Análise das variáveis do processo de coagulação-floculação para clarificação de água superficial

Abstract

The coagulation-flocculation process is widely used for the clarification of surface water. When using aluminum based metal coagulants, the residual concentration of this metal should be taken into account. The use of coagulation-flocculation modeling can facilitate understanding of the process and assist in determining the operating conditions. In this work, we developed empiric models for the coagulationflocculation. The best developed model used artificial neural networks, it was used as model inputs, turbidity and absorbance at 254 nm of the raw sample, coagulation pH and coagulant dosage. The model output were turbidity, absorbance at 254 nm, dissolved organic carbon, residual aluminum concentration at the end of the process. For this model, we obtained R2 of 0.973 and MSE of 0.051 for the total adjustment of the data. The modeled outputs of the process may be minimized under different treatment conditions. Thus, multi-objective optimization was used to determine the pH and coagulant dosage conditions that minimize different process objectives. Despite the low number of training data, it was verified by the confirmatory experiments that the model has good predictability for turbidity, absorbance at 254 nm and residual aluminum. This shows that it is possible to quickly determine the operating conditions of the coagulation-flocculation process.