Sentient Science partnered with Amazon Web Services to host DigitalClone, Sentient’s cloud-based software platform, which provides predictive analytics and life extension actions to wind turbine equipment owners, operators and service providers.
DigitalClone predicts component life based on material characteristics and properties within field wind turbines. Sentient’s software brings prognostics to the market through materials science, providing visibility to crack nucleation, initiation and time to surface level damage.
The highly technical and scientific approach enables time and data-driven life extension action to align with business rules and improve financial outcomes for each field turbine. Sentient has a team of materials scientists, mechanical engineers and tribologists to build the DigitalClone models. Sentient’s Data Science team then integrates field and operational data to provide short-term life predictions of the current state of the asset.
Drivetrains are complex and to understand how the bearings and gears fatigue or crack, the scientists reverse engineer the properties of the asset to predict how that component will behave in its operational environment.
Well instrumented wind turbines provide patterns that Machine Learning models can pick up to predict when a crack has formed in a component, however, without materials science it’s typically too late to take life extension and cost-preventative action. DigitalClone integrates the materials models with ML or data science models that are trained from the operational data, SCADA, fault and events databases of each unique asset in the fleet.
Sentient partnered with AWS TensorIoT to accelerate its time to deployment in a “Cold Start” strategy. Sentient’s current customer base is 40,000 wind turbines with 20,000 deployed live in the platform. It takes time to prepare the data before it can be integrated into the software, so Cold Start enables Sentient to engage with customers on a 10x scale by building an architecture that provides immediate value.
Cold Start looks at the energy and grid control intelligent edge from wind farm geolocation, install date, climate, terrain and MW capacity to provide a failure rate assessment.
The Cold Start model can be applied to every wind turbine in the United States. The initial data analysis can be integrated with a full turbine model that can be trained using customer data.