betaFeedback

WorksPhysics and Astronomy

From laminar to chaotic flow via stochastic resonance in viscoelastic channel flow

Abstract

Recent research indicates that low-inertia viscoelastic channel flow experiences supercritical non-normal mode elastic instability from laminar to sustained chaotic flow due to finite-size perturbations. The challenge of this study is to elucidate a realization of such a pathway when the intensity of the elastic wave is too low to amplify velocity fluctuations above the instability onset. The study identifies two subregions in the transition flow regime at Weissenberg number Wi>WicWi>Wi_c, the instability onset. In the lower subregion at Wic≤Wi≤300Wi_c\leq Wi\leq 300, we discover periodic spikes in the streamwise velocity time series u(t)u(t) that appear in the chaotic power spectrum as low-frequency, high-intensity peaks resembling stochastic resonance (SR). In contrast, the spanwise velocity power spectrum, EwE_w, remains flat with low-intensity, noisy, and broad elastic wave peaks. The spikes significantly distort the probability density function of uu, initiating and amplifying random streaks and wall-normal vorticity fluctuations. The SR appearance is similar to dynamical systems where chaotic attractor and limit cycle interact with external white noise. This similarity is confirmed by presenting a phase portrait in two subregions of the transition regime. In the upper subregion at Wi>400Wi>400 the periodic spikes disappear and EwE_w becomes chaotic with a large intensity elastic wave sufficient to self-organize and synchronize the streaks into cycles and to amplify the wall normal vorticity according to a recently proposed mechanism.

Introduction

The transition from laminar to turbulent flow in Newtonian or to chaotic flow in viscoelastic fluids follows two distinct pathways. One pathway, common to both fluids, is attributed to a linear normal mode instability followed by a sequence of linear instabilities leading to turbulent or chaotic flow [1]. Linear stability analysis has been successfully applied and validated for various flow geometries with curved streamlines [1, 2, 3, 4]. It is important to note that this approach is only valid for Hermitian equations (or operators) characterized by normal eigenmodes, which are prone to linear instability. The onset of instability is determined by the critical eigenvalue of the most unstable normal eigenmode, which grows exponentially. Furthermore, nonlinear effects stabilize the critical eigenmode at a sufficiently large amplitude, thereby replacing a laminar flow that is stable at Wi<WicWi < Wi_c.

In the second scenario, which occurs in both Newtonian and viscoelastic parallel shear flows, turbulent or chaotic flows are induced by finite-sized external perturbations just above the instability threshold and persist under steady conditions. This type of instability, known as non-normal mode instability, occurs in a flow described by non-Hermitian equations characterized by both normal and non-normal modes, as first discussed for Newtonian parallel shear flows [5, 6, 7]. The study of the linear instability of Newtonian parallel shear flows uses a linearized version of the Navier–Stokes equation, called the Orr–Somm­erfeld equation, a classic example of non-Hermitian equations. Historically, this equation has been used extensively in its Hermitian approximation at Reynolds number Re≫1Re \gg1, leading to its linear stability at all ReRe [1]. More recently, its non-Hermitian form has been used to study the instability of Newtonian flows. This approach shows that while stable normal modes decay exponentially, unstable non-normal modes grow algebraically and can be amplified by several orders of magnitude over time [5, 6, 7].

A linear normal mode instability is driven by a deterministic instability mechanism based on the competition between a destabilizing force and a stabilizing factor, such as dissipation or external fields, especially when the former predominates [1, 8]. Specifically, in polymer solution flows with curved streamlines, linear instability arises due to “hoop stress”, an elastic stress generated by polymers stretched along curved streamlines by the velocity gradient. The hoop stress generates a destabilizing bulk force directed toward the center of curvature, which competes with relaxation dissipation of the elastic stress. The onset of linear elastic instability, denoted by the Weissenberg critical number WicWi_c, is determined by its mismatch [2, 3, 4]. Here, Wi≡λU/dWi \equiv\lambda U/d is the main control parameter in inertia-less, viscoelastic fluid flow and denotes the ratio of elastic stress, generated by polymer stretching due to its entropic elasticity to stress dissipation due to its relaxation, where UU is the mean flow velocity, dd is the characteristic vessel size, and λ\lambda is the longest polymer relaxation time. In the present experiment, we consider Wi≫1Wi \gg1 and the Reynolds number Re≡Udρ/η≪1Re \equiv Ud\rho/\eta\ll1, which corresponds to an extremely large elasticity number El=Wi/Re=λη/ρd2≫1El = Wi/Re = \lambda\eta/\rho d^2 \gg1, where ρ\rho and η\eta are the solution density and dynamic viscosity, respectively (see Materials and Methods for details).

However, the instability mechanism becomes ineffective at a zero-curvature limit [2, 9]. Therefore, viscoelastic parallel shear flows have been shown to be linearly stable over all WiWi and for Re≪1Re \ll1 [10, 11], which is analogous to Newtonian parallel shear flows at all ReRe. For the latter, however, the proven linear stability fails to explain Reynolds’ seminal experiments, where a transition from laminar to turbulent flow was observed in Newtonian pipe flow at finite ReRe [1, 6, 12]. Thus, linear stability does not necessarily imply global stability in either Newtonian [1, 6] or viscoelastic [12, 13, 14, 15, 16, 17] parallel shear flows. Similar to Newtonian flows, inertia-less viscoelastic parallel shear flows exhibit non-normal mode instability at Wi≫1Wi \gg1 and Re≪1Re \ll1 due to the non-Hermitian nature of the linearized elastic stress equation [12, 13, 14, 15, 16, 17]. These papers focus on theoretical investigations and numerical analyses of the non-normal unstable modes in viscoelastic channel and plane Couette flows only during the linear transient growth at Wi≫1Wi \gg1 and Re≪1Re \ll1. During the transient growth, the emergence of coherent structures (CSs), such as streamwise streaks and rolls, is observed. However, the linear growth model fails to predict the presence of nonlinearly stabilized sustained CSs.

The first experiments on viscoelastic parallel shear flows are carried out in pipes [18, 19, 20], then extended to straight square microchannels with strong pre-arranged perturbations at the inlet [21, 22, 23]. These experiments report the finding of significant velocity fluctuations above WicWi_c, which contradicts the theoretically proven linear stability of such viscoelastic flows [10, 11]. Subsequent experiments performed in our laboratory on viscoelastic straight channel flows at Wi≫1Wi \gg1 and Re≪1Re \ll1 with different external perturbation intensities [24, 25, 26, 27, 28, 29], find a supercritical instability at Wi≥WicWi \ge Wi_c leading directly to a sustained chaotic flow. This transition is characterized by the dependence of the friction factor, the normalized root mean square (rms) streamwise velocity and pressure fluctuations on Wi−WicWi-Wi_c, with scaling exponents that deviate from the 0.5 typical for normal mode supercritical bifurcation [1, 8]. Moreover, WicWi_c depends on the amplitude and structure of the external perturbations, in contrast to the normal mode bifurcation [1]. These two key features - the direct transition from laminar to chaotic flow and the dependence of WicWi_c on the intensity and spectral characteristics of the external perturbations - validate the non-modal elastic instability [12, 14, 30]. At Wi>WicWi > Wi_c, regardless of the strength and structure of the external perturbations, three chaotic flow regimes emerge: transition, elastic turbulence (ET), and drag reduction (DR), each accompanied by elastic waves just above WicWi_c. The experimental results show a universal scaling of the flow properties, a dependence of the elastic wave propagation velocity on Wi−WicWi-Wi_c, a power-law decay of the velocity power spectra, and a predominant presence of streaks [24, 25, 26, 27, 28, 29].

The existence of elastic waves in three flow regimes is the characteristic feature of viscoelastic chaotic channel flow. The waves are predicted to appear in the turbulent flow of viscoelastic fluid at Re≫1Re \gg1 and Wi≈1Wi \approx1, and are expected to be suppressed by large viscous dissipation at Re≪1Re \ll1 and Wi≫1Wi \gg1 [31]. Contrary to these predictions, the elastic waves were first discovered in the inertia-less viscoelastic flow at Re≪1Re \ll1 and Wi≫1Wi \gg1 in the ET flow regime. This discovery is characterized by distinct peaks in the spanwise velocity power spectra, with frequency and intensity varying with WiWi, and in particular by their propagation velocity in the direction of the streamwise velocity, which is characterized by a scaling law as a function of Wi−WicWi-Wi_c, as reported in Ref. [32]. Subsequent research has confirmed this phenomenon and established its universality in channel flows with different external noise intensities [4, 24, 25, 26, 27, 28, 29]. A simple physical explanation for the appearance of elastic waves is based on the analogy of the response of an elastic stress field to transverse perturbations, similar to the response of an elastic string when plucked [4].

Here, the experimental study focuses on the immediate vicinity above the onset of the supercritical elastic instability of inertia-less viscoelastic planar channel flow subjected to finite-size perturbations induced by the unsmoothed inlet and two small cavities in the upper wall at both channel ends. For Wi≥WicWi \ge Wi_c, the instability leads to a transition from laminar to sustained chaotic flow. Recent experimental evidence reveals the key role of elastic waves in amplifying wall normal vorticity fluctuations through resonant pumping of elastic wave energy withdrawn from the mean shear flow into the fluctuating wall normal vorticity [28]. The higher(lower) the elastic wave intensity, the higher(lower) the flow resistance and the rotational vorticity fluctuations [28]. This mechanism also helps to explain the transition from ET to DR first described in Ref. [33] and explained in Ref. [34]. Given the key role of elastic waves in promoting sustained chaotic flow, the main goal of the current experiment is to uncover the pathway to chaotic flow, in particular when the energy of the elastic wave is insufficient to initiate and amplify streaks and wall normal vorticity fluctuations above WicWi_c.

Results

We performed the experiment in a long channel of dimensions 500 (LL, xx: streamwise) ×\times 0.5 (hh, yy: wall-normal) ×\times 3.5 (WW, zz: spanwise) mm3^3, shown in Figure 1. The channel has an unsmoothed inlet and two small holes of 0.5 mm diameter in the top plate near the inlet and outlet, used for pressure measurements, which introduce finite-size perturbations, much weaker than strong pre-arranged perturbations used in Refs. [24, 25] and slightly weaker than in the channel discussed in Ref. [29]. The latter is determined by the rms of streamwise velocity fluctuations, urmsu_{rms}, at the inlet. A dilute aqueous solution with 64% sucrose concentration and 80 ppm high molecular weight polyacrylamide (PAA) is the same as used in Refs. [28, 29]. Its longest relaxation time (λ\lambda) is 13 s. In the Materials and Methods section, we discuss details of the experimental setup, preparation and characterization of the polymer solution, pressure (pp) fluctuations, and Particle Image Velocimetry (PIV) measurement techniques.

Schematic of the experimental setup

Figure 1. Schematic of the experimental setup. Flow in the long straight channel with the non-smoothed inlet and two small cavities in the top plate close to the inlet and outlet, used for absolute pressure measurement, is driven by compressed Nitrogen gas (from left to right in the schematic). Pressure, velocity, and flow discharge are measured by absolute pressure sensors, a high-speed and high spatial resolution camera with laser sheet illumination, and a balance interfaced with a computer, correspondingly.

Characterization of the elastic transition at Wi≥WiicWi \ge Wi_{ic}

The onset of the elastic instability at Wiic=150±10Wi_{ic} = 150 \pm10 and the characterization of the channel flow at Wi≥WiicWi \ge Wi_{ic} in three subsequent chaotic flow regimes, namely transition, ET, and DR, are described in Ref. [28]. These three flow regimes are well characterized by the dependence of the friction factor, normalized pressure, and streamwise velocity fluctuations on Wi/Wiic−1Wi/Wi_{ic} - 1 at the downstream location l/h=380l/h = 380, as reported in the same reference [28]. The corresponding exponents are found to be universal for the studied viscoelastic channel flows with different external perturbations [24, 26, 29]. Examining urmsu_{\mathrm{rms}} and rms pressure fluctuations, prmsp_{\mathrm{rms}}, versus WiWi below and above WiicWi_{ic}, one finds their continuous growth without any clear indication of the instability onset as shown in Figure 2(a) and Figure 2(c). Algebraic fits of the normalized streamwise velocity fluctuations, urms/urms,lamu_{\mathrm{rms}}/u_{\mathrm{rms,lam}}, and pressure fluctuations, prms/prms,lamp_{\mathrm{rms}}/p_{\mathrm{rms,lam}}, versus Wi/Wiic−1Wi/Wi_{ic} - 1 up to Wi=280Wi = 280 yield exponent values of 0.85 and 1.10, respectively. These values are consistent with those obtained in the previous studies, where the exponents were obtained for data taken in the entire transition regime as shown in Figure 2(b) and Figure 2(d) [24, 26, 27, 29]. Here urms,lamu_{\mathrm{rms,lam}} and prms,lamp_{\mathrm{rms,lam}} are rms velocity and pressure fluctuations in laminar flow, respectively.

Characterization of elastic instability and transition regime as a function of $Wi$

Figure 2. Characterization of elastic instability and transition regime as a function of WiWi in a limited range above WicWi_c. (a) rms stream-wise velocity fluctuations, urmsu_{\mathrm{rms}}, (b) normalized rms stream-wise velocity fluctuations, urms/urms,lamu_{\mathrm{rms}}/u_{\mathrm{rms,lam}}, (c) rms pressure fluctuations, prmsp_{\mathrm{rms}}, (d) normalized rms pressure fluctuations, prms/prms,lamp_{\mathrm{rms}}/p_{\mathrm{rms,lam}}, versus WiWi in linear coordinates. The black dashed line is the power-law fit of urms/urms,lamu_{\mathrm{rms}}/u_{\mathrm{rms,lam}} and prms/prms,lamp_{\mathrm{rms}}/p_{\mathrm{rms,lam}} versus (Wi/Wic−1)ϵ+1(Wi/Wi_c-1)^{\epsilon+1} with ϵ=0.85\epsilon=0.85 and 1.10, respectively, above and in close vicinity to WicWi_c.

Temporal evolution of stream-wise velocity fluctuations and low-frequency periodic spikes

By monitoring the temporal evolution of the velocity at the channel center at l/h=380l/h = 380 by PIV, followed by a slight pressure increase at the inlet at t=0t = 0 s, the streamwise velocity u(t)u(t) in laminar flow at Wi=139Wi = 139 starts to grow algebraically up to t∼100t \sim100 s, as shown in Figure 3(a). This growth pattern is in stark contrast to the exponential growth in a case of normal mode instability [1]. At t>100t > 100 s, the growth rate of u(t)u(t) decreases sharply, with u(t)u(t) reaching a saturation point after t≈150t \approx150 s at Wi=169>WiicWi = 169 > Wi_{ic}. During the transient growth and subsequent saturation, the streamwise velocity fluctuations, u′=u−umeanu' = u - u_{\mathrm{mean}}, show a slight increase, in agreement with the observations in Figure 4(a) and Figure 4(b). In particular, at t≈100t \approx100 s, large and almost periodic negative spikes appear and persist until the end of time series (Figure 3(a)). Here uu and umeanu_{\mathrm{mean}} are the streamwise velocity and the mean streamwise velocity averaged over a narrow spanwise extent around the channel centerline, respectively.

Temporal evolution of the streamwise velocity and low-frequency periodic spikes

Figure 3. Temporal evolution of the streamwise velocity u(t)u(t), above Wic=150±10Wi_c=150\pm10 and low-frequency periodic spikes. (a) Streamwise velocity u(t)u(t), and (b) its corresponding partially averaged velocity fluctuations urmsp(t)u_{\mathrm{rms}}^p(t) over each 1 s at 100 fps at l/h=380l/h=380 due to a pressure increase at t=0t=0 s. The flow varies from Wi=139Wi=139 to Wi=169Wi=169. (c) Streamwise velocity, (d) pressure power spectra at low normalized frequencies, λf\lambda f, in lin-log coordinates at three WiWi values. Note the sharp peaks of velocity and pressure fluctuations. (e) WiWi dependence of the normalized periodic spikes frequency, λfp\lambda f_p, above WicWi_c to about the edge of their existence.

Time evolution of velocity fluctuations witnessing a sudden change of inlet pressure

Figure 4. Time evolution of velocity fluctuations witnessing a sudden change of inlet pressure. (a) u(t)u(t) after pressure increase at the inlet at t=0t = 0, resulting in WiWi growth from 140 to 204. The grey line presents the raw data taken with 100 fps of stream-wise velocity as a function of tt, u(t)u(t), the black presents u(t)u(t) averaged over 1 s, and the red dashed line is the fit A×tϵ+u0A \times t^{\epsilon} + u_0, with fitting parameters A=0.0384A = 0.0384, ϵ=0.0829\epsilon= 0.0829, and u0=6.52u_0 = 6.52 mm/s. (b) Seven cases of u(t)u(t) include two scenarios: from laminar to laminar (WiWi: 141→\rightarrow224 and 132→\rightarrow138) and from laminar to transition flow regime at Wi>WicWi > Wi_c (WiWi: 139→\rightarrow169, 148→\rightarrow180, 143→\rightarrow198, 140→\rightarrow204, and 140→\rightarrow224). (c) streamwise velocity at a fixed location as a function of time, u(t)u(t), during 50 s at different Wi=169Wi = 169, 180, 198, 204, 224.

To enhance the visibility of these spikes, u(t)u(t) is partially averaged over each 1 s interval (100 measurement samples), and the resulting partially smoothed data is plotted as urmsp(t)u_{\mathrm{rms}}^{p}(t) in Figure 3(b). It can be seen that urmsp(t)u_{\mathrm{rms}}^{p}(t) shows only a marginal increase at t∼0t \sim0 s, with much more pronounced periodic spikes at t≈100t \approx100 s. In Figure 3(c) and Figure 3(d) of the streamwise velocity power spectra, EuE_u, and pressure power spectra, EpE_p, presented in lin-log coordinates, show narrow spike peaks in the low frequency range for three values of WiWi up to 280. However, at Wi>300Wi > 300 the energy of these spikes is barely detectable and eventually disappears. Thus, the spikes exist only in the lower subrange of the transition flow regime from WiicWi_{ic} up to Wi≈300Wi \approx300. Finally, the dependence of the normalized spike frequency, λfp\lambda_{fp}, obtained from EuE_u, as a function of WiWi is shown in Figure 3(e) along with a fitted relation λfp∼(Wi/Wiic−1)0.45\lambda_{fp} \sim(Wi/Wi_{ic} - 1)^{0.45}.

Statistical properties of velocity fluctuations

Figure 5(a) shows the probability density functions (PDFs) of normalized streamwise velocity fluctuations, P((u−umean)/urms)P((u-u_{\mathrm{mean}})/u_{\mathrm{rms}}), over a range of WiWi from 52 to 280 below and above WiicWi_{ic}. In particular, the first deviations from Gaussian PDFs appear for negative values of (u−umean)/urms(u-u_{\mathrm{mean}})/u_{\mathrm{rms}} at Wi=160Wi = 160, just above WiicWi_{ic}, with these deviations becoming more pronounced up to Wi=280Wi = 280. This is highlighted in Figure 5(c) by showing the PDFs for Wi≥WiicWi \ge Wi_{ic}, with black arrows indicating the primary changes from the Gaussian PDF as WiWi increases. As can be seen, the PDFs of the positive fluctuations decrease, while the PDFs of the negative fluctuations increase significantly. This is due to the spikes at lower u(t)u(t) values, which are smaller than umeanu_{\mathrm{mean}} at Wi>WiicWi > Wi_{ic} (Figure 3(a) and Figure 4). The inset in Figure 5(d) compares PDFs at Wi=52<WiicWi = 52 < Wi_{ic} and Wi=253>WiicWi = 253 > Wi_{ic}, where the latter is fitted by two Gaussian PDFs to emphasize the distortion in the PDF. The main plot in Figure 5(d) illustrates the abrupt changes in skewness (SS) and flatness (kurtosis) (FF) corresponding to the third and fourth moments of P((u−umean)/urms)P((u-u_{\mathrm{mean}})/u_{\mathrm{rms}}) at Wi>WicWi>Wi_c. Remarkably, their deviations from zero precisely determine the instability onset at Wic=150Wi_c=150. In contrast, the PDFs of the normalized spanwise velocity fluctuations, P(w/wrms)P(w/w_{\mathrm{rms}}), show only small deviations from the Gaussian distribution within the error bars in Figure 5(b) for the same WiWi range (Figure 6(c) and Figure 6(d)). Moreover, their SS and FF remain zero both below and above WicWi_c (Figure 5(d)).

PDFs of normalized stream-wise and span-wise velocity fluctuations

Figure 5. PDFs of normalized stream-wise P((u−umean)/urms)P((u-u_{\mathrm{mean}})/u_{\mathrm{rms}}) and span-wise P(w/wrms)P(w/w_{\mathrm{rms}}) velocity fluctuations at the channel center and various WiWi values below and above WicWi_c. (a) P((u−umean)/urms)P((u-u_{\mathrm{mean}})/u_{\mathrm{rms}}) and (b) P(w/wrms)P(w/w_{\mathrm{rms}}) in the whole lower sub-range of WiWi, (c) P((u−umean)/urms)P((u-u_{\mathrm{mean}})/u_{\mathrm{rms}}) only above WicWi_c, and (d) skewness, SS, and flatness (kurtosis), FF, of P((u−umean)/urms)P((u-u_{\mathrm{mean}})/u_{\mathrm{rms}}) and P(w/wrms)P(w/w_{\mathrm{rms}}), respectively. Inset: P((u−umean)/urms)P((u-u_{\mathrm{mean}})/u_{\mathrm{rms}}) for Wi=52<WicWi=52<Wi_c and Wi=253>WicWi=253>Wi_c. The dashed black line is a Gaussian fit to the data at Wi<WicWi<Wi_c, and dashed green lines are Gaussian partial fits to the data at Wi>WicWi>Wi_c with extensions to negative and positive values of (u−umean)/urms(u-u_{\mathrm{mean}})/u_{\mathrm{rms}}, respectively.

PDFs of spanwise and streamwise velocities at channel center

Figure 6. PDFs of spanwise and streamwise velocities at channel center. PDFs of (u−umean/urms)(u-u_{\mathrm{mean}}/u_{\mathrm{rms}}) at (a) Wi<WicWi < Wi_c, and (b) Wi>WicWi > Wi_c. PDFs of w/wrmsw/w_{\mathrm{rms}} at (c) Wi<WicWi < Wi_c, and (d) Wi>WicWi > Wi_c.

Streamwise and spanwise velocity power spectra and elastic waves

At Wi<WicWi<Wi_c, the streamwise velocity power spectrum, EuE_u, shows a slight increase from nearly flat to slight growth toward lower normalized frequencies, λf\lambda f, reminiscent of a white noise spectrum (Figure 7(a)). However, above WicWi_c up to Wi=280Wi=280, EuE_u shows a significant increase, up to three orders of magnitude at λf≤10\lambda f\leq10 (Figure 7(a) and Figure 7(c)). This increase is accompanied by a sharp power-law decay at λf≥10\lambda f\geq10, with the exponent α\alpha ranging from -0.8 to -2.5 (see Figure 7(d)). Here α\alpha is denoted as the exponent of the power-law fit of EuE_u to λ⋅f\lambda\cdot f in the decay regime.

Energy spectra and properties of elastic waves

Figure 7. Energy spectra and properties of elastic waves. (a) EuE_u and (b) EwE_w vs. λf\lambda f at different WiWi from 52 to 280 in log-log and lin-log scales, respectively. (c) EuE_u vs. λf\lambda f at Wi>WicWi > Wi_c with power-law decays fitted by dashed lines with corresponding values of the decay exponents α\alpha and colors. (d) Dependence of α\alpha on WiWi. (e) Dependence of the normalized elastic wave frequency, λfel\lambda f_{\mathrm{el}}, on Wi>WicWi > Wi_c. (f) Dependence of the normalized elastic wave energy, Iel/wrms2I_{\mathrm{el}}/w_{\mathrm{rms}}^2, on WiWi at Wi>WicWi > Wi_c. The dashed black line is the fit, which gives a scaling relation: Iel/wrms2∼(Wi/Wic−1)0.2I_{\mathrm{el}}/w_{\mathrm{rms}}^2 \sim(Wi/Wi_c - 1)^{0.2}. In both (e) and (f) plots the last two points near the onset reach the experimental resolution limit.

In the lower subrange of Wi>WicWi>Wi_c up to Wi=280Wi=280, the spanwise velocity power spectra, EwE_w, plotted on a linear-logarithmic scale in Figure 7(b) and on a log-log scale in Figure 8, remain flat and similar to those observed at Wi<WicWi<Wi_c, which characterizes white noise. However, at Wi>WicWi>Wi_c, pronounced broad and noisy peaks associated with the elastic waves appear at the top of the flat power spectra (Figure 7(b)) and are visible in Figure 8.

Energy spectra of spanwise velocity

Figure 8. Energy spectra of spanwise velocity, EwE_w, versus normalized frequency, λ⋅f\lambda\cdot f in log-log scales at Wi<WicWi < Wi_c in laminar flow and up to Wi=3051Wi = 3051 in transition, ET and DR flow regimes. It is clearly notable that in the lower sub-range of the transition flow regime elastic wave peaks are not noticeable and an order of magnitude smaller than in the upper sub-range at Wi=415Wi = 415 in log-log coordinates.

Since the identification of the peak maximum location and the estimation of its height become unambiguous near WicWi_c due to a drastic increase in peak widths and fluctuations as WiWi approaches WicWi_c, the resolution of both λfel\lambda_{fel} and the normalized intensity of the elastic waves, Iel/wrms2I_{el}/w_{rms}^2, is limited to about ∼5\sim5 and ≈2×10−6\approx2 \times10^{-6} respectively (Figure 7(e) and Figure 7(e)(Figure 7f)). Isolated single peaks in Iel/wrms2I_{el}/w_{rms}^2 versus λf\lambda_f coordinates for Wi>WicWi > Wi_c provide better resolution (Figure 9). In addition, Figure 7(f) includes a power-law fit of the normalized elastic wave energy dependence on WiWi as Iel/wrms2∼(Wi/Wic−1)βI_{el}/w_{rms}^2 \sim(Wi/Wi_c - 1)^\beta with β=0.20\beta= 0.20. Remarkably, in the upper subrange of the transition regime at 400≤Wi⪅1000400 \leq Wi \lessapprox1000, and further in the ET and DR regimes, the elastic wave peaks in EwE_w grow by up to four orders of magnitude, as shown in Figure 8. The location of the peak maximum shifts more than tenfold to higher values of λf\lambda_f, and a power-law decay occurs with the scaling exponent increasing with WiWi. This observation reveals a clear distinction between two subranges in the transition flow regime.

Individual looks of elastic waves

Figure 9. Individual looks of elastic waves. Iel/urms2I_{\mathrm{el}}/u_{\mathrm{rms}}^{2} at Wi=146Wi=146, 160, 172, 190, 195, 253, and 280. Each dashed circle with the corresponding color indicates the appearance of one significant peak.

Observation of random streaks

One of the striking findings in inertia-less viscoelastic channel flows, especially when subjected to external perturbations of different amplitudes at Wi>WicWi > Wi_c, is the phenomenon of cyclic self-organized streaks synchronized by elastic waves [29]. To verify the streak synchronization by the elastic waves we explore the approach based on the velocity difference across the counter-propagating streak interface as a function of normalized time, t∗=tfelt^* = t f_{el}, developed in our laboratory and used in several experiments [24, 25, 26, 27, 29]. In all these experiments, at Wi>WicWi > Wi_c, we discovered cycling with t∗≃1t^* \simeq1 in all three flow regimes.

On the contrary, in the lower subrange at Wi>WicWi > Wi_c up to Wi=280Wi = 280, we reveal the emergence of random streaks shown in Figure 10 in two rows of 5 consecutive images each, illustrating the streaks as a function of t∗t^* at Wi=193Wi = 193 and 207. At Wi≤280Wi \leq280 only random streaks appear, although the number of streaks increases significantly with WiWi (Figure 11). These streaks lack the organized, cyclic nature observed at higher WiWi values. At Wi≥400Wi \geq400, the cycling of self-organized streaks synchronized by the elastic waves, whose energy increases by orders of magnitude, is in agreement with our previous experiments [24, 25, 26, 27, 28, 29].

Instantaneous images of streamwise velocity fluctuations

Figure 10. Instantaneous images of streamwise velocity fluctuations obtained from the full streamwise velocity uu by first subtracting the mean streamwise velocity profile, umean(z)u_{\mathrm{mean}}(z), and then divided by it. Thus, the normalized streamwise velocity fluctuations, u′(x,z)/umean(z)u'(x,z)/u_{\mathrm{mean}}(z), at Wi=193Wi=193 (upper 5 images) and Wi=207Wi=207 (lower 5 images) for 5 normalized times t∗=tfelt^*=t f_{\mathrm{el}} are presented for each WiWi. PIV measurements present images in Δz/W=0.67\Delta z/W=0.67 and Δl/h=4\Delta l/h=4 spatial window.

Number of streaks observed as a function of $Wi$

Figure 11. Number of streaks observed within 70 seconds as a function of WiWi. The appearance of streaks is identified by a contour plot of streamwise velocity fluctuations with high resolution via PIV. The color bar of streaks is presented for Wi>WicWi > Wi_c in the lower sub-range of transition flow regime at Wi=193Wi = 193 and 207.

Discussion

This experimental study attempts to elucidate the mechanism behind the supercritical, non-normal mode elastic instability in viscoelastic channel flow with negligible inertia initiated by external finite-size perturbations due to an unsmoothed inlet and two small holes, in the narrow range from Wi∼60Wi\sim60 to Wi∼300Wi\sim300 close to Wic=150Wi_c=150.

The first finding of this study is the existence of two subranges in the transition flow regime above WicWi_c with different flow properties, as detailed in the Results section. However, the most surprising discovery is the observation of low-frequency periodic spikes in u(t)u(t) and p(t)p(t) detected exclusively in the lower subrange at Wic≤Wi≤300Wi_c\le Wi\le300 (Figure 3 and Figure 4). The periodic spikes appear as sharp low-frequency peaks in the streamwise velocity, EuE_u, and pressure, EpE_p, power spectra presented in lin-log coordinates in Figure 3(c) and Figure 3(d), and only EuE_u in log-log coordinates in Figure 7(a) and Figure 7(c). In the latter, the low-frequency λf<1\lambda f<1 sharp peaks appear on top of the streamwise velocity fluctuation power spectrum characterized by a power-law decay at λf>1\lambda f>1, indicating chaotic flow in the streamwise direction. In contrast, the power spectrum of the spanwise velocity fluctuations, EwE_w, slowly increases at Wi>WicWi>Wi_c in the lower subrange up to Wi≈300Wi\approx300, where EwE_w still remains flat, similar to that observed at Wi<WicWi<Wi_c, indicating a white noise spectrum in the spanwise direction (Figure 7(b) and Figure 8). Thus, in the lower subrange, the chaotic streamwise velocity fluctuations coexist and interact with the spanwise white noise velocity fluctuations in the presence of low-intensity elastic waves. Furthermore, the periodic spikes presented and characterized in Results are reminiscent of stochastic resonance (SR), a phenomenon considered in autonomous dynamical chaotic systems interacting with external white noise in the presence of weak periodic modulation. In dynamical chaotic systems, the strange attractors found in practice are structurally unstable. This means that such a quasi-attractor exists as several regular chaotic attractors coexisting in the phase space of the system, which merge into one chaotic set at the critical value of the control parameter, the so-called attractor crisis [35]. Above the crisis, the merged attractor shows intermittent switching of the “chaos-chaos” type, where the ”deterministic stochastic attractor” is observed [35, 36]. The dynamical intermittent switching is a result of additive noise depending on its intensity, which can be synchronized by weak periodic modulation

resulting in SR [35, 36]. However, the deterministic SR is fundamentally different from the classical SR realized in a bistable system driven simultaneously by noise and a weak periodic signal [37]. To further substantiate the connection of the observations with deterministic SR, we characterize the flow dynamics by plotting the phase portrait in coordinates of the streamwise, (urmsp)2(u_{\mathrm{rms}}^p)^2, versus spanwise, (wrmsp)2(w_{\mathrm{rms}}^p)^2, velocity energies in the two subranges of the transition flow regime (Figure 12(a)). In the lower subrange above WicWi_c we find a one-dimensional streamwise velocity chaotic attractor interacting with spanwise velocity white noise and weak elastic waves, resulting in a SR periodic orbit at Wi=159Wi = 159, 190, and 280 (Figure 12(b)-Figure 12(d)). We note that in the lower subrange, the energy of the external white noise perturbations, (wrmsp)2(w_{\mathrm{rms}}^p)^2 at Wi≤WicWi \leq Wi_c increases slightly up to about 50%, while the chaotic streamwise velocity energy, (urmsp)2(u_{\mathrm{rms}}^p)^2, increases about tenfold. In contrast, in the upper subrange we detect a two-dimensional chaotic attractor of both $(u_{\mathrm{rms}}^p)^2 and (wrmsp)2(w_{\mathrm{rms}}^p)^2 that grows by more than two orders of magnitude compared to external perturbations in a laminar flow (Figure 12(a)).

Phase portraits of the instantaneous dynamics with increasing $Wi$

Figure 12. Phase portraits of the instantaneous dynamics with increasing WiWi. (a) (urmsp)2(u_{rms}^{p})^2 versus (wrmsp)2(w_{rms}^{p})^2 at six WiWi values representing three states: (i) laminar basin (Wi=93Wi=93); (ii) chaotic state of (urmsp)2(u_{rms}^{p})^2 along with periodic orbits and white noise of (wrmsp)2(w_{rms}^{p})^2 at Wi=159Wi=159, 190, and 280; and (iii) the pure chaotic state of both (urmsp)2(u_{rms}^{p})^2 and (wrmsp)2(w_{rms}^{p})^2 at Wi=415Wi=415 and 750. (urmsp)2(u_{rms}^{p})^2 and (wrmsp)2(w_{rms}^{p})^2 are computed from squared velocity fluctuations averaged over 1 second (100 points). Individual views of state (ii) are shown separately at (b) Wi=280Wi=280, (c) Wi=190Wi=190, and (d) Wi=159Wi=159.

Moreover, in the lower subrange above WicWi_c, the presence of periodic spikes leads to significant deviations from Gaussian distributions in the PDFs of (u−umean)/urms(u-u_{\mathrm{mean}})/u_{\mathrm{rms}} at the channel center (Figure 5(a), Figure 5(c), Figure 5(d), and Figure 6). Other notable features in the lower sub-range include the abrupt change from zero to high absolute values of the PDFs’ skewness, SS, and flatness (kurtosis), FF, from Wic=150Wi_c = 150 to Wi<300Wi < 300; the appearance of random streaks (Figure 11); and the observation of wall-normal vortex fluctuations already in the lower subrange (Figure 13). However, the synchronization at felf_{\mathrm{el}} of the self-organized cycling streaks and the enhancement of the wall-normal vortex fluctuations occur only in the upper subrange, at Wi≥400Wi \ge400, where the elastic wave energy, IelI_{\mathrm{el}}, increases by orders of magnitude compared to the lower subrange (Figure 7(b) and Figure 8).

Instantaneous snapshot of the 2D image of streamlines of velocity fluctuations and vortical fluctuations

Figure 13. An instantaneous snapshot of the 2D image of streamlines of velocity fluctuations field (u′,w′)(u', w') in xx–zz plane at the channel center overlapped with the field of the absolute value of the vortical part of the wall-normal vortex fluctuations, ∣ΩR∣|\Omega_R|, at Wi=210Wi=210. The fluid is seeded with red-fluorescent 0.2 μm0.2\,\mu\mathrm{m} traces of ∼0.15%\sim0.15\% w/w concentration (Thermo Scientific). The resolution is 512×512512\times512 pxl2^2 at 4000 fps. The PIV experimental setup is the same as that in Fig. 2A in Ref. [28]. The color bar of the vortical part of vorticity fluctuations is shown on the right.

It is also unexpected that the wall-normal vortex fluctuations are likely generated and supported by strong periodic spikes in the lower subrange, while the elastic waves take over their amplification only in the upper subrange at Wi≥400Wi \ge400 [28]. Remarkably, at Wi>300Wi > 300 the low-frequency periodic spikes disappear, the strongly distorted negative tail of the PDF of (u−umean)/urms(u-u_{\mathrm{mean}})/u_{\mathrm{rms}} becomes exponential, and the flow becomes chaotic in both the uu and ww velocity fields. The flow dynamics at Wi≥400Wi \ge400 are characterized by a two-dimensional chaotic attractor, and the scaling of the flow properties is found to be the same as in the transition flow regime of viscoelastic channel flows with different perturbation intensities, where the measurements are made with lower resolution and further away from WicWi_c [28, 29].

In conclusion, the presented experimental results reveal novel and unexpected features of the non-modal instability evolution above and near WicWi_c and provide strong evidence for the stochastic nature of the instability mechanism. Above WicWi_c, the low-frequency periodic spikes observed in the streamwise velocity time series, characterized by the sharp peaks in EuE_u in the lower subrange at Wic≤Wi≤300Wi_c \le Wi \le300 of the transitional flow regime, are reminiscent of SR. Based on the experimental phase portrait of the instantaneous dynamics, it is proposed that SR arises due to the interaction of streamwise velocity chaotic flow with spanwise velocity white noise in the presence of low-intensity elastic waves. This scenario is similar to the deterministic SR found in autonomous chaotic nonlinear dynamical systems, where deterministic SR is generated due to the interaction of a chaotic attractor with external white noise in the presence of a weak periodic signal at the fixed value of the control parameter [35, 36]. SR greatly increases the probability of slow streamwise velocity fluctuations and promotes the generation of wall-normal vorticity fluctuations. Moreover, since the elastic waves have low intensity at Wic≤Wi≤300Wi_c \le Wi \le300, which is insufficient to initiate and synchronize the streaks and amplify the wall-normal vorticity fluctuations, SR takes over the role of the elastic waves in initiating the random streaks and amplifying the wall-normal vorticity fluctuations.

The discovery of SR in the limited range of Wic≤Wi≤300Wi_c \le Wi \le300 is crucial for understanding the pathway above supercritical non-modal elastic instability to sustained chaotic flow. This phenomenon may have broader implications for various flows, including those in stress fields, such as magneto-hydrodynamic [38], viscoelastic electrolyte [39, 40], viscous solution of rods [41] and suspension of flexible fibers [42], active nematics [43] and bacterial and active turbulence [44] in parallel shear flows.

Finally, we would like to point out similarities and differences with the widely studied theoretically, numerically, and experimentally Newtonian pipe flow at high ReRe, especially in the last two decades [7, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54]. While Newtonian pipe flows have long been shown to be linearly stable, their global stability is not guaranteed. In fact, despite their proven linear stability, Newtonian parallel shear flows become unstable to finite-size external perturbations at finite ReRe [1]. Only about four decades ago it was realized that the discrepancy between the proven linear stability and the experimental observations is explained by the non-Hermitian Orr-Sommerfeld equation, the linearized Navier-Stokes equation for Newtonian parallel shear flows [1]. At finite ReRe, this equation is prone to transient algebraic time growth of non-normal modes, whereas linear eigenmodes decay exponentially, since the flow is linearly stable [1, 5, 6, 7]. The stochastic nature of the non-normal mode instability, recognized since Reynolds’ seminal experiments, has only recently been experimentally demonstrated in viscoelastic parallel shear flows [48, 53, 55]. The dependence of RecRe_c on the external perturbation intensity contradicts the normal mode linear instability onset independent of perturbation intensity [1]. However, the pathway from laminar flow to sustained turbulence via subcritical non-modal instability in Newtonian pipe flow at Re≫1Re \gg1 by convective and absolute instabilities is significantly different and more complicated [46, 48, 50, 51, 52, 53, 56, 57], than the pathway from laminar to sustained chaotic flow via supercritical non-normal-mode elastic absolute instability in viscoelastic channel flows at Wi≫1Wi \gg1, Re≪1Re \ll1 and El≫1El \gg1 [24, 26, 27, 28, 29]. As suggested in Ref. [29], the surprising difference between the flows is due to the presence of only one nonlinear advection term in the Navier-Stokes equation, as opposed to three nonlinear terms in the elastic stress equation, of which only one is an advection term.

Acknowledgments

We thank Guy Han and Rostyslav Baron for their help with the experimental setup. This work was supported in part by the Israel Science Foundation (ISF, grant #784/19).

Materials and Methods

Experimental setup and flow discharge measurements

The experiments are conducted in a straight channel of 500(L)×3.5(W)×0.5(h)500(L) \times3.5(W) \times0.5(h) mm3^3 dimensions, shown in Figure 1. The fluid is driven by N2_2 gas at pressure up to 10psi. The fluid discharge is weighed instantaneously, m(t)m(t), by a PC-interfaced balance (BPS-1000-C2-V2, MRC) to measure the time-averaged fluid discharge rate Q=⟨Δm/Δt⟩Q = \langle\Delta m/\Delta t \rangle to get the mean velocity U=Q/ρWhU = Q/\rho Wh. Then Wi=λU/hWi = \lambda U/h and Re=ρUh/ηRe = \rho Uh/\eta vary in the ranges (30, 300) and (0.005, 0.045), respectively. High resolution (0.1% of full scale) absolute pressure sensors located near the inlet and outlet at Δl/h≈10\Delta l/h \approx10 (HSC series, Honeywell) with range up to 5 psi are used to measure the pressure fluctuations. For Figure 3 and Figure 4, the container of the solution is immediately switched to the other N2_2 supply line with a slightly higher or lower pressure at t=0t = 0 and just after switching, WiWi is calculated when the flow discharge, when the flow system reaches equilibrium.

Polymer solution preparation and characterization

As the working fluid, a dilute polymer solution of high molecular weight polyacrylamide (Polysciences, Mw=18M_w = 18M Da) at a concentration of c=80c = 80 ppm (c/c∗≈0.4c/c^* \approx0.4 with the overlapping polymer concentration of c∗≈200c^* \approx200 ppm [58]) is prepared using a water-sucrose solvent with a sugar weight fraction of 64%. The properties of the solution are as follows: the solution density (ρ\rho) is 1320 kg/m3^3, the solvent viscosity (ηs\eta_s) is 0.13 Pa·s, and the total solution viscosity (η\eta) is 0.17 Pa·s. The ratio of solvent viscosity to total solution viscosity is ηs/(ηs+ηp)=0.765\eta_s/(\eta_s+\eta_p)=0.765, where ηp\eta_p is the polymer contribution to the solution viscosity. The longest polymer relaxation time (λ\lambda) is 13 seconds, obtained by the stress relaxation method [58]. The result is El=Wi/Re=λρ/ηd2=6.8×103El=Wi/Re=\lambda\rho/\eta d^2=6.8 \times10^3.

Imaging system and PIV measurements

We perform velocity field measurements at various distances l/hl/h downstream of the inlet using the Particle Image Velocimetry (PIV) method. The PIV setup consists of 3.2 μ\mum latex fluorescent particle tracers of ∼\sim1% w/w concentration (Thermo Scientific) illuminated by a

laser sheet of ≈50 μ\approx50\ \mum thickness over the central channel plane, i.e. the x−zx-z plane. A high speed camera (Mini WX100 FASTCAM, Photron) has a high spatial resolution and images of tracer pairs are acquired at 200 to 2000 fps. The OpenPIV software [59] is used to analyze u(x,z,t)u(x,z,t) and w(x,z,t)w(x,z,t) in the 2D xx-zz plane to record data for ∼O(15)\sim O(15) minutes, or ∼O(50λ)\sim O(50\lambda), for each WiWi to obtain sufficient statistics. For the velocity fluctuations in Figure 3, Figure 7, and Figure 4, we use spatial averaging over narrow windows with a resolution of 256 (xx) ×\times96 (zz) pxl2^2 with a 4x objective, which serves as a single point velocity measurement at the channel center. The window size for PIV is 32×\times32 pxl2^2 with 50% overlap and 200% search window size. For the velocity profile visualization in Figure 10, we keep the same scale but increase the spatial resolution to 1280×\times1280 pxl2^2.

References

References

  1. [1]P. G. Drazin and W. H. Reid, Hydrodynamic stability (Cambridge university press, 2004).
  2. [2]R. G. Larson, E. S. Shaqfeh, and S. J. Muller, A purely elastic instability in taylor–couette flow, J. Fluid Mech. 218, 573 (1990).DOI
  3. [3]E. S. Shaqfeh, Purely elastic instabilities in viscometric flows, Annu. Rev. Fluid Mech. 28, 129 (1996).
  4. [4]V. Steinberg, Elastic turbulence: an experimental view on inertialess random flow, Annu. Rev. Fluid Mech. 53, 27 (2021).DOI
  5. [5]P. J. Schmid, Nonmodal stability theory, Annu. Rev. Fluid Mech. 39, 129 (2007).DOI
  6. [6]L. N. Trefethen, A. E. Trefethen, S. C. Reddy, and T. A. Driscoll, Hydrodynamic stability without eigenval- ues, Science 261, 578 (1993).
  7. [7]R. Kerswell, Nonlinear nonmodal stability theory, Annu. Rev. Fluid Mech. 50, 319 (2018).DOI
  8. [8]P. C. Cross, M. C. & Hohenberg, Pattern formation outside of equilibrium, Rev. Mod. Phys. 65, 851 (1993).DOI
  9. [9]P. Pakdel and G. H. McKinley, Elastic instability and curved streamlines, Phys. Rev. Lett. 77, 2459 (1996).pubmed.ncbi.nlm.nih.gov/10061959
  10. [10]V. Gorodtsov and A. Leonov, On a linear instability of a plane parallel couette flow of viscoelastic fluid, J. Appl. Math. Mech. 31, 310 (1967).DOI
  11. [11]M. Renardy and Y. Renardy, Linear stability of plane couette flow of an upper convected maxwell fluid, J. Non-newton. Fluid Mech. 22, 23 (1986).
  12. [12]H. A. C. Sánchez, M. R. Jovanović, S. Kumar, A. Morozov, V. Shankar, G. Subramanian, and H. J. Wilson, Understanding viscoelastic flow instabilities: Oldroyd-b and beyond, J. Nonnewton. Fluid Mech. 302, 104742 (2022).
  13. [13]M. R. Jovanović and S. Kumar, Transient growth without inertia, Phys. Fluid 22, 023101 (2010).DOI
  14. [14]M. R. Jovanović and S. Kumar, Nonmodal amplification of stochastic disturbances in strongly elastic channel flows, J. Nonnewton. Fluid Mech. 166, 755 (2011).DOI
  15. [15]J. Page and T. A. Zaki, Streak evolution in viscoelastic couette flow, J. Fluid Mech. 742, 520 (2014).DOI
  16. [16]G. Hariharan, M. R. Jovanović, and S. Kumar, Localized stress amplification in inertialess channel flows of viscoelastic fluids, J. Nonnewton. Fluid Mech. 291, 104514 (2021).DOI
  17. [17]B. K. Lieu, M. R. Jovanović, and S. Kumar, Worst-case amplification of disturbances in inertialess couette flow of viscoelastic fluids, J. Fluid Mech. 723, 232 (2013).DOI
  18. [18]B. Yesilata, Nonlinear dynamics of a highly viscous and elastic fluid in pipe flow, Fluid Dyn. Res. 31, 41 (2002).DOI
  19. [19]B. Yesilata, Temporal nature of polymeric flows near circular pipe-exit, Polym-Plast. Technol. 48, 723 (2009).DOI
  20. [20]D. Bonn, F. Ingremeau, Y. Amarouchene, and H. Kellay, Large velocity fluctuations in small-reynolds-number pipe flow of polymer solutions, Phys. Rev. E 84, 045301 (2011).pubmed.ncbi.nlm.nih.gov/22181216
  21. [21]L. Pan, A. Morozov, C. Wagner, and P. Arratia, Nonlinear elastic instability in channel flows at low reynolds numbers, Phys. Rev. Lett. 110, 174502 (2013).pubmed.ncbi.nlm.nih.gov/23679736
  22. [22]B. Qin and P. E. Arratia, Characterizing elastic turbulence in channel flows at low reynolds number, Phys. Rev. Fluids 2, 083302 (2017).arxiv.org/abs/1609.08532
  23. [23]B. Qin, P. F. Salipante, S. D. Hudson, and P. E. Arratia, Flow resistance and structures in viscoelastic channel flows at low re, Phys. Rev. Lett. 123, 194501 (2019).DOI
  24. [24]N. K. Jha and V. Steinberg, Universal coherent structures of elastic turbulence in straight channel with viscoelastic fluid flow (2020), arXiv preprint https://arxiv.org/abs/2009.12258.
  25. [25]N. K. Jha and V. Steinberg, Elastically driven kelvin–helmholtz-like instability in straight channel flow, Proceedings of the National Academy of Sciences 118, e2105211118 (2021).DOI
  26. [26]R. Shnapp and V. Steinberg, Nonmodal elastic instability and elastic waves in weakly perturbed channel flow, Phys. Rev. Fluids 7, 063901 (2022).DOI
  27. [27]V. Steinberg, New direction and perspectives in elastic instability and turbulence in various viscoelastic flow geometries without inertia, Low Temp. Phys. 48, 492 (2022).DOI
  28. [28]Y. Li and V. Steinberg, Mechanism of vorticity amplification by elastic waves in a viscoelastic channel flow, Proc. Nat. Acad. Sci. U. S. A. 120, e2305595120 (2023).DOI
  29. [29]Y. Li and V. Steinberg, Elastic instability in a straight channel of viscoelastic flow without prearranged perturbations, Sci. Rep. 13, 1064 (2023).DOI
  30. [30]S. S. Datta, A. M. Ardekani, P. E. Arratia, A. N. Beris, I. Bishofberger, G. H. McKinley, J. G. Eggers, J. E. López-Aguilar, S. M. Fielding, A. Frishman, M. D. Graham, J. S. Guasto, S. J. Haward, A. Q. Shen, S. Hormozi, A. Morozov, R. J. Poole, V. Shankar, E. S. G. Shaqfeh, H. Stark, V. Steinberg, G. Subramanian, and H. A. Stone, Perspectives on viscoelastic flow instabilities and elastic turbulence, Phys. Rev. Fluids 7, 080701 (2022).arxiv.org/abs/2108.09841
  31. [31]E. Balkovsky, A. Fouxon, and V. Lebedev, Turbulence of polymer solutions, Phys. Rev. E 64, 056301 (2001).pubmed.ncbi.nlm.nih.gov/11736088
  32. [32]A. Varshney and V. Steinberg, Elastic alfven waves in elastic turbulence, Nat. Commun. 10, 652 (2019).
  33. [33]A. Varshney and V. Steinberg, Drag enhancement and drag reduction in viscoelastic flow, Phys. Rev. Fluids 3, 103302 (2018).DOI
  34. [34]M. V. Kumar, A. Varshney, D. Li, and V. Steinberg, Relaminarization of elastic turbulence, Phys. Rev. Fluids 7, L081301 (2022).
  35. [35]V. S. Anishchenko, A. B. Neiman, F. Moss, and L. Shimansky-Geier, Stochastic resonance: noise-enhanced order, Physics-Uspekhi 42, 7 (1999).
  36. [36]V. S. Anishchenko, A. B. Neiman, and M. A. Safonova, Stochastic resonance in chaotic systems, J. Stat. Phys. 70, 183 (1993).DOI
  37. [37]R. Benzi, A. Sutera, and A. Vulpiani, The mechanism of stochastic resonance, J. Phys. A: Math. Gen. 14A, L451 (1981).DOI
  38. [38]A. A. Schekochihin, Mhd turbulence: a biased review, J. Plasma Phys. 88, 155880501 (2022).arxiv.org/abs/2010.00699
  39. [39]G. Li, L. A. Archer, and D. L. Koch, Electroconvection in a viscoelastic electrolyte, Phys. Rev. Lett. 122, 124501 (2019).DOI
  40. [40]S. K. Das, G. Natale, and A. M. Benneker, Viscoelastic behavior of dilute polyelectrolyte solutions in complex geometries, J. Nonnewton. Fluid Mech. 309, 104920 (2022).DOI
  41. [41]L. Emmanuel, S. Musacchio, and D. Vincenzi, Emergence of chaos in a viscous solution of rods, Phys. Rev. E 96, 053108 (2017).DOI
  42. [42]C. Brouzet, G. Verhille, and P. Le Gal, Flexible fiber in a turbulent flow: A macroscopic polymer, Phys. Rev. Lett. 112, 074501 (2014).
  43. [43]A. Doostmohammadi, J. Ignés-Mullol, J. M. Yeomans, and F. Sagués, Active nematics, Nat. Commun. 9, 3246 (2018).DOI
  44. [44]R. Alert, J. Casademunt, and J.-F. Joanny, Active turbulence, Annual Review of Condensed Matter Physics 13, 143 (2022).
  45. [45]H. Faisst and B. Eckhardt, Sensitive dependence on initial conditions in transition to turbulence in pipe flow, J. Fluid Mech. 504, 343 (2004).DOI
  46. [46]H. Wedin and R. R. Kerswell, Exact coherent structures in pipe flow: travelling wave solutions, J. Fluid Mech. 508, 333 (2004).
  47. [47]R. Kerswell, Recent progress in understanding the transition to turbulence in a pipe, Nonlinearity 18, R17 (2005).DOI
  48. [48]J. Peixinho and T. Mullin, Decay of turbulence in pipe flow, Phys. Rev. Lett. 96, 094501 (2006).pubmed.ncbi.nlm.nih.gov/16606268
  49. [49]T. Mullin, Experimental studies of transition to turbulence in a pipe, Annu. Rev. Fluid Mech. 43, 1 (2011).DOI
  50. [50]B. Hof, J. Westerweel, T. M. Schneider, and B. Eckhardt, Finite lifetime of turbulence in shear flows, Nature (London) 443, 59 (2006).pubmed.ncbi.nlm.nih.gov/16957725
  51. [51]B. Hof, A. D. Lozar, D. J. Kuik, and J. Westerweel, Repeller or attractor? selecting the dynamical model for the onset of turbulence in pipe flow, Phys. Rev. Lett. 101, 214501 (2008).pubmed.ncbi.nlm.nih.gov/19113412
  52. [52]K. Avila, D. Moxey, A. De Lozar, M. Avila, D. Barkley, and B. Hof, The onset of turbulence in pipe flow, Science 333, 192 (2011).DOI
  53. [53]D. Barkley, Theoretical perspective on the route to turbulence in a pipe, J. Fluid Mech. 803, P1 (2016).DOI
  54. [54]M. Avila, D. Barkley, and B. Hof, Transition to turbulence in pipe flow, Annu. Rev. Fluid Mech. 55, 575 (2023).DOI
  55. [55]B. Hof, A. Juel, and T. Mullin, Scaling of the threshold of pipe flow turbulence, Phys. Rev. Lett. 91, 244502 (2003).
  56. [56]H. Faisst and B. Eckhardt, Traveling waves in pipe flow, Phys. Rev. Lett. 91, 224502 (2003).pubmed.ncbi.nlm.nih.gov/14683243
  57. [57]C. S. Paranjape, G. Yalnız, Y. Duguet, N. B. Budanur, and B. Hof, Direct path from turbulence to time-periodic solutions, Phys. Rev. Lett. 131, 034002 (2023).DOI
  58. [58]Y. Liu, Y. Jun, and V. Steinberg, Concentration dependence of the longest relaxation times of dilute and semi-dilute polymer solutions, J. Rheol. 53, 1069 (2009).DOI
  59. [59]A. Liberzon, T. Käufer, A. Bauer, P. Vennemann, and E. Zimmer, Openpiv/openpiv-python: Openpiv-python v0.23.6 (2021).

Paper details

vlnc:W_55mtr3t2
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