September 24, 2013 - From the September, 2013 issue

James Kelly on Emerging Value of Sensors, Data Analytics, and Grid Stability

In this MIR interview, James Kelly, formerly of Southern California Edison and now a strategic advisor to GRIDiant Corporation among other startups, discusses the role of Embedded Network Sensing (ENS) in generating data, how improved data analytics will affect the management of the electrical grid, and what the evolution of data portends for the physical worlds of industry and utility. As alternative energies and power sources gain support in state and local government, stronger tools for monitoring the electric grid will be necessary for balancing supply and demand.

James Kelly

“[When] we talk energy, we recognize increasingly that this is a world that will be governed by data, and that data has to be made into useful, usable information. This is that kind of breakthrough.” -Jim Kelly

Jim, you focused when last interviewed for VX News on the challenges of integrating renewables into the electric grid, as well as on how you have invested your talents and time on new smart grid technologies since retiring from Southern California Edison after 38 years. In this interview, begin please by commenting on the evolving role and value of embedded sensors and data analytics for utilities and alternative energy producers.

James Kelly: It’s a fascinating topic that I’ve gotten increasingly involved in while thinking about this trend that’s sweeping through our industrial processes and our personal lives. More and more, whether we realize it or not, what we do is being monitored in some form or fashion. 

If you’re carrying a smartphone, the odds are that someone can tell exactly where you are, and if they apply a little bit of analytics—which really means smart looks at the data and a little bit of math—then they could figure out at that time who you were with and probably what you were doing based on that locational information. They’d start to triangulate between your location, the location of others, where that point on the globe was, and perhaps even then tie back to what you spent or bought. They could start to put together a perhaps scary yet complete look at what you did in your day. And that doesn’t take any breakthrough technology: it takes the phone you are carrying in your pocket right now. 

If we think about that, and we expand it to a macro-scale, and we talk energy, we recognize increasingly that this is a world that will be governed by data, and that data has to be made into useful, usable information. This is that kind of breakthrough. When people think about data, they often confuse data with usable information. Data is just a bunch of stuff, and it’s not something I can necessarily do any good with unless I apply smart tools to it. When you think about this question, you break it into two pieces: the first is Embedded Network Sensing and this is a field in which tremendous work has been done by a number of academics—Dr. Monica Kohler at Caltech comes to mind—but also many others who have thought about how to put low-cost sensors throughout the world to monitor important things like, in Dr. Kohler’s case, seismic events. How can we understand them better and even someday accurately predict them? 

When you apply that same idea to energy, you realize that what we’d really like is to have a vast network of sensors that are smart enough to look for things on the electric grid, whether they are in your dishwasher or on a 500,000-volt transmission line. Then, we’d like to send that information back so that these smart tools and analytics can be applied to turn that into usable information. In a very short period of time, almost real time, we can react based on what’s happening in the grid. 

If you think about a very up-to-the-minute case, here we have California leading the nation in the adoption of renewable power technologies, and going forward that means the sun and the wind. The fundamental problem with the sun and the wind is that they are, by nature, intermittent. Cloud cover comes over and solar panels start producing less; the winds stop blowing, windmills produce less. These things change hundreds or thousands of times in the course of a day and yet the role of the grid operator is to ensure that when you flip the switch on your wall, the lights come on each and every time. There are incredibly high reliability expectations from us as consumers and from businesses of all kinds. It’s imperative to public safety. Imagine schools, hospitals, and traffic lights all going out because we can’t maintain the grid’s reliability. So how do you take fundamentally intermittent resources like the wind and the sun and match them to the demands of consumers with incredibly high reliability? That’s an example where I need tons and tons of data so that in any given second in the course of a day I know where all my sources of energy are, and I know what all of my uses are, at least in broad categories. Then, I figure out in real time how to match those two. In order to do that I need incredibly powerful computing with really smart mathematics, and that’s where this notion of grid analytics has risen at the highest level.

Growing attention is being paid to big data, both the generation of it and the analytics that give this data meaning. Will the next digital disruption be in the application of these technologies and data analytics to the industrial sector?

It will. I think we are already seeing it to a much greater degree than people realize. When people talk about a smart grid, I always say that we don’t need a smart grid because we already have one. What we want is a smarter grid because we’re already applying so much of this field of study to the management of the grid, which is an incredibly complex machine. It comes back to those fundamental tools. We need sensors out there gathering information, we need communication technologies that can cheaply and efficiently transmit the data back, and we’ve got to have computers so big, powerful, and cheap that we can process the data fast enough for it to matter. Data by itself means nothing. Data is a library with no readers in it. What matters is information.

Jeffrey Immelt, the long serving CEO of GE, is betting big on what he calls the “industrial internet”—bringing, as never before, digital intelligence to the physical world of industry. Is his focus for GE consistent with your analysis of the market potential? If so, what are the implications?

We are seeing it. We’re seeing a variety of different silos within industry, and hopefully they’re all going to come together in a wonderful way. We have a number of firms looking more at the issue of sensors—how do I monitor more things in our physical world? We’re also seeing firms concentrating more on the communication piece—how do I get the data back? Finally, we’re seeing firms asking, “once we have all this data, how do we make it into usable information quickly?” We’re seeing this in manufacturing processes. It helps cut costs to consumers by enabling smarter, better production with higher quality. In the electric grid, it enables us to improve reliability while reducing emissions, carbon footprint, and so forth, which is no small task.

IBM has a project called “Smarter Planet.” Cisco champions the “internet of things.” As noted, GE now is committed to the “industrial internet.” Extrapolate over the next few years how utilities and industry will adapt to having the capacity to amass and analyze more and more industrial data. 

You have both an obvious, primary consequence and a more subtle, secondary consequence. On the front end, the obvious one is that as more technology is applied by some of the industrial giants you mentioned, it enables the electric utility industry to adapt and adopt those technologies. Rather than having to create them in a laboratory, we can take the same technology that works for other devices and other processes throughout the industrial world and adapt them onto the grid so we can put a really cheap chip inside your dishwasher or inside your pool pump or at the meter. Smart meters are certainly a revolution right now. We take digital electronics, we put a smart chip in there and we suddenly have a whole new source of data that can be turned into information. 


The secondary consequence of that, which a lot of people have missed, is that we kind of have a double whammy in the electric utility industry. As more and more of the things in our world are connected, digital and communicating, they also are incredibly dependent upon reliable electric supply. We are now looking at industrial processes that are so sensitive that if that process loses power for even portions of a second, it could cost that manufacturer tens or hundreds of thousands of dollars in lost product. They need to reboot all the equipment. We’re not talking about steam engines now. We’re talking about making digital machines that demand incredibly high power quality. The predictions of the Ray Bradburys of the world are now coming true. At the same time that we’re facing more and more intermittent resources and challenges on the grid, we have customers who are properly demanding ever-greater levels of reliability and power quality. If the electric utility industry doesn’t incorporate this technology to enable that to happen, they are letting down customers and letting down our push towards progress. So this is really a big deal on both those fronts.

You said a year ago about the smart grid that we need good regulations and good economics and that good economics would trump everything, even if regulations were weak. Is that still your opinion?

It is. One of the things that we’re going to see, and that we’re already seeing, is that there will be many people with bold ideas that ultimately don’t pay. Frankly, we find that giving consumers, in all walks of life, data or even processed information that isn’t applicable to what they’re doing in their lives is really a fool’s errand. Trumpeted a bit in the trade press, there was this notion that as soon as we put in smart meters, everyone would have an application running on their home computer to sit and studiously monitor their energy consumption. What we found is most people just want it to work and don’t want it to cost too much. 

The same is true if you think about where we’ve gone in the whole digital world. When I started with computing, I had to learn to write code—ones and zeros and bits and bytes. Now kids would laugh at me for that. They’ve never even heard of that because the computer just takes care of it for you. It gives you the information you need, when you need it, and all you have to do is a little bit of basic configuration. Ultimately, that is where all of this has to go - to the point where it becomes relatively seamless and provides value, and that’s where the economics come in. If you build the smartest piece of software in the history of man and it does something that nobody gives a damn about, you will still go bankrupt because the economics will trump the cool science. Ultimately it must provide value. Looking for analytics for the grid that actually provide value and not just flash has been my pursuit over the last few months.

You now serve on the board of GRIDiant, one of the young companies you have advised since leaving SCE. Given your observations above regarding the evolving marketplace for big data, what induced you to become a part of this company?

At GRIDiant, you have the one-two punch needed in this space to make a difference. You have people who are incredibly smart with math and computers. All of this that we’ve been talking about relies on really high order math. We’re applying all these techniques for solving very difficult problems in very small amounts of time with massive data pouring in. Therefore, the math tools that underlie all this—the engines if you will—have to be really formidable. But you have to combine that with enough knowledge of how the grid works so that when you’re all done doing the fancy math, what pops up on the screen is stuff that people can use. It comes back and says, here’s your problem and here’s how to fix it, or here’s an opportunity to improve customer reliability for the lowest possible cost, or here’s the smartest thing you can do to get those customers who did have an outage back online as quickly and safely as possible. 

Combining those two is the secret ingredient to this really making a difference. And that’s what I liked about GRIDiant. We have experts in utility operations combined with people who are incredibly bright with math and computing. The firms that prosper, and I think there will be several in this space, will have that same combination: expertise in the industry they’re serving and the math and computer skills.

Lastly, the opportunities of distributive energy necessitate utilities to integrate and manage intermittent renewable energy into the grid. Is it your view that data analytics will contribute to better management of the grid? 

I think they will help manage it at two different levels. When we think about renewables, there are at least two big platforms. First are those sorts of large centralized renewable power plants that replace the coal plants, the nuclear plants, and the gas plants of old. So you go out and put up a couple thousands acres of photovoltaic panels, or you put up a gigantic wind farm as we see in the Tehachapis, and that’s really a central power station, if you will. We’ve got to manage that intermittency, in real time, between that and all of the other power stations. 

In addition, in not only California but around the world, in Germany for instance, we’re seeing renewable power come down to the neighborhood and the individual home. I’m thinking of rooftop solar, which is really exploding around the world. When you think about putting what are essentially power plants on people’s roofs, you’ve got a real challenge in operating the distribution grid because the grid—which was, for the most part, built for soldiers coming home from World War II into the new suburban developments that sprung up—was meant for a very specific, one-way function: power came to your house, it was metered, you consumed it. That was it. 

So now there is the notion of having a much more interactive two-way grid where you’re a participant in energy management, where you make choices about your consumption based on information given, that this is not a good time to turn on that pool pump because the system is short right now. Combined with the fact that you produce power and send it back on the same wires that sent the power to your house in the first place, making all of that synchronize so that the grid doesn’t fail and customers are able to make those choices without a reliability impact becomes a formidable challenge. It’s one that, frankly, we couldn’t have taken on even ten years ago without the smarter tools developed in that time—the computing, the Embedded Network Sensing, the analytics. These are the keys to an entirely new energy future that, when I started in the business 40 years ago, I never in my wildest dreams could have predicted. It would have been as fanciful as believing that we’d all be in flying cars, but here we are. We’re on the cusp of it.



© 2021 The Planning Report | David Abel, Publisher, ABL, Inc.