Fish canning industry wastewater variability assessment using multivariate statistical methods

Abstract Usually, fish canning industrial wastewaters have a highly variable composition over time. For a good performance of treatment processes it is necessary to limit that variation. However, extended wastewater monitoring, including all relevant analytical parameters, is expensive. This work proposes an efficient approach to minimize the analytical determinations number without compromising the global characterization goal. This way, fish canning industry wastewaters variability was assessed and interpreted through multivariate statistical tools application to analytical data obtained from a monitoring program carried out in a fish canning industry of northern Portugal. 23 physicochemical parameters were determined in 20 samples collected on an 8 months period. The results achieved by correlation analysis, principal component analysis (PCA) and cluster analysis (CA) led to the main water pollution sources identification and to the minimization of physical and chemical parameters number to be analyzed in order to achieve a correct wastewater characterization, at minimum cost. The main pollution sources proved to be the brine and eviscerating step waters. Dissolved organic carbon (DOC), total suspended solids (TSS), conductivity, pH, Ca2+, F− and one of the parameters SO42, NO3− and PO43− were identified as important parameters that must be monitored in order to obtain an accurate characterization allowing to define the most appropriate wastewater treatment.