Morphing Gas Turbine Component Design via Neural Network Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current design processes for gas turbine engine components are inefficient and costly due to the lack of predictive physics-based models, which hinders the ability to accurately simulate and optimize internal component geometries, making purely computer-driven simulation design unviable.
Innovation Solution
A hybrid process that combines empirical testing with computational optimization using a morphing component and a neural network system, where a test rig with morphing components iteratively adjusts geometries based on empirically determined performance parameters until optimized, allowing for efficient and accurate geometry optimization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If predictive physics-based models are used for designing gas turbine engine components, then design accuracy is improved, but computational cost and complexity increase prohibitively
Solution Approach 1:
The patent creates a digital twin or virtual model of the gas turbine engine that replicates the behavior of the physical system. This virtual model allows for accurate prediction and optimization of component geometries without requiring expensive and complex physics-based simulations for each design iteration, thereby maintaining design accuracy while reducing computational complexity.
Solution Approach 2:
The patent performs preliminary data collection and model training using physics-based simulations and experimental data before the actual design optimization process. By pre-training the neural network with comprehensive datasets, the system establishes accurate predictive capabilities in advance, allowing subsequent design iterations to use the pre-trained model rather than running full physics-based simulations each time.
2Productivity
If purely computer-driven simulation design is used, then design speed is improved, but reliability and accuracy deteriorate due to lack of predictive models
Solution Approach 1:
The patent introduces a neural network as an intermediary between the physical test rig and the design optimization process. The neural network learns from empirical data collected from the test rig and serves as a bridge that enables fast computer-driven design iterations while maintaining the reliability and accuracy of empirical measurements, combining the advantages of both experimental and computational approaches.
Solution Approach 2:
The patent implements a feedback loop where the neural network's predictions are continuously validated and refined against actual empirical data from the test rig. This feedback mechanism ensures that the computer-driven simulation maintains high reliability and accuracy by constantly correcting and improving its predictive models based on real-world measurements, while still enabling rapid design iterations.
3Measurement precision
If traditional iterative design processes are used, then design accuracy is improved, but time consumption and cost increase significantly
Solution Approach 1:
The patent replaces the traditional mechanical iterative design process (building physical prototypes, conducting manual tests, and repeating the cycle) with an automated system using neural networks and computer-driven optimization. This substitution maintains design accuracy by preserving the empirical testing component while dramatically reducing design cycle time through automated data collection, analysis, and optimization algorithms.
Solution Approach 2:
The patent introduces dynamic adaptability into the design process by using a neural network that can learn and adapt from empirical data in real-time. The system dynamically adjusts its predictive models and optimization parameters based on the data collected from the test rig, enabling accurate design iterations to be performed much faster than traditional static methods by continuously improving its understanding of the system behavior.
Data Source
AI summary
A process for designing an internal turbine engine component including operating a test rig incorporating a physical morphing component having a first geometry and generating a data set of empirically determined component performance parameters corresponding to the first geometry. Providing the data set of empirically determined component performance parameters to a computational optimization system and automatically. Determining a geometry optimization of the morphing component. Altering the geometry of the morphing component to match the geometry optimization. Reiterating operating the test rig and providing the data set of empirically determined component performance parameters.