 Hi, everyone. Thank you for coming on this kind of dismal day. Please turn off your cell phones. Please turn them all the way off. I'm going to now introduce to you, OK, I'm going to say this. Mads Almasalki. Did I say that right? I keep practicing. So Mads, I'm not going to say his last name again, is the L. Richard Fisher Professor of Electrical Engineering and Associate Professor in the Department of Electrical and Biomedical Engineering at the University of Vermont. His research interests lie at the intersection of power slash energy systems, mathematical optimization, and control systems, and focus on developing scalpable algorithms that improve responsiveness and resilience of energy and power systems. Prior to joining UVM, he was lead systems engineer at Energy Startup Company, route three technologies in Chicago, Illinois. Before that, he received his PhD from the University of Michigan in electrical engineering systems in 2013, and a dual major in electrical engineering and applied mathematics at the University of Cincinnati in Ohio in 2008. When he is not working, which I'm thinking isn't often, on energy problems or teaching, he spends his time with his amazing wife and their three wonderful children. Please give a warm welcome to Mads. It should be online now. OK, good. Now I come from above. Can I turn down? Oh, you're in charge. OK, good. I just look like I'm in charge, but Travis is in charge. So I'm really excited to be here in the dark and talk about renewable energy. As you may have heard, I am not a policy wonk. I'm not a politician. I am a researcher who is very interested in how to enable renewable energy through technology. Those technologies is what I term grid flexibility, and I'll explain what that means and how it figures into renewable energy. The other thing I want to mention is, as you heard, I've been in Ohio, I've been in Michigan, I've been in Illinois. When I joined the University of Vermont, I had never been to the Northeast, except for the interview. But my wife had never been either. But what was really exciting about Vermont is that it's well known for energy topics. The complex ones, the interesting ones, the challenging ones. And as I knew before I came, and what I know now, is Vermont is really kind of a little testbed, where things happen pretty quickly. Ideas are tested quickly, sometimes shut down, sometimes lifted up. And what we learn from Vermont is applied to the country as a whole. And so when I travel to different conferences around the country and around the world, they know about energy in Vermont. And they always have an eye on what's happening in Vermont. And so it's a really exciting place to do research. Because the questions we can ask at the university has a chance to actually bubble over the walls and into the communities we live in. And so, like many of you, I live five minutes from here, which is, I think, a South Burlington special. And so to highlight some of those exciting topics in Vermont, so this is across the US, the first efficiency utility, effectively selling what's called negative watts, savings on energy, was in Vermont. So that's efficiency Vermont out of the EIC. The first utility in the US to be deemed 100% renewable, deemed through political action, is Burlington Electric Department, first in the country. And Fast Company, which is a technology magazine, deemed that Green Mountain Power was the number one company in the world when it comes to energy innovation in 2018, number five in 2019. And this is in the world of all energy technology companies, including Tesla, Google, Amazon. And so Vermont has a footprint that far exceeds its little state. And we can also see that on a picture like this. What we see is Vermont as a state, how much power Vermont consumes during a day in April. So the green line is a cloudy day on Wednesday, April 5, 2003. The Monday, April 10, is a sunny day, both of them in spring. What is this difference? That's the solar panels on roofs in the fields across the state. That's a lot of solar. If we predict a cloudy day wrong, we're going to have a pretty big jump in power. And so if you see the y-axis here, the Vermont consumes roughly 6,700 megawatts of power at any given time on a cloudy day. And you can see that on a sunny day in the middle of the day, we get down to 200 megawatts. What has happened, actually, very recently, and I don't have the data for this, what actually has happened is that this blue line now goes down here. And so it happened for the first time ever. Vermont exported solar energy to New England. So Vermont, the sunny state, is now exporting solar energy to New England on a sunny day. That's pretty awesome. And so that's new. It's very exciting for us when we think about that's a pretty big achievement for a small state to do that, to basically become a generator for another section. And so part of that excitement around what Vermont can do within energy has bubbled into really good collaborations at the university. And so we have a relatively large Vermont style large. Interdisciplinary research group at UVM that works on all kinds of what we call resilient energy and autonomous systems, systems that can act and behave on their own and respond to certain challenges on their own. So that means we have to develop and design lots of algorithms. And when you design these algorithms, you need to know whether they work well, when they don't work well, and all kinds of other engineering problems. And so we've been lucky that we've been able to collaborate with lots of companies and federal partners in this endeavor. And so we have lots of local and national collaborators, industry collaborators. We have lots of federal funding partners. Kind of funny. I have worked with NASA on swarm satellites. Worked with the Sloan Foundation on energy projects in Alaska, Department of Energy. Worked with New York, Vermont. It's really fun. Pacific Northwest, the Washington state. We've also been really lucky that the work we have been developing through various academic papers, research projects, we have some patents that we filed. Actually, we're able to build a company out of it. And so that company was called Packetized Energy. So we formed this company in 2016 with a couple of my colleagues at the university. And that company was able to get other folks excited enough about our technology and about what we have achieved so far. So they actually acquired the company. And so Energy Hub, which is a large demand response provider in New York City, effectively acquired Packetized, hired all the employees, and opened up their R&D headquarters in Vermont as a result. And so Packetized Energy has now become Energy Hub. And we have an office down off Pine Street in Burlington. We, meaning them. I'm no longer working with them. But what's really exciting is that the algorithms we worked on, the students at the university, essentially now have access. The algorithms have access to 1.2 million devices across the country. And those devices basically help integrate renewables, which I'll explain in a second. Beyond kind of the translational research aspect, which has been really amazing experience, very hard work, if you ask my wife about it, she has a different story than I do. But everything is possible when you work together. The other things we're doing at the university is we're building research facilities. So we've shown the country. We've shown our funding partners, the federal agencies, that we can do really cool stuff at the university. So now we're starting to basically invest in our research infrastructure, going from algorithms to hardware, essentially. So at the McNeil plant, if you see solar panels popping up at the McNeil plant, that's our new research center. If you pop inside the new Discovery Building and go to the fourth floor where you can see the mountains and wind turbines out in the distance, we're building a new accelerated testing laboratory where we get to get the actual hardware in there, build models, gather data, better understand how they perform in extreme operating conditions. As the climate changes, the devices we put in the field, we need to know how they perform as climate changes. Extreme summers, extreme winters, extreme heat, extreme cold. So we're building that lab at the university, too. That's the hardware side of things. We're also going to build a laboratory facility for the software side of things. And so if you're familiar with campus, you may have heard of Torrey Hall. Torrey Hall is undergoing renovations at the moment, and at the top floor, we're gonna be given this entire floor to build a laboratory for software, which is kind of silly because software sits on a computer, not in a room. And so what does that lab look like? It's a wall of monitors. And so this is a drawing made by Chat T.P.T., right? This is not real. But the idea is going to be, we're gonna have a wall of monitors where we effectively gather data from our industry collaborators in Vermont. Velco transmission companies, different utility companies. We'll get this data in here, and we can build models of Vermont's grid, and we can see how those models, how the grid behaves in real time. And so we'll have a digital twin of Vermont's power system. And so that's the idea. It's coming 2024, which is exciting. And so that's a little bit about what we've been doing at university. And so as a professor, and I think Cindy told me I'm in charge, and so I get to throw in a pop quiz. And so it's a very simple, it's one question. It's one question. I think you're probably the right audience for this one. What's the greatest engineering achievement of the 20th century? And I'll take, you don't have to put your hands up. The internet is an option. Not according to the National Academy of Engineering. A good guess. I would have thought that one too, but it's not. Close, very close. So this machine that spans the entire continent, electricity I think was the 19th century, 18th century. So it's very close. Yes? The grid. So electrification was deemed the greatest engineering achievement of the 20th century. Do you know what happened roughly a year or so after this announcement? It did. The 2023 blackout happened. And so what we've been used to, and I guess now I appreciate the darkness. This is electrification, right? Electrification is the backbone of modern society. It's what enables healthcare systems. It's what enables the internet. What enables computers and transistors to function is they have access to affordable, reliable energy. And so when we talk about the 20th century, we need to appreciate that this is the greatest achievement of the 20th century. This is electrification. This is Tesla's vision, Nikola Tesla, not the new one. Nikola Tesla's vision of how power from far away can be transmitted long distance at high voltage all the way to people's homes and businesses. And so this is Tesla's vision of low voltage, safe electricity at the house, delivered from a distance from very large, efficient-ish thermal generators, at least large-scale generators. This was the 20th century. The 21st century is different. We heard someone in the crowd mention solar power. That's different, right? Suddenly, I don't need that generator to produce my electricity all the time. I have local generation. In Vermont, we have a lot of this. We saw that earlier. On the mountain tops, we have wind turbines. We have policies, right? The Yankee nuclear power plant, shut down. Not because it didn't work, but because politicians, politics, said it shouldn't be there. We have electric vehicles. Most houses, the grid was designed with a rule of thumb that said every home has a fridge. Let's design the grid to match a fridge in every house. An electric vehicle is like eight fridges. We have smart thermostats. I don't know how many of you have the Google Mests, Honeywell, all kinds of thermostats, making decisions in real time about how your room, your house, your floors are being heated and cooled. If you heard about GMP's battery programs, we have batteries popping up across the grid. This is not just Vermont, this is more nationally. And so it's very reasonable to say, the times are changing. What was the 20th century's greatest achievement? Doesn't look the same anymore. This is not the 20th century's greatest achievement. Things have changed. The answer is blowing in the wind. This is chat GPT. I can't tell if he's tucking it in or blowing it out. So this is generated with AI. I just asked it to make a picture of someone that looks like Bob Dylan, blowing at wind turbines with a city in the background, and it did this. It's quite beautiful, actually. This is another AI-generated image that really shows the crossroads that we're at. On the left is kind of the 20th century, a little bit of the 19th too. On the right probably won't look like this in the future. We're not gonna have wind turbines all over us. But we really are at a crossroad between a system that was and a system that will become. And I'm not talking about 20 years from now. I'm talking about in the next five, 10 years. And so the reason for this is that solar power today, electricity from solar PV, is the cheapest source of electricity in the history of humanity, period. If you want access to the cheapest electricity, solar PV is the way. Maybe not for you and your roof, but for you and your system, the electricity system we're in today. And so what is going to happen is that solar PV will be installed at a rate globally that will outmatch in about five to 10 years. It'll be the largest power plant. Sorry, the largest capacity for power in the world will come from solar. So IEA says 2027, let's say 2030. Capacity, right? We also know that at night the sun doesn't shine. So if you have a lot of capacity, it doesn't matter at night. So this isn't energy, this is power, which only works when the sun shines. Wind is also going to blow. More and more wind will be installed. But the fact of the matter remains, solar PV is the cheapest source of electricity ever in the history of humanity. And it's just gonna get cheaper. And so what this means is that we are on a path to decarbonize the power sector. Again, Ember is another, you can find reports for almost anything. These are very good reports, right? The source is IEA, which is the International Energy Agency. What they are arguing is that by 2040 we'll decarbonize the power sector. Which isn't unreasonable if the cheapest source of electricity is wind and solar. And it is. The other good news I have for you is there's lots of job opportunities, right? So if you're looking for a job, lots of options, maybe a grand kid or your kid. Electrification is taking off and workforce is advancing rapidly in energy sectors. So I think it's something like the US job market of the last year or two, roughly a 3% growth in jobs, workforce development, in energy is 4% on average. So there's enormous opportunities. And what that means is that the US, the green economy is growing, it's growing rapidly. These numbers are large. Say a whole lot, but they're very positive. It's a good thing, right? But unfortunately, so are climate challenges. If you look at this chart, what it tells you is billion dollar weather related disasters in the US. So each of those on the y-axis, you have the number of events. There's a number of cumulative events over a year. And so of course, as the year goes on, you only have more and more events. And so these things are always rising and then the colors represent different years. The gray ones are historical data. The black line is the average across the last many, many years. These are the last couple years, right? The weather is causing more and more billion dollar disasters. It's undisputable, it's just the number. Maybe inflation helps with this, of course, but it's certainly the number of events is quite significant. These are weather related events. Familiar with Puerto Rico in 17. Pakistan globally, world records being set every year. 2019 was another climate disaster. 2020 horrendous, Australian wildfires. 2021, there's some good news, right? Don't worry guys, it'll be slightly less hot than 2020. And they were right. Slightly less hot, what happened? 2020, the big freeze, the big freeze. The Texas grid went under. What was the cost of that? 100 billion. So good job, right? Less hot, it became a lot less hot. 2022, this is just, you can just find these, these are not that special. 2023, we started seeing these new surprising results. So we have the Hawaii fire, which was horrendous. Global temperatures are now exceeding what we thought was gonna happen, right? We're surprising climate scientists, that's not good. This should surprise us. We should not surprise them. And so we're setting all the wrong records. And so now that we've gone through these pictures in a row, I'm sad, right? So let's take a breather, take a pause. Also AI generated. As an engineer, the first thing I always ask, how can we fix this? My wife hate that because I apply what I do at work, sometimes at home. And so engineers are sometimes very bad. Problem solvers at home, very good problem solvers at work. And so the question becomes, what are solutions? Luckily the solutions are where we are going now, right? 10 years ago that went out of an obvious because the cost of solar and wind wasn't the cheapest. Now it is. And so the solution to drive down greenhouse gas emissions is lots and lots and lots of renewables, right? Conventional abatement technologies is kind of the formal word. Nuclear power is in here. Almost certainly will be part of the future, whether it's in Vermont or not, I don't know. Other things that will happen is that we'll have carbon removing technologies probably in the foreseeable future. But the matter of the fact is, we're gonna need not gigawatts, but terawatts of renewable generation across the country. And integrating terawatts of variable renewable power is really challenging. A lot of grid operators are not sure how we will do that. So technically speaking, it's a hard problem. Economically, it's trivial to take the cheapest source, put it on the grid, and it's clean. And so myself, having been a power engineer, a power systems engineer, we didn't used to be that popular at cocktail parties. Now we get invited to speak at these events. And so what's really happening is that power engineering is really becoming climate tech, climate technology. And I'll show you why in a second. Some of the challenges that renewables bring, so this is the McNeil solar array. We have data from it streaming to the university now. We have four days of, here you see four different days. It's kind of like my mood swing. On a sunny day, things look pretty good. Fairly clean energy coming at a fairly predictable path. On a cloudy day, imagine that like across the state. Imagine that picture, what I saw in the beginning, right? Across the state, it's a lot of dancing. What is that? This is me on a Friday night, I'm done. I'm tired. You can't supply anyone with this little energy. And then Saturday I come back up, but still cloudy. This is pretty challenging. This is a relatively large solar array at McNeil in Vermont, in Burlington, which is why we're studying the effects of challenging Vermont conditions. Trying to understand how weather affects it, how to improve solar integration, maybe by batteries, maybe by other means. And the other thing is that when things are varying on the supply side, the grid responds. If you have a grid and you have on one side sources coming in and the other side, people consuming the energy, the network itself will experience lots of swings. At the core of the grid we have today in the U.S., 60 Hertz is the language of love, right? If you are at 60 Hertz, the grid is good. In Europe that's 50 Hertz. This is a Danish outlet. I'm born in Copenhagen, Denmark. And so I always appreciate how the happiest people in the world have the happiest outlet. And so if we want to achieve 60 Hertz and we have wind and solar kind of blowing at will, uncontrollably all the time, we want to maintain that 60 Hertz somehow. And so we have to control that frequency. We have to manage things to cancel out the variability. And so think of the grid's frequency as kind of like your Fitbit heart rate. So the grid has a frequency, I have a heart rate. I can monitor my frequency, sorry, my heart rate. I can monitor my heart rate and there are good zones in the middle. There are zones when I do too many things and if it gets too low, that's not good either. And so there's kind of a range of frequencies that are good and there's a range that are bad. So if the frequency falls below this range or above this range, then the grid is in trouble. And so this is actual grid measurements of Burlington, Vermont, so Burlington, Vermont frequency basically. So it's supposed to be 60 Hertz, right? I don't know how many of you are aware of 60 Hertz signals. We know it's supposed to be 60 Hertz, but it never is 60 Hertz. And so this variability has always been there. When I turn on or off lights, if my lights were big enough, you could measure it. Every single little fluctuation in electricity consumption changes the frequency, which is amazing because we're all connected on this network and so me switching on a light bulb basically creates a mini-miniscule change in the grid's frequency. If we create too many changes too quickly, things will get too fast. The grid will move too fast or too slow. Usually it's too slow. So when you lose a generator or when inverters trip and you lose a lot of solar generation, then the frequency can dip very quickly. But it never really sits at 60 Hertz, right? That's a textbook exercise. And so what we need to do is as we put more and more solar on the grid, regulating this frequency becomes important. And so we need resources that can flex with the variability from solar and wind. And so we need flexibility to cancel out variability. And so I'm not sure your age group here. Salt and pepper, does that say anything? So what we need to do is we need to talk about flexibility. And so what's happening is the need for flexibility will rise the more renewables I have. And so there's some really neat reports again. International Energy Agency makes really nice plots. And so if you look at the flexibility needs of the grid in the U.S., what we have today, we need flexibility, this much flexibility. Going to 2030 with all that renewables, we will need roughly double the amount of flexibility. If you look at Europe, China, India, the answer is the same. More flexibility is needed to handle the variability. Roughly double, we need double the flexibility. Where does that flexibility come from? Historically, this flexibility that we see today, most of it comes from big power plants, thermal power plants that ramp up and down to ensure that the frequency stays close to 60 hertz. If we keep closing thermal power plants, they will not be the ones regulating the frequency. And so if we look at where will this flexibility come from? Today, most of these things, most of this, this is power plants. Power plants providing the flexibility. If we go into the future, what you see is that this little sliver here today becomes this big chunk tomorrow, tomorrow meaning 2030. And so power plants provide roughly 90% of flexibility needs that the grid needs today in 2050 demand response and batteries will need to make up 50% of that flexibility. That's a lot. And so when we look at where that comes from, we need to look at different sectors, right? We have the energy sector, we have transportation sector, electric vehicle charging will be super helpful. We have buildings that are electrifying. Yes, sir. What is the demand response? Very good question. So demand response is when your devices, so have you heard of the water heater program that GMP runs? Are you familiar with that? The GMP turns off demand at certain times of the day. BED does too, right? And so when demand responds to grid conditions by some kind of external control signal, that's demand response. And I'll share actually a little bit more of that in one second here. And so demand response is really people's electrical appliances responding to some kind of signal. GMP's signal is specifically designed to combat what's called peak demand. And so I'll show that in a second, I'll show it. And so what we want to do is we want to have renewable energy, but we want to use all those renewable kilowatt hours as efficiently as possible. And so we're gonna combine renewables and efficiency because then we need less renewables and it's less costly, less variable, less flexibility needs. Then we're going to make some of these appliances, whether they're electric vehicles, batteries, heat pumps, water heaters, we're gonna make them flexible. And by making them flexible, we can help kind of filter out the variability that comes from renewables. And so if you listen to what the government, the federal government wants, is we're gonna move from what's called 50 gigawatts of demand response, which is basically low appliances turning off in response to something. It's gonna move to become 200 gigawatts of something much more controllable, like a big, big battery across a very large geographical scale. And if you're managing 200 gigawatts of some form of demand response, 200 gigawatts is not my fridge, it's not my refrigerator, it's not my heat pump, it's not my air conditioner. It's millions of them. And so we need technology that can coordinate millions of appliances, effectively in real time. And so this is the answer. Demand response comes from controlling and regulating and managing water heaters, electric vehicle chargers, thermostats, heat pumps, grid edge batteries, pool pumps, irrigation pumps, don't have a lot of pool pumps in Vermont, but Florida has a lot. PV inverters, which is what connects your solar panels to your house, they can also do really cool things. Even refrigerators, all of these devices, most of them today, have chips in them, right? Computer chips, transistors that were mentioned earlier. Do you know why they have chips in them today? Besides the computer chips, do you know why they have apps? Everything has an app today. Do you know why? Because the sensors are dirt cheap. So connectivity is enabling demand response. Connectivity will enable flexible demand. So if you look at sensors, this is just over the last 15 years. Sensor prices have dropped dramatically. I was reading an article, this was a couple years ago, and there's this beautiful discussion in this article. This was a technical article at a journal. It's this beautiful discussion about how unrealistic an algorithm was, because back then, in the 1970s, when the article was written, that chip they needed for their algorithm cost $60,000. $60,000 for one chip. Definitely not gonna be installed on my refrigerator. Today, I can go on Amazon. I can buy that chip for $10. I can then install that chip in some kind of an internet of things device, some kind of a device, and attach that to my fridge. And so that's roughly a $100 solution from a $10 chip. But that's totally possible, financially feasible, and not a bad business model. This is basically what we did with packetize, is we used the fact that chips are dirt cheap today. And when chips are cheap, they're everywhere. We've seen it, right? My watch, my phone, my keys. Chips are everywhere. That means everything is connected. With a little bit of control, I can turn every device, every home, every neighborhood, every town, every state, every county can be turned into some kind of a dispatchable resource. We're done. I finished, I made it. And so what we're seeing today is, this is Burlington Electric, got a little grant from the government. And actually a very nice grant. Advancing what's called grid interactive efficient buildings in Vermont. Managing water heaters, managing thermostats and electric vehicles. I'm sure many of you have heard of the O'Brien neighborhood in South Burlington. O'Brien neighborhood is gonna be equipped with EV chargers, solar on the roof, batteries in the garage, heat pumps, all of which are dispatchable. Each house acts effectively like a little battery. In unison, the whole neighborhood becomes a dispatchable resource. And so that little house I have here that looks like a house of batteries, it's basically what our houses look like to the grid. Store energy for a short period of time. Maybe that energy is chemical in a battery. Maybe it's thermal in the air in our house. I can heat up, I can preheat my house so that I don't have to heat it in the next half an hour. If I know, maybe that there's not as much wind available, not as much solar available. And so you can dispatch resources to optimize economics, carbon, or eat as much renewables as you can. The other thing you can do is you can participate with this flexibility in wholesale energy markets. If I were to take my fridge and try to bit it into the market, horrible experience. I don't have enough flexibility on my own to do it. But if I take a Brian neighborhood or a hundred O'Brien neighborhoods across the state and I start playing in the electricity market, that's pretty valuable. There are all kinds of services today available for participating in wholesale markets. And you can generate roughly $100 to $1,000 per kilowatt flexible per year at a scale that matters, which is why you're seeing companies called Tesla, Sun Run, Generac, and Energy Hub all playing in these markets or participating to save either utilities money or to make money themselves. Pretty amazing. Every house is a battery. Every neighborhood is a battery. Every town is a battery. Every state becomes a battery. Of course, that's the good news. Other news is that it's challenging. It's not super easy to manage a bunch of devices in real time to match with prices on the grid, to match with CO2 signals, to match with the sun and the wind. It's hard. One thing we didn't mention yet is that when I turn off your water heater, if you take one shower, you might be okay. If you take another one, it's gonna be cold. So we recently upgraded our water heater to a natural, it's still natural gas, but more efficient. I could not figure out how to electrify. It's super hard if you have an older home. And so I ended up taking a cold shower the other day because I had to work. It was really cold, like headache, cold. And so taking cold showers makes you very upset. And so if you are managing people's water heaters or their heat pumps and they end up being inconvenienced and uncomfortable, that's a losing war. You will lose that every time as a technology company. If you don't manage your resources well, you can also interact with the grid incorrectly. Imagine if you're supposed to increase the number of megawatts and you go the other way accidentally. Maybe somebody hacks your software and you go the wrong way. And so the grid conditions become important as well. So we have human constraints, we have physical network constraints. Business models are still evolving. I said that our market signals today, the market signals tomorrow are somewhat unknown. We don't know exactly what those markets will look like. So those risks associated with the business themselves, like any business, I guess. Then there's the other boogeyman. Cyber security and data privacy. We've had wonderful discussions in the past on whether data privacy is dead or not. How many of you have credit cards? Credit card companies basically know everything about you. And so if we complain about, oh, GMP knows something about me, is that real or is it perceived? Facebook knows a lot about us if we have Facebook. And so what we have to do if we develop any of these flexible grid technologies is we have to respect the human in the loop. And the reason for that is that energy is cheap and probably in some sense getting cheaper, at least by what we see. And so there are lots of examples of these demand response programs that you can sign up for. So I don't know if anyone is participating in the Burlington Electric or the GMP water heater program. I think you get $2 a month or something, right? $2 a month, how much is a cup of coffee? How much is a hot dog? How much is a beer? Right, we have problems. If we expect people to be uncomfortable for $2 a month, that's not a lot of money. And so what we're seeing nationally is that all of these demand response programs, if they're not run properly, essentially die on their own. And so this is in national grid, this is in national grid in the US. Basically 10% of people override events, 25% override any eight hour events, and these programs lose roughly a third of their customers every year. There's a really nice study done looking at thermostat behaviors, how humans interact with thermostats. And show that 50% of 27,000 people, statistically speaking, half of people will be annoyed and uncomfortable with just a two degree Fahrenheit change to their set point within 30 minutes. So if you guys go and change your thermostats within 30 minutes, now I have no flexibility to play with in the wholesale market. You make my battery worse, but it's really me making the battery worse by not keeping you guys comfortable. And so when you have this problem of very little margin for error and not a lot of money, it's really what we call the creamy conundrum. Quality of service is really important. That has to be guaranteed. Just like if you drive an autonomous vehicle, no one will guarantee a 5% chance of an accident. If you drive a self driving vehicle, no one will get in a car if it has any chance really of crashing on film. Just like this, we don't wanna crash our virtual batteries. Crashing means you guys become uncomfortable. I become uncomfortable. And so what we do is we need to encode quality of service into the algorithm. The algorithms have to be aware of people's need for energy. And that then raises the question of why don't we encode quality of service into the technology? How much flexibility do we really have? If I have to guarantee you're always comfortable, certainly I have to give up on flexibility. So basically we have a battery when fully charges when everyone is comfortable. So then I have stuff I can play with. I can turn things off for a little bit when people are very comfortable. If we're all taking cold showers, there is no flexibility left in that system. And so you can see an example of this. This is actually a local Vermont program on water heaters. So this is the total water heater power consumption of that fleet of roughly 2,000 water heaters. How much power they consume. So 2,000 water heaters consume roughly a megawatt at 8 a.m. Here is when the command comes and shuts off the water heaters for three hours. So that these water heaters are not part of any peak consumption. What happens after you shut off water heaters for three hours? Water doesn't get hotter, it only gets colder. And so when you suddenly say, the demand response period is over, what happens? When you allow the water heaters to be normal again, they all say, I am too cold. I have a high need for energy and they all effectively turn on. Do you know why there are two peaks? Because only half of the water heaters are allowed to turn on in the beginning and the other half get half an hour extra. If they didn't do that, you'd have double the peak. So it'd go above the page. And then what happens afterwards? So that's called a cold snap. That happens with water heaters, with air conditioners. If you turn air conditions off in the summer, you have, it's a hot snap. But they'll just consume a lot of energy. So people's need for energy is real. And so when you encode that need for energy. So if you look at electric heaters. And so at the university, we wanted to ask the question, how many electric water heaters do I need to generate a megawatt of flexibility? Which means these water heaters can then go up and down. And then I wanted to ask, how much flexibility do I then get per water heater? And then what we see is that the flexibility per device. So this is like an average flexibility for a whole fleet of devices. At nighttime, when no one uses a hot water heater. Right, once I use a little bit of flexibility, if you don't use it, then I have to sit and wait for that water heater to heat back up again. And then I can use it again. And so if no one is using the water heater, I should have less flexibility than when people use the hot water heater. Because extracting water from it means it heats up. Which means I can actually turn it off for a little bit of time and turn it back on again. At nighttime, if I heat the water up and no one is using it, just hot water sits then I can't really heat it anymore. I can't boil the water in the water heater. It's not good. And so what's really interesting is with water heaters the flexibility you have changes with the time of day. It makes sense. What about batteries, right? If I have a certain grid service I want to deliver. So here's one megawatt plus minus a megawatt. I want to deliver this, you see the kind of the thick gray signals. I want to kind of match that gray signals. Because I can make money if I do a good job of matching up the black with the gray. And so with 500 batteries it's really hard to deliver a megawatt accurately. But the more batteries I have, the more accurately I can deliver this service and the more revenue I can generate. Of course with a million batteries it's very easy to deliver a megawatt. But it's also more expensive because you're investing in those batteries. So the more devices you have, the better. The longer you're delivering this flexibility, the longer you deliver this flexibility. So if you're going to participate in one hour or two hour chunks or three hour chunks, the less flexibility you can offer. And so what you have to do is you have to manage how much you want to bid into the market, how much risk you want to take. The more greedy you get the less you have. If you want to account for people's comfort. And so then what we did was we developed algorithms that could manage people's comfort. The algorithms we developed and were funded by the Department of Energy was called packetized energy management. And the idea was inspired by the internet. So when I'm streaming my voice on the internet, right? The sound audio wave doesn't make across the internet. It gets chopped up into small data packets that then gets shipped across a complicated network of routers using random access protocol. And so you packetize the data into small bits. You send the data across the internet and it's then managed through the network with random access protocols. That's basically how the internet works is chop things up into small pieces and ship them across the internet. When they all arrive at the end of their source they get put together again and then you hear my audio. And so what we wanted to do was very similar. The heating of water in a water heater. When it's cold, when it's hot. When you heat water up, you basically have this big chunk. It can take 15, 20 minutes to heat up a water heater. Maybe 30 minutes. That's a big bulky packet. What if I could do that instead of 30 minutes I could do it into small six minute chunks. Of course there's no flexibility yet. But I don't have to deliver these small chunks consecutively, right? I could space them apart in some smart way. If I can space them apart you still get the energy you need over a slightly longer duration. And so by deferring these packets just a slight fit, right? By having these gaps in here I can put other devices packets in there, right? I can not necessarily schedule them but I can make room for them. And suddenly by executing when these packets are being delivered to the device or when the device consumes those packets I can suddenly manage a fleet of resources. And so we implement it and test it and put them into the field, these algorithms, for air conditioners, batteries, electric vehicles and water heaters. And what happens with one device is not very interesting. But when you have lots of devices what you see is that we can use this idea of packetization to have demand meet the supply rather than supply meet the demand. And so these two things are not perfect. A mismatch between demand and supply is not a good thing. What we would love to do is when the sun shines and the wind blows we wanna absorb that supply by shifting demand. And so we packetize and we randomize with a little bit of control. Kinda like how the internet works. And so we can packetize that demand and then we can shift things in some way in time. And by shifting the packets around smartly which is what the control algorithms do we can match demand to supply. And we can do it in basically real time. And so what was really cool, the work we did at UVM was really that we could solve this super hard scheduling problem. If you were to schedule each of those packets to match under this curve, that's a very difficult problem. In theory, it's a very hard problem in practice because this is me, like when I move this packet from here under the red curve it's basically me asking you to change your shower schedule. That's what it would mean if I were to schedule it. But if you shower and I just shift your packet by a minute or 30 seconds I can shift things around elsewhere. And so we implemented a nice simulator. We went into the field, had about 150 homes in Vermont. Together with Velco, Vermont Electric Co-op, Green Mountain Power, BED, implemented this, tested the algorithms in the field and showed the Department of Energy that we could do it. Which is pretty cool. Then we, Department of Energy came back and said, nice job. Can you do it faster? So this is, what is how much is this? This is two hours but a relatively slow moving blue signal. So what we did was, we got two years. They doubled the funding for us and we then showed that we could not only track, we could not only provide flexibility on a minute by minute timescale but we could do it on a second by second timescale. And so these are actual tests of about 200 homes across Carolinas and Vermont. And we showed that we could deliver this. And basically, can you track this red line is the question. And you can measure performance through what's called a composite score. Perfect score is one. You could pay it if you're above 0.7. A thermal power plant is roughly around 80 to 90%. So 0.8, 0.9. We outperformed a thermal power plant with our fleet of 200 devices in different homes across the country. In real time, which is pretty cool. The only system that beats this virtual battery is a big megawatt scale physical battery. Like a big battery will outperform our virtual battery. And so we spent, we spent, this took years to get to this point but it's super cool to know that this technology we built in our labs with our student and spilled it into the state of Vermont and across the country outperforms thermal generators in providing flexibility. There are still lots of interesting issues. When you do this flexibility, you need to model how people use energy. I don't need to model how you shower. I need to model how people statistically shower. And so what that means is that people behave differently which means that this battery is not a deterministic resource. It's not something that I know I can measure. It's not a single entity. It's not a physical box. I can't measure it. And so it's some kind of stochastic entity. There's uncertainty in it. And if you model and control it, you can certainly do that but at some point you have to dispatch it. And dispatching uncertain resources is a really fun technical problem. So we've been working a lot on that. So what happens if you have a battery where you don't know the state of charge and you don't know how big it is? Exactly. And so how do those errors propagate? And so we saw the climate disasters that are happening. We saw the need for flexibility to integrate renewable. And so what we believe in our team and our group is that if we can make intelligent electrification work, if we can make flexibility work, demand response work, it'll matter a lot because it's gonna help move us to work a clean and affordable energy future. And so we see this as climate technology, what we're working on. We have some recent articles on the topic. So intelligent electrification is kind of a term we're kind of taking hold of and trying to espouse as much as possible within our community because we think it's the enabler of decarbonization. And so I've worked in this area for a while. Traditional demand response is kind of this big hammer where you shut things off for a certain amount of time. I think we're moving to more, it's a much more autonomous, a much more nimble system. And so the grid is really gonna move from 20th century into a brave new world. And so it's exciting to be part of that. So thank you so much. That's the part of it. You know, whether you've got 5G networks and we're making this a second reason to control, you know, doing 200,000 to scale it up to 2 million, 200 million. Yeah, that's a very good question. So one of the open questions is really how can we enable connectivity across large geographical ranges and really efficiently, really fast at communication? To do that. I mean, my phone right now is pretty quick at communicating with anyone. And so there's lots of connectivity opportunities. There's a lack of standards in this space. So the internet works because of standards, right? The internet works because every device knows how to communicate with each other. When I send that SMS to anyone in this room, you'll get it super quickly because there are standards for that. There are very few standards today for demand response or for flexible demand. And so that means that every single developer or OEM has their own implementation and version of how to communicate with each other. As we clean up the standards, we will leverage the connectivity backbone that we have already. I think Vermont is in a great position. Fiber optic communication networks is something that we pride ourselves on. Something we need to leverage and use more often and spread more. I would love to have South Burlington have a fiber optic option. Okay, why can't we use heat pump concept to gather and multiply solar energy on cloudy days? We can. So heat pumps, did I put, did I not have them in here? So heat pumps absolutely a part of the solution. And so basically anything that stores energy whether it's thermal or chemical will be part of a solution like this, front row. Yeah, I was just curious about how you're communicating with the individual devices. Do you require a dedicated internet connection to each one or can you overlay a frequency on the power grid in a higher frequency one? So what we did in the past was we leveraged either cellular communication or piggyback on the person's wifi signal. The amount of power or sorry, amount of communication requirements is super low. We're talking about kilobytes of data and today's internet connections are megabits of data. So there's a big gap. So public wifi works, racist concerns about security, cellular networks are more secure or more expensive. Most people don't know that AT&T and other carriers have special emergency network bandwidth set aside. So there's been lots of discussion of whether you can tap into that secure bandwidth to deliver these energy services. But yeah, we have not used power line carrier communications. There are some companies that do that. There are a few companies that do that. It just at some point you have to leave the house. I noticed that a lot of your research was obviously in the last three, four years. Did the Canadian wildfires affect how much solar energy you could get? So we don't have data for that because our solar arrays is too young. But we did hear from industry folks that it was noticeable. So I think a 10% reduction in solar output from the wildfires. So climate matters locally and globally. In the back. Where it's being generated on a sunny day versus where it's not being generated on a cloudy day as far as the Midwest and New England, whatever. I was introduced to direct current transmission when I was on honeymoon in the 80s in New Zealand where they had a huge dam producing huge power in a place where they didn't need it. And they put it in the ground in a six foot hole and took it to the North Island with DC power and with no line loss and no big swaths of overhead lines. Is that part of this plan or will your packetization take care of that? So no single solution will solve this problem. What we need is effectively in the kitchen sink. We need everything. So your high voltage DC terminals that you may have visited. So Highgate has one of those high voltage DCs. So we import power from Hydro-Quebec through high voltage DC lines, which is super neat because you can control a lot of that power flow in the Vermont network. And so controlling power flow on networks is a really important source of resilience, of reliability. And so absolutely going forward we will need a strategy on high voltage DC across the country. Vermont can't solve that problem on its own either. And so you're correct. If Vermont has a cloudy day, won't be a lot of solar. And you can't packetize for a whole day. I can't turn off your water heater for a whole day. And so the flexibility I talked about here was really on the minutely timescale, the intra-hourly timescale. So when clouds come and go, we can manage that. Batteries can manage that. But regionally you would need some kind of transportation power, power flow between states. And so ISO in England is doing that today. Going forward there needs to be a federal incentive and strategy around high voltage DC to connect different sources of wind and solar. And so that opens the question of transmission, constructing transmission, improving transmission infrastructures, which is as complicated as nuclear power. Sorry. That's okay. I have a couple of questions. One is I'm wondering what happens to the system if you lose the communications channel? And the second question is it seems I might be losing some flexibility here. And so if my relative visits and I need to have the heat higher, am I still gonna be able to do that? Yep. So on communications, so when we were developing this technology ourselves, we had logic in our devices that said if you lose communications, revert back to normal operations. And so a water heater is meant to heat water in a certain way. When you lose communications, basically what happens is the device exits the game and does its own thing as it always has done. So you wouldn't notice and you would still have the hot water you need. We would lose flexibility. And so as long as we didn't lose everyone at the same time, it's fine because distributed solutions are resilient to individual losses. Does that make sense? Like if I lose this one device, I still have this other fleet. And so these edge cases would be fine. But at scale it's a problem. So a cyber attack across the state, you would lose the entire fleet. You'll still have hot water but the company managing those resources wouldn't be managing them. Which is a problem. So cyber security is a concern. In theory, I think in practice there are lots of good things you can do to mitigate those concerns. You can't eliminate them but you can mitigate them. Yes. Yeah, I'm sure you're probably very familiar with S5, the Affordable Heat Act that's been recently passed. I know you said you're not a politician so I won't go there. Save myself. And that's coming right around the corner. I think 2025 they're supposed to have the final go-no-go on that. What's gonna have to happen to our grid to accommodate that? Because that seems like that's gonna be quite a change and increase load on our grid. So I can speak to the experiences I've had with those exact challenges. And so living in South Burlington in an older house with a hundred amp panel, I tried really hard to get an electric water heater, a heat pump water heater. I could not. I really tried. Super expensive. It's not worth spending $10,000 just so that I can actually buy another water heater. And so I ended up going with a more efficient natural gas. And so that's a real challenge. So if you're gonna make policies that says I can't have a natural gas generator, then you gotta pay for my upgrades on the panel. That's my personal opinion. The other question you had was, the other question you had was, what was the question was, if the grid, how can the grid handle the new heat? Good, good. So the air conditioners, so we have air conditioners because we don't have central, we don't have central HVAC, we just have air conditioners in the window. Those air conditioners are not very energy, they're pretty big loads, they're pretty large loads. If you have central HVAC systems, those are compressor-based and compressor-based HVAC systems are energy-hungry appliances. Heat pumps are far more efficient on average. Heat pumps do have a backup resistive heating element that's non-trivial. But the average energy requirement of a heat pump is lower than your air conditioners and HVAC systems. And so we have not seen or heard of major concerns that electrifying heating necessarily will cause any problems. The real concern is EV chargers because every time you plug in that car, that's gonna be 11 kilowatts. Whereas here, you're really trading off relatively air window units, let me be pretty small. It's not a big difference between them and the heat pumps. So I think the EV chargers are much bigger problem than the heating component of electrification. Thank you so much, this was very interesting.