Experimental Validation of Empirical Methods for Dynamic Stress Prediction in Turbomachinery Blades

Abstract

Turbomachinery blade fatigue life estimation requires reliable knowledge of actual static and dynamic stresses occurring within the blades. A common method for predicting dynamic stresses is to construct a finite element model of the blade and simulate the dynamic response to aerodynamic loads. Although this method is powerful and very useful, modeling errors (geometry, boundary conditions, stress concentrations, damping, etc.) may result in inaccurate stress predictions. Furthermore, unavoidable variability in manufacturing results in blade mistuning, which significantly affects stress amplification at resonance. This paper presents two empirical methods for predicting dynamic stresses in turbomachinery blades that include the actual effects of structural damping and mistuning. Both methods use strain gauge measurements from a blade modal test to obtain load to strain transfer functions, which are applied to predict the blade strain or stress response to a simulated load. The advantages and disadvantages of each method are discussed. The predictions of each method are compared with dynamic blade strain data acquired during a rotating test of a centrifugal compressor impeller.