (5)) to evaluate the impulse ratio instead of the equivalent bare-to-cased charges’ mass ratio, as follows:

I

I

c

b

=

0 . 5 0 . 5 + MC c
0
. 5
0 . 5 +
MC
c

(7)

Hutchinson also showed that his equations, which do not con- sider properties of the casing material and the explosive type (Eqs. (5) and (7)), are valid for very ductile casings. He further argued that these casings are accelerated up to their ideal Gurney velocity before they fracture [8]. In many cases, the casing fractures before it is fully accelerated by the energy available from the explosive, and for these cases Hutchinson proposed the following formula [8]:

I

I

c

b

=

2 γ − 1 ( ) 1 ⎛ R M 0 ⎞ + ⎛ 1
2 γ − 1
(
)
1 ⎛ R
M
0 ⎞
+
⎛ 1
+ M ⎞
2 ⎠
C
⎝ R
2
C c
cf

(8)

where γ is the heat capacity ratio, R 0 is the initial casing radius and R f is the radius at fracture. He mentioned that the radius R f can be estimated by the fracture strain of the casing material (however, he did not provide further details for this estimation). In his veri- fication with experimental results, using high speed cameras, R f / R 0 was assessed to be equal to 2 (i.e., the casing radius was increased to about twice its initial radius) [8]. In another research, Hutchinson [13] extended his approach to consider the casing material yield stress, as follows:

H. Grisaro, A.N. Dancygier/International Journal of Impact Engineering 94 (2016) 13–22

15

Journal of Impact Engineering 94 (2016) 13–22 15 Fig. 1. The experimental setup reported by Dunnet

Fig. 1. The experimental setup reported by Dunnet et al. (Fig. 3 from [2]). (Note that although the authors reported the usage of a gage at 10 m, they ignored its records

[2].)

I

I

c

b

=

γ − 1 1 M ⎛ σ y ⎞ ⎟ + M ⎞ ⎜ 2
γ
− 1
1 M ⎛
σ
y ⎞
+ M ⎞
2 C
c ⎝
γ
⎝ ⎜
⎛ 1
+
P 0
2 C
⎠ ⎟
c

(9)

where σ y is the casing yield stress and P 0 is pressure that depends on the explosive type. This equation was verified with experimen- tal data [13]. Further development of Eq. (9) by the same author yielded a similar formula, which takes into account a strain energy dissipation. However, these energy losses were found to be very small

[14].

Finally, the Unified Facilities Criteria (UFC) [3] recommends that for design purposes, the casing effect can be neglected (i.e., apply a conservative approach).

3. Numerical simulations of bare charges

The first step of the current work is to verify the simulation tech- nique we are using. This has been done for the relatively simpler

case of a detonated bare charge. Experimental data from Dunnet et al. include measurements from tests of both bare and cased charges [2]. The experimental setup is illustrated in Fig. 1. The explosive type of the bare charge was RX1100 and its mass was 1 kg. It had a cylindrical shape with a length-to-diameter ratio of 2, and it was detonated at the cylinder end. Due to the axial sym- metry applied in the experiments, it is possible to use 2D axisymmetric numerical simulations, rather than full and more cum- bersome 3D simulations. In these experiments, pressure gauges that measured the pressure–time history were set at 3, 4.5 and 6 meters from the charge centroid (the authors have also reported the usage of a gage at 10 meters; however, they ignored its records [2]). Thus, in order to compare the measurements at distances up to 6 meters, the computational domain should be larger than 6 meters. Based on the authors’ past experience (e.g. [15]), it was concluded that for this kind of problems, relatively small cells (~ 2 mm) are re- quired in the vicinity of the charge. However, simulations with a computational domain of a few meters with such small cells are ex- pected to be very expensive in terms of computational resources. As a result, the simulation has been divided into two phases: the first phase consisted of a relatively small computational domain (up to 500 mm from the charge center) with a fine mesh, while the second phase consisted of a larger domain (up to 10 meters from the charge center) and a coarser mesh. The results that were re- corded at the end of the first phase were the input for the second one. The simulations were conducted with the Ansys Autodyn hydro- code [16]. The size of the computational domain in the first phase

was 1000 × 500 millimeters, as shown in Fig. 2. The Eulerian mesh has been divided into 500 and 250 cells in the horizontal and vertical directions, respectively, which yielded a 2-mm cell size. The “C4” material model (with the “JWL” equa- tion of state, EOS) was chosen from the Ansys Autodyn material library, to simulate the explosive charge. The C4 and RX1100 charges comprise about 90% and 88% RDX, respectively. Thus, this was the best option for the simulation of the actual explosive material (which proved to be suitable when verified against the reported experi- mental data; see section 4 in the following text). The air has been modeled with the “Air” material model from the Autodyn library with an ideal gas EOS. For further details regarding the material models, the reader is referred to reference [16]. The detonation point in the charge was located at the cylinder end, as in the experi- ment. “Flow out” boundary conditions were applied to all boundaries (except for the symmetry line). Just before the front of the blast wave

for the symmetry line). Just before the front of the blast wave Fig. 2. The first

Fig. 2. The first phase of the numerical simulation.

16

H. Grisaro, A.N. Dancygier/International Journal of Impact Engineering 94 (2016) 13–22

Journal of Impact Engineering 94 (2016) 13–22 Fig. 3. The second phase of the numerical simulation.

Fig. 3. The second phase of the numerical simulation.

reached the boundaries, the results were mapped into the second phase of the simulation. Hence, these boundary conditions did not in-fact influence the results. The size of the computational domain in the second phase was 8 × 10 meters, as shown in Fig. 3, where the material models and the boundary conditions were the same as in the first phase. Nu-

1- and 0.477-kg charges. As expected, the lower mass yielded lower impulse values. However, when the scaled impulses are plotted against their scaled distances, the curves of the two masses fall on the same line, as can be seen in Fig. 7b, and hence, this result serves as an additional verification of the current calculations. It should be noted that any mass other than 477 gr should have yielded the

6).

merical gauges were placed along a line, perpendicular to the axis of the charge, as shown in Fig. 3. The blast wave is expected to prop- agate in the air, and at some point to reach the boundaries. At this moment, some of the material “flows out” out of the Eulerian mesh.

same (scaled) result. Here, the 477-gr mass has been chosen for the subsequent calculations of a cased charge (as will be shown in section

However, the numerical gauges, as noted above, are located along

5.

Numerical simulations of cased charges

a central vertical line, and it is therefore reasonable to assume that they are far enough to be affected from this material loss.

4. Results and verification

The first simulation was of a 1-kg charge (as in the experi-

ments). The propagation of the blast wave in the air in the first and second phases of the simulation is illustrated in Fig. 4, in terms of pressure contours. Fig. 5 shows an example of the overpressure (rel- ative to the atmospheric pressure) time history from a numerical gauge, located 3 meters from the charge center (see gauge #31 in Fig. 3). The overall shape of the calculated curve is as expected, i.e., zero overpressure until a certain (arrival) time, and then a sharp overpressure increase followed by an exponential decay and a neg- ative phase. The integration of this curve yields the blast wave impulse at this point. Similarly, the peak impulses were calcu- lated from numerical gauges that were set at various distances (here, every 100 mm; see dashed line in Fig. 3). The results of the simulation were first verified against pub- lished experimental measurements at various distances from the charge [2], as shown in Fig. 6. It is evident from the figure that there

is a good agreement between the calculated and measured peak im-

pulses. A second verification has been made by applying the scaling laws as follows: according to the scaling laws, two charges with dif- ferent masses will produce the same scaled impulse at the same scaled distance. Thus, an additional simulation with a charge of the

same type and shape as those of the 1-kg charge, but with a dif- ferent mass of 477 gr, has been conducted, using the same technique. Fig. 7a shows the absolute values of the resulted impulses from the

After the detailed simulations of bare charges and their verifi- cations, an effort has been made to consider the casing effect for the cases that were examined. It is noted that 3D simulations that include detailed modeling of the fragmentation process are very ex- pensive in terms of computational resources and time, because they require a very small cell size to properly model the casing materi- al and obtain correct fragmentation results (e.g. [15,17]). This requirement leads to the use of very small time steps and to time- consuming simulations. Yet, because this study has focused on the characteristics of the blast wave, detailed simulation of the casing fragmentation can be left out and therefore it is out of the scope of this work. Thus, although 2D simulations cannot model the casing fragmentation (which is clearly non-symmetric), they were applied in this study. Hence, for the calculations of the cased charges, a two- phase simulation, similar to that described above, has been conducted, with an additional Lagrangian mesh for the casing. The cell size in the Lagrangian mesh was relatively large, and it varied between 3 and 5 cells along the casing thickness in all the cases that were analyzed. As will be shown, after detonation the casing expands and at some point of time, the stresses and strains in the casing ma- terial are relatively high, leading to “numerical erosion” of the casing material. Although “erosion” is a numerical parameter (rather than a physical one), it enables modeling of the gases escape through the spaces generated in the casing. This is an adequate numerical pro- cedure especially because from the point of time when the casing ruptures, propagation of the blast wave is of the main interest, rather than the casing fragmentation itself. At this point, the first phase of the analysis ends. At the second phase, the results of the first phase

H. Grisaro, A.N. Dancygier/International Journal of Impact Engineering 94 (2016) 13–22

17

t=0.05 ms t=0.1 ms (a) t=4 ms t=6 ms
t=0.05 ms
t=0.1 ms
(a)
t=4 ms
t=6 ms

(b)

Fig. 4. Contours of overpressure in the (a) first and (b) second phases.

of overpressure in the (a) first and (b) second phases. Fig. 5. Overpressure and impulse time

Fig. 5. Overpressure and impulse time histories at 3 m from the charge.

and impulse time histories at 3 m from the charge. Fig. 6. Comparison between simulation results

Fig. 6. Comparison between simulation results and experimental data from Dunnet et al. [2].

18

H. Grisaro, A.N. Dancygier/International Journal of Impact Engineering 94 (2016) 13–22

100 90 1 Kg 80 0.477 Kg 70 60 50 40 30 20 10 0
100
90
1 Kg
80
0.477 Kg
70
60
50
40
30
20
10
0
Peak impulse (kPa-ms)

23456789

distance (m)

(a)

20 10 0 Peak impulse (kPa-ms) 23456789 distance (m) (a) (b) Fig. 7. Numerical results of

(b)

Fig. 7. Numerical results of the 1- and 0.477-kg charges: (a) absolute values (b) scaled values.

are mapped as initial conditions of the following analysis, where the casing is no longer considered, and the calculation continues according to the above process. In this way, the effect of the casing can be taken into account in the simulations with a minimum burden on the required com- putational resources. Verification of this methodology has been done with the measured impulses of cased charges that were reported by Dunnet et al. [2]. Fig. 8 shows an example of the first phase mesh, which includes a Lagrangian mesh for the casing. In the experi- ment, the casing was made from EN24 steel. In the absence of further details, the “steel 4340” material model from Autodyn library [16] was chosen to simulate the casing, because the yield stresses of the two types of steel are very similar (~ 700 vs. 790 MPa). This model employs the Johnson–Cook material model [18], with its constant parameters taken from the program’s default values (which are iden- tical to those proposed in [18]). An additional erosion criterion of a geometric strain = 5.0, has been adopted for this material. This rel- atively large value of the erosion criterion enables significant

distortion of the casing before its rupture starts the fragmentation process [19]. As mentioned in section 2, works on this subject indicate that when the casing is relatively ductile its rupture occurs at a rela- tively late time. At this phase, the velocity of the casing has already been stabilized at its maximum value and then the casing contin- ues to expand until it ruptures. In these cases (as opposed to brittle casings), a large portion of the detonation energy goes to the dis- tortion of the casing as kinetic energy (see also a detailed explanation by Hutchinson [10,13]). If however the casing is relatively brittle, fragmentation occurs at a relatively early stage, before the cas- ing’s velocity has reached a maximum value. In this case, less energy is dissipated as kinetic energy that accelerates the casing and the blast wave has higher impulse than that caused by a more ductile casing. This explains why in the former case of ductile casing the most dominant parameter is the M/C ratio. Indeed, in most avail- able models this ratio is the only parameter that controls the equivalent charge mass. Furthermore, in Hutchinson’s model [13],

charge mass. Furthermore, in Hutchinson’s model [13] , Fig. 8. Model of the first phase of

Fig. 8. Model of the first phase of the simulation with a casing.

H. Grisaro, A.N. Dancygier/International Journal of Impact Engineering 94 (2016) 13–22

19

Journal of Impact Engineering 94 (2016) 13–22 19 Fig. 9. Comparison of the peak impulses of

Fig. 9. Comparison of the peak impulses of the cased and bare charges with exper- imental data.

of the cased and bare charges with exper- imental data. Fig. 10. Calculation methodology of the

Fig. 10. Calculation methodology of the equivalent bare charge.

which takes into account other parameters (see Eq. (9)) the M/C ratio is still most dominant. The results of our simulations show that indeed fragmentation started after the casing velocity has been sta- bilized, which indicates a ductile behavior of the casing. Comparison of the numerical results with the experimental data is shown in Fig. 9. In all cases, the charge mass was 1 kg, and hence, the graph shows the scaled, as well as the absolute values (in SI units). It is evident that there is good agreement between the sim- ulation results and the experimental data. Note that the hollow circular markers in Fig. 9 represent the bare charge and that they were already plotted in Fig. 6. As expected, lower impulse values were obtained for larger casing masses, (for the same charge mass). In summary, the above methodology has been found to be reli- able and relatively effective in terms of computational time and resources for simulations of cased charges.

6. Assessment of an equivalent bare charge

As noted above, using the equivalent bare charge allows a sim- plified analysis for the estimation of blast wave parameters caused by cased charges. It is possible to numerically assess the equiva- lent bare charge mass, by applying the blast waves scaling laws, according to the following methodology: First, the impulse is cal- culated from the numerical simulations at various distances (i.e., at the locations of the numerical gauges), as described in section 3. Second, the impulses and the distances are scaled according to the scaling laws (they are divided by C 1/3 , where C is the charge mass) in order to generate curves of the scaled impulse against scaled dis- tance. This is done for both the bare and cased charges, where the curve of the bare charge serves as a “reference line”. It is noted that many studies refer to “TNT equivalent” parameters. However, this equivalency refers mostly to comparisons between the effects of various types of explosives. In this study, one needs to know the equation of the scaled impulse–distance curve of a reference bare charge (see following text). Therefore, instead of using a “refer- ence TNT curve”, and since we already have the results of the bare C4 charge, we can use it (directly) as the “reference curve” (and there is no need to refer to a “third” reference curve). Once a reference line is known, “equivalency factors” can be cal- culated by solving the following equation for C e , for all points on the scaled impulse–distance curve of the cased charge (refer also to Fig. 10):

I c 13 C e
I
c
13
C
e

= f

⎛ R ⎞ ⎜ ⎟ ⎝ 13 C ⎠ e
R ⎞
13
C
e

(10)

where I c represents values of the cased charge impulse at loca- tions set at distances R, f is the function of the “reference line” (i.e., of the bare charge scaled impulse vs. distance curve), C e is the mass of the equivalent charge, which solves Eq. (10), and the “equiva- lency factor” is equal to C e /C c (where C c is the mass of the cased charge). This solution assures that the equivalent bare charge C e , which substitutes the given cased charge, will produce impulses at given distances that, when scaled with respect to C e , will fall on the “ref- erence line”. This methodology is illustrated in Fig. 10. In order to find the above solution of the equivalent bare charge, the reference line needs to be known. The numerical results of the 1-kg bare charge were in good agreement with the experimental data (see section 4) and they could, therefore, be used as the ref- erence line. A fitted curve has been generated with the least square method for the calculated points of the bare-charge and its ana-

B , where A and

lytical expression takes the form I C

13 bb
13
bb

=

(

13 b
13
b

ARC

)

B are equal to 174.121 and 0.9088, respectively, with a very high coefficient of determination, r 2 = 0.999. Fig. 11 describes the equivalency factors C e /C c (where C c = 1 kg) as a function of the scaled distance, calculated with the above meth- odology, from the numerical results for three masses of the casing, 0.5, 2 and 5 kg. For the range used in the simulations, the figure in-

5 kg. For the range used in the simulations, the figure in- Fig. 11. Equivalency factors

Fig. 11. Equivalency factors (C e /C c ) at various distance.

20

H. Grisaro, A.N. Dancygier/International Journal of Impact Engineering 94 (2016) 13–22

Journal of Impact Engineering 94 (2016) 13–22 Fig. 12. Impulse against distance for the 477-gr bare

Fig. 12. Impulse against distance for the 477-gr bare charge and 1-kg charge with 2-kg casing.

the 477-gr bare charge and 1-kg charge with 2-kg casing. Fig. 14. Impulse ratios at various

Fig. 14. Impulse ratios at various distances.

dicates unique equivalency factors of 0.770, 0.477 and 0.337 for the 0.5, 2 and 5 kg bare charges, respectively (except for a minor de- viation at small distances for the 0.5-kg charge). Thus, a 1-kg charge, with a 2-kg casing is expected to produce the same impulses at the same distances as those caused by a 477- gr bare charge. The results of the simulation of a 477-gr bare charge are presented in section 3 and they are compared in Fig. 12 with the results of the 1-kg charges with the 2-kg casing. It is evident from the figure that the two sets of results converge to the same curve. This provides an additional verification for the method pre- sented above for the evaluation of the equivalent bare charge.

7. Comparison with available models

The above results of the equivalent bare charge masses are com- pared with available models in Fig. 13. As mentioned above, Hutchinson proposed an analytical expression for the coefficient f m in Eq. (6), yet predictions of this equation did not agree with ex- perimental data. However, Fig. 13 shows that Eq. (6) is still in good agreement with experimental data for a unique, fitted value of 0.87 for f m , as well as Fisher’s formula (Eq. (3)) and Hutchinson’s other model (Eq. (5)). Note that Eqs. (5) and (7) are identical (and were proposed by the same author). However, while Eq. (5) predicts the

equivalency factor for the mass of the charge, Eq. (7) predicts the impulse ratio (I c /I b ). All of the charge masses were equal to 1 kg in all simulations (and only the casing mass was changed). Therefore, it is possible to calculate the cased-to-bare charge impulse ratios. These I c /I b ratios are plotted in Fig. 14, which indicates a rather constant value for each mass of the casing. The calculated I c /I b ratios are compared in Fig. 15 with the three models proposed by Hutchison (Eq. (7), (8) and (9)). It can be seen in the figure that Eq. (7) is not in good agree- ment with the (verified) numerical results. Note that the same equation, for the prediction of the equivalent mass (Eq. (5)) did yield good agreement with the numerical results (as shown in Fig. 13). As for Eq. (8), there is no available analytical expression for the ratio between the initial casing radius and the radius at fracture, R 0 /R f , of the casing. Yet Hutchinson mentioned that from high-speed camera recordings, this ratio was about 0.5 (i.e., the casing radius at fracture was double that of the initial radius) [8]. Therefore, Eq. (8) has been plotted in Fig. 15 with R 0 /R f = 0.5, yet it is still not in good agreement with the numerical results. The yield stress of the casing material and the parameters P 0 and γ should be known for Eq. (9). Several sources mention that γ = 3 for this kind of problem [8,11,13,20] and therefore this value was adopted here. For an ex- plosive that comprises 88% RDX and 12% wax, P 0 was approximated

comprises 88% RDX and 12% wax, P 0 was approximated Fig. 13. Comparison of the equivalent

Fig. 13. Comparison of the equivalent bare charge factor with available models.

of the equivalent bare charge factor with available models. Fig. 15. Impulse ratio from the numerical

Fig. 15. Impulse ratio from the numerical results and available models.

H. Grisaro, A.N. Dancygier/International Journal of Impact Engineering 94 (2016) 13–22

21

as 11 GPa in Reference 13, and this value has been adopted here as well. In Dunnet et al. [2], the yield stress of the metal casing was not given and therefore, in the current calculations, it was taken as the default value from Autodyn’s Johnson-Cook constitutive model (0.792 GPa). Using the above values, Eq. (9) is found to be in good agreement with the numerical results (Fig. 15). Hutchinson [13] ap- proximated that the yield stress of the steel case in Dunnet et al. [2] was 0.95 GPa (based on stress hardening of the material due to

effects of strain rate and the temperature). Thus, Fig. 15 includes also

a plot of Eq. (9) with σ y = 0.95 GPa. As it can be seen, there is no significant difference between the two curves that were obtained with the two yield stresses.

7.1. Relation between masses and impulse ratios

Essentially, there is a relation between the equivalent bare charge- to-cased charge mass ratio, C e /C b , and the impulse ratio I c /I b . An analytical expression for this relation can be derived, if the func- tion of the scaled impulse against the scaled distance is known for the bare charge. In the current bare charge simulation, it was found that the scaled impulse takes the following form (see section 6):

I b 13 C b
I
b
13
C
b

= A

⎛ R ⎞ ⎜ ⎟ ⎝ 13 C ⎠ b
R
13
C
b

B

,

(11)

where A and B are constants (e.g., for the above simulations we found that A = 174.121 and B = 0.9088). The scaled impulse and dis- tance for a given cased charge do not lie on the same curve of the

bare charge. However, when they are scaled with respect to the equivalent charge C e (rather than C c ), they satisfy Eq. (11). Hence,

it is possible to write the following expression for the cased charge:

I c 13 C e
I
c
13
C
e

= A

⎛ R ⎞ ⎜ ⎟ ⎝ 13 C ⎠ e
R
13
C
e

B

(12)

Algebraic manipulations on Eqs. (11) and (12) lead to the fol- lowing formula:

I

I

c

b

=


C

C

e

b


3

1

(

B 1

+

)

(13)

Specifically for the current numerical results (B = 0.9088), this expression becomes:

I

I

c

b

=

C

⎜ ⎝

e

C

b

⎟ ⎠

0.6363

(14)

Eq. (14) shows that the impulse ratio and the mass ratio (of the equivalent and cased charges) are not similar, as opposed to Hutchin- son’s assumption.

8. Effect of the casing and explosive materials

leading to the escape of the explosive products through the frag- mented casing, and consequently, to higher impulses [8,13]. Results reported by Dunnet et al. demonstrate that there is a com- bined effect of the explosive type and the casing material on the equivalent charge [2]. These researchers conducted experiments with two types of explosive and two casing materials, steel and alumi- num (which has a lower yield stress). Although they did not report measured impulses, they did provide the equivalent bare charge factors for M/C = 0.5 and 2. When they used an explosive type

RX1100, these reported factors for the aluminum casings were higher than those reported for the steel casings (of the same casing masses). However, when another explosive type was used (RX1400), the re- ported factors for the aluminum casing were lower than those of the steel casing. Thus, in view of these opposite trends (Fig. 11 in Reference 2) in the experimental results, Hutchinson’s assump- tion regarding the effect of the casing ductility and yield strength [8,13] is not conclusive. Hence, it is interesting to check the numerical simulation predictions when the casing material model is changed. The material model “ALUMINUM” from Autodyn library was chosen to simulate the aluminum casing [16]. The model includes von Mises yielding criterion and the failure criterion was chosen to be

a 0.2 plastic strain. The above described two-phase simulation

technique has been adopted here to calculate the impulses at various distances from the charge. It should be noted that because the density of aluminum is lower than the density of steel, thicker casing had to be applied in order to yield the same casing mass (as it was in Dunnet et al.’s experiments [2]). These simulations were performed with 0.5, 2 and 5 kg aluminum casing masses. Unexpectedly, the impulses at various distances were the same for both types of casings, steel and aluminum (with maximum differences below 1%). One simulation, of the 2-kg aluminum casing mass, was calculated also with different values of the material yield stress (Autodyn’s default value of 290 MPa, as well as 100, 500 and 1000 MPa). In addition, a failure criterion of plastic strain of 0.12 was also examined. All of these modifica- tions led to very similar results (with up to 1% error). It is noted, however, that although the aluminum and steel casings that were examined in these simulations had different thicknesses and

different material models, they had the same total mass. In summary, our simulations lead to a clear observation: the casing mass has a pronounced influence on the impulses caused by the charge. This can be seen by the (almost) identical results that were obtained from simulations of two geometries (thicknesses) and material models, but with the same casing mass. These results were

not changed when the yield stress or the failure plastic strain cri- teria were modified (as opposed to Hutchinson’s predictions [8,13]). Yet, as stated above, in Dunnet et al.’s experimental data, the usage

of one type of explosive has led to higher values of an equivalent

bare charge, while lower values were obtained with another type

of explosive. Therefore, in our opinion, this issue has not been fully

resolved, and it needs further study.

Experimental data [2] and analytical models [11,13] suggest that the casing material and the explosive type also have an important influence on the casing effect (especially when the casing is not ductile), and not only the M/C ratio (as suggested in Eqs. (2)–(5)). According to Reference 21, the explosive type was found to influ- ence the casing effect, especially for highly oxygen-deficient compositions such as explosives that are partially filled with alu- minum. These explosives generate additional blast energy by interaction of flammable gases with the air. Hutchinson referred to the effect of the casing material on the blast wave [8,13]. He argued that when the casing is less ductile, its effect (which can also be a result of its relatively low yield strength) is manifested through ex- pansion and subsequent fragmentation at a relatively early stage,

9. Summary and conclusions

This paper deals with several aspects of the problem of a cased charge equivalency. A review of available models for assessment of the mass ratio is presented. Another pertinent parameter is the cased-to-bare impulse ratio, which is a direct measure of the effect of the casing on the resulted impulse. A numerical study is pre- sented of the ratio between the masses of equivalent bare and cased charges, which is required to produce similar impulses by both of them. A 2D numerical technique has been adopted to consider the charge and the casing material. First, simulations of bare charges were verified by experimental measurements and by the scaling laws of blast waves. Next, simulations of cased charges were per-

22

H. Grisaro, A.N. Dancygier/International Journal of Impact Engineering 94 (2016) 13–22

formed and their results were also verified by experimental data. The equivalent bare charge has been calculated from these results and by using the scaling laws. Thus, further verification of the cased charge simulation results was available by their comparison with results of the equivalent bare charge simulation. Based on these results, a relation between the equivalent charge ratio and the impulse ratio has been derived. Finally, simulations with different casing materials were performed to examine their possible effect on the mass of an equivalent charge. The following conclusions can be derived from the current study:

1. 2D axisymmetric models can be used to simulate the propaga- tion in the air of a blast wave from a cased charge. They are not suitable for the fragmentation of the casing. However, the frag- mentation process is out of the scope of the current study.

2. The simulation results have been compared with available models, where most of them consider only the effect of the casing-to- charge mass ratio. Fisher and Hutchinson’s formulas (Eqs. (3) and (5)) were found to be in very good agreement with the simu- lations’ data.

3. A derivation process for the relation between the equivalent bare- to-cased charge mass ratio, C e /C b , and the impulse ratio I c /I b is presented in Eq. (14). These two ratios are not equal to each other (as hypothesized in previous studies).

4. Simulations with an aluminum casing with the same mass as that of a steel casing (and therefore, of different thicknesses) yielded the same results. This indicates that the casing mass (nor- malized with respect to the mass of the charge) is a very dominant parameter in the blast performance and in the deter- mination of the equivalent bare charge. Furthermore, the results of the simulations were not sensitive to various values of the yield stresses that were used in the material model of these simula- tions. Experimental observations could have contributed better understanding of the material effects. However, the limited avail- able experimental results show opposite trends regarding the effect of the casing ductility and yield strength. Thus, the results of the current study, together with conflicting trends reported in the literature, indicate that the effects of the casing and ex- plosive properties require further investigation.

References

[3]

8.

9.