Performance-Based Geometric Tolerancing of Compressor Blades

Abstract

The relationship between statically measured geometric parameters (tolerances) and the aerodynamic performance of an airfoil are investigated in this paper. The goal is to determine which geometric parameters are critical to control during manufacturing, such that a blade will have acceptable aerodynamic performance. A probabilistic model of geometric variability for a three-dimensional blade is derived. Using this geometric model, probabilistic aerodynamic simulations are conducted to analyze the variability in aerodynamic performance. Tolerance optimization is then applied in which tolerance ranges are modified to best sort blades according to some arbitrary performance limit. The optimization is performed for several limits, expressed as a percent of nominal performance, to observe both which parameters best predict performance and the accuracy of that prediction at each limit. Two blade cases are considered, both based on the same compressor blade: the base compressor blade with nominal manufacturing noise; and a probabilistic redesign of the blade geometry designed to minimize the impact of manufacturing noise, also analyzed with nominal manufacturing noise. Results show the best static indicators of meanline performance are parameters concerning the LE of the airfoil, and the effectiveness of these parameters vary greatly depending on the chosen performance limit. In addition, it was shown that the optimized tolerances for the redesigned blade were consistently looser, or less restrictive, than those for the original blade population for a given performance limit. The differences in observed optimized tolerance ranges are small for less restrictive performance limits but at more aggressive performance limits, there is a 20–30% increase in tolerance range for the redesigned blade population.