Wind turbine condition monitoring


By the end of 2013, global wind power installed capacity had reached more than 318 GW, according to the Global Wind Energy Council. Although this growth is very encouraging, one of the challenges still facing the wind industry is premature component failures that lead to increased operation and maintenance costs, which increase the cost of wind energy. To make wind power more competitive, the reliability and availability of turbines have to be improved, especially as turbines increase in size and are installed offshore.

Various historical statistics have shown that the largest downtime driver and the most costly component to maintain throughout a turbine's 20 year design life is the gearbox. It is an industry-wide challenge that needs to be addressed by all parties along the gearbox supply chain. To help the industry improve gearbox reliability and availability, the National Wind Technology Center (NWTC) at the US Department of Energy's National Renewable Energy Laboratory (NREL) started a project called the Gearbox Reliability Collaborative (GRC). The main objective of the GRC is to understand the possible causes of premature gearbox failures and make recommendations for reducing these failures. Condition monitoring (CM) is one area of research conducted under the GRC.

One CM project was the NREL wind turbine gearbox CM round-robin study. The study was based on data collected from a gearbox with various CM systems. The gearbox was tested at the NWTC's 2.5 MW dynamometer testing facility and was then installed in a turbine for field testing. The gearbox experienced two oil loss events and suffered damage to its internal components during the field test. It was then brought back to NWTC's 2.5 MW dynamometer and retested. NREL invited its academic and industrial partners from around the world to participate in the study. Seven university partners and nine private sector partners agreed to join this study. The vibration data collected from the gearbox, along with its configuration information, were first shared with these partners as a blind study. The partners did not have prior knowledge of the extent or location of the damage in the test gearbox. They were given two months to analyze the shared data using whichever algorithms they had or could develop. After their diagnostic results were submitted to NREL, the actual damage information on the test gearbox was disclosed to them so they could further refine their results. The entire project is described in an NREL report [[1]] that contains detailed analysis algorithms and diagnostic results from eight out of the sixteen partners.

To better archive the research findings from this round-robin study and bring more benefits to the wind turbine CM community, NREL proposed a special issue on wind turbine CM to the Wind Energy (WE) journal editorial board and was approved. Both NREL and the WE journal editorial board hoped that the work reported in this special issue would provide richer and deeper research and development elements than those reported in the technical report [[1]]. NREL invited the eight contributors to the technical report [[1]] to submit articles, and, after careful review, five were accepted for publication. In addition, two more invited papers are included to set the stage for the special issue and provide a broader perspective to the conclusion.

The seven papers contained in this special issue cover a breadth of topics, ranging from discussions on technical and commercial challenges that may impede the use of wind turbine CM, to attempts to address vibration analysis applied to wind turbine CM from both academic and industrial perspectives, to making use of wind turbine CM techniques to support prognostics health and management of wind farms.

Yang et al.[2] presented a detailed analysis of technical and commercial challenges with existing wind turbine CM technology. It discusses the requirements for monitoring newly developed wind turbines. Future work needed to develop a reliable and cost-effective wind turbine CM system is discussed on the basis of a review of nondestructive techniques, currently available commercial wind turbine CM systems and signal-processing techniques.

Siegel et al.[3] provided a detailed comparison among various vibration-based CM algorithms, including frequency domain and cepstrum analysis, time synchronous averaging narrowband and residual methods, bearing envelope analysis and spectral kurtosis-based methods. A tabular set of results is presented to highlight the ability of each algorithm to accurately detect bearing and gear wheel component health.

Luo et al.[4] introduced a synchronous sampling and order analysis approach to detect bearing and gear damage features under constantly varying operational conditions as typically seen in wind turbines. The synchronous sampling is carried out in digital domain for multiple revolutions and at a relatively high rate. For the situation where only time history of shaft rotational speed is given, a procedure to synthesize a keyphasor is also presented.

Sawalhi et al.[5] adapted techniques developed for other applications to solve wind turbine gearbox diagnostics problems. Gear health diagnostics is made through spectrum and cepstrum analyses based on synchronous averages of raw vibration data at appropriate shafts and bearing condition diagnostics is made through whitening, filtering and enveloping analyses based on data obtained by subtracting all synchronously averaged signals from the raw. The analysis also involved obtaining synchronous averages for the individual planetary gears and for the sun gear by using software patented by the Australian Defence Science and Technology Organisation.

Sheldon et al.[6] applied vibration diagnostic algorithms developed and matured by the authors for Department of Defense applications to wind turbine gearbox CM. The results of these automated algorithms were also corroborated with visual spectral analysis. The algorithms and the results, along with some recommendations to conduct future tests and analysis, are presented.

Wang et al.[7] presented a fault component separation method that decomposed one channel of vibration measurements into a series of intrinsic mode functions by means of ensemble empirical mode decomposition and then performed an independent component analysis on the intrinsic mode functions to separate bearing defect-related signals from gear meshing signals. The effectiveness of the developed method is evaluated numerically and experimentally.

Haddad et al.[8] introduced a maintenance-options methodology to quantify the value of maintenance decisions made after a prognostic indication and an analysis of wind turbine maintenance data. The value of waiting until after a prognostic indication is determined by using a model that quantifies the benefit resulting from a prognostics and health management implementation. This allows the decision maker to delay maintenance actions. The presented work naturally places the focus of this special issue on wind turbine CM in a bigger framework of wind turbine/plant prognostics and health management, linking data with decisions.

Although the focus of this special issue is on vibration analysis, it is only one CM technique. Most of the papers from the NREL wind turbine gearbox CM round-robin study are based on vibration analysis. However, this does not imply that vibration analysis is a universal solution for all wind turbine CM needs. Other techniques, such as oil debris analysis, are also beneficial and in order to cover more turbine failure modes, a comprehensive approach must be taken.

The GRC and the gearbox CM round-robin projects were funded by the US Department of Energy under contract no. DE-AC02-05CH11231. I had the privilege to lead and manage the NREL wind turbine gearbox CM round-robin project, and I deeply appreciate the voluntary support from all 16 partners and their contributions. I would also like to give special thanks to the WE journal editorial board for agreeing to host this special issue. It is a great honor for me to serve as the guest editor for this special issue. I especially thank Paul Veers at NREL for his help and guidance through the entire process, and I greatly appreciate the support from Janis Hunt with logistics. I would also like to thank the authors for their contributions to this special issue and express sincere appreciation to the reviewers for their time and expertise in providing valuable comments to improve the quality of the papers contained herein.