Avaliação estatística de desempenho de estações de tratamento de água, em função da tecnologia, do porte e do tipo de manancial

Abstract

In this research were evaluated data from 45 water treatment plants, considering five different technologies, operated by three different concessionaires (from three Brazilian states). This large number of data, about 7 million observations, allowed the application of several statistical methods, aiming at comparing the performance of the plants. After the systematization of the database, it was possible to start the study, which began with the preliminary statistical analysis of the data, which included the identification of outliers, verification of the frequency distribution of the data and calculation of the descriptive statistics. The performance analyzes considered the comparison of the raw and treated water parameters of the water treatment plants by means of non-parametric and multivariate statistics, the evaluation of the treatment standards for treated water and the reliability analysis of the plants. All the analyzes considered the technology, size and type of source of each WTP. The database studied is very heterogeneous, varying the number of data per parameter of water quality and by plants, besides the monitoring frequencies. We evaluated 37 conventional stations, five of direct filtration, one of double filtration, one of complete cycle with flotation and one of oxidation followed by filtration. Regarding the size, 25 stations were small, 13 medium and 7 large. In relation to the spring, 31 seasons were of loquacious spring, 11 lentic and only three lentic / lotic. The grouping of the stations by means of the similar characteristics did not allow to identify a standard that could justify this performance. In addition, the results of the multivariate analysis did not indicate a pattern in the clusters formed in the dendrogram of raw water and treated water. The percentage of compliance with drinking standards (USEPA and Consolidation nº 5/2017) and WHO guidelines for the apparent color, pH, chlorine and fluoride parameters was 100% for most of the WTP analyzed. The turbidity parameter presented a mismatch in relation to the standards. The reliability analysis showed that the mean turbidity values observed in the water treated by almost all WTP would be insufficient to guarantee 95% compliance with the 0.1, 0.3 and 0.5 uT standards. Therefore, it can be said that the operation of the water treatment plant, specific for each WTP, seems to influence more the performance, than the other analyzed factors.