Application of Spectral Method for Vibration-Induced High-Cycle Fatigue Evaluation of an HP Turbine Blade

Abstract

Abstract A method of fluid-structure interaction coupling is implemented for a forced-response, vibration-induced fatigue life estimation of a high-pressure turbine blade. Two simulations approaches; a two-way (fully-coupled) and one-way (uncoupled) methods are implemented to investigate the influence of fluidsolid coupling on a turbine blade structural response. The fatigue analysis is performed using the frequency domain spectral moments estimated from the response power spectral density of the two simulation cases. The method is demonstrated in light of the time-domain method of the rainflow cycle counting method with mean stress correction. Correspondingly, the mean stress and multiaxiality effects are also accounted for in the frequency domain spectral approach. In the mean stress case, a multiplication coefficient is derived based on the Morrow equation, while the case of multiaxiality is based on a criterion which reduces the triaxial stress state to an equivalent uniaxial stress using the critical plane assumption. The analyses show that while the vibration-induced stress histories of both simulation approaches are stationary, they violate the assumption of normality of the frequency domain approaches. The stress history profile of both processes can be described as platykurtic with the distributions having less mass near its mean and in the tail region, as compared to a Gaussian distribution with an equal standard deviation. The fully-coupled method is right leaning with positive skewness while the uncoupled approach is left leaning with negative skewness. The directional orientation of the principal axes was also analyzed based on the Euler angle estimation. Although noticeable differences were found in the peak distribution of the normal stresses for both methods, the predicted Euler angle orientations were consistent in both cases, depicting a similar orientation of the critical plane during a crack initiation process. It is shown that the fatigue life estimation was conservative in the fully-coupled solution approach.