Abstract

In hydraulic engineering, a transition facilitates a change in the direction, slope or cross-section of an open channel or pipeline. This thesis focuses on expanding flows in open-channel expansions and hydraulic jumps. This thesis uses wall-resolved Large Eddy Simulation (LES) to study the two-phase turbulent flows in several three-dimensional (3-D) geometries, including non-prismatic open channels and sloping pipes. The LES predictions compare well with corresponding experimental results and benchmark solutions.

First, for a turbulent bistable flow approaching a straight-wall channel expansion, either of two stable flow states can occur, depending on the flow history. The thesis aims to reveal the ensemble-average flow characteristics and explore effective ways to control bistability. Turbulent eddies initiated by shear instability dominate those associated with sidewall-friction force, which is responsible for the occurrence of bistability. Fitting a simple hump at a flat-bottom expansion is an effective way to suppress bistability.

Next, hydraulic jumps in sloping pipes are investigated to achieve an improved understanding of jump behaviours driven by different discharges and slopes. Flow behaviours such as free-surface fluctuations and jump-toe oscillations resemble the classical hydraulic jump on horizontal floors. Depending on the discharge and slope, the resulting jump can be a complete or an incomplete jump. The latter causes flow choking downstream, which has severe consequences on drainage conditions in sewer pipes. The Okubo-Weiss parameter is a new way to subtly delineate the region of hydraulic jump.

Last, to study turbulent flows in a non-prismatic warped expansion using LES, the thesis discusses rigorous strategies for model setup, parameter selection and parametric value assignment. Mapping mean-velocity distributions from experimental data, combined with the spectral synthesiser approach for velocity fluctuations, gives a satisfactory inlet condition; alternatively, a 1/7th power-law for the mean-velocity, combined with the vortex method for the fluctuations, is acceptable.

Compared to a prismatic channel, a non-prismatic channel exhibits more complicated eddy motions and turbulence interactions. This thesis contributes to a systematic assessment of computational strategies, result visualisation, and analysis, all relevant to practical applications.