Predictive Analytics Preventing Cable Failure

ENMAX Power Corporation identified underground cable failures as the highest contributor to both total customer outage minutes and total customer outage frequency. In response to this finding, ENMAX Power developed an innovative predictive analytics model to proactively identify buried cables at high risk of failure.

The model is meant to improve the customer experience by addressing reliability concerns due to unplanned cable failures in high-risk areas and help mitigate costs associated with unplanned outages, which can take time to locate and repair.

The predictive model identified the top factors that best predict future faults and applied a ranking system similar to that of online search engines to prioritize results. Overall, the model used more than 1.5 million data points to “learn” from thirteen years of historical information. This involved data for each asset including physical features such as- burial method of the cables and the number of downstream customers, as well as dynamic features such as outage history, previous faults, and ambient relative humidity. Data gathered between January and July 2019 was used in the testing phase of the model, when ENMAX Power compared predicted outages to actual outages that occurred during that timeframe. The results revealed strong correlations and accuracy in the model’s predictions. On average, 70 per cent of the top cables predicted to fail experienced at least one outage.

ENMAX Power views digital predictive analytics as a way to optimize spending while also improving and maintaining the reliability of the grid to serve customers.