Download files
Abstract
Turbulent particle-laden flows are common in many engineering applications, such as pulverised coal or biomass combustors, particle-based solar energy receivers, etc. The interactions of particles and turbulence have a significant impact on the performance of these practical devices. However, a fundamental understanding of these interactions is still lacking. In this thesis, direct numerical simulations (DNS) of experimental particle-laden jet flows were reported with excellent agreement between the predictions and the experimental measurements, made possible by incorporating accurate experimental boundary conditions for both the particles and flow in the DNS. The simulations were performed at Stokes numbers (St) ranging from 0.3 to 11.2 using the point-particle approach in an Eulerian-Lagrangian framework. The DNS results show that lighter particles exhibit small-scale clustering already a few diameters downstream the nozzle, whereas this effect is less obvious for larger particles. The presence of particles has an attenuation effect on the turbulent velocity of the flow and this effect is enhanced as St increases. A velocity difference between the dispersed phase and the gas phase is present throughout the entire domain and is found to be the largest for the highest St case. A self-similar nature of the particle-laden jets is observed in terms of the flow mean and turbulent velocities as well as the particle concentration. A global analysis of the extent of preferential concentration reveals that when scaled appropriately, the particle clustering is maximal for St close to unity. An estimation of the dominant length scale of clustering shows that the particle clusters are under the influence of viscous effects. Further, a novel method is introduced to identify and accurately characterise the particle clusters from a two-dimensional (2D) thin layer at the centre of the jet. The proposed method is applied to the DNS case of St = 1.4, which is chosen based on its particular relevance in the study of industrial burners and solar thermo chemical reactors. A variety of cluster shapes and sizes are observed throughout the domain, with the characteristic dimensions distributed in a log-normal fashion. The strong anisotropic nature of the shear flow induces preferential orientation of the clusters so that they are aligned at oblique angles with the jet centre line. Overall, the conclusions of this thesis support and complement the prior experimental and numerical investigations. In the context of DNS investigation of turbulent two-phase flows in the TWC regime, a good overall correspondence between the DNS results and experiments were observed, which is motivating. The use of DNS revealed a number of unexpected and novel features that merit further investigations, especially in regards to preferential concentration of particles in the jet.