Optimal operation of water distribution networks using pressure management via non-dominated sorting differential evolution, shrimp algorithms and deep neural networks
Frontiers in Earth Science · 12.08.2026 · 14
DOI: 10.3389/feart.2026.1847899 ↗Kurz gefasst
Fachartikel in der Zeitschrift Frontiers in Earth Science, online erschienen am 12.08.2026. Verfasst von Yaser A. Nanehkaran und 5 weiteren Autor:innen. Die Arbeit nennt 44 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Water distribution networks (WDNs) are among the most critical urban infrastructure systems, especially in arid regions such as Iran, where water scarcity has intensified in recent years. Optimizing the operation of WDNs is essential to achieving sustainable, cost-effective water management. One of the key strategies for improving network efficiency is pressure management, which not only ensures reliable service delivery but also reduces operational and maintenance costs. This study presents a multi-objective optimization framework for the optimal placement of pressure-reducing valves (PRVs) to minimize both pressure deviations and associated costs. To achieve this, a hybrid simulation-optimization approach was employed. The NSDE (Non-dominated Sorting Differential Evolution) algorithm and the Shrimp optimization algorithm were applied as evolutionary metaheuristics, while a Deep Neural Network (DNN) model was trained to provide predictive insight and enhance solution robustness. Hydraulic simulations were performed using the EPANET software, and the proposed framework was applied to the real case of the Khorramshahr water distribution network. The results demonstrate that the inclusion of optimally located PRVs significantly improves pressure conditions. For example, installing a single PRV reduced excess pressure by 8.616%, while installing five PRVs reduced excess pressure by up to 55.262%. Further installation of six and seven PRVs yielded pressure reductions of 55.287% and 55.456%, respectively. A comparative analysis between the NSDE and Shrimp algorithms revealed that NSDE outperformed Shrimp in scenarios with 2 and 3 PRVs, whereas Shrimp performed slightly better in the single-PRV configuration. Overall, the integration of intelligent optimization algorithms and machine learning models, coupled with hydraulic simulation, provides a powerful toolset for improving the design and operation of water distribution systems. The findings highlight the practical potential of data-driven, pressure-management strategies to enhance the reliability and sustainability of urban water networks.
Zitieren
APA
Nanehkaran, Y. A., Zhu, L., Pusatli, T., Chen, J., Eum, Y., & Zhang, Y. (2026). Optimal operation of water distribution networks using pressure management via non-dominated sorting differential evolution, shrimp algorithms and deep neural networks. Frontiers in Earth Science, 14. https://doi.org/10.3389/feart.2026.1847899
Vancouver
Nanehkaran YA, Zhu L, Pusatli T, Chen J, Eum Y, Zhang Y. Optimal operation of water distribution networks using pressure management via non-dominated sorting differential evolution, shrimp algorithms and deep neural networks. Frontiers in Earth Science. 2026;14. doi:10.3389/feart.2026.1847899
BibTeX
@article{Nanehkaran2026,
author = {Nanehkaran, Yaser A. and Zhu, Licai and Pusatli, Tolga and Chen, Junde and Eum, Youngseob and Zhang, Yanni},
title = {Optimal operation of water distribution networks using pressure management via non-dominated sorting differential evolution, shrimp algorithms and deep neural networks},
journal = {Frontiers in Earth Science},
year = {2026},
volume = {14},
doi = {10.3389/feart.2026.1847899}
}
EUVIMED