Structural Optimization of a Pipeline Savonius Hydro Turbine Based on Broad Learning

Utilizing pipeline Savonius hydro turbines driven by the pressure difference in water flow in the irrigation pipelines to generate electricity can achieve stable and reliable charging of the sensor battery, thereby addressing autonomous energy supply concerns. However, due to the low energy capture efficiency of the Savonius turbine with a small pressure difference, it is necessary to optimally design the internal flow channel structure to improve its efficiency and achieve a more miniaturized overall equipment, which has extremely high engineering application value. This study investigates the Savonius turbine, installed within the irrigation pipeline operating under specific conditions (diameter: 60 mm; flow rate: 15 m3/h; inlet pressure: 0.2 MPa; outlet pressure ≤ 0.05 MPa). A validated Computational Fluid Dynamics numerical model was developed to generate 200 datasets for a subsequent structural optimization process utilizing a Bayesian-optimized broad learning technique. The optimal structural parameters for maximum efficiency are a deflector angle of 10°, an aspect ratio of 0.775, and three blades, resulting in a maximum efficiency of 21.73% at a rotational speed of 1098 r/min. For the maximum power coefficient, the optimal combination is a deflector angle of 10°, a height-to-diameter ratio of 0.9, and three blades, yielding a peak power coefficient of 0.1859 at a rotational speed of 600 r/min.