Modeling the dependence of coagulante consumption on gross water parameters: fitting a linear regression model

Abstract

The growing demand for drinking water has been gradually increasing and consequently the need for improvements in treatment as well. One of the determining factors of the quality of drinking water, both from a health and environmental point of view, is the amount of coagulant used in the water treatment process, since the dosage can vary according to the physical-chemical characteristics of the raw water. . In this sense, this work presents the initial development of a model capable of predicting amounts of coagulant for a given volume of raw water, considering specific water characteristics as independent variables. The methodological strategy adopted was multiple linear regression analysis performed via R software in three main stages: data collection and separation, modeling and model validation. The data were provided by the company CORSAN/RS, with a time frame from 2015 to 2019 corresponding to two cities. The results showed the existence of a statistically considerable relationship between the dependent variable (coagulant aluminum sulfate consumption) and the independent variables (turbidity, color, alkalinity, organic matter and added volume). The adjusted R² value of 0.95 indicates that the prediction model is statistically satisfactory. That is, the ease of adjustment combined with the good performance of the studied phenomena shows that the model presented initially can be applied to improve wáter treatment processes in stations of similar size.