Cavitation Erosion Workshop, May 27-28, Bassin d’Essais des Carenes, Val de Reuil, 2004 On the Vibratory Approach For Cavitation Erosion Monitoring in Hydraulic Turbines X. Escaler, E. Egusquiza Technical University of Catalonia M. Farhat, F. Avellan Swiss Federal Institute of Technology Thank you very much Mr. Chairman. Good morning Ladies and Gentleman, I would like to present to you this work about …

OUTLINE Introduction Leading edge cavitation Detection techniques Instrumentation Test cases: results & discussion Conclusions & Ongoing work The outline of my presentation is the following one: I will begin with a brief introduction about the motivation of the work. Then I will describe the prototype under investigation and the instrumentation. And after explaining the measuring campaign I will expose the results to finish with the conclusions.

INTRODUCTION Cavitation erosion in hydro turbines is increasing due to: Power concentration ( velocities   ) Operation far from b.e.p. (off-design) Detection during operation is of prime importance Predictive maintenance Extending intervals between erosion repairs Avoidance of erosive regimes Optimization of turbine operation Validation of turbine rehabilitation Acceptance tests Models + Prototypes Cavitation erosion in hydro turbines is a problem that is probably going to increase in a near future. This is due to the fact that current tendencies are to design new units with higher power concentration and to operate the existing turbines far from their best efficiency point. Therefore, the detection of erosive cavitation during prototype operation is of prime importance to be able to control its negative consequences. The traditional approach for erosive cavitation detection is based on the analysis of external vibrations measured on fixed parts of the machine. In that case, the best position to monitor cavitation induced vibrations in the prototype is found to be the turbine guide bearing pedestal. This is the case when dealing with inlet edge sheet cavitation on the runner blades (is the most erosive form of cavitation). However, this type of cavitation takes place on the rotating blades.

LEADING EDGE CAVITATION Attached cavitation on a 2D hydrofoil Cloud cavitation: unstable and highly erosive regime Shedding frequency follows a Strouhal like law

LEADING EDGE CAVITATION Attached cavitation on the blades of an hydraulic runner Results in strong erosion on the suction side Shedding frequency forced by hydrodynamic phenomena !

LEADING EDGE CAVITATION Unsteady incoming flow Guide vane wakes Von Karman vortex shedding Unsteady W and  Sheet fluctuation Pressure fluctuations Induced vibrations High frequency noise Strong erosion

LEADING EDGE CAVITATION Typical fixed measuring positions: Turbine guide bearing pedestal Wicket gate arm Draft-tube wall Signal is attenuated Relative quantification For absolute quantification transmissibility functions are needed Dynamic calibration The inconvenient of measuring vibrations on fixed parts such as bearing and guide vane is that the influence of the shaft rotation and of the effects of the fluid film and the water in the transmission path from the excitation location to the measuring positions are not well known. So, it is believed that a sensor located on the shaft should be advantageous since it has a mechanical link to the blades which permit to avoids these aspects. Therefore a measuring campaign on a water turbine prototype has been proposed to investigate this possibility.

DETECTION TECHNIQUES Monitoring cavitation-induced signals next to the runner Vibrations Acoustic emissions Detection techniques High-frequency content Amplitude demodulation of band-pass filtered signals Phase or synchronous averaging of envelopes Quantification techniques Dynamic calibration of transmission paths Inferring absolute intensities from impulsive vibrations induced by cavity collapses Comparison of estimates with lab results on a given material

DETECTION TECHNIQUES Amplitude demodulation of band-pass filtered signals Signal acquisition at a high sampling rate Band pass filtering above several kHz Envelope extraction by the use of the Hilbert transform Low-pass filtering and decimation Spectral analysis and phase averaging of resulting signals

INSTRUMENTATION Sensors Instrumented impact hammers Miniature high-frequency accelerometers Acoustic emission sensors Dynamic pressure sensors Instrumented impact hammers Conditioning units and filters Recording devices Digital tape recorders A/D converters Spectrum analyser Signal processing software

KAPLAN TURBINE 1 Max. output power = 73 MW # blades = 6, # guide vanes = 24 Cavitation erosion observed on blades Radial measurements in turbine guide bearing: Broadband vibration > 3 kHz Vibration amplitude is maximum at full output load

KAPLAN TURBINE 1 Signals band-pass filtered from 5 to 10 kHz. Envelope spectra indicates that wicket gate passing frequency predominates at maximum load Amplitude of significant frequencies are much higher in bearing closer to runner

KAPLAN TURBINE 2 Max. output power = 100 MW # blades = 5, # guide vanes = 20 Aerated draft tube High frequency accelerometers Guide bearing Wicket gates Tachometer

Upstream & Downstream levels KAPLAN TURBINE 2 Test conditions: Test Id a [%] P [MW] Upstream & Downstream levels H1 & H2 [m] Aeration T0 H1=31.02, H2=14.13 Open T1 29 12.5 T2 37 20 H1=30.96, H2=14.13 Closed T3 H1=30.93, H2=14.13 T4 49 36.5 H1=30.96, H2=13.92 T5 60 50.7 H1=30.92, H2=13.93 T6 70 55.6 H1=30.89, H2=13.96 T7 75 63.6 H1=30.88, H2=13.98 T8 80 71.3 H1=30.86, H2=14.00 T9 90 89.3 H1=30.85, H2=14.02 T10 100 H1=30.83, H2=14.08

KAPLAN TURBINE 2 Spectra of guide bearing and wicket gates envelopes

KAPLAN TURBINE 2 Phase averaging of the guide bearing acceleration envelope Zone I : a < 45 % Random Acceleration No modulation  NO CAVITATION Zone II : 45 < a < 80 % Strong & Periodic Impacts at the blade passing frequency Phase shift analysis  Mechanical contact of the blades  NO CAVITATION

KAPLAN TURBINE 2 Zone III : a > 80 % Strong modulation at the wicket gate passing frequency  ATTACHED CAVITATION Model tests showed attached cavitation for wicket gates opening > 85 %

FRANCIS TURBINE 1 2 Vertical Francis turbines Hydropower Plant A, Units 1 and 2 65 MW per unit, 250 rpm # blades = 15, # guide vanes = 24 High freq. Accelerometers Guide bearing Wicket gates Draft-tube Acoustic emission sensor Sampling freq. 100 kHz

FRANCIS TURBINE 1 Averaged RMS values in the band [1-49k] Hz Different behaviour in draft tube Similar trends in bearing and wicket gates Levels in wicket gates higher than in bearing Maximum levels at 30 and 60 MW Acoustic emission and acceleration qualitatively analogous

FRANCIS TURBINE 1 Auto-power Spectra of vibrations in Turbine Guide Bearing UNIT 1 UNIT 2 High frequency region is very sensitive to operating condition Maximum amplitudes and shapes of spectra depend on measuring position, sensor and unit

FRANCIS TURBINE 1 Auto-power Spectra of Envelope in the band [30k-40k] Hz UNIT 1 Wicket Gate Passing Freq. UNIT 2 Outlet Swirl Freq. Peaks appear at hydrodynamic frequencies At low loads, draft tube swirl frequency At high loads, wicket gate or blade passing frequency

FRANCIS TURBINE 2 Pitting Tests: Onboard Vibration: 4 polished stainless steel samples flush mounted (45 mm diameter) Onboard Vibration: 4 kHz accelerometer Onboard conditioning Signal transmission: Hollow turbine shaft Slip ring. Freq. coding Guide bearing vibration 50 kHz accelerometers

FRANCIS TURBINE 2 Dynamic Calibration : Instrumented hammer Multiple impacts on the runner blade Synchronous measurements of onboard and guide bearing vibrations Impact force & Response (in air) Transfer function and coherence

Downstream level, Hd [m] FRANCIS TURBINE 2 Test conditions: 40 min 205.9 206 72 M5 15 min 205.5 201 70 M4,4 219 75 M4,3 16 min 205.6 234 80 M4,2 17 min 205.7 247 85 M4,1 207.5 253 90 M3 41 min 207.4 223 78 M2 206.2 227 M1 Exposure Time Downstream level, Hd [m] P [MW] a [%] Test Id

FRANCIS TURBINE 2 Pitting tests: Attached cavitation with maximum erosion located in the cavity closure The cavity length depends on the downstream level Pits size increases with the cavity length Pits density decreases with the cavity length

FRANCIS TURBINE 2 Onboard vibration: No possible to derive the contribution of cavitation induced vibration from raw acceleration signals with valuable accuracy

FRANCIS TURBINE 2 Validation of the guide bearing location E0: Averaged power spectrum of onboard vibration envelope E1: Averaged power spectrum of guide bearing vibration envelope E0,p (E1,p): Averaged coherent power spectrum of onboard (guide bearing) vibration envelope referred to pressure fluctuation in the draft tube Validation of the guide bearing location Modulation at wicket gate passing frequency

FRANCIS TURBINE 2 Cavitation aggressiveness as a function of the guide vane opening Maximum erosion risk : 80% of guide opening

CONCLUSIONS Turbine guide bearing pedestal is the best fixed measuring point for detection of erosive leading edge cavitation on blades Guide bearing measurements correlate with onboard measurements on the blade High frequency accelerometers and acoustic emission sensors serve for detection Amplitude demodulation techniques are the best method to identify hydrodynamic frequencies governing cavity fluctuations

CONCLUSIONS Maximum erosive cavitation as a function of machine load is indicated by the amplitude of wicket gate passing frequency fv. Forces on the blades are well transmitted to the turbine guide bearing through the shaft Amplitude of fv is not sufficient to quantify in absolute terms erosive forces if transmission properties are not considered Transmissibility functions can be obtained with an instrumented hammer with good coherence up to 15 kHz

ONGOING WORK Detailed study of onboard measurements in various prototypes Improve and apply calibration techniques for cavitation aggressiveness quantification. Include the material resistance from laboratory measurements.