Monitoring the condition of membrane bioreactors with a combined PCA - FC algorithm

The aim of this study is to assess the membrane fouling condition of membrane bioreactors (MBR) in real-time using solely commonly measured transmembrane pressure (TMP) data. The proposed algorithm is based on a combination of functional principal component analysis (FPCA) and fuzzy clustering (FC) to extract the relevant trends and patterns from the data and allocate them. The algorithm was tested on 6 months of data from a lab-scale MBR. Two major trends were visualised in the data and these could be linked to reversible and irreversible fouling. The gradual evolution between normal, reversibly and irreversibly fouled membrane clusters is also clearly visible. By providing crucial real-time information on the membrane state from inexpensive TMP measurements, the proposed method proves promising as online fouling monitoring system and can form the basis for dynamic membrane operation control.