Hostname: page-component-7b7567c8c-z4zkx Total loading time: 0 Render date: 2026-11-16T02:59:07.200Z Has data issue: false hasContentIssue false

The effect of electrostatic charges on particle-laden duct flows

Published online by Cambridge University Press:  29 December 2020

Holger Grosshans*
Affiliation:
Physikalisch-Technische Bundesanstalt (PTB), Braunschweig, GermanyInstitute of Apparatus- and Environmental Technology, Otto von Guericke University of Magdeburg, Germany
Claus Bissinger
Affiliation:
Physikalisch-Technische Bundesanstalt (PTB), Braunschweig, Germany
Mathieu Calero
Affiliation:
Institute of Mechanics, Materials and Civil Engineering, Université catholique de Louvain, Louvain-la-Neuve, Belgium
Miltiadis V. Papalexandris
Affiliation:
Institute of Mechanics, Materials and Civil Engineering, Université catholique de Louvain, Louvain-la-Neuve, Belgium
*
†Email address for correspondence: holger.grosshans@ptb.de

Abstract

We report on direct numerical simulations of the effect of electrostatic charges on particle-laden duct flows. The corresponding electrostatic forces are known to affect the particle dynamics at small scales and the associated turbophoretic drift. Our simulations, however, predicted that electrostatic forces also dominate the vortical motion of the particles, induced by the secondary flows of Prandtl's second kind of the carrier fluid. Herein, we treated flows at two frictional Reynolds numbers ($Re_{\tau }= 300$ and 600), two particle-to-gas density ratios ($\rho _{p}/\rho =1000$ and 7500) and three Coulombic-to-gravitational force ratios ($F_{el}/F_{g}=0$, 0.004 and 0.026). In flows with a high density ratio at $Re_{\tau }=600$ and $F_{el}/F_{g}=0.004$, the particles tend to accumulate at the walls. On the other hand, at a lower density ratio, respectively a higher $F_{el}/F_{g}$ of 0.026, the charged particles still follow the secondary flow structures that are developed in the duct. However, even in this case, the electrostatic forces counteract the particles’ inward flux from the wall and, as a result, their vortical motion in these secondary structures is significantly attenuated. This change in the flow pattern results in an increase of the particle number density at the bisectors of the walls by a factor of five compared with the corresponding flow with uncharged particles. Finally, at $Re_{\tau }=300$, $\rho _{p}/\rho =1000$ and $F_{el}/F_{g}=0.026$ the electrostatic forces dominate over the aerodynamic forces and gravity and, consequently, the particles no longer follow the streamlines of the carrier gas.

Information

Type
JFM Papers
Copyright
© The Author(s), 2020. Published by Cambridge University Press
Figure 0

Figure 1.Symmetry planes of the duct and streamlines of induced Prandtl's secondary flow of the second kind. The arrows indicate the direction of the possible particle motion in the cross-section following the gas phase.

Figure 1

Figure 2.Sketch of the parameters used in the ray casting collision detection approach in the rest frame of particle 2.

Figure 2

Figure 3.(a) Graphic representation and dimensions of the flow domain. The arrow points in the direction of the flow. (b) Computational mesh in the cross-section, consisting of $60 \times 60$ cells.

Figure 3

Table 1.Parameters of the numerical simulations.

Figure 4

Figure 4.Temporal evolution of the normalized particle number density $C$ near the walls: (a) $Re_{\tau }=600$, $\rho _{p}/\rho = 1000$, $F_{el}/F_{g}= 0$; (b) $Re_{\tau }=600$, $\rho _{p}/\rho = 1000$, $F_{el}/F_{g}= 0.026$; (c) $Re_{\tau }=300$, $\rho _{p}/\rho = 1000$, $F_{el}/F_{g}= 0$; (d) $Re_{\tau }=300$, $\rho _{p}/\rho = 1000$, $F_{el}/F_{g}= 0.026$; (e) $Re_{\tau }=600$, $\rho _{p}/\rho = 7500$, $F_{el}/F_{g}= 0$; (f) $Re_{\tau }=600$, $\rho _{p}/\rho = 7500$, $F_{el}/F_{g}= 0.004$; (black solid line) $y^+$ or $z^+$$<1$ for $Re_{\tau }= 600$ and $y^+$ or $z^+$$<0.5$ for $Re_{\tau }= 300$: (blue dashed line) $y^+$ or $z^+$$<5$ for $Re_{\tau }= 600$ and $y^+$ or $z^+$$<2.5$ for $Re_{\tau }= 300$; (green dotted line) $y^+$ or $z^+$$<50$ for $Re_{\tau }= 600$ and $y^+$ or $y^+$$<50$ for $Re_{\tau }= 300$.

Figure 5

Figure 5.Instantaneous particle positions in the cross-section of the duct ($y$–$z$ plane), recorded once the flow has become statistically stationary: (a) $Re_{\tau }=600$, $\rho _{p}/\rho =1000$, $F_{el}/F_{g}= 0$; (b) $Re_{\tau }=600$, $\rho _{p}/\rho =1000$, $F_{el}/F_{g}= 0.026$; (c) $Re_{\tau }=300$, $\rho _{p}/\rho =1000$, $F_{el}/F_{g}=0$; (d) $Re_{\tau }=300$, $\rho _{p}/\rho =1000$, $F_{el}/F_{g}= 0.026$; (e) $Re_{\tau }=600$, $\rho _{p}/\rho =7500$, $F_{el}/F_{g}=0$; (f) $Re_{\tau }=600$, $\rho _{p}/\rho =7500$, $F_{el}/F_{g}=0.004$. The colours indicate the streamwise velocity of the particles; for each case, the red colour corresponds to the fastest and the blue colour to the slowest particle. For visualization purposes the particle size is scaled up and only one in every five particles is shown.

Figure 6

Figure 6.Mean particle concentration in arbitrary units, $\langle C \rangle$, in the cross-section of the duct: (a) $Re_{\tau }=600$, $\rho _{p}/\rho =1000$, $F_{el}/F_{g}=0$; (b) $Re_{\tau }=600$, $\rho _{p}/\rho =1000$, $F_{el}/F_{g}= 0.026$; (c) $Re_{\tau }=300$, $\rho _{p}/\rho =1000$, $F_{el}/F_{g}=0$; (d) $Re_{\tau }=300$, $\rho _{p}/\rho =1000$, $F_{el}/F_{g}= 0.026$; (e) $Re_{\tau }=600$, $\rho _{p}/\rho =7500$, $F_{el}/F_{g}=0$; (f) $Re_{\tau }=600$, $\rho _{p}/\rho =7500$, $F_{el}/F_{g}= 0.004$. Note that $\langle C \rangle$ is normalized by the maximum local concentration of each case, i.e. the vertical axes of the figures are scaled differently.

Figure 7

Figure 7.Mean normalized particle number density, $\langle C\rangle$, in four different slices: (a) $2.5 for $Re_{\tau }= 300$ and $5 for $Re_{\tau }= 600$; (b) $10 for $Re_{\tau }= 300$ and $20 for $Re_{\tau }= 600$; (c) $27.5$Re_{\tau }= 300$ and $55 for $Re_{\tau }= 600$; (d) $147.5 for $Re_{\tau }= 300$ and $295 for $Re_{\tau }= 600$. For visibility purposes, in (a) the horizontal axis is linear, whereas in (b)–(d) it is logarithmic. In each figure, the top horizontal axis corresponds to the flows at $Re_{\tau }= 300$ and the lower one to flows $Re_{\tau }= 600$; (black solid line) $Re_{\tau }= 600$, $\rho _{p} / \rho = 1000$, $F_{el}/F_{g}=0$; (black dashed line) $Re_{\tau }= 600$, $\rho _{p} / \rho = 1000$, $F_{el}/F_{g}= 0.026$; (red solid line) $Re_{\tau }= 300$, $\rho _{p} / \rho = 1000$, ${F_{el}/F_{g}=0}$; (red dashed line) $Re_{\tau }= 300$, $\rho _{p} / \rho = 1000$, $F_{el}/F_{g}= 0.026$; (blue solid line) $Re_{\tau }= 600$, $\rho _{p} / \rho = 7500$, $F_{el}/F_{g}=0$; (blue dashed line) $Re_{\tau }= 600$, $\rho _{p} / \rho = 7500$, $F_{el}/F_{g}= 0.004$.

Figure 8

Figure 8.Normalized mean particle velocity in a wall-normal direction in four different slices: (a) $2.5 for $Re_{\tau }= 300$ and $5 for $Re_{\tau }= 600$; (b) $22.5 for $Re_{\tau }= 300$ and $45 for $Re_{\tau }= 600$; (c) $35 for $Re_{\tau }= 300$ and $70 for $Re_{\tau }= 600$; (d) $147.5 for $Re_{\tau }= 300$ and $295 for $Re_{\tau }= 600$. In each panel, the top horizontal axis corresponds to the flows at $Re_{\tau }= 300$ and the lower one to the flows at $Re_{\tau }= 600$: (black solid line) $Re_{\tau }= 600$, $\rho _{p} / \rho = 1000$, $F_{el}/F_{g}=0$; (black dashed line) $Re_{\tau }= 600$, $\rho _{p} / \rho = 1000$, $F_{el}/F_{g}= 0.026$; (red solid line) $Re_{\tau }= 300$, $\rho _{p} / \rho = 1000$, ${F_{el}/F_{g}=0}$; (red dashed line) $Re_{\tau }= 300$, $\rho _{p} / \rho = 1000$, $F_{el}/F_{g}= 0.026$; (blue solid line) $Re_{\tau }= 600$, $\rho _{p} / \rho = 7500$, $F_{el}/F_{g}=0$; (blue dashed line) $Re_{\tau }= 600$, $\rho _{p} / \rho = 7500$, $F_{el}/F_{g}= 0.004$.

Figure 9

Figure 9.Instantaneous snapshots of the electric potential across the cross-section of the duct, induced by charges carried by the particles: (a) $Re_{\tau }=600$, $\rho _{p} / \rho =1000$; (b) $Re_{\tau }=300$, $\rho _{p} / \rho =1000$; (c) $Re_{\tau }=600$, $\rho _{p} / \rho =7500$.

Figure 10

Figure 10.Normalized mean particle velocity in the streamwise direction in four different slices as a function of the distance to the duct's wall: (a) $2.5 for $Re_{\tau }= 300$ and $5 for $Re_{\tau }= 600$; (b) $22.5 for $Re_{\tau }= 300$ and $45 for $Re_{\tau }= 600$; (c) $35 for $Re_{\tau }= 300$ and $70 for $Re_{\tau }= 600$; (d) $147.5 for $Re_{\tau }= 300$ and $295 for $Re_{\tau }= 600$. For visibility purposes, in (a) the horizontal axis is linear, whereas in (b–d) it is logarithmic. The upper axis in each figure relates to the cases of $Re_{\tau }= 300$ and the lower axis to the cases of $Re_{\tau }= 600$. In addition, in (d) the mean velocity of the undisturbed gaseous phase is displayed. (black solid line) $Re_{\tau }= 600$, $\rho _{p} / \rho = 1000$, $F_{el}/F_{g}=0$; (black dashed line) $Re_{\tau }= 600$, $\rho _{p} / \rho = 1000$, $F_{el}/F_{g}= 0.026$; (red solid line) $Re_{\tau }= 300$, $\rho _{p} / \rho = 1000$, $F_{el}/F_{g}=0$; (red dashed line) $Re_{\tau }= 300$, $\rho _{p} / \rho = 1000$, $F_{el}/F_{g}= 0.026$; (blue solid line) $Re_{\tau }= 600$, $\rho _{p} / \rho = 7500$, $F_{el}/F_{g}=0$; (blue dashed line) $Re_{\tau }= 600$, $\rho _{p} / \rho = 7500$, $F_{el}/F_{g}= 0.004$.

Figure 11

Figure 11.Scalability of the fluid solver and the fluid plus the particle and electrostatic solver. The straight dashed lone denotes the ideal linear scaling.

Figure 12

Figure 12.DNS of flow in a duct without particles: (a) $Re_{\tau } = 600$; (b) $Re_{\tau } = 300$. Comparison of the mean streamwise velocity at the centre plane of the duct with the DNS data of Huser & Biringen (1993) and Sharma & Phares (2006). The profiles for $Re_{\tau }= 600$ are computed on five different grid resolutions, as shown in the legend of the figure.

Figure 13

Figure 13.DNS of flow in a duct without particles: (a) $Re_{\tau } = 600$; (b) $Re_{\tau } = 600$; (c) $Re_{\tau } = 600$; (d) $Re_{\tau } = 300$; (e) $Re_{\tau } = 300$; (f) $Re_{\tau } = 300$. Comparison of the r.m.s. velocity components with the DNS data of Huser & Biringen (1993), Zhu, Yang & Chen (2009) and Sharma & Phares (2006). The profiles for $Re_{\tau }= 600$ are computed on five different grid resolutions, as shown in the legend of the figure. Note that Sharma & Phares (2006) did not provide data for $v^+_{rms}$.

Figure 14

Figure 14.Mean normalized particle number density, $\langle C\rangle$, replotted from figures 7(b) and 7(d) with additional cases considering a higher charge level: (a) $10 for $Re_{\tau }= 300$ and $20 for $Re_{\tau }= 600$; (b) $147.5 for $Re_{\tau }= 300$ and $295 for $Re_{\tau }= 600$; (black dotted line) $Re_{\tau }= 600$, $\rho _{p} / \rho = 1000$, $F_{el}/F_{g}=0.104$; (red dotted line) $Re_{\tau }= 300$, $\rho _{p} / \rho = 1000$, $F_{el}/F_{g}=0.104$; (blue dotted line) $Re_{\tau }= 600$, $\rho _{p} / \rho = 7500$, $F_{el}/F_{g}=0.017$.