River water quality management using a fuzzy optimization model and the NSFWQI Index

• Published in: Water S. A. Vol. 47; no. 1; pp. 45 - 53
• Main Authors: Ghorbani, Mohammad Kazem, Afshar, Abbas, Hamidifar, Hossein
• Format: Journal Article
• Language: English
• Published: Gezina Water Research Commission (WRC) 01.01.2021; Water Research Commission
• Subjects: ACO algorithm fuzzy set theory multiple pollutant; Algorithms; Analysis; Ant colony optimization; Fuzzy set theory; Fuzzy sets; Management; NSFWQI; Objective function; Optimization; Optimization models; Pollutants; Pollution load; QUAL2K model waste load allocation model; Quality control; Quality management; Quality standards; Random variables; River water; Rivers; Sanitation; Set theory; Simulation; Simulators; Waste load; Wasteload allocation; Water; Water pollution; Water quality; Water quality management; Water treatment; Iran
• ISSN: 0378-4738; 1816-7950; 1816-7950
• DOI: 10.17159/wsa/2021.v47.i1.9444

In this study, a novel multiple-pollutant waste load allocation (WLA) model for a river system is presented based on the National Sanitation Foundation Water Quality Index (NSFWQI). This study aims to determine the value of the quality index as the objective function integrated into the fuzzy set theory so that it could decrease the uncertainties associated with water quality goals as well as specify the river's water quality status rapidly. The simulation-optimization (S-O) approach is used for solving the proposed model. The QUAL2K model is used for simulating water quality in diferent parts of the river system and ant colony optimization (ACO) algorithm is applied as an optimizer of the model. The model performance was examined on a hypothetical river system with a length of 30 km and 17 checkpoints. The results show that for a given number of both the simulator model runs and the artificial ants, the maximum objective function will be obtained when the regulatory parameter of the ACO algorithm (i.e., q0) is considered equal to 0.6 and 0.7 (instead of 0.8 and 0.9). Also, the results do not depend on the exponent of the membership function (i.e., γ). Furthermore, the proposed methodology can find optimum solutions in a shorter time.