Are you an EPFL student looking for a semester project?
Work with us on data science and visualisation projects, and deploy your project as an app on top of Graph Search.
The hydropower sector has recently raised the interest in pump as turbine (PaT) that can be a valid trade-off between capital cost and performance in micro-scale installations. Nevertheless, the modest efficiency of PaTs often restricts their exploitation. In this paper, available experimental data are used to tune a numerical model which aims to investigate the effect of the pump cutwater design on the PaT performance to improve the hydraulic efficiency. Because of its finite thickness, the cutwater interferes with the flow at the runner inlet, and generates local flaws in the velocity field such as swirl and deviations of the streamlines which limit the machine performance. To identify the geometrical characteristics of the cutwater impacting on the PaT performances, different values of stretching and thickness of the cutwater are studied at variable inclination by computational fluid dynamics (CFD) simulations. A multivariate regression method is applied on the CFD results to build a surrogate model of the PaT hydraulic characteristics as a function of the geometrical parameters of the machine cutwater. Based on this model, an optimization problem is solved to identify the most advantageous geometrical asset of the PaT cutwater to maximize the efficiency. The results highlight that the length and the cutwater angle are the most affecting variables in favouring a tangential component at the entrance of the runner. The hydraulic efficiency peak of the optimized geometry results to be 86.3%, while the baseline configuration records an efficiency of 82.4%, and the Psi - phi characteristic moves the best efficiency point towards higher head (+7.5%) and lower discharge (-13.0%). The proposed methodology allows identifying the best geometrical characteristics of the PaT cutwater to maximize the performances while significantly reducing the computational time. (C) 2021 Elsevier Ltd. All rights reserved.
Jürg Alexander Schiffmann, Sajjad Zakeralhoseini
Elena Vagnoni, Alessandro Morabito