Analyse modale harmonique des roues de turbines hydroélectriques par l’étude des régimes permanents

Abstract

The emerging use of hydroelectric turbines as grid regulators and recent trends in data valuation justify current efforts to develop digital twins in the field. Indeed, monitoring tools enable real-time diagnosis of turbine conditions. For the runner, fatigue and cavitation are the two main failure modes. For a Francis runner, fatigue calculations depend on the initial state of micro defects on the blades in the connection areas between the crown and the belt, as well as the runner's loading history. Furthermore, the calibration of numerical simulation tools using experimental data is still relevant. Therefore, the characterization of the dynamic behaviour of a turbine-generator unit through modal analysis in operation remains a current research topic. The presented research works focus on the modal identification of hydroelectric turbine runners in steady-state operation. The project is in collaboration with IREQ (Hydro-Québec) and Andritz Hydro. The master’s thesis provides a concise literature review on the dynamic behaviour of turbine runners, modal identification methods and their associated uncertainties. The proposed methodology is based on the analysis of synchronous harmonics of the runner observed in strain gauge measurements, assuming that each harmonic excites a specific group of vibrational modes. Three main technical aspects are explored: modelling of a runner’s response to periodic excitations, system identification using Bayesian formalism, and operational modal analysis. Biases in the method, such as the omission of stochastic excitations and damping in the model have been identified. The method’s sensitivity to different excitation periodicities and the number of considered modes have been assessed and applicability limits were also identified. Additional knowledge is required to enhance the proposed method, particularly regarding the separation of stochastic and periodic signal components, understanding turbine runners damping applied to the method, and model selection tools.

Date	19 Nov 2023
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Antoine Tahan (Supervisor) & Jérôme Antoni (Co-supervisor)