Modelling of the ultrasonic disintegration of activated sludge N. Lambert*, I. Smets**, J. Van Impe**, R. Dewil* *KU Leuven, Department of Chemical Engineering, Process and Environmental Technology Lab, Jan Pieter De Nayerlaan 5 – BE-2860 Sint-Katelijne-Waver - Belgium (e-mail: nico.lambert@cit.kuleuven.be). **KU Leuven, Department of Chemical Engineering, Chemical and Biochemical Process Technology and Control Division – W. de Croylaan 46 – BE-3001 Heverlee - Belgium (e-mail:ilse.smets@cit.kuleuven.be) Abstract: Ultrasonic treatment of waste activated sludge is one of the possibilities to reduce excess sludge production through the mechanism of sludge disintegration and cell lysis. In the past, several attempts have been made to model the process of solubilisation of the particulate volatile suspended solid part of the activated sludge (VSS) into soluble COD (sCOD). However, the focus of these models was predominantly on predicting an efficiency factor for the release of sCOD (Disintegration Degree, DD COD ) and provided no information on the release of nutrients and the instantaneous reduction of VSS. Moreover, often insufficient influential variables were included in the model equations, making the models only applicable on the training dataset of their own experimental research. This paper, therefore, seeks to build a simple model, which contains all influential input variables, that can predict not only the sCOD release but also the nutrients release (ortho-PO 4 -P and soluble Kjeldahl nitrogen) and VSS reduction simultaneously. Therefore, in first instance, a Principal Component Analysis (PCA) is carried out on the input and output data matrix of obtained experimental observations that will be used as training data. In this way, certain correlated input variables and independent output variables can be removed from the model, in order to increase its simplicity and predictive nature. Then, the model is built on the basis of Partial Least Squares Regression (PLS-R) and a part of the observations is used to validate the predictive strength of the model. Keywords: wastewater treatment modelling, ultrasonic disintegration, excess activated sludge reduction. 1. INTRODUCTION Focused on considerably reducing the generation of excess sludge in activated sludge wastewater treatment plants, cell lysis and cryptic growth of microorganisms has been found very effective to reduce Waste Activated Sludge (WAS). Ultrasonication of WAS and RAS (Return Activated Sludge) is capable of destroying the flocs within their matrix of extracellular polymeric substances (EPS) to result first in much smaller flocs and afterwards in cell death and lysis (Xiaoxia Wang, 2010). Cell lysis is a process in which the cell disintegrates after which the organic, nitrogen and phosphorus constituents of the cell provide an autochthonous substrate that contributes to the organic loading of the wastewater. Growth on such lysis products is described as cryptic growth and results in a reduced overall biomass production (Low and Chase, 1999). In most of the scientific publications on the topic of ultrasonic sludge disintegration the efficiency of the process is assessed by the release of sCOD in the supernatant liquid of the activated sludge. It is assumed that the solubilisation of particular COD (pCOD) into soluble COD (sCOD) involves first deagglomeration of the active sludge flocs (floc disintegration), and, second, the release of the cellular material in the supernatant liquid as a result of the breakage of the cell wall (cell lysis) (Si-Kyun et al.). However, in current literature little attention is given to the release of nitrogen and phosphorous compounds in the supernatant liquid during the process of sludge disintegration although, e.g., Bougrier et al. (2005) highlight that nitrogen is mainly released as proteins and amino acids and that, at moderate specific energies, a solubilisation degree of 40% can be expected. Several previous research papers reported relevant models for single alkaline activated sludge treatment, single ultrasonic sludge disintegration and the combined treatment process (e.g., Li et al., 2010; Wang et al., 2005 and Kim et al., 2010). All these models use a different combination of independent predictive variables to determine the sCOD release in the supernatant liquid of the activated sludge. It is evident that there are still a lot of questions that remain unanswered. In the first place, it can be said that the choice of the appropriate predictor variables to describe the untreated activated sludge itself and to characterize the process conditions of the ultrasonic sludge treatment is of great importance to generate an appropriate ultrasonic sludge disintegration model. For example, it seems logical to characterize the treated sludge on the basis of both the initial (biomass quantity related) MLSS 0 and MLVSS 0 concentration. In this way it will be taken into account whether or not we are dealing with a more organic or more inorganic activated sludge or whether the sludge sample is highly thickened or not. The ultrasonic reactor design can be described by the ultrasonic density (D S - W/cm 2 ) and the type of ultrasonic horn or transducer can be quantified by the ultrasonic intensity (I S – W/mL). To have a global view on Preprints of the 19th World Congress The International Federation of Automatic Control Cape Town, South Africa. August 24-29, 2014 Copyright © 2014 IFAC 7122