Gas Turbine Rotor Assembly Clocking for Vibration Mitigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for minimizing vibration in gas turbine engines by adjusting clock angles do not necessarily achieve optimal vibration reduction due to the distribution of unbalance in module assemblies, which can excite different modal tendencies and amplify vibration at specific RPM ranges.
Innovation Solution
Utilizing artificial neural networks (ANNs) to predict vibration at locations of interest in gas turbine engines by training networks with data on contributors to unbalance, such as radial offset, squareness error, and residual unbalance, and optimizing clock angles and trim weights to mitigate vibration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If traditional methods are used to minimize unbalance in module assemblies by adjusting clock angles, then unbalance can be reduced to acceptable levels, but vibration cannot be optimally minimized due to distribution effects and modal tendencies
Solution Approach 1:
The patent changes the approach from minimizing unbalance magnitude to optimizing unbalance distribution by adjusting clock angles as key parameters. The neural network optimizes clock angle parameters to distribute unbalance in a way that avoids exciting critical modal tendencies, thereby minimizing vibration even when unbalance magnitude remains constant.
Solution Approach 2:
The patent replaces traditional mechanical balancing methods with an artificial neural network-based predictive system. Instead of relying on iterative mechanical adjustments and physical measurements, the ANN predicts vibration levels based on input parameters (unbalance contributors, clock angles) and provides optimized configurations, substituting physical trial-and-error with computational intelligence.
2Manufacturing precision
If clock angles are adjusted to minimize unbalance magnitude, then unbalance can be corrected, but the distribution of unbalance may still excite different modal tendencies and amplify vibration at specific RPM ranges
Solution Approach 1:
The patent applies preliminary action by using the neural network to predict and optimize clock angle configurations before final assembly. The system calculates optimal clock angles that pre-distribute unbalance to avoid critical speeds, preventing vibration amplification before the engine operates at problematic RPM ranges.
Solution Approach 2:
The neural network incorporates feedback mechanisms by using training data that includes relationships between unbalance distribution, clock angles, and vibration outcomes. The ANN learns from historical data how different clock angle configurations affect vibration across various RPM ranges, providing optimized recommendations that account for modal tendencies.
3Object-affected harmful factors
If neural networks are used to predict and optimize clock angles, then vibration can be minimized by optimizing unbalance distribution, but the complexity of the assembly process increases
Solution Approach 1:
The patent introduces an artificial neural network as an intermediary between unbalance measurement and clock angle determination. The ANN serves as a computational mediator that processes input parameters (unbalance contributors, geometric drivers) and outputs optimized clock angles, replacing complex manual calculations and iterative adjustments with a trained predictive model.
Solution Approach 2:
The neural network creates a virtual model or copy of the physical assembly system during training. The ANN learns the complex relationships between unbalance distribution, clock angles, and vibration by processing training data that represents various assembly configurations, allowing it to predict optimal settings without physical trial-and-error.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method of optimizing the assembly of rotating hardware of a gas turbine engine (GTE) includes obtaining a respective input data set (152) indicative of one or more contributors to unbalance for one or of a plurality of stages (304) of one or more modules (302A; 302B) of the GTE. For each of the module(s), one or more neural networks associated with the module are utilized to obtain, based on the respective input data set (152) for the module (302A; 302B), a set of optimized clock angles (168) for arranging the stages (304) of the module relative to each other to mitigate vibration of the GTE. Each of the first neural network(s) (150A) has been trained with training data including the contributor(s) to unbalance, and at least one rotor dynamics model that uses the training data and the set of clock angles (168) from the neural network(s) to predict vibration at one or more locations of interest in the GTE.