Automated Detection of Premature Flow Transitions on Wind Turbine Blades Using Model-Based Algorithms

Abstract

Defects on rotor blade leading edges of wind turbines can lead to premature laminar–turbulent transitions, whereby the turbulent boundary layer flow forms turbulence wedges. The increased area of turbulent flow around the blade is of interest here, as it can have a negative effect on the energy production of the wind turbine. Infrared thermography is an established method to visualize the transition from laminar to turbulent flow, but the contrast-to-noise ratio (CNR) of the turbulence wedges is often too low to allow a reliable wedge detection with the existing image processing techniques. To facilitate a reliable detection, a model-based algorithm is presented that uses prior knowledge about the wedge-like shape of the premature flow transition. A verification of the algorithm with simulated thermograms and a validation with measured thermograms of a rotor blade from an operating wind turbine are performed. As a result, the proposed algorithm is able to detect turbulence wedges and to determine their area down to a CNR of 2. For turbulence wedges in a recorded thermogram on a wind turbine with CNR as low as 0.2, at least 80% of the area of the turbulence wedges is detected. Thus, the model-based algorithm is proven to be a powerful tool for the detection of turbulence wedges in thermograms of rotor blades of in-service wind turbines and for determining the resulting areas of the additional turbulent flow regions with a low measurement error.