Probabilistic turbine blade thermal analysis of manufacturing variability and toleranced designs

Abstract

Manufacturing variability is likely the primary cause of a large scatter in the life of gas turbine hot-section components. This research deals with schemes to improve robustness through tolerancing input parameters in ranges of the distributions which make non-conformances more likely. The need for probabilistic analysis to investigate this problem is substantiated due to differences which arise when input parameters vary at different levels, for example the engine-to-engine and blade-to-blade level. Specifically, the importance of blade-to-blade level input parameters relative to engine-to-engine level input parameters becomes increasingly important for larger numbers of blades in a row. A framework for calculating the potential number of prevented non-conformances and the corresponding cost savings associated with various tolerancing schemes is presented. Specifically this research investigates manufacturing variability and its effect on first-stage turbine blades through the use of a parametric CAD model, automated CAD regeneration software, and a parametric finite element thermal model. Probabilistic analysis is performed using Monte Carlo simulation on both the finite element model as well as response surfaces built from the finite element model.

(cont.) Blade-to-blade cooling flow variability, especially as a result of film-hole diameter variability in critical locations is identified as the most likely candidate for parameter tolerancing. More promising is a combined two-factor tolerancing scheme which additionally tolerances gas path temperature.