By Noah Myrent, global head of monitoring at ONYX InSight
Renewable energy is on the rise, now more than ever. Thanks to advances in digitization and data analytics, wind turbines can be relied on as a robust source of energy. While in the past, wind turbines were largely unintelligent monoliths monitored periodically via on-site technician checks, wind farms are now creating vast oceans of data filled with potential.
The two traditional operations and maintenance (O&M) models for wind farms have been reactive and preventative maintenance. Reactive maintenance involves a “run to failure” model, replacing components when they are degraded to the point of causing problems with the day-to-day operations of the wind turbines affected.
Preventative maintenance relies on regularly scheduled replacements and repairs. It can be hard to deliver significant savings with this model because of the recurring, scheduled component costs in addition to the resources used for conducting the routine up-tower work. Laboratory results from oil analysis conducted in this manner, for example, can take a significant amount of time from the initial sample collection to having results in-hand, negating many potential benefits of monitoring in the first place. This cumbersome process means that O&M teams do not have the opportunity to proactively manage the performance of their wind turbines and thus minimize potential damage to each turbine’s components.
Up to 90% of the cost of offshore wind O&M is due to accessibility, reflecting the logistical challenges of transporting engineers and technicians to offshore sites. Through digitalization and a predictive maintenance regime, these costs can be significantly decreased by making turbine health information available remotely and ensuring that visits by engineers and technicians can be utilized more effectively to replace or repair multiple components at the same time — and reduce the frequency of these visits accordingly.
As it grows, the wind industry must collectively ensure that the levelized cost of energy (LCOE) stays competitive with alternative means of energy generation. This is especially important in light of the move to a merchant market in recent years, shifting financial risk from governments to energy producers. Despite the subsidies enjoyed by conventional energy around the world, well executed digitalization of wind energy assets can bring the LCOE below subsidized levels.
O&M teams have an opportunity to lead offshore wind forward and adopt the latest technologies to help maximize efficiencies across the assets they manage. O&M costs make up 60% of all operational expenditure (OPEX) on wind farms — and nearly two-thirds of this expenditure is unplanned. By combining real world engineering expertise with the latest advancements in artificial intelligence (AI) and machine learning (ML), firms can move to a predictive maintenance model and reduce O&M costs by up to 30%.
Shifting trends in software adoption for wind energy O&M
The wind energy industry has often been reluctant to adopt digital technology advances. But in a post-subsidy market, the need to improve efficiencies is far greater than it was previously. It has become increasingly common and affordable for wind turbines to be fitted with digital sensors which log invaluable data regarding the performance of the turbines. At this stage, wind farm owner-operators are failing to unlock the full potential of this data.
This is changing, however. Using the vast quantities of data generated by pre-installed sensors for detecting and measuring health indicators such as drive train vibration, O&M experts can work with data analysis specialists to train algorithms to detect issues in wind turbines before they emerge. When combined with the specialist knowledge of O&M professionals, these algorithms can be trained to diagnose problems with accuracy rates close to 99%. It is then possible to have an entire site of digitalized wind turbines, connected to the Internet of Things (IoT) and to each other, delivering performance and health data to remotely situated O&M teams.
Using AI means that the data analysis can be automated, so that large data sets, beyond the capacity of an engineering team to analyze in a timely fashion, can be scrutinized for trends that indicate health changes. Engineering expertise is vital to the process, as it ensures that the algorithms are trained properly to distinguish which trends were important, and what a particular trend signature means for a wind turbine’s performance and reliability.
Cloud computing reduces the cost barriers to this cutting-edge analysis even further, by enabling wind farm owner-operators to process large amounts of data and access it easily, providing a scale of computing resource that was previously unattainable within the wind industry. This allows in-house teams to run predictive maintenance at a lower cost, but still utilize the support of engineering expertise as needed. Across the 9 GW of wind capacity that ONYX InSight monitors, 20% of those assets have already adopting this DIY approach. The universal accessibility of the cloud means that the physical location of wind farms, O&M teams and expert engineers is a non-issue in relation to the monitoring and analysis of wind turbine performance.
The trend toward digitalization is certainly gaining momentum. In September 2019, ONYX hosted its annual North American Technical Symposium, gathering together leading experts in the wind O&M field, as well as owner operators in the industry. Among the attendees, 51% confirmed that their organization was aligned to adopt AI and ML to monitor their wind turbines in the near term, while 22% have already employed the technology in their O&M practices.
The risks of adopting AI and ML without the right digital tools
To add to the challenge, the technical element requires extra care. There is a tendency among some to jump on the AI bandwagon because AI and ML are exciting new technologies. This is made more difficult by views of AI and ML as cure-alls, capable of solving every problem and optimizing for every situation. The reality is that AI and ML are most effective when directed at specific, targeted problems, and the technologies need to be “trained” on data sets by experienced, expert wind industry O&M engineers before they can unleash their full power. Wind farm owner-operators must ensure that they can leverage this real-world expertise — and combine it with the right technology — before they make significant forays into the digital world.
When handled correctly, the wind energy industry stands to gain enormously from using AI and ML for predictive maintenance to create opportunities for savings. At the same time, renewable energy is set to become a significantly larger component of the global energy mix, and, in regions such as the UK and northern Europe, wind energy will become the largest energy source.
Consequently, wind energy infrastructure is becoming a critical part of the national energy grid. There have long been concerns about cyber-attacks on nuclear power plants in particular, but as wind energy becomes more prominent, it will present a more attractive target to rogue states, terror organizations and even cyber criminals. Increasing the connectivity of wind farms is a necessary step toward cementing wind power’s eventual status as one of the primary forms of energy generation in the 21st century, but it also makes wind infrastructure more vulnerable.
It is necessary to learn and grow from past outages such as the 2016 incident in Ukraine, caused by hackers targeting the national power grid. Cybercrime is no new concept, as hackers have been able to hold energy firms to ransom with viruses. If the wind farm has its servers infected with malicious data and is shut down, it could cost wind farm owners millions in lost revenue costs until the problem is resolved.
Cyber security may seem daunting, but fortunately, there are simple and relatively cost-effective steps that can be taken to reduce the risk of incursions. These include installing in-line firewalls between turbines, investing in operations technology (OT) security software, and ensuring data is encrypted to a high standard. These measures can have the added benefit of lowering insurance premiums as well.
It is clear that wind farm owners can’t afford to neglect AI and ML in their O&M paradigms. By staying ahead of emerging trends in wind O&M technology, companies can gain a strong competitive edge, cut their operational costs and reduce the LCOE of their assets. Forward-thinking wind farm owners can protect themselves better against the risks of a post-subsidy market, extend life of their assets, and properly protect against emerging asset risk — if they are prepared to intricately adopt advanced computing technology across their portfolios.
Noah Myrent is the Head of Global Monitoring for ONYX InSight, where he manages a global team that applies advanced signal processing, condition monitoring data analytics and best practices in predictive maintenance in order to routinely monitor thousands of wind turbines every day. From 2013-2015, Noah led the wind energy research team at Vanderbilt University’s Laboratory for Systems Integrity and Reliability. Noah’s background includes structural dynamics, condition monitoring, sensitivity analysis and rotor blade fault detection. He received his Bachelor’s in acoustical engineering from Purdue University in 2008 and earned his Master’s in mechanical engineering in 2013, also from Purdue.