An essential tool for WRRF modelling: a realistic and complete influent generator for flow rate and water quality based on data-driven methods

Abstract

Modelling, automation, and control are widely used for water resource recovery facility (WRRF) optimization. An influent generator (IG) is a model, aiming to provide the flowrate and pollutant concentration dynamics at the inlet of a WRRF for a range of modelling applications. In this study, a new data-driven IG model is proposed, only using routine data and weather information, and without need for any additional data collection. The model is constructed by an artificial neural network (ANN) and completed with a multivariate regression to generate time series for certain pollutants. The model is able to generate flowrate and quality data (TSS, COD, and nutrients) at different time scales and resolutions (daily or hourly), depending on various user objectives. The model performance is analyzed by a series of statistical criteria. It is shown that the model can generate a very reliable dataset for different model applications. HIGHLIGHTS A data-driven influent generator (IG) to create a long-term time series for flow and pollutant concentration at hourly and daily frequencies.; The IG application only needs WRRF routine data and weather information, without need for any additional data collection.; The IG also generates nutrient concentration time series.; A random walk stochastic process is added to better approximate observed temporal variability.;

Keywords

• data mining
• digitalization
• machine learning
• multivariate regression
• wastewater modelling
• water quality