Practical application and calibration of apparent half-saturation coefficients for aerobic granular sludge modelling: Limitations and value

Aerobic granular sludge (AGS) models for process optimization and control require balancing model complexity and accuracy. This study explores the strengths and limitations of using 0D-biofilm models based on apparent-half saturation coefficients (KS,app) and evaluates a recently proposed procedure, based on 1D-biofilm model outcomes (1D-biofilm-to-KS,app procedure), for calibrating KS,app usig routine monitoring data commonly available from full-scale facilities. Two case-studies were investigated, differing in reactor size and process conditions, both characterized by operation as sequencing batch reactors. The implications of using such estimates in 0D-biofilm AGS models were assessed, also considering the possibility of incorporating vertical gradients in the reactor model during the non-mixed anaerobic feeding phase. When implemented in a completely mixed batch reactor model, significant relative errors (30–82 %) were observed in predicting steady-state effluent concentrations (ammonium, phosphorous and biodegradable COD), with negligible improvements when vertical concentration gradients were included. Based upon sensitivity analysis outcomes, direct calibration of KS,app was carried out by fitting predicted vs. average measured effluent concentrations. The application of these new KS,app values under dynamic conditions proved that 0D-biofilm models calibrated using routine monitoring data may not reliably capture the typical variability of AGS processes (Theil's Inequality Coefficient > 0.36 for ammonium and phosphorous).