Large Eddy Simulation of turbulent dispersed multi-phase flows for engineering applications - development of Eulerian-Lagrangian algorithms

Chen, Boyang (2023). Large Eddy Simulation of turbulent dispersed multi-phase flows for engineering applications - development of Eulerian-Lagrangian algorithms. University of Birmingham. Ph.D.

Preview

Chen2023PhD.pdf
Text - Accepted Version
Available under License All rights reserved.
Download (26MB) | Preview

Abstract

This PhD thesis presents a numerical investigation of turbulent dispersed multiphase flows based on Large-Eddy simulation (LES). A wide range of engineering scenarios involve multiphase flow and require a fundamental understanding of the fluid mechanisms induced by different scales of motion. The numerical model is developed based on an in-house Computational Fluid Dynamics (CFD) code using a point-particle Eulerian-Lagrangian (PP-EL) algorithm and aimed for a universal numerical approach to simulate practical multiphase flows. The core of the thesis document is composed of three journal manuscripts where subsequent developments of this algorithm are applied to flows of engineering interest (i.e., chemical, ocean and environmental engineering) and validated versus experiments. The numerical results in each work remarkably agree with experimental results.

The first chapter starts by showing the first application of an Eulerian-Lagrangian strategy to the prediction of buoyancy-induced mixing with a thorough and successful validation. The aim of this research is to explore the optimal design of aerators in chemical reactors. This is achieved by quantifying the buoyancy-driven mixing produced by bubble screens and different arrangements of the individual plume, and examining the impact of gas flow rate and the depth of the reactor on the mixing time. The outcome of this work indicates a better performance provided by bubble screens, saving at least 20% of the energy usage in aeration.

The second chapter focuses on gravity currents driven by inertial particles (particledriven currents or turbidity currents). The solver uses an Eulerian-Lagrangian pointparticle algorithm to provide four-way coupling of the ambient fluid and the suspension of solid particles, with the interaction between particles being handled by a soft-sphere collision model. The results explore the dynamics of turbidity currents when compared to density-driven gravity currents, highlighting the influence of inertial particles on the propagation of the currents and the scales of motion of the coherent structures generated at the shear layer. The simulations reveal that the dissipation rate of the turbulent kinetic energy within the current is mostly due to the contribution of the settling of solid particles. It is shown as well that the removal of the lock gate affects the early flow development whereas has little influence on the front speed of the current.

Combining the implementations made in the first two chapters, the third chapter introduces a novel modelling tool for the activated sludge process (ASP) based on large-eddy simulation and multiphase Eulerian-Lagrangian coupling. The model uses an Eulerian-Lagrangian point-particle algorithm that respects the discrete nature of both sludge flocs and air bubbles. Four-way coupling is implemented, where the interaction between solid particles is handled by a soft-sphere collision model. Subsequently, the integrated model is used to simulate a realistic scenario within the aeration basin of a wastewater plant and explore its results across a wide parameter range (aerator distribution, dissolved Oxygen levels, air flow rate, sludge size, bubble size). The results indicate that the initial dissolved Oxygen levels within the basin (related to weather conditions and aeration frequency) are critical for sludge activation, with initial anoxic conditions being very taxing. For a given flow rate, bubble screens (i.e, more aerators) provide significantly better performance.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):