Probabilistic HCF Analysis for Gas Turbine Airfoils
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current design and validation practices for gas turbine engines do not effectively capture the inherent variability in high cycle fatigue (HCF) behavior of components, leading to inadequate risk assessment and potential component failure due to vibratory stress cycles.
Innovation Solution
A method and system utilizing computer processors to build flexible models and emulators based on parametric data, predicting HCF probability by accounting for mistuning and damping effects, and displaying results in a histogram to identify contributing parameters, thereby optimizing airfoil design and reducing stress variability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If deterministic, single-point assessment methods are used for HCF analysis, then the analysis process is simple and deterministic, but the inherent variability in HCF behavior cannot be captured
Solution Approach 1:
The patent transforms the HCF analysis from deterministic single-point assessment to probabilistic analysis by changing the fundamental parameter approach. It uses parametric data representing distributions of material properties, geometric variations, and loading conditions across multiple parameters simultaneously, enabling capture of inherent variability while maintaining computational tractability through systematic parameter variation
Solution Approach 2:
The patent creates emulators (surrogate models) that replicate the behavior of complex flexible models. These emulators are trained on data from detailed finite element models and then used to efficiently evaluate HCF risk across many parameter combinations, capturing variability without requiring computationally expensive full-scale simulations for every assessment
2Power
If higher rotor speeds and reduced axial spacing are implemented, then engine power and efficiency are improved, but HCF stress on components increases
Solution Approach 1:
The patent performs HCF risk assessment during the design phase rather than after component fabrication. By evaluating probabilistic HCF risk before final design decisions are made, engineers can identify potential failure modes and optimize designs to prevent HCF issues before they occur in high-power, high-stress operating conditions
Solution Approach 2:
The patent incorporates feedback loops where probabilistic HCF assessment results are fed back into the design optimization process. The system identifies which parametric variations most significantly affect HCF risk and uses this information to refine design parameters, creating an iterative process that continuously improves component resistance to HCF while maintaining power output
3Measurement precision
If probabilistic analysis with multiple flexible models and emulators is used, then HCF variability is captured accurately, but computational resources and analysis time increase
Solution Approach 1:
The patent segments the HCF analysis into distinct computational stages: (1) creating detailed flexible models for a limited set of representative parameter combinations, (2) training emulator models on this data, and (3) using emulators to efficiently evaluate HCF risk across the full parameter space. This segmentation allows accurate capture of variability in critical regions while using computationally efficient emulators elsewhere
Solution Approach 2:
The patent applies full detailed flexible model analysis only to representative cases needed for emulator training, rather than exhaustively analyzing every possible parameter combination. The emulators then handle the remaining evaluations with reduced computational effort, providing sufficient accuracy for design decisions without the excessive time cost of complete exhaustive analysis
Data Source
AI summary
A novel probabilistic method for analyzing high cycle fatigue (HCF) in a design of a gas turbine engine is disclosed. The method may comprise identifying a component of the gas turbine engine for high cycle fatigue analysis, inputting parametric data of the component over a predetermined parameter space into at least one computer processor, using the at least one computer processor to build a plurality of flexible models of the component based on the parametric data of the component over the predetermined parameter space, using the at least one computer processor to build a plurality of emulators of the component based on the plurality of flexible models, and using the at least one computer processor to predict a probability of HCF based at least in part on the parametric data of the component over the predetermined parameter space and the plurality of emulators.