Study on the Pressure Characteristics of Shock Wave Propagating across Multilayer Structures during Underwater Explosion

.e propagation of the shock wave across multilayer structures during underwater explosion is a very complex physical phenomenon, involving violent fluid-structure interaction (FSI) problems. In this paper, the coupled Eulerian–Lagrangian (CEL) method in AUTODYN is used to simulate the process of shock wave propagation and solve FSI problems. Firstly, the governing equation and the treatment of fluid and structure interface of the CEL method are briefly reviewed. Afterwards, two underwater explosion numerical models are established, and the results are compared with the empirical formula and experimental data, respectively, to verify the reliability of numerical solutions. .e results obtained by this method show good agreements with those of the empirical formula and experiment. Furthermore, the model of the multilayer structures composed of two hemispherical shells and the fluid filled between the shells subjected to underwater explosion is established, and the pressure characteristics of the shock wave propagating across the multilayer structures are analyzed regarding the wave reflection and transmission. Finally, the effects of the shell thickness and the filled fluid type among the multilayer structures on the wave reflection and transmission are studied.


Introduction
e shock wave produced by underwater explosion can cause serious damage to submarines and threaten survivability of submarines [1].erefore, knowledge on the pressure characteristics of shock waves is of particular importance to an overall understanding of the shock resistance of submarines.Although numerous studies have been conducted [2][3][4][5], there are still many tough problems to solve due to the violent interactions between the structure and the shock wave, especially for the understanding of the wave propagation in multilayer structures.Among them, the double-layer hemispherical shell is often used as the protective structure of submarines and other underwater vehicles.Hence, in this paper, the pressure propagation characteristics of the underwater explosion shock wave in two hemispherical shells with air or water among them are investigated.
Regarding shock wave propagation in structures, a great deal of researches have been performed.Taylor [6] assumed that the underwater explosion shock wave is a onedimensional planar, exponentially decaying pressure wave and adopted this planar wave as shock load to calculate the pressure and impulse on the wetted surface (the interface of water and structure).Although Taylor's solution can give accurate pressure time histories for a simple plate, it is difficult to capture the reflected wave from the multilayer structures.On this basis, Jin et al. [7] presented an analytical method to obtain interfacial pressure of coated plates under underwater weak shock waves.e method can take into account multiple reflections and transmissions in multilayer structures.But the major limitation to this method is that the shock wave should be weak enough to guarantee some assumptions they adopted that rigid body motion is negligible and the variation of the wave propagation time in each medium caused by the medium deformation is negligible.Recently, Chen et al. [8] applied one-dimensional cavitation theory and section-varying bar theory to discuss the propagation characteristics of stress waves in simple plates and stiffened plates (T profiles and I profiles).Considering the cavitation effect, the pressure and impulse on the wetted surface are predicted accurately by analytical solution.ey found that obviously pressure drops and oscillations occur on the wetted surface in the stiffened plates model due to the superposition of the reflected wave from the structure and the incident wave compared to the simple plate model.However, this analytical model is limited to the one-dimensional system.
Valuable researches on pressure characteristics of the shock wave in structures using numerical simulation have also been conducted.Otsuka et al. [9] utilized LS-DYNA3D, an analysis code using the arbitrary Lagrangian-Eulerian approach, to simulate the propagation process of the explosion shock waves produced by the two modes of detonation (a detonation cord and an emulsion explosive) in the three layers of air, water, and structure.ey found that as the air layer becomes thick, the peak pressure decreases in the water.However, the impulse remains unchanged since the duration of pressure is increased by the effect of a reflected wave.Yet, their simulation lacks a comparative study between the different material layers.Desceliers et al. [10] presented a new numerical hybrid method dedicated to the simulation of the transient elastic wave propagation in multilayer unbounded media, which can be fluid or solid, subjected to given transient loads. is method can capture clearly some profiles of the elastic waves including reflected waves, transmitted waves, and refracted surface waves.However, it is restricted to simple geometrical structures.Wang et al. [11] used the coupled Eulerian-Lagrangian (CEL) method in AUTODYN to investigate the shock wave propagation characteristics near the air-backed plate and water-backed plate.ey found that two pressure peaks occurred in these two numerical models, where the first peak is generated by the incident wave and the second peak is induced by the reflected wave from the plate due to the impedance of the water being lower than that of the plate.But in that work, they did not apply the CEL method to multilayer structures.Wu et al. [12] investigated the protection effects of a mitigation layer on the structure exerted by an underwater explosion with the help of the Mie-Gr€ uneisen mixture model and discussed the affections of the layer material (steel, neoprene, and polyethylene), layer thickness, and explosive-structure distance on protection effects.ey concluded that the acoustic impedance of the mitigation layer plays an important role in determining not only the property of shock loads but also the occurrence of protection effects of the mitigation layer, while the layer thickness and explosive-structure distance have a little effect on the main shock loads.Nevertheless, the loads of the inner shock wave in the mitigation layer are their main concern.In addition, other scholars mainly focus on structural response and damage characteristics [13][14][15][16][17][18][19][20][21].In a word, theoretical research based on certain assumptions is suitable for a structure with a simple geometry, while numerical simulation is easier to implement to deal with shock wave propagation problems in complex structures.
Owing to the presence of complex superposition of multiple waves, geometry with curvature, moving interface, and large plastic deformation, traditional theoretical research comes across many difficulties in solving shock wave propagation characteristics in multilayer structures.In order to tackle these problems, the CEL method in AUTODYN is adopted to study the propagation characteristics of the shock wave in the multilayer structures during underwater explosion in the present work.e shell thickness and the filled medium between the double shells are focused, and their influences on the propagation of the shock wave are analyzed.e relationships among the incident, the reflected, and the transmitted shock waves are discussed.

Theoretical Background
2.1.Coupled Eulerian and Lagrangian Method.Actually, the process of underwater explosion near structures is a complicated fluid-structure interaction (FSI) problem [22,23].Two sorts of methods are usually used to solve this problem, namely, the Lagrangian method and Eulerian method [11,24].Both of them have their own advantages and limitations.e Lagrangian method is suitable for describing the stress, strain, and deformation of the structure.However, when encountering large deformation problems, the Lagrangian mesh is severely distorted, which leads to adverse effects on calculation.Conversely, the Eulerian method can deal with large deformation problems and some flow problems.But, it cannot track accurately material interfaces.erefore, a coupled Eulerian-Lagrangian (CEL) analysis method which combines the advantages of these two methods is introduced to tackle FSI problems.e structure part is solved by the Lagrangian method, while the fluid part is solved by the Eulerian method.e FSI between the two parts is carried out at the interface [11,25]. is approach has been made possible in some hydrocodes, e.g., AUTODYN [25] and ABAQUS [26].Specifically, in the Lagrangian method based on the finite difference scheme, the material is attached to the mesh element which deforms with the flow of the material, and the mass of the element remains constant [25].In the Lagrangian solver, given the node position x � (x, y, z), the node velocity u, the element density ρ, the element volume V, the internal energy e, and the governing equations are as follows [25,27,28]: where ρ 0 , V 0 , m, and σ are the initial density, initial volume, mass, and total stress tensor, respectively.e stress tensor is separated into a hydrostatic component and a deviatoric component, which can be referred from the study in [29].

Shock and Vibration
A series of calculations in a Lagrangian subgrid are shown schematically in Figure 1 (the left column).A detailed description about this process can be referred from the study in [25,[30][31][32].
In the Eulerian solver, assuming that the fluid is inviscid, the explosion process is adiabatic and the body force is neglected [27,28,33,34].e governing equations can be written as follows [35,36]: where ρ is the density, u is the velocity, E is the total energy, P is the pressure, ρu is the mass flux, ρuu is the momentum flux, and ρEu is the energy flux.
In fact, the Eulerian method contains two steps to account for both the changes in the element solution caused by the source and the transport of the material through the mesh.More details can be found in [32,[37][38][39].
Besides, the calculation process of the Eulerian solver is shown in Figure 1 (the right column).ere are also many complicated works that need to be accomplished including the calculation of intersection points, intersecting lines, common areas, surface area, and three-dimensional volume.

Treatments of Fluid-Structure
Interface.e interaction between the structure and fluid is performed at the coupling surface.At the interface, the structure described by the Lagrangian view is regarded as the geometric and velocity boundary for the fluid.Conversely, the fluid is regarded as the pressure boundary for the structure [25,32,37].
For the Lagrangian structure element, the position x, velocity u, force F, and mass m are defined at the node, while the stress σ, strain ε, pressure P, energy e, and density ρ are defined at the center of the element.For the Eulerian fluid element, all variables are defined at the center of the element except the position x (Figures 2(a) and 2(b)). is definition mode facilitates the coupling between the Lagrangian structure and the Eulerian fluid [25].

Numerical Validation.
Firstly, the free-field underwater explosion model is established to validate the feasibility of the CEL method in AUTODYN [25] for simulating the shock wave pressure.e spherical TNT charge with a mass m r of 0.853 kg and a radius R 0 of 0.05 m is detonated in the cylindrical domain with a radius of 20R 0 and a height of 50R 0 , as shown in Figure 3(a).A nondimensional distance r hereinafter is defined to describe the distance from the measuring points to the TNT center, namely, r � R/R 0 , where R is the detonation distance.Different distances of r � 10 − 16 are chosen.e transmitting boundary condition is set around the computational domain to reduce the nonphysical reflection.In the present work, we adopt a two-dimensional axisymmetric model to simulate the process of free-field underwater explosion.
e density of TNT is 1,630 kg/m 3 , and the detonation velocity is 6,930 m/s.Besides, the state equation of Jones-Wilkins-Lee (JWL) hereinafter is used to simulate the TNT detonation process, and it can be written as equation (3), where P is the pressure produced by the explosive charges, V is the specific volume of detonation products over the specific volume of undetonated explosives, E is the specific internal energy, and A, B, R 1 , R 2 , and ω are material constants listed in Table 1 [25,44].e equation of state for water hereinafter can be defined as equation (4), where μ � ρ/ρ 0 − 1, in which ρ 0 is the initial density of water, i.e., ρ 0 � 1, 000 kg/m 3 .
e term e is the specific internal energy for water.A 1 , A 2 , A 3 , B 0 , B 1 , T 1 , and T 2 are constants defined in the AUTODYN material library.ese parameters are listed in Table 2 [25,44]: e water and explosive are modeled by the Eulerian subgrid.In order to test the convergence of the present CEL method, three mesh resolutions are used in the Eulerian part, 2.5 mm, 5 mm, and 10 mm, respectively.An evenly partitioned grid is used.e shock wave pressure versus time at r � 10 is measured, as shown in Figure 4. Clearly, the result of the mesh size 2.5 mm shows a little difference compared with the result of the mesh size 5 mm.e convergence rate of three resolutions is calculated to be 1.95 using their peak pressure.Generally, if the convergence rate is over 1, then it means this numerical method can give better results [45].In order to reduce the computational cost, the element size for this free-field underwater explosion model is 5 mm, and the number of elements is about 100,000.
e attenuation of shock wave pressure varying with time p(t) is given by the Zamyshlyayev empirical formula [3]:

Shock and Vibration
where P m is the peak pressure of the shock wave (Pa), θ is the time decay constant of the shock wave (s), W is the weight of the explosive charge (kg), R is the distance between the explosion center and measuring point (m), R 0 is the initial radius of the explosive (m), t p is the time when the bubble reaches its maximum radius (s), P h is the hydrostatic pressure near the explosion center (Pa), P atm is the atmospheric pressure (Pa), C is the sound speed of water (m/s), and h is the initial depth of the explosion center (m).e numerical model of free-field underwater explosion and the pressure contours at two typical instants are shown in Figure 3. e shock wave propagating in water is spherical (Figure 3(b)).When the shock wave arrives at the boundary, it is not reflected due to the enforcement of  [25,32,37].σ, ε, P, e, ρ, x, u, F, and m are the stress, strain, pressure, energy, density, coordinate, velocity, force, and mass, respectively.
4 Shock and Vibration the transmitting boundary condition, as shown in Figure 3(c).e time histories of the pressure at r � 10 and r � 13 obtained from CEL and Zamyshlyayev empirical formulas [3] are compared in Figure 5.It reveals that they accord with each other.If one defines the relative error of the peak pressure between the numerical P 0 and the Zamyshlyayev results P as ε � |P 0 − P|/P, the values can be listed in Table 3.We can find that the relative errors are within 10%.Although there are some oscillations caused by the strong discontinuity of the shock wave in the numerical results, the whole attenuation process and peak pressure values are in good agreement with the empirical formula.
In fact, during the shock wave propagating in multilayer media, the FSI process is involved.In order to further test the reliability of the CEL method in dealing with FSI problems, an experiment (this experimental case and relevant data are provided by the Institute of Fluid Physics, China Academy of Engineering Physics) is performed in a water tank with the size of 2 m × 2 m × 2 m. e depth of water is 1.6 m.As shown in Figure 6, the square test plate made from the mild steel Q235 is fixed on the wall of the tank.It has a length of 0.8 m and a thickness of 0.003 m. e cylindrical charge has a size of Φ 20 mm × 18 mm, and thus, the mass is 10 g. e charge is placed in the water, and its depth is D � 0.8 m. e axis of the charge is parallel with the plate, and the distance R 165.9   Table 2: e parameters of polynomial EOS for water [25,44].Shock and Vibration from the axis to the plate is about four times the charge radius R 0 , namely, R � 4R 0 .e experimental layout is illustrated in Figure 6.Two measuring points (D1 and D2) are placed on the side of the plate to record the displacement and velocity (Figure 7).e three-dimensional numerical model is established according to the layout of the experiment.e water and TNT are also modeled by the Eulerian subgrid.Note that the air in this numerical model can be ignored.
e steel is modeled by the Lagrangian subgrid, and its density is 7,830 kg/m 3 .e Johnson-Cook strength and fracture parameters for the steel are listed in Tables 4 and 5 (found in [46]).
e smallest element size of the Eulerian mesh is 5 mm, and it is 1.5 mm for the Lagrangian mesh.e Euler domain consists of about 3.8 million cells, and the Lagrangian domain contains 570,000 cells.
e displacement and velocity time histories calculated by the CEL method are compared with the experimental results in Figure 8.It is not hard to see from Figure 8 that they are in good accordance, especially the displacement time histories.erefore, the CEL method can simulate the FSI process well, and it can predict accurately the shock wave propagation in the complex model.

Numerical Model.
e two-dimensional axisymmetric numerical model of a double-layer hemispherical shell subjected to underwater explosion is established to investigate the propagation characteristics of the shock wave in multilayer media, as plotted in Figure 9. e outer shell has a radius of R 1 and a thickness of d 1 , while the inner shell has a radius of R 2 and a thickness of d 2 .e parameters related to distance are normalized by the charge radius R 0 , namely, and r � R/R 0 � 10−16.e thickness of the outer shell is changed to test the effects, so d 1 � 0.1−0.8 are selected with a constant of d 2 � 0.4.e mass of the spherical TNT charge, the charge radius R 0 , and the size of the Eulerian domain are the same as those of the free-field model in Section 2.3.e measuring points 1 and 3 are located near the outside surface of the outer shell and the inner shell, respectively.e measuring point 2 is located near the inside of the outer shell, and the distance from point 2 to the outer shell is d � 0.2R 0 .e shell structure is fixed around the boundary, and the transmitting condition is enforced around the fluid domain so as to overcome the nonphysical reflection.e parameters related to time hereinafter are normalized by the time decay constant θ when r � 10 in equation (8).
Medium 1 and medium 2 are water or air.e material parameters of water and charge are the same as those in the free-field model.Air is modeled by the ideal gas equation of state, which can be written as p � ρ/ρ 0 ((c − 1))e, where the specific heat ratio c is 1.4, the initial density of air ρ 0 is 1.225 kg/m 3 , and the specific energy e is 253.4 kJ/m 3 [25].
e density of steel is 7,830 kg/m 3 [25].Note that the Johnson-Cook strength parameters listed in Table 6 [25] for the steel are different from those of the experiment in Section 2.3.e element size for the Eulerian part is also 5 mm, the same as that of the free-field model, and more than four elements in the thickness direction are discretized for the Lagrangian part.us, the Eulerian domain contains about 100,000 cells, and the number of Lagrangian cells ranges from 4,230 to 8,940 due to different outer shell thicknesses.

Shock Wave Propagation
Process. Figure 10 shows the shock wave propagation process for the case of media 1 and 2 filled with water and air, respectively, at the detonation distance r � 10.Several typical instants during the process are presented.Obviously, they show a similar process after the charge detonation, and a spherical shock wave propagates to the outer shell surface (Figure 10(a)); when the wave arrives at the outer shell whose other side is backed to water, it is reflected and transmitted by the shell, as plotted in Figure 10(b) (the left column).Accordingly, when the wave arrives at the outer shell whose other side is backed to air, there exists a rarefaction wave reflected which travels backwards in water.Due to the much lower impedance of the air, the transmitted wave is weaker than the reflected wave.erefore, Figure 10(b) (the right column) cannot recognize the transmitted wave in the air-backed case; when the transmitted shock wave reaches the inner shell whose other side is backed to water, the shock wave is reflected and transmitted over the shell again in Figure 10(c) (the left column).Cavitation occurs near the outer shell in the air-backed case, as shown in Figure 10(c) (the right column).Also, the pressure of the reflected and transmitted waves on the inner shell in the water-backed case is obviously larger than that in the air-backed case.Later, the wave reflected from the inner shell propagates to the outer shell, and then it is reflected and transmitted again.Due to the complex superposition of the incident wave, the reflected wave, and the transmitted wave, the pressure becomes very complicated (Figure 10(d) (the left column)).At the same time, the cavitation region in the airbacked case is further expanding.
On the whole, because of the severe mismatch of the impedance between different media, the intensity of the transmitted wave of the shells backed to water is larger than that of the shells backed to air and the cavitation region near the outer shell occurs in the air-backed case.

Shock Wave Reflection.
Figure 11 shows the time history of the shock wave pressure at point 1. e outer shell is backed to water in case a and to air in case b.Meanwhile, different thicknesses of the outer shell are also taken into account, and the detonation distance r is 10.Note that medium 2 behind the inner shell has a little effect on the reflected pressure near the outer shell.As the results show, if d 1 > 0.4, with the increase of the outer shell thickness, the pressure of the reflected wave tends to be steady in both cases.By comparing the above two cases, it can be seen that the peak pressure of the reflected wave caused by the waterbacked shell is larger than that of the air-backed case, but the latter tends to be consistent with the former in value if d 1 ≥ 0.4.What is more, the pressure fluctuates more violently for the air-backed case.

Shock and Vibration
is can be explained from the perspective of wave propagations.e first interface for the wave arrival is the same for the two cases, namely, the interface between the water and the outside of the outer shell.erefore, the reflected wave pressure at this interface should be equal in both cases.However, at the next interface, the shock wave transmits from the inside of the outer shell to the water in case a while from the inside to the air in case b.Because the acoustic impedances of water and air are much lower than that of steel, there is a rarefaction wave reflected at this interface [47][48][49][50][51].
erefore, the intensity of the incident shock wave is unloaded.Moreover, the acoustic impedance of air is lower than that of water, the unloading effect in case a is less serious than that in case b, so the peak pressure of the reflected wave in case a is larger than that in case b. e stiffness of the outer shell, however, is improved with the increase of shell thickness.Hence, the intensity of the reflected wave in these two cases is gradually enhanced.To sum up, if the outer shell thickness d 1 ≥ 0.4, the medium behind the outer shell will affect the peak pressure of the reflected wave slightly.
e reflection coefficient of shock wave pressure is defined as λ r � P 1 /P 0 , where P 1 is the peak pressure of the reflected wave at point 1 and P 0 is the peak pressure at the same distance in the free-field underwater explosion model.In the present work, if the detonation distance is far enough, the numerical model can be regarded as a one-dimensional model.Based on Taylor's assumptions [6], we can obtain P 1 � P 0 + P r − ρcv, where P r is the pressure of the reflected wave and ρcv is the pressure of the rarefaction wave produced by the motion of the structure.ρ is the density of water, c is the sound speed of water, and v is the velocity of the outer shell.If the perfect reflection of the incident wave occurs [52,53], the pressure of the reflected wave would read P r � P 0 .erefore, P 1 should satisfy P 1 � 2P 0 − ρcv theoretically.And then, the reflection coefficient holds λ r � P 1 /P 0 � 2 − ρcv/P 0 .
Figure 12 illustrates the reflection coefficients of the shock wave varying with the thickness of the outer shell at the distance r � 10.Obviously, the reflection coefficients increase with the increase of shell thickness, but the trend gradually slows down for the two cases.Moreover, the coefficients of the air-backed case are almost the same as those of the water-backed case if d 1 ≥ 0.4.e reflection coefficients exceed 1 no matter what the medium is filled with at the back of the outer shell.It means that the presence of the outer shell enhances the peak pressure at point 1.
If the thickness d 1 � 0.4, the shock wave reflection coefficients varying with the distance r in these two cases are plotted in Figure 13.e coefficients decrease with the increase of detonation distance, as shown in Figure 13.e relationship between the reflection coefficients and the detonation distance is approximately linear for these two cases.In the present work, we can recast the Zamyshlyayev empirical formula, i.e., equation ( 6), as where k and α are the ratio parameter and exponent parameter, respectively.So, a relation between the reflection coefficient λ r and the distance r can be obtained as λ r � P 1 /P 0 � 2 − (ρcv/k)r α .As described in the literature [54], the rarefaction wave (i.e., ρcv) can be neglected at the early stage of shock wave impinging on the absolutely rigid structure.For the elastic shell, if we extract the velocities of the outer shell near point 1 from numerical results, listed in Table 7, we can find these velocities are positive for these cases.So, the rarefaction wave cannot be neglected.In this case, the reflection coefficient of peak pressure varying with the distance r can be obtained from the numerical simulation, as shown in Figure 13.One should notice that the decreasing trend of the coefficient is only suitable for the cases of close distances (i.e., 10 ≤ r ≤ 16).
e impulse reflection coefficient can be defined as I r � I 1 /I 0 , where I 1 is the pressure impulse at point 1 and I 0 is the pressure impulse at the same distance in the free-field underwater explosion model.Figure 14 shows the impulse reflection coefficients varying with shell thickness in the water-backed case and air-backed case at the detonation distance r � 10. e impulse reflection coefficients increase with the increase of shell thickness, and the tendency becomes gradually slow.If d 1 < 0.4, the impulse reflection coefficients of the air-backed case will be smaller than those of the water-backed case because the deformation of the outer shell in the air-backed case is more serious than that in the water-backed case.If the thickness d 1 ≥ 0.4, the coefficients of the air-backed case are larger than those of the water-backed case.Under this circumstance, the deformation of the outer shell is not the principal factor

Shock and Vibration
Table 6: e relevant parameters for material strength [25].Shock and Vibration influencing impulse, since the stiffness of the outer shell is high enough.e reason for this phenomenon is that the incident wave transmits over the outer shell in the waterbacked case, while the incident wave is mainly reflected in the air-backed case because of the impedance difference between water and air.
If the thickness d 1 � 0.4, the impulse reflection coefficients varying with different detonation distances in these two cases are plotted in Figure 15.
e coefficients approximately decrease linearly with the increase of detonation distance whatever the outer shell is exposed to water or air.It should be noticed that I r of the air-backed case changes faster than that of the water-backed case and varies in a wide range, which may be due to the fact that complex superposition of the wave system between the outer shell and the inner shell affects seriously the pressure of point 1 after t � 7.09θ, found in Figure 10.

Shock Wave Transmission.
When the shock wave transmits over the outer shell to the inner shell, the peak pressure and the energy of the shock wave gradually attenuate.In Figure 16, the pressure histories of the transmitted shock wave for different shell thicknesses are presented.Medium 1 between the two shells and medium 2 behind the inner shell are water or air, respectively.e detonation distance r is 10.In Figure 16(a), medium 1 and medium 2 are both water (noted as water-water), while medium 1 is water and medium 2 is air (noted as water-air) in Figure 16(b).Obviously, when the wave transmits over the  outer shell, a peak pressure, called the transmitted peak, occurs in these two cases, and then it attenuates quickly.With the attenuation of the pressure, the waves reflected by the outer and inner surfaces of the inner shell reach point 2 at about t � 6.30θ, as shown in Figure 17.Owing to the superposition of the transmitted wave and the reflected wave, it causes the pressure to rise and generates a second peak pressure, called the reflected peak, which might have a second impact on the shell (Figures 16(a      Shock and Vibration addition, both the peaks decrease with the increase of the outer shell thickness.It is worth noting that the reflected peak can be larger than the transmitted peak if d 1 ≥ 0.6.It indicates that the reflected wave produced by the inner surface of the outer shell plays a leading role under this circumstance.Media 1 and 2 are both air (noted as air-air) in Figure 16(c), while medium 1 is air and medium 2 is water (noted as air-water) in Figure 16(d).e peak pressures of the transmitted shock wave at point 2 are almost the same at the same shell thickness for both the cases, while they are about 1/330 times that of the water-water case.It may be due to the lower impedance of air which is about 1/5,000 times that of the water.If the shell thickness d 1 � 0.2 in these two cases, there are no values in pressure history curves when the time exceeds 9.46θ, which is due to the fact that the outer shell after deformation is over the position of the measuring point 2. With the increase of the shell thickness, the pressure attenuates gradually.Different from the water-water case or water-air case, the second peak later (at about t � 15.77θ) which may be due to the strong impedance mismatch.Also, the pressure time histories show no Shock and Vibration 13 differences for medium 2 filled with water or air behind the inner shell.
As seen in Figures 16(a)-16(d), it can be found that medium 1 has a great influence on the pressure of the transmitted wave.On the contrary, medium 2 behind the inner shell has few effects on the transmitted pressure.
e transmission coefficient λ t � P 2 /P 0 can be defined as the ratio of the peak pressure P 2 of the transmitted wave at point 2, i.e., the first peak in Figure 16, to the peak pressure P 0 of the incident shock wave in free field.e transmission coefficients of the shock wave varying with different shell thicknesses are plotted in Figure 18.As described in the analysis above, medium 2 behind the inner shell has little effect on the peak pressure of the transmitted shock wave.erefore, the cases in which medium 2 is water while medium 1 is water or air are hereinafter taken for examples.With the increase of the shell thickness, the transmission coefficients of the shock wave decrease exponentially for the two cases.Although the peak pressure P 2 of the air-backed case is much smaller than that of the water-backed case, the trend of transmission coefficients varying with different shell thicknesses is similar to that of the water-backed case.
Taking the thickness d 1 � 0.4 for example, the curves of the transmission coefficients varying with different detonation distances can be plotted in Figure 19.Unlike Figure 13, with the increase of the detonation distance, the transmission coefficients have a trend of increase by degrees.
e difference may be because the peak pressure P 2 is affected by the inner shell and the outer shell together, while P 1 is mainly affected by the outer shell, although the peak pressure P 1 and P 2 decrease as the detonation distance increases.
e total energy released by the TNT is 2703 kJ at t � 7.5θ, and the proportion of the energy absorbed by the outer shell accounting for the total energy is plotted in Figure 20.Because the TNT mass is quite small and the shell has not been destroyed, the energy absorbed by the outer shell is less.With the increase of shell thickness, the energy absorption proportion increases generally.Another feature is that the proportion of the energy absorbed by the outer shell can be up to 2% for the air-backed case while about 1% for the water-backed case when the shell thickness d 1 ranges from 0.1 to 0.8.For the same shell thickness, the energy absorption of the air-backed case is two to three times that of the water-backed case.
is may attribute to the larger deformation of the shell in the air-backed case.
Figure 21 shows the proportion of the energy transmitted over the outer shell versus different thicknesses.e transmission energy gradually decreases with the increase of shell thickness for the two cases.For the same shell thickness, the transmission energy of the air-backed shell is much smaller than that of the water-backed shell, which coincides with the above results in Figure 18. at is, the transmission coefficient λ t of the air-backed case is much smaller than that of the water-backed case.
Figure 22 shows the Mises stress contours of the outer and inner shells for the case of the outer shell thickness d 1 � 0.4.It can be seen that if medium 1 between the two shells is air, the maximum stress of the shell will be about 1.5 times that of the water-backed shell, and there are almost no stresses in the inner shell.erefore, it can be concluded that if the medium is filled with air between two shells, the inner shell is prevented from serious impact.

Comparison of Shock Wave Reflection and Transmission.
Actually, the shock wave pressure measured on the outer surface of the shell is from the superposition of the incident shock wave and the reflected shock wave.In order to study  Shock and Vibration the relationship between the reflection and transmission of the shock wave, the ratios λ 1 � (P 1 − P 0 )/P 2 of the peak pressure of the reflected shock wave to that of the transmitted shock wave are plotted in Figure 23.
e peak pressure of the reflected shock wave is the difference between the measured pressure P 1 near the surface of the outer shell and the incident pressure P 0 in free field.
Obviously, with the increase of the thickness, no matter what the outer shell is backed to, the ratios increase and tend to be slow when the thickness exceeds to some extents for the air-backed case.e reason for this phenomenon is that the peak pressure P 2 of the transmission wave decreases drastically with the increase of shell thickness (Figures 16 and  18).Another feature is that the ratios of the shell backed to air are much larger than those of the shell backed to water.
at is, the reflection proportion is dominant for the airbacked case. is can be explained by the fact which can be found in Section 3.4 that the peak pressure P 2 of the airbacked case is smaller than that of the water-backed case for two orders of magnitude.
Figure 24 shows the peak pressure ratios λ 2 � P 3 /P 1 at the measuring points 1 and 3 versus different outer shell thicknesses.Generally, the ratios decrease with the increase of the shell thickness no matter what media 1 and 2 are filled with.It indicates the reflection proportion increase as the shell thickens.In addition, the ratios of the water-backed cases (water-water case and water-air case) are obviously larger than those of the air-backed cases (air-air case and airwater case) which almost coincide with each other.ese accord well with the analysis in Sections 3.3 and 3.4.
After the directive wave propagating over the outer shell, the transmitted wave arrives at the inner shell, which can be regarded as a directive wave producing a new charge, i.e., the equivalent charge mass m e .at is to say, the effect of the transmitted wave on the inner shell is, to some extent, equivalent to that of a new charge on the inner shell.With the aid of the analysis in Section 3.3, m e can be transformed from a free-field component of incident peak pressure on the inner shell, namely, P 4 � P 3 /λ r , by Zamyshlyayev empirical formula (6).
Table 8 presents the ratios of the equivalent charge mass m e to the actual charge mass m r .e cases of the medium between the two shells filled with water are considered, and air-backed cases are absent because P 3 is too small.As the table shows, the equivalent charge mass for the inner shell can be decreased about 75% if the outer shell thickness is over 0.2 times the charge radius.With the increasing thickness, the equivalent charge mass decreases sharply.

Conclusions
e shock wave propagation between two hemispherical shells which are filled with different media is studied based on the coupled Eulerian-Lagrangian method in AUTODYN.Shock and Vibration e relationships among the incident, reflected, and transmitted waves are discussed.
e medium between two hemispherical shells will affect the peak pressure of the reflected wave generated by the outer shell slightly if the outer shell thickness is thicker than a certain level (here, it is over 0.4 times the charge radius).
e reflection coefficients of peak pressure and impulse increase with the increase of shell thickness and decrease linearly with the increase of detonation distance.Besides, the cavitation area caused by the reflected rarefaction wave near the outer shell occurs in the air-backed case.
When the wave transmits over the outer shell, a peak pressure, called the transmitted peak, occurs whatever the medium behind the outer shell is water or air.It is worth mentioning that a second peak, which is relatively compared to the transmitted peak, can be produced by the reflections of the inner shell, and it will be larger than the transmitted peak if the outer shell thickness reaches a certain extent (here, it is over 0.6 times the charge radius) for the water-water and water-air cases.e medium between the two shells has great influences on the transmitted wave, while the medium behind the inner shell has slight effects on the transmitted wave.With the increase of the shell thickness, the transmission coefficients of the shock wave decrease exponentially and increase linearly with the detonation distance whatever the outer shell is exposed to water or air.
For the same outer shell thickness, the energy absorption of the air-backed case is about two to three times that of the water-backed case.With the increasing thickness, the equivalent charge mass decreases drastically.e equivalent charge mass for the inner shell can be decreased about 75% if the outer shell thickness is over 0.2 times the charge radius.

Figure 4 :Figure 5 :
Figure 4: Shock wave pressure versus time from three different element sizes at r � 10.

Figure 7 :
Figure 7: e position of measuring points.

Figure 8 :Figure 9 :
Figure 8: Displacement and velocity time histories at D1 and D2 points (these experimental data are provided by the Institute of Fluid Physics, China Academy of Engineering Physics): (a) the displacement at D1 point; (b) the displacement at D2 point; (c) the velocity at D1 point; (d) the velocity at D2 point.

Figure 10 :
Figure 10: Comparison of the shock wave propagation when the double-layer shell was filled with water (the left column) and air (the right column).(a) t � 3.63θ.(b) t � 5.20θ.(c) t � 6.30θ.(d) t � 7.09θ.

Figure 11 :
Figure 11: e pressure at gauging point 1 for the cases of different thicknesses of the outer shell (dark color represents water and light color represents air).(a) Media 1 and 2 filled with water.(b) Media 1 and 2 filled with air.

1 Figure 12 :
Figure 12: e reflection coefficients of peak pressure for different outer shell thicknesses.

Figure 13 :
Figure 13: reflection coefficients of peak pressure for different detonation distances.

Figure 14 :
Figure 14: e reflection coefficients of impulse for different outer shell thicknesses.

r
Figure 15: e reflection coefficients of impulse for different detonation distances.

2 d 1 Figure 16 :
Figure 16: Transmitted pressure at gauging point 2 for different media filled between the shells (dark color represents water and light color represents air).(a) Water-water.(b) Water-air.(c) Air-air.(d) Air-water.

Figure 17 :
Figure 17: e pressure contours if the medium is water-water and water-air at the same time.(a) Water-water at t � 6.30θ.(b) Water-water at t � 7.09θ.(c) Water-air at t � 6.30θ.(d) Water-air at t � 7.09θ.
Figure 19:e transmission coefficients of peak pressure for different detonation distances.e water-backed case is plotted in the left axis, and the air-backed case is plotted in the right axis.

Figure 22 : 1 Figure 23 :
Figure 22: e Mises stress distribution of the double-layer shell structure.(a) Water filled between the two shells.(b) Air filled between the two shells.

1 Figure 24 :
Figure24: e peak pressures of P 3 /P 1 for different outer shell thicknesses.e water-backed case is plotted in the left axis, and the air-backed case is plotted in the right axis.

Table 3 :
Comparison of peak pressure.

Table 7 :
e time of the peak pressure occurring, the corresponding velocity of the outer shell, and the reflection coefficient with different detonation distances.

Table 8 :
e equivalent charge mass for the inner shell with different outer shell thicknesses.