Intelligent Utility Network Community

The mission of this network is to bring together utility folks who are in the thick of the digital utility transformation. This network incorporates the Mobile Professionals special interest group.

61,791 Subscribers

Article Post

Using Big Data to Predict Power Outages

Damage to electric distribution systems and long-haul transmission lines made national news this year as major storms and hurricanes caused outages that left millions of people in the dark.

Utilities may soon have a new tool in their arsenal to better prepare for such threats to their distribution networks.

Researchers at Texas A&M University developed a model that they say can predict a potential vulnerability to utility asset. The model also draws a map of where and when a possible outage may occur. At a practical level, the model and map allow trees in critical areas with the highest risk to be trimmed first.

Dr. Mladen Kezunovic of the Department of Electrical and Computer Engineering, along with graduate students Tatjana Dokic and Po-Chen Chen, developed the approach that can predict weather hazards, vulnerability of electric grids, and the economic impact of associated damage.

Data such as a utility company’s operational records, weather forecasts, altitude, and surrounding vegetation all can be used to customize the model. In fact, any kind of environmental data that has some relevance to the power system can be fed into the prediction framework.

Here’s how it works.

The first step is to determine the probability of a potential hazard, like a severe thunderstorm. 

Second, the vulnerability of specific utility assets is assessed. This is done by taking the weather probability from step one and predicting its impact on utility assets. 

Third, the model evaluates the impact of severe weather events and places a dollar amount on the likely costs of maintenance, replacement, and repair.

In short, the risk analysis tool uses big data to help predict the probability of events likely to happen in the near future. The financial part helps the utility develop an action plan for operators to execute.

The Texas A&M researchers used data from CenterPoint Energy and presented a proof of concept to the company. The next step is to implement the model on CenterPoint’s database and environment.

Besides the big data tool for outage prediction, CenterPoint Energy plans to use unmanned aerial vehicles—drones—as part of its Emergency Operating Plan. Such drones could help expedite the company's ability to assess damage to its electric transmission and distribution system following a major storm.

The utility had a chance to fly drones when thousands of electric power customers in Sealy, Texas, near Houston, lost power in late May when severe storms swept through the area. CenterPoint Energy crews worked overnight and used drones to restore power to more than 48,000 customers.

"We had the opportunity to test drone technology following severe weather in Sealy, Texas, and see potential for drones to play a key role in storm and disaster response," said Kenny Mercado, senior vice president of Electric Operations.

In the  end, drones and big data tools like the one developed at Texas A&M may not end storm risk, but they could help utilities better protect their transmission and distribution networks and get the power back on quickly.


So what exactly is the 'Big Data' that is being used? Unfortunately that has become an overused and abused term. While there are several categories of data mentioned, none would seem to qualify as 'Big Data'. While not discrediting their research I wonder what is novel about it. What distinguishes it from any of a number of other predictive analytics models that have been developed for outages? Delving into that would have added to the interest of this article.

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.

We are into power distribution business with thousands of km of 33, 22 and 11 kV underground cable network. It has 1200 nos. 11 kV feeders, each feeding @ 6 to 8 S/S. 11 kV network is in mesh topology with normally off points to have alternate feed during outages. There are 10000 plus cable sections connecting various DT stations. There are about 800 nos. cable faults per annum. Is there any methodology to predict the fault in particular section to take proactive measures and avoid large area outage.

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.