Compressor Blade Damage Prediction for Faster Operability Assessment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computational fluid dynamic (CFD) techniques for predicting damage in gas turbine compressor blades are time-consuming, labor-intensive, and not sufficiently accurate, leading to potential unnecessary condemnation of components and increased costs due to inaccurate predictions of blade damage impact on operability.
Innovation Solution
A computer-implemented method using machine learning algorithms, specifically artificial neural networks, to quantify damage in compressor blades by convoluting received data with damage parameters, determining the importance of these parameters, and performing optimization using Fourier-related transforms to improve prediction accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If computational fluid dynamic (CFD) techniques are used to model the effect of damage on compressor operability, then the analysis can be performed using detailed geometric modeling, but the process becomes time consuming and labor intensive taking days of expert time
Solution Approach 1:
The patent creates simplified geometric representations (copies) of damaged compressor blades that capture the essential damage features without requiring complete detailed geometry. These simplified models serve as adequate substitutes for full CFD modeling, enabling rapid assessment while maintaining sufficient prediction accuracy for engineering decisions.
Solution Approach 2:
The patent extracts only the critical damage features from the complete blade geometry, separating the essential damage characteristics (crack locations, sizes, orientations) from the full geometric detail. This extraction allows the damage assessment to focus on the most relevant features without the computational burden of processing complete geometric models.
2Measurement precision
If detailed geometric modeling is performed for damaged blades, then comprehensive damage assessment is possible, but the process becomes labor intensive requiring expert intervention
Solution Approach 1:
The patent implements an automated system that performs damage assessment without requiring expert intervention. The machine learning algorithm automatically processes measured damage data, applies the simplified geometric modeling approach, and generates operability predictions, replacing the need for expert analysts to manually perform detailed geometric modeling and assessment.
Solution Approach 2:
The patent replaces the manual expert-based mechanical process of geometric modeling and damage assessment with an automated computational system using machine learning algorithms. This substitution eliminates the need for expert human intervention while maintaining or improving assessment consistency and accuracy.
3Reliability
If CFD techniques are used for damage prediction, then detailed analysis can be performed, but the methods are not sufficiently accurate for predicting stall point and large separations caused by blunt blades
Solution Approach 1:
The patent changes the approach from detailed geometric parameter modeling to using damage feature parameters (crack size, location, orientation) combined with blade performance parameters. This parameter transformation enables the machine learning model to directly predict stall point and separation characteristics without requiring complex detailed geometric modeling, improving both accuracy and efficiency.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A computer-implemented method comprising: controlling input of at least a portion of a first training data set into a first machine learning algorithm, the first training data set including: data quantifying damage to a first compressor; and data quantifying a first operating parameter of the first compressor; executing the first machine learning algorithm; receiving data quantifying the first operating parameter as an output of the first machine learning algorithm; and training the first machine learning algorithm using: the received data output from the first machine learning algorithm; and data quantifying the first operating parameter of the first compressor, the trained first machine learning algorithm being configured to enable determination of operability of a second compressor of a gas turbine engine.