Contrast Enhancement in Thermographic Flow Visualization

Abstract

The distribution of the different flow regimes in the boundary layer of an airfoil significantly determines lift and drag [1] and, thus, directly influences the overall efficiency of wind turbines [2]. For the development and optimization of rotor blade profiles, a non-invasive and contactless measurement method is desired in order to investigate the boundary layer flow at wind turbine rotor blades in operation. In contrast to most conventional methods, thermographic measurements allow a non-invasive, contactless visualization of the boundary layer flow without preparing the rotor surface [3]. However, the key challenge for in-process measurements on wind turbine rotors is to cope with a low contrast to noise ratio in the thermographic images, because no active blade heating can be applied in field measurements to increase the necessary temperature difference between the surface and the boundary layer [4], and low contrast to noise ratios hinder a clear localization of the different flow regimes. In order to increase the image contrast and the distinguishability between the flow regimes, a time series of thermographic images is evaluated by means of the non-negative matrix factorization. This novel measurement approach is validated by means of experiments with a non-heated cylinder in cross-flow condition. Figure 1 shows (a) one resulting image of the non-negative matrix factorization in comparison with (b) the mean image of the time series as reference. As a result, an increase of the contrast to noise ratio between every pair of consecutive flow regimes is achieved. As an example, the contrast to noise ratio between the difficult to distinguish turbulent and turbulent separated flow regimes is increased by 35.7 %. This is achieved by reducing the systematic temperature gradients in flow direction within the flow regimes, leading to an improvement of the homogeneities by 62.55 % in average. Hence, a computer-aided image processing method such as the non-negative matrix factorization can increase the contrast between boundary layer flow regimes in thermographic flow visualizations and improve their distinguishability. Therefore it provides a new possibility to cope with the difficult measurement circumstances on wind turbines in operation.