16:15
Multiphase Flows 6
16:15
15 mins
#557
|
Sediment transport in a turbulent open-channel with macro-roughness elements
Michele Trevisson, Olivier Eiff, Yulia Akutina
Abstract: Supply-limited sediment transport in a turbulent channel flow with spherical macro-roughness elements is considered here as an idealized model for gravel-bed rivers under low flow conditions when the gravel are immobile. As observed by [1], [3], the transport rate of fine sediments is strongly correlated to the level of protrusion of the coarse immobile gravel. [4], studying the effect of different protrusion levels of a patch of hemispheres in a square arrangement with a fixed-sediment bed, show that a possible explanation for the suppression of sediment transport with increasing protrusion of the roughness elements, is the reduction of the effective shear stress acting on the grains as a consequence of a decrease in the number of sweep events reaching the grains. This effect is pursued here by considering a more realistic staggered pattern of spheres over a homogeneous bed with loose sediments. To this end, a 9 m long and 0.3 m wide tilting open-channel flume was used. To model the gravel bed, two layers of plastic spheres (R=1 cm) were fixed to the flume bed in staggered and symmetric patterns. Fine sand with a median diameter of 1.7 mm was used for the mobile sediment. To measure simultaneously the velocity field in between the upper layer of the spheres and above as well as the sediment bed-level, a stereo-PIV and a stereo-photogrammetry technique were used, respectively. Initial Shields numbers were set to about 0.08 with Re around 45, just above the critical shear stress, so that the initially uniformly covered bed starts to erode. The measurements were then performed at different times as the bed erodes. On the basis of shear stress partitioning theory (Raupach, 1992), it is hypothesized that, as the roughness elements protrude from the sediment bed, the portion of shear stress acting directly on the grains diminishes down to the critical value of particle motion. By analyzing the temporal evolution of area coverage of the bed, the possibility of reaching a static equilibrium was assessed.
The bed level measurements reveal that as the bed erodes below the spheres’ caps, the sediment transport rates are relatively high as long as the mean level of protrusion, defined by P=k/R, where k is the height of a protruding sphere over the sediment bed, remains below about P= 0.6. When the bed erodes further, the sediment transport rate reduce significantly and effectively stop near P= 1. To help explain this behaviour, the turbulent shear stress, τxz=u′w′, normalized by the stress acting on the sediments when the spheres are covered, τb, was analyzed (see Figures 1a-b for P=0.6 and 1, respectively), as well as a quadrant analysis. As seen in Fig. 1, for both protrusions, the shear stress at the level of the grains is lower than τb, suggesting sheltering by the spheres. At P=1 this sheltering effect is much stronger, resulting in the cessation of grain transport. At P= 0.6 the grains in the wake of the sphere are close to the mixing layer generated by the caps, resulting in the observed predominance of sweep events (Q4), known to effectively transport grains. For P= 1, instead, the shear stress at the level of the grains is negative as a result of a recirculation zone identified in the mean flow. Here Q1 events, i.e. outward interactions, are found to dominate the shear stress. While these events can be just as effective as sweeps in moving sediments [2], the particles remain trapped in the recirculation zone leading to the effective cessation of the sediment transport.
|
16:30
15 mins
#377
|
Homogeneous shear turbulence laden with finite-size spheroidal particles
Ali Yousefi, Mehdi Niazi Ardekani, Luca Brandt
Abstract: Self-sustaining wall-bounded shear flows are well characterized and correspond to physically realizable situations. However, they are inhomogeneous, and it is difficult to distinguish the effect of the presence of the wall, where fluctuations go to zero, and those due to the shear-induced turbulent production.
Therefore, investigating particle suspensions in a homogeneous flow configuration is of utmost importance to understand how a suspended phase modulates the turbulence.
Reproducing self-sustaining turbulence in a theoretically unbounded and homogeneous domain is one of the main difficulties in both experimental and numerical studies. Artificial forcing at a few low wavenumbers is a common way in numerical studies, however, it is difficult to judge how far into the cascade of energy, the effect of the forcing extends and therefore single out the role of a second phase. Homogeneous shear turbulence (HST) is perhaps the simplest flow configuration which shares the natural energy-generation mechanism of shear flows with the simplicity of homogeneity.
In the current study, we investigate turbulence modulation in the presence of finite-size spheroidal particles. We exploit \textit{shear-periodic} boundary condition to avoid re-meshing and keep the advantage of using a FFT-based Poisson solver for higher performance.
A direct-forcing immersed boundary method is used to account for the dispersed phase, together with lubrication correction and collision models for close range particle-particle interactions \cite{ardekani2017drag}.
These interface-resolved simulations give us a better understanding of the effect of concentration and shape of particles on the energy spectra and modulation of turbulence.
We study different shapes of spheroidal particles, with aspect ratios (polar to equatorial radius) $1/3$ (oblates) and $1$ (spheres), up to $20\%$ volume fraction, explaining how the particle dynamics affect the turbulence development in comparison to the single-phase flow.
Simulations begin with the initial velocity field obtained from a prescribed energy spectrum with a random phase to resemble homogeneous isotropic turbulence and evolve into steady state HST by the action of a given shear rate \cite{pumir1996turbulence}.
Preliminary results show that increasing the volume fraction of spherical particles, the mean steady-state value of the turbulent kinetic energy and the Taylor micro-scale Reynolds number decrease. This behavior is more noticeable in the case of oblate particles.
Detailed statistics of the fluid and the particle phase will be presented, focusing on the particle dynamics and the relation to the turbulence modulations.
|
16:45
15 mins
#412
|
Interface-resolved investigation of particle-laden turbulent channel flow in the point-particle limit
Pedro Costa, Luca Brandt, Francesco Picano
Abstract: More than thirty years have passed since the first direct numerical simulations of particle-laden turbulent flows using the point-particle approximation. This approximation assumes that the momentum exchange between the phases is well modeled by localized forces that mimic the presence of particles. Moreover, if the particle Stokes number is sufficiently high, the equations governing the particle dynamics are often simplified to a balance between inertia and drag. At present, it has become possible to simulate these flows without relying on the point-particle assumption. In other words, it is now possible to directly approximate the no-slip/no-penetration boundary condition on the surface of hundreds of thousands, or even millions of spherical particles. In the present work we revisit the problem of particle-laden turbulent channel flow in the point-particle limit, with insights from interface-resolved simulations. The numerical method uses an efficient immersed-boundary method, extended with closure models short-range particle-particle and particle-wall interactions (e.g. lubrication and collisions). We simulated different turbulent channel flow configurations with varying mass fraction, for fixed Reynolds number and particle size of about 3 viscous wall units. The goal is to mimic the so-called 1-way, 2-way and 4-way coupling regimes. The number of particles was varied from 500 to 500 000 particles, simulated on a mesh with 12 grid points over the particle diameter, corresponding to 13.5 billion grid points. We will present the results pertaining to these massive simulations, compared to those obtained from point-particle methods. In particular, we will show that – even for a very dilute flow – a widely-used 1-way coupling point-particle model fails in predicting basic quantities like the concentration profile and second-order moments of particle velocity. The causes for this basic discrepancy will be discussed.
|
17:00
15 mins
#576
|
TURBULENCE MODULATION BY INERTIAL PARTICLES IN A SWIRLING FLOW
Jérémy Vessaire, Romain Volk, Mickaël Bourgoin
Abstract: The dynamical behaviors of inertial particles in turbulence represents a fundamental and an open problem in multiphase flows. The lack of physical understanding of such phenomenon affects multiple applications where it remains a key obstacle to improve the optimization of industrial processes and modeling of numerous natural phenomena. In many practical situations, the carrier flow is turbulent and the particle mass loading is large, producing the dynamics of the continuous and dispersed phases which are strongly coupled [1}. Currently, for a given turbulent flow, there exists no general consensus on the impact of particles on energy injection and dissipation. Understanding how turbulence is modulated depending on the particles’ characteristics (relative density, size, volume fraction) is a great challenge both experimentally and numerically [2}.
Here, we present a study that aims at quantifying the modulation of a turbulent von K\'arm\'an flow of water seeded with glass particles at high volume fraction (up to 20%). In such conditions the system is expected to be a four way coupling regime [1}. The particles are suspended by a disc rotating at constant velocity at the top of a parallelepipedic tank with square section. At the highest considered fraction, the flow is completely opaque (classical optical diagnostics cannot not be used). Therefore, we characterize the turbulence modulation by quantifying (i) the global torque needed to sustain rotation together and (ii) the local information using high sensitivity pressure sensors placed at different heights on one side of the vessel. The combination of both diagnosis allows a multi-scale analysis of the turbulence modulation.
Independent of the volume fraction, when the particles are entirely suspended, the injected power scales as the cube of the rotation disk velocity, characteristic of a fully turbulent regime. The results highlight that power input always increases as particles are added. Further, the pressure spectra indicate turbulence is damped by the particles at small scale and enhanced at large scale, in agreement with the common knowledge of two phase flows. Influence of particle size and fluid viscosity will be discussed.
[1] Elghobashi, On predicting particle-laden turbulent flows S. Appl. Sci. 52, 4 (1994).
[2] Bourgoin and Xu, Focus on dynamics of particles in turbulence New J. Phys. 16, 085010 (2014).
|
17:15
15 mins
#292
|
Collapse of turbulence in particle laden channel flow at critical volume loading
Pradeep Muramulla, Viswanathan Kumaran, Partha Sarathi Goswami
Abstract: One of the important issues in the dynamics of particle laden turbulent suspensions is the effect of the particles on the gas phase turbulence \cite{Tanaka_2008}. Generally, two mechanisms are considered responsible for the turbulence modification. Turbulence augmentation happens due to the enhancement of fluctuations by wakes behind particles, whereas turbulence attenuation happens as result of increased dissipation due to the particle drag. To examine the turbulence modification mechanism, Direct Numerical Simulations of a particle gas suspension were performed at a Bulk Reynolds Number 3333, based on the channel width $h$ and the average gas velocity $\bar{u}$. The particle volume fraction is small, in the range $0-2 \times 10^{-3}$, the mass loading is varied in the range $0 -13.5$, and the particle Stokes number is $\tau_v/\tau_f$ is varied in the range $4.7-380$. Here,$\tau_v = \varrho_p d_p^2 / 18\mu (1+0.15Re_p^{0.687})$ is the particle relaxation time, with stokes inertial correction and $\tau_f = h/\bar{u}$ is the fluid time scale. The reverse force due to the particles on the fluid is modeled as a point force, because the particle diameter is smaller than the Kolmogorov scale. As the particle volume fraction is increased, a discontinuous decrease in the turbulent velocity fluctuations is observed at a critical volume fraction. There is a reduction, by 1 to 2 order of magnitude, in the mean square fluctuating velocities in all directions and in the Reynolds stress. At this critical volume fraction, mean gas velocity profile differs from fully developed channel flow with sharp increase in velocity at the centre of the channel. As a constant bulk flow condition is maintained, the pressure gradient and wall shear stress are modified. There is a dramatic collapse in rate of turbulent production at critical volume fraction. Though a modest increase in the rate of energy dissipation due to particle drag is observed, but this increase is smaller than the decrease in the rate of fluid turbulent energy production. It is also observed that, there is a decrease in the total energy dissipation rate when turbulence collapses at critical volume fraction. In an unsteady simulation, rate of decay in turbulent production with time is faster, compared to energy dissipation due to particle drag. Thus, the discontinuous collapse of turbulence attenuation appears to be due to a disruption of the turbulence production mechanism, and not due to the increased dissipation due to the particles.
|
17:30
15 mins
#229
|
Influence of the quiescent core region on inertial particle dynamics
Yucheng Jie, Helge Andersson, Guixiang Cui, Lihao Zhao
Abstract: Suspensions of particles with different inertia in wall-bounded turbulent flows are of fundamental significance in both industrial processes and natural phenomena. There are typical flow structures in wall-bounded turbulence, especially at high Reynolds numbers, and their influences on the dynamics of inertia particles need to be better understood. A typical flow structure named uniform momentum zone (UMZ) was discovered by Meinhart & Adrian (1995) and then further discussed by de Silva et al. (2016) in turbulent boundary layer flows. By applying a similar identification of UMZs to turbulent channel flows, Kwon et al. (2014) revealed that a large UMZ settled in the channel center and named it the quiescent core (QC). In the present work, a direct numerical simulation of turbulent channel flow at Re_\tau=600 with inertial spherical particles was performed to explore the effects of the QC region on particle dynamics. Figure 1(a) shows the instantaneous distribution of St=30 particles in a streamwise--wall-normal plane, where the white lines represent the boundary of the QC region (QC-boundary). The Voronoï method (Monchaux et al. 2010; Nilsen et al. 2013) was adopted to investigate the preferential concentration of particles, especially that around the QC-boundary. Particles with medium inertia tend to accumulate near the walls and in the channel core (see Figure 1b). Inertial particles are found to concentrate in the regions with relatively high streamwise velocity around the channel core (see Figure 1a). Moreover, quantitative analysis of mean Voronoï volume V around the QC-boundary is shown in Figure 1(c), where a sharp decrease of V appears, revealing an abrupt change of concentration. Statistics of particle translational motion are further studied and will be included in the presentation at the conference.
|
17:45
15 mins
#383
|
DENSE SUSPENSIONS FLOWING IN CHANNELS AT MODERATE REYNOLDS NUMBERS
francesco picano, pedro costa, luca brandt
Abstract: Suspensions of finite-size particles in flows at moderate/high Reynolds numbers are often found in the environment and in industrial applications.
In this work, we study dense suspensions of neutrally buoyant finite-size spherical particles in a plane channel at moderate Reynolds numbers using fully-resolved DNS. We will show that despite dense cases may show a friction factor close to corresponding more dilute cases or single phase flows their microscopic dynamics is strongly different since the law of the wall (log-layer) does not apply and we observe instead reduced Reynolds shear stresses and much smaller with respect to the particle induced stress. In the final paper, we will show difference and similarities between dense and dilute regimes at moderate Reynolds numbers.
|
|