Appraisal of Industrial Pollutants in Sewage and Biogas Production Using Multivariate Analysis and Unsupervised Machine Learning Clustering

Sewage composition analysis is important for understanding environmental impact and ensuring effective treatment processes. In this study, we employed multivariate analysis techniques to delve into the intricate composition of sewage. Specifically, we utilized Principal Component Analysis (PCA) and Detrended Correspondence Analysis (DCA) to uncover patterns and relationships among different types of sewage pollutants. Statistical analysis revealed that treatment stages did not consistently reduce all pollutant concentrations. Mechanical treatment failed to lower chlorides and sulfates, but was effective for ether extract and phenols. Moreover, total mechanical–biological treatment provided a significant, 91% reduction of the ether extract and phenols, while only reducing chlorides by 13% and sulfates by 22%. The multivariate analysis revealed significant differences between raw sewage and mechanically treated sewage. Totally treated sewage stood out as the key factor influencing the pollutants studied, particularly chlorides and sulfates. This finding emphasizes the critical role of comprehensive treatment processes in effective sewage management. Among the analysed substances, chlorides showed the strongest clustering potential, with an average Silhouette coefficient of 0.738, the highest observed. Phenols, on the other hand, exhibited lower Within-Cluster Sum of Squares (WCSS), suggesting their potential as an alternative parameter for evaluation.