Probablistic Lifetime Analysis for Turbine Blades Based on a Combined Direct Monte-Carlo and Respond Surface Approach

M. Voigt, R. Mücke, M. Oevermann, K. Vogeler

Further progress in the development of modern gas turbines for aero engines and electric power stations requires both continuous optimization on component and system level as well as the use of new and innovative technology. Thereby, the design is often pushed closer to the physical limits, which demands refined lifecycle models and improved prediction methods for the structural behavior under different design and off-design conditions.
Due to the considerable costs of real-life testing, the knowledge on structural behavi-or and failure mechanisms of gas turbine components is often gained from validated numerical models. To obtain a realistic physical and computational model, uncertain-ties inherent in the design as material properties, loading and the operation conditi-ons have to be considered in the modeling process.
Uncertainties can be accounted for by Probabilistic Design Methods (PDMs) based e.g. on the direct Monte Carlo Simulation (MSC) and the Response Surface Method (RSM). Contrary to deterministic analysis approaches, in PDMs the most important design parameters are defined by their probability distribution functions rather than by fixed constants. As a result, the total variety of lifetime or/and performance data, the probability of failure, as well as the structural sensitivity of the stochastic variables are evaluated.
However, today's use of PDM for practical engineering problems is rare since com-mon modeling requirements (e.g. complex geometry and loading conditions, para-metric modeling) challenge the applicability of probabilistic methods to full scale ana-lysis models.
In this paper probabilistic design methods are applied to the lifetime prediction of two cooled gas turbines blades. A combination of direct Monte Carlo Simulation and Response Surface Method is used allowing a significant reduction of required finite element calculations. Therefore, the computational effort, which typically is high for PDM, becomes now acceptable even for full scale analysis models. As a result of the blade analyses, the probability of failure as well as the sensitivity of lifetime with respect to heat transfer variables and material properties is presented.

ASME Turbo Expo GT2004-53439, 49th ASME International Gas Turbine & Aeroengine Technical Congress & Exposition, Vienna, Austria, June 14-17, 2004