9th IWA Symposium on Systems Analysis and Integrated Assessment 14-17 June 2015, Gold Coast, Australia Towards improved accuracy in modelling aeration efficiency by accounting for bubble size distribution dynamics Amaral, A.* , **, Bellandi, G.*, Mortier, S.T.F. C.*, Neves, R.**, and Nopens, I.* * BIOMATH, Department of Mathematical Modelling, Statistics and Bioinformatics, Ghent University, Coupure Links 653, 9000 Gent, Belgium ** MARETEC, Instituto Superior Técnico, Universidade de Lisboa, Av. Rovisco Pais 1, 1049-001 Lisboa, Portugal Keywords: oxygen transfer; population balance modelling; wastewater treatment Summary of key findings Aeration is the largest energy consumer in a wastewater treatment plant, which is why oxygen transfer optimization is fundamental if one wants to improve energy efficiency. Although oxygen transfer is strongly influenced by the bubble size distribution, most current aeration models do not include this influence explicitly and when they do, they assume a single average bubble size. Therefore, the motivation of this work is to investigate the impact of bubble size on oxygen transfer and subsequently initiate the development of an aeration model that accounts for the bubble size dynamics. Laboratory experiments were performed to study bubble size distribution dynamics at different air flow rates, based on which a population balance model (PBM) is formulated and simulated. This is the first step towards an improved aeration model which could substantially improve the prediction of oxygen transfer, contributing to more reliable and accurate optimization of oxygen transfer and consequently energy saving. Background and relevance Activated sludge is globally the most common biological treatment process for both municipal and industrial wastewater (Jenkis and Wanner, 2011). Aeration is crucial for this process and is the largest energy consumer in the plant (up to 45%) (Henze et al., 2007). Thus, oxygen transfer optimization is fundamental to deal with the important issue of energy efficiency. The bubble size distribution (BSD) is a dominant factor that influences the oxygen transfer, since it dictates the available interfacial area for gas-liquid mass transfer together with the gas hold-up (Dhanasekharan et al., 2005). However, aeration models commonly do not consider bubble size explicitly and if they do, they assume a single average bubble size (Nopens et al., 2015). This is an oversimplification as the BSD is spatially very dynamic. Most aeration models are  ! -based, where  !  is the overall gas-liquid mass transfer coefficient, a constant that can be divided in two parts: , the interfacial area for gas-liquid mass transfer and  ! , the local mass transfer coefficient (Wang, 2010). Taking into account that the interfacial area for gas- liquid mass transfer controls the oxygen transfer, the unrealistic assumption of  !  being constant can severely limit the accuracy of the aeration model and any optimisation study making use of it. The rigour with which aeration needs to be modelled is obviously driven by the modelling objective. Hence, the developments we propose here are intended for knowledge build-up as well as modelling goals that require a higher level of rigor in the aeration model. The interfacial area for gas-liquid mass transfer and, consequently, the BSD is not at all homogenous in a reactor. Bubbles grow in size due to the hydrostatic pressure drop while travelling from the bottom to the top of the reactor. Moreover, bubble coalescence occurs, further enlarging the bubble sizes. The latter process is more pronounced when the viscosity increases (Nopens et al, 2015; Wang, 2010) which is evidenced in membrane bioreactors where a substantial decrease in oxygen transfer efficiency is observed at elevated sludge concentrations. Population balance modelling (PBM) is a powerful tool able to predict BSD (Wang, 2010). The application of PBM is widespread in chemical engineering, however, information about its particular application to a wastewater treatment process is rather scarce (Nopens et al., 2015). The present work aims to (1) collect detailed experimental data on bubble behaviour and (2) initiate the development of a PBM able to predict the observed BSD dynamics.