Dynamic Characterization Of Hydroelectric Turbine In Transient Using OBMA And Phase-Shift Analysis
Quentin Dollon, Souheil-Antoine Tahan  1  , Jérôme Antoni  2  , Martin Gagnon  3  , Christine Monette  4  
1 : Ecole de Technologie Supérieure  (ETS)  -  Website
1100, rue Notre-Dame Ouest Montréal (Qc) H3C 1K3 Canada -  Canada
2 : Institut national des sciences appliquées de Lyon  (INSA Lyon)  -  Website
INSA Lyon
20 Avenue Albert Einstein, 69621 Villeurbanne cedex -  France
3 : Institut de Recherche d'Hydro-Québec [Varennes]
4 : Andritz Hydro Ltd

Design and exploitation of hydroelectric turbines relies on the knowledge of their dynamic behavior. This last enables one either to get a good assessments of life duration or to set up predictive-based maintenance sessions. In needed, two sources of information are mandatory to characterize the mechanical behavior of a structure: numerical simulations, and experimental data processing. The first field gives a whole and detailed analysis of the behavior in any expected regime, but needs to be validated by the second to be reliable, which is a straightforward consequence of the strong assumptions made to reduce computing burden. The second method aims at achieving in-situ measurements to extract dynamic features; these features will be highly fragmented, but in general closer to reality for a given measured operating condition. These two sources of information are then crossed to obtain a hybrid representation of the dynamic behavior.

However, one of the problems with experimental analysis is the cost of data acquisition. To reduce financial burden of measurements, the idea is to extract a maximum of information from transient records instead of several stationary records, which would make the measurement less time consuming. Our goal is to determine whether the signal processing is able to extract precise and suitable features from these transient measurements.

The purpose of this paper is to consider the possibility of extracting modal parameters of a Francis hydroelectric turbine in transient conditions by focusing on resonance regions generated by the interaction of a structural mode with a frequency-variant harmonic pressure pulsation. Especially when numerous modes are in the same bandwidth, this method separates them by exciting only matching shapes. The resonance retrieval is done using Order Tracking (method used to extract a specified harmonic from the signal), and a classical Modal Identification algorithm is then used to feature the isolated mode. Furthermore, a phase-shift analysis is made between captors measuring resonance in order to both localize the mode and determine the shape.



  • Poster
Online user: 57