Hostname: page-component-7b7567c8c-tnkh4 Total loading time: 0 Render date: 2026-11-16T16:19:20.680Z Has data issue: false hasContentIssue false

The flow field due to a sphere moving in a viscous, density-stratified fluid

Published online by Cambridge University Press:  08 January 2025

Ramana Patibandla
Affiliation:
Department of Applied Mechanics and Biomedical Engineering, Indian Institute of Technology Madras, Chennai 600036, India
Anubhab Roy
Affiliation:
Department of Applied Mechanics and Biomedical Engineering, Indian Institute of Technology Madras, Chennai 600036, India
Ganesh Subramanian*
Affiliation:
Engineering Mechanics Unit, Jawaharlal Nehru Center for Advanced Scientific Research, Bangalore 560064, India
*
†Email address for correspondence: sganesh@jncasr.ac.in

Abstract

In this paper, we study the disturbance velocity and density fields induced by a sphere translating vertically in a viscous density-stratified ambient. Specifically, we consider the limit of a vanishingly small Reynolds number $(Re = \rho U a/\mu \ll 1)$, a small but finite viscous Richardson number $(Ri_v = \gamma a^3g/\mu U\ll 1)$ and large Péclet number $(Pe = Ua/D\gg 1)$. Here, $a$ is the sphere's radius, $U$ its translational velocity, $\rho$ an appropriate reference density within the framework of the Boussinesq approximation, $\mu$ the ambient viscosity, $\gamma$ the absolute value of the background density gradient, g is acceleration due to gravity and $D$ the diffusivity of the stratifying agent. For the scenario where buoyancy forces first become comparable to viscous forces at large distances, corresponding to the Stokes-stratification regime defined by $Re \ll Ri_v^{1/3} \ll 1$ for $Pe \gg 1$, important flow features have been identified by Varanasi & Subramanian (J. Fluid Mech., vol. 949, 2022, A29) – these include a vertically oriented reverse jet, and a horizontal axisymmetric wake, on scales larger than the primary (stratification) screening length of ${O}(aRi_v^{-1/3})$. Here, we study the reverse-jet region in more detail, and show that it is only the central portion of a columnar structure with multiple annular cells concentric about the rear stagnation streamline. In the absence of diffusion, corresponding to $Pe = \infty$ $( \beta _\infty = Ri_v^{1/3}Pe^{-1} = 0)$, this columnar structure extends to downstream infinity with the number of annular cells diverging in this limit. We provide expressions for the boundary of the structure, and the number of cells within, as a function of the downstream distance. For small but finite $\beta _\infty$, two length scales emerge in addition to the primary screening length – a secondary screening length of ${O}(aRi_v^{-1/2}Pe^{1/2})$ where diffusion starts to smear out density variations across cells, leading to exponentially decaying density and velocity fields; and a tertiary screening length, $l_t \sim {O}(aRi_v^{-1/2}Pe^{1/2}[\zeta + \frac {13}{4}\ln {\zeta } + ({13^2}/{4^2})({\ln \zeta }/{\zeta })])$ with $\zeta = \frac {1}{2}\ln ({\sqrt {{\rm \pi} }Ri_v^{-1}Pe^3}/{2160})$, beyond which the columnar structure ceases to exist. The latter causes a transition from a vertical to a predominantly horizontal flow, with the downstream disturbance fields reverting from an exponential to an eventual algebraic decay, analogous to that prevalent at large distances upstream.

Information

Type
JFM Papers
Copyright
© The Author(s), 2025. Published by Cambridge University Press
Figure 0

Figure 1.A schematic (not to scale) of the disturbance flow fields due to a translating sphere (of radius $a$) in different physical scenarios. (a) Potential flow, (b) Stokes flow, (c) in the presence of weak fluid inertia and (d) with a weak background stratification. Here, $l_o = {O}(a Re^{-1})$, the Oseen length, is the scale at which inertial forces become comparable to viscous ones, and $l_c = {O}(a Ri_v^{-1/3})$, the stratification screening length, is the scale at which buoyancy forces become comparable to viscous ones (for large $Pe$). The focus of the current work is the flow field beyond $l_c$ in (d). The definitions of the non-dimensional numbers Re, Ri and Pe are as given in the fourth paragraph of § 1.

Figure 1

Figure 2.(a) Typical sizes and speeds of different groups/classes of marine particulate matter: Pseudoalteromonas haloplanktis (Wagner, Young & Lauga 2014), Chlamydomonas reinhardtii (More & Ardekani 2020), marine snow (relevant range of values indicated by a purple box, see Turner 2002), Volvox carteri (Drescher et al.2009), Diaptomus minutus (Malkiel et al.2003), Daphnia magna (Noss & Lorke 2014), Artemia salina (Houghton et al.2018), Euphasia pacifica (Kunze et al.2006) and Mastigias sp. (Katija & Dabiri 2009). The blue and red lines correspond to constant Reynolds and viscous Richardson numbers, respectively. Here, $U \propto a^{-1}$ and $U\propto a^3$ along the blue and red lines, respectively. Note that U and a are characteristic speed and size, of the marine particulate matter, respectively. (b) A classification of the marine particulate matter in (a), in dimensionless terms, on a parameter plane comprising the ratio of the stratification screening length to particle size ($l_c/a$) and the ratio of Oseen length scale to the particle size ($l_o/l_a$); refer to third paragraph of § 1 for definitions of $l_c$ and $l_o$.

Figure 2

Table 1.The parameter regimes considered in the previous literature. Here, $Re$, $Ri$ and $Pe$ are the Reynolds number, viscous Richardson number and Péclet number, respectively, and are defined in the third paragraph of § 1. The flow regime considered by each study is given in the fifth column: the Stokes-stratification regime corresponds to $l_c \ll l_o$ and the inertia-stratification regime to $l_c \gg l_o$. Here, $l_c$ is the stratification screening length and $l_o$ is the Oseen length (see para 2, § 1). The focus of the different studies can be classified broadly into the flow-field description, drag calculation, drift-volume calculation and mixing efficiency estimation. The nature of the work is also indicated via superscripts $A$ (analytical), $N$ (numerical) or $E$ (experimental).

Figure 3

Table 2.The large-$Pe$ wake region asymptotes; the wake region has a self-similar structure with the similarity variable defined by $\tilde {\eta } = \tilde {z}/\tilde {r}_t^{2/5}$.

Figure 4

Figure 3.Schematics of the flow field induced by a sphere translating vertically in a viscous density-stratified ambient. Panel (a) is based on VS22, and (b) is based on the present study. In both figures, the primary screening length appears as a dashed circle with radius $l_c \sim {O}(aRi_v^{-1/3})$, and diffusion smears out transverse density variations, downstream of the sphere, at distances beyond the secondary screening length, $l_s \sim {O}(aRi_v^{-1/2}Pe^{1/2})$. The tertiary screening length, $l_t \sim {O}(aRi_v^{-1/2}Pe^{1/2}[\zeta + \frac {13}{4}\ln {\zeta } + ({13^2}/{4^2})({\ln \zeta }/{\zeta })])$ with $\zeta = \frac {1}{2}\ln ({\sqrt {{\rm \pi} }Ri_v^{-1}Pe^3}/{2160})$, marks the length of the columnar structure on the right.

Figure 5

Figure 4.The downstream axial velocity disturbance ($\tilde {u}_z$) and streamline pattern for $\beta _\infty = 0$. Panel (a) shows $|\tilde {u}_z|$ plotted as a function of $\tilde {z}\,({>}0)$. The numerical results (bright-coloured solid curves), obtained from evaluating the inverse transform integral in (2.11), are compared withthe asymptotic approximation derived from (3.6) (pale-coloured dashed curves with dots added for better differentiation) for different non-zero $\tilde {r}_t$. The reverse-Stokeslet decay for $\tilde {r}_t = 0$ is shown as a solid grey line, while the (common) algebraic asymptote, $3240/\tilde {z}^7$, appears as a dashed grey line. Panel (b) shows the streamline pattern for $\tilde {z} > 1$, depicting the columnar structure behind the translating sphere. The boundary of the columnar structure, which is the red curve connecting the ends of the cell boundaries (black curves) above the wake region ($\tilde {z} \gtrsim 100$), is given by (3.8).

Figure 6

Figure 5.The absolute value of the axial velocity ($|\tilde {u}_z|$) (a) along the rear stagnation streamline; and (b) along the third (continuous), fifth (dashed) and seventh (dotted) zero crossings (of the Stokes streamfunction), for different $\beta _\infty$, including $\beta _\infty = 0$. For all non-zero $\beta _\infty$, the plots depict the transition from a reverse-Stokeslet decay ($3/2\tilde {z}$) to the far-field asymptote ($3240/\tilde {z}^7$) at large $\tilde {z}$. The secondary and tertiary screening lengths, marking the beginning and end of the exponential decay phase, are shown for each $\beta _\infty$.

Figure 7

Figure 6.Streamline patterns in an axial (half) plane for $\beta _\infty = (a)$$10$, (b) $1$, (c) $10^{-1}$, (d) $10^{-2}$, (e) $10^{-3}$ and (f) $10^{-4}$. The sphere is located at $(\tilde {r}_t,\tilde {z}) \equiv (0,0)$, with $\tilde {r}_t$ and $\tilde {z}$ being scaled by the primary screening length $l_c = {O}(aRi_v^{-1/3})$. Here, the dense blue bands mark the zero crossings of the Stokes streamfunction. The streamline patterns within the central regions, enclosed by the black dashed rectangles, are those originally obtained by VS22. The black continuous curves in (a,f) are wake boundaries corresponding to the small- and large$-Pe$ similarity solutions, respectively.

Figure 8

Figure 7.Streamline patterns in an axial (half) plane, and over an extended downstream region ($1 < \tilde {z} < 10^4$), showing the columnar structure for $\beta _\infty = (a)$$10^{-3}$, (b) $10^{-4}$, (c) $10^{-5}$ and (d) $0$. The sphere is at $(\tilde {r}_t,\tilde {z}) \equiv (0,0)$, with $\tilde {r}_t$ and $\tilde {z}$ being scaled using the primary screening length $l_c = {O}(aRi_v^{-1/3})$. The dense blue bands mark the zero crossings of the Stokes streamfunction; the region enclosed by black dashed rectangles correspond to the downstream patterns accessed in VS22. The inset in (b) is a zoomed-in version of the red dashed rectangle, showing fixed points at the end of recirculating cells that enable the transition from the cellular to a largely horizontal flow with increasing downstream distance. (e) Features of the columnar structures, for different non-zero $\beta _\infty$, compared with the limiting case of $\beta _\infty = 0$; horizontal dashed and dot-dashed lines indicate secondary($l_s$) and tertiary ($l_t$) screening lengths, respectively.

Figure 9

Figure 8.Isopycnal patterns in an axial (half) plane at $\beta _\infty = (a)$$10$, (b) $1$, (c) $10^{-1}$, (d) $10^{-2}$, (e) $10^{-2}$ and (f) $10^{-3}$. The sphere is located at the origin $(\tilde {r}_t,\tilde {z}) = (0,0)$. Here, $\tilde {r}_t$ and $\tilde {z}$ are non-dimensionalized using the primary screening length $l_c = {O}(aRi_v^{-1/3})$, and the dense bands correspond to the zero crossings (i.e. $\tilde {\rho }_f = 0$) of the density. The portions of the patterns in the region enclosed by the black dashed rectangles correspond to those obtained by VS22.

Figure 10

Figure 9.Isopycnal patterns in an axial half (plane) (and positive $\tilde {z}$) showing the downstream columnar structure at $\beta _\infty = (a)$$10^{-3}$, (b) $10^{-4}$, (c) $10^{-5}$ and (d) $0$. The sphere is located at the origin $(\tilde {r}_t,\tilde {z}) = (0,0)$. Both $\tilde {r}_t$ and $\tilde {z}$ are non-dimensionalized using the primary screening length $l_c = {O}(aRi_v^{-1/3})$. Here, the dense bands correspond to the zero crossings of perturbation density (i.e. $\tilde {\rho }_f = 0$). The isopycnal patterns in the regions enclosed by the black dashed rectangles are those obtained by VS22. (e) The columnar structures at various non-zero $\beta _\infty$, compared with the case of $\beta _\infty = 0$. The horizontal dashed and dot-dashed lines indicate secondary ($l_s$) and tertiary ($l_s$) screening lengths for different $\beta _\infty$.

Figure 11

Figure 10.(a) Schematic of the drift-volume calculation methodology, illustrating the total drift-volume evaluation as the sum of upstream and downstream components. (b) The initial locations of two material points (black stars) superimposed on the streamline pattern for $\beta _\infty = 0$; the red lines indicate the disturbance flow field encountered by these points during the sphere's translation. The non-dimensional drift displacements ($z/a$) of (c) point 1 and (d) point 2, plotted as a function of time ($tU/a$), for $Ri_v = 10^{-4}$. The red curve is the downstream drift displacement resulting from the forward integration, and the blue curve denotes the upstream displacement resulting from the backward integration; the corresponding dashed curves are the drift displacements in the absence of stratification. Black dashed lines mark the time when the points cross the stratification screening length $l_c$.

Figure 12

Figure 11.(a) Streamline and (b) isopycnal patterns in an axial (half) plane for $\beta _\infty = 10^{-5}$ (for $-14<\tilde {r}_t<0$) and $10$ (for $0<\bar {r}_t<14$); only the downstream region, corresponding to $\tilde {z} > 1$, is shown. Please note that the radial and axial coordinates for $\beta _\infty = 10^{-5}$ are scaled by the large-$Pe$ stratification screening length of ${O}[a(Ri_v)^{-1/3}]$, whereas those for $\beta _\infty = 10$ are scaled by the small-$Pe$ screening length of ${O}[a(Ri_vPe)^{-1/4}]$.

Figure 13

Figure 12.Contours of integration used in solving the $k$-integral in (A4). Here, (a) shows a contour in the upper half of the complex$-k$ plane, used when the argument of the complex exponentials in (A4) is positive (case I), whereas (b) shows a contour in the lower half of the complex$-k$ plane used when the argument of the complex exponentials in (A4) is negative (case II). The parameters $C_{L1}-C_{L4}$ are integrals along the real line, $R$ is the radius of the curves $C_{R2}$ and $C_{R4}$ and $\epsilon$ is the radius of curves $C_{R1}$ and $C_{R3}$. Here, $\kappa _0$ to $\kappa _4$ are zeros of the denominator in the integrand of (A4).

Figure 14

Figure 13.The behaviour of (a) first and second roots, and (b) third and fourth roots of the quartic polynomial ($\beta _\infty k^4+{\rm i}k^3\cos \theta + \sin ^2\theta$), as a function of $\theta$ and for various $\beta _\infty$, in the complex$-k$ plane. Here, $\theta$ varies from $0$ to ${\rm \pi} /2$ and increases in the direction shown by the arrows. The continuous and dashed lines correspond to root 1 (3) and 2 (4) in panel (a) (panel b), respectively.

Figure 15

Figure 14.Constant phase curves in the complex-$\hat {\lambda }$ plane from expression (B7). The blue, red and green curves (both dashed and continuous) correspond to saddle points 1, 2 and 3, respectively; however, we consider only the continuous curves. The red and blue continuous curves are the steepest-descent curves, whereas the green curve is the steepest-ascent curve.

Figure 16

Figure 15.(a) Comparison of the columnar structure boundary obtained in Appendix B.2 with the boundary drawn from ends of different recirculating cells in the streamline plot of $\beta _\infty = 0$ (blue curve). The red and black dashed curves are from the single and two term asymptotic solution from (B16) and (B20), respectively. (b) Comparison of the tertiary screening length obtained in Appendix C and that from numerically solving (C1) (blue curve). The red, green and black curves are from the single-, two- and three-termasymptotic solutions given in (C4), (C6) and (C7), respectively.