Component Life Estimation Using Probabilistic Crack Growth Models

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for estimating the life of electromechanical machine components, such as gas turbines, are overly conservative due to uncertainties in material properties and initial flaw sizes, leading to premature service intervals and increased costs.

Innovation Solution

A system and method that optimize crack growth rate models and temperature models using a joint optimization method, incorporating a scatter parameter to estimate component life, which accounts for non-normal distributions in material data through maximum likelihood estimates and temperature models.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If deterministic approach is used for fatigue crack life calculations, then conservative design is achieved, but component service lifetime is reduced and costs increase

Engineering Contradiction:
Improvedesign safetyVSAvoidcomponent service lifetime
Core Design Contradiction:
ReliabilityVSDuration of action of moving object

Solution Approach 1:

The patent transforms the deterministic design approach into a probabilistic approach by changing the fundamental parameter from single extreme values to statistical distributions of material properties, flaw sizes, and loading conditions. This allows for more realistic life predictions that account for variability without being overly conservative, thereby extending component service lifetime while maintaining adequate reliability.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent introduces probabilistic fracture mechanics as an intermediary methodology between deterministic design and actual component behavior. By using statistical distributions and reliability-based design frameworks, it mediates between the need for conservative safety margins and the desire to maximize component utilization, enabling optimized service intervals that prevent premature replacements.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Device complexity

If conventional least-squares regression is used for fitting crack growth data, then simple modeling is achieved, but accuracy is reduced due to non-normal distributions

Engineering Contradiction:
Improvemodeling complexityVSAvoidscatter estimation accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent changes the statistical parameter assumption from normal distribution to lognormal distribution for modeling scatter in crack growth data. This parameter change better captures the actual behavior of material property variations and crack growth measurements, significantly improving the accuracy of scatter estimation while maintaining reasonable modeling complexity through the use of established statistical methods.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3819806A1System, apparatus and method for estimating life of components
Publication Date: 2021.05.12 SIEMENS ENERGY GLOBAL GMBH & CO KG
  • EP3819806A1 patent drawingFigure 1
  • EP3819806A1 patent drawingFigure 2
  • EP3819806A1 patent drawingFigure 2A

AI summary

A method (200) for estimating life of a component includes obtaining (202) fracture data (108) corresponding to a component. The fracture data (108) comprises first dataset (128) corresponding to a threshold region where the crack in the component is dormant below a fatigue threshold. The method (200) further includes determining (204) initial estimates of parameters of a crack growth rate model and parameters of temperature models corresponding to the crack growth rate model based on the fracture data (108). The method (200) also includes computing (206) optimized parameters of temperature models corresponding to the crack growth rate model, and a scatter parameter via simulation of a joint optimization method using the initial estimates. The method (200) includes determining (208) a cumulative distribution function based on the optimized parameters and the scatter parameter and estimating (210) life of the component based on the cumulative distribution function.