The use of acoustic emission for the condition assessment of gas turbines: Acoustic emission generation from normal running

Mohammad S. Nashed, John A. Steel, Robert L. Reuben

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

Many approaches have been taken to the condition monitoring of gas turbines including performance analysis, vibration monitoring and lubricant debris monitoring. Acoustic emission monitoring has the potential to provide information about turbine operation and faults as they occur and has two possible advantages over other techniques. Firstly, because acoustic emission is sensitive to minor changes which do not necessarily involve whole-body motion, it is potentially able to reveal faults at an early stage. Secondly, because acoustic emission propagates over the structure from the source(s) to the sensor(s), it has the capacity to locate the source of any fault signal without intrusion. This paper explores the nature of the acoustic emission signals generated in a laboratory-scale gas turbine in order to extract and select features of the signal under normal running conditions and to establish the physical source(s) of this acoustic emission. A series of tests with the turbine running normally, either idling with the speed being controlled by fuel and air flow, or under load at fixed fuel and air flow with the speed being controlled by the amount of load applied, provided a range of conditions of gas flow through the turbine. An ancillary set of simplified tests with the free power turbine impeller jammed or absent was used to help distinguish between components of the acoustic emission associated with standing waves in various parts of the turbine and turbulence around the impeller. The results provide the first systematic study of fluid-induced acoustic emission in turbines and, as such, offer a baseline interpretation for acoustic emission generation in turbines and fluid machinery generally. Specifically, the findings will be compared with measurements made on the same turbine with simulated blade faults in a future publication.

Original languageEnglish
Pages (from-to)286-308
Number of pages23
JournalProceedings of the Institution of Mechanical Engineers, Part E: Journal of Process Mechanical Engineering
Volume228
Issue number4
DOIs
Publication statusPublished - Nov 2014

Keywords

  • Acoustic emission
  • gas turbine
  • condition monitoring
  • artificial neural networks
  • pattern recognition
  • signal processing
  • CAVITATION

Cite this