Solving the challenging puzzle of the Grimsel II runner vibration

Abstract

Abstract After 30 years of operation, the runners of the four 100 MW high-head Francis turbines successively developed fatigue cracks starting from the hub-side trailing-edge fillets. Assessment of alternating stress in normal turbine operation revealed that stress levels were not nearly large enough to explain the damages. Consequently, a test campaign comprising start and stop manoeuvres in addition to normal turbine operation was conducted in 2017. Strain gauge signals obtained on-board of the runner blades were recorded together with casing vibration and acoustic signals. Intense high-frequency vibration centred around 611 Hz was detected in three transient conditions of operation: at speed-no-load (SNL), during coast-down and during back-to-back pump start. Common properties of these conditions were very low to no discharge, strong secondary flow in the runner and speed between 50 and 100 %. Due to the extremely narrow bandwidth, three possible ways of excitation were considered and checked: rotor-stator interaction (RSI), rotor-dynamic instability and Von Kármán excitation. RSI was excluded due to speed-independent frequency. Strong sidebands in the stationary-frame frequency spectra - obviously caused by runner shroud deformation shapes ND1 through ND5 - excluded any feedback from runner side chamber pressure; thus leaving Von Kármán excitation as the only possible diagnosis. Accordingly, a very safe profile was selected and implemented for the runner blade trailing edges. Subsequent field-testing at two modified units confirmed that the vibration problem at the harmful operation points had hereby been eliminated.