Gearbox Bearings Research Articles

Condition Monitoring of Wind Turbine Gearbox Bearing Based on Deep Learning Model

Wind turbines condition monitoring and fault warning have important practical value for wind farms to reduce maintenance costs and improve operation levels. Due to the increase in the number of wind farms and turbines, the amount of data of wind turbines have increased dramatically. This problem has caused a need for efficiency and accuracy in monitoring the operating condition of the turbine. In this paper, the idea of deep learning is introduced into wind turbine condition monitoring. After selecting the variables by the method of the adaptive elastic network, the convolutional neural network (CNN) and the long and short term memory network (LSTM) are combined to establish the logical relationship between observed variables. Based on training data and hardware facilities, the method is used to process the temperature data of gearbox bearing. The purpose of artificial intelligence monitoring and over-temperature fault warning of the high-speed side of bearing is realized efficiently and conveniently. The example analysis experiments verify the high practicability and generalization of the proposed method.

The Effect of Lubricant Composition on White Etching Crack Failures

White etching cracks (WECs) are the dominant mode of failure for wind turbine gearbox bearings. These failures are characterized by subsurface initiation and local region of microstructural alterations adjacent to the crack faces. The definitive cause of WECs within the field is unknown, because of this laboratory replication has proved difficult. At a benchtop scale, specific lubricant formulations referred to as “bad reference oils” (BROs) are often employed to aid in the formation of WECs; however, exactly how these lubricants induce WECs is unknown. The present work intends to elucidate how these lubricants facilitate the formation of WECs by systematically varying the additives which are found in BROs and studying the effect that these additive combinations have on time until failure, as well as tribofilm development. It was found that the lubricant containing Zinc dialkyldithiophosphate alone led to the formation of WECs sooner than any lubricant studied. It was also documented that a lubricants frictional characteristics play a more dominant role than the tribofilm characteristics.

Prediction of Remaining Useful Life of Wind Turbine Bearings under Non-Stationary Operating Conditions

Wind-powered electricity generation has grown significantly over the past decade. While there are many components that might impact their useful life, the gearbox and generator bearings are among the most fragile components in wind turbines. Therefore, the prediction of remaining useful life (RUL) of faulty or damaged wind turbine bearings will provide useful support for reliability evaluation and advanced maintenance of wind turbines. This paper proposes a data-driven method combining the interval whitenization method with a Gaussian process (GP) algorithm in order to predict the RUL of wind turbine generator bearings. Firstly, a wavelet packet transform is used to eliminate noise in the vibration signals and extract the characteristic fault signals. A comprehensive analysis of the real degradation process is used to determine the indicators of degradation. The interval whitenization method is proposed to reduce the interference of non-stationary operating conditions to improve the quality of health indicators. Finally, the GP method is utilized to construct the model which reflects the relationship between the RUL and health indicators. The method is assessed using actual vibration datasets from two wind turbines. The prediction results demonstrate that the proposed method can reduce the effect of non-stationary operating conditions. In addition, compared with the support vector regression (SVR) method and artificial neural network (ANN), the prediction accuracy of the proposed method has an improvement of more than 65.8%. The prediction results verify the effectiveness and superiority of the proposed method.

The origins of white etching cracks and their significance to rolling bearing failures

Presence of white etching cracks has been widely associated with early failures of rolling bearings in a number of applications, with wind turbine gearbox bearings being the most frequently cited and practically significant example. Despite the recent research efforts, there is yet no universal agreement on the mechanisms of formation of these cracks and little direct evidence of their significance to bearing reliability. In an attempt to address this, this paper proposes a new theory on the origins and significance of white etching cracks. The paper provides systematic experimental evidence in support of this theory through rolling contact fatigue tests performed with AISI 52100 bearing steel specimens on a triple-disc machine over a wide range of contact conditions. The test results show that white etching cracks can be formed with base oils as well as commercially formulated transmission and engine oils. WECs were generated under slide-roll-ratios ranging from 0.05 to 0.3, under positive and negative sliding, different contact pressures and specific film thicknesses ranging from 0.1 to 0.7. No white etching areas were ever observed without the associate crack being present, and it was also shown that white etching areas themselves can be produced in a pure rubbing contact of bearing steels under both lubricated and unlubricated conditions. These results provide direct evidence that the steel transformations that exhibit themselves as white etching areas are formed through rubbing of the existing crack faces, and that the chemical composition of the lubricant and the magnitude and direction of sliding are not the primary driver of WEC formation, in contrast to literature. Instead, the results presented here show that WECs are formed through the action of a specific stress history in time via the following mechanism: (i) Short-lived high contact stresses, which can be caused by a number of factors, act in the initial stages of the component life to initiate early fatigue cracks, (ii) These high stresses cease and the contact subsequently operates under a relatively moderate nominal stress so that the early-formed cracks propagate relatively slowly and their faces rub and beat together for a prolonged period, (iii) White etching areas are formed around the existing cracks through this crack-face rubbing, (iv) Eventual contact failure occurs through pitting which can be caused by a white etching or a non-white etching crack depending on its location and propagation rate. Standard contact fatigue life models are unable to account for the adverse effects of this specific stress sequence owing to their use of linear damage accumulation rules, which leads to such failures being termed as ‘premature’. The proposed theory asserts that the white etching cracks are not the cause of ‘premature’ bearing failures, but that both the WECs and the earlier-than-predicted bearing failures are caused simultaneously by a third factor: a specific time history of applied contact stress.

Microscopic investigation of subsurface initiated damage of wind turbine gearbox bearings

Wind turbine gearbox bearings experience premature failures by White Structure Flaking (WSF), often occurs much earlier than their designed life of 20 to 25 years. This results in increased operational and maintenance costs due to unplanned maintenance and early replacement. The main causes and damage initiation mechanism of this premature failure are not fully understood; despite extensive research and investigation in recent years. In this paper, two planetary bearings from a failed gearbox of a multi-megawatt wind turbine are destructively investigated to characterize the subsurface microstructural damage and to understand damage initiation mechanism leading to surface WSF. The results show that the non-metallic inclusions are not the only initiator of subsurface damage. The microcracks are also initiated in the subsurface to form macrocracks which then propagate or connect to other macrocracks to reach the rolling contact surface causing WSF. The characterization of different forms of subsurface microstructural damage shows a close correlation of the maximum shear stress with the damage initiation. Butterfly wings are found to initiate from the compound type of non-metallic inclusions with low aspect ratio and to be associated with inclusion internal cracking in a direction approximately parallel to the axis of the maximum wing length.

Condition Monitoring and Fault Diagnosis of Wind Turbines Gearbox Bearing Temperature Based on Kolmogorov-Smirnov Test and Convolutional Neural Network Model

Wind turbine condition-monitoring and fault diagnosis have important practical value for wind farms to reduce maintenance cost and improve operating level. Due to the special distribution law of the operating parameters of similar turbines, this paper compares the instantaneous operation parameters of four 1.5 MW turbines with strong correlation of a wind farm. The temperature-power distribution of the gearbox bearings is analyzed to find out the main trend of the turbines and the deviations of individual turbine parameters. At the same time, for the huge amount of data caused by the increase of turbines number and monitoring parameters, this paper uses the huge neural network and multi-hidden layer of a convolutional neural network to model historical data. Finally, the rapid warning and judgment of gearbox bearing over-temperature faults proves that the monitoring method is of great significance for large-scale wind farms.

Reliability assessment of wind turbine bearing based on the degradation-Hidden-Markov model

Wind power develops very quickly in last decade to overcome the energy crisis and environment crisis. Mechanical components of wind turbines usually have characteristic with performance degradation that results in the declining reliability over time. Generally, the reliability data of equipment come from statistical analysis based on extensive experiments and operations. However, wind turbines, as expensive large-scale equipment with long lifetime, face with the dilemma of lacking enough statistical data, and leads to insufficiency reliability data for field operations and thus results in frequent wind turbine faults. A new reliability assessment method based on Hidden-Markov model considering performance degradation, called degradation-Hidden-Markov model, is proposed in this paper. The performance degradation rule of wind turbine component is derived using the monitoring data of performance parameters. Hidden-Markov model is improved by the performance degradation rule of the component to create a new time-correlated state transition probability matrix with degradation feature. The reliability curve is obtained using the state probabilities of the degradation-Hidden-Markov model. Thus, the presented method realizes the reliability assessment of component based on small sample data of wind turbine. Finally, the reliability assessment of a gearbox bearing of a 1.5 MW wind turbine by the degradation-Hidden-Markov model proves its validity.

Threshold Maps for Inclusion-Initiated Micro-Cracks and White Etching Areas in Bearing Steel: The Role of Impact Loading and Surface Sliding

Wind turbine gearbox (WTG) bearings can fail prematurely, significantly affecting wind turbine operational availability and the cost of energy production. The current most commonly accepted theory of failure mechanism is that the bearing subsurface is weakened by white etching crack (WEC) networks that eventually lead to the flaking away of material from the bearing surface. Subsurface damage due to rolling contact fatigue (RCF) is thought to be the main cause of premature failure, resulting from the initiation of micro-cracks, often at non-metallic inclusions or other material defects, which then propagate to the bearing surface. This study proposes a hypothesis that impact loading together with high levels of surface traction and contact pressure are important factors contributing to the initiation of micro-cracks and white etching areas (WEAs) at non-metallic inclusions which may lead to the formation of WEC networks. Both repeated impact and twin-disc RCF tests were designed to investigate inclusion-initiated micro-cracks and WEAs at subsurface in order to test this hypothesis. This led to the recreation of inclusion-initiated micro-cracks and WEAs in tested specimens, similar to the subsurface damage observed at inclusions in failed WTG bearing raceways. Tests were carried out to determine threshold levels of contact pressure, surface traction, and impact loading required for the formation of inclusion-initiated micro-cracks and WEAs.

Mapping of axial plastic zone for roller bearing overloads using neutron transmission imaging

Premature failure of wind turbine gearbox bearings is an ongoing concern for industry, with sudden overload events potentially contributing towards raceway damage, significantly hindering performance. Subsurface stresses generated along a line contact cause material yielding, and a probable crack initiation site. Currently, the ability to study subsurface plastic zone evolution using non-destructive techniques is limited. Neutron Bragg edge imaging is a novel technique, allowing for two-dimensional mapping of the Bragg edge broadening parameter, indicative of bulk plastic deformation. An experiment on the ENGIN-X strain scanning instrument, at the ISIS neutron source, UK, was setup for Bragg edge transmission imaging, to measure the effect of in situ loading on the raceway of a bearing, scaled-down from a traditional wind turbine gearbox bearing. Results demonstrate a strong correlation between load and the Bragg edge width, and allow for future experimental development in studying, not only the effect of overloads on fatigue life, but also the use of neutron imaging for evaluating plastic deformation in engineering components.

Robustness Test for Wind Turbine Gearbox Bearings

This paper introduces an innovative approach for the creation of the robustness test against specific failures of the planetary and HSS bearings (e.g. classical fatigue, smearing, micro-pitting, and lip fractures) in the wind turbine gearboxes. The introduced Bearing Robustness Test (BRT) considers the system-dependent characteristics (e.g. drive train design, interaction between components, assembly process, lubricant aging) and real dynamic load conditions, frequencies and sequence. The creation of the BRT is based on field and simulation data. The core element is the simulative approach for the determination of the relation between external wind and grid loads on the one side and local loads of the bearing on the other side. BRT aims the mapping of the most critical, but real, field load situations in the bearing test rig. By means of the BRT it is possible to evaluate the robustness of bearing against specific field conditions in the early stage of the product cycle and consequently to enhance the quality and to reduce the failure rate of the bearing.

Adaptive deep feature learning network with Nesterov momentum and its application to rotating machinery fault diagnosis

Abstract The effective fault diagnosis of rotating machinery is critical to ensure the continuous operation of equipment and is more economical than scheduled maintenance. Traditional signal processing-based and artificial intelligence-based methods, such as wavelet packet transform and support vector machine, have been proved effective in fault diagnosis of rotating machinery, which prevents unexpected machine breakdowns due to the failure of significant components. However, these methods have several disadvantages that make them unable to automatically and effectively extract valid fault features for the effective fault diagnosis of rotating machinery. A novel adaptive learning rate deep belief network combined with Nesterov momentum is developed in this study for rotating machinery fault diagnosis. Nesterov momentum is adopted to replace traditional momentum to enable declining in advance and to improve training performance. Then, an individual adaptive learning rate method is used to select a suitable step length for accelerating descent. To confirm the utility of the proposed deep learning network architecture, two examinations are implemented on datasets from gearbox and locomotive bearing test rigs. Results indicate that the method achieves impressive performance in fault pattern recognition. Comparisons with existing methods are also conducted to demonstrate that the proposed method is more accurate and robust.

Modelling of thermal stresses in bearing steel structure generated by electrical current impulses

This work is the study of one particular candidate for white etching crack (WEC) initiation mechanism in wind turbine gearbox bearings: discharge current impulses flowing through bearing steel with associated thermal stresses and material fatigue. Using data/results from previously published works, the authors develop a series of models that are utilized to simulate these processes under various conditions/local microstructure configurations, as well as to verify the results of the previous numerical studies. Presented models show that the resulting stresses are several orders of magnitude below the fatigue limit/yield strength for the parameters used herein. Results and analysis of models provided by Scepanskis, M. et al. also indicate that certain effects predicted in their previous work resulted from a physically unfounded assumption about material thermodynamic properties and numerical model implementation issues.

Initiation of sub-surface micro-cracks and white etching areas from debonding at non-metallic inclusions in wind turbine gearbox bearing

In this study, a failed planetary bearing from a wind turbine gearbox was destructively examined to investigate the initiation of micro-cracks and butterflies at non-metallic inclusions, and the effect of debonding between these inclusions and the steel matrix. The butterflies were scanned using Atomic Force Microscopy (AFM) to show the topography that could not be assessed by using other microscopy techniques. Nano-indentation test was conducted across a butterfly wing and a non-metallic inclusion to measure the hardness at the interface with the steel matrix. It was found that the White Etching Areas (WEA) in the region of the butterfly wing was a damaged material that showed tearing at the debonding gap between the inclusion and steel matrix. This study highlighted the effect of debonding on the initiation of micro-cracks, WEA and inclusion cracking. A direct relationship was found between the size of inclusions and the total length of inclusions and micro-cracks or butterfly wings. The depth of the observed sub-surface damage was correlated with the sub-surface stress distribution and these results suggested that surface traction could be an important contributing factor to the subsurface damage initiation.

Adaptive iterative generalized demodulation for nonstationary complex signal analysis: Principle and application in rotating machinery fault diagnosis

Abstract Effective identification of signal frequency contents and their time variability is a key to success in rotating machinery fault diagnosis under nonstationary conditions. Fine time-frequency resolution and free from both inner and outer interferences are necessary to achieve this purpose. Iterative generalized demodulation (IGD) can separate a nonstationary multi-component signal into constituent mono-components, and derive quality time-frequency representation based on Hilbert spectrum of each mono-component, thus offering an effective approach to nonstationary complex signal analysis. Nevertheless, it requires prior expertise knowledge about signal structure to ascertain true constituent components and construct proper demodulation phase functions. This leads to a major difficulty in real signal analysis tasks, where expertise knowledge is usually unavailable and signal time-frequency structure identification via visual observation is susceptible to noise interferences. To address this issue, the capability of surrogate test technique in recognizing true signal components is exploited, and thereby an adaptive iterative generalized demodulation (AIGD) is proposed. Proper demodulation phase functions corresponding to constituent components are derived accordingly, and Hilbert spectra of all constituent mono-components are superposed to generate time-frequency representation (TFR). This method features good merits, such as fine time-frequency resolution and free of both inner and outer interferences. Additionally, it neither relies on prior expertise knowledge about signals, nor needs to construct any basis function. It is highly adaptive to reveal the frequency contents and track their time variability of a signal, and provides an effective approach to nonstationary complex multi-component signal analysis. It is illustrated and validated via numerical simulations, lab experiments of a planetary gearbox and rolling bearings, and on-site tests of a hydraulic turbine in a hydraulic power plant.

A leakage-free resonance sparse decomposition technique for bearing fault detection in gearboxes

Most of rotating machinery deficiencies are related to defects in rolling element bearings. Reliable bearing fault detection still remains a challenging task, especially for bearings in gearboxes as bearing-defect-related features are nonstationary and modulated by gear mesh vibration. A new leakage-free resonance sparse decomposition (LRSD) technique is proposed in this paper for early bearing fault detection of gearboxes. In the proposed LRSD technique, a leakage-free filter is suggested to remove strong gear mesh and shaft running signatures. A kurtosis and cosine distance measure is suggested to select appropriate redundancy r and quality factor Q. The signal residual is processed by signal sparse decomposition for highpass and lowpass resonance analysis to extract representative features for bearing fault detection. The effectiveness of the proposed technique is verified by a succession of experimental tests corresponding to different gearbox and bearing conditions.

Toward Investigation of External Oil Flow From a Journal Bearing in an Epicyclic Gearbox

High loads and bearing life requirements make journal bearings the preferred choice for use in high-power, planetary gearboxes in jet engines. With the planet gears rotating about their own axis and orbiting around the sun gear, centrifugal forces generated by both motions interact with each other and create complex kinematic conditions. This paper presents a literature and state-of-the-art knowledge review to identify existing work performed on cases similar to external journal bearing oil flow. In order to numerically investigate external journal bearing oil flow, an approach to decompose an actual journal bearing into simplified models is proposed. Preliminary modeling considerations are discussed. The findings and conclusions are used to create a three-dimensional (3D), two-component computational fluid dynamics (CFD) sector model with rotationally periodic boundaries of the most simplistic approximation of an actual journal bearing: a nonorbiting representation, rotating about its own axis, with a circumferentially constant, i.e., concentric, lubricating gap. In order to track the phase interface between the oil and the air, the volume of fluid (VoF) method is used. External journal bearing oil flow is simulated with a number of different mesh densities. Two different operating temperatures, representing low and high viscosity oil, are used to assess the effect on the external flow field behavior. In order to achieve the future objective of creating a design tool for routine use, key areas are identified in which further progress is required.

Review of the damage mechanism in wind turbine gearbox bearings under rolling contact fatigue

Wind turbine gearbox bearings fail with the service life is much shorter than the designed life. Gearbox bearings are subjected to rolling contact fatigue (RCF) and they are observed to fail due to axial cracking, surface flaking, and the formation of white etching areas (WEAs). The current study reviewed these three typical failure modes. The underlying dominant mechanisms were discussed with emphasis on the formation mechanism of WEAs. Although numerous studies have been carried out, the formation of WEAs remains unclear. The prevailing mechanism of the rubbing of crack faces that generates WEAs was questioned by the authors. WEAs were compared with adiabatic shear bands (ASBs) generated in the high strain rate deformation in terms of microstructural compositions, grain refinement, and formation mechanism. Results indicate that a number of similarities exist between them. However, substantial evidence is required to verify whether or not WEAs and ASBs are the same matters.

Deep learning for automated drivetrain fault detection

AbstractA novel data‐driven deep‐learning system for large‐scale wind turbine drivetrain monitoring applications is presented. It uses convolutional neural network processing on complex vibration signal inputs. The system is demonstrated to learn successfully from the actions of human diagnostic experts and provide early and robust fault detection on both rotor bearing, planetary and helical stage gear box bearings from analysis of multisensor vibration patterns using only a high‐level feature selection. On the basis of data from 251 actual wind turbine bearing failures, we are able to accurately quantify the fleet‐wide diagnostic model performance. The analysis also explores the time dependence of the diagnostic performance, providing a detailed view of the timeliness and accuracy of the diagnostic outputs across the different architectures. Deep architectures are shown to outperform the human analyst as well as shallow‐learning architectures, and the results demonstrate that when applied in a large‐scale monitoring system, machine intelligence is now able to handle some of the most challenging diagnostic tasks related to wind turbines.

Failure analysis on abnormal wear of roller bearings in gearbox for wind turbine

Abstract The booming wind power industry as one of the predominant approaches of renewable energy resources is not independent of components reliability within the wind turbines, especially the rotating ones including blades, motors, gearboxes etc. Hence, case studies on failed parts of such components play an important role for failure prevention and deserves to be publicly reported for experience sharing. In this paper, a failure case concerning severe abnormal wear of the roller bearing's inner ring in the gearbox of one 1.5 MW wind turbine in China was reported. Since such roller bearings were imported from a foreign company and operated by a Chinese company, not only economic losses but also international commercial dispute would be induced if this case would not be immediately and properly solved. To this end, by means of comprehensive and systematic investigation into the base materials, process media, surface morphologies, micro-area compositions and even service environments, root causes of this failure were determined, detailed mechanisms were discussed, and pertinent countermeasures were proposed. Achievement of this paper would provide the solid evidence to distinguish the responsibilities for the failure, and would also help to prevent such failures of roller bearings with similar design in wind turbines.

Investigation of Microstructural Alterations in Low- and High-Speed Intermediate-Stage Wind Turbine Gearbox Bearings

Wind turbine drive train system failures have been the major source of down time and costly repairs. One of the main failure modes associated with gearbox failures is the formation of white etching cracks (WECs) in bearings. In this paper, low-speed intermediate shaft (LSIS) and high-speed intermediate shaft (HSIS) bearings are collected from failed gearboxes to examine microstructural alterations and evaluate damage. Butterfly wing formations are detected in LSIS bearings around nonmetallic inclusions at depths of up to about 260 µm from the contact surface and the measured span length ranging from 5 to 70 µm. Circumferential cracks oriented parallel to the contact surface are detected at depths greater than 300 μm. Irregular white etched cracks are revealed in HSIS bearings with random crack orientations. The crack depth is measured to be 7000 μm in the cross section perpendicular to the bearing axis. The bearing characterization revealed that microstructural alterations can occur with and without the formation of cracks. Evidence found supports the existing hypotheses that (1) white etch areas precedes crack formations and (2) cracks are prerequistes for white etch formation. The microstructural alterations detected in LSIS and HSIS bearings are observed to have different crack morphology and could be initiated by different mechanisms. The results presented in this paper can be used to develop materials-based prognostic life prediction models. This can be achieved by linking WECs and butterfly wings qualitative observations to turbine real operating conditions, bearing design, and manufacturing process.