Keeping Track of pH Sensors in Biological Wastewater Treatment systems

In the wastewater sector, water quality sensors constitute the most sensitive component to online monitoring and control systems. Consequently, the purported benefits of automation remain an elusive goal in many situations. This is especially true for decentralized wastewater resource recovery facilities (WRRFs) as studied in this work. Indeed, the use of frequent on-site sensor inspection and maintenance is typically economically prohibitive. For this reason, a tremendous amount of research has focused on the development of specialized algorithms for fault detection, isolation, and diagnosis. Such algorithms typically assume (i) that some historical normal data are available for algorithm tuning and (ii) that faults generally do not appear simultaneously in multiple sensors. In this work, we challenge these assumptions by means of a long-term study of wear-and-tear in pH sensors exposed to nitrified urine, a high-strength anthropogenic wastewater. Our results indicate that both assumptions are invalid, thereby preventing an effective use of the majority of fault detection algorithms available today.