Field data and numerical simulations reveal sediment erosion mechanisms in Pelton turbines, suggesting design improvements.
Sediment-induced erosion presents a significant challenge for high-head Pelton turbines in Himalayan hydropower projects, where rivers carry high concentrations of hard mineral particles. This study investigates the erosion mechanisms associated with sediment-laden flows using a combination of field data and numerical simulation. Sediment samples were collected from the Manal hydropower plant on the Giri River, a tributary of the Yamuna River in India. The samples were analyzed for size distribution, shape factor, and mineral composition, revealing a high abundance of hard minerals such as quartz and feldspar, with 90% of particles smaller than 280 μm and a mean shape factor of 0.71. Using these field-derived properties, a numerical study was conducted on a modified Pelton turbine injector design, which lies between the two conventional configurations—long spear-guided and blunt-body with short spear. The simulation employed a coupled volume of fluid and discrete phase model approach to capture multiphase interactions between air, water, and sediment particles. Erosion rates were evaluated using different models and by varying sediment concentrations, particle sizes, and operational heads. The results consistently identified the nozzle exit, needle tip, and needle head as the most erosion-prone regions. Higher sediment concentrations and operating heads significantly increased erosion, while larger particle sizes enhanced nozzle erosion but reduced needle erosion. Empirical regression models were developed for predicting maximum erosion rates, with their functional forms expressed as power-law relationships of particle size, sediment concentration, and head. This work highlights the importance of integrating field sediment characteristics into erosion models, contributing to enhanced design for Pelton turbines in abrasive, sediment-rich environments.
No takes yet. Share an insight, caveat, or question.
Vishwakarma et al. (2025) studied this question.