A Study on the Impact of Blade Tolerances on Turbine Performance

Abstract

When manufacturing blades, there is a need for tolerancing around nominal blade design. Tolerance design is a balance between turbine performance and cost efficiency; a tolerance band that is too wide will result in a large scatter in turbine characteristics, whereas a too narrow tolerance band leads to high manufacturing costs due to high scrap rates of blades not meeting requirements. Determining an optimal manufacturing tolerance is therefore an issue of great importance. The purpose of this thesis was to study the impact of geometric variability within manufacturing tolerances on turbine performance. This Master’s Thesis was conducted at the Department of Rotors at GKN Aerospace in Trollhättan, Sweden. In this thesis, a rational method has been developed to vary blade profiles of both stator and rotor blades within current manufacturing tolerances at GKN Aerospace. Blade profiles are modified using a sinusoidal profile variation, which allows variation of amplitude Ai, period ni, and phase angle 'i, to model manufacturing variations within profile tolerance limits. The method has been utilised to study the impact of simulated manufacturing tolerances on turbine performance through CFD simulations. Several design cases were produced by variations of the function used to produce sinusoidal profile variations, and evaluated in the project. It was seen that limiting curves for flow function Q were characterised by stator-rotor configurations for minimum-minimum and maximum-maximum profile tolerance, while limiting curves for efficiency where characterised by stator-rotor configurations for minimum-maximum and maximum-minimum profile tolerances. The impact of subsequent design cases for profile variations within manufacturing tolerances fell within these limiting curves. Certain blade design parameters, such as trailing edge thickness and unguided trailing edge, were identified as key parameters with respect to impact on turbine performance.