Abstract
Removal of algae present in domestic wastewaters of stabilization ponds can be a significant challenge in the supply of recycle water for re-use purposes. Algae can be present in high numbers, especially in warm to hot summer and autumn months diminishing water quality when most needed for reuse purposes. Consequently further treatment of wastewaters may be needed for removal of turbidity and algae, prior to recycled water supply. A large recycle water treatment plant in South Australia - the Bolivar DAFF plant utilizes alum as primary coagulant for algal and suspended solids removal. Alum dosing varies significantly throughout the year, depending on source water (from stabilization ponds) quality. Algae can be present at bloom levels (>10 6 cells per mL) and the water can be at very high pH (~ 10); and with high alkalinity, very high doses of alum may then be required. With this high variation in source water quality, adjustment of coagulant dosing is continually needed to be performed by operators, at times diurnally. This is done on the basis of meeting target treated water quality (of low turbidity) and minimizing costs of chemical use. Coagulant use at the plant has been highly variable, and with consideration that dosing at times, may not be at optimum. As part of continual process improvement, research was instigated to determine if model development could be achieved for prediction of coagulant dose requirements. Data available was sourced from documented treatment plant operations over a number of years where wastewater quality varied seasonally, and from laboratory investigations incorporating jar tests and water quality analyses. These water quality parameters included untreated water algae, pH, turbidity, ambient temperature and treatment plant daily processing volumes. The model developed is based on apparent relationships between each of the source water qualities with alum dose rates applied at the treatment plant. Data was highly scattered for most parameters and model development was in context that error was potentially in operational dose application, as well as in model fitting i.e. that actual dosing data was not presumed to be 'true' or optima, and general trend equations might best reflect the optima. From general trend relationships and fitted algorithms of each, a model was built where the outputs of each sub-model is averaged to provide a fitted or predicted value. Model options include selection of input variables (e.g. pH, turbidity, algal number) and weighting of importance to each parameter. Treatment plant operators have accepted the model as an additional tool for decision making. In this paper the need for and the design of the model are discussed.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have