Otimização do índice de qualidade de estação convencional de tratamento de água (IQETA) por meio de análise estatística multivariada

Abstract

This study focused on the reduction of the number of parameters that comprise the Water Treatment Plant Quality Index (WTPQI), applied to conventional treatment plants with horizontal sedimentation basins, in order to ease its application in real scale. The index, initially composed of nineteen parameters, was optimized by means of technical criteria and multivariate statistical tool, named Factorial Analysis. This technique allows to deal simultaneously with several variables, even when the theoretical model of relations between them is unknown. It suggests that if a technical criterion for variable selection is applied the model becomes as parsimonious as possible. The following procedures were carried out aiming the use of the Factorial Analysis: problem formulation, construction of the correlation matrix, definition of the method of the Factorial Analysis and the number of factors, spin factors, interpretation of factors and calculation of the factor loadings of variables or choice of the substitute variables. The sampling universe of this research covered a range of 21,600 hourly flow rate values from ten water treatment plants during the rainy season, usually from November to March. Based on the flow rate values, the variables involved in the water treatment process were obtained. However, there were employed the daily flow rate average values that restricted the dataset to 909 daily data, for each of the 15 variables. These variables submitted to Factorial Analysis refer to the 19 original ones minus two variables concerning rapid mix (velocity gradient and detention time) and two relating to the disinfection (detention time and number of compartments of the clearwell). As mentioned earlier, these variables were excluded due to technical criteria. With exception of Instruction Level and Jar test, the other variables are related directly or indirectly to the hourly flow rates. The Factorial Analysis was accomplished with the partial values of the WTPQI, i. e. the valueof the observed variable powers by the weight conferred by the experts.The new index (WTPQI12), composed of just 12 parameters, becomes easier to employ a tool to help the water treatment plant performance assessing. Despite the Spearmans correlation coefficient between the filtered water turbidity and the new index value was lower than for theoriginal index, this analysis shows the same tendency of the a higher WTPQI12 is usually associated with better conditions to produce a highest treated water quality