On the 16th and 17th September, ONYX InSight hosted its Annual 2019 North America Wind Turbine Technical Symposium in Denver, Colorado, the forum that brings together wind industry experts to share cutting edge developments in operations and maintenance (O&M).
The event gave wind farm owners, operators, engineers and technicians from companies including Duke, MidAmerican and PGE an in-depth insight into wind farm performance. The symposium enabled attendees to learn from the leading experience of speakers in O&M best practice to manage their assets as efficiently and cost effectively as possible.
The symposium covered the most common failures in wind turbines – and their root causes – from the perspectives of independent drivetrain, monitoring and lubrication experts. Core themes included:
- Challenges experienced in the delivery of re-powering
- Latest issues and mitigation in turbine failures and reliability
- Managing aging assets, including pitch, yaw, gearbox, and main bearings
- Due diligence and OPEX concerning operating asset acquisitions
- New developments in predictive maintenance
- The latest national labs researchon bearing failures
- How to achieve scheduled maintenance cost reductions
A poll aimed at assessing the views of leading figures in the sector was conducted among the attendees, on the subject of what the major challenges for the North American wind market are – and how to address these challenges.
The consensus of the gathered experts was that the most commonly observed failure modes on newer turbines were pitch system and blade failures. Almost two thirds of attendees said that they had seen pitch system failures. The results revealed that a wide range of common industry failure modes require further investigation, including pitch faults, data access to owners, tower vibration and blade failures.
Three-quarters of attendees agreed that identifying future risk was the main factor driving them to perform a Root Cause Analysis, followed by the need to support an insurance or warranty claim. Most respondents said that the bulk of their RCA activities were performed by independent engineers.
According to symposium attendees, the most important property to measure to determine the need for oil replacement was viscosity, followed by cleanliness. However, 65% of respondents remain to be convinced whether online oil sensing can deliver better value than standard lab analysis.
That said, online oil condition monitoring was selected by almost half of attendees as the technology which would best compliment vibration analysis – over traditional oil monitoring method such as particle counters and chip detectors. At the symposium, ONYX InSight Mechanical Engineer Jesse Graeter presented the results of our study on oil quality sensors.
Only 15% of respondents feel that oil particle counters are an indispensable part of system health monitoring, with a third believing they aren’t needed at all. Nearly half of attendees only considered oil particle counters cost-effective when used in large projects with less financial constraints.
Furthermore, over 60% of attendees have not acted to correct rotor mass imbalance, yet over a third have experienced rotor mass problems directly causing damage to components in the pitch system, drivetrain and blades.
FUTURE DEVELOPMENTS IN PREDICTIVE MAINTENANCE
60% of attendees said that in the last three years, they had more confidence to act on vibration monitoring recommendations, and almost as many agreed that operators could make significant cost savings using vibration monitoring, if implemented correctly.
Moreover, two-thirds agree that the area of planet bearing vibration monitoring is where the most improvement has been seen in the last five years.
Another key takeaway from the conference is that the engineers and technicians at the symposium would like to see predictive analytics technology applied to the issue of pitch bearings, with almost a third agreeing that this is the most vital problem. This supports the earlier agreement among the gathered experts that the pitch system is the most likely to fail.
Two-thirds of attendees are either already using AI and/or machine learning technology in their organisations or expect to start using it within the next three years. It is clear that there is increasing recognition by wind farm owners, operators and engineers that the future of O&M is increasingly data-driven.
Attendees found that the area where AI and machine learning can make the biggest impact is improved fault detection using SCADA data, and that the two best candidates for machine learning identification among wind turbine failures are drivetrain failure and temperature faults.