Modelling the nitrification in a full-scale tertiary biological aerated filter unit

Abstract

A wastewater biofiltration model is used to assess its capacity to reproduce the treatment behaviour of a plant-sized tertiary nitrifying biofilter unit. It is calibrated on two different types of datasets collected at the Seine-Aval biofiltration plant (Achères, France): grab samples at several heights inside the media bed and a long-term daily plant monitoring over a 1-year period. The model parameters are first calibrated to fit the dynamics observed in the media bed, after which the model is compared to the second dataset. Further parameter changes are then made if necessary and the model is once again compared to both datasets to ensure its ability to predict the treatment behaviour on both size scales. The calibrated model provides correct predictions for most observed nutrient variables for both datasets. An overestimation of the oxygen transfer under a summer, low ammonia load period however leads to a slight underestimation of the nitrifying efficiency of the biofilters. Statistical score computation corroborates the model accuracy as the mean error scores usually remain low. They also point to a certain weakness of the model regarding the suspended solids filtration. Both datasets are overall correctly modelled using a single parameter set. Most of this parameter set is close to or contained in value ranges found in the literature. The parameters related to aeration, however, seem to be slightly higher than what is reported elsewhere.