Performance evaluation of slow sand filters: a probabilistic approach

Abstract

The proposed thesis incorporates a statistical approach to evaluate the performance of slow sand filtration, with an emphasis on removal efficiency of indicator microorganisms. The data for the analysis of removal efficiency have been taken from 15 months study on a slow sand filters. Three pilot scale slow sand filters of 2m diameters were operated under different conditions of controlled parameters namely; filtration rate, sand bed depth and media grain size. The population of and fecal coliforms in the tertiary effluents have been used to calculate the risk involved in violation of effluent stndards promulgated by local regulating agencies. For this purpose different statistical distributions namely; Normal, Longnormal and Weibull distributions were employed. The Longnormal distribution was found to be the most appropriation distribution to fit the microbiological data. The removal efficiency data of microbiological indicators; total coliform, fecal coliform, standard plate counts and coliphages have been fitted on 3 different distributions. In case of removal efficiency the Type III maximum extreme value asymptotic distribution was proved to be the best in comparison to Normal and Lognormal distributions. The K-S test was used for goodness-of-fit of the distributions on 14 conditions of all four microorganisms. The characteristic parameters of the distribution; shape factor and scale factor, were used to calculate the mean and standard deviation of the data. This mean of the distribution have been used to develop the regression models. The regression equations of the removal efficiency were the function of filtration rate, sand bed depth and media grain size only. These equations are capable of predicting removal efficiency of different indicators under given conditions of controlled parameters.