Waste stabilisation ponds (WSP) are a popular form of wastewater treatment worldwide, especially for rural-based manufacturing plants and small community sewage treatment. Ponds offer a robust and operationally simple technology, which are inexpensive where land is available, and have the potential to provide a considerable degree of treatment. However the continued use of WSP is being undermined by their inconsistent performance relative to current discharge requirements, particularly with respect to suspended solids, pathogen and nutrient removal. In a climate of increased public awareness of pollution, and the ever more stringent environmental protection regulations, novel pond designs need to be developed, and existing ponds retro-fitted, to improve their performance. This dissertation investigated the hydraulic modelling of non-mechanically mixed ponds, and produced a modelling framework from which improved pond designs could be evaluated.
Computational fluid dynamics (CFD) simulations were used to develop models which were able to predict the hydraulics of arbitrarily shaped, non-mechanically mixed ponds under controlled conditions. The models represent an important departure from traditional pond modelling techniques, which are based on either historical experience or simple hydraulic and reaction models. The CFD approach overcomes the main limitation of these models, as it accounts for spatial variations of parameters within a pond such as fluid velocity, or pollutant concentration. This allows for the prediction of pond hydraulics based on the pond geometry (such as inlet configuration, pond shape or baffle placement), pond inlet boundary conditions and the fluid properties. Thus CFD models allow the rapid investigation of the effect of design modifications on pond performance.
The WSP models were designed using a two stage process. The first stage, a steady state simulation, calculated the velocity and turbulence fields for the pond; the second stage, a transient numerical tracer, utilised the underlying steady state results to calculate the advection and diffusion of a tracer species. The species concentration at the outlet was then integrated to produce residence time distributions (RTD) and other quantities which were used to characterise the pond hydraulics, and quantitatively compare the models with experimental results to assess the pond¡¦s performance. These techniques could be applied to any numerical pond flow model, and are a discerning test of the model¡¦s consistency.
RTD generated from two-dimensional (2-D) CFD simulations were compared to experimental RTD derived by Mangelson and Watters (1972). In one of the three geometries simulated, the 2-D CFD model successfully predicted the experimental RTD. However, the flow patterns in the other two geometries were not well described, due to the difficulty of representing a three dimensional (3-D) inlet in the 2-D CFD model. As no general relationship could be found for approximating a 2-D inlet in 3-D, full 3-D simulations were used to model the unsuccessful cases. The 3 D simulations provided much improved results, predicting all the major features of the RTD over the first residence time, and matching exponential decay of the RTD after this period. Due to the uncertainty in the exact experimental inlet dimensions, a range of inlet depths were simulated. This showed that the CFD model was sensitive to changes in the inlet configuration, and using the appropriate inlet depth, the simulated RTD matched the experimental results well. A sensitivity analysis of the effect of the inlet turbulent boundary conditions and tracer molecular diffusivity for the k-Õ turbulent model, showed the RTD was insensitive to these properties, thereby confirming similar results in related systems (Benelmouffok, 1989; De Vantier and Larock, 1987). This is significant for future pond modelling, as these properties are difficult to measure experimentally or predict reliably.
Tracer studies were performed in this dissertation on five full-scale pond systems. In Tasmania three identical sewage ponds with different inlet and baffle configurations were investigated. However wind conditions in this locality masked any effect of these modifications. Tracer studies were also performed on sugar mill ponds near Mackay. While the models predicted qualitative consistent RTD results, they did not match the experimentally measured RTD due to uncontrolled environmental mixing factors and the long residence times of these ponds. A preliminary investigation of the effect of wind mixing was undertaken by imposing a velocity to the top surface of the model. These results confirmed the strong influence of even small wind velocities due to the large surface area of the ponds.
Practical experience has indicated that the pond hydraulics are often the limiting factor in pond performance. Both experimental and simulation results have confirmed this through the presence of short circuiting and dead zones within the pond. Three baffle designs were assessed, all of which improved the pond hydraulics by either dispersing the inlet jet, or utilising the jet to generate specific pond mixing.
Finally the work in this thesis has highlighted a number of other areas for future investigation. These include reservations over the use of RTD to characterise full-scale pond hydraulics, and considerations regarding the most efficient use of the inlet mixing in ponds. The hydraulic models developed in this dissertation can be extended to include solids, stratification and reaction models, which would enable the direct validation of the model based on physical or chemical parameters. In addition, a coupled flow and reaction model would provide a tool that could be used to truly optimise pond performance. This offers the possibility of tailoring the design of ponds for specific reactions, such as improved biological nutrient removal.