We have a lot of it, but now what?

Today’s utilities are trying to move forward with digital transformation, but understanding and contextualizing massive amounts of data is proving to be difficult. It’s stopping some utilities in their tracks… or even worse making some utilities backtrack. Understanding what data means is fundamental to digital transformation and with the cognification of the grid, new data streams are causing utilities to furrow their brow and ask, “what is this data even telling me’?

Zpryme has been exploring this challenge in a series of surveys and reports, the most recent being a research report titled “The Autonomous Grid in the Age of Artificial Intelligence of Things”. As part of this study I called up several utility employees to ask them a few questions about how their organization is using artificial intelligence (AI) and the Internet of Things (IoT) today and what the plans are for these technologies moving forward.

In my first interview with the CIO of a large co-op in the southwest I asked what the biggest areas of opportunity for AI in his utility today are. His answer, not much to my surprise, centered around the same problem most utilities have: too much data, not enough information. He said, “Nine years ago it was very clear that there would be an opportunity to manage all the AMI data as it was coming in. Now, it’s every digital device that we deploy and asking ‘How do you make sense of the data?’ and ‘How do you know its normal? How do you trigger a response whenever you recognize an abnormality?’” These questions are not unique to this one southern coop. In fact, while attending a Zpryme workshop on using AI for strategic planning, I spoke with the Director of Enterprise analytics for a large muni and she her team also was having trouble making sense of their AMI and newly deployed sensor data.

The issue of big data paralysis exists because we as humans are not equipped to efficiently and effectively process how much data is being generated from the grid and customer’s homes. Luckily, there’s a lady just across the room that some may refer to as a renaissance woman. She can solve a variety a problems, including this too much data conundrum. Her name is artificial intelligence, but most people call her AI.

Some utilities are already using analytics for their AMI programs, but even the most advanced systems today can be limited in their ability to impact business operations. They can exist in silos and sometimes fail to produce meaningful information. Some analytics programs are inoperable with new systems and can’t synthesize data from countless disparate sources utilities need today. In an interview I conducted with a manager of Advanced Metering for a large IOU he spoke about this issue by saying, “For IoT the most obvious thing we’ve done is the meter…but the hardest thing is making the data available, there’s petabytes available. The key is to get in a system where people can get the data from disparate sources. Then you gotta make it available to everyone who needs it and the different types they need.”

The Internet of Things plays a huge role in data availability, and over the past two years utilities have significantly ramped up their use of IoT technology (Figure 1). In this chart from our recent report with SAS, we compared findings from a 2016 survey we conducted regarding the use of IoT/AI to our findings from the recent survey.  More utilities are using IoT since the survey and significantly more utilities (20%) have a specific and comprehensive strategy for IoT. This is increasing focus on IoT inherently means new data. In the interview the CIO, he mentioned how growing IoT is creating new challenges, “For IoT, we have deployed 150 to 200 modems in vehicles. We discovered some of those modems were hacked. One of the key things we are looking at is to secure these IoT devices…managing the devices and making them more secure.” The deployment and installation of new sensors and devices forces utilities to secure and control the grid in a more extensive and intricate way. More visibility means more accountability, which today’s utilities are struggling to deal with.

You may ask, but what about AI? Our survey of the industry only showed a modest increase over 2016 in the number of utilities using the technology. However, more importantly over half of utilities consider AI to be critical to their company’s future success. It’s reasonable to conclude more utilities have not moved forward with using the technology because they are unsure how best to proceed and not because they don’t see the value. The silver lining of too much data and not enough information conundrum is the potential for AI to solve the business challenges utilities face.

Sifting through our survey data, it is evident that AMI and distribution system data is the first biggest hurdle that many utilities are working through in order to prioritize greater reliability. Most enterprises are starting by layering AI on existing analytics and automation programs, which they have seen success with over the past few years. The best example is Distribution Automation (DA). The Advanced Metering Manager said, “We’re doing a lot with AMI data and other DA equipment, looking at asset failure predictions. This in conjunction with all of our sensors and IoT allows our to trouble shooting remotely which improves efficiency and outage and restoration time”. This use case of analytics for predicting asset failure is popular with most utilities today, but only a limited number of utilities have used AI in business areas outside of operations. Utilities are still exploring how to recognize value in other areas of the organization outside of operations.

The focus on operations stems from the expected benefits utilities expect to achieve from IoT and AI. The top two according to our survey are improved grid management and operations and improved power quality, reliability, and restoration resilience. From this standpoint, it makes sense to start with operations.

Transforming into the utility of future means more than just the modernization of operations, it requires a more holistic approach. It’s time to think bigger and break down barriers when it comes to information sharing across organizations. The convergence of IT/OT, which my colleague Chris Moyer addresses in his article “How Utilities Use Data Matters”, is creating new business structures where data moves and flows in currently unfamiliar ways. Utilities are only just starting to explore the potential of sharing data between discreate systems and departments. This attempt to break down silos using data is exactly what the Director of Enterprise Analytics for the large municipal utility shared with me in our recent AI workshop. Her team is exploring use cases for AI in strategic planning and trying to prove a business case for it. In building out the business case, they also wanted to link improved customer satisfaction to the implementation of AI for asset failure prediction. She said the first presented use case was too operationally focused and they needed more customer facing improvement. To explore the customer side, the utility is now doing an outage prediction use case as an extension of failure prediction. “The idea is to improve customer experience by communicating with greater accuracy the likelihood of any outage and expected duration, because customers take different action being out 5 min versus 30 min”, the director said.

The idea of improving customer service by improving reliability is not new, but how utilities do it, and more importantly proving how they do it is new. Tackling this via the digital transformation is what most utilities today have their sites set on. In Figure 9 from the report, we can see this reflected in primary planned uses of AI and IoT together. The top three are grid operations, customer engagement, and improved reliability.

Overall, AI has the potential to give utilities a competitive advantage through process efficiencies but the use cases are still being explored. Cognifying the grid is the next evolution of enterprise analytics. Starting with internal processes, utilities are leveraging enterprise analytics to support Artificial Intelligence uses cases across the entire value chain, from generation to the customer. Automation will help electric utilities better predict supply and demand, balance the grid in real time, reduce downtime, maximize yield, and improve end-users’ experience.

In my next article, I will explore in more depth the different types of AI and how utilities can build business cases to support its implementation.


To learn more about industry perspectives on AI and IoT download the full report, “The Autonomous Grid in the Age of the Artificial Intelligence of Things”. Zpryme surveyed 120 global utility professionals asking a series of questions over how they use and plan using AI, IoT, and Blockchain.