An adaptive metamodelling for multipollutant waste load allocation in a river–reservoir system; hydrodynamic simulation–optimization

Abstract

AbstractDischarge of nutrient load from aquaculture activities may pose an eutrophication risk to receiving waterbodies. In this study, monthly multipollutant waste load allocation (MPWLA) programmes are developed to determine the required treatment levels of pollutants released into a river–reservoir system. The objectives of the MPWLA programmes consist of minimizing the total wastewater treatment cost, maximizing the profit from fish production and improving water quality index. The CE‐QUAL‐W2 model with artificial neural networks is linked to the multiobjective particle swarm optimization algorithm to obtain the trade‐off surface in a surrogate‐based simulation–optimization framework. To tackle the ambiguities in quantifying the discharged pollutants, fuzzy membership functions are defined. The application of the proposed MPWLA programmes in the Behesht‐Abad River‐Reservoir system, Iran, has led to a reduction in phosphate concentration more than 3%, enhancement in dissolved‐oxygen levels in the ranges of 2% to 11.6%, and alleviation of Carlson Index between 3 to 7 units.