A novel application of genetic algorithm for synthesizing optimal water reuse network with multiple objectives

The current drive towards water conservation has compelled the process industries to seek various measures to minimize water. One such strategy is to design and synthesize an optimum water network that enables maximum reuse in between operations. Such optimization has been known to be challenging due to the presence of nonlinearities in the formulation that makes it difficult to solve using the standard deterministic techniques. Moreover, most of these techniques have focused on solving single objective problem involving freshwater usage while other decision factors such as treatment costs and network topologies are generally not accounted for as they would introduce extra complexities to the problem. In this paper, we propose to address optimal water reuse network as a multi-objective optimization problem-our strategy is to apply a well-known genetic algorithm (NSGA-II) for single- or multi-objective optimization of water reuse network synthesis in a total (integrated) setting. First, a superstructure entailing all reuse opportunities between the water-using processes and wastewater treatment units is formulated. Next, decision variables in the form of various stream flows are encoded as the initial population chromosomes and then subjected to a set of genetic operators (selection, mutation, and crossover) to generate the offspring chromosomes. Coupled with these operators is an elitism strategy to improve the convergence and spread of the solutions. A key benefit of this approach is that it can be equally applied for both single objective and multi-objective cases in a robust manner. To demonstrate the efficacy of the approach, four single-objective case studies are solved with the results compared with the reported global optima. Thereafter, two multi-objective case studies are solved to generate the complete Pareto set. The key advantage of the proposed approach lies in its ability to solve complex optimization problem without resorting to simplifying heuristics and optimization strategies that are characteristics of the literature approaches.