Predictive maintenance of wind turbine low-speed shafts based on an autonomous ultrasonic system

Abstract

Low-speed shafts breakage in wind turbines (WT) supposes, besides an elevated repair cost, an unattainable risk for workers integrity due to the induced rotor fall. Based on a Root Cause Analysis (RCA) an autonomous ultrasonic monitoring system has been developed with the aim of extending the shaft useful life safely beyond the 50% obtained when manual ultrasonic inspections are used.

This system consists of ad-hoc electronics, a specific firmware that includes the detection and assessment algorithm, and an annular array of transducers attached to the shaft by a specific mechanical holder. To estimate the transducers type, quantity and optimum location and to establish references, healthy shafts were analyzed and tests specimens with artificial defects were manufactured and studied reproducing the critical crack sizes estimated in the RCA. The firmware controls the entire system and is the responsible for autonomous diagnostics that launches preliminary alerts when detecting non-critical cracks. The system creates an “imminent failure” alert when the shaft has consumed up to 96% of its useful life. The system has been tested in real WTs with positive results, omitting no-relevant cracks and even detecting certain type of cracks not detected by manual inspections.