AES Builds Wind Turbine Failure Prediction and Smart Meter Models on H2O AI Cloud

AES Corporation, a leading renewable-energy company with global operations recently named one of the World's Most Ethical Companies for the ninth straight year and winner of the Edison Electric Institute's Edison Award, built its AI practice on the H2O AI Cloud, starting with three use cases: wind turbine predictive maintenance, energy bidding strategy for hydroelectric plants, and smart meters. The primary use case was wind-turbine predictive maintenance, where specialized and expensive cranes can cost upwards of $100k for every repair — but if maintenance can be predicted and planned, the costs are 2/3rds less, around $30k. 'We wanted to send our trucks to the towers that needed repair, with knowledge about specific maintenance needs. We also wanted to make sure they were equipped with the parts required to perform the maintenance. Before our AI solutions, we were working on a best effort basis, and trips rarely went exactly to plan,' said Otto of AES. The team built about a dozen models that provided greater than 90% accuracy, starting with existing data from wind turbine manufacturers and combining those with H2O AI Cloud. To make the outputs more useful for performance engineers, scenario analysis (Monte Carlo simulations) are included in the final outputs to answer: What do we think is going to fail? How long until it fails? Should we handle the issue with planned maintenance, or let it fail? The initiative immediately delivered cost savings and more consistent power delivery. The AES team then captured more data, added sensors to multiple wind turbine components, and automated oil sampling. AES now has 35 models in production helping predict failures across the wind turbine fleet. A smart meter project eliminated 3,000 non-essential technician truck rolls, saving about $1M a year. A separate vegetation management solution, which optimises tree trimming to reduce power line outages, cut the area that needed trimming and helped AES reduce its Customer Average Interruption Duration Index (CAIDI) by 10%. H2O's case study, which has since been updated, now says AES runs more than 150 models in production supporting 85 AI use cases, using H2O MLOps, H2O Wave and H2O Feature Store.