Hostname: page-component-7b7567c8c-tnkh4 Total loading time: 0 Render date: 2026-11-16T06:15:49.171Z Has data issue: false hasContentIssue false

Sedimenting-particle redistribution in a horizontal Couette

Published online by Cambridge University Press:  17 January 2024

M. Davoodi*
Affiliation:
Schlumberger Cambridge Research, High Cross, Madingley Road, Cambridge CB3 0HE, UK
A. Clarke
Affiliation:
Schlumberger Cambridge Research, High Cross, Madingley Road, Cambridge CB3 0HE, UK
*
†Email address for correspondence: MDavoodi@slb.com

Abstract

Flow between axially rotating concentric cylinders is well known to exhibit rich dynamics. Hence, Taylor instabilities have been studied, both experimentally and theoretically, for many years. Although usually studied in the abstract, such geometries arise in a range of practical situations including drilling, when a drilling fluid flow enters a well via a pipe that is the centre body and returns via the annulus between the pipe and the borehole wall. In drilling, the centre body rotates and the annular flow contains rock cuttings. Here, we report the development of an Eulerian-Eulerian solver, based on OpenFOAM, that solves for this cuttings transport problem in the presence of both gravity and Taylor vortices. To check the reliability of the solver, we conduct a set of experiments spanning a wide range of complex flow regimes. We show that the model successfully predicts, in all regimes, the observed complex redistribution of particulates. However, for suspension flows under viscously dominated conditions, high particle concentrations and in rectilinear flow, particle pressure and normal stress differences are sufficient to capture particle migration. Results show that in more complex flows exemplified by the Taylor–Couette flow studied here, more realistic predictions of non-Brownian particle migration require inclusion of forces arising through the relative velocity of the two phases including lift forces originating both from inertia and particle rotation.

Information

Type
JFM Papers
Copyright
© Schlumberger Cambridge Research Limited, 2024. Published by Cambridge University Press
Figure 0

Figure 1.Schematic of the Taylor–Couette geometry used for the numerical simulation. Cyclic boundary conditions are applied to the axial faces of the domain. The inner wall rotates and the outer is stationary. (a) Elevation defining dimensions, (b) perspective of the 3-D domain.

Figure 1

Figure 2.(a) Schematic layout of the flow loop for particle transport measurement with X-ray shadowgraphy. (b) Schematic cross-section of particle loader B in particle load position, (c) cross-section of particle loader B rotated to allow particles to fall into the Couette. Experiments are performed with the device in the position of panel (b) so that the concentric cylinders are unimpeded.

Figure 2

Figure 3.(a) X-ray imaging layout, (b) construction diagram to estimate X-ray absorption.

Figure 3

Figure 4.Measured mixing parameter for two average volume fractions (ϕave = 1.8 % and ϕave = 3.6 %) in aqueous glycerol (viscosity 17.7 mPa s).

Figure 4

Figure 5.Space-rotation diagram for the evolution of ϕave = 3.6 % particles in 17.75 mPa s aqueous glycerol. (a) Location of lines used to generate space-rotation plots, (b) bottom, (c) top. Here initiation of the redistribution of particles starts at approximately 170 rpm. Thereafter, the structure is remarkably stationary. Equivalent Ta numbers are indicated at the lower edge of each plot.

Figure 5

Figure 6.Still X-ray images obtained during the rotation rate sweep of ϕave = 10.8 % particles in aqueous 21.3 mPa s aqueous glycerol. Images are labelled with rotation rate in rpm and are averages of 10 frames. White indicates the presence of particles. Inset table shows corresponding dimensionless numbers: Taylor (Ta), Froude (Fr), Reynolds (Re) and revolution count (N). Panels (a) and (b) have brightness increased so that initial particle mixing can be seen.

Figure 6

Figure 7.Time-space diagrams of 2 s of video at 650 rpm (Ta = 187) for ϕave = 10.8 % particles in 21.3 mPa s aqueous glycerol: (a) x−y space elevation; (b) y-time at the position of the vertical yellow line in panel (a); (c) x-time at the position of the horizontal yellow line in panel (a). In panel (c), there can be seen a periodic perturbation along the time axis suggesting wavy vortex flow.

Figure 7

Figure 8.Mixing parameter plotted as a function of Taylor number for ${\varPhi _{ave}} = 3.6\,\%$. Experimental data from figure 4.

Figure 8

Figure 9.Projection of the numerical domain showing the solid bed distribution viewed from the underside of the Couette at different Taylor numbers for: (a) before the start point of the Taylor instability and mixing process at Ta = 7.5; (b) beginning of the Taylor vortex instability transition at Ta = 52 and (c) higher rotation rate at Ta = 90 when Taylor instability is established.

Figure 9

Figure 10.Solids distribution (greyscale) at sequential rotation speeds with superimposed streamlines of the fluid flow: (a) before the start point of the Taylor instability and the mixing process at Ta = 7.5; (b) beginning of the Taylor vortex instability transition at Ta = 52 and (c) higher rotation rate at Ta = 90, when the Taylor instability is established, at the plane z = 0.

Figure 10

Figure 11.Streamlines superimposed on (a) concentration distribution parameter and (b) drag force for Ta = 90 in the presence of Taylor vortex. Note the panels are taken from the bottom section of the Couette cell.

Figure 11

Figure 12.Formation of ring structures in the presence of Taylor vortices at (a) Ta = 60, (b) Ta = 70, (c) T = 90. Note here the isovolumetric concentration contours are set with different opacities to provide a better comparison with experimental results.

Figure 12

Figure 13.Formation of ring structures ((a–c) the top of the Couette and (d–f) the bottom of the Couette) as a function of rotation speed at (a,d) Ta = 60, (b,e) Ta = 70, (c,f) Ta = 90. Plots are cross-sections in the x–y plane at z = 0, −2 < x < 2 from (a–c) 4 < y < 5 and (d–f) −4 < y < −5.

Figure 13

Figure 14.Distributions of solid particles ((a) for sedimenting particles with $\gamma = 2.09$, correspondingly, and (c) for neutrally buoyant cases with $\gamma = 1$) and Archimedes force ((b) for sedimenting particles with $\gamma = 2.09$ and (d) for neutrally buoyant cases with $\gamma = 1$, correspondingly) at $Ta = 90$ at the (i) top and (ii) bottom of the Couette geometry.

Figure 14

Figure 15.(a) Variation of the mixing parameter in time for Ta = 30 and Ta = 200. Inset shows an expanded view of the long-time oscillatory behaviour for Ta = 200. (b) Particle distribution at $Ta = 200$ in the presence of wavy vortex flows; the pattern is time dependent (Supplementary material and movie 2).

Figure 15

Figure 16.Solid distribution at Ta = 90. Plots are in x–y plane at z = 0, −2 < x < 2 from (a–c) 4 < y < 5 and (d–f) −4 < y < −5. (a) and (d) In the presence of lift forces ${F_{Saff}}$ and ${F_{Mag}}$ as explained in (2.15) and (2.17), (b) and (e) using lift forces $F_{Saff}^\angle $ and ${F_{Mag}}$ as formulated in (2.16) and (2.17), (c) and (f) in the absence of any lift forces.

Figure 16

Figure 17.Variation of (a) relative velocity magnitude in vorticity direction, (b) Saffman lift force magnitude, (c) Magnus lift force magnitude and (d) total drag force magnitude at Ta = 90 for (i) sedimenting particles $\gamma = 2.09$ and (ii) neutrally buoyant particles $\gamma = 1$ plotted in the x–y plane at z = 0, −2 < x < 2 from (a–c) 4 < y < 5, i.e. the top of the geometry.

Figure 17

Figure 18.(a) Schematic layout of optical camera for this sequence of optical images. In the lower part of each image, we see the underside of the Couette, note the left-right asymmetry across the images created by imaging parallax. (b) An initially uniform bed of particles is perturbed by the shear flow induced by rotating the inner cylinder of the Couette (rotation speed in rpm indicated). (c) Approach to period doubling of particle banks seen at 230 rpm, Ta = 1188 in panel (b). Here, ϕave = 10.8 % in water.

Figure 18

Figure 19.(a) Optical image of two banks with indicated period, λΒ, showing shark-tooth appearance and viewed from underneath the particle bed. The fluid (water) flow is directed up the page in this image, i.e. the double end of the bank is in the direction of rotation of the Couette inner cylinder. (b) Measured periodicity of the particle banks divided by gap as a function of rotation rate and Taylor number.

Figure 19

Figure 20.(a) Data from figure 4 plot together with mixing data for a volume fraction ϕave = 10.8 % of particles in water. (b) Data plotted against a combined Froude–Shields number with n = 125 (see text).

Supplementary material: File

Davoodi and Clarke supplementary movie 1

Particle distribution in aqueous glycerol as a function of rotation rate (inset) Figure 7 dataset. The background derived from the fluid only filled couette is subtracted, the image inverted so that particles appear white and the contrast/brightness adjusted for visual enhancement.
Download Davoodi and Clarke supplementary movie 1(File)
File 8 MB
Supplementary material: File

Davoodi and Clarke supplementary movie 2

Numerically generated video of wavey vortex evolution at 644rpm, Ta=202 and a particle volume concentration of 3.6%.
Download Davoodi and Clarke supplementary movie 2(File)
File 4.6 MB
Supplementary material: File

Davoodi and Clarke supplementary material 3

Davoodi and Clarke supplementary material
Download Davoodi and Clarke supplementary material 3(File)
File 716.8 KB