Statistical and microbial analysis of bio-electrochemical sensors used for carbon monitoring at water resource recovery facilities

Abstract

Real-time carbon monitoring of wastewater using bio-electrochemical sensors coupled with advanced data analysis methods provides WRRFs with an opportunity for efficient wastewater quality monitoring and an early warning tool for plant upsets.

Integrating Bio-Electrochemical Sensors and Machine Learning to Predict the Efficacy of Biological Nutrient Removal Processes at Water Resource Recovery Facilities

“…Given the dynamic nature of anodic electroactive biofilm formation under different environments and operating conditions, it would not be possible to expect identical signal baselines from BESs installed in different unit processes and locations; this is due to differences in the microbial composition of anodic biofilms, biofilm thickness, and mass transfer considerations. It has also been reported that temperature variations can alter the BES signal baseline; therefore, it would not be possible to explicitly determine the measurement range of BESs applied in full-scale industrial environments and under uncontrolled conditions with a high degree of confidence. The full potential of the information-rich BES signal obtained from this study could be uncovered using advanced statistical and ML techniques, and was not a suitable replacement as a direct chemometric parameter for biodegradable C.…”

“…The consistently positive contribution likely indicates that the BES signal responded to naturally occurring available biodegradable C in the influent of the five-stage modified Bardenpho process. As the electroactive biofilm on the anode surface of BESs primarily consumes biodegradable organic compounds with a strong sensitivity to VFA, , a direct response between the concentration of soluble C and the BES signal intensity was expected. Because C metabolism is one of the main driving forces of these BNR processes, knowledge about the BES signal could never have a negative impact on the model output (MNR in the ANX1).…”

“…BES signals obtained from industrial-scale wastewater treatment unit processes respond to temperature variations and demonstrate trends and/or seasonalities. 6 The FB prophet additive model was selected to predict the BES signal as it can track nonlinear trends consistent with seasonalities and holiday effects in time-series data; the characteristics which also align well with centralized wastewater treatment applications. The FB prophet model was developed by Facebook's data science research team as an additive model and made public in 2017.…”

“…Bio-electrochemical sensors (BESs) provide real-time information about C concentration and the metabolism of readily biodegradable organic compounds in a broad range of wastewater treatment environments. 6,7 These devices leverage electroactive anode respiring biofilms to generate an amperometric signal that acts as a proxy for available biodegradable C. More than two decades of research has been dedicated to understanding the effectiveness of BESs for the recovery and detection of heavy metals, 8,9 environmental toxicity detection, 10 wastewater resource recovery, 11 and measurement of volatile fatty acids (VFAs) 12 and biochemical oxygen demand (BOD) under controlled laboratory conditions through linear correlation of bio-electric current and concentration. 13 pH and temperature variations, 14 wastewater composition, 15 and reaction time, 16 can have a major impact on the measurement performance of these biosensors.…”

“…Other operational variables, including pH and temperature variations, wastewater composition, and reaction time, can have a major impact on the measurement performance of these biosensors. A recent study focused on field deployment of BESs for C monitoring in industrial-scale WRRFs demonstrated that nonlinear and complex dynamics dominate the variance of the BES signals, elucidating the need for advanced analytical methods to unlock the full potential of BESs for real-time C monitoring …”