betaFeedback

WorksPhysics and Astronomy

Interfaces as transport barriers in two-dimensional Cahn-Hilliard-Navier-Stokes turbulence

Abstract

We investigate the role of interfaces as transport barriers in binary-fluid turbulence by employing Lagrangian tracer particles. The Cahn-Hilliard-Navier-Stokes (CHNS) system of partial differential equations provides a natural theoretical framework for our investigations. For specificity, we utilize the two-dimensional (2D) CHNS system. We capture efficiently interfaces and their fluctuations in 2D binary-fluid turbulence by using extensive pseudospectral direct numerical simulations (DNSs) of the 2D CHNS equations. We begin with nn tracers within a droplet of one phase and examine their dispersal into the second phase. The tracers remain within the droplet for a long time before emerging from it, so interfaces act as transport barriers in binary-fluid turbulence. We show that the fraction of the number of particles inside the droplet decays exponentially and is characterized by a decay time τξ∼R03/2τ_ξ\sim R_0^{3/2} that increases with R0R_0, the radius of the initially circular droplet. Furthermore, we demonstrate that the average first-passage time ⟨τ⟩\langle τ\rangle for tracers inside a droplet is orders of magnitude larger than it is for transport out of a hypothetical circle with the same radius as the initially circular droplet. We examine the roles of the Okubo-Weiss parameter ΛΛ, the fluctuations of the droplet perimeter, and the probability distribution function of cos⁡(θ)\cos(θ), with θθ the angle between the fluid velocity and the normal to a droplet interface, in trapping tracers inside droplets. We mention possible generalisations of our study.

Introduction

Droplet dynamics in multi-phase flows, both laminar and turbulent, poses considerable challenges for experimental, numerical, and theoretical studies: Not only must we resolve the dynamics of the immiscible fluids in a multi-phase flow, but we must also track the spatiotemporal evolution of the fluctuating interfaces that separate the coexisting fluids [1, 2, 3]. In particular, we must monitor mass- and momentum-transfer across these interfaces whose deformations can affect the flow topology, which has been studied for droplet-laden turbulent flows in Refs. 4,5 and for droplets in turbulent shear flows in Ref. 6. The interactions between particles and droplet interfaces in channel-flow turbulence has been reported in Ref. 7.

The Cahn-Hilliard-Navier-Stokes (CHNS) system of partial differential equations (PDEs) combine the Navier-Stokes equations for a Newtonian fluid [8, 9] and the Cahn-Hilliard equations for binary mixtures [10, 11, 12, 13, 14, 15, 16]. The CHNS system provides a natural framework for studying multi-phase flows [3, 17, 18, 19, 20, 21, 22, 23, 24], in general, and droplet dynamics, in particular. For example, the CHNS system has been used to uncover the multifractality of droplet-perimeter fluctuations in both conventional [25, 26] and active-fluid flows [27]; the spatiotemporal evolution of liquid-lens mergers has been studied by using a three-phase version of the CHNS PDEs in Ref. 28. We show that the role of interfaces as transport barriers in binary-fluid turbulence can be addressed in detail by monitoring the advection of Lagrangian tracer particles by CHNS flows. Such tracers have been employed widely in various forms of turbulence, including fluid and MHD turbulence, bacterial turbulence, and quantum turbulence [see, e.g., Refs. 29–37].

To illustrate how interfaces act as transport barriers, we carry out extensive direct numerical simulations (DNSs) of the two-dimensional (2D) CHNS equations with such Lagrangian tracers. Our DNSs lead to important insights that we summarise qualitatively below before we present the details of our work. If we initiate our DNS with tracers inside an initially circular droplet, which is advected by a turbulent flow, the tracers remain entrapped within the droplet for a long time before emerging from it. This demonstrates clearly that interfaces act as transport barriers for such tracers in binary-fluid turbulence. We show that the fraction of the number of particles inside the droplet decays exponentially and is characterized by a decay time τξ∼R03/2\tau_{\xi} \sim R_0^{3/2} that increases with R0R_0, the radius of the initially circular droplet. We calculate the average first-passage time ⟨τ⟩\langle\tau\rangle for tracers in a droplet and show that it is at least two orders of magnitude larger than it is for transport out of a hypothetical circle with the same radius as the initially circular droplet. We also elucidate the roles of the Okubo-Weiss parameter Λ\Lambda, the fluctuations of the droplet perimeter, and the probability distribution function of cos⁡(θ)\cos(\theta), with θ\theta the angle between the fluid velocity and the normal to a droplet interface, in trapping tracers inside droplets. The remainder of this paper is organised as follows. We introduce the CHNS model and the numerical methods that we use in Section II. We give a detailed presentation of our results in Section III. We end with a discussion of the significance of our results in Section IV.

Cahn-Hilliard-Navier-Stokes Equations

We consider the incompressible CHNS system of equations that couple the fluid velocity u\mathbf{u} with a scalar order parameter cc that changes continuously between the two fluid phases A and B. The fluid dynamics of u\mathbf{u} and cc is governed by the following CHNS PDEs[17, 18, 19, 25, 38, 39, 40] [we restrict our discussion to two spatial dimensions (2D)]:

∂tc+(u⋅∇)c=M∇2μ;(1)\partial_t c + (\mathbf{u} \cdot\nabla)c = M\nabla^2\mu; \tag*{(1)}
μ=(δFδc)=−32σϵ∇2c+24σϵc(1−c)(1−2c)(2)\mu= \left(\frac{\delta F}{\delta c}\right) = -\frac{3}{2}\frac{\sigma}{\epsilon}\nabla^2c + 24\frac{\sigma}{\epsilon}c(1-c)(1-2c) \tag*{(2)}
∂tω+(u⋅∇)ω=ν∇2ω−αω(3)\partial_t\omega+ (\mathbf{u} \cdot\nabla)\omega= \nu\nabla^2\omega- \alpha\omega \tag*{(3)}
+[∇×(μ∇c)]⋅e^z+fω;\qquad+ [\nabla\times(\mu\nabla c)] \cdot\hat{\mathbf{e}}_z + f^\omega;
∇⋅u=0;ω=(∇×u)⋅e^z;(4)\nabla\cdot\mathbf{u} = 0; \qquad\omega= (\nabla\times\mathbf{u}) \cdot\hat{\mathbf{e}}_z; \tag*{(4)}

the parameters ν\nu, α\alpha, and MM are the kinematic viscosity, bottom friction, and mobility, respectively; the fluid density ρ\rho is a constant that we set to 1. In Eq. (4) we employ the vorticity-streamfunction (ω−ψ\omega-\psi) formulation with

u=∇×(ψe^z)andψ=−∇−2ω.(5)\mathbf{u} = \nabla\times(\psi\hat{\mathbf{e}}_z) \quad\text{and} \quad\psi= -\nabla^{-2}\omega. \tag*{(5)}

The chemical potential μ\mu is derived from the variational derivative of the Landau-Ginzburg free-energy functional

F[c,∇c]=∫Ω[12σϵc2(1−c)2+34σϵ∣∇c∣2] dΩ;(6)F[c,\nabla c] = \int_{\Omega}\left[12\frac{\sigma}{\epsilon}c^2(1-c)^2 + \frac{3}{4}\sigma\epsilon\lvert\nabla c\rvert^2\right]\,\mathrm{d}\Omega; \tag*{(6)}

σ\sigma and ϵ\epsilon, which can be varied independently, are the surface tension and interface width, respectively. The first term is a double-well potential with minima at c=0c=0 and 11, which correspond to A-rich and B-rich bulk phases in equilibrium; and the second term gives the free-energy penalty for interfaces between these phases. The order parameter cc varies smoothly across an interface; c≃0c \simeq0 in phase A and c≃1c \simeq1 in phase B. Note that the scalar field cc is not a passive scalar (unlike the scalar in Ref. 41); cc affects the flow velocity u\mathbf{u} even as it is advected by u\mathbf{u}. If a strictly positive cc is required, then the term c2(1−c)2c^2(1-c)^2 in Eq. (6) should be replaced by [cln⁡(c)+(1−c)ln⁡(1−c)][c\ln(c)+(1-c)\ln(1-c)].

We generate turbulence by using a Kolmogorov-type forcing[42] of the form

fω=f0kfcos⁡(kfx);(7)f^\omega= f_0 k_f \cos(k_f x); \tag*{(7)}

f0f_0 and kfk_f are the forcing amplitude and forcing wavenumber, respectively.

Numerical methods

We carry out extensive direct numerical simulations (DNSs) of the coupled equations (1)-(4) by using pseudospectral methods[43], with periodic boundary conditions in a square box of size (2π×2π)(2\pi\times2\pi) with 1024×10241024 \times1024 collocation points. Our DNSs employ the semi-implicit exponential time difference Runge-Kutta-2 (ETDRK-2) method[44] for time integration and the 1/21/2-dealiasing scheme to remove aliasing errors. To resolve the interface well, we take three computational grid points at interfaces.

The spatiotemporal evolution of the fields in Eqs. (1)-(4) depend on the initial conditions (see below) and the non-dimensional control parameters given hereunder:

Reynolds number Re=L0U0ν;(8)\text{Reynolds number } Re = \frac{L_0U_0}{\nu}; \tag*{(8)}
Cahn number Cn=ϵL0;(9)\text{Cahn number } Cn = \frac{\epsilon}{L_0}; \tag*{(9)}
Weber number We=L0U02σ;(10)\text{Weber number } We = \frac{L_0U_0^2}{\sigma}; \tag*{(10)}
Peclet number Pe=L0U0ϵMσ;(11)\text{Peclet number } Pe = \frac{L_0U_0\epsilon}{M\sigma}; \tag*{(11)}
non-dimensional friction α′=αL0U0.(12)\text{non-dimensional friction } \alpha' = \frac{\alpha L_0}{U_0}. \tag*{(12)}

We use the following characteristic velocity, length, and time scales:

U=f0νkf2;L0=2πkf;T0=νkff0.(13)U = \frac{f_0}{\nu k_f^2}; \quad L_0 = \frac{2\pi}{k_f}; \quad T_0 = \frac{\nu k_f}{f_0}. \tag*{(13)}

For all our DNSs, we keep the following parameters fixed: Cn=0.01Cn=0.01, Re=1178Re=1178, We=30We=30, Pe=1445Pe=1445, and α′=0.004\alpha'=0.004.

Lagrangian tracers

To characterize the transport across a droplet’s interface, we introduce nn Lagrangian tracer particles inside the droplet[45]. The temporal evolution of a tracer, with position r0\mathbf{r}_0 at time t0t_0, is given by

dr(t)dt=v(r,t∣r0,t0)=u(r,t),(14)\frac{\mathrm{d}\mathbf{r}(t)}{\mathrm{d}t} = \mathbf{v}(\mathbf{r},t\mid\mathbf{r}_0,t_0) = \mathbf{u}(\mathbf{r},t), \tag*{(14)}

where v\mathbf{v} is the Lagrangian velocity, which we obtain from the Eulerian velocity field u\mathbf{u}[46, 47] [by using bilinear interpolation at off-grid points]. We use a first-order Euler scheme for the time integration of Eq. (14).

Initial conditions

We begin with an initially circular droplet of radius R0R_0 and with center at (x0,y0)=(π,π)(x_0,y_0)=(\pi,\pi):

c(x,y,t=0)=0.5[1+tanh⁡(R0−(x−x0)2+(y−y0)2ϵ/2)].(15)c(x,y,t=0) = 0.5\left[1+\tanh\left(\frac{R_0-\sqrt{(x-x_0)^2+(y-y_0)^2}}{\epsilon/2}\right)\right]. \tag*{(15)}

This field c(x,0)c(\boldsymbol{x}, 0) is used as an initial condition for Eqs. (1)-(4) in which the turbulent vorticity field ω\omega is generated by the forcing term in Eq. (7). The Reynolds number of the turbulent flow we use in our DNS is Re=1178\mathrm{Re} = 1178.

Results

We depict the spatiotemporal evolution of tracers by using pseudocolor plots of vorticity fields overlaid with the c=0.5c = 0.5 contour and tracer positions in Figure 1, at various representative values of the simulation time tt. [See the Videos V1-V3 in Appendix § 1.1.2 for the full spatiotemporal evolution.] For clarity, we limit the visualization to 6000 particles, although we use n=1024×128=131072n = 1024 \times128 = 131072 tracer particles in our DNSs for statistical analysis. If the tracers are not confined within a droplet, as shown in Figs.1(a)-(d), they disperse rapidly throughout the entire simulation domain. However, when we initiate our DNS with tracers inside a droplet, they remain entrapped within the droplet for a significant duration before emerging from it eventually, as we illustrate in Figs.1(e)-(h) and (i)-(l) for droplets with radii R0/L0=0.25R_0/L_0 = 0.25 and R0/L0=0.5R_0/L_0 = 0.5, respectively. By comparing Figs. 1(b), (f), and (j), captured at the same simulation time t=15t = 15, we see how rapidly the tracers spread out without a droplet [Fig. 1(b)] and how their spreading is constrained by the droplet boundary [Figs. 1(f) and (j)], so they move along with the droplet. Clearly, the droplet interface acts as a transport barrier.

Illustrative pseudocolor plots of $\omega$

Figure 1. Illustrative pseudocolor plots of ω\omega, with the c=0.5c = 0.5 contour shown in magenta, at different representative simulation times (increasing from left to right): t1=0t_1 = 0, t2=15t_2 = 15, t3=150t_3 = 150, and t4=300t_4 = 300. The tracers are shown as black points for (a)-(d) droplet-free flow, (e)-(h) a droplet with initial radius R0/L0=0.25R_0/L_0 = 0.25, and (i)-(l) a droplet with initial radius R0/L0=0.5R_0/L_0 = 0.5; the Weber number We=30\mathrm{We} = 30. We use n=1024×128=131072n = 1024 \times128 = 131072 tracer particles in our DNSs but, for the figures here, we show the positions of only 6000 particles. The vorticity ω\omega is normalized by its maximum absolute value for ease of visualization. [For the full spatiotemporal evolution, see the videos V1-V3 in Appendix § 1.1.2.]

In the limit of an infinitely thin interface, with a high value of σ\sigma, no Lagrangian tracers should come out of the droplet. In the CHNS system we consider, the interface is diffuse; therefore, tracers do escape from droplets over a sufficiently long period of time. The narrower the interface, the longer is this time. We illustrate below how the mean droplet-escape time changes with the width of the CHNS interface; and we discuss the numerical subtleties of taking the limit of vanishing interface thickness.

To quantify the impact of the interfacial barrier on particle transport, we present semi-log plots, in Figure 2(a), of the fraction of particles inside the droplet nin(t)/nn_{in}(t)/n versus time tt. For large tt, these plots are consistent with the exponential decay

Figure 2. (a) Semi-log plots illustrating the exponential decay of the fraction of particles nin(t)/nn_{in}(t)/n inside the droplet with the scaled time t/Tt/T [Eq. (13)] for different values of the scaled initial droplet radii R0/L0R_0/L_0 and at a fixed Weber number We=30We=30. (b) Log-log plot of the scaled decay time τξ\tau_\xi versus R0/L0R_0/L_0; the black line represents the power-law relationship τξ∼[R0/L0]3/2\tau_\xi\sim[R_0/L_0]^{3/2}. (c) Log-log plots of the complementary CPDF of the first-passage time τ\tau (see text). (d) Semi-log plots of the PDF of cos⁡(θ)\cos(\theta), with θ\theta the angle between u\mathbf{u} and the normal to the droplet’s interface (inset: a schematic diagram illustrating the angle θ\theta).

nin(t)n∼exp⁡(−t/τξ),(16)\frac{n_{in}(t)}{n} \sim\exp(-t/\tau_\xi), \tag*{(16)}

where the decay time τξ\tau_\xi depends on the scaled droplet radius R0/L0R_0/L_0; we hold the Weber number at We=30We=30. In Figure 2(b) we give a log-log plot of τξ\tau_\xi versus R0/L0R_0/L_0; this is consistent with [Fig. 2(b)]

τξ∼[R0L0]3/2.(17)\tau_\xi\sim\left[\frac{R_0}{L_0}\right]^{3/2}. \tag*{(17)}

We obtain τξ\tau_\xi from a local-slope analysis of the exponentially decaying tails of the curves in Figure 2(a) [see Fig. 5 in Appendix A] for details.

We have conducted two illustrative simulations with particles distributed uniformly at the centre of the simulation box within a radius r/L0=0.5r/L_0=0.5. (a) In the first simulation, we consider no droplet, effectively employing a hypothetical circular boundary (as depicted in Figure 1(a)). (b) In the second simulation, we introduce a droplet into the system (as shown in Figure 1(i)). At a simulation time of t=15t=15, the majority of the particles successfully cross the hypothetical boundary, with a radius of r/L0=0.5r/L_0=0.5, in the absence of a droplet, as illustrated in Figure 1(b). In stark contrast, the presence of the droplet boundary in Figure 1(j) acts as a transport barrier for the particles, causing them to be carried along with the droplet. To characterise the time particles spend inside a droplet, or a hypothetical circular boundary, we calculate their firstpassage-time τ\tau, which is the time required by a particle to cross the interface of the droplet or the boundary for the first time [46] for conventional fluid turbulence. We determine the complementary cumulative probability distribution function (CPDF) of τ\tau, namely, Q(τ)Q(\tau), by using the rank-order method46,48,49{}^{46,48,49}, for various droplet radii. In Figure 2(c) we give semi-log plots of Q(τ)Q(\tau) whose tails decay exponentially for all the radii we consider. We find that the average first-passage time ⟨τ⟩\langle\tau\rangle for the droplet with a scaled radius R0/L0=0.5R_0/L_0 = 0.5 is approximately 300 times longer than that for the case of a hypothetical circular boundary with a scaled radius of r/L0=0.5r/L_0 = 0.5.

Let us now examine the PDF P(cos⁡θ)P(\cos\theta) of cos⁡θ\cos\theta, where θ\theta is the angle between the velocity vector and the normal to the interface, as illustrated in the schematic diagram of a droplet (see the inset of Figure 2(d)):

cos⁡(θ)≡n^⋅u∣n^∣∣u∣;n^=−∇c∣∇c∣(18)\cos(\theta) \equiv\frac{\hat{n}\cdot\boldsymbol{u}}{|\hat{n}||\boldsymbol{u}|}; \quad\hat{n}=-\frac{\nabla c}{|\nabla c|} \tag*{(18)}

We plot P(cos⁡(θ))P(\cos(\theta)) in Figure 2(d) for various values of the scaled droplet radius R0/L0R_0/L_0 of the droplet. We find that, for small values of R0/L0R_0/L_0, the dominant peaks in P(cos⁡θ)P(\cos\theta) occur at cos⁡(θ)≃±1\cos(\theta)\simeq\pm1; this indicates that there is a tendency of velocity vectors to align or anti-align with n^\hat{n}. By contrast, for large values of R0/L0R_0/L_0, in addition the peaks at cos⁡θ≃±1\cos\theta\simeq\pm1, P(cos⁡θ)P(\cos\theta) exhibits a broad maximum around cos⁡θ≃0\cos\theta\simeq0. Thus, there is an increase in the probability that velocity vectors are aligned tangentially to the interface and not only perpendicular to it, so the particles are less likely to cross the droplet interface if R0/L0R_0/L_0 is large than when it is small. This suggests that, for droplets, it is useful to define a Cahn number Cnd≡ϵ/R0Cn_d\equiv\epsilon/R_0, which is based on the droplet radius and decreases as R0R_0 increases. Thus, as R0R_0 increases, CndCn_d tends 0, so τξ\tau_\xi increases (see the discussion below about the limit of vanishing interface thickness).

Next we characterise the interface, which acts as a transport barrier, by determining its scaled local curvature κR0\kappa R_0. For this we use the local-curvature algorithm in MATLAB50 to calculate the local curvature κ\kappa along the droplet’s interface, the c=0.5c = 0.5 contour. In Figure 3(a), we present an illustrative plot of κR0\kappa R_0 along this contour for a droplet with a radius of R0/L0=1R_0/L_0 = 1 at a representative time; the colors help us to distinguish regions of high and low curvature. To quantify curvature fluctuations, we calculate the PDFs of κR0\kappa R_0 for different droplet scaled radii R0/L0R_0/L_0 [semi-log plots in Figure 3(b)]. These PDFs reveal that the larger the droplets the more significant their fluctuations, which appear to be linked to the effectiveness of large droplets as transport barriers.

Figure 3. (a) A pseudocolor plot of the normalized curvature κR0\kappa R_0 along the droplet boundary (the c=0.5c = 0.5 contour) at a representative time. (b) Semi-log plots of the PDF of κR0\kappa R_0 for different values of R0/L0R_0/L_0. (c) Semi-log plots of the PDF of the Okubo-Weiss parameter Λ\Lambda (see text) for different radii; these PDFs are normalized with the root-mean-square values of Λ\Lambda. (d) Plot showing the time-averaged deformation ⟨Γ(t)⟩t\langle\Gamma(t)\rangle_t (see text) of the droplet for different scaled radii R0/L0R_0/L_0.

Curvature fluctuations are intrinsically linked to the deformation of the droplets. To quantify this deformation, we define a deformation parameter as follows25,27:

Γ(t)=S(t)S0(t)−1,(19)\Gamma(t) = \frac{S(t)}{S_0(t)} - 1, \tag*{(19)}

where S(t)S(t) is the perimeter of the droplet at time tt, and S0(t)S_0(t) denotes the perimeter of a circular droplet with an equivalent area at the same time. In Figure 3(d), we present the mean deformation ⟨Γ(t)⟩t\langle\Gamma(t) \rangle_t for various droplet radii; and we confirm that larger the droplet the more its deformation.

Not only do large droplets deform more than small ones, but they also induce significant changes in the flow topology. To investigate these droplet-induced alterations in the flow topology, we calculate the Okubo-Weiss parameter Λ\Lambda, which measures the relative importance of flow strain (Λ>0\Lambda> 0) and vorticity (Λ<0\Lambda< 0) and is defined as follows[5, 42, 51, 52, 53]:

Λ=D2−Ω2;D=∇u+∇uT2;\Lambda= D^2 - \Omega^2; \quad D = \frac{\nabla\boldsymbol{u} + \nabla\boldsymbol{u}^{T}}{2};
Ω=∇u−∇uT2;(20)\Omega= \frac{\nabla\boldsymbol{u} - \nabla\boldsymbol{u}^{T}}{2}; \tag*{(20)}

DD and Ω\Omega are the rate-of-deformation and rate-of-rotation tensor, respectively. In Figure 3(c), we present semi-log plots of the PDFs of Λ\Lambda for various values of R0/L0R_0/L_0. These PDFs reveal that the larger the droplets the more they affect Λ\Lambda; in particular, these droplets enhance the tails of the PDFs of the Okubo-Weiss parameter.

To illustrate how the mean droplet-escape time τξ\tau_\xi [see Eq. (16)] changes with the width of the CHNS interface, we have carried out representative DNS runs with different values of the interfacial width ϵ\epsilon. The semi-log plots of the fraction nin(t)/nn_{\mathrm{in}}(t)/n versus time tt in Figure 4 (a) show clearly that τξ\tau_\xi increases as ϵ\epsilon decreases (with the Reynolds number held fixed at Re=1178Re = 1178). A similar plot [Figure 4 (b)], with a fixed interface width characterised by the Cahn number Cn=0.01Cn = 0.01, shows that τξ\tau_\xi decreases as the Reynolds number ReRe increases. We expect that τξ→∞\tau_\xi\to\infty, in the limit of an infinitely thin interface, i.e. Cn→0Cn \to0; however, it is numerically challenging to approach this limit in the CHNS system at large ReRe.

Figure 4. Semi-log plots versus simulation time tt of the fraction nin(t)/nn_{\mathrm{in}}(t)/n for (a) different values of interface widths ϵ\epsilon (with fixed Re=1178Re = 1178 and R0/L0=0.5R_0/L_0 = 0.5) and (b) different values of the Reynolds number ReRe (with fixed Cahn number Cn=0.01Cn = 0.01 and R0/L0=0.5R_0/L_0 = 0.5).

Conclusions

We have shown that the CHNS system provides a natural theoretical framework for examining how interfaces in turbulent binary-fluid flows act as transport barriers. Lagrangian tracer particles and their spreading in such flows play crucial roles in our quantitative characterisations of these transport barriers. In particular, the fraction of particles inside a droplet decays exponentially and the decay time τξ∼R02/3\tau_\xi\sim R_0^{2/3}. The mean first-passage time ⟨τ⟩\langle\tau\rangle, which we obtain from the CPDF of τ\tau, increases dramatically if the tracers are inside a droplet relative to its value without a droplet. Perimeter and curvature fluctuations of a droplet’s interface increase with R0/L0R_0/L_0 and appear to enhance the ability of the interface to act as a transport barrier. Our studies also uncover the role of the PDF of cos⁡(θ)\cos(\theta) in enhancing the efficacy of this barrier. Our study of interfacial transport barriers in 2D binary-fluid flows can be generalised, mutatis mutandis, to the three-dimensional (3D) CHNS system, to multi-fluid flows [see, e.g., Refs.[17] and [28]], and to active-binary-fluid systems [see, e.g., Ref.[27]].

Transport barriers occur in various turbulent flows. Important examples are found in systems that show staircases such as those that occur in plasma physics [see, e.g., Refs.[54] and [55]] and in double-diffusive convection [see, e.g., Refs.[56] and [57]]. Analogies between the 2D CHNS system and 2D MHD have been explored in detail in Refs.[38] and [39]. Therefore, transport barriers in systems with such staircases can be studied fruitfully by using the Lagrangian methods that we have developed in our CHNS-based study of interfaces as transport barriers in binary-fluid turbulence.

Acknowledgments

We thank J.K. Alageshan, A.K. Verma, and K.V. Kiran for valuable discussions and the Science and Engineering Research Board (SERB) and the National Supercomputing Mission (NSM) [Grant No. DST/NSM/R&D HPC/Applications/2021/34] India for support, and the Supercomputer Education and Research Centre (IISc) for computational resources.

Data and code availability

Data from this study and the computer scripts can be obtained from the authors upon reasonable request.

Conflicts of Interest

No conflicts of interests, financial or otherwise, are declared by the authors.

Author Contributions

NBP and RP planned the research and analysed the numerical data; NBP carried out the calculations and prepared the tables, figures, and the draft of the manuscript; NBP and RP then revised the manuscript in detail and approved the final version.

Local-slope analysis

We obtain τξ\tau_{\xi} from a local-slope analysis of the exponentially decaying tails of the curves in Figure 2(a); we illustrate this analysis in Figure 5.

Semi-log plots and local slopes for particles inside a droplet

Figure 5. (a) Semi-log plots of the fraction of particles inside the droplet nin(t)/nn_{\mathrm{in}}(t)/n versus time tt, illustrating the regions (black lines) within which we evaluate the local slopes for different scaled radii R0/L0R_0/L_0. (b) Plots depicting the local slopes λ=τξ/T\lambda= \tau_{\xi}/T over selected regions in (a). (c) Plot of the local slopes τξ/T\tau_{\xi}/T versus R0/L0R_0/L_0, with one-standard-deviations error bars; the average value of the slope yields the decay time τξ\tau_{\xi} that we report in the main part of this paper.

Videos

  1. Video V1: Video showing the spatiotemporal evolution of the pseudocolor plots of the vorticity field overlaid with positions of the tracer particles (black points) as shown in Figure 1(a)-(Figure 1d) in the absence of droplets. Video URL: https://youtu.be/aXIuJAcYwac

  1. Video V2: Video showing the spatiotemporal evolution of the pseudocolor plots of the vorticity field overlaid with the c=0.5c = 0.5 contour (magenta line) and positions of the tracer particles as shown in Figure 1(e)-(Figure 1h) for the droplet of radius R0/L0=0.25R_0/L_0 = 0.25. Video URL: https://youtu.be/tAjNXb6Pz1I

  1. Video V3: Video showing the spatiotemporal evolution of the pseudocolor plots of the vorticity field overlaid with the c=0.5c = 0.5 contour (magenta line) and positions of the tracer particles as shown in Figure 1(i)-(Figure 1l) for the droplet of radius R0/L0=0.5R_0/L_0 = 0.5. Video URL: https://youtu.be/3wIBDXGkTvA

References

  1. [1]¹R. Scardovelli and S. Zaleski, “Direct numerical simulation of free-surface and interfacial flow,” Annual review of fluid mechanics 31, 567–603 (1999).DOI
  2. [2]²W. J. Rider and D. B. Kothe, “Reconstructing volume tracking,” Journal of computational physics 141, 112–152 (1998).DOI
  3. [3]³A. Roccon, F. Zonta, and A. Soldati, “Phase-field modeling of complex interface dynamics in drop-laden turbulence,” Physical Review Fluids 8, 090501 (2023).
  4. [5]⁵G. Soligo, A. Roccon, and A. Soldati, “Effect of surfactant-laden droplets on turbulent flow topology,” Physical Review Fluids 5, 073606 (2020).DOI
  5. [8]⁸C. Navier, “Mémoire sur les lois du mouvement des fluides,” Mémoires de l’Académie Royale des Sciences de l’Institut de France 6, 389–440 (1823).
  6. [9]⁹C. R. Doering and D. J. Gibbon, Applied analysis of the Navier-Stokes equations, 12 (Cambridge university press, 1995).DOI
  7. [10]¹⁰J. W. Cahn and J. E. Hilliard, “Free energy of a nonuniform system. i. interfacial free energy,” The Journal of chemical physics 28, 258–267 (1958).
  8. [11]¹¹J. W. Cahn and J. E. Hilliard, “Free energy of a nonuniform system. iii. nucleation in a two-component incompressible fluid,” The Journal of chemical physics 31, 688–699 (1959).
  9. [12]J. W. Cahn, “On spinodal decomposition,” Acta metallurgica 9, 795–801 (1961).
  10. [13]P. C. Hohenberg and B. I. Halperin, “Theory of dynamic critical phenomena,” Reviews of Modern Physics 49, 435 (1977).DOI
  11. [14]A. Onuki, Phase transition dynamics (Cambridge University Press, 2002).
  12. [15]S. Puri and V. Wadhawan, Kinetics of phase transitions (CRC press, 2009).
  13. [16]D. Lee, J.-Y. Huh, D. Jeong, J. Shin, A. Yun, and J. Kim, “Physical, mathematical, and numerical derivations of the cahn–hilliard equation,” Computational Materials Science 81, 216–225 (2014).DOI
  14. [17]D. Jacqmin, “Calculation of two-phase navier–stokes flows using phase-field modeling,” Journal of computational physics 155, 96–127 (1999).DOI
  15. [18]F. Magaletti, F. Picano, M. Chinappi, L. Marino, and C. M. Casciola, “The sharp-interface limit of the cahn–hilliard/navier–stokes model for binary fluids,” Journal of Fluid Mechanics 714, 95–126 (2013).DOI
  16. [19]F. Boyer and C. Lapuerta, “Study of a three component cahn-hilliard flow model,” ESAIM: Mathematical Modelling and Numerical Analysis-Modélisation Mathématique et Analyse Numérique 40, 653–687 (2006).DOI
  17. [20]S. Mirjalili, S. Jain, and A. Mani, “Modeling heat and mass transfer across interfaces in two-phase flows using phase-field methods,” .
  18. [21]A. Celani, A. Mazzino, P. Muratore-Ginanneschi, and L. Vozella, “Phase-field model for the rayleigh–taylor instability of immiscible fluids,” Journal of Fluid Mechanics 622, 115–134 (2009).
  19. [22]H. G. Lee and J. Kim, “Numerical simulation of the three-dimensional rayleigh–taylor instability,” Computers & Mathematics with Applications 66, 1466–1474 (2013).DOI
  20. [23]J. D. Gibbon, N. Pal, A. Gupta, and R. Pandit, “Regularity criterion for solutions of the three-dimensional cahn–hilliard-navier–stokes equations and associated computations,” Physical Review E 94, 063103 (2016).DOI
  21. [24]H. G. Lee and J. Kim, “Two-dimensional kelvin–helmholtz instabilities of multi-component fluids,” European Journal of Mechanics-B/Fluids 49, 77–88 (2015).DOI
  22. [25]N. Pal, P. Perlekar, A. Gupta, and R. Pandit, “Binary-fluid turbulence: Signatures of multifractal droplet dynamics and dissipation reduction,” Phys. Rev. E 93, 063115 (2016).DOI
  23. [26]N. Pal, Cahn–Hilliard–Navier–Stokes Investigations of Binary-Fluid Turbulence and Droplet Dynamics, Ph.D. thesis, Indian Institute of Science, Bangalore, India (2016).
  24. [27]N. B. Padhan and R. Pandit, “Activity-induced droplet propulsion and multifractality,” Physical Review Research 5, L032013 (2023).arxiv.org/abs/2209.04864
  25. [28]N. B. Padhan and R. Pandit, “Unveiling the spatiotemporal evolution of liquid-lens coalescence: Self-similarity, vortex quadrupoles, and turbulence in a three-phase fluid system,” arXiv preprint arXiv:2308.08993 (2023).arxiv.org/abs/2308.08993
  26. [38]X. Fan, P. Diamond, L. Chacón, and H. Li, “Cascades and spectra of a turbulent spinodal decomposition in two-dimensional symmetric binary liquid mixtures,” Physical Review Fluids 1, 054403 (2016).DOI
  27. [39]X. Fan, P. Diamond, and L. Chacón, “Chns: A case study of turbulence in elastic media,” Physics of Plasmas 25 (2018).DOI
  28. [40]P. M. Chaikin, T. C. Lubensky, and T. A. Witten, Principles of condensed matter physics, Vol. 10 (Cambridge university press Cambridge, 1995).
  29. [42]R. Pandit, D. Banerjee, A. Bhatnagar, M. Brachet, A. Gupta, D. Mitra, N. Pal, P. Perlekar, S. S. Ray, V. Shukla, et al., “An overview of the statistical properties of two-dimensional turbulence in fluids with particles, conducting fluids, fluids with polymer additives, binary-fluid mixtures, and superfluids,” Physics of fluids 29 (2017).DOI
  30. [43]C. Canuto, M. Y. Hussaini, A. Quarteroni, A. Thomas Jr, et al., Spectral methods in fluid dynamics (Springer Science Business Media, 2012).
  31. [44]S. M. Cox and P. C. Matthews, “Exponential time differencing for stiff systems,” Journal of Computational Physics 176, 430–455 (2002).DOI
  32. [45]C. C. Lalescu and M. Wilczek, “How tracer particles sample the complexity of turbulence,” New Journal of Physics 20, 013001 (2018).DOI
  33. [46]A. K. Verma, A. Bhatnagar, D. Mitra, and R. Pandit, “First-passage-time problem for tracers in turbulent flows applied to virus spreading,” Physical Review Research 2, 033239 (2020).DOI
  34. [47]R. Benzl, L. Biferale, R. Fisher, D. Lamb, and F. Toschi, “Inertial range eulerian and lagrangian statistics from numerical simulations of isotropic turbulence,” Journal of Fluid Mechanics 653, 221–244 (2010).
  35. [51]J. Weiss, “The dynamics of enstrophy transfer in two-dimensional hydrodynamics,” Physica D: Nonlinear Phenomena 48, 273–294 (1991).DOI
  36. [52]A. Okubo, “Horizontal dispersion of floatable particles in the vicinity of velocity singularities such as convergences,” in Deep sea research and oceanographic abstracts, Vol. 17 (Elsevier, 1970) pp. 445–454.DOI
  37. [53]B. Shivamoggi, G. van Heijst, and L. Kamp, “The okubo–weiss criterion in hydrodynamic flows: geometric aspects and further extension,” Fluid Dynamics Research 54, 015505 (2022).DOI
  38. [54]P. Diamond, V. Lebedev, D. Newman, and B. Carreras, “Dynamics of spatiotemporally propagating transport barriers,” Physics of Plasmas 2, 3685–3695 (1995).DOI
  39. [55]A. Ashourvan and P. H. Diamond, “On the emergence of macroscopic transport barriers from staircase structures,” Physics of Plasmas 24 (2017).DOI
  40. [56]T. Radko, Double-diffusive convection (Cambridge University Press, 2013).
  41. [57]R. Ouillon, P. Edel, P. Garaud, and E. Meiburg, “Settling-driven large-scale instabilities in double-diffusive convection,” Journal of Fluid Mechanics 901, A12 (2020).DOI

Paper details

vlnc:W_k4shrtzx
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 3

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