In the mid-1960s as the Vietnam War raged, Secretary of Defense, Robert McNamara attempted to use data and numbers to gauge the success of the United States’ war effort. He famously said, “Measure what is important, don’t make important what you can measure”. Unfortunately, this data-driven axiom was not applied to the gruesome tally of human causalities or the “body count”. McNamara thought the best way to measure the progress of the conflict was how much pain the U.S. was inflicting on their enemy. He failed to appreciate that this was not the most metric that mattered most to the Viet-Cong or even the American generals on the ground. In his 1977 book, The War Managers, retired Army general Douglas Kinnard castigated the obsession with the body count as a data point. “A mere 2 percent of America’s generals considered the body count a valid way to measure progress. ‘A fake—totally worthless,’ wrote one general in his comments. ‘Often blatant lies,’ wrote another. ‘They were grossly exaggerated by many units primarily because of the incredible interest shown by people like McNamara,’ said a third.” In today’s information age data has become one of the most pressure commodities a company can have. Businesses have learned vital lessons from the mistakes of the Vietnam era Department of Defense. It is crucial to measure what is important, but it is now almost more essential to comprehensively understand how that data will impact an organization beyond the point it was gathered at. As utilities capture more and more data from IoT sensors and devices throughout the grid it will become increasingly difficult to understand how each data point relates to each other, and what is truly important. It is at this point of data overload that artificial intelligence (AI) becomes the catalyst for turning information into actionable decision making.
The utility industry has been using sensors to collect and manage grid and asset data for thirty years. However, the process of data collection accelerated after 2009 with extensive grid modernization initiatives ushering advanced metering infrastructure and outage management system investments. In 2016 Zpryme conducted a survey that explored perceptions and deployment of IoT devices, “Autonomous Grid, Machine Learning and IoT for Utilities”. At the time only 43% of utilities reported using IoT technology. In the last two years, that number has grown by a third with 57% of utilities now having deployed connected sensors and devices (Figure 1). This growth is significant for two key, related reasons. First, the growth in IoT devices will mean that utilities have a significantly larger understanding of what is happening in their grid in real-time, because of the vast amounts of data they are able to collect. Second, this new data has the potential to overwhelm most utilities ability to process it.
Source: The Autonomous Grid in the Age of the Artificial Intelligence of Things
The pace of AI adoption was already behind the deployment of IoT going into Zpryme’s first study, and it has not kept pace over the past couple of years. While a majority of utilities see AI as critical to their future success, this has not translated into widespread usage as of yet. Only slightly more than a fourth of all utilities are currently using AI. “The Autonomous Grid in the Age of the Artificial Intelligence of Things” report explores the causes of this, and why it is so crucial for utilities to start coordinating their IoT and AI programs. From a process perspective, it makes sense to have the equipment to capture the data first, and then implement the advanced algorithmic software to analyze it.
How can data scientist at utilities make the case to stakeholders, customers, shareholders, and regulators? The biggest pathway towards proving an ROI will come when IT and OT professionals experience business challenges, where the standard human approach to data analysis is too cumbersome or slow, and AI provides a pathway to overcome the problem. A distribution manager from a northeastern US utility described one of those challenges liked this, “The biggest challenge is integration with our old assets. That’s an IT problem because they need to figure out how to utilize the software to better manage them, and it’s an OT problem because they need to acquire the right resources”
AI can tackle this IT and OT problem simultaneously, through advanced algorithms by applying logic to these disparate data sets. Then using real-time IoT data, AI-based machine learning will look for optimal resource and asset management results that distribution managers can apply. It is the ability to solve business challenges like this that results in 55% of utilities recognizing that a tightly coupled AI and IoT strategy is crucial to the future success of the industry (Figure 2).
In a world of human biases, it is often easy to fall into the trap that Secretary of Defense McNamara did by making important what we can measure. In a business world where we can increasingly measure everything, it becomes all the more imperative that we use AI to make better decisions.
The “The Autonomous Grid in the Age of the Artificial Intelligence of Things” report from Zpryme was based on a survey of 120 global utility professionals conducted in 2018. Respondents were asked a series of questions about how they are using AI, IoT, and Blockchain and how they planned to use these technologies together in the future.
