Due to manufacturing tolerances, wear during operation or regeneration processes like maintenance operation, the structural properties of turbine blades deviate from design condition to reference blades. This deviation usually causes higher vibration amplitudes and as a consequence a lower service life expectation. Many different calculation methods can be used to simulate these increased amplitudes of mistuned blades. The major resulting problem is on the one hand to capture the occurring deviation of the eigenfrequencies from the reference blade and on the other hand to incorporate these real deviations in simulations. Solving these problems with a simplified experimental setup will make it possible to predict the maximum amplitude and to avoid costly experiments in a rotating turbine.
The aim of the paper is to verify a simulation of the vibration amplitude by experiments using a reduction method to calculate a mistuned system in reasonable time. The results of the chosen simulation are compared to experiments in a rotating turbine.
To reduce the number of degrees of freedom of the full finite-element model and the computational effort, a multi-step reduction method is used. In the simulation, the centrifugal force, the structural damping, the steady static pressure on the blades, and the mistuning are considered. To find the occurring deviations of each manufactured blade, an experimental modal analysis is performed for every single blade in a non-rotating setup with the eigenfrequencies of every single blade as an output. The single-stage results of the simulation are subsequently compared to experiments in a 5-stage air turbine in which the vibration amplitudes and the eigenfrequencies of every blade in the last rotor blade row are measured by a tip-timing system.