Nourani V, Najafi H, Nikofar B, Huang J J. Optimizing reservoir operations with Z-numbers: Addressing uncertainty and reliability. Journal of Hydrology. 2025;656:132903.
This study presents a novel optimization framework for reservoir management under uncertainty, integrating the emerging concept of Z-numbers—a new generation of fuzzy logic that incorporates both uncertainty and reliability in data interpretation. Conducted on the Alavian Dam, the research combines Feed-Forward Neural Networks (FFNN) with the Lower and Upper Bound Estimation (LUBE) method to enhance inflow prediction accuracy and improve decision-making in water release operations.
Comparative analyses demonstrate that the Z-number-based optimization outperforms traditional fuzzy and crisp methods in terms of reliability, vulnerability, and sustainability. Specifically, the approach satisfied 92% of downstream water demand, minimized total deficiencies to 17.13 MCM, and reduced vulnerability to 21%, offering a more stable and sustainable operational strategy.
By integrating artificial intelligence and uncertainty modeling, this work highlights the transformative potential of Z-number theory for water resource systems. It provides decision-makers with a robust and interpretable framework for optimizing reservoir operations amid fluctuating environmental and hydrological conditions, ultimately contributing to resilient and sustainable water management practices.
About the researchers
Prof. Vahid Nourani is a distinguished scholar affiliated with the Faculty of Civil and Environmental Engineering at Near East University, Nicosia, and the Center of Excellence in Hydroinformatics at the University of Tabriz, Iran. He is internationally recognized for his pioneering work in hydroinformatics, water resources management, and artificial intelligence applications in hydrology. At Near East University, Prof. Nourani contributes to advancing research and education in data-driven modeling and sustainable water management, bridging engineering innovation with environmental resilience. With more than 300 scientific publications and numerous international collaborations, his work continues to shape the global discourse on AI-assisted hydrological modeling and uncertainty analysis.
Dr. Hessam Najafi is a researcher at the College of Environmental Science and Engineering and the Sino-Canada Joint R&D Centre for Water and Environmental Safety at Nankai University, China. His expertise includes environmental systems modeling, fuzzy logic, and sustainable water management under uncertainty.
Dr. Bagher Nikofar is with the Faculty of Civil Engineering at the University of Tabriz, specializing in hydraulic structures, flood control, and optimization in water resource engineering.
Dr. Jinhui Jeanne Huang, also from Nankai University, focuses on environmental sustainability, hydrological processes, and water resource safety.
Together, the team’s collaboration connects Near East University with leading international institutions, promoting innovative AI-based solutions for resilient and sustainable water-resources management. For collaboration and inquiries: [email protected]
Abstract
Accurately estimating water resource potential and optimizing its utilization under uncertainty is a critical challenge in reservoir management. While traditional fuzzy logic methods address some uncertainties, they often fail to adequately represent the reliability of data. In this context, Z-numbers, as a new generation of fuzzy logic, hold notable potential for describing the uncertainty of human knowledge, as they consider both constraint and reliability of data. This study developed an optimization framework leveraging Z-numbers to address uncertainties in operational data and improve decision-making for the Alavian Dam. Inflow predictions were generated using a Feed-Forward Neural Network (FFNN) combined with the Lower and Upper Bound Estimation (LUBE) method, providing a robust basis for optimization. The performance of the Z-number-based optimization was evaluated against classical fuzzy logic and a crisp method solved using genetic algorithm (GA). Performance indicators such as reliability, vulnerability, and sustainability were employed to evaluate these methods. The results show that the Z-number method achieved the highest performance, meeting 92% of downstream water demand, compared to 85% for classical fuzzy logic and 78% for the crisp method. Moreover, the Z-number approach minimized total deficiencies to 17.13 MCM, reduced vulnerability to 21%, and improved the sustainability index to 14, outperforming other methods. The rule curve for optimal operation parameters, derived using Z-number modeling based on inflow predictions via the LUBE method, demonstrated the potential to guide reservoir operations effectively by addressing uncertainties in inflow predictions and release planning. These findings underscore the potential of Z-number theory in enhancing reservoir operation under uncertainty, providing robust and reliable solutions for effective water resource management.
