The dynamical behavior of almost neutrally buoyant finite-size rigid fibers or rods in turbulent channel flow is studied by direct numerical simulations. The time evolution of the fiber orientation and translational and rotational motions in a statistically steady channel flow is obtained for three different fiber lengths. The turbulent flow is modeled by an entropy lattice Boltzmann method, and the interaction between fibers and carrier fluid is modeled through an external boundary force method. Direct contact and lubrication force models for fiber-fiber interactions and fiber-wall interaction are taken into account to allow for a full four-way interaction. The density ratio is chosen to mimic cellulose fibers in water. It is shown that the finite size leads to fiber-turbulence interactions that are significantly different from earlier reported results for point like particles (e.g., elongated ellipsoids smaller than the Kolmogorov scale). An effect that becomes increasingly accentuated with fiber length is an accumulation in high-speed regions near the wall, resulting in a mean fiber velocity that is higher than the mean fluid velocity. The simulation results indicate that the finite-size fibers tend to stay in the high-speed streaks due to collisions with the wall. In the central region of the channel, long fibers tend to align in the spanwise direction. Closer to the wall the long fibers instead tend to toward to a rotation in the shear plane, while very close to the wall they become predominantly aligned in the streamwise direction.

An Eulerian model was developed for turbulent gas-particle flow that takes into account the influence of particles on the gas-phase turbulence. For the description of the particle-phase stress the kinetic theory of granular flow and the simpler Hinze model were adopted. A K-ω model was used as the gas phase turbulence model. The difference between one- and two-way coupling was investigated for different particle volume fractions and particle diameters. It was found that particles with a much higher density than the fluid substantially affect the gas-phase in turbulent channel flow for particle volume fractions as low as 10 -4. The models with the particle-phase stress described by the kinetic theory of granular flow and the simpler Hinze model produce similar results for particles with small response times but deviate for larger response times. The study shows that two-way coupling and the turbophoretic effect must be taken into account in models even at relatively low particle volume fractions.

An Eulerian turbulent two phase flow model using kinetic theory ofgranular flows for the particle phase was developed in order to studyevolving upward turbulent gas particle flows in a pipe. Themodel takes the feedback of the particles into account and its resultsagree well with experiments. Simulations show that the pipe length required for particle laden turbulent flow to become fully developed is up to five times longer than an unladen flow. To increase theunderstanding of the dependence of the development length on particlediameter a simple model for the expected development length wasderived. It shows that the development length becomes shorter forincreasing particle diameters, which agrees with simulations up to aparticle diameter of 100 μm. Thereafter the development lengthbecomes longer again for increasing particle diameters because largerparticles need a longer time to adjust to the velocity of the carrierphase.

The correlation correlation between the fluctuating particle and gas velocity in isotropic turbulence is studied with a set of stochastic differential equations taking into account both particle-particle collisions and the particle feedback on the turbulence. The principal aim of this work is to use the Langevin equations to formulate closures for two-fluid gas-particle flow models. Using Ito calculus we derived solutions for the turbulent kinetic energy of the particle phase and the particle-gas velocity correlations. If particle-particle collisions and particle feedback on the turbulence are neglected the new relations approach the ones derived by Tchen and Hinze but if these effects are included additional terms in the relations appear. In this study we only use a very simple model for the particle-particle collisions. The new relation and the classical relation of Tchen and Hinze for the particle turbulent kinetic energy as well as a relation based on the kinetic theory of granular flows have been implemented in a two-fluid model for turbulent gas-particle flow in a channel in order to make comparison for different particle Stokes numbers. Results show that while the two-fluid model using Hinze's relations only gives good results for small Stokes numbers, the new relation yields significant improvements for a large range of Stokes numbers.

An Eulerian turbulent two phase flow model in cylindrical coordinates was developed in order to study evolving turbulent gas particle flows in a downward pipe flow. The model takes the feedback of the particles into account. Simulations shows good agreement with experiments and showed that the pipe length required for particle laden turbulent pipe flow to become fully developed is four times longer than for unladen flows, even for rather low mass loadings. The accumulation of particles in the near wall region showed a nonmonotonic behaviour and was found to be strongest for St=0.5.

Computations of the turbulent flow through plane asymmetric diffusers for opening angles from 8degrees to 10degrees have been carried out with the explicit algebraic Reynolds stress model (EARSM) of Wallin and Johansson [J. Fluid Mech. 403 (2000) 89]. It is based on a two-equation platform in the form of a low-Re K - omega formulation, see e.g. Wilcox [Turbulence Modeling for CFD, DCW Industries Inc., 1993]. The flow has also been studied experimentally for the 8.5degrees opening angle using PIV and LDV. The models under-predict the size and magnitude of the recirculation zone. This is, at least partially, attributed to an over-estimation of the wall normal turbulence component in a region close to the diffuser inlet and to the use of damping functions in the near-wall region. By analyzing the balance between the production and dissipation of the turbulence kinetic energy we find that the predicted dissipation is too large. Hence, we can identify a need for improvement of the modeling the transport equation for the turbulence length-scale related quantity.