Prediction of odor complaints at a large composite reservoir in a highly urbanized area: A machine learning approach

Abstract

Odorous compound emissions and odor complaints from the public are rising concerns for agricultural, industrial, and water resource recovery facilities (WRRFs) near urban areas. Many facilities are deploying sensors that measure malodorous compounds and other factors related to odor creation and dispersion. Focusing on the Metropolitan Water Reclamation District of Greater Chicago's (MWRDGCs) Thornton Composite Reservoir (7.9 billion gallon capacity), we used meteorological, operational, and H2S sensor data to train a 3‐day advance‐warning predictor of local odor complaints, so as to implement targeted odor prevention measures. Using a machine learning approach, we bypassed difficulties in modeling both physical dispersion and human perception of odors. Utilizing random forest algorithms with varied settings and input attributes, we find that a small network of H2S sensors, meteorological data, and operational data are able to predict odor complaints three days in advance with greater than 60% accuracy and less than 25% false‐positive rates, exceeding MWRDGC’s standards required for full‐scale deployment. Practitioner points A random forest algorithm trained on H2S, weather, and operations data successfully predicted odor complaints surrounding a large composite reservoir. Thirty‐two data attribute combinations were tested. It was found that H2S sensor data alone are insufficient for predicting odor complaints. The best predictor was a Random Forest Classifier trained on weather, operational, and H2S readings from the reservoir corner locations. This study demonstrates odor complaint prediction capability utilizing a limited set of data sources and open‐source machine learning techniques. Given a small network of H2S sensors and organized data management, WRRFs and similar facilities can conduct advance‐warning odor complaint prediction.

Advanced machine learning application for odor and corrosion control at a water resource recovery facility

“…In the past decades, ML models have been successfully applied in the wastewater industry to tackle issues that are challenging using conventional mathematical and kinetic models. For instance, the concentration of wastewater quality indicators (Granata et al, 2017;Kundu et al, 2014;Liu & Xiao, 2019;Quan et al, 2018;Verma et al, 2013;Zhao et al, 2017), H 2 S level (Zounemat-Kermani et al, 2019), odor concentration (Kang et al, 2020), odor complaints (Mulrow et al, 2020), concrete corrosion loss (Zounemat-Kermani et al, 2020), and VFAs concentration (Kazemi et al, 2020;Tay & Zhang, 2000) were successfully predicted using different types of ML models. Liu and Xiao (2019) used back propagation artificial neural network (ANN) to provide accurate prediction of ammonia-nitrogen (NH 3 -N) in wastewater.…”

Gaseous emissions from wastewater facilities

“…A machine learning approach using random forest algorithm trained on H 2 S, weather, and operations data to predict odor complaints 3 days in advance was successfully demonstrated with >60% accuracy and <25% false‐positive rates, suggesting potential applications in WRRFs and similar facilities (Mulrow et al., 2019).…”

Artificial Intelligence-Assisted Prediction of Effluent Phosphorus in a Full-Scale Wastewater Treatment Plant with Missing Phosphorus Input and Removal Data

“…Compared to the traditional dynamic–mechanistic models that require a thorough understanding of underlying reactions and system-specific calibration, ML and DL models are easier to apply by identifying the input–output data relationship without reflecting the physical, biological, or chemical processes. , The key applications of ML and DL models for wastewater treatments include predicting key parameters, detecting anomalies, precise dosage control, and advanced automated control. ,− For example, the random forest (RF) model was used to accurately predict the effluent phosphorus concentration in the studies of full-scale WWTPs with high coefficients of determination ( R 2 ). , However, phosphorus removal with alum, which is commonly used in small-scale WWTPs, has not been well studied using ML/DL models. In particular, some small-scale WWTPs lack the use of real-time sensors and do not monitor all of the key parameters such as the influent phosphorus concentration, making it challenging to establish ML/DL models .…”