An investigation of free surface hydraulic structures using large eddy simulation and computational fluid dynamics

Publication Type:
Issue Date:
Full metadata record
Files in This Item:
Filename Description Size
Thumbnail01front.pdf1.75 MB
Adobe PDF
02whole.pdf40.81 MB
Adobe PDF
The work presented in this dissertation is essentially a thesis in three distinct parts (single fluid validation, two fluid validation and data analysis) rather than the established approach for the development of a novel computational fluid dynamics solver. First, the progression is a traditional one, in which an existing technique was applied to a new area and subsequently extended. Second, from detailed analysis of the large volume of data generated in the validation process, a number of insights were gained into the flow features of the prototypes investigated that extended beyond a traditional validation study and discovered a number of new physical phenomena. Previous researchers have used monotonically integrated large eddy simulation (MILES) methods to investigate a range of flows including turbulent decay in rotary valve engines and rocket body dynamics. MLES methods have the distinct advantage over standard LES simulation techniques in that they promise to provide similar levels of detail and accuracy but at a fraction of the computational cost. However, to the author’s knowledge these techniques have not been applied to the prototype problem of this thesis: cylinders in cross flow without and with free surfaces. Hence the raison d’étre of this thesis: to apply a faster yet equally accurate CFD method to a free surface problem via a validated single fluid investigation. Specifically, the progression was to first validate the method against a single right square cylinder in cross flow without a free surface and then to extend the method to a right circular cylinder in cross flow with a free surface. With the right square cylinder without free surface the research focussed on the extensively studied configuration of a two dimensional square cylinder at a Reynolds number of 22 × 10³. Despite the agreement of the validation parameters with published data, detailed examination of the flow field revealed inconsistencies in the modelled results. In particular the power spectrum decay of the data appear too “easy” to obtain, indicating possible flaws in the theoretical basis, while correlation data apparently supports a conclusion that the previous assumption of four diameters domain width is too narrow to provide an uncorrelated flow region. The free surface physics of the circular cylinder model was captured with the volume of fluid method and was applied to Reynolds number flows based on cylinder diameter of between 27 × 10³ and 54 × 10³. These flows, at the provided grid resolution, push the lower boundary of what can be called MILES, yet interpretations of the results indicates that the model is accurately capturing the physics of the flows.
Please use this identifier to cite or link to this item: