Comparative Experimental Determination of Mistuning of a Bladed Turbine Wheel Using Different Evaluation Approaches

Abstract Random mistuning of integrally manufactured turbine and compressor wheels can lead to severely high magnitudes in blade vibration. This is caused by unavoidable imperfections due to manufacturing processes or material inhomogeneities and the resulting deviation in geometry and structure between the individual blades. In order to be able to calculate the vibration response of bladed wheels correctly the actual mistuning should be quantified most accurately and taken into account e.g. in simulation models. Since mistuning is usually described as a frequency deviation of the nominal value measuring procedures have been developed in order to measure existing wheel hardware. An industry-suited procedure has been introduced by Kühhorn and Beirow [1] which modifies a conventional experimental modal analysis by isolating the blade that is excited and measured from all remaining ones. Therefore, additional masses are applied in order to cause a detuning and thus an isolated vibration of the blade under consideration. This enables the often coupled frequency response function peaks to decouple and obtain only one single peak which marks the actual frequency mistuning of the blade. Indeed, the procedure has proved to be accurate if isolated blade mode families are considered. In other cases, blade-disk-coupling can still be a problem and falsify the evaluation. The reason for this is an unfavourable application of the additional masses which most often cannot be avoided and may lead to an inaccurate or even erroneous mistuning value. However, Zhou et al. [2] have developed a novel approach to calculate an error term in order to correct the measured frequency mistuning. In this paper a radial turbine wheel is subjected to the blade frequency tests. The evaluation is carried out by using both, the conventional evaluation of the frequency response function and the novel approach. The error term is calculated and the mistuning distributions were corrected. Finally, the impact and measurement falsification of additional masses when testing a radial turbine wheel is discussed.