The Performance of Models for the Prediction of Coagulant and pH Control in Water Treatment

  • Larissa Lopes Gomes Universidade Federal da Bahia
  • Karla Oliveirea Esquerre Universidade Federal da Bahia
  • Robson Magalhães Universidade Federal da Bahia
Keywords: Neural Network, Multiple Linear Regression, Jar Test

Abstract

Coagulation is an important step in water treatment. Doses of coagulants and pH controllers in the treatment of water are predicted by the jar test. The test is performed in a laboratory and mean rates obtained will be applied to the plant. Factors, such as changes in rainfall indexes, may change water quality and new adjustments are required for coagulant doses and pH controllers. Current assay proposes the application of models based on neural networks and linear regression so that modifications in doses of sodium hydroxide (NaOH) and aluminum sulfate (Al2SO4) could be performed in real time. The best prediction result was obtained by the neural network model with optimizer.

Author Biographies

Larissa Lopes Gomes, Universidade Federal da Bahia
Graduanda em Engenharia Química Departamento de Engenharia Química (DEQ) da Universidade Federal da Bahia – UFBA, Salvador (BA), Brasil; Bolsista de Iniciação Científica pelo CNPQ
Karla Oliveirea Esquerre, Universidade Federal da Bahia
Docente do departamento de Engenharia Química e dos Programas de Pós-Graduação em Engenharia Industrial (PEI) e Mestrado em Meio Ambiente, Águas e Saneamento (MAASA) da Universidade Federal da Bahia – UFBA, Salvador (BA), Brasil
Robson Magalhães, Universidade Federal da Bahia
Docente do departamento de Engenharia Química e do Programa de Pós-Graduação em Engenharia Industrial (PEI) da Universidade Federal da Bahia - UFBA, Salvador (BA), Brasil
Published
2014-11-06