IoT for Electric Power: Smart Grid was the Beginning

Jeffrey S. Katz
January 10, 2017


For many electric power utilities, the Smart Grid was their first Internet of Things project. Just as some early smart grid projects were started before the term became popular, and were originally known as the intelligent utility network, or advanced distribution automation, so too was smart grid an early version of internet of things for utilities.

This is not surprising, considering utilities have more "things" distributed over a wider area, as a vast, complex, interconnected machine, than almost any other industry. Smart Grid embodies a subset of IoT principles, and some of the more advanced projects made use of today's IoT concepts, before groups such as the Industrial Internet Consortium popularized them. Applying IoT to the changing world of distributed and renewable energy generation is but one of the IoT business cases. Utilities often start with an IoT strategy in the context of their already familiar smart grid projects. As they begin to encompass more of the aspects of IoT, such as analytics platforms, IoT cloud, sensor fusion, and data governance, they may use these principles in continuation of smart grid, either in different jurisdictions, or for advanced functions such as distributed intelligence.

The most plentiful "thing" in electric power IoT is the smart meter. There were 400 million such devices estimated to have been installed worldwide by 2014, with an expected growth to 925 million by 2020 [1]. Similarly, equipment revenue for smart grid sensors is expected to expand by an order of magnitude in a similar period. However, smart lighting, estimated to be 46 million units in 2015, is projected to grow to 2.54 billion units by 2020, thus outpacing the smart meter. Just think of the number of municipal street lighting systems that are becoming not only smarter, but becoming Wi-Fi hot spots and other citizen service points. The total number of connected devices managed by utilities may be 1.53 billion in 2020, which is more than triple that in 2013.

Lessons learned

So what to do? A few lessons learned already from such projects can be summarized as:

  • Everything goes well until the project starts to look for and use the data. Generations of data acquisition systems, SCADA historians, "standard" data repositories, engineering data warehouses, data lakes, and more often mean there is no consistent format for the years of data that form an experience base.
  • Systems co-exist, but automation requires integration. An IoT approach is part of a business strategy for interoperability. Classic Enterprise Service Buses have been a good approach, but may not provide the desired data ubiquity.
  • Governance of data can stall the project. Even if the problem suggested in the first bullet is not a speed bump in a utility, continuing to properly curate new IoT data often does not get the attention it deserves, leading to unexpected project costs down the road.

Once the data is under control, there are often business cases based on optimization, such as root cause determination and fault location, or maximizing use of renewable energy. Experience tells us that optimization is in the eye of the beholder (read purchaser). The electric grid is interconnected and has many coupled effects. An IoT project should not be a surprise to the rest of the company, nor should individual optimization projects be undertaken without thinking of some overall supervisory system that keeps a watchful eye on the local optimizations, so the whole system benefits. Often this is the place where cognitive computing merges with IoT.

Another potential IoT danger is that while the traditional utility is working on an IoT strategy, new competitors are forming innovative business models using IoT from the beginning. Think of Uber, or Nest.

IoT connectivity needs security for a critical infrastructure industry such as electric power. Therefore, security is a starting point, not a non-functional requirement tossed back to IT. Security in IoT also includes privacy and trust, not just concern about hackers.

One should also think of Internet of Things by parsing the phrase. "Internet" is not necessarily the public Internet. There is the more select Internet2 for example. There are regional systems proposed for utilities such as the Eastern Interconnect Data Sharing Network. "Things" include people; causing some to use the phrase "Internet of Everything." Technicians, trucks, poles, consumers, wind turbines are 'things' in the broader view of IoT. Measurements as well as simulation results are all useful data.

Thinking points

Some points for IoT thinking that can be derived from these implications:

  • If you are still talking about IT/OT convergence in your company, then you may not be ready for IoT. Alternatively, to look at it another way, IoT is already converged.
  • Develop end-to-end trusted networks.
  • Insist on encrypted storage at the cloud. Test your vendor.
  • Look for cloud center interconnections that are not over the public Internet.
  • Take notice of what is new in IT:
    • Agile development
    • Concerns about the customer and their smartphone being smarter than enterprise IT
    • Vendor equipment having more intelligence than can be used in a utility
  • Think of proofs of concept as in-house engineering staff education. See for example the IoT recipes at IBM's Developer Works [2] (and the IoT Foundation Quick Start [3]).

Brief mention has been made of cognitive computing [4] and IoT in a supervisory role among optimization software. Software at the edge is not new, however, significant intelligence at the edge is. Classic programming embodies the programmer's view at the time of the coding, rigidly defined. However, the 'edge' is not monitored as well as data center IT. Often it is helpful to think of the centralized software component as the adult, and the edge computing as the child. The total system can learn what is normal, from both the system operation and security point of view. It informs the child computers of global conditions. It helps detect emergent behavior between distributed intelligent systems.

A short Energy and Utilities IoT 'to do' list:

  • Focus on security [5]
  • Implement Cloud Computing and Big Data and Analytics pilot projects
  • Re-invent the end-user experience (customer or technician)
  • Embrace innovation
  • Plan holistically
  • Pilot, learn, adapt

A closing remark about IoT analytics. Thinking that every analytic based on IoT data has already been done can lead to lack of innovation. Listen one weekend to the most popular show on National Public Radio, "Car Talk." This is a great demonstration of how symptoms, diagnoses, and correlations are not all obvious, even though the automobile has been around longer than the computer, and is a similar vintage to widespread electrification. Data discovery tools can provide unexpected ROI, as well as new insights and shorter paths to fault resolution and optimal operation.










Jeffrey S. KatzJeffrey S. Katz is the Chief Technology Officer of the Energy and Utilities industry at IBM. He is a Senior Member of the Institute of Electrical and Electronics Engineers. He is a member of the IBM Academy of Technology. He is a co-chair of the Industrial Internet Consortium's Energy group, and is a member of the Internet2 working group on the Internet of Things.

He was a co-chair of the IEEE 2030 Standard on Smart Grid Interoperability Guidelines IT Task Force. He is on the Advisory Board of the Advanced Energy Research and Technology Center. He was appointed to the IEEE Standards Association Standards Board for 2014. He is an Open Group Distinguished IT Specialist.

He can be reached at