Technical Report 1997:11 ISSN 0348-467X ISRN KTH/MEK/TR--97/11--SE

Large Eddy Simulation of Particulate Turbulent Channel Flows

Koji Fukagata

May 1997 Technical Reports from Royal Institute of Technology Faxen Laboratory S-100 44 Stockholm, Sweden

Fukagata, K. 1997 Large Eddy Simulation of Particulate Turbulent Channel Flows Faxn Laboratory e Royal Institute of Technology S-100 44 Stockholm, Sweden

Abstract
This thesis deals with numerical simulations of particulate turbulent channel ows. Turbulent velocity eld is simulated using large eddy simulation (LES) and the trajectories of particles are tracked by solving a particle equation of motion. The methodology used is at rst assessed through comparisons between predictions and the available data for a turbulent deposition of small particles in a channel. The prediction is in a good agreement with an empirical relation by Wood1 and simulations by Li & Ahmadi2. The assessment is done also with comparisons on the statistics of 70 m copper particles in a turbulent channel ow at Re = 180. The modication of uid ow due to particles is neglected, i.e. one-way coupling is considered. Very good agreement is found with a direct numerical simulation by Rouson & Eaton3 and a large eddy simulation by Wang & Squires4. The development time used in Wang & Squires is found to be short for obtaining a fully developed ow from the given initial condition. The statistics of 70 m copper particles and 50 m glass particles are accumulated after the ow has fully developed, and compared with the earlier data documented in the literature. Some discrepancies are indicated. The statistics are also used to evaluate dierent terms in the averaged Eulerian particle momentum equation, based on which a simplied version of the equation is derived. At last, simulations for the same parameters are performed taking into account the modication of uid velocity eld due to the presence of particles, i.e. two-way coupling. Statistics are compared with the cases of one-way coupling. Suppression of the turbulence is observed in the cases of 50 m glass particles and 70 m copper particles at mass ow rates of 0.4 and 3.0, respectively. An equation describing the relation between the modulation of uid stresses and interphase stress is derived. The thesis consists of a survey on the development time for the ows, force balances in the particle phase and modulations in the carrier uid turbulence due to the presence of particles followed by three papers, describing specic results.

WOOD N. B., 1981, J. Inst. Energy, 76, 76-93. LI A. & AHMADI G., 1993, Int. J. Engng Sci., 31, 435-451. 3 ROUSON D. W. I. & EATON J. K., 1994, Proceedings of the 7th Workshop on Two-phase Flow Predictions, Erlangen, Germany. 4 WANG Q. & SQUIRES K. D., 1996, Phys. Fluids, 8, 1207-1223.
2

ii

Preface
This thesis deals with numerical simulations of particulate turbulent channel ows, using large eddy simulation and Lagrangian particle tracking methodology. The thesis is based on the following three papers (updated from the original thesis). Paper 1: FUKAGATA K., ZAHRAI S. & BARK F. H., 1997, Large eddy simulation of particle motion in a turbulent channel ow appeared partly in two papers: Fukagata, K., Zahrai, S. & Bark, F. H., 1997. Large eddy simulation of particle motion in a turbulent channel ow. Proc. 1997 ASME Fluids Eng. Div. Summer Meeting (CD-ROM), Paper No. FEDSM97-3591, 1-6. Fukagata, K., Zahrai, S. & Bark, F. H., 2004. Dynamics of Brownian particles in a turbulent channel ow. Heat Mass Transfer 40, 715-726. Paper 2: FUKAGATA K., ZAHRAI S. & BARK F. H., 1998, Force balance in a turbulent particulate channel ow, Int. J. Multiphase Flow 24, 867-887. Paper 3: FUKAGATA K., ZAHRAI S. & BARK F. H., 1998, Fluid stress balance in a turbulent particulate channel ow. Proc. 3rd Int. Conf. Multiphase Flow (CD-ROM), Paper No. 157, 1-8.

Other reports related to this thesis: FUKAGATA K., 1995, Simulation of particle motion in a turbulent velocity eld, part I, Technical Report, SECRC/KB/TR-96/162E, ABB Corporate Research. FUKAGATA K. & ZAHRAI S., 1996, Simulation of particle motion in a turbulent velocity eld, part II, Technical Report, SECRC/B/TR-96/107E, ABB Corporate Research.

a)

z+

x+ b)

z+

x+ Figure 1: Modication of turbulent structure due to 50 m glass particles at mass ow ratio of 0.4. See Paper 3. a) one-way coupling; b) two-way coupling. Black point, particle. Red, high uid velocity; blue, low uid velocity.

iii

a)

z+

x+ b)

z+

x+ Figure 2: Modication of turbulent structure due to 70 m copper particles at mass ow ratio of 3.0. See Paper 3. a) one-way coupling; b) two-way coupling. Black point, particle. Red, high uid velocity; blue, low uid velocity.

iv

Contents
1 Introduction 1.1 Particulate turbulent ow . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Previous work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 Aim of this study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Overview of 2.1 Paper 1 2.2 Paper 2 2.3 Paper 3 the papers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1 3 5 7 7 8 8 11 13 17 47 77

Acknowledgement Bibliography A Paper 1 B Paper 2 C Paper 3

vi

CONTENTS

Chapter 1 Introduction
1.1 Particulate turbulent ow

Two-phase ow can be found almost everywhere from the processes in the nature such as the clouds on the sky, which have existed since long time before mankind appeared on the earth, to the industrial processes such as in the pipes in nuclear reactors which were realised by assembling modern technology. As well as most of the single phase ows, two-phase ows are often turbulent. Since perhaps the rst scientic observation on the turbulence by da Vinci in 15th century, turbulence has been an attractive research area. Numerous experiments were carried out to understand the characteristics of turbulence and to model it mathematically. The eort put on turbulence research has brought two monumental achievements in the turbulent theory, i.e. the mixing length model and the Kolmogorov spectrum. Despite the enormous eort put on the subject, turbulence has not fully been understood. According to Tennekes & Lumley (1972), turbulence can be characterized by irregularity, diusivity, large Reynolds number, three-dimensional vorticity uctuations, dissipation, continuum and turbulent ows are ows. Among others it is the irregularity that makes the deterministic approach dicult. This nature of turbulence will be a hurdle also when, for example, turbulent ows including many small particles are considered. Even if the particle equation of motion is simple, the motion of the particles becomes complicated due to underlying irregularity of carrier ow. One exception is the case when the particles have innitely large inertia compared to the uid and do not interact with the carrier uid. In that case particles can be treated similarly to the theory for gas molecules, see Sundaram & Collins (1997). In most cases which are of interest, however, particles do not have innitely large inertia and do interact with the carrier uid. Therefore, similarly to the approach for single phase turbulent ow, statistical methods are to be relied on. Turbulent diusivity is known to cause rapid mixing of the containments such as heat and particles. The diusivity of the particles in a turbulent ow is of great interest when a ltering, separating or mixing apparatus is designed, see e.g. van der Velde (1996). 1

CHAPTER 1. INTRODUCTION

After the invention of computers, many researchers examined to predict the statistics, such as the mean velocity, of turbulent ows by solving the averaged Navier-Stokes equation. Since the Navier-Stokes equation involves an advection term which is non-linear, extra unknowns called Reynolds stresses appear in the averaged equation. Therefore a relevant closure model is needed to close the equation. One of the most popular closure models is the k model (Jones & Launder, 1972), which is based on the mixing-length theory. The Reynolds stress is approximated to be the product of the turbulent viscosity and the mean strain rate. The turbulent viscosity is assumed to be proportional to k 2 / , where k is the turbulent kinetic energy and is the dissipation rate which are obtained by solving the transport equations for k and . Although the k model has some disadvantages such that it cannot predict a highly anisotropic ow or a ow under rotation, it is widely used for industrial purposes owing to its simplicity and numerical stability. During the passed few decades, due to the rapid development of computational ability, numerical simulations of turbulent ows without high degree of modeling has been made possible. Direct numerical simulation (DNS) uses very ne computational meshes such that the smallest eddies comparable with the Kolmogorov length scale can be captured. In DNS, the Navier-Stokes equation is solved as it is and there is no need for any closure models. Large eddy simulation (LES) uses also ne meshes, but coarser than those in DNS, such that relatively large energetic eddies can be resolved according to the averaged Navier-Stokes equation. The relatively large eddies in this context designate the same sizes or larger than those comparable with the integral length scale in the ow. The smaller, subgrid scale, eddies are considered to be independent of the geometry of the ow and isotropic. Therefore they can be modeled using subgrid scale model. DNS and LES have frequently been used to investigate pure turbulence. However, recently, these methods are also applied to more complex ows such as turbulent two-phase ows. Two-phase ows consist of two dierent phases, e.g. gas and liquid, gas and particle or liquid and particle. In general these dierent phases interact each other and change their shape of the surface and transit from one ow pattern to another. For this reason, two-phase ows are usually divided into dierent ow regimes, which are treated separately. For example, gas-liquid two-phase ows in a pipe can be classied depending on the gas ow rate and the liquid ow rate, e.g. bubble ow, slug ow, churn ow, annular ow, dispersed ow etc.. The interest in this thesis is focused on the dispersed ow, to be more specic, uid-particle ows consisting of a uid as a continuous phase and spherical solid particles as dispersed phase. As the carrier ow considered here is turbulent, the particulate ows considered in this study are often called particulate turbulent ows.

1.2. PREVIOUS WORK

1.2

Previous work

To understand the behavior of an ensemble of particles in a ow of a continuous phase, it is at rst necessary to have a knowledge about the motion of one single particle in the ow. Basset (1888), Boussinesq (1903) and Oseen (1927) separately examined the motion of a sphere in a uid settling down under gravitational force. The particle equation of motion developed by them is called Basset-Boussinesq-Oseen (BBO) equation. Tchen (1947) extended the BBO equation such that it could describe the particle motion in an unsteady ow. Extension was also examined for non-uniform ows, however, inconsistency was pointed out by Corrsin & Lumley (1956). Later, Buevich (1966) & Riley (1971) made a correction in the Tchens expression and Maxey & Riley (1983) presented a particle equation of motion, which is often used today, consisting of the following terms: 1. drag force 2. gravitational force 3. uid acceleration 4. added mass 5. history eect A correction term due to the velocity curvature of the carrier uid, Faxn correction e (Faxn, 1922) is included in the drag force, the added mass and the history eect e appearing in the expression by Maxey & Riley (1983). Under certain conditions, contributions from the following terms may be taken into account: 6. lift force 7. Brownian motion 8. Cunningham slip correction The lift force may be included when the shear rate of carrier ow is large, i.e. Samans lift force (Saman, 1965), or when the particles are rotating quickly. The Brownian motion and the Cunningham slip correction are of importance in the cases where the diameter of particles is comparable to the mean free path of carrier uid molecules. When the particle to be considered is much heavier than uid e.g. p /f 103 , the particle equation of motion can reduce from dimensional analysis to that consisting of only drag force and gravitational force. The reduced particle equation of motion normalized by the Stokes number, includes only two parameters, the Stokes number, which is the ratio of particle time scale to uid time scale, and the drift parameter, which reects the gravitational eect. The eects of these two parameters on the diusivity of particles are studied and summerized by Stock (1996).

CHAPTER 1. INTRODUCTION

There are mainly two dierent ways to deal with the behavior of large number of particles in a ow, the Eulerian-Eulerian approach and the Eulerian-Lagrangian approach. In the Eulerian-Eulerian approach, the particle phase is treated as a continuum and the equations for all phases are solved at the same time. In this case, closure models are needed also for the particle phase equation similarly to the equation for uid. In the Eulerian-Lagrangian approach, the trajectories of particles are tracked separately according to the particle equation of motion, which is expressed in the Lagrangian frame. The simplest case to be studied is a particulate ow with a homogeneous turbulence. Shih & Lumley (1986) reported on second-order closure modeling of particle dispersion in a decaying homogeneous turbulence and the simulation was found to be in good agreement with the experimental data by Wells & Stock (1983). Direct numerical simulation and Lagrangian tracking of motion of particles in a decaying homogeneous turbulence was performed by Elghobashi & Truesdell (1992) and the important quantities such as Lagrangian velocity auto-correlations and diusivity of the particles were studied. Shear ows and jet ows are the next simplest cases because solid boundaries need not to be taken into account. Simonin, Deutsch & Boivin (1995) developed a second-order closure model of particle uctuating motion in a turbulent shear ow and the time evaluation of the various correlations were found in good agreement with the data from large eddy simulation carried out by them. A turbulent particulate jet ow was simulated by Chen & Wood (1985) using a k model. Eect of the particle size and particle loading on the dispersion was studied by comparing the simulation data with available experimental data by, e.g. Boguslawski & Popiel (1979), Wall, Subramanian & Howley (1982) and Moderrass, Tan & Elghobashi (1984). For the practical applications, such as predictions of behaviors of particulate ows appearing in industrial systems, it is necessary to understand the motion of particles in geometries with rigid walls. The simplest examples with such geometries are straight pipes or channels bounded with two parallel walls. In fact most of the experiments, except jet, are carried out in such pipes or channels. To observe particle motion, laser Dopper anemometry, by e.g. Maeda, Hishida & Furutani (1980), Lee & Durst (1981), Tsuji, Morikawa & Shiomi (1984), Kulick, Fessler & Eaton (1994) and Kaftori, Hetsroni & Banerjee (1995), or high-speed video, by e.g. Rashidi, Hetsroni & Banerjee (1990) and Ni o & Garcia (1996), has been used. It n was observed in the experiments that the carrier uid turbulence are suppressed by suspension of small particles and enhanced by large particles. Hetsroni (1989) and Gore & Crowe (1989) summerized the data from the available experiments and discussed the criteria between the enhancement and the suppression of turbulence. It was also found that the particle with nite, small inertia tend to concentrate in the regions with low shear in the streaky turbulent structure near the wall. Theoretical studies about these wall-bounded particulate ows using numerical simulations have also been reported. The main research interest may be 1) to clarify the phenomena of deposition and entrainment of particles near the wall and 2) to provide statistical moments of particles. Ounis, Ahmadi & McLaughlin (1991)

1.3. AIM OF THIS STUDY

reported on the dispersion and the deposition of very small particles using DNS and Lagrangian particle tracking. The Browinan motion due to interaction between particles and uid molecules was also taken into account. With the same methodology, the entrainment process was studied by Soltani & Ahmadi (1995). Deposition in a pipe ow was simulated by Wijttewaal & Oliemans (1996). In Wang & Squires (1996a), LES was used for simulations of particle deposition in a channel instead of DNS. Li & Ahmadi (1993) studied the deposition rate of particles in a channel using a quasi-turbulent uid velocity eld and Lagrangian particle tracking. The predicted deposition velocity was in good agreement with the empirical relation by Wood (1981). The method was also applied to the prediction of deposition rate in a complex geometry (Li, Ahmadi, Bayer & Gaynes, 1993). Swailes & Reeks (1994) used a kinetic theory for high inertia particles to simulate the particle deposition. The statistics of relatively high inertia particles obtained from DNS and LES have been reported by several researchers. Pedinotti, Mariotti & Banerjee (1992) carried out a simulation of motion of light particles in a horizontal channel using DNS and Lagrangian particle tracking. The simulation data was compared with the experimental data by Rashidi, Hetsroni & Banerjee (1990). Pan & Banerjee (1996) took into account two-way coupling, i.e. the particle motion modies the uid velocity eld, and the further comparison with the experiment by Rashidi et al. (1990) was done. Rouson & Eaton (1994) and Wang & Squires (1996b) performed DNS and LES, respectively, to simulate the experiments by Kulick et al.. Two-equation models, i.e. Eulerian-Eulerian approach, of the particulate turbulent channel ow have also reported by Rizk & Elghobashi (1989), Bolio & Sinclair (1995) and Cao & Ahmadi (1995). As an example, comparison between the model by Rizk & Elghobashi and experiments by Maeda et al. (1980) and Tsuji et al. (1984) shows that the model by Rizk & Elghobashi has predicted the behavior of particles and uid reasonably well. For more accurate prediction, however, further improvement of the existing models is likely to be needed.

1.3

Aim of this study

The nal goal of this work is to study the eect of presence of particles on the turbulent structure near the wall and to make a simple, more accurate model to predict such ows. For these purposes, a computational code to calculate the motion of particles in a turbulent channel ow was developed based on the large eddy simulation code developed by Zahrai, Bark & Karlsson (1995). The code was prepared to account for the modulation of the turbulent uid velocity eld due to the presence of particles. Various statistics both of the uid and of the particles are calculated, which will provide the necessary data for the modeling. This thesis is organized in the following way: Overviews of the papers, Paper 1, Paper 2 and Paper 3, are presented in Chapter 2. Specic topics are discussed in these papers. Each paper is in a complete form and can be read also as an independent article. The papers are attached to this introductory note.

CHAPTER 1. INTRODUCTION

Chapter 2 Overview of the papers

2.1 Paper 1

The methodology was validated through comparisons with available data in literature. In the rst part in the paper, statistics of very small particles in a turbulent chan+ nel ow were presented. The nondimensional particle relaxation time, p , ranged from 104 to 101 . The deposition velocity was compared with the result of quasi-turbulence simulation by Li & Ahmadi (1993) and the empirical relation suggested by Wood (1981). While the deposition velocities obtained using a time-dependent turbulent velocity eld were in good agreement with those predicted by Li & Ahmadi (1993) and Wood (1981), a simulation using a frozen velocity eld could not predict reasonable deposition velocities except in the ranges of particle diameter where the Brownian motion was dominant or where the inertia of particle was large. In the second part, similarly to the direct numerical simulation carried out by Rouson & Eaton (1994) and the large eddy simulation by Wang & Squires (1996b), 70 m copper particles were considered. The particle relaxation time nondimension+ alized by the viscous scale, p , was 810. The Reynolds number of the carrier uid based on the friction velocity and the channel half-width, Re , was 180. Simulation was started distributing 250000 particles homogeneously in the channel. The initial velocities of the particle were set to the local uid velocities. Similarly to Wang & Squires, a nondimensional time of 6 /u was taken for the development time, i.e. the time necessary to obtain a statistically steady particulate ow. The statistics such as mean velocity and RMS levels of the velocity uctuations were accumulated for a time period of 6 /u after the development time mentioned above. Excellent agreement was found between this study and the data in the literature, both in the mean particle velocity prole and the velocity uctuations. However, a continuous build-up of particles was observed during the accumulation of statistics. As a simple interpretation of the latter phenomena, the development time used in the literature was considered to be short. Simulations were also carried out for 50 m glass particles. The particle relax+ ation time, p , was 117. Since the development time was not reported in Wang & 7

CHAPTER 2. OVERVIEW OF THE PAPERS

Squires, an estimated value was used in this simulation. Again, excellent agreement was found in the statistics compared with Wang & Squires. The simulations were done both with 250000 samples and with 22000 samples. Although Wang & Squires reported that at least 250000 particles were necessary to obtain a continuous representation of statistics in this problem, these two simulations in this study showed that there were no quantitative dierence in the obtained statistics between the case with 250000 samples and 22000 samples.

2.2

Paper 2

The statistics of fully developed turbulent ows with 70 m copper particles and 50 m glass particles in a channel were calculated using large eddy simulations. In order to obtain fully developed ows, the development time for the 70 m copper particles and 50 m glass particles were chosen as 12 /u and 8.4 /u , respectively, which were longer than those used in earlier studies. The statistics of particles were compared with the data presented in Paper 1 and in Wang & Squires (1996b) which used a shorter development time. Both in the cases of 70 m copper particles and 50 m glass particles, slight decrease of the mean particle velocity and slight increase of the particle number density were observed near the wall. The statistics were used for a detail study of force balance and interphase forces. The Reynolds decomposition based on ensemble average was applied to the Eulerian particle momentum equation. Each term in the decomposed equation was obtained using the statistics. In the direction normal to the wall, the mean drag force was found to be balanced by the transport of particle ux due to turbulence. The mean drag force was mainly contributed from the term due to the drift velocity and the term arising from drag velocity correlation. In the case of 50 m glass particles, where the particle Reynolds number was low, the drag - velocity correlation term was also very small. In the streamwise direction, the mean drag force was considered to be balanced by a sum of the turbulent transport of particle ux and the gravitational force. The mean drag force could be approximated by the term due to mean relative velocity between particles and uid. Based on these investigations mentioned above, simplied versions of averaged momentum equations for particle phase were presented.

2.3

Paper 3

Two-way coupling simulations, i.e. simulations where the modication of carrier uid ow due to the presence of particles are taken into account, were performed for the same parameters considered in Paper 2. The forces from particles to uid were modeled as the sum of drag forces acting on particles with negative sign, and added as a body force to the Navier-Stokes equation.

2.3. PAPER 3

Equations were integrated during the development time before the accumulation of statistics in order to ensure that the data were representing fully developed ows. The accumulated statistics were compared with the data for the one-way coupling simulations presented in Paper 2. The presence of 50 m glass particles at mass ow rate of 0.4 did not change the mean uid velocity. The streamwise RMS velocity decreased only near the wall, but the other RMS velocity components and the Reynolds stress decreased in the whole channel. The streaky structure of nearwall turbulence was modulated. The two-point velocity correlations showed that the streaks had become longer than those in undisturbed ows. With 70 m copper particles at mass ow rate of 3.0, the centerline uid velocity was found to be double as that of undisturbed uid. The mean uid velocity prole deviated signicantly from the logarithmic law. The streamwise RMS velocity was at comparable level with that of undisturbed ow. The RMS velocities in the other directions and the Reynolds stress were found to have values close to zero. The turbulent structure was also destroyed. A relation between the uid stresses and the interphase stress were derived following the treatment for deriving the linear relation of stresses in a pure turbulent channel ow (Tennekes & Lumley, 1975). The resulting equations were studied term by term using the accumulated statistics.

10

CHAPTER 2. OVERVIEW OF THE PAPERS

Acknowledgement
I wish to thank my supervisors, Professor Fritz H. Bark and Dr. Said Zahrai, for the kind and strict supervisions. I also appreciate the opponent at the presentation of this thesis, Professor Alain Lin at Institut National des Sciences appliques de Toulouse, France. e e For the economical support, Faxn Laboratory, which is a cooperative research e center supported by KTH, NUTEK and a number of Swedish companies including ABB Corporate Research, and Axel och Margaret Ax:son Johnsons Stiftelse are greatly acknowledged. Finally, I would like to thank all the colleagues at KTH, ABB Corporate Research and University of Tokyo. I owe all of them for my fruitful time in Sweden.

11

12

ACKNOWLEDGEMENT

Bibliography
[1] BASSET A. B., 1888, Treatise on Hydrodynamics, Vol. 2, Chap. 22, pp. 285297, Deighton Bell, London. [2] BOGUSLAWSKI L. & POPIEL C. O., 1979, Flow structure of the free round turbulent jet in the initial region, J. Fluid Mech., 99, 531. [3] BOLIO E. J. & SINCLAIR J. L., 1995, Gas turbulence modulation in the pneumatic conveying massive particles in vertical tubes, Int. J. Multiphase Flow, 21, 985-1001. [4] BOUSSINESQ J., 1903, Theorie Analytique de la Chaleur, Vol. 2, p. 224, LEcole Polytechnique, Paris. [5] BUEVICH Y. A., 1966, Fluid Dynam., 1, 119. [6] CAO J. & AHMADI G., 1995, Gas-particle two-phase turbulent ow in a vertical duct, Int. J. Multiphase Flow, 21 1203-1228. [7] CHEN C. P. & WOOD P. E., 1985, A Turbulent Closure Model for Dilute Gas-Particle Flows, Can. J. Chem. Engng. 63, 349-360 [8] CORRSIN S. & LUMLEY J., 1956, Appl. Sci. Res. A, 6, 114. [9] ELGHOBASHI S. & TRUESDELL G. C., 1992, Direct simulation of particle dispersion in a decaying isotropic turbulence, J. Fluid Mech., 242, 655-700. [10] FAXEN H., 1922, Vereinfachte Darstellung der verallgemeinerten Greenschen Gleichungen fr die konstante Translationsbewegung eines starren Krpers in u o einer zhen Flssigkeit, Ark. Mat., Astron. o Fys., 17, 1-5. a u [11] GORE R. & CROWE C., 1989, Eect of particle size on modulating turbulent intensity, Int. J. Multiphase Flow, 15, 279-285. [12] HETSRONI G., 1989, Particles-turbulence interaction, Int. J. Multiphase Flow, 15 (5), 735-746. [13] JONES W. P. & LAUNDER B. E., 1972, The prediction of laminarization with a two-equation model of turbulence, Int. J. Heat Mass Transfer, 15, 301. 13

14

BIBLIOGRAPHY

[14] KAFTORI D., HETSRONI G. & BANERJEE S., 1995, Particle behavior in the turbulent boundary layer. I. Motion, deposition, and entrainment, Phys. Fluids, 7 (5), 1095-1106. [15] KULICK J. D., FESSLER J. R. & EATON J. K., 1994, Particle response and turbulence modication in fully developed channel ow, J. Fluid Mech., 277, 109-134. [16] LEE S., & DURST F., 1982, On the motion of particles in turbulent duct ows, Int. J. Multiphase Flows, 8, 125-146. [17] LI A. & AHMADI G., 1993, Deposition of aerosols on surfaces in a turbulent channel ow, Int. J. Engng Sci., 31 (3), 435-451. [18] LI A., AHMADI G., BAYER R. G. & GAYNES M. A., 1993, A digital simulation method for estimating dust deposition in a complex geometry passage, Advances in Electronic Packaging, ASME, EEP-Vol.4-1, 411-417. [19] MAEDA M., HISHIDA K. & FURUTANI T., 1980, Optical measurements of local gas and particle velocity in an upward owing dilute gas-solids suspension. Polyphase Flow and Transport Technology, Century 2-ETC, pp. 211-216. [20] MAXEY M. R. & RILEY J. J., 1983, Equation of motion for a small rigid sphere in a nonuniform ow, Phys. Fluids, 26 (4), 883-889. [21] MODERRASS D., TAN H. & ELGHOBASHI S., 1984, Two-component LDA measurement in two-phase turbulent jet, AIAA J., 22, 624-630. [22] NINO & GARCIA, 1996, Experiments on particle-turbulence interactions in the near-wall region of an open channel ow: implications for sediment transport, J. Fluid Mech., 326, 285-319. [23] OSEEN C. W., 1927, Hydrodynamik, p. 132, Leipzig. [24] OUNIS H., AHMADI G. & McLAUGHLIN J. B., 1991, Dispersion and deposition of Brownian particles from point sources in a simulated turbulent channel ow, J. Colloid Interface Sci., 147 (1), 233-250. [25] PAN Y. & BANERJEE S., 1996, Numerical simulation of particle interactions with wall turbulence, Phys. Fluids, 8 2733-2755. [26] PEDINOTTI S., MARIOTTI G. & BANERJEE S., 1992, Direct numerical simulation of particle behaviour in the wall region of turbulent ows in horizontal channels, Int. J. Multiphase Flow, 15 (6), 927-941. [27] RASHIDI M., HETSRONI G. & BANERJEE S., 1990, Particle-turbulence interaction in a boundary layer, Int. J. Multiphase Flow, 16 (6), 935-949. [28] RILEY J. J., 1971, Ph.D Thesis, The Johns Hopkins University, Baltimore, Maryland.

BIBLIOGRAPHY

15

[29] RIZK M. A. & ELGHOBASHI S. E., 1989, A two-equation turbulence model for dispersed dilute conned two-phase ows, Int. J. Multiphase Flow, 15 (1), 119-133. [30] ROUSON D. W. I. & EATON J. K., 1994, Direct numerical simulation of particles interacting with a turbulent channel ow, Proceedings of the 7th Workshop on Two-phase Flow Predictions, Erlangen, Germany. [31] SAFFMAN P. G., 1965, The lift on a small sphere in a slow shear ow, J. Fluid Mech., 22 (2), 385-400. [32] SHIH T.-H. & LUMLEY J. L., 1986, Second-order modelling of particle dispersion in a turbulent ow, J. Fluid Mech., 163, 349-363. [33] SIMONIN O., DEUTSCH E. & BOIVIN M., 1995, Large eddy simulation and second-moment closure model of particle uctuating motion in two-phase turbulent shear ows, 9th International Symposium on Turbulent Shear Flows, Kyoto, Aug. 1993, 85-115. [34] SOLTANI M. & AHMADI G., 1995, Direct numerical simulation of particle entrainment in turbulent channel ow, Phys. Fluids, 7 (3), 647-657. [35] STOCK D. E., 1996, Particle Dispersion in Flowing Gases 1994 Freeman Scholar Lecture, J. Fluids Engng., 118, 4-17. [36] SUNDARAM S. & COLLINS L. R., 1997 Collision statistics in an isotropic particle-laden turbulent suspension. Part 1. Direct numerical simulations, J. Fluid Mech., 335, 75-109. [37] SWAILES D. C. & REEKS M. W., 1994, Particle deposition from a turbulent ow. I. A steady-state model for high inertia particles, Phys. Fluids, 6 (10), 3392-3403. [38] TENNEKES H. & LUMLEY J. L., 1975, A rst course in turbulence, MIT Press, Cambrige. [39] TCHEN C.-M., 1947, Mean value and correlation problems connected with the motion of small particles suspended in a turbulent uid, Ph.D thesis, De technische hogeschool te Delft. [40] TSUJI Y., MORIKAWA Y. & SHIOMI H., 1984, LDV measurement of an air-solid two-phase ow in a vertical pipe, J. Fluid Mech., 139, 417-434. [41] WIJTTEWAAL W. S. J. & OLIEMANS R. V. A., 1996. Particle dispersion and deposition in direct numerical and large eddy simulations of vertical pipe ows, Phys. Fluids, 8, 2590-2604. [42] Van der VELDE H., 1996, On turbulent reacting dilute gas-particle ows in cyclones.

16

ACKNOWLEDGEMENT

[43] WALL T. F., SUBRAMANIAN V. & HOWLEY P., 1982, An experimental study of the geometry mixing and entrainment of particle laden jets up to ten diameters from the nozzle, Trans. Inst. Chem. Eng., 60, 231. [44] WANG Q. & SQUIRES K. D., 1996a, Large eddy simulation of particle deposition in a vertical turbulent channel ow, Int. J. Multiphase Flow, 22 (4), 667-683. [45] WANG Q. & SQUIRES K. D., 1996b, Large eddy simulation of particle-laden turbulent channel ow, Phys. Fluids, 8 (5), 1207-1223. [46] WELLS M. R. & STOCK D. E, 1983, The eects of crossing trajectories on the dispersion of particles in a turbulent uid, J. Fluid Mech., 136, 31-62. [47] WOOD N. B., 1981, The mass transfer of particles and acid vapour to cooled surfaces, J. Inst. Energy, 76, 76-93. [48] ZAHRAI S., BARK F. H. & KARLSSON R. I., 1995, On anisotropic subgrid modeling, Eur. J. Mech., B/Fluids, 14 (4), 459-486.