ENVI Inform in Action: Wind Turbine Inspections

Problem:

Wind energy is a fast-growing industry that accounts for ten percent ^[1] of the US electricity supply. At the end of 2023, there were 73,352 turbines ^[2] in the United States with a rated wind capacity of 141 Gigawatts (GW). Conservative estimates predict 3,000 turbines will be added annually with the Department of Energy forecasting that by 2030, the total wind capacity in the US will grow to 224.07 GW; a 62% increase.

Wind farms are located in remote areas. The harsh and often extreme weather and environmental conditions they encounter, such as hail, lightning, and dust, can cause significant blade damage that directly impacts the performance and shortens the turbine's usable life. Decreased turbine efficiency can result in up to a 25% decrease in a wind turbine's Annual Energy Production (AEP).

In the absence of preventative care, maintenance can cost a company upwards of $50,000 per turbine per year. Between the decrease in a turbine's AEP, and the risk of high maintenance costs, wind farm operators are turning to drones, imagery, and machine learning to prioritize maintenance efforts quickly and proactively.

BladeEdge's Capture Assurance Tool (BECAT) was specifically designed and configured for wind turbine inspections. To ensure pilots capture a complete collection of images before leaving a site, BECAT validates that the captured data contains the relevant GPS and image metadata, and ensures the data has the proper overlap and pixel spacing for downstream analysis. BECAT also makes it easy for pilots to package the data to be uploaded for downstream analysis.

Once a solution for data capture assurance was complete, the next step was creating a process to automate the inspection workflow. The NV5 Custom Solutions team trained machine learning models using data labeled by BladeEdge domain experts to create an automated workflow.

The workflow was deployed to ENVI Inform. ENVI Inform is an automated monitoring service that processes and analyzes large volumes of data and delivers insights. In this case, ENVI Inform automatically preprocesses the thousands of images captured in the field, runs the machine learning (ML) algorithms to detect damage on the high pressure, low pressure, trailing edge, and leading edge sides of the wind turbine blades. The damages detected in the individual images are then translated to the specific location on the blade. The inspection results are published to the BladeEdge portal where clients can view inspection results within a matter of minutes, and the end-user has data to make informed decisions.

Results:

With BECAT, UAV inspectors have an easy way of reviewing the data captured in the field and no longer need to make expensive return trips. BECAT ensures pilots capture a complete and quality dataset the first time, while not over-collecting. Using the BladeEdge controller application for automated flight software, UAV inspection times were reduced by more than 75%. Experienced pilots can complete an inspection within 15 minutes, and BladeEdge has confidence the data captured is optimized for downstream analytics.

ENVI Inform provides BladeEdge with an operational workflow that takes the images captured in the field, prepares the data for processing, performs the analysis, and delivers the results to their web portal 24 hours a day and 7 days a week. With ENVI Inform, BladeEdge was able to create the wind industry's first image-based Autonomous AI solution that continues to deliver 98.8% accuracy in automatically locating damage. As a result of these efforts, BladeEdge has achieved more than a 3X improvement in operational efficiencies and a 4X improvement in cost savings.

Learn more about ENVI Inform and see if an automated monitoring service would help your mission. If you want to move your geospatial projects forward, faster, reach out to our Custom Solutions team of remote sensing and subject matter experts.

[1] US Energy Information Administration (eia.gov)
[2] US Wind Turbine Database (USWTDB_V6) https://eerscmap.usgs.gov/uswtdb