Elsevier

Ultrasonics

Volume 41, Issue 9, March 2004, Pages 765-773
Ultrasonics

Diagnosis of continuous rotor–stator rubbing in large scale turbine units using acoustic emissions

https://doi.org/10.1016/j.ultras.2003.09.002Get rights and content

Abstract

Continuous rubbing between the shaft and surrounding seals or end-glands of electricity generating turbine units can escalate into very severe vibration and costly rotor damage. Therefore such rotor–stator contacts require early diagnosis so as to minimize the financial consequences of any unplanned shutdowns. Acoustic emissions (AEs) or stress wave monitoring at the bearings has been identified as a sensitive non-destructive monitoring technique for such rub conditions [Electr. Eng. Jpn. 110(2) (1990); IEEE Proc. 6 (2000) 79; Hall and Mba, 14th International Congress on Condition Monitoring and Diagnostic Engineering Management (COMADEM’2001), Manchester, UK, 2001, p. 21]. However, experimental results from real turbines have been scarce. This paper presents a diagnosis of continuous rotor–stator rubbing in an operational 500 MW turbine unit via high frequency AE measurement within a 100 KHz–1 MHz ultrasonic band. As detailed by Sato [Electr. Eng. Jpn. 110(2) (1990)] and reported in this paper the onset of a continuous rub contact at a seal/gland was revealed by a sinusoidal modulation within the raw ‘rf’ AE response. By synchronous measurement at adjacent bearings, an estimation of the location of the rub was calculated using the phase delay between the adjacent AE modulations. Importantly, the AE diagnosis was closely corroborated by post-inspection of the turbine rotor.

Introduction

To maximise efficiency, the seals and end-glands along the length of modern turbine units are in very close radial proximity to the rotating shaft. This is illustrated for a low-pressure section of a 500 MW unit in Fig. 1. It is therefore unsurprising that contacts or rubs can readily occur between the high-inertia 50 Hz shaft and the surrounding components. Factors that have been attributed to the onset of rubbing in rotating plant [1] include thermal effects, foundation movement, component movement, rotor unbalance or misalignment. Regardless of the exact relationship between cause and effect, the existence of rubbing is of great concern to the rotor dynamist engineer as it can develop into significant mechanical vibration leading to costly rotor damage. Rotor–stator rubbing may be broadly classified as either partial or continuous. The partial type describes brief intermittent contacts between the rotor and stator. Often these are at apparently random locations along the turbine and may be innocuous where contacts are light and the seals are backed by high torsion springs. However, partial rubs can be more detrimental to the health of the turbine when they occur at a single shaft location so as to make contact with the stator on every revolution of the shaft. Such once per-revolution partial rub conditions can lead to localized regions of increased temperature on the shaft that may cause it to bend [4]. Alternatively, continuous rotor–stator rubbing describes more sustained rotor–stator interaction over many shaft rotations and are always of concern to the rotor dynamics engineer. Such full annular conditions have a profound effect upon rotor dynamics. Diagnosis of this important type of rub condition via acoustic emissions (AEs) has been proposed [1], [2], [3] and is the topic of this paper.

Presently, the only non-destructive technique widely available for identifying shaft-seal rubbing in large scale turbines is vibration monitoring of the bearing pedestals. This involves analysis of vibration data from accelerometers operating within a 0–400 Hz band permanently attached to the bearing housing. However, this approach has not proved to be effective in detecting the very earliest stages of rotor–stator rubbing. Moreover, its success in identifying more severe rubbing is highly dependent upon expert interpretation from the rotor dynamist as increased vibration has a number of potential causes and can only be attributed to rubbing in the most well developed cases. Therefore a more sensitive monitoring technique that could provide a direct and unambiguous indication of the onset of rotor–stator rubbing conditions is extremely valuable.

Frictional rubbing between metallic components cause a release of transient broadband stress wave energy (SWE) referred to as AEs. It is considered that this SWE within a 100 KHz–1 MHz band can propagate along the shaft and across the various component interfaces within the bearing, so as to be registered by an AE sensor attached to the bearing housing. Clearly acoustic attenuation at such high frequency and extraneous background noise registered by the AE sensor in large scale rotating machinery has contributed to a low signal-to-noise ratio at the sensor and has necessitated adequate signal processing to increase the probability of rub detection. However, it is argued that the AE wave modes across the shaft propagate successfully without significant attenuation. Rayleigh surface waves are thought to be predominant for rub induced AE waves across the shaft. These are elliptical waves slower than both longitudinal and transverse modes and they penetrate no more than a couple of wavelengths.

To examine the AE response at the bearings during continuous rubbing a low acoustic noise test-rig capable of emulating continuous shaft-seal rubbing was employed for preliminary tests. In addition to validating that AE waves from a continuous rub can be detected at the bearings, these tests revealed that continuous rubbing has a significant effect upon the rotor dynamics that is manifested by a characteristic modulation in the measured AE response. In addition, previous work [1] has suggested that a more detailed estimation of the exact location of the rub along the rotor can be obtained by simultaneous AE measurement at each of the supporting bearings, as indicated in Fig. 1. This paper presents validation of this AE technique for a real operational 500 MW steam unit. Significantly, very specific AE diagnosis of a rub close to one of the bearings was vindicated by post-inspection of the rotor.

Section snippets

AE monitoring of rotating machinery

Although AE, introduced by Kaiser in 1950 [5], has traditionally been used in monitoring defects in static or loaded structures or materials, it has found increasing application in non-destructive condition monitoring of rotating machinery of many different sizes, loads and operational speeds. However for large scale rotating machinery, AE is better established in slow-speed plant incorporating rolling-element bearings [6]. Reasons for this include the obvious unsuitability of conventional

The AE system

The AE signal measurement system employed for this study is shown in Fig. 2. This incorporates wideband piezoelectric Physical Acoustic Corp© WD sensors differentially connected to a 20/40/60 dB gain pre amplifier for measurement within the 100 kHz–1 MHz band. The separate pre amp incorporated a plug-in analogue high-pass filter (100 KHz–1.2 MHz) to suppress low frequency acoustic noise components and exhibited better temperature performance than could be achieved using an integral

Preliminary laboratory experiment

Prior to AE measurements upon the real turbine, the AE system was applied to the journal bearings of a laboratory test rig illustrated in Fig. 3. Although this test rig is much smaller than the rotors of a real turbine and could only rotate at a maximum speed of ∼2500 rev/min, it served as a very low acoustic noise environment in which the nature of AE response from continuous rubbing could be studied. Continuous shaft-seal rub conditions were simulated between the journal bearings using the

AE diagnosis of operational turbine unit

Further to the laboratory experiment, a number of operational turbine units were regularly monitored using the AE system over a period of 12 months. One of these AE tests was conducted on a 500 MW unit operating at full load. A schematic of this turbine comprising of one high-pressure (HP), one intermediate pressure (IP) and three low-pressure (LP) coupled rotor stages is shown in Fig. 7. In addition, the supporting journal bearings are labelled from bearing-1 at the HP end to bearing-10 at the

Conclusions

Ultimately, a new condition monitoring technique for such economically important industrial machinery is only accepted after extensive validation and testing within the field. As such, this paper constitutes a contribution to AE based condition monitoring of turbines as it presents validated AE diagnosis of the important condition of continuous rubbing in real operational plant. The AE diagnosis of the operational unit was supported by the preliminary experiments conducted on continuous rubbing

Acknowledgements

The authors wish to express their gratitude to Dr. Mark Smart, Dr. Rob Herbert and Mr. Chris White of Innogy PLC for sponsoring the project into AE monitoring of large scale turbines and allowing access to operational units.

References (9)

  • I. Sato

    Rotating machinery diagnosis with acoustic emission techniques

    Electr. Eng. Jpn.

    (1990)
  • C.B. Board, Stress wave analysis of turbine engine faults, Aerospace Conference, IEEE Proceedings (Cat. No.00TH8484) 6...
  • L.D. Hall, D. Mba, The detection of shaft-seal rubbing in large scale turbines using acoustic emission, 14th...
  • A. Muszynska

    Rotor–stationary element rub-related vibration phenomena in rotating machinery-literature survey

    Shock Vibrat. Dig.

    (1989)
There are more references available in the full text version of this article.

Cited by (37)

  • Analysis on accuracy improvement of rotor-stator rubbing localization based on acoustic emission beamforming method

    2014, Ultrasonics
    Citation Excerpt :

    Wang and Chu [13] extracted the rub fault feature through AE signal waveform analysis technology, and developed the AE rubbing positioning method based on wavelet decomposition. Hall and Mba [14,15] diagnosed the rotor–stator rubs of turbine units by using AE technology. Deng et al. [16] researched the noise reduction for rubbing AE signals in the rotating machinery.

  • Evaluation of the use of envelope analysis and DWT on AE signals generated from degrading shafts

    2012, Materials Science and Engineering: B
    Citation Excerpt :

    The acoustic emission (AE) technique in the condition monitoring of rotating machinery has been widely spread in recent years. This is particularly true for bearing fault diagnosis [1–7], seal rubbing [8] and crack growth [9–11]. The main drawback with the application of the AE technique is an attenuated signal, so the AE sensor has to be close to its source.

View all citing articles on Scopus
View full text