Endesa: Spanish Power Grid Predicts Outages with AWS
Spanish Power Grid, provides critical energy and utilities for the nation. The variation in geography and weather in Spain meant that power generators need to provide stable output in many different situations, creating complex operational requirements for the company. Outages were highly undesirable and fined by the hour by the national regulator.
Challenge
They approached Colibri to explore a predictive maintenance and analytics solution. The aim of the project was to create a more data driven approach to planning maintenance and engineering resources around the country, where engineers are deployed more intelligently, and there is less waste into part replacements.
Solution
Colibri implemented a cutting edge, real-time internet of things (IoT) analytics solution for Spanish Power Grid. We first ingested data from all power substations, tracking readings from transformers, electricity sensors, and environmental sensors. This data was pulled centrally using an AWS data pipeline.
The data volume was enormous - there was real time readings of power, voltage, draw, and more; weather factors like wind speed, direction, and temperature also went into the data lake and centralized. In addition, we helped Spanish Power Grid to enrich its own data with open source weather data.
With all the data sources in place, we transformed, cleaned, and normalized the data, then sent it to the business intelligence platform.
Results
With the new analytics system, our ML engineers put in place a new big data machine learning solution to predict failures and output fluctuations.
As a result, management and executives were able to draw insights like never before. We highlighted top 10 issues across all substations real time, and predicted the likelihood of failure in the next 7 days for all components.
We materially reduced outages experienced by the power grid, and delivered a fantastic ROI in terms of cost reductions on component replacements and engineering time.