SPH studies of complex fluid flow in pipes and stirred vessels

Lian, Xue (2025). SPH studies of complex fluid flow in pipes and stirred vessels. University of Birmingham. Ph.D.

Preview

Lian2025PhD.pdf
Text - Accepted Version
Available under License All rights reserved.
Download (7MB) | Preview

Abstract

Fluid dynamics is extremely vital in the 21\(^{st}\) century and to develop it, both experimental and numerical studies are necessary. Complex flows in industrial devices present many challenges, especially from a modeling point of view. Research is needed to improve the understanding and prediction of the flow behavior of solid-liquid flow systems and to investigate novel numerical modeling to guide the design of flows in different geometries. This thesis focuses primarily on numerical modeling and includes three main research areas. First, the Smooth Particle Hydrodynamics (SPH) method, coupled with the Discrete Element Method (DEM), was employed to study the influence of various parameters, including particle size, shape and density, pipe Reynolds number, and carrier fluid rheology on the Lagrangian migration trajectories of solid particles. Second, a coupled SPH-DEM model was developed to predict and obtain the velocity and solid concentration distributions of the particle-laden flows in the pipes, as well as information on flow field details, such as particle angular velocities, slip velocities, and vorticity of the liquid. Finally, the SPH method was applied to simulate Newtonian and non-Newtonian fluid flows in mechanically agitated stirred vessels with different impellers and Reynolds numbers. The velocity fields provided by SPH agree well with the Positron Emission Particle Tracking (PEPT) experimental data.

The investigation focuses on the inertial migration of a individual solid particle within a horizontal pipe flow, encompassing a broad range of tube Reynolds numbers spanning 350 to 3500. To model the flow dynamics, a coupled approach employing SPH in conjunction with the DEM is employed. Initially, the study focused on the behavior of single-phase flows comprising both Newtonian and non-Newtonian fluids, with a keen emphasis on validating the SPH-DEM predicted equilibrium radial position of a single particle. This validation process involves experimental validation utilizing a pipe flow loop configuration equipped with digital imaging technique. Subsequently, the validated SPH-DEM methodology is utilized to explore the impact of various parameters, including particle size, particle density, pipe Reynolds number, carrier fluid rheology, and particle shape, on the equilibrium radial position of a solitary particle within the flow. The insights garnered from this comprehensive investigation are expected to offer valuable contributions to the understanding and characterization of particle-laden flows across a diverse array of engineering applications.

A Lagrangian–Lagrangian particle-based SPH–DEM numerical approach was applied to simulate the horizontal laminar pipe flow of coarse particle–liquid food suspensions. SPH–DEM was able to predict the single-phase flow of various non-Newtonian fluid rheologies including the carrier fluid used to convey food particles with a high degree of accuracy, as validated by analytical solutions of radial velocity profiles. The particle–liquid flow simulations were also successfully validated using experimental measurements obtained by a technique of PEPT. The radial particle velocity profiles, radial solid phase distributions, as well as particle passage time distributions were accurately predicted at solid loadings varying from 10 to 40vol.%. Radial particle velocity profiles are asymmetric, with the maximum being located above the centerline. Such asymmetry disappears at high solid concentrations, and the maximum moves to the center. Simulations yielded detailed information on the local dynamics of the two-phase flows investigated, including translational as well as rotational particle slip velocities, flow pressure field, and fluid vorticity. While local translational particle slip velocities, as expected for nearly neutrally buoyant particles, were vanishingly small, particle angular (slip) velocities were significant. Spin is positive (anticlockwise) in the top half of the pipe cross section and negative (clockwise) in the bottom half. Particle spin is fastest closest to the wall where liquid velocity gradients are highest, reducing to zero at the center of the pipe before switching direction. As the solid load increases, close to the particle packing fraction, the particles pack so densely that their motion (e.g., rotation or spin) is limited. Spin becomes negligible except near the walls. The capability of the SPH-DEM methodology was also successfully validated in non-Newtonian single-phase as well as in two-phase particle-liquid flows by comparing the local phase velocity flow field, radial particle distribution and particle passage times in three-dimensional (3D) simulations. The simulations also yielded accurate predictions of flow pressure drop. In addition, detailed information was afforded on local particle spin, fluid pressure and carrier fluid vorticity. The results demonstrate the high capability of the proposed numerical framework to predict the complex features of complex particle-liquid flows in pipes.

As is known, the mixing characteristics of non-Newtonian fluids in a mixing vessel are quite different from Newtonian fluids mixing flows due to the special relationship between viscosity and shear rate. The pseudo cavern and cavern are, respectively, the classic mixing features of shear-thinning fluids and viscous fluids. The accuracy of the SPH model in simulating different mixing features of non-Newtonian fluids within stirred vessels was validated through comparisons with experimental data obtained from particle image velocimetry (PIV), PEPT, and planar laser-induced fluorescence (PLIF) measurements. Flow fields generated by the mixing of Newtonian and non-Newtonian fluids in single-phase flows with different Reynolds numbers in transitional and laminar flow regimes are well predicted by the SPH model. Detailed 3D distributions of liquid-phase velocities can be captured using the SPH model. However, due to the complex flow mechanisms and shortcomings of SPH model, SPH modelling at high Reynolds numbers, especially turbulent flows, is still a great challenge for SPH at present.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):