Reutilização de água em processos industriais : uma abordagem metaheurística

Abstract

In this work, synthesis of reuse water networks in industrial processes has been investigated by applying an improved Particle Swarm Optimization (PSO) method. A superstructure that represents the whole possible reuse water network configurations was defined in order to formulate a wide set of optimization models. In order to lead with both discrete- and continuous-type variables in Nonlinear Problems (NLP) and Mixed Integer Nonlinear Problems (MINLP), an improved PSO method that was coded in Maple 14® software was proposed, whose main goal is to seek the synthesis of the reuse water networks by demanding some criteria of optimization. The minimization of the freshwater flowrate, total number of interconnections and the total annual cost for the industrial plant were tested as criteria. By using concepts of Pinch Analysis, violations on the mass transfer processes are mainly avoided as well as a suitable reduction on the search space in the framework of the PSO method. In addition, improvement on the performance in the optimality of the reuse water network synthesis could be attained. As a strategy on the handling of the restricted search space, an external static penalty approach of the objective function was incorporated, resulting in an improvement of the optimization procedure. Deterministic methods such as the generalized reduced gradient and branch and bound that are available in GAMS/CONOPT and GAMS/SBB local solvers were also used to obtain the reuse water network synthesis, whose results are compared with those obtained by the PSO method. In order to show the applicability and effectiveness of the improved PSO method, four study cases previously reported in the literature and representing water allocation problems (WAP) with single and multi-component systems were analyzed. As results of the reuse water network synthesis, optimized values of freshwater flowrate, wastewater flowrate and maximum pollutant concentrations in the inlet and outlet of each process stream are provided. Demanding the minimization of the freshwater flowrate, the PSO method results were similar to those obtained by the GAMS solvers as well as those reported in the literature, warranting the minimum freshwater flowrate recommended by the Pinch Analysis. On the other hand, when the minimization of total number of interconnections is demanded, the PSO method has provided the best solutions for the synthesis of the reuse water networks as compared with those obtained by the GAMS solvers and other algorithms found in the literature. In addition, all reuse water networks optimized by the PSO method are economical and technical viables, with reuse water flowrate values above 2 ton h-1. In respect to the minimization of the total annual cost, the PSO method results on the synthesis of the reuse water network has attained the lower total annual cost than those results obtained by the GAMS solvers and allowed to perform an adequate economical analysis in relation to the water consumption in industrial processes. Based on all tested criteria, the improved PSO method results are thus better than that obtained by GAMS solvers, showing its flexibility and efficiency on the methaheuristic approach in solving water network reuse synthesis problems. Finally, an association of Pinch Analysis, mathematical programming and improved PSO method-based WAP optimization has been adequate to provide global optimum reuse water networks for all minimization criteria tested.