Generation of (synthetic) influent data for performing wastewater treatment modelling studies

The success of many modelling studies strongly depends on the availability of sufficiently
long influent time series - the main disturbance of a typical wastewater treatment plant (WWTP) -
representing the inherent natural variability at the plant inlet as accurately as possible. This is an
important point since most modelling projects suffer from a lack of realistic data representing the
influent wastewater dynamics. The objective of this paper is to show the advantages of creating
synthetic data when performing modelling studies for WWTPs. This study reviews the different
principles that influent generators can be based on, in order to create realistic influent time series. In
addition, the paper summarizes the variables that those models can describe: influent flow rate,
temperature and traditional/emerging pollution compounds, weather conditions (dry/wet) as well as
their temporal resolution (from minutes to years). The importance of calibration/validation is
addressed and the authors critically analyse the pros and cons of manual versus automatic and
frequentistic vs Bayesian methods. The presentation will focus on potential engineering applications
of influent generators, illustrating the different model concepts with case studies. The authors have
significant experience using these types of tools and have worked on interesting case studies that they
will share with the audience. Discussion with experts at the WWTmod seminar shall facilitate
identifying critical knowledge gaps in current WWTP influent disturbance models. Finally, the
outcome of these discussions will be used to define specific tasks that should be tackled in the near
future to achieve more general acceptance and use of WWTP influent generators.