Hostname: page-component-7b7567c8c-wxvwp Total loading time: 0 Render date: 2026-11-16T10:08:24.419Z Has data issue: false hasContentIssue false

Early-time jet formation in liquid–liquid impact problems: theory and simulations

Published online by Cambridge University Press:  11 October 2018

R. Cimpeanu*
Affiliation:
Mathematical Institute, Andrew Wiles Building, Radcliffe Observatory Quarter, Woodstock Road, Oxford OX2 6GG, UKDepartment of Mathematics, Imperial College London, South Kensington Campus, London SW7 2AZ, UK
Madeleine Rose Moore
Affiliation:
Mathematical Institute, Andrew Wiles Building, Radcliffe Observatory Quarter, Woodstock Road, Oxford OX2 6GG, UK
*
†Email address for correspondence: radu.cimpeanu@imperial.ac.uk

Abstract

We perform a thorough qualitative and quantitative comparison of theoretical predictions and direct numerical simulations for the two-dimensional, vertical impact of two droplets of the same fluid. In particular, we show that the theoretical predictions for the location and velocity of the jet root are excellent in the early stages of the impact, while the predicted jet velocity and thickness profiles are also in good agreement with the computations before the jet begins to bend. By neglecting the role of the surrounding gas both before and after impact, we are able to use Wagner theory to describe the early-time structure of the impact. We derive the model for general droplet velocities and radii, which encompasses a wide range of impact scenarios from the symmetric impact of identical drops to liquid drops impacting a deep pool. The leading-order solution is sufficient to predict the curve along which the root of the high-speed jet travels. After moving into a frame fixed in this curve, we are able to derive the zero-gravity shallow-water equations governing the leading-order thickness and velocity of the jet. Our numerical simulations are performed in the open-source software Gerris, which allows for the level of local grid refinement necessary for a problem with such a wide variety of length scales. The numerical simulations incorporate more of the physics of the problem, in particular the surrounding gas, the fluid viscosities, gravity and surface tension. We compare the computed and predicted solutions for a range of droplet radii and velocities, finding excellent agreement in the early stage. In light of these successful comparisons, we discuss the tangible benefits of using Wagner theory to confidently track properties such as the jet-root location, jet thickness and jet velocity in future studies of splash jet/ejecta evolution.

Information

Type
JFM Papers
Copyright
© 2018 Cambridge University Press 
Figure 0

Figure 1.(a) The general configuration of two circular droplets of the same liquid just before impact. The fluid in each droplet is coloured with different shades of grey for ease of viewing. The upper droplet is moving downwards with speed $V_{+}$ and has a radius of curvature $R_{+}$. The lower droplet is moving upwards with speed $V_{-}$ and has radius of curvature $R_{-}$. The droplets are surrounded by gas. (b) The symmetric limit in which $V_{+}=V_{-}$ and $R_{+}=R_{-}$, i.e. $R=1,V=1$. (c) The deep pool limit in which $V_{-}=0$, $R_{-}=\infty$, i.e. $V=0,R=\infty$.

Figure 1

Figure 2.The Wagner asymptotic structure based on the small time scale $\unicode[STIX]{x1D6FF}$. The fluid in each droplet is coloured with different shades of grey for ease of viewing. The horizontal coordinates of the upper and lower turnover points are denoted by $\pm d_{\pm }(\hat{t})$ respectively. Note that the problem is symmetric about the $y$-axis.

Figure 2

Figure 3.A close-up of the right-hand jet root to highlight the main quantities of interest local to the turnover region. The fluid in each droplet is coloured with different shades of grey for ease of viewing. The turnover points $\boldsymbol{x}_{j\pm }$ as defined in (3.10) are taken to be the points at which the curvature of the free surface is locally maximised. The thickness of the jet is taken to be the distance between the turnover points and is denoted by $J(\hat{t})$. The angle $\unicode[STIX]{x1D703}_{jet}(\hat{t})$ is a measure of the local angle the jet makes to the horizontal. Specifically, we take $\unicode[STIX]{x1D703}_{jet}(\hat{t})$ to be the angle formed when we take the local tangent to the interface between the fluids at the point at which the line joining the turnover points intersects the interface.

Figure 3

Figure 4.The leading-order outer problem for the velocity potentials $\hat{\unicode[STIX]{x1D719}}_{\pm 0}$. We have integrated the dynamic boundary conditions with respect to time and applied the leading-order initial condition to obtain the boundary conditions $\hat{\unicode[STIX]{x1D719}}_{\pm 0}=0$ on the free surfaces and $\hat{\unicode[STIX]{x1D719}}_{+0}=\hat{\unicode[STIX]{x1D719}}_{-0}$ on the interface. The problem is supplemented by the far-field conditions that $\hat{\unicode[STIX]{x1D719}}_{+0}\sim -{\hat{y}}$ as $\hat{x}^{2}+{\hat{y}}^{2}\rightarrow \infty$ and $\hat{\unicode[STIX]{x1D719}}_{-0}\sim V{\hat{y}}$ as $\hat{x}^{2}+{\hat{y}}^{2}\rightarrow \infty$. The free surfaces have initial profiles given by ${\hat{h}}_{+0}=\hat{x}^{2}/2$ and ${\hat{h}}_{-0}=-\hat{x}^{2}/(2R)$ respectively. This problem is co-dimension two in the sense that $\hat{d}_{0}(\hat{t})$ must be solved for as part of the problem.

Figure 4

Figure 5.The leading-order jet-root curve for $\unicode[STIX]{x1D6FF}=0.1$ and various values of $R$ and $V$. Note that $V=0$ corresponds to a stationary lower droplet and $V=1$ to symmetric impact speeds, while $1/R=1$ represents identical droplets and $1/R\rightarrow 0$ allows the lower droplet to approach a half-space.

Figure 5

Figure 6.The leading-order inner problem for the velocity potentials, $\unicode[STIX]{x1D6F7}_{\pm 0}$, upper and lower free surfaces $H_{\pm 0}$ and vortex sheet $\unicode[STIX]{x1D6E5}_{0}$. Dividing streamlines in each fluid are denoted by the dashed lines. They meet on the vortex sheet at a stagnation point. The far-field jet thickness is denoted by $H_{j}(\hat{t})$.

Figure 6

Figure 7.Direct numerical simulation set-up, with the background image illustrating the finite computational domain at $t=0$ for a test case described by $R=2$ and $V=1$. The inset focuses on details of the grid refinement at later times ($t=0.05$, once impact has taken place and the jet is formed), when temporal and spatial adaptive criteria based on changes in velocity, vorticity and interfacial location are imposed.

Figure 7

Figure 8.Horizontal velocity $U(x,y)$, vertical velocity $V(x,y)$ and vorticity $\unicode[STIX]{x1D714}(x,y)$ inside a selected jet-root region for an example solution characterised by $R=2$, $V=1$ at dimensionless time $t=0.1$. (a) Liquid–liquid interfaces are shown, while the inset depicts the region of interest illustrated in the figures on the right-hand side. Scale bars corresponding to $10~\unicode[STIX]{x03BC}\text{m}$ are present on the lower right-hand corner of (b–d).

Figure 8

Figure 9.Comparison between analytical prediction (3.21) (lines) and numerical results (symbols) for two families of tests. (a) Relative impact velocity coincides ($V=1$), while $R$ is varied from a case with identical drop size to the limiting case $R\rightarrow \infty$, where a drop impacts a liquid pool. (b) The relative drop size is set to 1 and the relative impact velocity $V$ is varied from 0.0 (stationary drop) to 4.0.

Figure 9

Figure 10.Jet-root thickness $J^{\ast }$ evolution in time for three different cases in $R$ and $V$. The symbols represent data from direct numerical simulations, while the underlying lines present evolutions proportional to $(t^{\ast })^{a}$ that best fit the early stages of the flow. The solid line corresponds to $a=3/2$ (the Wagner prediction), while the dashed line indicates linear growth in time as given by $a=1$ (the inviscid prediction of Josserand & Zaleski (2003)).

Figure 10

Figure 11.Summary of instantaneous jet angle $\unicode[STIX]{x1D703}_{jet}$ evolution for two families of tests. (a) Relative impact velocity coincides ($V=1$), while $R$ is varied from a case with identical drop sizes to the limiting case $R\rightarrow \infty$, where a drop impacts a liquid pool. (b) The relative drop size is set to 1 and the relative impact velocity $V$ is varied from 0 (stationary drop) to 4.

Figure 11

Figure 12.Summary of the jet-root evolution for the case described by $R=1$ and$V=2$. (a) The identified lower and upper turnover points along with the associated jet-root coordinates, with the symbols denoting the numerically obtained results and the continuous line illustrating the analytical prediction as given by (3.21). (b) The inviscid estimate of the jet angle, $\unicode[STIX]{x1D6FC}$ previously derived as expression (3.24). (c) Instantaneous jet angle $\unicode[STIX]{x1D703}_{jet}$ obtained numerically using distance criteria based on including either 1, 4 or 16 virtual interface points ahead of the jet root in order to calculate the angle the jet makes with the horizontal axis.

Figure 12

Figure 13.Summary of the jet-root evolution for the case described by $R=3$ and $V=0$. (a) The identified lower and upper turnover points along with the associated jet-root coordinates, with the symbols denoting the numerically obtained results and the continuous line illustrating the analytical prediction as given by (3.21). (b) Jet-root region angle $\unicode[STIX]{x1D6FC}$, with the analytical result previously derived as expression (3.24). (c) Instantaneous jet angle $\unicode[STIX]{x1D703}_{jet}$ obtained numerically using distance criteria based on including either 1, 4 or 16 virtual interface points ahead of the jet root in order to calculate the angle the jet makes with the horizontal axis.

Figure 13

Figure 14.(a) Evolution of the jet for $R=2$ and $V=1$ from the time of its creation at $t^{\ast }=2.4\times 10^{-6}\,\text{s}$ to $t^{\ast }=1.24\times 10^{-5}\,\text{s}$ in increments of $1.0\times 10^{-6}\,\text{s}$. The full shape up to the jet tip is extracted from the direct numerical simulations, while the coordinates of the jet root as obtained analytically are also presented. (b) Dimensional horizontal velocity $U^{\ast }(x^{\ast },t^{\ast })$ for each time step and comparison to the leading-order Wagner solution (3.34). Both a direct comparison (dotted line) and a simple correction to account for the presence of the entrapped gas bubble (dashed line) are illustrated.

Figure 14

Figure 15.Evolution of the jet thickness for the case described by $R=2$ and$V=1$ illustrated at five instances in time, from $t^{\ast }=6.7\times 10^{-6}$  s to $t^{\ast }=1.47\times 10^{-5}$  s in increments of $2.0\times 10^{-6}$  s. The grey symbols indicate results extracted from direct numerical simulations, while the solid lines represent the associated analytical predictions as given by expression (3.34). The leading-order jet thickness as given by analysing the inner region, previously derived as (3.29), is also presented in each case using white squares.