Turbomachine Blade Repair Using ML-Optimized Blend Geometry
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for repairing turbomachinery airfoils are inadequate in ensuring that the resonant natural frequencies of the repaired blades remain within design specifications, which can lead to structural resonance and fatigue failure.
Innovation Solution
A machine learning-based approach is used to determine the optimal blend geometry and compensatory blends to adjust the natural frequencies of the airfoil blades, ensuring they stay within design limits by removing damaged material and adding material at specific locations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Shape
If material is removed around the damaged region to blend the repair, then the aerodynamic smoothness is improved, but the blend dimensions are limited by resonant frequency requirements
Solution Approach 1:
The patent applies parameter changes by modifying the blend radius and blend width dimensions to simultaneously achieve aerodynamic smoothness and maintain resonant frequency compliance. The optimization process adjusts these geometric parameters within specific ranges (blend radius: 0.02-0.08 inches, blend width: 0.1-0.3 inches) to resolve the contradiction between shape quality and structural reliability
Solution Approach 2:
The patent employs preliminary action by performing finite element analysis and resonant frequency calculations before finalizing the blend geometry. This allows the repair design to anticipate and prevent resonant frequency violations, ensuring both aerodynamic smoothness and structural reliability are achieved in the preliminary design stage
2Ease of repair
If larger blend dimensions are used to repair more damage, then the repair capability is improved, but the resonant natural frequencies may change beyond design limits causing structural resonance
Solution Approach 1:
The patent implements feedback by using an iterative optimization process that calculates resonant frequencies based on proposed blend geometries, compares them against design limits, and adjusts the blend parameters accordingly. This closed-loop approach enables larger repair dimensions while maintaining frequency compliance through continuous feedback validation
Solution Approach 2:
The patent applies dynamics by making the blend geometry adaptive to the specific damage configuration and blade characteristics. The optimization process dynamically adjusts blend radius, width, and depth parameters based on the damaged area size, location, and the blade's modal properties, allowing maximum repair capability within frequency constraints
3Reliability
If blending operations are performed to re-contour the damaged region, then the aerodynamic performance is improved, but the manufacturing complexity increases
Solution Approach 1:
The patent applies local quality by concentrating the blending operations in specific localized regions around the damage site rather than extensive area-wide blending. The optimized blend geometry focuses material removal and addition only where necessary to restore aerodynamic contours, reducing overall manufacturing complexity while maintaining aerodynamic performance
Data Source
AI summary
The present disclosure is directed to a method and system for repairing a blade for a turbomachine that has been damaged due to foreign object impact. Machine learning algorithms are used to determine an amount of material to be removed from one or more damage locations and determine an amount of material addition to the damaged portion(s) and/or at other locations on the blade to keep the natural frequencies of the repaired blade within design limits during operation.