Hostname: page-component-7b7567c8c-tnkh4 Total loading time: 0 Render date: 2026-11-16T10:33:24.240Z Has data issue: false hasContentIssue false

Lubricated viscous gravity currents

Published online by Cambridge University Press:  10 February 2015

Katarzyna N. Kowal*
Affiliation:
Institute of Theoretical Geophysics, Department of Applied Mathematics and Theoretical Physics, University of Cambridge, Wilberforce Road, Cambridge CB3 0WA, UK
M. Grae Worster
Affiliation:
Institute of Theoretical Geophysics, Department of Applied Mathematics and Theoretical Physics, University of Cambridge, Wilberforce Road, Cambridge CB3 0WA, UK
*
†Email address for correspondence: k.kowal@damtp.cam.ac.uk
Rights & Permissions [Opens in a new window]

Abstract

We present a theoretical and experimental study of viscous gravity currents lubricated by another viscous fluid from below. We use lubrication theory to model both layers as Newtonian fluids spreading under their own weight in two-dimensional and axisymmetric settings over a smooth rigid horizontal surface and consider the limit in which vertical shear provides the dominant resistance to the flow in both layers. There are contributions from Poiseuille-like flow driven by buoyancy and Couette-like flow driven by viscous coupling between the layers. The flow is self-similar if both fluids are released simultaneously, and exhibits initial transient behaviour when there is a delay between the initiation of flow in the two layers. We solve for both situations and show that the latter converges towards self-similarity at late times. The flow depends on three key dimensionless parameters relating the relative dynamic viscosities, input fluxes and density differences between the two layers. Provided the density difference between the two layers is bounded away from zero, we find an asymptotic solution in which the front of the lubricant is driven by its own gravitational spreading. There is a singular limit of equal densities in which the lubricant no longer spreads under its own weight in the vicinity of its nose and ends abruptly with a non-zero thickness there. We explore various regimes, from thin lubricating layers underneath a more viscous current to thin surface films coating an underlying more viscous current and find that although a thin film does not greatly influence the more viscous current if it forms a surface coating, it begins to cause interesting dynamics if it lubricates the more viscous current from below. We find experimentally that a lubricated gravity current is prone to a fingering instability.

Information

Type
Papers
Copyright
© 2015 Cambridge University Press 
Figure 0

Figure 1.Schematic diagram illustrating the side profile of the flow of two superposed thin films of viscous fluid spreading under gravity on a rigid horizontal surface.

Figure 1

Figure 2.Solutions of the full system of partial differential equations (PDEs) (2.27)–(2.35) for two-dimensional spreading with $\mathscr{M}=10\,000$, $\mathscr{D}=0.1$ and $\mathscr{Q}=0.2$. (a) Profile thicknesses as functions of $x$ at dimensionless times $t=1.1,2,3$. Surface heights are solid curves and interfacial heights are dashed. (b) Frontal positions and (c) source thicknesses as functions of $t$ for the upper (bold solid) and lower (bold dashed) films. The solutions of the PDEs (2.27)–(2.35) in (b,c) undergo an initial transient and converge to the self-similar solutions (2.38)–(2.47) marked by thin black horizontal lines as $t\rightarrow \infty$. The $y$-axes in (b,c) are scaled time-dependently with respect to the long-time similarity solution.

Figure 2

Figure 3.Similarity solutions and asymptotic results in two dimensions. Solutions of (2.38)–(2.47) for the thicknesses of both currents are presented as solid curves. Asymptotic solutions (2.48) and (2.49) near the two fronts are presented as dashed bold curves. Parameter values used are $\mathscr{M}=100$, $\mathscr{D}=1$ and $\mathscr{Q}=0.2$.

Figure 3

Figure 4.Solutions of the full system of PDEs (3.8)–(3.16) for axisymmetric spreading with $\mathscr{M}=10\,000$, $\mathscr{D}=0.1$ and $\mathscr{Q}=0.2$. (a) Profile thicknesses as functions of $x$ at dimensionless times $t=1.01,2,3$. Surface heights are solid curves and interfacial heights are dashed. (b) Frontal positions as functions of $t$ for the upper (bold solid) and lower (bold dashed) films. (c) Thickness of the upper film at $r=r_{L}$ (bold solid) as a function of $t$. The solutions of the PDEs (3.8)–(3.16) in (b,c) undergo an initial transient and converge to the self-similar solutions (3.19)–(3.28) marked by thin black horizontal lines as $t\rightarrow \infty$. The $y$-axes in (b,c) are scaled time-dependently with respect to the long-time similarity solution.

Figure 4

Figure 5.Similarity solutions and asymptotic results in axisymmetry. Solutions of (3.19)–(3.28) for the thicknesses of both currents are presented as solid curves. Asymptotic solutions (3.29) and (3.30) near the two fronts are presented as dashed bold curves. Parameter values used are $\mathscr{M}=100$, $\mathscr{D}=1$ and $\mathscr{Q}=0.2$.

Figure 5

Figure 6.Schematic of our experimental set-up.

Figure 6

Table 1.Parameter values used in our experiments.

Figure 7

Figure 7.Bottom view of experiment D at $t=0,60,180,420$  s after injection of salt solution. The full experiment is shown in supplementary movie 1 (http://dx.doi.org/10.1017/jfm.2015.30).

Figure 8

Figure 8.Side view of experiment D at $t=0,60,180,420$  s after injection of salt solution. The full experiment is shown in supplementary movie 2.

Figure 9

Figure 9.Comparison of our experimental data for experiments A–H for $r_{N}$ ( filled symbols) and $r_{L}$ (open symbols) to solutions for $r_{N}$ (solid) and $r_{L}$ (dashed) of the full system of PDEs (3.8)–(3.16) for axisymmetric spreading. The dash-dot curves show $r_{N}$ as a function of $t$ in the unlubricated case. The vertical dotted lines indicate the time at which there is a first sign of a fingering instability in the experiments.

Figure 10

Figure 10.Experimental values for $r_{L}(\mathscr{M}\mathscr{Q}gQ_{0}^{3}{\it\nu}^{-1})^{-1/8}$ versus time. The solid line is a least-squares power-law fit.

Figure 11

Figure 11.Regimes (I)–(IV) in parameter space for two-dimensional currents. Here $\mathscr{D}$ is held fixed and bounded away from zero.

Figure 12

Table 2.Scalings in two-dimensional and axisymmetric settings.

Figure 13

Figure 12.Profile thicknesses of two-dimensional two-layer gravity currents in similarity coordinates for a range of different parameters with $\mathscr{D}=1$ fixed. Upper surface heights are solid and interfacial heights are dashed. The dotted (red online) curves in (a–c) are the surface heights that the upper fluid would have if the lower fluid were not present. The dotted (blue online) curve in (d) is the surface height that the lower fluid would have if the upper fluid were not present. The parameter values used are $\mathscr{M}=1$, $\mathscr{Q}=0.1$ in (a), $\mathscr{M}=100$, $\mathscr{Q}=2$ in (b), $\mathscr{M}=100$, $\mathscr{Q}=0.1$ in (c) and $\mathscr{M}=0.05$ and $\mathscr{Q}=1$ in (d). (a–d) correspond to regimes (I)–(IV), respectively.

Figure 14

Figure 13.Profile thicknesses of axisymmetric two-layer gravity currents in similarity coordinates for a range of different parameters with $\mathscr{D}=1$ fixed. Upper surface heights are solid and interfacial heights are dashed. The dotted (red online) curves in (a–c) are the surface heights that the upper fluid would have if the lower fluid were not present. The dotted (blue online) curve in (d) is the surface height that the lower fluid would have if the upper fluid were not present. The parameter values used are $\mathscr{M}=1$, $\mathscr{Q}=0.1$ in (a), $\mathscr{M}=100$, $\mathscr{Q}=2$ in (b), $\mathscr{M}=100$, $\mathscr{Q}=0.1$ in (c), and $\mathscr{M}=0.05$ and $\mathscr{Q}=1$ in (d). (a–d) correspond to regimes (I)–(IV), respectively.

Figure 15

Figure 14.(a,b) Front positions and (c,d) source thicknesses of two dimensional currents on regular and log–log axes, all versus flux ratio. Parameter values used are $\mathscr{M}=50$ and $\mathscr{D}=1$, corresponding to regimes (I)–(III). The solid (red online) line in (a,b) corresponds to the extent of the upper more viscous current, if the lower less viscous layer were not present. The solid (red online) line in (c) corresponds to the thickness of the upper more viscous current if the lower less viscous layer were not present. The power laws (5.1), (5.5), and (5.8) for ${\it\eta}_{L}$ and $f_{0}$ with respect to $\mathscr{Q}$ are shown as triangles.

Figure 16

Figure 15.(a) Front positions and (b) source thicknesses of two dimensional currents on log–log axes versus viscosity ratio, for $\mathscr{Q}=0.001$ and $\mathscr{D}=1$, corresponding to regime (I). The power laws (5.1a−d) for ${\it\eta}_{L}$ and $f_{0}$ with respect to $\mathscr{M}$ are shown as triangles.

Figure 17

Figure 16.Two-dimensional thickness profiles in similarity coordinates for varying $\mathscr{D}\in \{0.05,0.25,0.5,1\}$ with $\mathscr{M}=10,\mathscr{Q}=0.5$. Surface heights are solid curves and interfacial heights between the two fluids are dashed. The $\mathscr{D}=0$ singular limiting solution is shown as bold curves for the two layers.

Kowal and Worster supplementary movie

Movie of the bottom view of one of our experiments. The lubricant (dyed green) was introduced 334 seconds after the start of the experiment. The movie is sped up by a factor of one hundred.

Download Kowal and Worster supplementary movie(Video)
Video 517.9 KB

Kowal and Worster supplementary movie

Movie of the side view of one of our experiments. The lubricant (dyed green) was introduced 334 seconds after the start of the experiment. The movie is sped up by a factor of one hundred.

Download Kowal and Worster supplementary movie(Video)
Video 614.6 KB