Individual Pitch Actuator Monitoring of Offshore Wind Turbines

Individual Pitch Actuator Monitoring for Offshore Wind Turbines

EPSRC Prosperity Partnership – “A New Partnership in Offshore Wind”

Dr Yanhua Liu and Professor Ron Patton, University of Hull

Offshore wind turbines have large rotor diameters and high towers for high energy capture. The significant development of offshore wind power in recent years has led to a significant decrease in the levelized cost of energy (LCoE) of this form of renewable energy. However, unexpected malfunction and failures of turbine components especially pitch system will result in expensive repairs and typically months of machine unavailability, thus increasing the operation and maintenance (O&M) costs and threatening to increase the LCoE. Hydraulic pitch systems (actuators & sensors) are prone to faults, affecting the turbine power output and system stability, which contribute approximately 22% of the annual turbine downtime just after the electrical subsystem. Therefore, it is of fundamental importance to design appropriate pitch actuator monitoring strategy to obtain the fault information and compensate the fault effects. During my research work, the following pitch actuator monitoring strategies are studied. The effectiveness of the proposed strategies is verified in the FAST 5MW NREL wind turbine model.

(1) Fault monitoring for pitch actuator stuck (PAS) faults. After PAS occurs, the pitch actuation output will stay constant no matter what the pitch reference is. Moreover, the pitch measurement will be a fixed value. A fault detection and isolation (FDI) strategy using a Kalman filter is proposed. The pitch system output estimates are generated from the Kalman filter and a residual is used to detect the faults.

(2) Fault monitoring for pitch incipient dynamic changing faults. Hydraulic leakage due to improper management of oil, pump wear from continuous operation& high air content in oil will cause pitch actuator system has changed pitch system dynamics. This leads to slow pitching and unstable outputs. A step-by-step sliding mode observer is adopted obtain the pitch fault estimates.

(3) Fault monitoring for pitch sensor faults. Pitch sensor faults lead to incorrect pitch position readings. Hence, the pitch control system cannot guarantee that each blade is in the reference position. 4 different sensor faults areconsidered including (i) biased sensor, (ii) stuck sensor output, (iii) total and (iv) partial sensor faults. A robust “unknown input observer” using H infinity optimization theory is used for accurate sensor fault estimation.