Abstract:

Proper identification of groups of water pollutants can be helpful in preventing water pollution and protect vital water resources. In this paper data on chemical industrial wastewater are treated using Principal Factor Analysis (PFA) and Cluster Analysis (CA) to identify of groups of water pollutants. For untypical data Spectral Cluster Analysis (SCA) is conducted. The background input and output data has been inventoried as follows: suspended solids (matters), biochemical oxygen demand (BOD₅), chemical oxygen demand (COD), zinc (Zn), organic chrome (Cr), copper (Cu), organic iron (Fe), sulfides, volatile phenols, free cyanides, oils & fats. The principal output reports presented in the study consist of the factorization in different coordinate systems and factorization on a plane obtained from the multidimensional scaling for the comparison purposes with the grouping results (PFA), as well as of the illustration of grouping of pollution (SCA). To verify the quality of grouping two quality indexes (grouping measures) were computed: Caliński-Harabasz index and Silhouette index. PFA and SCA produced similar groups of pollutants. Three groups were clustered: suspended solids, BOD₅ and organic iron pollutions in the first cluster, free cyanides and copper pollutions in the second cluster and COD, sulfides and organic chrome pollutions in the third cluster. The main difference concerns the grouping of volatile phenols and oils & fats pollutions. The factorization in different coordinate systems and factorization on a plane obtained from the multidimensional scaling for the comparison purposes with the grouping results also support the observed results. The quality indexes indicate that SCA is more suitable for the identification of groups of wastewaters. The results obtained using the proposed tools are discussed and usefulness of the proposed methods is assessed.