betaFeedback

WorksPhysics and Astronomy

Significance of secondary baroclinic hydrodynamic instability on mixing enhancement in shock-bubble interaction

Abstract

Different strength of hydrodynamic instability can be induced by the variations in the initial diffusion of shock bubble interaction (SBI), while the influence of hydrodynamic instability on variable-density mixing in SBI remains unclear. The present study aims to investigate the hydrodynamic instability of SBI through high-resolution numerical simulations. To isolate each factor within this instability, a circulation control method is employed to ensure consistent Reynolds number Re and Peclect number Pe. An examination of the morphology of the bubbles and vorticity dynamics reveals that the hydrodynamic instability can be characterized by positive circulation. Through vorticity budget analysis, the positive circulation is dominated by the baroclinic torque. Therefore, the identified hydrodynamic instability is labeled as secondary baroclinic hydrodynamic instability (SBHI). Based on the dimensional analysis of vorticity transport equation, a new dimensionless parameter, the secondary baroclinic vorticity (SBV) number, is proposed to characterize the strength of SBHI. Regarding mixing characteristics, cases with stronger SBHI exhibit higher mixing rates. Indicated by the temporal-averaged mixing rate with different SBV numbers, a scaling behavior is revealed: the mixing rate is increased proportionally to the square of SBV numbers. It is widely recognized that unstable flow can also be induced by a high Reynolds number Re. The distinction and connection of SBHI and high Re unstable flow are further studied. The scaling behavior of the mixing rate in SBHI shows distinct from the Reynolds number Re scaling: the mixing can hardly be altered effectively in the limit of large Re, the mechanisms of which are inherently different with respect to the stretch term, closely relating to the principal strain and the alignment angle between the scalar gradient and the principal strain axis.

Introduction

The evolution of the interface between fluids of different densities subjected to shock can be classified as the Richtmyer–Meshkov (RM) instability [1, 2, 3]. This instability is characterized by wave patterns, vortex dynamics, and turbulent mixing, making it an intricate fluid phenomenon. A typical example of the RM instability is the interaction between a shock wave and a cylindrical or spherical gas bubble with a density contrast from the ambient gas, which is known as shock bubble interaction (SBI) [4]. The misalignment of the pressure gradient (∇p\nabla p) and density gradient (∇ρ\nabla\rho) leads to the generation of baroclinic vorticity, depositing on the surface and triggering the formation of a primary vortex within the bubble. Subsequently, secondary instability arises and breaks the interface into smaller structures. Throughout the evolution of the bubble morphology, mixing between the bubble and the ambient gas takes place simultaneously. The complex evolution of the interface in SBI make it to be a typical case for testing the high-resolution numerical method [5, 6]. The investigation of mixing also holds a central position in SBI research due to its significant implications in inertial confinement fusion (ICF) [7], supernovas [8], and supersonic combustion [9, 10].

In a simplified conceptualization, the mixing in SBI can be outlined as follows: generally, following shock interaction, a large-scale primary vortex is formed. Driven by this main vortex, the initially concentrated bubble is tangentially stretched along the interface with the ambient air, leading to an increase in material line length. Simultaneously, as the interface undergoes stretching, the scalar gradient deposited along it is amplified, thereby intensifying molecular diffusion until the bubble is completely mixed with the surrounding air. This simplified mixing process is consistent with the one proposed by Villermaux et al. [11]. According to this simplified mixing conception, the material line stretch rate and the bubble area serve as two critical parameters characterizing mixing, and have been widely used in SBI mixing research. Early experiments conducted by Jacobs [12] utilized the bubble area to depict the extent of mixing and revealed a constant mixing rate by investigating the temporal derivative of the bubble area ratio. Yang [13] employed the material line stretch rate to characterize the mixing rate, and discovered an exponential stretching of the material line in the early stage of SBI mixing. Kumar [14] also conducted experiments on SBI of SF6_6 cylinders using the planar laser-induced fluorescence (PLIF) technique to explore the material line stretch rate, and a strong dependence of the stretch rate on the configuration and orientation of the cylinders was observed. Niederhaus [15] studied the mixing process using the mixedness defined by the bubble volume fraction in three-dimensional shock spherical interactions through numerical simulation. Ou and Zhai et al. [16] investigated the effect of aspect ratio on mixing in shock elliptic cylinder interaction. The bubble area, mean volume fraction and the mixing rate were presented to illustrate the increase of aspect ratio can promote mixing. Yu [17] discovered the amplification effect of secondary baroclinic vorticity on the stretch rate and proposed a model to predict the variable-density mixing time.

Moreover, mixing is a process characterized by the molecular diffusion of mass fraction YY. Hence, in addition to the material line stretch rate, mixing indicators can also be directly defined from the mass fraction. In the experiment on the single vortex mixing, Cete­gen [18] proposed the mixedness f=4Y(1−Y)f = 4Y(1 - Y) to quantify the extent of mixing, which reaches maximum value when the bubble and surrounding air are in a 1:1 state. From the derivation of the transport equation for mixedness ff, Buch et al. [19] put forward the scalar dissipation rate (SDR) χ=∂Y∂xi∂Y∂xi\chi= \frac{\partial Y}{\partial x_i}\frac{\partial Y}{\partial x_i} to describe the mixing rate in turbulent shear flow, with ∂Y∂xi\frac{\partial Y}{\partial x_i} representing the scalar gradient. Tomkins [20] utilized PLIF experiments of shock interaction with SF6_6 cylinders, and observed that the SDR is primarily distributed on the bridge structure. Shankar et al. [21] studied the effect of the presence of a third species on the evolution of SBI by using SDR as the mixing rate. Li et al. [22] put forward a Gauss model to predict the late time evolution of SBI based on the definition of the mixedness. In consideration of the difference between variable density mixing and passive scalar mixing, Yu et al. [23] defined the density accelerated mixing rate. Through an investigation of the scaling behavior of this mixing rate with the Reynolds number ReRe and Schmidt number ScSc [24], it was discovered that with a constant Péclect number PePe, the mixing rate experiences minimal alteration with increasing Reynolds number once ReRe reaches a large value: (χ∗/Peα)≈βRe0(\chi^*/Pe^\alpha) \approx\beta Re^0. This scaling behavior was also confirmed in homogeneous isotropic turbulence mixing [25]. Furthermore, the underlying mechanism of the scaling behavior can be attributed to the influence of flow on the mixing rate. Based on the derivation of the transport equation of SDR, Buch et al. [19] proposed the stretch term −2∂Y∂xiSij∂Y∂xj-2\frac{\partial Y}{\partial x_i}S_{ij}\frac{\partial Y}{\partial x_j} as the primary source of the growth of SDR, which has been extensively examined in various types of mixing, including shock-turbulence interaction [26, 27, 28], isotropic turbulence mixing [29], and under-expanded jets [30]. The characteristic of the stretch term is summarized in the recent review [31].

With the interesting phenomenon that the initial diffusion can affect mixing even in simple shear flow described by Villermaux [11] in Figure 1a, the initial diffusion also widely exists in SBI research. In previous SBI experiments, two common techniques have been used to create the gas cylinder: the membrane technique, such as soap films [16, 32], and the membraneless technique, such as jets [12, 14]. These techniques result in different states of initial diffusion, as indicated by the width of the diffusive layer. Different from the findings of Villermaux in Figure 1a, the variations in initial diffusion can lead to different strength of hydrodynamic instability, subsequently altering the evolution of the bubbles, as illustrated in Figure 1b [33]. Therefore, Li [33, 34] conducted a study on the impact of initial diffusion on mixing by examining the material line stretch rate and the bubble area. However, the total circulation was not controlled in these existing studies, resulting in variations in the Reynolds number ReRe and Péclect number PePe across the cases, which in turn made it difficult to distinguish each factor within this hydrodynamic instability. Therefore, the influence of the hydrodynamic instability on mixing remains unclear and can be summarized in the following three questions:

  1. What is the source of the hydrodynamic instability induced by initial diffusion?

  2. What is the effect of this instability on mixing?

  3. What is the mechanism underlying the effect of this instability on mixing?

Schematic illustration of the effect of initial diffusion on mixing in simple shear flow

(a)

Mixing in shock SF$_6$ cylinder interaction

(b)

Figure 1. The effect of initial diffusion on (a) mixing in simple shear flow [11], and (b) mixing in shock SF6_6 cylinder interaction [33].

By investigating the aforementioned three questions, it was discovered that the hydrodynamic instability induced by the initial diffusion, which is characterized by a defined secondary baroclinic vorticity (SBV) number Ωsbv\Omega_{sbv}, can enhance mixing by amplifying the SDR stretch term −2∂Y∂xiSij∂Y∂xj-2\frac{\partial Y}{\partial x_i}S_{ij}\frac{\partial Y}{\partial x_j}. This paper is organized as follows: In Section 2, we provide an overview of the numerical method and the setup of the cases designed to control circulation. § 1.2 presents the numerical results and discussion, and the aforementioned three questions are elaborated in this section. Finally, our study is summarized, and potential future work is discussed in Section 4.

Numerical method and setup

Numerical method

In this study, the governing equations used for simulating two-dimensional SBI are the compressible Navier-Stokes equations with multiple components, expressed as:

{∂ρ~∂t~+∂(ρ~u~i)∂x~i=0,∂ρ~u~i∂t~+∂(ρ~u~iu~j)∂x~j=−∂p~∂x~i+∂σ~ij∂x~j,∂(ρ~E~)∂t~+∂(ρ~u~iH~)∂x~i=−∂q~i∂x~i+∂(u~jσ~ij)∂x~i,∂(ρ~Y~m)∂t~+∂(ρ~u~iY~m)∂x~i=∂∂x~i(ρ~D∂Y~m∂x~i),(m=1,2,⋯ ,s−1).(1)\left\{ \begin{aligned} \frac{\partial\tilde{\rho}}{\partial\tilde{t}}+\frac{\partial(\tilde{\rho}\tilde{u}_i)}{\partial\tilde{x}_i}&=0,\\ \frac{\partial\tilde{\rho}\tilde{u}_i}{\partial\tilde{t}}+\frac{\partial(\tilde{\rho}\tilde{u}_i\tilde{u}_j)}{\partial\tilde{x}_j}&=-\frac{\partial\tilde{p}}{\partial\tilde{x}_i}+\frac{\partial\tilde{\sigma}_{ij}}{\partial\tilde{x}_j},\\ \frac{\partial(\tilde{\rho}\tilde{E})}{\partial\tilde{t}}+\frac{\partial(\tilde{\rho}\tilde{u}_i\tilde{H})}{\partial\tilde{x}_i}&=-\frac{\partial\tilde{q}_i}{\partial\tilde{x}_i}+\frac{\partial(\tilde{u}_j\tilde{\sigma}_{ij})}{\partial\tilde{x}_i},\\ \frac{\partial(\tilde{\rho}\tilde{Y}_m)}{\partial\tilde{t}}+\frac{\partial(\tilde{\rho}\tilde{u}_i\tilde{Y}_m)}{\partial\tilde{x}_i}&=\frac{\partial}{\partial\tilde{x}_i}\left(\tilde{\rho}\mathcal{D}\frac{\partial\tilde{Y}_m}{\partial\tilde{x}_i}\right), \end{aligned} \right. \qquad(m=1,2,\cdots,s-1). \tag*{(1)}

Here, (⋅~)(\tilde{\cdot}) indicates dimensional parameters. The symbols ρ~,u~i,p~,E~,H~\tilde{\rho},\tilde{u}_i,\tilde{p},\tilde{E},\tilde{H} denote the density, velocity, pressure, energy, and enthalpy, respectively. Furthermore, Y~m\tilde{Y}_m refers to the mass fraction of species mm, and t~\tilde{t} and x~i\tilde{x}_i represent the corresponding time and space coordinates.

The viscous stress tensor and heat flux are defined as:

{σ~ij=μ(∂u~i∂x~j+∂u~j∂x~i−23δij∂u~k∂x~k),q~i=−λ∂T~∂x~i,λ=CpμPr,(2)\left\{ \begin{aligned} \tilde{\sigma}_{ij}&=\mu\left(\frac{\partial\tilde{u}_i}{\partial\tilde{x}_j}+\frac{\partial\tilde{u}_j}{\partial\tilde{x}_i}-\frac{2}{3}\delta_{ij}\frac{\partial\tilde{u}_k}{\partial\tilde{x}_k}\right),\\ \tilde{q}_i&=-\lambda\frac{\partial\tilde{T}}{\partial\tilde{x}_i},\\ \lambda&=\frac{C_p\mu}{Pr}, \end{aligned} \right. \tag*{(2)}

where μ\mu and λ\lambda denote the dynamic viscosity and heat conduction coefficient, respectively. The constant-pressure specific heat is represented by CpC_p, and the Prandtl number is set as a constant, i.e., Pr=0.72Pr=0.72. The kinetic viscosity is defined as ν=μ/ρˉ\nu=\mu/\bar{\rho}, the average density ρˉ=(ρ1′+ρ2′)/2\bar{\rho}=(\rho'_1+\rho'_2)/2 corresponds to the average of the post-shock ambient air density ρ1′\rho'_1 and post-shock helium bubble density ρ2′\rho'_2. These two parameters are obtained from one-dimensional shock dynamics [4]. In the multi-species transport equations as described in Eq. (1), D\mathcal{D} represents the constant Fickian diffusivity. During the SBI simulation, the dynamic viscosity μ\mu is set as a constant of 50×10−6 Pa⋅s50\times10^{-6}\,\mathrm{Pa\cdot s}, and the constant Fickian diffusivity D\mathcal{D} is set as 90×10−6 m2/s90\times10^{-6}\,\mathrm{m^2/s}.

In this study, the compressible Navier-Stokes equations are numerically solved using an in-house high-resolution code, ParNS3D, which has undergone extensive validation [17, 35, 36, 37]. Temporal discretization is performed using the third-order Total Variation Diminishing (TVD) Runge-Kutta method. The fifth-order Weighted Essentially Non-Oscillatory (WENO) scheme is employed for the discretization of convection terms, while the second-order central difference method discretizes the viscous terms. The grid resolution study of the numerical results is provided in Appendix A1. Furthermore, the reliability of these results is verified through budget analysis, which will be presented in subsequent sections.

Initial conditions and dimensionless parameters

The initial conditions for the numerical simulation are depicted in Figure 2. The computation domain encompasses a two-dimensional space in a Cartesian coordinate system, with X×Y=[0,Lx]×[0,Ly]X \times Y = [0, L_x] \times[0, L_y], where the length is Lx=30L_x = 30 mm and the height is Ly=5.5L_y = 5.5 mm. A shock of strength Ma=2.4Ma = 2.4 is positioned at xshock=0.7x_{shock} = 0.7 mm. The pre-shock air has thermodynamic states of 1 atm pressure and 293 K temperature prior to being impacted by the shock. The post-shock air conditions (u1′u_1^{\prime}, ρ1′\rho_1^{\prime}, P1′P_1^{\prime}, and T1′T_1^{\prime}) are calculated from the Rankine-Hugoniot equation. The helium bubble, centered at the position xbubble=4.5x_{bubble} = 4.5 mm, consists of a core region with a radius of RcoreR_{core} and a diffusive layer with a width of WW. Then, the radius of the bubble is R=Rcore+WR = R_{core} + W. A new geometric parameter, ξ=W/Rcore\xi= W/R_{core}, is defined to represent the initial diffusion degree of the bubble. The diffusive layer is positioned at the edge of the core region, as illustrated in the inserted figure in Figure 2. Based on this configuration, the distribution of mass fraction of helium is set as:

Schematic of the initial conditions of shock cylindrical bubble interaction

Figure 2. Schematic of the initial conditions of shock cylindrical bubble interaction.

YHe(r)={Ymax,r<Rcore,Ymaxe−α[(r−Rcore)/W]2,Rcore≤r<R,0.r≥R.(3)Y_{He}(r) = \begin{cases} Y_{max}, & r < R_{core}, \\ Y_{max}e^{-\alpha[(r-R_{core})/W]^2}, & R_{core} \le r < R, \\ 0. & r \ge R. \end{cases} \tag*{(3)}

where Ymax=1.0Y_{max} = 1.0 and α=5\alpha= 5, and YHeY_{He} is simplified as YY in the following content.

This distribution is consistent with that reported in Refs. [17, 24, 35]. The leftmost boundary is set as an inlet, while the upper and rightmost boundaries are conditioned with a fourth-order extra-interpolation to prevent pseudo-pressure reflection wave interference with the concerned flow structures. The symmetric boundary is applied at the bottom of the computational zone. Previous research has shown that the strength of hydrodynamic instability can be influenced by the initial diffusion condition of the bubble [33, 34]. To manage this instability, the ratio ξ=W/Rcore\xi= W/R_{core} is set at values of 0.1, 1.0, 5.0, and 10.0, respectively.

Following the interaction of the shock wave with the bubble over a brief interval, a nearly constant total circulation is obtained from the spatial integration of the vorticity ω~=∇×u\widetilde{\omega} = \nabla\times\boldsymbol{u}. The integration domain lies within the bubble, where the mass fraction of helium exceeds 0.05 (Y≥0.05Y \ge0.05):

Γ=∬Y≥0.05ω~ dV~.(4)\Gamma= \iint_{Y \ge0.05} \widetilde{\omega}\,\mathrm{d}\widetilde{V}. \tag*{(4)}

Once the total circulation is determined, the non-dimensional Reynolds number can be defined as Re=Γ/νRe = \Gamma/\nu [38]. Given the constant average density, ρ‾=(ρ1′+ρ2′)/2\overline{\rho} = (\rho_1^{\prime} + \rho_2^{\prime})/2, and constant dynamic viscosity, μ\mu, the kinetic viscosity ν=μ/ρ‾\nu= \mu/\overline{\rho} remains the same throughout the cases. Therefore, the careful selection of the bubble’s geometric parameters, RcoreR_{core}, WW and RR, is essential to control the circulation Γ\Gamma to manage the Reynolds number ReRe, as discussed in Appendix A2. Finally, these parameters for the cases with the same Reynolds number ReRe, but different strength of hydrodynamic instability, are confirmed and listed in Table 1. With the same circulation Γ\Gamma, the equality of another dimensionless parameter related to mixing, the Péclet number Pe=Γ/DPe = \Gamma/\mathcal{D} [39] is also guaranteed.

ξ\xiRcoreR_{core} (mm)WW (mm)RR (mm)
0.12.0700.2072.277
1.01.3901.3902.780
5.00.5902.9503.540
10.00.3403.4003.780

Table 1. Geometric parameters for the two-dimensional SBI cases in this paper

Another aspect considered in this section is the selection of characteristic time and length to normalize the flow field for the cases outlined in Table 1, each with different length scales. The details of this selection process are depicted in Appendix A3. The appropriate characteristic length xx and time tt considering the diffusive layer are defined as:

{x∗=RReff,t∗=RReffΓ,Reff=Γu1′,(5)\left\{ \begin{aligned} x^{*} &= \sqrt{R R_{eff}}, \\ t^{*} &= \frac{R R_{eff}}{\Gamma}, \\ R_{eff} &= \frac{\Gamma}{u_1^{\prime}}, \end{aligned} \right. \tag*{(5)}

ReffR_{\mathrm{eff}} indicates the effective radius. Utilizing the definition in Eq. (5), the characteristic velocity and vorticity are

{u∗=x∗t∗=ΓRReff,ω∗=u∗x∗=ΓRReff.(6)\begin{cases} u^{*} = \dfrac{x^{*}}{t^{*}} = \dfrac{\Gamma}{\sqrt{RR_{\mathrm{eff}}}},\\ \omega^{*} = \dfrac{u^{*}}{x^{*}} = \dfrac{\Gamma}{RR_{\mathrm{eff}}}. \tag*{(6)} \end{cases}

Furthermore, the characteristic density ρ∗\rho^{*} can be given as the average density according to the definition of kinetic viscosity ν=μ/ρˉ\nu= \mu/\bar{\rho}:

ρ∗=(ρ1′+ρ2′)/2.(7)\rho^{*} = (\rho_{1}^{\prime}+\rho_{2}^{\prime})/2. \tag*{(7)}

By using these definitions, the physical parameters x~,t~,u~,ω~,ρ~\tilde{x},\tilde{t},\tilde{u},\tilde{\omega},\tilde{\rho} are normalized as

{x=x~x∗,t=t~−t0t∗,u=u~u∗,ω=ω~ω∗,ρ=ρ~ρ∗.(8)\begin{cases} x = \dfrac{\tilde{x}}{x^{*}},\\ t = \dfrac{\tilde{t}-t_{0}}{t^{*}},\\ u = \dfrac{\tilde{u}}{u^{*}},\\ \omega= \dfrac{\tilde{\omega}}{\omega^{*}},\\ \rho= \dfrac{\tilde{\rho}}{\rho^{*}}. \tag*{(8)} \end{cases}

where t0t_{0} is the moment when the shock wave passes through the right edge of the bubble. Once the physical parameters are non-dimensionalized, the numerical results with different length scales listed in Table 1 become comparable. Consequently, the influence of the hydrodynamic instability induced by initial diffusion on VD mixing in two-dimensional SBI can be investigated, as demonstrated in the next section.

Results and discussions

Hydrodynamic characteristics of SBHI

Temporal morphology of the bubble

Using the dimensionless time tt in Eq. (8), we qualitatively demonstrate the temporal evolution of the bubble’s morphology. The density and vorticity contours for the cases with different diffusive layer ratios ξ\xi are presented in Figure 3 and Figure 4, respectively. It is evident that the bubble’s morphology across different cases exhibits similar characteristics at the same dimensionless time, thus confirming the validity of the dimensionless time definition.

Temporal evolution of the density contour for four cases with different diffusive layer ratio

Figure 3. The temporal evolution of the density contour of the four cases with different diffusive layer ratio ξ\xi.

Temporal evolution of the vorticity contour for four cases with different diffusive layer ratio

Figure 4. The temporal evolution of the vorticity contour of the four cases with different diffusive layer ratio ξ\xi. The isoline of Y=0.05Y = 0.05 is plotted as the black dashed-dot line.

Furthermore, these density and vorticity contours illustrate that SBI is a complex physical phenomenon strongly coupled with wave patterns, bubble deformation, and vortex dynamics. Regarding the wave patterns captured by the density contour in Figure 3, the first relevant shock-related structure is the transmitted shock (TS), attributed to the faster propagation speed of the shock wave in the light bubble compared to the ambient air. The second structure is the reflected rarefaction wave (RRW), a result of the shock’s interaction with the light bubble. Additionally, two canonical shock wave structures, Mach stem (MS) and triple points (TP), are also observed. The density contour in Figure 3 also reveals flow structures with respect to the bubble’s deformation. The bubble is primarily stirred by a main vortex (MV), along with the presence of deformed trailing bubble (TB). The air jet (AJ) connects the bubble’s downstream edge, forming the bridge structure (Br), as reported in studies on shock-heavy bubble interactions [20]. The irregular interface, stemming from small-scale structures, is observed in the cases of ξ=0.1\xi= 0.1 and ξ=1.0\xi= 1.0, which is indicative of the presence of the hydrodynamic instability. Conversely, such instability is not evident in the cases of ξ=5.0\xi= 5.0 and ξ=10.0\xi= 10.0. The vorticity-related structures are depicted in the vorticity contour shown in Figure 4. Initially, at t=0t = 0, the baroclinic vortex (BV) is located at the bubble edge, arising from the misalignment of the pressure gradient (∇p~\nabla\tilde{p}) and the density gradient (∇ρ~\nabla\tilde{\rho}). The width of the diffusive layer causes variation in the distribution of BV at t=0.0t = 0.0. As the bubble evolves, secondary baroclinic vorticity (SBV) emerges, manifested as dominant-negative vorticity intensifying and rolling up with the positive vorticity. Subsequently, BV and SBV are aggregated together to form the main vortex (MV) structure. The vorticity deposited at the slipstream (SS) forms small-scale vortex, driven by the Kelvin–Helmholtz instability. The presence of small-scale vorticity structures in the cases of ξ=0.1\xi= 0.1 and ξ=1.0\xi= 1.0 also provides evidence of the occurrence of hydrodynamic instability, which is not evident in the cases of ξ=5.0\xi= 5.0 and ξ=10.0\xi= 10.0.

Vorticity dynamics in SBHI

After conducting a qualitative study of the morphology of bubble, we perform a qualitative analysis to explore the hydrodynamic characteristics of the instability induced by initial diffusion. Two fundamental parameters, the total circulation Γ\Gamma, and the compression rate η\eta, are presented firstly:

{Γ=∬Y≥0.05ω~ dV,η=∬Y≥0.05dV~V~0=V~bubbleV~0.(9)\begin{cases} \Gamma= \iint_{Y\geq0.05}\tilde{\omega}\,dV,\\ \eta= \dfrac{\iint_{Y\geq0.05}d\tilde{V}}{\tilde{V}_{0}} = \dfrac{\tilde{V}_{\mathrm{bubble}}}{\tilde{V}_{0}}. \tag*{(9)} \end{cases}

The circulation Γ\Gamma signifies the strength of the main vortex (MV) in Figure 3 and Figure 4, while the compression rate η\eta reflects the primary shock compression. The temporal evolution of Γ\Gamma and η\eta is depicted in Figure 5. Due to the careful case

Comparison of the temporal variation of the circulation and compression rate

Figure 5. Comparison of the temporal variation of the circulation Γ\Gamma and the compression rate η\eta

design detailed in Appendix A2, after the dimensionless moment t=0t=0, when the shock wave propagates through the right edge of the bubble, the circulation Γ\Gamma and compression rate η\eta stabilize at approximately 1.7 m2/s1.7\ \mathrm{m^2/s} and 0.40.4, respectively.

Moreover, in Figure 4, we observe that the primary discrepancy within the vorticity field lies in the intensity of SBV. Thus the total circulation Γ\Gamma is decomposed to the positive part Γ+\Gamma^{+} and negative part Γ−\Gamma^{-}:

Γ=Γ++Γ−,Γ+=⟨ω~⟩∣ω~>0=(∬Y≥0.05ω~ dV~)∣ω~>0,Γ−=⟨ω~⟩∣ω~<0=(∬Y≥0.05ω~ dV~)∣ω~<0.(10)\begin{aligned} \Gamma&= \Gamma^{+}+\Gamma^{-},\\ \Gamma^{+} &= \langle\widetilde{\omega}\rangle_{\lvert\widetilde{\omega}>0}=\left(\iint_{Y\geq0.05}\widetilde{\omega}\,\mathrm{d}\widetilde{V}\right)_{\lvert\widetilde{\omega}>0},\\ \Gamma^{-} &= \langle\widetilde{\omega}\rangle_{\lvert\widetilde{\omega}<0}=\left(\iint_{Y\geq0.05}\widetilde{\omega}\,\mathrm{d}\widetilde{V}\right)_{\lvert\widetilde{\omega}<0}. \tag*{(10)} \end{aligned}

In Figure 6, following t=0t=0, both the positive and negative circulation, denoted as Γ+\Gamma^{+} and Γ−\Gamma^{-}, exhibit identical growth rates, which confirms the conservative nature of the total circulation Γ\Gamma. It is also evident that at t=2t=2, the positive circulation Γ+\Gamma^{+} reaches its peak value, mirroring the behavior of the negative circulation Γ−\Gamma^{-}. Due to the growth rate and peak value of the positive circulation Γ+\Gamma^{+} originate from the secondary baroclinic vorticity (SBV) shown in Figure 4, as will be validated by the budget analysis shortly, this hydrodynamic instability resulting from the initial diffusion is labeled as the secondary baroclinic hydrodynamic instability (SBHI), and can be measured by the discrepancy between the peak value of positive circulation and the total circulation Γsbv+=(Γ+)∣peak−Γ\Gamma^{+}_{\mathrm{sbv}}=(\Gamma^{+})_{\lvert\mathrm{peak}}-\Gamma.

Temporal evolution of the circulation, positive circulation and negative circulation

Figure 6. Temporal evolution of the circulation Γ\Gamma, positive circulation Γ+\Gamma^{+} and negative circulation Γ−\Gamma^{-}. Γsbv+=(Γ+)∣peak−Γ\Gamma^{+}_{\mathrm{sbv}}=(\Gamma^{+})_{\lvert\mathrm{peak}}-\Gamma indicates the discrepancy between the peak value of positive circulation and the total circulation at t=2t=2.

We further analyze the positive circulation budget to identify the source of Γsbv+\Gamma^{+}_{\mathrm{sbv}}. The vorticity transport equation is expressed as:

Dω~iDt=−(∂u~k∂x~k)ω~i+S~ijω~j+ϵijk1ρ~2∂ρ~∂x~j∂p~∂x~k+ν∂2ω~i∂x~k2(11)\frac{D\widetilde{\omega}_{i}}{D t}=-\left(\frac{\partial\widetilde{u}_{k}}{\partial\widetilde{x}_{k}}\right)\widetilde{\omega}_{i}+\widetilde{S}_{ij}\widetilde{\omega}_{j}+\epsilon_{ijk}\frac{1}{\widetilde{\rho}^{2}}\frac{\partial\widetilde{\rho}}{\partial\widetilde{x}_{j}}\frac{\partial\widetilde{p}}{\partial\widetilde{x}_{k}}+\nu\frac{\partial^{2}\widetilde{\omega}_{i}}{\partial\widetilde{x}_{k}^{2}} \tag*{(11)}

The first term on the right side of Eq. (11), −(∂u~k∂x~k)ω~i-\left(\frac{\partial\widetilde{u}_{k}}{\partial\widetilde{x}_{k}}\right)\widetilde{\omega}_{i}, corresponds to the compression term resulting from the velocity divergence ∂u~k∂x~k\frac{\partial\widetilde{u}_{k}}{\partial\widetilde{x}_{k}} and is denoted as T~divU\widetilde{T}_{divU}. The second term S~ijω~j\widetilde{S}_{ij}\widetilde{\omega}_{j} represents the vortex stretch term, and this term vanishes in two-dimensional flow. The third term ϵijk1ρ~2∂ρ~∂x~j∂p~∂x~k\epsilon_{ijk}\frac{1}{\widetilde{\rho}^{2}}\frac{\partial\widetilde{\rho}}{\partial\widetilde{x}_{j}}\frac{\partial\widetilde{p}}{\partial\widetilde{x}_{k}} constitutes the baroclinic torque originating from the misalignment of pressure gradient ∇p~\nabla\widetilde{p} and density gradient ∇ρ~\nabla\widetilde{\rho} and is indicated as B~\widetilde{B}. The fourth term ν∂2ω~i∂x~k2\nu\frac{\partial^{2}\widetilde{\omega}_{i}}{\partial\widetilde{x}_{k}^{2}} represents the viscous term and is labeled as T~ν\widetilde{T}_{\nu}. In two-dimensional SBI cases, the vorticity ω~i\widetilde{\omega}_{i} only aligns with the ZZ coordinate, so it can be expressed as ω~\widetilde{\omega}. Then, Eq. (11) is simplified to:

Dω~Dt=T~divU+B~+T~ν.(12)\frac{D\widetilde{\omega}}{D t}=\widetilde{T}_{divU}+\widetilde{B}+\widetilde{T}_{\nu}. \tag*{(12)}

According to the definition of positive circulation in Eq. (10), the time derivative of Γ+\Gamma^{+} is:

DΓ+Dt=D⟨ω~⟩∣ω~>0Dt=⟨Dω~Dt⟩∣ω~>0+⟨(∂u~k∂x~k)ω~⟩∣ω~>0=⟨Dω~Dt⟩∣ω~>0−⟨T~divU⟩∣ω~>0.(13)\begin{aligned} \frac{D\Gamma^{+}}{D t}=\frac{D\langle\widetilde{\omega}\rangle_{\lvert\widetilde{\omega}>0}}{D t}=\left\langle\frac{D\widetilde{\omega}}{D t}\right\rangle_{\lvert\widetilde{\omega}>0}+\left\langle\left(\frac{\partial\widetilde{u}_{k}}{\partial\widetilde{x}_{k}}\right)\widetilde{\omega}\right\rangle_{\lvert\widetilde{\omega}>0} \\ &=\left\langle\frac{D\widetilde{\omega}}{D t}\right\rangle_{\lvert\widetilde{\omega}>0}-\left\langle\widetilde{T}_{divU}\right\rangle_{\lvert\widetilde{\omega}>0}. \tag*{(13)} \end{aligned}

Substituting the vorticity transport equation Eq. (12), Eq. (13) transforms to:

DΓ~+Dt~=⟨B~⟩∣ω~∣>0+⟨T~v⟩∣ω~∣>0=⟨B+⟩+⟨Tv+⟩(14)\frac{D\tilde{\Gamma}^{+}}{D\tilde{t}}=\left\langle\tilde{B}\right\rangle_{\left|\tilde{\omega}\right|>0}+\left\langle\tilde{T}_{v}\right\rangle_{\left|\tilde{\omega}\right|>0}=\left\langle B^{+}\right\rangle+\left\langle T_{v}^{+}\right\rangle \tag*{(14)}

The left-hand side of Eq. (14) can be normalized as following by employing the characteristic vorticity, time, and length in Eq. (5) and Eq. (6):

DΓ~+Dt~=D∬Y≥0.05ω~ dV~Dt~∣ω~>0=D∬Y≥0.05(ωω∗) d(x∗2V)D(tt∗)∣ω>0=Γ2RReffD⟨ω⟩∣ω∣>0Dt=Γ2RReffDΓn+Dt.(15)\begin{aligned} \frac{D\tilde{\Gamma}^{+}}{D\tilde{t}}=\left.\frac{D\iint_{Y\geq0.05}\tilde{\omega}\,d\tilde{V}}{D\tilde{t}}\right|_{\tilde{\omega}>0}=\left.\frac{D\iint_{Y\geq0.05}(\omega\omega^{*})\,d(x^{*2}V)}{D(tt^{*})}\right|_{\omega>0} \\ &=\frac{\Gamma^{2}}{RR_{\mathrm{eff}}}\frac{D\langle\omega\rangle_{\left|\omega\right|>0}}{Dt}=\frac{\Gamma^{2}}{RR_{\mathrm{eff}}}\frac{D\Gamma_{n}^{+}}{Dt}. \tag*{(15)} \end{aligned}

Γn+=∬Y≥0.05ω dV=⟨ω⟩∣ω∣>0\Gamma_{n}^{+}=\iint_{Y\geq0.05}\omega\,dV=\langle\omega\rangle_{\left|\omega\right|>0} indicates the dimensionless positive circulation, which can also relate to the total circulation by:

Γ+=∬Y≥0.05(ωω∗) d(x∗2V)D(tt∗)=Γ∬Y≥0.05ω dV=Γ⟨ω⟩∣ω∣>0=ΓΓn+.(16)\begin{aligned} \Gamma^{+}=\iint_{Y\geq0.05}\frac{(\omega\omega^{*})\,d(x^{*2}V)}{D(tt^{*})} \\ &=\Gamma\iint_{Y\geq0.05}\omega\,dV=\Gamma\langle\omega\rangle_{\left|\omega\right|>0}=\Gamma\Gamma_{n}^{+}. \tag*{(16)} \end{aligned}

Subsequently, Eq. (14) is transformed to:

DΓn+Dt~=RReffΓ2(⟨B+⟩+⟨Tv+⟩)=⟨B+⟩+⟨Tv+⟩.(17)\begin{aligned} \frac{D\Gamma_{n}^{+}}{D\tilde{t}}=\frac{RR_{\mathrm{eff}}}{\Gamma^{2}}\left(\left\langle B^{+}\right\rangle+\left\langle T_{v}^{+}\right\rangle\right) \\ &=\left\langle B^{+}\right\rangle+\left\langle T_{v}^{+}\right\rangle. \tag*{(17)} \end{aligned}

In Eq. (17), DΓn+Dt\frac{D\Gamma_{n}^{+}}{Dt} can be simplified as:

DΓn+Dt=D⟨ω⟩∣ω∣>0Dt=⟨DωDt⟩∣ω∣>0+⟨ω∇⋅u⟩∣ω∣>0=⟨∂ω∂t⟩∣ω∣>0+⟨u⋅∇ω⟩∣ω∣>0+⟨ω∇⋅u⟩∣ω∣>0=⟨∂ω∂t⟩∣ω∣>0+⟨∇⋅(ωu)⟩∣ω∣>0=∂⟨ω⟩∣ω∣>0∂t.(18)\begin{aligned} \frac{D\Gamma_{n}^{+}}{Dt}=\frac{D\langle\omega\rangle_{\left|\omega\right|>0}}{Dt}=\left\langle\frac{D\omega}{Dt}\right\rangle_{\left|\omega\right|>0}+\left\langle\omega\nabla\cdot u\right\rangle_{\left|\omega\right|>0} \\ &=\left\langle\frac{\partial\omega}{\partial t}\right\rangle_{\left|\omega\right|>0}+\left\langle u\cdot\nabla\omega\right\rangle_{\left|\omega\right|>0}+\left\langle\omega\nabla\cdot u\right\rangle_{\left|\omega\right|>0} \\ &=\left\langle\frac{\partial\omega}{\partial t}\right\rangle_{\left|\omega\right|>0}+\left\langle\nabla\cdot(\omega u)\right\rangle_{\left|\omega\right|>0}=\frac{\partial\langle\omega\rangle_{\left|\omega\right|>0}}{\partial t}. \tag*{(18)} \end{aligned}

By utilizing the relationship that the spatial integration of the divergence of vorticity flux ⟨∇⋅(ωu)⟩=0\langle\nabla\cdot(\omega u)\rangle=0 if the integration boundary is ω=0\omega=0 [24], we can replace the unavailable material derivative with the easily computed time derivative. Finally, the simplest form of Eq. (17) is:

DΓn+Dt=∂⟨ω⟩∣ω∣>0∂t=⟨B+⟩+⟨Tv+⟩.(19)\frac{D\Gamma_{n}^{+}}{Dt}=\frac{\partial\langle\omega\rangle_{\left|\omega\right|>0}}{\partial t}=\left\langle B^{+}\right\rangle+\left\langle T_{v}^{+}\right\rangle. \tag*{(19)}

The time derivative of the normalized positive circulation Γn+\Gamma_{n}^{+} and its corresponding source terms are depicted in Figure 7. It is evident that the green line, representing the sum of terms on the right-hand side of Eq. (19) ⟨Tright⟩\langle T_{\mathrm{right}}\rangle, closely matches the red line depicting the time derivative of the dimensionless positive circulation, DΓn+Dt\frac{D\Gamma_{n}^{+}}{Dt}. This observation validates the reliability of numerical results in capturing vorticity dynamics within the grid, as demonstrated in Appendix A1.

Time derivative of the normalized positive circulation and corresponding source terms

Figure 7. Time derivative of the normalized positive circulation Γn+\Gamma_n^{+} and the corresponding source terms in Eq. (19).

within the grid, as demonstrated in Appendix A1. It can also be observed before t=2t=2, the positive baroclinic torque ⟨B+⟩\langle B^{+}\rangle dominates the growth of Γ+\Gamma^{+}; after t=2t=2, the negative viscous term ⟨Tv+⟩\langle T_{v}^{+}\rangle becomes the major factor hindering the growth of Γ+\Gamma^{+}. This observation explains why the moment when Γ+\Gamma^{+} reaches its peak value occurs at t=2t=2. The temporal evolution of ⟨B+⟩\langle B^{+}\rangle of different cases is compared in Figure 8, and this comparison reveals the variation in Γsbv+\Gamma_{sbv}^{+} arises from different baroclinic torque ⟨B+⟩\langle B^{+}\rangle. In summary, we can conclude that the hydrodynamic instability induced by the initial diffusion can be labeled as the SBHI here, as it originates from the baroclinic torque ⟨B+⟩\langle B^{+}\rangle, and we can measure the strength of SBHI by the discrepancy between the peak value of positive circulation and the total circulation Γsbv+\Gamma_{sbv}^{+}.

Comparison of the temporal evolution of baroclinic torque

Figure 8. Comparison of the temporal evolution of baroclinic torque ⟨B+⟩\langle B^{+}\rangle, which is the primary source term of positive circulation Γ+\Gamma^{+}.

Effect of SBHI on mixing

Mixedness dynamics of SBHI

In section 3.1, the hydrodynamic instability induced by the initial diffusion, which has been shown to originate from the baroclinic torque, is detonated as the SBHI. In this section, we focus on analyzing the effect of SBHI on the mixing characteristics. Two mixing parameters are employed to characterize the mixing process of SBI. Firstly, the mixedness ff is used to describe the local mixing extent of the mixture. Cetegon et al. [18] defined mixedness ff by the equation:

f=4Y(1−Y).(20)f=4Y(1-Y). \tag*{(20)}

YY represents the mass fraction of the relevant gas component, chosen as helium in this study. The mixedness ff reaches its maximum when the mixture is in a state where helium and the ambient air mix in a 1:1 proportion.

Secondly, the scalar dissipation rate (SDR) is considered. Buch et al. [19] defined SDR χ~\tilde{\chi} as:

χ~=∂Y∂x~i∂Y∂x~i.(21)\tilde{\chi}=\frac{\partial Y}{\partial\tilde{x}_i}\frac{\partial Y}{\partial\tilde{x}_i}. \tag*{(21)}

∂Y∂x~i\frac{\partial Y}{\partial\tilde{x}_i} indicates the spatial gradient of the mass fraction YY, referred to as the scalar gradient in the subsequent content. The SDR is the square of the modulus of the scalar gradient and represents the local mixing rate according to the mixedness transport equation:

DfDt~=8Dχ~+D∂2f∂x~j2+4(1−2Y)Dρ~∂ρ~∂x~j∂Y∂x~j.(22)\frac{Df}{D\tilde{t}}=8\mathcal{D}\tilde{\chi}+\mathcal{D}\frac{\partial^{2}f}{\partial\tilde{x}_{j}^{2}}+4(1-2Y)\frac{\mathcal{D}}{\tilde{\rho}}\frac{\partial\tilde{\rho}}{\partial\tilde{x}_{j}}\frac{\partial Y}{\partial\tilde{x}_{j}}. \tag*{(22)}

The first term in Eq. (22), 8Dχ~8\mathcal{D}\tilde{\chi}, labeled as the SDR term T~SDR\tilde{T}_{SDR}, explains why SDR χ~\tilde{\chi} represents the local mixing rate. The second term, D∂2f∂x~j2\mathcal{D}\frac{\partial^{2}f}{\partial\tilde{x}_{j}^{2}} is the diffusion term T~diff\tilde{T}_{diff}, which describes the transport of the mixedness ff by molecular diffusion. The third term, 4(1−2Y)Dρ~∂ρ~∂x~j∂Y∂x~j4(1-2Y)\frac{\mathcal{D}}{\tilde{\rho}}\frac{\partial\tilde{\rho}}{\partial\tilde{x}_{j}}\frac{\partial Y}{\partial\tilde{x}_{j}}, arises from the production of the density gradient ∂ρ~∂x~j\frac{\partial\tilde{\rho}}{\partial\tilde{x}_{j}} and the scalar gradient ∂Y∂x~j\frac{\partial Y}{\partial\tilde{x}_{j}}, and is denoted as T~den\tilde{T}_{den}.

The temporal evolution of SDR (χ~\tilde{\chi}) contour and mass fraction (YY) contour is shown in Figure 9. Due to the similarities in the morphology of χ~\tilde{\chi} and YY in the cases of ξ=5.0\xi=5.0 and ξ=10.0\xi=10.0, only the mixing field of ξ=5.0\xi=5.0 is illustrated. In the early stage of SBI mixing, a region with a high mixing rate is observed around the bridge (Br) structure, this observation is consistent with the previous experiments conducted by Tomkins [20]. Subsequently, during the formation of the main vortex (MV), mixing is intensified by the stirring effect, leading to a high mixing rate inside the MV. As the mixing process approaches completion in the late stage, the SDR diminishes to a relatively low level. It is evident that the SBHI, characterized by the irregular interface originating from small-scale structures, amplifies the magnitude of SDR, thereby enhancing the mixing rate. This qualitative conclusion is supported by the quantitative analysis of mixing parameters in the following content.

Temporal evolution of SDR contour and mass fraction contour for three cases with different diffusive layer ratio

Figure 9. The temporal evolution of SDR χ~\widetilde{\chi} contour (upper) and mass fraction contour YY (lower) of three cases with different diffusive layer ratio (a) ξ=0.1\xi= 0.1, (b) ξ=1.0\xi= 1.0 and (c) ξ=5.0\xi= 5.0. The isoline of Y=0.05Y = 0.05 is plotted as red dashed-dot line.

To facilitate a quantitative comparison of mixing parameters, it is essential to transform the terms in Eq. (22) into the dimensionless form. Using the scaling method presented in Eq. (8), each term in Eq. (22) can be normalized as:

{DfDt~=DfD(τ∗t)=ΓRReffDfDt,8Dχ~=8D∂Y∂x~j∂Y∂x~j=8D∂Y∂xj∗∂Y∂xj∗=8DRReffχ,D∂2f∂x~j2=D∂2f∂(x∗xj)2=DRReff∂2f∂xj2,4(1−2Y)Dρ~∂ρ~∂x~j∂Y∂x~j=4(1−2Y)DRReff∂ρ∂xj∂Y∂xj.(23)\begin{cases} \frac{Df}{D\tilde{t}}=\frac{Df}{D(\tau^{*}t)}=\frac{\Gamma}{RR_{eff}}\frac{Df}{Dt},\\ 8\mathcal{D}\tilde{\chi}=8\mathcal{D}\frac{\partial Y}{\partial\tilde{x}_{j}}\frac{\partial Y}{\partial\tilde{x}_{j}}=8\mathcal{D}\frac{\partial Y}{\partial x_{j}^{*}}\frac{\partial Y}{\partial x_{j}^{*}}=8\frac{\mathcal{D}}{RR_{eff}}\chi,\\ \mathcal{D}\frac{\partial^{2}f}{\partial\tilde{x}_{j}^{2}}=\mathcal{D}\frac{\partial^{2}f}{\partial(x^{*}x_{j})^{2}}=\frac{\mathcal{D}}{RR_{eff}}\frac{\partial^{2}f}{\partial x_{j}^{2}},\\ 4(1-2Y)\frac{\mathcal{D}}{\tilde{\rho}}\frac{\partial\tilde{\rho}}{\partial\tilde{x}_{j}}\frac{\partial Y}{\partial\tilde{x}_{j}}=4(1-2Y)\frac{\mathcal{D}}{RR_{eff}}\frac{\partial\rho}{\partial x_{j}}\frac{\partial Y}{\partial x_{j}}. \tag*{(23)} \end{cases}

Substituting these normalized terms to Eq. (22), the dimensionless mixedness transport equation is:

DfDt=8Peχ+Pe∂2f∂xj2+4Pe(1−2Y)1ρ∂ρ∂xj∂Y∂xj=TSDR+Tdiff+Tden.(24)\begin{aligned} \frac{Df}{Dt}=8Pe\chi+Pe\frac{\partial^{2}f}{\partial x_{j}^{2}}+4Pe(1-2Y)\frac{1}{\rho}\frac{\partial\rho}{\partial x_{j}}\frac{\partial Y}{\partial x_{j}} \\ &=T_{SDR}+T_{diff}+T_{den}. \tag*{(24)} \end{aligned}

In this equation, Pe=Γ/DPe=\Gamma/\mathcal{D} is the Péclet number, which remains constant due to the uniform circulation Γ\Gamma across the cases.

The spatial integration of mixedness ⟨f⟩=∬Y≥0.05f dV\langle f\rangle=\iint_{Y\geq0.05}f\,dV is employed to reflect the global mixing extent, the corresponding transport equation is:

D⟨f⟩Dt=⟨DfDt⟩+⟨f(∂uk∂xk)⟩=⟨TSDR⟩+⟨Tdiff⟩+⟨Tden⟩+⟨TdivU⟩=⟨TSDR⟩+⟨Tden⟩+⟨TdivU⟩.(25)\begin{aligned} \frac{D\langle f\rangle}{Dt}=\left\langle\frac{Df}{Dt}\right\rangle+\left\langle f\left(\frac{\partial u_{k}}{\partial x_{k}}\right)\right\rangle\\ &=\langle T_{SDR}\rangle+\langle T_{diff}\rangle+\langle T_{den}\rangle+\langle T_{divU}\rangle\\ &=\langle T_{SDR}\rangle+\langle T_{den}\rangle+\langle T_{divU}\rangle. \tag*{(25)} \end{aligned}

where ⟨Tdiff⟩=0\langle T_{diff}\rangle=0 owing to the mixedness flux ∇f\nabla f equals to 0 at the boundary the integration region [23].

Figure 10 displays the temporal evolution of ⟨f⟩\langle f\rangle. The diffusive layer ratio ξ\xi reflects the initial diffusion of the bubble, hence the initial global mixedness ⟨f⟩\langle f\rangle at t=0t=0 is larger in cases with larger ξ\xi. Subsequently, there is a decrease of ⟨f⟩\langle f\rangle in the cases with larger ξ\xi during the early stage, and then an increase throughout the mixing process. Additionally, it can be observed that in cases with a smaller diffusive layer ratio ξ\xi, representing stronger SBHI, the increase of ⟨f⟩\langle f\rangle is more pronounced, thus confirming the quantitative conclusion obtained in Figure 9.

Temporal evolution of the spatial integration of mixedness

Figure 10. Temporal evolution of the spatial integration of mixedness ⟨f⟩\langle f\rangle.

The budget analysis of Eq. (25) in Figure 11 presents the sources contributing to the growth of global mixedness ⟨f⟩\langle f\rangle. The time derivative of global mixedness D⟨f⟩Dt\frac{D\langle f\rangle}{Dt}, denoted by the red line, closely aligns with the green line representing ⟨Tright⟩\langle T_{right}\rangle, the sum of the terms on the right-hand side of Eq. (25). This observation validates the trustworthiness of the numerical results in terms of the resolution of mixedness dynamics within the grid, as detailed in Appendix A1. Notably, the primary source of mixedness growth is the SDR term ⟨TSDR⟩\langle T_{SDR}\rangle, depicted by the black line. This phenomenon explains the physical meaning of SDR χ\chi as a measure of mixing rate. In contrast, the density term ⟨Tden⟩\langle T_{den}\rangle illustrated by blue line approaches 0, while the velocity divergence term ⟨TdivU⟩\langle T_{divU}\rangle shown by orange line is predominantly negative throughout the mixing process, particularly during the early stage, contributing to the decrease in ⟨f⟩\langle f\rangle in cases with large ξ\xi.

Time derivative of the spatial integration of mixedness

Figure 11. Time derivative of the spatial integration of mixedness ⟨f⟩\langle f\rangle defined in Eq. (20) and the corresponding source terms in Eq. (25).

This budget analysis prompts a modification in the definition of global mixedness:

⟨f∗⟩=⟨f⟩−∫0t⟨f(∂uk∂xk)⟩dt′,(26)\langle f^* \rangle= \langle f \rangle- \int_0^t \left\langle f\left(\frac{\partial u_k}{\partial x_k}\right)\right\rangle\mathrm{d}t', \tag*{(26)}

The transport equation of this parameter is:

D⟨f∗⟩Dt=⟨TSDR⟩+⟨Tden⟩(27)\frac{D\langle f^*\rangle}{Dt} = \langle T_{SDR}\rangle+ \langle T_{den}\rangle \tag*{(27)}

The temporal evolution of the increment of the modified global mixedness, ⟨f∗⟩Δ=⟨f∗⟩−⟨f∗⟩∣t=0\langle f^*\rangle_\Delta=\langle f^*\rangle-\langle f^*\rangle|_{t=0}, is shown in Figure 12. Furthermore, the time derivative of the modified global mixedness ⟨f∗⟩\langle f^*\rangle and its corresponding source term are illustrated in Figure 13. Additionally, the SDR term ⟨TSDR⟩\langle T_{SDR}\rangle and time derivative of modified mixedness are presented in Figure 14. From these figures, it can be deduced that the mixing rate, as indicated by the SDR term ⟨TSDR⟩\langle T_{SDR}\rangle, is enhanced by SBHI, consequently leading to a larger increment of the modified global mixedness ⟨f∗⟩Δ\langle f^*\rangle_\Delta.

Comparison of the time evolution of the SDR term and the time derivative of the modified mixedness

Figure 14. Comparison of the time evolution of the SDR term ⟨TSDR⟩\langle T_{SDR}\rangle and the time derivative of the modified mixedness d⟨f∗⟩/dt\mathrm{d}\langle f^*\rangle/\mathrm{d}t.

Temporal evolution of the increment of spatial integration of modified mixedness

Figure 12. Temporal evolution of the increment of spatial integration of modified mixedness ⟨f∗⟩Δ\langle f^*\rangle_\Delta.

Time derivative of the spatial integration of modified mixedness

Figure 13. Time derivative of the spatial integration of modified mixedness ⟨f∗⟩\langle f^*\rangle defined in Eq. (26) and the corresponding source terms in in Eq. (27).

Scaling behavior of mixing rate on SBV number Ωsbv\Omega_{sbv}

The observations in the aforementioned research indicate the SBHI, quantified by Γsbv+\Gamma^+_{sbv}, enhances the mixing rate as described by the SDR term ⟨TSDR⟩\langle T_{SDR}\rangle in Eq. (27). This section introduces a new dimensionless number Ωsbv\Omega_{sbv} to capture the strength of Γsbv+\Gamma^+_{sbv}, essentially measuring the intensity of SBHI. Additionally, the scaling behavior of the mixing rate in relation to the SBV number Ωsbv\Omega_{sbv} is demonstrated.

As discussed in § 1.2.1.2, the difference in Γsbv+\Gamma^+_{sbv} originates from the variations of ⟨B+⟩\langle B^+\rangle in Eq. (14). Hence, dimensional analysis is employed here to propose the SBV number Ωsbv\Omega_{sbv}.

The detailed form of Eq. (14) is:

DΓn+Dt=D(Γ+/Γ)Dt=⟨B+⟩+⟨Tv+⟩=RReffΓ2(⟨ϵijkρ~2∂ρ~∂xj∂p~∂xk∣ω~>0⟩+⟨ν∂2ω~∂xk2∣ω~>0⟩).(28)\frac{D\Gamma_n^+}{Dt}=\frac{D(\Gamma^+/\Gamma)}{Dt}=\langle B^+\rangle+\langle T_v^+\rangle =\frac{RR_{eff}}{\Gamma^2}\left(\left\langle\left.\frac{\epsilon_{ijk}}{\tilde{\rho}^2}\frac{\partial\tilde{\rho}}{\partial x_j}\frac{\partial\tilde{p}}{\partial x_k}\right|_{\tilde{\omega}>0}\right\rangle+\left\langle\left.\nu\frac{\partial^2\tilde{\omega}}{\partial x_k^2}\right|_{\tilde{\omega}>0}\right\rangle\right). \tag*{(28)}

Using the scaling method in Eq. (8), the viscous term in this equation can be normalized:

⟨Tv+⟩=RReffΓ2∬ω~>0ν∂2ω~∂xk2dV~=RReffΓ2∬ω>0ν∂2ω∗∂xk∗2x∗2dV=∬ω>0νΓ∂2ω∂xk2dV=⟨1Re∂2ω∂xk2⟩ω>0.(29)\langle T_v^+\rangle=\frac{RR_{eff}}{\Gamma^2}\iint_{\tilde{\omega}>0}\nu\frac{\partial^2\tilde{\omega}}{\partial x_k^2}\mathrm{d}\tilde{V} =\frac{RR_{eff}}{\Gamma^2}\iint_{\omega>0}\nu\frac{\partial^2\omega^*}{\partial x_k^{*2}}x^{*2}\mathrm{d}V =\iint_{\omega>0}\frac{\nu}{\Gamma}\frac{\partial^2\omega}{\partial x_k^2}\mathrm{d}V =\left\langle\frac{1}{Re}\frac{\partial^2\omega}{\partial x_k^2}\right\rangle_{\omega>0}. \tag*{(29)}

This dimensionless form indicates the Reynolds number ReRe measures the viscous effect on the evolution of dimensionless positive circulation Γn+\Gamma_n^+.

Because the density ρ~\tilde{\rho} and pressure p~\tilde{p} distributions vary across the cases with different diffusive layer ratio ξ\xi, the den-

sity gradient ∂ρ~∂x~i\frac{\partial\tilde{\rho}}{\partial\tilde{x}_i} and pressure gradient ∂p~∂x~i\frac{\partial\tilde{p}}{\partial\tilde{x}_i} need to be modeled to propose Ωsbv\Omega_{sbv}.

Regarding the density gradient, employing the canonical correlation between mass fraction and density in multispecies miscible flows [40]:

1ρ~=1−Yρ1′+Yρ2′,(30)\frac{1}{\tilde{\rho}}=\frac{1-Y}{\rho'_1}+\frac{Y}{\rho'_2}, \tag*{(30)}

the relationship between the density gradient and the scalar gradient can be established as:

∂ρ~∂x~i=ρ~2ρ2′−ρ1′ρ1′ρ2′∂Y∂x~i.(31)\frac{\partial\tilde{\rho}}{\partial\tilde{x}_i}=\tilde{\rho}^{2}\frac{\rho'_2-\rho'_1}{\rho'_1\rho'_2}\frac{\partial Y}{\partial\tilde{x}_i}. \tag*{(31)}

Substituting this expression of density gradient into Eq. (28), the dimensionless baroclinic torque transforms to:

⟨B+⟩=RReeffΓ2⟨ϵijk1ρ~2∂ρ~∂x~j∂p~∂x~k⟩ω~>0=RReeffΓ2∬ω~>0ϵijk1ρ~2∂ρ~∂x~j∂p~∂x~kdV~=RReeffΓ2∬ω~>0ϵijkρ2′−ρ1′ρ1′ρ2′∂Y∂x~j∂p~∂x~kdV~.(32)\begin{aligned} \langle B^+\rangle &=\frac{RRe_{eff}}{\Gamma^2}\left\langle\epsilon_{ijk}\frac{1}{\tilde{\rho}^{2}}\frac{\partial\tilde{\rho}}{\partial\tilde{x}_j}\frac{\partial\tilde{p}}{\partial\tilde{x}_k}\right\rangle_{\tilde{\omega}>0}\\ &=\frac{RRe_{eff}}{\Gamma^2}\iint_{\tilde{\omega}>0}\epsilon_{ijk}\frac{1}{\tilde{\rho}^{2}}\frac{\partial\tilde{\rho}}{\partial\tilde{x}_j}\frac{\partial\tilde{p}}{\partial\tilde{x}_k}d\tilde{V}\\ &=\frac{RRe_{eff}}{\Gamma^2}\iint_{\tilde{\omega}>0}\epsilon_{ijk}\frac{\rho'_2-\rho'_1}{\rho'_1\rho'_2}\frac{\partial Y}{\partial\tilde{x}_j}\frac{\partial\tilde{p}}{\partial\tilde{x}_k}d\tilde{V}. \tag*{(32)} \end{aligned}

The scalar gradient ∂Y∂x~j\frac{\partial Y}{\partial\tilde{x}_j} is modeled first. Similar to the definition of flame width between the mass fraction from YaY_a to YbY_b in the research of premixed-flame by Hamlinton et al. [41]:

δ(Ya,Yb)=∫YaYb∣∂Y∂n∣−1dY=∫YaYbχ~−1/2(Y)dY,(33)\delta(Y_a,Y_b)=\int_{Y_a}^{Y_b}\left|\frac{\partial Y}{\partial n}\right|^{-1}dY=\int_{Y_a}^{Y_b}\tilde{\chi}^{-1/2}(Y)dY, \tag*{(33)}

the definition of average width of bubble’s mass fraction is:

δˉ=δ(0,1)=∫01χ~−1/2(Y)dY=∬χ~−1/2dV~,(34)\bar{\delta}=\delta(0,1)=\int_{0}^{1}\tilde{\chi}^{-1/2}(Y)dY=\iint\tilde{\chi}^{-1/2}d\tilde{V}, \tag*{(34)}

by which the scalar gradient is scaled as:

∂Y∂x~i=1δˉ∂Y∂xi.(35)\frac{\partial Y}{\partial\tilde{x}_i}=\frac{1}{\bar{\delta}}\frac{\partial Y}{\partial x_i}. \tag*{(35)}

The normalization of the pressure gradient is based on the assumption of a single vortex. Illustrated in Figure 15, the pressure contour suggests that the pressure field of SBI resembles that of a single vortex. By establishing a polar system with the main vortex (MV) core as the coordinate origin, the pressure gradient can be estimated using the Lamb-Oseen vortex model [17, 42]:

The pressure contour and vorticity contour in the frame of the main vortex core

Figure 15. (left) The pressure contour and (right) vorticity contour in the frame of main vortex core of the SBI case with diffusive interface ratio ξ=0.1\xi=0.1 at time t=1.3t=1.3.

∂p~∂r~=ρ~Γ24π2r~3,(36)\frac{\partial\tilde{p}}{\partial\tilde{r}}=\frac{\tilde{\rho}\Gamma^2}{4\pi^2\tilde{r}^3}, \tag*{(36)}

therefore, the pressure gradient ∂p~∂x~k\frac{\partial\tilde{p}}{\partial\tilde{x}_k} is scales as:

∂p~∂x~i=ρ∗Γ24π2(ReeffR)3/2∂p∂xi=ρ1′+ρ2′2Γ24π2(ReeffR)3/2∂p∂xi.(37)\frac{\partial\tilde{p}}{\partial\tilde{x}_i}=\frac{\rho^*\Gamma^2}{4\pi^2(Re_{eff}R)^{3/2}}\frac{\partial p}{\partial x_i}=\frac{\rho'_1+\rho'_2}{2}\frac{\Gamma^2}{4\pi^2(Re_{eff}R)^{3/2}}\frac{\partial p}{\partial x_i}. \tag*{(37)}

As the baroclinic torque B+B^+ is deposited on the bubble’s area with positive vorticity, the characteristic length in Eq. (5) is not appropriate to normalize dV~d\tilde{V} in Eq. (32), the area A=∬ω~>0dV~A=\iint_{\tilde{\omega}>0}d\tilde{V} is defined to normalize dV~d\tilde{V}, represented as: dV~=AdVd\tilde{V}=AdV. With these normalized parameters, Eq. (32) adopts the following form:

⟨B+⟩=RReeffΓ2∬ω~>0ϵijkρ2′−ρ1′ρ1′ρ2′∂Y∂x~j∂p~∂x~kdV~=RReeffΓ2∬ω~>0ϵijkρ2′−ρ1′ρ1′ρ2′∂Y∂(δxj)ρ1′+ρ2′2Γ24π2(ReeffR)3/2∂p∂xkAdV=(ρ2′−ρ1′)(ρ1′+ρ2′)ρ1′ρ2′A4π2δˉ2RReeff∬ω~>0ϵijk∂Y∂xj∂p∂xkdV=(σ−1)(σ+1)2σA4π2δˉ2RReeff⟨∂Y∂xj∂p∂xk⟩ω~>0.(38)\begin{aligned} \langle B^+\rangle &=\frac{RRe_{eff}}{\Gamma^2}\iint_{\tilde{\omega}>0}\epsilon_{ijk}\frac{\rho'_2-\rho'_1}{\rho'_1\rho'_2}\frac{\partial Y}{\partial\tilde{x}_j}\frac{\partial\tilde{p}}{\partial\tilde{x}_k}d\tilde{V}\\ &=\frac{RRe_{eff}}{\Gamma^2}\iint_{\tilde{\omega}>0}\epsilon_{ijk}\frac{\rho'_2-\rho'_1}{\rho'_1\rho'_2}\frac{\partial Y}{\partial(\delta x_j)}\frac{\rho'_1+\rho'_2}{2}\frac{\Gamma^2}{4\pi^2(Re_{eff}R)^{3/2}}\frac{\partial p}{\partial x_k}AdV\\ &=\frac{\left(\rho'_2-\rho'_1\right)\left(\rho'_1+\rho'_2\right)}{\rho'_1\rho'_2}\frac{A}{4\pi^2\bar{\delta}^{2}\sqrt{RRe_{eff}}}\iint_{\tilde{\omega}>0}\epsilon_{ijk}\frac{\partial Y}{\partial x_j}\frac{\partial p}{\partial x_k}dV\\ &=\frac{(\sigma-1)(\sigma+1)}{2\sigma}\frac{A}{4\pi^2\bar{\delta}^{2}\sqrt{RRe_{eff}}}\left\langle\frac{\partial Y}{\partial x_j}\frac{\partial p}{\partial x_k}\right\rangle_{\tilde{\omega}>0}. \tag*{(38)} \end{aligned}

In this equation, σ=ρ2′/ρ1′≈0.083\sigma=\rho'_2/\rho'_1\approx0.083 is the density ratio to the post-shock helium and ambient air.

Substituting the dimensionless baroclinic torque in Eq. (38) and viscous term in Eq. (29), Eq. (28) transforms to:

D(Γ+/Γ)Dt=(σ−1)(σ+1)2σA4π2δˉ2RReeff⟨∂Y∂xj∂p∂xk⟩ω~>0+⟨1Re∂2ω∂xk2⟩ω~>0.(39)\frac{D\left(\Gamma^+/\Gamma\right)}{Dt}=\frac{(\sigma-1)(\sigma+1)}{2\sigma}\frac{A}{4\pi^2\bar{\delta}^{2}\sqrt{RRe_{eff}}}\left\langle\frac{\partial Y}{\partial x_j}\frac{\partial p}{\partial x_k}\right\rangle_{\tilde{\omega}>0}+\left\langle\frac{1}{Re}\frac{\partial^2\omega}{\partial x_k^2}\right\rangle_{\tilde{\omega}>0}. \tag*{(39)}

Referring to the budget analysis of the growth of positive circulation in Figure 7, Γsbv+\Gamma^+_{sbv} is primarily influenced by the baroclinic torque ⟨B+⟩\langle B^+\rangle. Furthermore, as the Reynolds number ReRe remains constant across the cases, the viscous effect on the variance of Γsbv+\Gamma^+_{sbv} is neglected. Hence, Γsbv+Γ\frac{\Gamma^+_{sbv}}{\Gamma} is estimated as:

Γsbv+Γ∼∣∫02(σ−1)(σ+1)2σA4π2δˉRReeff dt∣(40)\frac{\Gamma^+_{sbv}}{\Gamma} \sim\left|\int_0^2 \frac{(\sigma-1)(\sigma+1)}{2\sigma}\frac{A}{4\pi^2\bar{\delta}\sqrt{RRe_{eff}}}\,dt\right| \tag*{(40)}
=(1−σ)(1+σ)2σ14π2RReeff∫02Aδˉ dt.(41)= \frac{(1-\sigma)(1+\sigma)}{2\sigma}\frac{1}{4\pi^2\sqrt{RRe_{eff}}}\int_0^2\frac{A}{\bar{\delta}}\,dt. \tag*{(41)}

A new dimensionless SBV number Ωsbv\Omega_{sbv} is put forward:

Ωsbv=(1−σ)(1+σ)2σ14π2RReeff∫02Aδˉ dt,(42)\Omega_{sbv}=\frac{(1-\sigma)(1+\sigma)}{2\sigma}\frac{1}{4\pi^2\sqrt{RRe_{eff}}}\int_0^2\frac{A}{\bar{\delta}}\,dt, \tag*{(42)}

where the area AA and integration ∫02Aδˉdt\int_0^2 \frac{A}{\bar{\delta}}dt are determined based on their evolution curve plotted in Figure 16. This definition of Ωsbv\Omega_{sbv} has similarity to the SBV model proposed in our previous research [17]:

Comparison of the time evolution of the area and the time integrated area over the mean width

Figure 16. Comparison of the time evolution of (a) the area AA over which the positive vorticity ω+\omega^+ is distributed, and (b) the time integrated result of area over the mean width of the bubble’s mass fraction ∫(A/δˉ)dt\int(A/\bar{\delta})dt.

Ωsbv=∣At+∣Dπδ2(11−ΔUtbD−1).\Omega_{sbv}=\frac{\lvert A_t^+\rvert D}{\pi\delta^2}\left(\frac{1}{1-\Delta U_t b_D}-1\right).

However, the previous model lacks the capacity to measure Γsbv+Γ\frac{\Gamma^+_{sbv}}{\Gamma} since it only models the baroclinic torque distributed in the Br structure shown in Figure 4.

The validation is required to assess the reasonability of the definition of Ωsbv\Omega_{sbv} in Eq. (42). The validation process involved the use of Table 2 and Figure 17 (a), which present the values of Ωsbv\Omega_{sbv} and the corresponding Γsbv+/Γ\Gamma^+_{sbv}/\Gamma in different cases. It’s observed that the ratio between Γsbv+/Γ\Gamma^+_{sbv}/\Gamma and Ωsbv\Omega_{sbv} remains almost constant, indicating a linear correlation between these two variables. Consequently, the SBV number Ωsbv\Omega_{sbv} demonstrates effectiveness in quantifying Γsbv+/Γ\Gamma^{+}_{sbv}/\Gamma, and therefore has the capacity to measure the strength of SBHI.

Dimensionless SBV number versus circulation ratio and scaling of the time averaged SDR term

Figure 17. (a) The dimensionless number Ωsbv\Omega_{sbv} versus the ratio of the positive circulation Γsbv+\Gamma^+_{sbv} and the total circulation Γ\Gamma. (b) The scaling of the time averaged SDR term ⟨TSDR⟩\langle T_{SDR}\rangle on the dimensionless number Ωsbv\Omega_{sbv}.

ξ\xiΩsbv\Omega_{sbv}Γsbv+/Γ\Gamma_{sbv}^{+}/\Gammaratio1_1⟨TSDR⟩‾\overline{\langle T_{SDR}\rangle}ratio2_2
0.10.7581.321.7410.06520.111
1.00.5070.921.8150.03080.113
5.00.3700.621.6760.01610.109
10.00.3280.561.7070.01290.109

Table 2. The SBV number Ωsbv\Omega_{sbv}, dimensionless positive circulation at t=2t=2 Γsbv+/Γ\Gamma_{sbv}^{+}/\Gamma, ratio1_1 is defined as (Γsbv+/Γ)/Ωsbv(\Gamma_{sbv}^{+}/\Gamma)/\Omega_{sbv}, the time-averaged mixing rate ⟨TSDR⟩‾\overline{\langle T_{SDR}\rangle}, and ratio2_2 is defined as ⟨TSDR⟩‾/Ωsbv1.918\overline{\langle T_{SDR}\rangle}/\Omega_{sbv}^{1.918}.

Once the SBV number Ωsbv\Omega_{sbv} is proposed, a scaling analysis of the mixing rate’s dependency on Ωsbv\Omega_{sbv} is presented. As shown in Eq. (27), the global mixing rate is represented by the SDR term ⟨TSDR⟩\langle T_{SDR}\rangle. The time-averaged mixing rate is expressed as:

⟨TSDR⟩‾=1T∫0T⟨TSDR⟩ dt,(43)\overline{\langle T_{SDR}\rangle}=\frac{1}{T}\int_{0}^{T}\langle T_{SDR}\rangle\,dt, \tag*{(43)}

where T=5.6T=5.6 is the total SBI evolution time. The scaling behavior of ⟨TSDR⟩‾\overline{\langle T_{SDR}\rangle} with respect to the SBV number Ωsbv\Omega_{sbv} is illustrated in Table 2 and Figure 17 (b). It is observed that the time-averaged mixing rate ⟨TSDR⟩‾\overline{\langle T_{SDR}\rangle} is positively correlated with Ωsbv\Omega_{sbv}. This positive correlation demonstrates the enhancement effect of SBHI on the mixing rate, which can be summarized as:

{Γsbv+Γ∼Ωsbv,⟨TSDR⟩‾∼Ωsbv2.(44)\begin{cases} \frac{\Gamma_{sbv}^{+}}{\Gamma}\sim\Omega_{sbv},\\ \overline{\langle T_{SDR}\rangle}\sim\Omega_{sbv}^{2}. \tag*{(44)} \end{cases}

The underlying mechanisms of the scaling behavior

The scaling of mixing rate on Reynolds number ReRe

In the previous section, turbulent-like flow resulting from SBHI is observed to have the capacity to enhance mixing. However, previous research [23, 24, 25] have suggested that the unstable flow originating from high Reynolds number ReRe does not significantly contribute to mixing when the Péclet number PePe is held constant. In this study, additional simulations are conducted to investigate the scaling behavior of the mixing rate with respect to the Reynolds number ReRe. The geometric parameters outlined in Table 1 remains unchanged to ensure that the additional cases have the same SBV number Ωsbv\Omega_{sbv} as the original set. However, the dynamic viscosity μ\mu is varied in order to obtain different Reynolds numbers Re=Γ‾μ=ρ∗Γ‾ΓRe=\frac{\overline{\Gamma}}{\mu}=\frac{\rho^{*}\overline{\Gamma}}{\Gamma}. The parameters for the original and additional cases are set as in Table 3.

ξ\xiRcoreR_{core} (mm)μ\mu (Pa · s−1^{-1})Ωsbv\Omega_{sbv}ReRe
0.12.0705.0×10−55.0\times10^{-5}0.75871060
0.12.0707.5×10−57.5\times10^{-5}0.75847373
1.01.3905.0×10−55.0\times10^{-5}0.50771060
1.01.3906.4×10−56.4\times10^{-5}0.50755516
5.00.5905.0×10−55.0\times10^{-5}0.37071060
5.00.5907.4×10−57.4\times10^{-5}0.37048274
10.00.3405.0×10−55.0\times10^{-5}0.32871060
10.00.3401.0×10−51.0\times10^{-5}0.328355300

Table 3. The parameters for the original and additional cases

The increment of modified mixedness ⟨f∗⟩Δ\langle f^{*}\rangle_{\Delta} and SDR term ⟨TSDR⟩\langle T_{SDR}\rangle are presented in Figure 18. Based on the results depicted in the figure, it is evident that there is no significant alteration in ⟨f∗⟩Δ\langle f^{*}\rangle_{\Delta} and ⟨TSDR⟩\langle T_{SDR}\rangle as a result of variations in Reynolds number ReRe. The time-averaged mixing rate ⟨TSDR⟩\langle T_{SDR}\rangle is calculated according to the SDR term evolution in Figure 18(b), and the scaling behavior of ⟨TSDR⟩\langle T_{SDR}\rangle on Reynolds number ReRe is summarized in Figure 19.

Temporal evolution of modified mixedness and SDR term

Figure 18. Comparison of the temporal evolution of (a) the increment of modified mixedness ⟨f∗⟩Δ\langle f^{*}\rangle_{\Delta} and (b) SDR term ⟨TSDR⟩\langle T_{SDR}\rangle of SBI cases with different Ωsbv\Omega_{sbv} and Reynolds number ReRe.

Scaling of the time averaged SDR term and flow fields

Figure 19. The scaling of the time averaged SDR term in the form of ⟨TSDR⟩‾\overline{\langle T_{\mathrm{SDR}}\rangle} on ReRe numbers and Ωsbv\Omega_{\mathrm{sbv}} numbers.

The left-hand side of Figure 19 demonstrates the distribution of ⟨TSDR⟩\langle T_{SDR}\rangle with respect to the Reynolds number ReRe and SBV number Ωsbv\Omega_{sbv}. Solid symbols correspond to the original cases, while empty symbols correspond to the additional cases. It is apparent that when the SBV number Ωsbv\Omega_{sbv} is held constant, the values of ⟨TSDR⟩\langle T_{SDR}\rangle remain nearly unchanged among cases with different ReRe.

The right-hand side of the figure presents the vorticity and SDR contour for cases with Ωsbv=0.758\Omega_{sbv}=0.758 and 0.3280.328. As the Reynolds number ReRe increases, the stable vortex filament breaks into small scale vortex in the cases with Ωsbv=0.758\Omega_{sbv}=0.758, and the stable slipstream (SS) transform into unstable SS, and the vorticity is observed to be enhanced in the cases with Ωsbv=0.328\Omega_{sbv}=0.328. These observations demonstrate that an increase in ReRe can lead to an unstable flow field. However, as shown in the left-hand side of the figure, this instability has minimal impact on enhancing mixing, which differs from the effects of SBHI as represented by Ωsbv\Omega_{sbv}.

The distinct mechanisms of two scaling behavior

As observed in the preceding subsection 3.3.1, the scaling behavior of the mixing rate with respect to Ωsbv\Omega_{sbv} and ReRe exhibits distinct characteristics. In the current subsection, we revisit these scaling behaviors to investigate their underlying mechanisms. The mixing rate is represented as ⟨TSDR⟩=⟨8Peχ⟩\langle T_{SDR}\rangle=\langle8Pe\chi\rangle, and therefore, the evolution of SDR χ\chi is examined here through a budget analysis of the SDR transport equation, and the normalized form is:

DχDt=−2∂Y∂xiSij∂Y∂xj+2Pe∂Y∂xi∂∂xi(1ρ∂ρ∂xj∂Y∂xj)+2Pe∂Y∂xi∂2∂xj2(∂Y∂xi)=Tstretch+Tden+Tdiff.(45)\begin{aligned} \frac{D\chi}{Dt}={}&-2\frac{\partial Y}{\partial x_i}S_{ij}\frac{\partial Y}{\partial x_j}+2Pe\frac{\partial Y}{\partial x_i}\frac{\partial}{\partial x_i}\left(\frac{1}{\rho}\frac{\partial\rho}{\partial x_j}\frac{\partial Y}{\partial x_j}\right)\\ &+2Pe\frac{\partial Y}{\partial x_i}\frac{\partial^2}{\partial x_j^2}\left(\frac{\partial Y}{\partial x_i}\right)=T_{stretch}+T_{den}+T_{diff}. \tag*{(45)} \end{aligned}

In this equation, Tstretch=−2∂Y∂xiSij∂Y∂xjT_{stretch}=-2\frac{\partial Y}{\partial x_i}S_{ij}\frac{\partial Y}{\partial x_j} is the stretch term, referring to the increase/decrease of SDR by stretching the scalar gradient ∂Y∂xi\frac{\partial Y}{\partial x_i}. Tden=2Pe∂Y∂xi∂∂xi(1ρ∂ρ∂xj∂Y∂xj)T_{den}=2Pe\frac{\partial Y}{\partial x_i}\frac{\partial}{\partial x_i}\left(\frac{1}{\rho}\frac{\partial\rho}{\partial x_j}\frac{\partial Y}{\partial x_j}\right) is induced by the density gradient ∂ρ∂xi\frac{\partial\rho}{\partial x_i}, thus is labeled as the density term. Tdiff=2Pe∂Y∂xi∂2∂xj2(∂Y∂xi)T_{diff}=2Pe\frac{\partial Y}{\partial x_i}\frac{\partial^2}{\partial x_j^2}\left(\frac{\partial Y}{\partial x_i}\right) is diffusion term, which changes SDR by the molecular diffusion.

By applying the spatial integration ⟨χ⟩=∬χdV\langle\chi\rangle=\iint\chi dV, Eq. (45) can be reformulated in the integrated form as follows:

D⟨χ⟩Dt=⟨DχDt⟩+⟨χ(∂uk∂xk)⟩=⟨Tstretch⟩+⟨Tden⟩+⟨Tdiff⟩+⟨TdivU⟩.(46)\begin{aligned} \frac{D\langle\chi\rangle}{Dt}&=\left\langle\frac{D\chi}{Dt}\right\rangle+\left\langle\chi\left(\frac{\partial u_k}{\partial x_k}\right)\right\rangle\\ &=\langle T_{stretch}\rangle+\langle T_{den}\rangle+\langle T_{diff}\rangle+\langle T_{divU}\rangle. \tag*{(46)} \end{aligned}

In this equation, ⟨TdivU⟩=⟨χ(∂uk∂xk)⟩\langle T_{divU}\rangle=\left\langle\chi\left(\frac{\partial u_k}{\partial x_k}\right)\right\rangle arises from the velocity divergence and is therefore referred to the velocity divergence term.

Figure 20 displays the time derivative of SDR and the corresponding source terms in Eq. (46). The time derivative of SDR D⟨χ⟩Dt\frac{D\langle\chi\rangle}{Dt}, indicated by the red line, closely corresponds with the green line representing ⟨Tright⟩\langle T_{right}\rangle, which is the sum of the terms on the right-hand side of Eq. (46). This observation validates the reliability of the numerical results concerning the resolution of SDR dynamics within the grid, as discussed in Appendix A1. The primary source of SDR growth is the black line representing the stretch term ⟨Tstretch⟩\langle T_{stretch}\rangle, whereas the diffusion term ⟨Tdiff⟩\langle T_{diff}\rangle depicted by blue line primarily diminishes SDR. The density term ⟨Tden⟩\langle T_{den}\rangle and the velocity divergence term ⟨TdivU⟩\langle T_{divU}\rangle are nearly negligible compared to the other terms and are therefore not presented in this figure.

Time derivative of scalar dissipation rate and corresponding source terms

Figure 20. Time derivative of scalar dissipation rate χ\chi and the corresponding source terms in Eq. (46) for different SBI cases: (a) Ωsbv=0.758,Re=71060\Omega_{\mathrm{sbv}}=0.758, Re=71060, (a′) Ωsbv=0.758,Re=47373\Omega_{\mathrm{sbv}}=0.758, Re=47373, (b) Ωsbv=0.507,Re=71060\Omega_{\mathrm{sbv}}=0.507, Re=71060, (c) Ωsbv=0.370,Re=71060\Omega_{\mathrm{sbv}}=0.370, Re=71060, (d) Ωsbv=0.328,Re=71060\Omega_{\mathrm{sbv}}=0.328, Re=71060, (d′) Ωsbv=0.328,Re=355300\Omega_{\mathrm{sbv}}=0.328, Re=355300.

Based on the budget analysis of SDR depicted in Figure 20, the primary source of SDR growth, TstretchT_{stretch}, further investigated to unravel the underlying mechanisms behind the two scaling behaviors mentioned above. Consistent with previous studies [26, 27, 28, 29, 30], we decompose the stretch term using the eigenvectors of the strain rate tensor as follows:

Sij=ΛivivjT,S_{ij}=\Lambda_{i}v_i v_j^{T},
Λij=diag⁡{si}=siδij.(47)\Lambda_{ij}=\operatorname{diag}\{s_i\}=s_i\delta_{ij}. \tag*{(47)}

Here, sis_i represents the iith eigenvalue of the strain rate tensor SijS_{ij}, labelled as the principal strain rate. The eigenvalues are ordered as s1>s2>s3s_1>s_2>s_3, with viv_i denoting the corresponding eigenvector. Moreover, the scalar gradient ∂Y∂xi\frac{\partial Y}{\partial x_i} is expressed as:

∂Y∂xi=∣∇Y∣ei=χei.(48)\frac{\partial Y}{\partial x_i}=|\nabla Y|e_i=\sqrt{\chi}e_i. \tag*{(48)}

In this equation, eie_i is the unit vector aligned with the scalar gradient ∂Y∂xi\frac{\partial Y}{\partial x_i}. Consequently, the stretch term can further be transformed as follows:

Tstretch=−2∂Y∂xiSij∂Y∂xj=−2(χei)TΛijvivjTχej=−2χΛij(viTei)vjTej=−2χsiδijλiλj=−2χsiλi2.(49)\begin{aligned} T_{stretch}&=-2\frac{\partial Y}{\partial x_i}S_{ij}\frac{\partial Y}{\partial x_j}=-2\left(\sqrt{\chi}e_i\right)^T\Lambda_{ij}v_i v_j^T\sqrt{\chi}e_j\\ &=-2\chi\Lambda_{ij}(v_i^Te_i)v_j^Te_j=-2\chi s_i\delta_{ij}\lambda_i\lambda_j\\ &=-2\chi s_i\lambda_i^2. \tag*{(49)} \end{aligned}

Here, λi=viTei\lambda_i=v_i^Te_i is dot product of the unit vector eie_i and the eigenvector viv_i, reflecting the alignment of the scalar gradient ∇Y\nabla Y with the eigenvector viv_i corresponding to the iith principal strain rate sis_i.

In two-dimensional flow, s3=0s_3=0 and λ12+λ22=1\lambda_1^2+\lambda_2^2=1. Owing to the eigenvalues sis_i are ordered by s1>s2s_1>s_2, s1s_1 is the principal strain in the stretch direction, while s2s_2 is the principal strain in the compress direction. The stretch term can be expressed as:

Tstretch=−2χsiλi2=−2χ(s1λ12+s2λ22)=−2χ[s2+(s1−s2)λ12].(50)T_{\mathrm{stretch}}=-2\chi s_i\lambda_i^2=-2\chi\left(s_1\lambda_1^2+s_2\lambda_2^2\right) =-2\chi\left[s_2+(s_1-s_2)\lambda_1^2\right]. \tag*{(50)}

Additionally, it is observed from the budget analysis in Figure 20 that the velocity divergence term TdivU=χ(∂uk∂xk)T_{\mathrm{div}U}=\chi\left(\frac{\partial u_k}{\partial x_k}\right) is nearly negligible. Thus, the velocity divergence ∂uk∂xk=s1+s2≈0\frac{\partial u_k}{\partial x_k}=s_1+s_2\approx0, yielding the final form of the stretch term TstretchT_{\mathrm{stretch}} is:

Tstretch=−2χ[s2+(s1−s2)λ12]=χ[2s1(1−2λ12)].(51)T_{\mathrm{stretch}}=-2\chi\left[s_2+(s_1-s_2)\lambda_1^2\right] =\chi\left[2s_1\left(1-2\lambda_1^2\right)\right]. \tag*{(51)}

By eliminating SDR χ\chi from Eq. (51), the modified stretch term reflecting the stretch effect of flow on scalar gradient is given by:

Tstretch∗=2s1(1−2λ12),(52)T_{\mathrm{stretch}}^*=2s_1\left(1-2\lambda_1^2\right), \tag*{(52)}

from this expression, it’s evident that the stretch rate Tstretch∗T_{\mathrm{stretch}}^* increases with a large principal strain rate s1s_1 and a small λ1\lambda_1. A small value of λ1\lambda_1 implies that the scalar gradient ∇Y\nabla Y is distant from the eigenvector v1\boldsymbol{v}_1 corresponding to principal strain s1s_1. Additionally, we employ two statistical methods, spatial integration ⟨⋅⟩=∬⋅ dV\langle\cdot\rangle=\iint\cdot\,\mathrm{d}V and the probability density function P(⋅)P(\cdot), to investigate this modified stretch term Tstretch∗T_{\mathrm{stretch}}^*.

Figure 21 (a) illustrates the temporal evolution of the spatial average stretch term ⟨Tstretch∗⟩\langle T_{\mathrm{stretch}}^*\rangle. The solid lines, corresponding to ⟨Tstretch∗⟩\langle T_{\mathrm{stretch}}^*\rangle for cases with the same Reynolds number ReRe, are ordered from the red solid line to the pink solid line, indicating a notable increase in the spatial average stretch term ⟨Tstretch∗⟩\langle T_{\mathrm{stretch}}^*\rangle for cases with larger Ωsbv\Omega_{sbv}. However, there is no significant variation between the dashed lines and the solid lines of the same color. This observation suggests that, for the same Ωsbv\Omega_{sbv} value, an increase in the Reynolds number ReRe does not result in a significant increase in the spatial average stretch term ⟨Tstretch∗⟩\langle T_{\mathrm{stretch}}^*\rangle. The PDF depicted in Figure 21 (b) further supports this observation. An increase in Ωsbv\Omega_{sbv} shifts the PDF curves to the right, indicating a greater distribution of Tstretch∗T_{\mathrm{stretch}}^* in the region of large positive values. However, comparison of the solid and dashed curves shows no significant changes as ReRe increases. These findings regarding the modified stretch term are consistent with the previously mentioned scaling behaviors of the mixing rate ⟨TSDR⟩\langle T_{SDR}\rangle.

Comparison of the temporal evolution and probability density function of the modified stretch term

Figure 21. Comparison of (a) the time evolution of the space averaged modified stretch term in the form of ⟨Tstretch∗⟩\langle T_{\mathrm{stretch}}^*\rangle and (b) the probability density function (PDF) of the modified stretch term P(Tstretch∗)P(T_{\mathrm{stretch}}^*), PDF(n)PDF(n) denotes a neglectable variation of the PDF curve.

Expanding on the observation of the stretch term Tstretch∗T_{\mathrm{stretch}}^* and referring to Eq. (51), we further analyze the principal strain rate s1s_1 and alignment λ1\lambda_1. The temporal evolution of the spatial average principal strain rate ⟨s1⟩\langle s_1\rangle is presented in Figure 22 (a). Here, the solid lines, representing ⟨s1⟩\langle s_1\rangle for cases with the same Reynolds number ReRe, are arranged from the red solid line to the pink solid line, demonstrating that the spatial average strain rate ⟨s1⟩\langle s_1\rangle is greater for cases with larger Ωsbv\Omega_{sbv} values. However, there is only a slight variation between the dashed and solid lines of the same color. This finding suggests that, for the same Ωsbv\Omega_{sbv} value, an increase in the Reynolds number ReRe results in a finite increase in the spatial average principal strain ⟨s1⟩\langle s_1\rangle. This observation is further supported by the PDF curves shown in Figure 22 (b). It is apparent that the solid lines with larger Ωsbv\Omega_{sbv} values shift to the right, indicating a greater distribution of s1s_1 in the region of large positive values. Additionally, the dashed lines closely resemble the solid lines of the same color, suggesting that flows with the same Ωsbv\Omega_{sbv} but different ReRe have similar distributions of s1s_1. In summary, the SBHI, as reflected by Ωsbv\Omega_{sbv}, amplifies the principal strain s1s_1, while the unsteady flow caused by large ReRe sightly alters the principal strain s1s_1.

Time evolution and probability density function of the principal strain

Figure 22. Comparison of (a) the time evolution of the spatial averaged principle strain in the stretch direction in the form of ⟨s1⟩\langle s_1\rangle and (b) the probability density function (PDF) of the principle strain in the stretch direction P(s1)P(s_1). →(s)\to(s) indicates the PDF curves shift to the right slightly.

As for the alignment λ1\lambda_1, Figure 23 (a) displays the temporal evolution of the spatial average alignment ⟨λ1⟩\langle\lambda_1\rangle, differing from the observation in the principal strain ⟨s1⟩\langle s_1\rangle, the solid lines, representing ⟨λ1⟩\langle\lambda_1\rangle for cases with the same Reynolds number ReRe, are ordered from the pink solid line to the red solid line, indicating that the spatial average alignment ⟨λ1⟩\langle\lambda_1\rangle decreases with larger Ωsbv\Omega_{sbv} values. Furthermore, the variations between the dashed lines and solid lines of the same color are nearly neglectable, suggesting that an increase in the Reynolds number ReRe does not alter the spatial average alignment ⟨λ1⟩\langle\lambda_1\rangle. The PDF of λ1\lambda_1 is further plotted in Figure 23 (b). The peak of the solid curves shifts to the left with the increase of Ωsbv\Omega_{sbv}, implying a greater distribution of λ1\lambda_1 in the region with smaller value. The dashed lines are coincide with the solid lines of the same colour, which indicates the Reynolds number ReRe hardly alter the distribution of λ1\lambda_1.

Time evolution and probability density function of the alignment

Figure 23. Comparison of (a) the time evolution of the space averaged alignment of the scalar gradient ∇Y\nabla Y with the principle strain in the stretch direction ⟨λ1⟩\langle\lambda_1\rangle and (b) the probability density function (PDF) of the alignment of the scalar gradient ∇Y\nabla Y with the principle strain in the stretch direction P(λ1)P(\lambda_1), PDF(n)PDF(n) denotes a neglectable variation of the PDF curve.

Based on the research findings outlined above, we can summarize the distinct mechanisms of two scaling behaviors as follows:

  1. The strength of SBHI is measured by the SBV number Ωsbv\Omega_{sbv}. The increase in Ωsbv\Omega_{sbv} results in greater s1s_1 and smaller λ1\lambda_1, leading to the amplification of the stretch term Tstretch∗T^*_{stretch} in accordance with Eq. (52). The amplification of stretch rate leads to a higher growth rate of SDR χ\chi, signifying an increased mixing rate. Consequently, the increment of mixedness ⟨f∗⟩Δ\langle f^* \rangle_{\Delta} can attain a larger value, which means the enhancement in the mixing between the helium in the bubble and the ambient air. These logical relationships can be summarized as:

Ωsbv↑⇒{s1↑λ1↓⇒Eq. 51Tstretch∗↑⇒Eq. 45χ↑⇒Eq. 27⟨f∗⟩Δ↑.(53)\Omega_{sbv} \uparrow\Rightarrow\left\{\begin{aligned} s_1 \uparrow\\ \lambda_1 \downarrow\end{aligned}\right. \xRightarrow{\text{Eq. 51}} T^*_{stretch} \uparrow\xRightarrow{\text{Eq. 45}} \chi\uparrow\xRightarrow{\text{Eq. 27}} \langle f^* \rangle_{\Delta} \uparrow. \tag*{(53)}
  1. In the case of the unstable flow caused by high Reynolds number ReRe, differing from SBHI, a large increase of ReRe results in a finite alteration of principal strain s1s_1 that is consist with previous research [43], while the alignment λ1\lambda_1 undergoes no significant change. As a consequence, there is finite variation in the stretch term Tstretch∗T^*_{stretch}. Consequently, the mixing rate SDR χ\chi and the increment of mixedness ⟨f∗⟩Δ\langle f^* \rangle_{\Delta} are neglectable, which means the mixing between the helium in the bubble and the ambient air can hardly be enhanced effectively by increasing the Reynolds number ReRe. These logical relationships can be summarized as:

Re↑⇒{s1↑(s)λ1(n)⇒Eq. 51Tstretch∗↑(s)⇒Eq. 45χ(n)⇒Eq. 27⟨f∗⟩Δ(n),(54)Re \uparrow\Rightarrow\left\{\begin{aligned} s_1 \uparrow(\mathrm{s}) \\ \lambda_1 (\mathrm{n}) \end{aligned}\right. \xRightarrow{\text{Eq. 51}} T^*_{stretch} \uparrow(\mathrm{s}) \xRightarrow{\text{Eq. 45}} \chi(\mathrm{n}) \xRightarrow{\text{Eq. 27}} \langle f^* \rangle_{\Delta} (\mathrm{n}), \tag*{(54)}

in this equation, ↑\uparrow (s) indicates the increment of the parameter is slight, and (n) denotes a neglectable variation of the parameter.

Conclusions and future work

In this paper, we investigate the significant effect of the hydrodynamic instability resulting from the initial diffusion on variable density mixing in shock cylindrical bubble interaction (SBI). To ditinguish each factor within this instability, we design a series of cases with varying degrees of the hydrodynamic instability, while ensuring that the total circulation Γ\Gamma is controlled to maintain constant values for the Reynolds number Re=Γ/νRe = \Gamma/\nu and the Péclet number Pe=Γ/DPe = \Gamma/\mathcal{D} by the circulation control method.

Firstly, we investigate the hydrodynamic characteristics of the hydrodynamic instability by analysing the temporal evolution of the morphology of the bubble and vorticity dynamics. Through the examination of positive circulation Γ+\Gamma^+ and the associated budget analysis, we ascertain that the hydrodynamic instability induced by initial diffusion arises from the baroclinic torque ⟨B+⟩\langle B^+ \rangle. Therefore, this instability is categorized as SBHI, and the magnitude of which can be measured by the difference of the peak value of positive circulation and the total circulation: Γsbv+=(Γ+)∣peak−Γ\Gamma^+_{sbv} = (\Gamma^+)_{|peak} - \Gamma.

Secondly, we study the effect of SBHI on mixing by checking the mixedness ff and SDR χ\chi. Through the temporal evolution of the increment of global modified mixedness ⟨fΔ∗⟩\langle f^*_{\Delta} \rangle and the corresponding budget analysis, we discern that the mixing rate, as reflected in the SDR term ⟨TSDR⟩\langle T_{SDR} \rangle, is enhanced by SBHI. To further describe this mixing enhancement behavior, a new dimensionless parameter, the SBV number Ωsbv\Omega_{sbv}, is put forward to qualify the strength of SBHI. By examining the temporal-averaged mixing rate TSDRT_{SDR} in cases with different Ωsbv\Omega_{sbv}, we observe a positive correlation between the mixing rate ⟨TSDR⟩\langle T_{SDR} \rangle and Ωsbv\Omega_{sbv}, yielding ⟨TSDR⟩∼Ωsbv2\langle T_{SDR} \rangle\sim\Omega_{sbv}^{2}.

Finally, our findings delineate the mechanisms underlying the enhancement effect of SBHI on mixing. Previous research has suggested that the unstable flow resulting from a high Reynolds number ReRe is less effective in altering mixing. The underlying mechanisms of these two scaling behaviors are elucidated by analyzing the stretch term ⟨Tstretch⟩\langle T_{stretch} \rangle in Eq. (46), which governs the growth of the mixing rate ⟨χ⟩\langle\chi\rangle. This term is further decomposed into the principal strain s1s_1 and the alignment λ1\lambda_1. Through an evaluation of the stretch term Tstretch∗T^*_{stretch} using two statistical methods, spatial integration and probability density function, we unravel the mechanisms underlying these two scaling behaviors: the SBHI, characterized by Ωsbv\Omega_{sbv}, can enhance mixing by increasing the principal strain s1s_1 and decreasing the alignment λ1\lambda_1, while the unstable flow resulting from high ReRe cannot effectively enhance mixing due to the finite alteration of the principal strain s1s_1 and the absence of significant change in alignment λ1\lambda_1. These mechanisms are summarized as Eq. (53) and Eq. (54).

The current study demonstrates the significant enhancement effect of SBHI on mixing. Consequently, adjusting the initial diffusion is an effective method for controlling mixing, making it a potential approach for enhancing mixing in practical engineering flow scenarios, such as oblique shock jet interactions. The reason why SBHI can increase the principal strain s1s_1 and decrease the alignment λ1\lambda_1 deserves further investigation in the future.

This work was supported by the Postdoctoral Fellowship Program of China Postdoctoral Science Foundation (GZC20231566), Sichuan Science and Technology Program (2022NSFSC0834) and Natural Science Foundation of Shanghai (25ZR1402273). We would like to express our gratitude to Prof. Pan Shucheng for raising the question at the Conference on the instability of the fluid interface and multi-species turbulence, which inspired this paper.

Conflict of interest The authors declare that they have no conflict of interest.

Appendix

Grid resolution study

Appendix A1 introduces the grid resolution study of the simulation, and the case with Ωsbv=29.92\Omega_{\mathrm{sbv}} = 29.92 and Re=71060Re = 71060 is listed as an example. Since the second order differential of SDR is sensitive to mesh resolution, it’s crucial to select the grid size to ensure the numerical results are resolved. Consistent with our previous works [23, 24], in order to guarantee the grid size can capture the small-scale structures in the SBI configuration of this paper, the mesh Reynolds number ReΔ=u∗Δ/νRe_{\Delta} = u^{*}\Delta/\nu and the mesh Péclect number PeΔ=u∗Δ/DPe_{\Delta} = u^{*}\Delta/\mathcal{D}, are introduced, where u∗u^{*} is the characteristic length defined in Eq. (5), and Δ\Delta signifies the grid size. In previous research [23, 24], the grid resolution criteria dictate that: ReΔ≤140Re_{\Delta} \le140 and PeΔ≤23Pe_{\Delta} \le23. For this study, the grid size of Δ=3.0 μm\Delta= 3.0\,\mu\mathrm{m}, with the total grid number of [Nx,Ny]=[10000,1832][N_x,N_y] = [10000,1832] is employed, resulting in the corresponding mesh Reynolds number and mesh Péclect number which match this criteria: ReΔ=81Re_{\Delta} = 81 and PeΔ=21Pe_{\Delta} = 21.

In order to verify if this grid size is enough to obtain a resolved numerical result, other two meshes with greater and smaller grid size are designed to compare the mesh resolution, the grid settings are displayed in Table A24:

Δ\Delta (μm\mu\mathrm{m})[Nx,Ny][N_x,N_y]ReΔRe_{\Delta}PeΔPe_{\Delta}
12.5[3200, 520]33888
3.0[10000, 1832]8121
2.5[12000, 2199]6818

Table A4. The mesh settings applied to validate the grid resolution

Qualitative comparisons among the three mesh resolutions are presented in Figure A24. By comparing the vorticity and SDR contour of the three different resolutions, it is evident that the small-scale structures are smeared by numerical viscosity in the coarse mesh with a grid size of Δ=12.5 μm\Delta= 12.5\,\mu\mathrm{m}, whereas these structures are clearly captured in the fine meshes with grid sizes of Δ=3.0 μm\Delta= 3.0\,\mu\mathrm{m} and 2.5 μm2.5\,\mu\mathrm{m}. Additionally, a general agreement is observed between the vorticity and SDR in these two fine meshes.

Vorticity contour and SDR contour for three different mesh resolutions

Figure A24. Vorticity contour (top) and SDR contour (bottom) with black dashed isoline of Y=0.05Y = 0.05 of three different mesh resolutions. (a) Δ=12.5μm\Delta= 12.5\mu\mathrm{m}; (b) Δ=3.0μm\Delta= 3.0\mu\mathrm{m}; (c) Δ=2.5μm\Delta= 2.5\mu\mathrm{m}

We then conducted a quantitative analysis to check the mesh resolutions. The quantities related to the large scale structures are firstly examined. In Figure A25 (a), the temporal evolution of the normalized geometric scales L=L~/x∗L = \tilde{L}/x^{*}, H=H~/x∗H = \tilde{H}/x^{*} is presented, where L~\tilde{L} represents the length of the bubble, and H~\tilde{H} represents the height. Additionally, Figure A25 (b) depicts the circulation Γ\Gamma. These figures illustrate that the geometric scales and circulation are independent of grid size, indicating that the qualities related to large scale structures can be accurately captured using the present fine mesh settings.

Comparison of normalized geometric scales and circulation

Figure A25. Comparison of the (a) normalized geometric scales and (b) circulation Γ\Gamma obtained from three different meshes. The black dashed line in Fig (a) is the isoline of Y=0.05Y = 0.05.

The present research focuses on the investigation of positive circulation Γ+\Gamma^{+}, mixedness ff, and SDR χ\chi. These quantities are influenced by small-scale structures, hence, ensuring mesh independence of large-scale structures alone is insufficient, the grid resolution of small-scale structures also needs to be carefully considered. However, as indicated by the observation of the vorticity contour in Figure 4, the presence of small-scale structures signifies a strong instability within the bubble. Consequently, achieving mesh independence of small-scale structures in this type of simulation is relatively challenging. Thus, budget analysis is applied to evaluate the resolution, which has been previously utilized in research on turbulent channel flow [44].

The transport equations for the positive circulation Γ+\Gamma^{+}, mixedness ff, and SDR χ\chi are given by Eq. (19), Eq. (25), and Eq. (46), respectively. Figure A26 presents the budget analysis of these equations. Taking the positive circulation budget in the first row of Figure A26 as an example, the temporal derivative of positive circulation DΓ+Dt\frac{\mathrm{D}\Gamma^{+}}{\mathrm{D}t} and the corresponding terms in the coarse mesh with a grid size of Δ=12.5 μm\Delta= 12.5\,\mu\mathrm{m} are illustrated in Figure A26 (a1a_{1}). The deviation between the red line representing DΓ+Dt\frac{\mathrm{D}\Gamma^{+}}{\mathrm{D}t} and the green line depicting the sum of the terms on the right-hand side of Eq. (19), indicates that the numerical viscosity introduced by the coarse mesh is relatively large compared to the kinematic viscosity ν\nu. Consequently, the evolution of the positive circulation Γ+\Gamma^{+} is significantly influenced by the numerical viscosity, therefore the coarse mesh lacks the capacity to effectively resolve the vorticity dynamics.

Budget analysis of the transport equations

Figure A26. The budget analysis of the transport equations of (a1, b1, c1) the positive circulation Γ+\Gamma^{+} given by Eq. (19); (a2, b2, c2) the mixedness ff given by Eq. (24); (a3, b3, c3) the SDR χ\chi given by Eq. (46). aia_i is the results obtained form the coarse mesh with grid size of Δ=12.5 μm\Delta= 12.5\,\mu\mathrm{m}; bib_i corresponds to the fine mesh with grid size of Δ=3.0 μm\Delta= 3.0\,\mu\mathrm{m}; and cic_i corresponds to the fine mesh with grid size of Δ=2.5 μm\Delta= 2.5\,\mu\mathrm{m}

Similar analyses for the fine meshes with a grid size of Δ=3.0 μm\Delta= 3.0\,\mu\mathrm{m} and 2.5 μm2.5\,\mu\mathrm{m} are depicted in Figure A26 (b1b_{1}) and (c1c_{1}). In contrast to the results obtained from the coarse mesh, the red line from the fine meshes aligns well with the green line, which means that the numerical viscosity introduced by the fine meshes can be neglected. Consequently, the vorticity dynamics can be considered resolved by these two fine meshes. Additionally, the transport equations of mixedness and SDR, Eq. (25) and Eq. (46), are examined in the medium and bottom rows of Figure A26. With the results being similar to the positive circulation equation, we can make the conclusion that these two fine meshes have the ability to resolve the mixedness and SDR dynamics, whereas the coarse mesh does not.

Considering both the computational cost and simulation accuracy, we choose the mesh with the grid size of Δ=3.0 μm\Delta= 3.0\,\mu\mathrm{m} in the present study. The budget analysis in Figure 7, Figure 11 and Figure 20 confirms the numerical results obtained from this mesh are resolved in the other cases in this paper.

Circulation control method

The details of the selection of geometric parameters are displayed here. The cases are intricately designed to explore the impact of hydrodynamic instability induced by initial diffusion on mixing in two-dimensional SBI. As detailed in Section 2, the dimensionless number Re=Γ/νRe = \Gamma/\nu must remain consistent throughout the cases, therefore the total circulation Γ\Gamma also needs to remain constant. The ratio ξ=W/Rcore\xi= W/R_{\mathrm{core}} has been set at values of 0.1, 1.0, 5.0 and 10.0 to manage the hydrodynamic instability, thus the radius of the core region of the bubble RcoreR_{\mathrm{core}} are designed here to control the total circulation Γ\Gamma.

According to previous research [45, 46, 47, 48] on the circulation model:

ΓPB=2uiR(1−ui′2Wiln⁡(ρ1ρ2)),ΓYKZ=4RWiP1′−P1ρ1′ρ2−ρ1ρ2+ρ1,ΓSZ∝2γ1/21+γ(1−σ−1/2)(1+1Ma+1Ma2)(Ma−1)⋅R,(55)\begin{aligned} \Gamma_{\mathrm{PB}} &= 2u_{i}R\left(1-\frac{u_{i}^{\prime}}{2W_{i}}\ln\left(\frac{\rho_{1}}{\rho_{2}}\right)\right),\\ \Gamma_{\mathrm{YKZ}} &= \frac{4R}{W_{i}}\frac{P_{1}^{\prime}-P_{1}}{\rho_{1}^{\prime}}\frac{\rho_{2}-\rho_{1}}{\rho_{2}+\rho_{1}},\\ \Gamma_{\mathrm{SZ}} &\propto\frac{2\gamma^{1/2}}{1+\gamma}(1-\sigma^{-1/2})\left(1+\frac{1}{Ma}+\frac{1}{Ma^{2}}\right)(Ma-1)\cdot R, \tag*{(55)} \end{aligned}

where WiW_{i} is the velocity of the incident shock, γ\gamma is the specific heat ratio, and MaMa is the Mach number, the total circulation is directly proportional to the radius of the bubble: Γ∝R\Gamma\propto R. However, due to varying mass fraction distributions caused by different ratios of ξ\xi, Γ\Gamma cannot remain constant while maintaining the same radius RR. Therefore, based on the configuration illustrated in Figure 2 and the mass fraction distribution in Eq. (3), the average radius RavgR_{\mathrm{avg}} is defined by the following equations:

YavgπR22=∬Y dV,ρavg=ρ2Yavg+ρ1(1−Yavg),ρavgπRavg22=∬ρ dV.(56)\begin{aligned} Y_{\mathrm{avg}}\frac{\pi R^{2}}{2} &= \iint Y\,\mathrm{d}V,\\ \rho_{\mathrm{avg}} &= \rho_{2}Y_{\mathrm{avg}}+\rho_{1}(1-Y_{\mathrm{avg}}),\\ \rho_{\mathrm{avg}}\frac{\pi R_{\mathrm{avg}}^{2}}{2} &= \iint\rho\,\mathrm{d}V. \tag*{(56)} \end{aligned}

The first round simulation maintains the same RρavgR_{\rho\mathrm{avg}}, and the geometric parameters along with the corresponding total circulation Γ\Gamma (as shown in Figure A27) are compiled in Table A5.

ξ\xiRcoreR_{\mathrm{core}} (mm)YavgY_{\mathrm{avg}}RρavgR_{\rho\mathrm{avg}} (mm)Γ\Gamma (m2^{2}/s)
0.11.7500.9001.8201.430
1.01.2900.4971.8201.560
5.00.5800.2761.8201.620
10.00.2380.3401.8201.660

Table A5. Geometric parameters for the two-dimensional SBI cases in the first round simulation.

Comparison of the time variation of the total circulation $\Gamma$ of two-dimensional SBI cases in the first round simulation

Figure A27. Comparison of the time variation of the total circulation Γ\Gamma of two-dimensional SBI cases in the first round simulation.

In Table A5 and Figure A27, the circulation of SBI with a constant RρavgR_{\rho\mathrm{avg}} is relatively consistent across the cases, but still varies. Utilizing the relationship Γ∝R\Gamma\propto R, RcoreR_{\mathrm{core}} is adjusted based on the circulation in the first round simulation, and the corresponding geometric parameters in the second round simulation are listed in Table A6. As demonstrated in Figure A28, the total circulation Γ\Gamma resulting from the provided parameters in Table A6 approximates 1.7 m2/s1.7\ \mathrm{m}^{2}/\mathrm{s}. Thus, the geometric parameters in the second round simulation are selected in this study to ensure the maintenance of the same Reynolds number ReRe across the cases.

ξ\xiRcoreR_{\mathrm{core}} (mm)RR (mm)Γ\Gamma (m2^{2}/s)
0.12.0702.2771.740
1.01.3902.7801.680
5.00.5903.5401.660
10.00.2383.7401.660

Table A6. Geometric parameters for the two-dimensional SBI cases in the second round simulation.

Comparison of the time variation of the total circulation $\Gamma$ of two-dimensional SBI cases in the second round simulation

Figure A28. Comparison of the time variation of the total circulation Γ\Gamma of two-dimensional SBI cases in the second round simulation.

Characteristic time and length

The details regarding the selection of the characteristic time and length to normalize the flow field is introduced in this subsection.

As discussed in the previous research [24, 35], the characteristic time can be determined by the post-shock air velocity:

t∗=Ru1′,(57)t^{*}=\frac{R}{u_{1}^{\prime}}, \tag*{(57)}

or the total circulation:

t∗=R2Γ.(58)t^{*}=\frac{R^{2}}{\Gamma}. \tag*{(58)}

Defining the initial moment t0t_{0} is the physical time at which the shock wave passing the right edge of the bubble shown in Figure 3, the dimensionless time is

t=(t~−t0)/t∗(59)t=(\tilde{t}-t_{0})/t^{*} \tag*{(59)}

The mass fraction contour of the four cases in this paper at t=3.0t=3.0 scaled by two characteristic time definition in Eq. a(5) is shown in Figure A29.

Mass fraction contours of the four cases

Figure A29. The mass fraction contour of the four cases at dimensionless time t=3.0t = 3.0. (a), (b), (c), (d): time is scaled by the post-shock air velocity: t∗=Ru1′t^{*} = \frac{R}{u'_{1}}. (a′), (b′), (c′), (d′): time is scaled by the total circulation: t∗=R2Γt^{*} = \frac{R^{2}}{\Gamma}.

In Figure A29, the characteristic time defined in Eq. a(3) is suitable for normalizing the flow field, whereas the definition in Eq. a(4) is less appropriate. The invalidation of the second definition can be attributed to the effect of the width of the diffusive layer on the characteristic length scale. By defining the effective radius Reff=Γ/u1′R_{eff} = \Gamma/u'_{1}, the appropriate characteristic time t∗=R/u1′t^{*} = R/u'_{1} can be expressed as t∗=ReffR/Γt^{*} = R_{eff}R/\Gamma, which essentially replaces the length RR in Eq. a(4) with RReff\sqrt{RR_{eff}}. Therefore, the suitable characteristic length and time considering the diffusive layer are defined as in Eq. (5).

References

  1. [1]1 R. D. Richtmyer, “Taylor instability in shock acceleration of compressible fluids,” Communication on Pure and Applied Mathematics, vol. 13, no. 2, pp. 297–319, 1960. [Online]. Available: https://doi.org/10.1002/cpa.3160130207
  2. [2]2 E. Meshkov, “Instability of the interface of two gases accelerated by a shock wave,” Fluid Dynamics, vol. 4, no. 5, pp. 101–104, 1969. [Online]. Available: https://doi.org/10.1007/BF01015969
  3. [3]3 M. Brouillette, “The Richtmyer–Meshkov instability,” Annual Review of Fluid Mechanics, vol. 34, no. 1, pp. 445–468, 2002. [Online]. Available: https://doi.org/10.1146/annurev.fluid.34.090101.162238
  4. [4]4 D. Ranjan, J. Oakley, and R. Bonazza, “Shock-bubble interactions,” Annual Review of Fluid Mechanics, vol. 43, pp. 117–140, 2011. [Online]. Available: https://doi.org/10.1146/annurev-fluid-122109-160744
  5. [5]5 S. Pan, X. Y. Hu, and N. A. Adams, “High-resolution method for evolving complex interface networks,” Computer Physics Communications, vol. 225, pp. 10–27, 2018. [Online]. Available: https://doi.org/10.1016/j.cpc.2018.01.001
  6. [6]6 S. Pan, L. Han, X. Hu, and N. A. Adams, “A conservative interface-interaction method for compressible multi-material flows,” Journal of Computational Physics, vol. 371, pp. 870–895, 2018. [Online]. Available: https://doi.org/10.1016/j.jcp.2018.02.007
  7. [7]7 J. D. Lindl, R. L. McCrory, and E. M. Campbell, “Progress toward ignition and burn propagation in inertial confinement fusion,” Phys. Today, vol. 45, no. 9, pp. 32–40, 1992. [Online]. Available: https://doi.org/10.1063/1.881318
  8. [8]8 R. I. Klein, C. F. McKee, and P. Colella, “On the hydrodynamic interaction of shock waves with interstellar clouds. 1: Nondrative shocks in small clouds,” The Astrophysical Journal, vol. 420, pp. 213–236, 1994. [Online]. Available: https://doi.org/10.1086/173554
  9. [9]9 F. E. Marble, E. E. Zukoski, J. W. Jacobs, G. Hendricks, and I. Waitz, “Shock enhancement and control of hypersonic mixing and combustion,” AIAA paper, vol. 1981, p. 1990, 1990. [Online]. Available: https://doi.org/10.2514/6.1990-1981
  10. [10]10 B. Yu, M. He, B. Zhang, and H. Liu, “Two-stage growth mode for lift-off mechanism in oblique shock-wave/jet interaction,” Physics of Fluids, vol. 32, no. 11, p. 116105, 2020. [Online]. Available: https://doi.org/10.1063/5.0022449
  11. [11]11 E. Villermaux, “Mixing versus stirring,” Annual Review of Fluid Mechanics, vol. 51, pp. 245–273, 2019. [Online]. Available: https://doi.org/10.1146/annurev-fluid-010518-040306
  12. [12]12 J. W. Jacobs, “Shock-induced mixing of a light-gas cylinder,” Journal of Fluid Mechanics, vol. 234, pp. 629–649, 1992. [Online]. Available: https://doi.org/10.1017/S0022112092000096
  13. [13]13 J. Yang, T. Kubota, and E. E. Zukoski, “Applications of shock-induced mixing to supersonic combustion,” AIAA journal, vol. 31, no. 5, pp. 854–862, 1993. [Online]. Available: https://doi.org/10.2514/3.11696
  14. [14]14 S. Kumar, G. Orlicz, C. Tomkins, C. Goodenough, K. Prestridge, P. Vorobieff, and R. Benjamin, “Stretching of material lines in shock-accelerated gaseous flows,” Physics of Fluids (1994-present), vol. 17, no. 8, p. 082107, 2005. [Online]. Available: https://doi.org/10.1063/1.2031347
  15. [15]15 J. H. Niederhaus, J. Greenough, J. Oakley, D. Ranjan, M. Anderson, and R. Bonazza, “A computational parameter study for the three-dimensional shock–bubble interaction,” Journal of Fluid Mechanics, vol. 594, pp. 85–124, 2008. [Online]. Available: https://doi.org/10.1017/S0022112007008749
  16. [16]16 J. Ou and Z. Zhai, “Effects of aspect ratio on shock-cylinder interaction,” Acta Mechanica Sinica, vol. 35, no. 1, pp. 61–69, 2019. [Online]. Available: https://doi.org/10.1007/s10409-018-0819-3
  17. [17]17 H. Liu, B. Yu, B. Zhang, and Y. Xiang, “On mixing enhancement by secondary baroclinic vorticity in a shock–bubble interaction,” Journal of Fluid Mechanics, vol. 931, p. A17, 2022. [Online]. Available: https://doi.org/10.1017/jfm.2021.923
  18. [18]18 B. M. C. Dexter and N. Mohamad, “Experiments on liquid mixing and reaction in a vortex,” Journal of Fluid Mechanics, vol.
  19. [19]19 K. A. Buch and W. J. Dahm, “Experimental study of the fine-scale structure of conserved scalar mixing in turbulent shear flows. part 1. sc [dbl greater-than sign] 1,” Journal of Fluid Mechanics, vol. 317, pp. 21–71, 1996. [Online]. Available: https://doi.org/10.1017/S0022112096000651
  20. [20]20 C. Tomkins, S. Kumar, G. Orlicz, and K. Prestridge, “An experimental investigation of mixing mechanisms in shock-accelerated flow,” Journal of Fluid Mechanics, vol. 611, pp. 131–150, 2008. [Online]. Available: https://doi.org/10.1017/S0022112008002723
  21. [21]21 S. K. Shankar, S. Kawai, and S. K. Lele, “Two-dimensional viscous flow simulation of a shock accelerated heavy gas cylinder,” Physics of Fluids (1994-present), vol. 23, no. 2, p. 024102, 2011. [Online]. Available: https://doi.org/10.1063/1.3553282
  22. [22]22 Y. Li, Z. Wang, B. Yu, B. Zhang, and H. Liu, “Gaussian models for late-time evolution of two-dimensional shock–light cylindrical bubble interaction,” Shock Waves, pp. 1–16, 2019. [Online]. Available: https://doi.org/10.1007/s00193-019-00928-w
  23. [23]23 B. Yu, H. Liu, and H. Liu, “Scaling behavior of density gradient accelerated mixing rate in shock bubble interaction,” Physical Review Fluids, vol. 6, no. 6, p. 064502, 2021. [Online]. Available: https://doi.org/10.1103/PhysRevFluids.6.064502
  24. [24]24 B. Yu, L. Li, H. Xu, B. Zhang, and H. Liu, “Effects of reynolds number and schmidt number on variable density mixing in shock bubble interaction,” Acta Mechanica Sinica, vol. 38, no. 6, p. 121256, 2022. [Online]. Available: https://doi.org/10.1007/s10409-022-02011-9
  25. [25]25 D. Buaria, M. P. Clay, K. R. Sreenivasan, and P. Yeung, “Turbulence is an ineffective mixer when schmidt numbers are large,” Physical Review Letters, vol. 126, no. 7, p. 074501, 2021. [Online]. Available: https://doi.org/10.1103/PhysRevLett.126.074501
  26. [26]26 Y. Tian, F. A. Jaberi, Z. Li, and D. Livescu, “Numerical study of variable density turbulence interaction with a normal shock wave,” Journal of Fluid Mechanics, vol. 829, pp. 551–588, 2017. [Online]. Available: https://doi.org/10.1017/jfm.2017.542
  27. [27]27 Y. Tian, F. A. Jaberi, and D. Livescu, “Density effects on post-shock turbulence structure and dynamics,” Journal of Fluid Mechanics, vol. 880, pp. 935–968, 2019. [Online]. Available: https://doi.org/10.1017/jfm.2019.707
  28. [28]28 X. Gao, I. Bermejo-Moreno, and J. Larsson, “Parametric numerical study of passive scalar mixing in shock turbulence interaction,” Journal of Fluid Mechanics, vol. 895, p. A21, 2020. [Online]. Available: https://doi.org/10.1017/jfm.2020.292
  29. [29]29 M. Danish, S. Suman, and S. S. Girimaji, “Influence of flow topology and dilatation on scalar mixing in compressible turbulence,” Journal of Fluid Mechanics, vol. 793, pp. 633–655, 2016. [Online]. Available: https://doi.org/10.1017/jfm.2016.145
  30. [30]30 R. Buttay, G. Lehnasch, and A. Mura, “Analysis of small-scale scalar mixing processes in highly under-expanded jets,” Shock Waves, vol. 26, pp. 193–212, 2016. [Online]. Available: https://doi.org/10.1007/s00193-015-0599-7
  31. [31]31 X. Han, B. Yu, and H. Liu, “Mixing mechanisms in the view of mixing indicators: from passive-scalar mixing to variable-density mixing,” Metascience in Aerospace, vol. 1, no. 1, pp. 1–37, 2024. [Online]. Available: 10.3934/mina.2024001DOI
  32. [32]32 J. Ding, T. Si, M. Chen, Z. Zhai, X. Lu, and X. Luo, “On the interaction of a planar shock with a three-dimensional light gas cylinder,” Journal of Fluid Mechanics, vol. 828, pp. 289–317, 2017. [Online]. Available: https://doi.org/10.1017/jfm.2017.528
  33. [33]33 D. Li, B. Guan, and G. Wang, “Numerical study on shock-accelerated heavy gas cylinders with diffusive interfaces,” Acta Mechanica Sinica, vol. 35, no. 4, pp. 750–762, 2019. [Online]. Available: https://doi.org/10.1007/s10409-019-00867-w
  34. [34]34 D. Li, B. Guan, and G. Wang, “Effects of interface diffusion and shock strength on shock-accelerated sf6 cylinder,” Physics of Flu- ids, vol. 34, no. 7, p. 076109, 2022. [Online]. Available: https://doi.org/10.1063/5.0099353
  35. [35]Z. Wang, B. Yu, H. Chen, B. Zhang, and H. Liu, “Scaling vortex breakdown mechanism based on viscous effect in shock cylindrical bubble interaction,” Physics of Fluids, vol. 30, no. 12, p. 126103, 2018. [Online]. Available: https://doi.org/10.1063/5.0150463
  36. [36]G. Liang, B. Yu, B. Zhang, H. Xu, and H. Liu, “Hidden flow structures in compressible mixing layer and a quantitative analysis of entrainment based on lagrangian method,” Journal of Hydrodynamics, vol. 31, no. 2, pp. 256–265, 2019. [Online]. Available: https://doi.org/10.1017/jfm.2020.295
  37. [37]C. Liu, Z. Wang, B. Yu, B. Zhang, and H. Liu, “Optimal excitation mechanism for combustion enhancement of supersonic shear layers with pulsed jets,” International Journal of Hydrogen Energy, vol. 45, no. 43, pp. 23674–23691, 2020. [Online]. Available: https://doi.org/10.1016/j.ijhydene.2020.06.217
  38. [38]A. Glezer, “The formation of vortex rings,” Physics of Fluids (1958-1988), vol. 31, no. 12, pp. 3532–3542, 1988. [Online]. Available: https://doi.org/10.1063/1.866920
  39. [39]P. Meunier and E. Villermaux, “How vortices mix,” Journal of Fluid Mechanics, vol. 476, pp. 213–222, 2003. [Online]. Available: https://doi.org/10.1017/S0022112002003166
  40. [40]C. Tomkins, B. Balakumar, G. Orlicz, K. Prestridge, and J. Ristorcelli, “Evolution of the density self-correlation in developing richtmyer–meshkov turbulence,” J. Fluid Mech, vol. 735, pp. 288–306, 2013. [Online]. Available: https://doi.org/10.1017/jfm.2013.430
  41. [41]P. E. Hamlington, A. Y. Poludnenko, and E. S. Oran, “Interactions between turbulence and flames in premixed reacting flows,” Physics of Fluids, vol. 23, no. 12, 2011. [Online]. Available: https://doi.org/10.1063/1.3671736
  42. [42]N. Peng, Y. Yang, J. Wu, and Z. Xiao, “Mechanism and modelling of the secondary baroclinic vorticity in the richtmyer–meshkov instability,” Journal of Fluid Mechanics, vol. 911, 2021. [Online]. Available: https://doi.org/10.1017/jfm.2020.1080
  43. [43]S. B. Pope and S. B. Pope, Turbulent flows. Cambridge university press, 2000.DOI
  44. [44]J.-P. Tang, Z.-H. Wan, N.-S. Liu, and X.-Y. Lu, “Statistical properties of pressure-hessian tensor in a turbulent channel flow,” Journal of Fluid Mechanics, vol. 934, p. A22, 2022. [Online]. Available: https://doi.org/10.1017/jfm.2021.1038
  45. [45]J. Picone and J. Boris, “Vorticity generation by shock propagation through bubbles in a gas,” Journal of Fluid Mechanics, vol. 189, pp. 23–51, 1988. [Online]. Available: https://doi.org/10.1017/S0022112088000904
  46. [46]J. Yang, T. Kubota, and E. E. Zukoski, “A model for characterization of a vortex pair formed by shock passage over a light-gas inhomogeneity,” Journal of Fluid Mechanics, vol. 258, pp. 217–244, 1994. [Online]. Available: https://doi.org/10.1017/S0022112094003307
  47. [47]R. Samtaney and N. J. Zabusky, “On shock polar analysis and analytical expressions for vorticity deposition in shock-accelerated density-stratified interfaces,” Physics of Fluids A: Fluid Dynamics, vol. 5, no. 6, pp. 1285–1287, 1993. [Online]. Available: https://doi.org/10.1063/1.858618
  48. [48]H. Liu, B. Yu, H. Chen, B. Zhang, H. Xu, and H. Liu, “Contribution of viscosity to the circulation deposition in the richtmyer–meshkov instability,” Journal of Fluid Mechanics, vol. 895, 2020. [Online]. Available: https://doi.org/10.1017/jfm.2020.295

Paper details

vlnc:W_yk38dsy5
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents