Water and the Circular Economy
Project Round
Project Number
512 - 001
Research Organisation
Water Futures

Quantifying Pathogen Removal in Australian Activated Sludge Plants

The Challenge

The performance of activated sludge plants in producing recycled water fit for its end use is influenced by numerous factors. Accurate models for predicting and validating pathogen reduction are a valuable tool in designing plants to meet regulatory and safety requirements. In this project several modelling strategies were considered and models were developed and compared to live pilot plant performance. Water recycling scheme developers working to comply with new guidelines on specified pathogen reduction levels and management of associated risks will find this study and its results useful in designing and operating ASPs.

The Project

The project began with an extensive literature review and a survey of data from Australian activated sludge plants.  This supported an analysis which linked operating parameters to reported pathogen reductions.

An activated sludge pilot plant was commissioned to demonstrate steady state operating conditions and provide a basis against which to test various models.

A rational for the selection of indicators for pathogen removal was developed and experiments with the pilot plant provided insights into the effect of variability in plant conditions on pathogen removal.

Finally a comparison of the performance of predictive models was conducted.  This linked operating inputs to monitored data.

The Outcome

Regression modelling showed promise as a predictive tool; however superior performance was obtained with a neural network model. Neural Tools (2009) from Palisade Corporation running on the readily available MS Excel spreadsheet software formed the basic platform for the neural network model which is comprised of algorithms
developed during this project.  In contrast to the regression models, the neural network does not require preliminary analyses or data transformations and is easier to use. The results of this project demonstrate that accurate real time predictions can be achieved in steady state operation of activated sludge plants.

Supporting documents