Compressor Stall Precursor Classification via Correlation Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing compressor assemblies in turbo machines face challenges in detecting and distinguishing between different types of stall or surge conditions, leading to potential damage, uncommanded engine shutdowns, and reduced performance, as current methods fail to effectively prevent or mitigate these conditions before they occur.
Innovation Solution
A computer-implemented method and system that compares data sets obtained over different periods to determine correlation factors, removes mean values, and classifies stall precursors as spike, modal, or combination stall precursors, allowing for the generation of control signals to adjust compressor operation and prevent stall or surge conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If compressor assemblies operate at high loading to maximize performance, then productivity and power output are improved, but the risk of compressor stall and surge increases
Solution Approach 1:
The system performs preliminary detection of stall precursors by continuously monitoring compressor operation parameters and comparing them against learned patterns from training data. This early detection enables preventive action before actual stall or surge occurs, allowing the compressor to operate at high loading safely by intervening only when necessary.
Solution Approach 2:
The system implements a feedback mechanism where compressor performance data is continuously collected, analyzed, and used to adjust control parameters in real-time. The control system receives feedback about stall precursor detection and automatically adjusts compressor operation to maintain stable conditions while maximizing productivity.
2Reliability
If general stall prevention methods are applied, then reliability is improved, but the ability to address specific stall types (spike, modal, combination) is lost
Solution Approach 1:
The system segments stall detection into distinct categories (spike stall, modal stall, and combination stall) by analyzing specific characteristics of compressor data. Each stall type is identified through pattern recognition of unique signatures in the monitored parameters, enabling targeted prevention strategies for each specific stall type rather than using a single general approach.
Solution Approach 2:
The system applies different analysis methods and thresholds for detecting different stall types based on their local characteristics. Spike stall detection focuses on rapid pressure changes, modal stall on oscillatory patterns, and combination stall on mixed signatures. This localized approach to detection precision enables reliable differentiation and targeted response for each stall type.
3Measurement precision
If complex data analysis is performed to detect and classify stall precursors, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The system uses pattern copying from historical training data to identify current stall precursors. Instead of implementing complex real-time analysis of all possible stall modes, the system compares current compressor data against pre-recorded patterns of spike stall, modal stall, and combination stall. This pattern-matching approach achieves high detection precision while keeping the device complexity manageable through reuse of established data patterns.
Solution Approach 2:
The system transforms raw compressor data into specific parameter representations that highlight stall precursor characteristics. By converting complex multi-dimensional data into focused parameters such as pressure ratio changes, flow rate variations, and their rates of change, the system achieves precise stall detection while simplifying the processing requirements through parameter transformation.
4Reliability
If real-time monitoring and classification of stall precursors is implemented, then reliability is improved, but use of energy increases
Solution Approach 1:
The system uses a universal data processing framework that handles multiple stall types (spike, modal, combination) through a single integrated analysis pipeline. The same core processing architecture and algorithms are used regardless of which stall type is detected, eliminating the need for separate dedicated systems for each stall type. This multi-functionality reduces overall energy consumption while maintaining comprehensive monitoring capability.
Data Source
AI summary
A turbo machine, a computer-implemented method, and a computer system for operating a compressor assembly are provided. The method includes comparing a first data set and a second data set to determine a first correlation factor, comparing the first correlation factor to a first threshold that at least partially determines whether a stall precursor exists, removing mean values from the first data set and the second data set, comparing the first data set and the second data set each removed of mean values to determine a second correlation factor, and comparing the second correlation factor to the first threshold, and classifying the stall precursor as either a spike stall precursor, a modal stall precursor, or a combination stall precursor.