COSTA, A. B.; http://lattes.cnpq.br/1691129669628462; COSTA, Adriana Barbosa da.
Resumo:
A continuous study for improving the treatment of wastewater and the effluent disposal is
necessary in order to deal with increasingly stringent environmental laws in this field.
Wastewater treatment plants can be considered as highly non-linear systems, due to the
existing disturbances as well as the interaction of a considerable number of process variables. In such a context, the study, optimization and control of these plants are essential for the proper operation of the process with respect to requirements. Several optimization methods are proposed in the literature and, their implementation for engineering applications can be significantly improved by the use of metamodels representing the rigorous model of the process starting from computational data. The present work deals with the development of metamodels, such as the Kriging model, a wastewater treatment process. To this end, the steps of data sampling, through Latin Hypercube Sampling, parameter estimation and validation are performed. The proposed methodology is based on the generation of computational data through the rigorous model of the Benchmark Simulation Model No. 2, implemented in Simulink®, and the optimization of the process using of the Kriging metamodels. These models obtained through the rigorous process data show a high accuracy and the
computational effort of the optimization methods. The Sequential Quadratic Programming
and Genetic Algorithm are used for the optimization task, as well as the generation of the Real Time Optimization model. The achieved results on benchmark model demonstrate the potentiality of the proposed methodology, to minimize the process cost while obeying the effluent restrictions of the treated wastewater.