Regional sediment characterization and curve reconstruction for hydropower infrastructure design and turbine wear management
Hydro-abrasive turbine wear caused by sediment passage is a severe operational and economic threat to run-of-river hydropower plants (HPPs) in steep, mountainous regions, leading to rapid efficiency decline and high turbine maintenance costs. This study investigated the spatial and seasonal variability of sediment grain-size distributions (GSD) across 11 HPPs in Rize, Türkiye, to establish a regional baseline and a sediment characterization and GSD reconstruction framework relevant to sediment management. A total of 53 GSD curves were evaluated from regulator sediments during winter and summer periods. Samples were analyzed using conventional granulometric parameters, compositional data analysis (CoDa), principal component analysis (PCA), unsupervised clustering, and paired seasonal tests. The representativeness of composite mixed samples was also assessed at the HPP level to evaluate cost-efficient sampling protocols. Ridge, partial least squares, and Gaussian process regression were compared with regional median and logarithmic interpolation benchmarks under three partial-measurement scenarios using leave-one-HPP-out cross-validation to evaluate reconstruction accuracy at unmeasured sieve sizes. No statistically significant seasonal differences were detected in sediment compositions ( p = 0.650). This result indicates that a consistent winter–summer contrast was not established in the sampled dataset; it does not identify the processes controlling sediment variability or erosive hazards. PCA and clustering identified four exploratory sediment groups with contrasting gravel, sand, and fine-material proportions, providing a descriptive basis for comparing deposited sediment among the sampled locations. Crucially, Under Scenario 2, using measurements at five sieve sizes, logarithmic interpolation reconstructed the unmeasured GSD values with high accuracy (R 2 = 0.978; MAE = 1.706 percentage points), outperforming the evaluated machine-learning models. These findings support a five-sieve reconstruction approach for preliminary sediment screening, although its benefits in terms of testing time, labour, and cost require direct evaluation.
Authors
- Hasan Oğulcan Marangoz (ORCID: https://orcid.org/0000-0003-0215-6159)
- Veli Süme (ORCID: https://orcid.org/0000-0001-8251-2461)
- Rasoul Daneshfaraz (ORCID: https://orcid.org/0000-0003-1012-8342)
- Enver Yılmaz (ORCID: https://orcid.org/0000-0002-6434-3526)
- Elshad Aliyev
Institutions
- University of Maragheh (IR)
- Recep Tayyip Erdoğan University (TR)
Publication Details
- Journal
- Scientific Reports
- Published
- 2026-11-21
- DOI
- https://doi.org/10.1038/s41598-026-80372-w
- Primary Topic
- Erosion and Abrasive Machining
- Type
- article
- Field-Weighted Citation Impact
- 0.00