Turbine Temperature Sensor Failure Detection by Peer Comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting temperature sensor failures in gas/steam turbine systems are inefficient, requiring manual examination of temperature graphs and event logs, which is time-consuming and prone to errors due to the large number of sensors and varying event messages across different turbine vendors.
Innovation Solution
A method that calculates characteristic values from measurement data of temperature sensors, identifies a sensor as defective if its value exceeds the maximum value of others by a predetermined factor, allowing for rapid detection without additional hardware or modifications to the turbine system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual examination of temperature graphs and event logs is used to detect sensor failures, then detection accuracy can be maintained, but the detection time and operator workload increase significantly
Solution Approach 1:
The system performs self-diagnosis by automatically comparing temperature measurements from multiple sensors against expected relationships, enabling the turbine control system to identify sensor failures without operator intervention. The control system calculates expected temperatures for redundant sensors and detects deviations automatically.
Solution Approach 2:
The manual mechanical process of examining graphs and logs is replaced with an automated computational system that uses algorithms to compare sensor readings, calculate expected values, and identify failures through electronic data processing rather than human visual inspection.
2Measurement precision
If manual examination of temperature graphs is used to detect sensor failures, then comprehensive analysis can be performed, but operator workload and complexity of the process increase
Solution Approach 1:
The control system autonomously performs the detection function by automatically comparing sensor readings, calculating expected temperatures, and identifying failures without requiring operator effort. The system serves itself by monitoring its own sensor health status.
Solution Approach 2:
The control system integrates multiple functions into a single automated process: it monitors temperature readings, compares them against expected relationships, detects sensor failures, and can trigger alarm functions, replacing multiple manual operations with one universal automated system.
3Extent of automation
If event log checking is used to detect thermocouple failures, then detection can be automated, but reliability decreases due to varying event messages from different vendors and incomplete reporting
Solution Approach 1:
The unreliable text-based event log checking is replaced with a robust computational comparison method that uses mathematical relationships between sensor readings. This substitution eliminates dependence on vendor-specific message formats and incomplete reporting by using direct physical relationship validation.
Solution Approach 2:
The system uses feedback from multiple temperature sensors to validate each other's readings. By continuously comparing actual measurements against expected values derived from other sensors and known thermal relationships, the system creates a self-validating detection mechanism that is independent of external event logging systems.
Data Source
Figure 1
Figure 2
AI summary
There is described a method of detecting a temperature sensor failure in a turbine system, the method comprising (a) obtaining (102) individual measurement values from each temperature sensor in a group of temperature sensors, (b) calculating (104) a characteristic value for each temperature sensor in the group based on the measurement values for the corresponding temperature sensor, (c) selecting (106) a first characteristic value among the calculated characteristic values, (d) determining (108) a first maximum value as the maximum of the characteristic values except for the first characteristic value, and (e) determining (110) that the temperature sensor corresponding to the first characteristic value is defective if the first characteristic value is larger than the first maximum value multiplied by a predetermined factor. Further, a corresponding device, system, computer program and computer program product are described.