Compressor Blade Damage Prediction for Gas Turbine Operability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computational fluid dynamic (CFD) techniques for predicting gas turbine engine compressor blade damage are time-consuming, labor-intensive, and not sufficiently accurate, leading to potential unnecessary condemnation of components and increased costs due to inaccurate predictions of operability.
Innovation Solution
A computer-implemented method using machine learning algorithms, specifically artificial neural networks, is trained with data quantifying damage to compressor blades and operating parameters, enabling determination of compressor operability by inputting damage data and receiving output parameters for predicting stalling throttle coefficient and pressure rise characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If CFD techniques are used to model the effect of damage on compressor operability, then detailed analysis of blade damage impact is achieved, but the process becomes time-consuming and labor-intensive taking days of expert time
Solution Approach 1:
The patent applies preliminary action by pre-training machine learning models using extensive CFD simulation data and experimental measurements before actual damage assessment. The model is prepared in advance with a comprehensive database of damaged and undamaged compressor configurations, enabling rapid predictions without requiring time-consuming CFD analysis during the actual assessment phase.
Solution Approach 2:
The patent creates a simplified computational model that copies the essential physics and behavior of the complex CFD system. The machine learning model replicates the damage-impact relationships learned from detailed CFD simulations, providing approximate predictions that are sufficiently accurate for operational decisions but computationally much faster.
2Measurement precision
If CFD techniques are used to model large separations caused by blunt blades, then detailed flow analysis is achieved, but the current industry CFD methods are insufficiently accurate for predicting stall point
Solution Approach 1:
The patent replaces the complex mechanical CFD computation system with a machine learning-based predictive system. The ML model has learned the complex flow physics and separation patterns from training data, substituting iterative numerical solutions with direct mathematical predictions that are both faster and more accurate for stall point prediction.
Solution Approach 2:
The patent transforms the problem from solving complex partial differential equations to evaluating learned parameter relationships. The model uses input parameters (damage geometry, operating conditions) to directly predict output parameters (stall point, performance degradation) based on patterns learned during training, avoiding the need to resolve complex flow fields.
3Reliability
If inaccurate predictions of blade damage impact are made, then components may be condemned unnecessarily, but this results in significant cost penalties and increased waste
Solution Approach 1:
The patent implements feedback by comparing machine learning predictions with actual operational outcomes and CFD validation data. The model continuously refines its predictions based on discrepancies between predicted and actual performance, reducing false condemnations and improving the accuracy of operability determinations over time.
Data Source
AI summary
A computer-implemented method controls input of at least a portion of a first training data set into a first machine learning algorithm. The first training data set includes data quantifying damage to a first compressor and data quantifying a first operating parameter of the first compressor. The first machine learning algorithm is executed, and data quantifying the first operating parameter is received as an output of the first machine learning algorithm. The first machine learning algorithm is trained 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 is configured to enable determination of operability of a second compressor of a gas turbine engine.