Gaining access to offshore wind turbines remains largely dependent on favourable weather and wave heights, especially now that new projects are being constructed further from shore to take advantage of better wind resources.
Although helicopters can be used to transfer technicians, tools and components to the site, they are expensive to operate and their deployment can also be limited by bad weather.
As a result, the industry is exploring developments in areas such as data analysis, robotics and even artificial intelligence (AI).
A recent report by DNV GL found significant advantages for wind power developers and operators from these technologies, including the use of robots for surveying, AI for due diligence and inspections, and self-driving transport to deliver components.
The big challenge for offshore wind is to remove the need to enter the turbine. Ideally, the machine should be fixed remotely, or its faults should be predicted or prevented through component analysis. AI plays a large role in this process.
According to Nicolaj Mensberg, head of asset integrity management at Ørsted (formerly Dong Energy), the 312MW Borkum Riffgrund 1 project in Germany’s North Sea, is accessible from sea only half the time. "We don’t go out there unless we absolutely have to," he said.
Smart studies
The University of Manchester is heading up a study into how new technologies can bring down maintenance costs. Titled the Home Project, backers include leading OEMS and the UK seabed landlord, the Crown Estate.
The project’s scope includes the use of big data and sensors, advanced digital simulation (particularly in floating foundation design), the integration of wind through better forecasting, and the use of robotics for operation and maintenance services.
Project leader Mike Barnes says there a number of applications that could already be employed, not least the fact that much of a turbine’s data is not currently used.
There is also scope to use existing component performance data to analyse the condition of other turbine parts.
"The study is looking at enhancing state-of-the-art modelling to enable better ‘what if?’ scenarios to be run; looking at robotics for wind farm, subsea cable, and platform maintenance; looking at big data options and advanced sensing and sensor techniques," Barnes said.
US company Sentient Science has developed complex material-based models to better understand the stresses individual components face, right down to the level of grinding on individual bearing.
This allows the company to discover the weakest points of a specific asset, to the point of whether the same component could have different weaknesses on other makes of turbine.
This advanced understanding of a component’s make-up allows the company to predict how a part would perform in specific conditions or machines. It can be used to create a better picture of when a turbine might fail, and what might cause it.
Henrik Bæk Jørgensen, Head of Product Management at MHI Vestas remarked upon their use of smart data. "Traditionally the wind turbine industry, including MHI Vestas, has been relying on CMS and SCADA data with 10 minute average values to predict failures and to improve turbine performance. MHI Vestas has, together with Vestas, developed a SMART Fast Data solution, which collects data from approximately 1000 sensors in a V164-9.5 MW turbine between 1 and 50 times per second. This gives a very detailed insight into the health of the turbine resulting in increased availability and energy production."
Rise of the robots
New, far-from-shore projects in the North Sea will require technicians to be permanently stationed on floating bases. Robotics could help them in a number of ways.
Improved sensors and modelling will provide a good idea of the turbine’s condition, though closer inspection might be required. Drones are already being used to monitor blades of onshore turbines, but this is usually dependent on the pilot being close at hand to visually control the drone. Offshore, though, presents a different challenge.
Ideally, the drone could be sent out to the turbine and controlled from the mothership. But a delay in transmission of even a few hundred milliseconds can be problematic.
Another area where robotics could help is bringing components offshore. Transport logistics is an area with plenty of scope of improvement. A typical situation might be the transport of a specific part that the offshore engineer did not have.
Pros and cons of AI
Those hoping that AI will soon make the need for human wind technician redundant, will be disappointed. It will be some time before robots are fixing faulty wind turbines.
DNV GL has highlighted a number of concerns over AI deployment. Comparing self-learning AI to a black box, it said that even "on a basic level" we don’t understand how it works.
"We can see that once trained they give the right answers in general, but we can’t open up the system and see how it came to a particular result," it reported
From a safety perspective this puts enormous pressure on correctly setting up the initial training. It is hard to foresee the consequences if human bias finds its way into how the machine is programmed to train itself.
Then there is the socio-economic impact of handing jobs over to machines. Beyond this, there is a fear that the industry could effectively be deskilling itself.
Cutting the costs of operations and maintenance, with its consequent effect on the levelised cost of electricity (LCOE), is a compelling driver for offshore wind power developers and operators. The role that smart technology will play in this is only just beginning.