From afar, wind power appears to be the epitome of elegant simplicity: turbines dotting pastoral settings and shorelines with blades spinning gently in the wind.
Wind farm owners, of course, know differently.
Today’s wind turbines are complex and expensive pieces of machinery. Modern turbines feature blades longer than a football field, cost more than a million dollars, and incorporate hundreds of sensors to interface with technicians and engineers located miles away. A single wind farm may have thousands of turbines and generate gigawatts of renewable energy for nearby communities.
The challenge is to keep them turning.
The Cost of Running a Wind Farm
It’s estimated that a third of the cost-of-doing-business for a wind farm owner is implementing measures to avoid equipment failures and turbine shutdowns. These range from scheduled and unscheduled maintenance to inspection and repair. Complicating matters is that many turbines were installed when the modern wind power industry first took hold 20 years ago, leaving many aging wind turbines out of warranty and near the end of their projected lifespans.
However, owners can reduce maintenance costs and extend the longevity of their turbines by effectively leveraging data they already have.
Using Data to Reduce Costs and Extend Turbine Life
There are two key sources of data to consider. First, there are sensors that monitor the wind turbine’s ongoing health and day-to-day operations, reporting metrics such as the blade rotation rates, the amount of electricity being generated, and various temperatures inside the machine. Ever-growing volumes of this data are stored on a server, waiting to be analyzed in order to ensure that the turbine is performing as expected.
The second source is what’s known as transactional data. It’s the data generated when a technician is dispatched to the field to work on a turbine. This category includes procedures that workers follow, the photos they take, their notes, and all the other ways activities are documented in the field.
Today, most wind farms are not equipped to optimally manage either type of data. Due to the overwhelming amount of sensor data, wind farms can’t afford the number of analysts required to take full advantage of it, especially with traditional analytics solutions. When it comes to transactional data, the mix of paper notes and digital photography makes it difficult to access as well as challenging to integrate with digital sensor data.
The key is combining the two sets of data to help wind farms ensure that maintenance is predictive rather than reactive, thereby reducing repair costs and downtime. The transactional data adds context to the sensor data that the sensor data alone is not able to provide.
Today’s increased computational power and capacity to store massive amounts of data in the cloud now enables artificial intelligence (AI) to support wind power operations. While a limited number of asset managers are beginning to analyze sensor data with AI, they generally don’t incorporate transactional data from the field. This is a missed opportunity, since the greatest benefit is gained by bringing both types of data together and analyzing them as a complete picture with AI.
AI Assisted Blade Inspection and Repair
mCloud is aggressively pursuing this integration, particularly with blade inspection and repair. Inspection is the gateway to the blade maintenance process. It starts with autonomous drone photography, progresses to AI-based image processing, and ultimately data-driven decision making regarding what repairs are needed this season. Once a repair is approved, wind farms need to allocate capital and manage technicians with hands-free headsets and wireless network-enabled tablets to properly complete the repair.
A wind farm may also find that a piece of equipment does not require repair right away. In that case, inspection reports can be archived in a historian so they can be compared year-on-year to understand damage propagation. This cumulative blade inspection data can then be combined with the sensor information to inform AI analytics and predictive models, maximizing annual energy production (AEP).
The promise of AI is exponential returns, since it constantly learns from repetitive data and outcomes, offering predictions of what will happen next with repair and inspection decisions. With AI, wind farms will greatly benefit from predictive models that can look years into the future, letting operators know in advance, for example, when a blade or gearbox is expected to fail. This foresight enables the owner to earmark funds and order inventory and services at the best price, supporting cost-effective, efficient, and consistent power generation via predictive rather than reactive maintenance.