betaFeedback

WorksPhysics and Astronomy

Bubbles-induced transition to elasto-inertial turbulence

Abstract

Interface-resolved direct numerical simulations are performed to investigate bubble-induced transition from laminar to elasto-inertial turbulent (EIT) state in a pressure-driven viscoelastic square channel flow. The Giesekus model is used to account for the viscoelasticity of the continuous phase while the dispersed phase is Newtonian. Simulations are performed for both single and two-phase flows for a wide range of the Reynolds (Re) and the Weissenberg (Wi) numbers. It is demonstrated that injection of bubbles into a laminar viscoelastic flow introduces streamline curvature that is sufficient to trigger an elastic instability leading to a transition to a fully EIT regime. The temporal turbulent kinetic energy spectrum shows a scaling of -2 for this multiphase EIT regime. It is shown that, once the flow is fully transitioned to a turbulent state by the injection of bubbles, the drag increases for all the cases. It is also observed that bubbles move towards the channel centreline and form a string-shaped alignment pattern in the core region at the lower values of Re=10 and Wi=1. In this regime, the flow exhibits an intermittent behaviour, i.e., there are turbulent like fluctuations in the core region while it is essentially laminar near the wall. Unlike the solid particles, it is found that increasing shear-thinning effect breaks up the alignment of bubbles. Interestingly, the drag remains slightly lower in this intermittent regime than the corresponding laminar state.

Introduction

Adding a minute amount of long-chain flexible polymers can drastically change mechanical properties of an otherwise Newtonian fluid and makes it viscoelastic. Stretching of polymer molecules causes elastic stresses and gives a memory to the fluid flow, resulting in a range of exotic and often counter-intuitive behaviours such as rod climbing, elastic recoil and drag reduction in turbulent flows [8]. The exact mechanism behind many of the exotic phenomena exhibited by the viscoelastic fluids are yet to be fully explored. A striking example of such phenomena is sustaining a chaotic, turbulent like fluid motion observed at a low Reynolds number [19, 42, 8, 41]. In case of a classical Newtonian turbulent flow, the flow instabilities arise from the non-linear convective terms of the Navier-Stokes equations, and therefore a very high Reynolds number is always required to sustain a turbulent flow field. In case of a viscoelastic fluid flow, however, the non-linear coupling between viscous and viscoelastic stresses are capable of triggering and sustaining a chaotic motion even in the absence of inertia [8]. This turbulent flow state at a Reynolds number below the threshold value of inertial turbulence (IT) is often referred as the “elasto-inertial turbulence” (EIT). When the Reynolds number is further reduced to a negligibly smaller value, it turns out that high enough viscoelastic stresses are still capable of sustaining a chaotic turbulent-like state, referred as the elastic turbulence (ET), which is purely dominated by the elastic effects [42]. The present study mainly focuses on the elasto-inertial instability and transition to the elasto-inertial turbulence where, as the name implies, the chaotic flow state is caused and sustained by both elastic and inertial effects.

In the inertialess limit, only nonlinearity arises from the elastic stress that becomes anisotropic in a shear flow and results in the hoop stress acting on the fluid in the direction of the streamline curvature. This hoop stress provides the driving force for the elastic instability and triggers a transition to the elastic turbulent (ET) state even in the absence of inertia [42]. Since the discovery of ET in inertialess flows with curved streamlines by Groisman & Steinberg [19], a non-linear instability leading to a similar kind of chaotic state in planar shear flows was suspected and investigated by Morozov & van Saarloos [30] and Morozov & Van Saarloos [29]. These studies suggested that a non-linear instability can also be triggered in a planar shear flow when a sufficient flow curvature is induced via external perturbations. To realize this, the initial perturbations have been usually induced in the form of blowing/suction at the wall or by placing an obstacle in the channel to trigger the instability and attain an EIT state [12].

There are different pathways which can be followed to achieve the EIT state in a wall bounded flow. Samanta et al. [36] have demonstrated that the transition threshold to inertial turbulence can be controlled by varying the polymer concentration. They computed the puff lifetimes to examine the effects of viscoelasticity on transition and differentiate the EIT states from the Newtonian turbulence in a pipe flow. They found that polymer additives delay the subcritical transition to IT conforming to their drag reducing effects in turbulent channel flows. They also demonstrated that, as the shear rate is increased, a separate instability occurs at a much lower Reynolds number than the inertial turbulence and triggers a very different type of disordered motion of elasto-inertial turbulence. Foggi Rota et al. [16] showed that a reverse pathway can also be followed to achieve the EIT state. That is, viscoelasticity is introduced into an already turbulent Newtonian flow and then the Reynolds number is reduced gradually. Unlike the Newtonian flow, which would immediately re-laminarize below a certain threshold value of ReRe, the viscoelastic flow is capable of maintaining its turbulent state even at a Reynolds number as low as Re=0.5Re = 0.5 in a rectilinear channel. Moreover, at a low Reynolds number, if some external perturbations are superimposed on a laminar viscoelastic flow field, these instabilities can grow and achieve the EIT state under a certain parametric setting [17, 3, 2].

The dynamic features of EIT differ greatly from the inertial turbulence, which is the focus of many contemporary investigations. Sid et al. [40] have demonstrated that the essential dynamic structures of EIT exist even in a two-dimensional (2D) flow. More recently, Dubief et al. [11] have identified four distinct regimes of the stress field in a 2D channel flow by using the FENE-P viscoelastic model. These regimes were named as chaotic, chaotic arrowhead, intermittent arrowhead and stable arrowhead regimes based on constant contours of first normal stress difference (N1N_1) and pressure field. A sharp pressure gradient was observed within the flow field, similar to the one across a shock wave, promoting the shape of an arrowhead. In this 2D flow field, it was reported that an increase in the polymer concentration promoted stability while an increase in the domain length promoted chaos, indicating the role of large scales in the dynamics of EIT. The drag was found to be increasing in all these EIT regimes when compared to the corresponding laminar states. In a 3D periodic channel flow, Dubief et al. [14] observed an alternating train of positive and negative QQ-iso-surfaces (second invariant of velocity gradient tensor) near the channel wall during transition to the EIT state. Interestingly, these patterns sustained on a smaller scale unlike the inertial turbulence even when the viscoelastic flow got fully developed. These patterns were interpreted as the regions of local rotation (Q>0Q > 0) and dissipation of turbulent kinetic energy (Q<0Q < 0), and were found to change their alignment from streamwise to spanwise direction during low-drag events. Another striking difference between IT and EIT is the reversed energy cascade. Unlike the inertial turbulence where the energy transfer occurs from large to small scales, energy transfer of turbulent elastic energy to turbulent kinetic energy is observed to be from smaller to larger scales in EIT [13].

Despite an ever increasing interest in this area in recent years, many fundamental aspects related to EIT are still elusive. In the present study, we focus on two such aspects. First, instead of providing any external perturbations to the flow, direct numerical simulations of concentrated viscoelastic flows are performed in a square-shaped channel by gradually increasing the fluid elasticity while keeping the Reynolds number fixed at a low value for which the corresponding Newtonian flow is fully laminar to examine the conditions for attaining an EIT state. Secondly, a novel mechanism of injecting bubbles into a laminar viscoelastic flow is investigated to ascertain whether the presence of bubbles can trigger a transition to an EIT state in this straight channel. For this purpose, extensive interface-resolved direct numerical simulations are performed using a finite-difference/front-tracking method [45]. The log-conformation method [15] is employed to handle the stiff constitutive equations of the viscoelastic model (Giesekus) at high Weissenberg numbers.

The numerical method and the computational setup are described in § 2 & § 3, respectively. The results are presented and discussed for a single-phase and then for the multiphase flows in § 4, followed by the conclusions in § 5.

Governing Equations and Numerical Method

The flow equations and the Giesekus model are presented within the framework of the finite-difference/front-tracking (FT) method. The front-tracking method used in the present study was originally developed by Unverdi & Tryggvason [45]. Izbassarov & Muradoglu [21] added a capability to this FT method to simulate viscoelastic two-phase systems in which one or both the phases could be viscoelastic. This robust and high fidelity method has been extensively used in our several previous works involving laminar (e.g., Izbassarov & Muradoglu [22], Izbassarov & Muradoglu [23], Naseer et al. [32], Naseer et al. [33]) as well as turbulent multiphase flows (e.g., Ahmed et al. [1], Izbassarov et al. [20]). Referring to the coordinate system shown in Figure 1 and employing the one-field formulation, the Navier-Stokes equations are written as

The computational domain and coordinate system

Figure 1. The computational domain and the coordinate system considered in the present study. The constant contours of QQ-criterion are shown around the bubbles (red colour) at a value of Q=25Q = 25 (Re=1000Re = 1000, Wi=10Wi = 10, Ca=0.01Ca = 0.01, β=0.1\beta= 0.1).

ρ∂u∂t+ρ∇⋅(uu)=−∇p−dp0dyj+∇⋅τ+∇⋅μs(∇u+∇uT)+∫Aσκnδ(x−xf) dA(1)\rho\frac{\partial\mathbf{u}}{\partial t}+\rho\nabla\cdot(\mathbf{u}\mathbf{u})=-\nabla p-\frac{d p_{0}}{d y}\mathbf{j}+\nabla\cdot\boldsymbol{\tau}+\nabla\cdot\mu_{s}(\nabla\mathbf{u}+\nabla\mathbf{u}^{\mathrm{T}})+\int_{A}\sigma\kappa\mathbf{n}\delta(\mathbf{x}-\mathbf{x}_{f})\,dA \tag*{(1)}

where u\mathbf{u}, τ\boldsymbol{\tau}, pp, ρ\rho and μs\mu_{s} are the velocity vector, the polymer stress tensor, the pressure, and the discontinuous density and solvent viscosity fields, respectively. A pressure gradient −dp0dyj-\frac{d p_{0}}{d y}\mathbf{j} is applied to drive the flow and it is adjusted dynamically to keep the flow rate constant, where j\mathbf{j} is the unit vector in the yy-direction. The effect of surface tension is added as a body force term on the right-hand side of the momentum equation where σ\sigma is the surface tension coefficient, κ\kappa is twice the mean curvature and n\mathbf{n} is a unit vector normal to the interface. As the surface tension acts only on the interface, δ\delta represents a three-dimensional Dirac delta function with the arguments x\mathbf{x} and xf\mathbf{x}_{f} being a point at which the equation is evaluated and a point at the interface, respectively. The momentum equation is supplemented by the incompressibility condition

∇⋅u=0.(2)\nabla\cdot\mathbf{u}=0. \tag*{(2)}

Viscoelasticity of bulk liquid is modelled using the Giesekus model [18]. Since this model accounts for the polymer–polymer interactions, it is a suitable choice to simulate concentrated polymer solutions [46]. In the Giesekus model, the polymer stress tensor τ\boldsymbol{\tau} evolves by

τ=μpλ(B−I),(3)\boldsymbol{\tau}=\frac{\mu_{p}}{\lambda}(\mathbf{B}-\mathbf{I}), \tag*{(3)}

where μp\mu_{p} is the polymer viscosity, λ\lambda is the polymer relaxation time, B\mathbf{B} is the conformation tensor and I\mathbf{I} is the identity tensor. The conformation tensor evolves by

∂B∂t+u⋅∇B−∇uT⋅B−B⋅∇u=1λ[(1−α)I+(2α−1)B−αB2],(4)\frac{\partial\mathbf{B}}{\partial t}+\mathbf{u}\cdot\nabla\mathbf{B}-\nabla\mathbf{u}^{\mathrm{T}}\cdot\mathbf{B}-\mathbf{B}\cdot\nabla\mathbf{u}=\frac{1}{\lambda}[(1-\alpha)\mathbf{I}+(2\alpha-1)\mathbf{B}-\alpha\mathbf{B}^{2}], \tag*{(4)}

where α\alpha is the mobility factor representing the anisotropy of the hydrodynamic drag exerted on the polymer molecules. Due to the thermodynamic considerations, α\alpha is restricted to 0≤α≤0.50\leq\alpha\leq0.5 [37]. When α=0\alpha=0, the Giesekus model reduces to the Oldroyd-B model.

At high Weissenberg numbers, these highly non-linear viscoelastic constitutive equations become extremely stiff, which makes their numerical solution a challenging task. Artificial diffusion terms may be added to the discretized viscoelastic model equations to alleviate this problem. However, it has been found that, when the model equations are discretized using the conventional numerical schemes, this artificial diffusion method may alter the mathematical nature of the constitutive equations from hyperbolic to parabolic, affecting the integrity of the simulated EIT physics, especially at high Weissenberg numbers [12]. The log-conformation method offers an alternative solution to this problem and it is used in the present study. In this approach, an eigen decomposition is employed to re-write the constitutive equation of the conformation tensor [15, 24]. It is found that, although it is computationally more expensive than the artificial diffusion method, the log-conformation retains the hyperbolic nature and thus removes the stiffness of the viscoelastic model equations, allowing robust and accurate numerical solutions by preserving large-scale features in the simulations of elastic turbulence as shown recently by Yerasi et al. (2024). The interested readers are referred to Fattal & Kupferman [15] for the details of the procedure. The flow equations (Eqs. (1) and (2)) are solved fully coupled with the Giesekus model equation (Eq. (3)). A QUICK scheme is used to discretize the convective terms in the momentum equations while second-order central differences are used for the diffusive terms. For the convective terms in the viscoelastic equations, a 5th5^{th}-order WENO-Z (Borges et al. [4]) scheme is used. An FFT-based solver is employed to solve the pressure Poisson equation. Since the pressure equation is not separable due to variable density in the present multiphase flow, the FFT-based solvers cannot be used directly. To overcome this challenge, a pressure-splitting technique presented by Dong & Shen [10] and Dodd & Ferrante [9] is employed. A predictor-corrector scheme is used to achieve second-order time accuracy as described by Tryggvason et al. [43]. The details of the front-tracking method can be found in the book by Tryggvason et al. [44] and in the review paper by Tryggvason et al. [43], and the treatment of the viscoelastic model equations in Izbassarov & Muradoglu [21] and Izbassarov et al. [24]. The present numerical scheme is second order accurate both in space and time.

Computational setup

Figure 1 shows the computational domain which is a square channel with the dimensions of 2h×12h×2h2h \times12h \times2h in the xx, yy and zz directions, respectively, where hh is half of the channel width. Periodic boundary conditions are applied in the streamwise (yy) direction whereas the other two directions (xx and zz) have no-slip/no-penetration boundary conditions. The length of the channel is set to 12h12h so that a sufficient length is available for the bubbles to exhibit their effects in the viscoelastic flow field.

The flow conditions are characterized by the following non-dimensional numbers defined as

Re=ρouohμo,Wi=λuoh,Ca=μouoσ,β=μsμo,(5)Re = \frac{\rho_o u_o h}{\mu_o}, \quad Wi = \frac{\lambda u_o}{h}, \quad Ca = \frac{\mu_o u_o}{\sigma}, \quad\beta= \frac{\mu_s}{\mu_o}, \tag*{(5)}

where ReRe, WiWi and CaCa are the Reynolds, Weissenberg and capillary numbers, respectively. Further, β\beta is the ratio of the solvent viscosity (μs\mu_s) to the zero shear viscosity (μo\mu_o) of the viscoelastic fluid. In Eq. (5), ρo\rho_o is the density of the bulk fluid and uou_o is the average velocity in the channel. The viscosity ratio is fixed at β=0.1\beta= 0.1, representing a highly concentrated polymer solution, for all the results presented in this paper.

The present study covers the Reynolds and the Weissenberg numbers spanning in the ranges of 10≤Re≤100010 \le Re \le1000 and 1≤Wi≤10001 \le Wi \le1000. Note that the highest value of Re=1000Re = 1000 is still less than the minimum value required to sustain inertial turbulence in a channel flow [34]. For these low values of ReRe, the grid resolution near the wall cannot be determined by the classical wall criterion of x+<1x^{+} < 1 and z+<1z^{+} < 1. As the bubbles are also injected into the flow, despite a low ReRe, a sufficiently fine grid is required to resolve the bubble interfaces and to capture the essential flow features of this chaotic viscoelastic flow regime. Therefore, a grid size of 320×960×320320 \times960 \times320 is used in the xx, yy and zz directions even for Re=10Re = 10 case. For a reference, this grid resolution is sufficient to achieve x+<0.8x^{+} < 0.8 and z+<0.8z^{+} < 0.8 in an inertial turbulent channel flow at Re=5600Re = 5600 (Reτ=180Re_{\tau} = 180). For the multiphase cases, there are 64 grid cells per equivalent bubble diameter with this selected grid size.

The flow is driven by applying dpo/dyd p_o/d y in the negative yy-direction. This external pressure gradient is adjusted dynamically at each time step to keep the flow rate (and thus the bulk Reynolds number) constant in the channel. To normalize various quantities reported in this study, hh is used as the length scale, uou_o as the velocity scale and h/uoh/u_o as the time scale. The normalized quantities are denoted by the superscript *. The stresses are normalized by ρovτ2\rho_o v_{\tau}^{2}, where vτ=τ‾w/ρov_{\tau} = \sqrt{\overline{\tau}_w/\rho_o} with τ‾w\overline{\tau}_w being the average total wall shear stress. Once the flow reaches a statistically steady state, the combined force due to the shear stresses at the four walls of the channel is balanced by the force due to the applied pressure gradient dpo/dyd p_o/d y. In this multiphase flow, the density and viscosity of the bubble are denoted by ρi\rho_i and μi\mu_i, respectively. The density and the viscosity ratios are set to ρo/ρi=10\rho_o/\rho_i = 10 and μo/μi=80\mu_o/\mu_i = 80, respectively. These comparatively small ratios are used to enhance numerical stability and thus relaxing the time step restrictions. Apparently, this low density ratio may seem unrealistic when compared to a liquid-air system. However, a higher density ratio does not affect the bubble dynamics as has been reported by Bunner & Tryggvason [6] for a Newtonian flow. Further simulations are also performed here to check the effects of the density and viscosity ratios on the viscoelastic multiphase flows considered in this study. As documented in the Appendix, the results are not very sensitive to a further increase in the density and viscosity ratios beyond ρo/ρi=10\rho_o/\rho_i = 10 and μo/μi=80\mu_o/\mu_i = 80.

Results and discussion

Simulations are first performed for a single-phase flow by gradually increasing the Reynolds number from Re=10Re = 10 to Re=1000Re = 1000. Subsequently, the flow is made more viscoelastic by gradually increasing the Weissenberg number in the range of 1≤Wi≤10001 \le Wi \le1000 at a fixed Reynolds number to ascertain whether the flow attains an EIT state without introducing any explicit external perturbations. To facilitate the development of second normal stress difference (N2N_2), a square-channel (duct) is used and a small shear-thinning effect is also incorporated by setting α=0.001\alpha= 0.001 for the Giesekus model. A concentrated polymer solution (β=0.1\beta= 0.1) is selected to follow the pathway identified by Samanta et al. [36] for the earlier possible transition to an EIT regime. After identifying the range of WiWi for which the single-phase flow remains essentially laminar, bubbles are subsequently injected into fully developed viscoelastic laminar flows at Wi=1,5Wi = 1, 5 and 1010 while keeping ReRe constant to ascertain whether the presence of bubbles can trigger a flow instability. The bubble volume fraction is kept constant at 3% for all the cases. The capillary number is fixed at a low value of Ca=0.01Ca = 0.01 to maintain approximately spherical bubble shapes. The resultant single-phase and multiphase flows are presented and discussed in the following sections.

EIT in a single-phase flow

Figure 2 summarizes the flow states in the ReRe-WiWi space. At a low value of Re=10Re = 10, the single-phase flow remains fully laminar for up to Wi=1000Wi = 1000. Two points are selected in the computational domain to collect all the components of velocity and viscoelastic stresses at each time step to monitor their evolution to ascertain the attainment of a statistically steady-state solution. One of the ‘numerical probes’ is located at the centre of the channel (x∗=1x^*=1, y∗=6y^*=6, z∗=1z^*=1) while the other one at 25% distance (x∗=0.5x^*=0.5, y∗=6y^*=6, z∗=1z^*=1) from one of the walls. Once the flow reaches a statistically steady state, the simulations are continued at least for another 3λ3\lambda time units, during which full flow fields are stored for a further statistical analysis.

Figure 2. Various flow states in the ReRe-WiWi space. (Ca=0.01Ca = 0.01, β=0.1\beta= 0.1)

The simulations are then repeated for Re=100Re = 100. While gradually increasing WiWi, it is observed that the flow remains laminar till Wi=100Wi = 100. However, for Wi≥100Wi \ge100, instabilities start to appear in this low ReRe flow. These instabilities start to grow with time, and fluctuations are observed ultimately in all three components of velocity akin to a typical turbulent signal. Viscoelastic stresses also start to fluctuate in time domain, as collected by the numerical probes at two different locations of the channel. It is important to emphasize that no explicit perturbations are provided from outside the system to initialize any instabilities in the flow. Only for Wi≥100Wi \ge100, the viscoelastic stresses are high enough to trigger an instability that ultimately leads to a chaotic flow regime. As the flow rate is kept constant for a fixed ReRe, a sudden increase in the applied pressure gradient is also observed once the flow is transitioned from laminar to this chaotic state, indicating a rapid increase in the drag. The characteristics of this chaotic flow regime for Wi≥100Wi \ge100 will be discussed in detail in the subsequent paragraphs.

When ReRe is increased further to 1000, the transition to a chaotic flow regime starts to appear at a lower value of Wi≥50Wi \ge50. This ReRe is still smaller than the minimum Recr≈1100Re_{\mathrm{cr}} \approx1100 required for sustaining the inertial turbulence in a channel flow (Owolabi et al. [34]). As WiWi is increased further, the turbulent intensity of this chaotic regime increases with a change in statistical quantities as well.

Figure 3a shows the velocity vectors in a vertical cutting xzxz-plane of the viscoelastic duct flow once it reaches a statistically steady state for the Re=1000Re = 1000 and Wi=1000Wi = 1000 case. The instantaneous and time-averaged flow fields are plotted in left and right panels. of Figure 3a, respectively. As seen, the instantaneous flow field exhibits a chaotic flow regime while the time-averaged velocity field shows an almost similar secondary flow pattern with eight vortices as also observed in the viscoelastic laminar duct flow with the shear-thinning effect [33] (Li et al. 2015). This secondary flow field is developed in the duct flow primarily due to the presence of a second normal stress difference N2=τˉzz−τˉxxN_2 = \bar{\tau}_{zz} - \bar{\tau}_{xx}. The instantaneous and time-averaged contour plots of N2N_2 are shown in the left and right panels of Figure 3b, respectively. The instantaneous plot confirms the chaotic flow pattern while the time-averaged one resembles the typical distribution of N2N_2 in a laminar duct flow. As the EIT regime is associated with long-elongated sheets of the first normal difference (N1N_1), the distribution of N1N_1 is also shown in Figure 3c in the yzyz-plane at the centre of the duct for the various combinations of ReRe and WiWi for which the single-phase flow becomes chaotic. Dubief et al. [12] categorized different EIT regimes based on the structures of N1N_1 for a 2D channel flow. For a low value of Re=100Re = 100, a similar ‘chaotic regime’ is observed in the present 3D duct flow as shown in Figure 3c. At a higher value of Re=1000Re = 1000 but with a moderate value of Wi=100Wi = 100, this regime is shifted to ‘chaotic arrowhead’ and once the Weissenberg number is increased to a very high value of

Instantaneous and time-averaged velocity vectors, normal stress differences, and Q-criterion iso-surfaces

Figure 3. (a) Instantaneous (left) and time-averaged (right) snapshots of velocity vectors showing the secondary flow field at Re=1000Re = 1000, Wi=1000Wi = 1000 (b) Instantaneous (left) and time-averaged (right) snapshots of second normal stress difference (N2N_2) at Re=1000Re = 1000 and Wi=1000Wi = 1000 (c). Contours of first normal stress difference (N1N_1) are shown in a statistically steady EIT regime for different values of ReRe and WiWi. (d) Iso-surfaces of QQ-criterion are shown close to the bottom wall of the channel in a statistically steady EIT regime for different values of ReRe and WiWi. The red and gray colours represent the positive and the negative values, respectively. The contours are plotted at ±5\pm5, ±0.05\pm0.05 and ±0.005\pm0.005 for (Re,Wi)=(1000,1000)(Re, Wi) = (1000, 1000), (1000,100)(1000, 100) and (100,1000)(100, 1000) cases, respectively.

Wi=1000Wi = 1000, the long-elongated sheets of N1N_1 break down completely and the flow starts to resemble a typical inertial turbulence. The iso-surfaces of the second invariant of velocity gradient tensor (QQ-criterion) is depicted in Figure 3d to show the corresponding change in the near-wall vortices for the same combinations of ReRe and WiWi. At a very high value of ReRe and WiWi, the positive and negative contours of the QQ-criterion show a chaotic pattern near the wall. With decreasing ReRe or WiWi, the spanwise elongated pattern of positive and negative contours of the QQ-criterion starts to appear indicating that the signatures of elastic turbulence become more and more apparent in the EIT regime once the elastic effects dominate over the inertial effects. In the absence of any perturbations from outside the system, the flow is laminar at a lower value of WiWi in the present scenario, as shown in Figure 2. Therefore, a clear pattern of alternating contours of QQ-criterion is not visible here as also observed by Dubief et al. [12].

Figure 4 shows various quantities used to evaluate turbulent characteristics of this single-phase chaotic flow regime for various combinations of ReRe and WiWi. The first one is the 1D energy spectrum plotted for Wi=100Wi = 100 and Wi=1000Wi = 1000 at Re=1000Re = 1000 (Figure 4a). The energy axis is normalized by 12ρouo2\frac{1}{2}\rho_o u_o^2 and the frequency axis by uo/hu_o/h. At the higher value of Wi=1000Wi = 1000, the kinetic energy spectrum shows a typical −5/3-5/3 scaling at the large scales but it shifts to −4-4 at the smaller scales (dissipation range). These values are consistent with the results obtained in a single-phase channel flow at lower values of ReRe by Mukherjee et al. [31] and Foggi Rota et al. [16]. Once WiWi is decreased to Wi=100Wi = 100 while keeping ReRe the same, the scaling of energy spectrum starts to deviate from −4-4. At this comparatively higher value of ReRe (but still lower than the minimum one required for the inertial turbulence to sustain in a Newtonian flow), the turbulent characteristics exhibit the manifestation of both the inertial and the elastic effects. Figure 4c shows the energy spectra obtained for Re=100Re = 100 and Re=1000Re = 1000 at Wi=1000Wi = 1000. As seen, the flow is dominated by the elastic effects at Re=100Re = 100. Hence, the maximum range of energy spectrum gets closer to the scaling of −4-4. It can be inferred from these scales that once the turbulent flow is entirely governed by inertia (classical inertial turbulence), the energy spectrum assumes a slope of −5/3-5/3 whereas once the turbulent regime is dominated by the elastic effects with negligible inertia, as in ET, the spectrum exhibits a slope of −4-4. Any combination of ReRe and WiWi in between these two regimes assumes a slope in between these two limits.

Energy spectra, shear-stress components, and turbulent kinetic energy distributions

Figure 4. The 1D energy spectrum for (a) (Re,Wi)=(1000,1000)(Re, Wi) = (1000, 1000) & (Re,Wi)=(1000,100)(Re, Wi) = (1000, 100) and for (c) (Re,Wi)=(1000,1000)(Re, Wi) = (1000, 1000) & (Re,Wi)=(100,1000)(Re, Wi) = (100, 1000). The distribution of the components of the average shear stress from the channel wall towards the centre in the mid-plane for (b) (Re,Wi)=(1000,1000)(Re, Wi) = (1000, 1000) (top panel) & (Re,Wi)=(1000,100)(Re, Wi) = (1000, 100) (bottom panel) and for (d) (Re,Wi)=(1000,1000)(Re, Wi) = (1000, 1000) (top panel) & (Re,Wi)=(100,1000)(Re, Wi) = (100, 1000) (bottom panel) (e, f, g) The distribution of turbulent kinetic energy in the mid-plane of the channel for different values of ReRe and WiWi. In the insets of respective figures, the average elongation of polymer molecules represented by tr⁡(Bˉ)\operatorname{tr}(\bar{B}) are shown for the same mid-planes.

The total mean shear stress in the mid-plane comprises three components: the viscous shear due to mean flow τˉN\bar{\tau}_N, the Reynolds stress due to velocity fluctuations τˉR\bar{\tau}_R and a polymeric stress τˉP\bar{\tau}_P. Once the flow reaches a statistically steady state, the applied pressure gradient must balance the total shear stress at the four walls of the channel. The three components of shear stress (τˉN=μo∂uˉ∂z\bar{\tau}_N = \mu_o \frac{\partial\bar{u}}{\partial z}, τˉR=−ρou′v′‾\bar{\tau}_R = -\rho_o \overline{u'v'}, τˉP=τˉyz\bar{\tau}_P = \bar{\tau}_{yz}) are plotted for the central plane of the duct (x∗=1x^* = 1) for different values of ReRe and WiWi (Figure 4b & Figure 4d). The fluctuating and the time-averaged quantities are denoted by prime (′') and overbar (ˉ\bar{}), respectively. Note that the averaging is performed both in time and in streamwise direction since the flow is expected to be homogeneous in the streamwise direction once a statistically steady state is reached. The stress components are normalized by the local shear stress at the wall of the respective plane. For high values of Re=1000Re = 1000 and Wi=1000Wi = 1000, it is observed that the stress balance in the mid-plane of the channel resembles that of a classical inertial turbulence. The viscous stress is maximum at the wall and decays to zero at the channel centre whereas the Reynolds stress is zero at the wall, gradually increases to its peak value, and then goes to zero at the channel centre. The viscoelastic stress also follows the viscous stress profile and becomes maximum at the wall and zero at the channel centre. As a result, the total shear stress decays linearly from the wall towards the channel centre (Figure 4b). Once WiWi is reduced while keeping Re=1000Re = 1000, the magnitude of Reynolds stress becomes negligibly small compared to viscous and polymeric stresses. Although the same scaling of the energy spectrum and the turbulent signal are indicative of a turbulent flow field at this comparatively lower value of Wi=100Wi = 100, the Reynolds stress becomes negligibly small compared to the overall shear stress in the mid-plane. These not-so-smooth profiles of different shear stress components indicate a clear departure of the turbulent flow field from the inertial one. A similar situation is observed at lower inertia as well. For instance, at a low Reynolds number (Re=100Re = 100) and a high Weissenberg number (Wi=1000Wi = 1000), the turbulent flow field shows almost the same scaling of energy spectrum but the total shear stress remains to be dominated by the viscous and polymer stress components (Figure 4d). The distribution of

the turbulent kinetic energy (TKE) K=0.5(u′2‾+v′2‾+w′2‾)\mathcal{K} = 0.5(\overline{u^{\prime2}}+\overline{v^{\prime2}}+\overline{w^{\prime2}}) is shown in Figure 4e, Figure 4f & Figure 4g at the mid-plane where TKE is found to be maximum near the walls and decays to a minimum value at the centre. Interestingly, the average elongation of polymer molecules quantified by the trace of the conformation tensor tr⁡(B‾)\operatorname{tr}(\overline{\mathbf{B}}) shows a similar trend for different values of WiWi as shown in the insets of Figure 4e, Figure 4f & Figure 4g, confirming the fact that turbulent kinetic energy is mainly produced by the elastic effects in this EIT regime. It is worth mentioning that the value of the applied pressure gradient decreases as WiWi increases for a constant value of ReRe, indicating a decrease in the drag force. This trend is consistent with the reversed pathway; that is, when polymer molecules are added into an already turbulent Newtonian flow, drag is reduced with an increase in WiWi (Serafini et al. [38]).

It is important to highlight that exceptionally high Weissenberg numbers required to trigger flow instability in this rectilinear channel may seem impractical. However, it should be noted that WiWi is defined based on the average flow velocity in the channel. As shear rate decreases significantly towards the channel centre in this duct flow, the effective local Weissenberg number is actually much lower than the nominal value of WiWi, which is based on the average flow velocity.

Turbulent kinetic energy

The turbulent kinetic energy (K=12ui′ui′‾\mathcal{K} = \frac{1}{2}\overline{u_i'u_i'}) in a viscoelastic flow evolves by

∂K∂t=−u‾j∂K∂xj−12∂ui′ui′uj′‾∂xj−1ρoui′∂p′∂xi‾−ui′uj′∂u‾i∂xj‾+12∂∂xj(νs∂ui′ui′‾∂xj)−νs∂ui′∂xj∂ui′∂xj‾+1ρoui′∂τij′∂xj‾+ui′fi′‾.(6)\begin{aligned} \frac{\partial\mathcal{K}}{\partial t} ={}&-\overline{u}_j\frac{\partial\mathcal{K}}{\partial x_j} -\frac{1}{2}\frac{\partial\overline{u_i'u_i'u_j'}}{\partial x_j} -\frac{1}{\rho_o}\overline{u_i'\frac{\partial p'}{\partial x_i}} -\overline{u_i'u_j'\frac{\partial\overline{u}_i}{\partial x_j}} \\ &+\frac{1}{2}\frac{\partial}{\partial x_j}\left(\nu_s\frac{\partial\overline{u_i'u_i'}}{\partial x_j}\right) -\nu_s\overline{\frac{\partial u_i'}{\partial x_j}\frac{\partial u_i'}{\partial x_j}} +\frac{1}{\rho_o}\overline{u_i'\frac{\partial\tau_{ij}'}{\partial x_j}} +\overline{u_i'f_i'} . \tag*{(6)} \end{aligned}

where A\mathcal{A}, Q\mathcal{Q}, R\mathcal{R}, P\mathcal{P}, D\mathcal{D}, ϵ\epsilon, Wp\mathcal{W}_p and B\mathcal{B} represent advection by mean flow, transport by velocity fluctuations, transport by pressure, production by mean flow, viscous diffusion, viscous dissipation, polymer work and work done by body force, respectively. In a viscoelastic turbulent flow, Wp\mathcal{W}_p may change its sign and can serve as either dissipation or production depending upon the signs of the polymer stress fluctuations and fluctuating velocity gradients as has been observed in both experimental (Ptasinski et al. [35]) and numerical (Dallas et al. [7]) works.

To get better insight of the EIT regime in this duct flow, these terms are averaged in time and in the streamwise (yy) direction once the flow reaches a statistically steady state. The constant contours of all the terms in Eq. (6) are shown in Figure 5 at two different values of ReRe in a vertical cutting xzxz-plane except for the body force term that is zero for this single-phase flow. For Re=1000Re = 1000, the advection by mean flow (Figure 5a-left) shows small regions of positive peaks at the four corners and towards the centre of the duct. A pattern of four negative peaks is also observed slightly away from the walls at this high Reynolds number EIT flow. However, once the Reynolds number is lowered to Re=100Re = 100, the magnitude of the advection term becomes negligibly small. Similarly, a symmetric pattern of transport term is observed for the higher Reynolds number (Figure 5b-left) and the same term becomes negligibly small at lower inertia. At Re=1000Re = 1000, the positive and negative peaks of the transport term remain closer to the channel walls except at the corners where it remains approximately zero. An opposite trend is observed for the pressure term (Figure 5c). Once the flow is dominated by the inertial effects (Re=1000Re = 1000), the pressure term remains significant while, in an elastically dominated EIT regime (Re=100Re = 100), it shows a symmetric pattern of positive and negative zones, however, with a negligibly small magnitude. The positive peaks of the pressure term extend from four corners of the duct till the centre with negative peaks in the vicinity of the channel walls. Turbulence production by the mean flow is mainly produced near the walls except at the four corners (Figure 5d) and becomes zero in the most of the central region of the duct with a symmetric pattern at Re=1000Re = 1000. However, in the elastically dominated EIT flow (Re=100Re = 100), this pattern changes with four regions of positive peaks away from the walls. As the turbulence is mainly produced by the elastic effects at Re=100Re = 100, the magnitude of the production term also remains negligibly small. Diffusion term (Figure 5e) shows positive peaks adjacent to the walls except at the corners, immediately followed by the negative peaks before decaying to zero for Re=1000Re = 1000 whereas, for the Re=100Re = 100 case, it is positive near the walls and remains chaotic in the most of the remaining central portions of the duct. The dissipation term shows a similar pattern for both the cases (Figure 5f), i.e., it is maximum at the walls except in the corners and becomes zero towards the duct centre. The most important term in this EIT regime is the work done by the elastic force (Figure 5g) as it can contribute both towards generation or dissipation of turbulence. At higher inertia (Re=1000Re = 1000), the positive peak of the polymer work is observed to be closer to the four walls of the channel. It indicates that the viscoelastic stresses produce turbulence away from the walls

Contours of advection by mean flow, transport by velocity fluctuations, transport by pressure, production by mean flow, viscous diffusion, viscous dissipation, polymer work and time derivative of TKE

Figure 5. Contours of (a) advection by mean flow, (b) transport by velocity fluctuations, (c) transport by pressure, (d) production by mean flow, (e) viscous diffusion, (f) viscous dissipation, (g) polymer work and (h) time derivative of TKE are shown in a vertical cutting xzxz-plane for (left) Re=1000Re = 1000, Wi=1000Wi = 1000 and (right) Re=100Re = 100, Wi=1000Wi = 1000 cases, respectively.

but act as a dissipating force closer to the walls of the channel. The same term remains negligible at the centre. Interestingly, at lower inertia where the elastic effects dominate the EIT regime, the polymer work contributes towards turbulence production at the four corners and towards the central region of the duct. The dissipative role of polymer work is mainly observed at the four walls except the corners. The viscous dissipation and the pressure terms follow the same pattern, as seen in Figure 5f & Figure 5c, respectively. Finally, the summation of all the terms in Eq. (6) approaches zero confirming that a statistically steady state is reached, as shown in Figure 5h.

Notably, the distribution of various components of TKE in the present EIT regime of a highly concentrated polymer solution at low ReRe is markedly different than that in a dilute viscoelastic turbulent duct flow at high ReRe as studied by Shahmardi et al. [39], where the turbulent statistics of the flow were entirely dominated by the inertial effects.

EIT in a multiphase flow

For a given Reynolds number, a significantly high WiWi is always required to achieve the EIT state in a single phase flow in the absence of any external perturbations, as has been shown in the previous section. When ReRe is low (e.g., Re=10Re = 10), even the Weissenberg number as high as 1000 cannot make the flow unstable (Figure 2). Similarly, the single-phase flow is found to be essentially laminar at the lower values of WiWi even when Reynolds number is as high as Re=1000Re = 1000.

We next examine the effects of bubble injection into the viscoelastic laminar channel flow. For this purpose, simulations are performed for three relatively low values of Wi=1Wi = 1, 5 and 10 at each value of Re=10Re = 10, 100 and 1000. Note that the the single phase flow remains laminar for these combinations of WiWi and ReRe. Calculations are first carried out for the single-phase flow until a statistically steady state is reached. Then, spherical bubbles are injected instantaneously into the flow with a volume fraction of 3%. The bubbles are initially injected randomly and uniformly in the entire duct. The capillary number is fixed at Ca=0.01Ca = 0.01 to keep the bubbles nearly spherical. The state of this multiphase flow field is continuously monitored at each time step using the two numerical probes as for the single-phase case. By introducing the bubbles into the flow, the flow is found to achieve a chaotic state even for the Reynolds and Weissenberg numbers as low as Re=10Re = 10 and Wi=5Wi = 5. Time histories of instantaneous velocity components are plotted in Figure 6a for the Re=1000Re = 1000 and Wi=5Wi = 5 case to show the transition from a laminar to an EIT state after the injection of bubbles. Bubbles are injected at about t∗=400t^* = 400. As

Evolution of flow velocity and contours of vorticity magnitude for single-phase and multi-phase regimes

Figure 6. (a) Evolution of flow velocity and its transition to turbulent state once bubbles are injected into the flow (b) The contours of vorticity magnitude in an xzxz-plane at the centre of the domain for the single-phase and multi-phase regimes (Re=1000Re = 1000, Wi=5Wi = 5, β=0.1\beta= 0.1, Ca=0.01Ca = 0.01)

seen, the signals are very smooth until the injection of bubbles indicating a laminar flow. Once the bubbles are injected into the flow, all three components of the velocity exhibit random fluctuations indicating a transition to the EIT state. These fluctuations grow with the passage of time and ultimately the entire flow becomes fully turbulent. This turbulent flow state is visualized by plotting constant contours of vorticity (ω=∇×u\omega= \nabla\times u) magnitude in Figure 6b for the single-phase and multiphase regimes in a vertical cutting xzxz-plane at the centre of the duct. Once the flow reaches a statistically steady state after the injection of bubbles, the simulations are continued at least for another 10 flow-through time units (10×12h/uo10 \times12h/u_o) to collect data for further statistical analysis.

fig:7a shows the distribution of first normal stress difference (N1N_1) in the mid-plane for three different values of Re=10Re = 10, Re=100Re = 100 and Re=1000Re = 1000 at Wi=10Wi = 10. A ‘chaotic arrowhead’ regime is observed for the all three values of ReRe at this relatively low WiWi for which the single-phase flow remains fully laminar. Note that all bubbles are of the same size in Figure 7a and the seemingly different appearance of the bubble sizes is simply caused by the out-of-plane bubbles at this particular time instant. The constant contours of QQ-criterion at the walls for the same three cases are depicted in Figure 7b, showing a chaotic pattern similar to the one observed in the single-phase cases at low value of WiWi (Figure 3d). As the instability is solely triggered by the presence of bubbles, the spanwise elongated structures of QQ-criterion observed by Dubief et al. [12] by superimposing the isotropic turbulence as an initial condition to achieve the EIT regime are not observed in the present scenario. The contours of QQ-criterion at the wall are shown in Figure 7c for Re=10Re = 10, 100 and 1000 at Wi=1Wi = 1. It is observed that, for this low Wi=1Wi = 1, the flow remains laminar at the walls for Re=10Re = 10 and Re=100Re = 100 cases. The chaotic flow regime remains confined to the central region of the duct away from the walls. The positive and negative structures of QQ-criterion are only visible at the walls once plotted at negligibly smaller values (i.e; ±1×10−5\pm1 \times10^{-5}). For Re=1000Re = 1000, however, a chaotic flow regime is observed even at Wi=1Wi = 1 as seen by the contours of QQ-criterion (Figure 7c).

Contours, iso-surfaces, and bubble distributions for different values of $Re$ and $Wi$

Figure 7. (a) Contours of N1N_1 in the mid-plane for different values of ReRe at Wi=10Wi = 10. Iso-surfaces of QQ-criterion at the bottom walls of the channel for the multiphase EIT regime at (b) Wi=10Wi = 10 and at (c) Wi=1Wi = 1 for different values of ReRe. For (b), the contours are plotted at ±0.00001\pm0.00001, ±0.001\pm0.001 and ±0.01\pm0.01 for Re=10Re = 10, 100 & 1000 cases with red colour (positive) and grey (negative), respectively. For (c), the contours are plotted at ±0.00001\pm0.00001 for Re=10Re = 10, 100 and at ±0.1\pm0.1 for Re=1000Re = 1000 case. (d) Distribution of bubbles coloured by the magnitude of streamwise flow velocity in the channel for different values of ReRe at Wi=1Wi = 1

The bubble distributions in the duct are depicted in Figure 7d. It is interesting to see that, although the bubble distribution is essentially uniform in the other cases, the majority of the bubbles are aligned in the centre of the duct forming a string-shaped pattern for the lower values of Re=10Re = 10 and Wi=1Wi = 1, which also explains the reason that the disturbances remain confined to the core region of the duct away from the walls for this case. In the case of non-colloidal spherical solid particles, Won & Kim [47] have experimentally shown that although the main driving force for the lateral migration of particles is the first normal stress difference, particle alignment is actually promoted by the shear-thinning effect for a particular rheological setting. In the present case of bubbles, the role of shear-thinning effect in the alignment of bubbles is investigated by performing another simulation at Re=10Re = 10 and Wi=1Wi = 1 with a higher shear thinning and the results are shown in Figure 8. The change in the effective viscosity (μe\mu_e) of the flow due to shear-thinning effect at different shear rates (γ˙\dot{\gamma}) is governed by the concentration of polymers (β\beta) and the mobility factor (α\alpha) in the Giesekus model (Figure 8a). The slope of viscosity versus shear rate plot is controlled by the mobility factor α\alpha and the final value of viscosity at a high enough shear rate is determined by the concentration of polymers β\beta (Yoo & Choi [48]). The vertical line in Figure 8a indicates the shear rate at the wall for the mid-plane of the duct for the Re=10Re = 10 and Wi=1Wi = 1 case. Once the mobility factor is increased to α=0.1\alpha= 0.1, a distribution of bubbles is observed instead of the string-shaped aggregation at the centre of the duct (Figure 8c). We thus conclude that, unlike the solid particles, a higher shear-thinning effect of the ambient fluid breaks up the alignment of bubbles at least for this particular rheological setting. Further investigation is required to reveal the exact mechanism behind the role of shear-thinning effect on

Change in effective viscosity, velocity signal, and bubble distribution

Figure 8. (a) The change in the effective viscosity of a Giesekus fluid in a concentrated polymer solution. The vertical line in (a) shows the value of shear rate at the wall for the central yzyz-plane of the duct. (b) The velocity signal collected from the central point (x∗=1,y∗=6,z∗=1)(x^* = 1, y^* = 6, z^* = 1) of the domain for α=0.001\alpha= 0.001. In the inset, the same signal is shown for α=0.1\alpha= 0.1 case. (c) The distribution of bubbles in the duct for different values of shear thinning parameter α\alpha (Re=10,Wi=1,Ca=0.01,β=0.1Re = 10, Wi = 1, Ca = 0.01, \beta= 0.1)

different forces acting on the bubbles and causing them to align in the central region of the duct.

As the flow instability is mainly triggered by the presence of the bubbles, the chaotic flow also remains restricted to the core region of the duct once the bubbles are aligned, as shown by the contours of QQ-criterion in Figure 9a. At this low values of Re=10Re = 10 and Wi=1Wi = 1, even when the bubbles remain randomly distributed in a flow with high shear-thinning effect, the instability remains confined to immediate surroundings of the bubbles only. The velocity signals collected from the central point of the duct (x∗=1,y∗=6,z∗=1)(x^* = 1, y^* = 6, z^* = 1) for different values of α\alpha are shown in Figure 8b. The velocity shows random fluctuations in all three components when the bubbles are aligned close to the central point of the duct. Once the bubbles are randomly distributed, the velocity components show negligible fluctuations at the central point as shown in the inset of Figure 8b. When the inertia is increased (Re=100Re = 100) at the same low value of Wi=1Wi = 1, majority of the bubbles accumulate in the core region in the form of small clusters. The flow still remains laminar at the walls due to low inertia and the presence of few bubbles there. Finally, once the inertia is very high (Re=1000Re = 1000), the bubbles become randomly distributed in the entire duct and, as a result, a fully chaotic flow regime is observed in the entire domain (Figure 9a) with a significant increase in the friction drag.

Iso-surfaces, energy spectra, shear-stress components, and turbulent kinetic energy for different Reynolds numbers

Figure 9. (a) Iso-surfaces of QQ-criterion are shown at different values of ReRe for Wi=1Wi = 1. The contours are plotted at 0.003, 0.3 and 3 for Re=10Re = 10, 100 and 1000 cases, respectively. (b,c,d) 1D energy spectrum, (e,f,g) the distribution of various components of shear stress and (h,i,j) turbulent kinetic energy (KK) are plotted in the mid-plane for Re=10Re = 10 (left), Re=100Re = 100 (middle) and Re=1000Re = 1000 (right) cases, respectively (β=0.1\beta= 0.1 and Ca=0.01Ca = 0.01).

fig:9b shows 1D energy spectrum of the streamwise component of flow velocity for Wi=1Wi = 1, Wi=5Wi = 5 and Wi=10Wi = 10 at Re=10Re = 10. Once the flow reaches a statistically steady state, the velocity signal is collected from the point away from the centre (x∗=0.5,y∗=6,z∗=1)(x^* = 0.5, y^* = 6, z^* = 1) to avoid the noise due to the bubble cluster. The flow velocity inside the bubbles

is filtered out from the signal. A slope of −2-2 is observed for the Wi=5Wi = 5 & 10 cases when the flow is chaotic in the entire domain. This slope is the same as the temporal spectrum scaling of centreline velocity observed by Lellep et al. [26] indicating a similar chaotic state. The energy spectra of Wi=5Wi = 5 and Wi=10Wi = 10 cases overlap each other, showing a similar flow state for both the cases in the presence of bubbles. As the flow is not fully chaotic at Wi=1Wi = 1 and its signal shows a very high intermittency, the same is manifested in its spectrum where the slope does not match with the slopes of Wi=5Wi = 5 & 10 cases. For the Re=100Re = 100 cases, the slope for Wi=1Wi = 1 gets closer to the slopes in the Wi=5Wi = 5 & 10 cases and finally at Re=1000Re = 1000, all the three cases show a similar slope of −2-2 (Figure 9d). This slope of −2-2 is greater than the classical slope of −5/3-5/3 in the inertial turbulence but still less than −4-4 reported for a purely elastic turbulence in the absence of inertia (Datta et al. [8]) or at a very high value of Wi=1000Wi = 1000 observed in the single-phase cases (Figure 4a,Figure 4c). As this multiphase EIT regime is achieved at the lower values of WiWi, it is governed by both inertia and viscoelasticity, and hence the slope remains in between these two limits.

The distribution of different components of shear stress from the channel wall towards the centre are shown in Figure 9e, Figure 9f & Figure 9g in the mid-plane of the duct for Re=10Re = 10, 100 & 1000 cases, respectively. As the flow is multiphase, an additional contribution from the surface tension force of the bubbles (τˉS\bar{\tau}_{S}) is added towards the total shear stress balance. However, this contribution remains much smaller than the remaining components as the bubble volume fraction is a mere 3%. Similar to the single-phase flow cases, these stress components are normalized by the local shear stress at the wall of the mid-plane. As seen in the figure, the chaotic flow regime is dominated by the viscoelastic stress (τˉP\bar{\tau}_{P}) followed by the viscous stress due to the mean flow (τˉN\bar{\tau}_{N}). These two stress components are highest at the wall and decay to zero towards the channel centre where the shear rate is minimum. The contribution of the Reynolds stress (τˉR\bar{\tau}_{R}) remains negligible for the Re=10Re = 10 and Re=100Re = 100 cases. The Reynolds stress component τˉR\bar{\tau}_{R} starts to show its effect once the inertia becomes high enough in the Re=1000Re = 1000 cases. As the source of flow instability is the curved streamlines around the bubble in this multiphase flow regime (McKinley et al. [28]), the transition to the EIT state greatly depends upon the distribution of bubbles across the channel. The role of viscoelasticity in promoting the bubble migration towards the channel centre or towards the wall is governed by the relative change in the convective time scale of the bubbles and the polymer relaxation time as has been explained by Bothe et al. [5]. A similar observation was also made in our earlier work as well (Naseer et al. [33]). The distribution of turbulent kinetic energy (TKE) for all these three values of ReRe shows a peak value near the wall and minimum at the channel centre, as shown in Figure 9h, Figure 9i & Figure 9j. By increasing the value of WiWi, the magnitude of TKE also increases following the same qualitative trend in agreement to our previous study of viscoelastic turbulent single phase case (Izbassarov et al. (2021b)).

Once the flow reaches a steady state, the applied pressure gradient is balanced by the total shear stress at the walls and the average wall shear stress can be simply computed as τˉw=−h2dpody\bar{\tau}_{w} = -\frac{h}{2}\frac{d p_{o}}{d y}. The change in the drag is quantified by ΔD=(τˉw−τw0)/τw0\Delta D = (\bar{\tau}_{w} - \tau_{w0})/\tau_{w0}, where τw0\tau_{w0} is the wall shear stress of the corresponding Newtonian laminar flow for the same Reynolds number. Figure 10 summarizes the percentage change in drag for all the multiphase simulations. The single-phase viscoelastic laminar flow has lower drag as compared to the Newtonian laminar flow for the same value of ReRe. Understandably, as the inertia increases in this shear thinning viscoelastic fluid once the flow is made viscoelastic, the drag is reduced. For a particular value of WiWi, the percentage reduction in drag is found to be the same for every value of ReRe. However, the drag increases significantly once a viscoelastic flow transitions to a chaotic turbulent state after injecting the bubbles (Fig. 10). In particular, a rapid increase in the drag is observed at Re=1000Re = 1000 once the flow becomes turbulent for all three values of WiWi. It is interesting to observe that, although the drag increases for the Re=10Re = 10 & 100100 cases at Wi=5Wi = 5 & 1010 upon transition to the EIT state, it still remains less than that of the respective Newtonian laminar state. Another interesting feature is also observed at Wi=1Wi = 1 for the Re=10Re = 10 & 100100 cases. Once bubbles are injected into the flow, minor fluctuations remain confined to the core region away from the walls where the bubbles collect and form a string-shaped pattern (Re=10Re = 10) or small clusters (Re=100Re = 100). The flow remains essentially laminar closer to walls and in the major portion of the duct. In this intermittent regime, the drag is reduced as compared to the single-phase laminar state. At a higher inertia (Re=1000Re = 1000), the flow becomes fully chaotic in the entire channel, and thus the drag is increased significantly even for the Wi=1Wi = 1 case (Fig. 10). Furthermore, it is interesting to note that for (Re,Wi)=(10,10)(Re, Wi) = (10, 10) and (100,10)(100, 10) cases, the flow has become completely chaotic and the EIT state is achieved by injecting the bubbles into the flow, however, the drag is still less than the corresponding Newtonian laminar state, which is primarily attributed to the shear-thinning effect of viscoelastic fluid (Fig. 10).

Figure 10. The percentage change in drag measured by the change in total shear stress at the wall (τw\tau_w) as compared to the Newtonian laminar flow once the flow is made viscoelastic and is subsequently transitioned to turbulent state by injecting the bubbles

TKE in a multiphase EIT regime

To get further insight into the multiphase EIT regime, all the terms in Eq. (6) are integrated in the streamwise direction and averaged in time after the flow reaches a statistically steady state. The results are plotted in Figure 11 for Re=100Re = 100 and Re=1000Re = 1000 at Wi=10Wi = 10. Compared to the single phase case shown in Figure 5, a prominent effect of bubbles is observed in smoothing sharp gradients near the wall in all the terms that contribute to TKE. We note, however, that comparison between single and multiphase cases should be interpreted only qualitatively since WiWi is two orders of magnitude higher in the single phase case. Figure 11 shows that the magnitudes of all terms are amplified as ReRe is increased from 100 to 1000 but the amplification is more pronounced in advection (A\mathcal{A}), turbulent transport (Q\mathcal{Q}), production by mean flow (P\mathcal{P}) and viscous diffusion (D\mathcal{D}) terms. There are also some qualitative differences in these terms in the Re=100Re = 100 and Re=1000Re = 1000 cases. The peak values of A\mathcal{A}, Q\mathcal{Q}, P\mathcal{P} and ϵ\epsilon occur in the corners and in the middle of the side walls for Re=100Re = 100 and Re=1000Re = 1000, respectively. In contrast to the single phase case, the maximum value of R\mathcal{R} occurs in the central portion of the duct due to presence of bubbles (Fig. 11c). Significant viscous dissipation observed in the channel centre is also attributed to the presence of bubbles there. The negative peaks of polymer work for the multiphase case shows that it still contributes towards the production instead of dissipation in some portions of the domain (Figure 11g). The body force term contributes significantly to the TKE budget now due to the work done by the surface tension (Figure 11h).

Contours of transport and budget terms for multiphase cases

Figure 11. Contours of (a) advection by mean flow, (b) transport by velocity fluctuations, (c) transport by pressure, (d) production by mean flow, (e) viscous diffusion, (f) viscous dissipation, (g) polymer work and (h) body force terms are shown in a vertical cutting xzxz-plane for (left) Re=100Re = 100, Wi=10Wi = 10 and (right) Re=1000Re = 1000, Wi=10Wi = 10 multiphase cases, respectively.

Conclusions

Direct numerical simulations of single and multiphase viscoelastic fluid flows are performed in a square-shaped channel and the complex dynamics of elasto-inertial turbulent regime are explored without introducing any explicit perturbations from outside the system. The bubble interfaces are fully resolved using a finite-difference/front-tracking method coupled with the Giesekus model equations to simulate chaotic flow regime of a shear-thinning viscoelastic fluid. Instead of using conventional numerical schemes to discretize the constitutive equations by introducing artificial diffusion, a computationally expensive but robust log-conformation method is employed for the present DNS to preserve the integrity of EIT physics (Yerasi et al. 2024). The Reynolds number is varied by two orders of magnitude to cover a vast range below the threshold value of the inertial turbulence in the channel flow. Then the Weissenberg number is varied by three orders of magnitude at each of these values of Reynolds numbers to explore the conditions for triggering transition to a turbulent regime in this rectilinear channel flow. Subsequently, unlike any previously tested methods, bubbles are injected into the flow under the conditions for which a single-phase flow remains fully laminar to check whether the presence of bubbles can trigger an instability to achieve the EIT flow regime. A constant flow rate is maintained corresponding to each ReRe by dynamically adjusting the applied pressure gradient. As different pathways can be followed in ReRe-WiWi space to achieve EIT (Samanta et al. [36]), a concentrated polymer solution (β=0.1\beta= 0.1) is used in the present study for which the EIT regime is triggered much earlier than the inertial turbulence as the Reynolds number is increased.

It is observed that beyond Wi≥100Wi \ge100, a single-phase flow becomes unstable when Re≥100Re \ge100. Instabilities start to grow in this straight channel even without introducing any discrete external perturbations. These instabilities grow with the passage of time, and ultimately a chaotic flow regime is observed in the entire domain. The threshold value of WiWi to trigger instability in the flow reduces with an increase in ReRe. For Re=1000Re = 1000, the critical value of WiWi to trigger transition from laminar to turbulent state is found to be 50. When the value of ReRe is low (Re=10Re = 10), a Weissenberg number as high as 1000 can not make the flow unstable. The energy spectra of this single-phase EIT regime reveal −4-4 scaling at the higher frequencies (dissipation range) while the classical −5/3-5/3 scaling is observed at the lower frequencies (inertial range) as the flow is dominated by both inertia as well as the elasticity.

A novel mechanism of achieving the EIT state by injecting the bubbles into the flow under the conditions for which a single-phase flow remains laminar is tested and verified at a Reynolds number as low as 10 and a Weissenberg number as low as 5. The curved streamlines across the bubbles interfaces provide the necessary condition to trigger an instability even when the viscoelastic stresses are not very high and the single-phase flow remains completely laminar. The energy spectra shows a scaling of −2-2 for this low WiWi multiphase EIT regime. Thus, it is concluded that the scaling of energy spectra varies from −4-4 to −5/3-5/3 depending upon the relative dominance of inertia and elasticity, i.e, −4-4 for a purely elastic turbulence and −5/3-5/3 for a purely inertial turbulence. All the other values of energy scaling would fall in between these two limits depending upon the relative strength of inertia (ReRe) and elasticity (WiWi). It is also observed that the drag increases for all the situations where the laminar viscoelastic flow is fully transitioned to a turbulent state by injecting the bubbles into the flow. However, at the lower values of Re=10Re = 10 & 100, this higher value of skin friction drag of the EIT regime still remains lower than that of the corresponding laminar Newtonian flow for the same Reynolds number. Once the inertia and the elasticity are both low, the bubbles are aligned in the core region of the duct forming a string-shaped pattern (Re=10Re = 10, Wi=1Wi = 1) or clusters (Re=100Re = 100, Wi=1Wi = 1), and the flow remains laminar in the wall region. For these cases, the friction drag reduces even further instead of increasing in this highly intermittent flow. Unlike the solid particles, it is found that a higher shear-thinning effect breaks up the alignment of bubbles.

The idea of achieving the EIT state at a very low value of ReRe and WiWi just by injecting the bubbles into the flow can find many potential applications involving processes where mixing, heat transfer or other transport phenomena are of primary importance. The present study would intrigue experimental verification as it has been demonstrated that the requirement of WiWi to achieve the EIT state in a multiphase flow is low and realistic.

Data Availability

Data can be shared upon request.

Acknowledgment

We acknowledge financial support from the Scientific and Technical Research Council of Türkiye (TUBITAK) [Grant Number 124M335] and the Research Council of Finland [Grant Number 354620].

Declaration of Interest

The authors declare no conflict of interest.

Appendix

Further simulations are performed to examine the effect of density and viscosity ratios on the multiphase flows considered in this study. For this purpose, the migration of a single bubble is examined in the pressure-driven viscoelastic channel flow, and the density and viscosity ratios are varied in the ranges of 10≤ρo/ρi≤4010 \le\rho_o/\rho_i \le40 and 10≤μo/μi≤16010 \le\mu_o/\mu_i \le160. The other parameters are fixed at Re=10Re = 10, Wi=1Wi = 1, Ca=0.01Ca = 0.01 and β=0.1\beta= 0.1. The results are shown in Figure 12. It is observed that the effects of density and viscosity ratios are negligible when ρo/ρi≥10\rho_o/\rho_i \ge10 and μo/μi≥80\mu_o/\mu_i \ge80, as seen in Figure 12a and Figure 12b, respectively. Therefore, the density and viscosity ratios are set to ρo/ρi=10\rho_o/\rho_i = 10 and μo/μi=80\mu_o/\mu_i = 80 in all the results presented in this paper. In a concentrated polymer solution (e.g., β=0.1\beta= 0.1) as considered in the present study, the viscosity ratio may have an impact on the bubble dynamics as the actual ratio can be as high as 10810^8. However, it is shown here that beyond μo/μi≥80\mu_o/\mu_i \ge80, the difference in the bubble path becomes less than 1% (Figure 12).

Figure 12. Effects of density and viscosity ratios. Evolution of a single bubble path in the wall-normal (zz) direction in a concentrated polymer solution at (aa) different density and at (bb) different viscosity ratios (Re=10Re = 10, Wi=1Wi = 1, Ca=0.01Ca = 0.01, β=0.1\beta= 0.1).

References

  1. [1]Ahmed, Zaheer, Izbassarov, Daulet, Lu, Jicai, Tryggvason, Gretar, Muradoglu, Metin & Tammisola, Outi 2020 Effects of soluble surfactant on lateral migration of a bubble in a pressure driven channel flow. International Journal of Multiphase Flow 126, 103251.DOI
  2. [2]Beneitez, Miguel, Page, Jacob, Dubief, Yves & Kerswell, Rich R. 2024 Multistability of elasto-inertial two-dimensional channel flow. Journal of Fluid Mechanics 981, A30.arxiv.org/abs/2308.11554
  3. [3]Beneitez, Miguel, Page, Jacob & Kerswell, Rich R. 2023 Polymer diffusive instability leading to elastic turbulence in plane couette flow. Physical Review Fluids 8, L101901.DOI
  4. [4]Borges, Rafael, Carmona, Monique, Costa, Bruno & Don, Wai Sun 2008 An improved weighted essentially non-oscillatory scheme for hyperbolic conservation laws. Journal of Computational Physics 227, 3191–3211.
  5. [5]Bothe, Dieter, Niethammer, Matthias, Pilz, Christian & Brenn, Günter 2022 On the molecular mechanism behind the bubble rise velocity jump discontinuity in viscoelastic liquids. Journal of Non-Newtonian Fluid Mechanics 302, 104748.DOI
  6. [6]Bunner, Bernard & Tryggvason, Grétar 2002 Dynamics of homogeneous bubbly flows part 1. rise velocity and microstructure of the bubbles. Journal of Fluid Mechanics 466, 17–52.DOI
  7. [7]Dallas, Vassilios, Vassilicos, J Christos & Hewitt, Geoffrey F 2010 Strong polymer-turbulence interactions in viscoelastic turbulent channel flow. Physical Review E 82 (6), 066303.DOI
  8. [8]Datta, Sujit S, Ardekani, Arezoo M, Arratia, Paulo E, Beris, Antony N, Bischofberger, Irmgard, McKinley, Gareth H, Eggers, Jens G, López-Aguilar, J Esteban, Fielding, Suzanne M, Frishman, Anna & Others 2022 Perspectives on viscoelastic flow instabilities and elastic turbulence. Physical Review Fluids 7 (8), 080701.arxiv.org/abs/2108.09841
  9. [9]Dodd, Michael S & Ferrante, Antonino 2014 A fast pressure-correction method for incompressible two-fluid flows. Journal of Computational Physics 273, 416–434.DOI
  10. [10]Dong, Suchuan & Shen, Jie 2012 A time-stepping scheme involving constant coefficient matrices for phase-field simulations of two-phase incompressible flows with large density ratios. Journal of Computational Physics 231 (17), 5788–5804.DOI
  11. [11]Dubief, Yves, Page, Jacob, Kerswell, Rich R, Terrapon, Vincent E & Steinberg, Victor 2022 First coherent structure in elasto-inertial turbulence. Physical Review Fluids 7 (7), 073301.DOI
  12. [12]Dubief, Yves, Terrapon, Vincent E & Hof, Björn 2023 Elasto-inertial turbulence. Annual Review of Fluid Mechanics 55, 675–705.
  13. [13]Dubief, Yves, Terrapon, Vincent E & Soria, Julio 2013 On the mechanism of elasto-inertial turbulence. Physics of Fluids 25 (11).DOI
  14. [14]Dubief, Yves, White, Christopher M, Shaqfeh, Eric SG & Terrapon, Vincent E 2011 Polymer maximum drag reduction: A unique transitional state. arXiv preprint arXiv:1106.4482.arxiv.org/abs/1106.4482
  15. [15]Fattal, Raanan & Kupferman, Raz 2005 Time-dependent simulation of viscoelastic flows at high weisenberg number using the log-conformation representation. Journal of Non-Newtonian Fluid Mechanics 126 (1), 23–37.DOI
  16. [16]Foggi Rota, Giulio, Amor, Christian, Le Clainche, Soledad & Rosti, Marco Edoardo 2024 Unified view of elastic and elasto-inertial turbulence in channel flows at low and moderate reynolds numbers. Physical Review Fluids 9 (12), L122602.
  17. [17]Garg, Piyush, Chaudhary, Indresh, Khalid, Mohammad, Mariani, Shankar, & Subramanian, Ganesh 2018 Viscoelastic pipe flow is linearly unstable. Physical Review Letters 121 (2), 024502.DOI
  18. [18]Giesekus, Hanswalter 1982 A simple constitutive equation for polymer fluids based on the concept of deformation-dependent tensorial mobility. Journal of Non-Newtonian Fluid Mechanics 11 (1-2), 69–109.DOI
  19. [19]Groisman, Alexander & Steinberg, Victor 2000 Elastic turbulence in a polymer solution flow. Nature 405 (6782), 53–55.pubmed.ncbi.nlm.nih.gov/10811214
  20. [20]Izbassarov, Daulet, Ahmed, Zaheer, Costa, Pedro, Vuorinen, Ville, Tammisola, Outi & Muradoglu, Metin 2021a Polymer drag reduction in surfactant-contaminated turbulent bubbly channel flows. Physical Review Fluids 6 (10), 104302.DOI
  21. [21]Izbassarov, Daulet & Muradoglu, Metin 2015 A front-tracking method for computational modeling of viscoelastic two-phase flow systems. Journal of Non-Newtonian Fluid Mechanics 223, 122–140.DOI
  22. [22]Izbassarov, Daulet & Muradoglu, Metin 2016a A computational study of two-phase viscoelastic systems in a capillary tube with a sudden contraction/expansion. Physics of Fluids 28 (1).
  23. [23]Izbassarov, Daulet & Muradoglu, Metin 2016b Effects of viscoelasticity on drop impact and spreading on a solid surface. Physical Review Fluids 1 (2), 023302.DOI
  24. [24]Izbassarov, Daulet, Rosti, Marco E, Ardekani, M Niazi, Sarabian, Mohammad, Hormozi, Sarah, Brandt, Luca & Tammisola, Outi 2018 Computational modeling of multiphase viscoelastic and elastoviscoplastic flows. International Journal for Numerical Methods in Fluids 88 (12), 521–543.DOI
  25. [26]Lellep, Martin, Linkmann, Moritz & Morozov, Alexander 2024 Purely elastic turbulence in pressure-driven channel flows. Proceedings of the National Academy of Sciences 121 (9), e231885121.DOI
  26. [28]McKinley, Gareth H, Pakdel, Peyman & Öztekin, Alparslan 1996 Rheological and geometric scaling of purely elastic flow instabilities. Journal of Non-Newtonian Fluid Mechanics 67, 19–47.DOI
  27. [29]Morozov, Alexander & Van Saarloos, Wim 2019 Subcritical instabilities in plane poiseuille flow of an oldroyd-b fluid. Journal of Statistical Physics 175, 554–577.
  28. [30]Morozov, Alexander N & van Saarloos, Wim 2005 Subcritical finite-amplitude solutions for plane couette flow of viscoelastic fluids. Physical Review Letters 95 (2), 024501.pubmed.ncbi.nlm.nih.gov/16090685
  29. [31]Mukherjee, Siddhartha, Singh, Rahul K, James, Martin & Ray, Samriddhi Sankar 2023 Intermittency, fluctuations and maximal chaos in an emergent universal state of active turbulence. Nature Physics 19 (6), 891–897.DOI
  30. [32]Naseer, Hafiz Usman, Ahmed, Zaheer, Izbassarov, Daulet & Muradoglu, Metin 2023 Dynamics and interactions of parallel bubbles rising in a viscoelastic fluid under buoyancy. Journal of Non-Newtonian Fluid Mechanics 313, 105000.
  31. [33]Naseer, Hafiz Usman, Izbassarov, Daulet, Ahmed, Zaheer & Muradoglu, Metin 2024 Lateral migration of a deformable fluid particle in a square channel flow of viscoelastic fluid. Journal of Fluid Mechanics 996, A31.DOI
  32. [34]Owolabi, Bayode E, Poole, Robert J & Dennis, David JC 2016 Experiments on low-reynolds-number turbulent flow through a square duct. Journal of Fluid Mechanics 798, 398–410.
  33. [35]Ptasinski, PK, Nieuwstadt, FTM, Van Den Brule, BHAA & Hulsen, MA 2001 Experiments in turbulent pipe flow with polymer additives at maximum drag reduction. Flow, Turbulence and Combustion 66, 159–182.DOI
  34. [36]Samanta, Devranjan, Dubief, Yves, Holzner, Markus, Schäfer, Christof, Morozov, Alexander N, Wagner, Christian & Hof, Björn 2013 Elasto-inertial turbulence. Proceedings of the National Academy of Sciences 110 (26), 10557–10562.DOI
  35. [37]Schleininger, Gilberto & Weinacht, RJ 1991 A remark on the giesekus viscoelastic fluid. Journal of Rheology 35 (6), 1157–1170.
  36. [38]Serafini, F, Battista, F, Gualtieri, P & Casciola, CM 2022 Drag reduction in turbulent wall-bounded flows of realistic polymer solutions. Physical Review Letters 129 (10), 104502.DOI
  37. [39]Shahmardi, Armin, Zade, Sagar, Ardekani, Mehdi N, Poole, Rob J, Lundell, Fredrik, Rosti, Marco E & Brandt, Luca 2019 Turbulent duct flow with polymers. Journal of Fluid Mechanics 859, 1057–1083.DOI
  38. [40]Sid, Samir, Terrapon, VE & Dubief, Y 2018 Two-dimensional dynamics of elasto-inertial turbulence and its role in polymer drag reduction. Physical Review Fluids 3 (1), 011301.DOI
  39. [41]Singh, Rahul K, Perlekar, Prasad, Mitra, Dhruvabidtya & Rosti, Marco Edoardo 2024 Intermittency in the not-so-smooth elastic turbulence. Nature Communications 15 (1), 4070.DOI
  40. [42]Steinberg, Victor 2021 Elastic turbulence: An experimental view on inertialess random flow. Annual Review of Fluid Mechanics 53, 27–58.DOI
  41. [43]Tryggvason, Grétar, Bunner, Bernard, Esmaeeli, Asghar, Juric, Damir, Al-Rawahi, N, Tauber, W, Han, J, Nas, S & Jan, Y-J 2001 A front-tracking method for the computations of multiphase flow. Journal of Computational Physics 169 (2), 708–759.DOI
  42. [44]Tryggvason, Grétar, Scardovelli, Ruben & Zaleski, Stéphane 2011 Direct numerical simulations of gas–liquid multiphase flows. Cambridge University Press.
  43. [45]Unverdi, Salih Ozen & Tryggvason, Gretar 1992 A front-tracking method for viscous, incompressible, multi-fluid flows. Journal of computational physics 100 (1), 25–37.
  44. [46]Varchanis, Stylianos, Haward, Simon J, Hopkins, Cameron C, Tsapopoulos, John & Shen, Amy Q 2022 Evaluation of constitutive models for shear-banding wormlike micellar solutions in simple and complex flows. Journal of Non-Newtonian Fluid Mechanics 307, 104855.DOI
  45. [47]Won, Daejin & Kim, Chongyoup 2004 Alignment and aggregation of spherical particles in viscoelastic fluid under shear flow. Journal of Non-Newtonian Fluid Mechanics 117 (2-3), 141–146.DOI
  46. [48]Yoo, Jy & Choi, H Ch 1989 On the steady simple shear flows of the one-mode giesekus fluid. Rheologica Acta 28, 13–24.DOI

Paper details

vlnc:W_c46knedv
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents