Asset modelling challenges in the wind energy sector

McMillan, David and Dinwoodie, Iain Allan and Wilson, Graeme and May, Allan and Hawker, Graeme (2014) Asset modelling challenges in the wind energy sector. In: CIGRE Session 2014, 2014-08-24 - 2014-08-30.

Full text not available in this repository.Request a copy

Abstract

The research area of asset management and cost reduction for wind energy is increasing in importance as over 100,000 turbine assets are currently installed worldwide. Such large numbers of turbine assets, with a growing offshore component, drive demand for research into improved decision support for wind farm planners and operators - as the operational risk for the assets is removed from OEMs (while under warranty) and is transferred to the power utility owners (post-warranty). The current and future demand for research in this field has thus been driven by the needs of owner/operators and central government policy formultation. The huge potential energy available from offshore wind in particular, is well known. The most crucial aspect of future deployment levels is the economic recovery of this potential energy. The long-term viability of the offshore wind power as a large scale energy solution hinges on reduction of the cost of energy, as a purely cost-based energy market would currently produce a gas- and coal- based system. The areas of research identified by the authors to aid this transition can be summarised into 3 key themes: 1. Asset and meteo-ocean modelling to support offshore operations and maintenance The first modeling innovation is in the advanced planning of logistics for offshore operations and maintenance. Research produced by Ernst and Young in 2009 has shown that far from reducing O&M costs through learning processes, the real and expected O&M spend for offshore wind has increased year on year: this is backed up by more recent anecdotal evidence. Wind farm owners - often utilities with no experience of offshore operations – whish to improve their knowledge of how to plan for the big risks such as vessel chartering for maintenance, spares provision and location/type of O&M service base. The unique blend of modeling methodologies utilized by the authors build on existing research on Markov chain reliability models applied to wind power assets, and with an increased focus on accurate statistical characterization of weather conditions (particularly working constraints such as wave heights and wind speeds). This advanced approach enables key questions regarding site accessibility in winter, jack up vessel strategy, and life cycle cost issues to be addressed in a way more comprehensive and innovative than any other approach in the literature. This work has far-reaching implications for utilities and offshore wind farm owners, representing a step change in modeling accuracy compared to existing industry standard tools. The future for this work stream will involve this work converging with research activity on condition based maintenance and decision support. 2. Analysis of wind farm maintenance data for asset life cycle optimization The second area of innovative research is analysis of wind turbine SAP and other work order and maintenance data to investigate possible links between failure events and weather conditions. Such analysis will also aid wind asset engineers in their anticipation of failure modes, traceability of faults via improved failure reporting, and calculation of risk exposure to specific failure mechanisms. Current analysis methods are primitive and ripe for improvement by porting data analysis methods from other sectors. Current research by the authors focuses on developing algorithms to mine these rich data sets and establish new understanding of failure patterns with respect to environmental conditions. Industrial interest in this area of work is growing as more assets exit 3-5 year warranty guarantees and the risk associated with serial defect, wear issues, and end of life ramping of failure rates becomes the responsibility of the wind farm owner/operator. Increased knowledge of asset deterioration and failure behaviour can also drive feedback to manufacturers, as well as providing a bargaining chip in 3rd party O&M contract negotiations. Academic novelty is also apparent as most research to date has focused on analysis of turbine SCADA/ condition monitoring data streams: the potential of SAP-type systems to provide a rich source of data to calibrate failure models and understand the impact of environmental conditions on reliability has not been adequately explored. Although at a less advanced stage of maturity compared with the offshore logistics work, this work stream again has the potential to be fused with parallel work on condition monitoring and asset management. 3. Role of condition monitoring in decreasing OPEX As wind turbines increase in size and move offshore, operations and maintenance procedures need to be optimised to increase reliability, safety and maximise cost effectiveness. The true economic value of a condition based maintenance strategy for offshore wind farms has yet to be calculated. Many existing studies have focused on either the functionality of the turbine, or on the structural integrity, but not both. This theme involves closing the research gap which exists between reliability modelling of wind turbine functionality and modelling of structural integrity. The aim is to produce a comprehensive approach which will facilitate to quantify the overall economic benefits of condition monitoring. With a case study, the industrial application of the approach will be shown and the benefits will be demonstrated. This modelling allows operators to examine a condition based maintenance approach that theoretically allows reduced costs over both preventive and corrective maintenance strategies. There have been several studies into the possible benefits and cost advantages of using a condition based maintenance strategy. However, few have examined the implications of false alarms. Investigating false alarms or ignoring false positives in a remote offshore environment will incur costs that may alter the cost benefit of condition monitoring systems. Probabilistic models are used in the paper to determine the possible benefits of using condition monitoring systems and the detect that false positives and negatives have on the reliability of the system. The methods used include Markov chains, Monte-Carlo simulations and time-series modelling.

ORCID iDs

McMillan, David ORCID logoORCID: https://orcid.org/0000-0003-3030-4702, Dinwoodie, Iain Allan ORCID logoORCID: https://orcid.org/0000-0001-9090-1256, Wilson, Graeme, May, Allan ORCID logoORCID: https://orcid.org/0000-0001-5900-8179 and Hawker, Graeme ORCID logoORCID: https://orcid.org/0000-0003-2876-4371;