Resource Center

Back to podcast homepage

Innovations in Energy Efficiency with Peter Kelly-Detwiler

Season 2: Episode 5

Listen now on Apple | Spotify | SimpleCast

What is the role of digitalization in energy?

Peter Kelly-Detwiler joins us on Clean Integration to take a look at the interesting and increasing trend of the digital grid over the last decade, what problems it solves today, and consider how a decentralized digital grid may increase energy efficiency in the years to come.

Peter has worked to solve some of energy’s best problems throughout the U.S. and around the world. He is the Co-Founder of NorthBridge Energy Partners, LLC, an independent consulting organization with expertise and perspective on U.S energy markets, and the author of the book, “The Energy Switch.”

With a 30-year career in energy, Peter has developed a background in both electrical markets and end-use technologies. This gives him unique and valuable insight into the interplay of technology on both sides of the meter.

(This transcript has been edited and condensed for print.)

[4:10] Why do you think the energy transformation is so big, and how does it relate to the efficiency, security, and reliability of the grid?

Peter: Sure. Well, if you step back and say, “First of all, what’s the inherent problem we’re trying to solve?” There are the three Ds: decarbonization, digitalization, and decentralization. The latter two, the digitalization and decentralization, are natural outgrowths of technology, but the first part, the decarbonization, that’s essentially humanity steering our ship, our collective society, in a different direction than we have in the past.

Back in the industrial revolution, there were a relatively small number of players in the space and we were also smaller as a species versus now where we’re closing in on 10 billion of us. We’ve got to decarbonize everything in order to keep temperatures from exceeding, say, two degrees centigrade, which was what the Paris Accord was looking at. Now, scientists are saying, maybe it has to be 1.5.

Putting that in perspective, Don Wooder, who’s the chief scientist who wrote the “National Climate Review” that was delivered to the Trump administration — I was at a conference where he spoke and he said, “What’s a couple degrees centigrade?” He goes, “Flip it the other way. Cool the earth six degrees centigrade.” And we were in Chicago when he was saying this, “Where would we be?” And we all looked at each other in confusion. He says, “We would be sitting under probably half a mile of ice, sitting in the middle of the Laurentian ice edge.” That’s how much six degrees can mean in one direction. So, you warm up the planet two or three to four degrees in another direction, you create massive instability, many parts of Africa and other places are simply not habitable anymore. You create massive potential for human migration flows. We’re looking at two million people crowding into Poland right now, or in Ukraine and other parts of Europe. That could be nothing compared to what happens within the next few decades if we continue to destabilize our climate.

So we are in a race that many people are still not fully aware of to push the carbon atom out of our industrial processes, out of our personal processes (heating homes, lighting our homes, driving our cars, everything). There are some estimates that this will be north of $100 trillion between now and 2050, in terms of the changes in business models, investments in electric vehicles, maybe modular nuclear, certainly more solar, wind, batteries, hydrogen, etc. So yes, this thing is absolutely staggering in the scope and scale of what we’re trying to accomplish.

John: Incredible. You alluded to Poland because we’re recording this in the midst of politics and economics, and so forth. The Russians have invaded Ukraine. It’s really to some extent re-energized the conversation around the carbon transition. Let’s talk about energy security. Any perspectives on that? It’s not meant to be a political comment.

[9:50] Can you make the case for us that there is an increasing confluence between IT and the energy grid?

Peter: Oh, sure. So first of all, the grid is now a cyber-physical entity. In our old traditional utilities, we had a supervisory control and data acquisition, a SCADA system, which was an industrial control system that controlled our power plants and our bulk power system, transmission, transformers, all that.

What’s new now though, is we start to develop this grid edge. You have more than a million rooftop solar arrays in a place like California. In Hawaii now, 80% of the new solar panels that go in have batteries attached. California, in certain areas, it’s 60%. New York’s going to start to see a pop in that area. Massachusetts, other places around the country, then electric vehicles, and so on.

Now, vehicles are simply batteries on wheels, conceptually. The critical piece is that the challenge of power is about how much, where it’s located, and the duration it lasts. So it’s a kilowatt. Where is that kilowatt? Where do you need it? And then how long do you need it? Kilowatt-hours. So it boils down to location and then just tons of data around the where, the what, the how long, and at what price.

And prices can be super varied. Texas has thousands of prices in the grid at any given time based upon location and constraints of transmission getting power from point A to point B. So now you start to bring in all these end-use devices that are responding to market conditions. First of all, you’ve got to monitor them all the time and know that they’re there. And then, you start to activate them.

So that Ford F-150 Lightning, that’s a 110-kilowatt-hour battery or 150-kilowatt hours. To put that in perspective, the average home uses about 30 kilowatt-hours per day. So that one pickup truck could power a home for say three to five days, and indeed is capable of doing so. So this past week, Pacific Gas and Electric announced they’re going to work with Ford to figure out ways to make that vehicle interact with the grid and stabilize the grid.

So fast forward to a future where the F-150 is roughly 5% of all the vehicles sold in this country (900,000 in a typical year of the 17 million cars sold in the U.S) — imagine a whole bunch of those plus a whole lot of other vehicles, electrified, charging at certain times, releasing power and bidirectional flow back into the grid at other times, and water heaters, rooftop solar, batteries. Pretty soon, you’re talking about potentially billions of transactions across the United States.

One utility, Pacific Gas and Electric estimated maybe four devices per household with five million households. So there are 20 million devices. Pretty soon you get all kinds of power flowing both ways. You have to monitor all that stuff. This becomes one of the biggest data plays in the world in order to effectively manage the grid and make it as efficient as we can. Because one other piece, John. The grid is sized right now for the maximum demand. It’d be like if we said, anybody who wants to fly on Thanksgiving can jump on a plane and we’re going to build the airplane fleet to accommodate all those travelers, even though the rest of the year we’re only going to fly those planes half full.

The grid is like that. It’s sized for peak demand. Those hottest days when air conditioning is running. And instead of telling people, “Shut off stuff,” we say, “We’ll build the grid bigger.” So it’s only utilized at a 50% average capacity factor. Now, if we can harness all those distributed devices to change the way we consume power, store power at periods of time, and release it at other times. If we could reduce or eliminate that top 1% of peak demand like eight hours of peak, we could save roughly 8% or 9% of our total infrastructure costs.

So as we build the grid out in the future, if we can make it more efficient, we can reduce everybody’s costs. And it will be built out. The CEO of the Electric Power Research Institute recently commented that if the Biden Energy Plan were implemented to decarbonize the U.S. economy, we would have to triple the size of today’s power grid. So if that indeed happens, it’s incumbent upon us to figure out how to use data, better architecture, and harness all these devices to make a more efficient cyber-physical machine going forward in the future. It’s a huge challenge, but it’s also an enormous opportunity if we do it right.

John: Peter, I love that. If I can play it back. The grid of the past was assuming it had a demand for power. It has dispatchable data centers, not green but fossil fuel facilities that come on and stay on, and they tell them when they come on and you’ve got kind of this dumb user, basically. To some extent, on the other end, that’s just a one-way relationship but once you start to change the mixture of the grid to have more intermittent resources, distributed energy resources, solar panels on the roof, batteries as part of that architecture, there can now be a two-way relationship between the grid and the consumers or the demand, if you will, such that those resources can actually become grid-level resources.

And it introduces more flexibility into the grid, and then, therefore, you need more than just a simple SCADA system turning things on and off. You’re now developing a much more sophisticated system that needs to have brains to figure out how to manage, architect, model, and project out the use of those different technologies. I find that fascinating because you’re right. These new electric cars, if you plug them all in or you know when they’re going to be plugged in, you can use an AI machine to determine when to charge them, and then that’s a resource for the grid.

And there’s lots of discussion around this concept of a virtual power plant where all of those batteries wake up at some point when the grid needs more power and puts it back. Power is coming back to the grid on a schedule or dispatch. That’s potentially very powerful and introduces the need for more than just physical infrastructure. Now you need digital infrastructure to control that. Is that a good way to summarize?

Peter: Indeed. You have encapsulated that quite well. And then there’s one more piece of that, which is, today, we’re talking about gasoline prices. It was $3 a gallon maybe three weeks ago, and now it’s $4.50. So everybody’s complaining, rightfully so, about a 50% increase. Think about electricity. In a place like Texas, when they had that freeze last year, the average price in Texas before the freeze had been around 3.3 cents a kilowatt-hour, the wholesale price for electricity.

When that freeze occurred for those four days, the price of electricity pegged out at $9 a kilowatt-hour. So, almost 300 times. So imagine if gas prices, instead of $3 were $900 a gallon. That’s what can happen with electricity. So because there are these potentially big price swings, now, Texas basically has limited the price to a $5,000 cap. But that’s the kind of price volatility. There’s no other commodity on the planet, therefore it really makes sense to digitize these devices and make them market-aware so that they respond.

John: You’re alluding to my next question, which is just business models. So if you look at the grid, utilities make power, power plants make power, and consumers buy the power. And there’s a very simple relationship there. What new business models do you see popping up now in this new digital world? One example you touched on is not only can I buy the power, I can actually sell the power back. What else are you seeing out there?

Click here to listen to this episode

[17:55] What new business models do you see popping up in the energy space?

Peter: Yeah. So the first ones we saw were the solar companies coming in and basically before solar was dominated by these solar companies, there wasn’t a lot of it going on because they hadn’t solved the financing problem, and solar panels, 20, 30,000, $40,000 for a decent size installation on the home.

Then they came in and said, “Oh, we’ll be the financial intermediaries. So we’ll buy the panels and we’ll lease them to you or sell you power under the power purchase agreement.” And for a while, 80% of all new solar coming in places like California, Massachusetts, and other markets was third-party finance. So that’s been around. Now, those companies are starting to move into storage. You see a company like Generac that used to just do backup generation. Now they’re buying batteries and putting those in. You see Inverter Companies, the devices that take the solar or the battery, DC power and drop it into the AC grid, converting it into alternating current.

Now the Inverter Companies start to buy batteries. So you start to see this convergence of companies building these things and then creating these virtual power plants. So they reflect in one direction and create value for the customer. Then they turn around in another direction, aggregate those devices and sell to the utilities or to the wholesale market. And the federal energy regulatory commission, the federal regulator, just promulgated a rule a year and a half ago that told all of the competitive markets they have to create operating models that compensate vendors of distributed devices that are aggregated to a hundred kilowatts that they have to be treated like any other wholesale asset in the marketplace.

You start to see, “Oh, okay, now there’s price formation capabilities.” So you now have, for example, electric school bus operators saying, “Oh, our buses are going to drop kids off or pick them up in the morning from six drop them into the school, six to nine. Then the bus will go dormant and sit in the lot. And then one o’clock, go to the school, drop the kids off between then and four. So a place like California, what’s going to happen?

The buses will absorb all the energy from the sunlight in the middle of the day when there’s already so much of it that prices sometimes go negative. Then, in the evening, when prices sometimes double or more, and all these power plants have to ramp up to serve the evening load, they have to double the amount of generation in California from say four to nine. The buses can release power and make money. And then they’ll charge a little bit in the morning, say three or four, before they pick up the kids, and then they’ll use energy dropping off the kids. Then they’ll absorb that from the sun the next day and keep doing it.

So you’re starting to see all the school buses, Thomas, Bluebird, BYD, etc, all announced bidirectional, capable buses. So power can flow in both directions. You’re starting to see business models like Fermata, which is a company that does vehicle to grid. They announced this summer that a Nissan Leaf in Burrillville, Rhode Island at a water treatment plant delivered energy from the battery to the plant so the plant didn’t have to buy it from the grid in a demand response type program. That single Nissan Leaf made $4,200 this summer. And a school bus in Massachusetts, in Beverly, MA — one school bus participated with National Grid, the local distribution utility, 50 times this summer and delivered three megawatt-hours of energy back to the grid. So you’re going to start to see all these new constellations of devices and vendors architecting these solutions.

And the real critical piece of that John is, if you have a whole bunch of different players and they’re not coordinating what’s happening between the distribution utility, wholesale market operator, the grid operator, and the vendors to create the necessary situational awareness, especially when power is flowing in both directions (which it didn’t use to do), we’re going to have some real challenges. So we have to migrate the intelligence out to the grid edge, and then probably create some kind of permission-based approach so that everybody who needs to see the data with the requisite latencies, the speed with which they need to see it can tap into that and pull it into their systems that can automate most of the behavior and decisions necessary. And it all has to be super cyber-secure.

John: I’ve heard these companies that are pooling these types of resources together, and then they turn around and do a much larger scaled contract with the grid, this kind of virtual power plants, and so forth. And then they build all the technology you’re talking about that collects the data and sends customized devices to those different endpoints. So it can collect the data and we know how much storage is available or power is available, et cetera. I think it’s fascinating and also interesting how that’s all being added retroactively. We’re coming back to legacy infrastructure and adding all this cool new tech and digital, I suspect IoT’s part of that as well, right?

Peter: Yes. Very much so.

John: Yeah. I think that the whole concept of energy-efficient systems at scale and how that can fit into the consumer market, and having the energy consumer market essentially participate as a wholesale provider in the electrical system is something that’s going to be exciting to see. Where the electric car can actually earn you some money, by participating in things that would only be the domain of large grid-scale companies and so forth.

[28:23] What’s your perspective on building a scalable demand response program for green energy?

Peter: Sure. There was a little bit of DR already when I came on with some paper mills in New England, but essentially, I had a great team. I had 35 really talented, dedicated people. We essentially built our capability in-house to about 850 megawatts of dispatchable load, which would be a decent size nuclear plant, one unit in a nuclear plant. Then we bought sea power and essentially doubled our size. But most of that, John was us calling up a customer with say two hours of lead time saying, you got to shut down for six hours, maybe three times a year, maybe five times a year. Contractually, it was different, but these were formal programs with the grid operators, the independent system operators, and the regional transmission operators. That’s kind of like the baby steps of DR because it proved that there were elastic responses, customer responses with respect to price, and changing behavior, with respect to price.

But that’s really the baby steps, the training wheels, when chairman Chartajee announced in October of 2020, I believe it was for order 22, he commented on that. He cited two studies, one of them at 60 gigawatts of flexible resource capability, and another one at 380 gigawatts. Now to put that in perspective, our grid had around 1,177 gigawatts of installed capacity in the whole grid. And he commented that it’s a small, but quote, “Mighty resource.” Now the real challenge is that those flexible assets are out there. The water heaters, the electric vehicles, all that stuff. The question is how much of it is cost-effectively capable of being commercial. And then how do we make sure again, that we have that data system to make it happen, but let’s take one example. So many, many homes in this country have water heaters and it’s basically a giant thermal batteries.

We use electricity and we have a resistance heater. So we put the electricity through a wire that doesn’t like it and resists it and creates heat that warms up the water. And nowadays, most of the time we heat up that water without regard to when we heat it and the impact it’s having on the grid. In theory, we could actually go quickly on and off and provide even frequency regulation, like fast response services that the grid needs, devices responding really quickly to that 60 hertz of frequency in the grid by doing or not doing something.

Now, the problem is, it’s very difficult to do it cost-effectively in a retrofit situation because you have to go and talk to all the customers and then put a device in that governs the behavior of that water heater. On a go-forward basis though, there are only half a dozen major water heaters in this country. So we could, for example, equip them all the same way. An iPhone has like a USB port or something, and say, we’re going to make these all grid capable. So the real challenge right now is yeah, there are plenty of these devices, but the question is how do we break them down into the sub-sectors? How much of it is going to be retrofit activity versus how much of it is going forward, new tech that we can quickly make grid DR grid flexible, and capable?

Ultimately if you fast forward, say 10 years, water heaters only last about a decade or a decade and a half. So you could roll out that population, and turn it over. So part of the challenge is as quickly as possible, to start to build these new capabilities in, in anticipation of the value they can create for society. And there’s a bifurcation here, which is that end-use devices weren’t built to service the grid. They were built to do something else: to warm a house, cool a house, heat water, do something, cars, electric, fuel move people around.

Batteries themselves are quite different, though. They were designed specifically to absorb or release power. So you’re starting to see programs like in Hawaii, they’re now paying customers who put batteries in with their solar panels a really good tariff in the evening to deliver energy back to the grid. So batteries are a completely different subset than harnessing all these other end uses for the DR that we need. I think there’s a lot of potential here, but I think we need to take a very holistic view.

FERC order 2222 will help because the grid operators very soon have to come back with their operating models. Then there will be some tussle with the distribution utility saying, well, we need more visibility into what’s going on. And at some point, there will be some kind of a politically palatable settlement. And then this thing will start to move forward. Once the rules of the road are clear for vendors, I think they’ll ramp up and amp up their capabilities. And then obviously all the manufacturers need to make these devices capable from the get-go so they can play in this grid in the future.

John: Got it. Yeah. I get your point. It’s hard to take these existing consumer products or consumer infrastructure products and turn them into smart devices. But if you understand where the grid is going, we’ve had things like grid one, grid two, grid three, grid four, this whole evolution of what the grid is and the definition of it, then you can start building the end units differently and incorporating technology to support their participation in a more flexible grid. And that’s how you feel we get to the scales that we need is what you’re saying?

Peter: Yes. I think you really have to make all the new devices, as soon as possible, make them grid flexible and capable.