Hostname: page-component-7b7567c8c-z4zkx Total loading time: 0 Render date: 2026-11-16T14:50:53.454Z Has data issue: false hasContentIssue false

Vortex merging and splitting events in viscoelastic Taylor–Couette flow

Published online by Cambridge University Press:  08 August 2022

Jose M. Lopez*
Affiliation:
Physics Department, Universitat Politècnica de Catalunya, Campus Nord UPC, 08034 Barcelona, Spain
*
†Email address for correspondence: jose.manuel.lopez.alonso@upc.edu

Abstract

Recent experiments have reported a novel transition to elasto-inertial turbulence in the Taylor–Couette flow of a dilute polymer solution. Unlike previously reported transitions, this newly discovered scenario, dubbed vortex merging and splitting (VMS) transition, occurs in the centrifugally unstable regime and the mechanisms underlying it are two-dimensional: the flow becomes chaotic due to the proliferation of events where axisymmetric vortex pairs may be either created (vortex splitting) or annihilated (vortex merging). In this paper, we present direct numerical simulations, using the finitely extensible nonlinear elastic-Peterlin (FENE-P) constitutive equation to model the polymer dynamics, which reproduce the experimental observations with great accuracy and elucidate the reasons for the onset of this surprising dynamics. Starting from the Newtonian limit and increasing progressively the fluid's elasticity, we demonstrate that the VMS dynamics is not associated with the well-known Taylor vortices, but with a steady pattern of elastically induced axisymmetric vortex pairs known as diwhirls. The amount of angular momentum carried by these elastic vortices becomes increasingly small as the fluid's elasticity increases and it eventually reaches a marginal level. When this occurs, the diwhirls become dynamically disconnected from the rest of the system and move independently from each other in the axial direction. It is shown that vortex merging and splitting events, along with local transient chaotic dynamics, result from the interactions among diwhirls, and that this complex spatio-temporal dynamics persists even at elasticity levels twice as large as those investigated experimentally.

Information

Type
JFM Papers
Copyright
© The Author(s), 2022. Published by Cambridge University Press
Figure 0

Table 1.Number of radial nodes ($m_r$) and axial Fourier modes ($m_z$) used in the simulations depending on the aspect ratio $\varGamma$ of the system.

Figure 1

Figure 1.Space–time plot representing the magnitude of the radial velocity $u$ at mid-gap along the axial direction $z$. Panels (a) and (b) correspond to a simulation performed at $Re = 95$ where, starting from a Newtonian TVF, $El$ was slowly increased with time ($El = 10^{-3}t/Re$). Note that the time $t$ has been replaced by the corresponding $El$ values on the horizontal axis. Panel (a) shows the variation of $u$ from Newtonian flow (i.e. $El = 0$) up to the largest $El$ value simulated ($El = 0.50$), whereas panel (b) shows in more detail the range of $El$ values for which a complex spatio-temporal dynamics take place. Panel (c) illustrates the spatio-temporal dynamics when $El$ is kept constant after the VMS regime is achieved. The case exemplified corresponds to $El = 0.32$. Red and blue areas indicate outflows and inflows respectively. Note that periodic boundary conditions are used in $z$.

Figure 2

Figure 2.Structural variation of the TVF pattern as $El$ increases from the Newtonian case (i.e. $El = 0.00$) up to values near the onset of spatio-temporal dynamics. Each panel shows a colour map of the radial velocity $u$ in a meridional plane $(z/d,r/d) \in [0,4{\rm \pi} ] \times [3.35,4.35]$, where red regions indicate outflows (i.e. positive velocity), blue regions represent inflows (i.e. negative velocity) and the zero velocity has been set as grey. The colour scale is based on the maximum and minimum values of each case and hence differs among the different panels. There are four positive and four negative contours evenly distributed across the full range of values in each case: 1. $El = 0.00$, $u$ in $[-0.037,0.059]$; 2. $El = 0.03$, $u$ in $[-0.043,0.047]$; 3. $El = 0.05$, $u$ in $[-0.048,0.029]$; 4. $El = 0.10$, $u$ in $[-0.048,0.014]$; 5. $El = 0.15$, $u$ in $[-0.082,0.015]$; 6. $El = 0.25$, $u$ in $[-0.072,0.011]$. The system is shown rotated by $90^{\circ}$ in the counterclockwise direction with respect to its original position. The location of the inner and outer cylinders, $r_i$ and $r_o$, respectively, as well as the locations of the top ($z/d = 4{\rm \pi}$) and bottom ($z/d =0$) of the system are indicated in the bottom panel.

Figure 3

Figure 3.Axial profiles of the elastic force $F_e$ (left panels) and its associated work $F_e u$ (right panels) obtained at the mid-gap for $El$ values matching those of the last three panels in figure 2. The elastic force is calculated as $F_e = ({(1-\beta )}/{Re}) (\partial _r T_{rr} + {(T_{rr}-T_{\theta \theta })}/{r} + \partial _z T_{rz})$. A dashed line has been added at $F_e u = 0$ to help identify inflows ($F_e u > 0$) and outflows ($F_e u < 0$).

Figure 4

Figure 4.Profiles of the radial velocity $u$ at mid-gap along the $z$-axis. Panel (a) shows profiles for $El < 0.12$, where profound changes in the structure of the flow pattern are observed in figure 2, whereas panel (b) focuses on the range of $El$ values ($El \geq 0.12$) where the variation in the profiles is small. Note that in both panels $u$ is normalized with the largest absolute value among all cases which is obtained for $El = 0.12$ and corresponds to $|u_{max}| = 0.0824$.

Figure 5

Figure 5.Quantification of the changes in the structure and strength of the vortices as $El$ increases. Panel (a) shows the variation in the axial extent of inflows and outflows at the midgap in units of the gap-width, $d$, as the fluid becomes more elastic. Panel (b) displays the ratio between the maximal values of $u$ in the outflows and inflows as a function of $El$. The dashed line indicates the approximate value this ratio seems to approach asymptotically as $El$ increases.

Figure 6

Figure 6.Variation of the integral energy budgets with increasing $El$ before the spatio-temporal dynamics sets in. Here, $\mathcal {P}$ denotes the kinetic energy production by the inertial forces, $\epsilon$ stands for the viscous dissipation and $\varPi _e$ is the work done by the elastic stresses. The dotted (orange) vertical line indicates the $El$ value at which the transition between the low and high $El$ regimes takes place. Hereafter, these regimes are denoted as centrifugally and elasticity dominated regimes, respectively.

Figure 7

Figure 7.Comparison between the flow streamlines $\psi$ in the centrifugally (top) and elastically (bottom) dominated regimes. Positive (negative) values are represented as red (blue), whereas the zero value is shown as grey. The colour scale is based on the maximum and minimum values of each case and hence differs between the panels. There are four positive and four negative contours evenly distributed across the full range of values in each case: 1. $El = 0.00$, $\psi$ in $[-0.660,0.660]$; 2. $El = 0.25$, $\psi$ in $[-0.295,0.295]$. The system is shown rotated by $90\,^{\circ}$ in the counterclockwise direction.

Figure 8

Figure 8.Colour maps of $u$ illustrating the axial oscillation of the vortex pairs that emerges from the Hopf bifurcation. Letters from A to E are used to show the correspondence between each panel and the time series shown in figure 9. Positive (negative) velocity is represented as red (blue), whereas the zero velocity is shown as grey. There are four positive and four negative contours evenly distributed in $u \in [-0.0780,0.0095]$. The system is shown rotated by $90\,^{\circ}$ in the counterclockwise direction. Note that the flow patterns shown in ${\rm A}$ and ${\rm E}$ are identical, whereas those shown in B and D only differ in that their outflows are antisymmetric.

Figure 9

Figure 9.Periodic motion arising from the instability of the stationary pattern of DW. (a) Time series of the axial velocity $w$ for $El = 0.3$ calculated at $r/d = 4.15$ and $z/d = {\rm \pi}$ . The circles indicate time instants for which colour maps of $u$ are presented in figure 8. (b) Power spectral density (PSD). The frequency is normalized with the elastic frequency, $f_e$.

Figure 10

Figure 10.Colour maps of $w$ illustrating the standing wave causing the axial oscillation of the flow pattern. The top panel shows $w$ for a stationary state at $El = 0.25$, whereas the middle and bottom panels correspond to the states B and D indicated in figure 9. Positive (negative) values of $w$ are shown as red (blue), whereas that for $w=0$ is shown as grey. From top to bottom, there are ${\rm four}$ positive and ${\rm four}$ negative contours evenly distributed in $w \in [-0.030,0.030]$, $[-0.026,0.030]$ and $[-0.030,0.026]$, respectively. Two additional contours at $w = \pm 0.003$ are also added to help visualize the axial waves. The system is shown rotated by $90\,^{\circ}$ in the counterclockwise direction.

Figure 11

Figure 11.Colour maps of $u$ illustrating the simultaneous occurrence of two merging events at the beginning of the VMS regime. They correspond to a simulation performed at constant $El = 0.32$, whose space–time diagram was shown in figure 1(c). The dashed (green) rectangle in the latter figure indicates the time window that is illustrated in the current figure. Positive (negative) values of $u$ are shown as red (blue), whereas $u=0$ is shown as grey. In each panel there are ${\rm four}$ positive and ${\rm four}$ negative contours evenly distributed in $u \in [-0.071,0.013]$. The system is shown rotated by $90\,^{\circ}$ in the counterclockwise direction.

Figure 12

Figure 12.Colour maps of $u$ illustrating the flow patterns associated with the chaotic dynamics that appears interspersed between slow merging events and splitting events. The dynamics displayed in this figure corresponds to the simulation with constant $El=0.32$, and more specifically, to the time window shown as a solid line (brown) rectangle in figure 1(c). Positive (negative) values of $u$ are shown as red (blue), whereas $u=0$ is shown as grey. There are ${\rm four}$ positive and ${\rm four}$ negative contours evenly distributed in $u \in [-0.073,0.013]$. The system is shown rotated by $90\,^{\circ}$ in the counterclockwise direction.

Figure 13

Figure 13.Colour maps of $u$ illustrating the occurrence of a splitting event. The example shown corresponds to the simulation with constant $El=0.32$ and it is indicated as a dash-dotted (purple) rectangle in figure 1(c). Positive (negative) values of $u$ are shown as red (blue), whereas $u=0$ is shown as grey. There are ${\rm four}$ positive and ${\rm four}$ negative contours evenly distributed in $u \in [-0.077,0.015]$. The system is shown rotated by $90\,^{\circ}$ in the counterclockwise direction.

Figure 14

Figure 14.PSD obtained from simulations in the VMS regime at three distinct values of $El$. The left panels show the spectra in a linear–log plot, with the frequency normalized with the elastic frequency, $f_e$, whereas the right panels show the spectra in log–log scale and the frequency is non-dimensionalized with the inverse of the advective time.

Figure 15

Figure 15.Space–time diagrams showing the magnitude of $u$ at the mid-gap along the $z$ direction in simulations where the axial length of the system and/or the number of vortex pairs of the initial condition were varied with respect to the simulations shown in figure 1. The cases illustrated correspond to (a) $\varGamma = 10$ using an initial condition with ${\rm five}$ vortex pairs, (b) $\varGamma = 14$ using an initial condition with ${\rm eight}$ vortex pairs and (c) $\varGamma = 14$ using an initial condition with ${\rm six}$ vortex pairs. As in the simulations presented in figure 1, $El$ has been slowly increased with time ($El = 10^{-3}t/Re$) and the latter has been replaced by its corresponding $El$ value in the horizontal axis of the space–time plots. Red and blue areas indicate outflows and inflows respectively. Note that periodic boundary conditions are used in $z$.

Figure 16

Figure 16.Space–time diagram showing the magnitude of $u$ at the mid-gap along the $z$ direction in a simulation where $El$ was slowly decreased with time at the same rate as it was increased in the simulations previously presented ($El = 10^{-3}t/Re$). The parameters used in this simulation are the same as those used in the simulation shown in § 3.1. The initial condition corresponds to the state obtained in the simulation where $El$ was slowly increased with time for $El = 0.392$ (indicated by an orange dashed line in figure 1a). The flow states observed when $El$ decreases differ from those obtained when $El$ increases, evidencing the existence of hysteresis. Red and blue areas indicate outflows and inflows, respectively. Note that periodic boundary conditions are used in $z$.

Figure 17

Figure 17.Contribution of the elastic stresses to the integral kinetic energy budget $\varPi _e$ as a function of the maximum polymer extension $L$ in the range of $El$ values where the transition between the centrifugally and elastically dominated regimes takes place.

Figure 18

Figure 18.Space–time diagram showing the magnitude of $u$ at the mid-gap along the $z$ direction in simulations where the maximum polymer extension was set to $L=70$ (a) and $L=250$ (b). The rest of the parameters were kept as in § 3.1. Red and blue areas indicate outflows and inflows, respectively.

Figure 19

Figure 19.Example of the dynamics observed in the simulations when $\alpha$ (i.e. the rate of increase in $El$) is larger than $1.25 \times 10^{-3}$. (a) Space–time diagram showing the magnitude of $u$ at the mid-gap along the $z$ direction in a simulation where $\alpha = 1.6 \times 10^{-3}$. The rest of the parameters are as in § 3.1. (b) Space–time plot obtained when $El$ is held constant at $0.5$ in the simulation shown in (a). The dashed (red) rectangles demarcate regions of transient chaotic dynamics. Red and blue areas indicate outflows and inflows, respectively.

Figure 20

Figure 20.Colour map of $u$ illustrating the instantaneous flow structure at $El = 0.5$ in a simulation where $\alpha = 5\times 10^{-3}$. Positive (negative) values of $u$ are shown as red (blue), whereas $u=0$ is shown as grey. There are $10$ contours evenly distributed in $u \in [-0.065,0.015]$. The system is shown rotated by $90\,^{\circ}$ in the counterclockwise direction. Dashed (green) rectangles are used to highlight the small scale vortices that emerge near the inner cylinder.

Figure 21

Figure 21.Variation of the axially and time averaged angular momentum current ($J^{\omega }$) and its contributing terms, $J^{\omega }_c$, $J^{\omega }_d$ and $J^{\omega }_p$, obtained near mid-gap, as $El$ increases. The (orange) vertical dotted lines demarcate the different flow regimes observed in the present study: I. the centrifugally dominated regime, II. the elastically dominated regime characterized by steady patterns of equally spaced DW, III. the elastically dominated regime characterized by spatio-temporal dynamics.

Figure 22

Figure 22.Variation of the angular velocity current, $J^{\omega }$, and its components, $J^{\omega }_c$, $J^{\omega }_d$ and $J^{\omega }_p$, with $El$. Panels (a) and (b) correspond to simulations where the VMS regime takes place, whereas panel (c) exemplifies a case where this regime does not occur. More specifically, the data shown in panels (a) and (b) correspond to the simulations whose space–time diagrams are illustrated in figure 15(b) and (c), whereas the data shown in panel (c) corresponds to the simulation illustrated in figure 18(a).