AI and Technology: Rethinking energy consumption

NARRATOR    Welcome to the DNV Talks Energy podcast series. Electrification, rise of renewables and new technologies, supported by more data and IT systems, are transforming the power system. Join us each week as we discuss these changes with guests from around the industry.

MATHIAS STECK    Hello and welcome to the tenth series of the DNV Talks Energy podcast. I’m your host, Mathias Steck. In this series we take a fresh look at the role businesses play in lowering the world’s carbon emissions and how they can work with governments, policymakers, and other key decision-makers to transition faster to a clean energy future. We open this new series by taking a look at the big picture in terms of global emissions and how new AI tools can automatically use energy when it’s at its cleanest. I speak to Gavin McCormick, founder of non-profit technology company WattTime. He tells me how a hackathon event helped him discover the power of software and data to dramatically reduce global emissions. Gavin, who is a behavioural economist, also encouraged a shift in mindset from blaming energy producers and users for emissions to empowering them to make smarter choices. We hope you enjoy the episode. Gavin, to get us started, it would be great if you could briefly introduce WattTime and yourself. What I would be especially interested in is, what is your personal motivation to engage in the energy transition?

GAVIN MCCORMICK    Sure. So, I think something a little bit unusual about me is that I did not intend to found WattTime. My backstory is that I was a PhD student at UC Berkeley studying energy and behavioural economics and my professor discovered a very interesting technique to measure how much different wind farms reduce emissions. So, it was not obvious to me that some wind farms, if you build them, then what will happen is, a coal plant will be replaced and other wind farms, if you build them, what would happen is, for example, another wind farm would be replaced. So, it’s a very different impact on the environment. So, I was working on a paper supporting this, and I lived in San Francisco and a colleague of mine said, you can’t live in San Francisco and never attend a hackathon. So, for fun I showed up to a hackathon with a bunch of Facebook and Google software engineers and described the work I was doing, and they said, wait a minute, if you know which wind farms are most effective, we could write software that would make it possible to deliberately use electricity at times and places that are better for the environment. You could know what time you should run your dishwasher or use your air conditioner and it would be cleaner. I thought this was a fun little hobby at first, but what I watched is that having studied this topic for nine years – in nine hours we were able to write software that could scan power grids, find out this information, and automatically inform devices of which times were better. And I was just so struck by the energy and enthusiasm of software engineers to volunteer to help with the energy transition, as opposed to the sometimes the slow speed of academia. I got really hooked on the idea of, can we move faster together to transition by getting software engineers more involved in the climate movement?

MATHIAS STECK    Right. I would like to dive a little deeper there because I watched a couple of your YouTube videos and you make a slightly controversial statement there at the first sight where you were saying, if I turn off my fridge or my light to save some energy, I’m not necessarily doing something useful to the environment. Can you explain that?

GAVIN MCCORMICK    Yes, it’s so surprising to me. I mean, I spent my whole life thinking about energy efficiency and conserving energy. Of course, that’s the same thing as helping the environment. What I’m seeing is that we now have so much renewable energy in parts of the grid, there are times when there’s too much renewable energy in the power grid. And if you don’t have energy storage to save it, it has to just get thrown away. So, now one-third of the time in California, for example, we actually have more wind or solar power than the grid needs and if you use energy at that exact moment, all that happens is, they waste a little bit less wind or solar. So, amazingly, that means if you conserve energy at that time, it actually doesn’t do anything for the environment. You aren’t reducing pollution, you’re just having a little bit less wind thrown away. That’s becoming more and more common as we begin to succeed at the energy transition, there’s more and more renewables.

MATHIAS STECK    Gavin, at WattTime you talk a lot about clean energy and dirty energy. Can you explain what you mean by these terms and why that matters?

GAVIN MCCORMICK    Yes. We use those terms very deliberately because one of the things we’ve learned is, people can get very lost in minor disagreements about what type of energy is best. So, for context, I’m a behavioural economist and I’ve spent a lot of time studying, how do grid experts take in new information and act? One of the things we learned is, it is so possible for two people who are both focused on the energy transition to get lost in an argument. Is nuclear power a good thing or a bad thing? Is hydro environmentally friendly? Well, what about if it’s a small hydro facility? What we’ve learned is, if you… instead of saying, I am going to define what is clean and tell you what to do, if you flip it and say, what do you think this is best and what would you like to do about it, it makes people move to action much faster. One reason I think that’s very interesting to us is, we’ve learned – here in America we’ve done a lot of studies and found that if you ask someone what clean energy means to them and you make it clear that they will have a choice to do something about it, 99% of Americans say that what they would like is renewable energy. That’s a very significant number because that is far more Americans than even believe in climate change. So, in a very politicized era where not everyone would agree to say that they care about the energy transition, what we’re seeing is, actually everybody agrees on doing something about the energy transition. If you just make it a little bit less about me telling you what I think and a little bit more about, okay, you tell me what you want to do, and you mean that in an authentic way, actually many more people agree than is commonly known.

MATHIAS STECK     Gavin, I’m aware that you developed the technology “Automated Emissions Reduction”. Could you briefly explain what that actually is and what it does?

GAVIN MCCORMICK     Sure. The idea is, if certain times now using energy actually doesn’t cause environmental harm, it would be really nice to use energy at those times. And I think it’s not generally known how many devices on the power grid now have an internet connection. Many people have heard of the Internet of Things and smart thermostats and electric vehicles, but probably are not aware of just how many there are now. So, we now, I believe, just crossed 26 billion smart devices all over the world that have an internet connection and can be set to automatically use energy at certain times. So, our idea of “Automated Emissions Reduction” is, let’s detect, what are those moments where there’s too much renewable energy? What are those devices that you can automatically set to use energy at certain times? Let’s have them use energy at the cleanest times that are convenient instead of using random times. So, my favourite example is a refrigerator. Okay, most people don’t have a smart refrigerator yet, but they’re becoming more and more common and soon many people will have one. Most people don’t care what time their refrigerator uses energy. They just care that their food is cold. In fact, there’s a lot of randomness in when a refrigerator turns on, turns off, is cooling, and “Automated Emissions Reduction” would say, why doesn’t the refrigerator specifically cool at moments that there’s clean energy instead of just doing it at random?

MATHIAS STECK     So, if we look at the big picture in terms of the implementation of this technology you’ve talked about throughout a larger grid, what does it require for all refrigerators to know, “I have to use energy at this particular time?” Does this create a new challenge? What is needed to roll this out at the energy ecosystem level?

GAVIN MCCORMICK    So, what this looks like for a consumer level is just like when your phone automatically updates or your laptop automatically updates. It’s becoming much more common that your car could automatically update or your thermostat could automatically update, but in order to do that successfully, you would need to have software that is scanning every power grid in the world every five minutes and detecting, “Is right now a clean time or a dirty time?” It would need a forecast of, “Is the next five minutes going to be cleaner or dirtier?” You would need a way for that intelligence to get to the software companies and even more importantly perhaps, you would need every one of those software companies and the electric utilities and the regulators to understand how the whole system works. So, it’s a surprisingly small project from a software perspective, but this is so different from how power grids have been run before, that for the whole world to begin thinking this way. There’s a lot of understanding that needs to happen in order for people to begin realizing that increasingly certain times are much cleaner than other times.

MATHIAS STECK    There is another aspect where I would like to drill a little deeper. You said this way of running a power grid would be very different from how we have done this in the past. I think there are also different ways how we look at emissions. You could say if there is a grid that is 50% wind and 50% conventional, then the emissions are just some average out of that emissions the generation is emitting, but we have this concept of average emissions versus marginal emissions and your technology makes use of this. Could you explain that in a bit more detail to us?

GAVIN MCCORMICK    Sure. So, I would say in the past, who were the people that spent most time counting emissions? It was carbon accounting companies and corporate sustainability teams and their tools were primarily focused on, whose fault are the emissions? So, there’s a long history of averages being used as a way to decide, if two companies live in a power grid and they each use half the electricity, whose fault is it that there are emissions? You could say, okay, let’s cut it down the middle. If they both use half the electricity, then half the pollution belongs to this company and half the pollution belongs to that company, but that framework is very focused on blame. It’s very focused on the idea of, it’s important that we understand who is to blame for the pollution. And a very different way of looking at it is to say, how can we solve pollution? If you look at it that way, you can instead say, okay, if this company takes this action, what will happen in the real world as a result? If they build a wind farm, if they install an energy efficiency programme, if they change  electric vehicles, how is the world different in the real world? That idea would be called marginal emissions and it didn’t use to be such a big deal, but now that we are starting to see bigger and bigger changes in power grids, now that we’re seeing times, for example, that there’s no pollution at all, knowing what’s the real effect of doing something is becoming more and more important and we call that marginal emissions.

MATHIAS STECK    I would like to come to your background also as a behavioural economist. We have the COVID-19 pandemic. We are in the middle of it. Consumption patterns have changed significantly in the corporate world, but also in the private world. Can you see any proof, if you use the concepts you have just explained to us, that there was in fact a real impact of COVID-19 on the electricity consumption patterns?

GAVIN MCCORMICK    Yes, very much so. So, it’s actually a wonderful example of how marginal emissions work. So, for example in New York state in America, they’re their own power grid, so we’re monitoring what’s going on with all their pollution and all their electricity use. We saw about a 7% drop in the electricity that people in New York state use from COVID, so that’s great. A little bit more people working at home saved about 7% electricity. We saw about a 45% drop in the pollution. That’s a huge difference, 7% drop in use, 45% drop in pollution. What we saw is, it’s because, which are the power plants that turned off? It was the very dirtiest power plants of all, that were the ones that actually were turned off when people began using less energy. So, we think that’s not going to entirely last. A lot of patterns are likely to go back to normal and to some extent, nobody can know for certain, but we are seeing clear evidence that some of the changes, for example, in transportation patterns, look like they’re permanent. We’re seeing evidence in countries like China where transportation – the pandemic is practically over in China – and we are seeing that transportation patterns are not going totally back to normal. So, we think some of the benefits of people using less electricity and energy are probably here to stay.

MATHIAS STECK    Coming back to the energy transition and what technologies can do to this, you may be aware that DNV has also an Energy Transition Outlook where we unfortunately come to the conclusion year after year that the carbon budget we have, will be consumed much earlier than it should be. What do you think these new technologies - which allow us smart use of electricity -how much can they help to actually change the situation?

GAVIN MCCORMICK    I think they can have a really significant impact and I keep coming back to what Al Gore once said about climate change progress. It always starts slower than you think, but when it really gets going, it always moves faster than you think. I think we’re going to see something similar here. So, for example, in the first generation of “Automated Emissions Reduction” algorithms, we got about a 1% reduction in pollution through software. That’s nice, but it’s not a big deal. On the latest generation of software in California, we saw a 100% reduction in pollution in energy storage. There was completely no carbon footprint from energy storage running this software. Now, we don’t think it’s all going to be 100% as of today, but I think that’s a really good example of how the technology is just continuing to grow more powerful as we also have more and more renewable energy that is being wasted and you can use in this way. So, we think it is possible to have about 300 million tonnes of CO2 per year reduced with this kind of technology, and with related technologies like optimizing wind farms better with this technology, there’s even potential for four gigatonnes of CO2 reduction per year. So, for context, that’s as much energy as Europe uses in terms of emissions. So, we think that over time there’s really significant potential.

MATHIAS STECK    So, if you would be a utility and you want to get that going, what would you need to do to get there and get there fast with your clients?

GAVIN MCCORMICK    I think one of the things that we’re pointing out here is that we’re talking about clever uses of data and so you need two pieces of the puzzle. First of all, you need the data. So, you need to know, okay, which times are the cleaner times and where, because it depends on the location. The other thing, though, is that you need help to understand, what are the programmes you can do with this? There are so many different things you could do with this data. One thing you could do is, you could have a demand response programme that instead of just paying attention to moments when energy is very peak, it can also think about moments when energy is very dirty. And it’s pretty easy, if a utility really understands the data and has expert guidance, to change out a demand response programme to get about 72 times more emissions savings. Another thing you can do is begin to optimize electric vehicles and smart thermostat programmes. A third example would be redesigning energy efficiency programmes to deliberately save more energy at times of day that are more important. So, thinking about things like, if you retrofit a house and it is more energy efficient, well, does the house tend to use more energy in the morning or the night? These are important questions to know. So, we think that the one-two punch for a utility is getting data from an organization like WattTime and then any guidance on how to think about those data. And that’s actually one reason we’re so excited about the DNV partnership because we think that’s a really good way for organizations to get guidance on, great, I have information, what do I do with it?

MATHIAS STECK    Yes. Maybe you could tell us a little bit more about that, the DNV partnership, and how you are exactly contributing there to what we try to do to give clients a better insight also on their emission data.

GAVIN MCCORMICK    Sure. So, one of the really interesting things about this partnership is that DNV just has so much expertise in so many pieces of the energy transition. So, we often work with companies, for example, that specialize in electric vehicles. So, that’s great. We can clean up electric vehicles, but that’s just one slice of a much much bigger energy transition story. What we think is so interesting about DNV is, people come for everything from optimizing wind farms to thinking about utility planning. There’s a wide variety of different topics for which DNV general expertise about energy paired with emissions expertise, allows for redesign of programmes. The reason I think redesign is so interesting is that what I’ve learned as just as a climate activist is, it’s really hard to start something from scratch. If you have an idea for building something entirely new, it’s going to take years, but if you can find those opportunities where utilities, governments, companies have a programme that, with the right expertise from people they’re already trusting, could be doing more impact faster with a slight tweak - we think that’s how you can really accelerate the energy transition fast. So, that’s why we’re really excited.

MATHIAS STECK    Right. So, I’m obviously excited as well here. I wouldn’t want to miss out to talk about another topic where you are engaged and this is the Google “AI Impact Challenge” to measure carbon output from large power plants. Can you tell us a bit more about that project and what the impact of that may be?

GAVIN MCCORMICK    Sure. So, the connection is, if you think about, how is it that we are able to know what are the times when renewable energy is being thrown out, it’s because we need access to very high quality data on what’s going on in power grids. I didn’t know this when I started this project, but the United States has far better open public data on power grids than anyone else in the world. It’s funny. We don’t always think of America as the leader in environmental innovation these days, but back in the 1970s the United States EPA created a system to monitor every fossil fuel power plant in the country every hour and provide the pollution and generation information to the public. As a result, technologies like ours have been possible. When we went to go build this in other countries, what we discovered is, there isn’t the same quality of open monitoring. There are some countries that have some information, but it’s not a global system of that. So, we partnered with a lot of other non-profits who are interested in also accelerating the energy transition with a lot of data. For example, Carbon Tracker in the UK identifies, what are coal plants that would be just more profitable to just shut them down, they’re just losing money? And they realized that they wished there was a system like the US EPA had in every country in the world, same as we did. So, we applied to Google together as a group of non-profits and said, what if we were able to apply artificial intelligence to satellite imagery of the entire world to look from space at what’s going on in power plants and produce a global system to measure what’s going on at every power plant all day long all over the world, and make that data available to the public so that technologies like ours would be possible, but also regulators could know what’s going on in power grids, the Paris Agreement negotiators could have a clear sense that their data have been validated externally, and a lot of different applications could just move a little faster because we had very high quality global monitoring of all power plants.

MATHIAS STECK    Right. You just mentioned stakeholders from the regulation side already. We have now an unparalleled insight into energy systems. What would you think the impact on how we are organized in the energy industry could be? I’m thinking about, now we have countries looking after their grids, maybe continents. What you have just described would open up much more global interconnection of the whole energy ecosystem. What would be your hunch how the future would look here like?

GAVIN MCCORMICK    I think that there’s a lot of really interesting things going on about integrating the markets of trading electricity and those have generally been ahead of integrating the environmental component of power grids. I think what this project opens up is understanding - suppose that you are consuming power in Spain. Some of your power is coming from France. Some of your power is coming from Portugal. The more you understand about the environmental implications of making certain choices, the more you can begin to say, how do we not just build a few wind farms in Spain? How do we make sure that those wind farms have the maximum effect possible, and instead of just replacing and turning off wind farms in France, they’re turning off coal plants and helping us all move faster together? We think this type of data is going to make it possible for grid operators and regulators to think about not, what’s the effect on my little town, but, what is the effect I have on my neighbours and the effect they have on me, and how can we have a system that is optimized together to have even more benefit from similar regulations?

MATHIAS STECK    Gavin, you have just given us such a fantastic outlook and vision about what’s happening in the future. To get a bit more reality to this, what do you think will happen in the next ten years?.

GAVIN MCCORMICK    I think we’ve seen even in the last five years, particularly in the United States, just a very rapid move from what I would call the first phase of the energy transition to the second phase, which is really moving away from coal. So, in the last five years we’ve moved from power grids where about half the time you’re getting coal energy if you flip on a light switch and about half the time you’re getting natural gas, to it’s almost always natural gas and increasingly often it’s renewables. We think in the next five to ten years you’re going to increasingly see a change that’s already starting to happen in Europe and maybe in California and the US, to most of the time when you flip a light switch you’re actually turning on renewable energy. We think that’s very significant because it’s going to mean that the energy transition changes from a story about, let’s just build more renewable energy, to, we actually have enough renewable energy, now it’s a question of optimization, it’s a question of timing. So, I think ten years from now, in most places in the world, it will be possible to have no carbon footprint at all if you are able to use software to have energy at exactly the right times and I think that is a bigger change from how it works today than people have yet realized. I think it’s all… We’re seeing that it’s starting to move very fast.

MATHIAS STECK    Thank you, Gavin, for these great insights. I have one final question to you coming back to DNV’s Transition Faster Together campaign. What would you think is the most impactful enabler for us to actually transition faster together to a clean energy future?

GAVIN MCCORMICK    I think the most impactful thing I’ve seen is to move away from thinking, particularly in corporate sustainability and any organization that’s setting their own sustainability goals, to move away from just thinking about, how do I clean up my carbon footprint, to, how do I clean up Earth’s carbon footprint, and realizing that with better data now there are many opportunities, for example, to build a wind farm - maybe not across the street if you are in Sweden, but instead build it in Poland. If you have a multinational company, for example, with that capability, often the same wind farm can be several times more emissions-reducing in the real world. And there are these opportunities for big emissions savings by just thinking a little bit less about my carbon footprint and a little bit more about my effect on Earth’s carbon footprint. We think there’s a lot of low-hanging fruit to move very fast that way.

MATHIAS STECK    Thank you very much, Gavin, for these great insights.

GAVIN MCCORMICK    Well, thank you so much for spending the time today. Really exciting and take care.

MATHIAS STECK    Thanks for joining us for this week’s episode. It was fascinating to hear about WattTime’s work and listen to Gavin’s insights into the role of data in helping to redefine how we think about energy consumption. In next week’s episode of DNV Talks Energy, we speak with Antonio Cammisecra, Head of the Global Infrastructure and Networks business line at Enel. He’ll discuss some of the megatrends in renewable energy as well as how investment in energy infrastructure will help drive the energy transition. To hear more podcasts in the series, please visit dnv.com/talksenergy .