LAUSR

Abstract Aging infrastructure and funding problems continue to plague water supply networks worldwide. The failure risk of water mains is posing a considerable threat to drinking water security in urban centers, necessitating performance assessment strategies. The prediction of networks' condition index has seldom been studied, while a novel automated method can be developed to reasonably address this issue, thus promoting sustainability-based strategies. This paper presents an integrated framework for the assessment and quantification of the water main condition index. The Arequipa region in Peru consists of eight provinces; this region is chosen to exemplify the proposed framework due to the rapid pace of urbanization, making water infrastructure even more crucial. The novelty of this study includes innovation of research concept as it restructures the water network condition assessment by exploiting the hybridization technique of two potent intelligent systems; namely, the adaptive neuro-fuzzy inference system and the fuzzy inference system. These systems are employed sequentially to accomplish computational simulations and reasoning consolidations to generate automatically one condition index. Thus, making up for the lack of integrated research in water piping system. This research also gives prominence to enhancing water sustainability via incorporating the physical factors (pipe attributes) together with operational and environmental factors, represented by the impact of leaching and disinfection byproducts on people's health. The neuro-fuzzy processor is designed to predict each province's condition index via the grid partitioning and hybrid algorithm. The hybrid algorithm uses a combination of backpropagation and least-squares regression to optimize and tune the fuzzy parameters. The fuzzy consolidator indicated that the region's network condition index is 63.1, which reveals a medium condition for the Arequipa region water networks. The integrated framework is validated by conducting a comparative analysis with the multiple linear regression model. The results reveal better performance of the proposed framework since it demonstrates higher R2 of 0.9145. This study promotes multidimensional applications in urban sustainability, environmental sustainability, and urban water management systems.