Hostname: page-component-7b7567c8c-2n9c9 Total loading time: 0 Render date: 2026-11-16T08:11:53.579Z Has data issue: false hasContentIssue false

Triangular instability of a strained Batchelor vortex

Published online by Cambridge University Press:  06 March 2026

Aditya Sai Pranith Ayapilla*
Affiliation:
Graduate School of Information Sciences, Tohoku University , Sendai 980-8579, Japan
Yuji Hattori
Affiliation:
Institute of Fluid Science, Tohoku University, Sendai 980-8577, Japan
Stéphane Le Dizès
Affiliation:
Aix Marseille Université, Centrale Méditerranée, IRPHE, CNRS, Marseille France
*
Corresponding author: Aditya Sai Pranith Ayapilla, ayapilla.aditya.sai.pranith.t6@dc.tohoku.ac.jp

Abstract

We investigate the triangular instability of a Batchelor vortex subjected to a stationary triangular strain field generated by three satellite vortices, in the presence of weak axial flow. The analysis combines theoretical predictions with numerical simulations. Theoretically, the instability arises from resonant coupling between two quasi-neutral Kelvin modes with azimuthal wavenumbers $m$ and $m+3$ with the background strain. Numerically, we solve the linearised Navier–Stokes equations around a quasi-steady base flow to identify the most unstable modes, and compare their growth rates and frequencies with theoretical predictions for a Reynolds number $\textrm{Re} = 10^4$ and a straining strength $\epsilon = 0.008$. In the absence of axial flow, only the mode pair $(m_A, m_B) = (-1,2)$ (and its symmetric counterpart) is unstable. However, we show that additional combinations such as $(0,3)$, $(1,4)$ and $(2,5)$, which are otherwise strongly damped by the critical layer in the absence of axial flow, also become unstable once axial flow exceeds a certain threshold, as the critical-layer damping is significantly reduced. Furthermore, we show that the most unstable mode in the no-axial-flow case, originating from the second branch of $m = -1$ and the first branch of $m = 2$, becomes less unstable as axial flow increases. It is eventually overtaken by a mode from the first branches of both wavenumbers, which then remains the dominant unstable mode across a wide range of axial flow strengths, Reynolds numbers and straining strengths. A comprehensive instability diagram as a function of the axial flow parameter is presented.

Information

Type
JFM Papers
Copyright
© The Author(s), 2026. Published by Cambridge University Press
Figure 0

Figure 1.Flow configuration of the hub vortex and the three satellite vortices.

Figure 1

Figure 2.Comparison of the axial vorticity and axial velocity of the quasi-steady base flow obtained from DNS (solid black line) and theory (red dashed line) for $\textrm{Re} = 1000$ and $\epsilon = 0.008$. Panels (a) and (b) show the leading-order term and the triangular straining correction of the axial vorticity, respectively. Panels (c) and (d) present the corresponding leading-order and correction terms for the axial velocity.

Figure 2

Figure 3.Effects of axial flow on the dispersion curves ($\textrm{{Re}}(\omega )$ vs $k$) of Kelvin modes of the Batchelor vortex, computed at $\textrm{Re} = 10^4$ by integration along the real axis. Each point represents a mode, with its greyscale intensity corresponding to the damping rate $-\textrm {Im}(\omega )$: darker points indicate lower damping ($=0$) and lighter points higher damping ($=0.1$). Top panels show results for $W_0 = 0$ (Lamb–Oseen vortex) and bottom panels for $W_0 = -0.1$. Results are shown for (a) $m = -1$, (b) $m = 2$, (c) $m = 0$ and (d) $m = 3$.

Figure 3

Figure 4.Effects of axial flow on the crossing points of the dispersion curves ($\textrm{{Re}}(\omega )$ vs $k$) for $m_A = -1$ and $m_B = 2$. Results are shown for (a) $W_0 = 0$, (b) $W_0 = -0.1$ and (c) $W_0 = -0.3$. Each point represents a mode, with its greyscale intensity corresponding to the damping rate $-\textrm {Im}(\omega )$: darker points indicate lower damping ($=0$) and lighter points higher damping ($=0.1$).

Figure 4

Figure 5.Plots of the epicyclic frequencies $\omega ^+$ and $\omega ^-$ (solid lines) and the critical frequency $\omega _c$ (dashed line) as functions of the radial coordinate $r$. Results are shown for (a) $m = 2$, $kW_0 = 0$, with an inset showing a zoomed-in view of the indicated region; (b) $m = 2$, $kW_0 = -4$. The blue and red shaded areas represent the domains of regular neutral core modes and singular neutral core modes, respectively. In (b) the yellow shaded region corresponds to regular ring modes. Hatched regions denote the regions where modes are localised. The dotted region in the inset of (a) highlights where singular neutral modes encounter a critical point given by $\omega _c$.

Figure 5

Figure 6.Comparison of the structures of a regular neutral mode and a singular neutral mode obtained at $\textrm{Re} = 10^4$, $m = 2$, $kW_0 = 0$. (a) Regular neutral mode: 2-D contours of axial disturbance vorticity for the mode with $k = 10.0$, $\textrm {{Re}}(\omega ) = 3.138$, $\textrm {Im}(\omega ) = -0.020$; (b) singular neutral mode: $k = 1.76$, $\textrm {{Re}}(\omega ) = 0.407$, $\textrm {Im}(\omega ) = -0.045$; (c) radial distribution of the absolute value of the eigenfunction $|u(r)|$, with the black curve corresponding to the mode in (a) and the red curve to the mode in (b).

Figure 6

Figure 7.Comparison of the structures of a regular core mode and a regular ring mode, obtained at $\textrm{Re} = 10^4$, $m = 2$, $kW_0 = -4$. (a) Core mode: 2-D contours of axial disturbance vorticity for the mode with $k = 10.0$, $\textrm {{Re}}(\omega ) = -3.252$, $\textrm {Im}(\omega ) = -0.016$; (b) ring mode: $k = 10.0$, $\textrm {{Re}}(\omega ) = 0.719$, $\textrm {Im}(\omega ) = -0.011$; (c) radial distribution of the absolute value of the eigenfunction $|u(r)|$, with the black curve corresponding to the mode in (a) and the red curve to the mode in (b).

Figure 7

Figure 8.Domain of possible resonance between Kelvin modes of the Batchelor vortex, shown for the azimuthal wavenumber pairs $(m_A, m_B) = (-1, 2)$, $(0, 3)$ and $(1, 4)$, indicated in black, red and blue, respectively. Solid lines represent resonance domains involving core modes, while dashed lines correspond to ring modes of $m_A$. In the $(-1,2)$ domain, the black shaded region indicates resonances involving singular modes of $m = 2$ only; the mid-grey region corresponds to resonances involving singular modes of both $m = -1$ and $m = 2$; and the pale grey region corresponds to resonances involving singular modes of $m = -1$ only. The black dotted line at $kW_0 = 0$ represents the Lamb–Oseen vortex case.

Figure 8

Figure 9.Resonant modes A1, A2, etc. and their growth rates at $W_0 = -0.1$. Top panels: dispersion curves showing $\textrm {{Re}}(\omega )$ vs $k$ for (a) $(m_A, m_B) = (-1, 2)$ and (b) $(0, 3)$. Red curves represent $m_B$ modes, blue curves represent $m_A$ modes, with colour transitions indicating increasing damping (from 0 to $-0.05$). Resonant modes are marked with black circles at the crossing points. Grey background regions in the top panels mark the core mode resonance domain predicted by the asymptotic theory. Bottom panels: growth rate curves $\sigma$ vs $k$; solid black lines show the viscous case ($\textrm{Re} = 10^4$), dashed lines show the inviscid case and red circles indicate DNS results.

Figure 9

Figure 10.Axial flow strength $W_0 = -0.1$; 2-D structures of the modes labelled A1, A2, etc., obtained from DNS, showing contours of axial vorticity in the $x$-$y$ plane. Red regions indicate positive vorticity, while blue regions indicate negative vorticity. Domain: $-2 \lt X,Y \lt 2$. The hub-vortex boundary is indicated with a dotted circle for reference.

Figure 10

Figure 11.Resonant modes B1, B2, etc. and their growth rates at $W_0 = -0.2$. Top panels: dispersion curves showing $\textrm {{Re}}(\omega )$ vs $k$ for (a) $(m_A, m_B) = (-1, 2)$, (b) $(0, 3)$, (c) $(1, 4)$ and (d) $(2, 5)$. Red curves represent $m_B$ modes, blue curves represent $m_A$ modes, with colour transitions indicating increasing damping (from 0 to $-0.05$). Resonant modes are marked with black circles at the crossing points. Grey background regions in the top panels mark the core mode resonance domain predicted by the asymptotic theory, while orange regions represent the resonance domain for ring modes of $m_A$. Bottom panels: growth rate curves $\sigma$ vs $k$; solid black lines show the viscous case ($\textrm{Re} = 10^4$), dashed lines show the inviscid case and red circles indicate DNS results.

Figure 11

Figure 12.Axial flow strength $W_0 = -0.2$; 2-D structures of the modes labelled B1, B2, etc., obtained from DNS, showing contours of axial vorticity in the $x$–$y$ plane. Red regions indicate positive vorticity, while blue regions indicate negative vorticity. Domain: $-2 \lt X,Y \lt 2$. The hub-vortex boundary is indicated with a dotted circle for reference.

Figure 12

Figure 13.Resonant modes C1, C2, etc. and their growth rates at $W_0 = -0.3$. Top panels: dispersion curves showing $\textrm {{Re}}(\omega )$ vs $k$ for (a) $(m_A, m_B) = (-1, 2)$, (b) $(0, 3)$ and (c) $(1, 4)$. Blue–green curves represent $m_B$ modes, black–brown curves represent $m_A$ modes, with colour transitions indicating increasing damping (from 0 to $-0.05$). Resonant modes are marked with black circles at the crossing points. Grey background regions in the top panels mark the core mode resonance domain predicted by the asymptotic theory, while orange regions represent the resonance domain for ring modes of $m_A$. Bottom panels: growth rate curves $\sigma$ vs $k$; solid black lines show the viscous case ($\textrm{Re} = 10^4$), dashed lines show the inviscid case and red circles indicate DNS results.

Figure 13

Figure 14.Axial flow strength $W_0 = -0.3$; 2-D structures of the modes labelled C1, C2, etc., obtained from DNS, showing contours of axial vorticity in the $x$–$y$ plane. Red regions indicate positive vorticity, while blue regions indicate negative vorticity. Domain: $-2 \lt X,Y \lt 2$. The hub-vortex boundary is indicated with a dotted circle for reference.

Figure 14

Table 1.Composition of the numerically obtained modes for $W_0 = -0.1$, $-0.2$ and $-0.3$. The second column lists the mode labels (A, B, C) corresponding to the respective $W_0$ values. The third and fourth columns indicate the primary azimuthal wavenumber pairs $(m_A, m_B)$ and their corresponding branches $(l_A, l_B)$, and any secondary resonant pairs, if present. Columns 5–11 show the energy distribution among different azimuthal wavenumbers $m$, representing the relative contribution of each $m$ component to the total mode energy.

Figure 15

Table 2.Summary of the primary resonant modes obtained for $W_0 = -0.1$, $-0.2$ and $-0.3$. For each mode, the table lists the axial wavenumber $k$, frequency $\omega _{\textit{dns}}$ obtained from DNS, theoretical frequency $\omega _{{th}}$, growth rate $\sigma _{\textit{dns}}$ from DNS at $\textrm{Re} = 10^4$, viscous growth rate $\sigma _{\text{th}}$ from theory at $\textrm{Re} = 10^4$ and inviscid growth rate $\sigma _{{th}}^{(\infty )}$ from theory.

Figure 16

Figure 15.Distributions of the eigenfunctions $|u(r)|$ vs $r$ for modes B5, B8, C2, C3, C4 and C5, where the modes corresponding to $m_A$ are ring modes. Coloured lines represent DNS results, while black lines correspond to eigenfunctions obtained by solving (4.3).

Figure 17

Figure 16.Characteristics of the resonant modes as a function of $W_0$ for $(m_A, m_B) = (-1, 2)$, corresponding to the branches $[l_A, l_B]$ indicated in the legend: (a) axial wavenumber $k$, (b) frequency $\textrm {{Re}}(\omega )$, (c) damping rate of $m_B$$\textrm {Im}(\omega _B)$, and (d) damping rate of $m_A$$\textrm {Im}(\omega _A)$.

Figure 18

Figure 17.Same as figure 16 but for $(m_A, m_B) = (0, 3)$.

Figure 19

Figure 18.Same as figure 16 but for $(m_A, m_B) = (1, 4)$.

Figure 20

Figure 19.Same as figure 16 but for $(m_A, m_B) = (2, 5)$.

Figure 21

Figure 20.Panels (a), (b), (c) and (d) show the maximum inviscid growth rate as a function of $-W_0$ for (a) $(m_A, m_B) = (-1, 2)$, (b) $(0, 3)$, (c) $(1, 4)$ and (d) $(2, 5)$. Panels (e), (f), (g) and (h) display the corresponding maximum growth rates at $\textrm{Re} = 10^4$. The black dashed line indicates the maximum growth rate of the first principal mode $[1,1]$ and the blue dashed line corresponds to the second principal mode $[2,2]$.

Figure 22

Figure 21.Growth rate $\sigma$ contours on the $(k, W_0)$ plane for $(m_A, m_B) = (-1, 2)$, corresponding to mode branches $[l_A, l_B]$ as labelled near each island. (a): inviscid case with $\epsilon = 0.008$; (b): viscous case with $\textrm{Re} = 10^4$ and $\epsilon = 0.008$; (c): inviscid case with $\epsilon = 0.001$. The corresponding resonant axial wavenumbers as a function of $-W_0$ for those modes are also plotted as thin white dashed lines for reference.

Figure 23

Figure 22.Growth rate $\sigma$ contours on the $(k, W_0)$ plane for $(m_A, m_B) = (0, 3)$, corresponding to mode branches $[l_A, l_B]$ as labelled near each island. (a): inviscid case with $\epsilon = 0.008$; (b): viscous case with $\textrm{Re} = 10^4$ and $\epsilon = 0.008$; (c): inviscid case with $\epsilon = 0.001$. The modes $[2,1]$ and $[2,2]$, which exhibit two local maxima in growth rate and distinct ‘tail’ regions, are highlighted in red and blue, respectively, for clarity. The corresponding resonant axial wavenumbers as a function of $-W_0$ for those modes are also plotted as thin white dashed lines for reference.

Figure 24

Figure 23.Growth rate $\sigma$ contours on the $(k, W_0)$ plane for $(m_A, m_B) = (1, 4)$, corresponding to mode branches $[l_A, l_B]$ as labelled near each island. (a): inviscid case with $\epsilon = 0.008$; (b): viscous case with $\textrm{Re} = 10^4$ and $\epsilon = 0.008$; (c): inviscid case with $\epsilon = 0.001$. The modes $[2,1]$ and $[2,2]$, which exhibit two local maxima in growth rate and distinct ‘tail’ regions, are highlighted in red and blue, respectively, for clarity. The corresponding resonant axial wavenumbers as a function of $-W_0$ for those modes are also plotted as thin white dashed lines for reference.

Figure 25

Figure 24.Growth rate $\sigma$ contours on the $(k, W_0)$ plane for $(m_A, m_B) = (2, 5)$, corresponding to mode branches $[l_A, l_B]$ as labelled near each island. Left panel: inviscid case with $\epsilon = 0.008$; centre panel: viscous case with $\textrm{Re} = 10^4$ and $\epsilon = 0.008$; right panel: inviscid case with $\epsilon = 0.001$. The modes $[2,1]$ and $[2,2]$, which exhibit two local maxima in growth rate and distinct ‘tail’ regions, are highlighted in red and blue, respectively, for clarity. The corresponding resonant axial wavenumbers as a function of $-W_0$ for those modes are also plotted as thin white dashed lines for reference.

Figure 26

Figure 25.Growth rate $\sigma$ contours on the $(k, W_0)$ plane for the principal modes $[1,1]$ and $[2,2]$, shown for $(m_A, m_B) = (-1, 2)$ (grey), $(0, 3)$ (blue), $(1, 4)$ (red) and $(2, 5)$ (purple). Left panel: inviscid case with $\epsilon = 0.008$; centre panel: viscous case with $\textrm{Re} = 10^4$ and $\epsilon = 0.008$; right panel: inviscid case with $\epsilon = 0.001$.

Figure 27

Figure 26.Transition of the modes corresponding to $m_A$ from core modes to ring modes for (a) $(m_A, m_B) = (0, 3)$, (b) $(1, 4)$ and (c) $(2, 5)$. Top panels: frequency versus $-kW_0$ for the resonant modes, overlaid on the core and ring-mode domains predicted by asymptotic theory. Bottom panels: maximum inviscid growth rate versus $-kW_0$, with stars indicating the transition points from core to ring modes. Labels and colour scheme follow those used in figures 16–20.

Figure 28

Figure 27.Top: evolution of the 2-D axial disturbance vorticity field of the mode $(-1,2,[1,1])$, obtained theoretically at $\textrm{Re} = 10^4$, as $W_0$ varies from $0$ to $-0.4$. Panels (a–l) correspond to values of $-W_0$ equal to $0$, $0.025$, $0.05$, $0.075$, $0.1$, $0.15$, $0.2$, $0.25$, $0.3$, $0.35$, $0.375$ and $0.4$, respectively. Bottom: amplitude ratio of $m_B$ to $m_A$ as a function of $W_0$.

Figure 29

Figure 28.Contours of the maximum growth rate in the $(W_0, \textrm{Re})$ parameter space, computed for different modes identified by $(m_A, m_B, [l_A, l_B])$ as indicated in the legend. The left panel corresponds to $\epsilon = 0.008$ and the right panel to $\epsilon = 0.001$.

Figure 30

Figure 29.Contours of the maximum growth rate in the $(W_0, \epsilon )$ parameter space for $\textrm{Re} = 10^4$. Modes identified by $(m_A, m_B, [l_A, l_B])$ are indicated in the legend.