Water and the Circular Economy
Project Round
Project Number
72M - 7104
Research Organisation
SA Water Corporation

Quantification of Pathogen Removal in Australian Activated Sludge Plants (Phase 2)

The Challenge

Regulations for the validation of Active Sludge Plants (ASPs) currently require individual demonstration of pathogen reduction over the envelope of operating conditions that the plant may encounter. This is a cumbersome and expensive exercise.

The cost of safe and effective validation of ASPs can be reduced with the development of a reliable method for predicting pathogen reduction online to replace the need for collecting actual pathogen reduction validation data.

Development of prediction methods was undertaken via three tasks.  First, a thorough going literature survey aimed at illuminating world’s best practice in validation. Second, review of recorded Australian ASP data sets ensured actual local conditions were properly considered. Third, data was collected from pilot plant trials for analysis under normal and upset conditions to measure the accuracy of predictions of pathogen reduction.

The Project

Two predicative models for pathogen reduction in activated sludge processing were investigated with the aim of validating pathogen reduction predictions for a range of plant operating conditions.

Current literature was reviewed to summarise the findings of national and international studies.  Three key factors were found to influence pathogen reduction within the activated sludge process. These were; adsorption of pathogens to suspended solids, loss of pathogen infectivity and the efficiency of suspended solids removal through clarification processes.

The review of national data sets informed the design of the pilot plant facility and its accompanying experimental monitoring program.  It was found that few existing ASP facilities had collected data sets for the full range of pathogens of concern. The most commonly available data were for final effluent E. coli levels, as this is generally monitored for license and compliance purposes. Only one study reported on viruses separately. Most studies used indicator organisms for detecting viruses and also for parasites where they were included. The existence of correlations of pathogen populations with physical and chemical measures justified the proposed pilot plant experiments to fully characterize these relationships so that they can be used as predicators of ASP performance.

The experimental plan for the pilot plant studies aimed to identify reliable easily measured indicators for common pathogens and to investigate the effects of varying sludge age and increased flow through the plant.

The data analysis involved three components; comparison of data between trials; exploratory data analysis; and development of predictive models for each pathogen group monitored. The predictive model development involved the use of both covariance techniques and neural networks.

The Outcome

Nine pilot plant experiments were conducted to study the effects of sludge age and high flow conditions and determine reduction statistics for each pathogen and its indicator.

In matching prediction to measured performance, the covariance approach showed its possible utility as a predictive tool.  However, the neural network approach has better computational performance, lacks the need of complex preliminary analysis and does not require reduction of data into analysable subsets.  This ease of application recommends the use of neural nets as the superior predictive tool. Palisade’s Neural Tools software (Palisade Corporation 2009), allows live predictions to be made within a Microsoft Excel spreadsheet. With further development, online measurements of independent physical and chemical variables could be processed through a spreadsheet like this to provide a continuous reduction statistic for each pathogen of concern.

The predictive model approach offers a ‘surrogate’ for actual reduction value measurements, as pathogen reduction can be modelled using easily monitored online operating data. The neural net model could be further developed to ensure on line quality at critical control points and significantly reduce the work load associated with validating ASP performance.

This project follows on from quantifying pathogen removal in Australian activated sludge plants.

Supporting documents