The signal processing technique has achieved several recorded and verified successes on a laboratory-scale turbomachine blade-vibration test setup
Photo by: Darlene Creamer
Having developed a new signal-processing technique for the vibration monitoring and testing of turbomachinery, the University of Pretoria’s (UP’s) Centre for Asset Integrity Management (C-AIM) research facility aims to create industry partnerships to develop this technique to maturity and commercialisation.
The focus of C-AIM, established as part of the Department of Mechanical and Aeronautical Engineering, is to integrate its analysis and testing capability in assessing the structural integrity and performance of physical assets. This is then paired with sound scientific research to inform asset management decisions in the context of asset life cycles, UP C-AIM director Professor Stephan Heyns tells Engineering News.
Part of C-AIM’s research includes vibration monitoring, with a variety of rotating machine fault simulators from the Sasol Laboratory for Structural Mechanics Rotating Machinery at UP being used for the accelerated testing of machine elements and the development of new condition-monitoring algorithms.
“A notable project deals with using turbomachine blade-tip timing measurements for blade root damage detection. The technique performs significantly better than all the published algorithms . . . currently known in the literature,” Heyns highlights.
Blade-tip timing is used to monitor blade vibration and uses probes of various designs to detect the arrival time of individual rotor blades at a number of points around the rotor casing. The technique forms part of several online, nonintrusive condition-monitoring techniques focusing on vibration signals, or pressure fluctuations that are observed in the turbine blade testing.
UP PhD student in mechanical and aeronautical engineering Dawie Diamond developed the new statistical signal-processing technique for blade-tip timing measurements over the past two years. He uses the blade-timing approach in the analysis of synchronous vibration of turbine blades during a constant turbine shaft speed.
Diamond explains that this vibration is one of several vibration conditions of turbomachinery, which can include flutter, which occurs when there is aerodynamic instability and synchronous vibration.
“This statistical technique is based on Bayesian statistics (a system for describing epistemological uncertainty using the mathematical language of probability) that calculates the probability that a blade is in resonance, or vibrating, at one of its natural frequencies.
“It determines the size of the vibration in a probabilistic manner,” Diamond says, further explaining that by determining the size of the vibration, technicians can determine whether the blade is being excessively damaged, as well as the remaining blade life cycles.
Diamond highlights that the technique was successfully used in the Sasol laboratory from July to November 2014, with several recorded and verified successes on a laboratory-scale turbomachine blade-vibration test setup. The technique was also tested against other published methods, with the results providing an error of only 0.2% – a significant increase in accuracy – whereas another testing method provided an error of about 5% to 6%, he points out.
However, he acknowledges the challenges in testing practicalities, which can be verified only on industry-sized and -applied turbomachinery; hence, the drive for access to industry turbines to conduct active field research and complete the technique’s methodologies.
“Nevertheless, the research work has developed to such an extent that we believe we have a solution, which, in collaboration with industry partners, can be brought to commercial feasibility,” Heyns adds, believing that this will be a leap forward in understanding the conditions of turbomachinery.
He expects field testing to start within the next two years and stresses that, as turbomachinery, particularly turbines, lies at the heart of any power production process, it is crucial to ensure good integrity of these machines.
However, as these turbomachines are operated for extended and uninterrupted timeframes – for up to six or eight years – without being opened and inspected, Heyns points out that this eliminates easy access and the possibilities of nondestructive testing.
Diamond, therefore, emphasises the benefits of the signal-processing technique for industry, which would be to nonintrusively monitor the blades’ vibration and enable technicians to know, at any stage, the operational state of the machinery and, consequently, to implement any crucial operational changes if required.
“It would create a significant economic benefit, reducing plant maintenance costs and downtime, and preventing failures,” he concludes.