Scotland is famous for many things, but consistently fine weather is not one of them. The first week of August saw parts of the country experience winds that you’d expect to find in the North Sea in January. The weather station on top of the Cairngorms mountains – where, incidentally, the highest wind speed in UK history was recorded in 1986 (173mph) – registered top speeds of 115mph.
While causing considerable inconvenience and ruining some people’s holiday plans, these freak winds had a welcome side-effect. On Sunday 7 August Scotland’s wind turbines produced 36,545MWh of electricity, equivalent to 106% of the country’s electricity demand. This was the first time that enough wind has been generated to fulfil all of Scotland’s needs and is a possible major milestone in the efforts of wind power advocates to promote it as a pillar of generation and not just a supplementary technology.
Of course, such days don’t happen very often. With the conundrum of how to store wind-generated energy remaining unsolved, most wind farm operators clamor for the most accurate weather readings and forecasts to ensure their turbines can take advantage of optimum wind speed and to better predict supply and demand fluctuations on the grid. A suite of apps released in May by GE Renewable Energy for its Digital Wind Farm product should improve the ability to do this, as well as a number of other data-driven operations.
Predicting the future to optimise wind farm operations
Digital Wind Farm was first unveiled in May 2015, perhaps the first time that the Internet of Things was applied to wind generation. It essentially places wind turbines on a computer network through which their performance is monitored and tweaked by a series of sophisticated software programmes. Data collected by these apps is analysed and any insights derived go into the development of the next-generation of software, this latest suite being a case in point.
“Our goal in creating Digital Wind Farm apps is to help our customers drive more production, higher profitability, better availability and reduced risk,” says Dave Vernooy, executive digital product manager for GE Renewable Energy.
“That usually means helping them find ways to better interpret and act on their operational data. Not every customer’s needs are exactly the same, so it’s important that we work with them throughout the ideation and development process. That way we know we’re building a digital solution that will actually be useful. In general, we’re trying to build digital solutions that will help their turbines and their operations run better tomorrow than they do today.”
The first component of the new suite is the Energy Forecasting application. It consists of a real-time analytics engine, which forecasts wind generation potential every five minutes over the space of an hour; a day-ahead function, which projects wind generation in hourly intervals over a 24-hour period; and a seven-day forecasting tool which attempts to calculate wind generation potential each hour over the course of a week. Using these forecasts, the wind farm owner should be able to create more accurate financial models and be able to better identify wind ramps (where the wind direction changes suddenly) during the day.
“We worked with Exelon to develop our Energy Forecasting application,” says Vernooy. “It has now been deployed at four Exelon wind farms in the US and our customer is reporting between 1%-3% higher profitability for each site.”
To compliment this are two apps designed to improve wind farm operations. Digital Plan of the Day combines a variety of data sets such as SCADA information, diagnostic readings and weather metrics to help organise daily maintenance schedules. Then there is the second generation of the PowerUp Services app, which GE Renewables claims can increase wind farm output by up to 10%. It helps the facility make small software and hardware adjustments based on performance data, but with greater sensitivity than its predecessor.
“The new version includes an iterative tuning process that helps continually lock-in the most appropriate settings based on the most current information available,” says Vernooy. “It is fundamentally about having the turbines respond more dynamically to their local, real-time environment based on a combination of sensors, machine-specific models and controls.”
Predictive maintenance the long-term goal
The two remaining apps deal with asset performance management. One called Diagnostics crunches operational data to predict possible anomalies in performance, while Prognostics uses operating, maintenance and inspection data to judge the reliability of a wind turbine’s components. These apps are a step towards the holy grail of predictive maintenance, being able to identify and fix problems before they occur. In Vernooy’s view, we are finally reaching the point where there is enough data and sufficiently sophisticated analysis techniques to make predictive maintenance a reality. These two apps, he believes, could result in reduced maintenance costs of up to 10%.
“We’re getting closer and have seen significant improvements the last few years, but we still have a long way to go,” he says. “Predictive maintenance relies on both a deep understanding of the underlying reliability drivers as well as a comprehensive data set against which to refine that understanding. The scale of the wind industry is now at the point where this is becoming viable, and it will open up new possibilities in the way wind farms are managed.”
The new apps are currently being put through their paces in a series of pilot projects and will be available to customers later this year. If the expected efficiency savings are realised, it would force home the argument that you don’t need unseasonably windy conditions to justify the use of wind turbines.