Hostname: page-component-7b7567c8c-tnkh4 Total loading time: 0 Render date: 2026-11-16T07:58:52.893Z Has data issue: false hasContentIssue false

The effect of turbulence on mass transfer rates between inertial polydisperse particles and fluid

Published online by Cambridge University Press:  15 July 2019

Ewa Karchniwy*
Affiliation:
Department of Energy and Process Engineering, Norwegian University of Science and Technology, Kolbjørn Hejes vei 1B, 7491 Trondheim, NorwayInstitute of Thermal Technology, Silesian University of Technology, Konarskiego 22, 44-100 Gliwice, Poland
Adam Klimanek
Affiliation:
Institute of Thermal Technology, Silesian University of Technology, Konarskiego 22, 44-100 Gliwice, Poland
Nils Erland L. Haugen
Affiliation:
Department of Energy and Process Engineering, Norwegian University of Science and Technology, Kolbjørn Hejes vei 1B, 7491 Trondheim, NorwaySINTEF Energi AS, Sem Saelands vei 11, 7034 Trondheim, Norway
*
†Email address for correspondence: ewa.karchniwy@polsl.pl

Abstract

The current work investigates how turbulence affects the mass transfer rate between inertial particles and fluid in a dilute, polydisperse particle system. Direct numerical simulations are performed in which all scales of turbulence are fully resolved and particles are represented in a Lagrangian reference frame. The results show that, similarly to a monodisperse system, the mass transfer rate between particles and fluid decreases as a result of particle clustering. This occurs when the flow time scale (based on the turbulence integral scale) is long relative to the chemical time scale, and is strongest when the particle time scale is one order of magnitude smaller than the flow time scale (i.e. the Stokes number is around 0.1). It is also found that for larger solid mass fractions, the clustering of the heavier particles is enhanced by the effect of drag force from the particles on the fluid (momentum back-reactions or two-way coupling). In particular, when two-way coupling is accounted for, locations of particles of different sizes are much more correlated, which leads to a stronger effect of clustering, and thus a greater reduction of the particle–fluid mass transfer rate.

Information

Type
JFM Papers
Copyright
© 2019 Cambridge University Press 
Figure 0

Figure 1.Kinetic energy spectrum; black: $Re_{\unicode[STIX]{x1D706}}\approx 350$, red: $Re_{\unicode[STIX]{x1D706}}\approx 150$ (cases A and M in table 1). The wavenumbers $k_{L}$, $k_{\unicode[STIX]{x1D706}}$ and $k_{\unicode[STIX]{x1D702}}$ correspond to the integral, Taylor and Kolmogorov scales, respectively.

Figure 1

Table 1.Simulation parameters. The total number of grid points varies between $64^{3}$ and $512^{3}$, depending on $Re_{\unicode[STIX]{x1D706}}$ and $Da$. In all cases, the Schmidt number is equal to 0.2. The parameter $Da_{2}$ quantifies the effect of particle clustering and is explained in a later sub-section.

Figure 2

Figure 2.Particle size distributions used in the current work (for $Re_{\unicode[STIX]{x1D706}}=150$, $\overline{\mathit{St}}_{L}=0.1$).

Figure 3

Figure 3.Normalised decay rate as a function of Damköhler number obtained for different particle size distributions for $\overline{\mathit{St}}_{L}=0.1$ (cases D, J, K).

Figure 4

Figure 4.Comparison of the normalised decay rate obtained with and without particle back-reaction for simulations with $\overline{\mathit{St}}_{L}=0.05$ (a), 0.35 (b) and 1.2 (c) (cases C, E, F).

Figure 5

Figure 5.Particle number density for different Stokes numbers (case H). (a–c) Simulations with two-way coupling between particles and fluid; (d–f) only one-way coupling (no back-reactions from particles on fluid) is accounted for. White corresponds to zero particle number density, while black represents a particle number density that is 23 times larger than the average.

Figure 6

Figure 6.Particle number density for different Stokes numbers (case Q), one-way coupling, $Re_{\unicode[STIX]{x1D706}}\approx 350$.

Figure 7

Table 2.The correlation number between different Stokes numbers ($St=0.04$, 0.2 and 1), for simulations with and without back-reactions.

Figure 8

Figure 7.The auto-correlation function of the particle number density for simulations with and without back-reactions and for different particle sizes (case H).

Figure 9

Figure 8.The p.d.f. of $n_{p}/\overline{n}_{p}$ for different particle sizes, $Re_{\unicode[STIX]{x1D706}}\approx 150$, compensated distr., (a,b) with back-reaction, (c,d) no back-reaction, (a,c) $\overline{\mathit{St}}_{L}=1.2$, (b,d) $\overline{\mathit{St}}_{L}=0.05$, in all cases $Da\approx 35$ (cases C and F).

Figure 10

Figure 9.The p.d.f. of $X/\overline{X}$, (a) $\overline{\mathit{St}}_{L}=1.2$, (b) $\overline{\mathit{St}}_{L}=0.005$, compensated distribution, $Da\approx 35$ (cases C and F).

Figure 11

Figure 10.The p.d.f.s of (a) $n_{p}/\overline{n}_{p}$ and (b) $X/\overline{X}$ for different particle sizes, $Re_{\unicode[STIX]{x1D706}}\approx 350$, compensated distr., with back-reaction, $\overline{\mathit{St}}_{L}=0.04$, $Da=25$ (case N).

Figure 12

Figure 11.Normalised decay rate as a function of Damköhler number obtained for different $\overline{\mathit{St}}_{L}$ using a compensated distribution for (a) $Re_{\unicode[STIX]{x1D706}}\approx 150$ (cases A, B, D, F) and (b) $Re_{\unicode[STIX]{x1D706}}\approx 350$ (cases L, M, O, P).

Figure 13

Figure 12.Damköhler number $Da_{2}$ as a function of Stokes number, all cases with compensated and Dirac distributions.