Hostname: page-component-7b7567c8c-z4zkx Total loading time: 0 Render date: 2026-11-17T12:00:29.125Z Has data issue: false hasContentIssue false

Asymptotic study of linear instability in a viscoelastic pipe flow

Published online by Cambridge University Press:  31 January 2022

Ming Dong*
Affiliation:
State Key Laboratory of Nonlinear Mechanics, Institute of Mechanics, Chinese Academy of Sciences, Beijing 100190, PR China
Mengqi Zhang*
Affiliation:
Department of Mechanical Engineering, National University of Singapore, 9 Engineering Drive 1, 117575 Singapore
*
†Email addresses for correspondence: dongming@imech.ac.cn, mpezmq@nus.edu.sg
†Email addresses for correspondence: dongming@imech.ac.cn, mpezmq@nus.edu.sg

Abstract

It is recently found that viscoelastic pipe flows can be linearly unstable, leading to the possibility of a supercritical transition route, in contrast to Newtonian pipe flows. Such an instability is referred to as the centre mode, which was studied numerically by Chaudhary et al. (J. Fluid Mech., vol. 908, 2021, p. A11) based on an Oldroyd-B model. In this paper, we are interested in expanding the parameter space investigated and exploring the asymptotic scalings related to this centre instability in the Oldroyd-B viscoelastic pipe flow. It is found from the asymptotic analysis that the centre mode exhibits a three-layered asymptotic structure in the radial direction, a wall layer, a main layer and a central layer, which are driven by pure viscosity, axial and/or radial pressure gradient, and a combined effect of viscosity and elasticity, respectively. Depending on the relations of the control parameters, two regimes, the long-wavelength and short-wavelength centre instabilities, emerge, for which the central-layer thicknesses are of different orders of magnitude. Our large-Reynolds-number asymptotic predictions are compared to the numerical solutions of the original eigenvalue system, and favourable agreement is achieved, especially when the parameters approach their individual limits. In addition to revealing the dominant factors and their balances, the asymptotic analysis describes the instability system by reducing the number of control parameters, and furthermore explaining the collapse of the numerical results for different re-scalings.

Information

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

Figure 1.Projections of the critical neutral curve for $\beta =0.9$ and $m=0$ in the $Wi$–$Re$ (a) and $Wi$–$k$ (b) planes.

Figure 1

Figure 2.A sketch of the multi-layered structure for long-wavelength centre modes in a viscoelastic pipe flow. The critical layer appears only for near-neutral low-concentration centre modes.

Figure 2

Table 1.Summary of the asymptotic regimes, including the scaling relations of the control parameters, similarity parameters and thickness of each asymptotic layer in viscoelastic pipe flows.

Figure 3

Figure 3.Dependence of the real (a) and imaginary (b) parts of $c_1$ on $W_1$ for $\beta =0.65$, 0.8 and 0.9. Continuous curves, asymptotic predictions from § 3.1; open squares, EOS solutions for $(k, Re)=(0.5,2000)$; filled circles, EOS solutions for $(k,Re)=(0.5,20000)$. The thin horizontal line in (b) represents $c_{1,i}=0$.

Figure 4

Figure 4.Dependence of $c_{1,i}$ on $W_1$ (a) and $\bar W$ (b) for different $\beta$ values, where the thin horizontal line represents $c_{1,i}=0$, and the dots in (a) mark the second peaks of $c_{1,i}$. In (b), the black curve with circles denotes the low-concentration asymptotic prediction from § 3.2.

Figure 5

Table 2.Key parameters characterising the instability property.

Figure 6

Figure 5.Transverse distribution of the eigenfunctions for $\beta =0.85$. (a) Lower-branch neutral point of mode I, where $W_1=3.87$ and $c_1=-0.316$; (b) upper-branch neutral pint of mode I, where $W_1=8.85$ and $c_1=-0.0373$; (c) lower-branch neutral point of mode II , where $W_1=10.57$ and $c_1=-0.0392$; (d) upper-branch neutral point of mode II, where $W_1=34.3$ and $c_1=-0.00848$.

Figure 7

Figure 6.Dispersion relations of the neutral (red curves) and most unstable (blue curves) modes in the $W_1$–$\beta$ (a), $c_{1,r}$–$\beta$ (b) and $c_{1,i}$–$\beta$ (c) spaces. The pink dashed lines are the low-concentration asymptotic predictions.

Figure 8

Figure 7.Comparison of the dispersion relations between the asymptotic predictions (continuous curves) and EOS solutions (symbols). (a, b) Neutral modes in the $W_1$–$\beta$ and $c_{1,r}$–$\beta$ spaces, respectively. (c,d,e) The most unstable modes in the $W_1$–$\beta$, $c_{1,r}$–$\beta$ and $c_{1,i}$–$\beta$ spaces, respectively.

Figure 9

Figure 8.Dependence of the real (a) and imaginary (b) parts of $\hat c_1$ on $W_3$ for $\beta =0.65$, 0.8 and 0.9 in the short-wavelength regime, where $R_3=25$.

Figure 10

Figure 9.Dependence of the real (a) and imaginary (b) parts of $\hat c_1$ on $\bar W_3$ for $\beta =0.65$, 0.8, 0.9 and 0.99 in the short-wavelength regime, where $R_3=25$.

Figure 11

Figure 10.Dependence of the real (a) and imaginary (b) parts of $\hat c_1$ on $R_3$ in the short-wavelength regime, where $\bar W_3=\sigma W_3=1.071$. The thin dashed lines represent the low-concentration prediction for $R_3=\infty$. The inset in (a) shows a zoom-in plot for $-0.05<\hat c_{1,r}<0$.

Figure 12

Figure 11.(a) The neutral (red) and the most unstable (blue) curves in the $W_3$–$\beta$ plane for $R_3=25$; (b) the neutral curve (red) in the $R_3$–$\beta$ plane for $\bar W_3=1.071$. The pink dashed lines represent the predictions in the low-concentration regime. The symbols in (a) are the EOS results for $k=3$.

Figure 13

Figure 12.Neutral curve in the $R_3$–$\bar W_3$ plane in the low-concentration short-wavelength regime, where the dashed line represents the prediction in the limit as $R_3\rightarrow \infty$, and the filled circle indicates a representative neutral case, $(R_3,\bar W_3)=(5.9,1.071)$.

Figure 14

Figure 13.Comparisons between the asymptotic predictions and numerical results given by Chaudhary et al. (2021) (solid lines), where the red and green circles are the asymptotic predictions (5.3a,b) of the lower- and upper-branch neutral curves, respectively, and the blue squares are the asymptotic predictions from § 4.1. (a,b) $\beta =0.65$, (c,d) $\beta =0.9$.

Figure 15

Figure 14.Neutral curves in the $k[(1-\beta )E]^{1/2}$–$Re[(1-\beta )E]^{3/2}$ plane. Solid lines, numerical results given by Chaudhary et al. (2021); red circles, asymptotic prediction from (5.6); blue squares, low-concentration asymptotic prediction.

Figure 16

Table 3.Re-scale of the control parameters in figure 15.

Figure 17

Figure 15.Eigenfunctions for a representative long-wavelength configuration, where the coloured lines are from figure 25 of Chaudhary et al. (2021), and the circles denote the asymptotic prediction given by § 3.1 with $W_1=1.06$.

Figure 18

Table 4.Re-scale of the control parameters in figure 16.

Figure 19

Figure 16.Eigenfunctions for a representative short-wavelength configuration, where the coloured lines are from figure 24 of Chaudhary et al. (2021), and the circles denote the asymptotic prediction given by § 4.1 with $W_3=0.925$ and $R_3=245$.

Figure 20

Figure 17.$(\textit {a})$ Comparison of the eigenspectra between a viscoelastic pipe flow ($Re=5000$, $Wi=106.1$, $\beta =0.5$, $k=1$) and a Newtonian pipe flow ($Re=5000$, $\beta =1$, $k=1$). $(\textit {b})$ Comparison of the viscoelastic eigenspectra between our results and those in Garg et al. (2018), where the parameters are the same as the blue triangles in (a). The unstable centre viscoelastic modes are marked in both panels.

Figure 21

Figure 18.Contours of the perturbations $\hat u_z$ (a,b) and $\hat c_{11}$ (c,d) of the centre (a,c) and wall (b,d) modes for $Re=5000$, $Wi=106.1$, $\beta =0.5$, $k=1$. The complex eigenvalues for the centre and wall modes are $\omega =0.9948+0.0001116{{\rm i}}$ and $0.4024-0.06671{{\rm i}}$, respectively.

Figure 22

Figure 19.Eigenfunctions of $\hat u$ and $\hat v$ for a near-neutral low-concentration mode. (a) Solution from § 3.1 with $\beta =0.99$, $W_1=63$ and $\bar c_1=-0.183+0.004{{\rm i}}$; (b) solution from § 3.2 with $\bar W=0.64$ and $\bar c_1=-0.172+0.004{{\rm i}}$.

Figure 23

Figure 20.Dependence of the growth rate on $\beta$ for the large-$W_1$ mode II instability.